Sardinian Syntax (Romance Linguistics)

cover

next page >

Cover

title: author: publisher: isbn10 | asin: print isbn13: ebook isbn13: language: subject publication date: lcc: ddc: subject:

Sardinian Syntax Romance Linguistics Jones, Michael Allan. Taylor & Francis Routledge 0415049229 9780415049221 9780203993804 English Italian language--Dialects--Italy--Sardinia--Syntax, Italian language , Sardinia (Italy) 1993 PC1982.J66 1993eb 457/.9 Italian language--Dialects--Italy--Sardinia--Syntax, Italian language , Sardinia (Italy)

cover

next page >

< previous page

page_i

next page >

Page i Sardinian Syntax Sardinian Syntax presents the first comprehensive synchronic description of Sardinian syntax available in English. Michael Jones combines a detailed coverage of the language with a theoretical approach that draws on contemporary linguistic theory. The book provides not only an extensive reference grammar but an invaluable source of information for all linguists whose interests extend beyond the world’s major languages. Distinctive properties of Sardinian, such as the use of infinitives with nominative subjects, the prepositional accusative, the syntax of impersonal constructions, and the factors determining auxiliary choice, are investigated in such a way as to allow synchronic comparison with other Romance languages. The author makes a rare attempt to apply a theoretical framework to the comprehensive description of a language, and the result is a book that goes beyond standard grammatical description to make an original contribution to syntactic theory. Michael Allan Jones is Lecturer in Linguistics and French at the University of Essex. His work on the Sardinian language has been published in The Romance Languages edited by Martin Harris and Nigel Vincent.

< previous page

page_i

next page >

< previous page

page_ii

next page >

Page ii Other books in the Romance Linguistics series include: Structures and Transformations Christopher J.Pountain Studies in the Romance Verb eds Nigel Vincent and Martin Harris Weakening Processes in the History of Spanish Consonants Raymond Harris-Northall Spanish Word Formation M.F.Lang Tense and Text Dulcie Engel Variation and Change in French John Green and Wendy Ayres-Bennett Latin Syntax and Semantics Harm Pinkster Thematic Theory in Syntax Robin Clark Tense and Narrativity Suzanne Fleischman Comparative Constructions in Spanish and French Syntax Susan Price Latin and the Romance Languages in the Early Middle Ages Ed. Roger Wright The Rhaeto-Romance Languages John Haiman and Paola Benincà Tener+Past Participle Catherine E.Harre Also of interest: The Romance Languages Martin Harris and Nigel Vincent

< previous page

page_ii

next page >

< previous page

page_iii

next page >

page_iii

next page >

Page iii Sardinian Syntax Michael Allan Jones

LONDON AND NEW YORK

< previous page

< previous page

page_iv

next page >

Page iv First published in 1993 by Routledge 11 New Fetter Lane, London EC4P 4EE This edition published in the Taylor & Francis e-Library, 2005. To purchase your own copy of this or any of Taylor & Francis or Routledge’s collection of thousands of eBooks please go to www.eBookstore.tandf.co.uk. Simultaneously published in the USA and Canada by Routledge 29 West 35th Street, New York, NY 10001 © 1993 Michael Allan Jones All rights reserved. No part of this book may be reprinted or reproduced or utlilized in any form or by any electronic, mechanical, or other means, now known or hereafter invented, including photocopying and recording, or in any information storage or retrieval system, without permission in writing from the publishers. British Library Cataloguing in Publication Data a catalogue record for this title is available Library of Congress Cataloging in Publication Data applied for ISBN 0-203-99380-2 Master e-book ISBN ISBN 0-415-04922-9 (Print Edition)

< previous page

page_iv

next page >

< previous page

page_v

next page >

page_v

next page >

Page v For Laura and Catherine

< previous page

< previous page

page_vi

next page >

page_vi

next page >

Page vi This page intentionally left blank.

< previous page

< previous page

page_vii

next page >

Page vii Contents

Editorial statement Preface List of abbreviations

xi xii xiv

Introduction 0.1 Aims and methodology 0.2 Theoretical framework 0.3 The sociolinguistic status of Sardinian 0.4 Orthography 1 The sentence 1.1 Constituent order 1.1.1 Structure of the clause 1.1.2 Constituent-order variations 1.2 Propositional attitude 1.2.1 Negation 1.2.2 Questions 1.2.3 Imperatives and hortatives 2 The noun phrase 2.1 Components of the NP 2.1.1 The noun 2.1.2 Determiners and quantifiers 2.1.3 Numerals 2.1.4 Adnominal adjectives 2.1.5 Possessives 2.2 Structure of the NP 2.2.1 Preliminary assumptions 2.2.2 Prenominal elements 2.2.3 Postnominal elements

< previous page

1 1 3 5 8 13 13 13 16 21 21 24 26 31 31 31 34 39 41 44 45 45 46 50

page_vii

next page >

< previous page

page_viii

Page viii 2.2.4 Order of postnominal adjectives 2.2.5 NPs without determiners 2.2.6 Prepositional accusatives 2.2.7 Headless NPs 2.2.8 Extraposition of the head noun 3 The verb phrase 3.1 Morphosyntax of the verb 3.1.1 Verbal flexion 3.1.2 Perfective and progressive aspect 3.1.3 Periphrastic future and conditional constructions 3.2 Transitivity of the verb 3.2.1 Thematic roles and grammatical functions 3.2.2 Impersonal constructions 3.2.3 Unaccusative verbs 3.2.4 Existential uses of áere and éssere 3.2.5 Copular verbs 3.2.6 Pronominal verbs 3.2.7 Passive constructions 3.2.8 Pronominal voice 3.3 Auxiliary verbs 3.3.1 Choice of perfective auxiliary 3.3.2 Syntactic properties of perfective and progressive constructions 3.3.3 Modalverbs 3.3.4 The future use of áere 3.3.5 ‘Semi-auxiliaries’ 3.4 Structure of the VP 3.4.1 Auxiliaries and preverbal modifiers 3.4.2 Complements and postverbal modifiers 4 Minor predicate categories 4.1 Adjectives and adverbs 4.1.1 Morphology 4.1.2 Functional types 4.2 Degrees of comparison 4.2.1 Comparison of inequality 4.2.2 Superlative degree 4.2.3 Comparison of equality 4.3 Prepositions and adverbs of time and place 4.3.1 Basic uses of prepositions 4.3.2 Prepositions and their complements

< previous page

page_viii

next page > 52 56 65 69 76 80 80 80 82 89 94 94 100 106 113 114 119 124 127 130 130 137 142 146 149 154 154 158 165 165 165 168 173 173 178 179 180 180 185

next page >

< previous page

page_ix

Page ix 4.3.3 Location, goal and source 4.3.4 Temporal uses of prepositions 4.3.5 Deictic adverbs of place and time 5 Pronouns 5.1 Disjunctive pronouns 5.1.1 Personal pronouns 5.1.2 Demonstrative, interrogative and negative pronouns 5.1.3 Pronominal uses of quantifiers and indefinite expressions 5.2 Clitics 5.2.1 Forms and basic uses 5.2.2 Order and compatibility of clitics 5.2.3 Datives of interest 5.2.4 Ethic datives and locatives 5.3 Coreference and binding 5.3.1 Preliminary remarks 5.3.2 Binding properties of pronouns 5.3.3 Reciprocal and complementary relations 6 Subordinate clauses 6.1 Complement and adverbial clauses 6.1.1 Finite clauses 6.1.2 Mood 6.1.3 Infinitival clauses 6.1.4 Infinitival constructions with causative verbs 6.1.5 Imperfect subjunctives and inflected infinitives 6.1.6 Nominal infinitives 6.1.7 Participial clauses 6.1.8 Interrogative complements 6.2 Other subordinate clauses 6.2.1 Relative clauses 6.2.2 Free relatives and similar constructions 6.2.3 Conditional and concessive constructions 7 Constituent-order variations 7.1 Reordering processes 7.1.1 Left-dislocation 7.1.2 Right-dislocation 7.1.3 Inversion 7.1.4 NP fronting 7.1.5 Fronting of other constituents

< previous page

page_ix

next page > 189 193 195 199 199 199 203 208 213 213 218 221 230 239 239 240 245 247 247 247 251 260 270 278 282 285 291 293 293 300 305 312 312 312 318 327 332 338

next page >

< previous page

page_x

Page x 7.1.6 Fronting and other preposing phenomena 7.2 Stress, focus and related issues 7.2.1 Assignment of stress and focus 7.2.2 Focus and illocutionary force

Conclusion Notes Bibliography Index

< previous page

next page > 345 351 351 355 362 367 373 376

page_x

next page >

< previous page

page_xi

next page >

Page xi Romance Linguistics Editorial Statement Routledge publish the Romance Linguistics series under the editorship of Martin Harris (University of Manchester) and Nigel Vincent (University of Manchester). Romance Philology and General Linguistics have followed sometimes converging paths over the last century and a half. With the present series we wish to recognise and promote the mutual interaction of the two disciplines. The focus is deliberately wide, seeking to encompass not only work in the phonetics, phonology, morphology, syntax and lexis of the Romance languages, but also studies in the history of Romance linguistics and linguistic thought in the Romance cultural area. Some of the volumes will be devoted to particular aspects of individual languages, some will be comparative in nature; some will adopt a synchronic and some a diachronic slant; some will concentrate on linguistic structures, and some will investigate the sociocultural dimensions of language and language use in the Romance-speaking territories. Yet all will endorse the view that a General Linguistics that ignores the always rich and often unique data of Romance is as impoverished as a Romance Philology that turns its back on the insights of linguistics theory.

< previous page

page_xi

next page >

< previous page

page_xii

next page >

Page xii Preface The idea of writing a book on the syntax of Sardinian evolved as I was conducting research for my contribution to Martin Harris and Nigel Vincent’s collective volume on The Romance Languages (Jones 1988a). While much of the relevant information on the phonology, morphology and lexis of Sardinian could be gleaned from published sources, I quickly discovered that the syntax of Sardinian was relatively uncharted territory, which, moreover, concealed many treasures for the linguist. I therefore set about exploring this territory for myself. The fruits of this research soon outgrew the scope of the chapter for The Romance Languages and the project of a more extensive study devoted entirely to syntax was conceived. I would like to thank Martin Harris and Nigel Vincent for encouraging me to embark on this project and for their continuing support and assistance at every stage in its completion. The bulk of the fieldwork for this study was conducted during a period of sabbatical leave from the University of Essex in the Autumn of 1986 and during subsequent Summer vacations. In the course of these visits to Sardinia I had the pleasure of meeting many scholars and lay-specialists in Sardinian language and culture, notably Eduardo Blasco-Ferrar, Diego Corraine, Ines Loi Corvetto, Enzo Espa and Massimo Pittau. I thank them for their interest in my work, for the fruitful discussions we have shared and for their assistance in various other respects. I owe a very special debt to Serafino Spiggia and Dionigi Panedda who took me under their wing during my visits to Sardinia and who generously gave up their time to check through the thousands of examples on which this study is based. Their vigilance and their understanding of what I was trying to achieve undoubtedly saved me from a number of embarrassing errors. My other main informants, Maddalena Taras-Jones, Giuseppa Taras and Salvatore Taras, played a still more fundamental role since it is

< previous page

page_xii

next page >

< previous page

page_xiii

next page >

Page xiii through my personal relationship with them that I acquired my practical command of Sardinian, such as it is. I take this opportunity of thanking them for this and for their patience in answering what must have seemed endless tedious queries which I would not have dared to impose on anyone else. I would also like to thank the countless other friends and acquaintances in Sardinia who, sometimes unwittingly, acted as informants on a more casual basis. This book has evolved in a rather eclectic fashion involving the continuous interplay between detailed analysis of particular phenomena and the building up of a coherent picture of the grammar as a whole against the background of developments in syntactic theory. Undoubtedly, the most difficult task was that of bringing all the bits and pieces together and presenting them under reasonably sensible chapter headings. The final result is something of a compromise between the goals of explanatory adequacy and scientific rigour on the one hand and comprehensiveness and user-friendliness on the other. One of the least satisfactory aspects of this compromise is that I was unable to incorporate many of the important theoretical developments which took place during the period of this research, since to have done so would have entailed continuous revision of those parts of the research which were already completed. I hope nevertheless that this account approaches the syntax of Sardinian in a manner conducive to further investigation in the light of recent theoretical developments and, conversely, that scholars whose interest in Sardinian is not based on current generative theory will be able to benefit from this description without being daunted by the conceptual framework within which it is couched. Throughout this research I have benefited greatly from comments and criticisms by many of my fellow linguists. I am grateful to all those who provided me with feedback and who encouraged me in the belief that this enterprise was worthwhile, particularly (to mention but a few) Richie Kayne, Adrian Battye, Luigi Burzio, Andrew Radford, Luigi Rizzi and Nicolas Ruwet. While acknowledging my debt to all those who have helped me to bring this work to completion, I accept full responsibility for any errors or shortcomings which remain. Finally, on a more personal note, I would like to thank my parents and my wife Enam for their encouragement and moral support.

< previous page

page_xiii

next page >

< previous page Page xiv Abbreviations acc. adj. AG aux. Camp. com. CONJ dat. DET fam. f. imperf. ind. inf. It. LOC Log. Log.-Nuor. m. NEG nom. Nuor. obl. pers. perf. pol. pp. prog. refl. sg. subj. TH

< previous page

page_xiv

next page >

accusative adjective Agent auxiliary Campidanese comitative conjunction dative determiner familiar feminine imperfect indicative infinitive Italian Location Logudorese Logudorese-Nuorese masculine negative nominative Nuorese oblique person perfective polite past participle progressive reflexive singular subjunctive Theme

page_xiv

next page >

< previous page

page_1

next page >

Page 1 Introduction 0.1 AIMS AND METHODOLOGY The syntax of Sardinian is a rather neglected area of research. Most of the available information is to be found in general surveys of the language such as Porru (1811), Spano (1840), Wagner (1951), Pittau (1972), Atzori (1982), Blasco-Ferrer (1984, 1986) and Jones (1988a). Whereas there are a number of extensive studies of other aspects of Sardinian such as lexis (cf. Wagner 1960–4), morphology (cf. Wagner 1938–9) and phonetics and phonology (cf. Wagner 1941, 1984; Virdis 1978 and Contini 1985), there are no comparable studies devoted specifically to syntax. Moreover, while evidence from other Romance languages and dialects has had a profound influence on the development of syntactic theory, particularly within the generative paradigm, references to Sardinian data are conspicuous by their absence, presumably because the relevant facts are not widely known or have not been analysed in sufficient detail for their theoretical significance to be assessed. The principal aim of this study is to fill the gaps identified above, by providing a description of Sardinian syntax which is comprehensive enough to qualify as a reference grammar but which is formulated within a conceptual framework which allows individual facts to be studied as part of a coherent system. To a large extent, these objectives complement each other in so far as the adoption of an explicit framework imposes an attention to empirical details (e.g. the precise range of categories affected by a given phenomenon) which might be overlooked in an informal description and also provides a means of investigating the interaction between different syntactic processes. Nevertheless, there are also cases where the degree of abstraction necessary for a theoretical explanation of a particular phenomenon is superfluous to an adequate description of the data and, for the

< previous page

page_1

next page >

< previous page

page_2

next page >

Page 2 ‘uninitiated’ reader, may even obscure the generalisation which the theory is designed to capture. Moreover, in practice, theoretical research typically involves some degree of idealisation or selectivity which is incompatible with our aim of presenting a comprehensive description of Sardinian Syntax—whereas the theoretician can legitimately concentrate on a range of phenomena about which he has something interesting to say, our aim of comprehensivity requires us to deal with facts which we may not be able to explain in a theoretically insightful way. Bearing in mind these considerations, we have attempted to achieve a compromise between ‘explanatory adequacy’ and factual description, making use of theoretical concepts and formal representations in so far as they serve to elucidate the data under discussion while avoiding issues of a theory-internal nature. This type of analysis requires access to negative data (i.e. evidence of sentences or constructions which do not form part of the language). Consequently, we have relied heavily on the intuitive judgements of native speakers, our general strategy being to formulate initial hypotheses (based on our own observations, attested written examples, information provided in other grammars, etc.) and to test these hypotheses by eliciting informants’ judgements of appropriate constructed examples, revising and retesting our hypotheses as necessary. For this purpose, we selected a small group of five principal informants whom we were able to consult extensively on the whole range of phenomena investigated in this study. We also solicited judgements from a much wider range of speakers on a casual basis with regard to particular points of grammar of a more contentious nature (for example, cases where our principal informants diverged or hesitated in their judgements). A further resource which we have exploited is a corpus of written prose texts (see Bibliography) which we have used first, as a basis for formulating initial hypotheses, and second, as a means of checking the intuitive judgements of our informants against attested usage. The empirical evidence presented in our account consists mainly of constructed examples with the judgements elicited from our informants. As far as possible we have attempted to extrapolate a consensus from the intuitions of individual informants, indicating divergence by prefixing the relevant examples with a question mark and/or by appropriate comments in the text. Occasionally we have used attested written examples, mainly to illustrate matters of style or usage which require an authentic context or, in some cases, to corroborate or qualify the evidence provided by our informants.

< previous page

page_2

next page >

< previous page

page_3

next page >

Page 3 0.2 THEORETICAL FRAMEWORK Our approach to the syntax of Sardinian is a generative one in the sense that we assume that the infinite range of utterances which are possible in a given language is the product of a finite set of interacting rules and principles which constitute the speaker’s knowledge of the language. Within this perspective, the task of the linguist is to define this system of rules and principles in a manner consistent with observable data (and also with certain assumptions concerning the nature of the language faculty) rather than simply to illustrate or classify representative examples of the external manifestations of this system. Our study conforms to the spirit of this enterprise in that we have endeavoured to provide insights into the internalised grammatical system or, at least, to give sufficiently detailed empirical evidence to enable other linguists to gain such insights. The particular model which we have adopted as a basis for our description is a broad version of the Government-Binding theory presented in Chomsky (1981). We have used this model primarily as a heuristic framework rather than an object of inquiry in its own right and, in order to make our account intelligible to readers who may not be familiar with this framework, we have cut a number of theoretical corners. Thus, for example, although we have made use of empty categories as a means of representing ‘missing elements’ and have given an informal characterisation of the way in which they are interpreted, we have not systematically attempted to identify the type of empty category involved or the means by which it is licensed according to the theory. With regard to syntactic structure, our main requirement is a system of representation which allows us to indicate which elements of a sequence form constituents and to state generalisations regarding the order of elements within a given constituent. For these purposes, the framework of Chomsky (1981) is adequate. Specifically, we assume the following canonical schema for the structure of phrasal categories, where X stands for any of the major lexical categories: (1) a XP (SPECIFIER) X′ b X′ (MODIFIER) X′ (MODIFIER) c X′ X (COMPLEMENT) Starting from the bottom, the head X and its complement (if any) combine to form a category X′, the complement normally following the head in Sardinian, as shown in (1c). To this category, optional modifiers (e.g. adjectives, relative clauses, adverbs, etc.) may be

< previous page

page_3

next page >

< previous page

page_4

next page >

Page 4 adjoined to the right or left depending on their properties, as in (1b). Finally, the symbol XP represents the complete phrase or ‘maximal projection’, the position SPECIFIER being occupied by items such as determiners, quantifiers and degree elements. We take the schema in (1) as an idealised model which is subject to language-specific or construction-specific variation (i.e. we take it as a basis for the analysis of phrase structure rather than as a model which must be strictly adhered to). We follow Chomsky (1981) in treating sentences as a separate system (see Chapter 1 for discussion), in contrast to more recent proposals whereby sentences are analysed as phrasal projections of inflectional or tense features (for example, as in Chomsky (1986)). The amount of structural detail which we have given in our account depends largely on the extent to which it is useful for the purpose of description. Often the relevant facts can be adequately described in terms of linear order or fairly gross syntactic structure, in which case we have used representations (usually by means of labelled bracketing) which indicate only the major constituents which are relevant to the point under discussion. Where we have proposed more detailed structural analyses, our proposals should be taken as an attempt to capture certain generalisations within the framework outlined above, not as an implicit argument against alternative analyses which might be available within a more sophisticated model. In addition to surface structure, which corresponds to the form of sentences as they are actually attested, we assume an underlying level of representation in which elements occur in canonical positions determined by their semantic function within the sentence. Thus, to take a simple example, we postulate that the interrogative sentence (2a) has essentially the same underlying structure as the corresponding declarative sentence (2b), represented approximately in (2c), in which Juanne ‘John’ occurs in the canonical subject position and itte ‘what’ in the direct object position: (2) a Itte videt Juanne? ‘What does John see?’ b Juanne videt su libru. ‘John sees the book.’ c [s [NP Juanne] [Vp videt [NP itte]]] To derive the surface form, the interrogative item itte must be moved to the beginning of the clause and the subject Juanne must be placed in a postverbal position (see 1.2.2 and 7.1.3 for detailed discussion). It is generally assumed that the processes which map underlying

< previous page

page_4

next page >

< previous page

page_5

next page >

Page 5 structures on to surface structures are governed by various formal constraints. We shall not discuss these directly, but we have endeavoured to formulate our description in accordance with them. The schemata in (1) define a range of potential syntactic positions within a phrase. However the instantiation of particular positions is dependent on other factors, particularly properties of the head (X); for example, the COMPLEMENT position in (1c) can only be instantiated as an NP if X is a preposition or transitive verb. Within the framework assumed here, such dependencies are expressed (partially) in terms of θ-roles (or thematic roles) which characterise the semantic relation between a predicate and its argument(s) (e.g. ‘Agent’, ‘Patient’, etc.—see 3.2.1). Thus, the COMPLEMENT position in (1c) is available only if X is an item which assigns a θ-role to this position. In addition, overt NPs are subject to a syntactic licensing requirement which is formulated in terms of the notion of Case. Although overt Case inflection is restricted to certain pronouns in Sardinian (as in English), it is assumed that all overt argument NPs must bear a Case feature which is assigned by certain elements under particular structural conditions, the details of which we shall leave for discussion in relation to relevant data. The fundamental point is that if an NP is assigned a θrole but cannot receive a Case feature by virtue of its structural position, it must remain null (in which case its referential properties must be recovered from other elements in the sentence) or the NP itself or the Case-assigning item must move so that the requisite structural conditions on Case assignment are satisfied. 0.3 THE SOCIOLINGUISTIC STATUS OF SARDINIAN A question which we have avoided so far is whether our object of inquiry, the syntax of Sardinian, is sufficiently coherent as to be amenable to the type of investigation which we have outlined above. We do not wish to take a stance on the question of whether the dialects of Sardinian merit the status of an independent ‘language’, whatever this might mean, though we categorically reject the view that they are simply degenerate varieties of Italian. Nevertheless, there are a number of questions regarding the sociolinguistic status of Sardinian which are directly relevant to this study. The first concerns variation between the dialects of Sardinian. For the purposes of this study we define Sardinian as the family of dialects spoken in Sardinia except for Alghero (Catalan speaking), Calasetta and Carloforte (Ligurian), the area of Gallura along the northern

< previous page

page_5

next page >

< previous page

page_6

next page >

Page 6 coast where a variety of Southern Corsican is spoken and the town of Sassari where a hybrid dialect evolved in the Middle Ages as a result of close contact between the local population and the maritime powers of Pisa and Genoa. The dialects of Sardinian proper have also been influenced to a greater or lesser extent by other Romance languages and dialects (principally Pisan, Genovese, Catalan and Spanish and, more recently, Standard Italian) which have given rise to considerable regional differences, particularly between the Campidanese dialects spoken in the southern half of the island and the Logudorese-Nuorese dialects spoken in the northern-central areas. These differences are particularly striking in the domains of phonology, morphology and lexis. From a syntactic point of view Sardinian appears to be much more homogeneous, though this perception may be due in part to the fact that the syntax of Sardinian has not been investigated as extensively as other aspects of the language and to the more general fact that syntactic differences are less amenable to direct observation (for example, it is rather easier to document dialectal differences with regard to lenition of obstruents, the forms of the imperfect or the word for ‘butterfly’ than to investigate potential differences in the structural position of auxiliary verbs). Rather than attempting to formulate a description based on data covering the whole of the Sardinianspeaking community, we have concentrated on giving a detailed analysis of varieties which are generally recognised as conservative, principally from the central-eastern (Nuorese) area. Similarly, we have standardised the morphology, lexis and spelling of our examples in accordance with the usage and pronunciation of our principal informants, except in the case of examples taken from written sources which we have reproduced in their original form (see 0.4 for discussion of the orthographic conventions used in this study). In adopting this policy, we recognise that some of our claims may not be valid for all varieties of Sardinian. In cases where other descriptive studies reveal significant dialectal differences or present evidence which conflicts with our own observations, we have drawn attention to the relevant facts. However, the available evidence suggests that such differences are largely confined to matters of morphosyntax. Even if it should turn out that syntactic differences between dialects are more fundamental than we have supposed, we hope that this analysis will serve as a useful framework for the investigation of parameters of variation. A further issue is the influence of Standard Italian. Nowadays almost all speakers of Sardinian (including all our informants) are also fluent speakers of Italian, which they regard as one of their

< previous page

page_6

next page >

< previous page

page_7

next page >

Page 7 native languages. Indeed, in domains such as education, the media, administration and commerce, Italian is used exclusively, with the result that Sardinian is largely confined to informal conversation among friends and family. In view of this situation and the substantial similarities between Sardinian and Italian, some degree of interference between the two languages is to be expected. This effect can be seen in the emergence of a regional variety of Italian which incorporates many salient features of Sardinian and which is widespread even among Sardinians who do not speak Sardinian; see Loi-Corvetto (1983) for detailed description. With regard to the influence of modern Italian on Sardinian, we draw a distinction between genuine influences (features of Italian which have become part of accepted Sardinian usage) and cases where a speaker’s knowledge of Italian interferes with his intuitions or performance on a particular occasion. Since we wish our description to reflect contemporary usage, influences of the former type must be recognised (in the same way as the earlier influence of Catalan or Spanish), but the possibility of interference of the latter type raises important questions concerning the admissibility of certain types of evidence. This problem arises particularly in cases which rely on negative evidence. Consider, for example, a situation where positive evidence indicates that Sardinian has a construction X which corresponds to a different construction Y in Italian. Positive evidence is sufficient to establish that the two languages differ in so far as Sardinian allows X whereas Italian does not. Moreover, if informants systematically reject constructions of the type Y in Sardinian, we can conclude that construction Y does not form part of the grammar of Sardinian, assuming that the judgements of these informants are representative. The problem arises when judgements of examples of type Y in Sardinian vacillate (for example, the construction is accepted by a minority of informants, perhaps with some hesitation)—are we to take such judgements as evidence that construction Y is marginally possible in Sardinian or can they be attributed to interference from Italian? A similar problem occasionally arises with attested written examples, particularly in cases where the subject matter or the level of discourse demand a style which goes beyond the ‘normal’ resources of Sardinian as a medium for casual communication between friends or relations. To take an extreme example, Pira (1983:42) uses the phrase “s’issoro assimizzaresi e contriaresi”, literally ‘the their resembling each other and opposing each other’ (i.e. ‘the similarities and oppositions between them’). This phrase occurs within the context of a discussion of the nature of the linguistic sign, not the sort of

< previous page

page_7

next page >

< previous page

page_8

next page >

Page 8 discussion which would typically be conducted in Sardinian. The syntax of this example is quite clearly Italian; in Sardinian, possessive items (e.g. issoro ‘their’) always follow the head and clitic pronouns are never suffixed to infinitives, even in cases like this where the infinitive has a nominal function similar to that of the gerund in English (though, as a lexicographic convention, citation forms of pronominal verbs are sometimes given with the reflexive clitic suffixed to the infinitive—e.g. pentíresi ‘repent’—as in Spano (1852)). Although this example can be dismissed as a case of interference, there are other examples, involving more subtle aspects of grammar, where the choice is less clear. Later on the same page we find the example “limbar divessas usan paraular divessas” ‘different languages use different words’, which conflicts with the generalisation that indefinite NPs without a determiner do not normally occur in the preverbal subject position (see 2.2.5 for discussion). Here it is not clear whether the use of limbar divessas is simply an imitation of Italian diverse lingue or whether it constitutes a genuine exception to the generalisation in question. Having raised the problem of possible interference from Italian, we acknowledge that we have no principled strategy for dealing with it. As far as possible, we have attempted to ascertain the extent to which ‘aberrant’ data is representative by consulting a wider range of native-speakers. Sometimes, informants’ comments on other informants’ judgements have provided useful clues; for example, a comment like ‘Some people say things like that, but it is not real Sardinian’ suggests Italian influence which is resisted by more conservative speakers whereas a flat rejection of the type ‘No, that is pure Italian’ suggests that we are dealing with a case of interference. In presenting raw data we have endeavoured to reflect our informants’ judgements faithfully, but have exercised some discretion in extrapolating generalisations from this evidence. In particular, where examples which show Italian influence conflict with otherwise valid generalisations or with the predictions of our analysis, we have noted the relevant facts but have not attempted to modify our analysis in order to accommodate them. 0.4 ORTHOGRAPHY Sardinian has no generally accepted system of spelling, though numerous attempts have been made to establish one (see Spano 1840; Pittau 1978; Farina and Mingione n.d.). Since there is no variety of Sardinian which is acknowledged as a standard, individual writers

< previous page

page_8

next page >

< previous page

page_9

next page >

Page 9 tend to base their spelling on a particular dialect (usually their own). Second, whereas some writers adopt a phonetically-based approach, attempting to represent actual pronunciation within the limits of the standard alphabet, others aim at a more abstract, phonological representation which assigns a single spelling to a given word regardless of context and, in some cases, reflects etymology rather than current pronunciation. At a more trivial level, there is considerable variation in the choice of characters which are used to represent a given segment or phoneme, particularly with respect to consonants. The spelling conventions which we have adopted are a compromise between a phonemic representation of the dialect of our principal informants (namely the variety of Nuorese spoken in the Bitti-Lula area) and typographical convenience, with some concessions to orthographic traditions in so far as they exist. These conventions are outlined below, together with brief notes on some of the more common practices adopted by other writers, some of which will be encountered in written examples which we have cited—these are given in their original orthography and are distinguished from our own examples by the use of double quotation marks (both within the text and in numbered examples). The transcription of vowels poses no particular problems. Sardinian has a classic five-vowel system, with three degrees of aperture and a front-back distinction for high and mid vowels, and can thus be adequately represented by the characters a, e, i, o and u . Diphthongs are represented by sequences of vowels. Mid vowels are subject to a process of metaphony whereby they assume a mid-high pronunciation when the vowel of the following syllable is high—this effect is not reflected in our spelling. In Campidanese, mid vowels are systematically raised to /i/ and /u/ in final syllables and before other vowels and are transcribed as / and u in examples from Campidanese. There is also a tendency, which we have indicated wherever it occurs, for mid vowels to become high in pre-tonic position (e.g. cunténtu ‘content, happy’), leading to a change in the stem vowel in the paradigms of some verbs (e.g. issíre ‘to go out’ vs. ésso ‘I go out’). An acute accent over a vowel is used to indicate word stress when it falls on a syllable other than the penultimate, and in other cases where the position of stress is relevant to the discussion (as in the examples just given). We have also indicated word stress to distinguish cases where the second of two adjacent vowels constitutes a separate, stressed syllable (e.g. faínas ‘chores’, paúra ‘fear’) from those where the two vowels form a diphthong (e.g. áinu ‘donkey’, fáulas ‘lies’). Within the consonant system an important feature of Sardinian

< previous page

page_9

next page >

< previous page

page_10

next page >

Page 10 phonology is lenition of stops and fricatives in intervocalic position. Our general policy is to represent intervocalic consonants in their non-lenis forms. This accords with the pronunciation of the Bitti-Lula dialect, where lenition is restricted to voicing of fricatives and fricativisation of voiced stops. Thus, intervocalic s is pronounced [z], and d, g and b are pronounced respectively as [ð], [γ] and [v] (or [ß]), but other consonants have their normal phonetic value. However, in many dialects (including all varieties of Logudorese and Campidanese) voiceless stops become voiced fricatives, and voiced stops and fricatives are elided. Moreover, these effects are typically represented in spelling. Thus, corresponding to meta ‘much, many’, locu ‘place’, ape ‘bee’, pede ‘foot’, ovu ‘egg’, etc., we find the spellings meda, logu, abe, pe(e) (Camp. pei ), ou, etc. in Logudorese and Campidanese texts. In those cases where elision does occur in the dialect adopted here (e.g. caddu ‘horse’, jeo ‘I’, nue ‘cloud’), this is reflected in our spelling, even though the non-elided forms are attested in other Nuorese dialects (cabaddu, dego, nube). In most dialects, the etymological distinction between /v/ and /b/ has been neutralised in favour of the latter, but in the dialect adopted here it is generally maintained in initial position, though neutralised in other contexts (giving [v] intervocalically and [b] before or after a consonant other than /r/). Our spelling follows the pronunciation of this dialect, reflecting etymology in word-initial position, but not necessarily in other contexts: e.g. vénnere ‘come’, bíere ‘drink’, cumbín-kere ‘convince’, dévere ‘must’. Stops and fricatives derived historically from Latin geminates or clusters do not normally undergo lenition or elision in any dialect. Such cases are indicated in our spelling by gemination of the consonant. Voiceless stops transcribed as geminates are not phonetically geminate, and in the dialect presented here are identical to those represented by a single character (e.g. latte ‘milk’ and latu ‘flat’ are pronounced [late] and [latu])— in these cases, orthographic gemination is simply a means of indicating that the segment does not undergo lenition in other dialects (e.g. Log. [late] vs. [laðu]). Although this practice is quite common, it is not adopted by all writers. Other consonants transcribed here as geminates do have a characteristically long pronunciation, though in the case of stops and fricatives this is difficult to establish with certainty since the effects of lenition preclude direct comparison with corresponding simple consonants in the same context. Note that -dd- has a retroflex pronunciation (as opposed to simple d which is dental) and corresponds historically to Latin -LL- (e.g. nudda ‘nothing’). Also -bb- typically

< previous page

page_10

next page >

< previous page

page_11

next page >

Page 11 corresponds to Latin -QU- (e.g. abba ‘water’) (retained as /kw/ in Campidanese and transcribed accordingly: e.g. ácua) . The transcription of velar stops before front vowels is a rather contentious issue. In the Logudorese-Nuorese dialects, /k/ and /g/ retain their velar articulation before /i/ and /e/, whereas in Campidanese they become palatal affricates, as in Italian. In principle there is no reason why these phonemes should not be represented by the same characters (e.g. c and g ) regardless of context, as in Latin. In practice, the influence of Italian spelling conventions is so pervasive that written sequences of the type -ci-, -gi- are perceived as representing palatal affricates, with the result that in Logudorese-Nuorese texts the absence of palatalisation is indicated systematically by means of special characters (typically k or ch for /k/ and gh for /g/), whereas in Campidanese c and g are used for both velar and palatalised variants. Despite reservations, we have bowed to tradition on this matter, and use k and gh before front vowels but c and g elsewhere: e.g. kentu ‘hundred’, ghirare ‘return’, but cantare ‘sing’, gattu ‘cat’. In accordance with widespread practice, we use dz and tz to represent the alveolar affricates /dz/ and /ts/ (though some writers neutralise the voicing distinction by using z for both, often geminated to zz intervocalically). In addition, [dz] occurs as a variant of the palatal glide /j/, which we represent as j. Etymologically, /dz/ normally derives from Latin /lj/ (e.g. mudzere < MULIEREM ‘wife’) whereas /j/ corresponds to Latin /j/ or /dj/ (e.g. janna < JANUAM ‘door’, oje < HODIE ‘today’)—the Campidanese counterparts to these two phonemes are respectively /ll/ and the palatal affricate /(d3/, usually written as g followed by / or e, elided intervocalically (e.g. mulleri, genna, oi). We use the digraph th to represent a phoneme which is pronounced as a voiceless dental fricative [θ] by conservative Nuorese speakers but is gradually being supplanted by its Campidanese variant [ts]—Logudorese speakers have the variant [t], which does not undergo lenition and is typically written as tt. The palatal nasal /n/, which occurs mainly in loan words from Italian, is represented as gn as in Italian. Among the contextually determined phenomena which some writers have tried to encode, but which are not represented in our transcription, we draw attention to the following: elision of wordfinal /t/ before a consonant (usually indicated by an apostrophe); addition of an unstressed epenthetic vowel (identical to that of the preceding syllable) when a word ending in a consonant occurs in absolute final position; partial neutralisation of word-final /s/ and /r/,

< previous page

page_11

next page >

< previous page

page_12

next page >

Page 12 whereby /-s/ → [-r] (or [-1]) before a voiced consonant and /-r/ → [ [-s] before a voiceless consonant; modification of word-initial consonants according to the final segment of the preceding word (e.g. lenition or elision of initial stops or fricatives after a vowel). However, in accordance with established practice, we have indicated (by means of an apostrophe) elision of the final segment of certain function words before following vowels (for example, elision of the final vowel of determiners and clitic pronouns and of the final /n/ of the negative particle non: s’ómine ‘the man’, cuss’ ómine ‘this man’, l’ appo fattu ‘I did it’, no’est veru ‘it is not true’).

< previous page

page_12

next page >

< previous page

page_13

next page >

Page 13 1 The sentence 1.1 CONSTITUENT ORDER 1.1.1 Structure of the clause For the purposes of our discussion we assume that the underlying order of principal elements within the clause is subject-verbcomplement(s), with the verb and its complements (if any) forming a verb phrase (VP), whereas the subject noun phrase is external to this constituent, as represented in (1):

(1) The position of other elements such as adverbial expressions and auxiliary verbs will be discussed in subsequent chapters, as will the internal structure of NP, VP and other categories which occur within these phrases. Our claim that Sardinian has an underlying subject-verbcomplement order is based largely on the intuition that this is the canonical order of elements in declarative main clauses, as in (2): (2) a Su pitzinnu dormit. ‘The child sleeps.’ b Juanne leghet su libro. ‘John reads the book.’ c Sos pastores murghen sas berbekes. ‘The shepherds milk the sheep.’

< previous page

page_13

next page >

< previous page

page_14

next page >

Page 14 d Su mere at datu su dinari a su theraccu. ‘The master gave the money to the servant.’ Although alternative orders are possible (see 1.1.2 for a general overview and Chapter 7 for a more detailed analysis), they tend to be exploited for particular stylistic or discursive effects (such as focusing or emphasis of a constituent) and in some cases are subject to syntactic constraints which can best be accounted for in terms of deviation from the canonical structure in (1). Following Chomsky (1981) and our remarks in 0.2 (p. 4), we postulate two further refinements to the structure in (1), which we shall simply mention at this stage. First, we assume that specifications of tense and mood and person/number features, which are ultimately expressed by verb inflection, are encoded under a separate INFL node directly dependent on S (see Chomsky 1981), though in practice we shall only make use of the INFL node when it is directly relevant to the matter in hand—in other cases we shall use simplified representations of the type in (1) where the verb is given in its inflected form under the V node. Second, we postulate a higher projection of the S category, labelled S′, which includes a COMP position whose primary function is to accommodate ‘complementisers’ (such as the particle ki ‘that’ which typically introduces finite complement clauses) but also provides a natural site for various types of fronted items (cf. 1.2.2). Thus the full structure of the clause can be represented as in (3):

(3) Like most of the other Romance languages, with the notable exception of French, Sardinian is a null-subject (or ‘pro-drop’) language; that is, the subject of a finite clause can be omitted in circumstances where languages like English or French would require a subject pronoun: (4) a Dormo. ‘I sleep.’ b Leghet su libru. ‘He/she reads the book.’

< previous page

page_14

next page >

< previous page

page_15

next page >

Page 15 c Murghen sas berbekes. ‘They milk the sheep.’ We shall not take a stand here on the various theoretical approaches which have been proposed in the literature to account for the nullsubject phenomenon,1 but we shall retain two leading ideas which are common to these approaches. The first is the formal assumption that in sentences like those in (4) the subject NP node is present in the structure but is devoid of lexical content (i.e. it is an empty node, represented as ); e.g. (4a) has the structure (5):

(5) The second leading idea is that the null-subject phenomenon is related in some way to a rich system of person/number inflection on the verb (all Sardinian verbs have six distinct person/number endings in each finite tense and mood). Intuitively, either the verb inflection itself has a status akin to that of a subject pronoun, or it allows the empty subject to function as a personal pronoun by making the person/number features recoverable. Overt subject pronouns are typically only used when there is some element of emphasis or contrast, also in the third person to make gender distinctions which cannot be conveyed by verb inflection alone. When a subject pronoun is used, it must have specific (typically human) reference. In particular, there is no overt subject pronoun corresponding to impersonal ‘it’ in English—i.e. the subject position is always empty in cases such as (6): (6) a Proet. ‘It is raining.’ b Paret ki Maria est maláida. ‘It seems that Mary is sick.’ Complements of the verb and other categories within the VP cannot be omitted freely, but must be represented by a clitic pronoun:

< previous page

page_15

next page >

< previous page

page_16

next page >

Page 16 (7) a Juanne lu videt. ‘John sees him/it’ b Maria lis at datu su dinari. ‘Mary gave them the money.’ c Su pitzinnu nd’est cuntentu. ‘The child is happy with it.’ Clitics precede finite verbs and infinitives but follow imperatives and present participles (they cannot be attached to past participles or to bare infinitives governed by a modal auxiliary; see 3.3.2, 3.3.3). The circumstances under which disjunctive object pronouns are used instead of (or, in some cases, as well as) clitics are roughly the same as those which determine the use of subject pronouns, except that disjunctive pronouns can be used non-emphatically as complements of certain prepositions for which there is no corresponding clitic; see 5.1.1 for detailed discussion. 1.1.2 Constituent-order variations Variations on the canonical SVO order can be described in terms of three basic processes: dislocation, fronting and subject inversion. These phenomena will be discussed in detail in Chapter 7. Below we give an informal outline of their essential properties so that we can refer to them in discussion of other syntactic phenomena. Dislocated sentences are constructions in which an NP or PP is placed outside the core of the sentence with a resumptive pronominal element (usually a clitic) within the sentence. Dislocated phrases may be placed either before the sentence (left-dislocation) as in (8) or after it (right-dislocation) as in (9): (8) a Cussu libru, 1’appo lessu metas vias. [that book it I+have read many times] ‘That book, I have read it many times.’ b A Núgoro, non bi so mai andatu. [to Nuoro NEG there I+am never gone] ‘To Nuoro, I have never been there.’ (9) a L’appo lessu metas vias, cussu libru. ‘I have read it many times, that book.’ b Non bi so mai andatu, a Núgoro. ‘I have never been there, to Nuoro.’ Dislocated items are usually separated from the rest of the sentence by a pause (indicated here by a comma), though this is less systematic in cases of right-dislocation.

< previous page

page_16

next page >

< previous page

page_17

next page >

Page 17 Indefinite plural or non-count nouns which function as direct objects can be dislocated with the partitive clitic nde as the resumptive pronoun; e.g. the sentence Bimus meta abba ‘We drink a lot of water’ can be rephrased as in (10): (10) a (De) abba, nde bimus meta. ‘(Of) water, we drink a lot of it.’ b Nde bimus meta, de abba. ‘We drink a lot of it, of water.’ Note that the quantified noun (abba in (10)) must be introduced by the preposition de ‘of when rightdislocated (optionally when leftdislocated). Also, the quantifier (meta in (10)) is not dislocated along with the rest of the NP but remains in the direct-object position. The quantifier may be absent, in which case we get a ‘some…’ interpretation (‘any…’ in negative sentences) as in the following examples which correspond to Bimus abba ‘We drink water’. (11) a (De) abba, nde bimus. ‘(Of) water, we drink (some).’ b Nde bimus, de abba. ‘We drink some, of water.’ Apart from the constructions in (10)-(11), dislocated NPs are always definite. Subject NPs can also be dislocated, with the verb inflection (or the empty subject) functioning as the resumptive pronominal element: (12) a Maria, credío ki fit inoke. ‘Mary, I thought that she was here.’ b At accabbatu su travallu, Pretu. ‘He has finished the work, Peter.’ In (12a) we have chosen an example with an embedded subject to illustrate the effects of left-dislocation more clearly, but we assume that left-dislocation of the subject is also possible in simple sentences like (13) even though it does not affect word-order (for evidence in support of this assumption, see 7.1.1): (13) Pretu, at accabbatu su travallu. ‘Peter, he has finished the work.’ Fronted constructions superficially resemble left-dislocated constructions except that there is no resumptive pronoun and no pause after the fronted expression: (14) a Cussu libru appo lessu. [that book I+have read]

< previous page

page_17

next page >

< previous page

page_18

next page >

Page 18 b A Núgoro so andatu. [to Nuoro I+am gone] However, they have a very different interpretation. Whereas the function of left-dislocation is, roughly, to establish or confirm the dislocated element as the topic of the sentence (i.e. as the entity about which a statement is being made), fronted constituents always have a focus interpretation (expressing the part of the statement which is new or particularly noteworthy) similar to that of clefted elements in English; thus, the examples in (14) would translate as ‘It is that book that I have read’ and ‘It is to Nuoro that I went’. Fronting can apply to a wide variety of phrases and for this reason provides a useful diagnostic for constituenthood. In particular, fronting can apply to predicative categories, such as AP (adjective phrase) and infinitival or participial VPs dependent on an auxiliary, which cannot normally be clefted in English: (15) a Troppu grassu est. lit. ‘Too fat he is.’ b Dormire keljo. lit. ‘To sleep I want.’ c Andatu a Núgoro est. lit. ‘Gone to Nuoro he has.’ For arguments that such cases involve fronting of the AP or VP rather than postposition of the copula or auxiliary, see 7.1.5. Fronting of predicative categories such as AP or VP is particularly common in yes/no questions (see 1.2.2, 7.2.2) and also in answers to questions where in English one would typically give an elliptical reply (for example, (15b) is roughly equivalent to ‘Sleep!’ as a short answer to a question such as ‘What do you want to do tonight?’). Fronting also differs from left-dislocation in cases involving indefinite NPs. When an indefinite NP is fronted, the quantifier or deter-miner is moved along with the rest of the phrase (compare example (10a) above): (16) Meta abba amus bitu. lit. ‘A lot of water we have drunk.’ An important syntactic restriction on fronting (which again does not apply to left-dislocation) is that the preverbal subject position must be empty. Thus, if an expression is fronted, the subject must either remain unspecified (as in (14)-(16)) or must be placed in a postverbal position: (17) a *Troppu grassu Juanne est.

< previous page

page_18

next page >

< previous page

page_19

next page >

Page 19 b Troppu grassu est Juanne. ‘John is too fat.’ As we have seen (example (12b)), one way of achieving the latter effect is by means of right-dislocation of the subject. However, there are instances of postposed subjects which do not appear to be the result of dislocation. Typically, right-dislocated phrases are unstressed and refer to entities which are already fairly prominent in the discourse whereas corresponding phrases in final position in nondislocated constructions typically bear main stress and contribute new information. Thus, (18a) might be appropriate in a conversation about the whereabouts of the builder, whereas (18b) would be more natural as part of a report on the people whom I met while I was out: (18) a L’appo vistu, su mastru de muru. b Appo vistu su mastru de muru. ‘I have seen the builder.’ Note now that examples like (19) have both prosodic patterns and their associated discourse interpretations: (19) At telefonatu su mastru de muru. [has telephoned the builder] With stress on telefonatu and a pause (not indicated here) before su mastru de muru, (19) would be an appropriate answer to a question such as ‘Has the builder been in touch?’, whereas with no pause and with main stress on su mastru de muru a more suitable context would be ‘Did anyone telephone while I was out? ’ We postulate that on the first reading, (19) is an instance of right-dislocation of the subject, but on the second interpretation it is the result of an inversion process which places the subject in a postverbal position similar to that of a direct object (recall that dislocated items are placed outside the core of the sentence). The difference between inversion and right-dislocation of the subject is not always absolutely clear since, in many cases, they yield the same linear sequences and can only be distinguished by reference to prosody or context.2 This makes it rather difficult to determine or illustrate the particular syntactic properties of each process simply by means of acceptability judgements of constructed examples. Nevertheless, there are at least two syntactic differences between these two processes which we shall simply state here pending a more detailed discussion in 7.1.3. First, inversion, but not right-dislocation, appears to be inhibited by the presence of a postverbal complement, particularly a direct object. The second difference is that inversion

< previous page

page_19

next page >

< previous page

page_20

next page >

Page 20 can apply to both definite and indefinite subjects, whereas dislocation is possible only with definite subjects. This restriction on dislocation is to be expected in so far as null subjects have the same status as definite pronouns. With regard to inversion, we should note that although inversion of an indefinite subject is possible, as in (20a), an impersonal construction of the type in (20b) is often preferred in such cases: (20) a An telefonatu metas pessones. [have telephoned many people] b B’at telefonatu metas pessones. [there has telephoned many people] ‘Many people have telephoned.’ Whereas in (20a) the finite verb an is third person plural in agreement with the inverted subject metas pessones, the verb in the impersonal construction (20b) is always third person singular and must be accompanied by the clitic bi (for detailed discussion see 3.2.2). With regard to the subject-inversion phenomenon, there is a subclass of intransitive verbs which deserves special mention; namely, verbs whose subject is a Theme (or Patient) and which select éssere ‘be’ rather than áere ‘have’ as their perfective auxiliary. In some recent analyses of related languages such as Italian and Spanish (see for example Jaeggli 1982, Rizzi 1982, Burzio 1986), it has been argued that the subject of such verbs (which we shall refer to as ‘unaccusative’ verbs, though some linguists use the term ‘ergative’) originates in the postverbal position normally occupied by the direct object NP and is moved, optionally, into the preverbal subject position in essentially the same way as the underlying object in passive sentences, whereas with ‘normal’ intransitive verbs of the telefonare type which take an agentive subject, the subject is base-generated in the preverbal position, the inversion construction being derived by a transformation which adjoins the subject to the VP. Thus, adopting this approach for Sardinian, (19) above, repeated as (21a), would be derived from (21b) whereas (22a) would be derived from (22b): (21) a At telefonatu su mastru de muru. b Su mastru de muru at telefonatu. ‘The builder has telephoned.’ (22) a Su mastru de muru est arrivatu. b Est arrivatu su mastru de muru. ‘The builder has arrived.’ (lit. ‘…is arrived.’) The facts of Sardinian seem to be generally consistent with this

< previous page

page_20

next page >

< previous page

page_21

next page >

Page 21 approach, though our informants’ judgements about some of the examples which might positively support this analysis are rather uncertain. Some of the implications of this approach are explored in 3.2.3. 1.2 PROPOSITIONAL ATTITUDE 1.2.1 Negation Sentences are negated by placing the negative particle non before the verb. The distribution of non is very similar to that of preverbal clitic pronouns; i.e. it can occur before finite verb forms and before infinitives (except ‘bare’ infinitives governed by a modal auxiliary or causative verb; cf. 3.3.3, 6.1.4) but not before participles or imperatives (for negation of imperatives, see 1.2.3): (23) a Custu vinu non m’aggradat. ‘This wine does not please me.’ b Provo a non rúghere. ‘I try not to fall.’ c Non keljo issire. ‘I do not want to go out.’ d Non so ascurtande. ‘I am not listening.’ e No’est arrivatu. ‘He has not arrived.’ (24) a *Keljo no’issire. cf. (23c) b *So no’ ascurtande. cf. (23d) c *Est no’ arrivatu. cf. (23e) Unlike clitic pronouns, however, non can be used without a verb in elliptical constructions such as (25) :3 (25) a Appo comporatu sa petha ma non su pane. ‘I have bought the meat but not the bread.’ b Keljo dormire, non travallare. ‘I want to sleep, not to work.’ Particularly characteristic of Sardinian is the use of non with de+ noun in constructions of the following type: (26) a Non de abba, abbardente est. [not of water brandy it+is] ‘It is not water, but brandy.’

< previous page

page_21

next page >

< previous page

page_22

next page >

Page 22 b Non de frates, paren nimicos. [not of brothers they+seem enemies] ‘Far from being brothers, they are like enemies.’ When the negative item is used on its own (i.e. as the equivalent of English ‘no’) it is realised as no or, for greater emphasis, nono (there is also a polite form nossi, an abbreviation of no signore ‘no sir’). The corresponding affirmative items are emmo, eja ‘yes’ (polite form sissi) and for emphasis ello ‘yes indeed, of course’. The above items can be used in reported speech preceded by the complementiser ki or, in some dialects, ca: (27) T’appo natu ‘I told you yes/no.’ They can also be used to affirm or negate a preceding phrase, as in (28): (28) Isse travallat una die emmo, s’áttera nono. [he works one day yes the other no] ‘Some days he works, others he does not.’ The particle non is also used with negative pronouns and adverbs such as those listed in (29): (29) neune ‘nobody’, nudda ‘nothing’, neddue ‘nowhere’, mai ‘never’, ne… ne ‘neither…nor’, prus ‘no longer’, mancu ‘not even’, perunu, -a ‘none, not any’. (30) a Juanne non pessat a neune. ‘John does not think about anyone.’ b No’appo comporatu nudda. ‘I have not bought anything.’ c Non semus andatos a neddue. ‘We did not go anywhere.’ d No’appo mai vistu una cosa gai. ‘I have never seen a thing like that.’ e No’amus mandicatu ne petha ne casu. ‘We ate neither meat nor cheese.’ f Non lu faco prus. ‘I will not do it any more.’ g Non faeddan mancu su sardu. ‘They do not speak even Sardinian.’ h Non b’ at perículu perunu. There is no danger.’

< previous page

page_22

next page >

< previous page

page_23

next page >

Page 23 However, non does not occur when the negative pronoun or adverb precedes the verb (i.e. as a preverbal subject or a fronted item): (31) a Neune est vénnitu. ‘Nobody came.’ b Mai appo vistu una cosa gai. ‘Never have I seen a thing like that.’ c Mancu su sardu faeddan. ‘Not even Sardinian do they speak.’ On the other hand, non must be present when a negative pronoun occurs as an inverted subject: (32) No’est vénnitu neune. ‘Nobody came.’ The item prus ‘no longer’ differs from the other items discussed above in that it cannot be used alone outside the scope of non, but must be preceded by mai ‘never’ in such cases: (33) Mai prus faco cussu. ‘Never again will I do that.’ It is possible for a simple sentence to contain more than one negative pronoun or adverb: (34) No’appo mai datu nudda a neune. ‘I have never given anything to anyone.’ When one negative item precedes the verb with other negative items following it, non may be present or absent, but with a difference in meaning: (35) a Neune at mai peccatu. ‘Nobody has ever sinned.’ b Neune no’ at mai peccatu. ‘Nobody has never sinned.’ When non is omitted, subsequent negative items are interpreted as being within the scope of the first whereas when non is present, the first item is negated independently of the rest. In cases of the second type, a periphrastic construction such as (36), in which each negative item occurs within its own clause, is generally preferred: (36) Non b’ at neune ki no’ at mai peccatu. There is nobody who has never sinned.’ We may also note the following negative constructions with kene ‘without’:4

< previous page

page_23

next page >

< previous page

page_24

next page >

Page 24 (37)

a Custa tsíkkera est kene lavata. [This cup is without washed] ‘This cup has not been washed.’ b So kene abba. [I+am without water] ‘I have no water.’ c So issitu kene tuncare sa janna. ‘I went out without closing the door.’ Adverbs and pronouns interpreted as variables within the scope of kene are realised as the negative items listed in (29): kene faeddare a neune ‘without speaking to anyone’, kene nudda ‘without anything’, kene salutare mancu su babbu ‘without greeting even his father’, kene dormire mai ‘without ever sleeping’, kene pane ne abba ‘without bread or water’ (note that ne ‘nor’ does not occur before the first conjunct governed by kene ). 1.2.2 Questions Yes/no questions can be formed simply by means of intonation; essentially by failing to produce a full pitch descent on the syllable which bears main sentence stress. As we noted in 1.1.2, fronting of a constituent (especially a predicative element such as a verb or adjective) is particularly common in yes/no questions, though it is not specifically a question-formation process. Thus, sentences like those in (38) would typically be interpreted as questions—‘Are you happy?’, ‘Have you eaten?’, ‘Has John telephoned?’ (see 7.2.2): (38) a Cuntentu ses. [happy you+are] b Mandicatu as. [eaten you+have] c Telefonatu at Juanne. [telephoned has John] Sardinian has an interrogative particle a which can be prefixed to yes/no questions under certain conditions: (39) a A mi vattus cudda tassa? [a me you+fetch that glass] ‘Will you fetch me that glass?’ b A bi venit Juanne? [a there comes John] ‘Is John coming?’

< previous page

page_24

next page >

< previous page

page_25

next page >

Page 25 This particle is used predominantly, but not exclusively, in questions which are to be interpreted as requests (as in (39a)), invitations, offers, etc.: (40) a A keres vénnere a domo mea? ‘Do you want to come to my house?’ b A ti lavo cussos prattos? ‘Should I wash those plates for you?’ We postulate that interrogative a is a complementiser, in the sense that it occurs under the COMP node in (3) above, though it can only introduce main clauses. An important restriction on constructions with this particle is that the preverbal subject position must be empty; i.e. if the subject is overtly specified, it must be postposed as in (39b) above: (41) *A Juanne bi venit? [a John there comes] cf. (39b) Also, when a is present, no item may be fronted:5 (42) *A telefonatu at Juanne cf. (38c) Moreover, interrogative a cannot co-occur with negation: (43) *A no’est arrivatu Juanne? [a NEG is arrived John] ‘Has John not arrived?’ Questions which request a specific piece of information are normally introduced by the appropriate interrogative adverb or pronoun (e.g. one of the items in (44)) or a phrase containing such an item: (44) cale ‘which’, cando ‘when’, cantu ‘how much, how many’, comente ‘how’, itte ‘what’, kie ‘who’, proitte ‘why’, ube ‘where’. (45)a Cando sun arrivatos? ‘When did they arrive?’ b Kin kie est issitu Juanne? ‘With whom did John go out?’ c Itte as fattu? ‘What did you do?’ We assume that in underlying structure the interrogative expression occupies the position which is relevant to its semantic role within the sentence (i.e. the position occupied by the phrase which supplies the requested information in an appropriate reply) and is moved to

< previous page

page_25

next page >

< previous page

page_26

next page >

Page 26 the COMP position, as indicated in the derivation of (45c) given in (46): (46) Interrogative expressions may remain in their underlying position, as in (47), but such sentences can only be interpreted as ‘echo questions’ or expressions of surprise, as in English: (47) a Sun arrivatos cando? ‘They arrived when?’ b Juanne est issitu kin kie? ‘John went out with whom?’ c As fattu itte? ‘You did what?’ Movement of the interrogative expression to the COMP position, like fronting of a non-interrogative item (see example (17a) and the use of interrogative a (see example (41) above), precludes the presence of an overt preverbal subject—compare the following example with (45b) and (47b): *Kin kie Juanne est issitu. Also, the three phenomena just mentioned are mutually incompatible; in particular, movement of an interrogative expression precludes both fronting of another constituent and the use of a: (48) a *Cando telefonatu at Juanne? cf. (38c) [when telephoned has John] b *Proitte a bi venit Juanne? cf. (39b), (41) [why a there comes John] 1.2.3 Imperatives and hortatives Positive commands are expressed by means of the imperative verb forms, which, in regular cases, are identical to the second person forms of the present indicative except that the final -s is omitted;6 e.g. Mandica! (sg.), Mandicate! (pl.) ‘Eat!’, vs. Mandicas (sg.), Mandicates (pl.) ‘You eat’; for detailed discussion of verb inflection, see 3.1.1: (49) a Veni a inoke! ‘Come here!’ (singular) b Ascurtate su ki so nande! ‘Listen to what I am saying!’ (plural) The third person singular of the present subjunctive may be used as a

< previous page

page_26

next page >

< previous page

page_27

next page >

Page 27 polite imperative form when addressing persons of higher social status (i.e. those who would be addressed as vosté, see p. 200): (50) Mandiket kin nois! ‘Eat with us!’ Genuine imperative forms like those in (49) cannot be negated. In negative commands the second person forms of the present subjunctive are used: (51) a Non vendzas a inoke! ‘Do not come here!’ (singular) b No’ascurtetas! ‘Do not listen!’ (plural) The present subjunctive is also used affirmatively in various types of hortative or optative utterances: (52) a Deus m’assestat! ‘God help me!’ b Bon proe ti facat! ‘Much good may it do you!’ Such sentences can be introduced (optionally) by the complementisers ki ‘that’, or si ‘if or by the interjection ancu: 7 (53)

ti falet unu lampu! ‘May lightning strike you!’ Inversion of the subject is usual in such constructions, but not obligatory. Also, constituents can be fronted, with the complementiser ki remaining in place: (54) Unu lampu ki ti falet! The co-occurrence of the complementiser with a fronted item in such examples is perhaps related to the similar phenomenon found in exclamations such as the following: (55) a Itte bellu ki ses! ‘How beautiful you are!’ b Itte burdellu k’ an fattu! ‘What a noise they made!’ Exhortations of the ‘let us …’ type can be expressed by the first person plural of the present subjunctive or indicative:

< previous page

page_27

next page >

< previous page

page_28

next page >

Page 28 (56)

a Andemas a mare! (subjunctive) b Andamus a mare! (indicative) ‘Let us go to the sea!’ The use of the subjunctive in such cases is characteristic of the more conservative dialects. Although indicative expressions such as (56b) might be analysed syntactically as straightforward statements or questions—‘We shall go to the sea’, ‘Shall we go to the sea?’—which are interpreted pragmatically in the same way as genuine first person imperatives, evidence concerning the position of clitics (see below, examples (60)) suggests that they can also have a distinct imperative status. In positive, second person imperatives of the type in (49), clitics always follow the verb:8 (57) a Naramílu! Tell it to me!’ (singular) b Abbaitátebos! ‘Look at yourselves!’ On the other hand, clitics always precede the verb in negative imperatives and, as our earlier examples show, optative/hortative sentences of the type in (52)–(53): (58) a Non mi lu nies! ‘Do not tell it to me!’ (singular) b Non bos abbaitetas! ‘Do not look at yourselves!’ With the other cases discussed above (polite imperatives (50) and first person imperatives (56)), the facts are less clear. Blasco-Ferrer (1984:254) cites the following attested examples from Campidanese of encliticisation to polite imperatives (with elision of the final -t of the verb desinence) and intimates that this is the norm: “Fazzamí su praxeri!” (lit.) ‘Do me the favour!’, “Nerimíddu!” Tell it to me!’, “Lesamí!” ‘Let me go!’ However, our own (Nuorese-Logudorese) informants accepted only procliticisation with these forms, as in (59): (59) a Lu piket! ‘Take it!’ b Mi lu porret! ‘Pass it to me!’ It is not clear whether this difference is dialectal9 or whether it simply reflects uncertainty of usage, related perhaps to the pragmatic incongruity between an imperative command and the subservience of the speaker implicit in the use of the polite form.

< previous page

page_28

next page >

< previous page

page_29

next page >

Page 29 With first person plural imperatives, encliticisation is the norm, with elision of the final -s of the verb desinence: (60) a Pikémalu! (subjunctive) b Picámulu! (indicative) ‘Let us take it!’ We do find cases where the clitic precedes the indicative form, e.g. Lu picamus!, but in the light of our earlier comment on (56b) it seems reasonable to classify these syntactically as declaratives or interrogatives —‘We shall take it’, ‘Should we take it?’ Two sets of facts support this view. First, the construction with the proclitic, like declaratives in general, can be modified by modal adverbs such as fortzis ‘perhaps’, whereas this is not generally possible with genuine imperatives including the forms in (60): *Fortzis pícalu!, *Fortzis pikémaluf, *Fortzis picámulu!, but Fortzis lu picamus ‘Perhaps we shall take it’. Second, interrogative a is frequently prefixed to the indicative construction, suggesting that these are syntactically yes/no questions with the illocutionary force of requests or invitations (see 1.2.2), but in such cases clitics must always precede the verb: A lu picámus? ‘Should we take it?’, but *A picámulu? Finally, we may note the following construction with the infinitive: (61) A la trattare bene! [to her treat well] ‘Treat her well!’ This construction is used for commands or exhortations which are addressed not to some specific person or persons but rather to whoever may be concerned; a more exact translation of (61) might be ‘May she be treated well!’ or ‘Let everyone treat her well!’ The item a which introduces this construction is not the interrogative particle discussed in 1.2.2 (such expressions do not have the flavour of questions) but the complementiser or preposition which introduces the infinitival complement of many verbs and adjectives (see 6.1.3). There is a certain parallel between the construction in (61) and hortatives introduced by ki with the subjunctive, as in (53), in that both can be used as complements of verbs such as kérrere ‘want’. Alongside (61) we have the complex sentence (62),10 just as the subjunctive hortative in (63a) is an exact replica of the complement in (63b): (62) Keljo a la trattare bene. ‘I want (people) to treat her well.’

< previous page

page_29

next page >

< previous page

page_30

next page >

Page 30 (63) a Ki su maritu la trattet bene! ‘May her husband treat her well!’ b Keljo ki su maritu la trattet bene. lit. ‘I want that her husband treat her well.’

< previous page

page_30

next page >

< previous page

page_31

next page >

Page 31 2 The noun phrase 2.1 COMPONENTS OF THE NP 2.1.1 The noun Most nouns end in a vowel, -a, -u, or -e (raised to -/ in Campidanese). In the first two cases the vowel gives a fairly reliable indication of the gender of the noun. Nouns ending in -a are almost all feminine whereas those in -u are predominantly masculine: fémina (f.) ‘woman’, mesa (f.) ‘table’, rosa (f.) ‘rose’, annu (m.) ‘year’, prattu (m.) ‘plate’, vinu (m.) ‘wine’. However, the ending -e provides no clue as to gender: frate (m.) ‘brother’, frore (m.) ‘flower’, ómine (m.) ‘man’, dente (f.) ‘tooth’, die (f.) ‘day’, nue (f.) ‘cloud’, sorre (f.) ‘sister’. Historically, nouns of the above types are derived from Latin accusative forms ending in -AM, -UM, -EM by elision of the final -M, the desinence -AM being restricted to feminine nouns in Latin (first declension), -UM predominantly masculine or neuter (second declen-sion) and -EM occurring with both masculine and feminine nouns (third and fifth declensions). Latin neuter nouns systematically be-come masculine in Sardinian, which has a binary gender system. With this proviso, Latin gender is preserved with remarkable regularity. Thus the gender of nouns in -e can usually be determined with reference to etymology. Conservation of Latin gender is particularly apparent in certain feminine nouns ending in -u, derived from Latin feminine accusative forms of the fourth declension: acu (f.) ‘needle’, ficu (f.) ‘fig’ (fruit and tree), manu (f.) ‘hand’. Also, domo ‘house’ is feminine (even though -o is typically a masculine ending), being derived from the ablative form of DOMUS (f., fourth declension). Final -o in masculine nouns borrowed from Italian or Spanish is usually raised to -u: cantzellu (m.) ‘gate’, cuníliu (m.) ‘rabbit’.

< previous page

page_31

next page >

< previous page

page_32

next page >

Page 32 However, in some cases, final -o is retained (except in Campidanese), as in oro (m.) ‘gold’, moro (m.) ‘Moor, dark-skinned person’. We may also note the following masculine nouns in -o derived directly from Latin: coro ‘heart’ (
< previous page

page_32

next page >

< previous page

page_33

next page >

Page 33 a natural gender distinction: tziu ‘uncle’ vs. tzia ‘aunt’, fidzu ‘son’ vs. fidza ‘daughter’, pitzinnu ‘boy’ vs. pitzinna ‘girl’, puddu ‘cock’ vs. pudda ‘hen’—an exception to this pattern is gattu ‘cat’ which is used in both the masculine and feminine without any change in the final vowel: unu gattu ‘a tom-cat’ vs. una gattu ‘a she-cat’. There are, of course, many cases where the natural gender distinction is lexicalised, often with one member of the pair functioning as the unmarked term: caddu ‘horse’ (male or unmarked) vs. ebba ‘mare’, crapa ‘goat’ (female or unmarked) vs. becu ‘billy goat’. In addition to the vowel-final nouns discussed above, there are a number of nouns which end in a consonant (-s or -n). These are derived from Latin neuter or masculine nouns of the 3rd declension whose accusative form ends in a consonant other than -m and are consequently masculine in Sardinian: corpus (m.) ‘body’, lapis (m.) ‘pencil’, letamen (m.) ‘manure’, numen (m.) ‘name’, samben (m.) ‘blood’, semen (m.) ‘seed’, tempus (m.) ‘time’ (also, sitis (m.) ‘thirst’, which is feminine in Latin). To this list, we may also add the days of the week lunis (m.) ‘Monday’, martis (m.) ‘Tuesday’, and mércolis (m.) ‘Wednesday’ (the other members of this paradigm end in -u (m.) or -a (f.): jóvia (f.) ‘Thursday’, kenápura (f.) ‘Friday’ (lit. ‘pure meal’), sábbatu (m.) ‘Saturday’, and doménica (f.) ‘Sunday’). The plural is formed by adding -s to the singular form, with change of final -u to -o (except in Campidanese): sas rosas ‘the roses’, sos annos ‘the years’, sos frates ‘the brothers’, sas domos ‘the houses’, sos thilleris ‘the taverns’. When the noun ends in -en, -es is added: sos númenes ‘the names’. Nouns ending in -s do not add -s in the plural, but -us is changed to -os (except in Campidanese): sos lapis ‘the pencils’, sos tempos ‘the times’. There are no irregular plural noun forms. There are of course many nouns which do not show a number distinction: pluralia tantum nouns such as fórtikes (f.) ‘scissors’, conos (m.) ‘nausea’, tzicolitos (m.) ‘hiccups’; non-count abstract or mass nouns such as sitis (m.) ‘thirst’, virgondza (f.) ‘shame’, abba (f.) ‘water’, mele (m.) ‘honey’; inherently collective nouns such as dzente (f.) ‘people’, robba (f.) ‘things in general, belongings, produce, etc.’ (also, more specifically ‘livestock’, typically sheep: Su pastore tentat sa robba The shepherd looks after the sheep’), and ákina ‘grapes’ (which can only be used in the singular with collective reference—e.g. we cannot say *duas ákinas ‘two grapes’). Apart from fossilised examples such as ákina, a characteristic feature of Sardinian is the use of singular count nouns denoting fruit, vegetables, insects and small animals with plural or collective reference:

< previous page

page_33

next page >

< previous page

page_34

next page >

Page 34 (1) a So ispidzolande sa patata. ‘I am peeling the potatoes (lit. ‘the potato’).’ b Tunca sa janna ki no’intret sa thíthula. ‘Shut the door so that the mosquitoes (lit. ‘mosquito’) do not come in.’ Apart from certain pronouns (see 5.1.1) nouns do not inflect for Case. However, a relic of genitive Case survives in expressions such as the following involving body-part nouns: ocri tortu ‘cross-eyed’ (lit. ‘crooked of eye’), pili ruju ‘red-haired’, murri longu ‘long-faced’. Often such expressions are written as a single word (e.g. ocritortu) suggesting that the phenomenon may belong to the domain of wordformation rather than sentence-syntax. 2.1.2 Determiners and quantifiers The definite article forms are su (m. sg.), sa (f. sg.), sos (m. pl.), sas (f. pl.), except in Campidanese which has a single plural form is for both genders. Demonstrative articles show a three-way deictic contrast which can be characterised roughly as in (2): (2) a custu, -a, -os, -as (close to speaker) b cussu, -a, -os, -as (close to addressee) c cuddu, -a, -os, -as (remote from both speaker and addressee) The forms in (2b) are also used when referring back to entities mentioned earlier in the discourse: (3) A connoskes su babbu de Maria? No, no’ appo mai incontratu cuss’ ómine. ‘Do you know Mary’s father?’ ‘No, I have never met that man.’ All three types of demonstratives can be used without a following noun, in the manner of pronouns, with animate or inanimate reference (see 5.1.2 for examples and detailed discussion). The singular indefinite article forms are unu, -a, which are also used as the numeral ‘one’. These items can be used in the plural, unos, -as, usually accompanied by a numeral (e.g. unas vinti berbekes ‘some twenty sheep’—i.e. ‘about twenty sheep’). In some dialects, particularly Campidanese, these forms can be used with the item cantu (normally a noun ‘piece’ or an interrogative quantifier ‘how much’) to express indeterminate quantity: unas cantu (de) berbekes

< previous page

page_34

next page >

< previous page

page_35

next page >

Page 35 ‘some sheep’. The optional presence of de ‘of suggests that this may be a partitive construction of the type discussed below in 2.2.2 (examples (39)-(44)). Note, however, that unas agrees in number and gender with berbekes, not with cantu which is invariable in this construction (cf. Blasco-Ferrer 1986:102–3, 210). Indefinite quantification can also be expressed by carki ‘some’ (invariable). For some speakers, this quantifier always requires the following noun to be singular; for these speakers, sentence (4) is ambiguous: ‘Buy me some bottles of wine’ or ‘Buy me some bottle (or other) of wine’: (4) Cómporami carki ampulla de vinu. However, other speakers indicate plural reference by using the plural form of the noun: carki ampullas de vinu ‘some bottles of wine’. Unlike English some, carki cannot be used to quantify non-count nouns: (5) *Cómporami carki vinu. ‘Buy me some wine.’ Indefinite quantification of non-count nouns is typically expressed by the absence of an overt determiner, as in (6a), or by means of a partitive construction such as (6b) whose syntactic properties will be discussed in 2.2.2: (6) a Dáemi casu. ‘Give me cheese.’ b Dáemi (un) adziccu de casu. ‘Give me a bit of cheese.’ Other indefinite quantifiers are meta ‘many, much’, pacu ‘few, little’, tantu ‘so many, so much’, troppu ‘too many, too much’, paritzos ‘several, tzertos ‘certain’. As the glosses indicate, meta, pacu, tantu and troppu can quantify singular non-count nouns as well as plural count nouns, whereas paritzos and (as sole determiner) tzertos can only occur with the latter, though the singular form tzertu can be used after an indefinite article with a singular count noun (e.g. unu tzertu libru ‘a certain book’), tzertu being arguably an adjective rather than a determiner in this case. With regard to inflection, pacu and tantu agree in number and gender with the quantified noun (with the endings -u (m. sg.), -a (f. sg.), -os (m. pl.), and -as (f. pl.)). Paritzos and (as a determiner) tzertos inflect only for gender, with the feminine ending -as. Meta normally takes an -s in the plural but shows no gender distinction, whereas troppu is invariable.

< previous page

page_35

next page >

< previous page

page_36

next page >

Page 36 Typically the determiners mentioned above precede the head noun. However, meta and pacu can also occur postnominally, particularly as a means of emphasising the quantity: (7) a Amus bitu vinu. b Amus bitu vinu ‘We drank much/little wine.’ When meta and pacu follow a plural noun, there is some uncertainty regarding number agreement. Some speakers require the plural form (as when these items occur prenominally), while others treat them as invariable in this position: libros meta or libros metas ‘many books’, libros pacu or libros pacos ‘few books’. The principal interrogative determiners or quantifiers are cale ‘which’ and cantu ‘how much, how many’. Cantu always agrees in number and gender with the quantified noun (like tantu mentioned above). Cale is invariable for gender but shows some uncertainty with regard to number inflection. When cale is used without a head noun, -s must be added to indicate intended plural reference: (8) a Cale keres? ‘Which one do you want?’ b Cales keres? ‘Which ones do you want?’ However, speakers differ with respect to number inflection when calequalifies an overt plural noun: (9) Cale(s) libros keres? ‘Which books do you want?’ The item itte ‘what’ (invariable) typically occurs alone (ostensibly as an interrogative pronoun, but see 5.1.2), but it can also function as an interrogative determiner with an overt head noun, which may be animate or inanimate (even though itte on its own can only have inanimate reference), but is typically non-count, as in (10): (10) a Itte vinu as comporatu? ‘What wine did you buy?’ b Itte dzente as vistu? ‘What people did you see?’ The negative item nudda ‘nothing’ (also invariable) can be used as a determiner in the same way as itte:

< previous page

page_36

next page >

< previous page

page_37

next page >

Page 37 (11)

a No’appo comporatu nudda vinu. ‘I did not buy any wine.’ b No’appo vistu nudda dzente. ‘I did not see any people.’ In such cases, the use of nudda typically serves to deny a contextual presupposition (e.g. as answers to questions such as those in (10)). A similar effect can be achieved by the item perunu, which always takes a singular non-count noun and agrees with it in gender. Like meta and pacu, perunu can precede or follow the noun, the emphatic effect being rather stronger in the latter case. (12) a No’appo iscrittu peruna líttera. b No’appo iscrittu líttera peruna. ‘I did not write any letter.’ The principal universal quantifier is tottu ‘all’, which may take either a non-count noun or a plural count noun introduced by a definite or demonstrative article: (13) a Pretu at mandicatu tottu su pane. ‘Peter ate all the bread.’ b Juanne at lessu tottu cussos libros. ‘John has read all those books.’ c Tottu sas berbekes sun thuccatas. ‘All the sheep have run away.’ When this item quantifies a subject NP (overt or null) or an accusative clitic, it can be ‘floated’ to the postauxiliary position or to a position following the main verb: (14) a Juanne los at tottu lessos. b Juanne los at lessos tottu. ‘John has read them all.’ (15) a (Sas berbekes) sun tottu thuceatas. b (Sas berbekes) sun thuccatas tottu. The sheep/they have all left.’ This item can also be used pronominally: (16) a Appo mandicatu tottu. ‘I ate everything.’

< previous page

page_37

next page >

< previous page

page_38

next page >

Page 38 b Appo faeddatu kin tottu. ‘I spoke with everyone.’ In most dialects, tottu is invariable for both number and gender in all the above contexts. However, in some Nuorese dialects a plural form tottus (invariable for gender) is used in some cases where this item occurs in isolation (see Farina 1973:270). The examples given by Farina are the following: (17) a “Sun tuccàos tottus.” b “Sun tottu tuccàos.” They all ran away.’ Our own research suggests that this plural form can be used in cases like the (b) examples in (15)-(16), but not in the other examples given above, though our informants’ judgements are rather unclear. What looks like a feminine plural form of tottu is found in cases where this item is followed by a numeral: (18) tottas tres féminas ‘all three women’ In some dialects tottas is used even with masculine nouns, as in (19a), though there are some speakers who use tottus as in (19b): (19) a tottas tres ómines b tottus tres ómines ‘all three men’ It seems unlikely that cases like (19a) involve a gender neutralisation in favour of the feminine. Possibly the vowel /a/ is a relic of the Latin conjunction AC, tottas being derived historically from something like tottu AC sos/sas, a formula which survives in Campidanese: “totu a (is) duus” (Blasco-Ferrer 1986:119). A similar situation is found with the dual quantifier ambos ‘both’, which is typically followed by the numeral duos ‘two’ (feminine: duas). For some speakers, this item shows a gender opposition, as in (20a, b), whereas others use ambas even in the masculine, as in (20c) (sometimes with contraction of ambas duos to ambaduos): (20) a ambos duos ómines b ambas duas féminas c ambas duos ómines ‘both men/women’ Ambos can also be used without the numeral and, in this case, always shows gender agreement: ambos ómines, ambas féminas ‘both (of the) men/women’.

< previous page

page_38

next page >

< previous page

page_39

next page >

Page 39 Note that, unlike tottu, ambos is not followed by a definite or demonstrative article. This quantifier (with or without the numeral) can be ‘floated’ in much the same way as tottu, though it appears to be less natural in the post-auxiliary position than tottu: (21) a Ambas (duas) pitzinnas sun vénnitas. b ?Sas pitzinnas sun ambas (duas) vénnitas. c Sas pitzinnas sun vénnitas ambas (duas). ‘Both of the girls came.’ (22) a ?Las appo ambas (duas) salutatas. b Las appo salutatas ambas (duas). ‘I greeted them both (f.)’ Other universal quantifiers, of a distributive type, are dondzi and (in Nuorese) cada ‘every’. These items are invariable and always take a singular count noun: (23) Juanne at lessu, libru. ‘John has read every book.’ 2.1.3 Numerals The basic elements of the cardinal numeral system are given in Table 2.1, which is arranged in such a way as to show how these elements are combined to form complex numerals. Complex numerals are formed by concatenating these elements in the order given. However, in some cases, elements of the complex numeral must be conjoined by e ‘and’. The basic generalisation is that simple concate-nation (without e) is only possible with elements from adjacent columns: e.g. milli dukentos baranta kimbe (=1,245), but mill’e baranta kimbe (=1,045), milli dukentos e unu (=1,201), trekentos e dekesette (=317)—note that for the purposes of this generalisation the numbers 10–19, including those beginning with deke-, belong to the same category as ‘units’. In addition, e is required after milione(s) and midza and also between the items milli and kentu: unu milione e baranta midza (=1,040,000), tremidza e kimbikentos (=3,500), mill’e kentu (=1,100), mill’e kentu baranta (=1,140). The form una midza is used only as the final element in a complex numeral after milione(s): unu milione e una midza (=1,001,000), duos miliones e una midza (2,001,000), but unu milione e milli dukentos (1,001,200). Higher multiples of -kentos are sometimes used in preference to forms in milli: dóiki kentos (=1,200). When milione(s) and midza

< previous page

page_39

next page >

< previous page

Page 40 Table 2.1 Cardinal numerals (Logudorese-Nuorese forms) Millions Thousands Hundreds unu milione 1,000,000 milli (una midza) 1,000 kentu, -a duos miliones 2,000,000 duarmidza 2,000 dukentos, -as tres miliones 3,000,000 tremidza 3,000 trekentos, -as etc. battormidza 4,000 battoskentos, -as etc. kimbikentos, -as seskentos, -as settekentos, -as ottokentos, -as novekentos, -as

< previous page

next page >

page_40

page_40

100 200 300 400 500 600 700 800 900

Tens vinti trinta baranta kimbanta sessanta settanta ottanta novanta

20 30 40 50 60 70 80 90

Units/teens unu, -a duos, -as ties battor kimbe ses sette otto nove deke úndiki dóiki tréiki battórdiki bíndiki séiki dekesette dekeotto dekenove

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19

next page >

< previous page

page_41

next page >

Page 41 occur as the final element in a complex numeral, the quantified noun is introduced by de ‘of, obligatorily with milione(s), optionally with midza: unu milione de francos ‘a million lira’, duos miliones de ómines ‘two million men’, tremidza (de) vaccas ‘three thousand cows’. When unu is the final element, the quantified noun assumes the singular form: trintuna vacca ‘thirty-one cows’, kent’e unu cane ‘a hundred and one dogs’ (cf. English more than one dog/*dogs). The numerals unu, duos, and forms in -kentos (but not kentu on its own) agree in gender with the quantified noun, when used alone or as part of a complex numeral: una pitzinna (f.) ‘one girl’, duas domos (f.) ‘two houses’, trekentas vaccas (f.) ‘three hundred cows’, dukentas baranta duas berbekes (f.) ‘two hundred and forty-two sheep’. The forms milione and midza have inherent gender (milione (m.), midza (f.)); consequently it is these forms (rather than the quantified noun) which determine the gender of the preceding numeral in expressions like duos miliones de féminas ‘two million women’, kimbekentas midza de ómines ‘five hundred thousand men’. The ordinal numerals are essentially the same as in Italian apart from minor phonological changes: primu, sicundu, tertzu, cuártu, cuíntu, sestu, séttimu, ottavu, nonu, dékimu, etc. These forms inflect for gender and number with the usual endings (-u, -a, -os, -as) . However, apart from primu and sicundu (also úrtimu ‘last’), these forms are rarely used in colloquial speech. In their place we find a periphrastic construction consisting of the definite article followed by de ‘of and the appropriate cardinal numeral and (optionally) a quantified noun: (24) a su de duos (ómines) ‘the second (man)’ b sa de duas (féminas) ‘the second (woman)’ Note that when the quantified noun appears, it takes the plural form (as determined by the numeral) even though the referent of the expression as a whole is singular. Along with the use of the singular after numerals ending in -unu mentioned above, this suggests a general tendency for the head noun to agree in number with the last numerical element within the determiner expression. 2.1.4 Adnominal adjectives Adjectives follow the nominal pattern of inflection, giving two inflectional classes:

< previous page

page_41

next page >

< previous page

page_42

next page >

Page 42 I -u, -a, -os, -as II -e, -es Adjectives agree in number and gender with the noun which they modify, though the gender distinction is not overt with type II adjectives. Attributive adjectives almost always follow the noun which they modify; indeed, the postnominal position is obligatory with many adjectives whose counterparts in other Romance languages typically precede the noun: una mákkina nova vs. *una nova mákkina ‘a new car’, un’ómine vetzu vs. *unu vetz’ómine ‘an old man’, una tassa manna vs. *una manna tassa ‘a large glass’, una domo minore vs. *una minore domo ‘a small house’. However, there are a few attribu-tive adjectives which can precede the noun: bellu ‘beautiful’, bonu ‘good’, bravu ‘good, well-behaved’, bruttu ‘ugly’, póveru ‘poor’, santu ‘holy’. In prenominal position, such adjectives convey an affective attitude of appreciation or deprecation rather than describing an inherent property of the referent. Thus, in una fémina bella/brutta the adjective simply describes physical appearance, but in una bella-/brutta fémina it conveys the idea that the person is admired or despised on grounds of character or behaviour as well as appearance. Similarly, postnominal póveru indicates lack of wealth, whereas prenominally it means ‘to be pitied’; unu sant’ómine may refer to a man who one considers virtuous or saintly, whereas un’ómine santu must denote one who has been canonised or, at least, has taken holy orders; bonu usually follows the noun with the meaning ‘good to eat or drink’ (vinu bonu vs. *bonu vinu ‘good wine’), but can occur in either position with abstract nouns (e.g. una bona idea or un’idea bona ‘a good idea’). The adjective grandu, used in some dialects (mainly Campidanese) with the meaning ‘great, important’, always precedes the noun and is invariable (cf. Blasco-Ferrer 1986:90): una grandu festa ‘a great feast’. Another possible case of an adjective which can only occur prenominally is the item bette (sometimes béttedde or bettédde ) ‘big, huge’, which is used predominantly in exclamations: (25) a Itte bette cane! b *Itte cane bette! ‘What a huge dog!’ However, it is not absolutely clear that this item is an adjective. Wagner (1960–4:200) gives the following examples where bette is followed by de (often contracted to ’e), which suggest that it is

< previous page

page_42

next page >

< previous page

page_43

next page >

Page 43 functioning as an epithet noun (cf. English a whopper of a lie, a hulk of a lad ): (26) a “bette de vaula” ‘huge lie’ b “una bett’e domo” ‘a huge house’ c “unu bette de pittsinnu” ‘a big boy’ This analysis is supported by etymology, bette being derived from Latin VECTIS ‘lever’ or ‘post’, also ‘phallus’ (Wagner 1960–4:200). If bette in (25a) is actually a contraction of bette de as in (26b), this would explain the ungrammaticality of (25b). Note, however, that in Wagner’s examples in (26b, c) the preceding article shows a gender distinction determined by the modified noun (domo (f.) vs. pitzinnu (m.)), a fact which is consistent with the view that bett’e or bette de is an adjective. This item is classified as an adjective by Pittau (1972:70), who also mentions a plural form bettes, corroborated by some of our informants who accept examples like (27): (27) Itte bettes canes! ‘What huge dogs!’ Other informants rejected not only (27) but also variants with the uninflected form (e.g. *Itte bette (de) canes!) claiming that this item could not be used with a plural noun—as indeed seems to be the case with the epithet construction in English: *What whopper(s) of dogs! Interestingly, the informants who rejected (27) were also uncertain about gender agreement in examples like (26b, c) and reported that they would only use bette in exclamations with itte or without an article, as in (25a) and (26a), where the problem of agreement does not arise. For these speakers, we postulate that bette has conserved its status as a noun (so that (25a) should be transcribed as Itte bett’e cane ) whereas for speakers who accept (27), it appears that this contracted form has been reanalysed as an adjective agreeing in number with the modified noun. Examples like (26b, c) represent an intermediate stage in which bette appears to function as a noun but is transparent with respect to agreement of the article. Note that the agreement phenomenon in these examples is very similar to that observed in 2.1.2 (pp. 34–5) concerning the form unos cantu as in unas cantu (de) berbekes ‘some sheep’. As well as affective adjectives of the type discussed above,

< previous page

page_43

next page >

< previous page

page_44

next page >

Page 44 adjectives with a specifier-like function always precede the noun. This class includes the ordinal numbers primu ‘first’, sicundu ‘second’, etc. (also próssimu ‘next’ and úrtimu ‘last’), cardinal numbers when preceded by a determiner (cuddos tres libros ‘those three books’) and adjectives such as átteru ‘other’, mattessi (invariable) ‘same’, mesu (invariable) ‘half and únicu ‘single, only’: s’áttera manu ‘the other hand’, sa mattessi cosa ‘the same thing’, unu mesu litru ‘half a litre’, s’unicu problema ‘the only problem’. Note finally that the comparative/ superlative forms medzus ‘better/best’ and pejus ‘worse/worst’ (both invariable) always precede the noun. 2.1.5 Possessives The pronominal possessive forms are meu ‘my’, tuo ‘your’ (sg. fam.), suo ‘his, her, its’, nostru ‘our’, vostru (pl. and sg. pol.), issoro ‘their’. To this list we may add the item antzenu ‘another’s’ whose properties are similar to those of the personal forms. Apart from issoro, which is invariable (sas berbekes issoro ‘their sheep’), these items agree in number and gender with the noun denoting the possessed entity: meu, -a, -os, -as\ tuo, -a, -os, -as, etc. In the third person there is no distinction with respect to the gender of the possessor; when such a distinction needs to be made, a periphrastic construction with de and a disjunctive personal pronoun can be used: su libru de isse ‘his book’ vs. su libru de issa ‘her book’ (lit. ‘the book of him/her’). The construction with de is also used for non-pronominal possessives: su libru de Maria ‘Mary’s book’, sa domo de su duttore ‘the doctor’s house’. Possessive forms always follow the noun and typically co-occur with a determiner: su libru meu ‘my book’, sa robba antzena ‘another person’s property’. Determiners and quantifiers other than the definite article can co-occur with possessives: unu frate meu ‘one of my brothers’, tottu sos amicos tuos ‘all your friends’, metas berbekes meas ‘many of my sheep’, dondzi (úrtimu) libru meu ‘every (last/single) book of mine’. The definite article is typically omitted when the head noun is a kinship term or term of endearment and is modified by a pronominal possessive: frate tuo ‘your brother’, coro meu ‘my dar-ling’ (lit. ‘my heart’). This construction is used particularly with first and second person possessives. When the understood possessor is in the third person and anaphoric to some entity prominent in the discourse, typically the subject of the clause, the definite article is normally used without a possessive: Pretu est andatu kin su frate ‘Peter went with his (Peter’s) brother’.

< previous page

page_44

next page >

< previous page

page_45

next page >

Page 45 2.2 STRUCTURE OF THE NP 2.2.1 Preliminary assumptions For the purposes of the present account, the basic structure of the NP can be represented as in (28):

(28) The symbol N’ denotes a subconstituent which is not a well-formed NP in its own right. On this basis, tottu is distinguished from the other articles and quantifiers discussed in 2.1.2 (represented as DET in (28)) in that the expressions which follow tottu can be used independently as NPs (e.g. as subjects or direct objects) whereas those which follow the other determiners cannot (except in certain circumstances discussed in 2.2.5): (29) a tottu [NP su casu] ‘all the cheese’ b tottu [NP sos ómines de Orune] ‘all the men from Orune’ (30) a sos [N′ caddos k’appo vistu] ‘the horses that I saw’ b dondzi [N′ libru k’appo lessu] ‘every book that I read’ c una [N′ líttera dae Maria] ‘a letter from Mary’ The features [+spec] and [+aff] are mnemonic labels used to distinguish between the two types of adjective which can precede the noun; respectively, specifier-like adjectives (cardinal and ordinal numerals and items like átteru ‘other’, mattessi ‘same’, etc.) and the restricted class of attributive adjectives which can occur prenominally

< previous page

page_45

next page >

< previous page

page_46

next page >

Page 46 with an affective value (e.g. bellu ‘beautiful’, bonu ‘good’, bette ‘huge’, etc.)—see 2.1.4. The symbol X indicates the possible structural positions of complements and modifiers which follow the noun (e.g. possessives, APs, PPs, relative clauses, complement clauses, etc.). Questions concerning the precise structural position of each of these elements will be discussed in 2.2.3. The possible expansions of NP and N’ are defined by the following schemata: (31) (32) 2.2.2 Prenominal elements Of the various categories which can occur prenominally, the only one which can be iterated is ADJ[+spec]. This is permitted by recursive application of part (i) of rule (32), yielding structures such as the following: (33) a [NP cussos [N′ tres [N′ átteros [N caddos]]]] ‘those three other horses’ b [NP cussos [N′ átteros [N′ tres [N caddos]]]] ‘those other three horses’ c [NP sas [N′ duas [N′ úrtimas [N pádzines]]]] ‘the two last pages’ d [NP sas [N′ úrtimas [N′ duas [N pádzines]]]] ‘the last two pages’ There do not appear to be any absolute restrictions on the order of items of this class. Technically, recursive application of rule (31i) should also be possible, allowing indefinite iteration of tottu, though in practice this does not happen, presumably for semantic reasons. Note also that a stipulation must be added to the effect that the NP following tottu must be definite: (34) a *[NP tottu [NP carki libros]] ‘all some books’ b *[NP tottu [NP metas canes]] ‘all many dogs’

< previous page

page_46

next page >

< previous page

page_47

next page >

Page 47 Note that the grammar in (31)-(32) does not allow a sequence of items of the class A[+aff] in prenominal position, since (32ii) cannot apply recursively. Thus, if a noun is modified by more than one adjective of this class, all except one must be placed after the noun, with the proviso that some items of this class (e.g. bette) cannot occur postnominally: (35) a *unu bellu bette cane b *unu bette bellu cane c *unu bellu cane bette d unu bette cane bellu ‘a huge beautiful dog’ However, the rules in (32) correctly allow adjectives of the [+aff] type to follow adjectives of the [+spec] type: (36) a un’átteru bette cane ‘another huge dog’ b cussas duas bravas pitzinnas ‘those two good girls’ The rules in (32) introduce prenominal adjectives as a lexical category rather than as part of the phrasal category AP. This formulation reflects the fact that prenominal adjectives cannot be followed by a complement or be qualified by a degree modifier. In the case of adjectives of the [+spec] type, this restriction may be attributed to the inherent semantic properties of the items in question. However, adjectives of the [+aff] type can be modified provided that they follow the noun: (37) a una pitzinna meta bella b ??una meta bella pitzinna ‘a very beautiful girl’ Rule (31ii) allows only one determiner per NP, correctly excluding phrases such as *custos metas libros ‘these many books’, *sos carki frores ‘the some flowers’, etc. As already noted, tottu is exceptional in that it co-occurs with a definite or demonstrative article, hence the hypothesis that it is introduced by the special rule (31i) rather than under the DET node via (31ii). We assume that cardinal numerals function as [+spec] adjectives when preceded by a determiner (as in (38a)) but can occur under the DET node when no other determiner is present (as in (38b)):

< previous page

page_47

next page >

< previous page

page_48

next page >

Page 48

(38) The claim that numerals function as determiners in phrases like (38b) is based on the observation that NPs introduced by a bare numeral can occur freely in contexts which do not normally allow NPs with no determiner (see 2.2.5 for examples and discussion). Cardinal numerals and most other indefinite determiners also allow a partitive construction with de+ definite NP, as in (39): (39) a Una de sas berbekes est maláida. ‘One of the sheep is sick.’ b Appo pérditu duos de sos libros meos. ‘I have lost two of my books.’ c Metas de custas pitzinnas sun galanas. ‘Many of these girls are pretty.’ As a matter of execution, let us assume that the sequence de+NP is a PP complement of a null head noun and that this null N acts as a quantified variable ranging over the indefinite set denoted by the following NP, as in (40):

(40)

< previous page

page_48

next page >

< previous page

page_49

next page >

Page 49 When carki ‘some’ or dondzi ‘every’ occur in this construction, they must be followed by the proform unu/-a, which we assume to occupy the same position as the null head

in (40): (41) a Cark’una de sas berbekes sun maláidas. ‘Some of the sheep are sick.’ b Cark’una de sas berbekes est maláida. ‘One or other of the sheep is sick.’ c Dondz’una de sas berbekes est maláida. ‘Every one of the sheep is sick.’ A similar partitive construction is used with quantity nouns such as adziccu and pacu ‘little’, paju ‘couple’, dudzina ‘dozen’ and the numerals midza ‘thousand’ and milione ‘million’ (see 2.1.3). The NP governed by de is introduced by a definite or demonstrative article when referring to a portion of a specific set, otherwise this NP occurs without an article: (42) a un’adziccu de (custu) vinu ‘a little (of this) wine’ b una dudzina de (cuddos) ovos ‘a dozen (of those) eggs’ For such constructions, we propose the following structure:

(43) Note that the quantity noun determines the number and gender of the determiner (and any modifiers which may be present—e.g. un’áttera (f. sg.) dudzina (f. sg.) de ovos (m. pl.) ‘another dozen eggs’) but agreement of elements outside the NP is determined by the features of the embedded class noun, as in (44) where the auxiliary is plural and the past participle isparitos is masculine plural in agreement with ovos (m. pl.) rather than dudzina (f. sg.):

< previous page

page_49

next page >

< previous page

page_50

next page >

Page 50 (44)

Una dudzina de ovos sun isparitos. ‘A dozen eggs have disappeared.’ 2.2.3 Postnominal elements In contrast to prenominal elements, the linear order of items which follow the noun cannot be defined strictly in terms of syntactic categories, but appears to be determined by the following tendencies: (45)a Expressions which bear a close semantic relation to the noun tend to precede those which are more loosely related to the noun. b Syntactically simple expressions precede more complex ones (e.g. single words precede phrases, which precede clauses). The generalisation in (45a) is partly derivable from the general principle of X-bar syntax whereby complements occur as sisters of the head (under the lowest single-bar node) whereas modifiers are adjoined to the single-bar projection. Thus, in (46) the complement PP de frantzesu ‘of French’ must precede the modifier PP de Núgoro ‘from Nuoro’: (46)a unu professore de frantzesu de Núgoro b *unu professore de Núgoro de frantzesu ‘a teacher of French from Nuoro’ However, there are other cases where the effects of (45a) cannot be attributed straightforwardly to general structural principles. Consider the following examples: (47) a su frate mannu meu *su frate meu mannu ‘my big (=elder) brother’ b su frate meu riccu *su frate riccu meu ‘my rich brother’ c sa mákkina ruja mea sa mákkina mea ruja ‘my red car’ The order of elements in (47a) appears to be due to the fact that mannu is used with a special sense ‘elder’ which it can only assume in collocation with certain kinship nouns. If mannu is taken as a normal attributive adjective with the sense ‘large in stature’, the reverse order is required, as with riccu in (47b). Examples (47b, c) suggest that the distinction between inalienable and non-inalienable possession is relevant for the purposes of (45a), the inalienable possessor

< previous page

page_50

next page >

< previous page

page_51

next page >

Page 51 in (47b) being semantically more closely related to the head noun than the possessor in (47c) which is on a par with the attributive adjective ruju. Conceivably, this gradation might be couched in structural terms, frate mannu being analysed as a complex noun, the inalienable possessor meu in (47a, b) as a sister to the head N under the minimal N’ (in so far as it represents an argument of the relational noun frate), whereas non-inalienable possessives (mea in (47c)) and attributive adjectives are adjoined to N’ in the X position introduced by rule (31ii). Alternatively, we might maintain that all the postnominal items in (47) are modifiers adjoined to N’ with the order of adjunction determined by semantic considerations. We leave this question open, the important point being that the order of postnominal expressions is not statable in terms of syntactic category (e.g. possessive vs. attributive adjectives). The effects of (45b) are illustrated in (48): (48) a sa mákkina ruja de Juanne *sa mákkina de Juanne ruja ‘John’s red car’ b su cane meu k’ an furatu *su cane k’ an furatu meu ‘the dog of mine that they have stolen’ c s’ómine de Orune k’appo addoppiatu ?s‘ómine k’appo addoppiatu de Orune ‘the man from Orune that I met’ Although it is marginally possible for a relative clause to precede a modifier PP (contrary to (45b)), as shown in (48c), this order is not possible when the PP is a complement, in which case (45a) also comes into play: (49) su professore de frantzesu k’appo addoppiatu *su professore k’appo addoppiatu de frantzesu ‘the teacher of French that I met’ When the generalisations in (45) conflict, principle (45b) normally takes precedence, as in the following examples where the structurally simpler expressions precede the more complex ones even though the latter bear a closer semantic relation to the head noun according to the criteria adopted in our discussion of earlier examples: (50) a unu professore vetzu de frantzesu *unu professore de frantzesu vetzu ‘an old teacher of French’

< previous page

page_51

next page >

< previous page

page_52

next page >

Page 52 b su frate meu prus mannu *su frate prus mannu meu ‘my eldest brother’ In general, it appears that a postnominal expression consisting of a single word (e.g. a possessive or an unmodified adjective) cannot be separated from the head noun by a phrase consisting of more than one word. Exceptions to this generalisation are examples such as the following, where the N+de+N sequences can plausibly be analysed as compound nouns: (51) a una camisa de notte bianca *una camisa bianca de notte ‘a white nightshirt’ b unu mastru de linna abbistu *unu mastru abbistu de linna ‘a clever carpenter’ (lit. ‘master of wood’) c petha de andzone arrusta ?petha arrusta de andzone ‘roast lamb’ (lit. ‘roast meat of lamb’) 2.2.4 Order of postnominal adjectives When a noun is modified by more than one postnominal adjective, there are two possible constructions; (a) the adjectives are simply concatenated or (b) they are conjoined by e ‘and’ (or ma ‘but’ if there is an implied contrast between the two adjectives). We assume that constructions of the first type are derived by recursive application of (32i) with selection of AP in the position X: (52) [NP una [N′ [N′[N′ mákkina] manna] nova]] [a car big new] ‘a big new car’ In principle, this system allows an infinite number of adjectives to follow the noun, though, in practice, sequences of more than two adjectives do not usually occur. There are restrictions on the order of postnominal adjectives, but not of the sort which can be captured by phrase-structure rules which refer to discrete subclasses of adjectives (such as those proposed for prenominal adjectives in 2.2.2). Generally, we find that basic, common adjectives which can be used with a wide range of nouns tend to precede adjectives which denote more specific properties:

< previous page

page_52

next page >

< previous page

page_53

next page >

Page 53 (53)

a un’ arrantzolu mannu vellenosu [a spider big poisonous] b ?un’ arrantzolu vellenosu mannu [a spider poisonous big] Also, adjectives which denote permanent or typical properties usually precede those which indicate transient or incidental properties: (54) a unu mutzighinu biancu bruttu [a handkerchief white dirty] b ?unu mutzighinu bruttu biancu [a handkerchief dirty white] Interestingly, one of our informants pointed out that (54b) is mildly contradictory in so far as a handkerchief which is dirty is not, strictly speaking, white, but there is no contradiction in (54a) if mutzighinu biancu is taken to refer to a notional class (handkerchieves which are supposed to be white) with bruttu specifying a distinctive property of the intended referent. Similar factors, mediated by considerations of a more subjective nature, may be at work in (55) in that the size of a car has greater practical relevance than its colour, so that ‘big car’ is a more plausible subtype of car than ‘red car’, whereas ruju is more appropriate as a means of distinguishing the intended referent from other members of this class: (55) a una mákkina manna ruja [a car big red] b ?una mákkina ruja manna [a car red big] On the other hand, in (56) the colour term is preferred as the first adjective, perhaps because shirts are more naturally subclassified in terms of colour than of size: (56) a una camisa bianca longa [a shirt white long] b ?una camisa longa bianca [a shirt long white] The above examples are intended to be illustrative of the sort of factors which determine the order of postnominal adjectives and to substantiate our hypothesis that this order cannot be stipulated in terms of semantic subclasses such as colour, size or quality. A more rigid restriction on concatenation of postnominal adjectives is that only the final adjective may be modified (e.g. by a degree adverb):

< previous page

page_53

next page >

< previous page

page_54

next page >

Page 54 (57) a una mákkina manna tottu nova [a car big absolutely new] b *una mákkina tottu nova manna [a car absolutely new big] c *una mákkina meta manna nova [a car very big new] d ??una mákkina meta manna tottu nova [a car very big absolutely new] This conforms to our earlier claim (p. 52) that postnominal phrases consisting of a single word cannot be separated from the head noun by a phrase consisting of more than one word. In some cases, this restriction may override the ‘natural’ order of adjectives as determined by pragmatic considerations, as in (58) (compare (55b)): (58) una mákkina ruja meta manna [a car red very big] The above restrictions do not apply when postnominal adjectives are conjoined as in (59): (59) una mákkina manna e nova [a car big and new] In this construction, the adjectives are on much more of an equal footing, with the result that order is more flexible (e.g. reversal of the adjectives in (59) does not make any appreciable difference to acceptability or interpretation), though in some cases a particular order may be exploited to achieve a ‘crescendo’ effect, as in (60) where placement of delitziosa before either of the other adjectives would create an anticlimax: (60) una turta manna, tunda e delitziosa [a cake big round and delicious] Also, when the conjunction is ma ‘but’, the order of adjectives may make a difference to meaning; e.g. (61a) might well be taken as a compliment, but (61b) would not: (61) a Est una fémina vetza ma bella. [(she+) is a woman old but beautiful] b Est una fémina bella ma vetza. [(she+) is a woman beautiful but old] We assume that the constructions in (59)-(61) do not involve recursion within the NP system, but are accounted for by a general principle which allows any phrase to be expanded as a conjunction of

< previous page

page_54

next page >

< previous page

page_55

next page >

Page 55 phrases of the same category, in this case a rule of the form AP—> AP* CONJ AP, where AP* stands for an arbitrary number of APs. In conjoined constructions of this type, modification by a degree adverb is not restricted to the final adjective (cf. (57) above): (62) un‘ómine meta forte e abbistu [a man very strong and clever] Our informants consistently maintained that the degree modifier (e.g. meta) in such cases can only modify the first adjective, repetition of meta being required if the second adjective is also intensified (un’ómine meta forte e meta abbistu). This suggests that the constructions in question involve conjunction of the phrasal (AP) node rather than conjunction of A or A’; i.e. structures of the form [AP meta[A’ A CONJ A]] are not possible. Pursuing our suggestion that conjoined postnominal adjectives (or APs) are on a more equal footing than those which are adjoined recursively to the N’ node, we may note that adjectives which may assume a special meaning with certain nouns (e.g. mannu with the sense ‘elder’) lose this property when conjoined with other adjectives. Thus, (63) can only mean ‘I have a brother who is big (in stature) and strong’, not ‘a strong elder brother’: (63) Appo unu frate mannu e forte. Similarly, the contradiction between biancu and bruttu observed in (54b) but absent in (54a) re-emerges in (64): (64) ?unu mutzighinu biancu e bruttu [a handkerchief white and dirty] Interestingly, these remarks do not apply to the following examples, where mannu can mean ‘elder’ and biancu is compatible with the other adjectives: (65) a une frate mannu, forte e abbistu [a brother elder strong and clever] b unu mutzighinu biancu, bruttu e crentiatu [a handkerchief white dirty and stained] This is consistent with our analysis if we postulate that the first adjective is not conjoined with the other two, but is adjoined independently to the minimal N′, as in (66): (66) [NP unu [N′ [N′ frate mannu] [ AP forte e abbistu]]]

< previous page

page_55

next page >

< previous page

page_56

next page >

Page 56 In other words, these examples involve a combination of the recursive adjunction and conjunction strategies discussed above. 2.2.5 NPs without determiners NPs which lack a determiner can be divided into two types: I cases where omission of the determiner is conditioned entirely by internal properties of the NP; IIcases where omission of the determiner is contingent on factors external to the NP. Clear instances of the former type are personal pronouns and certain types of proper nouns (forenames and names of towns). Surnames accompanied by a professional title or other form of address also lack a determiner, but the article is obligatory with such titles when there is no surname: (67) a Segnora Medda est andata kin duttore Pira a vídere a professore Lai. ‘Mrs Medda has gone with Dr Pira to see Professor Lai.’ b Cussa segnora est andata kin su duttore a vídere su professore. ‘That lady has gone with the doctor to see the professor.’ This distinction also extends to vocative uses: e.g. (O) su duttore, non mi fakites una puntura! ‘Doctor, do not give me an injection!’ vs. (O) duttore Pira, non mi fakites una puntura! A further example of type I is the omission of the definite article before kinship or endearment terms accompanied by a pronominal possessive (cf. 2.1.5): (68) Frate meu at faeddatu kin mannedda tua. ‘My brother talked with your grandmother.’ A more restricted class of kinship terms (referring to parents or grandparents) can occur without a determiner even in the absence of a possessive when the family relationship is with the speaker: (69) Mama at telefonatu a manneddu. ‘Mother telephoned grandfather.’ Note that inclusion of the definite article before manneddu in (69) would imply a a kinship relation with some other person prominent in the discourse, typically the subject: Mama at telefonatu a su manneddu ‘Mother telephoned her grandfather’.

< previous page

page_56

next page >

< previous page

page_57

next page >

Page 57 Turning now to cases of type II, where omission of the article is conditioned by factors external to the NP, the determiner is frequently omitted after certain prepositions. With the locative prepositions a ‘to’, in ‘in, on, at’ and dae ‘from, out of, such expressions have a semiidiomatic flavour—though possible with a large number of nouns, the construction is not fully productive: e.g. in domo ‘at home’, a lettu ‘to bed’, dae cukina ‘from the kitchen’, in mesa ‘at table, on the table’, in bidda ‘in town’, in televisione ‘on television’, a iscola ‘to school’, in continente ‘on the (Italian) mainland’, but *in mesedda ‘on the little table’, *in libru ‘in the book’, *a universitate ‘to the university’, *in isula ‘on the island’. Also, some nouns appear to be selective with respect to the prepositions which they can take in such constructions; e.g. we have Semus in dzardinu ‘We are in the garden’ and Semus issitos dae dzardinu ‘We went out of the garden’, but our informants reject *Semus andatos a dzardinu ‘We went to the garden’. Pronominal possessives may occur in such constructions with most, but not all, nouns: (70) a Andamus a domo nostra. ‘Let us go to our house.’ b Vae a lettu tuo! ‘Go to your bed!’ c *Ses in mesa nostra. ‘You are at our table.’ Possessives of the type de+NP are not generally allowed in such constructions except in expressions such as alin domo de Juanne ‘to/at John’s house’ (often contracted to alin doe Juanne) and locutions involving a noun denoting a part or dimension of an object (e.g. in fundu de su mare ‘at the bottom of the sea’) where the sequence P + N+de may be regarded as a complex preposition (cf. 4.3.2). Modifiers other than possessives are excluded: (71) a *Andamus a domo manna. ‘Let us go to the big house.’ b *Juanne est in domo accurtz’a mare. ‘John is in the house by the sea.’ Names of countries or regions normally omit the article after locative prepositions, though the definite article is required in other contexts: (72) a Semus in Sardigna. ‘We are in Sardinia.’

< previous page

page_57

next page >

< previous page

page_58

next page >

Page 58

b Sun andatos a Frantza. ‘They went to France.’ c Est ghiratu dae Corsica. ‘He returned from Corsica.’ (73) a Sa Sardigna est un’isula bella. ‘Sardinia is a beautiful island.’ b Connosco bene sa Frantza. ‘I know France well.’ Similarly, names of languages or dialects omit the article after in, but require the article in other contexts: (74) a Custu libru est iscrittu in tedescu. ‘This book is written in German.’ b Cussa paráula venit dae su catalanu. ‘This word comes from Catalan.’ c Su sardu est una limba antica. ‘Sardinian is an ancient language.’ The determiner is also omitted after a or in in expressions denoting means of locomotion: in mákkina ‘by car’, in aéreo ‘by plane’, a pede ‘on foot’, a caddu ‘on horseback’. The preposition de may take an NP without a determiner in partitive constructions like una dudzina de ovos ‘a dozen eggs’ or unu cantu de petha ‘a piece of meat’ (see above 2.2.2) and in various expressions such as the following: una domo de linna ‘a wooden house’ (lit. ‘a house of wood’), unu corfu de fuste ‘a blow with a stick’ (lit. ‘a blow of stick’), un’ómine de onore ‘a man of honour’, un’istória de amore ‘a story of love’, su mere de domo ‘the master of the house’. NPs governed by kene ‘without’ usually lack a determiner unless they have specific reference: (75) a Semus kene mákkina. lit.‘We are without car.’ ‘We have no car (at the moment).’ b So issitu kene crae(s). ‘I went out without (my) key(s).’ c Bio su caffé kene latte. ‘I drink coffee without milk.’ Count nouns after kene normally take the singular form (as in (75a)) unless there is a natural expectation that there should be more than one of the items referred to: e.g. una domo kene barcones ‘a house without windows’. The article is normally omitted in certain types of predicate nominal, notably those which denote professions:

< previous page

page_58

next page >

< previous page

page_59

next page >

Page 59 (76)

a Marieddu est pastore. ‘Marieddu is a shepherd.’ b Juanne est diventatu sordatu. ‘John became a soldier.’ c Lu credío prítteru. lit.‘I believed him priest.’ ‘I thought he was a priest.’ The same construction is possible with other nouns modified by adjectives of an evaluative type: (77) a Custa est notítzia importante. ‘This is important news.’ b Teresa est diventata bella pitzinna. ‘Teresa has become a lovely girl.’ c Ti credío pitzinneddu bravu. ‘I thought you were a good little boy.’ Also as predicate nominals we find ‘bare’ nouns modified by a pronominal possessive: (78) a Custu est lettu meu. ‘This is my bed.’ b Cuddu est libru tuo. ‘That is your book.’ The NPs in (76)-(78) cannot be used referentially: *Appo vistu pastore ‘I saw a shepherd’, *Prítteru est vénnitu ‘A priest came’, *Appo faeddatu a bella pitzinna ‘I talked to a lovely girl’, etc. In particular, the possessive constructions in (70) and (78) have a much more restricted distribution than the superficially similar constructions with kinship nouns discussed above (see example (68)): *Lettu meu est cómmodo ‘My bed is comfortable’, *Appo lessu libru tuo ‘I read your book’ vs. Frate meu est riccu ‘My brother is rich’, Appo vistu a sorre tua ‘I saw your sister’. Verbs of possession, when negated, can also take direct objects without a determiner, under conditions similar to those observed with kene in (75): (79) a Non b’amus mákkina. ‘We have not got a car.’ b Maria non juket bonette. ‘Mary has not got a hat (on).’ Similarly, the determiner may be absent in yes/no questions, particularly when the direct object is fronted:

< previous page

page_59

next page >

< previous page

page_60

next page >

Page 60 (80)

a Mákkina b’as? ‘Have you got a car?’ b Bonette jukías? ‘Did you have a hat on?’ Plural and non-count nouns can occur without a determiner as objects of verbs and prepositions: (81) a Appo mandicatu casu. ‘I ate cheese.’ b Laura est toddende frores. ‘Laura is picking flowers.’ c No’amus comporatu pane. ‘We did not buy (any) bread.’ d Mandico sa pasta kin casu. ‘I eat pasta with cheese.’ e Semus ponnende vinu in ampullas. ‘We are putting wine in bottles.’ However, NPs of this type cannot occur in subject position: (82) a *Sórrikes an mandicatu su casu. ‘Mice ate the cheese.’ b *Vinu at crentiatu sa tiadza. ‘Wine stained the tablecloth.’ c *Pitzinnas sun arrivatas. ‘Girls have arrived.’ Thus, although the absence of a determiner in (81) is due principally to an internal property of the NP (the plural or non-count nature of the head noun) it is nevertheless dependent on external factors. The restriction observed in (82) holds for all determinerless NPs of type II. We therefore propose the informal generalisation in (83): (83) Determinerless NPs of type II cannot function as subjects. Note that this restriction does not apply to the type I cases discussed at the beginning of this section (i.e. personal pronouns, names of persons or towns, kinship terms modified by a possessive). For determinerless NPs of type II, one might be tempted to conclude that the restriction in (83) amounts to a requirement that such NPs must be governed by a lexical category (a verb or a preposition), subject NPs being governed by the non-lexical category INFL. However, NPs of this type can be fronted, as in (84), to a position which is presumably ungoverned: (84) a Pastore est. ‘He is a shepherd.’

< previous page

page_60

next page >

< previous page

page_61

next page >

Page 61 b Lettu meu est. ‘It is my bed.’ c Pane appo comporatu. ‘I have bought bread.’ d Frores so toddende. ‘I am picking flowers. In examples like those in (85), the initial NP can be analysed as the fronted complement of the impersonal construction in (86) rather than as a subject, as shown by the absence of agreement between this NP (plural) and the verb (singular), see 3.2.2: (85) a Pitzinnas b’ at arrivatu. ‘Girls have arrived.’ b Tenores b’ at cantatu. ‘Tenors have sung.’ (86) a B’at arrivatu pitzinnas. ‘There have arrived girls.’ b B’at cantatu tenores. ‘There have sung tenors.’ Predicate NPs without a determiner can be dislocated, with the masculine singular accusative clitic lu as the resumptive pronoun, though a fairly ‘loaded’ discourse context is required: (87) a Ma, dottoressa, ja lu ses, non veru? b Ma, ja lu ses, dottoressa, non veru? ‘But, surely you are a doctor (f.), aren’t you?’ On the other hand, indefinite NPs with plural or non-count nouns and negated or questioned NPs of the type in (79)-(80) are normally introduced by de when dislocated: (88) a Non nd’amus comporatu, de pane. ‘We did not buy any bread.’ b De frores non nd’appo tódditu. ‘I did not pick any flowers.’ c Non bi nd’amus, de mákkina. ‘We have not got a car.’ d De mákkina, non bi nd’as? ‘Have you not got a car?’ The generalisation proposed in (83) has a number of consequences. In particular, we must assume that cardinal numerals can occur under the DET node as well as functioning as [+spec] adjectives, as proposed in 2.2.2, since they can readily introduce subject NPs:

< previous page

page_61

next page >

< previous page

page_62

next page >

Page 62 (89) Tres sórrikes an mandicatu tottu su casu. Three mice ate all the cheese.’ Similarly, we must conclude that the items tzertos ‘certain’ and átteros ‘other’, in the plural, can function as determiners (though in the singular they can only function as [+spec] adjectives preceded by a determiner: una tzerta nave ‘a certain ship’, un ’áttera nave ‘another ship’): (90) Tzertas naves arrivan su mandzanu, átteras naves arrivan su sero. ‘Some ships arrive in the morning, other ships arrive in the evening.’ The noun adziccu (used to indicate a small quantity with non-count nouns—cf. 2.2.2) has an idiosyncratic property which accords with the restriction in (83). In direct object position and after prepositions this item can occur with or without the indefinite article: (91) Appo mandicatu (un’) adziccu de pane kin (un’) adziccu de casu. ‘I ate a little bread with a little cheese.’ However, in subject position, the indefinite article is obligatory: (92) Un‘adziccu de pane est rutt’a terra. ‘A bit of bread fell on the floor.’ A similar subject-object asymmetry is found with meta ‘much, many’ and pacu ‘little, few’. As noted in 2.1.2, these items can occur postnominally (as well as in the canonical, prenominal determiner position): (93) a Juanne at bitu vinu pacu. ‘John drank little wine.’ b Amus vistu turistas meta(s). ‘We saw many tourists.’ However, the postnominal use of these items appears to be much less acceptable in subject position: (94) a ?Vinu pacu est restatu. ‘Little wine is left.’ b ?Turistas meta(s) sun vénnitos. ‘Many tourists came.’ Some of our informants rejected examples like (94) completely, whereas others found them acceptable, but much less natural than their counterparts with prenominal meta or pacu.

< previous page

page_62

next page >

< previous page

page_63

next page >

Page 63 One way of accounting for the deviance of (94) would be to postulate that postnominal meta and pacu are not determiners (dominated directly by NP) but adjectives adjoined to the N′ node, so that the NPs in (93)(94) are determinerless plural or non-count NPs like those in (81)-(82) except that they contain an adjective indicating quantity. According to this analysis, the relative unacceptability of (94) would follow from the restriction in (83). Note also that NPs with postnominal meta and pacu can be fronted, just like determinerless NPs of type II (cf. (84)-(85)): (95) a Vinu pacu b’ at restatu. lit.‘Wine little there has remained.’ There is little wine left.’ b Turistas meta(s) b’ at vénnitu. lit.Tourists many there has come.’ ‘Many tourists have come.’ An advantage of this account is that we do not need to postulate a DET position to the right of N’. Moreover, the adjectival status of postnominal meta and pacu is independently supported by examples like the following where these items are conjoined with other adjectives (example (96b) is taken from a nursery rhyme): (96) a Occannu sa patata est paca e minoredda. ‘This year the potatoes (lit. the potato) are few and tiny.’ b pilos de seta, longos e meta ‘hair of silk, long and plentiful’ Nevertheless, the fact that examples like (94) are marginally acceptable for some speakers may indicate that their deviance is pragmatic rather than syntactic. Given our earlier observation (p. 36) that the postnominal use of meta and pacu tends to emphasise the notion of quantity, it is natural that they should occur predominantly in positions which are typically associated with focus and main stress (for example, clausefinal or fronted NPs rather than preverbal subjects). A similar problem arises with the negative ‘determiner’ perunu which can also occur postnominally (see p. 37). The existence of a subject-object asymmetry with this item is more difficult to establish independently of other factors since NPs containing this item tend to occur postverbally, within the scope of the negative operator non, in any case—as in (97):

< previous page

page_63

next page >

< previous page

page_64

next page >

Page 64 (97)

a No’appo lessu perunu libru. b No’appo lessu libru perunu. ‘I did not read any book.’ Nevertheless (98a) appears to be significantly worse than (98b): (98) a *Libru perunu m’est piághitu. b Perunu libru m’est piághitu. ‘No book pleased me.’ Given that singular count nouns can occur without a determiner in certain negative contexts, (see (75) and (79)), it is possible that postnominal perunu functions as an adjective which emphasises the negation of the NP (in much the same way as English single in I did not read a single book) rather than as a determiner. However, we do not have independent evidence (of the sort given in (96) for meta and pacu) that perunu can function as an adjective. Note also that the use of ‘bare’ count nouns in negative contexts is possible only with verbs of possession (as in (79)). Consequently, libru perunu in (97b) is not strictly analogous to genuine cases of negated nouns without a determiner. We therefore leave open the status of postnominal perunu and the reason for the deviance of (98a). In terms of structure, we may envisage three possible ways of representing NPs which lack a determiner. One possibility is that the determiner position is simply absent, giving structures of the type [NP [N′… N…]]. We assume that this analysis is valid for all cases of type I (proper names, personal pronouns, kinship nouns followed by a possessive). A second possibility is that the determiner node is present, but is left empty: NP[DET θ] [N′…N…]. This approach has some plausibility for cases where the missing determiner contrasts with the definite article. Thus, we may identify θ as the form of the indefinite article which occurs with plural and non-count nouns, whereas with singular count nouns it functions either as a negative or interrogative polarity item (rather like English any) or it indicates that the NP is to be interpreted predicatively rather than referentially. A final possibility is that in some cases the NP node itself is absent (i.e. the nominal expression is simply an N or N′), a structure which may be appropriate for semi-idiomatic collocations of the type a lettu ‘to bed’, in domo ‘at home’, etc. We do not wish to claim that these three structural analyses are ‘correct’, but we shall tentatively adopt them as a means of distinguishing between different types of NPs which lack a determiner.

< previous page

page_64

next page >

< previous page

page_65

next page >

Page 65 2.2.6 Prepositional accusatives The term ‘prepositional accusative’ refers to the phenomenon illustrated in (99), where a direct object is introduced by the preposition a (see Rohlfs 1971 for an overview of this phenomenon in the Romance languages and Bossong 1982 for discussion of the phenomenon in Sardinian): (99) Vido a Juanne. ‘I see John.’ It is important to distinguish this use of a from cases like (100) where a introduces an indirect object: (100) Cussu piaghet a Juanne. ‘That pleases John.’ Whereas replacement of Juanne by a clitic in (99) yields the accusative form lu, the dative form li is required in (100): Lu vido ‘I see him’, Li piaghet ‘It pleases him’. In cases like (100), a is selected as a property of the verb and must be present regardless of the nature of the NP. On the other hand, the presence of accusative a (as in (99)) is conditioned primarily by properties of the NP (but see 3.2.1). Essentially, accusative a may precede direct objects of all verbs, including áere ‘have’ and verbs which also take a dative object (Tue as a Maria ‘You have Mary’, Appo presentatu a Juanne a Lukia ‘I introduced John to Lucy’), and also the complements of some prepositions (see 4.3.2). However, accusative a does not occur before all types of NP: (101) a Tue as (*a) una mudzere. ‘You have a wife’ b Appo mandatu (*a) sa líttera a Olbia. ‘I sent the letter to Olbia.’ c Vido (*a) su cane. ‘I see the dog.’ The conditions under which a is prefixed to a direct object NP can be formulated as follows: (102) a Accusative a occurs obligatorily before determinerless NPs of type I (as defined in 2.2.5). b For some speakers, accusative a occurs optionally before definite NPs with human reference. There is a substantial overlap between the two cases in (102) since

< previous page

page_65

next page >

< previous page

page_66

next page >

Page 66 most NPs which fall within condition (a) have definite human reference (e.g. proper names, personal pronouns, kinship terms), as in (99) and the following examples where a is obligatory: (103) a Appo vistu a segnora Ledda. ‘I saw Mrs Ledda.’ b Non connosco a issa. ‘I do not know her.’ c Istimo a frate meu. ‘I love my brother.’ d Non connoskío a manneddu. ‘I did not know (my) grandfather.’ However, accusative a is likewise obligatory for NPs which satisfy (102a) but do not refer to humans (e.g. names of animals or places): (104) a An furatu a Ércole. ‘They have stolen Hercules (e.g. a horse).’ b Appo vistu a Nápoli. ‘I have seen Naples.’ On the other hand, the use of accusative a with NPs which conform to (102b) but not (102a) is at best optional, often marginal, and for some speakers totally excluded. The most readily accepted examples of this type are singular NPs consisting solely of a definite article and a noun which function as conventional titles for particular individuals in a given social context (e.g. su re ‘the King’, su mere ‘the boss’, su duttore ‘the doctor’, su postinu ‘the postman’, etc.): (105) a An assassinatu (a) su re. ‘They have assassinated the King.’ b Maria at vistu (a) su duttore. ‘Mary saw the doctor.’ The use of accusative a with such NPs tends to be more prevalent when the NP is fronted or left-dislocated, as in (106), perhaps because it provides an overt indication that the initial NP is an object rather than a subject: (106) a (A) su re an assassinatu. b (A) su re, l’an assassinatu. On the other hand, accusative a is much less acceptable with plural NPs and NPs which rely on deictic cues or the presence of a modifier in order to identify the intended referent: (107) a Appo vistu (?a) sos sordatos. ‘I saw the soldiers.’

< previous page

page_66

next page >

< previous page

page_67

next page >

Page 67 b Non connosco (?a) cuddu duttore. ‘I do not know that doctor.’ c Maria at iskiriatu (??a) su professore vetzu. ‘Mary chose the old teacher.’ d Appo addoppiatu (??a) s’ómine k’at iscrittu cussu libru. ‘I have met the man who wrote that book.’ Non-human NPs introduced by a definite determiner never allow accusative a: (108) Appo vistu (*a) su cane. ‘I saw the dog.’ The evidence presented so far suggests that condition (102a) represents the ‘core’ cases of the prepositional accusative phenomenon. Given the structural distinctions between different types of determi-nerless NPs envisaged at the end of 2.2.5, this condition can be stated in purely structural terms, as in (109): (109) Accusative a is obligatory before direct object NPs which lack a determiner position. This formulation correctly excludes the use of accusative a before NPs for which we have postulated an empty determiner position: (110) a Juanne est (*a) duttore. ‘John is a doctor’ b Non b’amus (*a) re. ‘We do not have a king.’ c Appo invitatu (*a) dzente. ‘I have invited people.’ The cases of accusative a defined by the semantically based condition (102b) represent a much more peripheral phenomenon. Even the ‘optimal’ uses in (105)-(106) are not acceptable to all speakers. Moreover, the fact that these examples involve ‘titles’, which are similar in function to proper names, suggests that this use of accusative a is an analogical extension of the core use with proper names. Thus, although statistically an overwhelming proportion of prepositional accusatives refer to humans, human reference itself plays a very marginal role in triggering the presence of accusative a. There is however one range of cases which appears to undermine this conclusion. Demonstratives and the quantifier tottu, when used without a following nominal expression, require accusative a when the referent is human but do not allow accusative a otherwise: (111) a Non connosco a custu. ‘I do not know this (person).’

< previous page

page_67

next page >

< previous page

page_68

next page >

Page 68

b Appo salutatu a tottu. ‘I greeted everybody.’ (112) a Appo fattu custu. ‘I did this.’ b Appo mandicatu tottu. ‘I ate everything.’ Similarly, the interrogative and negative pronouns kie ‘who’ and nemos or neune ‘nobody’ require accusative a, but their non-human counterparts itte ‘what’ and nudda ‘nothing’ do not take the prepositional accusative: (113) a A kie as vistu? ‘Who did you see?’ b No’appo vistu a nemos/neune. ‘I saw nobody.’ (114) a Itte as vistu? ‘What did you see?’ b No’appo vistu nudda. ‘I saw nothing.’ Ostensibly, in these cases the [±human] feature is decisive in determining the obligatory presence or absence of accusative a. However, in 5.1.2 we shall present an analysis which does not require us to draw this conclusion but which accounts for the data in (111)-(114) in terms of the conditions proposed in (102). The essence of this analysis is that kie, nemos/neune and, with human reference, tottu and the demonstratives are pronouns occupying the head N position (i.e. determinerless NPs of type I which take accusative a by virtue of condition (102a)), whereas itte, nudda and the non-human instances of tottu and the demonstratives are determiners or quantifiers which govern an empty category (i.e. [NP DET[N,θ]] or [NP Q[NPθ]]) and consequently do not fall within either of the conditions in (102). We shall defer detailed discussion of this account until section 5.1.2. For the moment we may note that the determiner status of itte and nudda is independently motivated by examples like itte mákkina ‘what car’ and nudda vinu ‘no wine’ (see above 2.1.2) whereas there is no evidence that kie and nemos/neune can function as anything but pronouns: *kie dzente ‘who people’, *nemos/neune dzente ‘nobody people’.

< previous page

page_68

next page >

< previous page

page_69

next page >

Page 69 2.2.7 Headless NPs There are a variety of cases where an NP may lack an overt head noun. As a working hypothesis we assume that these conform to the structure (28) proposed in 2.2.1 except that the the N′ node (or possibly N) is left unexpanded. This empty category can be viewed as a null proform similar in function to the overt proform one in English. In this section we shall concentrate on cases where the determiner is a definite article or a demonstrative. Indefinite determiners can occur in such constructions, but typically the partitive clitic nde (which we take to correspond to the missing head) must be attached to the verb, see 5.2.1 (pp. 216–18). Definite articles can introduce headless NPs in which the empty element is accompanied by a modifier, as in the following examples for which we propose a structure of the form [NP DET[N' [N' θ] X]]: (115) a Cudda mákkina est prus manna de sa de Juanne. ‘That car is bigger than John’s.’ b Sa pinna tua est niedda ma sa mea est ruja. ‘Your pen is black but mine is red.’ c Imbetzes de sa meledda manna, appo mandicatu sa minore. ‘Instead of the big apple, I ate the small one.’ d Custu vinu est medzus de su k’appo comporatu eris. ‘This wine is better than that which I bought yesterday.’ Examples like the following, involving prenominal adjectives with a specifier-like function, can also be analysed as headless NPs with the structure (116) a Custa mákkina est prus manna de s’áttera. ‘This car is bigger than the other one.’ b Appo comporatu custas iscarpas e Juanne at comporatu sas mattessi. ‘I bought these shoes and John bought the same ones.’ (i.e. shoes of a similar pattern).’ Prenominal adjectives of the affective type appear to be impossible in headless NPs. For instance, we do not have *su bette ‘the huge one’, and the phrase su póveru cannot have the meaning associated with the prenominal use of póveru (‘to be pitied’) but can only refer to someone who is indigent (though arguably póveru can be analysed as a noun (‘pauper’) in this case, a matter to which we shall return,

< previous page

page_69

next page >

< previous page

page_70

next page >

Page 70 p. 75). If prenominal affective adjectives are sisters to the head N, as proposed in 2.2.1, this restriction can be accounted for by postulating that the empty element is always an N′, not simply an N. However, we also find cases like (117) where the empty element is accompanied by a PP which is presumably a complement (hence a sister of N) rather than a modifier: (117) Su professore de frantzesu at faeddatu kin su de italianu. ‘The teacher of French talked with the one of Italian.’ Such examples suggest that the empty element can be simply an N. Headless NPs of the sort illustrated above are also possible with demonstrative articles, though the range of possible modifiers is more restricted. In particular, headless demonstratives cannot be modified by a pronominal possessive, though PP possessives are sometimes possible, particularly with cuddu ‘that’ whose effect in such cases seems to be (mildly) pejorative rather than deictic: (118) a *Custa mákkina est prus manna de cudda tua. ‘This car is bigger than that of yours.’ b Custa mákkina est prus manna de cudda de Juanne. ‘This car is bigger than that one of John’s.’ Also, demonstratives do not allow the headless construction with a complement PP: (119) *Cuddu professore de frantzesu at faeddatu kin custu de italianu. That teacher of French talked with this one of Italian.’ At the end of section 2.2.5 we proposed that ‘bare’ demonstratives (and certain other proforms) with nonhuman reference should be analysed as headless NPs consisting of a determiner and a null N′ (i.e. [NP DET[N′ θ]] If this analysis is correct, we may conclude that demonstratives differ from definite articles in that they can govern an N‘which is completely empty whereas definite articles require the N′ to contain a modifier or a complement (e.g. we do not have *Appo vistu su lit. ‘I saw the’ on a par with Appo vistu cussu ‘I saw that’). A related difference is that modifiers with definite articles must have a restrictive interpretation (conveying information which is necessary in order to identify the intended referent) whereas with demonstratives the modifier may be appositive (providing extra information about the entity which has been identified independently). Thus, in the following examples, where the object NP refers to, say, handkerchieves, bruttos in (120a) must denote a property which distinguishes

< previous page

page_70

next page >

< previous page

page_71

next page >

Page 71 the handkerchieves in question from other (clean) handkerchieves, whereas in (120b) the adjective can be interpreted as asserting the additional information that the handkerchieves in question are dirty: (120) a Non pikes sos bruttos. ‘Do not take the dirty ones.’ b Non pikes cussos bruttos. ‘Do not take those dirty ones.’ If appositive modifiers are external to the N′ (e.g. adjoined to NP), both differences can be subsumed under a structural condition to the effect that definite articles must govern an N′ which contains overt descriptive material, whereas demonstratives need not. In most of the examples given so far, the null proform has an antecedent from which it takes its semantic value (for example, in (115a) the phrase sa de Juanne is understood as referring to a car by virtue of the earlier NP cudda mákkina ‘that car’). In cases like (120), with the reading assumed above, there is no antecedent within the sentence, but the reference of the headless NP is determined by a discourse topic which may or may not have been mentioned explicitly in the preceding discourse. In some cases, there is no contextual cue (even of an implicit nature) clear examples being headless relatives introduced by the definite article su (invariable in this case) where the relative clause itself provides all the information necessary to establish reference: (121) a Juanne at pérditu su k’aíat vintzu. ‘John has lost what he had won.’ b Su k’as natu m’at offesu. ‘What you said offended me.’ After some verbs which can take either an object NP or a complement clause, this construction is more or less interchangeable with the indirect question construction introduced by itte ‘what’ (cf. 6.1.8): (122) a No’isco su ki devo fákere. b No’isco itte devo fákere. ‘I do not know what I must do.’ (Note, however, that itte cannot replace su ki in cases like (121).) In the absence of any contextual cue, there is an interesting correlation between number and the [±human] distinction. The plural expressions in (123) allow a generic or variable interpretation which is not dependent on an antecedent in the discourse (for example, ‘whoever stole the money’ in (123b)):

< previous page

page_71

next page >

< previous page

page_72

next page >

Page 72 (123)

a Sos ki mandican troppu s’ingrassan. ‘People who eat too much get fat.’ b Sos k’ an furatu su dinari an a éssere punitos. ‘The people who stole the money will be punished.’ However, the singular expression in (124) must be interpreted as picking out a particular member of a set identified in the preceding context: (124) Su k’at furatu su dinari at a éssere punitu. ‘The one who stole the money will be punished.’ Conversely, with non-human expressions of this type, the generic, contextually independent interpretation illustrated in (121) is possible only in the singular. Thus, in (125) the value of the missing N′ must be recoverable from context (e.g. as jocátolos ‘toys’): (125) Juanne at pérditu sos k’aíat vintzu a sa dzostra. ‘John lost the ones that he had won at the fair.’ For further discussion of the referential properties of headless relative constructions, see 6.2.2. The correlation between plurality and free human reference can also be seen in headless NPs modified by a possessive. Such NPs can occur with the full range of forms and referential properties (singular or plural, human or non-human, masculine or feminine) when the value of the missing head is recoverable from discourse context. However, the masculine plural forms of such NPs also have a conventional interpretation, denoting a collective kinship relation, which is independent of context (e.g. sos meos ‘my family’, sos de Juanne ‘John’s family’): (126) a Sos meos m’an telefonatu. ‘My family telephoned me.’ b Appo invitatu sos de Juanne. ‘I invited John’s family.’ Similarly, locative PPs introduced by de can define a collective entity in plural headless NPs: (127) a Sos de Lodé non sun meta furbos. ‘People from Lodé are not very crafty.’ b Amus invitatu sos de Lodé. ‘We have invited the people (e.g. our friends) from Lodé.’ Demonstratives are not normally possible in such cases, except for

< previous page

page_72

next page >

< previous page

page_73

next page >

Page 73 cuddu in the pejorative use mentioned with regard to (118b). PPs and possessives cannot generally be used in the singular with an independent interpretation of the sort found in free relatives like (121) (for example, su meu ‘mine’ must take its referent from an antecedent and cannot have an autonomous reading parallel to that of su k’appo ‘what I have’). However, the possessive item antzenu ‘other people’s’ does allow such an interpretation, with non-human reference as in free relatives like (121): (128) Non si devet invidiare s’antzenu. ‘One should not covet other people’s possessions.’ Headless NPs with possessives can also be used predicatively, as in (129), where the subject acts as the antecedent: (129) a Cussa pinna est sa mea. ‘That pen is mine.’ b Cudda mákkina est sa de Juanne. ‘That car is John’s.’ We also find similar examples where the definite article is absent (e.g. Cussa pinna est mea, Cudda mákkina est de Juanne), but in this construction we assume that the possessive is not a modifier within a headless NP but functions as the complement of the copular verb. In headless NPs with PP modifiers the preposition is always de when the determiner is a definite article; e.g. we do not have expressions like *su kin unu bonette ‘the one with a hat’, *su supra sa mesa ‘the one on the table’, *su in su calassu ‘the one in the drawer’, etc. In such cases, an appropriate relative clause would normally be used: su ki juket unu bonette ‘the one who has a hat’, su k’est supra sa mesa ‘the one that is on the table’, su ki b’ at supra sa mesa ‘what(ever) is on the table’, etc. Note also that demonstratives can sometimes be modified by a PP headed by an item other than de, particularly if the PP is interpreted as appositive rather than restrictive: cuddu kin unu bonette ‘that one with a hat’, custu supra sa mesa ‘this one on the table’, etc. A similar restriction, though of a semantic or pragmatic nature, is also observed in cases involving attributive adjectives, as in (120) above. Intuitively, in order to modify an empty N′ the adjective must denote some fairly concrete property which the addressee can easily recognise as distinguishing the intended referent from other potential referents, typical examples being colour terms and other ‘basic’ adjectives, particularly those which imply a binary paradigmatic contrast: e.g. mannu ‘big’, minore ‘small’; longu ‘long’, curtzu ‘short’;

< previous page

page_73

next page >

< previous page

page_74

next page >

Page 74 vetzu ‘old’, novu ‘new’, dzóvanu ‘young’; bruttu ‘dirty, ugly’, nettu ‘clean’, bellu ‘beautiful’, etc. Adjectives which express more subtle, subjective properties or which are inherently superlative are much less natural: ? s’ interessante ‘the interesting one’ (book), ?su diffítzile ‘the difficult one’ (problem), ?s’ enorme ‘the enormous one’ (lorry), ?su delitziosu ‘the delicious one’ (dish), etc. Also, the presence of a degree modifier such as meta tends to reduce acceptability: (130) Imbetzes de cudda mákkina, appo comporatu sa (?meta) manna. ‘Instead of that car, I bought the (very) big one.’ On the other hand, modification by the comparative/superlative item prus enhances acceptability, allowing the use of some adjectives which would otherwise be unnatural: (131) a M’an offertu duos libros, e jeo appo lessu su prus interessante. ‘They offered me two books, and I read the more interesting one.’ b Juanne at fattu tottu sas provas, fintzas sa prus diffítzile. ‘John did all the tests, even the most difficult one.’ The above restrictions concerning the range of adjectives which can modify an empty N′ do not apply in headless NPs which are contextually independent. Thus, although adjectives like interessante and diffítzile cannot readily be used to pick out a particular referent from a previously mentioned set (unless modified by prus as in (131)), they can be used with the definite article su (invariable) in a manner similar to that of the free relatives in (121): (132) a S’interessante est ki ses inoke. ‘What is interesting (important) is that you are here.’ b Su diffítzile est de cumbínkere a Juanne. ‘The difficult thing is to convince John.’ c Non poto fákere s’impossibile. ‘I cannot do the impossible.’ Similarly, the adjective coradzosu ‘courageous’ is odd in examples like (133a) (unless we addprus), but is perfectly natural when used in the plural with a generic interpretation which is not dependent on context, as in (133b): (133) a ?Duos ómines fin gherrande, e su coradzosu at vintzu. ‘Two men were fighting, and the courageous one won.’ b Sos coradzosos vinken semper. ‘The courageous always win.’

< previous page

page_74

next page >

< previous page

page_75

next page >

Page 75 The generic, plural construction in (133b) appears to be possible with all adjectives which can be used to describe human beings. There is, however, a more restricted class of adjectives which can be used in the singular, with independent specific reference: (134) Custa ricca at datu milli francos a su vetzu. ‘This rich woman gave a thousand lira to the old man.’ There is some evidence that the ‘adjectives’ in such constructions are actually nouns; i.e. there is a restricted class of items which can be used both as adjectives and as nouns (with the meaning ‘person who is adj.’): e.g. maccu ‘mad, a madman’, maláidu ‘sick, a sick person’, póveru ‘poor, a pauper’ etc. (and corresponding feminine forms such as ricca in (134)). First, these items can occur with the full range of determiners, including those such as dondzi ‘every’ which cannot be used in ‘genuine’ headless constructions (*Dondzi k’appo comporatu fit guastu ‘Every one that I bought was rotten’—in such cases dondzi must be accompanied by unu ‘one’, as in dondzunu k’appo compor-atu; see 5.1.3): (135) Su guvernu devíat dare una pensione a dondzi vetzu. ‘The government should give a pension to every old person.’ Similarly, these items can occur with indefinite determiners such as metas ‘many’ without the partitive clitic nde, as in (136a), but nde is obligatory in cases like (136b) which are similar to headless NPs in that they imply reference to some previously mentioned class (e.g. men, dogs, pigs,…): (136) a In cussu locu amus vistu metas maláidos. ‘In that place we saw many sick people.’ b In cussu locu nde amus vistu metas maláidos. ‘In that place we saw many sick ones.’ We suggest that maláidos is a noun in (136a) but an adjective in (136b), modifying an empty N′ which, in indefinite NPs, must be bound by the clitic nde. A similar distinction between nominal and adjectival uses of these items can be seen in cases where the item in question is accompanied by a degree modifier, such as prus ‘more’ or meta ‘very’, which can only modify adjectives. For instance, if we modify either of the relevant items in (134), a contextually dependent interpretation is imposed: (137) a Custa prus/?meta ricca at datu milli francos a su vetzu. ‘This richest/very rich (f.) one gave a thousand lira to the old man.’

< previous page

page_75

next page >

< previous page

page_76

next page >

Page 76 b Custa ricca at datu milli francos a su prus/?meta vetzu. ‘This rich woman gave a thousand lira to the oldest/very old one (m.).’ The marginal status of the versions of (137) with meta reflects the general restriction noted with regard to (130) above. The important point is that a specific, contextually independent interpretation for the items in question is excluded in just those cases where they must be analysed as adjectives modifying a null N′. Note finally that if the items in question are nouns in cases like (134), the correlation between number and human versus non-human reference in examples involving relative clauses, PPs and possessives can be generalised to cover adjectives as well. Thus, we may postulate the following generalisation: (138)In constructions of the form [NP DET θ X] where the value of θ is not recoverable from an antecedent, θ has the default value [+human] when the NP is plural, but [—human] when the NP is singular. To this generalisation we must add the caveat that in some cases the default value of may be unavailable for other reasons. For example, as we have noted, PPs and possessives (other than antzenu in (128)) do not appear to allow an interpretation parallel to that of free relatives (e.g. (121)) where θ has the non-human default value, so that singular headless NPs with such modifiers can only have a contextually dependent interpretation. 2.2.8 Extraposition of the head noun A further type of NP construction is illustrated in (139), where the head noun (or N′) is introduced by de and is placed to the right of a modifier or complement which, in other constructions, would normally follow the head noun: (139) a Préstami sa tua de pinna. ‘Lend me your pen.’ b Keljo comporare sa ruja de mákkina. ‘I want to buy the red car.’ c An furatu sos de Juanne de libros. They have stolen John’s books. d Devimus mandicare sos k’amus comporatu eris de ovos. ‘We must eat the eggs which we bought yesterday .’ This construction is fairly common in colloquial speech and is used

< previous page

page_76

next page >

< previous page

page_77

next page >

Page 77 primarily to emphasise the modifier (as indicated by italics in the translations) at the expense of the head noun (which is typically redundant and always unstressed). Note that the postposed head noun has the same number features as it would in its canonical position. In particular, the use of the singular in cases like (139a, b) suggests that these constructions are not analogous to partitive constructions such as una de sas pinnas meas ‘one of my pens’ where the plural form is used to denote the set of entities to which the intended referent belongs. We propose that these constructions are essentially headless NPs of the type discussed in 2.2.7, accompanied by a phrase which specifies the value of the null head in much the same way that rightdislocated phrases retrospectively identify the referent of a pronoun within the core of the sentence (see 7.1.2). The range of determiners and modifiers which can occur in this construction is identical to that found in headless NPs. Thus, preverbal adjectives of the [+spec] type are possible and, if the determiner is a demonstrative (but not a definite article), no modifier need be present: (140) a Mústrami s’atteru de libru. ‘Show me the other book.’ b Keljo comporare cudda de mákkina. ‘I want to buy that car.’ Determiners like dondzi ‘every’ which cannot govern a null head do not allow extraposition of the head either: (141) *Appo lessu dondzi de libru. ‘I have read every book.’ A further similarity is that the extraposed element appears to be able to function as an N rather than an N′, as in (142a), (see our earlier discussion of (117)), though other examples such as (142b) show clearly that extraposition can also apply to N′: (142) a Appo faeddatu kin su de frantzesu de professore. ‘I talked with the teacher of French.’ b Préstami sa tua de pinna ruja. ‘Lend me your red pen.’ There is a striking similarity between the constructions under discussion and cases of right-dislocation such as (143) where the dislocated expression de vinu specifies the value of the clitic nde which in turn represents the head N′ quantified by meta:

< previous page

page_77

next page >

< previous page

page_78

next page >

Page 78 (143)

Nde appo bitu meta, de vinu. ‘I drank a lot of wine.’ It would be tempting to analyse the constructions under discussion as instances of the same phenomenon, the only difference being that when the determiner is definite the null N′ is identified directly with the de phrase rather than through the intermediary of an overt clitic. However, there is an important structural difference between the two types of construction (hence our use of the distinct term ‘extraposition’ for the cases in question). In examples like (144) it is clear that the dislocated phrase is detached from the NP with which it is construed (e.g. adjoined to S—see 7.1.2), as can be seen in cases where this NP is fronted: (144) a Meta nde appo bitu, de vinu. b *Meta de vinu nde appo bitu. However, in cases of extraposition, the de phrase can be fronted along with the rest of the NP: (145) a Appo postu sa manna de ampulla in mesa. ‘I put the big bottle on the table.’ b Sa manna de ampulla appo postu in mesa. This evidence suggests that extraposed heads are adjoined to the NP, roughly as in (146), rather than to S (see Radford forthcoming for some discussion within a more recent theoretical framework): (146) [NP[NP DET(X) θ (X)][ de N′]] Nevertheless, although judgements are often rather hesitant, it is also possible for the de phrase to be detached from the rest of the NP: (147) a Sa manna appo postu de ampulla in mesa. b Sa manna appo postu in mesa de ampulla. It seems unlikely that the fronting process can simply extract the minimal NP in (146) leaving the rest of the NP in its original position, partly because of examples like (147b) where the de phrase is not in the normal direct object position, but also because of cases like (148) where the fronted sequence is not a constituent according to the structure in (149), which is simply (146) governed by a preposition: (148) Kin sa mea keljo iscríere de pinna. ‘I want to write with my pen.’ (149) [PP kin [NP [NP saθ mea] pinna]] Rather we assume, tentatively, that there is some process which

< previous page

page_78

next page >

< previous page

page_79

next page >

Page 79 allows the extraposed phrase to be reanalysed as a constituent of the dominating VP (with some freedom of order with respect to other constituents of the VP) prior to the application of movement rules such as fronting. The possible reordering effects of this process are exemplified, independently of other movement phenomena, in (150) which some speakers find acceptable, though less natural than (145a): (150) ?Appo postu sa manna in mesa de ampulla. In this connection we may note that detachment of the de phrase is not generally possible from the subject position: (151) a *Su tuo travallat bene de theraccu. ‘ Your servant works well.’ b *Su meu at appeddatu de cane. ‘ My dog has barked.’ This would follow automatically from the reanalysis approach envisaged above given that the subject NP is not dominated by VP. However, some apparently similar examples are accepted much more readily: (152) a Su meu sink’est andatu de theraccu. ‘My servant has gone away.’ b Su tuo no’est ghiratu de cane. ‘ Your dog has not returned.’ Although judgements regarding such examples are uncertain in some cases, it appears that the verbs which are most acceptable in this construction belong to the class of verbs which take éssere as their perfective auxiliary. In 3.2.3 we shall argue that the subjects of such verbs (i.e. ‘unaccusative’ verbs) are represented as objects within the VP in underlying structure and are subsequently moved to the preverbal position. Thus, examples like those in (152) can be accommodated if we assume that reanalysis of the de phrase as part of the VP applies before this movement takes place. Similarly, detachment of the de phrase from the subject is possible in ‘Middle’ constructions like (153) whose subject can also be claimed to have derived status (cf. 3.2.8): (153) Sas meas non si toccan de pinnas. ‘My pens are not to be touched.’

< previous page

page_79

next page >

< previous page

page_80

next page >

Page 80 3 The verb phrase 3.1 MORPHOSYNTAX OF THE VERB 3.1.1 Verbal flexion Regular verbs divide into three conjugation classes identified by the infinitive ending: I - áre, II - ere (with stress on the preceding syllable), III—íre. Class I corresponds to the Latin first conjugation (-ARE), class II to the second and third conjugations (-ĒRE, -ĔRE) and class III to the fourth conjugation (-IRE). The normal person inflections are given in Table 3.1. Table 3.2 lists full paradigms for regular verbs of each conjugation (with cantare ‘sing’, tímere ‘fear’ and pulire ‘clean’ as representative examples). Table 3.1 Person and number inflections 1 2 3 Singular -o (-e, -a) -s -t Plural -mus -tes (-dzis) -n There are many verbs, particularly in class II, whose paradigms are irregular, at least in part. In particular, many verbs have strong past participle forms: e.g. fattu (from fákere ‘do, make’), lessu (from léghere ‘read’), lintu (from línghere ‘lick’), postu (from pónnere ‘put’), etc. As Table 3.2 shows, there are just two inflectional tenses; present and imperfect. Some dialects also have past perfective paradigms of various types, but these are highly localised and untypical of the language as a whole. From a historico-morphological point of view, we can posit the same binary tense system in the subjunctive.

< previous page

page_80

next page >

< previous page Page 81 Table 3.2 Regular verb paradigms I Infinitive cantare Present Indicative canto cantas cantat cantamus cantates cantan Imperfect Indicative cantaío cantaías cantaíat cantaíamus cantaíadzis cantaían Past participle cantato Present participle cantande

< previous page

next page >

page_81 II tímere

III pulire

timo times timet timimus timites timen

pulo pulis pulit pulimus pulites pulin

timío timías timíat timíamus timíadzis timían

pulío pulías pulíat pulíamus pulíadzis pulían

tímitu

pulitu

timende

pulinde

page_81

I Subjunctive cante cantes cantet cantemas cantetas canten Subjunctive cantárepo cantares cantaret cantáremus cantáredzis cantaren Imperative canta cantate

II

III

tima timas timat timemas timetas timan

pula pulas pulat pulemas puletas pulan

timerepo tímeres tímeret timeremus timeredzis tímeren

pulírepo pulires puliret pulíremus pulíredzis pulire

time timite

puli pulite

next page >

< previous page

page_82

next page >

Page 82 However, syntactically, the so-called ‘imperfect subjunctive’ forms typically function as inflected forms of the infinitive rather than as past tense forms of the subjunctive, except with áere and éssere (see 6.1.5). Other distinctions of tense or aspect are expressed by means of auxiliary verbs, as described in the following sections. Present participles are invariable, but past participles inflect for number and gender under certain circumstances (see 3.1.2) with the endings -u, -a, -os, -as (as with type I adjectives—see 2.1.4). The principal dialectal variations (other than those which are due to purely phonological factors) can be summarised as follows. In southern Nuorese dialects (including Nuoro itself) the imperfect indicative paradigm for class I verbs is cantabo, -bas, -bat, -bamus, -bates, -ban and in classes II and III the first and second plural forms are timiabamus, -bates; puliabamus, -bates. A similar paradigm occurs in Campidanese with elision of -b- (cantabat > cantát) and mutation of the theme vowel in the first and second plural forms in classes II and III (timiabamus > timemus) . Many other Campidanese variants can be characterised in terms of elision of an intervocalic consonant: e.g. elision of -r- in the infinitive (cantái, tími(ri), pulí(ri)), of m- and -t- in first and second plural forms (cantáus ‘we sing’, cantáis ‘you (pl.) sing’, etc.) and of -t- in the past participle (cantáu, tímiu, pulíu)— these past participle forms are also standard in the dialect of Nuoro and surrounding areas. Conversely, Campidanese conserves the ending -nt (usually followed by an epenthetic vowel) in third person plural forms. A further feature of the Campidanese dialects which has direct syntactic relevance is that they lack the imperfect subjunctive forms given in Table 3.2 but substitute forms derived from the Latin pluperfect subjunctive (with the endings -essi, -essis, -essit, -éssimus, -éssidis, -essint) which can only be used as past tense subjunctives and do not have the inflected infinitive function typical of their Logudorese-Nuorese counterparts. Finally we may note that in Campidanese and Logudorese the class II present participle ending -ende (Campidanese -endi ) is generalised to all three conjugations (cantende, pulende). 3.1.2 Perfective and progressive aspect Perfective aspect is expressed by either áere ‘have’ or éssere ‘be’ followed by the past participle. The factors determining the choice of auxiliary verb will be discussed in detail in 3.3.1; for the moment, we can take áere as the unmarked auxiliary, éssere being used with some intransitive verbs and verbs accompanied by an accusative reflexive

< previous page

page_82

next page >

< previous page

page_83

next page >

Page 83 clitic. When the auxiliary is éssere the past participle must agree in number and gender with the subject, but with áere the participle agrees only with a third person accusative clitic: (1) a Sas pitzinnas sun arrivatas. ‘The girls have arrived.’ b Las appo vistas. ‘I saw them (f.).’ c Nos an vistu. ‘They saw us.’ As the translations given in (1) suggest, this tense form does not necessarily carry any implication that the event described is relevant to the present (i.e. it can correspond to the simple past in English as well as to the present perfect). The pluperfect is formed in the same way, with the auxiliary in the imperfect: (2) a Sas pitzinnas fin dza arrivatas. ‘The girls had already arrived.’ b Sos óspites aían dza mandicatu. ‘The guests had already eaten.’ A double perfective construction with two perfective auxiliaries (the first of which is in the imperfect) is sometimes used in conditional constructions in place of the straight pluperfect (typically only in the condition clause, though some speakers also accept this form in the consequent clause; see 6.2.3 for further discussion): (3) a Si los aío áppitos kérfitos, los aío (áppitos) comporatos. ‘If I had wanted them, I would have bought them.’ b Si fidzis istatos vénnitos prus kithu, aíadzis (áppitu) mandicatu kin nois. ‘If you had come earlier, you would have eaten with us.’ Note that in this construction both past participles undergo the agreement processes described above. Progressive aspect is expressed by éssere (in the present or imperfect) with the present participle: (4) a So travallande. ‘I am working.’ b Los fippo leghende. ‘I was reading them.’ This construction is extremely common and is often used in

< previous page

page_83

next page >

< previous page

page_84

next page >

Page 84 preference to the simple present or imperfect when describing actual (rather than typical or habitual) situations in the present or past, sometimes even with stative verbs: (5) a So pessande k’est tempus de dormire. lit. ‘I am thinking that it is time to sleep.’ b Non ti so cumprendende. lit. ‘I am not understanding you.’ ‘I do not understand what you are saying.’ Sometimes istare ‘be, stay’ is used with the present participle (like stare in Italian): (6) a Istaío travallande. ‘I was working.’ b Istan fakende su mándicu. ‘They are preparing the meal.’ This construction has a much more clearly ‘progressive’ value than the construction with éssere. It corresponds roughly to English be in the process of …and cannot be used with stative verbs. There are also a number of syntactic differences between the two progressive constructions, which we shall consider in 3.3.2. Perfective constructions with the auxiliary in the present tense (which we shall refer to as ‘present perfect’ on morphological grounds) can be used to present events occurring at some indeterminate time in the past as having direct implications in the present, as in (7) where the parenthesised propositions following the translations suggest plausible implications: (7) a Juanne est andatu a Núgoro. ‘John has gone to Nuoro.’ (i.e. He is no longer here.) b Appo pulitu sa mesa. ‘I have cleaned the table.’ (i.e. It is now clean.) However, they can also be used to describe events occurring at some particular point in the past (which may or may not be specified by an adverbial expression) and which have no obvious relevance to the present: (8) a Juanne est andatu a Núgoro su mandzanu e est ghiratu su sero. ‘John went to Nuoro in the morning and returned in the evening.’

< previous page

page_84

next page >

< previous page

page_85

next page >

Page 85 b Appo pulitu sa mesa; poi appo fattu su mándicu. ‘I cleaned the table; then I prepared the meal.’ The difference between (7) and (8) can be explicated in terms of Reichenbach’s (1947) concept of a reference-time (R) which is potentially distinct from both the time of the event (E) and the moment of speech (s), i.e. the present. In (7), the reference-time is identical to the moment of speech and the event is presented as occurring at some time prior to this reference point (i.e. the event is viewed from the perspective of the present), whereas in (8) the reference-time is a contextually determined point in the past (for example, the point reached in the narrative, optionally specified by an adverbial expression) and the event-time is presented as identical to this reference time. Using the symbol ‘>’ to denote temporal precedence and ‘=’ to denote simultaneity, (7) and (8) can be represented respectively as in (9) and (10): (9) E>R; R=S (10) E=R; R>S Within this framework, the normal use of the pluperfect illustrated in (2) can be represented as in (11); i.e. the event is presented as occurring prior to a reference-time in the past (indicated by dza ‘already’ in (2)): (11) E>R; R>S At this point we should mention that in literary narrative style the pluperfect is often used, in preference to the present perfect, to describe events which are not presented in relation to a present or past referencetime (cf. Blasco-Ferrer 1986:150). For example, in Spiggia’s (1982) translation of G. Deledda’s novel Elias Portolu, the narrative is presented exclusively in the pluperfect (corresponding to the simple past in the Italian original) and the imperfect, whereas the present perfect is confined to dialogue. This use is illustrated in the following passage where it is clear from context, and the expression un’ora appustis ‘an hour later’, that the events described are not to be located at some time prior to the reference-time established by the preceding narrative:1 (12)“Un’ora appustis 1’aian muttiu chi esseret andau derettu derettu a domo sua. Fit tuccau currende, tremendesi che zuncu.”

< previous page

page_85

next page >

< previous page

page_86

next page >

Page 86 ‘An hour later they called him for him to go straightaway to his house. He left in a hurry, shaking like a leaf (lit. ‘reed’).’ [EP: 190] This literary use of the pluperfect is reminiscent of the ‘past historic’ in French and can perhaps be attributed to a tendency to avoid morphologically present forms in texts which describe events in a selfcontained, fictional past. The principal difference between the perfective constructions discussed above and the simple imperfect forms is that with the former the event is viewed from the outside as an unanalysable unit (either looking back from a later point in time, as in (9) and (11), or from a reference-time which is coextensive with the event-time, as in (10)), whereas with the imperfect the event is viewed from the inside; i.e. the event-time is presented as extending (possibly indefinitely) beyond the reference-time, as represented in (13): (13) E…R…E; R m S The notational distinction between a time-interval (represented by the subscripted suspension points in (13)) and a time-point does not necessarily correspond to a difference in real time. Thus, the situations described in (14) have the same duration, but differ in terms of their temporal relation to other events within the discourse: (14) a Appo travallatu in Olbia pro tres annos. ‘I worked in Olbia for three years.’ b Travallaío in Olbia pro tres annos. ‘I was working in Olbia for three years.’ Sentence (14a), with a reading of the type in (10),2 would be appropriate as part of a summary of the speaker’s life-history in which each period is presented as a closed episode (e.g. (14a) might be followed by a statement of the type ‘and then I emigrated to Belgium’), whereas (14b) would be more natural as a preface to a description of particular events which occurred during this period (e.g. Una die, appo natu a su mere …‘One day I said to the boss…’). Regardless of the actual duration of the event described, the present perfect (in its past narrative use, represented in (10)) presents events as occurring consecutively, so that the reference-time advances with each use of the present perfect, whereas the imperfect presents situations as occurring in parallel with respect to a static referencetime. Thus, the events in (15) are presented in chronological order, whereas those in (16) are understood as occurring simultaneously and could be interspersed in any order in the discourse in (15):

< previous page

page_86

next page >

< previous page

page_87

next page >

Page 87 (15) Juanne s’est pesatu kithu. S’est vestitu. At abbertu su barcone. ‘John got up early. He got dressed. He opened the window.’ (16) Su sole nk’ issíat. Sos pudzones cantaían. Juanne si sentíat alligru. ‘The sun was rising. The birds were singing. John felt happy.’ There are various ways in which the relatively extended duration of the event described by the imperfect can manifest itself. When the verb denotes a state, process or activity, the use of the imperfect simply indicates that the state of affairs holds at and beyond the reference point. With verbs which normally denote instantaneous events, the imperfect can be used to provide a context for another event which is presented as occurring within this instant, as in (17): (17) Comente sa janna s’istrempaíat in palas meas, mi so ammentatu ki aío irmenticatu sa crae. ‘As the door slammed behind me, I remembered that I had forgotten my key.’ More typically, with such verbs the imperfect induces an iterative interpretation, as in (18) (compare the perfective version Sa janna s’est istrempata ‘The door slammed (once)’): (18) Sa janna s’istrempaíat. ‘The door kept slamming.’ In some cases, the event-time may be taken to include the periods immediately before and after the event which is literally denoted by the verb, as in (19), where the speaker’s arrival might be construed as occurring at any time from the announcement of the train’s departure to the moment when it is clear of the station: (19) Cando so arrivatu a s’istatzione, su trenu sink’andaiat. ‘When I arrived at the station, the train was leaving.’ Conversely, the use of perfective forms with some verbs which normally denote protracted states may restrict the event-time to the moment when the state comes into effect, as in (20) where iskire ‘know’ assumes an ingressive meaning ‘realise’ or ‘learn’: (20) Appo iskitu ki Maria fit maláida. ‘I realised/learned that Mary was ill.’ The imperfect is also used to describe habitual events, as in (21):

< previous page

page_87

next page >

< previous page

page_88

next page >

Page 88 (21) Cando fippo minore, sa dzente lavaíat sos pannos in su riu. ‘When I was young, people used to do their washing in the river.’ We assume that this use can be subsumed under the general representation given in (13), except that the reference time R, instead of corresponding to the particular point reached in the narrative, is perceived as a protracted (and possibly open-ended) time-span, specified by the time clause cando… in (21). This habitual interpretation is parallel to the iterative interpretation of (18) except that repetition of the event is presented as occurring regularly within and beyond the reference period rather than in rapid succession around the reference point. The simple present has essentially the same range of aspectual properties as the imperfect, but the reference-time is associated with the moment of speech S rather than with some time in the past. Thus, the simple present can be used as in (22) to describe an ongoing event or state (where the reference-time is identified with the immediate present) or as in (23) to describe a habitual event or a long-term state: (22) a Su pitzinnu keret una meledda. ‘The boy wants an apple.’ b Proet. ‘It is raining.’ (23) a Custu camminu andat a Orune. ‘This road leads to Orune.’ b In Iscótzia proet meta. ‘In Scotland it rains a lot.’ The progressive construction with éssere and the present participle is used primarily in cases where the reference-time is perceived as being punctual (either the present moment or the point reached in a past narrative): (24) a Su pitzineddu fit brincande supra su lettu. ‘The little boy was jumping (up and down) on the bed.’ b Su trenu sinke fit andande. ‘The train was leaving.’ c Est proende. ‘It is raining.’ Although the progressive construction is often preferred in such cases, the use of the simple present or imperfect is also possible, as

< previous page

page_88

next page >

< previous page

page_89

next page >

Page 89 examples (17)–(18) and (22) illustrate. Also, as noted in (5) above, the progressive can be used with stative verbs to emphasise the fact that the state is a transient one which obtains at the particular time of reference. Note, however, that éssere itself cannot be used in the progressive: *Maria est essende brava ‘Mary is being good’—in such cases a construction with fákere ‘make, do’ and a nominalised adjective is used: Maria est fakende sa brava . The progressive construction cannot usually have a habitual interpretation (e.g. Est proende in Iscótzia can only mean ‘It is raining in Scotland (at this moment)’), however a habitual interpretation is possible when the progressive is accompanied by the adverb semper ‘always’: Est semper proende in Iscotzia ‘It is always raining in Scotland’. We may also note cases like Occannu est proende pacu ‘It has not rained much this year’ (lit. This year it is raining little’) where the progressive is used to make a general statement about a relatively short-term situation. 3.1.3 Periphrastic future and conditional constructions A noteworthy feature of Sardinian verb morphology is the absence of synthetic future or conditional paradigms. The present simple tense can be used with future reference, particularly when accompanied by an appropriate time adverbial. Use of the simple present is the norm in sentences with first or second person subjects which express promises, requests, suggestions, etc. (essentially cases where the speaker or hearer has some control over the realisation of the event): (25) a Vendzo a ti vídere sa kitta ki vénit. ‘I will come to see you next week.’ b A mi cantas cussa cathone. ‘Will you sing me that song.’ c Andamus a su monte cras. ‘Let us go to the mountain tomorrow.’ However, in statements of a predictive type, the normal way of indicating future time is to use the periphrastic construction áere ‘have’+ a+infinitive: (26) a An a túndere sas berbekes cras. ‘They will shear the sheep tomorrow.’ b Su trenu at a arrivare a sas deke. ‘The train will arrive at ten o’clock.’ c At a próere cras. ‘It will rain tomorrow.’

< previous page

page_89

next page >

< previous page

page_90

next page >

Page 90 d Amus a rúghere si no’istamus attentos. ‘We will fall if we are not careful.’ Note that the áere+a formula does not carry any modal connotations such as intent or obligation. This can be seen by comparing the examples in (26) with similar sentences where futurity is expressed by the modal verb dévere ‘must’: (27) a Deven túndere sas berbekes cras. ‘They have to shear the sheep tomorrow.’ b Su trenu devet arrivare a sas deke. ‘The train is supposed to arrive at ten o’clock.’ c Devet próere cras. ‘It is supposed to rain tomorrow.’ d ?Devimus rúghere (si no’istamus attentos).’ ‘We must fall (if we are not careful).’ When the subject is agentive, as in (27a), dévere typically implies obligation by some higher authority or necessity determined by some other circumstance (e.g. in (27a), someone is coming to collect the wool the next day). No such implications are present in (26a), which simply expresses an expectation regarding a future event. With nonagentive subjects, as in (27b,c), dévere often presents an event as one which should happen in principle (e.g. according to the train timetable or the weather forecast) rather than as one which will actually happen. For instance, (27b) would not be appropriate if the train is scheduled to arrive at nine o’clock but one knows that trains are always late, whereas (26b) would be perfectly natural in such a situation. Note finally that in some cases, such as (27d) (with or without the condition clause), the use of dévere is decidedly odd, presumably because the modal properties of dévere are pragmatically incompatible with the event described. However, áere+a in (26d) is not subject to any restrictions of this sort. The above observations concerning dévere apply only to the regular present tense forms (devo, deves, devet, devímus, devítes, deven). However, dévere also has a ‘short form’ present paradigm (except in the first person singular) whose semantic properties are much closer to áere+a: des (2 sg.), det (3 sg.), demus (1 pl.), dedzis (2 pl.), den (3 pl.). An important restriction on these short forms is that they can only be used when the following infinitive is áere or éssere (as auxiliaries or main verbs). The similarity between these short forms and áere+a, and the contrast with the regular forms of dévere, can be seen in cases where these items express future perfect

< previous page

page_90

next page >

< previous page

page_91

next page >

Page 91 aspect, as in (28) where an a and den are interchangeable with no appreciable difference in meaning: (28) Sos óspites éssere thuccatos prima de arrivare nois. ‘The guests will have left before we arrive.’ However, substitution of the regular form deven gives a reading something like The guests are supposed to have left before we arrive’. Note also that both áere+a and the short forms of dévere can be used to express conjectures about the present or recent past as well as the future: (29) a Juanne éssere arrivatu como. ‘John will have arrived by now.’ b Sa mákkina non funtzionat; sa battería éssere guasta. ‘The car does not work; the battery must be faulty.’ Turning now to the conditional, we find a striking symmetry between conditional and future when the lexical verb assumes a simple tense form. Just as the simple present can be used with future reference, so the imperfect can be used with either a ‘future-in-thepast’ or genuine conditional value: (30) a Tappo natu ki rughías. lit. ‘I told you that you fell.’ ‘I told you that you would fall.’ b Si fippo riccu, compraío una domo manna. lit. ‘If I was (imperf.) rich, I bought (imperf.) a big house.’ ‘If I were rich, I would buy a big house.’ However, this symmetry does not extend fully to the periphrastic constructions discussed above. In particular, in the Logudorese-Nuorese dialects, the áere+a construction cannot be cast in the imperfect (i.e. the propositions in (30) cannot be expressed as in (31)), though such constructions are possible in Campidanese (cf. Blasco-Ferrer 1986:124): (31) a *T’ appo natu ki aías a rúghere. b *Si aío a éssere riccu, aío a comporare una domo manna.

< previous page

page_91

next page >

< previous page

page_92

next page >

Page 92 A closer correspondence between conditional and future is found in constructions with dévere, which has two conjugations in the imperfect just as in the present. The regular forms (devío, devías, devíat, devíamus, devíadzis, devían) always have modal properties of the sort noted in (27), whereas the short forms (dio, dias, diat, díamus, díadzis, dian) function as straightforward conditional or future-in-the-past auxiliaries. However, the short imperfect forms differ from the short present forms in that they can co-occur with all verbs (not just áere and éssere) filling, as it were, the gap left by the unavailability of imperfect forms of áere+a. The semantic differences between the the short and regular forms are illustrated in the following examples, which reflect the contrasts noted between áere+ a and dévere in (26)–(27): (32) a T’appo natu ki dian túndere sas berbekes oje. ‘I told you they would shear the sheep today.’ b Su mere at natu ki devían túndere sas berbekes oje. ‘The master said they had to shear the sheep today.’ (33) a T’appo natu ki su trenu diat arrivare a sas deke. ‘I told you that the train would arrive at ten o’clock.’ b M’an natu ki su trenu devíat arrivare a sas nove. ‘They told me that the train was supposed to arrive at nine.’ (34) a Non semus issitos ca credíamus ki diat próere. ‘We did not go out because we thought it would rain.’ b An natu ki devíat próere; imbetzes at fattu tempus bonu. ‘They said it was going to rain, but it has been fine.’ (35) a Iskío ki dias rúghere si no’istaías attentu. ‘I knew you would fall if you were not careful.’ b ?Iskío ki devías rúghere (si no’istaías attentu). ‘I knew you had to fall (if you were not careful).’ In hypothetical conditional constructions, the short imperfect forms of dévere can be used as an alternative to the imperfect of the lexical verb, as in (30b), in both the condition and consequent clauses, or just in the consequent clause (see 6.2.3 for further discussion): (36) Si ricco, dio comporare una domo manna. ‘If I were rich, I would buy a big house.’ As expected, use of the regular imperfect of dévere in either clause implies some element of obligation or necessity:

< previous page

page_92

next page >

< previous page

page_93

next page >

Page 93 (37) a Si devío andare a tháthari, piccaío su trenu. ‘If I had to go to Sassari, I would take the train.’ b Si non fis riccu, devías travallare comente sos átteros. ‘If you were not rich, you would have to work like other people.’ In addition to the expressions discussed above, there are various other periphrastic formulae which can be used to indicate future time, though again with certain modal implications. Alongside the áere+a construction, áere can be used with de and an infinitive. This construction is only possible with agentive subjects and always implies that the subject is under some obligation (possibly selfimposed) to perform the action: (38) a Appo de pulire sa domo. ‘I have to clean the house.’ b *Su trenu at de arrivare a sas deke. ‘The train has to arrive at ten o’clock.’ Unlike áere+a, the áere+de formula can be used in the imperfect: (39) Non potío vénnere ca aío de pulire sa domo. ‘I could not come because I had to clean the house.’ Essere can be used in the present or imperfect with an infinitive introduced by de (in a manner similar to one of the uses of English be going to) to indicate an action which is planned by the subject entity, but not necessarily realised (indeed, in the imperfect the intended action is usually construed as having been thwarted): (40) a Si ses de cantare cussa cathone, cántala como. ‘If you are going to sing that song, sing it now.’ b Fimus de andare a mare, ma poi at cumintzatu a próere. ‘We were going to go to the sea, but then it started to rain.’ A further construction with éssere, which may be construed as a future periphrasis of sorts, though in practice it serves to negate a proposition regarding the past, is éssere+galu ‘yet’+ a+infinitive: (41) Semus/fimus galu a mandicare. lit. ‘We are/were yet to eat.’ ‘We have/had not eaten yet.’

< previous page

page_93

next page >

< previous page

page_94

next page >

Page 94 3.2 TRANSITIVITY OF THE VERB 3.2.1 Thematic roles and grammatical functions In this section we shall consider the relationship between the semantic roles of arguments of the verb (i.e. ‘thematic roles’ in the sense of Gruber (1965) and Jackendoff (1972), or θ-roles (Chomsky, 1981)) and the grammatical functions which encode them (e.g. subject, direct object, indirect object). For the purposes of this discussion we adopt a basic three-way classification of θ-roles (with finer distinctions where appropriate): Agent (AG), Theme (TH) and Location (LOC). The Agent is the entity (usually animate) who causes the change of state or performs the action denoted by the verb. The Location role encompasses loci of a figurative or abstract nature as well as physical location; for example, it includes notions such as Possessor, Experiencer, Perceiver, Cogniser, etc., which we indicate collectively by the mnemonic feature [+animate], as opposed to physical location which is indicated by [−animate]. A further distinction within the broad Location class, orthogonal to the [±animate] distinction, arises in cases where the verb denotes a change of state. Where appropriate, we use the terms ‘Source’ and ‘Goal’ to denote respectively Location prior to and after the change of state. Thus, within the figurative [+animate] domain, we can characterise as Goal the subjects of verbs such as ‘receive’, ‘notice’, ‘realise’, etc., as Source the subjects of verbs like ‘lose’, ‘forget’, etc., whereas the subjects of verbs like ‘possess’, ‘see’, ‘like’, ‘want’, ‘believe’, etc. are intantiations of static Location. The Theme role is likewise fairly heterogeneous—we use this term to denote the entity of which some property is asserted or which undergoes a change of state or position. Also, by extension, it is the role which complements the various types of figurative location outlined above; for example, it is the role associated with a possessed entity, a phenomenon which is perceived or experienced, a proposition which is believed or communicated, etc. On the whole, the correspondence between θ-roles and grammatical functions conforms to a pattern familiar in other Romance languages and, indeed, in English and other Indo-European languges. In particular, verbs which involve an Agent systematically assign this role to the subject. Locative arguments of the [−animate] type tend to be realised as indirect objects, where the preposition indicates a particular spatial relation and/or a distinction between Goal, Source and static location:

< previous page

page_94

next page >

< previous page

page_95

next page >

Page 95 (42) a Juanne at postu su libru supra sa mesa. ‘John put the book on the table.’ b Su pastore est ghiratu dae su monte. ‘The shepherd has returned from the mountain.’ Similarly, with verbs which denote communication, transfer of property, etc., the animate Location (Goal or Source) is typically expressed by a dative indirect object (introduced by a and replaceable by a dative clitic such as li or lis ): (43) a Maria at datu unu regalu a su pitzinnu. ‘Mary gave a present to the boy.’ b Lukia at natu carki cosa a su mere. ‘Lucy said something to the boss.’ c Su professore at mustratu su cuadru a s’istudiente. ‘The teacher showed the picture to the student.’ d Gavini at furatu unu libru a su duttore. ‘Gavin stole a book from the doctor.’ There are a few verbs which select a direct object with a Location role, with the Theme expressed, if at all, as an indirect object introduced by de ‘of or kin ‘with’: e.g. (with a [−animate] Location) prenare su lavandinu (de abba) ‘fill the sink (with water)’, carricare su carru (de tricu) ‘load the cart (with corn)’, cucudzare su lettu (kin unu letholu) ‘cover the bed with a sheet’, cropicare sa patedda (kin unu prattu) ‘cover the pan (with a plate)’, and (with a [+animate] Location) suffruire carcunu (de latte) ‘supply someone (with milk)’, avvertere/ammunire carcunu (de su perículu) ‘warn someone (of the danger)’, munire carcunu ‘give someone a warning’, addescare carcunu ‘feed someone’, briare/juliare/salutare carcunu ‘scold/call/greet someone’. The choice of Location as direct object is associated with semantic effects familiar in other languages. When the direct object denotes a physical location (prenare ‘fill, carricare ‘load’, cucudzare/ cropicare ‘cover’) the Theme entity is understood as occupying the whole of the location, and the spatial relation is implicit in the meaning of the verb. Verbs of ‘giving’ (suffruire ‘supply’, addescare ‘feed’) or communication (avvértere ‘warn’) which select the Receiver as direct object typically imply that the Theme entity is something which the Receiver ought to have or know about. Also, in many cases the nature of the Theme is implied by the nature of the verb, clear cases being addescare, where the understood Theme is ‘food’, and illocutionary verbs such as briare ‘scold’, juliare ‘call’, munire ‘warn’, salutare ‘greet’ where the verb indicates the nature of the utterance which is communicated. The verbs cucudzare and cropicare can also

< previous page

page_95

next page >

< previous page

page_96

next page >

Page 96 be included in this class in so far as they make certain presuppositions about the nature of the Theme: cucudzare=‘cover with a substance or material’ (usually to conceal or protect), cropicare=‘cover with a lid (or object with a similar function)’. With verbs of this type, the Theme is not usually specified independently but, when it is, it is presented under the guise of an instrumental expression (i.e. as the complement of kin ‘with’), whereas verbs which do not incorporate information about the nature of the Theme (e.g. prenare ‘fill, carricare ‘load’, suffruire ‘supply’, avvértere ‘warn, inform’) allow the Theme to be specified freely as the complement of de. The distinction between direct objects and dative complements is partially obscured by the prepositional accusative phenomenon discussed in 2.2.6 and also by neutralisation of the accusative vs. dative distinction with first and second person clitics (mi, ti, nos, bos) and the reflexive clitic (si). Perhaps for this reason, datives occur predominantly with verbs which also take a direct object, as in (43) above. Two-place verbs involving a [+animate] Location fall into two principal classes; those which select the Theme as direct object and the Location as subject, as in (44), and those which express the Location as a dative complement with the Theme occurring as the superficial subject, as in (45): (44) a Juanne at vistu un’ arrantzolu. ‘John saw a spider.’ b Maria keret cussu libru. ‘Mary wants that book.’ c Lukia juket una mákkina. ‘Lucy has a car.’ (45) a Su casu piaghet a sos sórrikes. lit. The cheese pleases to the mice.’ ‘Mice like cheese.’ b Custu libru lis interessat. ‘This book interests them.’ c Custa mákkina non lis cumbenit. ‘This car does not suit them.’ Verbs of the type in (45) have a number of properties which, in the recent theoretical literature, have been associated with a class of ‘unaccusative’ verbs (see particularly Perlmutter 1978 and Burzio 1986) whose superficial subjects are claimed to have direct object status at a more abstract, underlying level of representation—in particular, these verbs are among those which take éssere ‘be’ as their perfective auxiliary. This matter will be discussed in greater detail in

< previous page

page_96

next page >

< previous page

page_97

next page >

Page 97 3.2.3, but for the moment we may note that if this analysis is adopted, cases like (45) conform to the generalisation that datives typically cooccur with (underlying) direct objects. It is difficult to find clear cases of genuinely intransitive verbs (other than unaccusatives) which take a dative complement. Verbs which express a relation between an Agent and a [+animate] Location tend to show some ambivalence or uncertainty with regard to the grammatical function of the latter argument. For example, faeddare ‘talk’ allows the object to be realised as either an accusative or a dative clitic according to dialect (the use of the accusative, as in (46a), being the more common): (46) a Los appo faeddatos. b Lis appo faeddatu. ‘I talked to them.’ In cases like (47), which are acceptable for all speakers, a might be analysed either as a dative marker or as an instance of the prepositional accusative discussed in 2.2.6 since proper names must be introduced by a even when they function as direct objects: (47) Appo faeddatu a Juanne. ‘I talked to John.’ However, when the object is an NP of the type which does not normally take the prepositional accusative (e.g. an indefinite NP), a is still required even for those speakers who use an accusative clitic as in (46a) (or, more typically, an alternative construction with kin ‘with’ is used): (48) a *Appo faeddatu unu duttore. b Appo faeddatu a/kin unu duttore. ‘I talked to/with a doctor.’ A rather similar situation is found with the verb pessare ‘think’ which requires a before all types of object NP and takes a locative clitic bi when referring to inanimate entities, but takes an accusative clitic with human reference: (49) a So pessande a cussu problema. ‘I am thinking about that problem.’ b So pessande a una pitzinna. ‘I am thinking about a girl.’ (50) a Non bi pesso prus. ‘I do not think about it any more.’

< previous page

page_97

next page >

< previous page

page_98

next page >

Page 98 b Non la pesso prus. ‘I do not think about her any more.’ Thus, both of these verbs tend to be treated as transitive when the object is a clitic with human reference, but require a form of the full object NP which might be analysed either as dative or as an extension of the prepositional accusative. The dative form is used more consistently (i.e. with both clitics and full NPs) to indicate the optional Goal entity with verbs like sorrídere ‘smile’: (51) a Appo sorrísitu a unu pitzinneddu. ‘I smiled at a little boy.’ b Lis appo sorrísitu. ‘I smiled at them.’ However, in such cases, the Goal expression is arguably an adjunct rather than a subcategorised complement. Moreover, as we shall see in 3.3.1, verbs like sorrídere show certain peculiarities in reflexive constructions which suggest that the Goal expression does not behave consistently like a true dative. On the basis of the above evidence, the use of the dative can be characterised in terms of two criteria. First, from a semantic point of view, the dative corresponds to the range of θ-roles which we have classified as [+animate] Location. Second, from a syntactic viewpoint, the dative is used almost exclusively to mark the second object of the verb in underlying structure (i.e. with transitive and unaccusative verbs). With two-place verbs which describe movement or static location, the Location argument is typically expressed by a PP and the Theme as superficial subject, though arguably an underlying direct object in so far as verbs of this type belong to the unaccusative class (for example, they take éssere as their perfective auxiliary): (52) a Juanne est andatu a Olbia. ‘John went to Olbia.’ b Maria est ghirata dae su dzardinu. ‘Mary returned from the garden.’ c Gavini est abbarratu in domo. ‘Gavin stayed at home.’ There are a few verbs which express the Location as a direct object, as in (53): (53) a Su sordatu at lómpitu sa frontiera. ‘The soldier reached the frontier.’

< previous page

page_98

next page >

< previous page

page_99

next page >

Page 99 b Juanne at lassatu sa bidda sua. ‘John left his home town.’ c Sa crapa at brincatu su riu. ‘The goat jumped across the river.’ There are numerous verbs which can be used both transitively (with a causative sense) or intransitively with the Theme as subject (to describe a change of state). In many cases, the intransitive use of the verb is signalled by a reflexive clitic, as in (55) (see 3.2.6 for further discussion): (54) a Maria at abbertu/tuncatu sa janna. ‘Mary opened/closed the door.’ b Pretu at secatu sa camba. ‘Peter broke the branch.’ (55) a Sa janna s’est abberta/tuncata. ‘The door opened/closed.’ b Sa camba s’est secata. ‘The branch broke.’ However, some verbs are used intransitively without a reflexive clitic: (56) a Juanne at budditu s’abba. ‘John has boiled the water.’ b Su pastore at brujatu sa linna. ‘The shepherd has burnt the wood.’ (57) a S’abba at budditu. ‘The water has boiled.’ b Sa linna at brujatu. ‘The wood has burnt.’ Note that in cases like (57), where the verb is non-pronominal in its intransitive use, the perfective auxiliary is áere, a fact which raises interesting questions regarding the general issue of unaccusativity in so far as the relationship between (56) and (57) lends itself to an analysis whereby the superficial subject in (57) is an underlying direct object just as in the causative construction in (56), but such an analysis would conflict with our assumption that unaccusative verbs select éssere as their auxiliary (see 3.3.1 for further discussion). Some verbs of movement can also be used both transitively and intransitively, as in (58)–(59): (58) a Juanne at issitu su mutzighinu dae sa butzacca. ‘John took his handkerchief out of his pocket.’

< previous page

page_99

next page >

< previous page

page_100

next page >

Page 100

b Maria at ampilatu sos jocátolos a su suffitu. ‘Mary took the toys up to the attic.’ (59) a So issitu dae sa cukina. ‘I went out of the kitchen.’ b Lukia est ampilatu a su suffitu. ‘Lucy went up to the attic.’ Note that the problem raised with respect to the verbs in (57) does not arise in these cases since these verbs select éssere in their intransitive use (just like those in (52)) and can thus be classified as unaccusative. The unaccusative verb mórrere ‘die’ can also be used transitively with the meaning ‘kill’, at least in its past participle form, as in (60b): (60) a Su cane est mortu eris. ‘The dog died yesterday.’ b Su pastore at mortu su mariane. ‘The shepherd killed the fox.’ The transitive use of this verb can be analysed as a case of suppletion in so far as the normal verb for ‘kill’, ukkíere, appears to lack a past participle form. 3.2.2 Impersonal constructions The term ‘impersonal’ is used here to denote constructions in which no θ-role is assigned to the subject position and which contain no phrase which can be characterised as a subject in any syntactic sense. In more concrete terms, verbs used impersonally always lack an overt subject and, when finite, occur only in the third person singular. Our discussion in this section will focus on the syntax of such constructions with particular reference to the choice of perfective auxiliary and the distinction (envisaged in 3.2.1) between ‘normal’ intransitive verbs and unaccusative verbs, the latter being essentially verbs which occur underlyingly in impersonal constructions (in the sense that they do not assign a θ-role to the subject position) but which allow the underlying object to assume subject status at the surface level. One class of verbs which may be regarded as intrinsically impersonal is composed of ‘weather’ verbs such as próere ‘rain’, nivare ‘snow’, tronitare ‘thunder’, which can be plausibly analysed as verbs which do not assign any θ-role, the nature of the Theme being incorporated as part of the meaning of the verb. With such verbs, the perfective auxiliary is always áere:

< previous page

page_100

next page >

< previous page

page_101

next page >

Page 101 (61)

At próitu meta eris. ‘It rained a lot yesterday.’ There are a few other verbs, notably kérrere (with the sense ‘be necessary’) and mancare ‘be missing’, which occur predominantly in impersonal constructions, as in (62), where the verb has third person singular inflection even though its only argument is plural: (62) a Bi keret tres ovos. ‘Three eggs are necessary.’ b Bi mancat duos buttones. ‘Two buttons are missing.’ In view of the lack of agreement between the verb and the postverbal NP, we assume that this NP is not, in any syntactic sense, a subject. The item bi3 (normally a locative clitic, though it has a variety of pleonastic uses; see 5.2.4) is characteristic of impersonal constructions in which the verb is followed by a complement NP, a point to which we shall return shortly. However, with kérrere and mancare, bi can be replaced by a dative clitic denoting an ‘interested party’: (63) a Nos keret unu milione de francos. ‘We need a million lire.’ b Mi mancat duas dentes. ‘I have two teeth missing.’ When these verbs are used as in (62) or (63), they take áere as their perfective auxiliary: (64) a B’at kérfitu tres ovos pro fákere cussas gatheddas. ‘Three eggs were needed to make those fritters.’ b M’at kérfitu duas oras pro arrivare a inoke. ‘It took me three hours to get here.’ c B’at mancatu milli francos. ‘A thousand lire were missing.’ However, when the NP following these verbs is definite, a different pattern emerges—the verb must agree with the complement NP and selects éssere as the perfective auxiliary (with agreement between the past participle and the postverbal NP): (65) a Bi mancan sos buttones. ‘The buttons are missing.’ b Bi sun kérfitos cussos ómines. ‘Those men were needed.’ Moreover, when this definite NP is an item which inflects for Case, it assumes the nominative form:

< previous page

page_101

next page >

< previous page

page_102

next page >

Page 102 (66)

Non bi keres tue inoke. ‘You are not needed here.’ We postulate that the definite Theme NPs in (65)–(66) have the status of subjects (in contrast to the indefinite Themes in (62)–(64)), even though they typically follow the verb. The general pattern observed above is by no means peculiar to kérrere and mancare. Let us first consider the the range of constructions available with other (more typical) verbs which take éssere as their auxiliary. With such verbs, the subject can readily follow the verb, as in (67), and, pursuing the unaccusative analysis envisaged in 3.2.1, we postulate that this represents the underlying order of elements with these verbs: (67) a (Bi) sun vénnitos issos. ‘They have come.’ b (Bi) sun ruttas sas tassas. ‘The glasses fell down.’ c Si (bi) mores tue, no’isco itte fakimus. ‘If you die, I do not know what we shall do.’ More specifically, we assume that unaccusative verbs assign a Theme role to the direct object position in the same way as typical transitive verbs, but are unable to license the Theme NP by assigning an accusative Case feature to it. Consequently, this NP must assume subject status to acquire nominative Case (as seen when the Theme is an NP of the type which shows overt Case inflection, as in (67c) where tue is a nominative form), which is manifested by person and number agreement of the finite verb. As the examples in (67) show, this ‘subjectivisation’ process does not necessarily entail movement to the preverbal subject position, though this option is available: Issos (bi) sun vénnitos, Sas tassas (bi) sun ruttas, Si tue (bi) mores, …. However, if the Theme NP is indefinite, movement of the NP to the canonical, preverbal subject position is an almost obligatory correlate of the subjectivisation process. Whereas the sentences in (68) are fully acceptable, their counterparts in (69) with a postverbal subject are judged to be rather peculiar, though such constructions do occur occasionally in spontaneous discourse: (68) a Tres pitzinnas (bi) sun vénnitas. ‘Three girls came.’ b Carki tassas (bi) sun ruttas. ‘Some glasses fell down.’

< previous page

page_102

next page >

< previous page

page_103

next page >

Page 103

c Metas sordatos (bi) sun mortos. ‘Many soldiers died.’ (69) a ?(Bi) sun vénnitas tres pitzinnas. b ?(Bi) sun ruttas carki tassas. c ?(Bi) sun mortos metas sordatos. More typically, if the indefinite Theme NP remains in postverbal position, an impersonal construction (with no agreement between the finite verb and the postverbal NP) is preferred: (70) a B’at vénnitu tres pitzinnas. b B’at ruttu carki tassas. c B’at mortu metas sordatos.4 Apart from the lack of agreement of the finite verb (and the past participle) with the postverbal NP in (70), there are two other features of these constructions which are worthy of note. First, the perfective auxiliary must be áere even though the verbs in question require éssere in ‘personal’ constructions (where the Theme NP has subject status), as in (67)–(68) and, indeed, the marginal examples in (69). Second, the presence of the clitic bi (nke in some dialects) is obligatory in (70) whereas in (67)–(69) bi is a purely incidental element which may denote a specific location, though often it has a discursive function which is hard to identify precisely (see 5.2.4 for further discussion), but, in any case, its presence is optional. This difference in the status of bi can be seen in the following examples: (71) a *In ube bi sun mortos cussos sordatos? ‘Where did those soldiers die?’ b In ube b’ at mortu metas sordatos? ‘Where did many soldiers die?’ The ungrammaticality of (71a) can plausibly be attributed to a conflict between the locative clitic bi ‘there’ and the interrogative locative expression in ube ‘where’. By the same token, the acceptability of (71b) suggests that bi does not have a locative value (duplicating the interrogative expression) in the impersonal construction. We hypothesise that the three properties of impersonal constructions mentioned above are systematically related. First, the use of áere in (70) shows that choice of auxiliary is not determined simply as a lexical property of the verb. Rather, selection of éssere appears to correlate with ‘subjectivisation’ of the Theme of an unaccusative verb (as manifested by verb-agreement and nominative Case). Second, the obligatory presence of bi in impersonal constructions suggests that

< previous page

page_103

next page >

< previous page

page_104

next page >

Page 104 this item provides an alternative means of licensing the postverbal Theme when it is indefinite and is not ‘subjectivised’. Thus, in (70), where the postverbal NP is not ‘subjectivised’, bi is obligatory and we have the unmarked auxiliary áere whereas in (67)–(69) the auxiliary is éssere and bi is optional (with a locative or pseudo-locative function). The precise manner in which bi licenses the postverbal Theme in impersonal constructions is a largely theoretical matter which is beyond the scope of this study. From a semantic perspective, we may postulate that bi functions as an existential operator which must have scope over an indefinite NP (thus accounting for the impossibility of impersonal constructions with definite NPs: *B’at vénnitu cussas pitzinnas. Those girls came’). The fact that this existential function is fulfilled by a locative clitic is, perhaps, rather natural in so far as existence can be equated with location within some universe of discourse (cf. existential constructions with there in English and, more particularly, il y a lit. ‘it there has’ in French). With regard to Case as a licensing feature, we might postulate either that bi assigns a Case feature to the NP within its scope (though we have no empirical evidence which would enable us to decide which Case is assigned since there are no indefinite expressions which display an overt Case distinction) or that indefinite NPs whose existence is predicated are exempt from the general requirement that all NPs must be licensed by a Case feature. Returning now to the constructions with kérrere and mancare in (62)–(66) the general pattern of correlation between auxiliary choice and ‘subjectivisation’ is the same as that which we have observed with more typical unaccusative verbs. The particularity of these two verbs is first that they strongly favour the ‘impersonal’ strategy for licensing the Theme and do not generally allow the Theme to move to the preverbal subject position and, second, that the impersonal construction is possible with a dative clitic in place of bi (cf. (63)), a possibility which is not generally available with other unaccusative verbs; e.g. *M‘at ruttu tres tassas vs. Cuddas tassas mi sun ruttas ‘Three/those glasses fell down on me’ where the dative (glossed by ‘on me’) is construed as referring to an entity who is affected in some way by the event (as opposed to a reading of the type ‘fell on top of me’). A possible explanation for the latter property is that the ‘existential operator’ function which we have ascribed to bi is inherent in the meaning of impersonal kérrere and mancare (which we might characterise as modalised expressions of existence), with the.result that bi is not necessary to license the Theme NP. This view is supported by the fact that there are other verbs of existence (notably capitare and

< previous page

page_104

next page >

< previous page

page_105

next page >

Page 105 sutzédere ‘happen’) which also allow a dative clitic in place of bi in impersonal constructions: (72) a B’at capitatu/sutzessu metas cosas istranas. ‘Many strange things have happened.’ b M’at capitatu/sutzessu metas cosas istranas. ‘Many strange things have happened to me.’ A difficulty with this approach is that bi is required, even with existential verbs, in the absence of a dative clitic. In other words, it seems that all verbs in their impersonal use must be accompanied by a clitic of some kind, which is normally bi but can be a dative if the verb has the existential properties which enable it to license the Theme NP directly. Let us now consider the range of constructions which are possible with ‘normal’ (i.e. non-unaccusative) intransitive verbs, which systematically take áere as their perfective auxiliary, taking ballare ‘dance’ as a representative example. We assume that, with such verbs, the subject is base-generated in the canonical preverbal position, as in (73): (73) a Cussas pitzinnas an ballatu. ‘Those girls danced.’ b Tres pitzinnas an ballatu. ‘Three girls danced.’ The subject can be moved into the postverbal position by a general process of inversion (see 7.1.3 for detailed discussion), as in (74): (74) a An ballatu cussas pitzinnas. b An ballatu tres pitzinnas. However, when the subject is indefinite, the impersonal construction with bi and lack of agreement is possible and generally preferred: (75) B’at ballatu tres pitzinnas. ‘Three girls danced.’ This construction is not possible when the NP is definite: *B’at ballatu cussos pitzinnas. There is a subtle pragmatic distinction between (74b) and (75) in that (74b) presupposes a situation of dancing and asserts either that the dancers were girls or that there were three of them whereas (75) would be more appropriate as a means of introducing the whole situation as new information. We see from the examples in (73)–(75) that the range of constructions with ‘normal’ intransitives is essentially the same as for

< previous page

page_105

next page >

< previous page

page_106

next page >

Page 106 unaccusative verbs, as in (67)–(70), except that there is no difference in choice of auxiliary. This follows from the assumption that subjects of ‘normal’ intransitives are base-generated in the canonical, preverbal position and thus do not undergo the ‘subjectivisation’ process which we have associated with selection of éssere. There is, however, a technical problem with the impersonal construction (75). If the subject tres pitzinnas is licensed in its underlying position (by nominative Case), we would expect the general inversion process to yield only (74b), rather than the construction with bi, which we have characterised as a means of licensing an NP which lacks Case. A possible solution to this problem is to assume that the person and number features which ultimately appear on the verb are selected independently of the features of the subject NP, but assignment of nominative Case is possible only when the two sets of features agree. Thus, we postulate that (76), with a plural subject and singular verbform, is a well-formed underlying structure but is ruled out at the surface level because the subject lacks Case: (76) *Tres pitzinnas at ballatu. lit. ‘Three girls has left.’ However, if the subject is indefinite, it can move into the postverbal position (by the general process of inversion) where it can be licensed by the clitic bi, as in the cases discussed above, giving the surface sentence (75). 3.2.3 Unaccusative verbs In our discussion so far we have simply assumed that there exists a class of unaccusative verbs (i.e. verbs whose superficial subject is an underlying object) which we have identified solely on the basis of auxiliary choice. In this section we shall discuss this and other criteria in greater detail in order to determine the membership of this class of verbs in Sardinian. As a basis for this discussion, we propose the following criteria which are representative of the general claims which have been made regarding unaccusative verbs in the relevant theoretical literature (see particularly Perlmutter 1978 and Burzio 1986): (77)a In languages which show alternation of perfective auxiliary, unaccusative verbs typically select be rather than have . b Unaccusative verbs denote movement, state or change of state and assign a Theme role to the superficial subject.

< previous page

page_106

next page >

< previous page

page_107

next page >

Page 107 c Superficial subjects of unaccusative verbs participate in syntactic phenomena which are characteristic of direct objects. With regard to property (77a), this is a particularly important criterion for Sardinian since the facts concerning impersonal constructions discussed in 3.2.2 show that choice of auxiliary is at least partially dependent on structural factors and cannot be stipulated as a purely lexical property. Criterion (77b) is ambiguous between a weak interpretation (all unaccusative verbs have the semantic properties in question) and a strong interpretation (all verbs with these semantic properties are unaccusative). The weak version of (77b) appears to hold for Sardinian. Although human subjects of movement verbs such as andare ‘go’, arrivare ‘arrive’ and vénnere ‘come’ might be characterised as Agents in so far as their referents are usually (though not always) the immediate instigators of the event, they also conform systematically to our definition of Theme since they always denote an entity which undergoes a change of location. Whether the strong version of (77b) also holds is a matter which we raised briefly in 3.2.1 in connection with verbs like buddire ‘boil’ and brujare ‘burn’ which can be used transitively (with the Agent as subject and the Theme as direct object) or intransitively with the Theme as superficial subject. As we noted in our earlier discussion, the intransitive uses of such verbs are prima facie candidates for an unaccusative analysis, except for the fact that they take áere as their auxiliary: (78) a S’abba at budditu. ‘The water has boiled.’ b Sa linna at brujatu. ‘The wood has burnt.’ There are many other verbs which have the semantic properties outlined in (77b) but which take áere as their auxiliary; a representative sample is given in (79): (79) accabbare ‘finish’, assutare ‘dry’, buddire ‘boil’, brujare ‘burn’, cambiare ‘change’, cumintzare ‘begin’, créskere ‘grow’, sanare ‘heal’, etc. If we adopt the strong version of (77b), which classifies these verbs as unaccusative, we must weaken criterion (77a) to the effect that all verbs which take éssere are unaccusative but not all unaccusative verbs take éssere. On the other hand, if we take selection of éssere as

< previous page

page_107

next page >

< previous page

page_108

next page >

Page 108 a defining property of unaccusative verbs, we are forced to adopt the weak version of criterion (77b). In principle this dilemma might be resolved by appealing to criterion (77c), but in order to apply this criterion we must identify syntactic phenomena which clearly distinguish between unaccusative and intransitive verbs. A well-known phenomenon of this type in Italian concerns the use of the partitive clitic ne. Belletti and Rizzi (1981) note that this clitic can be used to represent a quantified postverbal subject if the verb is unaccusative, but not otherwise: e.g. Ne arrivano molti There arrive many (of them)’ but *Ne ballano molti There dance many (of them)’. This test is difficult to apply directly to Sardinian given that indefinite subjects in postverbal position tend to be slightly infelicitous (see examples in (69)). For the corresponding impersonal constructions, there appears to be some correlation between unaccusativity and the use of partitive nde, but our informants’ judgements are rather less clear than those which are reported for Italian: (80) a Bind’arrivat metas. ‘There arrive many (of them).’ b ?Binde ballat metas. ‘There dance many (of them).’ Although such constructions are generally judged to be more acceptable with unaccusative verbs, the contrast is not sufficiently sharp to be used as a reliable diagnostic. Indeed, if we apply this test to the verbs in (79) we find similar differences in acceptability: (81) a Bind’at créskitu tres. ‘There grew three (of them).’ b ?Bind’at assutatu tres. ‘There dried three (of them).’ Not only is (81b) significantly worse than (81a) on the intended reading, it seems to be even less acceptable than (80b)—note that (81b) is acceptable with the transitive interpretation ‘He/she dried three (of them) there’. Other syntactic phenomena which constitute potential tests for unaccusativity are beset by similar uncertainties of judgement. For example, in 2.2.8 we noted that extraposition of a head noun from a preverbal subject occurs more readily with unaccusative verbs than with ‘normal’ intransitives: (82) a Su tuo no’est ghiratu de cane. ‘ Your dog has not returned.’

< previous page

page_108

next page >

< previous page

page_109

next page >

Page 109 b *Su meu at appeddatu de cane. ‘ My dog has barked.’ Although many of our informants perceived a clear constrast between examples like those in (82), there are others who were unable to make firm or consistent judgements on such examples. With verbs of the type in (79), judgements are even murkier. On the whole, (83a) seems to be on a par with (82a) whereas (83b) is less acceptable: (83) a Sos tuos an créskitu de frores. ‘ Your flowers have grown.’ b ?Su meu at assutatu de panurtzu. ‘ My towel has dried.’ From the data in (81) and (83) one might conclude that the distinction between unaccusative and ‘normal’ intransitive verbs cuts across the class of verbs in (79), créskere being unaccusative whereas assutare is intransitive. Note, however, that this is the worst possible conclusion from a theoretical point of view since it would force us to abandon the strong interpretation of both (77a) and (77b). An alternative conclusion is that the syntactic phenomena reviewed above are conditioned by pragmatic rather than purely structural factors. In this connection, we may note that both of the constructions in question have the effect of diverting focus away from the verb, to the quantifier in (80)–(81) and to the possessive in (82)–(83). It is therefore conceivable that the acceptability differences which we have observed reflect the extent to which particular verbs can be used to convey information which is presented as non-salient, verbs like arrivare and créskere being intuitively more amenable to such a use than ballare or assutare. Given our inability to find syntactic phenomena which constitute a clear and reliable diagnostic for unaccusativity, the status of the verbs in (79) must be based on criteria (77a) and (77b). If we classify these verbs as unaccusative, by virtue of their semantic properties, we must conclude that only a subset of unaccusative verbs take éssere as their auxilary, thus ruling out a purely structural account based on ‘subjectivisation’ of the underlying object. On the other hand, if we treat these verbs as ‘normal’ intransitives, we must conclude that unaccusative verbs constitute a subset of the verbs which have the semantic properties outlined in (77b). With these considerations in mind, let us now compare the problematic verbs in (79) with verbs which can be classified as unaccusative on the basis of both (77a) and (77b), a representative sample of which is given in (84), divided into three subclasses for ease of reference:

< previous page

page_109

next page >

< previous page

page_110

next page >

Page 110 (84)a abbarrare ‘stay, remain’, andare ‘go’, ampilare ‘go up’, arrivare ‘arrive’, diventare ‘become’, éssere ‘be’, falare ‘go down’, intrare ‘enter’, issire ‘go out’, istare ‘stay, live’, rúghere ‘fall’, vénnere ‘come’, etc. b aggradare ‘please’, (dis)piághere ‘(dis)please’, importare ‘concern, be important’, etc. c isparire ‘disappear’, náskere ‘be born’, mórrere ‘die’, etc. Note that all the verbs in (79) (and others which take a Theme as subject but select áere as their auxiliary) are one-place predicates whereas most of the verbs in (84) allow or require a complement in addition to the Theme (a locative or attributive complement in (84a), a dative complement in (84b)), as illustrated in (85): (85) a Juanne est abbarratu póveru. ‘John remained poor.’ b Maria est arrivata (a domo). ‘Mary has arrived (home).’ c Cussu regalu lis est aggradatu. ‘That gift pleased them (dat.).’ This difference is not absolute since the verbs in (84c) are also oneplace predicates. However such exceptions seem to be very limited in number; there are many other verbs which could be added to the lists in (84a, b), but the list in (84c) appears to be more or less exhaustive of the class of one-place verbs which take éssere (possible additions are capitare/sutzédere ‘happen’, though these also allow a dative complement as we noted in 3.2.2). Moreover, these verbs constitute a coherent semantic class in that they express coming into or going out of existence, unlike the verbs in (79) which simply denote a change of state. Indeed, in so far as there is an intuitive connection between existence and location (as we suggested in 3.2.2 with regard to the existential function of the locative clitic bi) the verbs in (84c) might be analysed as involving an implicit Source or Goal corresponding, very roughly, to the ‘real world’. The correlation between selection of éssere and an additional complement manifests itself in other more concrete ways. First, there are some verbs of movement such as cúrrere ‘run’ or volare ‘fly’ which normally take áere when used alone, as in (86), but which take éssere when accompanied by a Goal or Source complement, as in (87): (86) a Appo cúrritu meta. ‘I ran a lot.’

< previous page

page_110

next page >

< previous page

page_111

next page >

Page 111

b Su pudzoneddu at volatu pro sa prima via. ‘The little bird flew for the first time.’ (87) a So cúrritu a domo. ‘I ran home.’ b Su pudzoneddu est volatu dae su nidu. ‘The little bird flew from the nest.’ We postulate that these verbs are normal intransitives in (86) but have unaccusative status in (87). This is consistent with criterion (77b) in so far as these verbs can be analysed as activity verbs taking an agentive subject in (86) (on a par with verbs like ballare ‘dance’) whereas in (87) they can be construed as changeof-place verbs (like andare ‘go’ or vénnere ‘come’) whose superficial subject is a Theme. Second, there are some verbs of the type listed in (79) which normally take áere, as in (88a), but which take éssere when accompanied by an incidental dative which denotes the possessor of the Theme entity, as in (88b): (88) a Sos frores an créskitu. ‘The flowers have grown.’ b Sos pilos mi sun créskitos. [the hairs me (dat.) are grown] ‘My hair has grown.’ The nature of this relationship between selection of éssere and the existence of a complement other than the Theme is difficult to define precisely. Apart from the problem raised by the verbs in (84c), there is the problem that some verbs take éssere even when the additional complement is implicit (e.g. arrivare in (85b)) whereas with other verbs (like those in (87) and (88b)) éssere is selected only when the additional complement is overt. Moreover, in cases like (88b) the dative element is presumably not an argument of the verb. It would be extremely cumbersome to formulate these details as conditions on our basic generalisation that éssere is selected whenever the underlying object is subjectivised, especially since they involve semantic or θproperties of verbs which are not necessarily manifest in syntactic structure. It would, however, be natural for such conditions to play a role in the semantic definition of the class of unaccusative verbs (or uses of verbs), yielding a more restrictive definition than that postulated in (77b). In other words, we can achieve an optimally simple and natural overall account by maintaining the strong interpretation of (77a) and by modifying (77b) in line with our observations above so that it delimits a semantic class of verbs which coincides with

< previous page

page_111

next page >

< previous page

page_112

next page >

Page 112 selection of éssere. Given that the distinction between ‘normal’ intransitives and unaccusatives is essentially a matter of whether or not a θ-role is assigned to the underlying subject position, the facts which we have observed can be accounted for as follows: (89)No θ-role is assigned to the underlying subject if one or more of the following conditions obtain: (a) the verb has no θ-role to assign (e.g. weather verbs); (b) the verb denotes a change in existential status (e.g. the verbs in (84c)); (c) the verb is construed as expressing a relation between a Theme and some other entity or property (which need not be overtly specified) (e.g. the verbs in (84a, b)); (d) the Theme is qualified by a possessive dative (as in (88b)). If none of the conditions in (89) hold, a θ-role must be assigned directly to the subject. In particular, verbs which take a Theme as their sole argument (e.g. those in (79)) must assign the Theme role directly to the subject position except in the special cases covered by (89b) and (89d). Consequently, there is no ‘subjectivisation’ of an underlying object and, therefore, the marked auxiliary éssere is not selected. Our conclusion that verbs of the type in (79) are not inherently unaccusative should be regarded as tentative in so far as it is based on considerations of descriptive simplicity rather than on direct empirical evidence, and also because at least some verbs of this type display unaccusative properties in other, related languages (e.g. Italian). If this conclusion is correct, it therefore follows that the class of unaccusative verbs may vary from one language to another. Moreover, our analysis makes the claim that, in Sardinian at least, unaccusativity is not a purely lexical property but, in some cases, is induced by the syntactic context. The fact that verbs with a Theme ‘subject’ function as unaccusatives when accompanied by an overt complement or possessive dative may be related to the fact that this additional element predicates a property of the Theme. More specifically, these cases might be accounted for in terms of a general principle to the effect that such elements can only predicate an NP which occurs within the VP in the underlying structure; i.e. the Theme NP in examples like (87) and (88b) must be an underlying object in order to be predicated by the locative or dative expression. In the above discussion, we have considered only cases where the

< previous page

page_112

next page >

< previous page

page_113

next page >

Page 113 Theme is an NP. A slight problem for our account of auxiliary choice is posed by impersonal constructions such as (90) where the Theme is a clause: (90) a M’est pássitu ki Maria ésseret maláida. ‘It seemed to me that Mary was ill.’ b M’est piághitu a andare a incue. ‘It pleased me to go there.’ Note that in such constructions the auxiliary is always éssere. Consequently, our analysis requires us to assume that the Theme is an underlying object which has assumed the status of superficial subject. The problem is that there is no evidence to suggest that the complement clauses in (90) have subject status. Nor indeed is there any theoretical reason to assume that ‘subjectivisation’ has taken place in so far as clauses, unlike NPs, do not require a Case feature and therefore need not assume subject status in order to be licensed by nominative Case. Consequently, our analysis would lead us to expect áere in such constructions just as in cases where a Theme NP is licensed by the clitic bi rather than by nominative Case. We shall not attempt to solve this problem here. 3.2.4 Existential uses of áere and éssere Both áere ‘have’ and éssere ‘be’ can be used to assert the existence of an entity or to introduce an entity into the discourse, the choice between these two verbs being determined essentially by the definiteness or indefiniteness of the Theme NP: (91) a B’at metas frores in sa tanca. b ?Bi sun metas frores in sa tanca. ‘There are many flowers in the meadow.’ (92) a *B’at sos prattos in mesa. b Bi sun sos prattos in mesa. ‘There are the plates on the table.’ Áere can only be used when the following NP is indefinite and shows no agreement with this NP whereas with éssere the Theme NP is normally definite and triggers number and person agreement. Note that this is exactly the situation which we observed when these two verbs function as perfective auxiliaries qualifying an unaccusative verb with a Theme NP in the postverbal position (see 3.2.2, examples (67), (69)–(70)). The clitic bi (or nke; see p. 368, n.3) is obligatory in existential

< previous page

page_113

next page >

< previous page

page_114

next page >

Page 114 constructions with both verbs. However, the following examples which include an interrogative locative expression indicate that the status and function of this clitic is not the same in the two constructions: (93) a In ube b’ at metas frores? ‘Where are there many flowers?’ b *In ube bi sun sos prattos? ‘Where are the plates?’ Again this pattern is identical to that which we observed with impersonal constructions and constructions with a postverbal subject (see examples in (71)). In line with our discussion in 3.2.2, we propose that when bi occurs with éssere, as in (93b), it has a locative value whereas with áere, as in (93a), it functions as an existential operator which serves to license the postverbal NP. On the basis of the above data, we postulate that both éssere and existential áere are unaccusative verbs, but that they differ in terms of the means whereby the Theme NP is licensed. With áere, the Theme NP can only be licensed by occurring within the scope of the existential operator bi and thus must be indefinite. On the other hand, with éssere the Theme must assume subject status so that it can acquire nominative Case via agreement of the verb. This hypothesis is borne out by choice of perfective auxiliary in these constructions. In constructions with éssere, the perfective auxiliary is also éssere (due to subjectivisation of the Theme), whereas with áere (where no subjectivisation occurs) we have áere as the auxiliary: (94) a Bi sun istatos issos. ‘They were there.’ b B’at áppitu metas problemas. ‘There have been many problems.’ 3.2.5 Copular verbs Copular verbs are verbs such as éssere ‘be’, diventare ‘become’, abbarrare ‘remain’ and párrere ‘seem, look (like)’ which take an attributive complement (typically an AP or NP) which describes a property of the subject entity: (95) a Juanne est artu. ‘John is tall.’ b Maria diyentat una bella pitzinna. ‘Mary is becoming a beautiful girl.’

< previous page

page_114

next page >

< previous page

page_115

next page >

Page 115 c S’abba abbarrat fritta. ‘The water remains cold.’ d Cussa rocca paret un’elefante. ‘That rock looks like an elephant.’ In 3.2.3 we classified verbs of this type as unaccusative on the grounds that they take éssere as their perfective auxiliary (see (84a)). However there are two plausible underlying structures which are compatible with this analysis; either the Theme NP and the attributive expression are independent complements of the verb, as in (96a), or they might be analysed as forming a constituent, labelled X in (96b) which acts as the complement of the copula, a ‘small clause’ in the sense of Chomsky (1981): (96) Under either analysis, the sentences in (95) can be derived by moving the postverbal NP into the empty subject position (θ). Note that movement of the subject is obligatory in such cases, due to a general constraint which prevents postverbal subjects from co-occurring with other postverbal complements at the surface level (see 7.1.3). The choice between the underlying structures in (96) is largely a technical, theoryinternal question which we shall not attempt to resolve here. Nevertheless, it may be useful to consider some properties of copular constructions in Sardinian which are potentially relevant to this question. First, we may note that clitics corresponding to complements of the attributive expression can generally be prefixed to the copular verb (or its auxiliary in compound tenses), though with some restrictions, as in (97) where nde ‘of it’ represents a complement of cuntenta or su rnere: (97) a

Maria nde cuntenta. ‘Mary was/remained/became/appeared happy about it.’ b

Eliias nde su mere. ‘Elias was/remained/became/appeared the master of it (e.g. of the house).’

< previous page

page_115

next page >

< previous page

page_116

next page >

Page 116 The situation with abbarrare and diventare, particularly in (97a), is less clear than with éssere. For some reason (presumably semantic or pragmatic) these two verbs tend to be slightly unnatural with the sort of adjectives which normally take complements (for example, Maria est abbarrata/diventata cuntenta de sa situatzione is somewhat odd, though presumably syntactically well formed). A further complication is that nde can also be used as an adverb of reason (corresponding roughly to ‘thus’ or ‘for this reason’), with the result that the relevant cases can be interpreted as ‘Mary thus remained/became happy’, a reading which is not clearly distinguishable from the interpretation under discussion. Nevertheless, our informants’ comments on contextualised examples of the type in (97a) suggest that the cliticisation phenomenon under discussion is possible with these two verbs, though perhaps less readily than with éssere. The unacceptability of párrere in (97b) is rather surprising since there is nothing particularly odd about the non-cliticised variant Elias paríat su mere de sa domo ‘Elias looked (as though he was) the master of the house’, and párrere is clearly possible in (97a) with the intended interpretation. It is possible that the deviance of párrere in (97b) is due to pragmatic factors in so far as the phrase su mere de sa domo in the example just cited would normally be taken as symbolic of a certain lordly appearance or deportment (rather than alluding to mastership of a particular house), an interpretation which is arguably precluded when de sa domo is pronominalised. Given certain assumptions, the general possibility of cliticisation of the complement of the attribute may argue against underlying structures of the type in (96b). If the initial NP in the ‘small clause’ (Juanne in (96b)) is analysed as the subject of the ‘small clause’, we would expect the presence of this subject to block cliticisation phenomena of the type illustrated in (97) according to general theoretical principles which prevent certain syntactic processes from applying across a subject (for example, the Specified Subject Condition of Chomsky (1973, 1976); or the Binding principles of Chomsky (1981)). However, such principles would not come into play in the structure (96a) where the postverbal NP is not, in any obvious syntactic sense, the subject of the attributive expression. By the same token, if the version of (97b) with párrere is deemed to be ungrammatical (rather than simply pragmatically deviant), this judgement could be accounted for by postulating an underlying structure of the ‘small clause’ type in (96b) for just those cases where párrere occurs with an attributive NP. More generally, a ‘small clause’ analysis for copular constructions

< previous page

page_116

next page >

< previous page

page_117

next page >

Page 117 with párrere might be justified in terms of congruency with cases where this verb takes a finite complement clause. Thus, the underlying structure of a copular sentence such as Lukia (mi) paret maláida ‘Lucy seems ill (to me)’ would be parallel to that of (Mi) paret ki Lukia siat maláida ‘It seems (to me) that Lucy is ill’. Note, however, that there are significant semantic differences between the two constructions, particularly when the attributive expression is an NP. For example, (Mi) paret ki cussa rocca siat un’elefante ‘It seems (to me) that this rock is an elephant’ implies that the rock might actually be an elephant, whereas Cussa rocca (mi) paret un’elefante ‘This rock looks (to me) like an elephant’ simply asserts that there is a similarity between the two objects. We leave open the question of whether this semantic difference should be represented in terms of underlying structure. Let us now turn to ‘identificational’ constructioris such as (98) which differ from the ‘predicational’ constructions discussed above in that the postverbal NP identifies the entity described by the preverbal NP rather than asserting some property of this entity: (98) Su vinkitore est Máriu. ‘The winner is Mario.’ Possibly, such sentences can be derived from the same underlying structure as predicational constructions of the type in (95). Given an underlying structure of the form [θ est Máriu su vinkitore], sentence (98) can be derived by preposing the attributive NP, su vinkitore, whereas movement of the first NP, Máriu, yields the predicational sentence Máriu est su vinkitore ‘Mario is the winner’ as proposed above. Note, however, that in the identificational construction the inflection of the verb is determined by the postverbal NP, which shows nominative Case when it is a first or second person singular pronoun: (99) Su mere lit. ‘The boss am I/are you.’ ‘The boss is me/you.’ The analysis of identificational constructions is complicated in Sardinian by the availability of an alternative interpretation for examples like (98)–(99) whereby the attributive phrase is fronted to achieve a focusing effect similar to that of contrastive stress in English (e.g. ‘Mario is the winner (not the loser)’, ‘I am the boss (not the servant)’). With such interpretations, the sentences in (98)–(99) can

< previous page

page_117

next page >

< previous page

page_118

next page >

Page 118 be derived by the general fronting process discussed in 1.1.2 (see also 7.1.4, 7.1.5) which can apply to a wide variety of constituents and is not restricted to copular sentences (e.g. Appo lessu unu libru→Unu libru appo lessu ‘I have read a book’). However, with the identificational interpretation under discussion, stress and focus occur on the postverbal NP, a pattern which is not typical of fronted constructions. This difference in interpretation is not simply a matter of stress and focus. With the ‘focalised attribute’ reading, the postverbal NP can be omitted freely, but this is not possible with the identificational reading even when this NP is totally redundant (e.g. Su mere so is fine as a focused variant of So su mere ‘I am the boss’, but not as a variant of the identificational sentence in (99)). For some speakers, sentences like those in (100), with copular verbs other than éssere, can have an identificational reading of the type suggested in the translations, contrary to what we find in English (cf. Wasow 1977): (100) a Su vinkitore paret Máriu. ‘The winner seems to be Mario.’ b Su mere divento jeo. ‘The one who will become the boss is me.’ Again, these judgements cannot be attributed to the general fronting/focusing process, which would yield interpretations of the type ‘Mario seems to be the winner (not the loser)’, ‘I will become the boss (not the assistant)’, with stress on the initial NP. Although these interpretations are indeed possible, the availability of the readings indicated in our translations suggests that there is some process, distinct from the general fronting/focusing process and specific to copular constructions, which allows the attributive expression to occur preverbally yielding an identificational interpretation. An obvious hypothesis is that in the identificational construction the attributive NP moves into the preverbal subject position, just like the Theme NP in the predicational construction. Recall, however, that in the identificational construction it is the postverbal NP which enjoys subject status with regard to verb agreement and nominative Case. A further problem with this hypothesis is that in certain discourse contexts we find sentences like (101) which resemble identificational sentences in that the postverbal NP identifies the entity to which the property described by the preverbal expression applies, but this preverbal expression belongs to a syntactic category (e.g. AP) which cannot normally occur in subject position:

< previous page

page_118

next page >

< previous page

page_119

next page >

Page 119 (101)

Maccu ses tue. [stupid are you] ‘You are the one who is stupid.’ A similar construction is possible with verbs other than éssere, at least for those informants who accept the identificational reading for the examples in (100): (102) Maccu abbarras/divintas tue. ‘You are the one who is remaining/becoming stupid.’ For examples like (101) and (102) it seems reasonable to postulate a preposing process which places the attributive expression in a preverbal but non-subject position, but without the focusing effects induced by the general fronting process. This does not necessarily preclude the possibility that the initial NP in identificational constructions of the type in (98)–(100) does occupy the preverbal subject position. Nevertheless, in view of the similarities between the two types of construction (in terms of their interpretation and the fact that the verb agrees with the postverbal NP), it seems perverse to suppose that they differ in their syntactic structure. We shall return to this matter in 7.1.5. 3.2.6 Pronominal verbs We use the term ‘pronominal’ to denote constructions involving the pleonastic use of the reflexive clitic si (and its first and second person counterparts). In this section we shall be concerned only with cases where the reflexive clitic is selected as a lexical property of the verb (for other constructions with si which result from productive syntactic processes yielding a passive-like interpretation, see 3.2.8). Typical cases of the pronominal verb construction, which we mentioned briefly in 3.2.2, are illustrated in (103), where si signals the intransitive use (with Theme as subject) of verbs which can also be used transitively with an agentive subject, as in (104): (103) a Sa janna s’est abberta/tuncata. ‘The door opened/closed.’ b Sa camba s’est secata. ‘The branch broke.’ c Sa tzíkkera s’est candita. ‘The cup cracked.’

< previous page

page_119

next page >

< previous page

page_120

next page >

Page 120

d Sa tiadza s’est crentiata. ‘The tablecloth became stained.’ (104) a Maria at abbertu/tuncatu sa janna. ‘Mary opened/closed the door.’ b Pretu at secatu sa camba. ‘Peter broke the branch.’ c Lukia at canditu una tzíkkera. ‘Lucy cracked a cup.’ d Juanne at crentiatu sa tiádza. ‘John stained the tablecloth.’ Note that in pronominal constructions of the type in (103), which we shall refer to as ‘Neutral’ pronominal constructions, the Agent role is suppressed completely (the event or change of state being presented as occurring spontaneously), in contrast to the passive-like constructions which we shall discuss in 3.2.8. Selection of si in constructions of the type (103) appears to be an arbitrary lexical property of the verb. As we noted in 3.2.2, there are many verbs (e.g. assutare ‘dry’, brujare ‘burn’, buddire ‘boil’, etc.) which can be used both transitively and intransitively in the same way as those in (103)–(104) but which are nonpronominal in their intransitive, inchoative use: (105) a Sos pannos assutan medzus in su sole. ‘The clothes dry better in the sun.’ b Sa linna at brujatu. ‘The wood burnt.’ c S’abba at budditu. ‘The water boiled.’ Also there are some verbs which can be used both pronominally and non-pronominally, with some idiolectal variation but without any apparent difference in meaning: (106) a Sa ferita ‘The wound healed.’ b Sa mákkina ‘The car stopped.’ We leave open here the question of whether Neutral pronominal

< previous page

page_120

next page >

< previous page

page_121

next page >

Page 121 verbs like those in (103) are unaccusative (as argued for Italian by Burzio (1986)) or whether the Theme role is assigned directly to the subject position (as we suggested tentatively for verbs of the type in (105) in 3.2.4). Clear empirical evidence on this matter is difficult to find in so far as the choice of éssere as auxiliary can plausibly be attributed to the presence of si. We may also note that these verbs do not conform to the criteria for unaccusativity which we proposed in (89) in 3.2.3. Intransitive verbs formed from adjectives (with the meaning ‘be/become adj.’) are typically pronominal: (107) s’ammudare ‘become quiet’ (< mutu ‘quiet’), s’attristare ‘become sad’ (< tristu ‘sad’), s’imbetzare ‘grow old’ (< vetzu ‘old’), si cuntentare (de NP) ‘be content (with NP)’ (< cuntentu ‘happy’), s’immalaidare ‘become sick’ (< maláidu ‘sick’), s’immakkiare (de NP) ‘go crazy (about NP)’ (< maccu ‘mad’). When the subject is human, the distinction between Neutral-pronominal and ‘true’ reflexive constructions is somewhat blurred in some cases, as in the following examples which are ambiguous according to whether the subject is interpreted as an Agent as well as the Theme (e.g. in (108a) ‘John inflicted a wound on himself vs. ‘John got hurt’): (108) a Juanne s’est fertu. ‘John hurt himself.’ b Maria s’est brujata. ‘Mary burnt herself.’ We assume that with the involuntary, non-agentive reading such examples are Neutral-pronominal constructions. Further examples of pronominal verbs which take non-agentive, human subjects are given in (109): (109) s’abbidzare (de NP) ‘notice’, s’affetzionare (a NP) ‘become fond (of NP)’, s’arrenegare ‘become angry’, si fidare (de NP) ‘trust’, s’impudare (de NP) ‘be ashamed (of NP)’, s’indiosare (de NP) ‘to fall in love (with NP)’, s’irballare ‘be mistaken’, s’ispantare (de NP) ‘be astonished (at NP)’, s’offéndere (de NP) ‘take offence (at NP)’, si rassinare (a NP) ‘resign oneself (to NP)’, s’umbrare ‘take umbrage’, etc. With most verbs of this type, the Theme can only be expressed (if at

< previous page

page_121

next page >

< previous page

page_122

next page >

Page 122 all) as a prepositional complement. A possible explanation of this fact is that the reflexive clitic, though pleonastic, has the status of a direct object, thus preventing the Theme from being realised as a direct object. A notable exception to this generalisation is the verb s’ammentare which allows the Theme to be expressed either as a direct object or as the complement of de: (110) M’ammento (de) sa prima via ki so vénnitu a Sardigna. ‘I remember the first time that I came to Sardinia.’ With this verb, we postulate that the pleonastic clitic is dative, like the Experiencer complement of ammentare when used nonpronominally with the sense ‘remind’ as in (111): (111) Custa fotografía lis ammentat (de) cuss’occasione. ‘This photograph reminds them of that occasion.’ There are many pronominal verbs, of various types, which take an agentive subject: (112) si divértere/s’ispassiare ‘enjoy oneself, have fun’, s’imbreacare ‘get drunk’, si cuare/s’istikkire ‘hide (oneself)’, si cojuare (kin NP) ‘get married (to NP)’, si cuffessare ‘make confession (e.g. to a priest)’, si cummunicare ‘take communion’, si lamentare ‘lament’, si pentire ‘repent’ (also non-agentively with de+NP ‘regret’), si pasare ‘rest’, etc. Verbs of posture and some verbs of movement are also pronominal: (113) s’accurtziare (de NP) ‘approach’, s’allagare (de NP) ‘move away (from NP)’, s’antziare ‘rise’, s’arricare/si pesare ‘get up’, s’imbrenucare ‘kneel’, s’incurbare ‘bow, bend over’, si kitzire ‘move out of the way’, si corcare ‘lie down’, si móvere ‘move’, si sédere ‘sit down’, etc. Many of the verbs in (112)–(113) can also be used transitively (e.g. Juanne at cuatu su dinari ‘John hid the money’, Maria at corcatu su pitzinneddu ‘Mary laid the little boy down’). With such verbs, it is unclear whether the forms with a reflexive clitic should be treated as distinct pronominal verbs or as genuine reflexive uses of transitive verbs. Reflexive clitics may also occur with some unaccusative verbs accompanied by the Source clitic nke (sometimes nde):

< previous page

page_122

next page >

< previous page

page_123

next page >

Page 123 (114) a Lukia sink’est andata. ‘Lucy went away.’ b Appo vistu sa janna abberta e minke so intratu. ‘I saw the door open and went in.’ c Babbu sink’est ghiratu a sas úndiki.’ ‘Father came back at eleven o’clock.’ d Tziu Berte sink’est mortu eris.’ ‘Uncle Bert died yesterday.’ Note that with these verbs the reflexive clitic cannot be used without nke (or nde): *Lukía s’est andata/ghirata/intrata, etc. For the moment we leave open the question of whether such cases should be regarded as pronominal verbs in their own right or whether the reflexive clitic should be analysed as an ‘ethic’ dative accompanied by an incidental locative. This question will be discussed in detail in 5.2.4 in the general context of incidental uses of clitic pronouns. A similar question arises with various other verbs which can occur with or without a reflexive clitic, sometimes with a change in meaning and/or selection of complements, but without any change in the argument which is expressed as the subject (unlike cases of the type illustrated in (103)–(104)). A miscellaneous selection of such cases is given in (115): (115) (s’) assimidzare (a NP) ‘resemble’, detzídere (de INF) ‘decide (to INF)’ vs. si detzídere (a INF) ‘resolve (to INF)’, dormire ‘sleep’ vs. si dormire ‘go to sleep’, irmenticare NP or s’irmenticare (de NP) ‘forget’, isseare (NP) ‘wait (for NP)’ vs. s’isseare NP ‘expect’, rídere ‘laugh’ vs. si (nde) rídere (de NP) ‘mock, laugh at NP’. With (s’) assimidzare the two forms appear to be synonymous, though the pronominal form is the more usual and can perhaps be regarded as the Neutral-pronominal counterpart of the transitive use of this verb; assimidzare NP a NP ‘compare/liken NP to NP’. Similarly, the pronominal constructions with detzídere and dormire may be related to the causative transitive uses of these verbs (either as Neutral-pronominals or genuine reflexives): L’appo detzisu a lu fákere . ‘I persuaded him to do it’, Sa mamma est dorminde sa pipiedda The mother is getting the baby to sleep’. In other cases (e.g. isseare, rídere ) the difference in meaning may be simply due to an ‘intensifying’ effect induced by the presence of an ethic reflexive clitic (see 5.2.4 for detailed discussion and further examples).

< previous page

page_123

next page >

< previous page

page_124

next page >

Page 124 3.2.7 Passive constructions Passive constructions with éssere and the past participle are formed in much the same way as in other Romance languages. The entity corresponding to the subject of the active verb is either left unspecified or is introduced by the preposition dae ‘from’ (de in Campidanese), whereas the direct object of the active verb becomes the subject of the passive, determining the agreement features of the finite auxiliary and the participle(s): (116) a Custa domo est istata fraicata dae un’Italianu. ‘This house was built by an Italian.’ b Cussos libros non fin iscrittos in sardu. ‘These books were not written in Sardinian.’ Impersonal passives with a clausal complement are possible, but usually only in the perfect: (117) a *Est créitu/natu ki b’ at pantásimas in cussa conkedda. ‘It is believed/said that there are ghosts in that cave.’ b Est istatu detzisu ki Juanne sinke devíat andare. ‘It was decided that John should go away.’ c Est istatu provatu ki cussu no’est veru. ‘It has been proved that that is not true.’ However, impersonal passives of intransitive verbs are not possible even with a prepositional complement (e.g. there is no construction corresponding to French Il sera parlé de cette question lit. ‘It will be talked about that question’): (118) *At a éssere faeddatu de cussa kistione. ‘That question will be talked about.’ Moreover, there is no passive counterpart to impersonal constructions with bi and an indefinite postverbal NP of the type discussed in 3.2.2: (119) *B’at/est istatu fraicatu metas domos. lit. There has been built many houses.’ This suggests that éssere in its use as a passive auxiliary (as in its perfective and existential uses) always correlates with subjectivisation of an underlying direct object. It should be noted that the passive with éssere and the past participle is somewhat uncharacteristic of Sardinian and is rarely used in colloquial speech. Indeed, Pittau (1972:90) goes so far as to claim

< previous page

page_124

next page >

< previous page

page_125

next page >

Page 125 that in the dialect of Nuoro this construction does not exist. Attestations of this construction, including the following example by Pittau himself, suggest that this claim is rather exaggerated: (120)“in paritzos puntos de su romanzu traduttu kie leghet at s’impressione ki issu esseret istau iscrittu dae Grassie ‘e Ledda non in limba italiana, ma in limba sarda” ‘in many places in the translated novel the reader has the impression that it was written by Grazia Deledda not in Italian but in Sardinian’ (Pittau; Foreword to Spiggia 1982 [EP: 7]) Nevertheless, we may concur with Pittau’s judgement to the extent that the passive seems to be a learned construction confined to fairly formal registers. At the same time, there are other, more ‘authentic’ constructions in which the past participle is used with a passive function. One such case is the modal passive with kérrere (with the sense of ‘need’) which is used widely in all registers: (121) a Cussas faínas keren fattas prima de nos corcare. ‘Those chores need to be done before we go to bed.’ b Cussa dzente keret tímita. ‘Those people are to be feared.’ c Sa mákkina keret accontzata dae unu meccánicu. ‘The car needs to be repaired by a mechanic.’ The syntactic properties of this construction appear to be the same as those of the passive with éssere — note in particular that the past participle agrees with the superficial subject and that the Agent can be specified by dae +NP, as in (121c). Another passive construction which is typical of Sardinian is the use of the past participle with the preposition kene ‘without’, as in (122): (122) Custa camisa est kene lavata. [this shirt is without washed] ‘This shirt has not been washed.’ Note that in this construction the past participle agrees with the subject of éssere. However, unlike normal passives of the type in (116), the Agent cannot be specified, nor can the participle be accompanied by a complement or modifier: (123) a *Sa petha est kene mandicata dae sos pitzinnos. ‘The meat has not been eaten by the children.’

< previous page

page_125

next page >

< previous page

page_126

next page >

Page 126 b *Sa camisa est kene lavata in abba caente. ‘The shirt has not been washed in hot water.’ This situation is reminiscent of ‘un-passives’ in English: for example, The meat is uneaten (*by the children), The shirt is unwashed (*in hot water) . Various constructions are used in Sardinian as alternatives to the passive. One of these is the ‘Middle’ pronominal construction (e.g. Custa camisa si lavat in abba caente ‘This shirt can/should be washed in hot water’), which we leave for detailed discussion in 3.2.8. The topicalising effect of passive is often achieved by means of an active construction with left-dislocation of the direct object and (optionally) inversion of the subject, as in (124) which is roughly equivalent to (116a): (124) Custa domo, 1’at fraicata un’Italianu. [this house it has built an Italian] This construction is commonly used in preference to the passive, even in formal registers, as in the following attested examples: (125)a”Sa figura de sa coperta 1’at pintada [the illustration of the cover it has drawn Diego Asprone.” Diego Asprone] (acknowledgement on inner title page of Limbas 1) b”Su cheripighe sos pastores 1’usan [the propolis the shepherds it use pro meigare feritas de ispina e de lesolza.” to treat wounds of thorn and of knife] ‘Propolis (type of beeswax) is used by shepherds to treat scratches and cuts.’ [ABE: 11] Similarly, in place of the ‘short passive’ (i.e. without specification of the Agent), an active construction with null subject and third person plural inflection on the verb is often used: (126) Cussos libros, non los an iscrittos in Sardu. ‘These books, they were not written in Sardinian.’ The same formula can be used instead of impersonal passives such as those in (117)–(119): (127) a Creden/nan ki b’ at pantásimas in cussa conkedda. ‘They believe/say that there are ghosts in that cave.’

< previous page

page_126

next page >

< previous page

page_127

next page >

Page 127 b An detzisu ki Juanne sinke devet andare. ‘They decided that John should go away.’ c An provatu ki cussu no’est veru. ‘They have proved that that is not true.’ d An a faeddare de cussa kistione. ‘They will talk about this question.’ e An fraicatu metas domos. ‘They have built many houses.’ 3.2.8 Pronominal voice The reflexive clitic si can be used as a means of assigning generic reference to the understood subject, often with a modalising effect on the verb: (128) a In custa bidda si ballat meta. ‘In this village people dance a lot.’ b Non si dormit in custu lettu. ‘One cannot/must not sleep in this bed.’ c Non si andat in cussos locos. ‘One should not go to such places.’ In cases like (128) where the verb is intransitive or unaccusative, the subject is always null and the verb always occurs in the third person singular. The same pattern is observed when the verb takes a clausal complement, as in the following examples which correspond to the passive examples in (117) in 3.2.7 and can be glossed in the same way: (129) a Si credet/nat ki b’ at pantásimas in cussa conkedda. ‘It is believed/said that there are ghosts in that cave.’ b S’est detzisu ki Juanne sinke devíat andare. ‘It was decided that John should go away.’ c S’est provatu ki cussu no’est veru. ‘It has been proved that that is not true.’ However, with verbs which take a direct object NP, this direct object always assumes subject status, determining agreement of the verb, even though it may remain in postverbal position: (130) a Cussas cosas non si faken/*faket. b Non si faken/*faket cussas cosas. ‘Those things are not done.’ (131) a Su vinu si biet mandicande. ‘Wine should be drunk while eating.’

< previous page

page_127

next page >

< previous page

page_128

next page >

Page 128 b Si faeddat su gadduresu in Témpiu. ‘Gallurese is spoken in Tempio.’ These constructions, which we shall refer to as ‘Middle’ constructions (following Ruwet 1972), resemble passive constructions in so far as the logical subject of the verb is rendered implicit and the underlying direct object becomes the superficial subject. More specifically, we assume that the clitic si prevents the verb from assigning a θ-role to the subject position and from assigning accusative Case to the direct object. A similar analysis can be postulated for cases like (128a,b) and (129) except that there is no direct object NP which can be promoted to subject. In this respect, Middle constructions differ from passives, which can only occur with transitive verbs, as we observed in 3.2.7. The fact that the verb must agree with the underlying direct object, as shown in (130), indicates that this NP must assume subject status even when it remains in postverbal position. In other words, impersonal Middle constructions are only possible when there is no NP which can function as the syntactic subject. Note that this restriction also applies when the underlying direct object is indefinite (i.e. we do not have Middle counterparts of the impersonal constructions discussed in 3.2.2): (132) Si fráican/*fráicat metas domos accurtz’a mare. ‘Many houses are being built by the sea.’ This restriction is perhaps rather surprising in so far as agreement of the verb is optional in such cases in Italian. Note, however, that it follows directly from our characterisation of the respective functions of Middle si and impersonal bi, which are diametrically opposed to each other. Whereas bi allows an intransitive or unaccusative verb to occur with an indefinite postverbal NP without agreement, si requires the direct object of a transitive verb to be licensed by means of verb agreement. The Middle construction appears to be possible with all verbs which allow a human subject (except, for some reason, éssere), including many verbs which do not normally allow the passive even in languages where use of the passive is more widespread, as in the following example with áere ‘have’: (133) Sas cosas, cando s’an, sun bonas. [the things when si-have are good] ‘Things, when one has them, are good.’ Unlike the passive, the Middle construction does not allow the

< previous page

page_128

next page >

< previous page

page_129

next page >

Page 129 understood subject to be specified (e.g. by the preposition dae ). Moreover, whereas in passives without a dae phrase the understood subject is usually construed as indefinite (corresponding roughly to ‘someone’), the understood subject in the Middle construction typically has a generic or hypothetical value (referring to people in general or any individual who happens to be in the situation described). A possibly related difference is that passive sentences typically describe specific events or situations occurring at a specific time, whereas the Middle construction is used predominantly to make normative statements, characterising a situation or action as customary, recommended, proscribed, etc., as in (128), (130)–(131). However, the Middle construction is not restricted to statements of a normative kind. For example, it can be used with verbs of perception to describe a scene from the point of view of a hypothetical observer, as in the following attested examples: (134) a “In palas de sos ortos, allargu, s’intendiat su ferru de unu tadzu in filada…” ‘Behind the gardens, far away, one could hear the bells of a flock of sheep on the move…’ [EP: 32] b “Sos montes de Lugula sinche bidian inedda…” ‘The mountains of Lula could (now) be seen in the distance…’ [EP: 46] Other examples which have a more clearly eventive interpretation are given in (129) and (132) above. The limitations on this eventive use are not clear. However we may note that the use of a Middle construction in a case like *Su pane s’est mandicatu ‘The bread has been eaten’ is not possible and that the sentence Si mandicaíat meta pane can only have a habitual interpretation ‘People used to eat a lot of bread’ not the progressive interpretation ‘A lot of bread was being eaten’. Middle constructions resemble Neutral pronominal constructions of the type discussed in 3.2.6 (cf. examples (103)–(104)) in that the 6-role of the surface subject is the same as that of the direct object of the corresponding transitive verb. The essential difference between the two constructions is that the role of the subject of the transitive verb is always implicit in the Middle construction whereas it is suppressed completely in the Neutral construction. Moreover, as we have noted, the Middle construction appears to be fully productive

< previous page

page_129

next page >

< previous page

page_130

next page >

Page 130 whereas the Neutral-pronominal construction is lexically restricted. Consequently, verbs such as assutare ‘dry’ and buddire ‘boil’ which we characterised earlier as non-pronominal when intransitive can be used pronominally when an implicit Agent is involved, as in the following (compare examples (105) in 3.2.6): (135) a Sas tassas s’assutan kin custu panurtzeddu. ‘Glasses are to be dried with this towel.’ b S’abba si buddit prima de la bíere. ‘Water should be boiled before drinking it.’ By the same token, Neutral-pronominal verbs which also have an agentive-transitive use are potentially ambiguous between a ‘spontaneous event’ (Neutral) reading and a reading with an implicit Agent (Middle): (136) a Sa janna si tuncat cando b’ at ventu. ‘The door closes (by itself) when it is windy.’ (Neutral) ‘One should close the door when it is windy.’ (Middle) b Sa linna si secat cando est sicca. ‘Wood breaks when it is dry.’ (Neutral) ‘Wood should/can be broken/chopped when it is dry.’ (Middle) 3.3 AUXILIARY VERBS 3.3.1 Choice of perfective auxiliary The choice between áere ‘have’ and éssere ‘be’ as perfective auxiliary is determined either by the nature of the main verb or by the presence or absence of a reflexive clitic. Cases of the first type were discussed extensively in 3.2.2 and 3.2.3. To recapitulate, éssere is selected whenever the underlying object of an unaccusative verb assumes subject status (as evidenced by agreement of the verb). Similarly, as we noted in 3.2.7, éssere is selected in passive constructions, where the underlying direct object becomes the syntactic subject. In all other cases, áere is selected—i.e. with all transitive verbs, with intransitive verbs which assign a θ-role directly to the subject position, with impersonal ‘weather’ verbs and in impersonal constructions with the clitic bi. In this section we shall concentrate on the choice of auxiliary in constructions with a reflexive clitic and on the possibility of formulating a unified account of auxiliary selection which is applicable to both ranges of cases.

< previous page

page_130

next page >

< previous page

page_131

next page >

Page 131 Whenever a reflexive clitic functions as a direct object (expressing a genuine reflexive or reciprocal relation with the subject), éssere is selected: (137) a Juanne s’est vistu in s’isprecu. ‘John saw himself in the mirror.’ b Sos pitzinnos si sun madzatos. ‘The boys hit each other.’ Also, éssere is systematically selected with Neutral-pronominal verbs: (138) Sa janna s’est abberta. ‘The door opened.’ Similarly, to the extent that Middle constructions allow an eventive interpretation of the sort which is normally associated with perfective aspect, the auxiliary is always éssere: (139) S’est detzisu ki Juanne sinke devíat andare. ‘It was decided that John should go away.’ When the reflexive clitic functions as a dative (as an indirect object of the verb, a dative of interest or an ‘ethic dative’) and the verb is accompanied by a direct object, the auxiliary is always áere: (140) a Maria e Lukia s’an mandatu paritzas lítteras. ‘Mary and Lucy sent each other several letters.’ b Juanne s’at fraicatu una bella domo. ‘John built himself a beautiful house.’ c Jakeddu s’at mandicatu una meledda. ‘Jack ate an apple.’ Similarly, in constructions of ‘inalienable possession’ where the reflexive clitic represents the possessor of a body-part denoted by the direct object, the perfective auxiliary is áere: (141) a Tonina s’at fertu s’anca. ‘Tonina hurt her leg.’ b Su pitzinnu s’at brujatu su póddike. ‘The boy burnt his finger.’ Note also that attributive nominals in postverbal position do not trigger selection of áere, as shown by the following contrast: (142) a Elias s’est fattu prítteru. ‘Elias made himself (= became) a priest.’ b Frantziscu s’at fattu unu muntone de dinari. ‘Francis made himself a pile of money.’

< previous page

page_131

next page >

< previous page

page_132

next page >

Page 132 On the basis of the evidence presented so far, we might envisage either of the hypotheses in (143): (143) a Reflexive clitics trigger selection of éssere unless the verb is accompanied by a direct object. b Accusative (but not dative) reflexive clitics trigger selection of éssere. On grounds of simplicity, hypothesis (143b) is preferable, if it is tenable. Moreover, this hypothesis offers the prospect of a unified account of auxiliary selection whereby éssere is selected whenever there is a relation of ‘identity’ between the superficial subject and the underlying direct object. In the case of unaccusative and passive constructions, this ‘identity’ involves the status of the same NP at different levels of syntactic representation whereas in reflexive constructions it takes the form of anaphoric binding. An obvious way of deciding between the two hypotheses in (143) would be to look at reflexive constructions with verbs which take a dative complement without a direct object. However, as we noted in 3.2.1, clear examples of verbs of this type are difficult to find. In particular, we noted that verbs such as faeddare ‘talk’ and pessare ‘think’ tend to treat an NP complement as dative (introduced by a even when the NP is one which does not normally manifest the prepositional accusative phenomenon) but favour the use of an accusative clitic when the complement is pronominalised (with some dialectal variation in the case of fáeddare). In reflexive constructions, these verbs require the auxiliary éssere: (144) a Sas pitzinnas si sun faeddatas. ‘The girls talked to each other.’ b Sas pitzinnas si sun pessatas. ‘The girls thought about each other.’ In such cases, the reflexive clitic can plausibly be analysed as accusative, on a par with non-reflexive clitics (Los appo faeddatos/pessatos ‘I talked to/thought about them’), so that selection of éssere is consistent with both of the hypotheses in (143). With verbs such as sorrídere ‘smile’, which require the dative form for both full NPs and clitics denoting the Goal entity, the facts concerning auxiliary choice in reflexive constructions are rather less clear. The general consensus of our informants was that éssere is required in such constructions, with some hesitation with regard to agreement of the past participle: (145) a *Sas pitzinnas s’an sorrísitu.

< previous page

page_132

next page >

< previous page

page_133

next page >

Page 133 b Sas pitzinnas si sun sorrísitas/?sorrísitu. ‘The girls smiled at each other.’ However, it should be noted that our informants showed some difficulty in arriving at these judgements. Typically, they concluded in favour of examples like (145b) only when confronted with the alternative with áere (which they rejected unhesitatingly) and often offered paraphrases such as Sas pitzinnas an sorrísitu tra issas lit. The girls smiled among themselves’, which do not raise the question of auxiliary choice with a reflexive clitic, as more natural ways of expressing the proposition in question. On the other hand, there were no hesitiations regarding examples with a reflexive clitic such as Sas pitzinnas si sorridían The girls smiled at each other’ where the verb appears in a simple tense form. A possible way of construing the reservations of our informants regarding (145b) is that they are unable to make intuitive judgements on such examples on the basis of their ‘internalised grammar’, but, when pressed to make a decision, rely on reasoning of a more conscious kind (e.g. analogy with other examples). In reflexive constructions where the verb is accompanied by a complement clause, the facts concerning auxiliary choice are also rather murky in some cases. In examples like (146) there is a clear preference for éssere: (146) a Maria s’est abbidzata ki fit tardu. ‘Mary noticed that it was late.’ b Mi so cumbintu ki cussu fit veru. ‘I convinced myself that that was true.’ This is consistent with both the hypotheses in (143) in so far as the reflexive clitic can plausibly be analysed as an accusative and the complement clause appears to have the status of an indirect object, as shown by the fact that substitution of an NP such as carki cosa ‘something’ requires the presence of a preposition: e.g. s’abbidzare de carki cosa ‘notice something’, cumbínkere carcunu de carki cosa ‘convince someone of something’. Conversely, in constructions with nárrere ‘say’, where the complement clause can be replaced by a direct object NP and the reflexive appears to have dative status (e.g. nárrere carki cosa a carcunu ‘say something to someone’), there is a strong preference for áere, though some informants found the variant with éssere marginally acceptable: (147) a S’an natu ki non fit veru. b ??Si sun natos ki non fit veru. ‘They told each other/themselves that it was not true.’

< previous page

page_133

next page >

< previous page

page_134

next page >

Page 134 The verb s’ammentare ‘remember’ allows both constructions, as we might expect given that its nonpronominal counterpart (with the sense ‘remind’) can behave like both nárrere and cumbínkere (ammentare carki cosa a carcunu ‘remind someone of something’, ammentare carcunu de carki cosa ‘remember something’—see 3.2.6): (148) a Lukia s’at ammentatu ki devíat andare a Núgoro. b Lukia s’est ammentata ki devíat andare a Núgoro. ‘Lucy remembered that she had to go to Nuoro.’ However, there are other cases where the coherent pattern observed in (146)–(148) appears to break down. For example, dimandare ‘ask’ and promíttere ‘promise’, which behave like nárrere when used nonreflexively (dimandare/promíttere carki cosa a carcunu ‘ask someone for something’/‘promise something to someone’), tend to favour éssere in reflexive constructions with a complement clause: (149) a Isso si si fit veru. ‘They asked each other/wondered whether it was true.’ b Issos si de s’aduzare. ‘They promised each other to help each other.’ A similar problem arises with the verbs detzídere ‘decide’ and pessare ‘think’ which require éssere when accompanied by a reflexive clitic (yielding meanings of the type ‘resolve, make up one’s mind’) even though their complements ostensibly have direct object status when no reflexive is present (e.g. detzíderelpessare carki cosa ‘decide/think something’): (150) Mi ki non lu faco prus. ‘I have resolved that I will not do it any more.’ The above discussion of (146)–(150) rests on the assumption that the grammatical function of reflexive clitics (accusative vs. dative) and complement clauses (direct vs. indirect object) can be determined with reference to parallel constructions in which the element in question is replaced by an item which indicates its grammatical function explicitly (e.g. an NP or non-reflexive clitic). However, the evidence in (149)–(150) suggests that this assumption cannot be rigorously maintained under either of the hypotheses envisaged in

< previous page

page_134

next page >

< previous page

page_135

next page >

Page 135 (143). Under hypothesis (143a) the preference for éssere in (149)– (150) requires us to suppose that the complement clause does not have the status of a direct object (despite the parallelism with direct object NPs) whereas hypothesis (143b) forces us to make the further supposition that the reflexive clitic is in fact accusative. Recall, however, that variation between accusative and dative Case, according to whether the complement is a clitic or an NP, is independently attested with the verbs faeddare ‘speak’ and pessare ‘think (about)’; see 3.2.1, examples (46)–(50). It is therefore conceivable that the reflexive clitics in (149)–(150) are construed as accusative even though the corresponding non-reflexive clitics and NPs are dative. More precisely, we may postulate that the Case or grammatical function of neutral items such as reflexive clitics is determined along a syntagmatic rather than paradigmatic axis (for example, a reflexive clitic is perceived as dative only if it co-occurs with an element which can be readily recognised as an (underlying) direct object, regardless of the status of parallel complements which bear an identical θ-role with regard to the verb in question). Within this perspective, the uncertainties which we have noted in connection with the examples in (147) and (149) and the unexpected choice of éssere in (150) may be attributed to the fact that complement clauses (unlike corresponding NPs or non-reflexive clitics) are also neutral with respect to the distinction between direct and indirect object status and thus do not systematically confer dative status on the reflexive clitic. Returning now to the examples in (145), the judgements indicated follow straightforwardly from the hypothesis in (143a) whereas (143b) makes the wrong predictions if we assume that si is dative. At the same time, hypothesis (143a) fails to explain the reservations expressed by our informants with regard to the choice of éssere in such cases. According to hypothesis (143b) these reservations might be explained in terms of a conflict between the accusative status of the reflexive imposed by the ‘syntagmatic strategy’ envisaged above and the intuition that this element is not a subcategorised complement of the verb. Our observation that parallel sentences with a simple tense are accepted unreservedly suggests that this conflict need not be resolved as long as it has no other syntactic or morphological repercussions. Our informants’ willingness to suggest alternative paraphrases and, in some cases, their inability to decide whether or not the participle should agree may also be taken as evidence that they are unable to resolve this conflict on the basis of their ‘internalised grammar’ but, when pressed to make a judgement, either assimilate such cases to the familiar pattern attested in constructions

< previous page

page_135

next page >

< previous page

page_136

next page >

Page 136 where the clitic is clearly accusative (e.g. Si sun vistos ‘They saw each other’) or entertain a ‘compromise’ solution whereby the peculiar status of the reflexive clitic is signalled by lack of agreement of the participle. To summarise this rather inconclusive discussion, we have noted that neither of the hypotheses in (143) gives the right results in all cases, yet, curiously, they are equivalent in so far as the occurrence of a dative element (particularly a clitic) without an accompanying underlying direct object is in any case highly untypical of Sardinian. The essential claim which is common to both approaches is that, in the absence of an underlying direct object, reflexive clitics which correspond to dative complements behave in the same way as accusative reflexives in selecting éssere. The difference between these two approaches is that (143a) treats the assimilation of ‘unaccompanied’ dative reflexives to accusatives as part of the rule which determines selection of éssere whereas (143b) requires us to treat it as an independent phenomenon (i.e. apparent ‘unaccompanied’ dative reflexives are, in fact, accusative). In support of the latter view we have cited other instances where the distinction between accusatives and datives is unstable and have suggested that an account along these lines may help to explain the uncertainty of speakers’ judgements in some cases. Moreover, as we noted at the beginning of this section, this approach also offers the possibility of a unified account of auxiliary selection in both reflexive and non-reflexive constructions. Thus, although hypothesis (143a) appears to provide a more accurate description of the observed facts (particularly with regard to cases like those in (145)), (143b) may prove to have greater explanatory value in relation to wider aspects of Sardinian syntax. As a final observation, we may note that in Middle constructions, the auxiliary is always éssere even when the verb takes a complement clause. Note in particular the contrast between the following Middle construction and the reflexive/reciprocal example in (147): (151) S’est natu ki Juanne est meta riccu. ‘It has been said that John is very rich.’ The choice of éssere in such cases is consistent with either of the hypotheses envisaged in (143) provided that the complement clause (as the underlying direct object of nárrere ) assumes subject status, even though promotion to subject is not required for reasons of Case assignment in so far as clauses (unlike NPs) do not need to be licensed by a Case feature (see discussion of examples (90) in 3.2.3).

< previous page

page_136

next page >

< previous page

page_137

next page >

Page 137 3.3.2 Syntactic properties of perfective and progressive constructions In perfective constructions with áere and éssere, clitic pronouns are always placed to the left of the auxiliary: (152) a Los appo kircatos. ‘I looked for them.’ b Bi semus arrivatos. ‘We arrived there.’ Clitics typically occupy the same position in progressive constructions with éssere, as in (153a), but an alternative construction with the clitic suffixed to the participle is also possible, as in (153b): (153) a Juanne los fit kircande. b Juanne fit kircandelos. ‘John was looking for them.’ Perfective and progressive constructions also differ slightly in terms of the range of elements which can intervene between the auxiliary and the participle. The class of elements which can occur freely in this position in both constructions includes the floated quantifiers tottu ‘all’ and ambos ‘both’ (cf. 2.1.2); ‘focal’ adverbs such as fintzas ‘even’, mancu ‘not even’, solu ‘only’; and aspectual adverbs such as appenas ‘hardly, just’, belle ‘almost’, dza ‘already’, galu ‘yet, still’, mai ‘never’, semper ‘always’. However, manner adverbs and time adverbs which have a specific time reference (e.g. como ‘now’, eris ‘yesterday’, lunis ‘(on) Monday’, etc.) cannot normally occur in this position. In addition to the above restrictions based on semantic classes, intervening adverbs in perfective constructions must normally consist of a single word, complex expressions such as die die ‘all day’ or tottu su mandzanu ‘all morning’ being at best marginal in this position even though they are semantically similar to aspectual adverbs: (154) Appo travallatu. lit. ‘I have always/all day/all morning worked.’ However, this restriction does not apply in progressive constructions:

< previous page

page_137

next page >

< previous page

page_138

next page >

Page 138 (155) Fippo travallande. lit. ‘I was always/all day/all morning working.’ The situation with perfective éssere is rather less clear, a problem to which we return below. The facts concerning cliticisation and intervening phrasal adverbs appear to be related. In progressive constructions with cliticisation to the participle, phrasal adverbs can be inserted freely (just as in (155)), but if the clitic is attached to éssere we find the same restrictions as in the perfective construction: 156) a Fippo b

kircándelos.

Los fippo kircande. lit. ‘I was always/all day/all morning looking for them.’ This evidence suggests that progressive constructions can have two distinct syntactic analyses: progressive éssere can function as a bona fide auxiliary on a par with perfective áere (with obligatory clitic-climbing and tight restrictions on intervening adverbs) or as a main verb which takes the participial expression as its complement (allowing phrasal adverbs, but blocking clitic-climbing). In this connection, we may note that when istare is used instead of éssere to express progressive aspect (see 3.1.2), it always has the properties associated with main verb status: (157) a Istaío tottu su mandzanu kircándelos. ‘I was all morning looking for them.’ b *Los istaío kircande. ‘I was looking for them.’ The syntactic distinction proposed above also correlates with a semantic distinction. In 3.1.2 we observed that the progressive with éssere (but not istare ) is often used with stative verbs. However, in such cases clitics must attach to éssere not to the participle: (158) a Non ti so cumprendende.

< previous page

page_138

next page >

< previous page

page_139

next page >

Page 139 b *Non so cumprendendéti. lit. ‘I am not understanding you.’ In other words, the progressive is possible with stative verbs only when éssere is a true auxiliary. In cases where progressive éssere (or istare) functions as a main verb, we propose that the participial expression has a status similar to that of attributive expressions in copular constructions such as Lukia fit cuntenta de su regalu ‘Lucy was happy with the present’. Indeed, the range of adverbs which can intervene between éssere and the attributive expression appears to be the same as that noted in (156a), including phrasal aspectual expressions but not adverbs denoting a specific time (e.g. Lukia fit tottu su mandzanu/*eris cuntenta ‘Lucy was all morning/yesterday happy’), in contrast to infinitival complement constructions where time adverbs can intervene between the main verb and the infinitive (e.g. Provo como a travallare ‘I try now to work’). In our discussion of copular constructions in 3.2.5 we suggested that one way of representing the unaccusative status of éssere is to assume that the superficial subject originates as the subject of a ‘small clause’ which functions as the complement of éssere (see structure (96b), p. 115). Adopting this approach for progressive constructions where éssere acts as a main verb, we may postulate that (153b) above is derived from the structure (159), where the ‘small clause’ is labelled X: (159) In 3.2.5 we noted that the ‘small clause’ approach to attributive copular constructions is somewhat problematic in that the presence of the subject of the ‘small clause’ should prevent a complement of the attribute from being cliticised to the copula, an undesirable result in the case of attributive constructions (because of examples like Lukia nde fit cuntenta ‘Lucy was happy with it’) but exactly the right result for progressive constructions given our hypothesis that ‘clitic-climbing’ is possible only when éssere is a true auxiliary. Thus, in (159) the pronoun cannot be cliticised to fit because of the presence of the subject Juanne (or its trace), but must instead be cliticised to the participle. In progressive constructions where éssere is a true auxiliary, we assume that éssere and the participle form part of the same simple sentence and are not separated by a ‘small clause’ boundary. There are a variety of structures which are consistent with this assumption but, for reasons which we shall discuss in the following sections, we

< previous page

page_139

next page >

< previous page

page_140

next page >

Page 140 propose that the auxiliary is adjoined to the VP headed by the participle, as in (160): (160) [S Juanne [VP fit [VP kircande PRONOUN]]] In the absence of an intervening subject, the pronoun can be cliticised to the auxiliary fit as in (153a). With regard to our observations concerning phrasal adverbials such as tottu su mandzanu, we make no specific claims about their structural position, but suggest that their inability to occur after auxiliary éssere is a consequence of the relatively ‘tight’ syntactic relation between the auxiliary and the participle in (160) as compared with that which obtains in (159). There is nothing in our account so far which will prevent the pronoun in (160) from being cliticised to the participle, thus giving an alternative derivation for (153b). However, this possibility must be excluded if we are to maintain our conclusion concerning the use of the progressive with stative verbs in (158). For present purposes, let us simply assume a condition to the effect that clitics must be attached to the leftmost available verb within the limitations imposed by general locality constraints such as the constraint which prevents movement across subjects. Further evidence of the need for such a condition will be presented in 3.3.3. Returning now to perfective constructions, we assume that these have the same structure as progressive constructions of the type in (160) except that when the auxiliary is éssere and the main verb is nonreflexive, the subject originates in a postverbal position, as discussed in 3.2.3: (161) a ‘John has danced.’ b ‘John has arrived.’ With regard to intervening adverbials, perfective éssere typically behaves in the same way as áere (i.e. phrasal expressions are excluded): (162) Frantziscu est vénmtu a domo mea. lit. ‘Francis has always/every day come to my house.’ However, some verbs appear to allow intervening phrasal adverbs much more freely, more or less as in the progressive: (163) Maria fit tottu su mandzanu abbarrata in lettu. lit. ‘Mary was all morning stayed in bed.’ ‘Mary stayed in bed all morning.’

< previous page

page_140

next page >

< previous page

page_141

next page >

Page 141 We suggest that in such cases, the participle functions as an adjective –note that fit abbarrata is interpreted here as a past imperfect rather than as a pluperfect. A similar contrast can be seen between genuine passives like (164a), which show the same restrictions as perfectives with áere, and cases where the past participle functions as an adjective, as in (164b): (164) a ??Sa janna fit dondzi die abberta dae su theraccu. lit. ‘The door was every day opened by the servant.’ b Sa janna fit tottu su mandzanu abberta. lit. ‘The door was all morning open.’ Perfective and progressive aspect can be combined as in (165): (165) So istatu travallande. ‘I have been working.’ Note, however, that in such cases, clitics corresponding to complements of the lexical verb cannot be attached to the first auxiliary, but only to the present participle: (166) a *Juanne los est istatu kircande. b Juanne est istatu kircándelos. ‘John has been looking for them.’ In terms of the analysis presented so far, (166b) must have the underlying structure (167), where the progressive element istatu functions as a main verb: (167) By the same token, to account for the ungrammaticality of (166a) we must exclude underlying structures where perfective and progressive auxiliaries are combined, as in (168): (168) *[S NP [VP AUX perf [VP AUX prog VP]]] Note, however, that this cannot be achieved by a general condition allowing only one auxiliary per clause since we do find clauses with two perfective auxiliaries in certain types of conditional constructions (see 6.2.3 for further discussion): (169) Si 1’aío áppitu fattu,… lit. ‘If I had had done it,…’ Moreover, as we shall see in 3.3.3, both perfective and progressive auxiliaries can occur with certain modal verbs which have auxiliary status according to the criteria proposed above. We suggest that the

< previous page

page_141

next page >

< previous page

page_142

next page >

Page 142 ungrammaticality of (166a) is attributable to the suppletive nature of the item istatu and that this item is perceived here as the past participle not of éssere, but of istare which, as we have noted, can only function as a main verb. 3.3.3 Modal verbs The following verbs, when followed by a ‘bare’ infinitive (not introduced by a or de), have the same properties as the auxiliaries áere and éssere outlined in 3.3.2: kérrere ‘want’, dévere ‘must’, pótere ‘can’ and iskire ‘know (how to)’. The range of adverbials which can occur between these verbs and the infinitive appears to be the same as that described in 3.3.2 for perfective and progressive auxiliaries: (170) Juanne

cantare.

[John sing] In contrast, most other infinitive-taking verbs can be followed by the full range of adverbial expressions in (170): e.g. Provo semper/dondzi die/como a cantare ‘I try always/every day/now to sing’. Clitics which represent complements of the infinitive are always attached to the modal verb, never to the infinitive itself: (171) a

Juanne lu b

fákere.

*Juanne ‘John wants, etc. to do it’

< previous page

page_142

next page >

< previous page

page_143

next page >

Page 143 Conversely, with more typical infinitive-taking verbs such as provare ‘try’, clitics must immediately precede the infinitive (unless, of course, they represent complements or modifiers of the main verb): (172) a *Juanne lu provat a fákere. b Juanne provat a lu fákere. ‘John tries to do it.’ There are a number of other properties which distinguish these verbs from other verbs which take an infinitival complement. Perhaps the most obvious feature of these verbs is that they take a ‘bare’ infinitive whereas most other infinitive-taking verbs require a ‘particle’, a or de (see 6.1.3 for further discussion) before the infinitive, as shown in the above examples. The only other verbs which take a ‘bare’ infinitive are the causative verbs fákere ‘make’ and lassare ‘let’, which share many other properties of auxiliary verbs but also have further syntactic properties which merit separate discussion (see 6.1.4). A second property of modal verbs is that the infinitive cannot be independently negated (to simplify presentation, we shall give examples with only one of the verbs under discussion, but the observations may be taken as valid for the rest of this class): (173) a *Keljo non rúghere. ‘I want not to fall.’ b Provo a non rúghere. ‘I try not to fall.’ When modal verbs are used in the perfect, the choice of auxiliary is determined by the dependent infinitive, whereas in other infinitival constructions it is determined by the verb which governs the infinitive (e.g. áere with provare, but éssere with resissire ‘succeed, manage’): (174) a Appo/*so póttitu ballare. ‘I was able to dance.’ b So/*appo póttitu andare. ‘I was able to go.’ (175) a Appo/*so provatu a ballare. ‘I tried to dance.’ b Appo/*so provatu a andare. ‘I tried to go.’ c So/*appo resissitu a ballare. ‘I managed to dance.’ d So/*appo resissitu a andare. ‘I managed to go.’

< previous page

page_143

next page >

< previous page

page_144

next page >

Page 144 In Middle constructions such as (176) where the infinitive is followed by its underlying direct object, this NP determines the person and number inflection of the modal verb: (176) Si deven cantare cussas cathones. [si must (3.pl.) sing those songs] ‘Those songs must/should be sung.’ Given the analysis proposed in 3.2.8, this observation suggests that the postverbal NP has assumed the status of subject of the sentence. Indeed, it can be placed in the canonical subject position to the left of the modal verb: (177) Cussas cathones si deven cantare. However, in similar constructions with verbs like provare, the verb always has third person singular inflection and the NP cannot be placed in the canonical subject position: (178) a Si provat/*provan a cantare cussas cathones. b *Cussas cathones si provan a cantare. ‘One tries/should try to sing those songs.’ To account for the data in (178), we suggest that the element which is promoted to subject status is the whole infinitival complement rather than the object of the infinitive (cussas cathones). Finally, infinitives dependent on modal verbs can be readily fronted by the process described in 1.1.2, as can participles in perfective and progressive constructions (see also 7.1.5 for detailed discussion): (179) a Cantare keres? ‘Do you want to sing?’ b Cantatu as? ‘Have you sung?’ c Cantande ses? ‘Are you singing?’ However, complement clauses (finite or infinitival) cannot normally be fronted: (180) a *A cantare provas? ‘Do you try to sing?’ b *Ki as cantatu credo. lit. ‘That you have sung I believe.’ The above observations strongly suggest that sentences with the

< previous page

page_144

next page >

< previous page

page_145

next page >

Page 145 modal verbs under discussion have a monoclausal structure and that these verbs have a status similar to that of the perfective and progressive auxiliaries áere and éssere. On the other hand, sentences with other infinitive-taking verbs such as provare warrant a biclausal analysis whereby the infinitive (and its complements) constitute a separate clause which functions as the complement of the main verb. Within certain limitations (which appear to be semantic), modal verbs can occur freely with each other and with perfective and progressive auxiliaries, as in the following examples (see also (174) above): (181) a Juanne devet iskire annatare. ‘John must know how to swim.’ b Maria potet kérrere issire. ‘Mary may want to go out.’ c Non bi so kerrende andare. lit. ‘I am not wanting to leave.’ d Jakeddu los potet éssere iscriende. ‘Jack may be writing them.’ e Issos bi den éssere kérfitos andare. ‘They must have wanted to go there.’ f Cussas cosas, non las dias áere kérfitas inténdere. ‘Those things, you must not have wanted to hear them.’ Note that in (181d) progressive éssere functions as a genuine auxiliary (since it allows clitic-climbing, though the alternative Jakeddu potet éssere iskriendelos is also acceptable), thus demonstrating that progressive éssere as a true auxiliary is not intrinsically incapable of co-occurring with other auxiliaries (see our discussion of examples (166) in 3.3.2). There do not appear to be any absolute restrictions on the order of modal and aspectual auxiliaries of the sort found in English. Typically, when a modal is followed by an aspectual auxiliary (the short forms of dévere being possible in this case, see 3.1.3, p. 90), it has an epistemic interpretation (expressing degrees of probability), whereas after an aspectual auxiliary a root interpretation is preferred (expressing notions such as volition, ability, permission or obligation): (182) a Frantziscu devet/det áere fraicatu sa domo. ‘Francis must/will have built the house.’ b Frantziscu at dévitu fraicare sa domo. ‘Francis had to build the house.’

< previous page

page_145

next page >

< previous page

page_146

next page >

Page 146 Nevertheless, this generalisation is not absolute. In (183) a ‘retrospective obligation’ interpretation (‘The students are obliged to have read this book’) is possible in addition to the epistemic interpretation (‘It seems certain that the students have read this book’): (183) Sos istudientes deven áere lessu custu libru. ‘The students must have read this book.’ Similarly, there is a tendency for the first of two modal auxiliaries to have an epistemic reading (as in (181b)), though (181a) also allows a root interpretation for dévere (i.e. ‘John has to learn how to swim (before we let him take the boat out)’). For this reason, the modals which can occur most readily before another auxiliary are those which allow an epistemic interpretation (dévere ‘must’ and pótere ‘can’, but not kérrere ‘want’ or iskire ‘know’), though kérrere can sometimes be followed by another auxiliary to express desire with respect to a completed action or a capacity: (184) a Keljo áere finitu cussu travallu. ‘I want to have finished this work.’ b Keljo iskire faeddare su campidanesu. ‘I want to be able to speak Campidanese.’ 3.3.4 The future use of áere As we noted in 3.1.3, áere ‘have’ can be used with an infinitive introduced by a to express future time. In this use, áere shows many of the properties which we have associated with auxiliary verbs in the preceding sections. In particular, in constructions of this type clitic-climbing is obligatory, the dependent infinitive cannot be negated and in Middle constructions the direct object of the infinitive must assume subject status, determining the agreement of áere: (185) a L’appo a fákere. b *Appo a lu fákere. ‘I will do it.’ (186) *Appo a non pippare prus. ‘I will not smoke any more.’ (187) Non si an/*at a cantare cussas cathones. ‘Those songs will not be sung.’ With regard to the class of adverbial expressions which can intervene between áere and the infinitive, the restrictions are even more

< previous page

page_146

next page >

< previous page

page_147

next page >

Page 147 stringent than those which we have observed with other auxiliaries. In particular, simple aspectual adverbs such as semper ‘always’ and floated quantifiers appear to be excluded from this position, in addition to time or manner adverbs and phrasal expressions: (188) a *Appo semper a travallare. ‘I will always work.’ b *Issos an tottu a cantare. ‘They will all sing.’ However, ‘focal’ adverbs are accepted more readily: (189) Non los appo mancu a salutare. ‘I will not even greet them.’ On the other hand, future áere resembles main verbs which take an infinitive in that it normally requires a particle, a, before the infinitive, though some speakers allow omission of a particularly (but not exclusively) before verbs which begin with a vowel App’andare ‘I will go’, An éssere arrivatos ‘They will have arrived’. Possibly this is a lowlevel contraction phenomenon, though it does not generally occur with other verbs which take an infinitive introduced by a: *Prov’ an-dare ‘I will try to go’. With regard to fronting of the infinitive, the situation is rather unclear. Spano (1840:86, fn. 1) states that this phenomenon (which he refers to as postposition of the auxiliary) is quite frequent with future áere, giving the examples “a bider l’hamus” ‘we shall see it’, “benner hat a inoke” ‘he/she will come here’. However, our informants generally found these and other examples, such as (190), decidedly odd: (190) ??A travallare as cras? ‘Will you work tomorrow?’ It is possible that Spano’s remarks relate to archaic or poetic usage since, in the same footnote, he mentions the use by certain poets of future forms such as “timerhápo” ‘I shall fear’, “fagherhát” ‘he will fear’, where áere is suffixed to the infinitival stem, forms which are clearly not current in modern Sardinian. Nevertheless, we do not dismiss the possibility that fronting of the infinitive with áere is possible for some speakers. Unlike modal, perfective and progressive auxiliaries (which can be combined in any order within certain limitations), future áere cannot be preceded by another auxiliary: (191) a *Potto áere a vínkere. ‘I may be going to win.’

< previous page

page_147

next page >

< previous page

page_148

next page >

Page 148 b *Appo áppitu a iscríere cussa líttera. ‘I have been going to write that letter.’ Consequently, the criterion of ‘transparency’ regarding choice of perfective auxiliary (see 3.3.3 examples (174)–(175)) cannot be applied to this item. The restrictions illustrated in (191) are not simply a question of ordering of auxiliaries since future áere cannot occur as an infinitive dependent on a main verb, nor can it occur in the subjunctive: (192) a *Ispero de áere a víkere. ‘I hope that I will win.’ b *Mancari appat a próere,… ‘Even if it will rain,…’ Moreover, as we noted in 3.1.3, future áere cannot be used in the imperfect with a conditional value in the Logudorese-Nourese dialects (though such a use is possible in Campidanese). One way of capturing the restrictions illustrated in (191) is to treat future áere as the realisation of a [+future] tense feature under the INFL node in the structure [S NP INFL VP]. According to this analysis, the fact that future áere cannot be preceded by another auxiliary follows directly from our assumption that other auxiliaries are adjoined to VP. Also, this analysis allows the apparently defective nature of future áere to be accounted for in terms of restrictions on combinations of tense and mood features under INFL. Thus, (192a) can be excluded on the grounds that tense features cannot occur with the feature [+infinitive], and (192b) can be ruled out by a restriction preventing the combination of [+future] and [+subjunctive]. This analysis also fits in with the historical evolution of synthetic future paradigms in other Romance languages where forms derived from Latin HABERE have been fully grammaticalised as tense markers suffixed to the infinitive. In the case of Sardinian, we may postulate that this process of grammaticalisation has also occurred to the extent that áere has been reduced to the status of a tense marker under INFL (i.e. occupying the same underlying position as tense affixes) but has retained its morphological status as a verb rather than a suffix. The facts concerning clitics and Middle constructions illustrated in (185) and (187) are consistent with this analysis since áere is part of the same simple clause as the dependent infinitive even though the structural relation between these elements is different from that which obtains with other auxiliaries. The restrictions on intervening adverbs (see examples (188)) can perhaps be characterised in terms

< previous page

page_148

next page >

< previous page

page_149

next page >

Page 149 of a condition requiring adjacency between INFL and the first verb under VP, though it would need to be relaxed to accommodate examples like (189). Evidence in support of such a condition is presented in 3.4.1. In the case of synthetic tense forms, this condition can plausibly be construed as a constraint on the morphological process whereby tense markers under INFL are suffixed to the verb. Although this process does not apply to áere, the possibility of contracted forms such as App’andare (see p. 147) may be symptomatic of cliticisation of áere+a to the infinitive at the phonetic level. To the extent that this adjacency condition is justified even when affixation does not occur, it will also prevent fronting and indeed negation of the infinitive, as in (190) and (186). 3.3.5 ‘Semi-auxiliaries’ There are a number of verbs of an aspectual type, principally cumint-zare ‘begin’, accabbare/finire ‘finish’ and sikire ‘continue’ (lit. ‘follow’), which can function as main verbs taking an infinitival complement clause but also optionally show some of the properties associated with auxiliary verbs. With these verbs, cliticclimbing is optional: (193) a L’appo b

fákere.

Appo lu fákere. lit. ‘I have begun/finished/continued to do it.’ When clitic-climbing occurs, we find the tighter restrictions on intervening adverbials which we have seen to be characteristic of auxiliary constructions: (194) a L appo b

a fákere.

Appo cumintzatu a lu fákere. lit. ‘I have begun many times/now to do it.’ Negation of the dependent infinitive is often infelicitous, perhaps for pragmatic reasons, but plausible examples can be constructed (e.g.

< previous page

page_149

next page >

< previous page

page_150

next page >

Page 150 Cumintzo a non pippare prus ‘I begin not to smoke any more’). However, in such cases, clitic-climbing is blocked: (195) a *Lu cumintzo a non fákere prus. b Cumintzo a non lu fákere prus. ‘I begin not to do it any more.’ In Middle constructions, non-agreement with the underlying object of the dependent infinitive appears to be the preferred option (as in (196a)), but agreement (implying subjectivisation of the object NP) is also possible, particularly if this NP is moved into the preverbal subject position: (196) a Non si cumintzat a fákere cussas cosas. b ?Non si cumintzan a fákere cussas cosas. c Cussas cosas non si cumintzan a fákere. ‘One should not begin to do those things.’ The above evidence suggests that these verbs can function both as main verbs and auxiliaries. However, with respect to other criteria, they behave exclusively like main verbs which take an infinitival complement clause. First, and most obviously, they always require a particle (a or de) before the dependent infinitive, unlike the modal auxiliaries discussed in 3.3.3. Second, the infinitive cannot be fronted, like clausal complements of verbs such as provare ‘try’: (197) *A travallare cumintzas? ‘Will you begin to work?’ Finally, they do not display the ‘transparency’ with respect to choice of perfective auxiliary which we observed with modal auxiliaries, but always take áere: (198)

cumintzatu a ‘I have begun to sing/go.’ Interestingly, when the dependent infinitive is an unaccusative (éssere-taking) verb and the aspectual verb occurs in the perfective, clitic-climbing is not permitted: (199) *Bi cumintzatu a andare. ‘I have begun to go there.’ The evidence presented above suggests that these verbs are essen-

< previous page

page_150

next page >

< previous page

page_151

next page >

Page 151 tially main verbs which take an infinitival S' complement but can be reanalysed as auxiliaries, on the lines proposed by Rizzi (1986: ch. 1) for Italian (though we assume a different derived structure for reasons to be discussed in 3.4.1), thus permitting clitic-climbing and, in Middle constructions, subjectivisation of the underlying object of the infinitive. This reanalysis rule can be formulated approximately as in (200), leaving open the derived position of the particle a or de: (200) …[VP [S′ alde [S PRO INFL VP]]]… →…[VP V alde VP]… The facts in (194)–(195) can be accounted for by postulating that reanalysis is blocked by the presence of elements which cannot occur between ‘true’ auxiliaries and the dependent infinitive or participle. The data in (198)–(199) suggests that reanalysis is also blocked if it would entail a change in perfective auxiliary (i.e. Appo cumintzatu a andare is not a reanalysed construction, thus precluding clitic-climbing in (199)). This might be explained by assuming that the surface subject is also the underlying subject of the aspectual verb (i.e. these verbs are of the ‘control’ type rather than the ‘subjectraising’ type—a plausible assumption in so far as there are no clear instances of subject-raising with infinitives in Sardinian, see 6.1.3). Consequently, éssere cannot be selected since subjectivisation has not occurred, but the structure derived by reanalysis is one which requires éssere with genuine modal auxiliaries. It is not clear why fronting of the infinitive is excluded. We tentatively suggest a principle to the effect that reanalysed constituents must retain their syntactic cohesion, at least as far as the structural relation between the two verbs is concerned. A slightly different picture emerges with the verb torrare ‘return’. When followed by an infinitive (introduced by a) this verb can have either a movement interpretation with a purposive interpretation for the infinitive (‘come/go back to do something’) or an iterative interpretatiori (‘do something again’): (201) Torro a léghere cussu libru. ‘I am coming/going back to read that book.’ ‘I will read that book again.’ By and large, this verb functions as a main verb with the movement interpretation but shows auxiliary-like properties when it has the iterative reading. Thus, the two meanings of (201) are distinguished according to the position of the clitic in (202):

< previous page

page_151

next page >

< previous page

page_152

next page >

Page 152 (202)

a Torro a lu léghere. ‘I am going back to read it.’ b Lu torro a léghere. ‘I will read it again.’ Similarly, the presence of an adverbial which cannot intervene between an auxiliary and the dependent verb imposes a movement interpretation, as in (203a), but simple aspectual adverbs do not eliminate the ambiguity, as in (203b): (203) a Torro a travallare. ‘I (will) return every day/tomorrow to work.’ b Non torro mai a traallare. ‘I never return to work.’ ‘I will never work again.’ Moreover, clitic-climbing is incompatible with intervening adverbs of the type in (203a): (204) *Lu torro a léghere. lit. ‘I return every day/tomorrow to read it.’ With the movement interpretation, torrare is an unaccusative verb (taking éssere as its perfective auxiliary) but with the iterative reading it shows the same transparency with regard to auxiliary choice as modal auxiliaries (e.g. it takes áere if the following infinitive is not unaccusative): (205) a So torratu a léghere cussu libru. ‘I came/went back to read that book.’ b Appo torratu a léghere cussu libru. ‘I read that book again.’ There is a strict correlation between choice of auxiliary and the position of clitics, as shown in (206) where the interpretations are the same as in (205), with lu replacing cussu libru: (206) a b

torratu a lu léghere.

Lu torratu a léghere. Similarly, in Middle constructions, torrare may agree with the post-

< previous page

page_152

next page >

< previous page

page_153

next page >

Page 153 infinitival NP (as in (207)) or not (as in (208)) according to the pragmatically preferred interpretation: (207) Non si torran a fákere cussas cosas. ‘You must not do those things again.’ (208) Si torrat a salutare sos cumpandzos prima de sink’andare. ‘One should return to greet one’s friends before going away.’ Negation of the infinitive appears to be impossible under either reading: (209) *Torro a non pippare.’ lit. ‘I return not to smoke.’ The movement interpretation of (209) (‘I (will) come back not to smoke’) is presumably precluded for pragmatic reasons whereas the more plausible reading ‘I will not smoke again’ is ruled out in so far as it would require torrare to be treated as an auxiliary contrary to our observation that auxiliaries do not allow negation of the dependent infinitive. The facts regarding fronting of the infinitive are somewhat unexpected in that examples like (210) appear to allow (indeed prefer) the movement interpretation: (210) A travallare torras? ‘Are you going back to work?’ This may be related to the fact that purposive complements of other movement verbs, such as andare ‘go’, which do not display other properties of auxiliary verbs, can be fronted (unlike clausal complements of more typical infinitive-taking verbs such as provare ‘try’): A travallare andas? ‘Are you going to work?’ The position of the clitic in examples like (211) (attached to the infinitive rather than to torrare ) shows that torrare has main verb status in fronted constructions: (211) a A mi salutare torras? b *A salutare mi torras? ‘Will you return to greet me?’ Thus, for some reason, fronting of the infinitive appears to be impossible when torrare has auxiliary status, a problem which we have already raised in connection with verbs of the cumintzare type.

< previous page

page_153

next page >

< previous page

page_154

next page >

Page 154 3.4 STRUCTURE OF THE VP 3.4.1 Auxiliaries and preverbal modifiers In 3.3.2 we proposed that perfective and progressive auxiliaries are adjoined to the VP as in (212), an analysis which we have assumed for all other auxiliaries except future áere. (212) [VP V aux [VP V…]] This analysis is motivated by the following considerations. First, the absence of any systematic restrictions on the combination and order of auxiliaries suggests that these items are introduced by a recursive system. Second, the fact that the dependent infinitive or participate and its accompanying elements can be fronted together indicates that they form a constituent, thus precluding an analysis of the type in (213), advocated by Emonds (1978) for French and Rizzi (1986: ch. 1) for Italian, where the auxiliary and lexical verb form a verbal complex (represented here as V[#]): (213) [VP [V[#] Vaux V]…] Third, the facts concerning the position of clitics, agreement of the verb in Middle constructions, transparency with respect to choice of perfective auxiliary and the impossibility of negation of the dependent verb suggest that sentences containing auxiliary verbs have a monoclausal structure (i.e. the dependent verb and its complements and modifiers do not constitute a separate clause). In the preceding sections we noted that certain types of adverbs can intervene between auxiliary verbs and the dependent participle or infinitive, notably aspectual adverbs such as semper ‘always’, ‘focal’ adverbs like fintzas ‘even’ and the floated quantifier tottu ‘all’: (214) a An ballatu. ‘They have always/even/all danced.’ b Poten ballare. ‘They can always/even/all dance.’

< previous page

page_154

next page >

< previous page

page_155

next page >

Page 155 c Cumintzan a ballare. lit. ‘They begin always/even/all to dance.’ In examples involving more than one auxiliary, adverbs and quantifiers of this type can occur relatively freely between auxiliaries and between an auxiliary and a lexical verb (sometimes with a difference in semantic scope which may affect acceptability in some cases). A representative selection of possible combinations is given in (215): (215) a Juanne at semper kérfitu ballare. Juanne at kérfitu semper ballare. ‘John always wanted to dance.’ b Maria at fintzas cumintzatu a cantare. Maria at cumintzatu fintzas a cantare. ‘Mary even started to sing.’ c Issos deven tottu éssere arrivatos. Issos deven éssere tottu arrivatos. ‘They must have all arrived.’ This freedom of order can be captured by postulating that adverbs of this type (designated mnemonically as ADV [+F]) are also adjoined to the VP as in (216): (216) [VP ADV[+F] VP] Assuming that (216) can interact freely with (212), we can generate combinations of auxiliary verbs and adverbs of the type in question in any order, as in (215). The hypothesis that the adverb or floated quantifier forms a constituent with the following verb and its complements is borne out by fronted constructions such as the following: (217) a Semper kérfitu ballare an. lit. ‘Always wanted to dance they have.’ b Fintzas cumintzatu a cantare an. lit. ‘Even started to sing they have.’ c Tottu arrivatos deven éssere. lit. ‘All arrived they must be.’ Indeed, examples like the following show that these adverbs can be adjoined to other maximal projections which originate in the postverbal position, though in some cases tottu is more readily interpreted as quantifying the following NP or as a degree modifier of the following adjective rather than as a floated quantifier (e.g. Tottu su pane appo

< previous page

page_155

next page >

< previous page

page_156

next page >

Page 156 mandicatu ‘I have eaten all the bread’, Tott’ istrakku est ‘He is completely tired’): (218) a Semper a Núgoro andas. lit. ‘Always to Nuoro you go.’ b Fintzas prus abbistu de Juanne est. lit. ‘Even more clever than John he is.’ c Tottu sa mattessi cosa an comporatu. lit. ‘All the same thing they have bought.’ Thus, the schema in (216) can be seen as a special case of the more general schema in (219): (219) [XP ADV[+F] XP] In principle, (216) and (219) will also allow adverbs of the type in question to be adjoined to the highest VP in a clause, a possibility which we must exclude in view of examples such as the following: (220) a *Juanne semper ballat. ‘John always dances.’ b *Maria fintzas at cantatu.’ ‘Mary even has sung.’ c *Issos tottu travallan. ‘They all work.’ This can be achieved by means of a condition (suggested in 3.3.4) which requires the first verb within the VP to be adjacent to the INFL node, thus excluding structures of the type in (221): (221) *[S NP INFL [VP ADV [+F] [VP V…]]] In the case of finite sentences such as those in (220), this restriction can plausibly be construed as a requirement that tense and agreement features generated under INFL must be adjacent to the verb to which they are ultimately affixed, though we suggested in 3.3.4 that this restriction also holds in cases involving the future marker áere which is not affixed to the verb. It also appears to hold in genuine infinitival complement clauses, where the INFL node arguably contains no features, in so far as adverbs like semper cannot immediately precede the infinitival verb (e.g. *Provo a semper travallare ‘I try to always work’) though this may be due to a more general restriction which prevents any item other than clitics and the negative particle non from occurring between the complementiser a or de and the infinitive. Given the schema in (212), the types of adverbs which cannot

< previous page

page_156

next page >

< previous page

page_157

next page >

Page 157 intervene between an auxiliary and the following verb can be characterised as adverbs which cannot be adjoined to the VP as in (216). In the case of phrasal expressions such as dondzi die ‘every day’ and tottu su mandzanu ‘all morning’, let us assume that adjunction to VP is restricted to single lexical items. Manner adverbs such as lestru ‘quickly’ typically follow the verb which they modify (see 3.4.2) in much the same way as most adnominal adjectives follow the noun. The remaining types of adverbs which are excluded from the position indicated in (216) (e.g. time adverbs such as como ‘now’ and eris ‘yesterday’) can plausibly be characterised as modifiers of the sentence rather than of the verb and thus occur as constituents of S rather than VP (typically at the beginning or end of the sentence), though (as we shall see in 3.4.2) they can occur postverbally between elements which ostensibly form part of the VP. Similarly, our observations concerning negation can be accounted for by assuming that the negative particle non is a constituent of S which immediately precedes the INFL node (or possibly forms part of the INFL system). The structural analysis proposed in (212) and (216) is not the only one which is compatible with the observed facts. In particular, we may note that this account does not make use of the Specifier position within VP (i.e. [VP SPEC V′]) which is made available by the general X-bar framework, see 0.2. Consequently, it might be maintained that auxiliaries and the relevant adverbs are adjoined to V′ rather than to VP. The apparent absence of a Specifier of VP might be attributed in part to the condition requiring adjacency between INFL and the first verb, which entails that an element with Specifier status could only occur before a verb which is dependent on an auxiliary. However, since this is exactly the distribution which we have observed for aspectual and ‘focal’ adverbs, it would be possible to analyse these adverbs as Specifiers of VP provided that we also treat auxiliaries as verbs which take a VP complement rather than simply as adjuncts to the VP. Thus, the recursive pattern of such elements could be represented as in (222): (222) [VP [V′ Vaux [VP ADV[+F] [V′ Vaux VP]]]] Note, however, that the alternative analyses envisaged above fail to capture our observation that the adverbs which can intervene between an auxiliary coincide with the class of adverbs which can be adjoined to other maximal projections—i.e. they do not allow the distribution of these adverbs to be subsumed under the general schema proposed in (219).

< previous page

page_157

next page >

< previous page

page_158

next page >

Page 158 3.4.2 Complements and postverbal modifiers The order of postverbal elements within the VP is fairly flexible. At a very general level, the principles proposed in 2.2.3 regarding the order of postnominal elements apply also to the VP: (223)a Syntactically simple items tend to precede more complex constituents. b Elements which bear a close relation to the verb tend to precede those which are more loosely related to the verb. A further point to bear in mind is that items which occur in clause-final position (hence, typically, VP-final position) are usually interpreted as the focus of the sentence (conveying the information which is deemed to be most salient). The unmarked order of complements is: direct object-indirect object (PP)-complement clause (S′). For the sake of concreteness, let us assume that direct object NPs are generated as sisters of the V under the V′ node, whereas other complements are adjoined to the V′ as in (224):

(224) Note that the preferred order PP-S′ is not stipulated in (223), but can plausibly be attributed to the general principle (223a) above. We also assume that marked cases such as (225) where a particularly ‘heavy’ direct object follows an S′ or PP are the result of a stylistic reordering rule and do not require us to postulate an alternative structure to (224): (225) a Appo postu in su calassu tottu k’aías lassatu in mesa. ‘I put in the drawer everything you had left on the table.’ b Appo cumbintu de andare tottu sa dzente ki connoskío. ‘I persuaded to go all the people I knew.’ On the basis of the structure in (224), we may envisage three possible positions in which postverbal modifiers may occur, represented by X in (226): (226) a [V′ [V X (NP)]

< previous page

page_158

next page >

< previous page

page_159

next page >

Page 159 b [V′ V′ X] c [VPVP X] Case (226a) defines the position of elements which immediately follow the verb, before any complements including the direct object. Case (226b) characterises items which occur after the direct object but which may precede or follow other complements, whereas (226c) is appropriate to elements which follow all complements. The structural distinctions in (226) are intended mainly as a heuristic framework within which to express observations regarding the order of postverbal elements. We make no claim that adverbs or other expressions are strictly subcategorised in terms of these structures. It is possible that the ordering properties of at least some items can be accounted for directly in terms of general principles of the type mentioned at the beginning of this section without reference to syntactic structure. Position (a) in (226) is not generally available to phrasal expressions, possibly as a consequence of principles (223a) and (223b) above. The items which can occur most readily between the verb and the direct object are aspectual adverbs of the semper ‘always’ type and degree modifiers such as meta ‘a lot’, pacu ‘not much’ and troppu ‘too much’: (227) Cantan cussa cathone. lit. ‘They sing always/a lot/too much that song.’ As we noted in 3.4.1, adverbs of the semper type are (or can be) adjoined to the following constituent, as shown by fronted constructions such as (228): (228) Semper cussa cathone cantan. lit. ‘Always that song they sing.’ Consequently, the version of (227) with semper is not necessarily an instance of the structure in (226a). However, the analysis-in (226a) is appropriate for degree modifiers, which do not show evidence of adjunction to the following element: (229) * cussa cathone cantan. Note also that degree modifiers, unlike aspectual adverbs, do not readily precede the verb which they modify:

< previous page

page_159

next page >

< previous page

page_160

next page >

Page 160 (230) *An cantatu cussa cathone. Cases where an aspectual adverb occurs between complements of the verb can also be analysed in terms of adjunction to the following phrase, as in (231), rather than in in terms of the structure (226b): (231) Sa dzente offrit su vinu [PP semper [PP a sos óspites]]. ‘People offer wine always to the guests.’ However, aspectual adverbs can also occur in final position, particularly when the remainder of the VP consists solely of a finite verb: (232) a Juanne cantat semper. ‘John sings always.’ b Cantan cussa cathone semper. ‘They sing that song always.’ c Pessamus a cussas cosas semper. ‘We think of those things always.’ Such examples can be accommodated by postulating that aspectual adverbs can be adjoined to the right of VP (i.e. as in (226c)) as well as to the left. Examples like (232b, c) are somewhat less natural than their counterparts with semper preceding the complement (e.g. as in (227)), an observation which suggests that adjunction to the left of a complement is the preferred option whenever it is available. Degree modifiers occur still less readily after complements of the verb, but to the extent that examples like those in (233) are acceptable, we postulate that such modifiers can marginally occupy the position indicated in (226b): (233) a ?Cantan cussa cathone meta. ‘They sing that song a lot.’ b ?Pessamus a cussas cosas meta. ‘We think about those things a lot.’ ‘Focal’ adverbs such as fintzas ‘even’ have a similar distribution to aspectual adverbs (as we noted in 3.4.1 in connection with their occurrence in preverbal position). With these adverbs, the analysis in terms of adjunction to the following constituent is supported by semantic considerations in that the following constituent is clearly modified by the adverb; for example, fintzas gives a contrastive interpretation to su vinu in (234a) but to a sos pitzinneddos in (234b):

< previous page

page_160

next page >

< previous page

page_161

next page >

Page 161 (234) a Juanne dat fintzas su vinu a sos pitzinneddos. ‘John gives even wine to the little boys.’ b Juanne dat su vinu fintzas a sos pitzinneddos. ‘John gives wine even to the little boys.’ Like aspectual adverbs, ‘focal’ adverbs can be adjoined to the preceding VP, as in (235): (235) Juanne ballat fintzas. ‘John dances even.’ A slight complication with these adverbs is that they can, somewhat marginally, modify a preceding phrase other than a VP, as in Juanne fintzas ballat where fintzas gives a contrastive reading to Juanne (which must bear primary stress in this case) rather than to ballat recall that adjunction of fintzas to the following VP is excluded by condition (221) proposed in 3.4.1. Thus, in cases like (236) it is not clear whether the adverb modifies the whole VP or just the complement: (236) Maria iscríet a sos cumpandzos fintzas. lit. ‘Mary writes to her friends even.’ However, if the adverb is placed between the verb and the complement, it clearly modifies the latter: (237) Maria iscríet fintzas a sos cumpandzos. The most natural way of placing the verb within the scope of the adverb in such cases is to right-dislocate the complement as in (238): (238) Maria lis iscríet fintzas, a sos cumpandzos. Assuming that dislocated phrases are placed outside the VP (see 7.1.2), we may conclude that adverbs of this type can be adjoined to the right of any maximal projection, as in (239), but not to lower projections such as the V or V′ iscríet in (237). (239) [XP XP ADV focal] The preferred position for manner adverbs is that defined in (226b) (adjoined to V′), though they can also occur (less readily) between the verb and the direct object. Thus, (240a) is rather more natural than (240b), whereas the sentences in (241) are both completely acceptable (though with a different choice of focus: cuitande in (241a), a domo in (241b)): (240) a Pretu at sonatu sa kitterra abbellu. b Pretu at sonatu abbellu sa kitterra. ‘Peter played the guitar quietly.’

< previous page

page_161

next page >

< previous page

page_162

next page >

Page 162 (241)

a So cúrritu a domo cuitande. b So cúrritu cuitande a domo. ‘I ran home quickly.’ When the adverbial consists of a phrase (e.g. a PP or a modified adverb), it tends to favour a more peripheral position in the VP (perhaps as a result of the principles proposed in (223)), but placement of such items is noticeably less acceptable before a direct object than before an indirect object: (242) a ??Pretu at sonatu ‘Peter sang very quietly/with pleasure the guitar.’ b So cúrritu a domo. ‘I ran very quickly/with joy home.’ Given the structure proposed in (224) cúrritu in (242b) can be analysed as a V′ whereas sonatu in (242a) must be a V. Consequently, we postulate that phrasal manner adverbials must be adjoined to the V′ whereas lexical manner adverbs can occur as sisters of V. Phrasal aspectual adverbs such as tottu su mandzanu ‘all morning’ and dondzi die ‘every day’, show the same pattern: (243) a ??Pretu at sonatu tottu su mandzanu sa kitterra. ‘Peter played all morning the guitar.’ b So cúrritu dondzi die a domo. ‘I ran every day home.’ The items bene ‘well’ and male ‘badly’ differ from more typical manner adverbs in that they favour the immediate postverbal position even when a direct object is present, like the degree modifiers in (227) and (233): (244) a Pretu sonat bene/male sa kitterra. b ?Pretu sonat sa kitterra bene/malu. ‘Peter plays the guitar well/badly.’ Note that these items do not show the phenomenon of adjunction to the following complement which we have observed with aspectual adverbs such as semper: (245) * sa kitterra sonat. Consequently, the preferred position of these items can be defined in terms of the schema (226a).

< previous page

page_162

next page >

< previous page

page_163

next page >

Page 163 In 3.4.1 we proposed that time adverbs such as como ‘now’ and eris ‘yesterday’ should be analysed as sentence modifiers dominated by the S node. However, they can also occur between constituents of the VP. Placement of such items before a direct object is generally infelicitous, but they can occur before other complements: (246) a ??Juanne at cantatu eris cussa cathone. lit. ‘John sang yesterday that song.’ b Semus andatos eris a Núgoro. ‘We went yesterday to Nuoro.’ c Appo provatu como a ti telefonare. ‘I tried (Just) now to telephone you.’ d Appo datu sas craes eris a Lukia. lit. ‘I gave the keys yesterday to Lucy.’ These judgements are consistent with the hypothesis that such adverbs can be adjoined to V′. In terms of the positions defined in (226), our observations concerning the preferred distribution of adverbial expressions can be summarised as follows. Position (226a) (sister of V) appears to be restricted to lexical adverbs which modify the verb, principally degree modifiers and bene ‘well’ and male ‘badly’ (less typically other manner adverbs). Position (226b) (adjoined to V′) is the favoured position for manner adverbs and phrasal expressions which modify the verb, but is also available for time adverbs such as como ‘now’ and eris ‘yesterday’. With regard to position (226c) (adjoined to VP), the situation is rather more complicated in so far as this position is not readily distinguishable from that defined in (226b). The principal candidates for adverbs which occur in this position are the aspectual and ‘focal adverbs (e.g. semper ‘always’ and fintzas ‘even’). In examples where these items occur between constituents of the VP, we have suggested that they are adjoined to the following constituent, though in the case of aspectual adverbs we do not dismiss the possibility that they can also occur in the structural positions defined in (226a) or (226b). The conclusions reached above are broadly in line with the analysis proposed for NPs in 2.2.1 in so far as the typical position of modifiers can be defined in terms of adjunction to the X′ node. Moreover, if we abstract away from linear order with respect to the head, we can discern some similarity between those items which can occur as sisters of V and those which occur (prenominally) as sisters of N. In both cases, this position can only be occupied by single lexical items (though this restriction appears to hold more rigorously with

< previous page

page_163

next page >

< previous page

page_164

next page >

Page 164 prenominal adjectives). Also, although the semantic range of adverbs which can occur as sisters of V is more open-ended than the class of prenominal adjectives, we may note that the manner adverbs which occur most typically in this position (bene ‘well’ and male ‘badly’) have an ‘affective’ quality similar to that of attributive prenominal adjectives. Finally, the class of adverbs which appear to be adjoined to the right of VP also show a significant cross-categorial behaviour in that we have independent evidence that they can be adjoined to other maximal projections (though usually to the left).

< previous page

page_164

next page >

< previous page

page_165

next page >

Page 165 4 Minor predicate categories 4.2 DEGREES OF COMPARISON 4.1 ADJECTIVES AND ADVERBS 4.1.1 Morphology As we noted in section 2.1.4, there are two inflectional classes of adjectives: those of the type mortu, -a, os, -as ‘dead’, which show an overt gender distinction, and those like forte, -es ‘strong’, which vary only according to number. The first class is by far the more numerous. There are a handful of adjectives which can be assimilated to the first class except that their masculine singular ending is -o; notably grogo ‘yellow’, which apparently retains the ending -o even in the feminine in some dialects (e.g. Nuoro; Pittau 1972:70), and a few pejorative adjectives such as the following forms cited in Pittau (1972:70): bolloroddo ‘inflated, fat’, buluffo ‘cretinous’, maccocco ‘crazy’. We may also note the following invariable adjectives: mattessi ‘same’, paris ‘flat, level, even’ and the comparative/superlative forms medzus ‘better, best’ and pejus ‘worse, worst’. Adverbs do not show any inflectional morphology. The diminutive suffixes -eddu and -ittu, whose adnominal use was discussed in section 2.1.1, can be added to certain adjectives and adverbs. The quantitative effect of the diminutive is at its most transparent with items which can be glossed roughly by a formula of the type ‘…to a small extent’, where addition of the diminutive suffix yields the meaning ‘…to a very small extent’: minore, minoreddu ‘(very) small’, accurtzu, accurtzeddu ‘(very) near’, abbellu, abbelleddu ‘(very) slowly, gently, quietly’. Similarly, the item adziccu ‘rather’ which can function as a degree modifier with adjectives and adverbs has the diminutive form adzikkeddu which translates roughly as ‘just a little bit’: adzikkeddu ránkidu ‘just a little bit sour’. Another case is mannittu ‘big (e.g. of a child)’, from mannu

< previous page

page_165

next page >

< previous page

page_166

next page >

Page 166 ‘large’: e.g. Como ki ses mannittu, ti potes vestire solu solu ‘Now that you are (a) big (boy), you can get dressed all on your own’. In some cases, the diminutive can be iterated (always by means of the regular suffix -eddu, -edda, etc.), as in minoreddeddu and even minoreddeddeddu ‘(very) very very small’, also mannitteddu which is typically used with reference to children who are clearly in early infancy. With adjectives and adverbs (and to a lesser extent with nouns), the use of the diminutive often carries connotations of endearment and is frequently used when addressing children. With this function, the diminutive can often be used with adjectives which do not directly involve size or quantity; e.g. bellu, belleddu ‘beautiful, pretty’; póveru, poverittu/miskinu, miskineddu ‘poor’ (in the sense of ‘to be pitied’). The use of an intensifying suffix -íssimu (-a, -os, -as) with adjectives is sometimes encountered (e.g. bellíssimu ‘very beautiful), though it is regarded as an Italianism by many grammarians and tends to be avoided by more conservative speakers. From a historical point of view, it is clear that such forms are borrowed from Italian -issimo, etc., not descended directly from the Latin superlative -ISSIMUS. The innovation is not particularly recent, however; such forms are cited without any reservations by Spano (1840:46) and by Porru (1811:10–11). Also, the person/number endings of this suffix have been assimilated to Sardinian inflectional morphology (cf. the plural forms -íssimos, -íssimas as opposed to Italian -issimi, issime ) and are occasionally found attached to stems which do not have Italian cognates; e.g. manníssimu ‘very big’ (cf. mannu ‘big’, from Latin), baratíssimu ‘very cheap’ (cf. baratu ‘cheap’, from Catalan). Nevertheless, the use of this suffix is uncharacteristic of Sardinian. The intensifying effect is more usually achieved by means of the degree adverb meta ‘very’ or, even more characteristically, by iteration of the adjective or adverb: ruju ruju ‘very red’, abell’ abbellu ‘very gently, slowly, quietly’, kitho kitho ‘very early’, etc. Some adjectives can also be intensified by means of the adjective mannu ‘big’; e.g. Sas pitzinnas sun cuntentas mannas lit. ‘the girls are happy(f. pl.) great(f. pl.)’ ‘the girls are very happy’ (note that mannu always follows the adjective which it modifies and agrees with it in number and gender). Comparative degree of adjectives and adverbs is likewise expressed analytically, by means of the modifier prus ‘more’ (see section 4.2.1), with the exception of the forms medzus and pejus, which function as comparative/superlative forms of bonu ‘good’, bene ‘well’ and malu ‘bad’, male ‘badly’ respectively.1 There is no productive morphological process for forming adverbs

< previous page

page_166

next page >

< previous page

page_167

next page >

Page 167 from other parts of speech. The use of -mente as a suffix which forms adverbs from adjectives (e.g. fatzilmente ‘easily’) is more or less on a par with the case of -íssimu discussed above; that is, although it is gaining some currency through Italian influence, it is not characteristic of authentic Sardinian.2 A consequence of the lack of a productive adverb-formation process is that adverbs in Sardinian comprise a fairly restricted class as compared with English or the other Romance languages—with the result that Sardinian relies more heavily on periphrastic constructions such as prepositional phrases for the expression of notions such as manner or attitude. Iteration, mentioned above as an intensifying device, is also exploited as a means of creating adverbial expressions from other parts of speech: die die ‘all day’ (lit. ‘day day’), notte notte ‘all night’ (lit. ‘night night’), riu riu ‘along the river’ (lit. ‘river river’), curre curre/currende currende ‘in a hurry’ (from cúrrere ‘to run'). There are also a few items which function as both adjectives and adverbs; e.g. forte ‘strong(ly), loud(ly)’, lestru ‘quick(ly)’. Adjectives can be formed from other categories in a variety of ways. Particularly common is the use of the past participle as adjective: abbertu ‘open’ (from abbérrere ‘to open’), arrenegatu ‘angry’ (from arrenegare ‘to anger’), mortu ‘dead’ (from mórrere ‘to die’). Present participles are not normally used as adjectives, though some function as adverbs (e.g. cuitande ‘quickly’, from cuitare ‘to hurry’). There is no deverbal suffix corresponding to English -able (except, once again, for Italian calques)—in such cases Sardinian typically resorts to a pronominal verb construction: ki si potet mandicare ‘which can be eaten’ (i.e. ‘edible’). A fairly common denominal adjective suffix is -osu (occasionally attached to stems other than nouns): pilosu ‘hairy’ (from pilu ‘hair’), fracosu ‘smelly’ (from fracu ‘smell’), putzinosu ‘disgusting’ (from putzi, an interjection expressing disgust)—for other less productive adjectival suffixes, see Wagner (1951:339–64). Also noteworthy in this connection are compound adjectives of the type pilicanu ‘grey-haired’, ocritortu ‘crosseyed’, composed of a body-part noun stem with the desinence -i followed by an adjective (see 2.1.1). Note that, though semantically the adjectival element modifies the body-part, it agrees with the possessor of the body-part: una fémina ocritorta ‘a cross-eyed woman’, féminas ocritortas ‘cross-eyed women’ (ocru ‘eye’ is masculine). This suggests that such compounds have a morphological structure of the form [[[noun stem+i] adj. stem] inflection], where the agreement inflection is external to the combination of noun and adjective stems; e.g. [[[ocr+i] tort- ] -as]. Compound adjectives of this type are extremely

< previous page

page_167

next page >

< previous page

page_168

next page >

Page 168 common and can be used to describe transient, incidental properties as well as characteristic physionomic features: So tottu piliffustu ‘My hair is all wet’ (lit. ‘I am all wet-haired.’). 4.1.2 Functional types In this section we shall review the basic subtypes of adjectives and adverbs, classified loosely according to their semantic function, with brief comments on their distribution and other syntactic properties. The different types of adjective were discussed in 2.1.4. To summarise, we can make a basic distinction between attributive adjectives, which describe some property or quality of the entity in question, and a class of specifier-like adjectives (essentially numerals and items expressing identity or non-identity; e.g. mattessi ‘same’, átteru ‘other’). Generally, attributive adjectives can occur adnominally (after the noun) or predicatively. However, there are a few adjectives which can convey an ‘affective’ attitude on the part of the speaker and, with this reading, cannot be used predicatively, must immediately precede the noun and cannot be modified in any way. Specifier adjectives also precede the noun, before any affective adjectives which may be present. A further class of items which can plausibly be classed as adjectives are the possessive items (such as meu ‘my’) discussed in 2.2.3, which show the same distribution (postnominal and predicative) and the same agreement phenomena as attributive adjectives. Other items sometimes classified as adjectives (e.g. demonstratives and quantifiers) are treated here as determiners on distributional grounds (cf. 2.1.2; 2.2.1; 2.2.2). The traditional category ‘adverb’ is notoriously heterogeneous, embracing virtually every type of modifier other than adjectives. If we take ‘adverb’ to mean ‘modifier of the verb’, perhaps the clearest cases are items such as those in (1), which we shall refer to as ‘manner’ adverbs:3 (1) bene ‘well’, abbellu ‘gently, quietly, slowly’, forte ‘hard, loudly’, cuitande/lestru ‘quickly’, appostu ‘on purpose’, debadas ‘in vain’, gai ‘that way’, goi ‘this way’. These adverbs modify the verb in much the same way as attributive adjectives modify the noun and also show a striking similarity with regard to their position. They almost always follow the verb, their position relative to other complements and adjuncts within the VP being determined by essentially the same principles as those which govern the order of postnominal elements (see 3.4.2).

< previous page

page_168

next page >

< previous page

page_169

next page >

Page 169 A second class of items traditionally categorised as adverbs, and which we shall refer to heuristically as ‘modal’ adverbs, is illustrated in (2): (2) fortzis ‘perhaps’, tzertu ‘certainly’, abbisumeu ‘in my opinion’, diffattis/infattis ‘in fact’, antzis ‘indeed’, peró ‘however’, duncas ‘therefore’, imbetzes ‘instead’, tando ‘consequently’.4 Intuitively, these items modify the sentence as a whole (rather than simply the verb), expressing degrees of doubt or certainty, or relating the proposition to what has been asserted in previous discourse. Their syntactic distribution reflects their status as sentence modifiers; they occur in sentence-peripheral positions (usually at the beginning of the sentence) or, parenthetically, at major constituent boundaries (e.g. between the subject and the verb or, more rarely, between the verb and peripheral elements within the VP such as PPs or clauses). In this respect fortzis is exceptional since it can occur readily between the auxiliary and lexical verb or between the verb and the direct object. Also within this class of modal adverbs we may include the ‘hearsay’ adverb naki (a contraction of naran ki…or narat ki… ‘they say that…’) which is frequently interspersed in reported speech (with the same syntactic distribution as the items in (2) other than fortzis), but can also have a modal value roughly equivalent to ‘apparently’. It is also frequently used to prefix sentences which broach a new topic of conversation; e.g. Naki ses cojuande ‘I hear that you are getting married’. So-called adverbs of time and place will be discussed in detail in the second half of this chapter. As we shall see, there are a number of distinctions which cut across the time/place parameter. Items such as accurtzu ‘near(by)’, prima ‘before(hand)’, dzosso ‘down (below)’, etc. have essentially the same distribution as locative or temporal PPs and can be analysed as intransitive prepositions (see 4.3.2). Deictic adverbs such as inoke ‘here’, como ‘now’, cras ‘tomorrow’ show some affinities with NPs in that they can refer to specific times or places and, indeed, can be used as complements as well as modifiers, as in (3): (3) a Juanne istat accurtzu. ‘John lives nearby.’ b L’appo postu in dzosso. ‘I put it down below.’ c Veni a inoke! ‘Come here!’

< previous page

page_169

next page >

< previous page

page_170

next page >

Page 170 d So pessande a cras. ‘I am thinking about tomorrow.’ When these items function as modifiers, indicating the time or place at which the event described by the sentence occurs, they occupy the same positions as the modal adverbs in (2). However, within the time dimension, there are also a few items, referred to earlier as ‘aspectual’ adverbs, such as dza ‘already’ and galu ‘still’ (which express time relative to some reference point which may be distinct from the present) and semper ‘always’, mai ‘never’, prus ‘no longer’ (which quantify over periods of time), whose syntactic distribution is significantly different from deictic time adverbs such as como or cras. As noted in 3.4.1 and 3.4.2, these items tend to cling much more closely to the verb, occurring freely between the auxiliary and lexical verb or between the verb and its complement; also, evidence from fronted constructions suggests that they can be adjoined to complements of the verb (cf. pp. 155–6). Finally we may note a handful of items which ostensibly involve notions of time, but have the syntactic distribution of manner adverbs; e.g. kitho ‘early’, tardu ‘late’, addedie ‘by day’, addenotte ‘by night’, torra ‘again’ (from torrare ‘to return, to continue’). Degree adverbs such as those in (4) can modify a variety of categories: (4) meta ‘much, many, very’, pacu ‘little, few, not very’,5 troppu ‘too (much/many)’, tantu ‘so (much/many)’, adziccu ‘rather, a bit’, goi/gai ‘so, as much as this/that’. When used to modify nouns, these items have the status of determiners and in some cases show agreement inflection (see 2.1.2). With adjectives and adverbs denoting gradable qualities, they express the degree of the quality whereas with verbs they typically denote frequency or duration, or with some verbs, the intensity of the emotion or activity; e.g. Amus cantatu meta ‘We sang frequently/for a long time’, L’appo offesu meta ‘I offended him deeply’, Appo travallatu meta ‘I worked frequently/for a long time/hard’. These items normally precede nouns, adjectives and adverbs (though meta and pacu can sometimes occur after the head) but follow verbs. Meta, pacu, troppu and tantu can modify all the categories mentioned above, whereas adziccu cannot modify nouns directly but functions as a partitive noun: (Un) adziccu de casu ‘a bit of cheese’ (cf. 2.2.2). As a degree modifier, gai can only occur with adjectives and adverbs, as in (5):

< previous page

page_170

next page >

< previous page

page_171

next page >

Page 171 (5) a Est gai minoreddu ki no si videt mancu. ‘It is so small that you cannot even see it.’ b Est thuccatu gai cuitande ki non 1’appo póttitu arrejonare. ‘He left so quickly that I was not able to speak to him.’ However, gai and goi can be used as post-head modifiers of all categories with the reading ‘like this/like that’, sometimes yielding connotations of degree as in: unu piske mannu gai/goi ‘a fish as big as that/this (with gesture)’, unu cretinu gai ‘such a cretin’. After verbs gai only has a manner reading (=‘that way’); to get a degree interpretation ‘so much’, we must add meta: (6) Appo ballatu gai meta ki no’rego prus in ancas. ‘I have danced so much that I can no longer stand.’ Other items which may be classed as degree adverbs are tottu (used as a universal quantifier ‘all’ before NPs —cf. 2.1.2, but as a superlative marker ‘really, absolutely’ with other categories), belle ‘almost’ and nessi ‘at least’: Issa est tottu cuntenta ‘She, is really happy’, tott’ accurtzu’ ‘really near’, belle/nessi novant’ annos ‘almost/at least ninety years’, belle inibe ‘almost there’. Unlike the items in (4), these items can precede a VP governed by an auxiliary: (7) a L’appo tottu vantatu. ‘I really flattered him.’ b Semus belle arrivatos. ‘We have almost arrived.’ c No’appo áppitu tempus de 1’arrejonare, ma 1’appo nessi salutatu. ‘I did not have time to talk with him, but I just greeted him.’ The degree sense of nessi is apparent only when this item modifies an expression involving an explicit quantity; e.g. nessi kimbe vias ‘at least five times’, nessi sa metate de sa síndria ‘at least half of the watermelon’. With other expressions, as in (7c), nessi often has the meaning ‘just’ with a function similar to that of the adverbs in (8), which we refer to as ‘focal’ adverbs: (8) fintzas ‘even’, mancu ‘not even’, ebbia/petzi/solu ‘only’, puru ‘as well’. These items can be adjoined to the maximal projection of any

< previous page

page_171

next page >

< previous page

page_172

next page >

Page 172 category which is interpreted as the focus of the sentence in order to contrast the information actually asserted with assumptions which might be created by context. Thus, (9a,b) both assert that John ate the cheese, but relative to different assumptions about what John might be expected to have eaten: (9) a Juanne at mandicatu fintzas su casu.6 ‘John ate even the cheese.’ b Juanne at mandicatu solu su casu. ‘John ate only the cheese.’ Typically, fintzas , mancu, petzi and solu precede the modified phrase whereas ebbia and puru follow it, though neither generalisation is absolute. Unlike the degree modifiers in (5), the position of these items does not vary according to the category of the modified phrase except in so far as a general constraint prevents any modifier from being attached to the left of a finite VP (see 3.4.1). Other items which have a similar function, but a more restricted syntactic distribution, are francu ‘except’ and mascamente/massimamente ‘especially’, which typically modify appositive expressions which serve to amplify or moderate the assertion expressed in the core sentence: (10) a Su sardu si faeddat in tottu s’isula, francu in Gaddura. ‘Sardinian is spoken throughout the island, except in Gallura.’ b Sos sardos faeddan tottu s’italianu, francu sos prus vetzos. ‘All Sardinians speak Italian except for the very old.’ c Su sardu si faeddat in tottu s’isula, mascamente in sa Barbadza. ‘Sardinian is spoken throughout the island, especially in Barbagia.’ d Tottu sos sardos faeddan su sardu, mascamente sos pastores. ‘All Sardinians speak Sardinian, especially the shepherds.’ Mascamente and massimamente can also be used to qualify complements, but not subjects: (11) a Sos pitzinnos faeddan mascamente/massimamente s’italianu. ‘The children speak mainly Italian.’

< previous page

page_172

next page >

< previous page

page_173

next page >

Page 173 b *Mascamente/massimamente sos pitzinnos faeddan s’italianu. ‘Mainly the children speak Italian.’ 4.2.1 Comparison of inequality Apart from the synthetic forms medzus ‘better’ and pejus ‘worse’, comparison is expressed by means of the adverb prus ‘more’. The comparative of inferiority is expressed by means of prus pacu ‘less, fewer’ (lit. ‘more little/few’). Like the other degree adverbs listed in (4) above, prus and prus pacu occur postverbally when they modify the verb, but phrase-initially when they modify other categories: (12) a Oje, appo travallatu prus (pacu). ‘Today, I worked more/less.’ b Isse est prus (pacu) abbistu. ‘He is more/less clever.’ c Bi keret prus (pacu) abba. ‘More/less water is needed.’ d Cantate prus abbellu! ‘Sing more quietly!’ In the above examples the standard of comparison (i.e. the element or situation with which something is being compared) is implicit. In the remainder of this section, we shall be principally concerned with the syntax of constructions in which the standard of comparison is overtly expressed. Essentially, there are two ways in which the standard of comparison can be introduced—either by means of the preposition de or by a construction of the form ki non X, where ki is the finite clause complementiser, non the negative particle and X a reduced clause in which certain redundant elements are omitted. On the whole, the formula with de is preferred in what we may call ‘straightforward’ cases of comparison, whereas the ki non X construction is restricted to more marked cases where there is a potential ambiguity to be resolved or where the construction with de is excluded. Some typical cases of the construction with de are given in (13): (13) a Juanne est prus mannu de Maria. ‘John is taller than Mary.’ b Lukia curret prus cuitande de Arrita. ‘Lucy runs faster than Rita.’

< previous page

page_173

next page >

< previous page

page_174

next page >

Page 174 c Gavini sonat sa kitterra prus de Frantziscu. ‘Gavin plays the guitar more than Francis.’ d Gavini sonat sa kitterra prus de su sonette. ‘Gavin plays the guitar more than the accordion.’ e Elene at comporatu prus meledda de Frantzisca. ‘Helen bought more apples than Frances.’ f Elene at comporatu prus meledda de pira. ‘Helen bought more apples than pears.’ Given the facts in (13c-f), we would expect a sentence like (14) to be ambiguous, just like its English translation: (14) Juanne istimat a Maria prus de Gavini. ‘John loves Mary more than Gavin.’ Although some of our informants perceived this ambiguity, others felt that this sentence could only mean ‘John loves Mary more than he loves Gavin’. Somewhat less typical are cases like (15) where the complement of de is (or can be construed as being) compared with an indirect object, or (16) where de introduces a phrase other than an NP:7 (15) Juanne pessat prus a sa mamma de su babbu. ‘John thinks more of his mother than (he does of) his father.’ (16) a Gavini est prus riccu de abbistu. ‘Gavin is more rich than clever.’ b Lukia iskit prus cantare de ballare. ‘Lucy knows how to sing better than how to dance.’ c B’at prus calore in beranu de in attonzu. ‘It is hotter in Spring than in Autumn.’ The de phrase can occur in a variety of positions. When it immediately follows prus (or prus pacu) as in (13c, d), we assume that prus de NP forms a syntactic contituent, an AdvP of which prus is the head, occurring within the VP and modifying the verb. Note that for independent reasons prus de NP can only form a syntactic constituent when it modifies the verb; recall that degree modifers precede all categories other than verbs, but phrasal modifiers are excluded from the pre-head position. Thus, in all other cases, prus de X must be split up, with prus in the pre-head position and de X in some post-head position, as in (13a, b, e, f), (15), (16a, c). Even when prus does modify the verb, it can be separated from the de phrase, as in (17) which corresponds to (13d):

< previous page

page_174

next page >

< previous page

page_175

next page >

Page 175 (17) Gavini sonat prus sa kitterra de su sonette. However, (18) is much less acceptable as a variant of (13c): (18) ?Gavini sonat prus sa kitterra de Frantziscu. Informants report that (18) can only have the absurd reading where Frantziscu is the understood object of sonare (i.e. ‘Gavin plays the guitar more than he plays Francis’).8 This suggests the following parsing strategy: (19)Given a sequence prus X de Y, where X and Y are of the same syntactic category, interpret X and Y as the compared items. Despite appearances, (13e) is not a counterexample to this strategy since meledda is a noun (or N′) governed by the quantifier/determiner prus whereas Frantzisca is an NP in its own right. With regard to the structural relation between prus and the de phrase, evidence from fronting suggests that in many cases the de phrase may be either contained within the same constituent as prus, as in the (a) examples below (where the de phrase is fronted along with the rest of the expression modified by prus), or attached to some higher node (e.g. the VP) as in the (b) examples: (20) a Prus mannu de Maria est. (=13a) b Prus mannu est de Maria. (21) a Prus cuitande de Arrita curret. (=13b) b Prus cuitande curret de Arrita. (22) a Prus meledda de pira at comporatu. (=13f) b Prus meledda at comporatu de pira. However, the fronting test demonstrates that in (13e), unlike (13f), the de phrase must be outside the NP modified by prus: (23) a *Prus meledda de Frantzisca at comporatu. b Prus meledda at comporatu de Frantzisca. The crucial difference between the two cases is that in (13f) the complement of de is compared with an element within the NP whereas in (13e) it is compared with an element outside the NP. The contrast between (22a) and (23a) can be accounted for if we postulate a constraint which defines NPs as islands with respect to relations of comparison: (24) …Y…[ NP …prus…de X…]…Y…

< previous page

page_175

next page >

< previous page

page_176

next page >

Page 176 Thus, for sentence (13f) we can postulate that both of the structures in (25) are possible and semantically equivalent: (25) a Elene [VP at comporatu [NP prus meledda de pira]] b Elene [VP cúrntu at comporatu [NP prus meledda] de pira] Similarly, for (13e) we may assume that both structures in (26) are syntactically well formed, but (26a) is ruled out on semantic grounds since the constraint in (24) prevents Frantzisca from being compared with Elene: (26) a *Elene [VP at comporatu [NP prus meledda de Frantzisca]] b Elene [VP at comporatu [NP prus meledda] de Frantzisca] Consequently, the sequence prus meledda de Frantzisca cannot be fronted (cf. (23a)) since it is not a semantically coherent constituent. The effects of (24) can also be seen in constructions with the lexicalised comparative forms medzus and pejus. As we noted earlier, these adjectives normally precede the noun. However, they can also occur postnominally provided that they are accompanied by a complement indicating the standard of comparison, as in (27b): (27) a Juanne comporat [NP medzus vinu (de custu)] b Juanne comporat [NP vinu [AP medzus de custu]] c *Juanne comporat [NP vinu [AP medzus]] ‘John buys better wine (than this).’ But the adjective cannot be postposed in cases like (28): (28) a Juanne comporat medzus vinu de Gavini. b *Juanne comporat vinu medzus de Gavini. ‘John buys better wine than Gavin (does).’ In (28b) de Gavini cannot form a constituent with medzus, because if it did it would also be contained within the NP and so would be prevented from being compared with Juanne by the constraint (24). Thus, to get a well-formed interpretation consistent with (24) we would have to assume that de Gavini is outside the NP, as in (29), a possibility which is excluded since medzus can only occur postnominally when it has a complement, as seen in (27c): (29) * Juanne [VP at comporatu [NP vinu [AP medzus]] de Gavini] Turning now to comparative constructions of the type prus…ki non X, we may preface our discussion by remarking that this formula is perceived as being somewhat unwieldy in comparison to the de X

< previous page

page_176

next page >

< previous page

page_177

next page >

Page 177 formula and, as such, tends to be restricted to cases where de X is impossible or marginal, or where there is a potential ambiguity to be resolved. Within the first category, we may note that the ki non X formula is the only way of expressing comparison between finite verb forms: (30) a Juanne dormit prus ki non travallat. b *Juanne dormit prus de travallat. ‘John sleeps more (often) than he works.’ Similarly, the ki non formula must be used when comparing situations which differ with respect to more than one element, as in the following attested example: (31) “…un arvore juche’ pru’ ramos chi no s’ómine brazzos” [SINNOS: 26] ‘…a tree has more branches than a man (has) arms’ The ki non formula is also used quite readily, as an alternative to de X, where comparison is with an indirect object or a predicative expression such as an adjective or infinitive, as in the following examples (cf. (15) and (16) above): (32) a Juanne pessat prus a sa mama ki no’ a su babbu. ‘John thinks more of his mother than of his father.’ b Gavini est prus riccu ki no’ (est) abbistu. ‘Gavin is more rich than (he is) clever.’ c Lukia iskit prus cantare ki non ballare. ‘Lucy knows how to sing better than (how) to dance.’ d B’at prus calore in beranu ki no’in attonzu. ‘It is hotter in Spring than in Autumn.’ However, the ki non construction is generally avoided when comparison is with the subject or direct object NP (as in the examples in (13)), except as a means of avoiding ambiguity, as in the following cases (cf. (14)): (33) a Juanne istimat prus a Maria ki non l’istimat Gavini. ‘John loves Mary more than Gavin loves her.’ b Juanne istimat prus a Maria ki no’ (istimat) a Gavini. ‘John loves Mary more than (he loves) Gavin.’ With regard to the internal syntax of the ki non construction, we assume that the complement of ki non is basically a finite clause in which redundant elements are omitted under conditions which we shall now attempt to define. As in all types of finite clause, the subject

< previous page

page_177

next page >

< previous page

page_178

next page >

Page 178 can be freely omitted (and usually is), unless it contrasts with the the subject of the main clause. If the subject is retained, it must follow the verb, as in (33a) above. The finite verb can also be omitted when redundant (as in (31), (32) and (33b)), except when it is accompanied by a redundant complement, in which case the verb is retained and the complement is realised as a clitic (as in (33a)). Our informants tended to reject the possibility of omitting the whole of the redundant VP (e.g. l’istimat in (33a)) leaving only the subject behind: (34) *?Juanne istimat prus a Maria ki non Gavini. For this reason, (35b) is more acceptable, in spite of its clumsiness, than (35a), and in (36) we have a clear contrast according to whether the retained element is a subject or a direct object (references in parentheses are to corresponding sentences with de in place of ki non ): (35) a *?Lukia curret prus cuitande ki no’ Arrita. (13b) b ?Lukia curret prus cuitande ki non curret Arrita. (36) a *Elene at comporatu prus meledda ki no’ (13e) Frantzisca. b Elene at comporatu prus meledda ki non pira. (13f) Note also that when the verb is deleted leaving a complement behind, any prepositions which serve to indicate the grammatical function of the complement (including accusative d) must be retained—contrast (32a) and (33b) with corresponding examples with de (15) and (14) respectively, where the NP denoting the standard of comparison is treated simply as a complement of the preposition de regardless of the grammatical function of the compared item. Finally, we may note that, in many cases, sentences containing the sequence prus ki non X are much less acceptable than sentences in which prus is separated from its complement; e.g. (37) is significantly worse than (33b): (37) ?Juanne istimat a Maria prus ki no’ a Gavini. The reason for this is not clear. Moreover, not all cases where prus is adjacent to ki non X are deviant (see, for example, (30a)). 4.2.2 Superlative degree Superlative degree is expressed by placing prus, prus pacu, medzus or pejus within the scope of the definite article:

< previous page

page_178

next page >

< previous page

page_179

next page >

Page 179 (38) Custu mele est su ‘This honey is the sweetest/least sweet/best/worst.’ In superlative, postnominal adjective phrases, repetition of the definite article is optional: (39) su mele (su) prus durke ‘the sweetest honey’ Typically, prus (pacu) occurs immediately before the word which denotes the gradable quality (e.g. an adjective). However, one occasionally encounters examples such as the following where it is placed before an earlier item within an attributive NP: (40)a Oje est sa prus die fritta de occannu. [today is the most day cold of this year] Today is the coldest day of this year.’ b “Fît una die de festa in bidda mia, sa pius festa manna ‘e s’annu.” ‘It was a feast-day in my village, the biggest feast of the year.’ (lit.‘…the most feast big of the year.’) [BS: 15] c “…fît fosse su ajanu pius de calidade alta.” [he+was] perhaps the bachelor most of quality high] ‘…he was perhaps the most eligible bachelor.’ [BS: 41] It is not clear what is going on here. Possibly, such cases represent a compromise between the desire to place prus as close as possible to the definite article (to make the superlative reading clear) and the requirement that APs must follow the noun. 4.2.3 Comparison of equality Comparison of equality can be expressed by cantu, ke or comente. All of these items require accusative a before following pronouns, proper nouns, etc. (cf. 2.2.6). (41) Juanne est mannu ‘John is as big as Francis/me.’ Of the three items, cantu most clearly expresses comparison in terms

< previous page

page_179

next page >

< previous page

page_180

next page >

Page 180 of degree, and can be used attributively without an accompanying adjective, with the meaning ‘as big as’ or ‘as old as’: (42) Lukia est cant’a Maria. ‘Lucy is as big/old as Mary.’ Ke and comente can also be used without an adjective, but in this case they simply indicate similarity rather than degree: (43) Lukia est ke/comente a Maria. ‘Lucy is like Mary (in appearance, behaviour, etc.).’ A similar contrast is found when these items are used adverbially: (44) a Jeo appo travallatu cant’a tie. ‘I have worked as much (=‘as hard, long, frequently’) as you.’ b Jeo appo travallatu ke/comente a tie. ‘I have worked just like you/in the same way as you.’ Cantu (more rarely comente,9 but not ke) can also be used with a clausal complement, as in (45): (45) Isse est istupidu fit abbistu su babbu. ‘He is as stupid as his father was clever’ Note that, in such cases, the subject is placed after the verb, just as in comparative clauses introduced by ki non (cf. p. 178). Also, cantu can be used to compare APs: (46) “duos zovanos abiles e balentes cantu onestos, aderios a su dovere issoro” ‘two young men able and worthy as much as (they were) honest, dedicated to their duty’ [BS: 109] When the standard of comparison is an NP with generic reference (usually in idiomatic expressions or similes), ke is used in preference to cantu or comente, and the article is omitted: tétteru ke fuste ‘as stiff as a rod’, caente ke latte ‘as warm as milk’, etc. 4.3 PREPOSITIONS AND ADVERBS OF TIME AND PLACE 4.3.1 Basic uses of prepositions The principal uses of morphologically simple prepositions which can be followed directly by an NP are outlined below. Uses which are

< previous page

page_180

next page >

< previous page

page_181

next page >

Page 181 discussed elsewhere in this or other chapters are not illustrated here but are simply mentioned with references to the appropriate page or section. A ‘to, at’ This preposition is used primarily to express time or place with a ‘Goal’ orientation (see 4.3.3), as a dative marker introducing the indirect object of verbs of ‘giving’, ‘telling’, ‘showing’, ‘experiencing’, etc. (see 3.2.1) and as an accusative marker before direct objects consisting of a proper noun, pronoun, etc. (see 2.2.6). It is also used to indicate the logical subject of transitive infinitives in causative constructions (see 6.1.4). Other more idiomatic uses include the following: To indicate distance: Su mare est a tres kilómetros The sea is three kilometres away’. To indicate a resulting state: L’appo secatu a thadzos ‘I cut it into slices’, L’anfattu a sordatu They made him a soldier’. To introduce NPs denoting an identifying feature of some entity: L’appo connottu a sa voke ‘I recognised him by his voice’, una kitterra a dóiki cordas ‘a guitar with twelve strings’; similarly, with body-part nouns in expressions of posture, etc.: a math’a terra ‘with (one’s) stomach to the ground’ (i.e. ‘lying on one’s front’), So a thudda ritza ‘I have goosepimples’ (thudda=‘bristles (e.g. of a pig)’, ritza ‘standing up’), a manos ligatos ‘with hands tied’. With plural of certain action nouns (as attribute or manner adverbial): So a tziccolitos ‘I have hiccoughs’ (lit.‘…at hiccoughs’), Sun a vokes e a corfos ‘They are shouting and fighting’ (lit.‘…at voices and at blows’), L ’as fattu a brinkos ‘You did it haphazardly’ (from brincare ‘to jump’). To form miscellaneous adverbial expressions: (means) pesatu a latte ‘raised on milk’, (guise) vestitu a prítteru ‘dressed as a priest’, (attitude) a dispettu ‘out of spite’, (extent) prenu a cuccuru ‘full to the brim’, (position) a s’imbesse ‘upside-down, back to front’. In ‘in, on’ Like a, this preposition is used primarily in locative and temporal expressions (see 4.3.3, 4.3.4). It is also used figuratively in a variety of expressions of the type ‘in a state of…’: in amistate ‘on friendly terms’, in presse ‘in a hurry’, in férias ‘on holiday’. It is used with

< previous page

page_181

next page >

< previous page

page_182

next page >

Page 182 numerical expressions to quantify the subject, and as complement of verbs of ‘division’: Sun vénnitos in tres Three of them came’ (lit. ‘they came in three’), L’appo secatu in battor ‘I cut it in four’ (compare the similar use of a with non-numerical expressions: secatu a thadzos ‘cut into slices’). Dae ‘from’ (Logudorese-Nuorese only) 10 This item is used to indicate a locative source (see 4.3.3) or the beginning of a period of time (see 4.3.4). The source aspect of this preposition takes on a causative value in non-locative, adverbial expressions such as Est mortu dae s’assúccunu ‘He died from the shock’. This preposition also introduces the Agent in passive constructions (see 3.2.7) and, as an alternative to a, in causative constructions with a transitive infinitive (see 6.1.4) De ‘of’ Arguably, this preposition has no intrinsic semantic content. Generally, the relations which it expresses are those which are pragmatically the most salient given the properties of its complement with respect to other relevant items in the sentence. In particular, it is used to introduce Theme complements which, for syntactic reasons, cannot be realised as direct objects; for example, complements of adjectives and nouns (cuntentu de su travallu ‘happy with the work’, unu professore de frantzesu ‘a teacher of French’), complements of verbs whose direct object relation is fulfilled by some other item, including pleonastic si (see 3.2.1, 3.2.6), and complements of certain prepositions and locative-temporal adverbs (see 4.3.2). It also introduces the standard of comparison in comparatives of inequality (see 4.2.1). De has a partitive function after certain indefinite quantifiers and nouns which denote quantity (see 2.2.2), also before indefinite mass or plural nouns in dislocated position (see 1.1.2, 7.1.1, 7.2.2). This preposition also has a variety of attributive uses. With a human complement, such expressions often show ambiguities which can only be resolved pragmatically; for example, sa fotografía de Juanne can mean ‘the photograph belonging to John’ (possession), ‘the photograph depicting John’ (Theme), ‘the photograph taken by John’ (Agent). Such expressions can also be used predicatively with the same range of meanings Custa fotografía est de Juanne This photograph is John’s/of John/by John’. De can be used adnominally and predicatively to indicate ‘material’ or ‘origin’: un’ anneddu de oro

< previous page

page_182

next page >

< previous page

page_183

next page >

Page 183 ‘a gold ring’, Est de oro ‘It is (made) of gold’, vinu de Olíana ‘wine from Oliena’, un’ómine de Lodé ‘a man from Lodé’, Sun de Lodé ‘They are from Lodé’. Other attributive expressions with de can only be used adnominally: (quality) un’ómine de onore ‘a man of honour’, (type) un’arvore de péssike ‘a peach-tree’. Within this class, we may include compounds such as una camisa de notte ‘a nightshirt’, unu mastru de linna ‘a carpenter’ (lit. ‘master of wood’). Like dae (see above), de has a causative use: So murrinde de su sitis ‘I am dying of thirst’. There is a subtle semantic difference between this and the corresponding use of dae. Typically, dae introduces a specific state of affairs which literally causes the event denoted by the verb, whereas de also allows a figurative reading of the verb; for example, ‘I am so thirsty I could die’. Intro ‘inside, within’, supra ‘on, above’, sutta ‘under, below’ The primary function of these items is to denote spatial relations. They can be preceded by a, in, and dae, and followed by a or de to form complex locative expressions (see 4.3.2, 4.3.3). In addition, intro can be used to express a limitation on the time axis (see 4.3.4) and supra can have the figurative value ‘about’: unu libru supra sa Sardigna ‘a book about Sardinia’. Peri ‘across, through’, tra ‘between, among’ These are basically locative prepositions: So cúrritu peri sas tancas/ peri su padente ‘I ran across the fields/through the wood’, Orune est tra Núgoro e Vithi ‘Orune is between Nuoro and Bitti’. Peri can also be used with verbs of taking or holding, to indicate the part of the entity which is being grasped: L’appo tiratu peri su bratzu ‘I dragged him by the arm’. Tra can be used figuratively: Tra te e me, non b’ at segretos ‘Between you and me, there are no secrets’, tra su nárrere e su fákere …‘between saying and doing…’. Kin (cun) ‘with’ This item has essentially the same range of semantic functions as English ‘with’: (comitative) Juanne est issitu kin Maria ‘John went out with Mary’, Mandicamus sa petha kin coritholu ‘We eat meat with beans’, (instrument) L’appo accontzatu kin unu marteddu ‘I mended it with a hammer’, (reason) Kin custu calore, non potto dormire ‘With this heat, I cannot sleep’, Kin su fidzu maláidu, deven abbarrare in domo ‘With their son sick, they must stay at home’. It is also used with certain abstract nouns to form circumstantial adverbial

< previous page

page_183

next page >

< previous page

page_184

next page >

Page 184 expressions: kin piakere ‘with pleasure’, kin onore ‘with honour’, kin difficultate ‘with difficulty’. Kene fkentza, sena) ‘without’ This item functions as the negative of kin in all the uses described above: Est issitu kene su babbu ‘He went out without his father’, Non si potet mandicare petha kene pane ‘One cannot eat meat without bread’, Non lu potto secare kene su liputzu ‘I cannot cut it without the knife’, Kene custu calore, aíamus póttitu dormire ‘Without this heat, we would have been able to sleep’, kene piakere ‘without pleasure, reluctantly’, etc. Unlike kin, kene allows the article to be omitted before concrete nouns in some cases: So issitu kene crae ‘I went out without (the) key’ (for discussion, see 2.2.5, p. 58). Kene can be used predicatively (Semus kene pane ‘We have no bread’ (lit. ‘We are without bread’)) and also before past participles in negative passive constructions of the type Sa domo est kene pulita The house has not been cleaned’ (lit. ‘The house is without cleaned’) (see 3.2.7). Pro (po) ‘for’ This preposition has a variety of uses involving the notions of reason or purpose: (purpose) So ghiratu a domo pro ti víderelpro sa festa ‘I came home to see you/for the festival’, (benefactive) L’appo fattu pro te ‘I did it for you’, (reason) Pro cussu, non potío vénnere ‘Because of that, I could not come’. The ‘reason’ value of pro is often made more explicit by the phrase pro more de ‘on account of, in view of: Pro more de su tempus, non semus issitos ‘On account of the weather, we did not go out’. Pro can also be used with the sense ‘in exchange for’: Si usat de dare unu kilu de mele pro unu kilu de casu ‘It is customary to give a kilo of honey (in exchange) for a kilo of cheese’, L’appo vénditupro deke midza de francos ‘I sold it for 10,000 lire’. Perhaps as an extension of this use, pro can also mean ‘instead of: Appo piccatu s’abbardente pro s’abba ‘I took the brandy for/instead of water’. In addition, pro can be used with the sense ‘in the opinion of: Pro me, tue ses maccu ‘For me (=in my view), you are mad’. Sicunda ‘according to’ This item can be used in place of pro in cases like: Sicunda Juanne, at a próere cras ‘According to John, it will rain tomorrow’. It can also be

< previous page

page_184

next page >

< previous page

page_185

next page >

Page 185 used with the sense ‘depending on’: Sos prejos cámbian sicunda sa calitate Prices vary according to quality’. Cantu ‘like, as big as, as much as’, ke, comente ‘like, as’ In their comparative use described in 4.2.3, these items can be classed as prepositions since they can take an NP complement. However, cantu and comente have other non-prepositional uses (respectively as an interrogative quantifier ‘how much/many’ and adverb ‘how’). 4.3.2 Prepositions and their complements We assume that prepositions, like verbs, may differ according to the type of complements which they select. Thus, in addition to examples of the type illustrated in 4.3.1, where prepositions are directly followed by an NP, we postulate that some prepositions may take a clausal complement (S′) or a PP or, indeed, may occur intransitively without a complement. This assumption has certain consequences for the range of items which we take as belonging to the category of prepositions. For example, the class of intransitive prepositions may include some items traditionally classified as adverbs. Similarly, some so-called ‘subordinating conjunctions’ can be analysed as prepositions which take a clausal complement—though we shall defer discussion of such cases to our examination of subordinate clauses in Chapter 6. Within the theoretical framework adopted here (see 0.2), the ability of prepositions and verbs to take an NP complement can be attributed to their capacity to assign a Case feature to their complement, the leading idea being that all argument NPs must be licensed by such a feature. In Sardinian, NP complements of prepositions show a three-way Case contrast, as can be seen when this NP is a first or second person singular pronoun (the only NP forms which show overt Case inflection). The actual morphology of these pronouns varies somewhat from one dialect to another (see 5.1.1 for details), but the Case distinctions which they represent are fairly consistent across dialects. With most prepositions, these pronouns are realised as me and te, which we shall refer to as ‘oblique’ Case forms: kene te ‘without you’, dae te ‘from you’, prus mannu de me ‘bigger than me’, pro me ‘for me’, tra te e me ‘between you and me’, etc. The preposition kin takes a special ‘comitative’ Case: kin mecus/tecus. The item a in all of its uses (both as a genuine preposition and as an accusative marker) takes the forms mie and tie, which we refer to as ‘accusative-dative’.

< previous page

page_185

next page >

< previous page

page_186

next page >

Page 186 Some prepositions, notably the locatives intro, supra and sutta and the comparatives ke, cantu and comente, systematically take accusative a (just like transitive verbs) when the complement NP meets the conditions described in 2.2.6 (for example, when it is a pronoun or a proper noun): intro a mie ‘inside me’, supra a tie ‘on top of you’, coment’a Juanne ‘like John’, etc. For some speakers, the range of prepositions which take accusative a is more extensive, including some uses of pro and even de, as in the following attested examples: “Faghe comente ti naro, chi est mezus pro a mie” ‘Do what I tell you, for it is better for me’ [CF: 202], “…ca nde ischit pius de a nois” ‘ …for he knows more of it than us’ [ BS: 117]. A further complication with the locatives intro, supra and sutta is that they can be optionally followed by a preposition (usually de, but in some dialects a) even with NPs which do not normally require the prepositional accusative: intro (de) sa domo ‘inside the house’, sutta (de) su lettu ‘under the bed’. This may be related to the fact that, unlike the other prepositions discussed in 4.3.1, these three items can occur without any complement, sometimes with a specialised meaning which incorporates information about the nature of the entity which would otherwise be expressed by the complement; e.g. intro ‘inside, indoors’, supra ‘above, upstairs’, sutta ‘down below’. Indeed, there are many other items (many of which are traditionally classed as adverbs) which can occur without any complement or with an NP introduced by a preposition (indicated in parentheses): (47) accurtzu (a) ‘near’, addainnantis/indainnantis (de) ‘in front’, addaisecus/indaisecus (de) ‘behind’, allargu (dae) ‘far’, fattu (a, de) ‘after (person or thing)’, fora(s) (de) ‘outside’, intundu (de) ‘around’, (da)poil(ap)pustis (de) ‘after’, prima (de) ‘before’ (de) ‘before’. Following the approach envisaged at the beginning of this section, we may treat the items in (47) as intransitive prepositions. That is, they belong to the same syntactic class as the items discussed in 4.3.1, but lack the capacity to assign Case and consequently must occur without a complement or must be accompanied by a ‘dummy’ preposition whose function is to assign Case to the following NP. Within this perspective, intro, supra and sutta can be classed as prepositions which can be used transitively (assigning Case directly like the other items in 4.3.1) or intransitively (occurring without a complement or assigning Case via a ‘dummy’ preposition, like the items in (47)). Not all intransitive prepositions can be followed by a complement; in particular the items dzosso ‘down below, downstairs’ and susu ‘up

< previous page

page_186

next page >

< previous page

page_187

next page >

Page 187 above’ cannot be used to form any expressions of the type *susu de sa crea ‘above the church’, *dzosso de su lettu ‘under the bed’, etc. The item contra ‘against’ may be an example of the converse type in that it normally requires its complement to be introduced by a11 (suggesting that it is intransitive), but cannot normally be used without a complement: Ses semper contra a isse ‘You are always against him’, Cussu est contra a sa ledze ‘That is against the law’, *Jeo so contra ‘I am against’. It may be noted that many of the items in (47) begin with a- or in- . There are numerous complex expressions, functionally similar to the items in (47), which are formed by means of the prepositions a or in followed by a noun denoting a part or dimension of an entity:12 (48) a banda (de) ‘apart from’ (banda=‘side’), a caddu (de) ‘astride’ (caddu=‘horse’), a costadzu (de) ‘beside’ (cos-tadzu =‘side of rib-cage’), in fundu (de) ‘under’ (fundu= ‘bottom’), in mesus (de) ‘amidst’ (mesu=‘middle’), in palas (de) ‘behind’ (pala=‘back (of person/animal)’), in s’oru (de) ‘near, alongside’ (oru =‘edge’). With regard to the syntactic structure of these expressions, it is not immediately clear whether the ‘part’ noun goes with the preceding a or in to form a ‘complex preposition’ which takes de+NP as its complement, as in (49a), or whether a and in function as normal prepositions governing an NP composed of a ‘part’ noun and its complement, as in (49b): (49) a [PP [P in s’oru] [PP de su mare]] b [PP in [NP s’oru [PP de su mare]]] ‘at the edge of the sea’ (i.e. ‘by the sea’) According to (49b), such constructions resemble normal ‘possessive’ constructions which denote a partwhole relation (e.g. sa janna de sa domo ‘the door of the house’), and indeed the de+NP is frequently replaced by a possessive item when pronominal: (50) a Sedi a costadzu meu. ‘Sit down beside me.’ b Juanne fit in palas tuas. ‘John was behind you.’ Note that the possessive agrees in number and gender with the part noun (e.g. in (50b) both palas and tuas are feminine plural). This use might be taken as evidence in support for analysis (49b), except for the fact that such use is also common with items like those in (47) which contain no identifiable nominal element:

< previous page

page_187

next page >

< previous page

page_188

next page >

Page 188 (51) a . ‘Mary sat down in front of/behind me.’ b “…amus s’obbricu de los lassare a sos chi benin appustis nostru.” ‘…we have the obligation to leave them for those who come after us.’[Preface to EP: 17] c “Sos ogros sun puntados supra sua (a isse).” ‘The eyes are pointed at him.’ (Blasco-Ferrer 1986:188) Note that even in these cases, some form of agreement process seems to be at work, as shown in (51c) where supra is followed by a feminine singular form. A similar instance is found in Campidanese where, according to Lepori (1979:10), we have the expressions “acanta mia/tua” ‘near to me/you’, where mia and tua are feminine possessive forms even though acanta does not contain a feminine noun. What seems to be happening here is that supra and acanta are treated as feminine forms for the purposes of agreement by analogy with the general use of -a as a feminine desinence with nouns, adjectives, etc. Although the expressions in (48) are semantically fairly transparent, they are not strictly compositional in the way we might expect on the basis of structure (49b). In particular, we find cases of the type a costadzu de X, in palas de X, etc. where X denotes an inanimate object which does not have any part or dimension which could normally be identified by the nouns costadzu, palas, etc.; thus, costadzu de sa domo on its own is semantically deviant, but becomes meaningful when preceded by a: a costadzu de sa domo ‘beside the house’. Also, the semantic contrast between a and in (movement to a goal vs. static location: see 4.3.3), which is quite systematic before NPs, is neutralised in these expressions. These observations might be taken as evidence that alin and the part noun have been fused into a single meaningful unit (the complex preposition in (49a)) in which each item has lost its own semantic properties though each obviously contributes to the meaning of the expression as a whole. We shall not attempt to decide between the two analyses proposed in (49). It is quite conceivable that both structures are available for expressions of this type, with the further possibility that individual expressions may favour one structure over the other. A final point to note in connection with transitivity of prepositions

< previous page

page_188

next page >

< previous page

page_189

next page >

Page 189 is that the complement of certain prepositions and similar expressions can be expressed as a dative clitic: (52) a S’abba lis est akkirratu a supra. ‘The water fell down on them.’ b Juanne m’est cúrritu in fattu. ‘John ran after me.’ c Lis appo postu sa mesa indainnantis. ‘I put the table in front of them.’ This construction is possible only with prepositions which can be used intransitively; i.e. with intro, supra and sutta and items of the type in (47). Some of the expressions in (48) (e.g. in mesus and in palas) also allow this construction, but others do not, possibly for semantic reasons—generally this construction carries the implication that the dative entity is affected in some way by the event. We defer discussion of the syntax of these constructions until section 5.2.3, where we shall investigate them within the general context of the ‘dative of interest’ phenomenon. 4.3.3 Location, goal and source For the purposes of this discussion we assume that locative relations can be analysed in terms of two basic parameters; first, the spatial relation involved (e.g. location with respect to a point, surface, space or some particular dimension of an object,), second, an aspectual parameter (static location vs. ‘goal’ or ‘source’— movement to or from a place). It is well known that languages differ in the ways in which these two parameters can be expressed within the prepositional system. For example, English has a fairly rich system whereby individual prepositions can express both aspectual and spatial information (at least for the simpler spatial relations): e.g. at, to, from; on, on to, off; in, into, out of. In French, on the other hand, prepositions cannot express aspectual and spatial information simultaneously—e.g. the source preposition de does not specify any particular spatial relation whereas prepositions which do express spatial relations (e.g. à ‘at, to’, sur ‘on’ and dans ‘in’) are neutral with respect to aspectual distinctions. In Sardinian, the general tendency is for aspectual and spatial information to be encoded separately, aspect being expressed by in (static location), a (goal) or dae (source), whereas specific spatial information can be expressed by prepositions such as supra ‘on (top of)’, intro ‘in(side)’, sutta ‘under’, etc.

< previous page

page_189

next page >

< previous page

page_190

next page >

Page 190 The spatial neutrality of in, a and dae is illustrated in the following examples: (53) a Semus in . ‘We are in Nuoro/in France.’ b Sos prattos sun in sa mesa. ‘The plates are on the table.’ c Sa camisa est in su calassu. ‘The shirt is in the drawer.’ ‘I went to Nuoro/to France’ b Sun ampilatos a sa bóvita. (54) a So andatu a . ‘They climbed onto the roof.’ c L’appo ghettatu a su bidone. ‘I threw it into the waste-bin.’ (55) a So andatu dae Olbia a Núgoro. ‘I went from Olbia to Nuoro.’ b Su liputzu est ruttu dae sa mesa. ‘The knife fell off the table.’ c Semus issitos dae su thilleri. ‘We went out of the inn.’ Note in particular that in and a (unlike their cognates in other Romance languages) are not distinguished in terms of properties of the location entity (for example, whether it is conceived as a point or as an area or volume), but simply denote static location and goal respectively, the nature of the spatial relation being determined pragmatically or by other elements in the sentence such as the verb. Nevertheless, there are a couple of caveats which must be appended to this generalisation. First, the complements of some verbs (notably pónnere ‘put’) are usually introduced by in even though they represent a goal rather than a static location: (56) L appo postu in . ‘I put it on the table/in the waste-bin.’ Conversely, when a stative locative relation is defined with reference to an event rather than a physical object, a is used in preference to in:

< previous page

page_190

next page >

< previous page

page_191

next page >

Page 191 (57) Fimus a . ‘We were at the festival/at the mass.’ There are also some cases where usage concerning in and a is uncertain. For instance, with toponyms the distinction is typically made in terms of the location versus goal contrast, as in (53a), (54a) above, but some speakers base the distinction on a ‘town’ versus ‘country/region’ contrast, as in (58)—perhaps through Italian influence: (58) a . b . This use of in and a based on a spatial contrast is not typical of the Logudorese-Nuorese dialects, though it appears to be more prevalent in the Campidanese dialects. Indeed, Blasco Ferrer (1986:183) observes that Campidanese has a special particle me which can be used to specify the static locative sense of in: “Est sempiri me in domo” ‘(S)he is always at home’. The semantic distinction between a and in is also somewhat blurred in semi-idiomatic expressions involving a noun without an article, where the choice of preposition is subject to free or idiolectal variation or, in some cases, conflicts with the generalisation proposed above: e.g. éssere a bindza ‘to be at the vineyard’, éssere alin domo ‘to be at home’, rúghere alin terra ‘to fall to the ground’. Also, as we have noted, the semantic distinction is neutralised in complex expressions like those listed in (48) in section 4.3.2 where the choice between a and in is fixed for each expression (with due allowance for possible dialectal variation). The prepositions in, a and dae can be combined with intro, supra and sutta when specification of the spatial relation is required: (59) a Juanne est in intro (de) sa domo. ‘John is inside the house.’ b So cúrritu a supra (de) su ponte. ‘I ran onto the bridge.’ c Semus issitos dae sutta (de) sa mesa. ‘We came out from under the table.’ In these cases, the choice between in, a and dae is determined strictly

< previous page

page_191

next page >

< previous page

page_192

next page >

Page 192 by the Location/Goal/Source distinction. However, intro, supra and sutta can be used without in to specify a static location and, in some marginal cases, without a to specify a goal. In, a and dae are used in much the same way before interrogative and demonstrative adverbs of place: (60) a (In) ube ses? ‘Where are you?’ b A ube andas? ‘Where are you going?’ c Dae ube venis? ‘Where do you come from?’ Many items of this type have alternative forms in which in- has become lexically incorporated as part of the adverb and can be preceded by a or dae . Thus, alongside (60), we have the following paradigm: (61) a Inube ses? ‘Where are you?’ b A inube andas? ‘Where are you going?’ c Dae inube venis? ‘Where do you come from?’ Other items which show this phenomenon are the demonstratives ibe/inibe, cue/incue and cuddae/incuddae (for the semantics of these items, see 4.3.5). With all of these items, the forms incorporating in- are the more common and in some dialects represent the only possibility. Incorporation of in- is complete in all dialects with the item inoke ‘here’ (< Lat. HOC-CE); i.e. we do not find *oke, *a oke or *dae oke in any dialect:13 (62) a So inoke. ‘I am here.’ b Veni a inoke! ‘Come here!’ c Issimus dae inoke. ‘Let us get out of here.’ Conversely, lexical incorporation of in—does not occur with the items atterube ‘elsewhere’, tottube ‘everywhere’ and neddube ‘nowhere’: (63) a Sun (in) atterube. ‘They are elsewhere.’ b Non so andatu a neddube. ‘I did not go anywhere.’

< previous page

page_192

next page >

< previous page

page_193

next page >

Page 193 c Venin dae tottube. ‘They come from everywhere.’ A final point to note is that in expressions involving movement of a person to another person, the goal is specified by means of a periphrastic construction with the adverb ube ‘where’ and finite forms of éssere ‘be’:14 (64) Ando a ube est Juanne. ‘I am going to John.’ (lit.‘…to where John is.’) In this construction, éssere may occur in the past tense and normally inflects for number and person according to the following NP, which assumes nominative Case: (65) a So andatu a ube fin issos. ‘I went to them.’ (lit.‘…to where they were.’) b Veni a ube so jeo! ‘Come to me!’ (lit.‘…to where I am.’) Although this construction is clearly clausal in nature (with obligatory inversion of the subject), there is some evidence that it is becoming lexicalised (at least in the case where éssere is in the third person, present tense). In most dialects, this expression undergoes contraction: a ube est →/avé(s)/ (Camp. a undi esti →/andé(sti)/). In some cases this reanalysis has syntactic effects. Lepori (1979:12) gives the Campidanese example a nd’e mei with the reading ‘at/to my house, to me’, where e is a contraction of the third person singular of éssere and mei is an oblique form—suggesting that mei is perceived as the object of a preposition rather than as the subject of éssere (compare (65b) with nominative jeo ). Note also the following attested example (Nuorese dialect) where éssere is omitted: “Su sero fit torrau a ube Mallena” ‘In the evening he went back to Maddalena’ (lit.‘…to where Maddalena’) [EP: 187]. 4.3.4 Temporal uses of prepositions In expressions of time, in is used with nouns which denote a period of time whereas a is used with punctual expressions: in tríulas ‘in July’, in s’istíu ‘in summer’, but a sas otto ‘at eight o’clock’, a mesudie ‘at midday’, a s’impoddile ‘at dawn’. Thus, in their temporal uses, a and in can be characterised in terms of a ‘point’ vs. ‘extension’ distinction which we found to be of marginal relevance for their locative uses. In is also used to specify a future time in terms of a duration from the present or some other reference point:

< previous page

page_193

next page >

< previous page

page_194

next page >

Page 194 (66)

App’a ghirare in tres dies. ‘I will return in three days.’ Similarly, intro can be used to specify a period within which an event will occur: (67) Ti teléfono intro (de) s’ora. ‘I will telephone you within the hour.’ When a period of time is defined in terms of its beginning and end, dae is used for the former and a for the latter: dae sas nove a mesudie ‘from nine o’clock to midday’. When the end of the period is understood as the present or a reference time given by context, the a phrase is left out and dae assumes the value of ‘since’: (68) Semus inoke dae eris. ‘We have been here since yesterday.’ Conversely, if the beginning of the period is left unspecified, a is usually preceded by fintzas ‘even’: (69) Abbarramus fintzas a cras. ‘We are staying until tomorrow.’ Both dae and (fintzas) a can govern a clause introduced by cando ‘when’: (70) No’amus dormitu dae cando semus arrivatos (fintzas) a cando nonke semus andatos. ‘We did not sleep from when we arrived until (when) we left.’ Also, certain adjectives and adverbs can occur as complements of dae and (fintzas) a: dae kitho ‘from early’, (fintzas) a tardu ‘until late’, dae novu ‘again’ (lit. ‘from new’), dae minore ‘when young’ (e.g. Est emigratu dae minore a Frantza ‘He emigrated when young to France’), a mannu ‘when grown-up’ (e.g. Cussu l’appo imperatu a mannu ‘This I learned when grown-up’), dae meta ‘a long time ago, a long time before’ (lit. ‘from much’). There is no preposition corresponding to temporal ‘for’ as in ‘I waited for three hours’ or ‘I have not eaten for three days’. In such cases, the time expression is introduced by éssere and the verb is realised as a present participle or, if negated, by an infinitive introduced by kene ‘without’: (71) a Fippo tres oras isseande. lit. ‘I was three hours waiting.’

< previous page

page_194

next page >

< previous page

page_195

next page >

Page 195 b So duas dies kene mandicare. lit. ‘I am two days without eating.’ To specify the time reference of a past event in terms of the period separating it from the present (= English ago), a variety of constructions are used: (72) a L’appo accabbatu como tres annos. ‘I finished it three years ago.’ (lit. ‘…now three years.’) b Est náskitu duos mese faket. ‘He was born two months ago.’ (lit. ‘…two months it makes.’) c Semus arrivatos battor dies a oje. ‘We arrived four days ago.’ (lit. ‘…four days to today.’) The relation ‘before’ is expressed by prima or, more rarely, antis. Both of these items can be followed by a finite subjunctive clause introduced by ki or by an infinitive or NP introduced by de. Corresponding to ‘after’ we have (ap)pustis, (da)poi and, increasingly, dopo (from Italian). All three items can take a finite indicative complement clause introduced by ki. Dopo can be followed directly by an NP, but with (ap)pustis and (da)poi the NP must be preceded by de (see 4.3.2). With these items the non-finite complement construction parallel to prima+de+ infinitive is formed by means of the past participle: (73)

mandicatu, nos semus corcatos. ‘After eating, we lay down.’ (lit. ‘After eaten…’) For further discussion of time clauses, see Chapter 6. 4.3.5 Deictic adverbs of place and time The principal demonstrative adverbs of place are inoke, (in)cue, (in)cuddae and (in)ibe . Inoke ‘here’ denotes a place which is close to the speaker, but possibly remote from the hearer. The items (in)cue and (in)cuddae both indicate places remote from the speaker and, in the case of (in)cuddae, the hearer as well. The difference between these two terms is largely a matter of degree, (in)cuddae being used for locations which are more remote and/or less precise than those denoted by (in)cue. The item (in)ibe is used to refer back to some place which has been mentioned in preceding discourse.

< previous page

page_195

next page >

< previous page

page_196

next page >

Page 196 The above forms can be preceded by prepositions; not only in, a and dae as discussed in 4.3.3, but also preposition. which describe particular spatial relations (e.g. intro, peri, supra, sutta, tra, …): (74) a Istíkkiti intro inoke! ‘Hide in here!’ b L’appo postu supra incue. ‘I put it on there.’ In such cases, the preposition specifies the spatial relation between the Theme entity and the place denoted by the adverb. A slightly different effect is achieved when the adverb is followed by an intransitive preposition or other locative expression, as in (75): (75) a Abbarramus inoke foras. ‘Let us stay out here.’ b Dae incuddae supra si videt su mare. ‘From (right) up there you can see the sea.’ c Deus, k’est in cue supra… ‘God, who is up above…’ In these cases, the final element acts as a modifier of the demonstrative adverb, giving further specification of the place referred to or of the deictic orientation with respect to the interlocutors. Thus, whereas supra incue means something like ‘on or above some place remote from here’, incue supra means ‘at some place remote from here in an upward direction’. Deictic adverbs of time can be divided into two broad classes; those which make reference to natural periods of time and those which do not. Lexical adverbs of the first type all specify time in relation to the present, the relevant periods of time being days or years: oje ‘today’, cras ‘tomorrow’, pusticras/barrigadu ‘the day after tomorrow’, eris ‘yesterday’, janteris ‘the day before yesterday’, occannu ‘this year’, annovas ‘next year’, annupassatu ‘last year’, antiannu ‘the year before last’. Other time references are expressed by periphrastic formulae of the type custa kitta ‘this week’, sa kitta passata ‘last week’, sa kitta ki venit ‘next week’. The last two formulae are also used with days of the week: lunis passatu ‘last Monday’, kenápura ki venit ‘next Friday’. Names of days indicating a time within the current week are used without an article or preposition: So arrivatu martis, mink’ando sábbatu ‘I arrived on Tuesday, I am leaving on Saturday’. Deictic adverbs which indicate time without reference to natural periods can be divided into two subtypes. On the one hand, we have

< previous page

page_196

next page >

< previous page

page_197

next page >

Page 197 como (Camp. immoi) ‘now’ and tando (Camp. insara) ‘then’ which denote a time which is contemporary with some reference point, which in turn is specified as being either identical with (como) or distinct from (tando) the present. On the other hand, we have items such as derettu ‘immediately’, galu (Camp. ancora) ‘still, yet’, dza ‘already’ and luego ‘soon’. These items denote an event time or period which is non-identical with some reference point (either before or after), but the relation of this reference point to the present (identity or non-identity) is left open. Thus, dza indicates that the event occurs before a contextually determined reference point (represented by the bracketed expressions in (76)) which may be in the past, present or future: (76) a Juanne fit dza arrivatu (sábbatu passatu). ‘John had already arrived (last Saturday).’ b Juanne est dza arrivatu (como). ‘John has already arrived (now).’ c Juanne at a éssere dza arrivatu (cras). ‘John will have already arrived (tomorrow).’ Similarly, galu indicates that the specified state (or absence of event) holds throughout the period up to the reference point, which again may be past, present or future: (77) ‘(Two years ago) she was still a spinster.’ a (Duos annos faket) issa fit galu vakiana. b No’appo galu mandicatu. ‘I have not eaten yet.’ c (In dek’annos) amus a éssere galu inoke. ‘(In ten years time) we will still be here.’ Derettu and luego indicate an event time occurring shortly after a reference point, which need not be the present: (78) Cando appo lessu sa líttera, iskío ki devío andare derettu/luego. ‘When I read the letter, I knew I would have to go immediately/soon.’ As we have already noted (see 3.4.1), galu and dza (along with other aspectual adverbs like semper and mai) have a syntactic distribution which is very different from that of como, tando and other temporal expressions. Although derettu and luego show semantic affinities with galu and dza in that they denote time in a purely

< previous page

page_197

next page >

< previous page

page_198

next page >

Page 198 relative manner, their syntactic distribution is much closer to that of items of the como type; for example, they can occur sentence-initially or between the subject and finite verb or at the end of the sentence, but they do not readily occur between auxiliaries and lexical verbs.

< previous page

page_198

next page >

< previous page

page_199

next page >

Page 199 5 Pronouns 5.1 DISJUNCTIVE PRONOUNS 5.1.1 Personal pronouns The disjunctive forms of the personal pronouns show a four-way Case contrast in the first and second person singular; nominative for subjects, accusative/dative for pronouns governed by the preposition a, comitative for pronouns governed by kin ‘with’, and oblique for pronouns governed by other prepositions (see 4.3.2). The forms attested in the dialect in which our examples are cited are listed in Table 5.1. Note that when disjunctive pronouns function as direct objects, they are always preceded by a, by virtue of the prepositional accusative phenomenon discussed in 2.2.6. Consequently the items mie and tie can never occur without a, which is sometimes written as part of the pronoun: ammie, attie. Table 5.1 Disjunctive pronouns Singular Plural 1 2 3 1 2 3 m. f. m. f. Nom. jeo tue isse issa nois vois issos issas Acc./dat. mie tie isse issa nois vois issos issas Com. mecus tecus isse issa nois vois issos issas Obl. me te isse issa nois vois issos issas The Case distinctions given in Table 5.1 are not maintained fully in all dialects. In particular, Campidanese has a single second person singular form tui for all Cases, and in the first person singular the accusative/dative and comitative Cases are neutralised under a single

< previous page

page_199

next page >

< previous page

page_200

next page >

Page 200 form mimi. Even where the Case distinctions are maintained, the actual forms vary quite considerably. Thus, in Campidanese we have deu for jeo, mei for me, issu for isse, nosu for nois and bosastrus for vois. Within the Logudorese-Nuorese area we similarly find deo and dego in place of jeo, and issu instead of isse. 1 In the southern Nuorese dialects (particularly Nuoro itself) the first and second person singular forms are mime, tibe (acc./dat.) and mene, tene (obl.). As in other Romance languages, the second person plural forms are used with singular reference as polite forms (for example, when addressing strangers or elders). This usage extends to corresponding clitic and possessive forms and the use of the second person plural verb form without overt subject. A further pronoun vosté (Camp. fustei ), which functions syntactically as a third person singular form, is often used as a sign of respect when addressing persons of superior social status. Similarly, isse and (with plural reference) issos are sometimes used as respectful forms of address, as are the corresponding clitic and possessive forms. Unlike their clitic counterparts, the third person disjunctive forms are not normally used to refer to inanimate entities. This generalisation is not absolute, however, as is shown in the following attested example where issu clearly refers back to su rumanzu traduttu: (1)“in paritzos puntos de su rumanzu traduttu kie leghet at s’impressione ki issu esseret istau iscrittu dae Grassia ‘e Ledda non in limba italiana, ma in limba sarda” ‘in many places in the translated novel the reader has the impression that it was written by Grazia Deledda not in Italian, but in Sardinian’ (Pittau, Foreward to [EP: 7]) More typically, in cases where syntax or conditions of discourse require an inanimate entity to be designated by a disjunctive pronominal element, the demonstrative items custu, cussu and cuddu (and their feminine/plural variants) are used; see 5.1.2. We assume, as a working hypothesis, that the personal disjunctive pronouns listed in Table 5.1 are exhaustively dominated by the NP node and that they occupy the head N position within this constituent. Their syntactic distribution is essentially the same as that of other definite NPs except that their use is generally restricted to cases where a clitic form is unavailable, either for syntactic reasons or because of some element of emphasis which cannot be borne by a clitic. Typical cases of the first type are those where the pronoun functions as the complement of a preposition:

< previous page

page_200

next page >

< previous page

page_201

next page >

Page 201 (2) a Keljo andare kin tecus. ‘I want to go with you.’ b L’appo fattu pro issa. ‘I did it for her.’ c Maria est prus manna de me. ‘Mary is bigger than me.’ d Ses semper contra a mie. ‘You are always against me.’ In these examples, there are no clitics which can express the function of the pronoun. Consequently, the use of the disjunctive form does not necessarily convey emphasis. On the other hand, when the pronoun functions as a direct object, as in (3), or as an indirect (dative) object, as in (4), the use of a clitic form is strongly preferred: (3) a ?Appo vistu a issos. b Los appo vistos. ‘I saw them.’ (4) a ?Appo datu su dinari a issos. b Lis appo datu su dinari. ‘I gave the money to them.’ In such cases, the use of the disjunctive form always implies some element of contrast, as in the following examples: (5) a Appo vistu a issos, ma no’ a tie. ‘I saw them, but not you.’ b Imbetzes de dare su dinari a issos, lu dao a tie. ‘Instead of giving the money to them, I (will) give it to you.’ Disjunctive pronouns also occur as the complement of éssere ‘be’ in ‘identificational’ constructions (see 3.2.5): (6) a Su vinkitore est isse. ‘The winner is him.’ b Su mere so jeo. ‘The boss is me.’ Note that the pronoun assumes the nominative form and determines the agreement features of the verb. Disjunctive forms can also be used in addition to a clitic in dislocated constructions (see 7.1.1, 7.1.2): (7) a Los appo vistos, a issos. b A issos, los appo vistos. ‘I saw them.’

< previous page

page_201

next page >

< previous page

page_202

next page >

Page 202 The disjunctive pronoun in (7a) does not readily admit a contrastive interpretation of the type noted for the examples in (5), but serves rather to confirm the referent of los as the discourse topic. However, with first and second person forms, reduplication of the clitic by a disjunctive pronoun can be exploited for a contrastive effect, as in (8), which appears to be rather more natural than (5b): (8) Imbetzes de dare su dinari a issos, ti lu dao a tie. Possibly (8) is an instance of the ‘clitic-doubling’ phenomenon found in Spanish (see Jaeggli 1982 for detailed discussion). However, an alternative possibility is that ti in (8) is not simply a copy of a tie but functions as an ‘ethic’ dative denoting an ‘interested party’ who, in this case, happens to be identical to the receiver. Thus, (8) can be viewed as analogous to examples like Ti lu dao a frate tuo ‘I will give it to your brother’, where the clitic ti is clearly not a copy of the indirect object but serves to involve the hearer in the event (see 5.2.4 for further examples and discussion). This approach would explain why examples like (8) are possible only with first and second person pronouns, in so far as ethic datives appear to show the same restriction (see 5.2.4). If this analysis is tenable, we can postulate that (apart from cases of dislocation) clitic and disjunctive forms are mutually exclusive with the same grammatical function. The use of overt personal pronouns as subjects is likewise largely determined by discourse considerations such as contrast, as in (9): (9) Isse sink’est andatu, ma jeo so abbarratu inoke. ‘He went away, but I stayed here.’ However, the degree of salience which is sufficient to warrant the occurrence of an overt subject pronoun is less than that which is required in the case of disjunctive object pronouns. Often an overt subject pronoun is used when the syntactic subject is not the topic of discourse as established by preceding context or in order to indicate a shift in empathy. Thus, in so far as the first clause in (10) establishes the subject Juanne as the topic of the utterance (a judgement which might, of course, be altered by wider context), the use of the overt subject pronoun in (10a) favours the non-topical NP Gavini as antecedent whereas the null subject in (10b) is more likely to be interpreted as referring back to Juanne: (10) a Cando Juanne at vistu a Gavini, isse fit meta dispiághitu. ‘When John saw Gavin, he (= Gavin) was very unhappy.’

< previous page

page_202

next page >

< previous page

page_203

next page >

Page 203 b Cando Juanne at vistu a Gavini, fit meta dispiághitu. ‘When John saw Gavin, he (= John) was very unhappy.’ Similarly, in (11a) issa cannot normally be construed as coreferential with Maria, whereas the null subject in (11b) can refer either to Maria or to some other person who is established as the topic of discourse: (11) a Maria credet k’ issa est maláida. b Maria credet k’est maláida. ‘Mary thinks that she is ill.’ However, if issa in (11a) bears contrastive stress in the context of a proposition or assumption that Mary might believe that someone else is ill, coreference between issa and Maria is possible, as in English Mary believes that she herself is ill, though this interpretation can be made explicit by the use of a reflexive formula (see 5.3.2, examples (122)). 5.1.2 Demonstrative, interrogative and negative pronouns The demonstrative items custu, cussu and cuddu (and their feminine/plural variants) can be used without a following noun with the deictic properties described for their use as determiners in 2.1.2. ‘Bare’ demonstratives of this type can refer to animate (human) beings and to inanimate objects. The masculine forms are also used to refer to entities which lack grammatical gender; for example, propositions or events, or physical objects which have not been identified by means of a noun (as in (12c)): (12) a Cussu no’est veru. ‘That is not true.’ b Cussu est sutzessu cando fippo minore. ‘That happened when I was young.’ c Itte est custu? ‘What is this?’ When a ‘bare’ demonstrative functions as direct object, the human/non-human distinction is signalled systematically by the presence or absence of accusative a, as shown in (13): (13) a Appo vistu custu/cussu/cuddu. ‘I saw this/that thing.’

< previous page

page_203

next page >

< previous page

page_204

next page >

Page 204 b Appo vistu a custu/cussu/cuddu. ‘I saw this/that person.’ We find a similar pattern with interrogative and negative items; accusative a normally occurs before the personal forms kie ‘who(m)’ and neune/nemos ‘nobody’, but cannot occur with the inanimate forms itte ‘what’ and nudda ‘nothing’: (14) a A kie as vistu? ‘Who did you see?’ b No’appo vistu a neune/a nemos. ‘I did not see anyone.’ (15) a Itte as vistu? ‘What did you see?’ b No’appo vistu nudda. ‘I did not see anything.’ However, the interrogative item cale ‘which’, which can also be used without a following noun, typically occurs without accusative a even when it has human reference: (16) Cale as vistu? ‘Which (person/thing) did you see?’ In our general discussion of the prepositional accusative (2.2.6) we observed that although human reference does play a marginal role in allowing the (optional) occurrence of a before certain singular human NPs introduced by a definite determiner, the principal factor which triggers the obligatory occurrence of accusative a is the absence of a determiner (e.g. before proper names and personal pronouns of the type discussed in 5.1.1). Thus, on the face of it, the distribution of accusative a in (13)–(16) does not strictly conform to the pattern described in 2.2.6. If the pronominal items in these examples are pronouns (in the sense that they occupy the head N position—like the items discussed in 5.1.1) we would expect a to occur obligatorily in all cases, regardless of the nature of the referent, by virtue of the absence of a determiner. On the other hand, if the occurrence of a in these cases is triggered directly by the [+human] feature (as in the marginal uses of accusative a before NPs of the type [DET N′]—see 2.2.6. examples (105)–(107)), the fact that a is obligatory in (13b) would not be explained. Moreover, under this approach it is not clear why cale behaves differently from kie and neune/nemos. These observations lead us to suggest that the distribution of a in

< previous page

page_204

next page >

< previous page

page_205

next page >

Page 205 (13)–(16) is not determined directly by properties of the referent, but rather that it reflects differences in the syntactic category of the items involved. Specifically, we propose that those items which take accusative a are pronouns occupying the head N position, whereas those which do not are determiners which govern a null N′, thus bringing the facts in (13)–(16) into line with the general pattern whereby accusative a occurs with NPs which lack a determiner. Thus, cale, itte and nudda can be classified unambiguously as determiners, kie and neune/nemos as pronouns, whereas the demonstratives have pronoun status when referring to humans but determiner status otherwise. This classification of the interrogative and negative items is supported by the fact that cale and (more marginally) itte and nudda can function as determiners with an overt head N′ (which may be human or inanimate), but kie and neune/nemos can never be used in this way:2 (17)a cale mákkina ‘which car’, cale pastore ‘which shepherd’; itte problema ‘what problem’, itte professore ‘what teacher’; nudda risposta ‘no answer’, nudda duttore ‘no doctor’. b *kie professore lit. ‘who teacher’; *neune/nemos duttore lit. ‘nobody doctor’. With regard to the demonstratives, the idea that these items have a dual syntactic classification is relatively uncontroversial—clearly they are determiners when followed by an overt N′ and traditionally they are classed as pronouns when used alone. Consequently, our proposal does not introduce a new categorial distinction but simply redefines the criteria on which this distinction is based; essentially, pronoun status of ‘bare’ demonstratives is restricted to cases where they have human reference. The principal motivation for this redefinition is that it is consistent with an optimally simple account of the distribution of accusative a, as discussed above. However, in some dialects (mainly Logudorese), there is morphological evidence in favour of this proposal. In these dialects, the masculine singular forms custu, cussu and cuddu have the variants custu, cusse and cudde, but these variants appear to be used only as ‘bare’ demonstra-tives with human reference—i.e. as pronouns according to our proposal. Our conclusions regarding the correlation between syntactic status (as evidenced by the distribution of accusative a) and referential properties of the items discussed in this section are summarised in

< previous page

page_205

next page >

< previous page

page_206

next page >

Page 206 Table 5.2 Syntactic category and referential properties DET+N′ DET+ PRONOUN Demonstratives ±human -human +human itte ±human -human — kie — — +human nudda ±human -human — neune/nemos — — +human cale ±human ±human — Three generalisations emerge from the observations recorded in Table 5.2.: I When these items function as determiners accompanied by an overt N′, they are unmarked with respect to the [±human] distinction (a generalisation which is also valid for other deter-miners such as definite/indefinite articles, quantifiers, etc.). II When these items have the syntactic status of pronouns, they must have human reference (a fact which may be related to our observation in 5.1.1 that personal disjunctive pronouns typically refer to humans). III With the exception of cale, the use of these items as determiners governing a null N′ is restricted to cases of non-human reference. An alternative formulation of generalisation III is that, in the relevant cases, human reference must be expressed by means of a pronoun wherever possible, in which case the exceptional behaviour of cale might be accounted for by the absence of a pronoun which corresponds to cale in the way that kie and neune/nemos correspond to itte and nudda3 A further possibility which arises from the above discussion is that the referential properties [±human] of the items in question need not be stipulated, but can be derived from their syntactic classification by the generalisations I–III. Rather than classifying itte and nudda as non-human (a suspect claim in view of the examples in (17a) where these items are followed by a human noun) and kie and neune/nemos as specifically human forms, we can simply classify the former as determiners and the latter as pronouns. Thus itte and nudda occur as determiners with both human and non-human nouns (via generalisation I) but occur alone only with non-human reference (via generalisation III), whereas kie and neune/nemos can and must have human reference by virtue of their pronominal status (via generalisations I and III). The referential range of demonstratives can likewise be

< previous page

page_206

next page >

< previous page

page_207

next page >

Page 207 accounted for under the hypothesis that these items have both pronoun and determiner status. A further observation which may help to elucidate some of the issues raised above is that all the items discussed in this section can be followed by the adjective átteru ‘other’, though with different consequences regarding the prepositional accusative phenomenon. As we might expect, when cale, itte and nudda occur with átteru, they are not introduced by a: (18) a Cale átteru as vistu? ‘Which other (person/thing) did you see?’ b Itte átteru as vistu? ‘What else did you see?’ c No’appo vistu nudda átteru. ‘I saw nothing else.’ In these cases, it seems reasonable to suppose that cale, itte and nudda retain their determiner status and that átteru functions as an adjective modifying a null N. In comparable cases with kie and neune/nemos, accusative a is required: (19) a A kie átteru as vistu? ‘Who else did you see?’ b No’appo vistu a neune átteru. ‘I saw nobody else.’ These cases can be accommodated by postulating that átteru is a postposed adjective modifying the pronoun in the head N position. Interestingly, accusative a does not normally appear before demon-stratives followed by átteru even with human reference: (20) No’appo vistu a isse, ma appo vistu cudd’átteru. ‘I did not see him, but I saw that other (person).’ The first point to note about this example is that it confirms our earlier claim that the use of accusative a with demonstratives is not directly determined by the [+human] feature. On the other hand, in order to maintain our account of the distribution of accusative a, we must suppose that cuddu in (20) is a determiner which, moreover, governs a null N (if we are correct in treating átteru as an adjective), despite its human interpretation. What we would like to suggest is that this example illustrates the ‘whenever possible’ condition in the revised formulation of generalisation III above. The leading idea is that human reference in this case cannot be expressed by a demon-

< previous page

page_207

next page >

< previous page

page_208

next page >

Page 208 strative pronoun, because such pronouns cannot be modified by adjectives like átteru, so that cuddu must be treated as a determiner in the same way as cale, itte and nudda in (18). However, this proposed restriction cannot be stated in terms of pronouns in general, because of the examples in (19). Moreover, átteru can modify first and second person plural disjunctive pronouns: nois átteros ‘we others’, vois átteros ‘you others’. Nevertheless, this restriction does seem to hold for third person definite pronouns, a class which plausibly includes the demonstratives. Thus we do not find *isse átteru lit. ‘him other’ or *issos átteros lit. ‘them others’. In such cases, the definite article (or determiner) is used: s’átteru, sos átteros ‘the other(s)’. Given this restriction, we may take cuddu in (20) to be a determiner on a par with the definite articles su and sos in the examples just quoted, thus explaining the absence of accusative a in this case. 5.1.3 Pronominal uses of quantifiers and indefinite expressions The universal quantifier tottu shows the same properties with regard to accusative a as the demonstratives discussed in 5.1.2; i.e. it requires a when used alone with human reference, but does not take a otherwise: (21) a Appo salutatu a tottu. ‘I greeted everybody.’ b Appo mandicatu tottu. ‘I ate everything.’ Note that accusative a does not occur when tottu quantifies an overt NP: (22) a Appo salutatu tottu sos óspites. ‘I greeted all the guests.’ b Appo mandicatu tottu su pane. ‘I ate all the bread.’ Similarly, tottu does not take accusative a when it is accompanied by an object clitic, as in (23): (23) a Los appo salutatos tottu. ‘I greeted them all.’ b L’appo mandicatu tottu. ‘I ate it all.’ Pursuing the approach adopted in 5.1.2, we postulate that tottu is a pronoun in (21a) but a quantifier in all the other examples above. In particular, we assume that tottu has the same status in (21b) as in (22b) except that the quantified NP is null (i.e. we have the structure

< previous page

page_208

next page >

< previous page

page_209

next page >

Page 209 [NP tottu [NP θ }]), whereas in (21a) tottu occupies the head N position [NP [N tottu ]]. We also assume that the examples in (23) are parallel to those in (22) except that the quantified NP is realised as a clitic. Note that this classification conforms to the generalisations proposed in 5.1.2 (see p. 206) regarding the referential properties of determiners and pronouns: I when tottu quantifies an overt NP (including clitics), it is [+human]; II when it has pronoun status, it must have human reference; III when it quantifies a null NP, its reference is non-human. Similarly, the distribution of accusative a in (21)–(23) follows the pattern outlined in 5.1.2; a occurs regularly before the pronoun tottu but not before direct object NPs introduced by the quantifier tottu . In the case of tottu this categorial distinction is supported by syntactic evidence. As we pointed out in 2.1.2, tottu can be ‘floated’ to the position between the auxiliary and lexical verb in cases like (24): (24) a Los appo tottu salutatos. b L’appo tottu mandicatu. The same floating process can occur in (21b), but not in (21a) (even if a is omitted): (25) a *Appo (a) tottu salutatu. b Appo tottu mandicatu. In other words, it seems that ‘floated’ tottu can be [±human] when accompanied by a clitic, but must be [−human] when no clitic is present. In terms of our analysis, this rather untidy generalisation can be captured simply by postulating that the ‘floating’ process applies only to quantifiers, not to pronouns.4 Cardinal numerals (including unu ‘one’) and indefinite determiners such as meta ‘much, many’, pacu ‘little, few’ and the plural forms of átteru ‘other’ are not normally used alone with a pronoun-like function such as that found in English I bought several; three were broken; others were damaged . In the absence of an overt N′, these items are accompanied by the partitive clitic nde which, intuitively, acts as a substitute for the missing N′: (26) a Nde appo comporatu átteros. ‘I bought others (of them).’ b Nde sun vénnitos tres. ‘Three (of them) came.’ c Bi nd’at arrivatu pacos. ‘There arrived few (of them).’

< previous page

page_209

next page >

< previous page

page_210

next page >

Page 210 Partitive nde can only be associated with NPs directly governed by the verb (i.e. with direct objects, postverbal subjects of unaccusative verbs and complements of impersonal constructions, as in (26)), but not preverbal subjects or objects of prepositions: (27) a Metas (*nde) an cantatu. ‘Many (of them) sang.’ b (*Nde) appo faeddatu de tres. ‘I talked about three (of them).’ Although the variants of (27) without nde are marginally acceptable, such constructions tend to be avoided either by supplying an appropriate noun or NP to denote the set over which the quantifying expression ranges or, in the case of subjects, by recourse to an impersonal existential construction, as in (28): (28) Bi nd’at metas ki an cantatu. ‘There are many (of them) who sang.’ Alongside átteros, -as, as used in (26a), which we take to be a determiner in this case just as in NPs of the type átteros libros ‘other books’ (see 2.2.5), we also have the form atter ‘others’ which is used exclusively with human reference. This item bears all the hallmarks of a pronoun according to the criteria proposed above. In particular, atter does not co-occur with partitive nde and takes accusative a when used as a direct object: (29) No’appo vistu a isse, ma appo vistu a atter. ‘I have not seen him, but I have seen other people.’ This item is syntactically singular, though it is typically used with plural reference. It can also be used in interrogative and conditional constructions with a value similar to English ‘anyone’: (30) a B’at telefonatu atter? ‘Has anyone telephoned?’ b Si bi venit atter, náralis de colare prus tardu. ‘If anyone comes, tell them to pass by later.’ Note that atter in (30) does not necessarily serve to restrict reference to persons other than those who are already known to have telephoned or arrived (as would be implied in English by ‘anyone else’); rather, the use of this item seems to be motivated by the fact that the speaker and hearer (and possibly other people present) are naturally excluded as potential referents. The indefinite determiner carki ‘some’ cannot constitute an NP on

< previous page

page_210

next page >

< previous page

page_211

next page >

Page 211 its own, but must be combined with the item unu ‘one’, the resulting expression being frequently written as a single word carcunu (sometimes calicunu). This item has two distinct uses. When it occurs without an antecedent in the discourse, it has human reference (= ‘somebody’) and is invariable for both number and gender, as in (31): (31) Carcunu est vénnitu. ‘Somebody came.’ However, when it quantifies over a set of entities already mentioned in the discourse (or, possibly, present in the extralinguistic context), it may have inanimate reference and inflects for gender, as in (32): (32) Appo deke pinnas, ma carcuna no’iscrivet. ‘I have ten pens, but some (possibly only one) do not write.’ Note that in the second case carcunu is syntactically singular, though potentially plural in reference (as in (32)). For some speakers, when carcunu occurs as a direct object it takes accusative a in the first use and is accompanied by nde in the second: (33) Appo vistu a carcunu. ‘I saw somebody.’ (34) Appo deke pinnas, ma nde appo secatu carcuna. ‘I have ten pens, but I have broken some (one) of them.’ Judgements of such examples are somewhat uncertain; some speakers allow, or even prefer, the variants of (33)–(34) without a or nde. A similar situation is found with the distributive, universal quantifiers dondzi and cada ‘each, everybody’ which also must combine with unu when there is no following noun. In the absence of contextual cues, the items dondzunu and cadaunu must have human reference (= ‘everybody’), occur only in the masculine form and typically require accusative a when used as direct objects (though again there is some uncertainty of judgement on this point): (35) Appo salutatu a ‘I greeted everybody.’ On the other hand, when these items have an antecedent, they inflect for gender and may have inanimate reference (like the corresponding use of carcunu in (34)), but, unlike carcunu, they are not accompanied by the clitic nde:

< previous page

page_211

next page >

< previous page

page_212

next page >

Page 212 (36) Appo deke pinnas, ma appo secatu ‘I have ten pens, but I have broken every one.’ The phenomena illustrated in (31) to (36) may lend themselves to a structural account of the sort advocated above for other items, though some degree of caution is in order in view of the uncertainty concerning some of the relevant examples. The properties of these items when used to refer back to an antecedent are consistent with the view that these expressions are syntactically complex, with the quantifier element (carki, cada or dondzi) occupying the determiner position and unu, -a functioning either as a pro-form in the head N′ position or as an adjective modifying a null N (the availability of these two possibilities concerning the status of unu may be relevant to the variation with regard to the use of nde with carcunu in cases like (34)). On the other hand, the cases which do not involve an antecedent appear to favour an analysis which treats these expressions as single lexical items, pronouns occupying the head N position, at least for those speakers who require accusative a in cases like (33) and (35). The item unu, -a ‘one’ can be used on its own in the same way as other numerals and indefinite quantifiers such as meta, pacu, etc., typically with the partitive clitic nde, as in examples (26)–(28). It can also be introduced by the definite article to partition a set of entities, usually in contrast to a subset denoted by the item átteru: (37) Jeo pico s’unu, tue picas s’atteru. ‘I will take one, you will take the other.’ The distribution of unu is however more restricted than that of átteru. Apart from forms such as carcunu and dondzunu discussed above, unu can only be introduced by a definite article, or by no article at all, whereas átteru appears to allow the full range of determiners; e.g. átteru can be introduced by a demonstrative, but not unu: cudd’átteru ‘that other (one)’, *cudd’ unu ‘that one’. There is no pronoun corresponding to the generic use of English ‘one’. Typically, generic statements are rendered by means of a pronominal verb construction with the clitic si (see 3.2.8). Alternatively, generic reference can be expressed by the phrase sa pessone, lit. ‘the person’: (38) Cando sa pessone est maláida, non travallat. ‘When one is ill, one does not work.’ Although this phrase is compositionally transparent and can be

< previous page

page_212

next page >

< previous page

page_213

next page >

Page 213 treated as an ordinary NP rather than a pronoun, its generic use tends to be restricted to subject position in much the same way as generic ‘one’ in English; e.g. in (39), sa pessone can only be interpreted with specific reference: (39) ?Juanne non devet madzare sa pessone. ‘John should not hit the person.’ 5.2 CLITICS 5.2.1 Forms and basic uses The Nuorese-Logudorese personal clitic forms are listed in Table 5.3. In Campidanese, the initial /l/ of the third person non-reflexive forms is replaced by a retroflex geminate /dd/: ddu, dda, ddus, ddas, ddi and ddis. Also, in Campidanese the first and second person plural forms nos, bos are used only in postverbal position (with imperatives and present participles), but are replaced by si preverbally (the proclitic forms nosi and osi are attested in some dialects, though they appear to be becoming obsolete; see Blasco-Ferrer 1986:110). Table 5.3 Clitic pronouns 1 2 3 1 2 3 m. f. refl. m. f. refl. Acc. mi ti lu la si nos bos los las si Dat. mi ti li li si nos bos lis lis si As can be seen from Table 5.3, the third person reflexive is invariable for number, gender and Case, while the first and second person forms (which can also function as reflexives) are neutral with regard to the accusative/dative contrast. Note also that the gender distinction is neutralised with the dative forms li and lis though in some Campidanese dialects (just as in varieties of Spanish which exhibit the laismo phenomenon) a gender distinction is introduced by the use of the accusative forms dda and ddas to denote feminine indirect objects. A similar tendency to make a gender distinction in the dative can be discerned in Logudorese-Nuorese dialects in marginal cases where an invariable suppletive form bi is substituted for the dative clitic and the number and gender features of the indirect object are transferred to an accompanying accusative clitic (see 5.2.2 for examples and further discussion).

< previous page

page_213

next page >

< previous page

page_214

next page >

Page 214 The accusative clitics function as substitutes for direct object NPs, with animate or inanimate reference in the case of the third person forms. Dative clitics correspond, on the whole, to animate, indirect objects introduced by a. However, there are some cases where a dative clitic cannot be replaced by an animate NP introduced by a (cases which can be loosely categorised as ethic datives or datives of interest; see 5.2.3, 5.2.4). Generally, dative clitics (including first and second person forms) denote persons who are affected in some way by the event or state, for example beneficiaries, victims, receivers, experiencers. Thus, although movement verbs typically take a Goal complement introduced by a, this complement cannot normally be expressed by a dative clitic: (40) a *Juanne m’est vénnitu. ‘John came to me.’ b *Maria t’est andata. ‘Mary went to you.’ c *Pretu lis est arrivatu. ‘Peter arrived to them.’ However, such verbs can be used with a dative clitic in more figurative uses where they express an affective relation, as can arrivare with an inanimate subject, where the Goal is clearly perceived as a ‘receiver’: (41) a Cust’idea m’est vénnitu. ‘This idea came to me.’ b Cussa camisa t’ andat bene. ‘That shirt suits you.’ (lit.‘…goes to you well.’) c Cussa líttera lis est arrivata. ‘That letter arrived to them.’ (i.e. They received that letter.’) The reflexive si always refers back to the subject of the immmediate clause, either reflexively or reciprocally, except in Campidanese where si can also function as a non-reflexive first or second person plural clitic (see above) or as a suppletive dative form (replacing li or lis when the direct object is expressed as an accusative clitic; see 5.2.2). The first and second person clitics can also be used reflexively or, in the plural, reciprocally. In addition to the personal clitic forms given in Table 5.3, there are three ‘adverbial’ clitics, bi, nke (sometimes ke) and nde. Bi and nke are primarily locative clitics roughly equivalent to English ‘there’. Both of these items can be used, more or less interchangeably, to

< previous page

page_214

next page >

< previous page

page_215

next page >

Page 215 denote static location, though some speakers show a preference for one or other form. In cases of dynamic location, bi is used to denote a Goal (‘to there’) whereas nke generally has a Source orientation (‘from there’). However, the latter generalisation is confused somewhat by dialectal and idiolectal variation, in that speakers who favour nke to bi as a static Locative clitic also allow a Goal interpretation for nke. This is particularly apparent in the Campidanese dialects which lack the clitic bi and which use (n)ci (= nke) for all cases of location (static, source and goal). A further complicating factor is that, for many speakers, nde can be used as a Source clitic (as an alternative to nke), thus favouring the extension of nke to the expression of Goal. In other words, we seem to have a situation where bi and nke are vying with each other for status as the unmarked locative clitic while retaining respectively a Goal or Source orientation, but in dialects where nke has become established as the general locative clitic, it can also express a Goal relation either because of the absence of bi or because of the availability of nde as a Source clitic.5 As an extension of the genuine locative uses discussed above, bi and (for some speakers) nke function as substitutes for non-locative inanimate indirect objects normally introduced by a: (42) a Pesso a cussu problema. ‘I think about that problem.’ b Bi pesso. ‘I think about it.’ (43) a So resissitu a fákere cussu. ‘I succeeded in doing that.’ b Bi so resissitu. ‘I succeeded in it.’ In Logudorese-Nuorese, bi is also used with animate reference as a suppletive dative form (in the same way as si in Campidanese) to avoid unacceptable clitic sequences (see 5.2.2). In common with dative and reflexive clitics, bi and nke have a wide range of expletive uses which are difficult to define but appear to be motivated by considerations of discourse (see 5.2.4 for discussion). A further use of these clitics, which appears to be fully grammaticalised, is as an operator in impersonal, existential constructions of the type discussed in 3.2.2 (e.g. B’at vénnitu tres pitzinnas ‘There came three girls’). The third adverbial clitic nde can be loosely characterised as a proform which replaces phrases introduced by the preposition de. Thus, in cases like the following, nde can be used as a pronominal

< previous page

page_215

next page >

< previous page

page_216

next page >

Page 216 substitute for indirect objects introduced by de (indicated by the phrases in parentheses): (44) a Nd’appo faeddatu (de cussa kistione). ‘I talked about it (of that question).’ b Nde so cuntentu (de su travallu k’appo fattu). ‘I am happy with it (of the work which I did).’ Similarly, nde can denote the complement of a direct object NP which would normally be introduced by de: (45) Nd’appo lessu su primu capítulu (de su libru). ‘I read the first chapter (of the book).’ When nde functions as an indirect complement of the main predicate, its referent may be inanimate (as in (44)) or animate, though in the latter case a disjunctive pronoun introduced by de is generally preferred unless the antecedent is particularly prominent in the discourse, as in (46): (46) Mi piaghet Maria e nde so inamoratu. ‘Mary pleases me and I am in love with her.’ However, nde does not generally allow animate reference when it functions as the complement or adjunct of an object NP. In such cases a possessive adjective is used: (47) a Appo lessu su libru de Porru. b *Nde appo lessu su libru. c Appo lessu su libru suo. ‘I have read Porru’s/his book.’ The correlation between nde and the preposition de is also reflected in the partitive use of nde illustrated in (48): (48) a Nd’appo mandicatu meta (de custu pane). ‘I ate much (of this bread).’ b Nd’appo salutatu tres (de cuddas pitzinnas). ‘I greeted three (of those girls).’ c Nd’app’a comporare una (de cussas piras). ‘I will buy one (of those pears).’ As the examples in (48) show, the quantified element denoted by nde may be animate or inanimate, countable or non-countable. Note also that partitive nde may be used without an overt quantifier or numeral, with the same range of potential referents: (49) Nd’appo mandicatu. ‘I ate some (of it/them).’

< previous page

page_216

next page >

< previous page

page_217

next page >

Page 217 At this point, the correlation between nde and the preposition de begins to break down, at least at a superficial level, in so far as full indefinite NPs in direct object position are not introduced by de: (50) a Appo mandicatu petha. b *Appo mandicatu de (sa) petha. ‘I ate meat.’ Similarly, in cases where nde refers to an indefinite set or substance quantified by an overt determiner or numeral, it corresponds superficially to an N′ rather than an NP introduced by de:6 (51) a Juanne connosket pacos istudientes. Juanne nde connosket pacos. ‘John knows few (students).’ b Gavini at bitu meta vinu. Gavini nd’at bitu meta. ‘Gavin drank a lot (of wine).’ A further instance where nde fails to correspond to de is in the locative source use mentioned above, as in (52) where nde corresponds to dae rather than de:7 (52) a Juanne nd’est issitu. b Juanne est issitu dae/*de sa domo. ‘John came out of it/out of the house.’ When used partitively or as a substitute for a complement of a noun (i.e. in those cases where nde replaces a part of an NP), nde can only be related to an NP which is governed by the verb; e.g. a direct object or a postposed subject. In particular, nde cannot be interpreted as corresponding to part of a preverbal subject or an indirect object: (53) a *Tres nde sun vénnitos. ‘Three of them came.’ b *Su primu capítulu nd’est interessante. ‘The first chapter of it is interesting.’ (54) a *Cussu libru nd’est piághitu a battor. ‘That book pleased four of them.’ b *Nde so pessande a su primu capítulu. ‘I am thinking about the first chapter of it.’ To express the propositions in (53b) and (54b), the clitic nde is simply omitted (Su primu capítulu est interessante, So pessande a su primu capítulu) whereas (53a) and (54a) would normally be expressed by an

< previous page

page_217

next page >

< previous page

page_218

next page >

Page 218 impersonal construction (Bi nd’at vénnitu tres ‘There came three of them’ Bi nd’at battor ki cussu libru lis est piághitu ‘There are four of them whom this book pleased’—for the syntax of the relative clause in the latter example, see 6.2.1). 5.2.2 Order and compatibility of clitics Clitics precede finite verb forms and infinitives but follow present participles and imperatives. In both positions, clitics occur in a fixed order given in Table 5.4. Table 5.4 Order of clitics I II III IV V VI mi lu ti li la si bi nke nde los nos lis las bos Essentially, clitics which are neutral with regard to the accusative/ dative distinction come first, followed by the adverbial clitics in the order bi, nke, nde, then the specifically dative forms and, finally, the accusatives. With regard to the adverbial clitics, Table 5.4 differs slightly from that given in Jones (1988a:337), where nde and nke were placed in the same column, suggesting that they cannot occur together. However, subsequent research reveals that such combinations are possible and indeed that all three adverbial clitics can occur together in certain circumstances, as in (55), where bi has an existential function (typical of impersonal constructions, see 3.2.2), nke denotes the Source and nde has a partitive value (e.g. Three mice came out of the hole’): (55) Bi nke nd’at issitu tres. ‘There came three of them out of there.’ There are a number of restrictions on the combination of accusative and dative clitics. When a clitic from column I is used as a direct object, a pronominal indirect object is normally realised as a disjunctive form rather than a dative clitic, as in (56), suggesting that there is a restriction which prevents items from column I from being combined with each other or with items from column V:

< previous page

page_218

next page >

< previous page

page_219

next page >

Page 219 (56) Mi so presentatu a . ‘I introduced myself to you/him.’ Nevertheless, this restriction is not absolute, as can be seen in (57a) where clitics from columns I and V are combined and in (57b, c) where we have two clitics from column I: (57) a Si li leghíat in cara ki teníat sa gurpa. lit. ‘It read itself to him in face that he had the guilt.’ ‘One could see in his face that he was guilty.’ b Maria si ti pompiaíat. ‘Mary was staring at you.’ c Non ti mi mandikes! ‘Do not eat me up!’ Clitic combinations of this type appear to be predominantly cases where the dative element is not an indirect object of the verb. Thus, in (57a) li is a ‘dative of interest’ which also serves to indicate the possessor of the body-part cara (see 5.2.3), whereas si in (57b) and ti in (57c) can be classed as reflexive ethic datives whose function appears to be to intensify the action denoted by the verb in some way (e.g. giving the reading ‘to stare’ for pompiare rather than simply ‘to look at’ in (57b)); see 5.2.4. It may be significant that these are precisely the functions of dative clitics which cannot normally be expressed by means of an NP introduced by a. The conditions governing the order of clitics from column I are not absolutely clear, but, as far as we can ascertain, order appears to be determined by person rather than grammatical function; generally the third person reflexive comes first, then second person clitics and finally first person forms.8 Thus, if we compare (58) with (57c), we see that the order of ti and mi remains the same though their grammatical functions are reversed: (58) Emmo, ti mi mandico. ‘Yes, I will eat you up.’ A further restriction on clitic combinations is revealed in cases where both the direct and indirect object are third person, nonreflexive. To our knowlege, no dialects allow straightforward combinations of clitics from columns V and VI (e.g. *li lu, *lis lu, *li los, etc.). In some dialects (e.g. Nuoro, see Pittau 1972:83) such combinations are represented by single composite items: liu (=li+lu or lis+lu ), lia (=li+la or lis+la), lios (=li+los or lis+los), lias (=

< previous page

page_219

next page >

< previous page

page_220

next page >

Page 220 li+las or lis+las). More typically, in such cases a suppletive clitic (bi in Logudorese-Nuorese, si in Campidanese) is used in place of the dative: (59) a Bi l’appo datu. (Log.-Nuor.) b Si dd’appo donau. (Camp.) ‘I gave it to him/her/them.’ In dialects which adopt bi as the suppletive dative form, it is quite common for number and gender features associated with the indirect object to be transferred to the accusative clitic, particularly in cases where the direct object is an entity which lacks inherent number or gender features (e.g. a proposition). Thus, although a sentence like Narrabílu! can in principle mean Tell it to him/her/them!’, with bi replacing li or lis (both of which are gender-neutral), many speakers prefer the reading where the indirect object is masculine singular (‘Tell it to him!’) and use formulae such as those in (60) in cases where the indirect object is feminine and/or plural: (60) a Narrabíla! ‘Tell it to her!’ b Narrabílos! ‘Tell it to them (m.)!’ c Narrabílas! ‘Tell it to them (f.)!’ Also attested are cases where suppletive bi occurs with the dative plural lis (ostensibly replacing the accusative clitic rather than the dative) in order to specify plural reference of the indirect object: (61) Narrabílis! ‘Tell it to them!’ Some grammarians appear to dismiss cases like (60) and (61) (particularly the latter) as analogical mistakes (see Blasco-Ferrer 1986:211, n. 42). However, the phenomenon is sufficiently widespread to deserve a more probing analysis. From a functional point of view, this usage can be seen as a response to the need to maintain a number distinction with respect to the indirect object (li versus lis) which would otherwise be neutralised by the suppletive form bi—the gender distinction in (60) being, as it were, an added bonus which results from the transfer of the number feature to the accusative clitic. From a more formal perspective, the phenomena in (60)–(61) may provide some support for the view espoused in much recent work within the Government-Binding framework that personal clitics should be

< previous page

page_220

next page >

< previous page

page_221

next page >

Page 221 treated as object agreement particles (akin to subject agreement inflection on finite verbs) rather than as pronouns attached to the verb. If clitics were simply pronouns, we would expect them to encode features such as Case, number and gender compositionally, just as disjunctive pronouns do. On the other hand, the spreading of such features across the clitic sequence in cases like (60)–(61) is reminiscent of the situation common in inflectional morphology where, for instance, tense/mood features and person/number features in verbal inflection often merge. Pursuing this approach, it might be argued that the sequence bi las in (60c) or bi lis in (61) has coalesced into a single particle which indicates that the verb has two definite complements one of which is plural (and in the former case is also specified as feminine). 5.2.3 Datives of interest The ‘dative of interest’ phenomenon is illustrated in (62) where a dative clitic presents an entity as being affected in some way by the event but does not fulfil a θ-role which is implicit in the meaning of the verb, pulire ‘clean’ and disinnare ‘draw’ being two-place predicates, unlike the verbs in (63) which select an indirect (dative) object as one of their arguments: (62) a Lis appo pulitu su lavandinu. ‘I cleaned the sink for them.’ b Ti disinno unu ritrattu. ‘I (will) draw you a portrait.’ (63) a Lis dao unu regalu. ‘I (will) give them a present.’ b Ti mustro una fotografía. ‘I (will) show you a photograph.’ We acknowledge that there are some cases, like those in (64), which are unclear with respect to this distinction: (64) a Ti cómporo unu regalu. ‘I (will) buy you a present.’ b Ti lego un’istória. ‘I (will) read you a story.’ c Ti canto una cathone. ‘I (will) sing you a song.’ Like the dative of interest cases in (62), these examples are semantically complete and syntactically well formed without the dative

< previous page

page_221

next page >

< previous page

page_222

next page >

Page 222 element. However, they also resemble genuine indirect object constructions like (63) in that the dative element can be realised as a+ NP, a possibility which is not readily available with datives of interest of the type in (62): (65) a Appo datu unu regalu a su pitzinnu. ‘I gave a present to the boy.’ b Appo mustratu sa fotografía a Maria. ‘I showed the photograph to Mary.’ (66) a Appo comporatu unu regalu a su pitzinnu. ‘I bought a present for the boy.’ b Appo lessu un’istória a sa pitzinnedda. ‘I read a story to the little girl.’ c Appo cantatu una cathone a sos óspites. ‘I sang a song to the guests.’ (67) a ??Appo pulitu su lavandinu a cudda fémina. ‘I have cleaned the sink for that woman.’ b ??Appo disinnatu unu ritrattu a su mastru. ‘I drew a portrait for the teacher.’ For the moment, we leave open the status of the dative elements in (64) and (66). A more fundamental issue is the extent to which the unacceptability of the non-clitic forms in (67) can be attributed to their status as datives of interest. In Sardinian, as in other Romance languages, a dative clitic is often used to identify the possessor of a bodypart denoted by the direct object: (68) a Lis appo lavatu sas dentes. ‘I brushed (lit. ‘washed’) their teeth.’ b Juanne m’at toccatu sa manu. ‘John shook (lit. ‘touched’) my hand.’ c T’app’a secare sos pilos. ‘I will cut your hair.’ In such cases, phrases of the type a+NP can be used in place of the clitic: (69) a Appo lavatu sas dentes a sos pitzinnos. ‘I brushed the boys’ teeth.’ b Juanne at toccatu sa manu a su mere. ‘John shook the boss’s hand.’ c App’a secare sos pilos a Maria. ‘I will cut Mary’s hair.’

< previous page

page_222

next page >

< previous page

page_223

next page >

Page 223 Intuitively, the verbs in (68)–(69), like those in (62), are two-place predicates which do not normally require an indirect object even though in these cases the presence of a dative element is indispensible for a natural interpretation (without it, the direct object can only be interpreted as referring to a disembodied body-part, as in ‘I brushed the teeth’). Standard tests for constituency show that the a phrase in (69) does not form part of the direct object NP. For instance, the sequence sas dentes a sos pitzinnos cannot be fronted as a unit, though sas dentes and a sos pitzinnos can be fronted separately: (70) a *Sas dentes a sos pitzinnos appo lavatu. b Sas dentes appo lavatu a sos pitzinnos. c A sos pitzinnos appo lavatu sas dentes. Similarly, the two items can be pronominalised independently, as in (71) where bi is the suppletive dative form corresponding to a sos pitzinnos and las corresponds to sas dentes: (71) Bi las appo lavatas. This example also shows that the dative clitics in (68) cannot be analysed as corresponding simply to an adnominal possessive phrase—compare constructions of the type su primu capítulu de su libru ‘the first chapter of the book’ where the de phrase can be pronominalised by nde (Nd’appo lessu su primu capítulu ‘I read the first chapter of it’) but the two elements cannot be pronominalised independently (*Nde l’appo lessu). Given that the datives in (68)–(71) are neither subcategorised complements of the verb nor adnominal modifiers, we suggest that they are essentially ‘datives of interest’ which also serve to identify the possessor of a body-part. Like the datives of interest in (62), they must denote persons who are affected (beneficially or adversely) by the event, as shown by the oddness of the following examples where the nature of the verb does not favour such an interpretation: (72) a ??Li vido sa cara. ‘I can see his/her face.’ b ??L’appo connottu sa voke. ‘I recognised his/her voice.’ In such cases, an adnominal possessive would be used: Vido sa cara sua, Appo connottu sa voke sua. Conversely, the use of an adnomi-nal possessive in place of the dative construction in cases like (68) suggests the absence of any affective relation, often giving a

< previous page

page_223

next page >

< previous page

page_224

next page >

Page 224 pragmatically anomalous effect; e.g. ?Appo lavatu sas dentes issoro ‘I washed their teeth’, Juanne at toccatu sa manu mea (this could only mean ‘John touched my hand’, perhaps accidentally). Moreover, the possessor of a body-part can be indicated by items other than a dative, for example, by a subject (if the verb denotes an ‘internal action’, where the body-part is under the direct control of the possessor9) as in (73), or in some cases by a direct or indirect object subcategorised by the verb, as in (74): (73) a Appo tuncatu sa bucca. ‘I closed my mouth.’ (lit. ‘I closed the mouth.’) b Maria at tuncatu sa janna kin su pede. ‘Mary shut the door with her foot.’ (lit. ‘…with the foot.’) (74) a Appo connottu a Gavini a sa voke. ‘I recognised Gavin by his voice.’ (lit. ‘…the voice.’) b Juanne at fertu a Maria a su bratzu. ‘John wounded Mary in the arm.’ c Appo datu unu corfu a su cubitale a Pretu. ‘I dealt Peter a blow in the elbow.’ Informally, body-part NPs which do not contain an overt possessive item must be linked to an antecedent which indicates the possessor, datives of interest being potential antecedents alongside subjects and complement NPs. Having argued that possessive datives are essentially datives of interest, we must now account for the fact that non-clitic expressions of the type a+NP can function as datives of interest in cases like (69) but not in cases like (67). We suggest that the acceptability of the non-clitic dative in (69) is related to its ancillary possessive function; more specifically, dative clitics can designate any entity which is affected by the event, whereas a+NP is subject to the additional constraint that it must enter into a salient possessive relation. A possessive interpretation for datives of interest is also possible (though not obligatory) with some NPs which do not refer to body-parts: (75) a T’appo pulitu sa domo. ‘I have cleaned the/your house for you.’ b Lis appo accontzatu sas iscarpas. ‘I mended the/their shoes for them.’ The NPs which allow this interpretation do not form a well-defined class but can be characterised pragmatically as denoting entities

< previous page

page_224

next page >

< previous page

page_225

next page >

Page 225 which are typically or frequently identified in terms of their possessor, unlike the direct objects in (62). When this possessive interpretation obtains, the use of a non-clitic dative of interest is reasonably acceptable, significantly more so than in (67): (76) a ?Appo pulitu sa domo a cudda fémina. ‘I have cleaned that woman’s house for her.’ b ?Appo accontzatu sas iscarpas a sos pitzinnos. ‘I mended the boys’ shoes for them.’ Note that the possessive interpretation is obligatory in (76), though it is optional in (75), an observation which supports our hypothesis that non-clitic datives of interest must be construed as possessors as well as affected entities. A rather different range of cases where a dative of interest has a possessive-like function is illustrated in (77): (77) a Su theraccu m’at ghettatu una tassa de vinu. ‘The servant poured me a glass of wine.’ b Lis appo secatu una thadza de pane. ‘I cut them a slice of bread.’ c T’appo tódditu carki frores. ‘I have picked you some flowers.’ In these examples, the dative does not serve to identify the direct object entity in terms of its possessor, but denotes the person to whom it is destined (i.e. loosely speaking, its ultimate possessor). In such cases, the use of a non-clitic dative is again reasonably acceptable, though a ‘clitic-doubled’ (dislocated) construction is often preferred (e.g. Su theraccu lis at ghettatu una tassa de vinu, a sos óspites is rather more natural than (78a)): (78) a Su theraccu at ghettatu una tassa de vinu a sos óspites. ‘The servant poured the guests a glass of wine.’ b Appo secatu una thadza de pane a sos pitzinnos. ‘I cut the boys a slice of bread.’ c Appo tódditu carki frores a cussa pitzinna. ‘I picked some flowers for that girl.’ We assume that the datives in (77)–(78) are not subcategorised indirect objects but instances of the dative of interest where the effect on the entity is most naturally construed as one of ‘possession’ and that it is this ‘possessive’ aspect which allows the possibility of a non-clitic dative in (78). Support for this hypothesis can be drawn from the fact that the datives in (78) must have the possessive

< previous page

page_225

next page >

< previous page

page_226

next page >

Page 226 interpretation outlined above, whereas alternative readings are potentially available for the examples in (77); for example, with a little imagination (77b) could be taken to mean that I cut a slice of bread to show them how it should be done, to save them the effort or even to spite them in some way, but no such interpretations are available for (78b). A further complication is that in some cases (such as (77c), (78c)) the dative can be interpreted as a ‘dispossessed’ entity (e.g. ‘I picked some flowers from your/the girl’s garden’). This possibility may be related to the fact that a dative of interest can have an adversative interpretation under appropriate pragmatic circumstances, as in (79): (79) T’appo secatu custa tzíkkera. lit. ‘I have broken you this cup.’ Also relevant is the fact that some verbs (e.g. furare ‘to steal’) assign a’dispossessive’ interpretation to their dative indirect object: (80) a Appo furatu carki frores a cussa pitzinna. ‘I stole some flowers from that girl.’ b T’appo furatu carki frores. ‘I stole some flowers from/for you.’ Note that the clitic in (80b) can denote either a source or a beneficiary, whereas a+NP in (80a) can only have the former interpretation (in contrast to the potentially ambiguous example (78c)). We postulate that furare (but not tóddere ‘to pick, pluck’) optionally selects an indirect object denoting the dispossessed entity; i.e. under the ‘source’ interpretation the datives in (80) are arguments of the verb. The availability of a benefactive reading for (80b) but not for (80a) bears out our general observation that clitics are more amenable to a ‘dative of interest’ reading than phrases of the type a+NP; the clitic in (80b) can be treated as an incidental dative even though it potentially fills an argument slot of the verb, whereas a+NP in (80a) must be treated as an argument of the verb. The potential ambiguity of (78c) follows from the fact that under both the benefactive and adversative interpretation a+NP has non-argumental status involving some notion of possession. All the examples of datives of interest presented so far involve transitive verbs. Indeed, the possibility of a dative of interest seems to be dependent on the presence of a complement of some sort. In particular, such datives (even clitics) do not occur readily with genuine intransitive verbs: (81) a *Juanne ti ballat. ‘John (will) dance for you.’

< previous page

page_226

next page >

< previous page

page_227

next page >

Page 227 b ??Su pitzinnu m’at cascatu. ‘The boy yawned for/at/on me.’ c *Maria lis at travallatu. ‘Mary worked for them.’ Possible exceptions are verbs denoting communicative gestures such as sorrídere ‘smile’ which readily take a dative clitic or a+NP, though it is not clear whether this element is an optional complement or a dative of interest (see 3.2.1 for further discussion of such verbs). Also, incidental dative clitics are generally acceptable (subject to pragmatic considerations) with unaccusative and (neutral) pronominal verbs: (82) a Su frate m’est mortu. ‘My brother died (on me).’ b Su pallone si l’est isgonfiatu. (l’= li) ‘The balloon deflated on him/her.’ Such cases are consistent with the condition suggested above, at least at an abstract level, given the hypothesis that the superficial subjects of these verbs occupy the direct object position in underlying structure (see 3.2.3, 3.2.6), as in the equally acceptable examples in (83): (83) a M’est mortu su frate. b Si l’est isgonfiatu su pallone. However, postposition (inversion) of the subject in cases like (81) does not appear to improve acceptability: (84) a *Ti ballat Juanne. b ??M’at cascatu su pitzinnu. c *Lis travallat Maria. This observation suggests that the complement which sanctions the presence of an incidental dative must occupy the postverbal position in underlying structure. As well as direct objects and unaccusative subjects, PPs containing an inalienably possessed item can also sanction an incidental dative clitic (compare the following with (81a, b)): (85) a Juanne t’ at ballatu supra sos pedes. ‘John danced on your feet.’ (i.e. ‘…trod on your feet when dancing.’) b Su pitzinnu m’at cascatu in murros. ‘The boy yawned in my face.’ Dative clitics can also be used to denote the possessor of prepositional objects with transitive and unaccusative verbs:

< previous page

page_227

next page >

< previous page

page_228

next page >

Page 228 (86) a Lis appo postu unu bonette in conca. lit. ‘I put to them a hat on head.’ ‘I put hats on their heads.’ b S’abba m’est arrivata fintzas a sos grinucos. lit. ‘The water arrived to me even to the knees.’ ‘The water came right up to my knees.’ Turning now to counterparts of (81)–(86) with non-clitic datives, the unacceptable examples (81), (84) remain unacceptable with a+ NP in place of the clitic, as we would expect. More significantly, the use of a+NP in place of the clitic in cases like (83) is marginal at best. When asked to judge examples such as (87), our informants systematically offered the clitic-doubled (dislocated) variants Su frate l‘est mortu, a Juanne, Su pallone si l’est isgonfiatu, a su pitzinnu as more natural alternatives: (87) a ?Su frate est mortu a Juanne. ‘John’s brother died (on him).’ b ??Su pallone s’est isgonfiatu a su pitzinnu. ‘The balloon deflated on the boy.’ The deviance of (87b) is perhaps to be expected in so far as su pallone is not an item which naturally invites reference to a possessor (cf. our earlier comparison of (62) and (75)), but the kinship term su frate in (87a) arguably ought to sanction an incidental a+ NP with a possessive interpretation. However, even with a bodypart NP as subject of an unaccusative verb,10 the use of a non-clitic dative is decidedly odd, whereas the corresponding clitic (with or without dislocated a+NP) is fine: (88) a ?Sos pilos sun créskitos a cussa fémina. b Sos pilos li sun créskitos (a cussa fémina). lit. ‘The hair has grown to her/to that woman.’ ‘That woman’s/her hair has grown.’ In addition, a+NP does not occur readily with body-part nouns governed by a preposition (compare (85)– (86)): (89) a ??Juanne at ballatu supra sos pedes a Maria. ‘John danced on Mary’s feet.’ b ??Su pitzinnu at cascatu in murros a sa mama. ‘The boy yawned in his mother’s face.’ (90) a ??Appo postu unu bonette in conca a sas pitzinneddas. ‘I put hats on the little girls’ heads.’ b ??S’abba est arrivata fintzas a sos grinucos a Gavini. ‘The water came right up to Gavin’s knees.’

< previous page

page_228

next page >

< previous page

page_229

next page >

Page 229 Restricting attention to the clear cases, it seems that a+NP can act as a dative of interest only when it can be naturally construed as the possessor of the entity denoted by the direct object NP. An intuitive way of interpreting this restriction is to treat the incidental a phrase as a secondary predication whose ‘understood subject’ must be a sister of the a phrase (in the sense that both must be immediate constituents of the same maximal projection, ignoring intermediate single-bar projections). This condition is satisfied when the ‘possessed’ NP is a direct object, but not when it is governed by a preposition as in (89)–(90). The reduced acceptability of examples like (87a) and (88a) might be attributed to the fact that this condition is satisfied in underlying structure (assuming the unaccusative analysis) but not at the surface level,11 since the preverbal subject is clearly not a sister of the a phrase. However, fronting of either the ‘possessed’ direct object or the a phrase does not appear to impair acceptability (see examples (70b, c) above). There are various ways in which these observations might be reconciled, but for present purposes we simply note that the effect on acceptability depends on the type of movement process which destroys the sisterhood relation at the surface level. Clearly, the sisterhood condition does not hold for incidental dative clitics even when they are interpreted possessively. A possible reason for this difference is that PPs (including the a phrases under discussion) are inherently predicative whereas clitic pronouns have an essentially referential function and do not need to be predicated of any element. This is consistent with our general observation that clitics can function as datives of interest even when no possessive relation is involved (e.g. as in (62)). In section 4.3.2 we noted that a dative clitic can be used to denote the understood object of certain spatial prepositions and similar expressions which can also be used intransitively: (91) a S’abba lis est akkirrata a supra. ‘The water fell down on them.’ b Juanne m’est cúrritu in fattu. ‘John ran after me.’ c Lis appo postu sa mesa indainnantis. ‘I put the table in front of them.’ Rather than postulating some syntactic process which allows pronominal objects of such prepositions to be cliticised directly in the form of dative clitics, we suggest that such constructions should be analysed in essentially the same way as those in (85)–(86) where the preposition governs a body-part NP which requires an antecedent to

< previous page

page_229

next page >

< previous page

page_230

next page >

Page 230 indicate its possessor. In other words, the primary function of the dative clitics in (91) is to denote an individual who is affected in some way by the event, but also serves to identify the locative entity implied by the preposition. The relevance of the affective relation is borne out by the fact that dative clitics cannot be construed as complements of the preposition with verbs which do not invite reference to an affected entity; compare (92) with (91c): (92) *Lis appo vistu una mesa indainnantis. ‘I saw a table in front of them.’ The leading idea here is that spatial prepositions define location with respect to a particular dimension or part of the entity denoted by the complement NP. In the absence of a complement (i.e. when the preposition is used intransitively) the entity in question is left unspecified, but the relevant dimension is conveyed by the preposition, just as in examples like (85)–(86) the body-part NP denotes a dimension of a ‘possessor’ who is not specified within the phrase itself. In both cases, this implicit element can take its reference from an incidental dative clitic which denotes an affected entity. The similarity between these two cases is highlighted by complex spatial expressions, such as in palas de…‘behind’ (lit. ‘in back of’), where the dimension is specified by means of a body-part noun, and by cases like supra sua ‘on top of him/her’ where the locative entity is expressed by a pronominal possessive (see 4.3.2). 5.2.4 Ethic datives and locatives A characteristic feature of colloquial Sardinian is the frequent use of incidental dative and locative clitics whose function appears to be stylistic rather than semantic; they are used to create a dramatic or empathetic effect rather than to denote a salient place or affected entity. We use the term ‘ethic’ as a heuristic label for such uses of these clitics, though it must be emphasised that the phenomena covered by this term are fairly heterogeneous. The archetypal case of the ethic dative is illustrated in (93): (93) a Gavini ti mandicat tres kilos de pasta in deke minutos. lit. ‘Gavin eats you three kilos of pasta in ten minutes.’ b Sansone t’arricaíat una mesa kin su sintzirineddu. lit. ‘Samson lifted you a table with his little finger.’ Although the clitic in (93) could conceivably be interpreted as denoting an individual who is affected by the event in the same way as the

< previous page

page_230

next page >

< previous page

page_231

next page >

Page 231 datives of interest discussed in 5.2.3 (e.g. ‘for your benefit’, ‘at your expense’), these examples have a more natural reading which involves a much looser relation between the referent of the clitic and the event described. Under this reading, available only with second person clitics (usually in the singular), the function of the clitic is stylistic rather than semantic; the inclusion of ti serves to make the event more vivid, inviting the hearer to imagine himself as a witness to the event. This use of the dative may be seen as an illocutionary extension of the general dative of interest phenomenon; although the hearer is not presented as being affected by the event itself, the speaker invites him to be affected (e.g. surprised or impressed) by the report of the event. This distinction is perhaps clearest in past tense examples such as (93b) where the hearer could not be involved in the event itself but could be affected by the report of the event (for example, if Sansone refers to the Samson of biblical fame). This use of the second person dative is restricted to sentences which have nonpunctual tense (simple present or imperfect) and which denote a hypothetical, rather than actual, event (typically a fairly spectacular event which serves to illustrate some capacity on the part of the subject entity). A further property which distinguishes these datives from those discussed in 5.2.3 is that they cannot act as resumptive clitics for a dislocated disjunctive pronoun: (94) *A tie, Sansone t’arricaíat una mesa kin su sintzirineddu. This example could only have the rather absurd dative of interest reading (‘For you, Samson used to lift a table with his little finger’) parallel to A tie, Juanne ti puliat su lavandinu dondzi die ‘For you, John used to clean the sink every day’. A further range of cases where dative clitics may have a wider function connected with the pragmatics of discourse is in sentences which express requests or offers. This is perhaps seen most clearly in examples like the following which contain two incidental datives: (95) a A mi lis lavas sas dentes? lit. ‘Will you wash to me to them the teeth?’ b Ti lis lavo sas dentes. lit. ‘I wash to you to them the teeth.’ In principle, either of the clitics in (95) could act as the antecedent for the body-part NP; in practice, it is the third person form lis which normally assumes this role, the first and second person clitics having a more peripheral function: ‘Will you wash their teeth for me?’, ‘I will wash their teeth for you’. This distinction is not restricted to cases

< previous page

page_231

next page >

< previous page

page_232

next page >

Page 232 which involve possession. Suppose the interlocutor is urging me to clean the sink for the benefit of some third party, this could be achieved by the utterance in (96a), to which an appropriate reply might be (96b): (96) a A mi li pulis su lavandinu? ‘Will you wash me the sink for him/her?’ b Eja, ti li pulo su lavandinu. (or: Ti bi lu pulo.) ‘Yes, I will wash you the sink/it for him/her.’ Clearly, the function of mi and ti in (95)–(96) is not radically different from that of datives of interest in general; they refer to entities which are portrayed as benefiting in some way from the realisation of the event in the sense that the requests in (a) elicit an action as a favour to the speaker whereas the offers in (b) propose an action intended as a favour to the hearer, even though the immediate beneficiary may be some third party. Our point is that this more peripheral type of benefactive relation appears to be restricted to clitics which refer to one of the interlocutors and is intimately tied up with the illocutionary force of requests and offers. Whether such uses should be assimilated to ‘datives of interest’ or to ‘ethic’ datives of the type in (93) or whether they constitute a third, intermediate class is a question which we leave open. However, we may note that these clitics (like the ethic datives in (93)–(94) but unlike more typical datives of interest) cannot be readily associated with a dislocated phrase introduced by a:12 (97) a *A mi lis lavas sas dentes, a mie? b *Ti lis lavo sa dentes, a tie. c A mi lis lavas sa dentes, a issos? d Ti lis lavo sas dentes, a issos. Within our umbrella category of ethic datives we may include some uses of reflexive clitics which, broadly speaking, invite the hearer to empathise with the subject entity. Some of these uses show affinities with datives of interest of the type discussed in 5.2.3 whereas others may be analysed as forming part of fully lexicalised pronominal verbs. Often a reflexive clitic can be adjoined to a verb to achieve an intensifying effect or to indicate deliberateness on the part of the agent. For instance, when a reflexive clitic is added to a verb such as pompiare ‘to look at’ or piccare ‘to take’ as in (102) we get a variety of possible readings which might be glossed respectively as ‘stare at, scrutinise, ogle,…’ and ‘choose, steal, grab,…’:

< previous page

page_232

next page >

< previous page

page_233

next page >

Page 233 (98) a Juanne s’at pompiatu cudda pitzocca. ‘John stared at (etc.) that young girl.’ b Pretu s’at piccatu cuddu libru. ‘Peter chose (etc.) that book.’ The use of reflexive clitics is particularly common, indeed almost conventional, with ‘ingestive’ verbs such as mandicare ‘to eat’ and bíere ‘to drink’: (99) a Maria s’at mandicatu su pane. ‘Mary ate the bread.’ b Gavini s’at bitu unu caffé. ‘Gavin drank a coffee.’ The semantic contribution of the clitic in these cases is negligible, to the extent that one might be justified in treating si mandicare and si bíere as pronominal verbs. However, the reflexive does appear to have a residual semantic content (comparable perhaps to that of the particle ‘up’ with such verbs in English) which shows up not so much in the actual interpretation of sentences in which the reflexive occurs as in the range of cases where it cannot appear. In ‘eventive’ sentences such as (99) the use of si, though optional, is generally preferred (it gives a more idiomatic ring to the utterance without adding to the meaning), but in generic or negative sentences such as those in (100) the inclusion of si is decidedly odd: (100) a Sos sórrikes (??si) mandican su casu. ‘Mice eat cheese.’ b Maria non (??s’) at mandicatu su pane. ‘Mary did not eat the bread.’ Informally, the reflexive clitic with these verbs appears to emphasise the actuality of the event. The connotations of deliberateness, intensification and actualisation in (98)–(100) might be derived as pragmatic effects of the fact that the Agent is also presented as a beneficiary of the action, so that it is difficult to view the action as being performed casually, unintentionally or as a matter of habit. At the same time, it may be significant that with the verbs in question the subject is not only an Agent but also a Perceiver or Receiver (an ‘animate location’ in the sense used in 3.2.1) so that the clitic serves only to amplify a reflexive relation which is already implicit in the meaning of the verb and, to this extent, is semantically redundant. With verbs which do not have this property, the presence of an incidental reflexive clitic introduces

< previous page

page_233

next page >

< previous page

page_234

next page >

Page 234 an additional affective relation of the sort which we have associated with datives of interest: (101) a Pretu s’at secatu una thadza de pane. ‘Peter cut himself a slice of bread.’ b Maria s’at lavatu sas dentes.’ ‘Mary brushed her teeth.’ c M’appo disinnatu unu ritrattu. ‘I drew a portrait for myself.’ (e.g. for my own amusement) Like the datives of interest discussed in 5.2.3, incidental reflexive clitics cannot be used with normal intransitive verbs. This restriction extends to intransitive uses of mandicare, bíere, etc. (e.g. Nos mandicamus can only mean ‘We eat ourselves/each other’). Incidental reflexives can occur with certain unaccusative verbs of movement, accompanied by the Source clitic nke (sometimes nde): (102) a Lukia sink’est andata. ‘Lucy went away.’ b Juanne at vistu sa janna abberta e sink’est intratu a intro. ‘John saw the door open and he went in.’ c Maria sink’est issita. ‘Mary went out.’ Sink’andare ‘leave, go away’ in (102a) can perhaps be regarded as a fossilised pronominal verb (cf. It. andarsene, Fr. s’en aller ); note that andare on its own cannot readily occur with a Source complement (pronominal or otherwise) unless the Goal is also specified: ?Nke so andatu, ?So andatu dae incue ‘I went from there’, but So andatu dae Olbia a Othieri ‘I went from Olbia to Othieri’). Nevertheless, the use of a reflexive clitic with nke to denote movement away from the speaker or a reference point established by context extends to many other movement verbs, a notable exception being vénnere ‘come’ whose intrinsic deictic orientation is the exact opposite of that which is induced by sinke . This deictic effect can be seen by comparing (102b) with the following example: (103) An bussatu a sa janna e Juanne (??sink’) est intratu a intro. There was a knock at the door and John entered.’ However, this deictic effect is not consistent, as can be seen from example (104) which is deictically neutral with or without sinke : (104) Babbu (sink’) est ghiratu a sas úndiki. ‘Father came/went back at eleven o’clock.’

< previous page

page_234

next page >

< previous page

page_235

next page >

Page 235 Leaving the question of deixis briefly to one side, the presence of sinke in examples like the above tends to connote some degree of deliberateness on the part of the subject entity or at least to invite empathy with the subject, and also to focus on the event itself rather than the resulting state. For instance, (102c) would be an appropriate answer to a question such as ‘What did Mary do when you insulted her?’ but not to an enquiry about Mary’s whereabouts, though Est issita ‘She has gone out’ would be fine in this context. Similarly, sinke is inappropriate in (105) (compare with (104)): (105) Babbu non (??sink’) est galu ghiratu. ‘Father has not returned yet.’ In this respect, sinke may be seen as compensating for the lack of a preterite tense form as distinct from the present perfect. Similarly, with some verbs sinke serves to distinguish the perfective, verbal use of the past participle from its adjectival use: e.g. Sink’est mortu ‘He died’ vs. Est mortu which could also mean ‘He is dead’. We hypothesise that the effects which relate to the attitude of the subject entity are conveyed by si (for example, deliberateness or empathy, as in (98)–(99) above) whereas those which relate to the event itself (for example, the aspectual effect) are attributable to nke. More specifically, we propose that nke presents the event from the perspective of a preceding situation. Thus, in (102b) si attributes some degree of motivation to John, whereas nke alludes to an earlier motivating circumstance (for example, the event described in the first conjunct). This complementarity between si and nke may help to explain why ethic si is almost always accompanied by nke with verbs of this type. Generally, ethic si occurs with verbs whose subject is an Agent (as in (98)–(99)), whereas the subject of unaccusative verbs is essentially a Theme. However, in so far as nke focuses on the event from the perspective of a previous situation, it may also be seen as enhancing agentivity in such a way as to permit the use of si. In a similar, tentative, vein, the deictic effect observed in (102)–(103) may be derived as a pragmatic inference from the respective functions of si and nke in that nke invites us to view the event as a development from a prior situation while si invites us to empathise with the subject, so that (102c) is appropriate in the context of events occurring at the location of Mary prior to her exit. This approach does not explain the absence of the deictic effect in (104), except to the extent that we might expect pragmatic inferences of this kind to be overridden by other factors (for example, properties of the verb) whereas it would

< previous page

page_235

next page >

< previous page

page_236

next page >

Page 236 be difficult to reconcile such cases with an account which simply equated sinke with the semantic value ‘away’. Nke can also be used alone with effects which are broadly in line with the characterisation presented above. Often nke creates a dramatic effect, particularly in reports of untoward events, in a manner rather similar to the locution ‘go and…’ in English: (106) ‘John (went and) broke a cup.’ a Juanne nk’at secatu una tzíkkera. b Maria nk’at istrempatu sa janna. ‘Mary (went and) slammed the door.’ Similarly, in the following attested example the presence of nke seems to contribute to the dramatic, assertive tone of the utterance (the speaker is criticising two men who have let a sheep escape while trying to shear it): (107) “Itte maccos, si fippo andau dego a cust’ora, nche fit istada tusa.” ‘What idiots, if I had gone at this time, it would have been (+nke) shorn.’ [EP: 83] However, it is easy to find attested examples in which nke occurs without any effect of this sort, or indeed any other clearly identifiable function: (108)a “Appustis de una mes’ora fin torraos a tuccare: Elias e sa mama nche fin abbarraos in coa: oramai nche fin belle arribaos; su sole nche fit a midade caminu e cuddu benticheddu istemperabat su calore.” ‘After half an hour they started off again: Elias and his mother stayed (+nke) at the back (of the procession); by now they had almost arrived (+nke); the sun was (+nke) half-way on its journey and the breeze tempered the heat.’ [EP: 47] b “Tzeledha che fit manna e fossis no aiat mancu fatu iscolas in nudha.” ‘Angela was (+nke) grown up and perhaps had not even had any schooling at all.’ [ALIVERTU 7]

< previous page

page_236

next page >

< previous page

page_237

next page >

Page 237 c “A pacu a pacu su locu si fit ammudau: nche fit tardu.” ‘Little by little the place became quiet: it was (+nke) late.’ [EP: 61] d “Bos fipo isettande da-e eris ca sa cotta nche l’hamus accabada janteris manzanu.’ ‘I had been waiting for you since yesterday because the baking we finished (+nke) it two days ago in the morning.’ [BS: 110] e “Sa missa nch’at a esser za finia.” ‘The mass will be (+nke) already finished.’ [BS: 111] f “Non m’as accatau ca nche fippo in atterube, pacas dies.” ‘You did not find me because I was (+nke) elsewhere, for a few days.’ [EP: 81] We have little to say about such examples except to suggest that the function of nke appears to be temporal or vaguely ‘contextualising’ rather than specifically locative. In particular, we may note that nke can co-occur with an interrogative locative expression: (109) “Ube sa balla nche fit tziu Martine?” ‘Where on earth (lit. ‘the bullet’) was (+nke) Uncle Martin?’ [EP : 80] The other locative clitic bi can also occur in a variety of contexts in which reference to a Location or Goal appears to be superfluous. For example, movement verbs are often accompanied by bi (though it can be omitted) even though the Goal is evident from context: Juanne non b’est arrivatu ‘John has not arrived’, A bi venis? ‘Are you coming?’ Also bi can co-occur with another locative expression, as in (110): (110) a Cuddos sordatos bi sun mortos in Russia. ‘Those soldiers died in Russia.’ b Babbu bi fit in su thilleri. ‘Father was in the public house.’ Nevertheless, in such cases bi appears to retain its locative characteristics since, unlike nke in (109), it cannot be used in conjunction with

< previous page

page_237

next page >

< previous page

page_238

next page >

Page 238 an interrogative locative expression (except in impersonal constructions where it has an existential function— see 3.2.2): (111) a *Ube bi sun mortos cuddos sordatos? ‘Where did those soldiers die?’ b *Ube bi fit babbu? ‘Where was father?’ There are also some incidental uses of bi which are associated with particular verbs. For some speakers, there is a strong tendency to place bi before possessive uses of áere ‘have’, particularly when the subject is null: (112) a B’appo una mákkina ‘I have a car.’ b Non b’amus pane. ‘We have no bread.’ In Campidanese, bi has become lexically incorporated in the verb ‘go’ bandái, an interesting phenomenon given that bi is obsolete in these dialects (see 5.2.1). A further case, of a rather different type, is the use of bi with fákere ‘do, make’ to indicate capacity as in (113a) (compare (113b)): (113) a Non bi lu faco. ‘I can’t do/manage it.’ b Non lu faco. ‘I don’t/won’t do it.’ This case is rather curious in that most of the ethic uses of clitics which we have reviewed appear to emphasise the actuality or deliberateness of the action in some way, whereas the effect of bi in (113) is just the reverse. As should be evident from the tentative tone of the discussion in this section, the functions of dative and locative clitics which we have labelled ‘ethic’ are extremely unclear in many cases. For both types of clitic there seems to be a continuum ranging from straightforward cases where the clitic functions as an indirect object (or, in the case of bi and nke, denotes a specific location) through to cases where the clitic appears to be purely expletive. The purpose of our discussion in this section has been to illustrate, rather than define, various uses which tend towards the expletive end of this spectrum and to suggest

< previous page

page_238

next page >

< previous page

page_239

next page >

Page 239 syntactic criteria for distinguishing these from the more straightforward uses. 5.3 COREFERENCE AND BINDING 5.3.1 Preliminary remarks Most of the items discussed in the previous sections are referentially dependent in the sense that they do not name a particular entity or provide a description of the properties of the referent beyond features of person, number, gender and certain deictic indications, but rely almost entirely on contextual cues (the availability of an overt antecedent or the presence of a salient entity within the domain of discourse) to establish reference. In this section we shall be mainly concerned with the structural conditions which determine possible coreference between pronouns and potential antecedents. Following Chomsky (1981) we draw a distinction between corefer-ence and a more restricted notion of ‘binding’. Coreference is taken here to be a symmetric relation which holds between all expressions which refer to the same entity whereas binding takes into account the structural relation between coreferential items. The relevant structural relationship, known as c-command, can be defined for present purposes as in (114): (114) X c-commands Y if every phrasal category which dominates X also dominates Y, and X itself does not dominate Y. Thus, X binds Y if X and Y are coreferential and X c-commands Y. Within this framework, NPs are subclassified into three types: anaphors (corresponding roughly to reflexive and reciprocal pronouns), pronominals (personal pronouns) and R-expressions (principally NPs headed by a common or proper noun). Anaphors are items which must be bound by an antecedent within a certain local domain (often referred to as the ‘governing category’) which we shall provisionally take to be the simple clause. Pronominals on the other hand cannot be bound within their local domain but they can be bound by an antecedent outside this domain. R-expressions cannot be bound within any domain. Thus, in the following English examples, where the relevant local domain is delimited by brackets, the anaphor himself can only refer back to John, the pronominal him can refer to Bill or to some unmentioned third party, but not to John, whereas the

< previous page

page_239

next page >

< previous page

page_240

next page >

Page 240 R-expression that man cannot be construed as referring to either John or Bill: (115) a Bill thinks that [John admires himself]. b Bill thinks that [John admires him]. c Bill thinks that [John admires that man]. Note that the condition on R-expressions is formulated in terms of binding, not coreference, and therefore does not exclude the possibility of an R-expression referring back to an earlier NP, as in (116) where that man can be interpreted as referring to Harry: (116) Sue admires Harry but Mary detests that man. Coreference is allowed in (116) because Harry does not c-command (and, therefore, does not bind) that man—Harry is dominated by at least two phrasal nodes (the VP and S of the first conjunct) which do not dominate that man—whereas in (115c) that man is c-commanded by both Bill and John. 5.3.2 Binding properties of pronouns Within the class of third person clitics, a clear distinction is maintained between anaphors and pronominals. The reflexive/reciprocal anaphor si must corefer with the subject of its immediate clause (except in Campidanese where it can also function as a first or second person plural form or a suppletive non-reflexive dative form; see 5.2.2, example (59b)) whereas the non-reflexive (pronominal) clitics lu, la, los, las, li and lis can only be bound by elements outside the immediate clause. Thus, in (117a) si must refer back to Gavini whereas lu in (117b) can refer to Juanne or to some other masculine entity not mentioned, but not to Gavini: (117) a Juanne credet ki Gavini si videt in s’ispreku. ‘John thinks that Gavin sees himself in the mirror.’ b Juanne credet ki Gavini lu videt in s’ispreku. ‘John thinks that Gavin sees him/it in the mirror.’ However, there is no morphological distinction between anaphors and pronominals in the case of first and second person clitics (mi, ti, nos and bos); for example, mi functions as an anaphor (referring back to the subject of the immmediate clause) in (118a) but as a pronominal in (118b, c) (referring back to an item outside the immediate clause or lacking an antecedent altogether): (118) a Juanne credet ki (Jeo) mi vido in s’ispreku. ‘John thinks that I see myself in the mirror.’

< previous page

page_240

next page >

< previous page

page_241

next page >

Page 241 b (Jeo) credo ki Gavini mi videt in s’ispreku. ‘I think that Gavin sees me in the mirror.’ c Juanne credet ki Gavini mi videt in s’ispreku. ‘John thinks that Gavin sees me in the mirror.’ Nevertheless, the distinction does manifest itself in the choice of perfective auxiliary (éssere ‘be’ with anaphors, áere ‘have’ otherwise) when the clitic functions as a direct object (see 3.3.1 for detailed discussion): (119) a (Jeo) mi so vistu in s’ispreku. ‘I have (lit. ‘am’) seen myself in the mirror.’ b Juanne m’at vistu in s’ispreku. ‘John has seen me in the mirror.’ The personal disjunctive pronouns also show some ambivalence with respect to the anaphor-pronominal distinction. In (120) isse can refer to Gavini, Juanne or to someone else (in contrast to the English counterpart where the reflexive form himself must be used if Gavini is taken to be the antecedent): (120) Juanne credet ki Gavini l’at comporatu pro isse. ‘John thinks that Gavin bought it for him/himself.’ Such examples can be disambiguated in favour of a reflexive reading by placing e tottu (lit. ‘and all’) or mattessi after the pronoun, as in (121) where the antecedent must be Gavini: (121) a Juanne credet ki Gavini l’at comporatu pro isse e tottu. b Juanne credet ki Gavini l’at comporatu pro isse mattessi. ‘John thinks that Gavin bought it for himself.’ In these examples, e tottu and mattessi appear to impose anaphor status on the pronoun which they modify (in much the same way as the suffix -self in English), thus requiring the pronoun to be bound within the simple clause. However, we may note that these expressions can be appended to disjunctive pronouns (particularly in subject position) with an effect on intended reference which cannot be accounted for in terms of the formal theory of binding, as in the following examples:13 (122)a Kie at secatu sa pinna de Juanne? Isse e tottu l’at fattu. ‘Who broke John’s pen? He himself did it.’ b “Pro Mialinu «la lingua scritta è la fotografia della lingua parlata» comente isse matessi at iscrittu.” ‘According to Michelangelo “the written language is a

< previous page

page_241

next page >

< previous page

page_242

next page >

Page 242 photographic image of the spoken language” as he himself wrote.’ [Sinnos: 11] In these examples, the inclusion of e tottu and mattessi identifies Juanne and Mialinu as the only possible antecedents; without these expressions isse could refer to some other individual in the extralinguistic or wider discourse context. However, this effect cannot be attributed to anaphor status of the proform since the antecedent does not c-command the proform, nor are the two elements contained within a local domain of the sort which we have envisaged. Consequently, it is reasonable to view the function of e tottu and mattessi in these cases in terms of the pragmatics of discourse rather than in terms of the formal theory of binding. The precise effect of these expressions on the referential potential of the pronoun is difficult to define, but on the whole they appear to favour coreference with an item in the immediate discourse context as opposed to a more remote antecedent or some person who is present in the extralinguistic context. Also, in cases like (122a), these expressions seem to have a contrastive function, selecting a referent which is pragmatically unexpected. It is therefore conceivable that the interpretation of the examples in (121) can be accounted for in terms of the pragmatic effect of e tottu and mattessi without having to postulate that these expressions impose anaphor status on the preceding pronoun. In this connection, it may be noted that the use of these expressions is generally precluded when pragmatic considerations favour corefer-ence with a local antecedent; for example, the addition of e tottu or mattessi would be infelicitous in (123): (123)Sos ladros an partitu su dinari tra issos. ‘The thieves shared the money among themselves (lit. ‘among them’).’ This pragmatic account of (121) is consistent with the hypothesis that personal disjunctive pronouns can have both pronominal and ana-phor status. An alternative approach to examples like (120) is to assume that these items function solely as pronominals and that the relevant binding domain is a constituent smaller than the simple clause (e.g. the PP), so that both of the potential antecedents in (120) are outside this domain. However, we shall not pursue this matter further. In 5.1.2 we argued that demonstratives referring to humans have the syntactic status of pronouns. From a semantic point of view they are also similar to personal pronouns in that they rely heavily on

< previous page

page_242

next page >

< previous page

page_243

next page >

Page 243 contextual information to establish reference. Often demonstratives are used without an antecedent to refer to persons in terms of their location with respect to the interlocutors; e.g. cuddu ‘the one/man over there’, custu ‘the one/man here’. This use of custu is generally perceived as pejorative in so far as the person referred to is physically present but is not treated as a participant in the conversation; for instance, a woman criticising her husband in his presence might do so as in (124): (124) Custu non faket mai nudda in domo. ‘This one never does anything in the house.’ Demonstratives can also be used to refer back to an antecedent, as in (125), but again the choice of a demonstrative, rather than the personal pronoun isse, has derogatory overtones (rather like epithets such as ‘that idiot’, ‘that guy’, etc.): (125) Vistu as a Pretu? Uffá, no’mi faeddes de cuddu. ‘Have you seen Peter? Oh, don’t talk to me about that one.’ However, these pejorative connotations are absent in cases where the use of a demonstrative pronoun serves to pick out the actual antecedent from a set of potential antecedents, as in (126): (126)a Su duttore meu m’at natu ki non fippo maláidu. Tando so andatu a un’átteru duttore e cussu m’at mandatu a s’ispidale. ‘My doctor told me that I was not sick. So I went to another doctor and this one sent me to the hospital.’ b Pretu e Juanne sun duos frates. Custu est pastore ma cuddu travallat in tzittate. ‘Peter and John are brothers. The latter is a shepherd but the former works in town.’ In such cases, custu and cussu select the closest antecedent whereas cuddu selects the more remote antecedent. Note that although the demonstratives in (126) are coreferential with an earlier expression, they are not bound by their antecedent since they are not c-commanded by it. Indeed, it appears that demon-stratives, unlike personal pronouns, cannot be coreferential with a c-commanding NP. Thus in (127) cuddu cannot be used as a derogatory alternative to isse referring back to Pretu:

< previous page

page_243

next page >

< previous page

page_244

next page >

Page 244 (127) a *Pretui credet ki pesso semper a cuddui. b Pretui credet ki pesso semper a issei. ‘Peter believes that I am always thinking of him.’ Similarly, in (128) cuddu and custu cannot be used to distinguish between potential c-commanding antecedents: (128)*Juanne i at natu a Gavinij ki custuj deviat munire in terra e ki cuddui diat lavare s’istelju. ‘Johni said to Gavirij that he, should sweep the floor and that hei would wash the dishes.’ The evidence in (127)–(128) suggests that within the binding framework adopted here, demonstrative pronouns have the status of R-expressions, not pronominals. Inanimate demonstratives, which we analysed in 5.1.2 as determiners governing an empty noun, have essentially the same deictic properties as human demonstrative pronouns, though without the pejorative connotations noted in (124)–(125). In particular, they appear to function as R-expressions; they can refer back to an antecedent provided that the c-command relation does not hold. Note that this claim is not contradicted by examples such as (129): (129) Custa cratea est prus cómmoda de cudda. ‘This chair is more comfortable than that one.’ Although custa cratea can be regarded as an antecedent of cussa in the sense that the semantic content of the latter is recovered from the former (a usage which we adopted in our discussion of quantified expressions in 5.1.3), the relation beween these two items is clearly not one of coreference. On the contrary, the two expressions denote distinct entities from within the class identified by the head noun of the antecedent expression (cratea). Thus, given that coreference is a necessary condition for binding, cudda is not bound in (129). The looser type of referential dependency illustrated in (129) does not appear to be subject to structural conditions such as c-command and locality. While in (129) custa cratea c-commands cuddu within a simple clause (local domain), the same type of dependency can be seen in cases like (130), where neither condition holds: (130) Su mastru m’at natu de léghere cuddu libru, ma credo ki appo a léghere custu.

< previous page

page_244

next page >

< previous page

page_245

next page >

Page 245 ‘The teacher told me to read that book, but I think that I will read this one.’ 5.3.3 Reciprocal and complementary relations Reciprocal relations between a subject and a direct or dative object can be expressed by means of the reflexive clitic si or the first or second person plural clitics nos and bos: (131) a Sas pitzinnas si sun madzatas. ‘The girls hit each other.’ b Nos semos salutatos. ‘We greeted each other.’ c Sos ómines si cantaían cathones. ‘The men sang songs to each other.’ Such constructions are potentially ambiguous between a reciprocal and a genuine reflexive interpretation (e.g. The girls hit themselves’ etc.). The reciprocal interpretation can be made explicit by adding an expression of the form s’unu…s’átteru as in (132): (132) a Sas pitzinnas si sun madzatas s’una s’áttera. b Nos semus salutatos s’unu s’atteru. c Sos ómines si cantaían cathones s’unu a s’atteru. Note that both components of this expression agree in gender with the antecedent, but there is no variation in number (the reciprocal expression always takes the singular form regardless of whether two entities or more are involved). When si represents a dative object, the preposition a is inserted between the two components, as in (132c). In cases where the antecedent is not a subject (as in (133a, b)) and/or the other element involved in the reciprocal relation is a nondative indirect object (as in (133b-d)), the reciprocal expression s’unu…s’átteru occurs without a reflexive clitic, with the appropriate preposition inserted between the two components: (133) a Juanne at presentatu sos óspites s’unu a s’atteru. ‘John introduced the guests to each other.’ b Appo postu sos prattos s’unu supra s’atteru. ‘I put the plates on top of each other.’ c Sas pitzinnas an ballatu s’una kin s’áttera. ‘The girls danced with each other.’ d Devimus travallare s’unu pro s’atteru. ‘We must work for each other.’

< previous page

page_245

next page >

< previous page

page_246

next page >

Page 246 Like each other in English, expressions of the type s’unu…s’átteru behave like anaphors: (134) *Sos duos frates credían ki Maria fit issita s’unu kin s’átteru. lit. ‘The two brothers thought that Mary had gone out with each other.’ The elements s’unu and s’átteru can also be used as separate constituents to denote complementary members of a contextually salient set, the article before unu being optional: (135) B’aíat duos frates. (S′) unu fit riccu. S’átteru fit póveru. ‘There were two brothers. One was rich. The other was poor.’ Similarly, the plural forms can be used to refer to complementary subsets, though in this case unos must be preceded by the definite article: (136) Sos unos fin adzuande sos átteros. ‘Some (of them) were helping the others.’ In such cases, the two subsets may overlap partially, or even completely, thus allowing a reciprocal (or semi-reciprocal) interpretation. Apart from the fact that s’unu always precedes s’átteru, there do not appear to be any restrictions on the structural relation between the two elements in terms of either locality or ccommand (e.g. in (135) they occur in separate clauses whereas in (136) sos unos c-commands sos átteros within a local domain). However the relation between these elements and the expression which denotes the complete set is subject to structural restrictions: (137) a Sos fratesi+j an detzisu ki s’unui deviat adzuare s’átteruj. ‘The brothers decided that one should help the other.’ b Su babbu de sos fratesi+j at vistu s’unui accurtzu de s’átteruj. ‘The father of the two boys saw one next to the other.’ c *Sos fratesi+j an vistu s’unui accurtzu de s’átteruj. ‘The brothers saw one next to the other.’ The generalisation appears to be that the antecedent (sos frates in (137)) cannot c-command the complementary expressions within the local domain. Thus, if we extend the notion of binding to cover cases of referential inclusion as well as strict coreference, these items have the same binding properties as pronominals in their complementary use.

< previous page

page_246

next page >

< previous page

page_247

next page >

Page 247 6 Subordinate clauses 6.1 COMPLEMENT AND ADVERBIAL CLAUSES 6.1.1 Finite clauses With the exception of indirect questions (which we shall discuss in detail in 6.1.8), all finite complement clauses, both indicative and subjunctive, are introduced by the complementisers ki or ca. In the LogudoreseNuorese dialects ki is the usual form (ca being used primarily in reason clauses, with the sense ‘because’, see below), but some speakers also allow ca as an alternative to ki in complements of verbs of saying or belief: (1) a Tappo natu ki/ca sa butteca fit tuncata. ‘I told you that the shop was closed.’ b Juanne at rispostu ki/ca non potíat vénnere. ‘John replied that he could not come.’ c Maria pessat ki/ca su trenu est in ritardu. ‘Mary thinks that the train is late.’ In the Campidanese dialects the use of ca with verbs of this type appears to be more systematic (see Blasco-Ferrer 1986:195–6 for some discussion). Finite complements of other types of verbs and of adjectives and nouns are introduced by ki in all dialects: (2) a Timo ki su trenu siat in ritardu. ‘I am afraid that the train may be late.’ b Mi so abbidzatu ki carki cosa no’ andaiat bene. ‘I noticed that something was wrong.’ (lit. ‘…did not go well.’) c So cuntentu ki ses arrivatu. ‘I am glad that you arrived.’

< previous page

page_247

next page >

< previous page

page_248

next page >

Page 248 d Su fattu ki ses inoke non mi dat annéu. The fact that you are here does not bother me.’ Note that when finite complement clauses have the same function as an indirect object (for example, complements of all adjectives and nouns, and of some verbs such as s’abbidzare in (2b), which requires de before a following NP), they are not introduced by a preposition. However, most types of adverbial clauses are introduced by a preposition or similar item which indicates the semantic function of the clause, in addition to the complementiser ki: (3) a Appo serrata sa janna pro ki non s’istrempet. ‘I locked the door so that it would not slam.’ b Keljo issire prima ki vendzat Juanne. ‘I want to go out before John arrives.’ c Gavini est arrivatu appustis ki semus thuccatos. ‘Gavin arrived after we left.’ Note that prepositions which require de before a following NP (such as prima and appustis in (3b, c), see 4.3.2) do not take this item when they introduce a finite clause. In other words, finite clauses can be governed by prepositions which have a clear semantic function but not by prepositions whose function is simply grammatical. Within the theoretical framework adopted here, this fact can be accounted for by assuming that finite clauses, unlike NPs, do not need to be Case-marked but do need to be introduced by an item which specifies their semantic function within the sentence as a whole. The analysis of pro etc. in (3) as prepositions allows us to treat these clauses as complement clauses, which differ from those in (2) only in that the governing element is a preposition rather than a verb, adjective or noun. This analysis can be extended to cases like (4) provided that we treat paris, appenas and ja 1 as prepositions which are subcategorised for an S′ complement (unlike the items in (3) they cannot govern an NP complement, even with de); (4) a Juanne est arrivatu paris ki jeo so issitu. ‘John arrived just as I went out.’ b Sos invitatos sinke sun andatos appenas k’amus accabbatu de mandicare. ‘The guests left as soon as we had finished eating.’ c Ja ki ses inoke, mi potes adzuare. ‘Since you are here, you can help me.’ Thus, the adverbial clauses in (3)–(4) can be analysed as adjunct PPs with the internal structure (5):

< previous page

page_248

next page >

< previous page

page_249

next page >

Page 249 (5) [PP rep[S′ [COMP Ki]S]] A different pattern is found in reason and conditional clauses introduced by ca and si: (6) a Devo cúrrere ca so in ritardu. ‘I must hurry because I am late.’ b Appo a pérdere su trenu si non curro. ‘I will miss the train if I do not hurry.’ We postulate that ca and si are complementisers which indicate the semantic function of the clause without need for a preposition. Their complementiser status is reflected in the fact that cross-dialectally they are interchangeable with ki in certain uses, as seen in (1) above and in Campidanese conditional clauses, where ki is generally used in preference to si, as shown in (7)—examples from Blasco-Ferrer (1986:203): (7) a “Si cheres trigu, lu depes semenare.” (Log.-Nuor.) b “Chi bollis trigu, ddu depis arai.” (Camp.) ‘If you want corn, you must sow it.’ Also, like ki these items can be followed by emmo ‘yes’ and nono ‘no’ which we may take to be proforms which replace affirmative and negative propositions of the category S: (8) a At a próere cras? Juanne credet ki/ca emmo, ma Maria at natu ki/ca nono. ‘Will it rain tomorrow? John thinks so, but Mary said not.’ b At a próere cras? Si emmo abbarramus in domo, ma si nono andamus a mare. ‘Will it rain tomorrow? If so we shall stay at home, but if not we will go to the sea.’ The complementiser status of si is further motivated by its use in introducing interrogative complement clauses as in (9): (9) No’isco si at a próere cras. ‘I do not know whether it will rain tomorrow.’ Note finally that ca in reason clauses may be preceded by proitte, which is normally an interrogative adverb ‘why’ but in this use can perhaps be classified as a preposition which selects an S′ like the items in (4): (10) Semus in disamistate proitte ca isse non m’at kérfitu véndere cuddu terrenu. ‘We are in (a state of) feud because he would not sell me that plot of land.’

< previous page

page_249

next page >

< previous page

page_250

next page >

Page 250 The items mancari ‘although’ and sicomente ‘since’ (reason sense) typically occur without the complementiser ki: (11) a Mancari custa petha siat tosta, est meta saporita. ‘Although this meat is tough, it is very tasty.’ b Sicomente tue non bi fis, semus ghiratos a domo. ‘Since you were not there, we returned home.’ On this basis, we tentatively classify these items as complementisers on a par with ca and si in (6). However, some speakers appear to treat these items as prepositions since they allow the use of ki with these items: Mancari ki custa petha siat tosta…, Sicomente ki tue non bi fis… (see Blasco-Ferrer 1984:263, 1986:203). The only other types of adverbial clauses which do not conform to the structure in (5) are those which are introduced by a WH- item, as in (12): (12) a Semus ghiratos cando at cumintzatu a próere. ‘We came back when it started to rain.’ b Cantan cussu cathone comente la cantaían sos antzianos. ‘They sing that song in the same way as (lit. ‘how’) their forefathers sang it.’ c L’appo accattatu ube l’as lassatu. ‘I found it where you left it.’ d Oje sa dzente non travallat cantu travallaíat prima. ‘Today people do not work as much as they did before.’ We may also note in passing that comente can be used with ki and the subjunctive in counterfactual comparative constructions such as (13): (13) Juanne fit dande sos órdines comente k’ ésseret su re. ‘John was giving orders as though he were the king.’ Given that comente can be used as a preposition with a comparative value in examples like Juanne faéddat comente unu re ‘John talks like a king’, the construction in (13) can be treated as a straightforward instance of the structure (5), not as a variant of the WH- construction in (12). At this stage, we shall not attempt to offer a syntactic analysis of the constructions in (12), but merely note that adverbial clauses introduced by WH- items, along with interrogative complement clauses (see 6.1.8), constitute a natural class of exceptions to the generalisation that an overt complementiser must be present in finite subordinate clauses. We shall return to constructions of this type in the context of our discussion of headless relative clauses in 6.2.2.

< previous page

page_250

next page >

< previous page

page_251

next page >

Page 251 6.1.2 Mood In this section, we shall be concerned primarily with the factors that determine the choice between the subjunctive and the indicative in subordinate clauses. One of the issues which will be central to this discussion is the extent to which mood features can be viewed as having an intrinsic semantic value as opposed to being simply grammatical features which are selected by the governing predicate. In main clauses the indicative is always used for statements and questions, the use of the subjunctive being restricted to hortative or optative utterances and negative imperatives, as in (14) (see 1.2.3): (14)a (Ancu/ki/si) ti falet unu lampu! ‘May lightning strike you!’ b Si Juanne ésseret inoke! ‘If only John were here!’ c “Si calicunu deppet morrer, non sian issos, ma dego.” ‘If someone must die, let it not be them, but me.’ [EP: 97] d Non bi andes! ‘Do not go there!’ In so far as mood in main clauses cannot be attributed to a governing predicate, the subjunctives in (14) must be taken as having an intrinsic semantic value, indicating that the clause denotes a non-actual situation which is desired by the speaker or which the speaker attempts to bring about or prevent by making the utterance. Turning now to complement clauses, we can distinguish three sets of cases: those where only the indicative is possible; those where only the subjunctive is possible; and a fairly extensive range of cases where either mood is possible (often with some semantic difference). The clear cases correspond closely to the semantic values which we have attributed to the two moods in main clauses. The indicative is used systematically (in affirmative sentences) in complements of verbs of saying or asking (except when used with the sense of ‘order’ or ‘request’), in modal predicates which express a high degree of probability, and in cognitive predicates such as iskire ‘know’ which are factive (i.e. they imply the truth of their complement even when negated): (15) a Juanne at natu ki Maria fit maláidu. ‘John said that Mary was ill.’ b Antoni at dimandatu si Gavini fit inoke. ‘Anthony asked if Gavin was here.’

< previous page

page_251

next page >

< previous page

page_252

next page >

Page 252 c Est veru/sicuru/craru ki su vinu faket male. ‘It is true/certain/clear that wine is bad for one. d Lukia iskit ki Frantziscu sink’est andatu. ‘Lucy knows that Francis has gone away,’ The clearest cases of predicates which systematically take the subjunctive are emotive predicates like kérrere ‘want’ which are nonimplicative (i.e. which do not imply the truth of the complement) and ‘manipulative’ verbs which denote actions intended to facilitate or impede the realisation of the situation described in the complement: (16) a Keljo ki Maria vendzat. lit. ‘I want that Mary come.’ b Tendzo ki abbarres inoke. ‘I insist that you stay here.’ c Preferisco ki m’adzues. ‘I prefer that you help me.’ d Non permitto ki bi andes. lit. ‘I do not allow that you go there.’ e Su dzúdike at ordinatu ki su débitu siat pacatu. ‘The judge ordered that the debt be paid.’ Within the class of ‘manipulative’ verbs, we may also include nárrere ‘say’ and dimandare ‘ask’ when used with the sense ‘tell, instruct’ or ‘request’: (17) a Appo natu ki colen prus tardu. lit. ‘I said that they call by later.’ ‘I said that they should call by later.’ b Su mere at dimandatu ki su travallu siat fattu luego. ‘The boss asked that the work be done soon.’ With verbs of the type in (16)–(17) the infinitive is generally strongly preferred when the subject of the complement is identical to a potential ‘controller’ in the main clause (see 6.1.3), as in many other languages: e.g. Keljo andare ‘I want to go.’ vs. *Keljo ki ande ‘I want that I go.’ However, in Sardinian the use of the infinitive (optionally inflected) encroaches on the use of the subjunctive even in cases where the subject of the complement is not identical with a potential controller in the main clause but is specified independently by a postverbal nominative NP (see 6.1.3, 6.1.5 for further discussion): (18) a Keljo a/de vénnere(t) Maria. ‘I want Mary to come.’ b Tendzo de abbarrare(s) tue. ‘I insist that you stay.’

< previous page

page_252

next page >

< previous page

page_253

next page >

Page 253 c Preferisco de m’adzuare(s) tue. ‘I prefer you to help me.’ d Non permitto de bi andare(s) tue. ‘I do not allow you to go.’ e Appo natu a colare(n) (issos). ‘I said for them to come.’ Abstracting away from the availability of the infinitive in such cases, the factors which determine the use of the subjunctive in (16)–(17) are closely parallel to the semantic properties of the subjunctive in main clauses. On the basis of the ‘clear’ cases discussed above, we can relate choice of mood in complement clauses to two independent semantic parameters. The first of these can be characterised, rather loosely, in terms of the truth value of the complement clause: the indicative is used in cases where the truth of the complement is entailed as a logical property of the governing predicate or the complement expresses a reported statement or question. As a heuristic device, let us refer to such cases by the feature [+actual]. The second parameter involves the existence of an emotional attitude or a manipulative relation with respect to the situation described by the complement, which we may subsume under a mnemonic feature [+affective] and which appears to correlate with the subjunctive. Thus, the complements in (15) can be characterised as [+actual,−affective] whereas those in (16)–(17) are [−actual, +affective]. More generally, we propose that the features [+actual] and [−affective] favour the indicative whereas [−actual] and [+affective] favour the subjunctive and that cases where either mood is possible result from a conflict between these features in such a way that the actual choice of mood reflects a subjective decision on the part of the speaker regarding the relative salience of these features, particularly with regard to the [±actual] feature. Consider first the negative counterparts of the examples in (15) above: (19) a Juanne no’ at natu ki Maria fit/ésseret maláidu. ‘John did not say that Mary was ill.’ b Antoni no’ at dimandatu si Gavini fit/*ésseret inoke. ‘Anthony did not ask if Gavin was here.’ c No’est veru/sicuru/craru ki su vinu faket/fakat male. ‘It is not true/certain/clear that wine is bad for one.’ d Lukia no’iskit ki Frantziscu sink’est/*siat andatu. ‘Lucy does not know that Francis has gone away.’

< previous page

page_253

next page >

< previous page

page_254

next page >

Page 254 In (19a) and (19c) we note that negation of a declarative verb such as nárrere ‘say’ or an implicative predicate permits the use of the subjunctive in the complement as an alternative to the indicative (compare (15a, c). This is consistent with our general hypothesis in that negation of the main clause cancels the [+actual] feature of the complement; in (19a) the complement is not a reported statement and in (19c) its truth is not implied. With factive verbs such as iskire ‘know’ in (19d) the complement retains its [+actual] feature, since its truth is implied even under negation, and thus allows only the indicative. In order to cancel the implication with negated iskire, an indirect question form is used: (20) Lukia no’iskit si Frantziscu sink’est andatu. ‘Lucy does not know whether Francis has gone away.’ In this case, as in (19b), only the indicative is possible even though the truth value of the complement is indeterminate, perhaps because this is an inherent property of interrogative clauses and thus does not need to be signalled by the use of the subjunctive. Complements of non-implicative cognitive verbs allow both moods: (21) a Juanne pessat ki Gavini est/siat in Casteddu. ‘John thinks that Gavin is in Cagliary.’ b Su duttore credet ki Lukia est/siat maláida. ‘The doctor believes that Lucy is ill.’ In these cases, the subjunctive generally conveys some element of doubt concerning the truth of the complement clause and is particularly favoured when the main verb is in the past tense (where there is a strong pragmatic implication that the proposition is false) or is negated, though the indicative is still possible: (22) a Pessaío ki Gavini fit/ésseret in Casteddu. ‘I thought that Gavin was in Cagliary.’ b Su duttore non credet ki Lukia est/siat maláida. ‘The doctor does not believe that Lucy is ill.’ In (22b) the use of the subjunctive suggests that the speaker holds the complement to be false (i.e. he concurs with the doctor’s judgement that Lucy is not ill) whereas the indicative suggests that the speaker considers the complement to be true in spite of the doctor’s belief. On the other hand, the indicative is obligatory with these verbs in cases like (23) where the complement corresponds to what is being asserted:

< previous page

page_254

next page >

< previous page

page_255

next page >

Page 255 (23)

Credo/pesso ki ses maccu. ‘I think that you are mad.’ A similar situation is found with párrere ‘seem’ which takes either the subjunctive or the indicative (with a general preference for the former) when it is used to attenuate the speaker’s commitment to the truth of the complement: (24) (Mi) paret ki su trenu est/siat in ritardu. ‘It seems (to me) that the train is late.’ However, the indicative is obligatory in cases like (25) where párrere is semantically redundant but serves principally to broach a new topic of conversation (in rather the same way as I gather that …in English): (25) Paret ki ses/*sias cojuande. ‘It seems that you are getting married.’ With this verb, the use of the subjunctive is strongly favoured by the presence of a dative complement (which cannot occur in cases like (25), presumably because the dative item emphasises the subjective nature of the proposition. Modal predicates which express probability or possibility normally take the subjunctive, though the indicative is often found in casual speech: (26) a Est probábile ki Maria venit/vendzat. ‘It is probable that Mary will come.’ b Est possíbile/capassu ki su trenu est/siat in ritardu. ‘It is possible that the train is late.’ When possíbile or capassu is negated, the use of the subjunctive is even more highly favoured, though the indicative is still marginally possible: (27) No’est possíbile/capassu ki su trenu ?est/siat in ritardu. ‘It is not possible that the train is late.’ However, with impossíbile only the subjunctive is possible: (28) Est impossíbile ki su trenu *est/siat in ritardu. ‘It is impossible that the train is late.’ The examples in (26) can be characterised as [-actual,-affective] since the truth of the complement is not logically implied and there is no emotive or manipulative relation. Consequently either mood is possible, the indicative generally conveying a higher degree of

< previous page

page_255

next page >

< previous page

page_256

next page >

Page 256 probability. However, in (28) the truth of the complement is not simply indeterminate but is explicitly denied, thus excluding the use of the indicative as a means of suggesting probability. Nevertheless, when counterfactuality of the complement clause results from syntactic negation of the main clause, as in (27), it does not necessarily require use of the subjunctive. This phenomenon can be seen in other cases. In (19a) and (22) we noted that negated verbs of saying or belief can take either mood. However, verbs of these classes which are inherently negative require the subjunctive: (29) a Su duttore at negatu ki Lukia siat/*est maláida. ‘The doctor denied that Lucy is ill.’ b Dúbito ki su trenu siat/*est in ritardu. ‘I doubt that the train is late.’ In addition to the [−actual, −affective] cases discussed above, clauses which are [+actual, +affective] (i.e. complements of emotive predicates whose truth is presupposed) also allow either mood: (30) a M’aggradat ki Maria est/siat vénnita. ‘It pleases me that Mary has come.’ b Mi dispiághet ki su cane est/siat mortu. ‘It displeases me that the dog has died.’ c So cuntentu ki as/appas accabbatu su travallu. ‘I am happy that you have finished the work.’ Again we attribute the availability of both moods to a conflict between the semantic properties which are relevant to choice of mood in that the existence of an emotive relation favours the subjunctive whereas the presupposition of truth favours the indicative. In these cases the choice of mood makes no appreciable difference to meaning but appears to be largely a matter of style or personal preference. However, some predicates of this type can also be used nonimplicatively (particularly when the complement verb is in the simple present) to express an emotional attitude to a hypothetical situation, in which case the subjunctive is obligatory since the complement is [−actual] (not logically implied) and [+affective], both being features which favour the subjunctive. Thus (31a), with the subjunctive, is ambiguous between a factive and a hypothetical interpretation, whereas (31b) allows only the former interpretation: (31) a So cuntentu ki abbarres inoke. ‘I am happy that you are staying here.’ [+actual, +affective]

< previous page

page_256

next page >

< previous page

page_257

next page >

Page 257 ‘I am happy for you to stay here.’ (‘If you were to stay here, I would be happy.’) [−actual, +affective] b So cuntentu ki abbarras inoke. ‘I am happy that you are staying here.’ [+actual, +affective] Contrary to what we have observed in the case of cognitive and modal predicates, negation of the emotive predicate has no discernable effect on the choice of mood in the complement since negation affects only the nature of the emotional attitude and does not alter the logical properties of the governing predicate (i.e. the [±actual] and [+affective] features remain constant): (32) a Non so cuntentu ki as/appas accabbatu su travallu. ‘I am not happy that you have finished the work.’ [+actual, +affective] b Non so cuntentu ki abbarres inoke. ‘I am not happy that you are staying here.’ [+actual, +affective] ‘I am not happy for you to stay here.’ [−actual, +affective] c Non so cuntentu ki abbarras inoke. ‘I am not happy that you are staying here.’ [+actual, +affective] Choice of mood in adverbial clauses follows more or less the pattern which we have outlined for complement clauses. For instance, purposive clauses, which we may characterise as [−actual, +affective] (they denote situations which are not necessarily realised but are desired by the agent of the main clause), always occur in the subjunctive when finite: (33) Appo vaútu sa kiterra pro ki cantes/*cantas tue. ‘I have brought the guitar so that you can sing.’ Similarly, counterfactual circumstantial clauses take the subjunctive: (34) Imbetzes ki cantes/*cantas tue, canto jeo. ‘Instead of you singing, I will sing.’ On the other hand, reason clauses, whose truth is presupposed, take the indicative: (35) a Ja ki ses/*sias inoke, mi potes adzuare. ‘Since you are here, you can help me.’

< previous page

page_257

next page >

< previous page

page_258

next page >

Page 258 b Custu libru m’aggradat ca est/*siat interessante. ‘This book pleases me, because it is interesting.’ Time clauses which describe a situation prior to or simultaneous with the situation described in the main clause require the indicative: (36)a Appustis ki Maria est/*siat arrivata, nonke semus andatos. ‘After Mary arrived, we left.’ b Cando venis/*vendzas a domo mea, t’appo a cumbitare. ‘When you come to my house, I will show you hospitality.’ However, the subjunctive is required in time clauses which denote a situation which follows that described in the main clause: (37) a Devimus intrare prima ki proat/*proet. ‘We must go in before it rains.’ b Juanne est arrivatu prima ki Maria sinke siat prob-/*est andata. ‘John arrived before Mary left.’ c Sico a ti lu dimandare fintzas ki mi lu dies/*das. ‘I will keep asking you for it until you give me it.’ The choice of the indicative in (36) conforms to the pattern we have established since the truth of the time clause is implied and there is no affective relation. However, the obligatory use of the subjunctive with prima ‘before’ and fintzas ‘until’ is more problematic. In some cases it is possible to discern an affective relation which might favour the use of the subjunctive (e.g. adversative in (37a), purposive in (37c)). Also time clauses of this type differ from those in (36) in that their truth is not necessarily implied (as in (37a, c)) and in some cases the event described in the time clause is actually prevented by the event described in the main clause, as in (38): (38) Appo tentu sa tzíkkera prima ki siat/*est rutta a terra. ‘I caught the cup before it fell to the ground.’ Nevertheless, the subjunctive is required even when the event denoted by the time clause is assumed to have occurred and there is no affective connotation, as in (37b) above. Similarly, the indicative cannot be used in examples like (38) to force a reading of the type ‘I caught the cup and then it fell to the ground’. The fact that the indicative is not available in this way suggests that the subjunctive is selected as a property of prima and fintzas, though in some cases the use of the subjunctive is also semantically motivated according to the criteria which we have proposed. Counterfactual manner clauses introduced by comente+ki ‘as if, require the subjunctive (usually in the imperfect):

< previous page

page_258

next page >

< previous page

page_259

next page >

Page 259 (39) Su pitzinnu nos at trattatu comente ki ésseret/*fit su mere de domo. ‘The boy treated us as if he were the master of the house.’ On the other hand, manner clauses which make a comparison with actual events require the indicative (the complementiser ki being absent in this case; see 6.2.2): (40) Su pitzinnu faéddat commente faéddat/*faéddet su babbu. ‘The boy speaks like his father speaks.’ In relative clauses, when the NP refers to a real entity the indicative is required (as in (41a)), but the subjunctive is possible (and often preferred) when the NP has hypothetical reference (as in (41b)): (41) a Appo addoppiato un'ómine ki travallat/*travallet in Núgoro. ‘I met a man who works in Nuoro.’ b So kircande unu theraccu ki travallat/travallet bene. ‘I am looking for a servant who works well.’ Assuming that relative clauses are [-affective], the obligatory use of the indicative in (41a) correlates with the [+actual, −affective] features whereas the choice of mood in (41b) results from the conflict between [actual] and [-affective] as in other cases discussed above. The choice of mood in conditional and concessive clauses runs counter to the pattern which we have outlined above, in that conces-sive clauses, which are normally presupposed, take the subjunctive (as in (42a)), whereas conditional clauses, whose truth is not implied, usually take the indicative (as in (42b, c)): (42) a Mancari Juanne siat in domo, non potímus andare a lu vídere. ‘Although John is at home, we cannot go to see him.’ b Si Juanne est in domo, potímus andare a lu vídere. ‘If John is at home, we can go to see him.’ c Si Juanne fit in domo, potíamus andare a lu vídere. ‘If John was at home, we could go to see him.’ Possibly the use of the subjunctive in concessives like (42a) can be attributed to the fact that the use of this construction reflects a certain affective attitude on the part of the speaker with respect to the situation described in the concessive clause, namely an admission that it conflicts in some way with what is asserted in the main clause. In the case of conditionals, it might be argued that the complementiser si is itself a mood marker indicating an indeterminate truth value (cf.

< previous page

page_259

next page >

< previous page

page_260

next page >

Page 260 its use in indirect questions), thus rendering the inflectional mood redundant as a marker of the [−actual] value. In these constructions, degree of unreality (potential vs. hypothetical/counterfactual) is expressed by tense rather than mood (present in (42b) vs. imperfect in (42c). Nevertheless, inflectional mood does come into play in conditional constructions on the [±affective] dimension since the imperfect subjunctive can be used to express an optative [+affective] attitude to the condition situation: (43) Si Juanne ésseret in domo, potíamus andare a lu vídere. ‘If (only) John were at home, we could go to see him.’ For further discussion of concessive and conditional constructions, see 6.2.3. 6.1.3 Infinitival clauses Almost all infinitival complement clauses are introduced by the items a or de, which we shall refer to provisionally as ‘particles’. Exceptions to this generalisation are infinitival interrogatives of the type No’isco itte fákere ‘I do not know what to do’ (see 6.1.8), infinitives preceded by modal verbs, for which we have proposed a monoclausal analysis in which the modal verb has the status of an auxiliary (see 3.3.3), causative constructions with fákere ‘make’ and lassare ‘let’ (which we shall discuss in 6.1.4) and nominal infinitives introduced by the definite article which correspond roughly to the English gerund (e.g. su iscríere su rumantzu ‘the writing of the novel’), which we shall discuss in 6.1.6. In many cases the choice between a and de is largely arbitrary and subject to dialectal or idiolectal variation. However, when the infinitival complement fulfils an argument relation which can also be expressed by an indirect object NP, the pre-infinitival particle is consistently identical to the preposition which introduces the corresponding NP, as in (44)–(45): (44) a Mi so pentitu de éssere ghiratu gai tardu. ‘I regretted having returned so late.’ b Cussu liputzu servit a secare petha. ‘This knife serves to cut meat.’ c So andatu a comporare pane. ‘I went to buy bread.’ d So cuntentu de inténdere cussa notítzia. ‘I am glad to hear that news.’

< previous page

page_260

next page >

< previous page

page_261

next page >

Page 261 e Juanne est prontu a fákere tottu. ‘John is ready to do everything.’ (45) a Mi so pentitu de su k’appo fattu. ‘I regretted what I did.’ b Cussu liputzu non servit a nudda. lit. ‘This knife serves to nothing.’ (i.e. ‘This knife is useless.’) c So andatu a sa butteca. ‘I went to the shop.’ d So cuntentu de cussa notítzia. ‘I am happy about that news.’ e Juanne est prontu a tottu. ‘John is ready for everything.’ Also infinitival complements of movement verbs are introduced by a in the same way as Goal NPs: (46) a Juanne est issitu a juljare sos pitzinnos. ‘John went out to call the children.’ b Gavini est ghiratu a mandicare. ‘Gavin came/went back to eat.’ c Maria at mandatu su theraccu a comporare pane. ‘Mary sent the servant to buy bread.’ Intransitive prepositions which require de or a before a following NP (see 4.3.2) also require the same particle before an infinitive: (47) a Maria at serratu sa janna prima de si corcare. ‘Mary locked the door before going to bed.’ b Imbetzes de andare a iscola, est abbarratu in lettu. ‘Instead of going to school, he stayed in bed.’ c “Fin accurtzu a rugher: lis mancabat petzi s’occasione.” ‘They were close to falling (into sin): all they lacked was the opportunity.’ [EP : 123] Similarly, infinitival complements of abstract nouns are systematically introduced by de, which is also the preposition which typically introduces nominal complements of nouns: (48) a s’idea de mi fákere a prítteru ‘the idea of becoming a priest’ b su pessamentu de accabbare custu travallu ‘the worry of finishing this work’ c su fattu de éssere arrivatu ‘the fact of having arrived’

< previous page

page_261

next page >

< previous page

page_262

next page >

Page 262 In examples like the above we propose that the particles a and de are prepositions which mark the infinitival complement as an indirect object in exactly the same way as indirect object NPs, with the structure in (49): (49) [PP a/de [S? ?[S ? VP imf.]]] Note that if this analysis is correct, infinitival complements differ from finite complement clauses with respect to the marking of grammatical relations—recall that finite complement clauses are never introduced by prepositions which have a purely grammatical (Case-marking) function. The hypothesis that the pre-infinitival particles a and de are prepositions carries less conviction for cases where the infinitival clause has a semantic function analogous to that of a subject or direct object NP. For these cases we suggest that a and de form part of the infinitival clause, occurring in the COMP position with a complementiser function analogous to that of ki and ca in finite clauses, as in (50): (50) [S’ [COMP a/de] [S ? VPimf.]] Given this assumption, infinitival S′ has essentially the same distribution as NP; it can occur with the structure (50) as a subject (usually in postverbal position) or direct object, or as the object of a transitive preposition, in which case the COMP position must be empty as in (49). In subject clauses and complements of impersonal constructions, the usual form of the infinitival complementiser is a: (51) a Mi piaghet a annatare. lit. ‘It pleases me to swim.’ (i.e. ‘I like swimming.’) b Nos cumbenit a ghirare como. lit. ‘It suits us to return now.’ (i.e. ‘We should go back now.’) c Est diffitzile a faeddare custu dialettu. ‘It is difficult to speak this dialect.’ d “…naraian…ki a bucchiere in gherra er balentia.” ‘…they say that to kill in war is a triumph.’ [SINNOS: 61] e Fit istatu medzus a 1’iscríere in Italianu. ‘It would have been better to write it in Italian.’ Similarly, infinitival complements of éssere are usually introduced by a:

< previous page

page_262

next page >

< previous page

page_263

next page >

Page 263 (52) “Su corraggiu no’est a facher gherras…” ‘Courage is not making war…’ [SINNOS: 60] For some speakers, de is possible as a free variant in cases like (51)–(52). An extreme case of this variation can be seen in the following attested example where we have both a and de with the same verb (aggradare ‘please’) in the same sentence: (53)“Cando ippo animale a battor ancas m’aggradaiat a las ponnes tot’a cúrrere; e s’ippo testuine m’aggradaia’de andar’a bellu…” ‘When I was an animal with four legs it pleased me to run as fast as they could carry me; and if I was a tortoise it pleased me to go slowly….’ [SINNOS: 34] Also, there are a few verbs which take de systematically in impersonal constructions: (54) a Bisondzat de éssere forte. ‘It is necessary to be strong.’ b M’est capitatu/sutzessu de arrivare in ritardu. ‘It has happened to me to arrive late.’ c M’est pássitu de éssere maláidu. lit. ‘It seemed to me to be sick.’ ‘It seemed to me that I was sick.’ The choice of complementiser in object clauses appears to be arbitrary in many cases. One generalisation which appears to hold is that a is possible only with complements of verbs which are ‘futureoriented’ in the sense that they denote the onset, continuation or indirect causation of an event (e.g. cumintzare ‘begin’, sikire ‘continue’, cussidzare ‘advise’, nárrere ‘tell’, provare ‘try’, imparare ‘learn, teach’) or describe a mental state which anticipates the situation referred to in the complement (e.g. tímere ‘fear’). Some of the verbs just mentioned also allow de at least for some speakers: e.g. Appo natu a Juanne alde colare ‘I told John to call by’, Provo alde travallare ‘I am trying to work’, Timo alde issire ‘I am afraid to go out’. Moreover, there are many future-oriented verbs which always take de: e.g. detzídere ‘decide’, disidzare ‘desire’, isperare ‘hope’, permíttere ‘permit’, promíttere ‘promise’, propónnere ‘propose’. However, our generalisation holds more reliably in the opposite direction in that verbs which are not future-oriented regularly require the object clause to be introduced by the complementiser de. Thus verbs which

< previous page

page_263

next page >

< previous page

page_264

next page >

Page 264 denote the termination of a situation (e.g. accabbare and finire ‘finish’) take de, as do cognitive verbs such as those in (55): (55) a Credo de ti connóskere. lit. ‘I believe to know you.’ ‘I believe that I know you.’ b Maria iskit de éssere sa prus galana. lit. ‘Mary knows to be the prettiest.’ ‘Mary knows that she is the prettiest.’ The future orientation of a is illustrated by constructions with the verb pessare ‘think’ which allows only the complementiser de when used epistemically as in (56a) but can take either a or de when it expresses contemplation of a future action as in (56b): (56) a Pessaíat de éssere maláidu. lit. ‘He thought to be sick.’ ‘He thought that he was sick.’ b So pessande a/de comporare una mákkina. ‘I am thinking of buying a car.’ Note that this generalisation applies only to complements which have direct object status. For instance, the subject clauses in (51) have a though they are not future-oriented. Similarly resissire ‘succeed, manage’ takes a, though it describes the completion of an action, but the complement is clearly an indirect object with the structure in (49)—it is pronominalised by the locative clitic bi (Bi so resissitu ‘I managed it’); also the fact that resissire takes éssere as its perfective auxiliary indicates that the superficial subject is the underlying direct object according to the account proposed in 3.3.1 (So resissitu a lu fákere ‘I managed to do it’). The association between ‘future-orientation’ and the particle a in object clauses is reminiscent of the Goal value of a when used as a genuine preposition. Indeed, some of the instances of a which we have treated as complementisers may in fact be prepositions. One such case is the use of a in (56b) where the complement arguably corresponds to the indirect object in So pessande a su viadzu ‘I am thinking of the journey’ rather than the direct object in So pessande cosas malas ‘I am thinking bad things’. Significantly, the complement in (56b) can be pronominalised by the adverbial clitic bi whereas the accusative clitic lu can only correspond to the complement of pessare in its epistemic use illustrated in (56a). Nevertheless, there are some complements introduced by a which are clearly analogous to direct objects and are pronominalised by the accusative clitic lu (for

< previous page

page_264

next page >

< previous page

page_265

next page >

Page 265 example, the complements of nárrere ‘tell’ and cussidzare ‘advise’). The status of a with provare ‘try’ and tímere ‘fear’ is less clear. Pronominalisation of the complement with these verbs yields bi rather than lu though the redundant infinitive can simply be omitted: (57) a Provatu as a 1’accatare? Emmo, (b’) appo provatu. ‘Have you tried to find it? Yes, I have tried.’ b Non keres imparare a guidare? A (bi) times? ‘Don’t you want to learn to drive? Are you frightened to?’ Possibly bi in (57) is not a pronominal substitute for the infinitive but has an ethic function of the sort discussed in 5.2.4. Neither of these verbs can occur with an indirect object NP, though both can take a direct object NP, with the meaning ‘experience’ rather than ‘attempt’ in the case of provare (e.g. provare unu dolore ‘experience a pain’). The infinitive construction with provare can also be used with the ‘experience’ interpretation, as in the attested example (58a), or with an interpretation analogous to that of the gerundive construction in English (i.e. the action denoted by the complement is completed, but without achieving the desired result), as in (58b): (58) a “Probau as a pompiare unu ribale chi t’ at bintu…” ‘Have you had the experience of looking at a rival who has beaten you…’ [EP : 113] b Appo provatu a bi pónnere bendzina, ma fintzas gai sa mákkina no’ at funtzionatu. ‘I tried putting petrol in, but even so the car would not work.’ In these cases the complement can be pronominalised by the accusative form lu, suggesting that it has direct object status with a functioning as a complementiser. Note that these examples run counter to our generalisation in that they do not display the future-orientation normally associated with the complementiser a in object clauses. Notwithstanding such problem cases, the weight of evidence suggests that in object clauses the unmarked complementiser is de and that a is selected as a lexical property by certain futureoriented verbs, whereas in subject clauses a appears to be the unmarked complementiser. Apart from a and de, the only prepositions which can introduce an infinitive directly are kene ‘without’ and pro ‘for, in order to’, as in (59):

< previous page

page_265

next page >

< previous page

page_266

next page >

Page 266 (59) a So issitu kene serrare sa janna. ‘I went out without locking the door.’ b Appo telefonatu a Maria pro li dare sa notítzia. ‘I telephoned Mary to give her the news.’ Unlike a and de, these items can only occur in adjunct clauses. As an alternative to pro, the preposition a can sometimes be used in purposive clauses, as in (60): (60) So falatu a ti vídere. ‘I came down to see you.’ This use of a is possible only with verbs of movement (for example, a cannot replace pro in (59b)), suggesting perhaps that the infinitive functions as the Goal complement of the verb. However, purposive clauses with a can occur in addition to a Goal complement: (61) a Juanne est andatu a sa butteca a comporare pane. ‘John went to the shop to buy bread.’ b Babbu est ghiratu a domo a mandicare. ‘Father came home to eat.’ The purposive use of a is also subject to pragmatic restrictions which do not apply to pro : (62) a So intratu a sa butteca pro/?a m’istikkire. ‘I went into the shop to hide.’ b So falatu a bidda pro/?a ti fákere piakere. ‘I came down to the village to please you.’ Intuitively, a can only be used to denote an immediate purpose which must also be a reasonably ‘natural’ one, whereas pro has a wider function of expressing the motivation for the action described in the main clause. Note in this connection that, unlike pro, a cannot introduce negative purposive clauses, nor can it be modified by adverbs such as nessi ‘just’ in (63b): (63) a So issitu pro/*a non los vídere. ‘I went out in order not to see them.’ b So vénnitu a inoke nessi pro/*a ti salutare. ‘I came here just to greet you.’ With regard to the internal syntax of infinitive clauses, an important restriction is that the preverbal subject position must be empty. The infinitival constructions considered so far are of the ‘control’ type, where the understood subject of the infinitive is interpreted as being coreferential with an argument of the main clause (the ‘con-

< previous page

page_266

next page >

< previous page

page_267

next page >

Page 267 troller’). We assume that the conditions under which an argument can control the subject of an infinitive are determined by the semantic relation between the controller and the complement, though we shall not attempt to define these conditions. The possibilities for control in Sardinian are essentially the same as those found in infinitival and gerundive constructions in English except that Sardinian (like other Romance languages) allows control constructions with epistemic verbs such as crédere ‘believe’ and párrere ‘seem’ and verbs of ‘occurrence’ such as capitare, sutzédere ‘happen’, as in (54b, c) and (55) repeated below: (54) b M’est capitatu/sutzessu de arrivare in ritardu. ‘It has happened to me to arrive late.’ c M’est pássitu de éssere maláidu. lit. ‘It seemed to me to be sick.’ ‘It seemed to me that I was sick.’ (55) a Credo de ti connóskere. lit. ‘I believe to know you.’ ‘I believe that I know you.’ b Maria iskit de éssere sa prus galana. lit ‘Mary knows to be the prettiest.’ ‘Mary knows that she is the prettiest.’ In some cases the empty subject of the infinitive Has no overt controller but is interpreted as having generic or arbitrary reference: (64) a Est diffitzile a sonare sas launeddas. ‘It is difficult to play the pipes.’ b Babbu at natu a/de ghirare kithu. ‘Father said to come home early.’ c Sos duttores cussidzan de non pippare. lit. ‘The doctors advise not to smoke.’ For examples like (64), we assume that the empty subject is controlled by an implicit argument of the main predicate (the experiencer of difficulty, the receiver of the command or advice) which can be overtly expressed by means of a dative expression (e.g. M’est diffitzile a sonare sas launeddas ‘It is difficult for me to play the pipes’). In cases like (65) where the main predicate does not allow a complement which is a potential controller, we assume that the empty subject of the infinitive is simply assigned arbitrary reference: (65) a Est justu a si vindicare. ‘It is just to avenge oneself.’ b A nárrere cosas gai est una virgondza. ‘To say such things is a disgrace.’

< previous page

page_267

next page >

< previous page

page_268

next page >

Page 268 The precise analysis of the constructions in (64)–(65) is not crucial to our discussion. We simply note that in these cases the null subject of the infinitive receives an arbitrary, usually generic interpretation, which can be attributed to the absence of an overt controller in the main clause, a phenomenon which is widespread in other more familar languages. A more distinctive property of Sardinian infinitives is that the subject can be specified by a postverbal NP which has nominative Case, as shown in (66a) where this NP is a pronoun which shows overt Case inflection: (66) a Non keljo a vénnere tue. [NEG I+want to come you (nom.)] ‘I do not want you to come.’ b Appo tuncatu su barcone pro no’intrare sa thíthula. [I+have shut the window for NEG enter the mosquito] ‘I shut the window so that the mosquitoes would not come in.’ Moreover, under favourable pragmatic circumstances, the subject can be omitted while still preserving specific reference: (67) a Non keljo a secare cussu. ‘I do not want X to break that.’ b Appo tuncatu su barcone pro non s’istrempare. ‘I shut the window so that X would not to slam.’ In (67a) the subject of the infinitive (represented as X in the translation) is construed as non-coreferential with the subject of the main verb. Note that in this case, as in (66a), kérrere ‘want’ requires the complementiser a (or de for some speakers) whereas with a corefer-ential reading, it takes a bare infinitive: Non keljo secare cussu ‘I don’t want to break that’. Also in this use kérrere does not display the auxiliary properties discussed in 3.3.3; e.g. clitic-climbing does not occur: Non keljo a lu secare ‘I do not want X to break it’. Perhaps the most natural interpretation for (67a) is that X is the addressee, but it could also refer to a third party (for example, if the addressee is the parent of a child (= X) who is playing with the object in question) or be construed with generic reference (‘I do not want anyone to break that’). In (67b), X is most naturally taken as referring to the window, but this seems to be a pragmatic effect rather than an instance of control. Indeed, X can be taken as referring to some other entity (e.g. the door) in an appropriate context (for example, the draught from the window is in danger of making the door slam). The referen-

< previous page

page_268

next page >

< previous page

page_269

next page >

Page 269 tial properties of the missing subject in (67b) appear to be exactly the same as those of the overt pronoun it in the English sentence/ shut the window so that it would not slam, except that the missing subject is not restricted for number (for example, it could refer to several doors). We defer further discussion of the constructions in (66)–(67) to 6.1.5 where they will be considered in the light of constructions with inflected infinitives, which have essentially the same properties. Sardinian does not allow ‘accusative+infinitive’ constructions of the type I believe him to be intelligent nor are there any clear cases of infinitival ‘subjectraising’ constructions of the type John seems to have understood where the superficial subject of the main predicate is the logical subject of the infinitive but not an argument of the main predicate. Verbs which occur in such constructions in English allow only the control constructions illustrated in (68), and the construction with an attributive complement illustrated in (69) (see 3.2.5 for discussion): (68) a Mi paríat de sonnare. lit. ‘It seemed to me to dream.’ ‘It seemed to me that I was dreaming.’ b Credo de ti connóskere. lit. ‘I believe to know you.’ ‘I believe that I know you.’ (69) a Juanne paret cuntentu. ‘John seems happy.’ b Credío su pitzinnu maláidu. ‘I thought the boy ill.’ However, párrere ‘seem’ and transitive verbs of perception allow gerundive constructions which are perhaps analogous to infinitival cases of subject-raising and accusative+infinitive constructions in English and other languages (see 6.1.7 for examples and discussion). The only plausible instances of infinitival subject-raising constructions in Sardinian are sentences with modal verbs in their epistemic use (for example, expressing possibility as in (70) rather than capacity or permission) or with aspectual verbs like those in (71): (70) Su trenu potet éssere in ritardu. ‘The train may be late.’ (71) a S’abba at cumintzatu a buddire. ‘The water has begun to boil.’

< previous page

page_269

next page >

< previous page

page_270

next page >

Page 270 b Su teléfonu torrat a sonare. ‘The telephone is ringing again.’ (lit.‘…returns to ring.’) However, as we noted in 3.3.3 and 3.3.5, these constructions display syntactic properties which suggest a monoclausal analysis in which the modal or aspectual verb has the status of an auxiliary verb. Similarly, the only cases where the logical subject of an infinitive appears as an accusative are causative constructions with fákere ‘make’ and lassare ‘let’ which also show syntactic properties symptomatic of a monoclausal structure (see 6.1.4 for detailed discussion): (72) Appo fattu/lassatu dormire su pitzineddu. ‘I made/let the little boy sleep.’ If we are right in treating constructions with modal, aspectual and causative verbs as ‘special cases’ in so far as they have syntactic properties which suggest that the infinitive forms part of the main clause rather than constituting an independent complement clause, our claim that Sardinian lacks genuine infinitival subjectraising and accusative+infinitive constructions can be sustained. 6.1.4 Infinitival constructions with causative verbs In this section we shall be concerned with constructions in which the causative verbs fákere ‘make’ and lassare ‘let’ are followed by an infinitive. The basic properties of these constructions follow the pattern which is well documented in other Romance languages (e.g. as described for French in Kayne 1975). In such constructions the subject of the infinitive occurs in postverbal position, ostensibly as a direct object when the infinitive is intransitive or as an indirect object when the infinitive takes a direct object. In the latter case, the understood subject of the infinitive is introduced by either a or dae : (73) a Maria at fattu dormire su pitzinnu. [Mary has made sleep the boy] ‘Mary made the boy sleep.’ b Su politzottu at fattu issire sa dzente. [the policeman has made go-out the people] ‘The policeman made the people leave.’ c Juanneat fattu lavare s’istelju a/dae su theraccu. [John has made wash the dishes to/from the servant] ‘John made the servant wash the dishes.’ There is a slight semantic difference between the use of a and dae in

< previous page

page_270

next page >

< previous page

page_271

next page >

Page 271 cases like (73c). The use of a tends to imply that the understood subject of the infinitive is affected in some way by the action whereas the use of dae places emphasis on the realisation of the action denoted by the infinitive and portrays the understood subject of the infinitive as an instrument in the realisation of this action; thus the version of (73a) with dae may be rendered more accurately as ‘John had the dishes washed by the servant’. The distinction between direct and indirect object status of the understood subject of the infinitive is obscured somewhat by the prepositional accusative phenomenon discussed in 2.2.6, in that this NP is introduced by a when it consists of a proper noun or some similar item lacking a determiner, even though the infinitive is intransitive: (74) Faco/lasso travallare a Juanne. ‘I (will) make/let John work.’ However, replacement of the NP in question by a clitic pronoun shows that this item has accusative status and that the occurrence of a in (74) does not represent an extension of the indirect (dative) object strategy to subjects of intransitive infinitives: (75) a Lu faco/lasso travallare. (lu=accusative) b *Li faco/lasso travallare. (li=dative) ‘I make/let him work.’ Clitics which function as direct objects of the infinitive are attached to the governing causative verb: (76) L’appo fattu mandicare a su pitzinnu. ‘I made the boy eat it.’ This, along with the fact that the understood subject of the infinitive is realised as a complement of the causative verb and the fact that the infinitive is not introduced by a complementiser (a or de), suggests that such constructions have a monoclausal structure, at least at the surface level. Thus, although an underlying structure of the form (77) may be justified on semantic grounds, we postulate that the two clauses are fused together in such a way that the causative verb and the dependent infinitive form a complex verb with a shared argument structure, so that NP2 (the understood subject of the infinitive) is reanalysed as a complement of the complex verb (direct or indirect object according to whether the dependent infinitive has a direct object of its own). (77) [S NP1 FàZker/lassare [S NP2 [VP V imf (NP3)…]]]

< previous page

page_271

next page >

< previous page

page_272

next page >

Page 272 There are many ways of formalising this fusion process, but for the sake of concreteness we assume that the dependent infinitive is raised (along with its direct object if it has one) out of the complement clause (either to a position within the matrix VP, as proposed for French by Kayne (1975), or adjoined to the complement S, as argued by Rouveret and Vergnaud (1980)), yielding a surface structure of the form (78), where NP2 is treated as a direct or indirect object according to whether NP3 (the direct object of the infinitive) is present. (78) [S NP1 Fàkere/lassare Vinf (NP3) [S(a) NP2 [VP…]]] The hypothesis that the direct object is raised along with the infinitive is intended to account for the fact that the direct object can be cliticised to the causative verb (or to the preceding auxiliary if there is one). Note also that the dative or accusative subject can be realised as a clitic attached to the first verb: (79) a Los appo fattu cantare. ‘I made them sing.’ b Lis appo fattu mandicare su pane. ‘I made them eat the bread.’ However, other complements or adjuncts of the infinitive, which we assume remain within the embedded VP, cannot normally be cliti-cised to either the causative verb or the infinitive: (80) a *B’appo fattu andare sas féminas. *Appo fattu b’andare sas féminas. ‘I made the women go there.’ b *Nk’ appo fattu ghirare su pastore. *Appo fattu nke ghirare su pastore. ‘I made the shepherd come back from there.’ Note in particular that in constructions like (81) containing both a dative clitic and a dative NP introduced by a, the former must be interpreted as the subject of the infinitive and the latter as an indirect object (i.e. (81) cannot mean ‘John will make the lawyer write a letter to them’): (81) Juanne lis faket iscríere una líttera a s’avocatu. ‘John (will) make them write a letter to the lawyer.’ Following proposals by Kayne (1975) and developed in subsequent work within the generative framework, we assume that the cliticisa-tion process is subject to a general constraint which prevents an item

< previous page

page_272

next page >

< previous page

page_273

next page >

Page 273 within the domain of a subject from being moved to a position outside this domain. Thus, in the structure (82), which corresponds to the unavailable interpretation of (81) where the pronoun is an indirect object of iscríere, the pronoun occurs within the domain of a subject (s’avocatu) and cannot therefore be cliticised to the causative verb (or indeed the infinitive) which is outside this domain: (82) [S Junnane faket iscríere una líttera [S a S’avocatu [VP— a PRONOUN]]] Note that the restriction illustrated in (81) does not apply in constructions such as (83) where the understood subject of the infinitive is introduced by dae : (83) Juanne lis faket iscríere una líttera dae s’avocatu. ‘John will have a letter written to them by the lawyer.’ This fact can be reconciled with the approach outlined above if we assume that the dae phrase occurs within the embedded VP, the preverbal subject position in the complement clause being empty, as in (84): (84) [S Junnane faket iscríere una líttera [S θ[VP—a PRONOUN dae s’avocatu]]] Leaving details aside, we assume that s’avocatu, being within the VP, does not define a domain which includes the element PRONOUN and that the empty subject position does not qualify as a subject for the purposes of this condition. Consequently, PRONOUN is not in the domain of a subject (other than the subject of the main clause) and can thus be cliticised to the causative verb, as in (83). The verb-raising (or clause-fusion) process described above is obligatory with both fákere and lassare . In particular, unlike its counterparts in many other Romance languages, lassare does not permit constructions such as (85) where the subject of the infinitive occurs between the causative verb and the infinitive (i.e. in the normal subject position within the complement clause): (85) *Appo lassatu sos pitzinnos jocare. ‘I let the boy splay.’ Similarly, with lassare as with fákere, the understood subject of the infinitive must be realised as a dative (or complement of dae ) if the infinitive is transitive, and non-reflexive clitics must be attached to the first verb, not to the dependent infinitive: (86) a Mi l’an lassatu mandicare.

< previous page

page_273

next page >

< previous page

page_274

next page >

Page 274 b *M’an lassatu lu mandicare. ‘They let mee at it.’ Both fákere and lassare allow the understood subject of the infinitive to be omitted altogether. This construction occurs most readily with infinitives accompanied by a direct object and, in such cases, the missing subject typically has an existential interpretation, just as in agentless passives: (87) a Appo fattu accontzare sa mákkina. ‘I had the car repaired.’ b Juanne at fattu iscríere una líttera. ‘John had a letter written.’ Such examples can plausibly be treated as reduced variants of the construction with dae, analogous to agentless passives. As in the construction with dae, indirect objects of the infinitive can be cliti-cised to the first verb, though when the clitic is a dative it can also be interpreted as the subject of the infinitive, as in (88b): (88) a B’appo fattu istikkire su dinari. ‘I had the money hidden there.’ b Lis appo fattu iscríere una líttera. ‘I had a letter written to them.’ ‘I made them write a letter.’ There are other cases of missing subjects in causatives which cannot plausibly be treated as reduced forms of the dae construction. In particular, the subject can be omitted with intransitive verbs, which never allow specification of the Agent by means of dae (for example, (89a) cannot be analysed as a reduced version of (90)): (89) a Su vinu faket cantare. ‘Wine makes one sing.’ b Sa morte faket pessare a su destinu. ‘Death makes one think of (one’s) destiny.’ (90) *Su vinu faket cantare dae Juanne. lit. The wine makes sing by John.’ Sentences like those in (89) have a rather different interpretation from those with transitive verbs illustrated in (87). They are typically used to make normative statements rather than to report specific events and the missing subject is taken to be generic rather than existential. Also, they tend to be more felicitous when the subject of the causative verb is inanimate. Although we can construct examples

< previous page

page_274

next page >

< previous page

page_275

next page >

Page 275 such as (91) with an animate subject, they seem to favour an interpretation where causation is involuntary (for example, (91) means something like ‘Children are a source of amusement’ rather than ‘Children deliberately make one laugh’): (91) Sos pitzinneddos faken rídere. ‘Children make one laugh.’ This property can be seen in examples like (92) where the addition of an adverbial expression which normally implies agentivity leads to deviance: (92) ?Sos pitzinneddos faken rídere kin una pinna. ‘Children make one laugh with a feather.’ On the basis of these observations, we postulate two processes which allow omission of the subject of the infinitive in causative constructions. The first, illustrated in (87), is restricted to transitive verbs and allows an eventive, existential interpretation, analogous to that found in agentless passives. The second yields a generic interpretation for the missing subject and is contingent on a normative reading for the sentence as a whole. Note that the latter case is not restricted to intransitive verbs: (93) Su vinu faket nárrere istupidádzines. ‘Wine makes one say silly things.’ The distinction between these two cases is apparent in constructions where the direct object of the infinitive is a body-part NP. Note firstly that the logical subject of the infinitive can act as an antecedent for such an NP when it is realised as a dative but not when it is introduced by dae : (94) Su vinu at fattu pérdere sa conca a/*dae Juanne. ‘The wine made John lose his head.’ Similarly, in eventive causative constructions the missing subject cannot act as antecedent for the body-part NP, as we would expect if such cases simply involve omission of the dae phrase: (95) *Su vinu at fattu pérdere sa conca. ‘The wine made somebody lose his head.’ However, when the implicit subject is interpreted generically, as in (89), it can be construed as the possessor of the body-part: (96) Su vinu faket pérdere sa conca. ‘Wine makes one lose one’s head.’

< previous page

page_275

next page >

< previous page

page_276

next page >

Page 276 Thus the missing subject in (96) appears to correspond to the a phrase in (94). Note, however, that the missing generic subject does not block attachment of clitics to the causative verb, as shown in (97) where bi can be interpreted as a substitute for a su destinu in (89): (97) Sa morte bi faket pessare. ‘Death makes you think about it.’ Reflexive clitics are attached to the dependent infinitive when they refer to the subject of the infinitive: (98) Sa mama at fattu si lavare su pitzinnu. ‘The mother made the boy wash himself.’ However, if the reflexive refers back to the matrix subject, it is attached to the first verb, regardless of how the subject of the infinitive is expressed: (99) Maria s’est fattu cumpréndere (a/dae sos istudientes). ‘Mary made herself understood (by the students)’ ‘Mary made the students understand her.’ Note also that in (98) the subject of the infinitive is realised as a direct object even though the reflexive pronoun cliticised to the infinitive is interpreted as a direct object of the infinitive, whereas in (99) the reflexive direct object cliticised to the first verb causes the subject of the infinitive to assume indirect object status. In cases like (99) where the reflexive is a direct object, éssere is selected as the perfective auxiliary, but when the reflexive is an indirect object of the infinitive, we have áere, just as in simple sentences (see 3.3.1): (100) Juanne s’at fattu offrire unu muntone de dinari. ‘John had a pile of money offered to him.’ Note, however, that the past participle in (99) does not agree with the subject, similarly there is no agreement with the non-reflexive accusative clitic in cases like (101): (101) Los appo fattu accontzare. ‘I had them repaired.’ When the reflexive clitic forms part of a pronominal verb, it is typically omitted in the causative construction, as in (102) with s’isventare, which is a pronominal verb requiring a reflexive clitic in all other contexts: (102) Sa notítzia at fattu isventare sa pitzinna. ‘The news made the girl faint.’

< previous page

page_276

next page >

< previous page

page_277

next page >

Page 277 With pronominal verbs which also have a non-pronominal transitive use, omission of the clitic may give rise to ambiguity, as in (103): (103) a Maria at fattu tuncare sa janna. ‘Mary made the door close.’ ‘Mary made someone close the door.’ b Su duttore at fattu corcare su pitzineddu supra su lettu. ‘The doctor made the little boy lie down on the bed.’ ‘The doctor made someone lay the little boy down on the bed.’ Under the first interpretation, the infinitive functions as an intransitive pronominal verb, as in the finite sentences Sa janna s’est tuncata ‘The door closed’ and Su pitzinneddu s’est corcatu The boy lay down’. The second interpretation involves the transitive-causative use of these verbs (as in Carcunu at tuncatu sa janna ‘Someone closed the door’ and Carcunu at corcatu su pitzinneddu ‘Someone lay the boy down’) and thus implies the intervention of an intermediate Agent who is left unspecified. When the infinitive is construed as an intransitive pronominal verb (as in the first interpretations), the use of the analytic causative construction generally implies causation of a rather indirect sort which does not involve direct manipulation of the Patient by the Agent. Thus, (103a) would be natural in a situation where the closing of the door occurred as a consequence of some other action on Mary’s part (e.g. opening the window); if Mary simply closed the door in the normal way this would be more appropriately rendered by Maria at tuncatu sa janna. Also the analytic construction in (103a) is preferred with an inanimate subject: Su ventu at fattu tuncare sa janna The wind made the door close’. Similarly, (103b) on the first reading, suggests that the boy lay down under the doctor’s instructions, whereas Su duttore at corcatu su pitzinneddu supra su lettu would suggest that the doctor actually placed the boy on the bed (e.g. the boy was unconscious). There are some restrictions on the types of verbs which can occur in the complement of a causative verb. In particular, copular verbs and auxiliaries (including the modals) are excluded: (104)a *Su mere a fattu éssere infirmiera (a) sa fidza. lit. ‘The boss made his daughter be a nurse.’ b *Sa mama at fattu éssere jocande sas pitzinnas. lit. ‘The mother made the girls be playing.’ c *Su duttore at fattu pótere camminare su pitzinnu. lit. ‘The doctor made the boy be able to walk.’

< previous page

page_277

next page >

< previous page

page_278

next page >

Page 278 d *Sa música fakíat kérrere ballare sas féminas. ‘The music made the women want to dance.’ 6.1.5 Imperfect subjunctives and inflected infinitives In this section, we shall be principally concerned with the syntactic properties of the verb forms listed under the heading ‘imperfect subjunctive’ in Table 3.2 of section 3.1.1. The use of this label is based largely on the fact that these forms are derived directly and with little change from the imperfect subjunctive forms of Classical Latin. However, in terms of synchronic morphology, these forms can be analysed as infinitives to which the personal desinences—po,— s,— t,— mus,— dzis, and— n have been added, with a shift in stress in some cases. Pittau (1972:93) observes that in the dialect of Nuoro, these forms are used primarily as inflected variants of the infinitive, their use as past tense forms of the subjunctive being restricted to the verbs áere ‘have’ and éssere ‘be’ (including their auxiliary uses). This view is borne out by the judgements of our own informants, who readily accept the use of ‘imperfect subjunctives’ with prepositions and complementisers which normally only introduce infinitival clauses (see 6.1.3):2 (105) a Juanne nos at natu a coláremus. ‘John told us to call by.’ b Keljo cantare una cathone prima de sink’andaren. ‘I want to sing a song before they leave.’ c M’at serratu sa janna pro non intrarápo. ‘He locked the door on me so that I could not get in.’ At the same time, the imperfect subjunctive of áere and éssere can also occur with complementisers which introduce finite clauses (see 6.1.1): (106) a No credío ki ésseres inoke. ‘I did not think that you were here.’ b Si aerepo dinari meta, cumporaío cussa domo. ‘If only I had a lot of money, I would buy that house.’ c Mancari ésseret vetza, fit meta bella. ‘Although she was old, she was very beautiful.’ In parallel cases with verbs other than áere and éssere, the pluperfect subjunctive is normally used; i.e. the lexical verb is realised as a past participle introduced by the imperfect subjunctive of áere or éssere:

< previous page

page_278

next page >

< previous page

page_279

next page >

Page 279 (107) a Non kerío ki ésseret vénnitu a inoke. ‘I did not wish that he should come here.’ b Si esseremus arrivatos prus kitho, l’aíamus vistu. ‘If only we had arrived earlier, we would have seen him.’ c Mancari m’áeret piághitu, non l’aío comporatu. ‘Even though it pleased me, I would not have bought it.’ We occasionally find imperfect subjunctives of other verbs used with items which normally introduce finite clauses, but this appears to be confined to literary usage: (108)a “Nariana chi non b’aiat cosa chi no ischiret.” ‘They said that there was nothing that he did not know.’ [SINNOS: 50] b “Ma no m’ammento si si sono vinnicatos…lassendelu viu pro chi donzi notte sicheret a fachere su sonnu ch’appo contatu.” ‘But I do not remember whether they avenged themselves…by letting him live so that every night he would keep having the dream that I have described.’ [SINNOS: 86] The infinitival and finite uses of the ‘imperfect subjunctive’ both allow the subject to be overtly specified. When the subordinate clause is introduced by a finite complementiser (e.g. ki, si, mancari) the subject can occur preverbally, though inversion is generally possible, just as in other types of finite clause: (109) a Non credío ki Juanne ésseret inoke. ‘I did not think that John was here.’ b Si Gavini áeret iskitu cussu, non sinke fit andatu. ‘If only Gavin had known that, he would not have left.’ c Mancari su babbu bi l’áeret proibitu, Maria est andata a su ballu. ‘Although her father forbade her, Mary went to the dance.’ However, when the clause is introduced by an infinitival complemen-tiser or preposition, the overt subject must follow the verb: (110) a Non keljo a bi vénneres tue. ‘I do not want you to come.’ b Devo accabbare custu travallu prima de ghiraret su mere. ‘I must finish this job before the boss returns.’

< previous page

page_279

next page >

< previous page

page_280

next page >

Page 280 c Non credo de ésseret ghiratu Juanne. ‘I do not think that John has returned.’ Note that the postverbal subject in the inflected infinitive construction bears nominative Case, as can be seen from the nominative form tue in (110a). Also, in cases involving an auxiliary, the subject is placed after the main verb, as in (110c). The inflected infinitive is used primarily in cases where the subject is independently specified, as in (110), or its understood referent is distinct from that of a potential controller, as in (105b, c). It can also be used when the subject is coreferential with a controller, as in (105a), but it does not generally occur when control is obligatory by virtue of semantic properties of the main predicate as in (111): (111) ?Provo a travallarepo. ‘I am trying to work.’ However, inflected infinitives are absolutely excluded in causative constructions with fákere ‘make’ and lassare ‘let’ and after modal auxiliaries and the future auxiliary áere. This exclusion follows naturally if, as we have argued, the infinitive in these constructions forms part of the same clause as the finite verb at the surface level and if agreement features originate under the INFL node dominated by S. The possibility of specifying the subject by a nominative NP, as in (110), suggests that, in Sardinian, nominative Case is assigned by agreement features rather than by a tense specification. The fact that this subject must be postverbal follows from our earlier generalisation that infinitives in Sardinian never allow the preverbal subject position to be filled with overt material, though we would expect this stipulation itself to fall out from more basic principles. In slightly more abstract terms, it appears that agreement features on their own can only assign nominative Case to an NP within the VP, whereas assignment of nominative to the external (preverbal) position is dependent on the presence of finite tense features (including the subjunctive). Assuming that Case-assigners must govern the element to which Case is assigned, this generalisation can be expressed structurally by postulating that, in the absence of tense, agreement features must be affixed to the verb before Case is assigned (thus governing only positions within VP) but when they co-occur with tense they can remain under INFL (governing the external subject position) at the stage where Case is assigned. This proposal is reminiscent of early accounts of the so-called ‘pro-drop’ phenomenon in finite clauses (such as that presented in Chomsky 1981:253–75) which relate

< previous page

page_280

next page >

< previous page

page_281

next page >

Page 281 omission or inversion of the subject to affixation of inflectional elements at a stage prior to nominative Caseassignment. We shall not pursue this issue here (see Jones 1992 for detailed discussion), but it is interesting to note that the obligatory properties of inflected infinitive constructions in Sardinian closely resemble those which are optional in finite clauses in pro-drop languages—the subject is either null, with referential properties analogous to those of overt pronouns (i.e. not necessarily controlled), or is realised postverbally. So far we have assumed that the possible occurrence of overt subjects with inflected infinitives is due to the presence of agreement features which have the capacity to assign nominative Case. However, a serious problem for this assumption is posed by the fact, noted in 6.1.3, that uninflected infinitives can also take postverbal nominative subjects, as in (112), and in some circumstances allow a null subject to be interpreted with specific reference in much the same way as overt pronouns or null subjects of finite clauses, as in (113) —for discussion of the referential properties of the null subject in (113) see above pp. 268–9: (112) a Non keljo a vénnere tue. [NEG I+want to come you (nom.)] ‘I do not want you to come.’ b Appo tuncatu su barcone pro no’intrare [I+have shut the window for NEG enter sa thíthula. the mosquito] ‘I shut the window so that the mosquitoes would not come in.’ (113) a Non keljo a secare cussu. ‘I do not want X to break that.’ b Appo tuncatu su barcone pro non s’istrempare. ‘I shut the window so that X would not to slam.’ In other words, the obligatory pro-drop effects which we have associated with inflected infinitives can occur even when no agreement features are overtly present. A possible solution to this problem would be to postulate that these constructions contain abstract agreement features which have no phonetic realisation but which have the same syntactic properties as the overt affixes in constructions with inflected infinitives. Indeed, for the Logudorese-Nuorese dialects, it

< previous page

page_281

next page >

< previous page

page_282

next page >

Page 282 could be argued that the infinitives in (112)–(113) are simply reduced forms of the inflected infinitive in which the personal endings have been omitted. As far as we can ascertain, the syntactic environments under which the uninflected infinitive can be used as in (112)–(113) are the same as those in which the inflected infinitive can occur, except to the extent that the absence of person features in cases like (113) restricts the use of such constructions to contexts in which the understood subject can be readily identified by pragmatic means. Consequently, a rule of deletion (or non-realisation) of agreement features with infinitives could be formulated without reference to syntactic conditions. Such an approach is more difficult to justify for similar constructions in the Campidanese dialects. These dialects have an imperfect subjunctive paradigm which is derived from the Latin pluperfect subjunctive (— essi,—essis,—essit,—éssimus,—éssides,—essint) and which does not double as an inflected infinitive (i.e. these forms can only co-occur with finite complementisers). Indeed, there are no overtly inflected infinitive forms in Campidanese. Nevertheless, uninflected infinitives in Campidanese can occur in constructions of the type in (112) and (113), as in the following examples from Blasco-Ferrer (1986:158–9): (114) a “Cantamu prus a forti po m’intendi s’amanti.” ‘I sang louder for my lover to hear me.’ b “Deu bollu a ddu papai.” ‘I want X to eat it.’ In other words, uninflected infinitives in Campidanese can manifest the properties of inflected infinitives even though there are no overtly inflected infinitive forms from which they can be derived synchronically. 6.1.6 Nominal infinitives The term ‘nominal infinitive’ is used here to refer to the infinitival expressions illustrated in (115), which correspond roughly to gerundive expressions in English, as indicated in our translations: (115) a Su istare semper ritzu mi dat anneu. ‘Being always standing up annoys me.’ b S’iscríere cussu rumantzu l’at piccatu meta tempus. ‘Writing that novel took him a long time.’

< previous page

page_282

next page >

< previous page

page_283

next page >

Page 283 c “…sos ammentos fin troppo rechentes: sa festa, sa bisione de Isalle, su 1’aer piccada a groppa a sa ghirada.” ‘…the memories were too recent: the feast, the dream by the river Isalle, taking her on his horse on the way back.’ [EP : 80] d “…issu tempus chi so nande imur gai a ghisa ‘e piantas pro s’ischire ei su no ischire.” ‘…at the time I am talking about we were just like plants with regard to knowing and not knowing.’ [Sinnos: 26] In these constructions, the infinitive behaves like a verb with respect to complements and modifiers. For instance the infinitive can be modified by an adverb (e.g. semper ‘always’ in (115a)), but not by adjectives (though it can take an adjective as its complement, as in (115a)) and can govern a direct object NP (e.g. cussu rumantzu ‘that novel’ in (115b)). Moreover, it can be accompanied by a clitic pronoun and can occur as an auxiliary introducing a participle, as in (115c), and it can be negated, as in (115d). On the other hand, the infinitive behaves like a noun with regard to choice of specifier, being introduced by the definite article (occasionally a demonstrative), always in the masculine singular. The construction as a whole also behaves like an NP with regard to its distribution, as can be seen in (115c) where the nominal infinitive is conjoined with other NPs. In common with other infinitival constructions, the subject cannot be expressed by an NP in the preverbal position, but it can be specified by means of a possessive expression, as in the case of NPs: (116)a “…si nde ridiana de tutu s’andare e bennere e cúrrere e morrere de sos animales.” ‘(the trees)…laughed at all the going and coming and running and dying of the animals.’ [SINNOS: 27] b “…los toccaio chirchanne de cumprenne’ su secretu de su durare issoro…” ‘…I used to touch them (the trees) trying to understand the secret of their lasting…’ (i.e. ‘of their longevity’) [SINNOS: 27] The hybrid nature of the nominal infinitive construction can be characterised by postulating that the infinitive functions as the head of a VP which is embedded inside an NP, occupying the position

< previous page

page_283

next page >

< previous page

page_284

next page >

Page 284 which would normally be filled by the N′ constituent, as in the structures in (117) which correspond to the nominal infinitives in (115c) and (116b): (117) a [NP su [VP l’aer piccada a groppa a sa ghirada]] b [NP su [VP durare] issoro] Thus, the infinitive can be accompanied by the full range of complements and adjuncts which normally occur within the VP, but is also accompanied, in the peripheral positions, by items which normally occur within NP at a level outside N’ (determiners, quantifiers and possessives). Note that the VP and NP levels cannot be mixed. Whereas in NPs there is some flexibility regarding the order of possessives and other postnominal elements (see 2.2.3), possessives cannot intervene between the infinitive and its complements or modifiers in the nominal infinitive construction: (118) a sa mákkina nova mea b sa mákkina mea nova ‘my new car’ (119) a *su contu de sos avvennimentos suo b su contu suo de sos avvennimentos ‘his/her account of the events’ (120) a su andare abbellu suo b *su andare suo abbellu ‘his/her going slowly’ (121) a *su contare sos avvennimentos suo b *su contare suo sos avvennimentos ‘his/her relating the events’ We assume that (119a) and (121a) are deviant for the same reason (for example, because of some processing constraint which requires possessives to be associated with the nearest noun to the left). Abstracting away from this effect, the facts in (118)–(121) suggest either that possessives can occur within N′ as well as under NP or that some freedom of surface linear order is tolerated between elements belonging to different projections of the same category. However, the data in (120b), (121b) show that this tolerance does not extend across projections of different categories; i.e. the possessive must be external to the VP in terms of linear order as well as syntactic structure. Nominal infinitives cannot be inflected or accompanied by a nominative subject. This is consistent with our claim that the infinitival

< previous page

page_284

next page >

< previous page

page_285

next page >

Page 285 part of the expression is simply a VP (rather than S or S′) if, as we suggested in 6.1.5, the agreement features (overt or abstract) which license a nominative subject originate under the INFL node which is external to VP. 6.1.7 Participial clauses The present participle can be used to form adverbial clauses of a circumstantial nature. In pre-sentential position such clauses typically describe a situation which sets the scene for the event denoted by the main clause: (122) Caminande in su padente, appo vistu un’iscujátulu. ‘Walking in the wood, I saw a squirrel.’ The present participle can also be used presententially with a conditional or ‘reason’ function: (123) a Fakéndelu gai, non bi risessis. ‘Doing it like that (i.e. if you do it like that), you will not succeed.’ b Vidende ki Maria non bi fit, minke so andatu. ‘Seeing that Mary was not there, I went away.’ Generally, in colloquial usage, the subject of the participial clause is not overtly specified, as in (122)–(123), and is construed as being identical with the subject of the main clause. However, in more formal registers we find constructions such as (124)—perhaps calqued on Italian—in which the subject of the participle is realised as a nominative NP placed after the auxiliary: (124) No’aende jeo abbastantza dinari, Juanne at pacatu. ‘I not having enough money, John paid.’ Although the circumstantial participial clauses illustrated in (122)(124) typically precede the main clause, they can also occur sentencefinally, usually separated from the rest of the sentence by a pause or comma: (125) a Appo vistu un’isculjátulu, caminande in su padente. ‘I saw a squirrel, walking in the wood.’ b Non bi resessis, fakéndelu gai. ‘You will not succeed, doing it like that.’ c Juanne at pacatu, no’ aende jeo abbastantza dinari. ‘John paid, I not having enough money.’ In (125a, b), as in (122)–(123), the understood subject of the participle is identical to the subject of the main clause.

< previous page

page_285

next page >

< previous page

page_286

next page >

Page 286 The present participle can also be used, without an intervening pause or comma, after verbs of perception (such as vídere ‘see’ and inténdere ‘hear’ as in (126)) and other verbs which involve some element of perception, as in (127): (126) a Juanne at vistu su trenu arrivande. ‘John saw the train arriving.’ b Amus intesu sos tenores cantande. ‘We heard the tenors singing.’ (127) a Maria at accatatu unu ladru furande su dinari. ‘Mary found a thief stealing the money.’ b Appo addoppiatu su pastore ghirande a domo. ‘I met the shepherd returning home.’ In these cases the understood subject of the participle is the accusative NP (e.g. su trenu in (126a)). The participial expression in these examples (particularly those in (127)) can perhaps be analysed as a circumstantial adjunct, on a par with that in (122) except that the null subject of the participle is controlled by the direct object of the main verb, as in (128): (128) [S Maria[VP at accatatu unu ladrui [S θi furande su dinari]]] For the examples in (126) the accusative NP might plausibly be analysed not as the direct object of the main verb, but as the subject of the participial clause which functions as a complement of the verb denoting the perceived event, as in (129): (129) [S Junnane[VP at Vistu [S Su trenu arrivande]]] The basis for this distinction is that verbs such as accatare ‘find’ and addoppiare ‘meet’ normally require an object which is human or at least physical, thus forcing an adverbial interpretation for the participial expression (e.g. ‘Mary found a thief as he was stealing the money’, ‘We met the shepherd as he was returning home’) whereas verbs such as vídere ‘see’ and inténdere ‘hear’ readily allow an object which denotes an event (e.g. ‘John saw the arrival of the train’, ‘We heard the singing of the tenors’). In this connection, we may note that perception verbs like vídere and inténdere do not allow infinitival complements of the sort found in other Romance languages and English (e.g. we do not have *Juanne at vistu su trenu arrivare or *Juanne at vistu arrivare su trenu), suggesting that the present participle in these examples assumes a function typically associated with the infinitive in other languages. A further case which might be interpreted in the same vein is the

< previous page

page_286

next page >

< previous page

page_287

next page >

Page 287 use of the present participle after párrere ‘seem’ in examples like those in (130), which are reminiscent of ‘subject-raising’ constructions with an infinitival complement in more familiar languages (as we noted in 6.1.3, Sardinian lacks the infinitival ‘subject-raising’ construction, though párrere can take an infinitival complement whose subject is controlled by the dative NP). (130) a Juanne paret travallande. lit. ‘John seems working.’ b Su pitzinneddu paret timende carki cosa. lit. ‘The little boy seems fearing something.’ c Paret nivande. lit. ‘It seems snowing.’ However, the participle in this construction serves more clearly to express an apparent property of the subject entity than the infinitive in similar constructions in other languages; in this respect this construction is closer to expressions of the type look as though …than to seem with the infinitive in English. In particular, this construction is natural only with participles which can be construed as denoting a visible property, whereas the ‘seem’ construction can be used with any verb to attenuate the speaker’s commitment to the truth of the proposition. Thus (131) is odd in much the same way as the English sentence ?Mary looks as though she believes that Naples is in Sardinia (compare Mary seems to believe that Naples is in Sardinid ): (131) ?Maria paret credende ki Nápoli est in Sardigna. lit. ‘Mary seems believing that Naples is in Sardinia.’ The participle in this construction also retains some of its progressive value in that it must denote an impression regarding an ongoing situation, often with a counterfactual connotation, but cannot be used to describe tendencies or habitual events. This is particularly apparent in example (130c) which can only be used to describe a scene which resembles a snowstorm (e.g. a flurry of feathers), corresponding roughly to English It looks (almost) as though it is snowing, but cannot be interpreted in the same way as It seems to snow (often) nor even It seems to be snowing (now) —in such cases the finite construction with párrere is used: Paret ki nivet meta in Russia ‘It seems that it snows a lot in Russia’, Paret k’est nivande ‘It seems that it is snowing’. These observations suggest that the construction in (130) is analogous to copular constructions in which párrere takes an attributive complement (e.g. an AP as in Maria paret maláida ‘Mary seems ill)

< previous page

page_287

next page >

< previous page

page_288

next page >

Page 288 rather than to infinitival subject-raising constructions in English and other languages. The intuition behind this view is that with both attributive and participial complements párrere serves to predicate an apparent or illusory property of the subject entity, whereas verbs of the ‘seem’ type in the infinitival construction appear to modify the proposition expressed by the combination of the matrix subject and the infinitive complement. However, it is not clear whether this intuition can be captured in purely structural terms, given the possibility (envisaged in 3.2.5) that copular constructions with párrere are also derived by raising of the subject from a ‘small clause’ complement. A final, general observation we may make concerning the use of present participles as complements or adjuncts is that, unlike infinitives, they can never be introduced by prepositions (including items traditionally classified as subordinating conjunctions); for example, we do not have constructions of the type while walking in the wood or on hearing the news—as we have noted, simultaneous or causal circumstances of this sort are expressed simply by the present participle without a preposition (as in (122) and (123b)). This restriction does not apply to past participles, which can be used after appustis ‘after’ and its synonyms (pustis, dopo, dapoi): (132) Appustis kenatu, amus abbaitatu sa televisione. ‘After eating, we watched television.’ In this construction, the understood subject of the past participle is identical to that of the main clause. When the verb is transitive, it normally agrees in number and gender with the logical object, a fact which suggests that the past participle has a passive value and that the underlying object has the status of a subject, as in (133) where fattas takes the feminine plural form agreeing with sas faínas: (133) Appustis fattas sas faínas, Maria s’est corcata. lit. ‘After done the chores, Mary lay down.’ ‘The chores having been done, Mary lay down.’ Although in (133) the most likely interpretation is that the Agent in the time clause is Mary, this seems to be a purely pragmatic inference rather than the effect of a grammatical process of control, as can be seen in (134a) where the most likely Agent is the servant and in (134b) where the Agent is left unspecified: (134) a Appustis fattas sas faínas, amus lassatu andare su theraccu. ‘The chores having been done, we let the servant go.’

< previous page

page_288

next page >

< previous page

page_289

next page >

Page 289 b Appustis incuntzatu su tricu, at cumintzatu a próere.’ ‘The corn having been harvested, it started to rain.’ Interpretations of the type in (134) do not appear to be available when the participle is intransitive (the judgement of (135a) is based on a reading where sos tenores is the understood subject of the participle): (135) a *Appustis cantatu, amus datu a bíere a sos tenores. ‘After singing, we gave the tenors something to drink.’ b *Appustis kenatu, at cumintzatu a próere. ‘After eating, it started to rain.’ This difference can be accounted for if we assume that the understood subject of a past participle must be interpreted by means of control if and only if the participle lacks a syntactic subject. This condition obtains in (132) and (135), but not in the passive-like examples (134) where the underlying object (sas faínas) has assumed subject status. Consequently, the implicit Agent in (134) does not qualify as a subject and its reference is determined pragmatically. A construction similar to (134) is possible with unaccusative verbs (i.e. verbs which take éssere as their perfective auxiliary, see 3.2.3): (136) Appustis morta sa mama, Gavini s’est emigratu. ‘After his mother died, Gavin emigrated.’ Again, the participle agrees with the following NP. Note that past participles of non-unaccusative verbs do not allow the subject to be specified in this way: (137) *Appustis cantatu sos tenores, nonke semus andatos. ‘After the tenors had sung, we left.’ Note also that in the passive and unaccusative constructions, the ‘subject’ must follow the participle: (138) a *Appustis sas faínas fattas, Maria s’est corcata. lit. ‘After the chores done, Mary lay down.’ b *Appustis sa mama morta, Gavini s’est emigratu. lit. ‘After the mother died (pp.), Gavin emigrated.’ Past participles of transitive and unaccusative verbs can also be used without a preposition to form time clauses denoting a prior event. The past participle shows the same agreement phenomena as in (134) and (136) (suggesting again that the transitive construction should be treated as a passive), but in this case the ‘subject’ can precede or follow the participle:

< previous page

page_289

next page >

< previous page

page_290

next page >

Page 290 (139) a Fattas sas faínas, Maria s’est corcata. b Sas faínas fattas, Maria s’est corcata. ‘The chores having been done, Mary lay down.’ (140) a Arrivatu su beranu, sos frores an cumintzatu a créskere. b Su beranu arrivatu, sos frores an cumintzatu a créskere. ‘Spring having arrived, the flowers began to grow.’ This evidence suggests that the impossibility of a preverbal ‘subject’ in (138) is due to the presence of the preposition appustis rather than to an intrinsic property of past participles. Again this construction is not possible with ordinary intransitive verbs: (141) a *Cantatu sos tenores, nonke semus andatos. b *Sos tenores cantatu, nonke semus andatos. ‘The tenors having sung, we left.’ With unaccusative past participles, but not other intransitives, the subject can be omitted, and is controlled by the subject of the main clause—compare the following examples: (142) a Issitos dae su padente, amus rucratu su camminu mannu. ‘Having come out of the wood, we crossed the main road.’ b *Caminatu in su padente, amus rucratu su camminu mannu. ‘Having walked in the wood, we crossed the main road.’ A construction similar to (142a) is also possible with transitive/ passive past participles, though in this case the participle tends to be construed as denoting a conditional or motivating circumstance rather than a temporally prior event: (143) a Arrustita supra unu focu de linna, sa petha de andzone est delitziosa. ‘Roasted over a wood fire, lamb meat is delicious.’ b Offesa dae su babbu, Maria no’at kérfitu ghirare a domo. ‘Offended by her father, Mary did not want to return home.’ Past participles of transitive verbs can occur (again with a passive value) as complements of the verb kérrere, which is used in this case with the meaning ‘need’ or ‘should’ rather than its more usual sense ‘want’: (144) a Cussa mákkina keret accontzata. ‘That car needs (to be) repaired.’

< previous page

page_290

next page >

< previous page

page_291

next page >

Page 291 b Custa dzente keret tímita. ‘These people should be feared.’ Introduced by the preposition kene ‘without’, past participles of transitive verbs can also be used as complements of éssere in a construction which is roughly equivalent to a negative passive: (145) Sa petha est kene cotta. lit. ‘The meat is without cooked.’ ‘The meat has not been cooked.’ Neither of the constructions in (144)–(145) is possible with past participles of intransitive or unaccusative verbs. For further discussion of these constructions, see 3.2.7. 6.1.8 Interrogative complements Interrogative complement clauses of a polar nature (corresponding to yes/no questions) are introduced by the complementiser si when finite: (146) a Juanne at dimandatu si Maria fit arrivata (o nono). ‘John asked whether Maria had arrived (or not).’ b No’isco si custu libru est interessante (o nono). ‘I do not know whether this book is interesting (or not).’ c Lukia no’est sicura si at tuncatu sa janna (o nono). ‘Lucy is not sure whether she has closed the door (or not).’ d Gavini no’ at natu si babbu fit ghiratu (o nono). ‘Gavin did not say whether father had returned (or not).’ As we noted in 6.1.2, interrogative complements regularly take the indicative, even though their truth value is in doubt. The complemen-tiser si can only be used in finite complements, but with some governing predicates de can be used to introduce an infinitival interrogative complement: (147) a No’isco de b’andare (o nono). ‘I do not know whether to go there (or not).’ b Non so sicuru de mi corcare (o nono). ‘I am not sure whether to go to bed (or not).’ This construction appears to be restricted to complements of negated cognitive predicates and does not extend to genuine indirect questions: *Appo dimandatu de b’andare (o nono) ‘I asked whether to go or not’.

< previous page

page_291

next page >

< previous page

page_292

next page >

Page 292 Non-polar interrogative complements, which question some ele-ment within the proposition rather than its overall truth value, are introduced by the appropriate WH—item: (148) a Juanne at dimandatu kie est vénnitu. ‘John asked who came.’ b No’isco itte as fattu. ‘I do not know what you did.’ c Non so sicuru comente est sutzessu. ‘I do not know how it happened.’ d Gavini no’at natu proitte sink’est andatu. ‘Gavin did not say why he left.’ Such constructions are also possible in the infinitive, even with governing predicates which do not allow the polar infinitival interrogative: (149) a Juanne at dimandatu comente lu fákere. ‘John asked how to do it.’ b No’isco itte nárrere. ‘I do not know what to say.’ In such constructions the subject of the infinitive is null, controlled by an appropriate argument in the main clause. For this reason, the WH—item cannot correspond to the subject of the infinitive; e.g. we cannot say *Appo dimandatu kie vénnere to mean ‘I asked who should come’. A further restriction, which appears to hold in many other languages (including English), is that the infinitival interrogative complement cannot be introduced by proitte ‘why’: (150) *No’isco proitte t’invitare. ‘I do not know why to invite you.’ As in the case of direct questions (see 1.2.2), we assume that the WH—item originates in a position appropriate to its function within the clause and is moved to the COMP position as in the structures in (151) which correspond to (148b) and (149a) above: (151) a Non isco b Juanne at dimandatu We also assume that no more than one item or phrase may occur in the COMP position, thus accounting for the fact, noted in 6.1.1 and 6.1.3, that interrogative complements introduced by a WH—item are

< previous page

page_292

next page >

< previous page

page_293

next page >

Page 293 systematic exceptions to the generalisation that all subordinate clauses must be introduced by a complementiser (or a preposition in the case of infinitives). A further point to note is that WH- interrogative complements cannot be introduced by prepositions. For instance, the preposition de, which normally introduces complements of sicuru, does not occur before a WH —complement: (152) a Non so sicuru (*de) cando est arrivatu. ‘I am not sure (of) when he arrived.’ b Non so sicuru (*de) comente lu fákere. ‘I am not sure (of) how to do it.’ The absence of de in (152a) conforms to the generalisation that finite clauses cannot be introduced by a preposition with a purely grammatical function (So sicuru (*de) ki tenes rejone ‘I am sure that you are right’; see 6.1.1). However, the absence of de in (152b) conflicts with our generalisation in 6.1.3 that oblique infinitival complements require the preposition which is used to Case-mark a corresponding NP complement. In other words, interrogative WH—complements behave in the same way as finite complements with respect to the distribution of grammatical prepositions. 6.2 OTHER SUBORDINATE CLAUSES 6.2.1 Relative clauses Finite relative clauses can be divided into three structural types: Iclauses introduced by the complementiser ki with a gap in the position corresponding to the modified NP; IIclauses introduced by the complementiser ki with an overt pronominal element referring back to the modified NP; IIIclauses introduced by a relative pronoun with a gap in the position relevant to its grammatical function within the relative clause. These three types are illustrated in (153): (153) a su libru ki devo léghere (type I) ‘the book that I must read’ b sos pitzinnos ki lis appo datu sos jocátolos (type II) lit. ‘the boys that I gave the toys to them’ ‘the boys to whom I gave the toys’ c su professore kin su cale appo faeddatu (type III)

< previous page

page_293

next page >

< previous page

page_294

next page >

Page 294 ‘the teacher with whom I spoke’ Types I and II are by far the most common. We assume that all three types are derived from underlying structures of the following form, where PRONOUN represents a pronominal element which is coreferential with the whole NP: (154) [NP DET [N′ N′ [S′ [COMPki][S…PRONOUN…]]]] In constructions of types I and III, we take PRONOUN to be a relative pronoun (WH- item). As in interrogative constructions, this item or a phrase containing it is moved to the COMP position replacing the original complementiser ki, yielding constructions of type III. Alternatively, and more typically, the WH— expression is deleted (or left unrealised) and the complementiser ki remains in place, as in type I constructions. In type II constructions, PRONOUN is an ordinary personal pronoun which is realised in the same way as in independent clauses, for example, as a clitic (as in (153c)), a possessive or, more rarely, a disjunctive pronoun: (155) a s’autore k’appo lessu sos libros suos lit. ‘the author that I have read his books’ ‘the author whose books I have read’ b sa pitzinna ki so issitu kin issa lit. ‘the girl that I went out with her’ ‘the girl with whom I went out’ Since no WH—element is moved into the COMP position, the com-plementiser ki is retained, as in other subordinate clauses. Type I relative clauses occur predominantly when the modified NP corresponds to the subject or direct object of the relative clause, as in (153a) and the following examples: (156) a su libru ki m’interessat ‘the book that interests me’ b sos tenores ki sun cantande ‘the tenors that are singing’ c sa mákkina ki keljo comporare ‘the car that I want to buy’ d s’ómine k’appo vistu ‘the man that I saw’ In these cases, the deleted element consists solely of the relative pronoun, whose features are entirely recoverable from the NP as a whole. However, in casual speech we also find constructions of type I where the relativised element is an indirect object:

< previous page

page_294

next page >

< previous page

page_295

next page >

Page 295 (157)

a su calassu k’appo postu sas camisas lit. ‘the drawer that I put the shirts’ b s’ómine k’appo datu su dinari lit. ‘the man that I gave the money’ c sas cosas k’appo bisondzu lit. ‘the things that I have need’ In these cases, a preposition has apparently been deleted along with the relative pronoun. This possibility is restricted to prepositions such as in, a, and de which have little semantic content and which serve primarily to assign Case to the complement. In so far as such items are selected by the governing predicate, the missing prepositions in examples like (157) are recoverable according to purely linguistic criteria. On the other hand, we do not find examples such as (158) corresponding to (155b) above, which would involve deletion of a meaningful preposition (kin ‘with’) which expresses an adjunct relation and is therefore not recoverable: (158) *sa pitzinna ki so issitu ‘the girl that I went out (with)’ Relative clauses like those in (157) are more usually expressed by means of a type II construction in which the relativised item is represented as a clitic: (159) a su calassu ki b’appo postu sas camisas lit. ‘the drawer that I put the shirts there’ b s’ómine ki l’appo datu su dinari lit. ‘the man that I gave the money to him’ c sas cosas ki nd’appo bisondzu lit. ‘the things that I have need of them’ Indeed, type II relative clauses are most commonly used in cases where the pronominal element (PRONOUN in (154)) occurs in a position from which it cannot be extracted and subsequently deleted on its own (for example, as the complement of a preposition or as a possessive within a larger NP). To this extent, the use of a resumptive pronoun might be regarded as a ‘last-resort’ strategy which is exploited when the use of a type I construction is excluded or is marginal. However, constructions of type II are also possible when the relativised item is a direct object: (160) a su libru ki l’appo lessu ‘the book that I read’ b s’ómine ki l’appo vistu ‘the man that I saw’

< previous page

page_295

next page >

< previous page

page_296

next page >

Page 296 The use of a resumptive accusative pronoun appears to be rather more natural in a complement clause embedded within the relative clause, though again the movement-deletion strategy is not excluded: (161) su libru ki credo ki (l’)as lessu ‘the book that I think that you read’ When the complement clause is in the subjunctive, the use of a resumptive accusative pronoun tends to be preferred: (162) a unu libru ki non pesso ki l’appan mai lessu b ?unu libru ki non pesso k’appan mai lessu ‘a book that I do not think that they have ever read’ The situation regarding subject relativisation is less clear in so far as pronominal subjects are usually null in any case and are therefore indistinguishable from the gap left by movement and subsequent deletion of a WH—item. In simple relative clauses, the relativised subject cannot be realised as an overt pronoun, perhaps because of the emphatic effect associated with subject pronouns: (163) *s’ómine k’isse at furatu su dinari lit. ‘the mani that hei stole the money’ However, overt resumptive pronouns do appear to be possible as subjects of complement clauses within the relative clause, as an alternative to null subjects: (164) s’ómine ki credo ki (isse) at furatu su dinari ‘the man that I think stole the money’ Relative clauses of type III are uncharacteristic of colloquial Sardinian. In particular, the use of su cale (and its feminine and/or plural forms) illustrated in (153c) is restricted to formal or literary style and can perhaps be viewed as an imitation of Italian usage. Indeed, given our claim that ki in type I relatives is a complementiser, we may postulate that colloquial Sardinian does not have a set of relative pronouns. Expressions of the type su cale can function as subjects, direct objects or complements of prepositions, though as subjects or direct objects they appear to occur predominantly in nonrestrictive (appositive) relatives, as in (165a, b): (165) a Appo iscrittu una líttera a su presidente, su cale non m’at rispostu. ‘I have written a letter to the president, who has not replied to me.’

< previous page

page_296

next page >

< previous page

page_297

next page >

Page 297 b Cussa líttera, sa cale appo mandatu eris, bos contat tottu. ‘That letter, which I sent yesterday, tells you everything.’ c Sa mesa supra sa cale appo postu sa televisione est tunda. The table on which I put the television is round.’ Note that these forms are not preceded by accusative a when used as direct objects, as we would expect since they clearly contain an article. The only type III constructions which are current in informal registers are those where the relativised element has a locative function and is expressed by (in) ube ‘where’: (166) a sa domo (in) ube istat su duttore ‘the house where the doctor lives’ b sas biddas (in) ube colat su postale ‘the towns where the bus passes’ Thus, locative relatives exhibit all three construction types: type II as in (166), type III as in sas biddas ki bi colat su postale and, in very casual speech, type I as in sas biddas ki colat su postale . Inversion of the subject is common, though not obligatory, in relative clauses, particularly when the relative pronoun is overt as in (166) above. Inversion cannot apply in cases where the inverted subject could be construed as a direct object; e.g. (167a) cannot be interpreted as an inverted version of (167b): (167) a su pitzinnu k’at madzatu s’ómine ‘the boy that hit the man’ b su pitzinnu ki s’ómine at madzatu ‘the boy that the man hit’ However, inversion can apply in such cases when the subject is an NP of the type which takes accusative a when it has direct object status (see 2.2.6), as in (168a) which is distinguished from (168b) by the absence of a: (168) a su pitzinnu k’at madzatu Pretu ‘the boy that Peter hit’ b su pitzinnu k’at madzatu a Pretu ‘the boy that hit Peter’ Finite relatives normally occur in the indicative except when the existence of a referent for the modified NP is presented as hypothetical

< previous page

page_297

next page >

< previous page

page_298

next page >

Page 298 or is explicitly denied, in which case the subjunctive is generally preferred: (169) a Maria est kircande unu vistire ki I’ aggradet. ‘Mary is looking for a dress which pleases (subj.) her.’ b No’appo accatatu manc’una cratea ki siat cómmoda. ‘I have not found a single chair which is (subj.) comfortable.’ Sardinian also has infinitival relative clauses introduced by the item de, which we take to be a complementiser in this case, in which the subject is obligatorily null (with human reference, either generic or identified in terms of an entity which is salient in the discourse): (170) a Cussos sun sos libros de léghere. lit. ‘These are the books to read.’ ‘These are the books which should be read.’ b Appo fattu tottu sas cosas de fákere. lit. ‘I have done all the things to do.’ ‘I have done everything that needed to be done.’ As the translations of the above examples indicate, the infinitive in this construction generally carries a root modal interpretation, suggesting obligation. From a syntactic point of view, these constructions can be analysed in much the same way as finite relatives. Consider the following underlying structure: (171) [NP sos libros [S′ de [S léghere PRONOUN]]] To derive (170a), PRONOUN is moved to COMP and is deleted, leaving the complementiser de in place, just as in type I finite relatives. When the WH—item is overt, de (like ki) must be deleted: (172) a Kirco una borsa (in) ube pónnere su dinari. ‘I am looking for a bag in which to put the money.’ b Kirco una pinna kin sa cale iscríere una líttera. ‘I am looking for a pen with which to write a letter.’ Note that items of the su cale type cannot be used without a preposition in infinitival relatives: (173) *Kirco unu libru su cale léghere. lit. ‘I am looking for a book which to read.’ Relativised indirect objects can be realised as clitics or, in casual speech, be omitted altogether, just as in finite relatives: (174) a s’ómine de (li) dare su dinari ‘the man to give the money to’

< previous page

page_298

next page >

< previous page

page_299

next page >

Page 299 b su calassu de (bi) pónnere sa veste ‘the drawer to put the clothing in’ Infinitival relatives whose subject corresponds to the modified NP are somewhat marginal and are introduced by a instead of de. Such constructions appear to occur most naturally as complements of éssere: (175) a Est un’ómine a diféndere sas ideas suas. ‘He is a man to (= who will) defend his ideas.’ b Non sun dzente a fákere cosas gai. ‘They are not people to (= who would) do things like that’ Past and present participles can also be used to modify an NP with a function similar to that of relative clauses. This use of the past participle is restricted to unaccusative and transitive verbs, with a passive value in the latter case: (176) a Appo mortu su primu sórrike issitu dae sa tana. ‘I killed the first mouse come (= which came) out of the hole.’ b Appo lessu unu libru iscrittu dae un Italianu. ‘I read a book written by an Italian.’ c Sos letholos postos in su calassu sun nettas. ‘The sheets placed in the drawer are clean.’ There are no restrictions concerning the transitivity types of present participles which can be used adnominally, though typically such participles are non-stative: (177) a S’ómine travallande in cuddae ti keret faeddare. ‘The man working over there wants to talk to you.’ b Sa pitzinna biende su caffé est meta galana. ‘The girl drinking coffee is very pretty.’ c Su trenu arrivande como est in ritardu. ‘The train arriving now is late.’ The present participle in this construction clearly preserves its progressive value, identifying the intended referent in terms of an ongoing action or process in the immediate extralinguistic context. The modified NP always corresponds to the subject of the present participle.

< previous page

page_299

next page >

< previous page

page_300

next page >

Page 300 6.2.2 Free relatives and similar constructions The term ‘free relative’ is used here to refer to relative clause constructions in which the head of the modified expression is either null or consists of a pronominal element rather than a lexical item such as a common noun. Headless relatives typically consist of a definite article and a type I relative clause: (178) a Appo lessu su ki m’as mandatu. ‘I have read what you sent me.’ or ‘I have read the one that you sent me.’ b Appo intesu su k’est sutzessu. ‘I have heard what happened.’ We assume that the definite article occupies the determiner position of an NP which lacks an overt head and is modified by a relative clause of type I, as in (179): (179) [NP SOS libros[S′de[S léghere PRONOUN]]] When the definite article is masculine singular, as in (178), the expression can be either unrestricted in terms of its potential reference (as suggested by the translation with ‘what’) or it can be restricted to reference within a particular class usually defined by an antecedent in the preceding discourse. When the definite article is feminine and/or plural, only the latter interpretation is possible, its gender feature being determined by the antecedent. Thus, unless the wider discourse context provides an alternative antecedent, the headless relative in (180a) will be construed as referring to a particular letter, whereas that in (180b) can refer to any written material (e.g. a novel, poem, essay, etc.): (180) a Custa líttera est medzus de sa k’appo iscrittu eris. ‘This letter (f.) is better than the one (f.) I wrote yesterday.’ b Custa líttera est medzus de su k’appo iscrittu eris. ‘This letter (f.) is better than what (m.) I wrote yesterday.’ By the same token, (181) is deviant in the absence of an appropriate antecedent for the masculine plural expression: (181) *Custa líttera est medzus de sos k’appo iscrittu eris. ‘This letter (f.) is better than the ones (m.) I wrote yesterday.’

< previous page

page_300

next page >

< previous page

page_301

next page >

Page 301 These constructions can also be used to refer to human beings. In this case the masculine forms (both singular and plural) can have restricted or unrestricted reference, but the feminine forms normally require an antecedent unless the context precludes male reference, as in (182c): (182) a Su professore novu est prus furbu de su ki b’aíamus prima. ‘The new teacher is more crafty than the one we had before.’ b Sos ki travallan incue sun fortunatos. ‘The people who work there are lucky.’ c Sas ki sun gráidas non devían travallare. ‘People who are pregnant should not work.’ A similar construction is possible with demonstratives. In this case the range of potential referents is normally restricted in terms of an antecedent when inanimate, but can be either restricted or unrestricted with human reference: (183) a Cuddas meleddas sun prus durkes de custas k’appo comporatu. ‘Those apples are sweeter than these that I have bought.’ b Su professore novu est prus furbu de cuddu ki b’ aíamus prima. ‘The new teacher is more crafty than that one that we had before.’ c So kircande a cuddos k’an furatu sa mákkina mea. ‘I am looking for those who stole my car.’ Note that the demonstrative in (183c) must be introduced by accusative a. In the light of our discussion in 5.1.2, we postulate that when the demonstrative has human reference which is not restricted in terms of an antecedent, it occupies the head N position whereas in other cases it occupies the determiner position, the head N being null. On the other hand, the absence of accusative a in (184) suggests that definite articles in headless relatives always occupy the deter-miner position even when they have unrestricted human reference: (184) So kirkande (*a) sos ki an furatu sa mákkina mea. ‘I am looking for the people who stole my car.’ Both definite articles and demonstratives can be preceded by the universal quantifier tottu with essentially the same referential properties as those described above for the non-quantified cases:

< previous page

page_301

next page >

< previous page

page_302

next page >

Page 302 (185) Custu libru est prus interessante de tottu sos/cuddos k’appo iscrittu jeo. ‘This book is more interesting than all of those that I have written.’ b Su professore novu est prus furbu de tottu sos/cuddos ki b’aíamus prima. ‘The new teacher is more crafty than all those that we had before.’ c Custu libru est prus interessante de tottu su k’appo iscrittu jeo. ‘This book is more interesting than everything that I have written.’ d Tottu sos/cussos ki travallan incue sun fortunatos. ‘Everyone that works there is lucky.’ Tottu can also be used without an article with non-restricted inanimate reference: (186) Custu libru est prus interessante de tottu k’appo iscrittu jeo. ‘This book is more interesting than everything that I have written.’ The interrogative determiners cantu ‘how much, how many’ and cale ‘which’ can also introduce headless relatives, but with these items the complementiser ki must be absent: (187) a Potes mandicare cantu keres. ‘You can eat as much as you want.’ b Potes comporare cale ti piaghet. ‘You can buy whichever pleases you.’ Superficially these constructions resemble indirect questions, as in (188): (188) a No’isco cantu keres. ‘I do not know how much you want.’ b No’isco cale ti piaghet. ‘I do not know which pleases you.’ In indirect questions, we assume that the WH—item ( cantu or cale) originates in the complement clause and is moved to the COMP position, as discussed in 6.1.8. However, for the constructions in (187) we postulate that cantu and cale are base-generated in the

< previous page

page_302

next page >

< previous page

page_303

next page >

Page 303 determiner position of an NP which is modified by a relative clause, as in (189), just like the definite article in (179): (189) [NP [DET cantu/cale ] [N ] [S′ ki [S…WH-…]]] The intuitive motivation for this structural distinction is that the verbs in (188) require a propositional complement (expressible by a clause) whereas those in (187) require a complement which denotes a physical entity which must be realised as an NP. More concrete, syntactic evidence in support of this structural distinction is provided by cases where the element quantified by cantu is specified by the partitive clitic nde: (190) a No’isco cantu nde keres. ‘I do not know how much of it you want.’ b Nde potes mandicare cantu keres. ‘You can eat as much of it as you want.’ Assuming that clitic-placement occurs within the simple clause, the difference in the position of nde in (190) is naturally accounted for if we postulate underlying structures of the following form, where PRONOUN corresponds to nde: (191) a [S′ Non isco[S′ keres [NP cantu PRONOUN]]] b [S′ Potes mandicare [NP cantu PRONOUN[S′ ki keres WH-]]] The absence of the complementiser ki in (187) poses a slight problem for our analysis in that we have assumed so far that the complementiser is only deleted when it is replaced by an overt WH—expression. To accommodate such cases we must assume that the complementiser is also deleted when it is immediately preceded by a WH—expression. Further evidence in favour of the approach advocated above is provided by free relatives introduced by kie ‘who’, used here in the sense of ‘whoever’: (192) Ando a vídere a kie mi keret faeddare. ‘I will go and see whoever wants to talk to me.’ Note that kie in this construction must be introduced by accusative a when it introduces a direct object, in contrast to the superficially similar interrogative construction in (193): (193) No’isco (*a) kie mi keret faeddare. ‘I do not know who wants to talk to me.’ This contrast is naturally accounted for if we analyse kie in (192) as

< previous page

page_303

next page >

< previous page

page_304

next page >

Page 304 the head of the direct object NP modified by a relative clause, as in (194), whereas kie in (193) is the underlying subject of the interrogative complement which has been moved into the COMP position: (194) Ando a vídere a [NP [N kie] [S’ ki [S WH—mi keret faeddare]]] Again, we must assume that deletion of the complementiser ki is triggered by the presence of the immediately preceding WH—item kie rather than by movement of the unrealised relative pronoun into COMP. A similar analysis can be maintained for adverbial or oblique clauses introduced by WH—items, such as those in (195): (195) a Cando so intratu, issos sinke sun andatos. ‘When I entered, they left.’ b Lu devías fákere comente l’appo fattu jeo. ‘You should do it how (= the way that) I did it.’ c Semus andatos a (in) ube istat Juanne. ‘We went to where John lives.’ In these constructions, the clause introduced by the WH—adverb functions as a modifier or complement within the main clause, but the WH—adverb itself has a similar function within the dependent clause. Note also that the WH—adverb can be preceded by elements which are appropriate to its function in the main clause but not to its role in the dependent clause (e.g. in (195c) the presence of a ‘to’ indicates that (in) ube is a Goal complement of the movement verb andare rather than a complement of the stative verb istare ). A similar effect can be seen in adjunct expressions such as that in (196) where the presence of fintzas a indicates the end-point of a period which is compatible with the durative verb abbarrare ‘stay’ but not with the eventive verb issire ‘come/go out’: (196) Abbarro inoke fintzas a cando essis. lit. ‘I (will) stay here even to when (= until) you come out.’ These properties, which reflect those of relative clauses on nouns, can be accounted for by analysing the WH—adverb as the head of the construction modified by a clause containing a (phonetically unrealised) WH —item corresponding to the head adverb, as in (197): (197) [ADVP [ADV cando] [S′ ki [S so intratu [ADV WH—]]] The WH—item is moved to COMP where it is deleted, as is the

< previous page

page_304

next page >

< previous page

page_305

next page >

Page 305 complementiser ki due to the presence of the overt WH—item cando, as envisaged above for free relatives with cantu, cale and kie . Essentially the same analysis can be postulated for the adverbial clause in (198), except that ki is not deleted in this case since the head adverb como ‘now’ is not a WH—item: (198) Como ki ses arrivatu, potimus mandicare. ‘Now that you have arrived, we can eat.’ Headless relatives with definite or demonstrative articles or the quantifier tottu can take the infinitival form, in the same way as relatives with a lexical head (see 6.2.1, examples (170)): (199) a Custos sun sos de léghere. ‘These are the ones to read.’ b Cuddu libru est prus importante de custu de léghere como. ‘That book is more important than this one to read now.’ c Appo fattu tottu (su) de fákere. lit. ‘I have done everything to do.’ ‘I have done everything that needed to be done.’ However, the infinitive is not possible in free relatives and adverbial clauses introduced by WH—items (with or without the complemen-tiser de): (200) a *Appo comporatu cale (de) léghere. lit. ‘I bought which to read.’ b *Semus abbarratos (in) ube (de) dormire. lit. ‘We stayed where to sleep.’ c *Amus a mandicare cando (de) arrivare. lit. ‘We will eat when to arrive.’ In this respect, these constructions differ markedly from indirect questions and from adverbial clauses introduced by a preposition: No’isco cale léghere ‘I do not know which to read’, No’ isco (in) ube dormire ‘I do not know where to sleep’, Amus a mandicare prima de andare ‘We will eat before to go (= before we go)’. 6.2.3 Conditional and concessive constructions Conditional clauses are normally introduced by si ‘if in the Logudorese-Nuorese dialects, though ki is preferred in Campidanese. Generally, the tense of the condition clause is identical to that of the consequent, though they may differ in terms of aspect.

< previous page

page_305

next page >

< previous page

page_306

next page >

Page 306 Indicative present tense forms (including the present perfect and the future with áere) are used when the condition clause denotes a potentially realisable situation. The simple present is normally used when referring to events which are projected into the future: (201) a Si divento riccu, cómporo una bella domo. ‘If I become rich, I will buy a beautiful house.’ b Si su mere teléfonat, bi curro súbitu. ‘If the boss telephones, I will rush there immediately.’ Example (201b) can also be interpreted as a habitual statement ‘If (=whenever) the boss calls, I rush there immediately’. In such cases, when the future interpretation is intended, the use of future áere, in either or both clauses, is often preferred: (202) a Si su mere at a telefonare, bi curro súbitu. b Si su mere teléfonat, b’appo a cúrrere súbitu. c Si su mere at a telefonare, b’appo a cúrrere súbitu. ‘If the boss telephones, I will rush there immediately.’ Similarly, the explicit future form tends to be preferred with stative predicates, which do not readily allow a future interpretation for the present tense form: (203) a Si appo a éssere riccu, cómporo una bella domo. ‘If (ever) I am rich, I will buy a beautiful house.’ b Si appo a éssere riccu, appo a éssere cuntentu. ‘If (ever) I am rich, I will be happy.’ c Si Juanne venit, appo a éssere cuntentu.’ ‘If John comes, I will be happy.’ With stative predicates, the use of the simple present in the condition clause usually implies a present situation which is potentially actual, but not certain: (204) a Si Maria est in domo, potimus andare a la vídere. ‘If Mary is at home, we can go to see her.’ b Si tue bi keres andare, ti b’accumpandzo. ‘If you want to go there, I will accompany you.’ Similarly, the present perfect is used in the condition clause to denote an event which is potentially accomplished in the past: (205) a Si Maria sink’est andata, non la potimus vídere. ‘If Mary has left, we will not be able to see her.’ b Si Gavini at bitu tottu su vinu, nde cómporo átteru. ‘If Gavin has drunk all the wine, I will buy some more.’

< previous page

page_306

next page >

< previous page

page_307

next page >

Page 307 The imperfect indicative forms (including the pluperfect) are used when the condition is purely hypothetical or counterfactual: (206) a Si fippo riccu, comporaío una bella domo. ‘If I was rich I would buy a beautiful house.’ b Si fippo istatu riccu, aío comporatu una bella domo. ‘If I had been rich, I would have bought a beautiful house.’ c Si Juanne veníat, fakíamus sa festa. ‘If John came, we would celebrate.’ d Si Juanne fit vénnitu, aíamus fattu sa festa. ‘If John had come, we would have celebrated.’ e Si travallaío de prus, fippo prus cuntentu. ‘If I worked harder, I would be happier.’ f Si aío travallatu de prus, fippo istatu prus cuntentu. ‘If I had worked harder, I would have been happier.’ The simple imperfect is used to present imaginary situations in the present or future (as in (206a, c, e)) whereas the pluperfect has counterfactual past reference (as in (206b, d, f)). This hypothetical value of the imperfect and pluperfect indicative can also be seen in cases where there is no overt condit ion clause: (207) a (A postu tuo) comporaío una mákkina nova. ‘(In your position) I would buy a new car.’ b (A postu tuo) aío travallatu de prus. ‘(In your position) I would have worked harder.’ However, the imperfect can be used in conditionals with a purely past tense value, as in (208) which is ambiguous in the same way as (201b) above: (208) Si su mere telefonaíat, bi currío subitu. ‘If the boss telephoned, I would rush there immediately.’ ‘If (d whenever) the boss telephoned, I used to rush there immediately.’ The reduced imperfect forms of dévere ‘must’ (see 3.1.3) can also be used as a type of periphrastic conditional in both the condition and consequent clauses: (209) a Si Juanne diat vénnere, diamus fákere sa festa. ‘If John came, we would celebrate.’ b Si Juanne diat éssere vénnitu, diamus áere fattu sa festa. ‘If John had come, we would have celebrated.’ In counterfactual conditional constructions relating to the past, the

< previous page

page_307

next page >

< previous page

page_308

next page >

Page 308 normal pluperfect can be replaced, in either or both clauses, by a ‘double’ pluperfect in which the auxilary is repeated in its past participle form: (210) Si fis (istatu) vénnitu pruskitho, [if you+were (been) come earlier, aíamus (appitu) mandicatu imparis. we+had (had) eaten together] ‘If you had come earlier, we would have eaten together.’ The inclusion of this extra participle appears to be purely stylistic, having no effect on meaning, though its use is restricted to conditional constructions. According to Pittau (1972) this double pluperfect can also cooccur with reduced imperfect forms of dévere, thus giving the following range of possible ways of expressing the proposition ‘If you had gone to Cagliari, you would have concluded the matter’ (the item kke here is a pleonastic instance of the locative clitic nke; see 5.2.4): (211) a “Si fis (istau) andáu a CCasteddu, kk’aías (áppiu) finíu sacosa.” b “Si días ésser (istáu) andáu a CCasteddu, kke días áer (áppiu) finíu sa cosa.” (Pittau 1972:156–7) In hypothetical or counterfactual conditionals, the imperfect or pluperfect subjunctive can be used instead of the indicative in the condition clause (but not in the consequent clause), to give an optative effect similar to that conveyed by if only…in English: (212) a Si m’ésseret possíbile, abbaraío kin tecus. ‘If only it were possible for me, I would stay with you.’ b Si babbu ésseret istatu inoke, sa festa ki aíamus fattu! ‘If only father were here, what a feast we would have made!’ In such cases the consequent clause is frequently omitted: (213) a Si Juanne ésseret inoke! ‘If only John were here!’ b Si aeredzis travallatu de prus! ‘If only you had worked harder!’ In all the examples given above, the condition clause precedes the consequent clause, which is the unmarked order. The reverse order is also possible in most cases, though it tends to give less emphasis to the condition clause.

< previous page

page_308

next page >

< previous page

page_309

next page >

Page 309 The condition clause can be modified by fintzas ‘even’ as a means of implying that the situation described in the consequent clause has been or will be realised in spite of the situation described in the condition clause: (214) a Fintzas si proet, andamus a mare. ‘Even if it rains, we will go to the sea.’ b Fintzas si mi l’aían proibitu, l’aío fattu. ‘Even if they had forbidden me to, I would have done it.’ According to our informants, fintzas si normally takes the indicative, as in (214); in particular, it does not appear to be compatible with the optative use of the imperfect subjunctive illustrated in (212). On the other hand, mancari ‘even if, although’ always takes the subjunctive with a modal distinction indicated by tense similar to that found in indicative conditionals with si (see examples (201), (204)– (206)). With the imperfect or pluperfect subjunctive, mancari is equivalent to fintzas si; i.e. the clause which it introduces denotes a hypothetical or counterfactual situation: (215) a Mancari ésseret maláidu, no’andaío a lu vídere. ‘Even if he were ill, I would not go to see him.’ b Mancari mi l’áeren proibitu, l’aío fattu. ‘Even if they had forbidden me to, I would have done it.’ However, the use of the present (or present perfect) subjunctive normally gives a factual interpretation to the clause introduced by mancari, as with English although: (216) a Mancari siat maláidu, paret cuntentu. ‘Although he is ill, he seems happy.’ b Mancari mi l’appan proibitu, l’appo fattu. ‘Although they forbade me to, I did it.’ Other items which can introduce factive concessive clauses of the ‘although’ type are sende ki3 and belle ki. These items normally take the indicative and the choice of tense is based on temporal rather than modal factors: (217) a ki fit prus mannu de me, l’appo madzatu. ‘Although he was bigger than me, I hit him.’ b ki tendzo gana, non devo mandicare. ‘Although I am hungry, I must not eat.’

< previous page

page_309

next page >

< previous page

page_310

next page >

Page 310 The same pattern is found in reason clauses introduced by ja ki and sicomente ‘since’: (218) a fit prus mannu de me, non l’appo madzatu. ‘Since he was bigger than me, I did not hit him.’ b tendzo gana, mi mandico unu prattu de pasta. ‘Since I am hungry, I will eat a plate of pasta.’ With these items, the reason clause typically precedes the main clause, though the reverse order is also possible (as is also the case in conditional and concessive constructions). However, reason clauses introduced by the complementiser ca ‘because’ (sometimes preceded by proitte ‘why’) always follow the main clause: (219) a Non l’appo madzatu, (proitte) ca fit prus mannu de me. ‘I did not hit him, because he was bigger than me.’ b Mi mandico unu prattu de pasta, (proitte) ca tendzo gana. ‘I will eat a plate of pasta, because I am hungry.’ In cases where the main clause is negated, ca (without proitte) shows a potential scope ambiguity. In (219a) the ca clause is most naturally interpreted as being outside the scope of the negation (i.e. the size of the person in question constitutes the reason for my not hitting him), but in (220) the negation would normally be interpreted as denying the causal connection between the two clauses (i.e. I bought the book, but not because it was cheap), for purely pragmatic reasons (e.g. substitution of caru ‘expensive’ would favour the other reading): (220) No’appo comporatu custu libru ca fit baratu. ‘I did not buy this book because it was cheap.’ In practice, this ambiguity is resolved by intonation. When the reason clause is within the scope of the negation, as in (220), the whole sentence is treated as a single intonational phrase, with main stress at the end of the reason clause, but when the reason clause is outside the scope of the negation, as in (219), the two clauses constitute independent intonation phrases separated by a pause (indicated by a comma in (219)). We may also note that some speakers allow the use of proitte without ca to introduce reason clauses, with the same interpretive and prosodic properties observed for (219):

< previous page

page_310

next page >

< previous page

page_311

next page >

Page 311 (221) a Non l’appo madzatu, proitte fit prus mannu de me. ‘I did not hit him, because he was bigger than me.’ b Mi mandico unu prattu de pasta, proitte tendzo gana. ‘I will eat a plate of pasta, because I am hungry.’ The absence of ca in these cases can be accounted for by postulating either that proitte is treated as a complementiser for these speakers or that proitte retains its status as a WH—adverb, as in interrogatives, thus triggering deletion of the complementiser as proposed for the WH- adverbial clauses discussed in 6.2.2 (see examples (195)–(196)).

< previous page

page_311

next page >

< previous page

page_312

next page >

Page 312 7 Constituent-order variations 7.1 REORDERING PROCESSES 7.1.1 Left-dislocation The phenomenon of left-dislocation is illustrated in (1), where an element is placed before the ‘core’ part of the sentence and is duplicated by a clitic pronoun which indicates its function within the sentence: (1) a Su duttore, l’appo vistu. ‘The doctor, I have seen him.’ b Sa líttera, Maria l’at dza mandata. ‘The letter, Mary has already sent it.’ The function of left-dislocation is to establish or confirm the dislocated element (e.g. su duttore in (1a)) as the topic of the sentence. In some cases, it may also have the complementary effect of causing some other element to occur in sentence-final position where it can be naturally construed as the focus of the sentence (i.e. as the element which conveys new or particularly important information). Thus, in the canonical sentence Appo vistu su duttore, the natural focus is su duttore, whereas in (1a) this entity is presented as the topic while the verb assumes the status of focus. The prosodic properties of left-dislocated constructions suggest a rather loose syntactic relation between the dislocated phrase and the following sentence. Left-dislocated phrases are normally separated from the rest of the sentence by a pause, indicated by the comma in (1). Moreover, the dislocated element and the following clause constitute independent intonation phrases, each with its own primary stress, as indicated, very roughly, in (2a), whereas simple sentences

< previous page

page_312

next page >

< previous page

page_313

next page >

Page 313 of the canonical form contain only one item which bears main stress, as in(2b):1 (2) a b The position of left-dislocated phrases relative to other preclausal items reveals a paradox. In direct questions, the dislocated phrase must precede the WH—item: (3) a Su pane, in ube l’as postu? *In ube, su pane, l’as postu? ‘The bread, where did you put it?’ b A sos pitzinnos, itte lis as datu? *Itte, a sos pitzinnos, lis as datu? ‘To the boys, what did you give them?’ c Cudda líttera, a kie l’as mandata? *A kie, cudda líttera, l’as mandata? ‘That letter, to whom did you send it?’ Assuming that WH—items occur outside S but within S′ (e.g. in COMP, as proposed in 1.2.2), the dislocated phrases in (2a) must be adjoined to S′ or attached to some higher node. However, in finite subordinate clauses, left-dislocated items must follow the comple-mentiser, a fact which suggests that they are adjoined to S: (4) a Juanne at natu ki, sa líttera, l’aíat dza mandata. *Juanne at natu, sa líttera, ki l’aíat dza mandata. ‘John said that, the letter, he had already sent it.’ b Su mere m’at dimandatu si, sa líttera, l’aío dza mandata. *Su mere m’at dimandatu, sa líttera, si l’aío dza mandata. ‘The boss asked me whether, the letter, I had already sent it.’ c Su mere at a éssere arrenegatu, ca, sa líttera, non l’appo mandata. *Su mere at a éssere arrenegatu, sa líttera, ca non l’appo mandata. ‘The boss will be angry. because, the letter, I have not sent it.’ For the record, we may note that left-dislocation cannot occur in infinitival complements or indirect WH— questions:

< previous page

page_313

next page >

< previous page

page_314

next page >

Page 314 (5) a *Su mere m’at natu a, sa líttera, la mandare. b *Su mere m’at natu, sa líttera, a la mandare. ‘The boss told me, the letter, to send it.’ (6) a *No’isco kie, sa líttera, l’at mandata. b *No’isco, sa líttera, kie l’at mandata. ‘I do not know, the letter, who sent it.’ The problem raised by the examples in (2a)–(3) is not peculiar to Sardinian—it arises with left-dislocated constructions in other Romance languages and, to some extent, with topicalisation in English in so far as the topicalised phrase must follow the comple-mentiser in cases like I said that that letter I would not send even though other evidence suggests that the topicalised phrase occupies a position more peripheral than COMP (see Chomsky 1977). Consequently, we shall not attempt to resolve this matter here, but we shall concentrate on evidence from Sardinian which is relevant to questions concerning the nature of the leftdislocation phenomenon. A fundamental question is whether left-dislocation should be treated as a syntactic phenomenon at all. An alternative approach would be to view the dislocated expression as a separate, nonsentential utterance (not structurally linked to the ‘core sentence’, but related to it by general discourse principles) which typically announces an entity which is taken as the topic of the following sentence (as in (1) and (3)) but which can also be inserted parenthetically at certain major constituent boundaries (for example, before a subordinate S, as in (4)). Such an approach seems plausible at least for apparent instances of left-dislocation in spontaneous speech where the speaker enunciates the topic separately for reasons of performance, perhaps as an expression of surprise or simply to give himself time to formulate an appropriate statement about the entity in question. The question which then arises is whether there is a separate grammatical process of leftdislocation (definable in structural terms as part of the native speaker’s linguistic competence) which serves to alter the topic and focus relations within the proposition in the manner described above. Some insight into this question is provided by the use of accusative a and partitive de before left-dislocated NPs. When the leftdislocated item is a direct object NP of the appropriate type (e.g. a proper name, see 2.2.6), it is normally introduced by accusative a: (7) A Maria, l’appo vista. ?Maria, l’appo vista. ‘Mary, I saw her.’

< previous page

page_314

next page >

< previous page

page_315

next page >

Page 315 Similarly, indefinite plural or non-count nouns in left-dislocated position are usually introduced by partitive de: (8) a De vaccas, non nde appo vistu. ?Vaccas, non nde appo vistu. ‘Cows, I did not see any.’ b De vinu, nde appo bitu pacu. ?Vinu, nde appo bitu pacu. ‘Wine, I drank little of it.’ In so far as accusative a and partitive de have a purely syntactic function, their preferred presence in the above examples suggests that left-dislocation is governed by principles of syntax rather than of discourse. At the very least, the use of a in examples like (7) entails a prior decision on the part of the speaker to formulate the comment in such a way that the topic is represented by a direct (or possibly dative) object. On the other hand, the marginal acceptability of the versions without a or de is consistent with the discourse approach envisaged above, in that we would not expect an expression which acts as an independent utterance to display features appropriate to its syntactic function in the following utterance. Although it is difficult to elicit firm judgements from native speakers, it appears that the versions of (7)–(8) without a or de are more acceptable in contexts which are amenable to a performance account of the type envisaged above. More concretely, they appear to require a particularly long pause in the position of the comma and, typically, an interrogative contour (incomplete fall in pitch) on the dislocated expression. In particular, omission of the grammatical preposition seems to be favoured when the following sentence is introduced by a discursive element such as emmo ‘yes’: (9) a Maria… Emmo, l’appo vista. ‘Mary… Yes, I have seen her.’ b Pane… Emmo, nd’appo comporatu. ‘Bread… Yes, I bought some.’ When the left-dislocated phrase is an indirect object (e.g. dative or locative), the appropriate preposition is obligatory: (10) a A Orgósolo, non bi ando mai. *Orgósolo, non bi ando mai. ‘To Orgósolo, I never go there.’ b A sos pitzinnos, lis appo datu unu jocátulu. *Sos pitzinnos, lis appo datu unu jocátulu. To the boys, I gave them a toy.’

< previous page

page_315

next page >

< previous page

page_316

next page >

Page 316 c Dae cudda tana, unu sórrike nk’est issitu. *Cudda tana, unu sórrike nk’est issitu. ‘From that hole, a mouse came out of it.’ Versions of the above examples without the preposition are deviant even in circumstances where a performance (‘pause for thought’) account is plausible. In such circumstances, we often find ‘double’ dislocation, as in (11): (11) a Orgósolo…a inie, non bi ando mai. lit. ‘Orgosolo, to there, I never go there.’ b Sos pitzinnosz…a issos, lis appo datu unu jocátulu. lit. ‘The boys, to them, I gave them a toy.’ We propose, tentatively, that examples in which the dislocated NP is introduced by a preposition which indicates its syntactic function are instances of a syntactic dislocation process (where the dislocated phrase is structurally linked to the rest of the sentence) whereas examples which lack the relevant preposition are instances of the performance phenomenon described above. It is important to bear in mind the distinction between these two types of distinction when considering the structural properties of left-dislocation. Note, however, that this distinction does not, in itself, solve the problem raised above in connection with the examples in (3) and (4). In particular, we cannot maintain that examples in which the dislocated phrase precedes a WH—item are necessarily cases of the performance type since accusative a and partitive de can readily occur in such examples: (12) a A Maria, cando l’as vista? ‘Mary, when did you see her?’ b De pane, in ube nd’as comporatu? ‘Bread, where did you buy any?’ In 1.1.2, we noted that left-dislocation is also possible with subject NPs, with the null subject (or verbagreement) acting as the resumptive pronominal element, as can be seen most clearly in examples such as (13) where the dislocated NP corresponds to the subject of the subordinate clause: (13) Maria, credío ki fit inoke. ‘Mary, I thought that she was here.’ The use of an overt subject pronoun as the resumptive element tends to be avoided, perhaps because of the emphatic effects associated with disjunctive pronouns, so left-dislocation of the subject in simple sentences does not manifest itself in terms of the linear order of

< previous page

page_316

next page >

< previous page

page_317

next page >

Page 317 elements. Nevertheless examples like (14) can be pronounced with a stress and intonation pattern of the type in (2a) and can thus be taken as instances of left-dislocation: (14) Su duttore, no’est galu arrivatu. ‘The doctor, he has not yet arrived.’ Left-dislocation of the subject in simple sentences can be seen more clearly in cases like (15) where another dislocated element intervenes between the subject and the rest of the sentence: (15) Juanne, sa líttera, l’at dza mandata. ‘John, the letter, he has already sent it.’ Examples like (15) also illustrate the possibility of left-dislocating more than one element in the same sentence. Although in this particular case, dislocation of the first item could be an instance of the performance type, examples such as (16) where the first item is introduced by a functional preposition indicate that multiple leftdislocation of the syntactic type is possible, according to the criterion proposed above: (16) De vinu, a sos pitzinnos, non nde lis dao. ‘Wine, to the children, I do not give them any.’ This in turn suggests that syntactically left-dislocated items are introduced by a recursive system (e.g. adjunction to some projection of S) rather than generated under a designated node such as the TOPIC node proposed by Chomsky (1977) for topicalised phrases in English. As far as we can ascertain, the order of leftdislocated items is not fixed, though it may be affected by factors such as relative length, empathy or topicworthiness. Left-dislocation is not a clause-bound process, as can be seen from the following examples (also (13) above), where the phrase which precedes the main clause functions as an argument of the complement clause: (17) a Sa líttera, su mere m’at natu a la mandare. ‘The letter, the boss told me to send it,’ b Sa líttera, no’isco kie l’at mandata. ‘The letter, I do not know who sent it.’ c Sa líttera, Juanne at natu ki l’aíat dza mandata. ‘The letter, John said that he had already sent it.’ d Sa líttera, su mere m’at dimandatu si l’aío mandata. ‘The letter, the boss asked me if I had sent it.’ As we can see from (17b, d), left-dislocation can apply readily out of

< previous page

page_317

next page >

< previous page

page_318

next page >

Page 318 interrogative complements, which typically function as ‘islands’ with respect to movement processes. However, left-dislocation does appear to respect other island constraints. For instance, although some of the following examples may be marginally acceptable as instances of dislocation of the performance (‘pause for thought’) type, they are clearly much less natural than the examples in (17), a fact which suggests that adverbial clauses, relative clauses and con-junct clauses constitute islands with respect to syntactic leftdislocation: *Sa líttera, su mere at a éssere arrenegatu ca non l’appo mandata ‘The letter, the boss will be angry because I have not sent it’, *Su mere, mi so istikkitu cando l’appo vistu ‘The boss, I hid when I saw him’, *Custu libru, connosco s’ómine ki l’at iscrittu ‘This book, I know the man who wrote it’, *A Juanne, so issitu dae domo e l’appo vistu ‘John, I came out of my house and I saw him’. 7.1.2 Right-dislocation The phenomenon of right-dislocation is illustrated in (18): (18) a L’appo vistu, su duttore. ‘I have seen him, the doctor.’ b Maria l’at dza mandata, sa líttera. ‘Mary has already sent it, the letter.’ In right-dislocated constructions, the pause (indicated by the comma) is generally much less evident than in cases of left-dislocation and in written texts the comma is sometimes omitted (though we shall include it in all our examples). Also, the right-dislocated expression does not constitute an intonation phrase in its own right, but is relatively unstressed with a low, fairly flat intonation contour, roughly as in (19): (19) Some instances of right-dislocation can be viewed as incidental discourse phenomena, analogous to the performance cases of leftdislocation envisaged in 7.1.1, whereby the right-dislocated expression is added as an afterthought in order to clarify the referent of a pronoun. In cases where right-dislocation is exploited as a grammatical strategy for rearranging elements of the proposition in accordance with the organisation of the discourse, its function can be characterised in terms of defocalisation of the dislocated expression, with a concomitant shift in focus on to some other element in the

< previous page

page_318

next page >

< previous page

page_319

next page >

Page 319 sentence. Thus, in (18a) su duttore, which is the natural focus in the canonical sentence Appo vistu su duttore, is presented as given information and focus is shifted on to the verb vistu emphasising the fact that I actually saw the person in question. In this respect rightdislocation resembles left-dislocation (see our discussion of example (1a) in 7.1.1). However, by virtue of their position and their relative lack of stress, right-dislocated phrases cannot be used to introduce a new topic or to contrast the intended topic with other potential topics. In cases where the right-dislocated item is a phrase which would not normally occur in the clause-final, focused position in a sentence of the canonical form, the effect of right-dislocation is rather more subtle. Consider the following examples: (20) a Appo datu su jocátulu a sos pitzinnos. ‘I gave the toy to the boys.’ b L’appo datu a sos pitzinnos, su jocátulu. Both sentences in (20) allow, and indeed favour, focus on sos pitzinnos, as the final element within the core clause. They differ in that (20b) explicitly precludes the possibility of treating su jocátulu as part of the focus. Thus, whereas (20a) can also present the entire action as new information (i.e. the focused constituent is the VP datu su jocátulu a sos pitzinnos), the corresponding wide-focus reading for (20b) is one where the notion of doing something with the toy is given and the new information is the nature of this activity, specified by datu a sos pitzinnos. Even in cases where right-dislocation does not shift the focus on to another element, as in (20) on the reading where sos pitzinnos is the sole focus, it appears to give greater emphasis to the focused constituent. Thus (20b) would be more appropriate than (20a) in a situation where sos pitzinnos does not simply convey new information but is contrasted with some other expected receiver of the toy; see 7.2.1 for further discussion. The prosodic properties of right-dislocated constructions, and their associated discourse effects, provide the principal criteria for distinguishing right-dislocation of subjects from cases of inversion. Consider the examples in (21): (21) a b ‘The builder telephoned.’ Example (21a), which we take to be an instance of right-dislocation,

< previous page

page_319

next page >

< previous page

page_320

next page >

Page 320 places the focus on the verb yielding interpretations of the type The builder telephoned (rather than calling in person)’ or ‘The builder did telephone’. However, (22b) does not have the properties which we have associated with right-dislocation—the postverbal subject bears primary stress and is interpreted either as the focus itself, introducing the builder as a new entity in the discourse (for example, in response to a question like ‘Who telephoned today?’) or as part of a wider focus (for example, where the whole event is presented as new information). So far we have assumed that primary stress and focus are normally assigned to the final element within the core sentence. If we define the informal notion ‘core sentence’ as the minimal S, we may postulate that the inverted subject in (21b) forms part of the minimal S (see 7.1.3 for discussion) whereas right-dislocated phrases, including the subject in (21a), occur outside the minimal S. This hypothesis is borne out by the fact that inverted subjects (bearing primary stress and focus) must precede right-dislocated complements: (22) L’at mandata Juanne, cudda líttera. ‘John sent it, that letter.’ The converse order of subject and complement is possible, but only if the subject is unstressed and nonfocal, in which case it can be analysed as a dislocated element: (23) L’at mandata, cudda líttera, Juanne. Example (23) shows that right-dislocation, like left-dislocation, can apply to more than one element in the same clause. Further examples involving complements are given in (24), where datu bears main stress in both cases: (24) a Bi l’appo datu, su dinari, a s’avocatu. b Bi l’appo datu, a s’avocatu, su dinari. ‘I gave the money to the lawyer.’ The possibility of multiple right-dislocation and the absence of any clear restrictions on order suggests that right-dislocated phrases are introduced by a recursive system such as adjunction to S. Although we have argued that right-dislocated phrases are more peripheral than inverted subjects, we also have evidence that they are less peripheral than certain other phrases which may be plausibly construed as part of the S. In particular, right-dislocated phrases generally precede adverbial clauses:

< previous page

page_320

next page >

< previous page

page_321

next page >

Page 321 (25) a L’appo iscritta, sa líttera, pro protestare contra su guvernu. ??L’appo iscritta pro protestare contra su guvernu, sa líttera. ‘I wrote the letter to protest against the government.’ b L’appo intesa, sa notítzia, cando so andatu a su butteca. ??L’appo intesa cando so andatu a sa butteca, sa notítzia. ‘I heard the news when I went to the shop.’ Right-dislocated phrases can either precede or follow complement clauses and adjunct PPs: (26) a Lis appo natu, a sos duttores, ki s’iskina me dolíat. Lis appo natu ki s’iskina mi dolíat, a sos duttores. ‘I told the doctors that my back was hurting.’ b Lis appo dimandatu, a sos pitzinnos, de non gridare. Lis appo dimandatu de non gridare, a sos pitzinnos. ‘I asked the boys not to shout.’ (27) a L’appo vistu, su puale, in su dzardinu. L’appo vistu in su dzardinu, su puale. ‘I saw the bucket in the garden.’ b L’appo abberta, sa janna, kin unu crau. L’appo abberta kin unu crau, sa janna. ‘I opened the door with a nail.’ However, right-dislocated phrases cannot readily precede complement PPs: (28) a ??L’appo datu, su jocátulu, a sos pitzinnos. L’appo datu a sos pitzinnos, su jocátulu. ‘I gave the toy to the boys.’ b ??L’appo postu, su pane, supra sa mesa. L’appo postu supra sa mesa, su pane. ‘I put the bread on the table.’ In order to account for these judgements, let us assume that rightdislocated phrases are adjoined to the S node and that in those cases where another constituent follows the dislocated item, this constituent has undergone an ‘extraposition’ process which removes it from its canonical position within VP or S and adjoins it to the right of the maximal S, as shown in in (29) which corresponds to the examples in (27a): (29)

< previous page

page_321

next page >

< previous page

page_322

next page >

Page 322 This extraposition process can be thought of as a means of satisfying a requirement or tendency for certain constituents to occupy the absolute final position at the surface level. The evidence in (25)–(28) suggests that the preference for sentence-final position, relative to right-dislocated phrases, is virtually absolute with adverbial clauses, rather weaker with complement clauses and adjunct PPs, and negligible with complement PPs, a pattern which is reminiscent of that which we observed in 2.2.3 and 3.4.2 regarding the order of posthead phrases within NP and VP where the preference for a peripheral position seems to correlate with the syntactic complexity of the phrase in question and the ‘looseness’ of its semantic relation to the head. Our hypothesis that extraposition is involved in cases where the dislocated item is followed by another phrase is consistent with our assumptions regarding assignment of primary stress and focus. The prosodic properties of the examples in (26a) can be represented, again very roughly, as in (30): (30) a b In the non-extraposed version (30b), the complement clause and the main clause appear to be treated as a single intonation phrase with primary stress on the last element preceding the dislocated item (i.e. dolíat ) and with a range of possible foci extending leftwards from dolíat itself (‘I told him that my back was hurting’) to more inclusive domains (for example, a fuller description of the symptom, s’iskina mi dolíat, or the fact that I reported it natu ki s’iskina mi dolíat ). However, in the extraposed version (30a), focus is associated quite clearly with the matrix verb natu, giving an interpretation of the type ‘I told them about my backache’ (with the likely implication that the doctors ignored this symptom). Moreover, the extraposed clause in (30a) appears to function as an independent intonational phrase, assigning its own primary stress and focus to the final element, though with a lower overall pitch (not indicated in (30a)) reflecting its subordinate status. In 7.1.1 we noted that left-dislocation can occur across clause boundaries (see examples (17)). In the case of right-dislocation, decisive examples are hard to construct since the right-boundary of a complement clause almost always coincides with that of the main clause. Thus, in an example like Appo natu a su pitzinnu a non lu

< previous page

page_322

next page >

< previous page

page_323

next page >

Page 323 bíere, su vinu ‘I told the boy not to drink it, the wine’, we cannot tell whether su vinu is adjoined to the infinitival complement or to the main clause. However, to the extent that it is possible for an indirect object to follow an infinitival complement, as in Appo natu a non lu bíere a su pitzinnu ‘I said not to drink it to the boy’, we would expect to get (31a) if right-dislocation were unbounded whereas, in fact, the only possibility is the rather clumsy sentence (31b), where the dislocated item is adjoined to the complement clause: (31) a *Appo natu a non lu bíere a su pitzinnu, su vinu. b ?Appo natu a non lu bíere, su vinu, a su pitzinnu. We thus conclude that right-dislocation is a bounded process; i.e. right-dislocated items can only be adjoined to the immediate S node. A further point which emerges from (31) is that right-dislocation can occur within infinitival clauses, unlike left-dislocation (see the examples in (5)). Similarly, right-dislocation differs from leftdislocation in that it can apply in WH—complements (compare (6)): (32) No’isco kie l’at mandata, sa líttera. ‘I do not know who sent it, the letter.’ The boundedness of right-dislocation also entails that, in (33), the dislocated NP is adjoined to the adverbial clause, not to the main clause: (33) So ghiratu a domo cando las appo vistas, sas nues. ‘I returned home when I saw them, the clouds.’ This conclusion is borne out by the examples in (34), where the adverbial clause precedes the main clause: (34) a Cando las appo vistas, sas nues, so ghiratu a domo. b *Cando las appo vistas, so ghiratu a domo, sas nues. The range of phrases which can undergo right-dislocation appears to be determined by the availability of an appropriate clitic or null subject. Thus, PPs whose function can be represented by an adverbial clitic can be dislocated, as in (35): (35) a Bi so dza andatu, a sa butteca. ‘I have already gone there, to the shop.’ b Su sórrike nk’est issitu, dae sa tana. ‘The mouse came out of it, from the hole.’ c Nde so meta cuntentu, de cuddu regalu. ‘I am very happy with it, with that present.’ However, PPs which have no clitic counterpart, such as kin Maria

< previous page

page_323

next page >

< previous page

page_324

next page >

Page 324 and kene sa crae in the following examples, cannot be dislocated: Juanne est andatu kin Maria ‘John went with Mary’, L’appo abberta kene sa crae ‘I opened it without the key’. In particular, a disjunctive pronoun cannot be used as the resumptive pronoun, with or without repetition of the preposition in the dislocated phrase: (36) *Juanne est andatu kin issa, (kin) Maria. Indefinite expressions of a partitive type can be dislocated from the direct object position, with nde as the resumptive clitic, provided that the article or quantifier (if there is one) is left behind: (37) a Nde appo comporatu (meta), de pane. ‘I bought (much) bread.’ b Nde devímus léghere tres, de libros. ‘We must read three books.’ c Nde potes mandicare una, de meledda. ‘You may eat one apple.’ However, indefinite NPs cannot be dislocated as a whole, since there are no clitics which correspond to indefinite NPs which include an article; e.g. accusative clitics are always definite: (38) a *L’appo comporatu, meta pane. b *Los devímus léghere, tres libros. c *La potes mandicare, una meledda. By the same token, indefinite subjects cannot be left-dislocated, since null subjects are always interpreted as having definite reference: (39) a *At iscrittu cussu rumantzu, un italianu. lit. ‘He wrote that novel, an Italian.’ b *An secatu su jocátulu, tres pitzinnos. lit. ‘They broke the toy, three boys.’ We do not dismiss the possibility that examples like those in (39) are also ruled out by pragmatic factors, in so far as indefiniteness is incompatible with the defocalising effect of right-dislocation. Note that in (37) above, the presence of partitive de is obligatory. We suggested in 7.1.1 that this requirement holds also for cases of syntactic left-dislocation, exceptions being arguably instances of the performance type. However, with right-dislocation, this requirement is absolute even in cases where the dislocated phrase can be viewed as an afterthought which is added to clarify the reference of the pronoun. Similarly, accusative a is always obligatory before rightdislocated direct objects of the relevant type:

< previous page

page_324

next page >

< previous page

page_325

next page >

Page 325 (40)

Non l’appo vistu, a Juanne. ‘I have not seen John.’ The class of items which can be right-dislocated includes personal disjunctive pronouns. Examples involving first and second person singular pronouns raise an interesting complication. Although such examples can have the normal properties associated with rightdislocation (stress and focus on the final element before the dislocated item), as in (41a), it is also possible for the disjunctive pronoun to bear primary stress and focus, without a pause or intonational break, as in (41b): (41) a ‘I gave it to you.’ b ‘I gave it to you.’ We assume that in cases like (41b) the disjunctive pronoun is not dislocated, but remains in its canonical position within the VP. Possibly, (41b) is an instance of the so-called ‘clitic-doubling’ phenomenon, more widespread in certain other Romance languages such as Spanish, whereby certain types of complement NPs can or must be accompanied by a clitic with the same grammatical function. However, as we remarked in 5.1.1, it is also possible that ti in (41b) has an ethic function (denoting an interested party) as distinct from the indirect object (receiver) role expressed by a tie. Interestingly, third person disjunctive forms can also manifest the properties of a tie in (41b) when they are ‘doubled’ by the clitic bi (as a suppletive dative form, see 5.2.2) but not when the resumptive clitic is a normal dative form; e.g. issos in (42a) can bear main stress and focus, but (42b) appears to be possible only as a genuine case of right-dislocation: (42) a Bi l’appo datu a issos. ‘I gave it to them .’ b Cussa domo lis cumbenit, a issos. ‘This house suits them.’ The interest of this observation lies in the fact that bi has a range of ethic uses, analogous to those of first and second person clitics (see 5.2.4) which are not generally available to the normal third person dative clitics. At the same time, we should note that variants of (41b) and (42a) without the duplicate clitic are somewhat odd whereas (43),

< previous page

page_325

next page >

< previous page

page_326

next page >

Page 326 which corresponds to (42b) except that the clitic is absent and issos is focused, is fine: (43) Cussa domo cumbenit a issos. ‘That house suits them ’ This evidence suggests that although apparent cases of clitic-doubling are possible only when the clitic can be construed as having an ethic function, the use of an ethic clitic appears to have become grammaticalised as a means of ‘supporting’ a disjunctive pronoun whenever possible. Returning to genuine cases of right-dislocation, we have observed that they differ in a number of respects from cases of left-dislocation, principally in terms of their prosodic properties and discourse effects, apparent boundedness, and applicability within certain types of clause (e.g. infinitives and WH—complements). On the whole, these differences suggest that right-dislocated phrases are more closely related to the ‘coresentence’ than left-dislocated items, even if we abstract away from the ‘performance’ cases of leftdislocation. This impression might be expressed formally by postulating that rightdislocated phrases are adjoined to the S node (as we have proposed) but left-dislocated phrases are adjoined to S′, an analysis which is consistent with the fact that left-dislocated items precede WH—items in main questions (see 7.1.1 examples (3)) and with the order of leftdislocated items with respect to fronted constituents which we shall discuss in 7.1.4. For present purposes we shall adopt this analysis, while recognising that some of the differences mentioned above may be attributable to the obvious difference in linear order and its interaction with other principles of syntax, discourse and linguistic processing; for example, anticipatory expressions need to be emphasised whereas retrospective phrases reiterate information which is already given, unbounded displacement to the right may cause parsing difficulties which do not arise with displacement to the left, the presence of a preclausal phrase may be syntactically incompatible with other preclausal elements of a more functional nature (such as complementisers and WH—items). A fundamental similarity between the two types of dislocation, which distinguishes them from the other reordering processes which we shall discuss in the following sections, is that neither has any appreciable effect on the core sentence. From the point of view of prosody and discourse, the core sentence has its own main stress, intonation contour and focus. Syntactically, neither type of dislocation imposes any restrictions on the internal structure of the core

< previous page

page_326

next page >

< previous page

page_327

next page >

Page 327 sentence, which is always well formed and functionally complete, except to the extent that it must contain an appropriate pronominal expression. In particular, left-dislocation is compatible with the presence of a preverbal subject (unlike fronting, see 7.1.4, 7.1.5). Similarly, right-dislocation is not inhibited by the presence of postverbal complements (unlike inversion of the subject, see 7.1.3), except to the extent that some clausal or peripheral elements of the VP must be extraposed to the right of the dislocated phrase, as discussed above. 7.1.3 Inversion We argued in 7.1.2 that placement of the subject in a postverbal position can be achieved by two different syntactic processes, rightdislocation and inversion, which are distinguished primarily in terms of stress and focus assignment. Right-dislocated subjects are always unstressed and non-focal whereas inverted subjects bear primary stress and are interpreted as (part of) the focus. The examples on which this proposal was based were given in (21), repeated below, where (21a) involves right-dislocation of the subject and (21b) is an instance of inversion: (21) a b ‘The builder telephoned.’ Our hypothesis that the examples in (21) are the result of distinct syntactic processes, rather than a single process which can be exploited with different prosodic and discourse effects, is motivated by the fact that right-dislocation of complements systematically defo-calises the element concerned. Thus, if (21b) were an instance of dislocation, we would have to complicate our account of leftdislocation to allow a focal interpretation in just those cases where the affected element is a subject. Syntactic evidence in support of this distinction is provided by the fact that inversion is inhibited by the presence of a postverbal complement whereas right-dislocation is not. Thus, (44) is acceptable with the intonation contour and associated focus represented in (44a), but not with the inversion properties indicated in (44b):

< previous page

page_327

next page >

< previous page Page 328 (44)

page_328

next page >

a

b Examples like (45), where the postposed subject precedes the complement, are unacceptable under either interpretation: (45) *At telefonatu su mastru de muru a Juanne. Note that this restriction only applies when the complement occurs within the VP; for example, inversion is possible (with focal stress on the subject) if the complement is dislocated, as in (46): (46) a A Juanne, l’at telefonatu su mastru de muru. b L’at telefonatu su mastru de muru, a Juanne. Note also that this restriction applies with unaccusative verbs, such as piághere ‘please’, whose subject is underlyingly postverbal (see 3.2.3). Thus, although (47a) is well formed as an underlying structure, it is ungrammatical as it stands and must be ‘rescued’ either by raising the subject to the preverbal position or by removing the complement by dislocation or fronting: (47) a *Piághet su casu a sos sórrikes. b Su casu piaghet a sos sórrikes. c A sos sórrikes, lis piaghet su casu. d Lis piaghet su casu, a sos sórrikes. e A sos sórrikes, piaghet su casu. ‘Cheese pleases mice.’ For presentational convenience, we shall use the term ‘inversion’ to cover cases where the underlyingly postverbal subject of an unaccusative verb is not moved to the preverbal position as well as cases like (21b) where the subject is postposed from its underlying preverbal position. We shall return to the question of possible structural differences between the two cases later in this section. Complement clauses and adjunct phrases do not prevent inversion as long as they follow the postverbal subject: (48) a At natu Juanne ki Maria fit maláida. ‘John said that Mary was ill.’ b Nos at dimandatu su mere de non pippare. ‘The boss asked us not to smoke.’ c Est arrivatu Pretu kin unu sorrisu mannu. ‘Peter arrived with a big smile.’

< previous page

page_328

next page >

< previous page

page_329

next page >

Page 329 d Est vénnitu Juanne pro pacare s’affittu. ‘John came to pay the rent.’ Such cases can be accommodated by appealing to the extraposition process envisaged in 7.1.2 to account for examples where a rightdislocated phrase precedes such expressions (see examples (25)–(27) above). We have also found a few attested examples in which a fairly long subject is inverted to the right of a postverbal complement, as in (49): (49)a “A propositu de amicos, una die fin bènnios a saludare a Elias tottu cuddos de sa troppa chi l’aian iscaminau.” lit.‘Speaking of friends, one day came to greet Elias all those of the group who had led him astray.’ [EP: 43] b “Elias non podiat dormire: li pariat chi l’esseret arribau in fatza s’alenu caente de Mallena, corcada tra tzia Annedda e tziu Berte.” lit. ‘Elias could not sleep: it seemed to him that arrived in his face the warm breath of Maddalena, lying between Aunt Annedda and Uncle Berte.’ [EP: 61] Although the subjects in (49) appear to occupy the position typical of right-dislocated phrases, it is clear from the context that they require a focal interpretation of the sort associated with inversion. The acceptability of such examples appears to be related to a general stylistic tendency to place ‘heavy’ constituents in final position, which, in this case, overrides the restriction against inversion in the presence of a complement. We shall not propose a formal analysis of this phenomenon here, but simply note that all apparent violations of this restriction are cases where either the complement or the inverted subject itself is an element which favours a sentence-final position for stylistic reasons. A second syntactic difference between inversion and rightdislocation of the subject involves cases where the subject is indefinite. As we noted in 7.1.2, indefinite NPs cannot be dislocated. However, although an impersonal construction of the type discussed in 3.2.2 is generally preferred, straightforward inversion of indefinite subjects is possible, particularly (it seems) when the features of the subject coincide with the ‘default’ agreement features of the finite verb and past participle (i.e. third person singular and, with perfective unaccusatives, masculine):

< previous page

page_329

next page >

< previous page

page_330

next page >

Page 330 (50)

a At telefonatu una pitzinna. ‘A girl telephoned.’ b Est arrivatu unu politzottu. ‘A policeman arrived.’ Thus, whenever an indefinite subject occurs postverbally, it is always stressed and focal. Following analyses proposed for similar phenomena in other Romance languages (see, for example, Belletti and Rizzi 1981, Jaeggli 1982, Burzio 1986), we assume that inverted subjects with verbs of the telefonare type (i.e. those which are not unaccusative) are adjoined to the right of the VP, as in (51) :2 (51) [S [VP [VP nos at telefonatu] su duttore]] ‘The doctor telephoned us.’ For cases involving unaccusative verbs, such as arrivare, the simplest hypothesis would be to assume that the subject remains in its underlying position (as a sister to V under V′), as in (52): (52) [S [VP est [V′ arrivatu su duttore]]] ‘The doctor arrived.’ Direct empirical evidence in support of this structural distinction is difficult to obtain because of the general incompatibility between inversion and the presence of a postverbal complement.3 Moreover, this distinction seems to preclude a uniform formulation of this restriction since the position of the inverted subject relative to that of a potential complement is not the same in the two cases. An alternative approach would be to postulate, contrary to what we have assumed so far, that the NP under V′ in (52) cannot receive nominative Case in situ; for example, because the VP node prevents it from being governed by INFL, which is external to VP. Accordingly, su duttore in (52) must move to the preverbal subject position to receive Case, from where it can be optionally postposed and adjoined to the VP in the same way as with verbs of the telefonare type, thus re-establishing the underlying order but with a surface structure of the type in (51). An advantage of this approach is that it permits a uniform account of the restriction concerning inversion and the presence of a postverbal complement, which can now be stated as a local structural constraint which rules out adjunction of an NP to the right of a VP which contains a postverbal constituent, as in (53): (53) *[VP [VP V XP] NP] Thus, underlying structures such as (54a) below are ungrammatical at

< previous page

page_330

next page >

< previous page

page_331

next page >

Page 331 the surface level because the postverbal subject lacks Case, whereas (54b) violates the condition in (53): (54) a *Est arrivata Maria a sa festa. b *Est arrivata a sa festa Maria. ‘Mary arrived at the feast.’ Empirical support for this approach is provided by impersonal existential constructions, of the type discussed in 3.2.2, which appear to respect a weaker version of the restriction discussed above. Although such constructions are impossible when the verb is followed by a direct object, as shown in (55), an indefinite NP which corresponds to the subject of an unaccusative verb can precede the indirect object, as in (56) which is the impersonal counterpart of (54a): (55) a *B’at cantatu tres tenores una cathone. b *B’at cantatu una cathone tres tenores. ‘There sang a song three tenors.’ (56) B’at arrivatu tres pitzinnas a sa festa. ‘There arrived three girls at the feast.’ In 3.2.2 we argued that the indefinite ‘subject’ in such constructions does not receive nominative Case from INFL (as evidenced by the lack of verbal agreement) but is licensed in some way by the clitic bi. Consequently, the considerations which we used to exclude (54a) (inability of INFL to assign Case to the subject within V′) simply do not apply to postverbal ‘subjects’ in the impersonal construction. We do not wish to insist on the validity of the account proposed above since it runs counter to the assumption made in most of the relevant literature on other languages that apparent cases of inversion with unaccusative verbs are a direct reflection of the underlying constituent order. Indeed, it conflicts with the assumption which we have adopted elsewhere in this book (for example, in our discussion of infinitives with nominative subjects in 6.1.5) that nominative Case can be assigned directly to a subject within the VP. Nevertheless, this approach does offer a descriptively coherent analysis of the clear cases, particularly with respect to the difference between impersonal and non-impersonal constructions, and, to this extent, may be viewed as a counter-argument to the consensus view that postverbal subjects of unaccusative verbs remain in their base-generated position within V′.

< previous page

page_331

next page >

< previous page

page_332

next page >

Page 332 7.1.4 NP fronting Fronting of constituents to the sentence-initial position (without a resumptive pronominal element of the sort found in left-dislocated constructions) is a widespread phenomenon in Sardinian and is possible with many types of phrases. However, for reasons which we shall discuss in 7.1.5, fronting of NPs shows a clear set of properties which are obscured to some extent in cases where this phenomenon involves other categories. For this reason, we shall base our preliminary discussion of this process on cases involving NPs in order to provide a working hypothesis within which other instances of fronting can be examined. The first point to note is that fronted NPs always bear primary stress and have a strongly focal interpretation similar to that conveyed by ‘clefting’ in English: (57) Su duttore appo vistu. lit. ‘The doctor I saw.’ ‘It is the doctor that I saw.’ In fronted constructions, the ‘core’ part of the sentence is unstressed with a fairly low, flat intonation contour, roughly as in (58): (58) Note in particular that the core part of the sentence does not constitute an independent intonation phrase with a stressed-focal element of its own—compare (58) with the intonation pattern proposed for leftdislocated constructions in 7.1.1, repeated in (59): (59) Similarly, fronted NPs are not separated from the rest of the sentence by a pause. All types of NP can be fronted, including indefinite NPs: (60) a Una mákkina appo comporatu. ‘A car I bought.’ b Pane appo mandicatu. ‘Bread I ate.’ c Meta vinu amus bitu. ‘A lot of wine we drank.’

< previous page

page_332

next page >

< previous page

page_333

next page >

Page 333 Note that mass and plural nouns without an article (such as pane in (60b)) are not introduced by partitive de when fronted. Similarly, the N′ cannot be fronted on its own (with or without de) leaving the indefinite article or quantifier behind: (61) a *De vinu amus bitu meta. b *Vinu amus bitu meta ‘Wine we drank much.’ However, the indefinite article or quantifier can be fronted on its own, provided that the N′ is dislocated (to the right or left), as in (62): (62) a Meta nde amus bitu, de vinu. b (De) vinu, meta nde amus bitu. We suggest that the fronted item meta in (62) is not simply the quantifier itself but rather an NP consisting of the quantifier and a null N′ which corresponds to both the clitic nde and the dislocated phrase; i.e. (62a) is derived from the dislocated structure (63) by fronting of the NP [meta ]: (63) [S [S ndei amus bitu [NP meta [N′ ]]] de vinui] This hypothesis is supported by analogous examples like those in (64), where the head N (or N′) has been extraposed by the process discussed in 2.2.8, since su tuo and sa manna cannot be analysed as constituents unless they contain a null head N or N′ given that possessives and attributive adjectives occupy the postnominal position: (64) a Su tuo appo madzatu, de cane. lit. ‘The yours I beat of dog.’ ‘It was your dog that I beat.’ b Sa manna appo comporatu, de mákkina. lit. ‘The big I bought of car.’ ‘It was the big car that I bought.’ On the basis of this evidence, we conclude that fronting, unlike dislocation, cannot apply to subconstituents of the NP. Example (62b) shows that fronting of an NP can co-occur with leftdislocation. This is shown more clearly in (65): (65) A sos pitzinnos, unu jocátulu lis appo datu. lit. ‘To the boys, a toy I gave them.’ ‘To the boys, it was a toy that I gave them.’ Note that the fronted item must follow the left-dislocated phrase:

< previous page

page_333

next page >

< previous page

page_334

next page >

Page 334 (66) *Unu jocátulu, a sos pitzinnos, lis appo datu. In this respect, fronted NPs resemble preposed WH—items, which must also follow left-dislocated phrases; see 7.1.1, examples (3). There are a number of other properties common to NP fronting and WH—movement which distinguish these two processes from leftdislocation. For example, fronting (like WH—movement) can only apply to one element within the same clause: (67) a *A sos pitzinnos unu jocátulu appo datu. lit. ‘To the boys a toy I gave.’ b *Unu jocátulu a sos pitzinnos appo datu. lit. ‘A toy to the boys I gave.’ Similarly, fronting and WH—movement are mutually exclusive: (68) a *A kie su jocátulu as datu? b *Su jocátulu a kie as datu? ‘To whom did you give the toy’ Also, fronted items preclude the presence of a preverbal subject, just like preposed WH—items in main clauses: (69) a *Su duttore Juanne at vistu. lit. ‘The doctor John saw. b *Cale duttore Juanne at vistu? ‘Which doctor did John see?’ In such cases, the subject must either remain unspecified or be dislocated or inverted: (70) a Su duttore at vistu. b Juanne, su duttore at vistu. c Su duttore at vistu, Juanne. d Su duttore at vistu Juanne. A further similarity is that fronted NPs, like WH—items, are incompatible with the interrogative particle a (see 1.2.2), even though fronting is often associated with an interrogative interpretation (cf. 7.2.2):4 (71) a *Su duttore a as vistu? b *A su duttore as vistu? ‘Did you see the doctor?’ (72) a *Cale duttore a as vistu? b *A cale duttore as vistu? ‘Which doctor did you see?’

< previous page

page_334

next page >

< previous page

page_335

next page >

Page 335 On the other hand, there are a number of differences between NP fronting and WH—movement. First, whereas fronted NPs are always stressed and focused, to the exclusion of other elements in the remainder of the sentence, WH—items do not normally bear the main stress except in echo-questions or in cases where the content of the rest of the clause is pragmatically presupposed. Clearer syntactic differences between the two processes are evident in cases where the moved element originates in a subordinate clause. Whereas WH—movement is ostensibly an unbounded process, as shown in (73), long-distance NP fronting is generally impossible out of finite complements, though it appears to be more acceptable when the complement is infinitival: (73) Cale libru as natu k’as lessu? ‘Which book did you say that you have read?’ (74) a *Cuddu libru appo natu k’appo lessu. ‘That book I said that I have read.’ b ?Cuddu libru nos an dimandatu de léghere. ‘That book they told us to read.’ c ?Cuddu libru appo provatu a léghere. ‘That book I tried to read.’ Conversely, NP fronting may occur within a declarative finite complement, in which case the NP is placed after the complementiser as in (75), whereas WH—movement can only apply in complements which are interpreted as indirect questions and is incompatible with the presence of a complementiser: (75) Juanne m’at natu ki su duttore aíat vistu. ‘John told me that the doctor he had seen.’ (76) Appo dimandatu a Juanne cale duttore aíat vistu. ‘I asked John which doctor he saw.’ Also, while indirect WH—questions can be infinitival (e.g. No’isco itte fákere ‘I do not know what to do’), NPs cannot be fronted within an infinitival clause regardless of their position relative to the complementiser: (77) a *Maria m’at natu a su duttore vídere. b *Maria m’at natu su duttore a vídere. ‘Mary told me the doctor to see.’ Finally, WH—movement within a complement does not preclude the presence of a preverbal subject (though inversion is often preferred),

< previous page

page_335

next page >

< previous page

page_336

next page >

Page 336 but a preverbal subject is absolutely excluded with fronted NPs even in subordinate clauses: (78) a No’isco in ube Juanne at istikkitu su dinari. ‘I do not know where John has hidden the money.’ b *Credo ki su dinari Juanne at istikkitu. ‘I think that the money John has hidden.’ We shall not attempt to draw any firm conclusions from this evidence concerning the structural position of fronted NPs, but it may be useful to consider some of the options which are available. First, we may note that the similarities illustrated in (65)–(72) all concern incompatibility of fronted NPs and WH—items with each other and other preverbal elements. A natural way of accounting for these similarities would be to assume that fronted NPs and WH—items occupy the same designated position (e.g. COMP), distinct from that of left-dislocated elements. However, other solutions are available which do not entail structural isomorphism between NP fronting and WH—movement. For instance, the facts in (65)–(72) might be subsumed under a general verb-second principle which is not formulated in terms of specific structural positions, though such a principle would have to ignore certain ‘peripheral’ elements such as left-dislocated phrases as in (65). Also, it is not clear how the contrast in (78) could be accounted for within this approach. Leaving this problem aside, it might be postulated that WH—items are moved to the COMP position whereas fronted NPs are adjoined to the left of S. This approach would immediately account for the position of the fronted NP (after the complementiser) in cases like (75). Moreover, it might provide the basis for a principled explanation of the difference with respect to boundedness illustrated in (73)–(74); for example, if apparent unbounded movement involves successive applications of movement within a local domain, we might appeal to a condition which prevents successive applications of adjunction rules. If fronted NPs are assumed to occupy the same position as WH—items (e.g. COMP or some position within the COMP system), then clearly the differences in (73)–(78) must be attributed to factors which are independent of the structural relation between the moved item and other elements in the sentence. It is possible that the apparent boundedness of NP fronting is a pragmatic, rather than syntactic, phenomenon. Given that fronting of an NP requires the remainder of the sentence to be completely defocalised and destressed, we might expect NP fronting to become progressively less acceptable in proportion to the syntactic complexity of the material

< previous page

page_336

next page >

< previous page

page_337

next page >

Page 337 which must be defocalised and the amount of information which it expresses. Thus, the data in (74) might be taken as evidence that defocalisation of a string containing an infinitival complement lies just within the threshold of acceptability but this limit is overstepped when the defocalised string includes a finite complement clause. Some support for this approach is provided by the fact that longdistance fronting appears to be more acceptable in cases where the defocalised material within the finite complement has little semantic content, as in (79) where it consists solely of the copular verb éssere: (79) ?Sa mudzere de Juanne credo k’est. ‘John’s wife I think that she is.’ Note that these pragmatic considerations will not restrict unbounded WH—movement since, as we observed above, WH—movement does not normally divert stress and focus away from elements within the propositional part of the sentence. A further point which may be relevant is that WH—items must always be moved to the COMP of the clause which is interpreted as a question. Thus, in examples like (73) we do not have the option of moving the WH—phrase to the COMP of the complement clause. However, NP fronting does not have a specific target position in this sense. Consequently there is no need for unbounded or successive applications of fronting in cases like (74a) since the focusing effect can be achieved simply by moving the NP to the beginning of the immediate clause, as in cases like (75). The fact that the fronted NP in (75) co-occurs with and follows the complementiser remains a problem, though, as we noted in 7.1.1, the same problem arises with left-dislocated phrases (see the examples in (4)) which appear to occupy a more peripheral position than either fronted NPs or WH—items. A possible solution to both of these problems would be to allow finite complementisers such as ki to occur before higher projections of S (e.g. S’), the leading intuition being that the essential function of complementisers is to mark the onset of material which is to be interpreted as part of the complement clause. Such a proposal would also account for the ungrammaticality of (77b), where the infinitival complementiser follows an element which ‘belongs’ to the complement clause, whereas (77a) can plausibly be excluded by a general condition that pre-infinitival complementisers cannot be separated from the verb by anything other than clitics. We offer no explanation for the fact that the preverbal subject must be null when an NP is fronted within a subordinate clause (as in (78b)) except to note that it reflects the pattern of fronting (and WH—movement) in main clauses, and to suggest that it is the possibility of

< previous page

page_337

next page >

< previous page

page_338

next page >

Page 338 an overt preverbal subject in indirect questions (e.g. (78a)) which stands in need of explanation. 7.1.5 Fronting of other constituents The fronting process described in the preceding section in relation to NPs can also apply to PPs, APs, ADVPs and to VPs governed by an auxiliary: (80) a Supra sa mesa l’appo postu. ‘On the table I put it’ b Istrakku ke poleddu so. ‘As tired as a donkey I am.’ c Lestru-lestru curren. lit. ‘Quickly quickly they run.’ ‘They run very fast.’ d Mandatu sa líttera appo. ‘Senttheletterlhave.’ However, complement clauses (finite or infinitival) cannot be fronted: (81) a *A cantare provo. ‘To sing I try.’ b *Ki su trenu est in ritardu credo. ‘That the train is late I believe.’ The only infinitives which can be fronted are those dependent on a modal auxiliary: (82) a Mandicare su casu keljo. lit. ‘Eat the cheese I want.’ ‘What I want is to eat the cheese.’ b Andare a Núgoro devo. lit. ‘Go to Nuoro I must.’ ‘What I must do is go to Nuoro.’ Given our argument in 3.3.3 that infinitives dependent on auxiliaries are simply VPs rather than clauses (S or S′), we postulate that fronting applies only to categories which are (maximal) projections of lexical heads. Moreover, our evidence suggests that it can apply to all such categories. All the examples in (80) and (82) have the same prosodic features as the cases of NP fronting discussed in 7.1.4; i.e. main stress and pitch-fall on the fronted item with a fairly low, level contour throughout the rest of the sentence. Similarly, the fronted item is interpreted as the focus of the sentence with an effect similar to that of cleft or

< previous page

page_338

next page >

< previous page

page_339

next page >

Page 339 pseudo-cleft constructions in English; e.g. It is on the table that I put it, Very fast is how they run. Fronting of a VP or an AP, as in (80b, d), tends to be more natural when the fronted item consists solely of the head V or A, in which case fronting is typically associated with an interrogative interpretation, as in (83); see 7.2.2 for further discussion: (83) a Mandicatu as? ‘Have you eaten?’ b Istraccu ses? ‘Are you tired?’ Also there are some cases where a sentence-initial PP or AP does not bear primary stress and focus—these will be discussed in 7.1.6. Whenever fronted items do bear primary stress and focus, they show the same syntactic restrictions as fronted NPs whch we described in 7.1.4. In particular, they cannot co-occur with fronted WH—items, interrogative a or preverbal subjects: (84) a *Proitte istraccu ses? ‘Why are you tired?’ b *Itte mandicatu as? ‘What have you eaten?’ (85) a *A istraccu ses? ‘Are you tired?’ b *A mandicatu as? ‘Have you eaten?’ (86) a *Supra sa mesa Juanne l’at postu. ‘On the table John put it.’ b *Istraccu Juanne est. ‘Tired John is.’ c *Ballatu Juanne at. ‘Danced John has. Like NPs, other phrases can be fronted within a finite complement clause and placed after the complementiser: (87) a Credío ki maláidu fis. ‘I thought that you were ill.’ b Appo natu ki arrivatos sun. ‘I said that they have arrived.’ They also show the same apparent properties of boundedness as fronted NPs, long-distance fronting out of a finite clause being

< previous page

page_339

next page >

< previous page

page_340

next page >

Page 340 generally impossible when the item in question is a complement of a full, lexical verb (as in (88)) but rather more acceptable when the remainder of the embedded clause consists only of a copular or auxiliary verb (as in (89)) or when the embedded clause is infinitival (as in (90)): (88) *Supra sa mesa appo natu ki l’appo postu. ‘On the table I said that I put it.’ (89) a ?Maláidu appo natu k’est. ‘III I said that he was.’ b ?Supra sa mesa credo k’est. ‘On the table I think that it is.’ c ?Travallande paret ki sun. ‘Working it seems that they are.’ (90) ?Supra sa mesa appo natu a lu pónnere. ‘On the table I said to put it.’ This evidence suggests that the fronting process illustrated in (80) is identical to the NP fronting process discussed in 7.1.4; i.e. there is a single fronting process which can apply to all types of XP. Our assumption that examples like those in (80b, d), (82) and (83) are derived by fronting of the VP or AP requires further discussion since some grammarians have characterised such constructions in terms of ‘postposition of the auxiliary’ (or copula); see particularly Pittau (1972:142–3) who cites examples of the type in (83) as evidence of the mobility of the verb in Sardinian (by which he clearly means mobility of the auxiliary or copula). Our principle reason for adopting the fronting approach to such constructions is the evidence in (84)–(89), which reflects exactly the same syntactic restrictions and possibilities as examples involving NP fronting, and the fact that these constructions exhibit the same prosodic properties. If examples like (83) were derived by postposition of the auxiliary or copula, it is not at all obvious why these similarities should obtain. Also, in so far as examples like (89c) are possible, they cannot be derived by postposing since the ‘postposed’ element paret ki sun is clearly not a constituent. Similarly, examples like (91a) involving two auxiliaries can only be derived by fronting of the lowest VP in (91b) according to the structural analysis of auxiliaries proposed in 3.4.1, since det éssere does not form a constituent: (91) a Arrivatu det éssere (su trenu). b Su trenu [VP det [VP éssere [VP arrivatu]]] ‘The train must have arrived.’

< previous page

page_340

next page >

< previous page

page_341

next page >

Page 341 Having argued that examples like those in (83) are derived by fronting, the next question is whether the fronted element is a phrase (VP or AP) or simply the lexical head (V or A). This question has some theoretical importance in view of recent claims (see for example Chomsky 1986 and Baker 1988) that the constraints on movement of lexical heads are fundamentally different from those which apply to movement of phrasal categories. Potential evidence in favour of the head-movement hypothesis is provided by examples like those in (92) where a complement of the fronted element is left behind: (92) a Arrivatu est a sa festa. lit. ‘Arrived he has at the feast.’ b Cuntentu est de su regalu. lit. ‘Happy he is with the present.’ This construction is also possible when the complement is a clause (finite or infinitival): (93) a Irmenticatu an ki oje est festa. lit. ‘Forgotten they have that today is a holiday.’ b Sicuru est ki Maria est in domo. lit. ‘Sure he is that Mary is at home.’ c Resissitu ses a abbérrere sa janna. lit. ‘Managed you have to open the door.’ d Prontos sun a ghirare a domo. lit. ‘Ready they are to return home.’ However, direct object NPs cannot be left stranded in this way unless they are introduced by accusative a: (94) a *Vistu an s’intzidente. lit. ‘Seen they have the accident.’ b Vistu an a Maria. lit. ‘Seen they have Mary.’ The examples in (92)–(94) have the same stress pattern as other fronted constructions (i.e. primary stress on the fronted item) and like the examples in (83) tend to be used predominantly as yes/no questions (e.g. ‘Has he arrived at the feast?’, ‘Is he happy with the present?’, etc.), though an emphatic declarative interpretation is also possible given an appropriate context (e.g. ‘He has arrived…’, ‘He is happy with the present’, etc.). According to the head-movement approach, the range of complements which can be stranded in this way may be defined in terms of

< previous page

page_341

next page >

< previous page

page_342

next page >

Page 342 Case. Following Baker (1988), let us assume that when a verb (or other head) is moved, it loses its capacity to assign Case. In examples like (92)–(93), fronting of the head is possible since the stranded complements are categories which do not require a Case feature (PPs or clauses). Similarly, examples like (94b) can be accommodated by adopting Jaeggli’s (1982) proposal for Spanish that accusative a provides a means of assigning accusative Case to the direct object independently of the verb. However, in (94a) the direct object must receive accusative Case directly from the verb, thus precluding fronting of the verb in this instance. An alternative approach is to postulate, contrary to appearances, that the fronted items in (92)–(94) are in fact maximal projections and that the stranding effect is due to a process which allows certain complements to be extraposed from the VP or AP prior to fronting (alternatively, the complements can be reanalysed as occurring outside the minimal VP or AP). Within this perspective, the Case-theoretical requirements envisaged above may be reformulated as a condition on the extraposition or reanalysis process; i.e. extraposition or reanalysis can only apply to complements which are not dependent for Case on the governing head. Thus, direct objects of the type in (94a) cannot be extraposed and must be fronted along with the rest of the VP, as in (95): (95) Vistu s’intzidente an. lit. ‘Seen the accident they have.’ This appeal to extraposition or reanalysis may seem somewhat spurious, but there is some empirical evidence to support this approach. Consider the following examples where the fronted element is accompanied by a modifier: (96) a Dza arrivatu est a sa festa. lit, ‘Already arrived he has at the feast.’ b Meta contentu est de su regalu. lit. ‘Very happy he is with the present.’ c Mai vistu appo a Maria. lit. ‘Never seen I have Mary.’ d *Dza vistu aían s’intzidente. lit. ‘Already seen they had the accident.’ Clearly, in these examples the fronted expression is not simply a lexical head. Moreover, according to the general schema proposed in 0.2 and elaborated with regard to the VP system in 3.4.1, modifiers of

< previous page

page_342

next page >

< previous page

page_343

next page >

Page 343 the type in (96) are external to the projection (V′ or A′) which contains the canonical complement position, as represented in simplified form in (97): (97) [XP…MODIFIER…[X′ X COMPLEMENT]] Thus, at least for examples like (96), the complement must be construed as occurring in a position more peripheral than the canonical complement position in order for the modifier and head to be fronted as a constituent. In other words, appeal to some sort of extraposition or reanalysis process is motivated for these cases. Alongside constructions of the type in (92)–(94) we find examples where the auxiliary is preceded by a clitic corresponding to the ‘stranded’ complement, which we take to be cases of rightdislocation: (98) a Arrivatu b’est, a sa festa. lit. ‘Arrived there he has, at the feast.’ b Cuntentu nd’est, de su regalu. lit. ‘Happy of it he is, with the present.’ (99) a Irmenticatu l’an, ki oje est festa. lit. ‘Forgotten it they have, that today is a holiday.’ b Sicuru nd’est, ki Maria est in domo. lit. ‘Sure of it he is, that Mary is at home.’ c Resissitu bi ses, a abbérrere sa janna. lit. ‘Succeeded at it you have, to open the door.’ d Prontos bi sun, a ghirare a domo. lit. ‘Ready for it they are, to return home.’ (100) a Vistu l’an, s’intzidente. lit. ‘Seen it they have, the accident.’ b Vista l’an, a Maria. lit. ‘Seen her they have, Mary.’ These examples have the same prosodic properties as those in (92)– (94) and allow the same range of possible interpretations (for example, they tend to be interpreted as yes/no questions). In these cases, the involvement of a process (right-dislocation) which ‘removes’ the complement from the VP or AP is signalled by the presence of a clitic. Thus, although the fronted elements in (98)– (100) appear to consist solely of a lexical head, they can plausibly be

< previous page

page_343

next page >

< previous page

page_344

next page >

Page 344 analysed as maximal projections which contain an empty complement node which corresponds to both the clitic and the dislocated phrase, as in (101) as a representation of (100a): (101) The fact that (100a) is grammatical, unlike (94a), shows that rightdislocation is not subject to the Case restriction envisaged for the putative extraposition or reanalysis process in (92)–(94). We also assume that extraposed or reanalysed complements occupy a less peripheral position than right-dislocated phrases (e.g. they are ad-joined to the VP or AP rather than to the S node). Leaving such detailed differences aside, we propose that the examples in (92)–(94) can be analysed in essentially the same way as (101), by fronting of a phrasal category from which the complement has been removed. Examples such as (80b, d) and (95) above show that fronting can apply to VPs and APs, and examples like (96) require us to postulate some sort of restructuring process. Moreover, such examples appear to have the same syntactic, prosodic and interpretive properties as those which ostensibly involve fronting of the lexical head. We therefore conclude that there is no motivation for a separate headfronting rule distinct from the general XP fronting process. A final observation which we may make in this connection is that inverted subjects cannot be fronted along with the verb, even when the verb is unaccusative; for example, application of fronting to (102a) can only yield (102b), not (102c): (102) a Est arrivatu Juanne. lit. ‘Has arrived John.’ b Arrivatu est Juanne. c *Arrivatu Juanne est. This is rather surprising if we adopt the ‘standard’ view that postverbal subjects of unaccusative verbs occur within the V′ at all stages of the derivation. However, the ungrammaticality of (102c) can be accounted for if we adopt our suggestion in 7.1.3 that subjects of unaccusative verbs must move out of the VP to acquire Case and can be ultimately adjoined to the VP by the inversion process which applies with ‘ordinary’ verbs, provided we make the additional assumption that inverted subjects are adjoined to the VP which includes the auxiliary, as in (103):

< previous page

page_344

next page >

< previous page

page_345

next page >

Page 345

(103) Given this structure, arrivatu Juanne cannot be fronted to give (102c) since it does not form a constituent. 7.1.6 Fronting and other preposing phenomena At the beginning of 7.1.5, we mentioned that certain phrases can occur sentence-initially without the focus or stress normally associated with fronted items. Typical examples are locative or temporal PPs, as in (104): (104) a Supra sa mesa appo vistu unu cantu de pane. ‘On the table I saw a piece of bread.’ b Dopo s’iverru venit su beranu. ‘After winter comes spring.’ c Peri sa janna est intratu Juanne. Through the door entered John.’ Although these examples can be realised and interpreted with the properties characteristic of fronted constructions, the more normal pattern is for the final element to bear primary stress and to be interpreted as focus. For these examples, it might be argued that no movement is involved, the sentence-initial position being simply one of the positions in which circumstantial adverbial expressions can be

< previous page

page_345

next page >

< previous page

page_346

next page >

Page 346 base-generated. However, we also find similar cases where the initial PP is a complement of the verb: (105) Supra sa mesa appo postu una tiádza de randa. ‘On the table I put a lace tablecloth.’ Moreover, adjectival complements of éssere can also be placed in initial position with a contrastive focus on the postposed subject, as in (106) which would be an appropriate retort to an allegation of madness on the part of somebody else: (106) Maccu ses tue! lit. ‘Mad are you.’ ‘ You are the one who is mad.’ Possibly, identificational constructions such as (107), where the verb agrees with the following NP rather than with the NP which is ostensibly the subject, can be analysed as instances of the same phenomenon (see 3.2.5): (107) Su mere ses tue! lit. ‘The boss are you.’ ‘ You are the boss.’ For expository purposes we shall use the term ‘preposing’ to refer to the phenomena illustrated in (104)– (106), and possibly (107), reserving the term ‘fronting’ for cases where the affected element bears primary stress and focus. Part of our task is to determine whether preposing and fronting are distinct syntactic processes or whether they are manifestations of the same syntactic phenomenon which can give rise to different prosodic and discursive effects. From the point of view of discourse, preposing shows greater affinities with dislocation than with fronting. In particular, preposing of an element which would normally occupy the final position allows some other item to bear primary stress and focus (for example, the direct object in (104a) and (105), and the inverted subject in (104b, c) and (106))—indeed in the latter cases preposing is a necessary precondition for inversion in so far as inversion is normally inhibited by the presence of a postverbal expression (see 7.1.3). In examples like (106), the effect of focus on the final element is clearly contrastive, but in cases like (104) the effect is more typically presentational, introducing some new entity into the discourse. Conversely, in (106) the preposed item must denote a property which is contextually presupposed whereas the preposed expressions of place or time in (104)–(105) may designate new situations which serve to contextua-

< previous page

page_346

next page >

< previous page

page_347

next page >

Page 347 lise the information expressed in the focal part of the sentence. Thus (104)–(105) could be used to initiate a discourse, whereas (106) (with primary stress on tue ) could only occur in a context such as the following: (108) A: Mudzere mea est macca. B: No. Maccu ses tue, ca non l’apprétzias. ‘My wife is mad.’ ‘No. You are mad, because you do not appreciate her.’ Another typical function of preposing is to establish a link between the event described in the rest of the sentence and the focus of the preceding utterance, as in the second sentence in (109): (109) Eris amus arrostitu unu porkeddu. Kin sos restos amus fattu unu brodu. ‘Yesterday we roasted a suckling pig. With the leftovers we made a broth.’ This effect is similar to that which can be achieved by left-dislocation, as in (110), interpreted as a continuation of the first sentence in (109): (110) …Sos restos, los amus datu a su cane. ‘…The leftovers, we gave them to the dog.’ Similarly, the combination of preposing and inversion for contrastive effect in (106) is mirrored by the reply in (111), where the presupposed element is left-dislocated: (111) A: Maria at fattu su brodu. B: No. Su brodu, l’at fattu Juanne. ‘Mary made the broth.’ ‘No. John made the broth.’ Left-dislocated PPs can also be used with a contextualising function analogous to the preposed expressions in (104): (112) Supra sa mesa, b’appo vistu unu cantu de pane. ‘On the table, there I saw a piece of bread.’ In examples like (104) the preposed phrase may be pronounced with a complete intonation contour of its own (like left-dislocated elements) or as part of the pretonic sequence of the sentence: (113)

< previous page

page_347

next page >

< previous page

page_348

next page >

Page 348 However, in cases like (106), only the latter pattern is possible: (114) The relevant factor here seems to be that in (104) the preposed expression is an adjunct whereas in (106) it is a complement. More generally, we postulate that the preposed expression and the rest of the sentence can be treated as separate intonation phrases only if the sentence is functionally complete without the preposed expression. Indeed, in examples like (105), where the preposed item is a complement PP, the prosodic structure in (113b) seems to be highly preferred. Note that this condition is always satisfied in cases of leftdislocation since the dislocated item is realised as a resumptive pronoun within the core sentence. From a syntactic viewpoint, preposing differs crucially from fronting in that preposed items can be followed by a preverbal subject, at least in some cases: (115) a Supra sa mesa Juanne at vistu unu cantu de pane. ‘On the table John saw a piece of bread.’ b Dopo s’iverru sos pastores tunden sas berbekes. ‘After the winter the shepherds shear the sheep.’ c Supra sa mesa Juanne at postu una tiádza de randa. ‘On the table John put a lace tablecloth.’ However, inversion of the subject seems to be obligatory if the verb lacks a postverbal complement; e.g. we do not have examples like those in (116) corresponding to (104b, c) and (106): (116) a *Dopo s’iverru su beranu venit. ‘After winter spring comes.’ b *Peri sa janna Juanne est intratu. ‘Through the door John came in.’ c *Maccu tue ses. ‘Mad you are.’ Moreover, in such cases the subject cannot be simply omitted. Thus, examples like the following can only be interpreted as instances of fronting, with main stress on the initial constituent:5 (117) a Dopo s’iverru venit. ‘It is after winter that it comes.’ b Peri sa janna est intratu. ‘It is through the door that he came in.’

< previous page

page_348

next page >

< previous page

page_349

next page >

Page 349 c Maccu ses. ‘You are mad .’ Whereas fronting requires the preverbal subject position to be empty, the restriction on preposing seems to be that the postverbal complement position must be occupied by an overt element, either the underlying complement as in (115) or an inverted subject as in (104b, c) and (106), which can be interpreted as the focus of the sentence. Thus, obligatory inversion in the context of a preposed phrase is motivated by a need to fill the postverbal position (to provide an appropriate focus) whereas in fronted constructions inversion is triggered by a need to vacate the preverbal subject position, a requirement which can equally well be satisfied by omission or dislocation of the subject. Preposed expressions which allow a preverbal subject also allow fronting of the complement, giving a more dramatic effect to the focus on the fronted phrase, which must follow the preposed expression and bear main stress: (118) Supra sa mesa, unu cantu de pane appo vistu. ‘On the table, a piece of bread I saw.’ Similarly, WH—phrases can be placed after preposed items of this type: (119) Supra sa mesa, itte as vistu? ‘On the table, what did you see?’ Examples like (118)–(119) appear to favour a prosodic pattern of the type in (113a) where the preposed element constitutes an independent intonation phrase (as indicated by the comma). Also, the preposed PP can be reiterated by the locative clitic bi, without any appreciable difference in terms of appropriateness within the discourse, in which case the construction can be analysed as an instance of left-dislocation. Although judgements are rather murky, the options just mentioned appear to be more highly preferred when the preposed PP is a complement of the verb; i.e. fronting or WH—movement of some other constituent seems to favour left-dislocation rather than simple preposing of the PP: (120) a Supra sa mesa, una tiádza de randa ?(b’) appo postu. ‘On the table, a lace tablecloth I put.’ b Supra sa mesa, itte ?(b’) as postu? ‘On the table, what did you put?’

< previous page

page_349

next page >

< previous page

page_350

next page >

Page 350 Preposing of attributive complements such as maccu in (106) appears to be incompatible with fronting or WH—movement of another element. The main point which we have tried to establish in the above discussion is that the various phenomena which we have subsumed under the label ‘preposing’ differ quite sharply from the fronting phenomenon described in 7.1.4, 7.1.5 in terms of their prosodic properties, their function within the discourse and the syntactic restrictions which govern them. On the whole, preposing shows greater affinities with leftdislocation. Indeed, it might be maintained that the constructions in (104), at least with the intonation pattern in (113a), are straightforward instances of left-dislocation, if we postulate that the resumptive clitic is only required when the core sentence would otherwise be functionally incomplete; i.e. the clitic is optional when the dislocated element is an adjunct. In the other cases discussed, the similarity with left-dislocation is rather more tenuous and less amenable to an analysis which treats preposing as a subcase of dislocation. In particular, preposing of the attribute in copular constructions such as (106)–(107) may well be a separate process in so far as it goes hand in hand with inversion to yield ‘mirror-image’ constructions which conform to a verb-second pattern, whereas left-dislocation and preposing of adjuncts do not systematically induce verb-second effects. Thus, if there is a verb-second constraint which operates within some syntactic domain, it might be concluded that preposed adjuncts and left-dislocated phrases occur outside this domain whereas preposed attributes (like fronted elements) occur within it. A difficulty with such a strict dichotomy is that preposed PP complements appear to induce verb-second effects in the context of a fronted constituent or WH—item (in so far as the examples in (120) are mildly deviant without the resumptive clitic) but can cooccur with a preverbal subject (as in (115c)), though judgements of the relevant examples are not always clear. Possibly this indeterminacy reflects a tendency to treat certain peripheral complements as if they were adjunct phrases, subject to pragmatic considerations which affect the sentence as a whole. We leave open the question of whether the various preposing phenomena discussed above constitute a homogeneous syntactic process, and we make no claims regarding the structural position of preposed items. We shall, however, assume that these preposing processes are fundamentally different from the fronting process described in 7.1.4, 7.1.5 and that they should be distinguished either in terms of the derived position of the element involved or in terms of

< previous page

page_350

next page >

< previous page

page_351

next page >

Page 351 the status of the corresponding empty category within the core part of the sentence. 7.2 STRESS, FOCUS AND RELATED ISSUES 7.2.1 Assignment of stress and focus Throughout our discussion so far we have assumed that stress and focus are typically associated with the final element of the clause. Some deviation from this general pattern is permissible, even in sentences of the canonical form. For example, some items with a low information content (notably carcunu ‘someone’ and carki cosa ‘something’) tend to be unstressed even in final position, as in (121) where primary stress would normally fall on intesu: (121) Su cane at intesu carcunu/carki cosa. ‘The dog heard someone/something.’ Conversely, some determiners and quantifiers can bear primary stress (indicated by italics in (122)) even though they occupy the initial position within the NP: (122) a Keljo cudda mákkina. ‘I want that car.’ b Appo comporatu duas collanas. ‘I bought two necklaces.’ The use of contrastive stress to focalise other non-final elements in canonical sentences is also marginally possible, as in (123), but the more usual strategy is either to front the focused item or to reorder elements in such a way that the focused item occurs in final position: (123) a ?Juanne at datu una mákkina a su frate. ‘John gave a car to his brother.’ b ?Lukía at furatu cudda collana. ‘Lucy stole that necklace.’ Preverbal subjects can bear primary stress and focus more readily, as in (124): (124)

‘ John did it.’ Possibly this is an instance of the phenomenon illustrated in (123). However, an alternative possibility is that the subject in (124) has

< previous page

page_351

next page >

< previous page

page_352

next page >

Page 352 undergone the fronting process described in 7.1.4 (though this has no effect on linear order) and is thus stressed and focused in the same way as other fronted items. We have no hard syntactic evidence to support this analysis, but there is no obvious reason to suppose that subjects cannot undergo fronting in the same way that they can undergo dislocation and WH—movement. For those cases where stress and focus do correlate with the final position, the notion ‘final element’ needs to be defined more carefully. As far as stress is concerned, we postulate that main sentencestress is assigned to the accented syllable of the last word in sentences of the canonical type. Focus, on the other hand, is clearly not a property of a particular syllable nor necessarily of a single lexical item, but may be construed as a property of a syntactic constituent or the information which it conveys. Thus, our leading generalisation may be formulated as a principle which associates focus with the final constituent of the sentence. However, often there may be several phrases which qualify as final constituents. For example, in (125) Juanne is clearly a final constituent, but so too is the more inclusive NP su frate de Juanne and the VP madzatu su frate de Juanne (we ignore here functional items such as determiners and prepositions): (125) Appo [VP madzatu [NP su frate de [NP Juanne]]] ‘I hit John’s brother.’ In fact, all of the constituents identified above are potential foci; for example, (125) is an appropriate reply to any of the following questions: ‘Whose brother did you hit?’, ‘Who did you hit?’, ‘What did you do?’ Whether the entire sentence can be interpreted as the focus is a question which we shall examine in 7.2.2. Note that main sentence-stress falls on Juanne regardless of which final constituent is the intended focus. Possibly, differences in the choice of intended focus can be conveyed by prosodic variations within the pretonic sequence, but we shall not investigate this matter here. The effects of some of the reordering processes discussed in the preceding sections follow naturally from the account proposed above for canonical sentences. For example, the fact that inverted subjects usually bear main stress and are interpreted as the focus or part of the focus (see 7.1.3) can be seen as a consequence of their sentence-final position. Left-dislocated constructions and the instances of preposing discussed in 7.1.6 also conform to this general pattern, at least as far as the ‘core’ part of the sentence is concerned. Right-dislocated constructions appear to deviate from this pattern in that the dislocated phrase, though final, is unstressed and non-focal. This problem

< previous page

page_352

next page >

< previous page

page_353

next page >

Page 353 can be overcome by assuming that the general principles of stress and focus assignment operate within a certain syntactic domain (e.g. the minimal S) and that right-dislocated items occur outside this domain (e.g. adjoined to S). However, the properties of fronted constructions are more difficult to accommodate within this pattern in that stress and focus are assigned exclusively to the initial phrase, which, in many cases, would occur in sentence-final position in a canonical construction. A possible solution to this problem, which we propose with some reservations, is to make a distinction between ‘unmarked’ or ‘default’ focus, associated with whatever constituent is final within the minimal S, and ‘marked’ focus realised as a [+focus] feature which can be assigned freely to any constituent and which typically induces fronting or, more marginally, attracts contrastive stress in situ (as in (122)–(123) above), but in either case precludes assignment of focus or stress to any other element by the default principle. A potential advantage of this approach is that it allows us to treat fronting and WH—movement as the same syntactic process (e.g. movement to COMP) in spite of the fact that WH—items do not necessarily bear primary stress (see 7.1.4, p. 335). We could postulate that this process applies to items which bear the [+focus] or [+WH] feature, or both (as in questions of the ‘echo’ type), and that primary stress is associated exclusively with the [+focus] feature in such cases. Intuitively, the focusing effect achieved by fronting is more ‘dramatic’ than that which is associated with a clause-final element. For instance, although unu mariane is most naturally interpreted as the focus in (126a) as well as in (126b), the former is a fairly neutral statement in which unu mariane represents the most salient information whereas the latter would be appropriate only in a context which invites a contrast (e.g. ‘It was a fox, not a dog’) or where the speaker wishes to draw particular attention to this entity (for example, to convey a sense of urgency): (126) a Appo vistu unu mariane. ‘I saw a fox.’ b Unu mariane appo vistu. Within the approach envisaged above, this difference could be accounted for in terms of the distinction between marked and unmarked focus. Alternatively, it might be attributed to the fact that fronting allows the intended focus to be identified more precisely than in cases where focus is associated with the final constituent. As we observed in example (125), potential foci may range from the

< previous page

page_353

next page >

< previous page

page_354

next page >

Page 354 minimal final constituent to more inclusive phrases which are final within the sentence. Similarly, in (126a) the VP vistu unu mariane is a possible focus (for example, within an account of the speaker’s experiences while walking in the woods). In this connection, we may recall our observation in 7.1.2 (see our discussion of examples (20) repeated below in (127)) that right-dislocation tends to enhance the effect of focus even though it does not change the phrase which is final within the minimal S: (127) a Appo datu su jocátulu a sos pitzinnos. ‘I gave the toy to the boys.’ b L’appo datu a sos pitzinnos, su jocátulu. Given that focus on sos pitzinnos in (127b), as in (127a), falls within the unmarked or default principle, this evidence suggests that the ‘dramatic’ effect correlates with narrowing of the potential focus (in contrast to information which is explicitly non-focal) rather than with the marked [+focus] feature. We may also note that in fronted constructions, the intended focus does not always coincide with the whole of the fronted phrase. In (128), main stress would typically fall on Juanne, but the intended focus can be either Juanne or su babbu de Juanne; for example, as a response to either ‘Whose father did you see?’ or ‘Who did you see?’: (128) Su babbu de Juanne appo vistu. lit. ‘John’s father, I saw.’ Moreover, stress can occur on an earlier element within the fronted constituent to give a contrastive effect: (129) Su babbu de Juanne appo vistu. ‘It was John’s father that I saw.’ (e.g. ‘not his brother’) More typical examples of this phenomenon are cases where contrastive stress occurs on a determiner or quantifier (as in canonical sentences, see (122) above): (130) a Cudda mákkina keljo. ‘I want that car.’ b Duas collanas appo comporatu. ‘I bought two necklaces.’ The above facts are consistent with the hypothesis that fronting is triggered by the presence of a [+focus] feature. Given that WH—movement often applies to constituents which include the WH—item itself, because syntactic constraints prevent extraction of the WH—

< previous page

page_354

next page >

< previous page

page_355

next page >

Page 355 item on its own (the so-called ‘pied-piping’ phenomenon), we may postulate that the entire NP must be fronted in (128)–(130) regardless of whether the [+focus] feature is assigned to this NP or to some element within it, since fronting cannot apply to subconstituents within the NP (see 7.1.4). Nevertheless, this evidence also suggests that the possibilities for assignment of stress and focus within fronted constituents are not fundamentally different from those which obtain within clauses, a similarity which is perhaps obscured in an account which distinguishes between marked and unmarked focus. However, we shall not pursue this matter here. 7.2.2 Focus and illocutionary force In 7.1.5 we pointed out that fronting of a predicative element (for example, a verb dependent on an auxiliary, or an attributive expression in a copular construction) is particularly prevalent in yes/no questions, as in the (a) examples below, though it is not specifically a question-formation process, as shown by the (b) and (c) examples, which constitute appropriate answers to the questions in (a): (131) a Comporatu l’as? ‘Did you buy it?’ b Emmo, comporatu l’appo. ‘Yes, I did buy it.’ c No, furatu l’appo. ‘No, I stole it.’ (132) a Maláiduses? ‘Are you sick?’ b Emmo, maláidu so. ‘Yes, I am sick.’ c No, istraccu so. ‘No, I am tired.’ (133) a Duttore est? ‘Is he a doctor?’ b Emmo, duttore est. ‘Yes, he is a doctor.’ c No, professore est. ‘No, he is a teacher.’ The examples in (c) show the typical focusing effect of fronting in that the fronted item expresses new information which is presented as contrastive. However, in the (b) examples the fronted item expresses

< previous page

page_355

next page >

< previous page

page_356

next page >

Page 356 information which is already given in the preceding context, though it must bear primary stress as in other instances of fronting (unlike the preposed elements discussed in 7.1.6). In order to accommodate such examples, we suggest that narrow focus on the predicate of a sentence can give rise to two possible effects: either contrast or emphasis on the semantic content of the predicate itself (as with other fronted items) or emphasis on the truth value of the sentence as a whole (as expressed in English by stress on the auxiliary or copula or by insertion of do, as in the translation of (131b)). Note that the latter effect is not peculiar to cases where the predicate is fronted, but can also be seen in examples like (134b) where exclusive focus on the verb is achieved by right-dislocation of the object: (134) a Proitte no’as mandatu cudda líttera? ‘Why didn’t you send that letter?’ b Ma l’appo mandata, sa líttera. ‘But I did send the letter.’ The possibility of interpreting narrow focus on the predicate as focus on the truth value of the sentence may help to explain the tendency for fronting of the predicate to occur predominantly in yes/no questions. In declarative sentences, focus on the truth value is normally superfluous in so far as we take it for granted that the speaker intends the statement to be taken as true. Consequently, when such a focus does occur in a declarative sentence, it requires a context in which the truth value is at issue; for example, as an answer to a yes/no question as in the (b) examples in (131)–(133) or as a denial of a presupposition as in (134). However, in yes/no questions like the (a) examples in (131)–(133), the truth or falsity of the proposition is clearly at issue since this is precisely the information which the hearer is expected to provide. A further factor which may favour fronting of the predicate in yes/no questions concerns the intonational cues which distinguish questions from statements. Essentially, yes/no questions are characterised by an incomplete fall in pitch on the stressed element, with continuation of a mid (rather than low) tone throughout the posttonic sequence (and, optionally, a slight rise or fall on the final unstressed syllable(s) of the sentence); see Contini (1985). With regard to the contour of the pretonic sequence, yes/no questions do not appear to differ significantly from statements. Within this perspective, fronting offers two functional advantages which we shall illustrate by the examples in (135): (135) a As telefonatu a su duttore?

< previous page

page_356

next page >

< previous page

page_357

next page >

Page 357 b Telefonatu as a su duttore? ‘Did you telephone the doctor?’ First, fronting shifts the locus of stress to the beginning of the sentence (to telefonatu in (135b) as opposed to duttore in (135a)), so that the intonational cues which signal interrogative force are presented much earlier (we do not have to wait until the end of the sentence to discover that it is meant as a question). By the same token, fronting extends the sequence over which the mid tone is sustained (i.e. as a su duttore in (135b) as opposed to the final unstressed syllable of duttore in (135a)), thus, perhaps, rendering the incompleteness of the pitch-fall more clearly perceptible. We do not wish to exaggerate the value of the functional advantages suggested above—speakers are clearly capable of producing and recognising an interrogative contour on examples like (135a). Nevertheless, there is a general tendency in yes/no questions to adopt a (non-canonical) formulation which shifts main stress to an early position in the sentence, even when an interrogative interpretation is explicitly signalled by the particle a (see below), as in the rightdislocated example (136): (136) (A) l’as telefonatu, a su duttore? In both (135b) and (136) the intonational effects described above coincide with narrow focus on the predicate, which we have postulated as being appropriate to yes/no questions. However, fronting of a nonpredicative element is also common in yes/no questions, as in (137), even though it defocalises the predicate—indeed, (137) appears to be more conducive to an interrogative interpretation than the canonical sentence (135a): (137) A su duttore as telefonatu? Although the formulation in (137) attaches some importance to su duttore (for example, it could not be interpreted as an inquiry as to how the doctor was contacted), the focusing effect is not so strong as to require the remainder of the sentence to be presupposed (e.g. (137) is appropriate in a wider range of contexts than the English clefted question Who was it that you telephoned?). We suggest that fronting of the object in (137) is at least partly motivated by the intonational considerations discussed above. In 1.2.2 we noted that Sardinian has an explicit interrogative particle (or complementiser) a which can be used to introduce yes/no questions, with some syntactic restrictions (for example, it is incompatible with a preverbal subject or a fronted item). The use of

< previous page

page_357

next page >

< previous page

page_358

next page >

Page 358 interrogative a is particularly favoured, indeed virtually obligatory, when the core sentence consists solely of a verb in a simple tense form (ignoring clitics and dislocated phrases): (138) a A lu tuncas (su barcone)? ‘Will you shut it (the window)?’ b A venis? ‘Are you coming?’ A possible reason for this is that, as well as indicating interrogative force, a acts as an overt element which facilitates assignment of a narrow focus to the following predicate, which we have characterised as being conducive to an interrogative interpretation. This suggestion brings us to a question which we raised briefly in 7.2.1; namely, whether the whole of the core sentence can count as a final constituent and thus be interpreted as the focus. The evidence in (138), where a is virtually obligatory, versus (136), where a can be omitted more freely, suggests that the finality condition on focus should be defined relative to some overt element which is non-final. In (136) this condition is satisfied by the auxiliary as which permits assignment of focus to the predicate telefonatu whereas in (138) interrogative a is required to fulfil this requirement. A further observation which may be relevant to this question is that the presence of interrogative a often favours a request interpretation, as in (139a), as opposed to a simple factual question, as in (139b): (139) a A fakes su brodu? ‘Will you make the broth?’ b Fakes su brodu? ‘Are you making the broth?’ According to the hypothesis envisaged above, a in (139a) allows focus to be assigned to the whole VP which denotes the requested action, whereas in (139b) the only possible focus is the direct object, thus favouring an interpretation where the general nature of the action is ‘given’ and some particular aspect of it is being questioned. The function and distribution of interrogative a is mirrored in declarative sentences by the emphatic particle ja (in some dialects da). This particle is etymologically related to the aspectual adverb dza ‘already’ (both being derived from Latin IAM ‘now, already’), but has lost its temporal value, being roughly equivalent to English ‘indeed’. Except in one respect, which we shall discuss presently, its distribution appears to be identical to that of interrogative a; it occurs in sentence-initial position (though it can be preceded by left-

< previous page

page_358

next page >

< previous page

page_359

next page >

Page 359 dislocated items) and is incompatible with a preverbal subject or a fronted element (including WH—items). Like interrogative a, this item provides a means of satisfying the requirement that the focused constituent must be preceded by some overt element. For example, sentences like Lu tunco ‘I (will) shut it’ and Vendzo ‘I am coming’, though syntactically well formed, are not appropriate answers to the questions in (138), indeed it is difficult to imagine any context in which Vendzo would be a natural utterance in its own right. However, addition of ja makes such sentences natural in this context, in the same way as a in the corresponding questions, by allowing assignment of focus to the verb:6 (140) a Ja lu tunco. ‘I (will) shut it’ b Ja vendzo. ‘I am coming.’ Further evidence in support of the view that ja permits assignment of focus to an element which would not otherwise be final relative to some other item is provided by sentences involving the verb iskire ‘know’, which is almost always preceded by ja, with right-dislocation of the complement clause, when it occurs in the first person, simple present form: (141) a ?Isco k’est tardu. b Ja l’isco, k’est tardu. ‘I know that it is late.’ There is nothing syntactically wrong with (141a), but its relative unacceptability may be due to a pragmatically inappropriate assignment of focus which is imposed by the syntax. If focus is assigned to a constituent which is final with respect to overt elements which precede it, the only choice of focus is the complement clause or some part thereof. However, it is reasonable to suppose that this choice of focus conflicts with the fact that the complement of iskire is normally presupposed and cannot therefore be asserted as new information. In (141b), however, the complement clause is defocalised by rightdislocation and the presence of ja permits focus to be assigned exclusively to iskire, thus yielding a pragmatically appropriate interpretation in which the speaker emphasises his awareness of the situation. In examples such as (140)–141, we postulate that ja occupies the same structural position as interrogative a; e.g. as a complementiser. However, unlike interrogative a, ja can be repeated at the end of the

< previous page

page_359

next page >

< previous page

page_360

next page >

Page 360 sentence after the focused element to give even stronger emphasis. In such cases, final ja is unstressed with low pitch, giving an intonational pattern similar to that of right-dislocated constructions, as in (142): (142)

‘I am coming.’ We shall not speculate on the structural position of final ja in such cases or on the reasons why final ja cannot, apparently, occur without initial ja, as in *Vendzo ja . From a discourse perspective, the prosodic similarity between examples like (142) and right-dislocated constructions may be significant in so far as in both types of construction, focus on the final element within the ‘core’ part of the sentence seems to be intensified by the presence of an unstressed element following the focused item. The evidence presented so far suggests that the emphatic effect of ja is due largely to its presence in particular positions which affects the operation of general principles of focus assignment, rather than to its intrinsic semantic content. We might characterise ja as a purely expletive particle which, when it occurs sentence-initially, permits focus to be assigned to an element which would not otherwise be ‘final’ in the required sense and, when repeated after the focused element, acts as an unstressed coda which enhances the salience of the prosodic cues which identify the focused item. However, there are other aspects of the use of ja which are not readily explicable in terms of this approach. First, it should be noted that sentences consisting of a verb alone (or accompanied by a clitic pronoun) are not always infelicitous as utterances in their own right. For example, imperatives such as Veni! ‘Come!’ or Tuncalu! ‘Shut it!’ are perfectly acceptable. Moreover, declarative sentences such as Lu tunco ‘I shut it’ or Travallo ‘I work’ are acceptable as answers to questions of the type ‘How do you stop the window slamming?’ or ‘What do you do all day?’, where focus is on the semantic content of the verb itself rather than on the truth value of the proposition. Furthermore, the use of ja would be singularly inappropriate in such cases, and indeed totally impossible in imperative sentences. It appears then that our proposed requirement that a final focused constituent must be preceded by some overt element is too strong and should be modified to apply only to cases of focus on the truth value. By the same token, ja seems to be used only in sentences of a confirmatory type, to emphasise the truth of a

< previous page

page_360

next page >

< previous page

page_361

next page >

Page 361 proposition which is given by context rather than to introduce new information. Initial ja can also occur, with the same effects, even though the main predicate is preceded by another item (an auxiliary or copular verb) and thus satisfies the ‘finality’ requirement for focus on the truth value even when ja is absent: (143) a Ja l’appo tuncatu (ja). ‘I have shut it.’ b Ja so vennende (ja). ‘I am coming.’ c Ja so cuntentu (ja). ‘I am happy.’ Moreover, final ja can occur after a constituent which is non-focal, as in (144): (144) Ja l’isco k’est tardu ja. ‘I know that it is late.’ Finally, we may note that ja is not restricted to declarative sentences. It can be used (initially and finally) in sentences with an interrogative intonation, interpreted as questions which anticipate an affirmative reply in much the same way as tag-questions in English, as in (145) which can be translated roughly as ‘You did shut the window, didn’t you?’: (145) Ja l’as tuncatu (ja), su barcone? Thus, ja appears to have an affirmative rather than declarative function and cannot be analysed as a marker of illocutionary force on a par with interrogative a. To summarise, yes/no questions and sentences which are emphatically affirmative can be explicitly marked as such by the particles a and ja . In addition, placement of the predicate in a position where it attracts exclusive focus (for example, by fronting of the predicate or by right-dislocation of the complement) has an effect which is particularly appropriate to such utterances. When the proposition consists solely of a finite verb (possibly accompanied by a clitic), the particle appears to serve both functions, identifying the sentence as interrogative or affirmative and, at the same time, providing a syntactic context which allows narrow focus to be assigned to the verb.

< previous page

page_361

next page >

< previous page

page_362

next page >

Page 362 Conclusion In terms of its overall syntactic organisation, Sardinian displays many features which are characteristic of the modern Romance languages as a whole, though not necessarily common to all of these languages or indeed exclusive to this language family. Sardinian is essentially a configurational language with a canonical SVO order and a tendency towards verb-second order when a non-subject occurs in sentenceinitial position. At the phrasal level, it shows a dominant head-initial pattern, indeed perhaps rather more rigorously than other Romance languages; for example, within the NP, pronominal possessives are systematically postnominal and the range of adjectives which can precede the noun is severely restricted. As in other varieties of Romance, the principle deviation from the configurational pattern concerns clitic pronouns, whose grammatical function is expressed by Case morphology rather than structural or linear order (and, conversely, whose order with respect to each other and to the verb is determined by considerations which are largely independent of their grammatical function). We also find extended uses of clitic pronouns which are fairly typical of general Romance: for example, the grammaticalised use of reflexive clitics with a broadly ‘detransitivising’ function as in passive-like ‘Middle’ constructions and in the formation of pronominal verbs, the possessive use of dative clitics and, at a more discursive level, the use of clitics in dislocated constructions to denote the grammatical function of elements whose semantic content is specified by phrases which have a more peripheral structural status. As a morpho-syntactic correlate of configurationality, Sardinian also reflects the general Romance evolution away from the synthetic system of Classical Latin to a more analytic system. Indeed, in certain respects, this change has gone further in Sardinian than in many other varieties of Romance. For example, the rich inflectional tense system of Latin has given way to a binary distinction (present versus past

< previous page

page_362

next page >

< previous page

page_363

next page >

Page 363 imperfect) in most Sardinian dialects, other temporal or aspectual distinctions being realised by periphrastic formulae (see 3.1.1–3.1.3); for example, compound perfect or pluperfect forms expressing punctual or narrative past tense, periphrastic future and conditional constructions with áere ‘have’ and the short forms of dévere ‘must’ with the infinitive, and the widespread use of progressive constructions (éssere ‘be’ with the present participle) in preference to the simple present or imperfect. Similarly, with respect to the expression of grammatical functions, the role of inflectional Case is fulfilled by means of prepositions, a tendency which applies to the accusative under certain conditions (cf. the prepositional accusative discussed in 2.2.6) as well as to ‘oblique’ Cases such as the dative, genitive and ablative—a notable exception to this tendency being the survival of the genitive in expressions such as ocritortu ‘cross-eyed’ mentioned in 2.1.1 and 4.1.1. Further evidence of analyticity within the prepositional system, though of a rather different type, can be discerned in locative PPs where specification of the spatial relation is expressed independently of the distinction between static location, source and goal (see 4.3.3). The tendency towards periphrasis can also be seen in the absence of a productive process for forming adverbs out of adjectives (such as suffixation of — mente in Italian). Among the typological features which Sardinian shares with a subset of Romance languages, perhaps the most obvious is the optional presence of an overt subject in finite clauses and the possibility of inversion which is not necessarily triggered by other syntactic processes (i.e. the so-called ‘pro-drop’ or ‘null-subject’ parameter, which distinguishes languages like Italian or Spanish from French). On the other hand, Sardinian resembles Italian and French, but not modern Spanish, in displaying an alternation in the choice of perfective auxiliary (‘have’ versus ‘be’) which, in non-reflexive constructions, appears to correlate with the phenomenon of ‘unaccusativity’. Against this ‘pan-Romance’ background, we have noted a number of syntactic phenomena which are, by and large, peculiar to Sardinian, though they may have approximate or superficial analogues in certain other Romance languages or dialects; for example, the fronting-focusing process discussed in 7.1.4–7.1.5, the impersonal constructions discussed in 3.2.2 and 3.2.4, and the use of the infinitive (inflected or not) with an overt nominative subject or a null subject which can be interpreted as having independent reference (see 6.1.3, 6.1.5). Other respects in which Sardinian differs rather more subtly from other Romance languages include the prepositional accusative phenomenon, which occurs in languages such as Spanish and

< previous page

page_363

next page >

< previous page

page_364

next page >

Page 364 Rumanian but which appears to be determined by rather different factors in Sardinian (see 2.2.6). A further difference concerns the choice of perfective auxiliary in reflexive constructions where Sardinian diverges from languages like French and Italian in that ‘be’ is only selected when the reflexive clitic is accusative, at least in the clear cases, though the issue is complicated by some degree of inconsistency in the distinction between accusative and dative complements, an inconsistency which may in turn be partly attributable to the prepositional accusative phenomenon (see 3.2.1, 3.3.1). The interest of Sardinian syntax is not confined to the ‘curiosity value’ of the more distinctive properties mentioned above. Throughout the preceding chapters we have endeavoured to investigate the interaction between different aspects of the syntactic system rather than simply documenting them as independent phenomena. On the one hand, the extensive discussion of familiar Romance traits in the theoretical literature has provided us with a framework within which to examine the more distinctive aspects of Sardinian syntax. For example we have appealed to the general correlation between unaccusativity and auxiliary choice in our discussion of matters as diverse as the mapping between θ-roles and grammatical functions (3.2.1), the properties of impersonal and existential constructions (3.2.2, 3.2.4) and certain restrictions on extraposition of the head noun (2.2.8). Similarly, in 6.1.5 we suggested that the properties of the ‘peculiar’ infinitival constructions referred to above can be accounted for in terms of certain approaches to the so-called ‘prodrop’ phenomenon in finite clauses. By the same token, however, the syntactic particularities of Sardinian may have wider implications for the analysis of general phenomena such as pro-drop and unaccusativity in other languages. For example, in 3.2.2 we attributed the selection of ‘have’ in perfective impersonal constructions with unaccusative verbs to the lack of agreement with the postverbal NP, but in superficially similar constructions in French, unaccusative verbs select ‘be’ despite the lack of verb-agreement. We offer no proposals as to how this apparent contradiction might be resolved, but merely suggest that the Sardinian evidence may be relevant to an explanatory account of the French constructions. There are other cases where distinctive aspects of Sardinian provide a direct empirical insight into more general issues which is not available in other related languages. For example, the fronting process discussed in 7.1.4–7.1.5 is a useful diagnostic for certain aspects of constituent structure which may not be immediately obvious in other languages. In particular, this process lends strong support for the view that auxiliary verbs

< previous page

page_364

next page >

< previous page

page_365

next page >

Page 365 (including modals) are external to a phrasal constituent which comprises the dependent verb and its complements, even though such constructions in Sardinian display many other properties (such as cliticclimbing) which some linguists have taken as evidence that AUX+V should be analysed as a constituent in other Romance languages (see 3.4.1). Similarly, in 5.1.2–5.1.3 the distribution of accusative a led us to postulate a distinction between ‘genuine’ pronouns (occupying the head N position) and determiners or quantifiers accompanied by a null nominal element, a distinction which correlates fairly systematically with referential properties; typically, ‘genuine’ pronouns are human in reference, but ‘bare’ determiners refer to non-human entities, whereas NPs consisting of a determiner with an overt noun can have either property depending on the semantics of the head noun. Although this analysis is motivated primarily by the language-specific effects of the prepositional accusative phenomenon, it is possible that some version of this hypothesis may be applicable to other languages which do not display this phenomenon. For example, at a morphosyntactic level, French appears to display a similar pattern in that disjunctive pronouns such as lui ‘him’ are typically human whereas demonstrative items such as cela ‘that’, ceci ‘this’ and free relatives of the type ce que…, which appear to consist of a determiner ce and a modifying expression (e.g. là ‘there’, (i)ci ‘here’), can only refer to non-human entities, but pronominal expressions such celui-là ‘that one’ which contain the pronoun (head noun) lui are neutral with respect to the [±human] distinction. Apart from theoretical issues of the sort outlined above, which merit further investigation within a more rigorous framework than that which we have adopted here, there is much empirical work still to be done on the syntax of Sardinian. As we pointed out in the Introduction, our account is based mainly on the judgements of a small number of native informants from a fairly limited geographical area. While we have taken pains to ensure that the facts presented are reasonably representative of this dialect area, we have not addressed the question of dialectal or sociolectal variation apart from brief comments on certain obvious differences which are documented in the descriptive literature. In 0.3 we offered the rather impressionistic view that the syntax of Sardinian is relatively homogeneous, at least in comparison with phonology, morphology and lexis. Whether this view is correct is a matter which can only be settled by further research. Possibly, there are minor empirical differences too subtle to be revealed by superficial analysis which, on closer investigation, may be symptomatic of more significant variation, as in the case of some

< previous page

page_365

next page >

< previous page

page_366

next page >

Page 366 of the above-mentioned differences between Sardinian and other Romance languages. By presenting a detailed analysis of one variety of Sardinian which makes specific, testable claims regarding the nature of individual syntactic phenomena and the ways in which they interact, we hope to have provided a framework within which such differences can be identified and their potential significance assessed.

< previous page

page_366

next page >

< previous page

page_367

next page >

Page 367 Notes NOTES TO CHAPTER 1 1 See for example Chomsky (1981, 1982); Jaeggli (1982); Rizzi (1982); Bouchard (1984). 2 Where this distinction is not crucial to the matter in hand, we shall use the neutral term ‘postposition of the subject’. 3 See also the comparative construction with ki+non discussed in 4.2.1 where the complement containing non frequently lacks a verb. 4 See 3.2.7 for discussion of the construction in (37a). 5 In this connection, it is important to distinguish interrogative a from the homophonous preposition, which of course can introduce fronted expressions in questions, as in: A Maria as telefonatu? [to Mary you+have telephoned] ‘Have you telephoned Mary?’ 6 Some verbs also have a truncated imperative form derived by deleting the post-tonic sequence of the second person singular imperative; e.g. abbá (= abbarra ‘stay’), mi (= mira ‘look’), te (= tene ‘hold’). These items tend to be used as interjections rather than as genuine imperative verb forms. 7 The item ancu has no obvious equivalent in English (the closest approximation might be the locution ‘would that…’). It is used only to introduce optative/hortative sentences such as (53). It does not appear to have the complementiser status which we attributed to interrogative a in 1.2.2 since it can co-occur with the complementiser ki: Ancu ki Deus m’assestat! (= (52a)). 8 Note that when a single clitic follows the verb, as in (57b), the verb retains its normal stress pattern, but when more than one clitic is added, as in (57a), main stress occurs on the penultimate syllable of the whole sequence. 9 In Blasco-Ferrer (1986:115) we find further examples of encliticisation in Campidanese— “Neri-sí-ddu fustei!” ‘Tell it to him’, “Aporri-mí-dda sa manu!” ‘Give it to me, your hand!’—but corresponding Logudorese examples are given as: “Bi lu neret vosté!” (procliticisation) and “Isterret-mí-la sa manu” (encliticisation).

< previous page

page_367

next page >

< previous page

page_368

next page >

Page 368 10 For discussion of constructions of the type in (62) where the understood subject of an infinitive is not coreferential with an element in the main clause, see 6.1.3. NOTES TO CHAPTER 3 1 The Italian original is as follows, with the simple past throughout: “Ma un’ora dopo lo avvertirono che andasse a casa tuo; ed egli corse, pallido e col cuore in tumulto.” (Deledda 1941:191) 2 Examples like (14a) can also have an interpretation of type (9), where the event-time extends right up to the present: ‘I have worked in Olbia for three years’. Typically the imperfect implies that the situation described does not extend to the present, but this is a pragmatic implicature which can be overridden: e.g. Travallaío in Olbia e, in fattis, bi travallo semper ‘I was working/used to work in Olbia and, indeed, I still work there’. 3 As an alternative to bi the other locative clitic nke (see 5.2.1, 5.2.4) may be used. As far as we are aware these two clitics are interchangeable in all the constructions discussed in this section, though for ease of exposition we shall use bi exclusively in our examples. 4 The intended reading for (70c) ‘Many soldiers died’ is perhaps rather difficult to obtain out of context because of the transitive use of the past participle mortu with the sense ‘kill’ (see above, p. 100) which gives the alternative interpretation ‘He/she/it killed many soldiers (there)’. A clearer example would be B’at isparitu metas sordatos ‘Many soldiers disappeared’. NOTES TO CHAPTER 4 1 Analytic constructions with bonu and malu are sometimes encountered, particularly in the superlative use without a head noun; for example, Custu est su prus bonu ‘This is the best (one)’. 2 The following adverbs in -mente are acknowledged as being genuine Sardinian forms: comente ‘how’ and mascamente, massimamente ‘especially’. 3 Within this class we also include items which express other circumstantial notions such as ‘intention’ or ‘attitude’ rather than ‘manner’ in the strict sense. 4 Tando can also have a temporal value ‘then, at that time’; cf. 4.3.4. 5 Pacu is often used sarcastically with a sense similar to that of meta, in much the same way as ‘not half in colloquial English: Est pacu abbistu ‘Heisn’t half clever’. 6 Note that items like fintzas can readily modify a direct object NP, whereas in English the same effect is more typically achieved by placing even before the VP: John even ate the cheese. However, as in English, such items cannot be placed before an NP governed by a preposition (including accusative a): *Est cuntentu de fintzas su travallu ‘He is happy even about the work’, *Appo vistu a fintzas Juanne ‘I saw even John’. In such cases, the focal adverb is placed before the PP: Est cuntentu fintzas de su travallu, Appo vistu fintzas a Juanne.

< previous page

page_368

next page >

< previous page

page_369

next page >

Page 369 7 For some speakers, the item ke (more typically used in comparatives of equality with the reading ‘as’, see 4.2.3) is preferred to de in examples such as (15)–(16). When ke is used, the preposition must be repeated: Juanne pessat prus a sa mamma ke a su babbu (= 15). 8 This example also has a reading where de Frantziscu is a possessive and the standard of comparison is unspecified: ‘John plays Francis’s guitar more’. 9 Again, the use of comente suggests similarity or parallelism rather than comparison of degree; e.g. ‘He is stupid, just as his father was clever’. 10 In Campidanese, the uses of dae outlined in the text are expressed by de. 11 The facts here are difficult to establish in so far as in the spoken language the final— a of contra is always merged with the putative preposition a, yielding contr’a su ledze, etc. However, in the written language the preposition is transcribed fairly systematically as a separate word. 12 We adopt the working assumption that the items in (47) are syntactically simple (though forms beginnining with a—or in—may be morphologically complex) whereas the expressions in (48) are syntactically complex. This assumption is borne out to some extent by orthographic usage in that the items in (47) tend to be written as single words whereas the expressions in (48) are typically spaced as indicated. Nevertheless, orthographic conventions do not provide a reliable guide since we do find items in (47) spelt as separate words (e.g. in dainnantis, in tundu, etc.). Moreover, within the class of items listed in (48), one which is commonly spelt as a single word is in s’oru (issoru), even though it is perhaps the most syntactically transparent (cf. the presence of the definite article), whereas a caddu is always written as two words even though it does not readily admit a compositional analysis of the type suggested in (49b) below. 13 Note that forms which incorporate in—cannot be preceded by in to indicate static location: *in inube, *in inibe, *in inoke, etc. 14 This construction is not normally used to describe ‘professional visits’ of the type ‘go to the doctor’s/to the barber’s’, which are rendered simply by the use of a: andare a su duttorela su barbieri. NOTES TO CHAPTER 5 1 In some dialects issu and isse coexist, with the former used predominantly as a nominative form and the latter after prepositions (sometimes with se as a variant). 2 The judgements concerning examples with neune in (17b) are less clear than for those with kie and nemos . Many dialects have a further negative item nessunu (Camp. nisciunu) which, like Italian nessuno, can be used pronominally with human reference or as a determiner with either human or inanimate nouns: nessunu duttore ‘no doctor’, nessuna risposta ‘no reply’. It is possible that for some speakers neune can be used in the same way as nessunu. 3 The semantic property which distinguishes cale from kie and ‘bare’ itte is that cale elicits a referent from within a specified set such as might be delimited by the previous use of a common noun, whereas kie and itte range over an open set. 4 At first sight, cases of ‘floating’ of tottu from subject position appear to

< previous page

page_369

next page >

< previous page

page_370

next page >

Page 370 violate this restriction, since tottu can clearly have human reference in the following example: Sun tottu arrivatos ‘They all arrived’. Such examples can perhaps be accommodated by appealing to the idea that in nullsubject languages the person/number inflection on the verb has a similar function to object clitics. Thus, tottu in the above example can be construed as quantifying an NP which is identified by the verb inflection, just as in (23) the quantified NP is represented by a clitic. 5 Blasco-Ferrer (1986:112) suggests that the use of nde as a Source clitic is predominantly a Logudorese feature; a rather curious state of affairs in so far as we might expect that the pressure to exploit an alternative Source expression would be greater in the Campidanese dialects, where the absence of bi precludes any other means of contrasting Source with Goal. Our own research confirms the widespread use of nde as a Source clitic in Logudorese-Nuorese, but we leave open the question of whether it is restricted to these dialects. 6 Note, however, that the preposition de does occur in cases where an indefinite N’ is dislocated: Juanne nde connosket pacos, d’istudientes ‘John knows few, of students’, Gavini nd’at bitu meta, de vinu ‘Gavin drank a lot, of wine’. 7 Once again, the apparent reluctance towards use of nde (or ndi) as a locative Source clitic in Campidanese (cf. n. 5) is rather puzzling in so far as de is used as a Source preposition in this dialect, subsuming Logudorese-Nuorese dae (see 4.3.1). 8 The following attested example, where the second person clitic precedes si, suggests that this generalisation is not absolute: “Est una cosa chi pote’fachere un’átteru chi ti s’assemizzet…” [Sinnos: 125] ‘It is a thing which someone else who resembles you may have done…’. This example also contradicts our proposal that clitic sequences from column I are restricted to cases where the use of a dative clitic is unavoidable since ti is a genuine indirect object which could be expressed by a disjunctive form:… ki s’ammidzet a tie… It may be relevant that si in this example is part of the pronominal verb s’ammidzare rather than a referential item, though our attempts to test for this difference with native informants have not yielded any conclusive results. For the present, we simply mention this example as evidence that there may be some indeterminacy in the order of clitics from column I and the conditions under which such combinations are possible. 9 Where an action performed by the subject on a body-part involves the intervention of some other bodypart, a dative reflexive clitic must be used; e.g. *Juanne at lavatu sas dentes vs. Juanne s’at lavatu sas dentes ‘John brushed his teeth’. 10 See 3.2.3 for arguments that change of state verbs such as créskere are unaccusative when they cooccur with a dative element. 11 Failure to move the subject into the preverbal position in such cases also leads to ungrammaticality, even though the sisterhood condition is satisfied: *Sun créskitos sos pilos a cussa fémina. This is due to an independent constraint which prevents a postverbal subject from occurring with a postverbal dative even in cases which do not involve possession (see 7.1.3): *Piaghet su casu a sos sórrikes vs. Su casu piaghet a sos sórrikes or Lis piaghet su casu ‘Cheese pleases mice/them’.

< previous page

page_370

next page >

< previous page

page_371

next page >

Page 371 12 The use of pro ‘for’ instead of a in (97a, b) yields a better result—generally pro denotes a looser type of benefactive relation than a. Examples (97a, b) are also marginally acceptable with mi/ti as possessors of the teeth, though this requires a fairly loaded context in which the request or offer is made on behalf of the persons denoted by lis . In such a context the functions of mi/ti and lis in (95) could also be reversed. 13 Example (122b) is taken from B. Bandinu’s preface to Sos sinnos and Mialinu refers to the author (Michelangelo Pira). The quotation is in Italian. NOTES TO CHAPTER 6 1 Ja and ki in examples like (4c) are often spelt as a single word jaki, which might perhaps be analysed as a complementiser. The item ja is used on its own as an emphatic particle, roughly equivalent to ‘indeed’ (see 7.2.2), and is cognate with the adverb which we have transcribed as dza ‘already’. 2 These forms occur only in the Logudorese-Nuorese dialects. Camp-idanese has a different imperfect subjunctive paradigm (derived from the Latin pluperfect subjunctive) which cannot occur in the infinitival contexts illustrated in (105). Consequently, our observations concerning the use of ‘imperfect subjunctives’ as inflected infinitives do not apply to Campidanese. We shall return to the situation in Campidanese at the end of this section. 3 Sende here is a reduced form of the present participle of éssere ‘be’, essende, which can also be used as in (217). NOTES TO CHAPTER 7 1 The rudimentary intonation contours in (2) and elsewhere in this chapter represent only those features which are relevant to the syntax and interpretation of construction types, ignoring, for example, pitch variations within the pretonic or post-tonic sequence. A detailed examination of Sardinian intonation is given in Contini (1985). 2 We leave open the question of whether the subject is adjoined to the VP which includes the auxiliary or to the lower VP headed by the main verb (see 3.4.1 for a detailed discussion of the structural position of auxiliaries). 3 Other syntactic tests which have been used to justify this distinction do not yield conclusive results for Sardinian. For example, intuitions concerning the use of a partitive clitic to represent a postverbal subject are less clear than those reported for Italian by Belletti and Rizzi (1981) and in any case the relevant examples are impersonal constructions rather than instances of straightforward inversion (see 3.2.3, examples (80)– (81)). Similarly, the differential behaviour of unaccusative subjects with respect to the phenomenon of extraposition of the head noun discussed in 2.2.8 appears to reflect a structural distinction at the underlying rather than superficial level. 4 The item a in (71b) and (72b) is to be interpreted as the interrogative

< previous page

page_371

next page >

< previous page

page_372

next page >

Page 372 particle. Example (71b) is marginally acceptable if a is taken as the prepositional accusative marker (see 2.2.6). 5 This generalisation is not absolute. In contexts where the verb or its temporal or aspectual features are explicitly contrasted, examples of the type in (117) are possible: Peri sa janna ses intratu e peri sa janna deves issire Through the door you came in and through the door you must go out’, Maccu fis, maccu ses e maccu abbarras ‘Mad you were, mad you are and mad you will remain’. 6 For the purposes of this account, we treat clitics as part of the verb; e.g. lu in Lu tunco does not appear to count as a pre-focal element.

< previous page

page_372

next page >

< previous page

page_373

next page >

Page 373 Bibliography DESCRIPTIVE AND THEORETICAL STUDIES Atzori, M.T. (1982) Sardegna, Modena: Pacini. Baker, M. (1988) Incorporation: A Theory of Grammatical Function Changing, Chicago: University of Chicago Press. Belletti, A. and Rizzi, L. (1981) ‘The syntax of “ne”; some theoretical implications’, The Linguistic Review 1:117–54. Blasco-Ferrer, E. (1984) Storia linguistica della Sardegna, Tübingen: Max Niemeyer. ——(1986) La lingua sarda contemporanea, Cagliari: Della Torre. Bossong, G. (1982) ‘Der präpositionale Akkusativ im Sardischen’ in O. Winkelmann and M.Braisch (eds) Festschrift für Johannes Hubschmidt zum 65 Geburtstag, Berne: Francke. Bouchard, D. (1984) On the Content of Empty Categories, Dordrecht: Foris. Burzio, L. (1986), Italian Syntax, Dordrecht: Reidel. Chomsky, N. (1973) ‘Conditions on transformations’ in S.R.Anderson and P.Kiparsky (eds) A Festschrift for Morris Halle, New York: Holt, Rinehart & Winston. ——(1976) ‘Conditions on rules of grammar’, Linguistic Analysis 2:303–51. ——(1977) ‘On WH—movement’ in P.W.Culicover, T.Wasow and A. Akmajian (eds) Formal Syntax, New York: Academic Press. ——(1981) Lectures on Government and Binding, Dordrecht: Foris. ——(1982) Some Concepts and Consequences of the Theory of Government and Binding, Cambridge, Mass.: MIT Press. ——(1986) Barriers, Cambridge, Mass.: MIT Press. Contini, M. (1985) Etude de géographie phonétique et de phonétique instrumentale du sarde, Alessandria: Edizioni dell’Orso. Emonds, J. (1978) ‘The verbal complex V′—V in French’, Linguistic Inquiry 9: 151–75. Farina, L. (1973) Vocabulario nuorese-italiano, Sassari: Gallizzi. Farina, L. and Mingione, P.D. (n.d.) Ortografia sarda unificata, Nuoro: AR.P.E.F. Gruber, J.S. (1965) ‘Studies in Lexical Relations’, unpublished Ph.D. dissertation, Cambridge, Mass.: MIT. Jackendoff, R.S. (1972) Semantic Interpretation in Generative Grammar,

< previous page

page_373

next page >

< previous page

page_374

next page >

Page 374 Cambridge, Mass.: MIT Press. Jaeggli, O. (1982) Topics in Romance Syntax, Dordrecht: Foris. Jones, M.A. (1988a), ‘Sardinian’, in M.Harris and N.Vincent (eds) The Romance Languages, London: Croom Helm. ——(1988b) ‘Auxiliary verbs in Sardinian’, Transactions of the Philological Society 86, 2:173–203. ——(1992), ‘Infinitives with specified subjects in Sardinian’, in C.Lauefer and T.Morgan (eds) Theoretical Analyses in Romance Linguistics, Amsterdam: John Benjamins. ——(forthcoming), ‘The prepositional accusative in Sardinian: its distribution and syntactic repercussions’, in M.Maiden and J.C.Smith (eds) The Romance Languages and Current Linguistic Theory, Amsterdam: John Benjamins. Kayne, R.S. (1975) French Syntax, Cambridge, Mass.: MIT Press. ——(1991) ‘Romance clitics, verb movement and PRO’, Linguistic Inquiry 22, 4:647–86. Lepori, A. (1979) Prontuario di grammatica sarda. Variante campidanese, Cagliari: CUEC. Loi Corvetto, I. (1983) L’italiano regionale di Sardegna, Bologna: Zanichelli. Perlmutter, D. (1978) ‘Impersonal passives and the unaccusative hypothesis’, Berkeley Linguistics Society 4:57–189. Pittau, M. (1972) Grammatica del sardo-nuorese, Bologna: Patron. ——(1978) Pronunzia e scrittura del sardo logudorese, Sassari: Dessì. Porru, V.R. (1811) Saggio di grammatica sul sardo meridionale , Cagliari; repr. Sassari: Dessì, 1975. Radford, A. (forthcoming) ‘Head-hunting: on the trail of the nominal Janus’, in G.Corbett (ed.) Heads in Grammatical Theory, Cambridge: Cambridge University Press. Reichenbach, H. (1947) Elements of Symbolic Logic, London: Macmillan. Rizzi, L. (1982) Issues in Italian Syntax, Dordrecht: Foris. ——(1986) ‘Null objects in Italian and the theory of PRO’, Linguistic Inquiry 17:501–57. Rohlfs, G. (1971) ‘Autour de 1’accusatif prépositionnel dans les langues romanes’, Revue de linguistique romane 35:312–34. Rouveret, A. and Vergnaud, J.-R. (1980) ‘Specifying reference to the subject’, Linguistic Inquiry 11, 1:97– 202. Ruwet, N. (1972) Théorie syntaxique et syntaxe du français, Paris: Seuil. Spano, G. (1840) Orthografia sarda nazionale: ossia Grammatica della lingua logudorese paragonata all’ italiano, Cagliari; repr. Cagliari: Trois, 1974. ——(1852) Vocabulario sardo-italiano e italiano-sardo, 2 vols, Cagliari; repr. Bologna: Arnaldo Forni (1981). Virdis, M. (1978) Fonetica del dialetto sardo campidanese, Cagliari: Della Torre. Wagner, M.L. (1938–9) ‘Flessione nominale e verbale del sardo antico e moderno’, Italia Dialettale 14:93– 170 and 15:207–7. ——(1941) Historische Lautlehre des Sardischen, Halle: Max Niemeyer. ——(1951) La lingua sarda. Storia, spirito e forma, Berne: Francke. ——(1960–4) Dizionario etimologico sardo, Heidelberg: C.Winter. ——(1984) Fonetica storica del sardo (translation of Wagner 1941 with introduction and appendix, by G.Paulis), Cagliari: Trois.

< previous page

page_374

next page >

< previous page

page_375

next page >

Page 375 Wasow, T. (1977) Transformations and the lexicon’ in P.W. Culicover, T.Wasow and A.Akmajian (eds) Formal Syntax, New York: Academic Press. SARDINIAN TEXTS (Abbreviations in square brackets indicate the reference by which examples are cited.) Comitate San Paolo, (1986) Boghes sardas de e sa zittade, Sassari: Editrice Stamperia Artistica. (Selection of short stories and poems submitted for the San Paolo literary prize, 1978–82.) [BS] Corraine, D. (ed.) Limbas, Nuoro: private publication. (Journal devoted to Sardinian linguistics and related questions.) Deledda, G. (1941)—see Spiggia (1982). Enna, F. (ed.) (1983) Contos de foghile, Sassari: Gallizzi. (Anthology of Sardinian folk-tales with Italian translations.) [CF] Pira, M. (1983) Sos sinnos, Cagliari: Della Torre. (Novel.) [SINNOS] Puddu, M. (1986) Alivertu: Sa colonisatzione de unu pastore, Sant-uannisuerxu: Editziones Golosti. (Autobiographical novel.) [ALIVERTU] Spiggia, S. (1978) S’abe, Olbia (private publication by the author). (Short treatise on apiculture.) [ABE] ——(trans.) (1982) Elias Portolu, Nuoro: Istituto Superiore Regionale Etnografico.(Translation in Nuorese dialect of G.Deledda (1941) Elias Portolu, Milan: Arnoldo Mondadori.) [EP]

< previous page

page_375

next page >

< previous page

page_376

next page >

Page 376 Index

a (interrogative particle) 24–6, 334, 339, 357–9, 361 a (pre-infinitival particle) 29, 143, 150–1, 260–6, 299 a (preposition) 181, 182, 185, 187, 189–94; accusative 65–8, 132, 179, 186, 199, 203–13, 271, 297, 301, 303, 314–16, 324, 341, 365; dative 65, 132–4, 219, 222–9, 231, 232, 270–6, 315; dative vs accusative 96–8 [±actual], feature determining mood 253–60 accusative 96–8, 132–6, 185, 199, 286 see also prepositional accusative accusative+infinitive constructions 269 accusative-dative Case 185, 199 adjectives 165–80; affective 42–3, 45–7, 69, 168; comparison of 173–5; as complements of prepositions 194; modification of 53–5, 74–6; morphology of 165–6; as nouns 75–6; order of 45–7, 52–6; position of 176, 179; specifier-like 44, 45–8, 62, 69, 77, 168; within NP 41–7, 50–6, 176, 179, 362 adjuncts vs complements 321–2 adverbial clauses 248–50, 257–61, 265–6, 279, 281–2, 285, 288–90, 304–11, 320–2, 323–4; introduced by WH- items 304 adverbs 165–80, 195–8; aspectual 89, 137–41, 147, 154–7, 159–63, 170, 197–8, 358; as complements of prepositions 194; of degree 159, 170–1; focal 137, 147, 154–7, 160–1, 171–3; of manner 147, 157, 161–3, 137, 167, 168, 171; modal 169; morphology of 165–7; of place 103, 114, 169–70, 186, 192, 195–6; position of 137–41, 142, 146–7, 148–9, 152, 154–7, 158–64, 168–73, 198; of time 137, 147, 157, 163, 169–70, 196 áere ‘have’ 278–9; with de and infinitive 93; existential use 113–14; as future auxiliary 89–93, 146–9, 154, 156, 280; as perfective auxiliary, 82–3, 130–6, 140–1, 276 [±affective], feature determining mood 253–60 affixation of tense features 148, 156, 280–1 Agent 94–7, 107, 119–22, 130, 182, 233, 235, 274, 277, 288 agreement: of adjectives 42–3; of determiners 49–50; of numerals 41; of past-participle 83, 101, 276, 288, 289; of possessives 187–8; of quantifiers 35, 37–9; of verb 20, 49, 61, 101–6, 113–14, 117–19, 127–8, 144, 146, 150, 152–3, 193, 201 see also gender; number Algherese 5 ambos ‘both’ 38–9, 137 analycity 362–3

< previous page

page_376

next page >

< previous page

page_377

next page >

Page 377 anaphors 239–46 appositive modification 70–1, 73 appositive relative clauses 296–7 arbitrary reference 267–8 aspect 82–9, 235, 287, 299, 305–8, 363; with ethic datives 231; in Middle constructions 129 aspectual verbs 149–53, 269–70 Atzori 1 auxiliary verbs 18, 82–93, 130–57, 277, 340, 344–5, 361, 364–5; order of 141, 145, 147–8, 154; perfective, choice of 20, 82–3, 99–101, 103–4, 106–13, 114, 121, 130–6, 143, 148, 150, 152, 241, 276, 363– 4 see also áere; éssere; modal verbs Baker 341, 342 bare infinitive 142–3, 260, 268, 271 Belletti 108, 330 bi (adverbial clitic) 110, 113–14, 214–15, 264; existential use 20, 61, 101–6, 124, 128, 130, 331; locative 97, 103, 349; pleonastic uses 237–8, 265, 325; as substitute for dative clitic 213, 214, 220–1, 223, 325 binding 116, 132, 239–46 Blasco-Ferrer 1, 28, 35, 38, 42, 61, 85, 188, 191, 213, 220, 247, 249, 250 Bossong 65 boundedness 317–18, 322–3, 326, 335, 336–7, 339–40 Burzio 20, 96, 106, 121, 330 c-command 239, 242, 244, 246 ca (complementiser) 22, 247–50, 310–11 cale ‘which’ 36, 204–8, 302–3 Campidanese 6, 9–11, 28, 32, 33, 34, 38, 42, 82, 91, 124, 148, 188, 191, 193, 197, 199–200, 213, 215, 220, 238, 240, 247, 249, 282, 305 cantu ‘how much, how many’ 34–5, 179–80, 185, 186, 302–3 carki ‘some’ 34–5 Case 5, 34, 65, 113, 114, 132, 193, 344, 362, 363 Case assignment 102, 106, 128, 185, 186, 280–2, 293, 295, 330–1, 341–2; and clauses 136, 248 Case forms 185–6, 199–200, 213 Catalan 5, 6, 7, 32, 166 causative constructions 143, 181, 182, 260, 270–8, 280 Chomsky 3–4, 14, 94, 115, 116, 239, 280, 314, 317, 341 circumstantial clauses 285–6, 290 clitic-climbing 138, 141, 145, 146, 149–50, 152, 268, 365 clitic-doubling 202, 225, 228, 325 cliticisation, conditions on 116 clitics 200–2, 213–39, 294, 295, 298, 323–6, 362; adverbial 214–21; combination of 218–21; dative 101, 104–5, 325; forms 65, 96–8, 213; locative 103–4; pleonastic uses 215, 230–9, 265, 325–6; position of 8, 15–16, 28–9, 137, 142–3, 146, 149–50, 152, 154, 213, 218, 271–6; pronouns vs agreement particles 220–1, suppletion of 213, 214, 220–1, 223; with tottu 208–9 collective nouns 33–4 comente ‘how’ 179–80, 185, 186, 250, 258–9 comitative Case 185, 199 COMP 14–15, 25–6, 262, 292, 302–3, 313–14, 336–8, 353 comparative constructions 173–8, 179–80, 250 comparative forms 44, 74, 165, 166 complement clauses 133–6, 144–5, 180, 248–57, 261–85, 286–8, 290–3, 321–2, 338 complement position 3, 5, 50–2 complementisers 14, 22, 247–50, 278–9, 282, 294–9, 310–11, 336–8, 359; in main clauses 25, 27, 29–30; infinitival 29, 156, 262–5 complements vs modifiers 70, 321–2

complex prepositions 57, 187–8 compound adjectives 167–8 compound nouns 52, 183 concessive constructions 250, 259–60, 309 conditional 91–3, 148

< previous page

page_377

next page >

< previous page

page_378

next page >

Page 378 conditional constructions 83, 249, 259–60, 285, 305–9 conjunctions 52, 54–6 consonants 9–12 constituent order 13–21, 157, 312–61, 362 Contini 1, 356 contracted forms 12, 43, 57, 147, 149, 193, 219–20 contrastive stress 351 control 151, 252, 266–9, 280, 285–6, 288–90, 292 copular constructions 201, 287–8, 346, 350 copular verbs 73, 114–19, 139, 277, 340, 361 coreference 239–6, 266–9, 268, 280, 288–9, 294 counterfactual constructions 255–6, 306–9

dae ‘from’ 182, 183, 189–92, 194, 217; in causative constructions 270–6; in passives 124–5, 129 dative 95–8, 122, 131–6, 181, 214, 255; as complement of preposition 229–30; ethic 123, 131, 202, 214, 219, 230–6, 325–6; of interest 101, 104–5, 111, 131, 188–9, 214, 219, 221–32; of possession, 222–6, 362 de (pre-infinitival particle) 143, 150–1, 260–5, 291, 298, 305 de (preposition) 73, 95–6, 182–3, 186, 215–17; in comparative constructions 173–8; in ordinal expressions 41; partitive 35–6, 48–50, 58, 61, 76–9, 314–16, 324, 333; possessive 44 definitive articles 34; without head N 70–5; in superlatives 178–9 degree adverbs 53–5, 74–6, 170–3 see also comparative constructions; comparative forms deixis 34, 66–7, 169–70, 195, 196–7, 234–5 deletion of preposition 294–6, 315 demonstrative adverbs 192, 195–6 demonstratives 34, 67–8, 70–1, 72–3, 77, 200, 203–8, 242–5, 301 determiners 34–7, 45, 47–9, 77, 178–9, 282–5, 300–5; absence of 8, 35, 44, 56–64, 65–8, 180, 181–4, 191, 204–13, 333; vs pronouns 68, 205–8, 242–3, 301, 365 dévere ‘must’ 90–3, 142, 145–6, 363; short forms of 90–1, 307 dialectical variation 5, 8–11, 82, 91, 124, 132, 148, 191, 193, 199–200, 205, 213, 215, 219–20, 238, 240, 247, 260, 282, 305, 365–6 diminutives 32, 165–6 disjunctive pronouns 64, 66, 199–203, 218–19, 241–2, 294, 296, 316, 324–6, 325 dislocation 61, 201–3, 225, 228, 231, 312–27, 333, 346–7, 362 see also left-dislocation; right-dislocation double perfect 83, 141, 308 elision 28, 82 Emonds 154 emotive predicates 252–3, 256–7 empty category 3, 15, 48, 64, 68, 69–70, 79, 205, 208–9, 266, 273, 300 English 18, 43, 64, 69, 104, 126, 167, 174, 189, 209, 212–13, 233, 236, 239–40, 241, 246, 260, 265, 267, 269, 282, 286, 287, 288, 292, 314, 317, 332, 356, 357 epenthesis 82 epistemic modality 145–6, 269 epithets 43, 243 ergative see unaccusative verbs éssere ‘be’ 278, 337; as copular verb 114–19; existential use 113–14; with infinitive 93; in passives 124–5; as perfective auxiliary 82–3, 130–6, 140–1, 276; as progressive auxiliary 83–4, 137–41 ethic dative 123, 131, 202, 214, 219, 230–6, 325–6 event-time 85–9, 197 exclamatives 27 existential constructions 20, 61, 101–6, 108, 113–14, 210, 329, 331, 363, 364

extraposition 321, 327, 329, 342, 344;

< previous page

page_378

next page >

< previous page

page_379

next page >

Page 379 of head noun 76–9, 108–9, 333, 364 factivity 153, 251, 257, 309 see also presupposition fákere ‘do, make’ 143, 270–8, 280 Farina 8, 38 focus 18–19, 63, 109, 117–19, 157, 312, 318–20, 322, 324–6, 330, 332, 336–7, 338–9, 345–9, 351–61, 363 focus feature 353–5 free relatives 71–2, 300–5 French 14, 86, 104, 124, 154, 189, 270, 363, 364, 365 fronting 16, 17–19, 23, 24–7, 59–61, 63, 66, 78, 117–19, 144, 147, 149–50, 153, 154–6, 159, 326–7, 332– 45, 348–57, 363, 364–5; of NPs 332–8; of predicative categories 338–45, 355–7 future 89–93, 146–9; with áere 89–93, 305, 363; with dévere 90–1; expressed by present tense 89 future-in-the-past 91 future orientation 263–4 Gallurese 5–6 gender 31–9, 41–4, 165–7, 187–8, 203, 211, 213, 220–1, 245, 300–1 generic reference 71–2, 74, 127, 129, 180, 212–13, 267–8, 274, 298 genitive, survival of Latin 34, 363 Goal 94–5, 98, 110–11, 189–93, 214, 215, 234, 237–8, 261, 264, 266, 304 Government-Binding theory 3–5 grammatical functions 94–100, 132–6, 219, 286, 362; of complement clauses 133–6, 248, 260–5; of reflexive clitics 131–6 Greek 32 Gruber 94 habitual aspect 87–9, 129 head movement 341–4 head position 3–4, 200, 204–5, 283–4, 300, 301 headless NPs 48, 69–76, 300–5 headless relatives 71–2, 300–5 heavy constituents 50–2, 329 hortative constructions 27–30, 251 human reference 71–3, 76, 65–8, 128, 200, 203, 242–5, 298, 301, 365 identificational constructions 117–19, 201, 346 illocutionary force 355–61 imperatives 26–9, 81, 218, 360 imperfect 86–8, 91–3, 306 imperfect subjunctive 80–2, 260, 278–9, 308–9 impersonal constructions 15, 20, 61, 100–6, 108, 113–14, 128, 130, 210, 262–3, 329, 331, 363, 364 impersonal passives 124, 126 implication 253, 256 in (preposition) 181–2, 187, 189–94 inalienable possession 50–1, 131, 222–4, 227, 231, 275 incorporation of in 192–3 indefinite NPs 8, 17, 18, 20, 34–6, 60–3, 67, 128, 102–6, 113–14, 209–12, 216–17, 315, 324, 329–31, 332–3 indefinite quantifiers 209–12 indirect questions 254, 259–60, 291–3, 313–14, 323, 335, 337; vs headless relatives 71, 303 infinitival relatives 298–9, 305 infinitives 80–2, 142–54, 156, 194–5, 252–3, 260–85, 286–7, 291–3, 298–9, 305, 313–14, 323, 335, 338, 340; in main clauses 29–30; with nominative subject 268, 281–2, 284–5, 252–3, 331, 363 see also inflected infinitives INFL 14–15, 60, 148–9, 156–7, 280–1, 285, 330–1 inflected infinitives 82, 252–3, 278–82, 284, 363 interrogative adverbs 25, 192, 249, 250 interrogative constructions 4–5, 24–6, 254, 259–60, 291–3, 313–14, 323, 357–9 interrogative determiners 36, 302 interrogative particle 24–6 interrogative pronouns 25, 68, 204–8, 303–4

< previous page

page_379

next page >

< previous page

page_380

next page >

Page 380 interrogative quantifiers 36 intonation 24, 310–11, 312–13, 315, 316–17, 318–20, 325–8, 332, 338–9, 347–8, 350, 356–7, 360 inversion 16, 19–20, 23, 27, 105, 126, 178, 180, 227, 281, 297, 327–31, 334, 335–6, 344–5, 346, 348–50, 352; vs right dislocation 319–20, 327–30 iskire ‘know’ 142, 146, 252, 253, 359 istare ‘be, stay’ 84, 138–9, 141–2 Italian 5, 11, 20, 31, 32, 41, 84, 85, 108, 112, 121, 128, 151, 154, 166, 167, 195, 363, 364; influence of 6–8, 166, 167, 191, 285, 296; regional variety 7 iteration: as adverb formation 167; for intensification 166 itte ‘what’ 36, 68, 204–8

ja (emphatic particle) 358–61 Jackendoff 94 Jaeggli 20, 202, 330, 342 Jones 1, 218, 281 Kayne 270, 272 ke ‘as, than’ 179–80, 185, 186 kene ‘without’ 23–4, 58, 125–6, 133, 184, 265–6, 291 kérrere ‘want’ 29–30, 142, 146, 268, 290–1, 252; in impersonal constructions 101–2, 104; with passive function 125, 290–1 ki (complementiser) 22, 27, 29–30, 174, 176–8, 180, 247–51, 258–9, 293–6, 302–5, 337 kie ‘who’ 68, 204–8, 302–4 kin ‘with’ 95–6, 183–4, 185, 199 kinship nouns 50, 55, 56, 59, 64, 66, 228 laismo 213 lassare ‘let’ 143, 270–4, 280 Latin 10–11, 31, 33, 38, 43, 80–2, 148, 166, 192, 278, 282, 358, 362 left-dislocation 16, 17–18, 66, 126, 312–19, 320, 326, 332, 333–4, 347, 349, 352 Lepori 188, 193 Location 94–100, 102–5, 189–93, 215, 233, 237–8 locative prepositions 183, 186–93 Logudorese 6, 9–11, 28, 82, 91, 148, 191, 200, 205, 213, 215, 220, 249, 281–2, 305 Loi-Corvetto 7 manipulative verbs 252–3 manner adverbs 147, 157, 161–3, 137, 167, 168 manner clauses 258–9, 304 mannu ‘great’, as degree modifier 166 meta ‘much, many, very’ 35–6, 62–3, 166, 170 methodology 1–2, 7–8 Middle constructions 79, 126, 127–30, 131, 136, 144, 146, 150, 152–3, 362 Mingione 8 modal verbs 21, 90–3, 142–6, 150, 260, 269–70, 277, 338, 365 modifier position 3–4, 50–2 moment of speech 85–6, 88 mood 80–2, 148, 251–60; in conditional constructions 305–9; in hortatives, etc. 27–9; in relative clauses 296, 297–8 morphology: adjectives 41–2, 165–6; of adverbs 165–7; of of nouns 31–4; of verbs 80–2, 278, 282 natural gender 33 nde 214–18, 223; as complement 115–16; expressing Source 234; partitive 17, 61, 69, 75, 77–8, 108, 209–12, 303, 315, 324, 333; pleonastic uses of 122–3 negation 21–4, 25, 27–8, 36–7, 59, 63–4, 143, 146, 149–50, 153, 156–7, 310–11; and mood 253–4, 256–7 negative adverbs 22–4 negative commands 27–8, 251 negative determiners 36–7, 63–4

negative polarity 64 negative pronouns 22–4, 68, 204–8 Neutral-pronominal verbs 99, 119–23, 129–31, 227, 276–7

< previous page

page_380

next page >

< previous page

page_381

next page >

Page 381 nke (adverbial clitic) 103, 113, 214–15; pleonastic uses of 122–3, 234–7 nominal infinitive 8, 260, 282–5 nominative 101–2, 106, 114, 117–19, 193, 199–200, 201, 268, 280–2, 284–5, 285, 330–1 non-count nouns 33, 35, 60–3, 64, 315, 333 noun, morphology of 31–4 NP movement 115, 117–19 nudda ‘nothing’ 22–4, 36–7, 68, 204–8 null subject 14–15, 17, 18, 126–7, 177–8, 202–3, 266–9, 280–2, 286, 296, 298, 316–17, 323, 334, 335–6, 348–9, 363 number 33–9, 42–4, 62, 71–2, 74, 76, 77, 165–7, 187–8, 210, 211, 213, 220–1, 245, 300–1 numerals 34, 39–41, 44, 48, 61–2, 209, 212 Nuorese 6, 9–11, 28, 82, 91, 148, 165, 191, 193, 200, 213, 215, 219–20, 247, 249, 278, 281–2, 305 oblique Case 185, 199 optative constructions 27–30, 251, 260, 308 ordinal numerals 41, 44 orthography 6, 8–12

pacu ‘little, few’ 35–6, 62–3, 170 párrere ‘seem’ 114–19, 255, 269, 287–8 partitive constructions 17, 35, 48–50, 58, 62, 77, 124–7, 128, 130, 132, 141, 170, 216 passive constructions 182, 274, 288, 289–91, 299 past participle 184, 195, 288–91, 299; as adjective 141, 167, 235; agreement of 83, 124, 125, 132–3; forms 80–2 perfective aspect 82–7, 140–1 performance 314–16, 318, 326 Perlmutter 96, 106 personal pronouns 199–203 see also clitics; disjunctive pronouns phonology 9–12 pied-piping 354–5 Pittau 1, 8, 43, 124–5, 165, 219, 278, 308, 340 pluperfect 83, 85, 278–9, 282, 306 plural formation 33 pluralia tantum 33–4 polite forms 26–7, 28, 200 Porru 1, 166 possessives 8, 44, 50–2, 56, 57, 59, 70, 72–3, 168, 216, 223–4, 282–5, 294, 362; as complements of prepositions 187–8, 230 pótere ‘can, may’ 142, 146 predicate nominals 58–9, 73, 131 preposing 119, 345–51, 352 see also fronting prepositional accusative 65–8, 96–8, 132, 181, 199, 203–13, 271, 297, 301, 303, 314–16, 324, 341, 363–4, 365; after prepositions 179, 186 prepositions 95–8, 180–95; with clauses 248–51, 288, 293; vs complementisers 260–6; with determinerless nouns 57–8, 64; intransitive 169, 185, 186, 188–9, 196, 230, 261; locative 229–30; with past participle 288–90; temporal 193–5 present participles 83–4, 194–5, 285–8, 299; as adverbs 167 present perfect 84, 86–8, 305, 309 present tense 80–2, 88, 305, 309 presupposition 256, 257, 259, 356, 357, 359 see also factivity pro ‘for’ 184, 186, 265–6 pro-drop 14–15, 177–8, 280–2, 363, 364 see also null subject progressive 83–4, 88–9, 129, 137–41, 287, 299, 363 proitte ‘why’ 292, 310–11 pronominal verbs 8, 99, 119–23, 129–31, 233, 276–7, 362 pronominal voice see Middle constructions pronominalisation of clauses 264–5

pronominals 239–46 pronouns 199–246; Case forms of 185–6, 199–200, 213–14; vs determiners 205–8, 365; disjunctive vs clitic 16, 200–2

< previous page

page_381

next page >

< previous page

page_382

next page >

Page 382 proper nouns 56, 64, 66 purposive clauses 151, 257, 265–6 quantifier floating 37–8, 39, 137, 147, 154–6, 208–9 quantifiers 34–9, 333; as pronouns 208–13 quantity nouns 49–50, 62, 170 questions 4–5, 18, 24–6, 59–60, 339, 343, 355–8; indirect 254, 259–60, 291–3, 313–14, 323, 335, 337 R-expressions 239, 244 Radford 78 reanalysis 78–9, 151, 271, 342, 344 reason clauses 249–50, 257–8, 285, 310 reciprocal constructions 245–6 recoverability of deletion 294–6 reference-time 85–9, 194, 197 reflexive clitics 99, 121–2, 130–6, 213, 240, 364; in Middle constructions 127–30; omission of in causatives 276–7 pleonastic uses of 122, 232–6; in pronominal verbs 119–23; with reciprocal interpretation 245 register 124–5, 126, 279, 285, 296 Reichenbach 85 relative clauses 293–9, 300–5; mood in 259 relative pronouns 293, 296–7, 298 restrictive modification 70–1, 73 resumptive pronouns 16–17, 61, 77–8, 231, 293–9, 312, 316, 323–6, 332, 343, 348, 349–50 right-dislocation 16, 19–20, 77–8, 161, 318–30, 343–4, 352–3, 354, 356, 360 Rizzi 20, 108, 151, 154, 330 Rohlfs 65 Romance languages 1, 6, 14, 65, 148, 167, 190, 200, 267, 270, 273, 286, 314, 325, 330, 362–6 root modality 145–6, 298 Rouveret 272 Rumanian 364 Ruwet 128

s’unu…s’átteru ‘each other’ 245–6 Sassarese 6 scope: of adverbs 161; of existential operator bi 104, 114; of negative elements 23–4, 63–4, 310–11 semi-auxiliaries 149–53 sentence, structure of 4, 13–21 sentence stress 19, 77, 118, 203, 310–11, 313–14, 316–17, 318–20, 322, 324, 324–8, 330, 332, 336–7, 338–9, 341, 345–9, 351–61 si (reflexive clitic) 99, 121–2, 130–6, 213, 240, 364; in Middle constructions 127–30; omission of in causatives 276–7; pleonastic uses of 122, 232–6; in pronominal verbs 119–23; with reciprocal interpretation 245; as substitute for dative clitic 213, 214, 220 si ‘if’ 249, 259–60, 305–9; in hortative constructions 27 si ‘whether’ 291 small clauses 115–17, 139, 288 sociolinguistic factors 5–8 Source 94–5, 110–11, 182, 189–93, 215, 226, 234 Spanish 6, 7, 20, 31, 202, 213, 325, 342, 363 Spano 1, 8, 147, 166 Specified Subject Condition 116 specifier position 3–4, 157 stative verbs 87, 138–9, 306 structure 3–5, 50, 157, 161, 163–4, 342–3; of clause 4, 154, 248–9, 262, 313–14, 320–9, 326, 330, 336, 345; of nominal infinitives 283–4; of NP 45–79, 163–4, 283–4, 303; of PP 187–8, 262;

of VP 139–41, 151, 154–64, 330, 345 subject clauses 262 subject-object asymmetries 60–4 subject-raising 151, 269–70, 287–8 subjectivisation 102–6, 114, 124, 127–8, 136, 146, 150 subjunctive 80–2, 148, 195, 250, 251–60, 296, 297–8, 308–9; in main clauses 26–9, 251 subordinate clauses 247–311 subordinating conjunctions 185, 195, 288 superlative 44, 74, 165, 166, 178–9

< previous page

page_382

next page >

< previous page

page_383

Page 383 tense 80–9, 148, 231, 235, 309, 362–3; and Case assignment 280–2; modal uses 260, 305–9 Theme 20, 94–9, 106–7, 110–15, 119–22, 182, 235 Theta-roles 5, 94–100, 112, 119–22, 128, 130, 135, 221–2, 364 time clauses 194, 250, 258, 288–90, 304–5 topic 18, 126, 202–3, 312–15, 318 TOPIC position 317 tottu 37–9, 45–6, 67–8, 137, 154–6, 171, 208–9, 301–2, 305 transfer of number and gender features 220–1 transitivity 94–100 Tuscan 32 unaccusative verbs 20–1, 79, 96–100, 102–5, 106–13, 114, 120–1, 127, 130, 132, 139, 150, 152, 210, 227– 9, 234, 235, 289–90, 299, 328–31, 344–5, 363, 364 unu ‘a, one’ 34–5, 49, 211–12 verb agreement 20, 49, 61, 101–6, 113–14, 117–19, 127–8, 144, 146, 150, 152–3, 193, 201 verb inflection 15, 26, 80–2, 278, 282 verb raising 272–8 verb-second 336, 350, 362 verbal complex 154 verbs 80–158; causative 270–8; manipulative 252–3; of perception 286; stative 87, 138–9, 306 see also auxiliary verbs; modal verbs; unaccusative verbs Vergnaud 272 Virdis 1 vocative 56 vowels 9, 31–2 weather verbs 100–1, 130 Wagner 1, 42–3, 167 Wasow 118 WH—adverbs 304–5, 310–1 WH—items 250, 292–3, 294, 296, 302, 304–5, 310–1, 313–4, 316, 326, 335–7, 339, 349–50, 353, 354–5 WH—movement 25–6, 292, 294–8, 302–4, 334–8, 349–50, 353, 354–5 word stress 9 word-formation 34, 167–8 X-bar framework 3–5, 50, 157, 161, 163–4, 342–3

< previous page

page_383