Eurotyp: 8 Pragmatic Organization of Discourse in the Languages of Europe [Reprint 2012 ed.] 9783110892222, 9783110157536

The volume is a collection of papers reporting the results of investigations on the interaction of discourse and sentenc

352 95 3MB

English Pages 656 Year 2006

Report DMCA / Copyright

DOWNLOAD PDF FILE

Table of contents :
General Preface
Contributors
I. General issue
Introduction
Spoken and written language
II. From function to form
Sentence topics in the languages of Europe and beyond
Focus in the languages of Europe
Deixis and anaphora: Some case studies
Theticity
III. From form to function
The de-accenting of given information: A cognitive universal?
Interaction of syntactic and pragmatic factors on basic word order in the languages of Europe
Articles and article systems in some areas of Europe
Indexes
Subject index
Language index
Author index
Recommend Papers

Eurotyp: 8 Pragmatic Organization of Discourse in the Languages of Europe [Reprint 2012 ed.]
 9783110892222, 9783110157536

  • 0 0 0
  • Like this paper and download? You can publish your own PDF file online for free in a few minutes! Sign Up
File loading please wait...
Citation preview

Pragmatic Organization of Discourse in the Languages of Europe



Empirical Approaches to Language Typology Eurotyp

Editors Georg Bossong Bernard Comrie

Mouton de Gruyter Berlin · New York

20-8

Pragmatic Organization of Discourse in the Languages of Europe edited by Giuliano Bernini Marcia L. Schwartz

Mouton de Gruyter Berlin · New York

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

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

Library of Congress Cataloging-in-Publication Data Pragmatic organization of discourse in the languages of Europe / edited by Giuliano Bernini, Marcia L. Schwartz. p. cm. ⫺ (Empirical approaches to language typology. EUROTYP ; 20-8) Includes bibliographical references and index. ISBN-13: 978-3-11-015753-6 (hardcover : alk. paper) ISBN-10: 3-11-015753-5 (hardcover : alk. paper) 1. Europe ⫺ Languages. 2. Pragmatics. 3. Grammar, Comparative and general ⫺ Syntax. 4. Discourse analysis. I. Bernini, Giuliano. II. Schwartz, Marcia L. III. Series. P380.P73 2006 306.44094⫺dc22 2005031059

Bibliographic information published by Die Deutsche Bibliothek Die Deutsche Bibliothek lists this publication in the Deutsche Nationalbibliografie; detailed bibliographic data is available in the Internet at ⬍http://dnb.ddb.de⬎.

ISBN-13: 978-3-11-015753-6 ISBN-10: 3-11-015753-5 © Copyright 2006 by Walter de Gruyter GmbH & Co. KG, D-10785 Berlin. All rights reserved, including those of translation into foreign languages. No part of this book may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopy, recording or any information storage and retrieval system, without permission in writing from the publisher. Typesetting: Christoph Eyrich, Berlin. Printing and binding: AZ Druck und Datentechnik GmbH, Kempten (Allgäu). Printed in Germany.

The present volume is one of a series of nine volumes in which the results of the European research project “Typology of Languages in Europe” (EUROTYP) are published. The initiative for a European project on language typology came from a proposal jointly submitted to the European Science Foundation (ESF) by Johannes Bechert (University of Bremen), Claude Buridant (University of Strasbourg), Martin Harris (University of Salford, now University of Manchester) and Paolo Ramat (University of Pavia). On the basis of this proposal and following consultations with six experts the Standing Committee for the Humanities of the ESF decided to organize a workshop (Rome, January 1988), in which this idea was further explored and developed. The results of this workshop (published by Mouton, 1990) were sufficiently encouraging for the Standing Committee to appoint a preparatory committee and entrust it with the tasks of drawing up a preliminary proposal, of securing interest and participation from a sufficiently large number of scholars and of finding a suitable programme director. The project proposal formulated and sent out by Simon Dik (University of Amsterdam) as chair of this committee met with very supportive and enthusiastic reactions, so that the Standing Committee for the Humanities recommended the funding of a planning stage and the General Assembly of the ESF approved a year zero (1989) for an ESF Programme in Language Typology. During this planning phase all major decisions concerning the management structure and the organisation of the work were taken, i.e., the selection of a programme director, the selection of nine focal areas around which the research was to be organized, the selection of a theme coordinator for each theme and the selection of the advisory committee. The first task of the programme director was to draw up a definitive project proposal, which was supplemented with individual proposals for each theme formulated by the theme coordinators, and this new proposal became the basis of a decision by the ESF to fund the Programme for a period of five years (1990–1994). Language typology is the study of regularities, patterns and limits in crosslinguistic variation. The major goal of EUROTYP was to study the patterns and limits of variation in nine focal areas: pragmatic organization of discourse, constituent order, subordination and complementation, adverbial constructions, tense and aspect, noun phrase structure, clitics and word prosodic systems in the languages of Europe. The decision to restrict the investigation to the languages of Europe was imposed for purely practical and pragmatic reasons. In the course of the project an attempt was made, however, to make as much sense of this restriction as possible, by characterizing the specific features of European languages against the background of

vi

General Preface

non-European languages and by identifying areal phenomena (Sprachbünde) within Europe. More specifically, the goals of the EUROTYP project included the following: – to contribute to the analysis of the nine domains singled out as focal areas, to assess patterns and limits of cross-linguistic variation and to offer explanations of the patterns observed. – to bring linguists from various European countries and from different schools or traditions of linguistics together within a major international project on language typology and in doing so create a new basis for future cooperative ventures within the field of linguistics. More than 100 linguists from more than 20 European countries and the United States participated in the project. – to promote the field of language typology inside and outside of Europe. More specifically, an attempt was made to subject to typological analysis a large number of new aspects and domains of language which were uncharted territory before. – to provide new insights into the specific properties of European languages and thus contribute to the characterization of Europe as a linguistic area (Sprachbund). – to make a contribution to the methodology and the theoretical foundations of typology by developing new forms of cooperation and by assessing the role of inductive generalization and the role of theory construction in language typology. We had a further, more ambitious goal, namely to make a contribution to lingustic theory by uncovering major patterns of variation across an important subset of languages, by providing a large testing ground for theoretical controversies and by further developing certain theories in connection with a variety of languages. The results of our work are documented in the nine final volumes: Pragmatic Organization of Discourse in the Languages of Europe (edited by G. Bernini) Constituent Order in the Languages of Europe (edited by A. Siewierska) Subordination and Complementation in the Languages of Europe (edited by N. Vincent) Actance et Valence dans les langues d l’Europe (edited by J. Feuillet) Adverbial Constructions in the Languages of Europe (edited by J. van der Auwera) Tense and Aspect in the Languages of Europe (edited by Ö. Dahl) Noun Phrase Structure in the Languages of Europe (edited by F. Plank) Clitics in the Languages of Europe (edited by H. van Riemsdijk) Word Prosodic Systems in the Languages of Europe (edited by H. van der Hulst) In addition, the EUROTYP Project led to a large number of related activities and publications, too numerous to be listed here. At the end of this preface, I would like to express my profound appreciation to all organizations and individuals who made this project possible. First and foremost, I must mention the European Science Foundation, who funded and supported the

General Preface

vii

Programme. More specifically, I would like to express my appreciation to Christoph Mühlberg, Max Sparreboom and Geneviève Schauinger for their constant and efficient support, without which we would not have been able to concentrate on our work. I would, furthermore, like to thank my colleague and former assistant, Martin Haspelmath, and indeed all the participants in the Programme for their dedication and hard work. I finally acknowledge with gratitude the crucial role played by Johannes Bechert and Simon Dik in getting this project off the ground. Their illness and untimely deaths deprived us all of two of the project’s major instigators. Ekkehard König, Programme Director

Contents General Preface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

v

Contributors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xxiii I.

General issue

Giuliano Bernini Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

3

Jim Miller and M. M. Jocelyne Fernandez-Vest Spoken and written language . . . . . . . . . . . . . . . . . . . . . . . . . .

9

II.

From function to form

Elena Maslova and Giuliano Bernini Sentence topics in the languages of Europe and beyond . . . . . . . . . . . .

67

Jim Miller Focus in the languages of Europe . . . . . . . . . . . . . . . . . . . . . . . . 121 Yaron Matras and A. Machtelt Bolkestein Deixis and anaphora: Some case studies . . . . . . . . . . . . . . . . . . . . 215 Hans-Jürgen Sasse Theticity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 255 III.

From form to function

Alan Cruttenden The de-accenting of given information: A cognitive universal? . . . . . . . . 311 Rosanna Sornicola Interaction of syntactic and pragmatic factors on basic word order in the languages of Europe . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 357 Christoph Schroeder Articles and article systems in some areas of Europe . . . . . . . . . . . . . . 545

x

Contents

Indexes Subject index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 615 Language index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 619 Author index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 625

Contents detailed General Preface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

v

Contributors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xxiii

I.

General issue

Giuliano Bernini Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Jim Miller and M. M. Jocelyne Fernandez-Vest Spoken and written language . . . . . . . . . . . . . . . . . . . . . . . . . . 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2. Distinguishing spoken and written language . . . . . . . . . . . . . . . 2.1. Genres . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.1.1. Biber’s textual dimensions . . . . . . . . . . . . . . . . 2.1.2. Spontaneous spoken language, written language, and education . . . . . . . . . . . . . . . . . . . . . . . 2.2. Previous work on spoken and written language . . . . . . . . . . 2.2.1. General grammatical properties of spontaneous spoken language . . . . . . . . . . . . . . . . . . . . . . . . . 2.2.2. General discourse properties of spontaneous spoken language . . . . . . . . . . . . . . . . . . . . . . . . . 2.2.3. Analyses contra a special status for spontaneous spoken language . . . . . . . . . . . . . . . . . . . . . . . . . 3. Sentences and clauses . . . . . . . . . . . . . . . . . . . . . . . . . . 3.1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2. System sentences and text sentences . . . . . . . . . . . . . . . 3.3. Sentences in spoken texts . . . . . . . . . . . . . . . . . . . . . 3.3.1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . 3.3.2. Intuitions about sentences in spoken language . . . . . . 3.3.3. The sentence: A changing concept . . . . . . . . . . . . 3.4. Sentences and syntactic analysis . . . . . . . . . . . . . . . . . . 3.4.1. The clause as the central unit of syntactic analysis . . . . 3.4.2. Dependencies crossing sentence boundaries . . . . . . . 3.4.3. Sentences, clause, and distribution . . . . . . . . . . . . 4. “Syntactic fragments,” competence, and performance . . . . . . . . . .

3

9 9 9 9 10 11 13 13 14 15 16 16 16 19 19 20 20 24 24 27 27 29

xii

Contents detailed

5. 6. 7. 8.

Clause and clause complex . . . . . . . . . . . . . . . . . . . . . . Noun phrases . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Discourse . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Typology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8.1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . 8.2. Relative clauses, spoken and written language, and typology . 8.2.1. Typological overview . . . . . . . . . . . . . . . . . 8.2.2. The order of the head NP and the modifying relative clause . . . . . . . . . . . . . . . . . . . . . . . . . 8.2.3. How the head noun is reflected in the relative clause 8.2.4. English relative clauses . . . . . . . . . . . . . . . . 8.2.5. Relative clauses in Russian . . . . . . . . . . . . . . 9. Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Notes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

II.

. . . . . . .

. . . . . . .

33 35 36 44 44 46 46

. . . . . . .

. . . . . . .

46 48 49 54 56 56 61

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

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

67 67 68 71 71 73 76 80 83 83 84 88 93 96 100 100 103 108 113 114

From function to form

Elena Maslova and Giuliano Bernini Sentence topics in the languages of Europe and beyond . . . . . . . . . 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2. The hypothesis of existence for sentence topic: Pro and contra . . 3. Phenomenology of sentence topics: An overview . . . . . . . . . 3.1. Packaging topics . . . . . . . . . . . . . . . . . . . . . . . 3.2. Hanging topics . . . . . . . . . . . . . . . . . . . . . . . . 3.3. Topic constructions . . . . . . . . . . . . . . . . . . . . . 3.4. Non-nominal topics . . . . . . . . . . . . . . . . . . . . . 4. Semantic grounding of the topic-comment structure . . . . . . . . 4.1. Introductory remarks . . . . . . . . . . . . . . . . . . . . . 4.2. Primary participants . . . . . . . . . . . . . . . . . . . . . 4.3. Reference points . . . . . . . . . . . . . . . . . . . . . . . 4.4. Spatial frames . . . . . . . . . . . . . . . . . . . . . . . . 4.5. Three semantic templates vs. three aspects of topicality . . 5. Language-internal and crosslinguistic variation in topic encoding 5.1. Two dimensions of topic prominence . . . . . . . . . . . . 5.2. Competition between topic-encoding templates . . . . . . . 5.3. Interaction between topic-encoding templates . . . . . . . . 5.4. Conclusion: Universal constraints on topic encoding . . . . Notes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

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

xiii

Contents detailed

Jim Miller Focus in the languages of Europe . . . . . . . . . . . . . . . . . . 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . 2. Theory and data . . . . . . . . . . . . . . . . . . . . . . . . 2.1. Basic concepts . . . . . . . . . . . . . . . . . . . . . 2.2. Data collection . . . . . . . . . . . . . . . . . . . . . 3. Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4. Focus . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.1. Overview . . . . . . . . . . . . . . . . . . . . . . . . 4.2. Halliday . . . . . . . . . . . . . . . . . . . . . . . . 4.3. Chafe: Components of contrastiveness . . . . . . . . 4.4. Dik: Extended notion of focus . . . . . . . . . . . . . 4.5. Vallduví . . . . . . . . . . . . . . . . . . . . . . . . 4.6. Grosz and Sidner: Focus spaces and transition . . . . 5. New individuals and sets . . . . . . . . . . . . . . . . . . . 5.1. Focus spaces, file cards and entities . . . . . . . . . . 5.2. English . . . . . . . . . . . . . . . . . . . . . . . . . 5.2.1. New entities in written texts . . . . . . . . . 5.2.2. NP-clause . . . . . . . . . . . . . . . . . . . 5.2.3. see and know . . . . . . . . . . . . . . . . 5.3. German . . . . . . . . . . . . . . . . . . . . . . . . . 5.4. French . . . . . . . . . . . . . . . . . . . . . . . . . 5.5. Russian . . . . . . . . . . . . . . . . . . . . . . . . . 5.6. Hungarian . . . . . . . . . . . . . . . . . . . . . . . 5.7. Finnish . . . . . . . . . . . . . . . . . . . . . . . . . 5.8. Turkish . . . . . . . . . . . . . . . . . . . . . . . . . 5.9. Irish and Scottish Gaelic . . . . . . . . . . . . . . . . 5.10. Bulgarian . . . . . . . . . . . . . . . . . . . . . . . . 5.11. Topic-prominent and subject-prominent languages . . 6. New propositions . . . . . . . . . . . . . . . . . . . . . . . 6.1. English . . . . . . . . . . . . . . . . . . . . . . . . . 6.1.1. the thing is . . . . . . . . . . . . . . . . . . 6.1.2. see introducing propositions . . . . . . . . . 6.2. German . . . . . . . . . . . . . . . . . . . . . . . . . 6.3. Greek . . . . . . . . . . . . . . . . . . . . . . . . . . 6.4. Russian . . . . . . . . . . . . . . . . . . . . . . . . . 6.5. French . . . . . . . . . . . . . . . . . . . . . . . . . 7. Russian word order . . . . . . . . . . . . . . . . . . . . . . 7.1. Introduction: Siewierska and Bivon . . . . . . . . . . 7.2. Word order in spoken and written Russian and Polish 7.3. Clause position in written Russian . . . . . . . . . . .

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

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

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

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

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

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

121 121 122 122 123 127 129 129 130 130 131 132 133 134 134 135 135 135 136 137 138 139 142 142 142 142 143 144 145 145 145 146 147 148 148 151 151 151 152 154

xiv

Contents detailed

7.4. Word order as a highlighting device . . . . . . . 8. Reduction in salience . . . . . . . . . . . . . . . . . . 8.1. Overview . . . . . . . . . . . . . . . . . . . . . 8.2. Ellipsis in Russian . . . . . . . . . . . . . . . . 8.2.1. Null subject NPs . . . . . . . . . . . . 8.3. Null object NPs . . . . . . . . . . . . . . . . . 8.3.1. Null NPs in spoken and written Russian 8.4. Salience-reducing constructions . . . . . . . . . 9. Highlighting devices (nonpresentational) . . . . . . . 9.1. Discourse particles . . . . . . . . . . . . . . . . 9.2. Word order . . . . . . . . . . . . . . . . . . . . 9.3. Contrastive focus position . . . . . . . . . . . . 9.4. Syntax and intonation . . . . . . . . . . . . . . 9.5. Clefts . . . . . . . . . . . . . . . . . . . . . . . 9.5.1. Definition of cleft constructions . . . . 9.5.2. Languages without clefts . . . . . . . . 9.5.3. Languages with clefts . . . . . . . . . 10. Spontaneous map task dialogues . . . . . . . . . . . . 10.1. Introduction . . . . . . . . . . . . . . . . . . . 10.2. English . . . . . . . . . . . . . . . . . . . . . . 10.3. German . . . . . . . . . . . . . . . . . . . . . . 10.4. Catalan . . . . . . . . . . . . . . . . . . . . . . 10.5. Hungarian . . . . . . . . . . . . . . . . . . . . 10.6. Bulgarian . . . . . . . . . . . . . . . . . . . . . 11. Conclusions . . . . . . . . . . . . . . . . . . . . . . . Notes . . . . . . . . . . . . . . . . . . . . . . . . . . . .

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

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

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

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

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

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

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

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

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

156 157 157 158 158 160 162 162 165 165 166 166 170 171 171 174 179 188 188 189 190 193 198 200 203 205

Yaron Matras and A. Machtelt Bolkestein Deixis and anaphora: Some case studies . . . . . . . . . . . . . . . . . . . . 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2. Extratextual deixis, intratextual deixis and anaphora . . . . . . . . . . 3. Extratextual deixis, intratextual deixis, and anaphora in Classical Latin 3.1. Extratextual deixis . . . . . . . . . . . . . . . . . . . . . . . . . 3.2. The system of intratextual deixis and anaphora in Latin . . . . . 4. Situational deixis, intratextual deixis, and anaphora in modern Dutch . 4.1. Situational deixis in Dutch . . . . . . . . . . . . . . . . . . . . . 4.2. Intratextual deixis and anaphora in Dutch . . . . . . . . . . . . . 5. Deixis and anaphora in Romani and Modern Hebrew . . . . . . . . . . 5.1. Romani . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.2. Hebrew . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

215 215 216 220 220 221 230 230 231 239 239 246

Contents detailed

xv

6. Concluding remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . 249 Notes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 250 Hans-Jürgen Sasse Theticity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.1. The phenomenon . . . . . . . . . . . . . . . . . . . . . . . . . . 1.2. Previous research . . . . . . . . . . . . . . . . . . . . . . . . . 1.3. The present research strategy . . . . . . . . . . . . . . . . . . . 2. Constructions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.1. Formal types of “special” constructions in the domain of theticity 2.2. Distribution of formal types across languages . . . . . . . . . . . 2.3. Text frequency of “special” constructions . . . . . . . . . . . . . 2.4. Polysemy of constructions . . . . . . . . . . . . . . . . . . . . . 2.4.1. VS constructions . . . . . . . . . . . . . . . . . . . . . 2.4.1.1. Narrow focus . . . . . . . . . . . . . . . . . . . . . . . 2.4.1.2. Inversion . . . . . . . . . . . . . . . . . . . . . . . . . 2.4.2. SAcc and Split constructions . . . . . . . . . . . . . . . 2.5. Language-specific restrictions . . . . . . . . . . . . . . . . . . . 2.5.1. Global syntactic patterns . . . . . . . . . . . . . . . . . 2.5.2. Monoargumentality . . . . . . . . . . . . . . . . . . . . 2.5.3. Definiteness and animacy . . . . . . . . . . . . . . . . 2.5.4. Local restrictions . . . . . . . . . . . . . . . . . . . . . 3. Theticity and discourse functions . . . . . . . . . . . . . . . . . . . . 3.1. The relevant discourse-functional domains . . . . . . . . . . . . 3.1.1. The annuntiative function . . . . . . . . . . . . . . . . 3.1.2. The introductive function . . . . . . . . . . . . . . . . . 3.1.3. The interruptive function . . . . . . . . . . . . . . . . . 3.1.4. The descriptive function . . . . . . . . . . . . . . . . . 3.1.5. The explanative function . . . . . . . . . . . . . . . . . 3.2. Similarities and differences among the discourse functions . . . . 3.3. Coverage of functions by construction types in the different languages . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4. The “connective” complex . . . . . . . . . . . . . . . . . . . . . . . . 5. The impact of lexical semantics . . . . . . . . . . . . . . . . . . . . . 6. Summary, conclusions, and perspectives for future research . . . . . . Notes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

255 255 255 257 262 264 264 267 269 271 271 271 272 273 274 274 277 279 280 280 280 281 284 285 286 287 288 290 294 297 299 302

xvi III.

Contents detailed

From form to function

Alan Cruttenden The de-accenting of given information: A cognitive universal? 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . 2. De-accenting and given information . . . . . . . . . . . 3. De-accenting and re-accenting . . . . . . . . . . . . . . 4. Settings and responses . . . . . . . . . . . . . . . . . . 4.1. Setting-response 1 . . . . . . . . . . . . . . . . . 4.2. Setting-response 2 . . . . . . . . . . . . . . . . . 4.3. Setting-response 3 . . . . . . . . . . . . . . . . . 4.4. Setting-response 4 . . . . . . . . . . . . . . . . . 4.5. Setting-response 5 . . . . . . . . . . . . . . . . . 4.6. Setting-response 6 . . . . . . . . . . . . . . . . . 4.7. Setting-response 7 . . . . . . . . . . . . . . . . . 4.8. Setting-response 8 . . . . . . . . . . . . . . . . . 4.9. Setting-response 9 . . . . . . . . . . . . . . . . . 4.10. Setting-response 10 . . . . . . . . . . . . . . . . 5. Conclusions . . . . . . . . . . . . . . . . . . . . . . . . 5.1. Re-accenting in Romance and other languages . . 5.2. Contrast . . . . . . . . . . . . . . . . . . . . . . 5.3. Discourse and grammar . . . . . . . . . . . . . . A Albanian . . . . . . . . . . . . . . . . . . . . . . B Arabic (Tunisian) . . . . . . . . . . . . . . . . . C French . . . . . . . . . . . . . . . . . . . . . . . D German . . . . . . . . . . . . . . . . . . . . . . . E Greek . . . . . . . . . . . . . . . . . . . . . . . . F Italian . . . . . . . . . . . . . . . . . . . . . . . G Lithuanian . . . . . . . . . . . . . . . . . . . . . H Macedonian . . . . . . . . . . . . . . . . . . . . I Russian . . . . . . . . . . . . . . . . . . . . . . . J Spanish . . . . . . . . . . . . . . . . . . . . . . . K Swedish . . . . . . . . . . . . . . . . . . . . . . Notes . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

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

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

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

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

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

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

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

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

311 311 314 315 318 318 320 320 321 322 323 324 324 325 326 327 327 329 331 333 335 337 339 340 342 344 345 347 349 351 352

Rosanna Sornicola Interaction of syntactic and pragmatic factors on basic word order in the languages of Europe . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1. Theoretical and methodological foundations . . . . . . . . . . . . 1.1. A pragmatic study of word order . . . . . . . . . . . . . . 1.2. Word order between competence and performance . . . . .

. . . .

. . . .

. . . .

357 357 357 358

Contents detailed

1.3. 1.4. 1.5. 1.6.

Microscopy and macroscopy . . . . . . . . . . . . . . . . . . . The languages of the corpus . . . . . . . . . . . . . . . . . . . . The method of gathering data . . . . . . . . . . . . . . . . . . . Topology and syntax: Towards a geometry of the sentence . . . . 1.6.1. Notions of order . . . . . . . . . . . . . . . . . . . . . 1.6.2. Models of order as a result of movement . . . . . . . . . 1.6.3. The concept of position . . . . . . . . . . . . . . . . . . 1.6.4. Representation by fields . . . . . . . . . . . . . . . . . 1.7. The descriptive traditions . . . . . . . . . . . . . . . . . . . . . 1.7.1. Diversity in traditions of word order study . . . . . . . . 1.7.2. The d-configurationality model . . . . . . . . . . . . . . 1.7.2.1. Heltoft’s criticism of generative models . . . . . . . . . 1.7.2.2. Configurationality and d-configurationality . . . . . . . 1.8. Form and function in the study of word order . . . . . . . . . . . 1.8.1. Grammatical functions, semantic functions, pragmatic functions . . . . . . . . . . . . . . . . . . . . . . . . . 1.8.1.1. The syntactic and prosodic realization of pragmatic functions . . . . . . . . . . . . . . . . . . . . . . . . . 2. Word order in sentences with two arguments: The relationship between position, syntactic function, and pragmatic function . . . . . . . . . . . 2.1. TOPIC position . . . . . . . . . . . . . . . . . . . . . . . . . . 2.2. Languages with O in preverbal position in the basic order and FOCUS in preverbal position . . . . . . . . . . . . . . . . . . . 2.2.1. Basque and Turkish . . . . . . . . . . . . . . . . . . . . 2.3. Problematic cases concerning the relationship between position, grammatical function, and pragmatic function . . . . . . . . . . 2.3.1. Georgian . . . . . . . . . . . . . . . . . . . . . . . . . 2.3.2. Hungarian . . . . . . . . . . . . . . . . . . . . . . . . . 2.4. The interplay of position, grammatical functions, and pragmatic functions in languages with O in postverbal position . . . . . . . 2.4.1. General considerations about the position of the FOCUS constituent in SVO languages . . . . . . . . . . . . . . 2.4.2. SVO languages . . . . . . . . . . . . . . . . . . . . . . 2.4.3. SVO languages with highly flexible word order . . . . . 2.4.3.1. Estonian and Finnish . . . . . . . . . . . . . . . . . . . 2.4.3.2. The Slavonic languages . . . . . . . . . . . . . . . . . 2.4.4. SVO languages with weak word order flexibility . . . . 2.4.4.1. The Romance languages . . . . . . . . . . . . . . . . . 2.4.4.2. English . . . . . . . . . . . . . . . . . . . . . . . . . . 2.4.4.3. Some remarks on Italian and Spanish . . . . . . . . . . 2.4.5. V-2 languages . . . . . . . . . . . . . . . . . . . . . . .

xvii 361 361 363 364 364 365 366 368 368 368 370 370 372 374 375 377 379 379 379 381 384 384 387 389 389 391 393 393 397 403 403 409 411 413

xviii

Contents detailed

2.4.5.1. The role of scientific traditions in the formation of the V-2 concept . . . . . . . . . . . . . . . . . . . . . . . 2.4.5.2. The Germanic languages . . . . . . . . . . . . . . . . . 2.4.5.3. Vorfeld, Mittelfeld, Nachfeld . . . . . . . . . . . . . . . 2.4.5.4. Some non-V-2 structures . . . . . . . . . . . . . . . . . 2.4.5.5. German . . . . . . . . . . . . . . . . . . . . . . . . . . 2.4.6. V-initial languages . . . . . . . . . . . . . . . . . . . . 2.4.6.1. Irish and Welsh . . . . . . . . . . . . . . . . . . . . . . 2.4.6.2. Some problems concerning the relationship between position, grammatical function, and pragmatic function . 2.4.6.3. Concurrent SVO types . . . . . . . . . . . . . . . . . . 2.4.6.4. Properties of position P1 . . . . . . . . . . . . . . . . . 2.4.6.5. More on position P1 : A controversial type with NP in P1 2.4.6.6. A few concluding remarks . . . . . . . . . . . . . . . . 3. Word order in sentences with one-argument verbs . . . . . . . . . . . . 3.1. Monoargumentality and word order patterns . . . . . . . . . . . 3.1.1. Constructions with one-argument verbs . . . . . . . . . 3.1.2. Between the microscopic and the macroscopic . . . . . 3.2. SVO languages . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2.1. Indo-European languages . . . . . . . . . . . . . . . . . 3.2.1.1. The (X)VS pattern . . . . . . . . . . . . . . . . . . . . 3.2.2. The Balto-Finnic languages . . . . . . . . . . . . . . . 3.3. VSO languages . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.4. SOV languages . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.5. Hungarian . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.6. Between universal tendencies and typological conditions . . . . . 3.6.1. Structural factors and statistical regularities in unstable orders . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.6.2. Some problems with generative models . . . . . . . . . 3.6.3. Typological conditions . . . . . . . . . . . . . . . . . . 4. V or sentence in FOCUS . . . . . . . . . . . . . . . . . . . . . . . . . 4.1. V in FOCUS . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.2. Word order when sentence is all-in-FOCUS . . . . . . . . . . . 4.2.1. A few conclusions on typological properties of all-in-FOCUS structures . . . . . . . . . . . . . . . . . 5. The influence of semantic or textual features on word order . . . . . . . 5.1. The influence of the semantic features animacy, definiteness, and referentiality on word order . . . . . . . . . . . . . . . . . . . . 5.2. Textual relations . . . . . . . . . . . . . . . . . . . . . . . . . . 5.2.1. The effect of the features GIVEN and NEW on word order . . . . . . . . . . . . . . . . . . . . . . . .

413 414 416 417 419 429 429 429 431 432 434 437 442 442 442 445 447 447 449 452 452 453 455 456 456 457 460 461 461 464 466 466 466 470 470

xix

Contents detailed

5.2.2. The distribution of GIVEN and NEW in SVO languages 5.2.3. The distribution of GIVEN and NEW in VSO languages 5.2.4. The distribution of GIVEN and NEW in SOV languages 6. Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.1. Universal tendencies and the typology of word order . . . . . . . 6.1.1. The “written” vs. “spoken” parameter . . . . . . . . . . 6.1.2. The historical conditions . . . . . . . . . . . . . . . . . 6.1.3. Structural conditions . . . . . . . . . . . . . . . . . . . 6.1.4. Criteria for determination of the sentence space . . . . . 6.1.5. Some principles of word order organization . . . . . . . 6.2. Word order and the asymmetry of pragmatic functions . . . . . . 6.2.1. The relationship between position, grammatical function, and pragmatic function . . . . . . . . . . . . . . . . . . 6.2.2. Codification of FOCUS by means of word order . . . . 6.2.3. The relationship between the grammatical function O and FOCUS . . . . . . . . . . . . . . . . . . . . . . . . . . 6.2.4. Some possible typological correlations with the parameter “FOCUS in Pk ” . . . . . . . . . . . . . . . . 6.2.4.1. Hypothesis 1 . . . . . . . . . . . . . . . . . . . . . . . 6.2.4.2. Hypothesis 2 . . . . . . . . . . . . . . . . . . . . . . . 6.2.4.3. A third hypothesis . . . . . . . . . . . . . . . . . . . . Notes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Christoph Schroeder Articles and article systems in some areas of Europe . . . . . . . . . . 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2. The categories . . . . . . . . . . . . . . . . . . . . . . . . . . 2.1. Definiteness . . . . . . . . . . . . . . . . . . . . . . . . 2.1.1. Identifiability . . . . . . . . . . . . . . . . . . . 2.1.2. Identifiability based on anaphora . . . . . . . . . 2.1.3. Relationship of identity . . . . . . . . . . . . . . 2.1.4. Identifiability and situational context . . . . . . . 2.1.5. Generic NPs and proper names . . . . . . . . . . 2.2. Indefiniteness, specificity, referentiality, nonreferentiality 3. Articles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.1. Means of expression of (in)definiteness . . . . . . . . . . 3.2. Definite articles . . . . . . . . . . . . . . . . . . . . . . 3.3. Demonstratives and possessives as articles . . . . . . . . 3.4. Syntacticization trends of definite articles . . . . . . . . . 3.5. Indefinite articles . . . . . . . . . . . . . . . . . . . . . . 3.6. From numeral to indefinite article . . . . . . . . . . . . .

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

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

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

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

471 474 475 476 476 476 477 478 479 481 486 487 489 491 492 492 495 496 497

545 545 546 546 546 547 548 549 550 551 552 552 553 553 555 555 556

xx

Contents detailed

3.7. Emergent indefinite article . . . . . . . . . . . . . . . . . . . . 3.8. Pragmatic indefinite article . . . . . . . . . . . . . . . . . . . 3.9. Referential indefinite article . . . . . . . . . . . . . . . . . . . 3.10. Extended indefinite article . . . . . . . . . . . . . . . . . . . . 4. The survey . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.1. The north: Two definite articles . . . . . . . . . . . . . . . . . 4.1.1. North Frisian and Low German . . . . . . . . . . . . 4.1.2. Danish dialects of Jutland . . . . . . . . . . . . . . . 4.1.3. Standard Danish . . . . . . . . . . . . . . . . . . . . 4.1.4. Swedish, Norwegian, Faroese . . . . . . . . . . . . . 4.2. Excursus 1: Two definite articles in the Low German dialects of the North Sea region . . . . . . . . . . . . . . . . . . . . . . . 4.3. Transitions I . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.3.1. Adjectival endings . . . . . . . . . . . . . . . . . . . 4.3.2. Attributive and predicative forms of adjectives . . . . 4.3.3. Baltic languages . . . . . . . . . . . . . . . . . . . . 4.3.4. Balto-Finnic languages . . . . . . . . . . . . . . . . . 4.3.5. Slovene, Serbian, Croatian . . . . . . . . . . . . . . . 4.4. The Balkan Sprachbund and the south of it . . . . . . . . . . . 4.4.1. Macedonian . . . . . . . . . . . . . . . . . . . . . . . 4.4.2. Bulgarian . . . . . . . . . . . . . . . . . . . . . . . . 4.4.3. Albanian, Rumanian . . . . . . . . . . . . . . . . . . 4.4.4. Greek . . . . . . . . . . . . . . . . . . . . . . . . . . 4.4.5. Article reiteration in Greek and Maltese . . . . . . . . 4.4.6. Indefinite articles . . . . . . . . . . . . . . . . . . . . 4.5. Offshoots . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.6. The eastern margins of Europe: Possessive suffixes . . . . . . . 4.6.1. Uralic . . . . . . . . . . . . . . . . . . . . . . . . . . 4.6.2. Komi . . . . . . . . . . . . . . . . . . . . . . . . . . 4.6.3. 2nd person and 3rd person possessive suffixes . . . . . 4.6.4. Possessive suffixes as emerging articles . . . . . . . . 4.6.5. Indefinite articles in Uralic . . . . . . . . . . . . . . . 4.6.6. Turkic . . . . . . . . . . . . . . . . . . . . . . . . . . 4.6.7. Adverbs and adjectives of place and time . . . . . . . 4.6.8. Indefinite articles in Turkic . . . . . . . . . . . . . . . 4.7. Excursus 2: The possessive suffix 3 SG in Turkish – a definite article? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.7.1. Agreement . . . . . . . . . . . . . . . . . . . . . . . 4.7.2. Relational possessive . . . . . . . . . . . . . . . . . . 4.7.3. Compound marker . . . . . . . . . . . . . . . . . . . 4.7.4. Domains of use of possessive suffixes . . . . . . . . .

. . . . . . . . . .

557 557 558 559 560 560 560 562 562 564

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

565 570 570 571 573 575 576 577 577 578 579 582 583 584 584 585 585 586 587 588 591 591 592 592

. . . . .

594 594 595 596 596

xxi

Contents detailed

4.7.5. Categorial status of possessive suffixes 5. Conclusions . . . . . . . . . . . . . . . . . . . . . . . 5.1. Functions of definite articles . . . . . . . . . . . 5.2. Continua of development of articles . . . . . . . 5.2.1. Continuum 1: Indefinite articles . . . . 5.2.2. Continuum 2: Definite articles . . . . . 5.2.3. Continuum 3: Possessive suffixes . . . Notes . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

597 598 598 599 600 600 601 601

Indexes Subject index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 615 Language index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 619 Author index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 625

Contributors Giuliano Bernini Università di Bergamo Facoltà di Lingue e letterature straniere Piazza Rosate 2 24129 Bergamo Italy E-mail: [email protected] A. Machtelt Bolkestein passed away on October 21st, 2001, in Amsterdam, The Netherlands. Alan Cruttenden Professor emeritus of Phonetics, University of Manchester Fellow of the Phonetics Laboratory, University of Oxford Postal Mail: 31 Picklers Hill Abingdon Oxon OX14 2BB Great Britain E-mail: [email protected] M. M. Jocelyne Fernandez-Vest Linguistique Générale et Ouralienne E. P. H. E. – Sorbonne Esc. E, 1er étage 45-47 rue des Ecoles 75005 Paris France E-mail: [email protected] Yaron Matras School of Languages, Linguistics and Cultures University of Manchester Oxford Road Manchester M13 9PL Great Britain E-mail: [email protected]

xxiv

Contributors

Elena Maslova 2000 Walnut Ave, J307 Fremont, CA 94538 USA E-mail: [email protected] Jim Miller Department of Applied Language Studies and Linguistics University of Auckland Private Bag 92019 Auckland New Zealand E-mail: [email protected] Hans-Jürgen Sasse Institut für Linguistik Abt. Allgemeine Sprachwissenschaft Universität zu Köln 50923 Köln Germany E-mail: [email protected] Christoph Schroeder Istanbul Bilgi University Faculty of Science and Letters Ku¸stepe Campus ˙Inönü Cad. 28 34387 Si¸ ¸ sli/ Istanbul Turkey E-mail: [email protected] Rosanna Sornicola Dipartimento di Filologia Moderna Universita’ di Napoli Federico II Via Porta di Massa 1 80133 Napoli Italy E-mail: [email protected]

I.

General issue

Giuliano Bernini

Introduction

The eight contributions comprised in this volume of the EUROTYP series are devoted to the investigation of single areas in the field of the pragmatic organization of discourse in the languages of Europe. The contributions are grouped into three main parts, according to the issues addressed (general vs. particular) and to the approach adopted with respect to the dichotomy of function and form. In Chapter 1 (“Spoken and written language”) of Part I, which is devoted to “General issues,” Jim Miller and Jocelyne M. M. Fernandez-Vest investigate the different organization patterns of discourse resulting from the spoken vs. written medium of the transmission of verbal messages. The diamesic dimension is actually a continuum where different text types or genres distribute between the extreme points of impromptu speech on the “spoken” side of the continuum and high literature on the “written” side of the continuum in a non-clear-cut, but rather gradient way. The different factors conditioning language production along the diamesic dimension have consequences for the theoretical status of traditional units of analysis and for typology. As for traditional units of analysis, sentence is shown to be a low-level discourse unit mainly of written language; along the spoken side of the diamesic dimension and, most of all, at the endpoint of spontaneous spoken language, clause appears to be a more relevant unit of discourse analysis along with phrase, allowing a better understanding of the (apparent) fragmentary structure of spoken discourse. As for typology, different syntactic patterns appear to characterize the expression of the same function in spoken and written language, as in the case of conditional and relative clauses, among others. Occurrence in different languages and persistence over time of the distribution of different constructions for the same function makes the diamesic dimension a relevant parameter in the analysis of discourse organization in a typological approach. Part II (“From function to form”) comprises four contributions investigating in a problematic perspective some of the functions traditionally considered in studies of the pragmatic organization of discourse and the range of expressions found for them in European languages. The first function considered is that of topic, dealt with by Elena S. Maslova and Giuliano Bernini (“Sentence topics in the languages of Europe and beyond”). The contributors account in a unified fashion for the internal and crosslinguistic variation found for topic expressions in different languages and often thought to challenge the

4

Giuliano Bernini

validity of a relatively vague notion as that of topic as “what the sentence is about.” The topic status of a referent is claimed to be an aspect of the meaning of the sentence, rather than of the “packaging” of the sentence meaning, and to relate to the speaker’s perception and construal of reality. Different topic constructions share the function of opposing a referring expression to the rest of the sentence in a more or less explicit way, excluding it from the scope of the illocutionary operator (assertion or others). They may be subsumed under three major templates grounded in the relation between: the event described in the sentence and an entity related to one of its participants (as in so-called hanging topic constructions), the event described in the sentence and its spatial or temporal location (as in constituent fronting), and the event described in the sentence and its primary participant (as in subject-changing constructions). Different degrees of conventionalization of the three templates account for topic and subject prominence across languages, European languages being notoriously subject prominent. In the discourse organization within individual languages, the different templates are associated with low degrees of activation of the topic referent or with topic-comment relations not inferable from the context. Prominence given to some constituents and the information they carry in the unfolding of discourse is the key concept to all definitions of focus, critically surveyed by Jim Miller in the chapter on “Focus in the languages of Europe.” Prominence is therefore taken as the function allowing typological investigation in the broadest perspective, taking into account the various reasons for which constituents are made salient, be it introduction of new referents, contrast, or “exhaustive listing.” In this data-driven contribution, major attention is given to the grammatical means used for highlighting constituents on the basis of a sample of data drawn from various European languages, consisting of the oral instructions produced by two interlocutors in controlled conditions (the so-called “map task dialogues”), the responses to a questionnaire devised for this matter, and published corpora of spoken language. Written data are also considered. Major means found in the highlighting of constituents for different functions are clefts, particles, and rearrangement of word order. These means distribute differently on the diamesic dimension in some languages and furthermore, in a crosslinguistic perspective, they appear to concentrate in some areas of Europe, as in the case of clefts, characterizing Indo-European languages in the West of Europe (notably English), but unfound in Finno-Ugric and Turkic languages in the East of Europe. Cohesion in discourse is explored by Yaron Matras and A. Machtelt Bolkestein (“Deixis and anaphora”) in a selection of different languages with respect to extratextual and intratextual reference, including anaphora in the proper sense. On the one hand, extratextual reference is distinguished by factors such as distance/proximity and visibility, discussed at length in the literature. On the other hand, factors relevant for the choice of means of expression for intertextual reference are the type of referent referred to (participant or text segment, such as the content of a clause) and the

Introduction

5

degree of accessibility of the antecedent, mainly computed in terms of its distance from the anaphoric resumption, its syntactic status, and the potential competition with other referents. Distribution of deictic and anaphoric pronouns for intersentential anaphora in actual language use appears to result from the interplay of these factors: some favor, rather than oblige, the choice of a particular means of expression among the available ones and their effect may be better described in probabilistic terms on the basis of statistical considerations. Languages seem to differ considerably in the specific weight attributed to one or more factors in the selection of one form rather than another. Furthermore, pronouns used for extratextual reference appear also to be used for intratextual reference, but the relation between the two types of usage need further research in order to be accounted for in a straightforward way. A particularly sensitive place in discourse is where new entities are introduced or new events are announced. Utterances in which these general functions are expressed are often said to be “all new.” Within a philosophical tradition going back to the work of Franz Brentano and Anton Marty in the late nineteenth century, this kind of utterance is said to convey a particular kind of judgment called thetic, that is, a judgment by which one affirms/negates what is being represented as a whole situation. In the last chapter of Part II (“Theticity”), Hans-Jürgen Sasse investigates five construction types for thetic utterances found in European languages and compares them from a variety of perspectives: text frequency, polysemy, restrictions imposed upon them by different languages, discourse functions, and lexical semantics. Theticity appears to be a crosslinguistically comparable phenomenon, although theticity itself is denied the status of a category, being rather a conglomeration of similar presuppositional/assertional conditions correlating with similar semantic areas. Furthermore, the thetic-categorical distinction appears not to be straightforward. Two of the comparable constructions found for thetic utterances – one with accented subject and one with verb-subject order – predominate and distribute over two larger areas covering the North-West and the South-East of Europe, respectively. Part III (“From form to function”) comprises three contributions which investigate discourse functions of particular prosodic and (morpho)syntactic features. Prosody is investigated by Alan Cruttenden (“The de-accenting of given information: a cognitive universal?”), with particular regard to discourse continuity and the intonational correlates of salience/non-salience of the pieces of information conveyed in a message. The supposed universality of the de-accenting of given information is investigated crosslinguistically by means of a repetition test in order to guarantee data comparability. In different setting-response dialogue types, the same lexical item is repeated in the setting and in the response, establishing a sequence of new-given information in the particular context of each test dialogue type. Test subjects for different languages were asked to read the translations of the original English setting-response pairs into their language and the data so obtained were evaluated by independent analysts. Prosodic treatment of the given items in differ-

6

Giuliano Bernini

ent languages in terms of optional and/or obligatory de-accenting and re-accenting appears to be influenced by the interaction of general discoursal and grammatical factors with language-specific variation. De-accenting and re-accenting are favored in most dialogue types in some languages (e.g., German and Spanish, respectively), while other languages do not show a clear tendency towards the one or the other kind of prosodic treatment (e.g.. Italian and Swedish). As a consequence, de-accenting of given information cannot be claimed to be a simple cognitive universal. An integrated view of the interplay of the syntactic and pragmatic levels of analysis with respect to basic constituent order and its typology is aimed at by Rosanna Sornicola (“Interaction of syntactic and pragmatic factors on basic word order in the languages of Europe”). Word order types found in the languages of Europe are considered under different perspectives: the interplay of position, syntactic function, and pragmatic function with respect to sentences with two arguments and one argument; the effect of the particular cases of verb and all-sentence focus; the influence of semantic features (animacy, definiteness, and referentiality of nominal arguments) and of the given-new dichotomy as a textual feature. The thorough survey of word order patterns of European languages according to these perspectives allows the establishment of a set of principles which set a network of conditions regulating the organization of word order patterns. The principles constrain the organization of the “sentence space,” defined in terms of the relationship between constituents, constituent position, and the domain that contains them, barring, for example, the possibility for a constituent to occur in certain positions, even in marked constructions, as in the case of the space to the right of V in SOV languages. Particular attention is devoted to the interaction between constituency and the assignment of the focus function to O, that is, the constituent with the greater depth of embedding, discussed in detail with respect to the preverbal position of SOV languages, which could also result from the effect of semantic (e.g., animacy) or morphological (agglutination) features. This contribution opens a new perspective of typological relevance for the comparison of different word order patterns found in European (and non-European) languages. Articles as a morphological device primarily devoted to the expression of definiteness and indefiniteness, that is, to signaling the referents’ identifiability, are considered by Christoph Schroeder in the last chapter of Part III (“Articles and article systems in some areas of Europe”). The chapter surveys the articles and their systems found in languages of Northern Europe, of South-East Europe, and the Eastern margins of Europe in an areal perspective, taking into consideration the languages spoken in the interjacent areas with respect to other means used for the expression of (in)definiteness. The chapter contributes to the understanding of article systems that are significantly different from those known from the languages of Western Europe. Inventories with more than one series of definite articles serve to encode different types of reference. These might be dependent on whether the referent has already

Introduction

7

been introduced in the actual discourse or is unique in a certain situational context, as, for example, in some Low German dialects, or else express specific distinctions with regard to the location of the referent in terms of restricted or wider situational contexts, as in Macedonian. The latter functional pattern seems to characterize some systems of articles deriving from possessive suffixes, as in Komi. In the areas of Europe considered in this contribution, the functional ranges of article uses reflect three continua of grammaticalization: from the numeral for ‘one’ to the indefinite article, from a demonstrative pronoun to a definite article, and from possessive suffixes of 2nd and 3rd persons to a definite article. All of the contributions share a common functional-typological background and take into consideration the main factor contributing to the organization of linguistic material in discourse, that is, the opposition between spoken and written registers along the diamesic dimension, as discussed in Chapter 1. The diamesic dimension may play a crucial role in typological investigation of discourse organization. As, for example, Rosanna Sornicola points out in Chapter 7, standardized written registers of some European languages, for instance, German, may have been influenced by the literary tradition of classical languages and may show some discrepancies in word order with respect to spoken registers. Therefore, a careful treatment of these discrepancies is required in typology in order to arrive at reliable generalizations. One major area of contention in the study of the pragmatic organization of discourse is represented by the definition of the relevant functions and of the terms adopted to refer to them. The discussions carried out by the EUROTYP group devoted to the “Pragmatic organization of discourse” in five years of common research have not been able to settle the questions pertaining to the definition of functions and the use of a common terminology. Therefore, it seemed wiser to avoid any kind of tentative standardization of terminology across the chapters of this volume. Some chapters discuss in a problematic way the major functions used in research on pragmatic organization of discourse, trying to arrive at a definition which reflects the author’s own vision of the matter: cf. Chapter 2 on topic, Chapter 3 on focus, and Chapter 4 on theticity. In the remnant chapters, the authors define the way in which they use the controversial terms (topic, theme, focus, all new), which may be in accordance, or only partly in accordance, with the terminological use found in other chapters. In any case, the readers won’t be puzzled by non-congruent usage, but rather will be helped in their orientation across a field of research in which consistency among different scholars is hard to arrive at. In this perspective, each chapter contributes different viewpoints on the same aspects of pragmatic organization of discourse, highlighting the difficult areas of descriptive and theoretical relevance and giving a complex picture of this field of study, which reflects its factual and phenomenological complexity. This volume is the result of five years of common research during the EUROTYP program sponsored by the European Science Foundation, and of subsequent years of

8

Giuliano Bernini

elaboration and discussion of earlier versions of the chapters. Some chapters could be submitted earlier, some chapters could be submitted only later on, depending on the amount of exacting work requested by the treatment of their topics. Further unfortunate circumstances hindered the editorial work and the volume is now published with a considerable delay in the EUROTYP series. The members of the group met on ten occasions between 1990 and 1994 in Pavia (Italy), Amsterdam (Netherlands), at “Il Ciocco” near Lucca (Italy), Bremen (Germany), Edinburgh (United Kingdom), San Sebastián (Spain), Bergamo (Italy), Naples (Italy), Le Bischenberg near Strasburg (France), and a last time again in Pavia (Italy). The authors and I are grateful to all the institutions and the people who provided for a comfortable context of the meetings, and, in particular, the European Science Foundation for their organization and financial support. The authors and I are also grateful to the publisher Mouton de Gruyter and to the editors of the EUROTYP series, Bernard Comrie and Georg Bossong, for their support, help, and patience. The editing of the volume has received a fundamental contribution by the scrupulous revision of the co-editor Marcia L. Schwartz. The members of the group on Pragmatic Organization of Discourse were, besides myself, Johannes Bechert (Bremen), A. Machtelt Bolkestein (Amsterdam), Alan Cruttenden (Manchester), M. M. Jocelyne Fernandez-Vest (LACITO, Paris), Elena S. Maslova (Saint Petersburg), Yaron Matras (Manchester), Jim Miller (Edinburgh), Jean Perrot (Paris), Hans-Jürgen Sasse (Cologne), Christoph Schroeder (Bremen), Rosannna Sornicola (Naples), and Barbara Wehr (Mainz). Two members of the original group have left all of us: Johannes Bechert during the five years of common research, and A. Machtelt Bolkestein a few years afterwards. We all remember their scholarship and their contribution to the group work. This book is dedicated to their memory.

Jim Miller and M. M. Jocelyne Fernandez-Vest

Spoken and written language

1. Introduction The serious collection and analysis of spoken language corpuses began in the 1960s, but little attempt has been made to relate the findings to theoretical linguistics.1 In this chapter we argue that spontaneous spoken language differs in many respects from written language. The differences affect the general organization of discourse and all areas of syntax, from the elusiveness of sentences in spontaneous spoken language to the structure of noun phrases. The differences reach out from the core areas of syntax, morphosyntax, and discourse to areas such as historical change, language acquisition, and typology. We first show that the differences are extensive and deep and then use one major construction, relative clauses, to bring out the implications for typology. As will be argued in Section 8, the principal implication is that for a given language, the syntactic structures of spontaneous speech may fit one typological pattern, the syntactic structures of (formal) writing may fit another. Even more important, since this volume is concerned with discourse organization, is the fact that discourse organization is signaled by different devices in spontaneous speech and (formal) writing. The comments on spoken English are based on two corpuses. One is a corpus of 25,000 words of dialogue collected from pairs of informants as they were carrying out a task. The task involved drawing a route on a map, and the dialogues are referred to here as the “map task dialogues.” The second corpus contains 250,000 words of spontaneous conversation and is referred to as the “spontaneous conversations.” Data have also been taken from Macaulay (1991) and Milroy and Milroy (1993). The spoken Russian data is taken from Zemskaja (1973) and Lapteva (1976). The Sami data are from Fernandez-Vest (1987), the French and Finnish data from M. M. Jocelyne Fernandez-Vest’s database.

2. Distinguishing spoken and written language 2.1. Genres The title of this chapter embodies a straightforward distinction between spoken and written language which is untenable. There are no grammatical or discourse differ-

10

Jim Miller and M. M. Jocelyne Fernandez-Vest

ences that correspond with spoken vs. written texts. A first look at different texts quickly reveals that different genres must be recognized for both speech and writing: on the one hand, domestic conversation, dialogue in novels and plays, lectures, news broadcasts, discussion at academic conferences, legal speeches, sermons; on the other, personal letters, diaries, detective novels, poetry, academic monographs, “high” literature. And, of course, further distinctions can be drawn, say, within the category of high literature in English – the language of Jane Austen, the language of Charlotte Bronte, the language of Margaret Forster. Nonetheless, we will see below that spontaneous spoken language, especially conversation but also narratives and task-related dialogues, does have its own syntax and discourse organization. 2.1.1. Biber’s textual dimensions Biber (1988) proposes a more subtle approach. He draws up a list of grammatical constructions and categories: for example, yes-no questions, IT clefts, WHquestions, agentless passives, 3rd person pronouns, adjectives. The list is established on the basis of what occurs with a high frequency in texts. On the basis of its frequency in types of text, each factor is assigned a weighting on a scale from C1 to 1. For instance, yes–no questions are frequent in conversations – what Biber (1988: 395) characterizes as texts produced under conditions of high personal involvement and real-time constraints. They are assigned a weighting of .79. Features that are not salient are assigned a low weighting: for the same type of texts, present tense has a weighting of .42 and a word length of .71. On the basis of the factors and their cooccurrence, Biber (1988: 79–120) establishes various textual dimensions. The summary below gives the properties that cluster at either pole of each dimension. 1. Highly affective interaction and real-time constraints on language production vs. high informational content with time for editing 2. Narration of events vs. expository discourse 3. Explicit reference vs. situation-dependent reference 4. Overt signaling (for persuasive purposes) of the speaker’s/writer’s point of view, or overt assessment of the advisability and likelihood of an event vs. a lack of such signaling 5. Abstract, technical, and formal discourse vs. other types 6. Discourse that is informational but produced under real-time constraints vs. other types of discourse 7. The presence of hedges and qualifications vs. their absence

Spoken and written language

11

2.1.2. Spontaneous spoken language, written language, and education The problem highlighted by Biber is that different text types line up differently with respect to each dimension, and a clear-cut distinction between written and spoken language fails to emerge. In the first paragraph of this chapter we referred to spontaneous spoken language, by which we mean language produced impromptu in relatively relaxed circumstances by a range of speakers – not all with higher education or even with much secondary education – talking about events in their own or other people’s lives. The type of participants is crucial because of the effect of formal education and exposure to written language. Children in a literate society learn informal spoken conversational language as their native tongue. A proportion of children in, for example, Britain, listen to nursery rhymes, then short stories, then longer stories. From the age of five they are taught to read and write – not just to realize linguistic units as marks on paper, but to understand and use the structures and vocabulary of written English. This process lasts from age five to age sixteen at the very least and covers the language of personal narrative, description of scenes, reports of important public events, the language of modern and classical English literature, the description of experiments in science classes, the technical vocabulary and phraseology of mathematics, the sciences, modern studies, and so on. For some speakers, the process of learning to use (as opposed to understand) all these different types of written English continues until they are eighteen, and through their years in higher education. Not everybody is equally capable of combining clauses into well-integrated sentences with subordinate adverbial clauses, participial phrases, and relative clauses introduced by a preposition plus a WH-word. Not everybody possesses the same range of vocabulary and the same skill at using their vocabulary accurately and effectively. (NB “not everybody” can apply to the set of university graduates, to the set of people who have had any kind of further education, to the set of people who have stayed at school till the age of eighteen, or to the entire population of the United Kingdom). Biber’s spontaneous spoken language data comes from the Survey of English Usage and was collected from middle-class, university-educated males. Many of the males were academics and among the segment of the population most affected by formal written language.2 Investigators of spontaneous spoken speech in a number of countries (see references below) have discovered that such language typically has not just less complex structures but different structures; there are constructions that typically occur in spoken language but not in written language and vice versa. Biber maintains, correctly with respect to his data, that a clear-cut distinction between spoken and written language cannot be established. We claim that a clear-cut distinction can be established between typical spontaneous spoken language and typical formal written language: the constructions collected and described by the linguists listed below support that position. At the same time, we can subscribe to Biber’s dimensions,

12

Jim Miller and M. M. Jocelyne Fernandez-Vest

since the constructions that occur in spontaneous spoken language reflect properties such as situation-dependent reference, real-time constraints on language production, affective interaction, and so on. At the time of writing this chapter, Longman had just circulated material advertising their Grammar of Spoken and Written English (Biber et al. 1998). This material incorporates an essential distinction between repertoire and usage. In one quote, Geoffrey Leech declares that spoken and written English have the same repertoire of constructions; in another quote, Douglas Biber declares that usage differs from situation to situation. The two quotes are quite compatible: in principle, Leech can be construed as saying that the same constructions are available to users of English whether speaking or writing. In practice, Biber can be construed as saying that users make different choices depending upon, inter alia, whether they are speaking or writing. Leech’s assertion may be correct in principle but is actually unhelpful; de facto a good number of constructions that occur regularly in (spontaneous) speech do not occur in (formal) writing, and vice versa. Let us emphasize again that what we focus on in this chapter is not just spoken language but spontaneous spoken language – but note the discussion in Section 7 of just how far a planned speech can deviate from the written script if the speaker interacts with the audience rather than just reading out the script as though it were an unalterable text by some other writer. We believe that spontaneous spoken language deserves far more attention in linguistic research. It is what everyone acquires by the light of nature; it is what most people use most of the time; it is the source of much historical change. It is worth remarking here that the study of (spontaneous) spoken language is complicated by political and cultural attitudes. Consider the phenomenon of language variation. Variation that is mainly geographical is acknowledged, and linguists happily talk about, say, the ChiBemba dialect spoken in such and such a village or region. In contrast, variation between standard and nonstandard language brings problems which affect the analysis of spoken language because much of the study of spoken language has been based on nonstandard varieties or on varieties that differ from the standard written one. The difference in treatment is neatly reflected in the barbarous use of “substandard” instead of “nonstandard.” The approach taken here is that all spontaneous spoken language, whether standard or nonstandard, possesses certain syntactic properties: the sorts of structures observed in analyses of nonstandard English have parallels in the spoken standard Russian of professional people studied by Zemskaja (1973) and others.3, 4

Spoken and written language

13

2.2. Previous work on spoken and written language 2.2.1. General grammatical properties of spontaneous spoken language The considerations outlined in Section 2.2 are important when we consider the opposing views that emerged from earlier work on spoken and written language. It is now generally accepted that spoken language, especially spontaneous speech, is very different from written language. The differences spring from various properties of spontaneous spoken language which are listed below – but the pragmatic nature of the properties does not mean that the syntax of spontaneous spoken language is to be treated as resulting from performance error (see Section 4). (i) Spontaneous speech is produced in real time, impromptu and with no opportunity for editing, whereas written language is generally produced with pauses for thought and with much editing (ii) Spontaneous speech is subject to the limitations of short-term memory in both speaker and hearer: it has been said (by the psycholinguist George Miller) that the short-term memory can hold 7C=2 bits of information (iii) Spontaneous speech is typically produced by people talking face-to-face in a particular context (iv) Spontaneous speech, by definition, involves pitch, amplitude, rhythm, and voice quality (v) Spontaneous face-to-face speech is accompanied by gestures, eye-gaze, facial expressions, and body postures, all of which signal information (i) and (ii) are reflected in five linguistic properties. (a) Information is carefully staged, a small quantity of information being assigned to each phrase and clause (b) Spontaneous spoken language typically has far less grammatical subordination than written language and much more coordination or simple parataxis (c) The syntax of spontaneous spoken language is, in general, fragmented and unintegrated; phrases are less complex than phrases of written language; the clausal constructions are less complex (d) The range of vocabulary in spontaneous language is less than in written language (e) A number of constructions occur in spontaneous spoken language but not in written language, and vice versa The simple nature of phrases, the unintegrated nature of the syntax, and the smaller range of vocabulary are all made possible by (iii), since typically a lot of information is shared or present in the situation of utterance and does not need to be articulated. Furthermore, a certain quantity of information can be signaled by the ancillary systems mentioned in (iv) and (v).5

14

Jim Miller and M. M. Jocelyne Fernandez-Vest

2.2.2. General discourse properties of spontaneous spoken language Fernandez (1982: 259–260) draws attention to the use of repetition and inversion of word order in spontaneous speech as strategies of persuasion (and, we might add, attention-holding). She gives a good example of a stylistic scheme combining /repetition + iconical cohesion + circular cohesion/ from Finnish spoken by Samis: (1)

Finnish Alavieskassa ja Taivalkoskella | olen ollut. Alavieska-LOC(intern.) and Taivalkoski-LOC(extern.) I-have been Olen ollut / Nuorgamissa : : : Palokoskella I-have been Nuorgam-LOC(intern.) Palokoski-LOC(extern.) Karigasniemessä | olen ollu Rautuskaijissa | Karigasniemi-LOC(intern.) I-have been Rautuskaidi-LOC(intern.) olen ollu ja joka paikassa | olen ollu I-have been and every place-LOC I-have been

The first clause contains coordinated NPs in the locative case (internal/inessive, or external/adessive), in initial position before the verb. In the second clause the locative case NPs are in final position, simply juxtaposed (listing). The remaining three clauses all have the order locative case NP – verb; the NP in the final clause sums up and generalizes the locative case NPs – joka paikassa ‘every place.’ The boundary between the locative case NPs and the verb is further marked by two nonverbal signals. The first is gestural – a slap of the speaker’s hand on the table. The second is intonational: in each utterance, apart from the second one, which has the structure theme – rheme (I have been / in Nuorgam : : :, rising pitch / falling pitch), the portion to the left of the | carries falling pitch, whereas the portion to the right carries level pitch. The portions with the falling pitch are the rhematic part of the clause, and the portions with the level pitch are analyzed as post-rheme elements or mnemes (Fr. mnémème, see Fernandez-Vest 1994: 197 ff.) Consider the following two English texts. (2)

then he said why was I always trying to change him and I said probably because he’s such an obnoxious thoughtless selfish overbearing selfrighteous hypocritical arrogant loudmouthed misogynist bastard

(3)

Mr. X was tall bald hair to here and a beard he looked like Jesus he liked to think he did he liked to think he was Jesus he was horrible the most horrible man

(2) is taken from a humorous postcard (the words are spoken by one woman to another as they sit at the kitchen table chatting over a glass of wine and a cigarette) and the humor comes precisely from the fact that in such a relaxed setting, most people cannot produce even two or three well-chosen adjectives off-the-cuff, far less a

Spoken and written language

15

string of nine, and even the most fluent speakers are hard pressed to produce complex syntax at the right moment in stressful situations. Most of us suffer from l’esprit de l’escalier. (3) is taken from spontaneous conversation. It illustrates the combination of verbal and nonverbal communication, since hair to here is accompanied by a gesture – Mr. X is bald on top but has hair almost down to his shoulders. The remainder of (3) demonstrates how in spontaneous speech speakers make an assertion and then repeat it, not necessarily exactly but with additional information. Thus, the next clause carries the assertion that Mr. X looked like Jesus, while the following two clauses each carries the qualification that Mr. X liked to think he looked like Jesus. The penultimate clause conveys the judgment that Mr. X is horrible, while the final noun phrase intensifies the judgment; Mr. X is not just horrible but most horrible. (The change of syntax is to be expected; [Mr. X is] most horrible is unlikely to occur outside formal written English. Mr. X is the most horrible man sounds more natural, although the complement noun phrase has the air of a ready-made unit.) 2.2.3. Analyses contra a special status for spontaneous spoken language Some scholars have taken the opposite view that spontaneous spoken language has complex syntax. Poole and Field (1976) found spoken discourse to have a significantly greater degree of subordination, elaboration of syntactic structure, and use of adverbs. Halliday (1989: 76–91) maintained that written language is complex in that highly compact and simple syntactic constructions are loaded with many lexical items. Spoken language is complex in a different way, having intricate syntactic structure with a considerable proportion of subordinate clauses. Lexical items are spread over these subordinate clauses, reducing the lexical density of each one. Halliday illustrates the distinctions via the written sentence The use of this method of control unquestionably leads to safer and faster train running in the most adverse weather conditions and a possible spoken variant ||| If this method of control is used || trains will unquestionably (be able to) run more safely and faster || (even) when the weather conditions are most adverse ||| (Halliday 1989: 79). Unfortunately, various features mark this utterance as written or enunciated by someone who speaks like a book: the impersonal conditional clause, the adverb unquestionably, the noun phrases method of control and weather conditions, and the adjective phrase most adverse. A more plausible utterance would be: If you control the trains this way they’ll definitely run safer and faster – in really bad weather and all. Even Halliday’s (1989: 79) “more natural” spoken version contains sequences such as you can be quite sure that : : :, no matter how bad the weather gets, than they would otherwise, which are quite untypical of spontaneous spoken language. As Beaman (1984: 51) suggests – and cf. Biber’s dimensions – it is important to distinguish between the modality (spoken vs. written) and register (formality). She proposes that what have been treated as differences between spoken and written dis-

16

Jim Miller and M. M. Jocelyne Fernandez-Vest

course may in fact reflect differences in formality or planning time, and to these should be added the property mentioned in Section 2.2, exposure to higher education. Poole and Field gathered their data from undergraduate students in a formal setting, while the data cited by Halliday appears to have been invented by him and unfortunately reflects the complex nature of the spoken language of academics.6

3. Sentences and clauses7 3.1. Introduction The position defended here is that the sentence should be regarded as a low-level discourse unit of written language, that clauses and phrases are units of both spoken and written language, and that, as suggested by Halliday (1989), sequences of clauses in spoken language may form clause complexes, which do not have the structural integrity of sentences. Indeed Sornicola (1981) argues convincingly on the basis of Neapolitan Italian that much spontaneous language does not even have a syntactic structure in which phrases combine into clauses or clauses into integrated clause complexes. Rather, the structure consists of blocks of syntax (phrases) with little or no syntactic linkage and requiring from the listener a larger than usual exercise of inference based on contextual and world knowledge. Sornicola demonstrates that such fragments should be treated as the structures that speakers aim to produce in spontaneous speech, and not as the remnants of clauses that have fallen apart as a result of performance errors.8 Sornicola’s approach seems to deprive formal models of any solid unit for the analysis of interclause relations, reference, and anaphora, but alternative frameworks are now available, albeit relatively undeveloped, in the theories of rhetorical structure and discourse representation.

3.2. System sentences and text sentences We will assume a major distinction between language system and language behavior. The language system consists of the syntactic, morphosyntactic, semantic, phonological, and graphological principles controlling the generation of semantic and syntactic structures, the insertion of lexical items into the syntactic structures, and the realization of the structures as speech or writing. The products of speaking and writing are texts, which may be spontaneous or deliberately elicited by investigators. Hypotheses about particular language systems or the general nature of language systems are based on texts and intuitions. It is essential to distinguish units that can be recognized in texts from units that belong to the hypothesized language system. For instance, when a modern written text in English (or French, German, etc.) is examined, the analyst finds the text divided into units whose initial boundary is signaled

Spoken and written language

17

by a capital letter and whose final boundary is signaled by a full stop. These units, text sentences, are determined by the author of the text – if the text is a personal letter or an essay, the division into text sentences typically remains unchanged once the text has left the author’s hands, but if the text is, say, a book being prepared for publication, a subeditor may question some of the author’s decisions. For example, a particular author may have a subordinate clause of concession or a relative clause constituting a whole text sentence, whereas many copy editors prefer to include at least one main clause in each text sentence. In order to handle the patterns of constituent structure and dependency relations that manifest themselves during the analysis of texts, linguists establish units called sentences. These are what were referred to above as system sentences. System sentences do regularly map on to text sentences in written texts but not always. For instance, restrictive relative clauses are treated in linguistic analyses of English as embedded in noun phrases containing the nouns they modify. This enables the analyst to handle the dependency relation between relative clause and the noun it modifies, to capture what is indicated by distributional evidence – (determiner), noun and relative clause form a single constituent – and to compose the denotation and reference of that single constituent. In modern written English texts, however, it is not unusual to find a text sentence consisting entirely of a relative clause, while the noun it modifies is in another text sentence: The door was opened by a man. Who appeared to be about seven foot tall and six foot wide. Analysts have two approaches to choose from. They can allow their grammar to generate a single system sentence in which the relative clause is embedded inside the oblique object of the main clause, or they can allow their grammar to generate two separate system sentences: one the main clause and the other, the relative clause. The former approach allows dependency relations and compositionality to be handled straightforwardly, but requires the structure to be dis-integrated for mapping onto the surface structure. The latter does not require dis-integration but does call for devices to handle the dependencies and compositional relations that cross the sentence boundary. Strictly speaking, the different units should be clearly kept apart by means of different terms, such as “text sentence,” “system sentence,” “text clause,” “system clause,” etc., as in Lyons (1977). Here, “sentence” and “clause” will be used where it is clear from the context whether the unit belongs to text or to the language system. What text units can be recognized in spoken language? In written language, sentences and clauses (and phrases, paragraphs, etc.) are obvious in any text laid-out according to the conventions of the society in which it was written. Interestingly and importantly, the relevant conventions differ from society to society and from one period of time to another in the same society; the organization of clauses into sentences differs from Jane Austen to A. S. Byatt, and from British newspapers to French newspapers – but text sentences are clearly delimited in all these genres. The question of what units can be recognized in spoken language and are useful for its analysis is

18

Jim Miller and M. M. Jocelyne Fernandez-Vest

not so easily resolved. Some analysts maintain that sentences are not recognizable in spoken language, and others maintain that they are. System sentences are postulated by linguists in order to handle distribution and dependency relations, and must be retained if this goal is to be achieved. The status of system sentences in the language system of written English is not disputed, and is usually regarded as self-evident because of the clear delimitation of text sentences. It is not usually remembered that sentences are learned through the process of reading and writing, and are taught to the majority of language users, whereas clauses are acquired without specific teaching. Children in the early stages of primary school typically produce single-clause sentences and have to acquire the ability (partly by instruction, partly by reading) to combine a number of clauses into a sentence. Written texts are produced, and analyzed, not by the untutored, but by people who have been inducted over a long time into the conventions governing the organization of written texts in their society. The central problem is that it is far from self-evident that the language system of spoken English has sentences, for the simple reason that text sentences are hard to locate in spoken texts. Clauses are easily recognized: even where pauses and a pitch contour with appropriate scope are missing, a given verb and its complements can be picked out. Of course, one reply to the objection is that the system sentences employed by linguists need not correspond to text sentences. Against this, it can be argued that system sentences do not map onto text sentences in spontaneous language, because system sentences are based on the prototype concept of a sentence as containing at least one main clause and possibly other coordinated main clauses and/or subordinated clauses. (That this is indeed the prototype concept is easily verified by examining popular manuals such as Burton [1986], the literature in any generative framework or the training offered to school pupils). It can also be argued that in the language system, the essential locus of both dependency relations and distributional properties is in fact the clause (see Section 3.4). If spoken texts lack sentences, the language system must be analyzed as having clauses combining into clause complexes, as suggested by Halliday (1989). There are two major types of syntactic relationship: embedding and combining. Adverbial clauses only combine, that is, they are not part of any constituent in a matrix clause. Only relative and complement clauses can be embedded, since relative clauses are regularly part of an NP and complement clauses function as arguments to verbs. However, many occurrences of relative clauses cannot be treated as embedded, especially if, in dialogue, they occur in a different turn from the head noun and come from a different speaker. Relative clauses do occur as the sole constituents of sentences even in written English, although such syntactic arrangements might not be considered good style. English is not alone, since the same phenomenon is found from Classical Latin to Modern English. A French example, albeit of a special kind of relative construction, is given in (4b)(iv) and (v) below.9; 10

Spoken and written language

19

3.3. Sentences in spoken texts 3.3.1. Introduction This section briefly surveys the arguments for and against the sentence as a unit in spoken texts and as an analytical unit in accounts of spoken language. The case against has been stated most clearly by Halliday (1989: 66), who argues that the basic unit of syntax is the clause. Clauses occur singly or in complexes, and clause and clause complex are indispensable concepts for the study of both spoken and written syntax. Sentences in written language developed from the desire to mark clause complexes; the initial capital letter of the first word in a clause complex, and the full stop following the final word signal which clauses the writer wants the reader to construe as interconnected. Of course, clauses are also interconnected in spoken language; the difference is that interconnectedness is not signaled by adjacency nor even by the relevant clauses occurring in the same turn (in conversation) or under the same intonation contour (in narrative). A number of researchers recognize the problematic nature of the sentence in spoken language. Quirk et al. (1985: 47) state that the sentence boundaries can be difficult to locate “particularly in spoken language”11 and point out that the question “What counts as a grammatical English sentence?” does not always permit a decisive answer. They deal with the difficulty by avoiding any definition of sentence while continuing to use the term for a unit greater than the clause. Linell (1988: 54) reaffirms the lack of clear-cut sentences in spoken language and adds that talk consists of phrases and clauses loosely related to each other and combining into structures less clear and hierarchical than the structures dealt with in grammar books. Similar points had been made earlier by Brown et al. (1984: 16–18). Brown et al. and Linell are apparently satisfied to work with phrases and clauses, precisely the position adopted here.12 Sentences in spoken language are defended by Chafe and Danielewicz (1987: 94–96). They invoke “prototypical intonation units,” consisting of a single coherent intonation contour, possibly followed by a pause and stretching over a maximum of six words, which often constitute a clause but which may also constitute a phrase or simply a fragment of syntax. Chafe and Danielewicz (1987: 103) further say that speakers appear to produce sentence-final intonation when they judge that they have come to the end of some coherent content sequence. One difficulty with Chafe and Danielewicz’s account is that their sentences correspond more to short paragraphs than to the prototypical written sentence. The intonation contours they describe may encompass one or more main clauses, not conjoined but simply adjacent to each other. Conversely, the same type of intonation contour may encompass a mere phrase.13

20

Jim Miller and M. M. Jocelyne Fernandez-Vest

3.3.2. Intuitions about sentences in spoken language Wackernagel-Jolles (1971: 148–169) demonstrated that speakers do not share intuitions about what counts as a sentence in spoken language. She got groups of thirty to fifty final year undergraduate students at a German university to listen to recordings of narratives by native speakers of German. (The narratives had been prompted by questions from an interviewer). Each text was played through once to allow the students to accustom themselves to the speaker’s voice. They were then given a transcription of the recording, without punctuation. The text skipped to a new line only where there was a change of speaker. The recording was then played through a second time, and the students were asked to draw a line in the text wherever they thought a sentence ended. Agreement as to sentence endings ranged from 13 out of 20 in one text, to 6 out of 29 in another. The former text was the telling of a fairy tale; the latter, a panel-beater recounting his early life and his war experiences. Wackernagel-Jolles (1971: 149) comments that uninterrupted story-telling was especially conducive to clear intonation signals but that no correlation emerged between speed and clarity of pronunciation and degree of agreement. Speakers/writers, who as university students can doubtless organize their own written texts into acceptable sentences, were unanimous about the final boundary for less than half of the sentences in the texts.14 For them the sentence is a relatively fluid unit. 3.3.3. The sentence: A changing concept The view of the sentence as a relatively fluid discourse unit in written language fits with the fact that, in written English, text sentence boundaries vary from one historical period to another, from one genre to another, and from one individual to another. Moreover, text sentences vary from one language to another. In contrast, clauses (and phrases), which are central units of syntax, are not subject to such cultural variation and rhetorical fashion.15 As observed in Section 3.2, the concept of text sentence is not stable across cultural boundaries and can be manipulated to achieve particular stylistic effects. For instance, the French weekly L’Express has a house style that encourages phrases and subordinate clauses of all types to be presented as single text sentences, as exemplified in (4). (4)

a.

(i) (ii)

Ils sont de bonne foi. they are of good faith (i.e. sincere) Comme étaient de bonne foi ces ménagères engueulant as were of good faith these housewives shouting-at les refuzniks the refuseniks

Spoken and written language

b.

(i)

Certains ont invité les contestataires à certain ones invited the objectors to (ii) se plaindre auprès de Raissa Gorbatchev. to-complain to Raisa Gorbachev (iii) Avec quelques commentaires grossiers sur with some comments coarse on (iv) du secrétaire général. of-the secretary general (v) Ce qui n’ était pas très difficile. which not was not very difficult

21

aller go

la femme the wife

(4a)(ii) contains a subordinate adverbial clause of comparison constituting a complete sentence, namely Comme étaient de bonne foi ces ménagères engueulant les refuzniks. (4b)(iii) has a prepositional phrase as a separate sentence, while (4b)(v) has a relative clause as a separate sentence. The unusual segmentation, a strategic chunking of the information which imitates the rhythm of conversation, creates suspense and, as the writer probably hoped, the appropriate reaction from readers. The Russian weekly Argumenty i Fakty offers the examples in (5). (5)

a.

b.

Komandir soznatel’no idet na risk. (the)-captain consciously goes on risk ‘The captain consciously takes a risk.’ Nadejas’, cˇ to peregruzki ne budet. hoping, that overloading not will-be

In (5b) nadejas’ is a nonfinite verb form, a gerund. Equally interesting examples are in (6). (6)

a.

b.

Tol’ko za poslednij god Only in (the) last year po SNG sokratilsja in (the) CIS has fallen Na Ukraine i v In (the) Ukraine and in na polovinu. by (a) half

ob”em (the) volume na 30 %. by 30 % Rossii – na Russia – by

aviaperevozok v obšˇcem of air-journeys in general

tret’, v Tadžikistane – (a) third in Tadžikistane –

The interesting point about (6b) is that it is a complete sentence consisting of a gapping construction, and the constituent required in order to interpret the gap, sokratilsja, is in the previous sentence. (6b) is relevant not only as an example of a verbless sequence functioning as a sentence in a text, but as an example of a dependency carrying over from one sentence to another. This property means that in a generative analysis, the gap must be handled by a mechanism that can operate across

22

Jim Miller and M. M. Jocelyne Fernandez-Vest

sentence boundaries as well as across clause boundaries, and the example is another piece of evidence in support of the view that the abandonment of sentences as analytical units with respect to spoken language does not create new problems but requires mechanisms already required by existing problems. Further problems are posed by dialogue. Consider the examples in (7)–(8) from the map task dialogues. (7)

B:

right if you go from the front giraffe’s foot about hold on let me see if you go down about straight down about 6 cms you find the waterhole and it’s a big hole : : : with reeds round the side of it : : : and animals drinking out of it and it’s about it’s a an oval hole it’s about 2 cms wide north to south and from the side to side it’s about – 3 cms wide : : :

(8)

A: B: A:

you go down to the bridge uhuh to the left of the swamp? to the left of the swamp – taking a gentle curve southwest

(7) illustrates how in spontaneous spoken language information is carefully staged, in the sense of being spread out over different clauses. Most of the clauses are simple clauses and are simple in structure, though one clause has two prepositional phrases with a participial phrase inside one of the latter – with reeds round the side of it and (with) animals drinking out of it. It would be possible to gather the clauses into sentences, but various possibilities are open. For example, we might decide to have You find the waterhole and it’s a big hole : : : [and it’s about]. It’s an oval hole. It’s about 2 cms wide north to south. And from : : : Another possible version is You find the waterhole. And it’s a big hole : : : It’s about 2 cms wide north to south and from : : :. The basic difficulty is that in collecting the clauses into sentences, we rely on our ability to recognize clauses and on our knowledge of the stylistic conventions for written dialogue. As in (1), the intonation and pause boundaries do not coincide with the possible sentence boundaries, and to add to the difficulties, the prepositional phrases with reeds round the side of it and (with) animals drinking out of it are separated from the initial part of the clause and from each other by a long pause. It is in fact unclear whether these chunks should be analyzed as combining into a single clause. The analyst can combine the clauses into sentences, but the combining process is arbitrary and the sentences would not contribute to the analysis of the data as a coherent text. Coherence relations (say, as part of a discourse representation theory) must apply to clauses and indeed phrases, and sentences are not necessary.

Spoken and written language

23

Similarly, (7) could have been written as the compact, dense, syntactically integrated piece of prose in (9), but the characteristics of (9) cannot be invoked as criteria in the analysis of (7). It would be rather like taking a piece of written language, rewriting it, analyzing the rewritten piece, and presenting the analysis as pertaining to the original passage! (9)

It’s a big oval waterhole about 2 cms wide north to south and about 3 cms wide from side to side, surrounded by reeds and with animals drinking out of it.

(8) exemplifies another relationship that cannot reasonably be analyzed by invoking a single sentence. The free participial phrase taking a gentle curve southwest, modifies the clause produced in a previous turn by the same speaker. Participial phrases are discourse-dependent in the sense that the listener cannot interpret them without reference to a previous piece of text; at the very least, a subject has to be found for the participle itself. The nearest candidate for the subject is in the first line of (8), but this is not a reason for analyzing the participial phrase as belonging to one and the same sentence as you go down to the bridge. Note that speaker A was not interrupted in the process of producing a single sentence. You go down to the bridge is a completed utterance, with appropriate intonation. B signals acceptance of the instruction with uhuh, looks at the map and realizes that he needs more information: to the left of the swamp? Speaker A produces the participial phrase in response to speaker B’s question. The development of sentence structure in written language is discussed, for example, by Palmer and Guiraud. Palmer (1954: 119) remarks that complex sentences in written Latin prose were consciously developed by generations of writers and that the resulting body of rhetorical conventions had to be taught. Once the vernacular Romance languages had broken away from Latin, the organization of clauses into sentences had to be established for each vernacular Romance language as it began to be used as a vehicle for prose literature. Guiraud (1963: 113) observes that the Old French literary language was very close to the spoken language, having an essentially paratactic organization of clauses into larger units. Such a syntax “n’a jamais eu l’entraînement ou la pratique qui l’auraient pliée à l’expression d’une pensée élaborée; elle ignore l’articulation logique de la démonstration scientifique ou les méandres de l’argumentation philosophique” [has never had the training or practice that would have formed it to the expression of elaborated thought; it is unaware of the logical structure of scientific argument or the meanderings of philosophical discussion, translation by Jim Miller]. It is significant that Guiraud mentions the uses to which language is put by literate human beings; French syntax did not develop a complex, hypotactic organization of clauses by some mysterious process but through the conscious efforts of certain literate people to convert French into an instrument suitable for the purposes served by Classical Latin.16

24

Jim Miller and M. M. Jocelyne Fernandez-Vest

It must be emphasized that neither Palmer nor Guiraud suggests that the development of subordinate clauses was subsequent to the development of written Latin and written French. Rather, they allude to the organization of several or many clauses into a sentence and the way in which the conventions governing this organization were developed by the users of a given written language. They also allude to the special development of language by scientists for the accurate description of scientific data, experiments, and theories, which is quite in accord with the comments by Ong (1982) on the role of the Royal Society in Britain in the late seventeenth century in encouraging the emergence of a special scientific language and the logical presentation of hypotheses, data, and conclusions.

3.4. Sentences and syntactic analysis 3.4.1. The clause as the central unit of syntactic analysis We turn now to sentence and clause in linguistic analysis. The burden of the preceding discussion is that the sentence is not a unit that can be recognized in spoken texts or applied in their analysis. In contrast, the sentence is a prominent unit in written texts and requires a corresponding analytical unit. However, there is evidence that the clause should be taken as the major locus of distributional and dependency relations and not the (system) sentence. Crystal (1987: 94) provides a concise rendering of the definition of “sentence” provided by Bloomfield (1935: 170): “a sentence is the largest unit to which syntactic rules apply.” Interestingly, Bloomfield’s sentences each consist of a single finite clause. Bloomfield does treat the problem of two or more juxtaposed main clauses without pauses or intonation break by invoking a set of pitch phonemes, but this analysis is decisively countered by Matthews (1981: 30–34), on the grounds that intonation is continuous, the phonemic principle of sameness vs. distinctness does not apply, and there are no rules governing parataxis.17 In any case, the syntactic units (in spoken or written language) affected by the rules of distribution and dependency relations are the phrase and the clause. The clause is the locus of the densest dependency and distributional properties, although a few dependency relations cross clause boundaries, and, in written language, a few dependency relations cross sentence boundaries. That dependency relations cross clause boundaries could be interpreted as supporting the sentence as an analytical unit even in spoken language, but this is counteracted by the fact that dependency relations cross text sentence boundaries in written language. Because dependency relations cross text sentence boundaries, a complete grammatical theory must have a mechanism for specifying such dependencies, and whatever the mechanism is, it will undoubtedly be able to specify dependencies from clause to clause when the clauses are gathered, not into a sentence, but into a text.

Spoken and written language

25

Cross-clause dependencies are frequent in Classical Greek, as exemplified in (10)– (14). Let us note first that, like Russian, the densest network of dependencies is inside the Classical Greek clause. This is demonstrated in (10a), where the verb akoúousi assigns nominative case to its subject noun and accusative case to its direct object noun. (In contrast, in (10b) the verb khrâtai assigns dative case to its object noun, and the very assignment of dative case raises the question whether that verb takes a direct object or an oblique object.) In turn the nouns spread case, number, and gender to any dependent articles and adjectives: ándres assigns nominative, plural, and masculine to hoi and kakoí; and lógous spreads accusative, plural, and masculine to toús. Gunaikós (of the woman) is a feminine noun in the genitive case because of its relationship to lógous. Gunaikós spreads genitive, singular, and feminine to tês and sofês. (10) a.

b.

hoi kakoì ándres ouk the-NOM.PL.M evil-NOM.PL.M men-NOM.PL.M not akoúousi toùs tês sofês listen-to-PRES.3PL the-ACC.PL.M the-GEN.SG.F wise-GEN.SG.F gunaikòs lógous woman-GEN.SG.F word-ACC.PL.M ‘The evil men are not listening to the words of the wise woman.’ h¯e gun´¯e khrâtai toîs the-NOM.SG.F woman-NOM.SG.F use-PRES.3SG the-DAT.PL biblíois books-DAT.PL ‘The woman is using the books.’

As in English, complement-taking verbs in Classical Greek control the complementizer in the complement clause. Verbs of saying take hóti ‘that,’ as in (11), verbs of movement take hína ‘in order to,’ as in (12), verbs of inquiring take ei, as in (13), and certain specific verbs take hóp¯os, as in (14). (11) a. b.

(12) a.

b.

légei he/she-says eîpen he/she-said

hóti that hóti that

gráfei he/she-is-writing gráfoi he/she-was-writing

érkhetai hína íd¯e is coming in-order-that sees ‘He/she is coming to see.’ élthen hína ídoi came in-order-that sees ‘He/she came to see.’

26

Jim Miller and M. M. Jocelyne Fernandez-Vest

(13) a.

b.

(14) a.

b.

punthánetai ei akoúousi toùs lógous asks if hear the words ‘He/she is asking if they hear the words.’ epútheto ei akoúoien toùs lógous asked if hear the words ‘He/she asked if they heard the words.’ spoudázei hóp¯os akoúsetai toùs lógous hurries to hear the words ‘He/she is hurrying to hear the words.’ espoúdase hóp¯os akoúsetai toùs lógous hurried to hear the words ‘He/she hurried to hear the words.’

Note that although hóp¯os in (14a) and (14b) has been translated with ‘to,’ the complement clause is finite, akoúsetai being third person singular future. In addition to verbs selecting complementizers, there is another dependency crossing the clause boundary. When the verb in the matrix clause is past tense, the verb in the complement clause is in the optative mood. When the verb in the matrix clause is present tense, the verb in the complement clause is either indicative, after hóti in (11a) and ei in (13a), or subjunctive, after hína in (12a). When the verb in the matrix clause is aorist, the verb in the complement clause is in the optative mood.18 The exception to these changes in mood is hóp¯os in (14a) and (14b), which requires the verb to be future tense. In the last case it is hóp¯os and not the verb in the matrix clause that governs the occurrence of future tense. Dependencies do not cross clause boundaries into adverbial clauses. Certain combinations of adverbial clause and main clause appear to involve cross-clause dependencies, such as the rules in Classical Greek governing clause combinations expressing fulfilled or unfulfilled conditions as in (15). (15) a. b.

ei if ei if

toûto this toûto this

epoíoun, e¯ díkoun they-were-doing they-were-wrong épraxen, e¯ dík¯esan án they-had-done they-would-have-been-wrong particle

The English copula + adjective structure corresponds to a single verb in Greek. (15a) expresses a fulfilled condition; the conditional clause contains an imperfect verb, epoíoun, as does the main clause, e¯ díkoun. (15b) expresses a remote, unfulfilled condition. The conditional clause contains an aorist form, épraxen, and the main clause contains an aorist form with the particle án. Such examples, however (both the Classical Greek ones and their English equivalents), are not instances of dependencies crossing from clause to another. The syntactic constraints affect both the main and adverbial clauses, and the dependencies appear to be associated with the entire combination, rather than flowing from the main clause to the adverbial clause.

Spoken and written language

27

The above examples of cross-clause dependencies do not vitiate the proposition that the clause is the site of the densest network of dependencies. In each of the above examples, at most two dependencies cross the clause boundary, the selection of complementizer by the verb and the selection of mood in the complement clause. Inside each clause is a much greater number of dependencies. The verb controls the type of constituents it requires; some verbs allow two NPs, others allow three, and yet others allow only one; some verbs allow adjective phrases or PPs. In a given clause the verb assigns case to the dependent nouns and controls the choice of preposition in the PPs. Inside the phrasal constituents the head, N, A, or P, assigns case to its dependent constituents, and a head N also assigns gender and number. And there may be further PPs inside the NP and AP. Not all these dependencies flow directly from the verb, but they are all sited inside a given clause. 3.4.2. Dependencies crossing sentence boundaries To close this discussion of dependencies, let us consider the Russian text in (16), exemplifying dependencies crossing text sentence boundaries. (16) a. b. c. d.

Ètot portnoj byl krasivo starejušˇcij mužˇcina this tailor was handsomely growing-old man šest’ veˇcerov v nedelju stojal on za stolom, six evenings in week stood he at table, rezal šil proglažival švy utjugom. cut sewed smoothed seams with-iron. Zarabatyval den’gi. Voskresen’e provodil na ippodrome earned money. Sunday spent at racecourse

In (16) the first sentence has the full masculine singular NP ètot portnoj; the second sentence has the masculine singular pronoun on; and the third and fourth sentences have the subjectless masculine singular verbs zarabatyval and provodil. The properties “masculine” and “singular” are projected by ètot portnoj into the pronoun in the second sentence and then into the verb forms in the third and fourth sentences. 3.4.3. Sentences, clause, and distribution With respect to distribution, it is equally obvious that the classic distributional criteria for constituent structure apply within clauses rather than sentences. For example, in a recent introduction to transformational grammar (Radford 1988: 69–75), the vast bulk of the distributional evidence relates to single main clauses. Where there is more than one clause, one reason is that the additional material is needed to provide a convincing linguistic context; for example, in Down the hill John ran, as fast as he could the adverbial clause of manner, as fast as he could, lends naturalness to the

28

Jim Miller and M. M. Jocelyne Fernandez-Vest

fronted prepositional phrase down the hill. A second reason is that the extra clause is a relative clause or a complement clause, that is, clauses that are embedded inside arguments of the verb in the main clause. One example is He explained to her all the terrible problems that he had encountered, where the relative clause gives the necessary weight to the final noun phrase. The only clauses that have distribution inside a unit bigger than the clause are adverbial clauses. In written English and in relatively formal spoken English, adverbial clauses of time and reason, for example, can precede or follow the main clause with which they combine, but in informal spoken English, they tend to follow the main clause. That is, since their distribution even in written English is limited, they are no more than mild exceptions to the rule that the clause is the main focus of distributional properties. In any case, it is not clear that even adverbial clauses can be moved inside a sentence in written English. The difficulty is that not all subordinate clauses are equally subordinate, where subordination is measured in terms of possible constructions and word orders. Main clauses permit a large range of constructions – declarative, interrogative, imperative, tag questions – and a large range of word orders, whereas subordinate clauses vary in the extent to which they allow constructions other than declarative and word orders other than subject – verb – direct object. Consider the examples in (17). (17) a. b.

Because Aunt Norris came into the room, Fanny stopped speaking. Fanny stopped speaking because Aunt Norris came into the room.

In (17a) the adverbial clause of reason because Aunt Norris came into the room is at the front of the sentence and has limited structural possibilities. Into the room, for example, cannot be moved to the front of the clause, and the subject NP and the verb cannot be transposed, but these changes can be carried out in the because clause in (17b). Compare (18a) and (18b). (18) a. *Because into the room came Aunt Norris, Fanny stopped speaking. b. Fanny stopped speaking, because into the room came Aunt Norris. It has been suggested that because clauses have a different discourse function in sentence-initial and sentence-final positions (Chafe 1984). In sentence-initial position, because clauses (indeed, adverbial clauses in general) function as a guide to information flow, whereas in sentence-final position, adverbial clauses simply add something to the assertion conveyed by the main clause. In sequences of main clause – because clause, the two clauses almost function like coordinate clauses. The because clause may relate to one particular constituent in the main clause or may not relate directly to the main clause at all. That is, the un-subordinate nature of such because clauses, as evidenced by their syntactic flexibility, is accompanied by semantic flexibility.19 For instance, in Fiona isn’t coming to work today because her husband phoned up to say she was ill, the because clause presents, not the reason

Spoken and written language

29

for Fiona’s absence, but the reason for the speaker being able to state that Fiona is not coming in to work. For some speakers, adverbial clauses of reason following the main clause can even contain interrogative structures: for example, I’m not going to the party because who’s going to be there? Complement clauses are prima facie better examples of distribution inside the sentence, given that there appears to be no major difference in meaning between a sentence such as (19a), with a complement clause in sentence-initial position, and (19b), with the complement clause in sentence-final position. (19) a. b.

That the enemy was approaching the town apparently did not worry the inhabitants. It apparently did not worry the inhabitants that the enemy was approaching the town.

For written English it is indeed correct that complement clauses can occur at the beginning or end of sentences, but in spontaneous spoken English examples like (19a) are practically unknown. There are none in the corpuses of conversation and map task dialogues, and they are very rare even in formal spoken language such as is heard in serious discussion programs on radio and television. That is, in spontaneous spoken English, complement clauses are not mobile but fixed. To sum up, there is very little evidence to support either text sentences or system sentences in spontaneous spoken language. Planned or semi-planned spoken language is different, but is typically heavily influenced by the units and organization of written language. Much of the language system of a given language is medium-independent, but some is dependent, most obviously the complex syntactic constructions and the vocabulary that are typical of written language but not spoken language. Equally, there are constructions and vocabulary that occur in spontaneous spoken language but not in written language. The system differences can be kept to a minimum by appealing to the notion of discourse rules specific to a given medium. The discourse rules for written language map one or more clauses into sentences where appropriate; the discourse rules for spontaneous spoken language do not.

4. “Syntactic fragments,” competence, and performance Sornicola (1981) and Enkvist (1982) argue strongly that even apparently fragmented syntax should be treated on its own terms and not as the degraded realization of an ideal clause or clause complex, particularly as it is not always obvious what the ideal structure might be. Consider (20), from spontaneous spoken English. (20)

A: whose idea // was it B1: Charlie Richardson’s C: uhuh

30

Jim Miller and M. M. Jocelyne Fernandez-Vest

B2: they got one of the teachers that we always play jokes on / one of the young women / they got her to write it The sequence one of the young women is spoken on a much lower pitch than the surrounding material, and there is a clear tonic on we and write in B2. What syntactic structure in her competence was the speaker aiming at? It might have been they got one of the teachers that we always play jokes on to write it, but the inserted explanatory material one of the young women has disturbed the flow of the surface syntax. An alternative is that they got one of the teachers that we always play jokes on corresponds to one originally complete syntactic structure and that they got her to write it corresponds to another complete piece of syntax. That is, although the inserted material can be seen as interrupting the syntactic performance, it is equally possible that the text clauses correspond to the original abstract syntactic structures that the speaker had in mind. The same difficulty is posed by spontaneous spoken Russian. Consider (21), from Zemskaja (1973: 27). (21) a.

b.

moloko raznosit / ne prixodila ešˇcë? milk she-delivers / not she-came yet ‘The woman who delivers the milk, has she not come yet?’ u okna ležala / kapriznaja oˇcen’ at window she-lay / moody very ‘The woman in the bed by the window was very moody.’

The “missing” syntax in (21) is not just a noun but a noun and a relative pronoun. By “missing” is meant that in writing (21a) and (21b) have to be converted to (22) and (23). (22)

ženšˇcina, kotoraja moloko raznosit, ne prixodila ešˇcë? woman who milk she-delivers not she-came yet?

(23)

ženšˇcina, kotoraja u okna ležala, kapriznaja oˇcen’ woman who at window she-lay moody very

The essential point is that (21a) and (21b) must on no account be thought of as reduced versions of (22) and (23). Rather, the written examples have a very different clausal structure from the spoken ones. (22) and (23) consist of a main clause in which the subject NP contains a relative clause: ženšˇcina, kotoraja moloko raznosit and ženšˇcina, kotoraja u okna ležala. (21a) and (21b) consist of two main clauses juxtaposed: moloko raznosit and ne prixodila ešˇcë in (21a) and u okna ležala and kapriznaja oˇcen’ in (21b). From the perspective of English, one possibility that suggests itself is a headless relative construction. Could the Russian examples be equivalent to Has who delivers the milk come yet? and Who was in the bed by the window

Spoken and written language

31

was very moody? The attraction of the English examples is that, in spite of their infelicitous syntax, they provide subject NPs for the main clauses, namely who. This attraction does not transfer to the Russian examples, principally because the Russian equivalents of English headless relatives involve either a general noun such as cˇ elovek followed by a relative clause, or a correlative construction: for example, I will live where you live, with the headless relative where you live, corresponds to the Russian example in (24). (24)

Ja budu žit’ tam, gde ty budeš’ žit’ I will live there where you will live

The correlation is between tam and gde. The Russian equivalent of the one who is tot, kto. In order to analyze (21a) and (21b) as headless relatives, we would have to postulate a structure with a relative clause containing both an empty relative WH-NP operator and modifying an empty head NP. Since this structure provides no anchoring referent for the WH-NP operator, an analysis in terms of headless relatives lacks appeal. The analysis proposed above – that (21a) and (21b) consist of juxtaposed main clauses – is made possible by the frequent lack of subject NPs in spoken Russian. One of the striking features of Zemskaja’s data is the frequency of zero subject NPs and zero direct object NPs, especially in conversation. Even in written Russian an entity introduced by an overt NP in one clause can be referred to by zero subject NPs over five or six clauses, and even across sentence boundaries. It is clear that the speakers of (21a) and (21b) were referring to entities already mentioned or were treating the entities as highly given, and the absence of overt subject NPs is normal in this context. For (21a) an appropriate gloss is she delivers milk – has she come yet, and for (22b), she was in bed by the window – she was/is very moody. The occurrence of the feminine verb forms prixodila and ležala indicates that the speakers were referring to specific persons, women in both cases. A further example of unintegrated syntax that is typical of spontaneous spoken English is given in (25) from the map task dialogues. (25)

what you’re going to do – you’re going to go up past the allotments.

Weinert and Miller (1996) were taken to task by one referee for confusing competence and performance in their discussion of the structure of WH-clefts in English. According to the referee, (25) is the result of performance factors. But what would be the structure shared by (25) and the integrated WH-cleft of written English in (26)? (26)

What you are going to do is go up past the allotments.

The written construction has a copula but the spoken construction has not. The spoken construction has a second main clause with progressive aspect but the written

32

Jim Miller and M. M. Jocelyne Fernandez-Vest

construction has not. The spoken construction has an overt subject NP in the second clause but the written construction has not. A single integrated source structure would require a large number of unusual empty categories. Of course, perhaps the referee simply meant that speakers start to produce an integrated WH-clause, run into planning problems, cut the construction short after the WH-clause, and produce a main clause faute de mieux. This interpretation does not square with the fact that the context surrounding the WH-clefts in the map task dialogues and the spontaneous conversations display no symptoms of planning problems such as hesitations or repetitions. It also ignores the fact that the structure in (25) is easily analyzed as an information packaging instruction in the sense of Vallduví (1993). The WH-clause encodes an instruction to the listener to erase the information in the file “next action” and to prepare to enter fresh (and correct) information.20 (For the arguments supporting this analysis, see Weinert and Miller 1996) The WH-word points forward to some entity, whether proposition or concrete individual, which the speaker specifies in the second clause. Fernandez-Vest (1987: 686 ff.) observes that discourse particles play an important role in the presentation of disconnected phrases as a coherent message. In the Northern Sami example in (27) – an orally transmitted language until the 1980s – the particles connect chunks of syntax that do not by themselves make up a clause conveying a proposition, but correspond in fact to a dialogical style inserted in the monological narrative. (The chunks encoded with an exponent letter are in bold.) (27)

Sami [The anecdote is about láttánat (‘the landmen,’ i.e. non-Sami people), who enjoy fishing and wandering in the mountains, but freeze to death with the first drops of rain, as they are unable to start a fire.] ORAL VERSION Muhto maida datb dákkarc / báikegoddálaš boahtá gi lea ollu mehciid johtánd / dathanef galg arvingeh fidne dola gali . Naj i dask / mihkkigel go datm lea dola ožžonn dato datp galq i jáddat / dan galr i ajibeaivvisges+ (: : :) ‘But whata thenb such ac / local guy arrives who has a lot in the forest wanderedd / hee certainlyf yesg evenh in rainy weather+ / gets a fire yesi . Wellj in thisk / nothing nol when hem has fire-gotn heo of coursep sureq does not put it out / for surer nevers+ (: : :).’ WRITTEN VERSION 0 0 0 0 Muhto go boahtáa ollu mehciid johtánd báikegoddálaš, sonhane f galg’ 0 0 0 0 fidne arvingeh dola. Go sonm lea dola ožžon dolan , de+ láttáno i jáddat 0 dan olleges (: : : ) 0 0 ‘But when arrivesa a much-in the forest-having-wanderedd regional 0 e0 f0 g0 h0 guy, he certainly yes gets even in rainy weather a fire. When hem n0 + o0 s0 has got a fire , then the landman does not put it out at all (: : :).’

Spoken and written language

33

The interlocutive dimension, omnipresent in the Sami oral discourse, is exemplified here by the interrogative maid ‘what,’ responded in the following sentence by an elliptic comment – Na (opening particle ‘well, now’) i das mihkige ‘in this nothing no.’ The informant chose automatically to delete these dialogical signs (as well as many other discourse particles) in the written version. The final words in this section come from Heath (1985). He was writing about another phenomenon, but his sentiments are appropriate to the study of spoken language in general: There has been a recurrent tendency in much syntactic research to distinguish between an underlying, rather crystalline “grammar,” which then interacts in real speech with a distinct outer “psycholinguistic” component, the latter being especially concerned with short-term memory limitations, linear ordering of major clause constituents, resolution of surface ambiguities, etc. My view is that these two aspects of language are far more tightly welded to each other than it seems at first sight. (Heath 1985: 108)

5. Clause and clause complex A major difference between spontaneous spoken language and written language lies in the organization of clauses into clause complexes. The typical relationship between clauses in the former is hypotaxis, while in the latter, clauses are related both hypotactically and paratactically. Schulz (1973: 19–50) draws attention to the paratactic expression of causal relationships between clauses, pointing out that such relationships by no means call for subordinate clauses but can be expressed by particles, as in (28) and (29). (28)

Meistens, nachmittags, geh ich dann mit den Kinder raus, mostly in-the-afternoon go I then with the children out die müssen ja auch frische Luft haben. they must particle particle fresh air have ‘I usually go out with the children in the afternoon: : : because after all they must have fresh air.’

(29)

Da haben die nichts mit verdient. there have they nothing with earned ‘They earned nothing there.’ Die Kumpels, die hierher kamen, die hatten doch wenig the lads who here came they had particle little Geld. money ‘My mates who came here had little money.’

34

Jim Miller and M. M. Jocelyne Fernandez-Vest

Schulz glosses both examples by means of clauses introduced with weil ‘because’ and with the finite verb in clause-final position, the classical mark of subordinate clauses in German: Meistens gehe ich nachmittags mit den Kindern nach draußen : : :, weil sie frische Luft haben müssen and Damit verdienten sie nichts, weil die Kumpels, die hierher kamen, wenig Geld hatten. In the actual spoken examples the particles auch and doch signal that there is a relationship between the two clauses. Spontaneous spoken Russian exhibits a similar lack of hypotaxis, and even a lack of particles as functioning in (28) and (29). Lapteva (1976: 305–321) adduces many examples in order to demonstrate that clauses which are clearly in a discourse relation of subordination are regularly not in a relation of syntactic subordination.21 Examples are given in (30). (30) a.

b.

c.

d.

Ja ne uspel. Ja prišel tam uže ne I not was-in-time I arrived-PFV there already not prinimali. accept-IPFV-3PL ‘I was not in time. When I arrived there, they were no longer accepting [people, applications, : : :].’ Vy uezžali iz Moskvy doždik byl? you were-traveling out-of Moscow rain-DIM was? ‘When you were leaving Moscow, was it raining a little?’ cˇ to mne delat’ slesarja ne dozovešsja? what I-DAT do-INF joiner not call-2SG.REFL ‘What am I to do if you can’t get through to the joiner?’ K. vˇcera xvastalsja novuju palatku kupil. K. yesterday boast-IPFV.PST new tent buy-PFV.PST.3SG ‘K. was boasting yesterday that he had bought a new tent.’

The key property of the above examples is that the relationship between the clauses is not signaled by means of complementizers or particles. Even the notion of discourse subordination is not obviously applicable. In an English example such as When you left Moscow, was it raining? the adverbial time clause when you left Moscow can be interpreted but the interpretation is obviously incomplete; the event of leaving of Moscow is being used as a point of orientation for another event, but this other event is not specified. In (30b), however, vy uezžali iz Moskvy could be taken as a main clause, as could doždik byl.22 Fernandez-Vest (1994: 95–96), however, points out that intonation can signal the integration of two clauses as opposed to simple coordination or juxtaposition. Consider the French examples in (31).

Spoken and written language

(31) a.

35

Ţ

T’auras pas de dessert Ť t’es pas venu avec you won’t have any dessert (because) you didn’t come with nous us Il n’a pas plu Ť le linge est sec it didn’t rain (since) the washing is dry Ţ

b.

A special pitch pattern signals that t’es pas venu avec nous and le linge est sec are not free-floating clauses but are linked to the first clause in each example. Of course, the type of semantic link – time, reason, concession – does have to be reconstructed by the addressee but the information structural function is clear: a rhematic segment (falling intonation) is followed by a post-thematic one (flat intonation). To conclude this section, we note briefly that certain constructions occur in written language but typically do not occur in spontaneous spoken language. The examples in (32) are constructions from English. (32) a. b. c.

d. e. f.

g.

h.

Subordinate clauses introduced by although, since, as WH relative clauses, especially structures with the WH-form inside a PP or with a clause or clauses inside the relative clause. Gapping: Sue likes, and Bill hates, crosswords. Celia likes Van Gogh, and Bill – Rembrandt. Full gerunds: His having the book at all astonished me. Accusative + infinitive: I considered her to be the best candidate. Participial phrases: Sitting at the window, I noticed a car at the bank. Covered in confusion, he apologized. Certain conditional constructions: Were you to write to her, she would forgive you. Should you meet him, pass on my best wishes. Indirect questions: This is not just a case of whether the two words can combine.

Some of the constructions above will be picked up in Section 8 on typology.

6. Noun phrases Spontaneous spoken language and written language differ with respect to the structure of noun phrases. For English, the difference lies in the complexity of the phrase; noun phrases with more than one modifier, in addition to a determiner, are rare, and

36

Jim Miller and M. M. Jocelyne Fernandez-Vest

the modifier is typically a single adjective or a simple prepositional phrase.23 In other languages, however, different structures occur in spoken language. Zemskaja (1973: 254–255, 391) cites examples such as those in (33).24 (33) a.

b.

Tam est’ kraby banoˇcka odna there be-3SG crab-NOM.PL tin-NOM.SG one-NOM.SG ‘There’s one tin of crabs there.’ Modnyj kupili emu kostjum fashionable-ACC.M.SG buy-PFV.PST.3PL he-DAT suit-ACC.M.SG ‘They bought him a fashionable suit.’

In written Russian we would expect to find odna banoˇcka krab, with the genitive form krab. Zemskaja comments that kraby in (33a) is informationally important, hence the different constituent order. In (33b) modnyj modifies kostjum but is separated from it by the rest of the clause. Although there is no space here to discuss the matter in detail, it should be noted that a major question about constituent structure arises – namely, should the sequences kraby banoˇcka odna and modnyj kostjum be treated as deriving from solidary noun phrases or should they be analyzed as consisting of separate chunks – for example, kraby and banoˇcka odna – that are simply juxtaposed? Whatever the answer, it will certainly involve theories of lexical entries and the selection of anaphors, and appeals to the concept of scrambling are not likely to be helpful.25

7. Discourse It is perhaps surprising that spontaneous spoken language and written language should differ so much with respect to sentences, and to type and complexity of clause construction. The differences in discourse organization should be less unexpected; conversation, with turn-taking and continuous interaction between speaker and addressees, is quite different from written genres, except of course imitations of conversation in plays and novels. Planned spoken performances such as narratives, lectures, and political speeches have a general structure that runs parallel to the structure of written narratives, lectures, and speeches. The structure may well not be signaled in the same way. For instance, the narrator of an oral tale may mark a major change of event, place, and characters with a long pause and by a phrase such as “While all that was happening, : : :” or the classic cliché “Meanwhile, back at the ranch : : : .” Written narratives do not necessarily contain such continuity phrases, but major changes are signaled by gaps of different lengths and by the beginning of a new section or a new chapter. Speakers and writers do share certain tasks; they have to open texts by introducing the entities they want to talk or write about and the situations in which the entities participate; they have to signal the continuity of entities in the discourse by using

Spoken and written language

37

shorter NPs to refer to them, including null NPs; they have to introduce new entities to the original stock; when appropriate, they have to make entities more prominent or salient (NB “prominent” and “salient” are not used here as technical terms) than the other entities referred to in the text; and they have to indicate when a portion of text continues a general topic, changes the general topic, or expands or contradicts what has just been said or written. The interaction between speaker and addressees involves the speaker in tasks such as making sure that the addressees pay attention to the right entity at the right time, and checking whether the addressees follow the message and whether or not they agree with it. Formal written texts, such as novels, and especially expository texts, such as academic monographs and legal judgments, typically reflect adherence to general conventions such as beginning at the beginning of the matter and signaling continuity or change of topic, agreement, or contradiction, by the use of adverbial words and phrases such as furthermore, in contrast, nevertheless, while it is the case that : : :, and so on. The difference between spoken and written texts is clearly brought out by the comparison of written versions of papers at an academic conference with the spoken versions recorded as they were delivered (see Fernandez-Vest 1994: 144–158). The spoken versions were not given off the cuff but were anchored to a written version. One section of the published written version is given below.

(34)

la difficulté fondamentale (insurmontable?) the difficulty fundamental (insurmountable?) en traduction automatique in translation automatic ‘The fundamental (insurmountable?) difficulty in machine translation’ tient à ce qui fait la spécificité de tout texte – holds to that which makes the specificity of all text littéraire ou non – literary or not ‘lies in what constitutes the peculiar nature of any text – whether literary or not –’ écrit dans une langue, written in a language à savoir la formulation elliptique (: : :). to know the formulation elliptic (: : :) ‘written in a language, namely ellipsis (: : :).’ Or un automate ne peut travailler que sur l’explicite (: : :). now a machine not can work but on the explicit (: : :) ‘Now a machine can only handle what is explicit (: : :).’

38

Jim Miller and M. M. Jocelyne Fernandez-Vest

Le paradoxe est de devoir réconcilier le texte et l’automate. The paradox is to have-to reconcile the text and the machine ‘The paradox is having to bring together text and machine.’ The spoken text is 30 % longer than the written text. The speaker consulted his notes in order to keep to the logical order of topics and in order to provide correct documentary references, but the content and shape of the spoken text was determined by his interaction with the audience. As Fernandez-Vest (1994: 150) observes, the speaker’s presentation, unless it sticks strictly to its written version, does not require special analysis of the language of conference talks; the interaction with the audience has its roots in the practice of everyday conversation. The spoken text is given in (35). (35) a.

b.

c.

d.

e.

f.

alors/ la difficulté fondamentale / en fait / en traduction well the difficulty fundamental in fact in translation automatique automatic ‘Well, the basic difficulty in machine translation really’ elle est due à ce qui fait la spécificité de tout texte it is due to that which makes the specifity of all text écrit dans une langue naturelle / written in a language natural ‘is due to what constitutes the specific nature of any text written in a natural language’ que ce texte soit littéraire ou non / whether this text is literary or not ‘whether a given text is literary or not’ et qu’ on a évoqué / à plusieurs reprises / pendant and which one has invoked at several times during ces deux journées / these two days ‘and which has been mentioned several times during the past two days’ c’est à dire ce qui fait la spécificité d’un texte en that-is to say that which makes the specificity of-a text in langue naturelle language natural ‘that is to say what constitutes the peculiar nature of a natural language text’ c’est sa formulation elliptique that-is its formulation elliptic ‘is ellipsis’

Spoken and written language

g.

h.

i.

j.

k.

39

Hein/ le problème de l’ellipse / PART the problem of the-ellipsis ‘OK the problem of ellipsis’ la formulation elliptique avec le texte en langue naturelle / the formulation elliptic with the text in language natural ‘elliptical formulation with natural language text’ la linéarité / laconique / de la surface (: : :) the linearity laconic of the surface (: : :) ‘the laconic linearity of the surface text’ le texte n’étant autre chose que / the text not-being other thing than le signe le pointeur hein / the signe the pointer PART ‘the text being simply the sign the pointer OK’ qui suggère le sens plutôt qu’il ne le détaille. which suggests the meaning rather than it not it detail ‘which suggests the meaning rather than specifying it in detail.’

The NP specifying the new topic in line (a), la difficulté fondamentale en traduction automatique is not part of a clause; the following clause is separated from the NP by a short pause and has its own subject, elle, which picks up the text-initial NP. The NP is preceded by the particle alors signaling that the speaker is about to start; the first and important subpart of the NP, la difficulté fondamentale, is followed by another particle, en fait. This particle conveys the speaker’s knowledge that various major problems afflict machine translation, but that he is about to mention a difficulty which he regards as the most fundamental. At the same time, the particle makes la difficulté fondamentale stand out. In line (c) the phrase littéraire ou non in the written text is expanded into a full clause repeating the noun texte to indicate continuity of referent and giving accentual prominence to non. Line (d) ties in the about-to-bementioned fundamental difficulty to discussions at the conference: the difficulty is nothing bizarre but is well-known and the audience can relax. (Fernandez-Vest observes that the audience consisted of people who were linguists but not specialists of machine translation.) Line (e) is a second run-up to the mention of the fundamental difficulty, which is specified in line (f). The difficulty is respecified, with alterations, in lines (g)–(i). These lines are introduced by the particle hein, spoken with a rising pitch and followed by a short pause. The second occurrence of hein, in line (j), gives prominence to the important words signe and pointeur, and the significance of these words is spelled out in line (k). Other major differences in the discourse organization of spoken and written texts relate to the devices by which constituents, and the information they carry, are highlighted. We must make clear here and now that “highlighting” is not intended as a technical term. We are interested in any devices by which one constituent is made

40

Jim Miller and M. M. Jocelyne Fernandez-Vest

to stand out from the surrounding text: using the expressions with their everyday meaning, we say that constituents are highlighted, made salient or made prominent. We will avoid “focus,” which is a technical term with multiple uses which will be discussed in Chapter 3. Suffice it to say here that constituents can be highlighted because the speaker wants to mention an entity for the first time, reintroduce an entity already mentioned, or contrast one entity with another. Highlighting can be achieved by special word order, special constructions, particles, and of course pitch and volume. We will not be concerned with the latter two, although they are key properties of spoken language and are excluded from written language except insofar as a writer chooses to use italics or bold font or capital letters to hint at suprasegmental modulations. The highlighting devices discussed below consist of syntactic constructions or particles that are typical of spoken language – the examples are English, French, and Russian – but untypical of and excluded from written language. A common construction in spoken English (indeed, in the spoken varieties of many languages) has an NP followed by a complete clause containing a pronoun that picks up the referent of the NP. Examples from spontaneous conversation in English are given in (36). (36) a. b. c.

it’s not bad – ma Dad he doesn’t say a lot the driver he’s really friendly – you get a good laugh with him well another maths teacher that I dinnae get he must’ve corrected my papers

(36a) is a reply to the question what do you get discipline-wise which is part of a conversation about bringing up children. (36b) is a reply to the enquiry what are the people like to work with – the drivers and that? (36c) is part of a discussion about the speaker’s examination results. Note that the independent NPs in (36a) and (36b) are very simple, just a determiner and a noun, and that there are no indications of planning problems such as hesitation or repetition. (36c) does contain a complex NP, but this example also offers no indications of planning problems. The resumptive pronouns in the examples above are mostly subjects, with the exception of the oblique object with him. The pronoun can have any grammatical function; an example of a direct object is it in the book I lent you have you finished it yet? It is more likely that the construction is connected with the spreading of information over syntactic constituents in small doses, with the need to highlight entities being introduced into the discourse, and with the general constraint that constituents are kept simple in spontaneous spoken language. Of course, in (36b) the driver is given; it links back to a preceding utterance what are the people like to work with – the drivers and that but this specific driver is being mentioned for the first time. It is worth pointing out that the independent NP is not just a result of fronting; the link between the NP and the resumptive pronoun does not obey constraints such as the “complex NP constraint.” Consider (37).

Spoken and written language

(37)

41

The new boss did you hear the rumor that he’s leaving already?

Russian examples from Zemskaja (1973) are given in (38) and (39). (38) a.

b.

c.

a rebjata èti / kotorye igrajut v komandax / skol’ko but fellows these who play in teams how-many im let? to-them of-years ‘But these guys that play in the teams – how old are they?’ A kak ja uznaju kotoraja že Tat’jana? but how I will-recognize which-one particle Tat’jana ‘But how will I recognize which one is Tat’jana?’ Tanja / u nee belaja šapoˇcka s pomponom. Tanja at her white hat with pompom ‘Tanja has a white hat with a pompom.’

Russian also has a construction in which the independent NP is not picked up by a resumptive pronoun. Instead the independent NP sets a frame, and the NPs in the following clause relate to items in the frame. Consider (39). (39) a.

b.

c.

ved’ ržanoj xleb / vot ètot zapax specifiˇceskij you-know rye bread FOC that smell specific ostalsja s detstva. has-remained since childhood ‘You know, rye bread, that particular smell has stayed with me since I was a child.’ Sobaka / vsegda poly grjaznye. dog always floors dirty ‘When you have a dog/with a dog the floors are always dirty.’ Deti / bez šumu ne obojdeš’sja. children without noise not you-will-manage ‘Where there are children, you can’t avoid noise.’

Fernandez-Vest (1995) offers the French examples of a categoric (negative) general assertion (40) tempered by a more personal and specific one – detached topic + pragmatic particle (41) and the interesting example in (42) with two independent NPs. (40)

C’est pas le cas de tous les petits vieux. It is not the case of all the little old-men ‘It is not the case with all the little old men.’

(41)

Mon grand-père / bon / ça ne l’amusait pas tellement. my grandfather PART that not him-amused not so ‘My grandfather, well, it didn’t amuse him all that much.’

42 (42)

Jim Miller and M. M. Jocelyne Fernandez-Vest

Ta frangine, sa bague, c’est du toc. your sister her ring it’s of-the fake ‘Your sister’s ring, it’s fake.’

(42) could be rendered in (spoken) Scottish English as See your sister, see her ring, it’s fake. It could be rendered in other varieties of spoken English, including the standard, as You know your sister’s ring : : : or You see your sister’s ring : : : These are special constructions for the first mention of entities that are known to the speaker and hearer. Very common highlighting devices in spoken English are I mean/do you mean and like. The latter is exemplified in (43).26 (43)

A: B: A:

er I’m I’m not very sure ++ what I’m supposed to be doing em and then you have to go down again like I go past the collapsed shelter?

As mentioned above, both written and spoken English have WH-clefts, but the typical spoken WH- cleft differs from the written WH-cleft. The written one has a WH-clause followed by the copula, followed in turn by an infinitive phrase or a bare verb stem, as in What she did was to cut all the cuffs off his shirts/what she did was cut all the cuffs off his shirts. The typical spoken cleft has the WH-clause followed immediately by a main clause, as in (44). The two clauses can have the same or different subjects. (44)

no you had no food attached you got your meal hours of course but // what you did in the evenings you carried a / sandwich or two / and you had a little break in between

(45)

what you’re doing you’re going up past the market garden

(46)

I’ll give it a little stir because what happens – things tend to settle a bit

Spoken English also has a range of alternatives to the WH-cleft which do not typically occur in written texts. (47)

A: B: C:

What about Edinburgh do the people go up there oh yeah a lot oh aye especially at night they go to the pictures but the thing is if you go to the pictures if you go to the late show you’re you’ve to run for buses

(48)

A:

that’s the bad thing about the halls of residence there’s always people knocking on your door

(49)

A: B:

thing is he’s watching the man he’s not watching the ball right I see so is that the idea of this then? so you go straight to where I am instead of going round the picket fence?

Spoken and written language

43

In the map task dialogues many items are presented as new by means of know, as in know the bridge across the fast-flowing river. These utterances are understood as questions and contain only definite noun phrases. Thus, *know a bridge across the fast-flowing river.27 Spoken Russian has a range of highlighting devices that typically do not occur in written texts. (50) is an example of the particle vot. (50)

reka teˇcet / ona nazyvaetsja Kokkozka // Kokkozka // vot // river flows it is-called Kokkozka Kokkozka PART gornaja reka // mountain river ‘There’s a river it is called Kokkozka – Kokkozka, that’s it – a mountain river.’

Another construction involves a WH-question which the question-asker then answers. There are parallel constructions in spoken English and spoken French. (51)

On teper’ ved’ u nas cˇ to? he now particle at us what každyj den’ po utram truscoj begaet each day in mornings at-jog runs ‘Do you know what he’s doing now? He goes jogging every morning.’

(52)

A vy tuda kak? Èlektriˇckoj? but you there how by the electric train ‘But how did you get there? By the electric train?’

A third construction involves a construction analogous to the English reverse WHconstruction. (53)

A

B

A

Leša, slušaj Slyšiš’? Leša listen do you hear ‘Leša, listen! Can you hear me?’ Nu? particle ‘What?’ kopal. Takoe delo: zabud’, cˇ to bezxoz such thing forget that ownerless (grave) dug ‘Here’s what – forget you dug an ownerless grave.’

Spoken Russian also possesses a rudimentary IT- cleft. It consists of the neuter singular deictic form èto followed by a clause. There is never a copula between èto and the clause, which can be interpreted as indicating that there are two clauses: èto plus present tense copula, which is zero, and the following clause. The construction does not occur in formal written Russian. An example is given in (54).

44

Jim Miller and M. M. Jocelyne Fernandez-Vest

(54)

Bez cˇ etverti šest’ zvonit telefon without quarter six ring-3SG telephone Net, èto zvonjat u dveri. No this ring-3PL at door ‘At 5:45 the telephone rings. No, it’s someone ringing at the door.’

Finally, simple repetition is used to signal that a property is present in large quantity. (55)

/ i vot my šli-šli-šli / and particle we walked-walked-walked

A similar phenomenon can be observed in spoken English, not with verbs but with adjectives and adverbs: It’s a strong strong fish, Do you really really like me?28

8. Typology 8.1. Introduction That any typology of discourse must take account of the differences between spoken and written language will be clear from the preceding section and will be exemplified at various points throughout this volume. It will be useful here to spell out some of the consequences for typology of the clausal structures examined in Section 5. Since space is limited, the discussion will be based mainly on English and Russian. In Section 6 we said that this section would concentrate on one major construction, relative clauses. We demonstrate first that a whole range of syntactic constructions is affected, beginning with the conditional.29 Our spontaneous spoken English data has no examples of the construction in which the protasis is expressed by what looks like an interrogative structure, as in (56a) and (56b). (56) a. b.

Were you to write to her, she would forgive you. Should you meet him, pass on my best wishes.

These examples not only have interrogative word order in the first clause but contain an auxiliary verb (but never DO) and present the situation as remote. Spontaneous spoken English is in fact full of interrogatives functioning as protases, as in (57). (57)

Is he there? In that case I’m not going/Damn it I’m not going.

Where speakers do produce a conditional construction, the tense usage is typically different from the usage in written English. (58a) and (58b) are examples of the written construction, and (59a) and (59b) are examples of the spoken one.

Spoken and written language

(58) a. b.

45

If she came to see things for herself, she would understand. If she had come to see things for herself, she would have understood.

In spontaneous spoken English the past tense verb is frequently replaced with would + verb, and the pluperfect is replaced by would + have + participle. (59) a. b.

If she would come to see things for herself, : : : If she would have come to see things for herself, : : :

[=(58a)] [=(58b)]

The standard canon requires had come, as in (58b). Conditional clauses expressing events that can no longer happen also occur with the pluperfect replaced by had + have, the latter typically in its reduced form ‘ve. Cf. (60).30 (60) a. b.

I reckon I wouldn’t have been able to do it if I hadn’t ‘ve been able to read music [= hadn’t been able] you wouldn’t have got Mark’s place if you’d ‘ve come up last year [= had come up]

Quirk et al. (1985: § 1023) mention protasis clauses with would – I might have married her if she would have agreed – however, the interpretation is not ‘if she had agreed’ but ‘if she had been willing to agree.’ They list the example If I’d have seen her, I’d have told her only to say that informal American English speech may have matching modals – here, would = ‘d – in both clauses. The construction is clearly not just American English, and scholars undertaking a typology of conditional clauses run the risk of giving a misleading view of English if they draw their data from the major grammars of English (and other languages) without distinguishing between the spoken and written language, and without distinguishing between different genres of spoken and written language. Similar difficulties attach to participles and gerunds (not to mention ellipsis and oratio obliqua). Free participles as exemplified in Sitting at the window, I noticed a car at the bank and Covered in confusion, he apologized, are quite untypical of spontaneous spoken English. Sirotinina (1974: 25) observes that the corresponding Russian constructions are also not used in spoken Russian and cites a sentence from one of Pushkin’s letters to show that he had observed this in the 1820s. In typological work, spontaneous spoken English and Russian would have to be classified as not having free participle constructions, but the written languages as having them. The facts of spoken and written English and Russian in turn raise questions about languages like Turkish, questions which cannot be answered here.31

46

Jim Miller and M. M. Jocelyne Fernandez-Vest

8.2. Relative clauses, spoken and written language, and typology 8.2.1. Typological overview We begin by pointing out that the relative clause data to be used in this section is not new. The various types of relative clause structure have been treated at length in Lehmann (1984), and comments on the relative clause structures found in spoken language in Europe are to be found in grammars of particular languages and language groups; classic examples (different works are “classic” for different groups of linguists) are Murray (1873) and Meyer-Lübke (1900). The fact that one language may have two types of relative clause structure has been noted by, for example, Comrie (1981: 141) who refers to the gap (see (68) below) and pronoun-retention structures of Persian and the finite and nonfinite relative clause structures of English (see (79) below). What is highlighted here is the distribution of constructions over (spontaneous) speech and (formal) writing, a distinction that earlier scholars could not but be aware of, but which has gained in importance as more is discovered about the regular syntactic patterns in (spontaneous) speech and about the frequency of particular constructions. For instance, we will comment on the relative frequency and complexity of certain relative clause constructions exemplified in Section 8.2.4. Although we do not do so here, the various relative clause constructions could be analyzed with respect to varying degrees of integration, as discussed, for example, in Chafe and Danielewicz (1987) and can be seen as one part of a much larger picture of regular differences between spoken and written language. (For a discussion of English relative clauses and degrees of integration, see Miller and Weinert 1998: 108–111.) We pass now to a review of the properties relevant for a typology of relative clauses, following Comrie (1981: 131–157). 8.2.2. The order of the head NP and the modifying relative clause The relative clause can precede the NP, as in Chinese, or follow the NP, as in English. Examples are given in (61) and (62). (61)

Chinese n˘imen yòng de f˘azi you use particle method ‘the method that you use’

(62)

the book that you are reading

In the Chinese construction, the relative clause is separated from the NP it modifies by the particle de, which signals that the word, phrase, or clause-like sequence

Spoken and written language

47

preceding it is a modifier or an attribute, and that the subsequent expression is the center or head of the construction (Henne et al. 1977: 269) A third possibility is for the head NP to be inside the relative clause. Examples are given in (63). Note that Comrie deals only with restrictive relative clauses. (63)

Bambara tye be [n ye so min ye ] dyo man-the PRES I PAST house see build ‘The man is building the house that I saw.’

According to Comrie (1981: 138) the whole clause n ye so min ye functions as direct object of the main clause – Bambara has basic SOV word order and (63) has to be analyzed as having the constituents: auxiliary (be), direct object (n ye so min ye), and verb (dyo). Languages such as Bengali offer a fourth construction. Examples of this construction are appropriately translated into English by means of relative clauses, but it is quite different from the constructions exemplified in (61) and (62) above. In the latter, the relative clause and the noun it modifies belong together in one and the same NP; that is, the relative clause is inside the matrix clause containing the head noun, and the matrix clause may be main, as in (64a) or subordinate, as in (64b). (64) a. b.

I approve of the career you chose. I believe that Mary approves of the career you chose.

The career you chose is a single NP. The Bengali construction consists of two clauses, but neither is embedded in the other. Examples are in (65). (65) a.

b.

je cheleta amar bondhu, se esechilo WH boy-the (is) my friend he came ‘The boy who is my friend came.’ se esechilo, je cheleta amar bondhu he came WH boy-the (is) my friend [=(65a)]

(65a) and (65b) consist of the same two clauses, but in different orders. One clause contains the equivalent of a WH-pronoun in English, je cheleta amar bondhu, while the other contains the equivalent of a personal pronoun deictic, se esechilo. The je clause picks out a nonspecific (for the addressee) person, who is thereby introduced into the conversation and becomes given, while the se clause refers to that given person. An idiomatic English translation is The boy who is my friend came but a translation closer in spirit is some boy is friend – he came. The Bengali construction does the same work as the English one, namely, picking a particular entity out of a set of entities, but the goal is achieved by a different syntactic route.

48

Jim Miller and M. M. Jocelyne Fernandez-Vest

8.2.3. How the head noun is reflected in the relative clause In a relative clause structure with head noun and modifying relative clause, the head noun is reflected in various ways in the relative clause. a. The noun may occur unreduced in the relative clause. Comrie (1981: 140) cites the Bambara example in (63) and a Hindi example which is analogous to the Bengali examples in (65) above. (The Bengali examples are very different from the English relative clause construction, to the extent that it may be misleading to call the WHclause a relative clause. “Correlative construction” would be more appropriate and indeed this term is used by Comrie [1981: 139]). Nonstandard English (Trudgill 1983: 41) does offer examples such as he’s a man he likes his beer, with the pronoun subject of the main clause repeated as the subject of the relative clause, but no examples have been found with the head NP repeated in full inside the relative clause, along the lines of the house I saw the house was very offputting (= the house that I saw was very offputting). b. The relative clause may contain an ordinary personal pronoun bound to the head noun. An example is given in (66), taken from Comrie (1981: 141). (66)

Persian Man zan -i -r¯a [ke Hasan be u sibe zamini d¯ad] I woman -ACC that Hasan to her potato gave mišen¯asam know-PRES.1SG ‘I know the woman to whom Hasan gave a potato.’

c. The relative clause may contain a special relative pronoun, such as the WHpronouns in English, which are different from the ordinary personal pronouns. This structure is found throughout the Indo-European languages of Europe, as exemplified by the Russian sentence in (67). Written Russian has the classic Indo-European construction in which the relative clause contains a relative pronoun agreeing in gender and number with the head noun while taking its case from the verb in the relative clause. (67)

Mal’ˇcik kotoromu ona pomogala okazalsja boy-NOM who-DAT she-NOM help-PFV.PST turn out-PFV.PST neblagodarnym. ungrateful ‘The boy who she helped turned out to be ungrateful.’

In the relative clause kotoromu ona pomogala, kotoromu is masculine gender and singular number, as is mal’ˇcik. Kotoromu is also in the dative case, which is assigned to it by the verb in the relative clause, pomogala. d. The relative clause may contain a “gap”; that is, a verb whose valency is n nouns occurs in a given relative clause with only n-1 nouns. The missing noun corresponds

Spoken and written language

49

to the head noun modified by the relative clause. This construction is illustrated by the Tamil examples in (68). (Retroflex consonants are represented by a letter with a dot underneath them.) (68) a.

b.

c.

Tamil taaktar ponmanikku aaspattiriyil karatiyaal cooru ˙ ˙ ˙ doctor Ponmani-DAT hospital-LOC spoon-INST rice˙ kot:uttaar give-past-3sg ‘The doctor gave rice to Ponmani in hospital with a spoon.’ ponmanikku aaspattiriyil karatiyaal cooru kot:utta taaktar ˙ ˙ ˙ Ponmani-M/M hospital-M/M spoon-M/M rice˙ give-REL doctor ‘the doctor who gave rice to Ponmani in hospital with a spoon’ taaktar ponmanikku aaspattiriyil cooru kot:utta karati ˙ ˙ ˙ doctor Ponmani-DAT hospital-LOC rice˙ give-REL spoon ‘the spoon with which the doctor gave rice to Ponmani in the hospital’

8.2.4. English relative clauses The purpose of this section is to demonstrate that the place of English in the typology of relative clauses varies according to what type of English is considered. Formal written English has the classic Indo-European relative clause construction with a special relative pronoun, as in the examples in (69). The English variation in this construction is that preposition and WH-pronoun can occur together at the front of the relative clause, as in (69a) or the preposition can be stranded at the end of the relative clause, as in (69b). (69) a. b. c. d.

the lawyer from whom we received the assurances the lawyer who we received the assurances from the lawyer whom they consulted the department in whose files you found the missing documents

In less formal written English and in much spoken English we find the THAT relative clause construction. This is like the Tamil construction in that there is a relative clause marker, albeit a complementizer and not a special verb form, and the noun identical with the head noun is omitted. The noun phrases in (70) are alternative renderings of the Tamil examples in (68). (70) a. b.

the doctor that [ ] gave rice to Ponmani in hospital with a spoon the spoon that the doctor gave rice to Ponmani with [ ] in the hospital

In very informal English, constructions such as (70b) do not even have a preposition. (71a) is an example from the Scottish English conversations and (71b) was uttered during a radio discussion.

50

Jim Miller and M. M. Jocelyne Fernandez-Vest

(71) a. b.

of course there’s a rope that you can pull the seat back up [ ] I haven’t been to a party yet that I haven’t got home [ ] the same night

In more formal English we would expect (71a) to be : : : that you can pull the seat back up with, and for (71b) we would expect : : : that I haven’t got home from the same night. In the most formal English we would expect WH relative clauses; (71a) would be : : : with which you can pull the seat back up and (71b) would be : : : from which I haven’t got home the same night. Speakers of English use a third construction in which the relative clause is introduced by the complementizer THAT but contains a shadow pronoun. That is, this third construction is analogous to the Persian one exemplified in (66) and also to the relative clause construction of Hebrew. (72a) is from the Scottish English conversations, (72b) is from a corpus collected in Somerset by Ossi Ihalainen, (72c) was noted by Jim Miller during domestic conversation, and (72d), in a conversation between a shopkeeper and a customer. (72) a. b. c. d.

the girl that her eighteenth birthday was on that day was stoned, couldn’t stand up would those men I call their names step forward the spikes that you stick in the ground and throw rings over them an address which I hadn’t stayed there for several years

The relevant parts of the above examples are that her eighteenth birthday in (72a) as opposed to whose eighteenth birthday; I call their names in (72b) as opposed to whose names I call; throw rings over them in (72c) as opposed to throw rings over or over which you throw rings; and in (72d) which I hadn’t stayed there as opposed to which I hadn’t stayed at. (72d) contains which and not that but the relevant feature is the occurrence of the shadow pronoun there, which is pronominal but not a personal pronoun. The presence of her in (72a) contrasts with the absence of whose, and the presence of them in (72c) contrasts with the absence of which. Whose and whom are absent from nonstandard English and from most spontaneous spoken English. They are a hallmark of formal written English and occur in spoken English only if it is formal and planned, or if the speaker (like many academics and other professional people) has been greatly influenced by written English. Who uses the constructions in (72)? Comrie (1981: 140), discussing the example this is the road that I know where it leads, declares that this type of relative clause construction has a rather marginal existence in English, but the validity of this observation depends on whether “marginal” relates to written English or to grammars of English. The construction is regular in nonstandard English (which, after all, is spoken by the large majority of the population of the UK), but also occurs regularly in the spontaneous speech of educated speakers and can be heard, for example, in radio discussion programs. Like many other constructions in nonstandard English,

Spoken and written language

51

they are not new but can be traced back to Middle English: the Oxford English Dictionary has that same cock that Peter heard him crow. The significance of this OED example is twofold: it occurs in a written text, and the relative clause is not complex or compound and does not contain a possessive construction. Its occurrence in a written text is important because it demonstrates how long the construction has existed in English and that its occurrence is not caused by speakers having to cope with complicated syntax during unplanned speech production. In fact the construction type is widespread throughout Europe and possibly has always been available in spontaneous spoken language. Meyer-Lübke (1900: 702–703) cites Italian examples from Boccaccio, Ariosto and a French example from Aucassin et Nicolette, which bear witness to the antiquity of the construction. In contrast, the only modern examples he gives are from modern Provençal and it is more recent literature, such as Deulofeu (1981), which sets out examples from modern spoken French. Vondrák (1928: 483–484) cites both written examples from older stages of Slavic and examples that he noted down from spoken Polish and Czech, albeit more than seventy years ago, since the 1928 edition of his grammar was posthumous.32 Churlish though it may seem, it is essential to emphasize that what is central to the study of relative clauses in spoken and written language is not just the types of construction available across the languages of Europe, but the distribution of the constructions over genres and speakers and the position of the constructions with respect to the integration of clauses. The different types of relative clause construction range from the tightly integrated classical WH-type through the less integrated shadow – or resumptive – pronoun types to very unintegrated structures. (See comments on integration and genres in Section 8.2 and the comments on [72] below.) Finally, it is worth mentioning a construction in which relative clauses typically have no complementizer. Examples are given in (73). (73) a. b. c.

we had this French girl came to stay my friend’s got a brother used to be in the school there’s a man in our street has a Jaguar

(73a) and (73b) are taken from the corpus of Scottish English conversations, and (73c) was noted in the course of conversation. The formal English equivalents are: we had this French girl who came to stay, my friend’s got a brother who used to be in the school, and there’s a man in our street who has a Jaguar. The examples are all existential-presentative, either with THERE BE or NP HAVE (GOT). The above examples were produced by speakers of Scottish English, but the construction is common in spontaneous spoken English and examples are to be found in dialogues in novels – the earliest one in Jim Miller’s collection coming from an aristocratic character in Trollope’s The Duke’s Children. In fact the construction is much older. Lockwood (1968: 242) cites Shakespeare’s youth’s a stuff will not endure, which in modern English is youth’s a stuff that/which will not endure, and gives an example

52

Jim Miller and M. M. Jocelyne Fernandez-Vest

of a parallel construction from Otfrid’s ninth century Evangelienbuch to demonstrate that the structure existed in Germanic very early. A final two comments on relative clauses in spoken nonstandard English are in order. The examples in (72) can be thought of as one step away from the writtenlanguage construction in which a given relative clause is embedded in a noun phrase and integrated with the syntax of the main clause. The reduction in integration is reflected in the fact that the complementizer that is followed by a complete clause, that is, by a clause in which the verb has its full range of overt (appropriate) complement NPs. The examples in (72) have ordinary pronominal possessive, direct object and locational items – her birthday, their names, them, there. Trudgill (1983: 41) provides an example of a relative clause construction in which the subject slot in the relative clause is filled by an ordinary pronoun and in which the relative clause lacks a complementizer. These two properties in conjunction create the strong impression of two clauses merely juxtaposed and with no syntactic integration at all. (74)

He’s a man [ ] he likes his beer

Trudgill gives an example of a nonpresentative/existential construction in which the subject slot in the relative clause is empty. (75)

He’s a man [ ] likes his beer

Trudgill gives no indication of the geographical spread of the constructions in (74) and (75). There are no examples in the corpus of Scottish English conversation (cf., however, the example from Shakespeare cited above), but (74) and (75) are typical examples of the loosely integrated or completely unintegrated syntax of clause complexes in spontaneous spoken language. In contrast, (76) illustrates the occurrence of a third complementizer, as. (76b) demonstrates that as can function as the complementizer in complement clauses as well as in relative clauses, and the geographical spread of as is known approximately – southern England.33 (76) a. b.

He’s a man as likes his beer They say as he’s lost his nerve

A fourth complementizer is what, as in (77). This complementizer is widespread in Britain, although the boundaries of its distribution are unclear. Cheshire et al. (1993: 68), on the basis of a questionnaire issued to teachers and pupils in various urban schools, state that what appears to be the preferred relative pronoun in the urban centers of Britain today. Trudgill (1983: 41) provides (77a) and (77b) from a detective series set in the northern English town of Newcastle and written by a native of the town. What as a relativizer is not attested in the Scottish English conversations. (77) a. b.

He’s a man what likes his beer Thanks for the letter what you writ [Newcastle, in Spender, BBC1, 5 January 1993]

Spoken and written language

53

To conclude this brief overview of relative clauses in spontaneous spoken English, we may observe that two constructions cited in many discussions of relative clauses are very rare; when they do occur in speech, it is typically in the discourse of people who spend much time working with written English. The first is the non-restrictive relative, as in the boy, wo actually works in the yard, was saying : : : The alternative construction is simply coordinative, as in (78). (78)

The boy I was talking to last night – and he actually works in the yard – was saying it’s going to be closed down

The second rare construction is the infinitival relative clause. There are very few in the Scottish English corpus and the map task dialogues. There are no WH infinitival relatives – I’m looking for a place in which to set up the factory – and the ones that do occur are like spoken finite relative clauses in having no preposition in final position where one would be required in written English. Examples are given in (79), with the slot for the “missing” preposition enclosed in square brackets. (79) a. b. c. d. e. f.

eh Laurine – question to tell you – eh if you haven’t got the volcano – where do you go if you haven’t got the volcano I’ve got a place to start It’s not the ideal place to go [ ] for teenage drinking because there’s vandals and it’s a horrible place to live [ ] Saxone’s is a very good place to work [ ] can I have a mat to put the pizza down on

Note that (79f), with preposition, was produced by a university graduate. The complementizer that developed from the pronoun that, a typical IE source of complementizers. It is worth commenting that which is going the same way, witness which [B] in (80) from the Scottish English conversations. (80)

you can leave at Christmas if your birthday’s in December to February which [A] I think is wrong like my birthday’s March and I have to stay on to May which [B] when I’m 16 in March I could be looking for a job

The construction in (80) occurs in Dickens – for example, in the speech of Mr. Wegg in Our Mutual Friend – and was used throughout the second half of the nineteenth century in the humorous journal Punch as a marker of lower-class characters. Its principal characteristic is that which functions to connect two chunks of discourse and to signal that the second chunk relates to the first. What then is the English relative clause typologically? We must first recognize that the term “English relative clause” is merely a label for a set of constructions with the same function. We must also recognize that the English variant of the classical Indo-European relative clause construction belongs to written English. (This is not to deny that it occurs in speech; it occurs in planned formal spoken texts such as

54

Jim Miller and M. M. Jocelyne Fernandez-Vest

lectures, sermons, political speeches, and in the unplanned speech of highly educated people.) In spontaneous spoken English and in nonstandard English the typical relative clause construction is like the Persian or Semitic constructions, with an invariant complementizer. The clause following the complementizer may be like the Tamil construction in lacking a noun phrase that would occur if the clause were a main clause, or it may be like the Persian and Semitic constructions in having a shadow pronoun. 8.2.5. Relative clauses in Russian Like standard written English, standard written Russian possesses the classic IE relative clause construction with a WH relative pronoun. Prepositions are never stranded but occur at the front of the relative clause preceding the WH-pronoun. The latter agrees with the head noun in number and gender, but takes its case from the verb in the relative clause. In (81) kotoraja is feminine and singular, agreeing in these properties with kniga. It is nominative, a property assigned by soderžit. (81)

kniga, kotoraja soderžit èti svedenija (the) book which contains this information

(82a) is an example of fronted preposition and WH-pronoun, and (82b) is an example of an incorrect relative clause with a stranded preposition. (82) a.

kniga, v kotoroj ja našel èti svedenija (the) book, in which I found this information b. *kniga, kotoroj ja našel èti svedenija v

Informal spoken Russian possesses a relative clause construction analogous to the Bengali construction in (65). That is, it is more a correlative construction, consisting of two juxtaposed clauses. The first clause contains a WH-form and the second clause contains a definite deictic. The key to this construction is that the WH-forms are indefinite deictics, that is, speakers use them to point to a set of entities (the set may contain one or more members), introducing the entities into the discourse. The entities are known to the speaker but not to the addressee. The speaker introduces them by means of the first clause and assigns a property to them by means of the second clause. Consider the examples in (83), which are taken from Lapteva (1976: 144). (83) a.

kotorye vot klienty est’ u menja, i WH-NOM.PL particle customer-NOM.PL are at I-GEN and tem ja smotrju this-DAT.PL I-NOM look-PRES.1SG ‘I look after the customers who are mine.’

Spoken and written language

b.

55

kotorye mal’ˇciki lomajut, on vsex WH-NOM.PL boy-NOM.PL break-PRES.3PL he-NOM.SG all-ACC zabiraet catch-PRES.3SG ‘He catches all the boys who break things.’

The first clause in (83a) introduces the set of customers “belonging” to the speaker by means of kotorye : : : klienty est’ u menja, and the second clause picks up the reference by means of tem and assigns the property ‘I look after them.’ (83b) is interpreted analogously, although the pronoun in the second clause is vsex (all) and not tem. (84) is an example of a one-member set being introduced by the first clause, with the reference being picked up by means of an ordinary personal pronoun in the second clause. (84)

Ta, kotoraja zdes stojala that-NOM.F.SG WH-NOM.F.SG here stand-PST.3SG.F lampa, ja ee ne bral lamp-NOM.F.SG I it-ACC.SG not take-IPFV.PST.M.3SG ‘The lamp which was standing here, I didn’t take it.’

(85), also from Lapteva (1976: 144), demonstrates the use of the Russian WHforms as independent pronouns. (85)

v butylkax, ono segodnjašnee? A kotoroe but WH-NT.SG in bottle-LOC.PL it today’s ‘But the [milk] in the bottles, is it today’s?’

The examples above show that while written Russian has the classic Indo-European relative clause construction, spontaneous spoken Russian has a construction that can be analyzed as analogous to the correlative construction of Bengali. The correlative construction consists of two clauses linked by the occurrence of a WHpronoun in one and a definite deictic in the other, but spoken Russian offers examples of a relative clause construction with an even lower degree of cohesion. In (86) there are three clauses. The first clause is imperative (Idi v vannu), the second clause is also imperative (voz’mi tam platoˇcek), and the third clause looks like a main clause (na trube soxnet). This third clause has no subject NP but this is typical of spoken Russian and, to a lesser extent, of written Russian. The “missing subject” is platoˇcek, the information conveyed by this clause relating to the cloth. It is possible to find no intonation break between the second and third clauses, in which case the intonation can be taken as signaling cohesion between them, but the third clause could equally well have its own intonation.

56

Jim Miller and M. M. Jocelyne Fernandez-Vest

(86)

Idi v vannu, go-IMP into bathroom-ACC.SG.F voz’mi tam platoˇcek na trube soxnet take-IMP there cloth-ACC on pipe-LOC dry-PRES ‘Go into the bathroom and bring the cloth that is drying on the pipe.’

To sum up: spoken English relative clauses are like the relative clauses of Persian and Hebrew, and spoken Russian relative clauses are like the correlative construction of Bengali (and Hindi). Written English and written Russian have the classic IndoEuropean construction.

9. Conclusion Sections 3–7 have demonstrated the large differences in clausal syntax and discourse organization between spontaneous spoken language and written language. Were the examples drawn from only one language, skeptical readers might object that they represent merely performance phenomena. The occurrence of similar structures in different languages, their frequency and regularity in the language of different speakers in the English corpuses, and their persistence over long periods of time all indicate that they are not ephemeral performance phenomena but regular constructions of spontaneous speech. Section 8 demonstrates that the differences between spontaneous spoken language and written language are indeed relevant to typology, at least with respect to one construction; other relevant constructions are the passive, clefts, negation, combinations of protasis and apodosis. Finally, it is worth emphasizing that only examples (74)–(77) are from a nonstandard variety and that the rest of the data in this chapter relates to spoken standard English and spoken standard Russian. Nonstandard varieties of any language are typically spoken but not written, are usually spoken by larger groups of speakers than the standard, and should be taken into account by typologists. We need look no further than negation; standard English has single negation, as in We didn’t see anyone but all nonstandard varieties of English in Britain and North America have double negatives, as in we didn’t see no one. Where is “English” to be placed in a typology of negative constructions?

Notes 1.

The original version of this paper was written in the autumn of 1994 and revised in January 1995. This version of October 1999 benefits from comments by two referees and from further reflection on spoken and written language. The assertion that little attempt had been made to relate spoken data to problems of syntactic analysis was

Spoken and written language

2.

3.

4.

5. 6.

7.

57

true in early 1995; it must now be modified, but not thrown away, in response to the steadily growing number of papers and books on spoken data and theory. Equally, there is no suggestion that no serious work was carried out on spoken language before 1995. Serious work on the nature of spoken and written language and the relationship between them was carried out by Prague Circle linguists in the 1930s and scholars documented, for example, the evolution of early written Latin (taken as still close to the spoken language) into the complex clause and sentence structure of Golden Age Latin and similar evolutions in other languages. The problem of analyzing the syntax of spoken language, whether spontaneous or not, was left untackled, though to be fair, the problems were not noticed, and possibly not noticeable, until the development of genuinely portable tape- and cassette-recorders enabled investigators to record large quantities of data. We have not tried to establish an exhaustive list of even major contributors to the field, but refer to the work which has influenced us. To our personal knowledge, relevant work is in progress in the UK, North America, France, Italy, Germany, Finland, Russia, Poland, the Czech Republic, and elsewhere. The results of this work are published in a range of languages; while we could, in principle, read work published in the languages of the above countries, time simply does not permit exhaustive coverage and our references are inevitably, if regrettably, selective. The informants were also all white, a serious inadequacy with respect to present day considerations, but not surprising if we consider that the data was collected in the early- to mid-1960s in Britain. Interestingly, there is no corpus of, for example, spontaneous spoken standard English and no analysis. Quirk et al. (1985) includes spoken constructions but they are not sorted into standard and nonstandard. The layman’s notion of what constitutes acceptable spoken standard English is invariably based on the written language. The mixture of political and linguistic considerations that arise with respect to nonstandard varieties makes itself felt in the Eurotype Guidelines. Spoken languages such as Swiss German and Plattdeutsch are omitted, while Occitan, which has a medieval literature, is included. But Scots, which also has a good medieval literature, is omitted, along with the major nonstandard varieties of English. Upper and Lower Sorbian are distinguished, but not East and West Bulgarian. See the work of scholars such as Kroll (1977), Chafe (1982), Zemskaja (1973), Lapteva (1976), Sornicola (1981), Blanche-Benveniste (1991), and Milroy and Milroy (1985). It would be foolish to assert that spontaneous spoken texts never contain complex phrases or clauses of the sort given by Halliday. As Fernandez-Vest (1995) observes, even informal conversation contains rehearsed chunks of language which may range from a speaker’s favorite phrases to a narrative that the speaker has told several times before. Such rehearsed chunks may indeed be complex – but they are not typical. On the other hand, Halliday’s assertion about the considerable proportion of subordinate clauses does not apply at all to the corpus of spontaneous conversations nor to the narratives analysed in Macaulay (1991). For details see Miller and Weinert (1998: 14– 22). This discussion of sentences and clauses is necessarily brief. An extended discussion

58

8.

9.

10.

11.

12.

13.

Jim Miller and M. M. Jocelyne Fernandez-Vest can be found in Miller and Weinert (1998: 28–71), Fernandez-Vest (1995), and in the survey of “the written language bias” in Linell (1982). Sornicola’s analysis accords with the data presented in Zemskaja (1973), which has the additional advantage of belonging to informal spontaneous conversation from highly educated speakers. This is an advantage in that it weakens the possible objection that the structures under analysis owe some or all of their properties to their being instances of nonstandard English, Italian, and so on. Of course readers are entitled to ask why they should give up a well-developed system of sentential syntactic analysis in return for a system that exists only in the form of a promissory note. The answer is in several parts. Firstly, the analysis of written language still needs sentential syntactic analysis. There is no demand that such analysis be given up altogether, merely that limitations on its scope be recognized. Secondly, written language cannot be completely analyzed without discourse rules that specify the relationships between separate sentences – relationships that also hold between the clauses of speech. The relationships range from coreference to rhetorical structure ones such as condition, elaboration, and concession (cf. Matthiessen and Thompson 1988). Thirdly, the alternative to discourse rules is a set of rules mapping tightly integrated arrangements of system sentences onto loose arrangements of sentences (in writing) and clauses (in speech). Such a system of rules has not even been foreshadowed, far less explicitly proposed. The absence of any sentence marking in speech compared with an abundance of clear sentence marking in writing raises a further question as to the validity of the sentence in the analysis of spoken language. It is worthwhile remarking that the sort of discourse rules envisaged here already exist in prototype form in the work of Mann, Thompson, and others. (See Matthiessen and Thompson 1988; Mann and Thompson 1987; Hovy 1990). Matthiessen and Thompson (1988: 300) demonstrate two essential points: the same general relationships hold among clauses in clause combinations as among higher-level units of text, and clauses have the same combinatory structure as higher-level text units, namely, as members of a list or as nucleus with satellite. It is not clear why it should be difficult to recognize sentences in written English, provided a given writer has used capital letters and full stops according to convention. Whether readers find that writer’s usage stylistically acceptable is quite another question. Some analysts propose new analytical units for work on oral texts; the idea unit Kroll (1977), the turn unit of conversational analysis, the tone unit of Quirk et al. (1985) among others, the t-unit of Hunt (1966), and the utterance as defined by Crookes and Rulon (1985: 9). The details of these various alternative units need not concern us here, and syntactic structure does require phrases and clauses. Interestingly, similar difficulties occur in the transcripts of spoken Russian in Kapanadze and Zemskaja (1979) for what they call completed and uncompleted intonation. “Signalling completion of an utterance” appears to be equivalent to speakers signalling that they have come to the end of a coherent content sequence, as Chafe and Danielewicz (1987) put it. Such sequences may consist of a phrase and a fragment, two or more main clauses, or a main clause plus one or more subordinate clauses.

Spoken and written language 14.

15. 16.

17.

18.

19.

20.

21.

22.

59

The data from Wackernagel-Jolles not having been subjected to statistical analysis, it is impossible to state whether the students were in greater agreement on the sentence boundaries than chance would permit. For present purposes that question is not directly relevant, since the point being made is that even with a transcript, there was a conspicuous lack of agreement about the location of the sentence boundaries. It would be surprising if listeners were to disagree about the boundaries and the constituents of individual clauses, though admittedly the relevant results must come from an experiment that is yet to be conducted. For examples, see Miller and Weinert (1998: 41–45). M. M. Jocelyne Fernandez-Vest (1987: 642 ff.) points out that a similar observation can be made about Old Swedish legal language, which was based on the paratactic organization of clauses in the spoken language. Over time this legal language was elaborated and provided with the hypotactic organization of clauses typical of written language. While Matthews is correct about intonation and phonemic analysis, the importance of intonation and rhythm for spoken language must not be underestimated. It has been shown that the structure of oral discourse is based primarily on the segmentation of the sound chain into melodic units and prosodic groups. See Svartvik and Quirk (1980: 38–40) for English, and Morel and Danon-Boileau (1998) for French. Optative verb forms were used to express wishes (h¯emas sozoi ‘us he-save,’ i.e., ‘If only he would save us’) and occur in various irrealis constructions, such as the protases of conditional structures, expressing greater remoteness from reality than is expressed by subjunctive forms. Another approach to the constraints on order is to apply the concepts of theme and rheme to larger units than a single clause. Adverbial clauses in sentence-initial position can be seen as thematic, in sentence-final position as post-rhematic. FernandezVest (1994: 202–205) discusses the correlation between these positions and different types of prosody, although the situation is complicated by the constant entanglement in spontaneous spoken language of modus and dictum, which has been analyzed in terms of “polyphony.” Cf. Ducrot (1984: ch. 13). Vallduví (1994) likewise argues that right-detached constituents in Catalan are not afterthoughts to communicatively defective clauses but are part of a special information packaging instruction. The core of a given clause contains focal constituents but the constituents that form the ground are right-detached. Lapteva states that dialects and spontaneous spoken standard Russian share much syntactic structure. She is not investigating ‘weakly formed structures’ (slabooformlennye postroenija), which reflect the unplanned nature of spontaneous speech and present no general patterns, no typical meanings, and no internal system. Nor is she studying fixed expressions, almost totally predictable in terms of the situations in which they are used and their syntax. Rather, she is studying tipizirovannye struktury ‘typed structures,’ which allow a wide range of lexical items and syntactic modifications, being independent of context. They are very frequent and are subject to rules. The structure occurs in written Russian, not in highly formal registers but in, for instance, newspaper articles aimed at a wide audience. The example below is from an

60

Jim Miller and M. M. Jocelyne Fernandez-Vest article on grieving that appeared in the weekly Argumenty i Fakty. The clauses are both main clauses. (i)

23. 24.

25. 26.

27.

28.

29. 30.

Russian Podnimaeš’sja po lestnice, mel’kaet mysl’ : : : climb-2SG.IPFV up stair-DAT flash-3SG.IPFV thought ‘As you are going upstairs, the thought flashes into your head : : :’

For more details, see Miller and Weinert (1998: 164–169, 176–182). Elena Maslova (pers. comm.) points out that examples such as (33) are certainly more widespread in spoken than in written language but that they do occur in written language, certainly in poetry, but possibly also in prose. Blanche-Benveniste (1991: 103–111) provides a number of nominal structures from spoken French that are not found in written French. The analysis of like as a highlighter is controversial. A detailed account of the map task dialogues and some spontaneous conversation is given in Miller and Weinert (1995). The crucial fact is that like is not typically associated with planning difficulties of any kind. Moreover, there is another construction in which like occurs in clause-final position, which rules out any explanation in terms of planning problems. The declarative like-final construction is used to put forward an explanation for a previous statement that is felt by its producer as possibly causing puzzlement or disbelief. The interrogative construction seems simply to give prominence to a particular NP or even a whole clause. Know is equivalent to the Broad Scots ken (cognate with the German kennen). Ken introduces items into a discourse as in (i), converts, so to speak, the presentation of an entity as given into a presentation as new, or introduces a complete new topic of conversation. Ken never combines with you in this construction. (i)

ken John Ewan – he breeds spaniels

(ii)

the estate up at Macmerry – ken there’s a big estate there – it’s got a gamekeeper

(iii)

[after a long pause in the conversation] ken this wee lassie comes in with tea towels at ten to seven and I washed them – ken I’m wanting them boiled so they take a wee while you know

Hagège (1993: 150–156) observes that this reinforcement of meaning by the repetition of words (or syllables) is a type of morphosymbolism that occurs universally in natural languages and is especially productive in orally transmitted languages. Quirk et al. (1985: 1441) give examples such as A lovely, lovely chrysanthemum, The only remedy is work, work, work, and Everywhere I looked there were children, children. See also the discussion of typology and clause combining in Miller and Weinert (1998: 353–363). The same construction is also found in clauses introduced by wish, which likewise present an event as no longer possible.

Spoken and written language (i) 31.

32.

33.

61

I wish he’d ‘ve complimented me, Roger [= had complimented]

The need for further research is highlighted by the fact that, as observed by my theme group colleague Elena Maslova, the equivalent of the free participle construction is used by speakers of languages such as Yukagir, which, until recently, had no written variety. The Rila Monastery in Bulgaria is a World Heritage Site but typologists in need of additional reasons to visit it should know that they can look for an example of the shadow pronoun relative clause construction in one of the painted panels to the left of the main entrance. Miller saw it in 1975 but failed to record it. An equivalent construction has been noted in spontaneous spoken Finnish (both in the urban standard and in several dialects). The polysemic conjunction kun (the reduced kuin ‘as’ falls together with the temporal kun ‘when’) serves as a relativizer, as in (i)

naapurin tyttö kun toi minulle neighbor-GEN girl-NOM COMP bring-PST.PFV I-ALL ‘the neighbor’s girl as brought me’

The structure is discussed in Fernandez (1982: 220–230). With respect to information structure, this relativizer has been shown to share several of the characteristics of the pragmatic particles – cf. Ch. 3. It is a clitic, is prefixed to the initial constituent of a clause, and is a thematizer.

References Beaman, K. 1984

Coordination and subordination revisited: Syntactic complexity in spoken and written narrative discourse. In Coherence in Spoken and Written Discourse, Deborah Tannen (ed.), 45–80. Norwood, NJ: Ablex.

Biber, Douglas 1988 Variation across Speech and Writing. Cambridge: Cambridge University Press. Biber, Douglas, Stig Johanson, Geoffrey Leech, Conrad Susan and Edward Finegan 1998 Longman Grammar of Spoken and Written English. London: Longman. Blanche-Benveniste, Claire 1991 Le français parlé. Études grammaticales. Paris: Éditions du Centre National de la Recherche Scientifique. Bloomfield, Leonard 1935 Language. London: Allen and Unwin. Brown, G., A. Anderson, R. Shillcock and G. Yule 1984 Teaching Talk. Strategies for Production and Assessment. Cambridge: Cambridge University Press. Burton, S. H. 1986 ABC of Common Errors. Longman English Guides. Harlow: Longman. Chafe, Wallace L. 1982 Integration and involvement in speaking, writing and oral literature. In Oral and Written Language. Exploring Orality and Literacy, Deborah Tannen (ed.), 35–54. Norwood, NJ: Ablex.

62

Jim Miller and M. M. Jocelyne Fernandez-Vest

1984

How people use adverbial clauses. In Proceedings of the Tenth Annual Meeting of the Berkeley Linguistics Society, Claudia Brugman and Monica Macaulay (eds.), 437–479. Berkeley: Berkeley Linguistics Society. Chafe, Wallace L. and Jane Danielewicz 1987 Properties of written and spoken language. In Comprehending Oral and Written Language, Rosalind Horowitz and S. Jay Samuels (eds.), 83–113. New York: Academic Press. Cheshire, Jenny, Viv Edwards and Pamela Whittle 1993 Non-standard English and dialect levelling. In Real English: The Grammar of English Dialects in the British Isles, James Milroy and Lesley Milroy (eds.), 53–96. London, Longmann. Comrie, Bernard 1981 Language Universals and Linguistic Typology. Oxford: Basil Blackwell. Crookes, G. V. and K. Rulon 1985 Incorporation of Corrective Feedback in Native Speaker/Non-native Speaker Conversation. Technical Report No. 3, Center for Second Language Classroom Research, Social Science Research Institute. Honolulu: University of Hawaii. Crystal, David 1987 The Cambridge Encyclopaedia of Language. Cambridge: Cambridge University Press. Deulofeu, José 1981 Perspective linguistique et sociolinguistique dans l’étude des relatives en français. Recherches sur le français paré 3: 135–194. Ducrot, Oswald 1984 Le dire et le dit. Paris: Minuit. Enkvist, Nils Erik 1982 Impromptu speech, structure and process. In Impromptu Speech: A Symposium, N. E. Enkvist (ed.), 11–32. Åbo: Åbo Akademi Foundation. Fernandez, M. M. Jocelyne 1982 Le finnois parlé par les Sames bilingues d’Utsjoki-Ohcejohka (Laponie finlandaise) – Structures contrastives, syntaxiques, discursives. L’Europe de Tradition Orale 1, Paris: SELAF. Fernandez-Vest, M. M. Jocelyne 1987 La Finlande trilingue, 1 – Le discours des Sames – Oralité, contrastes, énonciation. Préface de Claude Hagège. Paris: Didier Erudition. 1994 Les particules énonciatives dans la construction du discours. Paris: Presses Universitaires de France. 1995 Morphogenèse orale du sens : de l’espace des langues aux objets de discours. In Oralité : invariants énonciatifs et diversité des langues. Intellectica 20 (1995/1), M. M. J. Fernandez-Vest (ed.), 7–53. Guiraud, Pierre 1963 Le moyen français. Paris: Presses Universitaires de France. Hagège, Claude 1993 The Language Builder. An Essay on the Human Signature in Linguistic Morphogenesis. Amsterdam: John Benjamins. Halliday, Michael A. K. 1989 Spoken and Written Language. Oxford: Oxford University Press. Heath, Jeffrey 1985 Discourse in the field: clause structure in Ngandi. In Grammar Inside and Outside the Clause: Some Approaches to Theory from the Field, Johanna Nichols and Anthony C. Woodbury (eds.), 89–110. Cambridge: Cambridge University Press.

Spoken and written language

63

Henne, Henry, Ole B. Rongen and Lars J. Hansen 1977 A Hanndbook on Chinese Language Structure. Oslo: Universitetsforlaget. Hovy, Edward 1990 Unresolved issues in paragraph planning. In Current Research in Natural Language Generation, Robert Dale, Chris Mellish and Michael Zock (eds.), 17–45. New York: Academic Press. Hunt, K. W. 1966 Recent measures in syntactic development. Elementary English 43: 732–739. Kapanadze, M. and E. O. Zemskaja 1979 Teksty. Moscow: Nauka. Kroll, Barbara 1977 Combining ideas in written and spoken English: a look at subordination and coordination. In Discourse across Time and Space, Southern California Occasional Papers in Linguistics 5, E. Ochs Keenan and T. L. Bennett (eds.), 69–108. Los Angeles: University of Southern California. Lapteva, O. A. 1976 Russkij razgovornyj sintaksis. Moscow: Nauka. Lehman, Christian 1984 Der Relativsatz. Typologie seiner Strukturen, Theorie seiner Funktionen, Kompendium seiner Grammatik. Tübingen: Gunter Narr. Linell, Per 1982 The Written Language Bias in Linguistics. Studies in Communication 2. Linköping: University of Linköping, Department of Communication Studies. 1988 The impact of literacy on the conception of language: the case of linguistics. In The Written World, R. Saljö (ed.), 41– 58. Berlin: Springer. Lockwood, W. B. 1968 Historical German Syntax. Oxford: Clarendon Press. Lyons, J. 1977 Semantics. Vol. 1. Cambridge: Cambridge University Press. Macaulay, Ronald K. S. 1991 ‘Coz it izny spelt when they say it’: displaying dialect in writing. American Speech 66 (3): 280–291. Mann, W. C. and S. A. Thompson 1987 Rhetorical structure theory: a theory of text organisation. ISI/RS-87-190. Marina del Rey, CA: Information Sciences Institute. Matthews, Peter H. 1981 Syntax. Cambridge: Cambridge University Press. Matthiessen, C. and S. A. Thompson 1988 The structure of discourse and subordination. In Clause Combining in Grammar and Discourse, J. Haiman and S. A. Thompson (eds.), 275–329. Amsterdam: John Benjamins. Meyer-Lübke, Wilhelm 1900 Grammaire des langues romanes. Tome troisième; Syntaxe [translated by Auguste Doutrepont and Georges Doutrepont]. Paris: H. Welter. Miller, Jim and Regina Weinert 1995 The function of LIKE in spoken language. Journal of Pragmatics 23: 365–393. 1998 Spontaneous Spoken Language. Syntax and Discourse. Oxford: Clarendon Press. Milroy, James and Lesley Milroy 1985 Authority in Language: Investigating Language Prescription and Standardisation. London: Routledge and Kegan Paul.

64

Jim Miller and M. M. Jocelyne Fernandez-Vest

Milroy, James and Lesley Milroy (eds.) 1993 Real English: The Grammar of English Dialects in the British Isles. London: Longman. Morel, Mary-Annick and Laurent Danon-Boileau 1998 Grammaire de l’intonation. L’exemple du français. Paris: Ophrys, Bibliothèque de Faits de langues. Murray, James A. H. 1873 The Dialect of the Southern Counties of Scotland. London: The Philological Society. Ong, Walter 1982 Orality and Literacy. London: Routledge. Palmer, L. R. 1954 The Latin Language. London: Faber and Faber. Poole, Millicent E. and T. W. Field 1976 A comparison of oral and written code elaboration. Language and Speech 19: 305–312. Quirk, Randolph, Sidney Greenbaum, Geoffrey Leech and Jan Svartvik 1985 A Comprehensive Grammar of the English Language. London: Longman. Radford, Andrew 1988 Transformational Grammar: A First Course. Cambridge: Cambridge University Press. Schulz, Gisela 1973 Die Bottroper Protokolle – Parataxe und Hypotaxe. Munich: Max Hueber. Sirotinina, O. B. 1974 Sovremennaja razgovornaja reˇc. Moscow: Prosvešˇcenie. Sornicola, Rosanna 1981 Sul parlato. Bologna: Il Mulino. Svartvik, Jan and Randolph Quirk 1980 A Corpus of English Conversations. Lund: liber/Gleerup. Trudgill, Peter 1983 Sociolinguistics: An Introduction to Language and Society. Harmonsdsworth: Penguin. Vallduví, Enric 1993 Information packaging: a survey. Research paper HCRC/RP-44, University of Edinburgh, Human Communication Research Centre. 1994 Catalan right-detachment and information packaging. Journal of Pragmatics 22: 573– 601. Vondrák, Wenzel 1928 Vergleichende slavische Grammatik. Band II: Formenlehre und Syntax. 2nd rev. ed. [O. Grünenthal]. Göttingen: Vandenhoeck and Ruprecht. Wackernagel-Jolles, B. 1971 Untersuchungen zur gesprochenen Sprache: Beobachtungen zur Verknüpfung spontanen Sprechens. Göppingen: Alfred Kümmerle. Weinert, Regina and Jim Miller 1996 Clefts in spoken discourse. Journal of Pragmatics 25: 173–202 Zemskaja, E. A. 1973 Russkaja razgovornaja reˇc’. Moscow: Nauka.

II.

From function to form

Elena Maslova and Giuliano Bernini

Sentence topics in the languages of Europe and beyond

1. Introduction1 The concept of sentence topic is one of the most controversial linguistic ideas. The reason for this seems quite clear: this concept seems too vague to be fruitfully applied to analysis of language-specific constructions, let alone crosslinguistic investigations. Yet there exists a range of “topic-related” syntactic phenomena, which provide the language-internal motivation for this concept and apparently make it impossible to discard it altogether. The variety of these phenomena can be preliminarily illustrated by the following conversation: (1) a. b. c. d.

English [inspector] “You are not to open any letters unless you recognize the handwriting,” he said. “Everything else we’ll look at first. As to the phone calls : : :” [woman] “Your sergeant said you’d have an arrangement to get my phone monitored” (Ruth Rendell, No more dying then. An Inspector Wexford Mystery.)

In spite of their obvious structural and functional diversity, these constructions display some essential similarities: a referring expression is more or less explicitly opposed to the rest of the sentence and thereby excluded from the scope of assertion. In this chapter, we take these similarities as the central piece of evidence in favor of the crosslinguistic relevance of the concept of sentence topic and assume, as a working hypothesis, that these and similar constructions instantiate the same universal phenomenon. Our goal is to demonstrate that this hypothesis opens the possibility to account, in a unified fashion, for language-internal and crosslinguistic variation in topic encoding and for universal constraints on this variation. Section 2 discusses this hypothesis in further detail and outlines the structure of the chapter. Although the chapter focuses on the languages of Europe, the so-called topic-prominent languages (in Li and Thompson’s [1976] sense) serve as the natural standard of comparison for any typological study of topic constructions and are therefore frequently invoked in the discussion. In addition, we draw evidence from some other languages whenever generally relevant typological options cannot be exemplified by data from European languages.

68

Elena Maslova and Giuliano Bernini

2. The hypothesis of existence for sentence topic: Pro and contra The most widely accepted definition of sentence topic is “what the sentence is about” (Hockett 1958: 21), or, in a more refined formulation suggested by Gundel, what “the speaker intends to increase the addressee’s knowledge about, request information about, or otherwise get the addressee to act with respect to” (Gundel 1988: 210). The most obvious arguments against this definition (as well as against the linguistic validity of the concept itself) is that it is vague and provides no rigorous criteria for identification of the topic of a given sentence, at least without scrutinizing a broader discourse context and/or speech situation (cf. Polinsky 1999: 572; Jacobs 2001 for recent discussions). However, the concept of sentence topic (in various terminological guises) would be employed again and again for description and explanation of certain linguistic phenomena. Informally, these are phenomena that cannot be appropriately accounted for by the apparatus of propositional semantics, but favor a description in terms of a context-sensitive choice of a syntactically prominent entity (= topic) that is in some sense kept outside the scope of assertion (or another illocutionary operator); Section 3 provides an overview of these phenomena, with particular reference to European languages. The chapter assumes, as a working hypothesis, that there exists a universal phenomenon of sentence topic – roughly speaking, a mental status of referent(s) with respect to the information conveyed by the sentence – which manifests itself in sentences where reference to an entity (topic) is in some way structurally separated from the remainder of the sentence (comment) and is thereby excluded from the scope of assertion.2 The topic-comment structure of a sentence is commonly construed not as an aspect of its meaning, but rather as an aspect of how this meaning is structured for the purpose of communicative interaction, which implies that the choice of topic is guided primarily by the speaker’s assumptions about the listener’s state of knowledge. This view is perhaps most picturesquely expressed in Chafe’s (1976: 28) metaphor comparing topics and other discourse-sensitive statuses with the packaging of toothpaste: the toothpaste exists and remains essentially the same independently of whether or not it is packaged, yet it has to be packaged to be delivered to the customers and conveniently used as intended. Similarly, the meaning of a sentence exists (in the mind of the speaker) independently of whether or not it is packed into a certain topic-comment structure, but it has to be “packaged” in an appropriate way to be conveniently processed and properly understood by the listener. Morphosyntactic phenomena providing evidence for this view are overviewed in Section 3.1. However, the analysis of the full range of topic-comment sentences demonstrates that the hypothesized topic status must belong to the meaning conveyed by the sentence, rather than just being an aspect of “packaging” of this meaning (see Sections 3.2– 3.3). In Section 4 we argue, in fact, for a stronger version of this view, namely, that the topic status is grounded in the perception/construal of reality: it is not only about

Sentence topics in the languages of Europe and beyond

69

structuring information for the purpose of being communicated and processed, but also about the structure of perceptions and thoughts. The central piece of evidence against a unified approach to sentence topic is the existence of multiple “topic constructions” with different functions both within a single language and across languages (cf., for example, the distinction between “topic” and “theme” in functional grammar [Dik 1989; Bolkestein 1993: 345–346; see also Jacobs 2001 for an elaborate argumentation to this effect]). To begin with, the existence of topicality as a universal category – as a mental status of an entity with respect to the information conveyed by the sentence – must not (and in fact hardly can) entail the existence of a uniform topic encoding in each and every language. More or less significant differences between functionally similar constructions competing within a single language is what keeps happening again and again in all domains of grammar, and thus what we would expect of topic-encoding constructions as well. The genuine problem with identification of topic-encoding constructions – and more generally, with the hypothesis of existence of topic in general – is, to put the things plainly, that there are obviously no topics in the “real world.” The topic status – assuming it indeed exists – is a language-internal, or, in slightly more cautious wording, mind-internal phenomenon without obvious counterparts in the perceived reality. It is apparently for this reason that the coexistence of alternative topic-encoding constructions in some (or even most) languages can be perceived as evidence against the existence/reality of topic status, in contrast to linguistic categories more obviously grounded in (our perception of) reality (like tense, semantic roles, etc.), for which the availability of different coding options within a single language, as well as crosslinguistic semantic differences, are easily accommodated by linguistic theories. For instance, our belief in the existence of time is firmly grounded in our perception of reality and cannot be shaken by language-internal and crosslinguistic variation in its grammatical encoding. With topics, it is obviously not the case. Nonetheless, if the topichood is recognized as an aspect of meaning, the existence of alternative topic-encoding options and the functional differences between topic constructions within and across languages cannot be plausibly construed as evidence against the existence of this meaning, since essentially similar variation in coding options exists for any other aspect of meaning. However, the lack of obvious “real world” counterparts of topics means that our understanding of what the topic status of a referent actually “means” (i.e., what it contributes to the meaning of the sentence) can only be based on the analysis of functions of language-specific topic constructions, at least until neural studies of the human brain are developed enough to provide independent evidence.3 The only palpable common property of all topic constructions is that they allow the speakers to activate a referent without including it into the scope of assertion (or another illocutionary operator), that is, to separate reference to an entity from the act of assertion. The only thing this can tell us about the hypothesized mental status of topic is that if

70

Elena Maslova and Giuliano Bernini

an entity has this status with respect to a piece of information, then activation of its mental representation is essential for interpretation of this information but is not (or need not be) a component of this piece of information (more precisely, of the proposition encoding this piece of information). However, different structural types of topic constructions have been described in terms of specific relations established between the topic and its comment: apart from (i) the “aboutness” topic function (below, ˛topic), the functions ascribed to topics include (ii) delimiting the frame within which the main predication holds (Chafe 1976: 50; Li and Thompson 1976; Jacobs 2001) and (iii) anchoring the predication in the context, thus linking it to the listener’s presumed state of knowledge (Mathesius 1929; Daneš 1964; Chafe 1976, 1987; Prince 1981; Gundel 1988; inter alia). The hypothesis of semantic grounding of topicality developed in Section 4 suggests that all aspects of topicality are in fact relevant for all types of topic constructions, that is, the hypothesized mental status in some sense invokes all three relations between an entity and a predication. We assume, however, that the intuitive notion of “aboutness” (˛-topic) is as close to the hypothesized mental status of topic as reasonably possible at the present state of knowledge and will occasionally invoke the aboutness-based intuitions in the following discussion, without attempting to define “aboutness” in more explicit and rigorous terms.4 Although the anchoring function is an inherent aspect of the hypothesized phenomenon of topic, language-specific topic constructions can impose different constraints on the assumed degree of activation of the topic referent in the listener’s mind by the time of utterance (Chafe 1976, 1987; Prince 1981; Geluykens 1989, 1992; Lambrecht 1994: 77–113; inter alia). Crosslinguistically, such constraints correlate, to a certain degree, with structural properties of topic constructions; for example, left-dislocations tend to be associated with a relatively low degree of activation. However, the language-internal and crosslinguistic variation in the degree of activation of topic referents for different topic constructions and, more broadly, in the ranges of discourse contexts in which these constructions are felicitous pose problems for the unified approach to sentence topic only if this concept itself is construed as a purely packaging phenomenon. The approach advocated here presupposes a distinction between the topic-comment structure as an aspect of the meaning to be conveyed, on the one hand, and the way this meaning is expressed, on the other. Whether and how the topicality of a referent is encoded, is expected to depend on the listener’s assumed state of knowledge and attention, as well as on the salience of the topic-comment structure in the overall meaning of the sentence (Section 5.1). In this respect, topicality does not differ from any other not fully grammaticalized meaning: the fact that the way a meaning is encoded depends on the context cannot and does not entail that different contexts correspond to different meanings. In Section 5, we demonstrate that a consistent distinction between the concept of sentence topic and language-specific constructions that can express this meaning allows for a unified explanation of language-internal and crosslinguistic variation in context-

Sentence topics in the languages of Europe and beyond

71

sensitive constraints associated with different topic constructions, on the one hand, and for attested constraints on such variation, on the other.

3.

Phenomenology of sentence topics: An overview

3.1. Packaging topics The first class of topic-related structural phenomena can be described in terms of “ packaging variants” ( Chafe 1976: 28) of the same contents: a sentence token consisting of a referring expression and an assertion is paradigmatically opposed to one or more propositionally equivalent sentences which would organize the same propositional contents in a different way, that is, single out another referring expression or none at all. The following examples from Russian illustrate some characteristic paradigmatic oppositions of this type (sentences in a. are authentic examples from Russian texts, sentences in b. represent alternative packaging options). (2) a.

b.

(3) a.

b.

Voice oppositions: Russian5 Ves’ nižnij ètaž tëtkin-ogo dom-a byl whole ground floor aunt:POSS-GEN house-GEN was zanjat restoran-om. occupied:M.SG restaurant-INST ‘The whole ground floor of the aunt’s house was occupied by a restaurant.’ (Mikhail Bulgakov, Master i Margarita) Restoran zanima-l ves’ nižnij ètaž tëtkin-ogo restaurant occupy-PST.SG.M whole ground floor aunt:POSS-GEN dom-a. house-GEN ‘The restaurant occupied the whole ground floor of the aunt’s house.’ Linear order: Russian Tak-uju že illjustrativn-uju funkci-ju ime-jut such-ACC same illustrative-ACC function-ACC have-PRES.3PL sxem-y R. Van Valin-a i W. Foley. scheme-PL.NOM R. Van Valin-GEN and W. Foley ‘The same illustrative role is played by R. Van Valin and W. Foley’s schemes.’ (Aleksandr Kibrik) Sxem-y R. Van Valin-a i W. Foley ime-jut scheme-PL.NOM R. Van Valin-GEN and W. Foley have-PRES.3PL tak-uju že illjustrativn-uju funkci-ju. such-ACC same illustrative-ACC function-ACC ‘R. Van Valin and W. Foley’s schemes have the same illustrative function.’

72

Elena Maslova and Giuliano Bernini

(4) a.

b.

Dislocation: Russian Postojannoe vnimanie k kul’tur-e mal-yx permanent attention to culture-DAT small-GEN nacional’nost-ej èto-mu nas uˇc-it otec. ethnicity-PL.GEN this-DAT us.ACC teach-PRES.3SG father ‘Permanent attention to the the culture of national minorities, that’s what Father teaches us.’ (Semen Lipkin, Dekada) nas postojann-omu vnimani-ju k Otec uˇc-it father teach-PRES.3SG us.ACC permanent-DAT attention-DAT to kul’tur-e mal-yx nacional’nost-ej. culture-DAT small-GEN ethnicity-PL.GEN ‘Father teaches us (to pay) permanent attention to the culture of national minorities.’

The following examples illustrate similar subparadigms of packaging variants from other European languages: (5) a. b. (6) a.

b.

(7) a.

b.

English (Foley and Van Valin 1985: 299) The man was kissed by the woman. The woman kissed the man. Hungarian (Kiss 1995: 208–209) Évát János várta a mozi el˝ott. Eve.ACC John waited the cinema in.front.of ‘Eve was waited for in front of the cinema by John.’ János Évát várta a mozi el˝ott. John Eve-ACC waited the cinema in.front.of ‘John waited for Eve in front of the cinema.’ German (Jacobs 2001: 642) Peter, ich habe ihn heute nicht getroffen. Peter I PFV.1SG he.ACC today NEG meet:PST.PART ‘Peter, I have not met him today.’ Ich habe Peter heute nicht getroffen. I PFV.1SG Peter today NEG meet:PST.PART ‘I have not met Peter today.’

In these and similar examples the structural differences between paradigmatic alternatives can be described in terms of singling out one component of the main proposition by putting it into a syntactically prominent position, which is syntagmatically opposed to the rest of the sentence (the subject position in (2) and (5), the clauseinitial position in (3) and (6), and the clause-external “topic” position in (4) and (7)), the approach most broadly applied in discussions of topic-comment structures.

Sentence topics in the languages of Europe and beyond

73

Although the very existence of packaging variants demonstrates a certain degree of independence between the topic-comment structuring and the propositional arrangement of information, the relation between these structures is constrained insofar as the “packaging” topic expression always specifies a variable in the propositional structure of the message (although not necessarily an argument variable of the main predicate). Indeed, it is this constraint that guarantees that alternative packaging variants with exactly identical propositional contents are possible at all (that is, that the topic expression can be present in a sentence with the same propositional contents in a nontopical structural position). Moreover, many language-specific constructions that ensure availability of alternative packaging variants impose further constraints on the relationship between the topic-comment structure and the propositional structure (most obviously, this concerns voice constructions). The existence of topic-oriented “packaging variants” with the same propositional contents gives one of the most important pieces of evidence in favor of the hypothesis of sentence topic and seems to indicate that the topic status resides not in the “contents” to be communicated (it apparently remains constant independently of which entity is chosen as the topic), but in the way these contents are encoded for the purpose of communication (i.e., for transmission of the information to the listener). In this sense, it is similar to other “packaging statuses” or “pragmatic roles” of NPs (e.g., to the role of narrow focus). According to this view, an entity would assume the topic status only in the context of the speaker-hearer interaction: it is a part of how the speaker “packages” the semantic contents for the listener and the choice of topic is thus guided primarily by the speaker’s assumptions about the listener’s state of knowledge (Chafe 1976, 1987; Prince 1981; Gundel 1988; inter alia).

3.2. Hanging topics The second class of topic-related phenomena comprises so-called “hanging” topics, that is, sentences in which the topic expression is juxtaposed to a clause-like component denoting the main proposition and does not specify a variable of this proposition; in other words, the semantic (“real world”) relation between the topic and the state of affairs described by the comment is not linguistically encoded and thus does not belong to the propositional structure of the sentence. This class of sentences received particular attention in linguistic typology due to so-called “topic-prominent” languages (Li and Thompson 1976), where this way of organizing sentences is fully conventionalized; the topic expression either bears an overt morphosyntactic topic marker (an adposition (8) or an affix (9)), or remains unmarked, as in (10).

74

Elena Maslova and Giuliano Bernini

(8)

Japanese (Chen 1996: 396) Nihon wa syuto ga sumi-yo-i. Japan top capital nom live-good-prs ‘As for Japan, its capital is a good place to live.’

(9)

Korean (Li and Thompson 1976: 468) Piheng1-nin 747-ka kh1-ta airplane-TOP 747-SBJ big-STAT ‘Airplanes, the 747 is big.’

(10)

Chinese (Chen 1996: 402) haixian, wo zui ai chi longxia. seafood I most love eat lobster ‘As for seafood, I like lobster most.’

Sentences with unmarked hanging topics appear to occur (at least) in informal registers of all languages, even if they are not fully conventionalized and are therefore avoided or even prohibited in more formal registers; the availability of this structure probably constitutes a universal property of “unplanned discourse” (Gundel 1988: 238–239); at least, whenever this sort of data is scrupulously investigated, some instances of hanging topics are likely to be found. The following examples illustrate this phenomenon for some European languages: (11)

Italian [looking at the fruit (casual conversation)] La frutta – t’ arrangi. DEF fruit 2SG.OBJ.CLT help.yourself:IMP ‘[If you like] fruit, help yourself.’ (field notes)

(12)

Russian (Miller 1992: 96) Sobak-a – vsegda pol-y grjazn-ye. dog-NOM.SG always floor-PL dirty-PL ‘The/a dog, the floors are always dirty.’

(13)

English (Lambrecht 1994: 193) [a sentence produced by a lecturer in an introductory linguistic course] Other languages, you don’t just have straight tones like that.

On the other hand, all European languages seem to have more or less conventionalized expressions that can be used to introduce topics without integrating them into the proposition; a list of such expressions from a number of languages is given in (14).

Sentence topics in the languages of Europe and beyond

(14)

75

Topic-introducing expressions in European languages: Italian per quanto riguarda T for what concerns T English as for/to T as regards/concerns T German was T angeht/betrifft what T concerns Dutch wat T betreft what T concerns Swedish på tal om T on mention about T i fråga om T in question about T Russian cˇ to do T what to T(GEN) cˇ to kasaetsja T what touches T(GEN) Hungarian T-val/vel kapcsolat-ban X-COM relation-LOC T illet˝o-en T(ACC) regarding-ADV

Whereas formally the topic expressions are integrated into the propositional structure by means of a relational expression (including verbs with meanings like concern, touch, regard, etc.), the actual semantic relation between the topic and the state of affairs described in the comment is not encoded, so in this sense the topics are “hanging,” for example: (15)

German (Jacobs 2001: 654) Was Goethes 250. Geburtstag betrifft: Das what Goethe’s 250 birthday concern.3SG DEF Stadttheater plant eine ungekürzte Faust-Aufführung. municipal.theater plan.3SG INDF unabridged Faust-performance ‘As for Goethe’s 250th birthday, the municipal theater plans an unabridged performance of “Faust”.’

(16)

Russian ˇ Cto kasaetsja professional’n-oj terminologi-i, to ja what touches professional-GEN terminology-GEN, then I sovsem prevrašˇca-jus’ v inostrank-u. completely turn.into-PRES.1SG in foreigner-F.DAT ‘As for work-related terminology, I am turning into a complete foreigner.’ (from an Internet chat) ˇ Cto že do praktiˇcesk-ogo programmirovani-ja – sprosite what PRT to applied-GEN programming-GEN ask:IMP:2PL ešˇcë, a javljaetsja li nauk-oj slesarnoe also CNJ constitutes INTR science-INST locksmith:ADJ delo. profession ‘As for software engineering, you can as well ask whether locksmithing is a science.’ (from an Internet chat)

a.

b.

76

Elena Maslova and Giuliano Bernini

(17)

Catalan (Vallduví 1995: 90) Quant al Joan i la Isidora no t’ho sé dir doncs as.for the Joan and the Isidora no IO.O know:1SG say since el Joan el veiem ben poc. the Joan OBJ see:1PL quite little ‘As for Joan and Isidora, I can’t say, since Joan we see very little of.’

In sentences of this class, the topic expression cannot be taken as something “singled out” from the main proposition, since it does not refer to any of its elements to begin with: if not for being in the topic position, it would have been just absent from the sentence. In other words, the exact semantic contents of a sentence with a hanging topic cannot be expressed in any other way: this would require integrating the topic expression into the proposition, which, if possible, would involve some specification of the semantic relation between the topic and the situation being described and thus inevitably modify the propositional contents of a sentence in a more or less significant way. This entails that hanging-topic sentences cannot be adequately described in terms of “packaging variants” of independently existing semantic contents. It follows that the topic status cannot be viewed as an aspect of “packaging” of pre-existing propositional contents. Rather, it must be a component of information/thought to be communicated; in other words, an entity assumes the topic status not after the propositional contents to be communicated is “created” (in the mind of the speaker), but rather before or in the course of the process of its creation (Kaseviˇc 1988: 238–240).6 In this respect, the phenomenon of topic saliently differs from that of focus: there seem to be no “hanging” foci in natural languages; in other words, a reference to an entity can be included into the scope of assertion of a sentence only if it constitutes a component of its propositional structure.

3.3. Topic constructions The central piece of evidence in favor of the hypothesis that hanging and packaging topics manifest the same phenomenon comes from the fact that they often instantiate the same language-specific construction. In topic-prominent languages, the construction used to introduce hanging topics also functions as the major strategy of singling out elements of the main proposition. In the following set of Japanese examples, the construction with the postposition wa singles out packaging topics in the first two sentences and introduces a hanging topic in the last one. (18) a.

Japanese (Shibatani 1990: 262, 275) Hi wa nobor-u. sun top rise-prs ‘The sun rises.’

Sentence topics in the languages of Europe and beyond

b.

c.

77

Hanako wa Taroo ga eigo o osieteita. Hanako top Taro nom English acc teaching:pst ‘Hanako, Taro was teaching English (to her).’ Tori wa mesu ga tamago o umu. bird top female nom egg acc lay ‘A bird, a female lays eggs.’

In European languages, the topic-introducing expressions like in (14) are certainly not constrained in terms of the relation between the topic expression and the main proposition; along with hanging-topic sentences, they can also be used for singling out elements of the main proposition, which are then cross-referenced within the comment. This is illustrated by the following examples: (19)

Irish (Ó Siadhail 1988: 196, 256) Ach maidir le fiaclóirí na cathrach seo, but as to dentists ART:GEN.F city:GEN this na hÉireann, ní dhéanfainn aon ART:GEN.F Ireland:GEN NEG do:COND:1SG one as-tu. from-3PL ‘But as far as the dentists of this city or the dentists of cerned, I would not trust them!’

nó fiaclóirí or dentists mhuinín trust

Ireland are con-

(20)

Italian Per quanto riguardava la rivoltella, di sicuro Enzo for what concerned the revolver:SG.F of sure Enzo l’ aveva ereditata. CLT.ACC.3SG.F had:3SG inherited ‘As for the revolver, Enzo had certainly inherited it.’ (field notes)

(21)

Russian ˇ Cto do ošibk-i, to Efremov sam pišet o nej v what to error-GEN then Efremov RFL writes about 3SG.OBL in predislovii k svoemu sobraniju soˇcinenij. introduction:OBL to RFL.POSS:DAT collection:DAT works:GEN ‘As for the error, Efremov writes about it himself in the introduction to his collected works.’ (from an Internet chat)

The same is true for constructions with unmarked clause-external topics, cf. (11)– (13) and (22)–(27).

78

Elena Maslova and Giuliano Bernini

(22)

Spanish (Downing 1997: 139) La radio no la suelo escuchar. DEF radio NEG it tend:1SG listen.to ‘The radio, I don’t usually listen to it.’

(23)

Italian Il tartaro il dentista lo toglie ma poi the tartar.SG.M the dentist CLT.ACC.3SG.M removes but then si riforma. RFL forms.again ‘The dentist removes the tartar, but it comes back again.’ (TV advertising spot: a cartoon appears on the screen, showing teeth and a scalpel removing the tartar from them.)

(24)

Italian (Benincà et al. 1988: 132) Noi altre monache, ci piace di sentir le storie per we others nuns us delights of hear:INF the stories for minuto. trivial.detail ‘As for nuns as we are, we enjoy hearing stories with their trivial details.’ (Alessandro Manzoni, Promessi Sposi, IX)

(25)

Russian (Fici Giusti et al. 1991: 321) Ta, kotor-aja zdes’ stojal-a that:NOM.F.SG which-NOM.F.SG here stood-F.SG lamp-a, ja eë ne bral. lamp-NOM.F.SG I 3SG.F.ACC NEG took (M.SG) ‘The lamp that used to stand here, I didn’t take it.’ (from a recording of oral speech)

(26)

English (Miller 1992: 95) The driver, you get a good laugh with him.

(27)

French (Cadiot 1992: 63) Les enfants j’ leur pardonne tout. the children I.CLT them.CLT forgive:1SG all ‘The children, I forgive them everything.’

More generally, if a language-specific construction can be used for introducing a hanging topic, it can also be used for singling out packaging topics; it simply does not impose any constraints on the role of topic in the proposition (or the existence thereof). Thus, insofar as hanging-topic sentences can be assumed to be a universal phenomenon, so can the sentence structure that combines a reference to an entity (topic expression) and a predicated proposition (comment), independently of

Sentence topics in the languages of Europe and beyond

79

whether or not the topic expression is integrated into the proposition, or, in other words, whether the semantic (“real world”) relation between the topic and the situation described by the proposition is encoded linguistically. The status of such structures as “constructions” – insofar as this term implies some degree of conventionalization – may be controversial in some languages. It can be hypothesized, however, that both structures with unmarked clause-external topic expressions and structures with topic expressions introduced by lexically transparent expressions like in (14) belong to the stock of what Harris and Campbell (1995: 54–56, 72–75) call “exploratory expressions”, that is, to the structures that, if not yet conventionalized, are still universally available and can be “pushed” (by circumstances) towards conventionalization at any time. Constructions of this class will be referred to below as h-topic constructions (where h stands for “hanging”). To sum up the discussion so far, the crosslinguistic availability of h-topic constructions and their ability to serve as a uniform topic encoding in topic-prominent languages constitutes a crucial piece of evidence in favor of the hypothesis of existence of sentence topic, whereas the existence of hanging-topic sentences demonstrates that the hypothesized mental status of topic cannot be adequately described in terms of “packaging,” but should be treated as an aspect of meaning conveyed by the sentence. On the other hand, many languages (in particular, the languages of Europe) have multiple topic constructions, some of which are constrained to packaging topics, or to a subset thereof (e.g., only to certain [types of] arguments of the main proposition). Such constructions fall in two major classes. In one class of constructions, the topic expression bears the same marking of its role in the propositional structure as in those packaging variants where the propositionally identical element is inside the scope of assertion; the topic encoding amounts to linear and/or intonational separation of the topic expression and the comment (f - topic constructions, where f stands for “fronting”); f -topic constructions are exemplified in (3) and (6) above. The second class subsumes so-called relation-changing constructions, or, more broadly, subject-predicate constructions (s- topic constructions; see (2) and (5)).7 The existence of such constructions – along with h-topic constructions – entails that some components of the propositional structure can be encoded as topics in two or more different ways; for example, the Russian sentence in (25) can be paraphrased as in (28), whereby both the propositional contents and (presumably) the topic-comment structure of the original example are retained: (28)

Russian Tu lamp-u, kotor-aja zdes’ stojal-a, that:ACC.F.SG lamp-ACC.F.SG which-NOM.F.SG here stood-F.SG ja ne bral. I NEG took (M.SG) ‘The lamp that used to stand here, I didn’t take it.’

80

Elena Maslova and Giuliano Bernini

This is an instance of f -topic construction: the only topic-indicating device is the (object-initial) linear order: the topic expression gets the case form determined by its role in the main proposition, and the comment contains no pronouns cross-referencing the topic. In Russian, then, the h-topic construction “competes” with the f topic construction for the encoding of packaging topics. The f -topic construction is considerably more conventionalized – whereas (25) is clearly perceived as resulting from the lack of “planning” and is only acceptable in informal colloquial speech, (28) is unmarked in this respect. The following pair of examples illustrates a similar competition between the h-topic construction and the s-topic construction: (29) a.

b.

Russian Politkorrektnost’ eë vzroslye izobre-li. political.correctness it:ACC adults:PL.NOM invent-PST.PL [Children bully one another everywhere.] ‘Political correctness, adults have invented it.’ (from an Internet chat) Politkorrektnost’ izobretena vzrosl-ymi. political.correctness invent:PASS adults-PL.INST ‘Political correctness has been invented by adults.’

Clearly, whenever alternative topic constructions are available, their functions are bound to differ to some extent. Apart from the register-based constraints, a specific construction can be constrained in terms of the role of the topic expression in the propositional structure of the sentence: in Russian, for example, the s-construction (as in (29b)) is available only for (a subset of) core arguments of the main verb; for other topics, the range of available coding options is limited to h-constructions and f -constructions. On the other hand, topic constructions can be associated with different types of discourse contexts; in particular, they can impose different constraints on identifiability and activation of the topic referent (Chafe 1976, 1987; Gundel et al. 1993; inter alia). As preliminarily discussed in Section 2, this variation does not contradict the hypothesis of existence for sentence topic, insofar as topicality is thought of as a component of meaning rather than an aspect of how this meaning is packaged for the purpose of communication. On the contrary, this hypothesis predicts the existence of alternative coding options and allows for a unified account of the attested constraints on language-internal and crosslinguistic variation of topic encoding. We return to this question in Section 5.

3.4. Non-nominal topics In the body of this chapter (Sections 4–5), we focus on the conceptually simplest and most common type of topic-comment sentences, where the topic referent is a

Sentence topics in the languages of Europe and beyond

81

stable entity referred to by a nominal expression and the comment is a description of a state of affairs. However, the hypothesis of existence for sentence topic does not rule out that this mental status can be assigned to a state of affairs, as in the following examples:8 (30) a.

b.

Italian (Benincà et al. 1988: 189) Che si parta tutti insieme, non è una bella that IMPR leave:SBJ:3SG all together NEG is a nice soluzione. solution ‘That we should leave all together isn’t a good solution.’ Che arrivavate oggi, non me l’ aveva detto nessuno. that arrive:IMPF:2PL today NEG to.me it had:3SG told nobody ‘Nobody had told me that you would arrive today.’

The ability of predicates to serve as topic expressions is more problematic, since they do not refer (cf. Lambrecht 1994: 76); accordingly, it is unclear what exactly can be thought of as the topic of such sentences. It seems, however, that the lexical predicate of the main predication in the topic position can activate a referent, namely, the state of affairs being described, whereby the statement conveyed by the comment is reduced to modality, most often, simple assertion or negation. This is possible if all relevant partipants of the state of affairs being described are recoverable from the context, so that the predicate (in an infinitival form) suffices to activate its mental representation. Consider the following example: (31)

Italian (field notes; casual conversation) A: l’ ha trovato? it.CLT have:3SG.POLITE found B: per ess-er-ci, c’ er-a ma er-a for be-INFIN-there there be:IMPF-3SG but be:IMPF-3SG chiuso. closed [A is waiting at the side of a street. B passes and asks A where he can find a tobacco shop in the neighborhood. A shows it to B. After a short time, B comes back to the spot where A is still waiting.] A: ‘Did you find it?’ B: ‘As to its being there, there it was, but it was closed.’

In this example, the infinitive in the topic position in B’s answer refers to the state of affairs of a tobacco shop being where A indicated, whereas the comment asserts that this state of affairs does indeed take place (compare, for example, the English translation, where the topic expression fully specifies the state of affairs). However, the description of the state of affairs (including the predicate) must be copied in the

82

Elena Maslova and Giuliano Bernini

comment, since otherwise it could not be construed as an assertion. In the following example from Russian, the state of affairs referred to by the predicate (the speaker having grown stupid) is negated in the comment.9 (32)

Russian Ja stal cˇ udakom, njan’ka : : : Poglupet’-to I became whimsical.INST nanny PFV.grow.stupid:INF-PTCL ja ešˇcë ne poglupel, Bog milostiv, mozgi na svoëm I yet NEG grew.stupid God merciful brains on their meste no cˇ uvstva kak-to pritupil-i-s’. place:LOC but feelings somehow dulled-PL-REFL ‘I have become whimsical, nanny. As for growing stupid, I have not yet, by the grace of God, my brain is at its place, but my feelings are dulled.’ (Anton Chekhov, Djadja Vanja)

Example (33) shows the use of an auxiliary instead of a lexical copy of the predicate. (33)

English She had been opening letters. The paper-knife was there on the desk. Carpenter seized it and drove it in. She may not have meant to kill, but kill she did. (James, P. D., A Certain Justice, London, Penguin, 1997, p. 269).

In examples (34)–(35), the comment asserts a modality expressed by a finite modal verb, without lexical copy of the topic predicate in (34) and with lexical copy of the topic predicate in (35). (34)

German Mein-e Kind-er versteh-en zwar Ladino, aber sprech-en my-PL child-PL understand-3PL PTCL Ladino but speak-INF könn-en sie es nicht. can-3PL they it NEG ‘My children understand Ladino, but they can’t speak it.’ (Ho¸s geldin. Die Türkei in kleinen Geschichten, Celal Özcan, Munich, Deutscher Taschenbuch Verlag, 1994, 130–131; example from the German translation of the Turkish original)

(35)

Spanish que no la Bueno, pero es una decisión well but is a decision:F.SG that NEG 3SG.F.ACC.CLT toma qualquiera, ¿eh? [: : :] Tomar-la se takes anybody eh take:INF-3SG.F.ACC.CLT IMPR puede tomar. can take:INF

Sentence topics in the languages of Europe and beyond

83

‘Well, but it is a decision that anybody can take: as for taking it, one can take it.’ (El habla culta de la ciudad de Buenos Aires. Materiales para su estudio. Tomo 1, Universidad Nacional de Buenos Aires. Facultad de Filosofía y Letras. Buenos Aires, 1987, p. 19). In some languages, predicates in the topic position can be expressed by bare infinitives, see (32), (33), (34); this is also possible in Hungarian (36) and in Yiddish (Lockwood 1995: 117). (36)

Hungarian (Jean Perrot, pers. comm.) figyel-ni figyel-t. pay.attention-INF pay.attention-PAST(3SG) ‘As to paying attention, he did pay attention.’

In other languages, infinitival topic expressions are accompanied by a preposition meaning ‘for,’ for example, in Italian (31) and French: (37)

French Ça sonna. Pour sonn-er, ça sonna. it rang:3SG for ring-INF it rang:3SG ‘[I called the jeweler from the nearest pub, Ranelagh 89–10.] It rang. As for ringing, it rang.’ (Léo Malet, Les enquêtes de Nestor Burma et les nouveaux mystères de Paris, Vol. 2. Paris, Robert Laffont, 1986, p. 195)

Whereas further crosslinguistic research is obviously needed for the in-depth understanding of this construction type, it seems plausible to assume, for the time being, that infinitives in topic positions can be used to encode states of affairs as topics.

4.

Semantic grounding of the topic-comment structure

4.1. Introductory remarks This section links the classification of topic constructions introduced in Section 3.3 with three distinct semantic (“real world”) relations between the state of affairs described in the comment and the topic referent: – s-topic constructions are grounded in the relation between event and its primary participant; – h-topic constructions are grounded in the relation between event and an entity closely related to one of its participants (“reference point”); – f -topic constructions are grounded in the relation between event and its spatial/ temporal frame.

84

Elena Maslova and Giuliano Bernini

In spite of their obvious differences, these semantics-based templates are similar in that they combine the description of an event with a reference to a (relatively) stable object, which is essential for topic-comment information structure. Crosslinguistic evidence suggests that these semantics-based structural templates are often employed beyond their primary semantic domains, and it is in such cases that their topic-encoding potential figures most prominently. Although these semantic templates are presumably universal, their topic-encoding potential obviously varies from language to language: it is often the case that in a given language, one template is extensively employed for topic encoding, while others are used in this function only marginally (if at all). In order to avoid possible misunderstanding, it should be stressed that it is not assumed that each of these structures is recruited for the topic-encoding function in all languages; the generic statements that may occur in what follows are to be understood as referring only to the universal availability of these options. Another necessary preliminary remark concerns affiliation of particular language-specific topic constructions with one or another structural template. It is not always the case that a single construction identified on the basis of language-specific properties can be affiliated with a single semantic template in an uncontroversial manner. These problems are discussed in Section 5.3.

4.2. Primary participants The first widely attested morphosyntactic phenomenon associated with the concept of topic is “grammatical subject.” The complex and crosslinguistically variable relations between “subject,” “topic,” and “actor” (or, in a more traditional frame of reference, grammatical, logical, and semantic subject) have been extensively discussed in the literature. Indeed, the well-established existence of a crosslinguistically relevant class of morphosyntactic structures that are grounded in the event structure and, at the same time, impose a certain “default” information structure serves as the starting point for the more general hypothesis on the phenomenon of sentence topic advocated here. This section represents, therefore, a summary of the previous findings (reformulated within the present framework), rather than a new contribution to the domain. Since the universal validity of the concept of “grammatical subject” – if defined in (morpho)syntactic terms – is a matter of considerable controversy, we shall, for the purposes of the present chapter, define “subject” as a morphosyntactic role – essentially, as a cluster of morphosyntactic features – that encodes the most agentive (in particular, the sole) participant of the event signified by the verb. For European languages, this semantically-oriented definition seems to cover essentially the same phenomenological domain as more technically sophisticated and syntactically-

Sentence topics in the languages of Europe and beyond

85

oriented definitions of subject do, so the change of the definition is not likely to pose any problems. The relevant morphosyntactic features would include case form (where applicable), verb agreement, and linear position, although the relative significance of these features in the identification of the subject NP obviously differ from language to language. Crucial for the further discussion is the assumption that, as a rule, the choice of a relational expression for the role of predicate determines a single element of its valence frame that has a privileged access to the morphosyntactic role of subject, although some other semantic roles can, in some languages, be encoded in the same way with appropriate additional marking within the predicate (most obviously, in passive constructions). This semantic role is referred to here as “primary participant” (PP) role, hence the label “PP-structure” below. It is widely acknowledged that the subject encoding of NP strongly correlates with topicality of its referent; that is, the overwhelming majority of subjects in discourse happen to be outside the scope of assertion and indeed refer to “what the sentence is about” in some intuitively clear sense (or, to put it in a somewhat weaker form, are most naturally interpreted as “what the sentence is about” without additional clues [Lambrecht 1994: 132]). The correlation between subjecthood and topicality seems so strong, indeed, that the aboutness relation – rather than the primary participant role – can be (and often was) used as the basis of a functional definition of subject:10 So far as I can see at present, the best way to characterize the subject function is not very different from the ancient statement that the subject is what we are talking about : : : it is likely that one of the main ways in which new knowledge is communicated – perhaps even the only way – is by identifying some particular as a starting point and adding to the addressee’s knowledge about it : : : We might call this the “addingknowledge-about” hypothesis regarding the functioning of subjects : : : (Chafe 1976: 46–47) There seem to be three universal factors contributing to this correlation: first, the primary participants are frequently human, and humans presumably universally tend to talk about themselves more frequently than about other things (Hawkinson and Hyman 1975; Comrie 1981: 116–123; Kozinskij and Sokolovskaja 1984; inter alia). Secondly, since the sole participants are also (by definition) primary, the primary participant is frequently just the only possible candidate for being the topic (cf. Lambrecht 1994: 132–133). Finally, the topic referent (if any) and the state of affairs to be described (along with the most appropriate relational expression for this description) are hardly chosen independently (in the speaker’s mind): the would-be PP-role (as opposed to a more peripheral role) of the topic referent is likely to be one of the factors that affect the choice of the appropriate propositional frame for the comment (the effect most clearly observed in the phenomenon of lexical conversion).

86

Elena Maslova and Giuliano Bernini

Obviously, these factors determine the correlation of the topicality of a referent with its (primary) participant role, rather than with a specific morphosyntactic encoding; in this sense, the association between topicality and subjecthood (taken as a cluster of morphosyntactic features) is but an epiphenomenal implication of the correlation between topicality and primary participation. The autonomous role of subject-predicate structure in topic-comment structuring of propositions can be visible only in sentences where the primary participant is not the topic referent. The first class of such sentences comprises sentences without structural topics, that is, so-called focus-presupposition (“subject-focus”) and thetic (or “broad focus”) sentences, for example: (38) a.

b.

Russian Èto sdelal ja! this make-PST.SG I ‘That’s me who did it.’ Zvonil mužˇcina. ring-PST.SG man ‘A MAN rang.’

Languages usually have special morphosyntactic structures and/or intonational patterns signaling that the referential identity (or kind affiliation) of the primary participant lies within the scope of assertion; in such structures, the primary participant can but need not retain the role of subject, that is, under the definition accepted here, the morphosyntactic features encoding primary participation (for an overview, see, for example, Sasse [1987 and this volume] or Kiss [1998]; clear examples of both alternatives are provided by the English translations of Russian sentences in (38)). The Russian examples in (38) illustrate an “intermediate” strategy most common in Russian (as well as in many other languages), namely, putting the primary NP in an unusual (in this case, clause-final) linear position. The NP thereby loses one morphosyntactic feature of subject, which plays a relatively marginal role in the identification of the primary participant in Russian11 and retains the others (the nominative case form and the control of verb agreement). Obviously, inasmuch as nontopical primary participants regularly lose some coding properties of primary participation, the language-specific correlation between these morphosyntactic features and topicality is strengthened beyond the level determined by the universal correlation between topicality and primary participation; accordingly, the role of these morphosyntactic features in the encoding of information structure is not confined to their role in the encoding of the participant role most commonly associated with topicality. According to Kiss’s recent study, nontopical primary participants consistently lose at least some coding properties of subjects in all European languages (Kiss 1998: 693–700). On the other hand, a language can have a more or less broad range of constructions that involve encoding of a nonprimary participant as the subject, and topicality

Sentence topics in the languages of Europe and beyond

87

often (or even universally) figures as one of the factors motivating the choice of such constructions in spite of their relative morphosyntactic complexity (see Comrie 1981: 116–123; Keenan 1985: 243–247; Foley and Van Valin 1985: 299–334). Sentences with a nonprimary referent “promoted” to the subject role and the primary participant (if specified at all) “demoted” to the clause periphery combine both phenomena. Notably, the choice of subject in such a sentence is more likely to be interpreted as a topic-encoding device than in an unmarked sentence with the primary participant in the subject role (see, for example, Davison [1984] about subject as one of the topic-defining positions and Gundel [1988] about “subject-creating” topic constructions). To give just one example of this well-known phenomenon, the Russian passive construction in (39a) strongly suggests that the queue is the topic of this sentence, whereby the primary participant (the speaker) is demoted to the status of reporting witness rather than the sentence topic and the (natural) protagonist of the story (an interpretation further supported by the overall context of the narrative, which is about a visit to the central office of KGB in the 1930s): (39) a.

b.

Russian Oˇcered’ byl-a mnoju zanjat-a ešˇcë s queue was-SG.F me.INST reserve.PASS-SG.F already from veˇcera. evening ‘The queue, I have been keeping my place since the evening.’ (Lydia Chukovskaya, Spusk pod vodu) Ja zanjal-a oˇcered’ ešˇcë s veˇcera. I reserved-SG.F queue already from evening ‘I have been keeping my place in the queue since the evening.’

The use of an unmarked (active) construction (39b) in the same context would be distinctly less marked in terms of topic encoding; although it is likely to be interpreted as information about the speaker, it does not preclude that the queue is ˛-topic (cf. Section 2). The reason for this difference in interpretation seems obvious enough: there can be several factors that might favor putting reference to the speaker into subject position: the semantic role of actor, first person, and, among others, topicality; accordingly, faced with an active sentence, the listener can assume that the first two factors (and not topicality) motivate the choice of subject. In contrast to this, the topic status seems to be the only possible motivating factor for encoding the queue as the subject; accordingly, the listener is led to infer that topicality must have overridden all other factors, and thus singling out the queue as the topic was indeed the speaker’s intention. According to the hypothesis advocated here, these observations suggest that the PP-structure can function as a topic-encoding device beyond its own semantic domain, that is, a linguistic structure grounded in a salient aspect of event structure (primary participant vs. action/state) can lend itself for the encoding

88

Elena Maslova and Giuliano Bernini

of information structure (topic vs. comment). Just as the very existence of such constructions in the grammar of a language indicates that the function of the subject role does not amount to primary participation, so the actual use of PP-structure beyond its semantic (role-oriented) domain articulates the information structure more clearly than the unmarked choice of subject. It should be stressed that this hypothesis is not intended as a hypothesis of diachronic development of specific subject-predicate constructions; that is, it does not imply that the “original” meaning of primary participation associated with the subject gets extended to the information-structuring function; in other words, the association of (language-specific) “subjects” with topicality is not taken to arise as a result of grammaticalization of, say, a passive construction. In fact, it is quite possible that a language-specific subject-predicate construction may result from conventionalization and grammaticalization of a former h-topic construction (Givón 1976; Shibatani 1991; inter alia), whereby the topic-encoding potential of this construction is diachronically reduced (rather than extended). Our idea is rather that this aspect of event structure (primary participant vs. state/action) – with its obvious crosslinguistic and presumably language-external (perception- and/or cognition-based) relevance – is one of the sources of the linguistic relevance of topic-comment structure, whereby the primary participant is aligned with the topic, and the action/state, with the comment. In this sense, the topic-comment information structure is grounded in the PP-structure, and this link does not depend on the language-specific degree of discourse flexibility of this structure. However, the use of the PP-structure beyond its core semantic domain (e.g., for secondary participants) certainly reinforces this link, which, diachronically, can work as a kind of positive feedback for language speakers and language learners and thus increase the topic-encoding potential of the corresponding structural template.

4.3. Reference points The class of h-topic constructions subsumes language-specific constructions that can be employed for introducing hanging topics (see Sections 3.2 and 3.3 for an overview of this class of constructions in general and in European languages in particular). The primary semantic motivation of the h-topic sentence structure seems to be, in effect, negative: its availability ensures the possibility to introduce a reference to the topic without encoding its semantic relation to the comment, or, at another level, without integrating this reference into the propositional structure of the sentence. It seems, however, that there is an important class of “real world” relations between entities and events that plays a special role in the “positive” semantic grounding of this class of constructions, namely, stable (pre-established) relations between the topic referent and a participant of the situation described in the comment, in particular, between

Sentence topics in the languages of Europe and beyond

89

the topic referent and the primary participant of this situation – roughly speaking, the same class of relations that can be encoded as (grammatical) possession. The special role of this class of entity-event relations with respect to h- topic constructions is determined by a combination of two properties: on the one hand, the entity does not play any role in the event and thus should not and cannot (in any direct way) be integrated into the propositional structure; on the other hand, the description of the event is likely to give relevant information about the entity. This class of relations between topic and comment is exemplified in (40)–(42); in (40) and (41a), the possessive-like relation is directly encoded by possessive cross-reference pronouns within the comment; in (41b)–(42), it must be inferred for the proper identification of the referent of subject NP (the package in (41b) and die jüngsten in (42)). (40) a.

b.

(41) a. b.

(42)

Russian ˇ Cto že do publik-i to eë otnošenie k what PRT to public-GEN then 3SG.POSS attitude to avtor-u možno opredelit’ kak neˇcto srednee author-DAT may(IMPRS) define as something middle meždu. “Blagodarju vas!” i “Pošël von!” between thank:1SG you:ACC and get lost ‘As for the public, its attitude towards the author can be described as anything between “I thank you!” and “Get lost!”’ (from an Internet chat) ˇ Cto že do fond-a finpomošˇc-i regionam, to what PRT to fund-GEN financial.help-GEN provinces:DAT then ego ob”ëm uveliˇcivaetsja po sravneni-ju s 2003 3M.POSS volume increases as comparison-DAT with 2003 god-om na 1.5 %. year-INSTR by 1.5 % ‘As for the fund for financing the provinces, its size increases by 1.5 % compared to 2003 : : :.’ (from an Internet chat) English As for the homeless, where did their food come from? (from an Internet forum) As for version 1.1, the package includes three utilities. (from an online description of a software package) German12 Schürr schrieb: Frankreich wolle er Schürr wrote(3SG) France want:SUBJ:3SG he Italien, die jüngsten seien ihm sehr Italy DEF youngest be:SUBJ:3PL he:DAT very

nicht, aber NEG but vertraut. familiar

90

Elena Maslova and Giuliano Bernini

‘Schürr wrote: France he didn’t want, but (as for) Italy, the youngest (authors) were very familiar to him.’ According to Langacker (1993: 24–26), both grammatical possession and the topiccomment relation can be viewed as a manifestation of the same reference-point relationship, or, as Langacker puts it, of “the ability to invoke the conception of one entity for purposes of establishing mental contact with another, i.e., to single it out for individual conscious awareness” (Langacker 1993: 5); it seems reasonable, therefore, to adopt his terminology in the formulation and discussion of our hypothesis and to refer to the structural template under discussion as reference-point (RP) structure. In fact, the concept of reference point is directly related to the topic-related function of anchoring the message in the shared knowledge (see Section 2). For our hypothesis, however, it is essential to stress that the “reference-point” relation between the topic and a participant of the situation being described (in contrast to the topic-comment relation) is based on a pre-established relation perceived/recognized by the speaker (and, presumably, the listener) as a “real world” one, and it is this relation that, according to our hypothesis, grounds the RP-structure in the perception and/or cognitive construal of reality. The crosslinguistic evidence for the special role of the possession-like class of relations between the topic and a component of the comment in the RP-structure is two-fold. Direct typological evidence comes from languages where RP-sentences with an explicitly grammatically encoded possession-like relation between the topic and an argument of the main proposition are more integrated into the grammar (i.e., conventionalized to a greater degree) than RP-sentences with genuinely “hanging” topics and/or RP-sentences featuring other relations between the topic and the comment. In Yukaghir, for example, sentences with a possession-like relation between the topic and an element of the comment constitute the only fully conventionalized class of RP-sentences with morphologically unmarked topic expression; the topic must be cross-referenced within the comment by a bound possessive pronoun (-gior -de- depending on the case form of the possessed), for example: (43) a.

b.

c.

Tundra Yukaghir (field notes) taN paj es’ie-gi taN pulut-ket joulos’-l’el-um. that woman father-3SG that old.man-abl ask-infr-3sg ‘That woman, her father asked that old man.’ taN pulun-die n’u:-gi Pottuo l’e-l. that old.man-dim name-his Pottuo be-SF ‘That old man, his name was Pottuo.’ taN paj ör-de-get met önme bojs’e šohie-j. that woman cry-her-abl my mind completely get.lost-3sg ‘That woman, her cry made me mad.’

Sentence topics in the languages of Europe and beyond

91

In Tz’utujil, the RP-structure (with unmarked preverbal topic expression) is conventionalized for topics that are either participants of the event being described (44a)– (44b) or are linked to a participant of this event by a possession-like relation (44c) (Aissen 1999). (44) a.

b.

c.

Tz’utujil (Aissen 1999: 170, 180) Ja r-me’al x-ok-i Malincha. the A3SG-daughter CMPL-enter-IV Malincha ‘His daughter played the role of Malincha.’ Ja w-xaayiil x-in-k’am el San Jwaan. the A1SG-wife CMPL-A1SG-take DIR San Juan ‘I took my wife to San Juan.’ Ja keej qas nim ruu-jeey. the horse very big A3SG-tail ‘The horse has a long tail.’

Moreover, RP-sentences featuring the possession-like relation between the topic and an element of the comment (“double-subject” sentences) appear to have a special grammatical status in topic-prominent languages as well (Kumashiro and Langacker 2003) and have been claimed to play a major role in grammaticalization of this structure in general and, consequently, in the rise of “topic prominence” as a typological state of language (Li and Thompson 1976: 480, 484). Some typical examples of such sentences from a topic-prominent language are given in (45). (45) a.

b.

Korean (Li and Thompson 1976: 468, 480) Pihengi-n1n 747-ka kh1-ta. airplane-TOP 747-SBJ big-stat ‘Airplanes, the 747 is big.’ John-1n meli-ka aph1-ta. John-TOP head-SBJ sick-stat ‘John has a headache.’

In this sense, the distribution of RP-sentences with regard to the RP-relation parallels the distribution of PP-sentences with regard to the PP-role: the availability of the RP-structure for “possessors,” just as the availability of the PP-structure for primary participants (“semantic subjects”) constitutes the typologically unmarked case. That is, if an h-construction is conventionalized at all, it is conventionalized for possessors. The second piece of evidence comes from the fact that the “possessor” of an entity participating in the situation is quite frequently the ˛-topic of a description of this situation (i.e., what this description is about), especially for such semantic subclasses of grammatical possession as whole-part (in particular, body part, see (43c), (45b)), kinship (43a), and arguments of relational nouns: for example, describing a

92

Elena Maslova and Giuliano Bernini

state of affairs involving someone’s body part is (more often than not) intended to increase the addressee’s knowledge about the person in question (rather than about their body part). More generally, the same factors that make a referent an appropriate reference point for an entity involved in the situation being described increase the likelihood of this referent being the ˛-topic of this description; to put it the other way around, it is often the case that adding information about a referent involves predicating something of a closely related entity. From the purely structural point of view, RP-sentences appear to instantiate the iconic strategy for encoding of this information structure. Compare, for instance, the example from Mandarin Chinese in (46) and its possible English counterparts in (47). (46)

Mandarin (Li and Thompson 1976: 468) nèike shù yèzi dà that tree leaves big ‘That tree, leaves are big.’

(47) a. b. c.

That tree has big leaves. The leaves of that tree are big. (As for) that tree, its leaves are big.

Assuming that the information to be added about ‘that tree’ is the information about the size of its leaves, not about its having leaves, syntactically and pragmatically “unmarked” English options are obviously non-iconic as far as information structuring is concerned: a have-based structure in (47a) indicates the topic by means of the PP-strategy, yet this is only possible because the statement is structured as a predication of possession, whereas the actual predication of property is packed into a noun phrase; conversely, a genitive-based structure (47b) does succeed in presenting the statement as a predication of property, but obviously fails to make the topiccomment structure syntactically transparent. The RP-structure in (47c) – which is, in effect, structurally equivalent to the Mandarin original – resolves this conflict, but at the cost of resorting to a less conventionalized and/or morphosyntactically and pragmatically more “marked” structure. It seems, indeed, that a relatively peripheral status of RP-sentences in European languages correlates with the existence of another conventionalized strategy for indicating topicality of reference points. In English – and, for that matter, in “ Standard Average European” – this alternative strategy is represented by have-based constructions (as exemplified in (47a)), which allow the speakers to employ the PP-scheme to encode this type of information structure (Langacker 1993; Aissen 1999);13 other European languages resort to a strategy based on spatial framing to accomplish the same information-structuring goal (see Section 4.4). This correlation can be construed as indirect evidence in favor of the hypothesis of the intrinsic link between the RP-structure and a presupposed relation between the ˛-topic and an argument of the main predication.

Sentence topics in the languages of Europe and beyond

93

On the other hand, the role of the RP-structure as a strategy of encoding the aboutness relation becomes particularly obvious if it is employed beyond its primary semantic domain, that is, for topics that are integrated into the main proposition (participate in the situation being described), especially for primary participants (as illustrated by sentences (b) in examples (48)–(51) for some European languages). (48) a. b.

(49) a. b.

(50) a. b. (51) a.

b.

French (Hagège 1985: 219) Les chiens mordent quand on les provoque. DEF dogs bite.3PL when IMPR them provokes Les chiens, ça mord quand on les provoque. DEF dogs this bites when IMPR them provokes ‘Dogs bite when provoked.’ German (Jacobs 2001: 658) Peter kommt morgen. Peter come.3SG tomorrow Peter, der kommt morgen. Peter he.DEM come.3SG tomorrow. ‘Peter is coming tomorrow.’ English (Givón 1976: 153) The wizard lived in Africa. Now the wizard, he lived in Africa. Russian samye pravil’nye. Pervye mysli first thoughts most right ‘The first thoughts are most true.’ Pervye mysli oni samye pravil’nye. first thoughts they most right ‘The first thoughts, they are most true.’ (from an Internet chat)

Here, again, the emerging picture is homomorphous to that discussed in Section 4.2 for the PP-structure: whereas an h-topic from the primary semantic domain need not encode the ˛-topic, the use of this structure beyond its primary semantic domain strongly suggests the aboutness relation between the topic and the comment. By the same token, the RP-encoding of the primary participant is a stronger indication of its topic role than the PP-encoding (cf. the a. and b. examples above).

4.4. Spatial frames The topic-encoding structures discussed so far do not cover a class of constructions that plays a major role in a number of European languages, namely, constructions

94

Elena Maslova and Giuliano Bernini

in which the semantic relation between the topic and the comment is encoded on the topic expression, by means of the same relational marker (a case marker and/or an adposition) that would be used for nontopical constituents in the same role (f constructions in terms of Section 3.3). This construction class is illustrated by the following examples: (52)

Italian (Bernini 1992: 44) Alla casa dell’ Impruneta ci andav-ano poco. to:def house of:def Impruneta loc.clit went-3PL few ‘To the house of Impruneta, they used to go there rarely.’

(53)

French (Cadiot 1992: 63) Aux enfants je pardonne tout. to:def children I forgive:1sg everything ‘To children I forgive everything.’

(54)

German Mir sind seit gestern noch eine Menge Fragen I:DAT are since yesterday still a quantity questions eingefallen. come.to.mind ‘A lot of questions have come to my mind since yesterday.’ (Jakob Arjouni, Happy Birthday, Türke! Ein Kayankaya-Roman)

(55) a.

b.

Russian Ètu poèmu Ivan Nikolaeviˇc soˇcinil, i this:acc poem:acc Ivan:nom Nikolaeviˇc:nom compose:pst:3sg and v oˇcen’ korotkij srok : : : in very short:acc time:acc ‘This poem Ivan Nikolaevich has composed, and in a short time : : :’ (Mikhail Bulgakov, Master i Margarita) V reljacionnoj grammatike osnovnoe vnimanie udeljaetsja in relational:obl grammar:obl main:nom attention:nom is:given ne procedure otoždestvlenija sintaksiˇceskix otnošenij : : : neg procedure:dat identification:gen syntactic:gen relations:gen ‘In relational grammar, attention is focused not on the procedure of the identification of syntactic relations : : :’ (A. Kibrik)

The morphosyntactic form of topic expression is determined by its role in the proposition, and the structuring of information is achieved by fronting, which can be accompanied by a break in intonation contour and/or a pause; in some languages, fronting of the topic expression requires cross-referencing of the topic within the comment (cf. the locative clitic in (52)).

Sentence topics in the languages of Europe and beyond

95

The hypothesis we would like to present in this section is that this topic-encoding strategy is grounded in the relation between the event and its spatial/temporal location, in the same way as the PP-strategy is grounded in the relation between the event and its primary participant, and the RP-strategy in the relation between the event and an entity related to one of its participants. This structure will be referred to as spatial-frame, or SF-structure. As in the preceding cases, the hypothesis is based on three types of considerations: (i) the description of a state of affairs can provide relevant information about the place/time where this state of affairs holds, whereby the place/time of an event assumes the ˛-topic status with respect to the description of this event; (ii) fronting of spatial/temporal “scene-setters” is less typologically marked than fronting of other components of the proposition (Paršin 1983, 1984); and (iii) the topic-encoding potential of the SF-structure is stronger (more visible) if it is applied beyond its (hypothesized) primary semantic domain. The first point can be illustrated by the following example: (56)

Russian V našej strane ateizm nikogo ne in our:obl country:obl atheism:nom nobody:acc neg udivljaet. surprise:pres:3sg ‘In our country atheism does not surprise anybody.’ (Mikhail Bulgakov, Master i Margarita)

Although the initial constituent of (56) specifies the location within which the situation being described takes place, the sentence is to be interpreted as adding to the addressee’s knowledge about “our country.” This interpretation is supported by the text from which the example is excerpted: the addressee of this remark extensively expresses his gratitude for very interesting and important information about the country. It goes without saying that not every clause-initial spatial/temporal adverbial can be reasonably interpreted as “what the sentence is about,” that is, the SF-structure used within its primary semantic domain need not encode the ˛-topic (see, however, Maienborn 1996; Jacobs 2001). However, the topic-encoding function of the SF-structure becomes more transparent if the locative relational expression is used metaphorically (as in (55b), (57), and (58)). (57)

English In English we can do something similar with certain temporal adverbs. (W. Chafe)

(58)

German (Jacobs 2001: 657) In meinem Traum war Peter ein Krokodil. in my dream was Peter a crocodile ‘In my dream, Peter was a crocodile.’

96

Elena Maslova and Giuliano Bernini

As in the previous cases, the SF-structure can be used beyond its primary (spatial/temporal) semantic domain for encoding topicality of participants of the situation being described, whereby the relational marker on the topic expression reflects the actual role of the topic in the situation; this is illustrated in (53) and (55a) above. A further piece of evidence in favor of topic-indicating potential of the SF-structure is given by languages that use locative expressions in predications of possession, cf. the following Russian examples: (59) a.

b.

Russian U zabora stoja-l-a lošad’. at fence-gen stand-pst-sg:fem horse:nom ‘A horse stood near the fence.’ U Ivana by-l-a kvartira. at Ivan:gen be-pst-sg:fem apartment:nom ‘Ivan had an apartment.’

This structure can also be employed if some information about an entity is provided by predicating a property of a related entity for which the topic can serve as a reference point, that is, the SF-structure can be used for topic-encoding beyond its primary semantic domain and extended to the semantic domain of RP-structure: (60) a.

b.

c.

Russian U Ivana bolit golova. at Ivan:gen sick:pres:3sg head:nom ‘Ivan has a headache.’ U menja ženilsja syn. at me:gen marry:pst:3sg son:nom ‘My son has married.’ Moj syn ženilsja. my son:nom marry:pst:3sg ‘My son has married.’

The sentences of this type are constructed in such a way as to encode the reference point for the subject as the topic (cf. the distinction between (60b) and (60c)), and it is the SF-structure that is employed for this purpose. As expected, the SF-topic of such sentences consistently coincides with the ˛-topic.

4.5. Three semantic templates vs. three aspects of topicality In Sections 4.2–4.4, we focused on the inherent association between three types of “real world” relations between entities and states of affairs (and the corresponding structural templates) and the intuitive concept of aboutness (˛-topic), which, as we assume together with the grammatical tradition, is the closest approximation to the

Sentence topics in the languages of Europe and beyond

97

hypothesized mental status of topic. In this section, we discuss their relationship to two other aspects of topicality mentioned in Section 2, namely, the functions of anchoring the predication in the context and delimiting its domain of applicability. The concept of topic as “given” goes back to the Prague Circle’s “theme vs. rheme” distinction; it was introduced to account for word order variations (Mathesius 1929; Daneš 1964), that is, in our framework, for the topic-encoding usage of the SFtemplate, and reflects the general observation that entities referred to by clause-initial constituents tend to be “given,” whereas new information is added later.14 In the modern accounts of topicality, this idea is elaborated and refined in terms of identifiability of topic referents to the listener and degree of their activation in the listener’s mind (Chafe 1976, 1987; Gundel et al. 1993; Lambrecht 1994; inter alia). As mentioned in Section 4.3, the relevance of this aspect of topicality to the RP-template follows directly from its primary reference-point function: an entity can serve as an appropriate reference point for something else only insofar as it is better anchored in the shared knowledge. The correlation between the degree of activation of the topic referent and the PP-template is determined by one of the universal correlations that link this template with topicality in the first place (see Section 4.2): primary participants are frequently humans, and thus central topics of discourse; accordingly, they are most likely to be previously mentioned (and thus activated). Strikingly, the SF-template is in fact less straightforwardly associated with the anchoring function; we will return to this issue in the end of this section. The concept of delimiting the domain of applicability of the comment was introduced to describe the functions of h-topic constructions, that is, of RP-topics in the conceptual framework introduced above. Li and Thompson (1976: 463–464), in their extensive discussion of topic-prominent languages, accept Chafe’s functional characterization of “Chinese-style” topics, which states that it “sets a spatial, temporal, or individual framework within which the main predication holds” (Chafe 1976: 50). Indeed, the RP-topic expression can limit the applicability of the main predication, in a clearly definable semantic sense: if it does, then the sentence does not entail its comment (Jacobs 2001: 656–658), that is, a sentence consisting of the comment only would be, at least without sufficient contextual clues, interpreted as applicable to a broader class of situations. For instance, consider an example of topic-comment sentence from (Li and Thompson 1976: 459): (61)

As for education, John prefers Bertrand Russell’s ideas.

The sentence does not entail its comment, insofar as the latter (without activation of the topic referent) is assumed to mean that John prefers Bertrand Russell’s ideas in all domains of knowledge enriched by Russell’s contributions. However, the concept of frame setting, if defined in purely semantic terms, does not apply straightforwardly to all instances of h-topic constructions. Sentence (62) (the English translation of German example (15)), according to Jacobs’ interpretation, does entail its comment:

98

Elena Maslova and Giuliano Bernini

the municipal theater plans a performance of “Faust” independently of whether this information is related to Goethe’s anniversary; consequently, Jacobs claims that the delimiting function cannot be viewed as an intrinsic property of the as-for construction (and nor of the German was X betrifft construction): (62)

As for Goethe’s 250th birthday, the municipal theater plans an unabridged performance of “Faust.”

However, the apparently obvious distinction between sentences like (61) and (62) becomes blurred if the context (in the most general sense of the word) is taken into account. On the one hand, if the comment of sentence (61) occurs as an independent sentence in the context of a conversation about education, the meaning conveyed by this sentence would be exactly the same as the meaning of the original sentence (with the RP-topic attached). On the other hand, (62) fails to delimit the domain of applicability of its comment only insofar as Goethe’s rendition of the story of Doctor Faust is taken to be the (only) one; insofar as the shared cultural knowledge of the interlocutors includes other versions of the story and does not single out Goethe’s version as the Faust, the RP-topic certainly delimits the set of potential referents of “Faust,” and thus the domain within which the main predication holds. To illustrate this point by another example, it is very likely that a sentence like [As for X] I like “Eugene Onegin” very much will be interpreted differently by different readers of this chapter: for some, X is most likely to be Alexander Pushkin, for others, Petr Tchaikovsky; and one cannot even exclude Ralph Fiennes as a possible topic of such a sentence. Depending on what the (explicit or implicit) topic is, the referent of “Eugene Onegin” would be the poem, the opera, or the movie with this title. The point is that a sentence like As for Pushkin, I like “Eugene Onegin” very much would seem, out of the context, to entail its comment (at least to our Russian readers) just as (62) seems to entail its comment, yet a sentence like I like “Eugene Onegin” very much can also be truthfully said by someone who has no idea of the referent of the first sentence and just likes the movie. This digression into a discussion of cultural differences, which may appear to have little to do with linguistics, is intended to demonstrate that the delimiting role of topics cannot be reduced to a purely semantic notion. In the case of RP-topics, it is inherently associated with their primary reference-point function, namely, with anchoring the proposition in the shared knowledge: the activation of a topic referent narrows down not the out-of-the-context propositional meaning of the comment, but the message that would be conveyed by the comment alone in the given context (where the context is taken to include both the current discourse and speech situation and the relevant cultural context). The concept of delimiting the domain of applicability has been defined in such a way that it cannot be straightforwardly applied to the PP-template, since a PP-topic is not semantically separated from its main proposition. If the contents of the comment is construed as the main proposition with unspecified primary participant, then

Sentence topics in the languages of Europe and beyond

99

the sentence obviously entails its comment and thus does not delimit the domain within which the proposition holds (for example, a sentence like The man kissed the woman entails something like The woman was kissed or Someone kissed the woman [cf. Jacobs 2001: 646, 656]). On the other hand, the comment of a PP-sentence is, as a rule, encoded in such a way as to indicate that the referential identity of the primary participant is specified (The man kissed the woman is not semantically identical to something like As for the man, someone kissed the woman.) In this sense, specification of primary participant certainly does impose additional truth conditions and thus delimits the individual domain within which the comment holds. Accordingly, PPtopics can be said to delimit the domain of applicability of their predications insofar as the comment of a PP-sentence differs from a sentence with unidentified subject reference (which seems to be the case in all European languages). Finally, the inherent affinity between this aspect of topicality and the SF-template seems to be beyond reasonable doubt: delimitation of the domain within which the state of affairs takes place can well be viewed with the primary function of sentenceinitial “scene-setters,” whether or not the template is used beyond this semantic domain in a given language. Since the existence of a specific spatial/temporal locus of the event is implied for most sentences (cf. Kratzer 1995), but its explicit specification is not, as a rule, syntactically obligatory, the primary scene-setting function determines the range of contexts where SF-sentences are likely to occur: these are the contexts where identification of the spatial/temporal domain of the state of affairs is essential for interpretation, but the spatial/temporal domain of the previously mentioned states of affairs is either broader or simply different. For example, the Russian sentence in (56) without the SF-topic expression could be interpreted as applying either more broadly than intended (e.g., as a statement concerning the world as a whole), or, in the context of a conversation about some other country, as a statement about that country. This results in a special type of anchoring of the sentence in the context associated with the SF-template: it is based not on a prior activation of the SF-topic itself, but rather on a prior (explicit or implicit) activation of other spatial/temporal locations. As a result, the SF-template is inherently associated with contrastivity; for example, fronting the reference to ‘our country’ in (56) implies the existence of other countries where atheism can be rare and therefore surprising. This aspect of meaning can be retained in sentences where the SF-template is used beyond its primary domain (cf. Chafe 1976; Paršin 1983; Büring 1999; inter alia). To sum up this section, each semantically grounded structure discussed in this section is inherently linked with one aspect of topicality: primary participants, with the aboutness function, spatial frames, with the delimiting (frame-setting) function, and reference points, with the anchoring function. In each case, however, one function triggers the others, so that each template comes to be associated with all three aspects of topicality. In our view, this corroborates our hypothesis that the hypothesized men-

100

Elena Maslova and Giuliano Bernini

tal status of topic joins together three linguistic templates directly grounded in our construal of reality.

5.

Language-internal and crosslinguistic variation in topic encoding

5.1. Two dimensions of topic prominence The core of our approach to the concept of sentence topic is formed by the idea that the topic-comment structure is an aspect of sentence meaning, that is, it is determined primarily by what the speaker has to say. This by no means implies that the speaker’s assumptions about the listener’s state of knowledge and attention play no role in the choice of topic, yet it is assumed to be no more (and no less) significant than the role of these assumptions in the choice of information to be communicated. For example, a sentence like (3a) (repeated here as (63a) for convenience) would hardly be appropriate if the listeners (or, in this case, the readers) were assumed to have no knowledge whatsoever of syntactic theories, the use of schemes in their presentation, and of the role-and-reference grammar and its authors in particular. The same is true for the alternative encoding of the same proposition in (63b), which differs from (63a) in that Van Valin and Foley’s schemes (rather than the role of linguistic schemes in general) is perceived as “what the sentence is about.” (63) a.

b.

Russian Tak-uju že illjustrativn-uju funkci-ju ime-jut such-ACC same illustrative-ACC function-ACC have-PRES.3PL sxem-y R. Van Valin-a i W. Foley. scheme-PL.NOM R. Van Valin-GEN and W. Foley ‘The same illustrative role is played by R. Van Valin and W. Foley’s schemes.’ (Aleksandr Kibrik) Sxem-y R. Van Valin-a i W. Foley ime-jut scheme-PL.NOM R. Van Valin-GEN and W. Foley have-PRES.3PL tak-uju že illjustrativn-uju funkci-ju. such-ACC same illustrative-ACC function-ACC ‘R. Van Valin and W. Foley’s schemes have the same illustrative function.’

The choice of one sentence structure over the other appears to be guided not by the author’s assumptions about his readers state of knowledge, but rather by what he actually wants to say. The structure of (63a) signals that the role of schemes in syntactic theories has the topic status, and the sentence is therefore understood as an additional illustration of the author’s general idea about this role. Although the choice of object-initial order is bound to have some impact on the processing of the

Sentence topics in the languages of Europe and beyond

101

sentence, the topic status of the object does not amount to a component of processing instructions, but rather belongs to the meaning to be processed. On the other hand, the choice of one or another construction for the encoding of the sentence is likely to be guided, to a large extent, by the speaker’s assumption about the listener’s state of knowledge and attention. In particular, the need for explicit and unambiguous encoding of the topic-comment structure is determined by two universal factors, (i) the assumed ability of the listener to infer the topic-comment structure from the context and/or the contents of the sentence, and (ii) the assumed degree of activation of the topic referent by the time of utterance. These factors are independent in the sense that if the topic referent is highly activated, the topiccomment structure can still be explicitly encoded if it cannot be inferred: (64)

German Und das tat er auch. and this did he too ‘[I have to wake up, he thought.] And he did it too.’ (Sven Regener, Herr Lehmann. Ein Roman.)

(65)

Italian (Berretta 1995: 142) A: Prendiamo un caffè? take:1PL a coffee B: No, il caffè lo prendiamo a casa. no the coffee OBJ.3SG.M take:1PL at home ‘Will we have coffee? – No. We’ll have coffee at home.’

In both cases, the listener would have been likely to infer, in the absence of the explicit encoding to the contrary, that the topic of the sentence is the primary participant. On the other hand, a low degree of activation of the topic referent would trigger explicit topic encoding even if the topic-comment structure can be inferred. This situation is neatly illustrated by the following Italian conversation: (66)

Italian (field notes) A: hai preso su la cassetta? have:2SG picked up the cassette B: eh? eh A: la cassetta l’ hai presa su? the cassette it:ACC have:2SG picked up [In the car, before leaving for a holiday] ‘A: Have you picked up the (audio)cassette? – B: Eh? – A: The cassette, have you picked it up?’

102

Elena Maslova and Giuliano Bernini

The two questions in this example have the same content, and can be assumed to have the same topic (the cassette). The cassette had already been the topic of the conversation preceding this excerpt, so the first speaker (A) initially assumes that this referent is sufficiently activated and B will have no problems in identifying the topic and inferring the topic-comment structure of the message. Since B does not understand the question, A apparently changes his assumptions and introduces the topic by means of a clause-external referring expression, which involves explicit encoding of the topic-comment structure. The degree of activation of topic-referent can influence the linear position of the topic expression. Although the most common linear position of structural topics is sentence-initial, they can also occur in the sentence-final position, as exemplified in (67). (67) a.

b.

c.

French (cf. also Lambrecht 1981: 1). Ils sont fous, ces Romains! they are mad those Romans ‘They’re mad, those Romans!’ Italian (field notes; casual conversation) le dicono proprio così per dir=le, le notizie. them say:3PL really so for tell:INF=them the news [A: While we were waiting at the tollgate, they said that there was a kilometer-long queue (implying that it wasn’t true).] B: ‘They just tell them in order to talk, the news items.’ German (from Die Zeit 28/3. 7. 91, p. 53) War eine zu heiße Sache, diese Affäre um den Pastor. was a too hot matter this affair around the pastor ‘[The pastor, who used to live here, certainly won’t check. And the cops won’t look into the mail box either. They never hurried in their investigation.] It was too hot a matter, this affair with the pastor.’

The referents of right-detached expressions are usually referred to within the main proposition (as witnessed by italicized pronouns in (67a) and (67b) and by the verb agreement in (67c)). The right-detached expression can be encoded for its role in the proposition, that is, such sentences can instantiate the SF-template, for example: (68)

Italian (filed notes; casual conversation) se mai- ci capiterò da quelle parti if ever there happen.to.go:FUT:1SG at those parts ‘If I ever happen to go there, in those districts’

Many authors pointed out differences in discourse contexts favoring sentence-initial and sentence-final topics: right-detached referring expressions appear to be limited to referents recoverable from the situation and/or immediate discourse context (e.g.,

Sentence topics in the languages of Europe and beyond

103

Gundel 1988: 229; Berretta 1995: 150–151; inter alia). According to Givón (1983), a right-detached NP recalls a discourse topic which is continuous for the speaker, but was introduced at such a distance that the hearer might have lost its track. A further example, drawn from Maltese, illustrates this point. (69)

Maltese ( Fischer and Jastrow 1980: 288) L-ewwel nett kien-u jagègnu-h b’ at-first completely be:PFV-3PL knead:IPFV:3PL-3SG.M with l-id-ejn, dawn it-tliet xkejjer. ART-hand-DU these ART-three sacks [From a narrative on bread baking] ‘At the very first they used to knead it with the hands, these three sacks.’

In this example, the third person object suffix -h ‘it’ refers to the flour gathered from three sacks, which were mentioned in the previous sentence, whereas the rightdetached NP recalls the flour by reference to the quantity involved in the kneading operation being described. This seems to suggest that the sentence-final position of topic expression is associated with contexts where the speaker might assume that the referent is sufficiently activated in the listener’s mind to be referred to by a pronoun.15 Oversimplifying the matters to some extent, the factors that are likely to trigger unambiguos topic encoding can be subsumed under a single parameter, which can be referred to as “topic prominence”: in this sense, the degree of topic prominence of a sentence is determined by the salience of the topic-comment structure in its overall semantic structure and the speaker’s assumption about the degree of activation of the topic referent in the listener’s mind and context-dependent inferability of the topiccomment structure.

5.2. Competition between topic-encoding templates The hypothesis outlined above entails that there are three universally available semantically-grounded templates for articulation of the topic-comment structure of the message; this provides the basis for language-internal variation in topic encoding. Assuming the intended meaning consists of a topic referent t and a comment most adequately conveyed by a proposition P, the speaker has a choice between four universally available options, which can be schematically represented as follows: (70)

PP-template: SF-template: RP-template: No topic encoding:

S(P0 (t)) S(R(t) + P) t + S(P) S(P)

S(: : :) signifies encoding of the information in the form of an independent sentence, R(t) indicates that the topic expression has the form required by its role (R) in the

104

Elena Maslova and Giuliano Bernini

proposition P, and P’ indicates that the proposition may need to be modified in order to encode t as the primary participant. This schematic paradigm is exemplified in (71). (71) a.

b.

c.

d.

Russian ˇ Cto do ošibk-i, to Efremov sam pišet o nej what to error-GEN then Efremov INTSF writes about 3SG.OBL v predislovi-i k svoemu sobrani-ju in introduction-OBL to RFL.POSS:DAT collection-DAT soˇcinenij. works:GEN ‘As for the error, Efremov writes about it himself in the introduction to his collected works.’ (from an Internet chat) Ob ètoj ošibk-e Efremov sam pišet v predislovi-i about this error-OBL Efremov INTSF writes in introduction-OBL k svoemu sobrani-ju soˇcinenij. to RFL.POSS:DAT collection-DAT works:GEN ‘Efremov writes about this error himself in the introduction to his collected works.’ Èta ošibk-a opisana samim Efremovym v this error-NOM described INTSF:INST Efremov:INST in predislovi-i k ego sobrani-ju soˇcinenij. introduction-OBL to his collection-DAT works:GEN ‘This error is described by Efremov himself in the introduction to his collected works.’ Efremov sam pišet o nej v predislovi-i k Efremov INTSF writes about 3SG.OBL in introduction-OBL to svoemu sobrani-ju soˇcinenij. RFL.POSS:DAT collection-DAT works:GEN ‘Efremov writes about it himself in the introduction to his collected works.’

Sentence (71a) is an authentic example found on an Internet forum and instantiates the RP-template. The topic referent (an error in one of the editions of Efremov’s book) is one of two topics of the discussion from which the example is extracted. Having said something about the other discourse topic, the author of this comment switches back to the error in question. In contrast to this, if the error had been established as the current topic of discussion and named as such in the preceding sentence, the speaker could choose not to encode it as a structural topic, but just to refer to it by means of a pronoun, the resulting sentence being identical to the comment of the original example (cf. examples (71a) and (71d)). The s-construction in (71c) (i.e., the PP-template) would be appropriate if the topic is continuous but has not yet been

Sentence topics in the languages of Europe and beyond

105

referred to as ošibka (for example, following a description of what is considered an error). The SF-template in (71b) would also sound natural in such a context, but it is more likely than the s-construction in the context of contrastive topic (for example, if the speaker wishes to imply that there are other errors described elsewhere or unnoticed by the author). Essentially the same range of discourse-sensitive factors, albeit in a somewhat different fashion, affect the output of similar competitions between coding options in other languages. Let us return to our first example, repeated here as (72). (72) a. b. c. d.

English [inspector] “You are not to open any letters unless you recognize the handwriting,” he said. “Everything else we’ll look at first. As to the phone calls : : :” [woman] “Your sergeant said you’d have an arrangement to get my phone monitored.” (Ruth Rendell, No more dying then. An Inspector Wexford Mystery.)

The PP-topic in (72a) is established as a continuous topic by the speech situation. The SF-template in (72b) explicitly contrasts two classes of letters which are to be dealt with in different ways; the form of the topic expression links it to the preceding sentence (insofar as everything else is to be understood as “other letters”). Finally, the (beginning of) RP-template in (72c) signals a switch to a different topic, which is obviously less activated by the preceding conversation than the SF-topic in (72b) but is also explicitly contrasted to the previous topic (“letters”). The tendencies exemplified by (71)–(72) appear to be universal and directly follow from the universal properties of topic-encoding templates outlined in Section 4. To begin with, the range of discourse contexts in which the PP-template can be felicitously used would include the contexts of continuous (i.e., highly activated and inferrable) topics, whereas the RP-template is more likely to be used for less accessible topic referents. This tendency is a direct implication of the inherent properties of these templates outlined in Sections 4.2–4.3. In most cases, the primary participant is a necessary component of the state of affairs, independently of whether or not it is overtly expressed: if not, it is normally taken to be “dropped” and its antecedent is to be looked for in the preceding context. In contrast to this, if no RP-topic is present, it is not “dropped,” it is just absent. For example, the sentence John prefers Bertrand Russell’s ideas (cf. (61)), even if produced when the listener is assumed to understand it as true only for education, would not be identified as an instance of h-topic construction with a “dropped” topic; similarly, (71d) would not be identified as an instance of the RP-template (identical to (71a) with topic “dropped”). Otherwise, we would have to describe almost all sentences taken from a coherent piece

106

Elena Maslova and Giuliano Bernini

of discourse in this fashion, namely, as “containing” a number of “dropped” reference points which are derivable from the preceding context (for example, all English sentences with anaphoric pronouns as instances of left-dislocation with “dropped” left-dislocated constituent). Although this approach may be in a sense helpful for analysis of discourse coherence, it does not seem reasonable for syntactic typology. Insofar as sentences are identified as instances of the RP-template if the topic expression is overtly present in a sentence, this template is bound to occur only in discourse contexts where the topic referent is not highly activated by the time of utterance. In contrast to this, the PP-template is not inherently constrained in terms of the degree of activation of the topic referent. The role of the SF-template in the overall system of topic encoding is subject to a more considerable crosslinguistic variation. On the one hand, it presupposes a higher level of integration of the topic expression into the propositional structure, insofar as their relation R has to be encoded morphosyntactically. As a result, it need not be associated with the points of topic discontinuity (in contrast to the RP-template) and can be employed to “compensate” for inherent semantic constraints on the PPtemplate. For example, the choice between the SF-template and the PP-template exemplified in (71b) and (71c) respectively in the context of continuous topic can be motivated by purely semantic considerations: assuming that the comment of the original example is the optimal expression of the speaker’s intended meaning (i.e., proposition P), the PP-template requires a slight modification of this meaning: in order to encode the error as the subject, the speaker must integrate it into the propositional structure by transitivizing the main verb, which happens to involve a change in aspect and tense as well (instead of the present imperfective of other alternatives, this one has a past perfective main predicate). Although the resulting difference between the encoding options amounts to delicate semantic nuances, this semantic distinction can be significant from the point of view of the speaker’s actual intentions and thus rule out the use of the PP-template. Obviously, this language-specific factor reflects an inherent property of the PP-template: depending on the relation R between t and P, the modification of P involved by the PP-encoding (referred to as P0 (t) in (70)) can turn out substantial enough to rule out this option and limit the competition to other alternatives. Although the scope of the resulting constraints on the use of PP-template vary from language to language (Keenan 1985: 243–247; Van Valin and Foley 1980; Foley and Van Valin 1985: 299–334), the very presence of such constraints is determined by the inherent properties of the PP-structure and is therefore universal. On the other hand, as described in Section 4.5, the SF-template is intrinsically associated with contrastive topics, namely, with contexts where the topic referent has to be explicitly singled out from a contextually determined set of entities that would be associated with other propositions (see Paršin 1983, 1984 for a detailed discussion of this issue). Insofar as the PP-structure is (relatively) broadly available for encoding of continuous topics, the SF-structure would tend to be constrained to contrastive

Sentence topics in the languages of Europe and beyond

107

topics (as in English, cf. (72b)). Note that a contrastive topic-comment structure is, by its very nature, highly salient and not inferrable from the context; accordingly, the SF-structure is likely to reflect a higher degree of topic prominence (in the sense defined in Section 5.1). To sum up the discussion so far, the inherent properties of semantically grounded templates entail the following mapping of the two dimensions of topic prominence on a hierarchy of the topic-encoding templates: (73)

reference point >> spatial frame >> primary participant

The RP-template is inherently associated with (relatively) low degrees of activation of the topic referent, whereas the SF-template and the PP-template are not intrinsically constrained in this respected. On the other hand, the RP-template and the SF-template are associated with topic-comment structures not inferrable from the context, whereas the PP-template is neutral in this respect. Finally, the choice of a topic-encoding template can be affected by the relative structural markedness of the corresponding constructions, or, in other terms, by the economy principle (Haiman 1983); most importantly, the considerations of economy would favor sentences where the topic is referred to once over those that contain a (nominal or pronominal) cross-referencing expression. Language-specific rules that require cross-referencing the topic generally follow the hierarchy in (73), which leads to a conflict between unambiguous encoding of the topic-comment structure (topic prominence) and markedness. To begin with, the comment of RP-template is, by definition, encoded as an independent sentence (which would occur in the context of highly activated topic); accordingly, it is most likely to require cross-referencing. No overt cross-referencing expression is needed only if the language allows zero discourse anaphora, which seems to be the case in all topic-prominent languages (Li and Thompson 1976: 466–471; Gundel 1988: 239–242). In other words, the RP-template can be consistently preferred over other templates only if it is never “penalized” by this structural markedness constraint; in the languages of Europe, this condition is obviously not satisfied. In contrast to this, the PP-topic is least likely to require crossreferencing by an independent pronoun. The SF-template occupies the intermediate position. In particular, it can require cross-referencing in some European languages (e.g., in Italian, see (52) and (74))16 but not in others (e.g., in Russian and in English, see (71b) and (72b)). (74)

Italian A Giovanni Maria non gli ha detto niente. to John Mary NEG he:DAT.CLT has told nothing ‘Mary has told nothing to John.’

If the SF-template does not require cross-referencing, structural markedness considerations favor this template over the RP-template; otherwise, they favor the PPtemplate over the other two templates.

108

Elena Maslova and Giuliano Bernini

5.3. Interaction between topic-encoding templates The choice of one topic-encoding template over the other can be affected by the presence of another topic in the intended meaning. For example, the SF-sentence in (71b) and the PP-sentence in (71c) differ in terms of the locus of the agent (Efremov) in the information structure: the SF-template construction allows for the encoding of the primary participant as “hyper-topic” (Kuno 1987: 17), as in the original example (71a): Efremov constitutes a more global topic, and within this general domain, one of his errors is singled out as the local sentence topic (Lambrecht [1994: 147–150] uses the terms “primary” and “secondary” topic[s] instead).17 The use of the PPtemplate for the object participant precludes the encoding of the primary participant as “hyper-topic”: in (71c), this referent must be included into the scope of assertion (in contrast to the original example). This difference can play a more or less significant role in the choice of encoding options depending on the speaker’s intentions. Generally, two structural templates can be combined with a single sentence to point to two different topic referents, so that the template encoding the local sentence topic “embeds” the template encoding the hyper-topic as its comment. Languagespecific constraints on such combinations appear to conform to the hierarchy of topic prominence in (73). The use of the RP-template for the encoding of the local topic tends to license a hyper-topic encoding in all languages, as illustrated by sentences (71a) (for Russian), (61) (for English), and (62) (for German). In contrast to this, the PP-template does not allow for the embedding of another topic-encoding template as its comment. The ability of the SF-template to embed the PP-template is subject to crosslinguistic variation: for example, this combination is possible in Russian (see (71b)) and in English (72b), but not in German. On the other hand, the SF-template can be easily “embedded” into the RP-template for hyper-topic encoding in Russian, but not in English. As a result, Russian allows for a combination of three hierarchically arranged topics, as demonstrated by (75). Here the discourse topic (“hypertopic”) is the enrollment advantages (encoded by means of the innermost PP-topic); within this domain, the speaker talks about the legal basis of these advantages (the law of Russian Federation, encoded by means of the next-level SF-topic); finally, the local sentence topic within this domain is the national-level competitions for gifted high school students, encoded as a RP-topic. (75)

Russian ˇ Cto kasaetsja obšˇcerossijsk-oj olimpiad-y, to po what concerns all.Russia:ADJ-GEN Olympics-GEN then PREP zakon-u RF l’got-y predostavljajutsja tol’ko law-DAT RF advantage-PL give:PASS.PL only pobeditel-jam. winner-PL.INST

Sentence topics in the languages of Europe and beyond

109

‘As for the national competition, the law of Russian Federation gives enrollment advantages only to the winners.’ (from an interview given by the rector of Moscow State University) To sum up, if the intended meaning invokes a hyper-topic, the encoding of local topic can be chosen in such a way as to allow for a hyper-topic encoding. Most commonly, this factor will disfavor the use of the PP-template if the local topic does not coincide with the primary participant and the hyper-topic does. In some languages, the paradigm of competing encoding options can be reduced by neutralization of the formal distinction between them. One type of neutralization can be referred to as “merged” topic encoding: in some languages, the morphosyntactic constructions corresponding to distinct coding options listed in (70) can turn out to be identical for some relations R between the topic expression and the proposition. The most trivial case of merged encoding is the identity of PP-encoding and SF-encoding with the no-encoding option for primary participants of P and spatial/temporal settings respectively. This situation was discussed in Section 4.2 with reference to example (39b): the use of an active construction does not ensure unambiguous encoding of the topicality of the primary participant. Accordingly, a relatively high degree of topic prominence would trigger the choice of a different template for unambiguous topic encoding. Moreover, for topic referents coinciding with primary participants the PP-template and the SF-template are merged in subjectinitial languages: whether the speaker chooses the PP-template or the SF-template, the result will be the same subject-initial structure. This, in effect, reduces the competition to two distinct candidates (roughly speaking, the RP-template vs. no encoding), which means that the RP-template emerges as the only option for unambiguous encoding of topic-comment structure for primary participants and spatial/temporal frames. The SF-template and the RP-template can merge into the same sentence structure if they impose similar requirements on cross-referencing the topic within the comment, that is, if their comments can have the same morphosyntactic form. Under this condition, the templates are merged in sentences where the role of topic in the propositional structure of the comment does not require overt marking. The following example from Italian illustrates this point: (76)

Italian Il coraggio, uno non se lo può dare. the courage one NEG REFL.DAT OBJ.3SG.M can:3SG give:INF ‘One cannot give himself courage.’ (Manzoni, I promessi sposi)

This sentence can instantiate either the SF-template (the topic expression has the same form as it would have within the comment) or the RP-template with unmarked topic (a similar example from Spanish is given in (22)). Furthermore, if the topic

110

Elena Maslova and Giuliano Bernini

coincides with the primary participant, all three templates can merge into a single sentence structure, as in (77a). (77) a.

b.

Italian Giovanni ha accettato un lavoro in un’officina meccanica. Giovanni has accepted a job in a shop machine-ADJ ‘John accepted a job in a machine-shop.’ ha accettato un lavoro in un’officina meccanica. has accepted a job in a shop machine-ADJ ‘He accepted a job in a machine-shop.’

On the one hand, (77a) instantiates the PP-template (i.e., the subject-predicate structure); on the other, it also instantiates the RP-template with unmarked topic, since its comment has the structure of an independent sentence with a highly activated subject referent (cf. (77b)). By the same token, it can be taken to instantiate the SF-template, since the topic expression has the same form as a nontopical subject. Another type of neutralization is triggered by double topic encoding, whereby different structural templates combined within a single sentence refer to the same topic. The following examples illustrate the combination of RP- and SF-topic encoding in Russian: the pronoun cross-referencing the RP-topic is fronted, so that the comment follows the SF-template with the same topic (see also (24) for a similar example from Italian and (49) for a German example where the RP-template is doubled by the PP-template merged with the SF-template). (78) a.

b.

Russian (from Internet chats) ˇ Cto kasaetsja Brèdbery, u nego est’ odin iz rann-ix what concerns Bradbury, at him exists one of earlier-PL.GEN rasskaz-ov The Small Assassin. story-PL.GEN the.small.assassin ‘As for Bradbury, he has an earlier story called “The Small Assassin.” ’ Politkorrektnost’ eë vzrosly-e izobre-li. political.correctness it:ACC adults-PL.NOM invent-PST.PL [Children bully one another everywhere.] ‘Political correctness, it was invented by adults.’

The English translations of these examples exemplify a similar combination of RPand PP-templates pointing to the same topic (see also (48)–(51)). In European languages, the double topic encoding is optional, that is, the embedded structural template can but need not point to the same topic, presumably depending on whether the sentence has a hyper-topic. It seems, however, that the double topic encoding can also be grammaticalized, so that the PP-template is obligatorily combined with the RP-template for topic encoding. This situation is apparently

Sentence topics in the languages of Europe and beyond

111

attested in Dinka, a Western Nilotic language described by Andersen (1991). Declarative clauses in Dinka contain a single preverbal noun phrase, which, according to Andersen’s description, encodes the sentence topic. The topic has some properties normally associated with subjects, for example, the verb must take a special passivelike form if the preverbal NP does not refer to the primary participant (Andersen 1991: 265), cf. (79a) and (79b). (79) a.

b.

Dinka, Western Nilotic (Andersen 1991: 272) m`oc a` -ku`al w´eN. ˜ ˜D-steal ˜ cow ˜ man ‘The man is stealing the cow.’ w´eN a` -ku´EEl mˆoc. ˜ ˜ ˜ ˜ ˜ cow D-steal:NTS man:GEN ‘The man is stealing the cow.’

In the present framework, this would indicate that topicality is encoded by means of the PP-template. On the other hand, the preverbal position is not constrained in terms of its role within the comment, for example, it can correspond to the location of the event being described (80a) and to the possessor (reference point) of a participant of this event (80b). (80) a.

b.

Dinka, Western Nilotic (Andersen 1991: 281, 284) r`oor a` -múuc mˆoc thín. ¨ ˜ ˜D-shoot:AP:NTS ¨ ˜ forest man:GEN PRO:LOC ‘The man is shooting in the forest.’ m`ari`aal a` -th´EEt ti´EEN-d`e m`eth. ˜ ˜ ˜ ˜ ˜D-beat:NTS ˜ ˜ Marial woman-3SG child ‘Marial’s wife is beating the child.’

Moreover, if the preverbal NP does not represent the primary participant, the postverbal part of the sentence can contain a coreferential proform (Andersen 1991: 278– 283), which indicates that the NP is clause-external. Cross-referencing the topic within the comment conforms to the general constraints on discourse anaphora: it is obligatory for locative topics (as in (80a)) and optional for possessors, instrumentals, and objects (cf. (79b) and (81)). (81)

Dinka, Western Nilotic (Andersen 1991: 276) y´Ok a´ a-kue´een dh´Ook (kê). ¨ ˜D:PL-count:NTS ˜ ˜ cows boy¨ 3PL ‘The boy is counting the cows.’

These properties appear to reveal the RP-template; indeed, Andersen’s claim is that it is a (syntactic) topic, rather than a subject. In our view, however, this construction

112

Elena Maslova and Giuliano Bernini

is best described in terms of grammaticalization of double topic encoding: the RPtopic must be simultaneously encoded as a PP-topic (hence the voice-like marking on the verb in sentences where it does not correspond to the primary participant). Thus, the language neutralizes the structural distinction and displays a uniform topic encoding that combines the properties of the RP-template and of the PP-template. The cases of “double” topic encoding and “merged” topic encoding can be consistently distinguished only insofar as the topic can be referred to twice within the sentence. For example, the Italian example in (77) has been described in terms of merging of the RP-template and the PP-template, yet it could as well be described in terms of obligatory double encoding, insofar as the use of the PP-template for topic encoding is obligatorily accompanied by the RP-encoding (insofar as the predicate of a subject-topic sentence is inevitably encoded as an independent sentence with a highly activated subject-topic). Note that essentially the same situation is observed in all topic-prominent languages, since they are characterized by common use of zero anaphora for established discourse topics (Li and Thompson 1976: 466–471; Gundel 1988: 239–242). A similar situation arises if the PP-template is merged with the SFtemplate for topics coinciding with primary participants and these templates cannot point to two different topics within the same sentence (as in German). This means, in effect, that the PP-encoding must be always “doubled” by the SF-encoding; accordingly the SF-template is invoked whenever the PP-template is used. Neutralization of the distinctions between topic-encoding options can have a significant impact on the overall distribution of competing templates in the language. One aspect of this impact was mentioned above: insofar as a specific topic-encoding template occasionally merges with the no-encoding option, it cannot be used for unambiguous encoding of the topic-comment structure and is therefore likely to be avoided in topic-prominent contexts. Alternatively, this can lead to obligatory doubling of topic encoding, whereby the unmarked (ambiguous) template is obligatorily accompanied by another topic-encoding template. On the other hand, the processing of a sentence instantiating a pair of merged/doubled topic-encoding templates increases discourse frequency of each template, which, in turn, is likely to promote conventionalization of the otherwise less frequent template and thus decrease its pragmatic markedness (cf. Dryer 1995). Indeed, each instance of a construction “entrenches” the underlying topic-encoding template in the minds of the interlocutors and thus increases its role in the language and makes it more readily available in other contexts and speech situations (Croft 2000: 72–73); in other words, an increase in discourse frequency serves as a positive “feedback” for the language speakers in the sense that it is likely to lead to a further increase in frequency.

Sentence topics in the languages of Europe and beyond

113

5.4. Conclusion: Universal constraints on topic encoding In this section, we attempted to demostrate how the inherent properties of alternative topic-encoding templates determine some universal constraints on crosslinguistic and language-internal variation in the domain of topic encoding. The RP-template obviously has a privileged status with respect to topic encoding, which manifests itself both in language-internal and in crosslinguistic variation: insofar as the semantic domains of competing templates overlap in a specific language, the RP-template is preferred in topic-prominent contexts; crosslinguistically, the RP-template can serve as the only strategy of topic encoding in topic-prominent languages in Li and Thompson’s (1976) sense (see Sections 3.2–3.3). This privileged status is determined, first and foremost, by the fact that both other topic-encoding templates are intrinsically constrained in terms of the role of the topic expression in the proposition conveyed by the comment: they cannot be used for encoding of hanging topic. Accordingly, the RP-template consistently wins the competition described in Section 5.2 if the relation between the topic and the state of affairs being described cannot be integrated into the propositional structure of the message by morphosyntactic means available in the language. By the same token, the RP-template is the only one that cannot merge with the no-encoding option and thus encodes the topic-comment structure unambiguously in all languages and independently of the relation between the topic referent and the state of affairs being described. On the other hand, due to the combination of factors described in Section 5.2, this template is intrinsically associated with the discourse contexts of topic discontinuity, where neither the topic referent nor the topic-comment structure of the message can be assumed to be inferable, that is, with the topic-prominent contexts. Consequently, it also consistently wins this competition in the contexts where the need for explicit and unambiguous encoding of the topic-comment structure and/or for activation of the topic referent overrides the considerations of structural markedness. However, it can emerge as the major topic-encoding template only if it does not require crossreferencing the topic within the comment, that is, if it never violates the markedness constraint against double reference to the same entity. In Section 5.3, we also suggested that the role of this template in the overall system of topic encoding is likely to be increased by the existence of constructions that neutralize the distinction between the RP-template and one or both of the other templates; this hypothesis seems to be corroborated by crosslinguistic studies (Geluykens 1989, 1992). The PP-template can be preferred over the RP-template insofar as (i) it is not ruled out by the intended meaning, that is, the topic referent can be integrated into the proposition as a primary participant without significant modifications of this proposition, and (ii) its application leads to a structurally simpler sentence. Under these conditions, the PP-template would be preferred in the contexts of relatively low topic prominence. The SF-template is characterized by a combination of relatively weak

114

Elena Maslova and Giuliano Bernini

semantic constraints (in effect, it is not available for hanging topics only) and similarly weak constraints on the degree of activation of the topic referent. Accordingly, it can compete with the PP-template for encoding of continuous topics and with the RP-template for encoding of salient and not inferable topic-comment structures. Its language-specific role can be increased by relative structural simplicity, by morphosyntactic merging with the other templates, and by merging of the PP-template with the no-encoding option, which restricts its ability to encode the topic-comment structure. Although a detailed analysis of how these factors interact with each other to determine language-specific preferences for one or another topic-encoding template is beyond the scope of this chapter, we hope to have shown that our approach to sentence topic as a semantically grounded aspect of sentence structure opens the possibility to account for the attested language-specific and crosslinguistic variation in topic encoding and for the universal constraints on this variation in a unified fashion.

Notes 1.

2.

3. 4.

5.

The earlier variants of most proposals put forward in this paper were extensively discussed at the meetings of the EUROTYP working group on the pragmatic organization of discourse, which helped us tremendously in developing and refining these proposals. Bernard Comrie and Georg Bossong read the first version of this paper and provided many helpful comments and suggestions, all of which have been taken into account in the final version. This hypothesis is not intended to imply that the topic-comment structure plays equally salient roles in all languages (cf., for example, Cowan 1995 for a discussion of this issue). Tomlin (1995) describes a crosslinguistic psycholinguistic experiment linking the mental status of topic with the state of the speaker’s own attention. In our view, the topic status cannot be reduced to semantic predication (cf. Jacobs 2001; Kiss 1998), since this would, by definition, exclude hanging-topic sentences (Section 3.2). The explication of aboutness in terms of “mental addressation” (Reinhart 1982; Vallduví 1995: 123–125; Jacobs 2001) seems to be based on an overly simplistic and “metaphorical” model of human memory and, in fact, does not provide any more explicit criteria for identification of topics than the intuitive notion of “aboutness.” Although this paper is primarily based on a corpus of authentic example sentences excerpted from natural discourse, at some points we have to resort to constructed examples in order to illustrate paradigmatic oppositions available in different languages. This includes some examples borrowed from linguistic literature (linguistic sources are indicated before the example sentence). References to primary sources (newspapers, Internet chats, fiction, etc.) are given in brackets after idiomatic translations. If the source is not explicitly indicated, the example (from Russian or Italian) is constructed on the basis of the native language competence of one of the authors.

Sentence topics in the languages of Europe and beyond 6. 7. 8.

9. 10. 11.

12.

13.

A similar view seems to be reflected in the traditional logical distinction between thetic and categorical judgments (see Sasse [this volume] for an overview). This classification does not imply that all language-specific topic constructions neatly fall in one or another class (see Section 5.3 for further discussion). According to Lötscher (1992: 129), adverbs can also be topics, as in Finanziell ist Otto sehr erfolgreich ‘Financially, Otto is very successful.’ The suggestion is plausible, since the adverb finanziell in the example quoted here may be paraphrased as, ‘Was die Finanzen angeht’/‘As to the financial side,’ etc. Our hypothesis does not preclude the possibility that the topic can be activated by means of an adverbial expression (see, in particular, Section 4.4), although the interpretation of adverbials as genuine topics can be controversial (Shibatani 1990: 275–278; Jacobs 2001: 656–657). The particle to attached to the topic expression in this example is not obligatory and also occurs with nominal topics. Cf. also Keenan’s suggestion to use “topichood” as one of the defining features of “basic subjects” (Keenan 1976: 318). The preverbal position emerges as a subject-marking feature only if both the subject and the direct object are referred to by nouns that have identical forms for nominative and accusative case. Taken from Klemperer, Victor (1995). Ich will Zeugnis ablegen bis zum letzten. Tagebücher 1933–1941. Berlin: Aufbau, p. 34. It might be worth noting that, unlike what happened in the case of Italien, integration of the topic object constituent Frankreich in the second sentence of the excerpt is signalled by second-positioning of the verb and by postverbal positioning of the subject er. Another possible way to subsume this type of information structure under the PPscheme is subject incorporation. This strategy is described for Chukchee (Nedjalkov 1977), for example: (i)

14.

15.

16. 17.

115

Ytlyg-yn q‘aa-v’e-g’e Father-SG:ABS deer-died-3SG ‘Father, (his) deer died.’

Major contributions in the research tradition of “functional sentence perspective” are collected in a volume edited by Sornicola and Svoboda (1991). For a general introduction to the theory and method of “functional sentence perspective,” see Eroms (1986). Such sentence structures can be then associated with the interpersonal or polite function of speech (cf. Aijmer 1989). Berretta (1995: 150–151) points out that in the Italian colloquial variety, right-dislocation might be losing its marked status, becoming a means for marking positive politeness. Examples such as Lo vuoi, un caffè? ‘Would you like it, a coffee?’ show construal of the yet unknown topic ‘the coffee’ as if it were known, a strategy claimed to appealing to the listener’s cooperation and to emphasize sharing of knowledge. In Italian, cross-referencing is not required only for some (“oblique”) relations R between the topic expression and the proposition. According to Vallduví (1995) and Kiss (1998: 691), a sentence can have two functionally equivalent topics; although it seems difficult to evaluate this claim on the basis of

116

Elena Maslova and Giuliano Bernini constructed examples, we hypothesize that the local vs. global opposition can be more or less sharp depending on the context, and the situation of equivalent topics is possible as a “limiting” case, where this opposition is neutralized. Indeed, there is nothing in the hypothesis of existence for sentence topic that would entail that only one entity per sentence can assume this mental status (in contrast to the “mental addressation” model, which explicitly excludes the possiblity of multiple topics [Reinhart 1982]).

References Aijmer, Karin 1989 Aissen, Judith 1999

Themes and tails: the discourse functions of dislocated elements. Nordic Journal of Linguistics 12: 137–154.

External possessor and logical subject in Tz’utujil. In External Possession, Doris L. Payne and Immanuel Barshi (eds.), 167–193. Amsterdam/Philadelphia: John Benjamins. Andersen, Torben 1991 Subject and topic in Dinka. Studies in Language 15 (2): 265–294. Benincà, Paola, Giampaolo Salvi, and Lorenza Frison 1988 L’ordine degli elementi della frase e le costruzioni marcate. In Grande grammatica italiana di consultazione, Vol. 1: La frase. I sintagmi nominale e preposizionale, Lorenzo Renzi (ed.), 115–225. Bologna: Il Mulino. Bernini, Giuliano 1992 The Sentence Topic in the Languages of Europe. EUROTYP Working Papers I/3. Programme in Language Typology, European Science Foundation. Berretta, Monica 1995 Ordini marcati dei costituenti maggiori di frase: una rassegna. Linguistica e filologia 1: 125–170. Bolkestein, A. Machtelt 1993 General ideas of functionalism in syntax. In Handbooks of Linguistics and Communication Science, Vol. 9.1: Syntax, Joachim Jacobs, Arnim von Stechow, Wolfgang Sternefeld, and Theo Vennemann (eds.), 339–349. Berlin: Walter de Gruyter. Büring, Daniel 1999 Topic. In Focus. Peter Bosch and Rob van der Sandt (eds.), 142–165. Cambridge: Cambridge University Press. Cadiot, Pierre 1992 Matching syntax and pragmatics: a typology of topics and topic-related constructions in spoken French. Linguistics 30 (1): 57–88. Chafe, Wallace L. 1976 Givenness, contrastiveness, definiteness, subjects, topics, and point of view. In Subject and Topic, Charles N. Li (ed.), 25–55. New York et al.: Academic Press. 1987 Cognitive constraints on information flow. In Coherence and Grounding in Discourse, Russell S. Tomlin (ed.), 21–52. Amsterdam: John Benjamins. Chen, Ping 1996 Pragmatic interpretation of structural topics in Chinese. Journal of Pragmatics 26: 389– 406. Comrie, Bernard 1981 Language Universals and Linguistic Typology. Chicago: University of Chicago Press.

Sentence topics in the languages of Europe and beyond Cowan, Ron 1995 Croft, William 2000 Daneš, František 1964 Davison, Alice 1984 Dik, Simon 1989

117

What are discourse principles made of? In Word Order in Discourse, Pamela Downing and Michael Noonan (eds.), 29–50. Amsterdam/Philadelphia: John Benjamins. Explaining Language Change. An Evolutionary Approach. London: Longman. A three-level approach to syntax. Travaux Linguistiques de Prague 1: 225–240. Syntactic markedness and the definition of sentence topic. Language 60: 797–846. The Theory of Functional Grammar. Part 1: The Structure of the Clause. Dordrecht: Foris.

Downing, Angela 1997 Discourse pragmatic functions of the theme constituent in spoken European Spanish. In Discourse and Pragmatics in Functional Grammar, John H. Connolly, Roel M. Vismans, Christopher S. Butler, and Richard A. Gatward (eds.), 137–163. Berlin/New York: Mouton de Gruyter. Dryer, Matthew S. 1995 Frequency and pragmatically unmarked word order. In Word Order in Discourse, Pamela Downing and Michael Noonan (eds.), 105–136. Amsterdam/Philadelphia: John Benjamins. Eroms, Hans-Werner 1986 Funktionale Satzperspektive. Romanistische Arbeitshefte 31. Tübingen: Niemeyer. Fici Giusti, Francesca, Lucyna Gebert, and Simonetta Signorini 1991 La lingua russa. Rome: La Nuova Italia Scientifica. Fischer, Wolfdietrich and Otto Jastrow (eds.) 1980 Handbuch der arabischen Dialekte. Porta Linguarum Orientalium, Neue Serie XVI. Wiesbaden: Harrassowitz. Foley, William A., and Robert D. Van Valin 1985 Information packaging in the clause. In Language Typology and Syntactic Description, T. Shopen (ed.), 282–364. Cambridge: Cambridge University Press. Geluykens, Ronald 1989 The syntactization of interactional processes. Some typological evidence. In Universals of Language, Michel Kefer and Johan van der Auwera (eds.), 91–103. Bruxelles: Éditions de l’Université de Bruxelles. (Belgian Journal of Linguistics 4.) 1992 From Discourse Process to Grammatical Constructions: On Left Dislocation in English. Amsterdam: John Benjamins. Givón, Talmy 1976 Topic, pronoun and grammatical agreement. In Subject and Topic, Charles N. Li (ed.), 149–188. New York et al.: Academic Press. Givón, Talmy (ed.) 1983 Topic Continuity in Discourse. A Quantitative Cross-Language Study. Amsterdam: John Benjamins. Gundel, Jeanette K. 1988 Universals of topic-comments structure. In Studies in Syntactic Typology, M. Hammond, E. A. Moravcsik, and J. R. Wirth (eds.), 209–242. Amsterdam: John Benjamins. Gundel, Jeanette K., Nancy Hedberg, and Ron Zacharski 1993 Cognitive status and the form of referring expressions in discourse. Language 69: 274– 307.

118

Elena Maslova and Giuliano Bernini

Hagège, Claude 1985 L’homme de paroles. Contribution linguistique aux sciences humaines. Paris: Librairie Arthème Fayard. Haiman, John 1983 Iconic and economic motivation. Language 59: 781–819. Harris, Alice C. and Lyle Campbell 1995 Historical Syntax in Cross-linguistic Perspective. Cambridge: Cambridge University Press. Hawkinson, Annie and Larry Hyman 1975 Hierarchies of natural topics in Shona. Studies in African Linguistics 5: 147–170. Hockett, Charles F. 1958 A Course in Modern Linguistics. New York: Macmillan. Jacobs, Joachim 2001 The dimensions of topic-comment. Linguistics 39 (4): 641–681. Kaseviˇc, Vadim B. 1988 Semantika, Sintaksis, Morfologija (Semantics, Syntax, Morphology). Moscow: Nauka. Keenan, Edward L. 1976 Towards a universal definition of “subject”. In Subject and Topic, Charles N. Li (ed.), 303–333. New York et al.: Academic Press. 1985 Passive in the world’s languages. In Language Typology and Syntactic Description, T. Shopen (ed.), 243–281. Cambridge: Cambridge University Press. Kiss, Katalin É. 1995 NP movement, operator movement, and scrambling in Hungarian. In Discourse Configurational Languages, Katalin É. Kiss (ed.), 207–243. New York/Oxford: Oxford University Press. 1998 Discourse-configurationality in the languages of Europe. In Constituent Order in the Languages in Europe, A. Siewierska (ed.), 681–729. Berlin/New York: Mouton de Gruyter. Kozinskij, Iosif S. and Natalja K. Sokolovskaja 1984 O sootnošenii aktual’nogo i sintaksiˇceskogo cˇ lenenija v sinxronii i diaxronii (On the relation between functional and syntactic sentence structure in synchrony and diachrony). In Vostoˇcnoje jazykoznanije: Grammatiˇceskoje i aktual’noje cˇ lenenije predloženija, V. M. Solncev (ed.), 63–76. Moscow: Nauka. Kratzer, Angelika 1995 Stage-level and individual-level predicates. In The Generic Book, G. N. Carlson and F. J. Pelletier (eds.), 125–175. Chicago: University of Chicago Press. Kumashiro, Toshiyuki and Ronald W. Langacker 2003 Double-subject and complex-predicate constructions. Cognitive Linguistics 14 (1): 1– 45. Kuno, Susumo 1987 Functional Syntax. Chicago: University of Chicago Press. Lambrecht, Knud 1994 Information Structure and Sentence Form. Cambridge Studies in Linguistics. Cambridge: Cambridge University Press. Langacker, Ronald W. 1993 Reference-point constructions. Cognitive Linguistics 4 (1): 1–38. Li, Charles N., and Sandra A. Thompson 1976 Subject and topic: a new typology of languages. In Subject and Topic, Charles N. Li (ed.), 457–490. New York et al.: Academic Press.

Sentence topics in the languages of Europe and beyond

119

Lockwood, W. B. 1995 Lehrbuch der modernen jiddischen Sprache. Mit ausgewählten Lesestücken. Hamburg: Buske. Lötscher, Andreas 1992 The pragmatics of nonreferential topics in German (and other languages). Linguistics 30 (1): 23–145. Maienborn, Claudia 1996 Situation und Lokation: Die Bedeutung lokaler Adjunkte von Verbalprojektionen. Tübingen: Stauffenburg. Mathesius, Vilém 1929 Zur Satzperspektive im modernen Englisch. Archiv für das Studium der modernen Sprachen und Literaturen 155: 200–210. Miller, Jim 1992

Null subjects and direct objects in Russian: remarks on topic continuity in Russian and English. In Topics, EUROTYP Working Papers I/2, Giuliano Bernini and David Ricca (eds.), 91–117. Programme in Language Typology, European Science Foundation.

Nedjalkov, Vladimir P. 1977 Possessivnost’ i inkorporacija v cˇ ukotskom jazyke (inkorporacija podležašˇcego) (Possessivity and incorporation in Chukchee [subject incorporation]). In Problemy lingvistiˇceskoj tipologii i struktury jazyka, Victor S. Xrakovskij (ed.), 108–137. Leningrad: Nauka. Ó Siadhail, Mícheál 1988 Learning Irish. An Introductory Self-Tutor. New Haven: Yale University Press. Paršin, Pavel B. 1983 Tema i topic: k sootnošeniju poniatij. In Voprosy vostoˇcnogho jazykoznanija, Vladimir M. Alpatov (ed.), 186–211. Moscow: Nauka. 1984

Topic i tože: granicy i interpretacija odnoj jazykovoj oppozicii (Topic and also: limits and interpretation of a linguistic opposition). In Vostoˇcnoe jazykoznanie: grammatiˇceskoe i aktual’noe cˇ lenenie predloženija, V. M. Solncev (ed.), 107–122. Moscow: Nauka.

Polinsky, Maria 1999 Review of Lambrecht 1994. Language 75: 567–582. Prince, Ellen F. 1981 Toward a taxonomy of given-new information. In Radical Pragmatics, Syntax and Semantics, Peter Cole (ed.), 213–255. New York: Academic Press. Reinhart, Tanya 1982 Pragmatics and linguistics: an analysis of sentence topics. Philosophica 27: 53–94. Sasse, Hans-Jürgen 1987 The thetic/categorical distinction revisited. Linguistics 25 (3): 511–580. Shibatani, Masayoshi 1990 The Languages of Japan. Cambridge: Cambridge University Press. 1991

Grammaticization of topic into subject. In Approaches to Grammaticalization, Vol. II: Focus on Types of Grammatical Markers, Elisabeth Closs Traugott and Bernd Heine (eds.), 93–133. Amsterdam: John Benjamins.

Sornicola, Rosanna and Aleš Svoboda (eds.) 1991 Il campo di tensione. La sintassi della Scuola di Praga. Naples: Liguori. Strawson, P. F. 1964

Identifying reference and truth values. Theoria XXX: 96–118.

120

Elena Maslova and Giuliano Bernini

Tomlin, Russel S. 1995 Focal attention, voice, and word order: an experimental, cross-linguistic study. In Word Order in Discourse, Pamela Downing and Michael Noonan (eds.), 517–554. Amsterdam: John Benjamins. Vallduví, Enric 1995 Structural properties of information packaging in Catalan. In Discourse Configurational Langagues, Katalin É. Kiss (ed.), 122–152. New York/Oxford: Oxford University Press. Van Valin, Robert D. and William A. Foley 1980 Role and reference grammar. In Current Approaches to Syntax. Syntax and Semantics, Vol. 13, E. Moravcsik (ed.), 329–352. New York: Academic Press.

Jim Miller

Focus in the languages of Europe

1. Introduction1 Chapter 1, “Spoken and written language,” draws a distinction between the repertoire of structures available in a particular language and the use that speakers make of a particular structure. This distinction was applied to (formal) writing vs. (spontaneous) speech and is relevant to any area of grammar where the set of constructions found in speech overlaps but does not coincide with the set of constructions found in writing. In the investigation of discourse organization it is not just relevant but crucial, and one of the author’s concerns has been not just to draw up lists of the devices by which discourse is organized, but to find a way of comparing the uses of devices across languages. Indeed the major difficulty for typological work on discourse organization is the need to analyze both the structure of discourse organization devices such as clefts and their function in discourse. The investigation of discourse functions is not one of the simpler analytical tasks in linguistics. The chapter deals with grammatical devices, ignoring intonation, although it is obviously central to the highlighting of constituents. What counts as a grammatical device turns out to be major constructions such as the clefts of English, particles such as the German doch, eigentlich, wirklich, and aber, special positions such as the preverbal position in Turkish, and word order. The major results of the investigation have to do with which languages have cleft constructions and which languages do not, which cleft constructions are available and used in which cleft-possessing language, and the use of particles and word order as alternatives to clefts. The availability and use of clefts declines across Europe from west to east, and the spoken language data on which this chapter is based indicates that non-neutral word orders are often (perhaps even typically) accompanied by particles. In order to compare, we need criteria and concepts, especially an understanding of focus, and we need data. Section 2 explains the approach to the key concepts and data collection. Section 3 provides a summary of the results. Section 4 offers an overview of focus. Section 5 deals with devices for highlighting new individuals, Section 6 with devices for highlighting new propositions. Section 7 describes the complexities of word order even within one language, Russian. Section 8 takes a brief look at reduction of highlighting – taking entities out of focus, and Sections 9 and 10 are the central sections on clefts in English and their equivalents in other languages;

122

Jim Miller

Section 9 deals with languages with clefts, and Section 10 deals with languages that do not have clefts. Section 11 contains the conclusions.

2. Theory and data 2.1. Basic concepts At the very beginning of the research a central theoretical problem arose: how should “focus” be interpreted? The answer we give to the question rests on the key assumption that we can recognize when constituents stand out from the surrounding constituents in a clause or sentence. We can ask what devices are available to the speakers and writers of a given language for making constituents stand out; we can ask what means they actually use according to corpus evidence; and we can ask what discourse purposes are served by a particular device. That is, the research supporting this chapter is data-driven. Since the various concepts of focus to be found in the literature will be briefly reviewed in Section 3, we say here simply that every concept of focus has to do with giving prominence to constituents and the information they carry, albeit for different reasons – the introduction of new entities or new propositions, the contrast of one entity with another, “exhaustive listing” (one particular entity and no other), or noncontrastive prominence. For typological purposes, it seems worthwhile taking as broad a perspective as possible; we will avoid adopting any one definition of focus and instead examine the devices which give prominence to constituents, whatever the reason. Consequently, readers should be prepared to come across constructions excluded by whatever definition of focus they might be thirled to. The noun focus and the verb focus carry too great a theoretical burden to be employed in a nontechnical sense. It will be necessary to appeal repeatedly to constituents standing out from the surrounding constituents, whether this is achieved by means of pitch, syntactic construction, word order, or particle. Constituents will be said to be given prominence, made prominent, highlighted, made salient, or given salience. Prominence, salience, and highlighting are not technical terms but are used in their everyday meaning. The other side of this theoretical coin is that we will also have to look at constituents that are made less prominent, defocused – typically because they refer to highly topical entities. The phrase “highly topical entities” brings us to the technical term “topic” and the associated term “theme.” The first point to make is that a highly topical entity (see Givón 1983) merely indicates that the entity has already been mentioned, possibly several times, in the immediately preceding text. Highly topical entities tend to be referred to by means of pronouns or zero NPs, with the referent being reestablished every so often by means of a full NP. New, nontopical entities tend to be referred

Focus in the languages of Europe

123

to by means of a full NP and are regularly highlighted by extra syntactic coding as described in Section 5. But choice of referring expression is affected by whether an entity is being mentioned for the very first time in a discourse or is being reintroduced after a previous mention. The choice of referring expression is also affected by whether speakers are participating in very relaxed conversation and referring to entities salient in the immediate context of situation, in which case pronouns may be used, or even zero NPs in a language such as Russian. As will emerge from the discussion of Russian in Section 8.2, this sort of defocusing occurs when an entity is deemed by the speaker to be salient in a situation, even without being mentioned. But different languages offer different possibilities for defocusing. The term “topic” itself will be avoided in this chapter, except for references to the topic of a paper or a book, or even the topic of a section or paragraph. The terms “theme” and “theme position” will be used occasionally, but in the usage associated with Halliday (1967, 1968, 1985). The theme of a clause is the first phrasal constituent in the clause; a phrasal constituent is a constituent occupying a slot that can be occupied by a single word or a sequence of words, such as if and in the event that, which are possible themes in conditional clauses. The theme of a sentence in a written text can be a phrase or a clause, subordinate or main. The crucial property of the Hallidayan theme is that it is defined syntactically and reflects the insight that speakers and writers do not put constituents into first position at random, but usually do so in order to achieve some goal connected with the discourse organization.2

2.2. Data collection A typology of focus requires knowledge of both the highlighting devices in a range of languages and of the discourse goals served by the use of a particular device in a particular text. A certain amount of information was available in existing descriptions and bodies of data, but it was essential to find a sound basis for comparing devices and discourse goals across languages. To this end we took as our point of orientation English dialogues produced during the performance of a task (see the account of this task below) in which the context was controlled as far as possible and analyzed in detail (see Miller and Weinert 1995; Weinert and Miller 1996). This very same task was subsequently performed by speakers of German, Catalan, Hungarian, and Bulgarian (see Note 1 for acknowledgments). In a further exercise, the author elicited data from informants by means of an oral questionnaire based on examples from the English dialogues, together with the contexts in which they were produced. This particular set of data is spoken, but the English highlighting devices, the cleft constructions, also occur in written language, though not necessarily serving the same discourse function. The corresponding devices in the other languages also occur in both speech and writing, with the same caveat. The reason for the caveat is

124

Jim Miller

that in the task-based English dialogues WH-clefts are used to draw a line under a complex piece of discussion and to open up a new set of instructions; reverse WH-clefts are used to drive home a particular point; and IT-clefts – very rare in the dialogues – are used to establish a contrast: X and not anything else. In other types of text, WH-clefts may be used simply to open a discussion, to catch the audience’s attention, and state the topic, while IT-clefts may be used merely as a highlighting device (see Section 6). A portion of the data discussed in this chapter is taken from written language, or at least could occur in both spoken and written texts, but the central set of data is spoken; the elicited data was spoken and the spontaneous dialogues are also spoken. Some of the spoken constructions also occur in formal written language, but the scope of the investigation did not permit the links between genre and discourse structures to be explored across the various languages. That much of the data is spoken requires no apology, given the argument in Chapter 1 to the effect that adequate typological work on discourse must at least distinguish between spontaneous speech and formal writing. An important feature of this chapter, which is discussed at more length below but is mentioned here as a forewarning, is that the central data on focus in Section 10 is merely data collected on one occasion under controlled conditions from a limited number of informants. Because of the difficulty in getting discourse data by questionnaire, we initially set out to compare data from three corpora – English map task dialogues, spontaneous English conversation, and various types of spontaneous spoken Russian as available in the transcripts published in Kapanadze and Zemskaja (1979) and as described in Zemskaja (1973) and Lapteva (1976). (Information on highlighting devices in written English and written Russian is available in grammars, although grammars typically provide very superficial analyses of discourse phenomena.) This comparison yielded interesting information on spontaneous spoken English and Russian and, in particular, on the limited role of word order, even in a language such as Russian with rich inflectional morphology. It is essential to emphasize here and now that an adequate typology of discourse organization depends on detailed contrastive analyses of the sort just mentioned, because it is only that type of investigation that yields an understanding of the function of particular syntactic constructions, word order, and particles in particular contexts. The fact that there is relatively detailed discussion of some texts in three or four languages and a superficial treatment of the other languages reflects the peculiar nature of discourse analysis in one language, let alone of a typology of discourse organization. Later, map task dialogues became available that had been collected from speakers of languages other than English, and it also became clear that the map task could be the basis of a questionnaire. The map task is carried out by two participants at a time. One participant has a map with landmarks, a starting point, and a route going

Focus in the languages of Europe

125

from the start to a finish point. The other participant has a map with landmarks and a starting point, but no route and no finish point. The two sets of landmarks largely overlap, but there are differences: a landmark may be missing from one map, or one map may have two occurrences of a particular landmark. The participants cannot see each other’s maps, and the participant who has the map with the route (the instruction giver, or IG) has to describe to the other participant (the instruction follower, or IF) how to draw the route.3 This task offers certain advantages. Researchers analyzing the dialogues know exactly which landmarks and labels are objectively given on the map and can distinguish between them and landmarks and labels that are not shared, but which are treated by one participant as if they were in fact given, that is, accessible to the other participant. The goals of the participants are known. The overall goal is for IGs to instruct IFs so that the latter draw a route from their starting point to a finish point in such a way that the IFs’ route matches the IGs’ route. Minor goals are to establish the nature and position of landmarks where the two maps do not coincide and to get the different sections of route drawn. It is clear where the participants move on from one subset of landmarks to another, and it is clear whether participants are giving an instruction for the first time, clarifying an instruction or requesting clarification, questioning a piece of information, and so on. Consequently, the discourse functions of syntactic constructions, pitch patterns, word order, and particles can be reliably analyzed. On the basis of the map task, an oral questionnaire was drawn up to investigate highlighting devices. Informants were invited to imagine themselves in various situations while playing the map task. The situations had actually occurred during the collection of the English map task dialogues, as had the English examples presented to the informants. Moreover, previous analysis of the English map task dialogues had established the discourse functions of the different cleft constructions in those dialogues (see the comments above, in Section 9.5.1, and in Weinert and Miller 1996); this understanding shaped the contextual information given to the informants who answered the questionnaire. The informants were not asked to translate the English examples; rather, they were asked what they would have said in the given situation. The map task examples were supplemented by other examples of English clefts, such as What I’m going to talk about today is X, which were included in order to explore different uses of cleft constructions. For each example an appropriate context was devised; for instance, the example above was said to have been uttered by a speaker beginning a talk. The use of a questionnaire has various disadvantages: it is impossible to determine whether a given informant has properly understood the situation; it is by no means certain that informants consult their intuitions correctly; and the influence of the English examples cannot be discounted. Both the Finnish informant and one of the Hungarian informants produced analogues of English WH-clefts. The Finnish

126

Jim Miller

informant commented that educated Finns with a good knowledge of English can be heard producing the Finnish analogue regularly, although it is not a construction recognized in Finnish grammars. The Hungarian informant felt initially that the Hungarian WH analogue was a possible structure but declared that no native speaker of Hungarian would use it. A similar comment was made by the Serbo-Croatian informant, who searched through written texts (children’s books, academic papers, instructions in crosswords and in magazines) for Serbo-Croatian analogues of IT-clefts but found none. Consultation with other native speakers of Serbo-Croatian living in Slovenia and Croatia revealed that none of them would ever use a Serbo-Croatian analogue of the English IT-cleft. Instead, the constituent to be highlighted is fronted. Nonetheless, the SerboCroatian informant insisted that an analogue of the English IT-cleft is possible and grammatical but that if it were used, say, in a Zagreb tram, heads would turn and possibly even roll. Finally, the German speaker who provided the IT-cleft constructions with es is a specialist in English, and the one German contributor to the German map task corpus who provided analogues of English WH-clefts had been living in the UK for two or three years. In spite of the drawbacks, certain pieces of reliable information were obtained. The English examples all involved some type of cleft construction: IT-clefts, as in It’s the bridge you avoid; WH-clefts, as in What you’re going to do is go up the side of the allotments, and RWH-clefts (reverse clefts), as in That’s where you should be. Informants were influenced by the English to the extent that, where possible, they provided a natural rendering by means of a cleft.4 But this brought out clearly the fact that some languages simply do not have cleft constructions, and that other languages have one or two but not the entire range of cleft constructions found in English. The spontaneous map task dialogues referred to above constitute a second, and important, source of data for the discussion of discourse devices; in addition to the large set of English dialogues, there are eight Catalan, fourteen German, six Hungarian, and ten Bulgarian dialogues. They are valuable, inter alia, because they provide a wider range of highlighting devices than supplied by the informants via the questionnaire. In particular, they demonstrate the use of particles and word order in German and Hungarian, and the fact that even where IT- and WH-clefts are available in principle, as in Catalan, speakers may not use them. The corpus of spontaneous spoken Russian in Kapanadze and Zemskaja (1979), Zemskaja’s (1973: 347–380) discussion of the data, and dialogue from contemporary short stories and plays has provided an extensive range of highlighting devices in spoken Russian.5 The paper does not purport to give a detailed discussion of the data from any one language; indeed space precludes such accounts and specialists in any one of the languages cited will find the discussion limited and possibly superficial. Fortunately, the limited nature of the discussion is irrelevant, since the goal of this chapter is to

Focus in the languages of Europe

127

provide reliable cross-language data as a first step towards a typology of highlighting constructions. Readers might object that the map task data is not natural, inasmuch as the participants are placed in a situation untypical of everyday intercourse: the task is unusual, the informants sit in a recording booth wearing microphones, and the range of vocabulary and syntax is likely to be limited. These objections have some force, but the fact is that cross-language comparative analysis of discourse cannot be placed on a sound basis except by sacrificing some naturalness in return for the fixed discourse properties described in Section 2.1. Another disadvantage is that the map task may not generate the contexts in which a given type of cleft appears. For instance, the German map task dialogues contain no WH-clefts except for the couple of examples mentioned above, which were produced by a native speaker of German who had been living in the UK for two or three years. On the basis of the German map task dialogues alone analysts would be tempted to dismiss the WH-clefts as arising from constant exposure to English. As will be reported in Section 9.5.3, German does indeed have a WH-cleft construction, but it has a different function from the English WH-clefts and occurs in formal discussion. (Note that this disadvantage applies only to languages that have cleft constructions but not the same range with the same discourse functions as the English ones. The languages in our sample that have no cleft constructions at all – Finnish, Hungarian, and Turkish – are unaffected. And for English and German the map task data is supplemented by data from corpora of conversation.) As argued in Chapter 1, the difference between spontaneous spoken language and written language bears on the analysis of any area of sentence grammar (not least the sentence) but is especially obvious with respect to the organization of discourse.6 For example, Zemskaja (1973: 382) observes that in literary Russian it is the end of an utterance that is more important thematically, whereas in spoken Russian it is the beginning of the utterance (typically corresponding to the end of clause in Zemskaja’s data). (We go further than Zemskaja and suggest that word order by itself is relatively unimportant for highlighting in spoken Russian but does have a role to play in written Russian.) She cites, though with qualification, the view of Kovtunova (1969: 60–61) that word order has a minimum role in signaling information structure (aktual’noe cˇ lenenie) in spoken Russian because of the role of context, gestures, and intonation.

3. Results The main conclusions that will emerge from this and the next chapter are the following.

128

Jim Miller

(1) A major role in highlighting is played by particles. All languages have particles, though some make more use of them than others. For instance, in the spontaneous map task dialogues, particles were frequent in German, Hungarian, and Catalan, but less so in English. Many of the particles are deictic. In English, particles are typical of spontaneous spoken language, whereas in, for example, German, Finnish, and Hungarian, particles are typical of both speech and writing. Interestingly, the particles constitute a category for which there is no entry in the Eurotype list of abbreviations.7 (2) Certain languages have clefts, while others do not. Among the former, only English possesses and regularly uses three cleft constructions. Some languages have clefts in principle but their speakers prefer to use particles – see the discussion of the Catalan spontaneous map task dialogues in Section 10.4. Clefts appear to be a feature of languages spoken on or near the western seaboard of Europe, an area which has to be interpreted liberally in order to include Italian. But as one moves east in Europe, clefts disappear. (3) Deixis plays an important role. The crucial pronouns in the various cleft constructions are deictic, and many particles are deictic. The deictic properties of individual particles or constructions will be pointed out as the discussion proceeds. (4) Changes in word order, that is, preserving the construction type but rearranging the words as in Below the castle there’s woods (declarative main clause but locative PP in first or theme position) have to do with the distinction between given and new in that the fronted constituent is given. Where the fronted constituent is to be highlighted, a number of languages, such as Russian and German, often combine fronting with the occurrence of a particle. In the latter two languages, the fronting alone highlights a deictic or personal pronoun, while in other languages, such as Hungarian, Greek, Serbo-Croatian, and Bulgarian, any fronted constituent is highlighted. Fronting, particularly of a deictic, is the discourse equivalent of reverse WH-clefts in English. (5) In all the languages sampled, both speakers and writers use special constructions to highlight constituents referring to new entities. The structures range from yesno interrogatives (Have you got a great viewpoint?) to special existential structures (There’s baboons on my map). Other devices include special particles, such as the Russian vot, but the particles in this function appear to be typical of spoken language. (6) New propositions have their own highlighting devices. In spoken language these are particles or semi-fixed constructions such as the English the thing is or thing is, or conversational strategies such as speakers asking a WH-question and answering it themselves: What did we do then? We just picked up the phone and called the police, She must be worth, what? several million pounds. Written language makes use of special constructions such as clefts – What I think about

Focus in the languages of Europe

129

this theory is that it has very limited application – or formulaic expressions as in Let us now consider the view that there are no differences between spoken and written language. (7) Contrastive and noncontrastive highlighting is achieved by means of particles in spoken language, but typically by means of special syntactic constructions in written language. For example, IT-clefts are a favorite focusing device in written English but are rare in our data, both the map task dialogues and the spontaneous conversations. Word order alone is seldom used for contrastive highlighting. It can be so used in VS clauses in Russian, but even Russian with its very rich morphosyntax prefers particles (possibly accompanied by changes in word order), overtly contrasted clauses, and other syntactic devices. (8) Spontaneous spoken language makes heavy use of pitch for highlighting. Written language can make use of underlining or italic and bold fonts to capture part, but only a part, of the role of pitch.

4. Focus 4.1. Overview As announced in the Section 2.1, our approach is data-driven, in that we are interested in devices that highlight constituents/give them prominence/make them salient in the basic sense of standing out from the surrounding constituents in a clause or sentence. How do these devices link in with the major concepts of focus that have been proposed over the last thirty years? The short answer is that all the concepts have something in common; they involve the possible highlighting of a constituent and the information it carries, albeit for different reasons. Highlighting serves contrastive and noncontrastive purposes, and the latter include highlighting for the purpose of fixing new entities, or re-fixing old entities, in the listener’s attention. In addition, the concept of focus is invoked in accounts of how a given speaker and listener shift attention from one entity to another in a given context, thereby shifting the reference of pronouns, as described in the classic paper by Grosz (1981). The core concept of contrastive focus as employed by Halliday (1967, 1968) and Chafe (1976) turns out to have much in common with the concept of focus as used by Dik (1980), and even with two apparently very different concepts – Grosz and Sidner’s (1986, 1990), which has to do with referential status, and Vallduví’s (1993), which has to do with the transmission of information. In order to accommodate the various concepts in one framework, we add the notion of communicative goal as deployed by Levy (1979) and bring in the further concepts of macro-focus, relating to high-level, major transitions in a text, and micro-focus, relating to low-level, minor transitions. The important notion here is “transition,” which is also at the heart of Grosz and Sidner’s work.

130

Jim Miller

4.2. Halliday Halliday (1967, 1968) posits information units which are realized as tone groups. A tone group contains one obligatory component, a tonic segment, which may be a whole word or part of a word, and at least one optional component, the pretonic segment. One information unit is realized as one tone group, and the number and size of the information units in a given utterance reflect how the speaker has divided the message into blocks of information. Within each information unit the speaker selects a certain element or elements as points of prominence within the message. For example, (1) can be realized as (2a) or (2b), where the double obliques mark tone group boundaries and small capitals mark the position of the tonic segment. (1) (2)

John saw the play yesterday. a. b.

//John// saw the play yesterday// // John saw the play yesterday//

“(Information) focus” assigns the function “new” to what is within its domain. The constituent specified as new is the one marked out by the speaker for interpretation as carrying nonderivable information, either cumulative to or contrastive with what has preceded. In this respect, Halliday’s notion of information focus is very close to the notion proposed by Vallduví (1993) – cf. Section 4.5 below. Halliday emphasizes that what is new is what the speaker chooses to present as new. In a later work, Halliday (1985: 277) states that one form of newness frequent in dialogue is contrastive emphasis. While this is correct, we will keep contrast apart from newness to facilitate our account of expressions that are employed for contrastive salience and expressions that are employed for making new entities salient.

4.3. Chafe: Components of contrastiveness Chafe (1976) extends Halliday’s work by asking explicitly what contrastiveness is. With respect to the example RONALD made the hamburgers Chafe spells out the factors involved: (i) background knowledge that a set of hamburgers was prepared; (ii) a set of possible candidates for the cook; (iii) assertion of which candidate is the correct one.8 Contrastiveness is manifested principally by the placement of higher pitch and stronger stress on the focus of contrast, although IT-clefts are an alternative device. Like Halliday, Chafe regards the pitch phenomena as of primary importance and the use of syntactic constructions as secondary, but Chafe limits focus to contrast, unlike Halliday, who regards contrastiveness as only one property of focus – focused, new information can be cumulative to existing information or in contrast with it.

Focus in the languages of Europe

131

4.4. Dik: Extended notion of focus Dik (1980: 210–229) extends the concept of focus and presents a taxonomy of focus types. He draws a distinction between shared and nonshared, or new, information – the distinction depending on a given speaker’s assessment – and, like Halliday, treats focus as usually, but not necessarily always, marking information as nonshared. In the interests of communication, the nonshared information is made salient, either by intonation, constituent order, focus markers such as particles, or special focus constructions. Dik provides a hierarchy of different types of focus – the examples are from Mackenzie and Keizer (1990). All the examples apart from parallel focus are unmarked, the tonic falling on the final accented lexical item, but they all have in common that the speaker uses the tonic to make information salient, presenting it as unshared or new. In Halliday’s scheme, the second and third clauses in the parallel focus example, (3b), each consist of two tone groups, for example, //John// was nice//, with a tonic not just on the final accented lexical item but also on the first lexical item. Examples of the different types of focus are given in (3). Obviously the placement of the tonic is relevant only to speech. Speech shares with written language the syntax of the different types of focus but in speech, the syntax might well be reduced; for instance, in (3c) a possible spoken retort is simply coffee, with the appropriate pitch pattern and an appropriate voice quality.9 (3)

a.

b. c.

d.

e.

f.

Completive focus: Where is John going? He’s going to the market / to the market. Parallel focus: John and Bill came to see me. John was nice but Bill was boring. Replacing focus: John bought rice. No, he bought coffee. Restricting focus: John bought coffee and rice. No, he only bought coffee. Expanding focus: John bought coffee. Yes, but he also bought rice. Selecting focus: Would you like coffee or tea? Coffee, please.

Essentially, Dik takes Halliday’s distinction between cumulative and contrastive focus, draws out the various reasons why a speaker might wish to highlight a con-

132

Jim Miller

stituent, and extends the range of devices to be recognized as highlighting. That is, the general concept of contrastive focus is all very well, but once one looks at spontaneous examples of spoken or written discourse, a range of sub-types emerges, some of which do not involve the selection of one and only one candidate as in Chafe’s scheme.

4.5. Vallduví The concept of focus as new has been incorporated into an explicit informationhandling model by Vallduví (1993). Consider (4). (4)

a. b.

What did Sally give Andrew for his birthday? She gave him a CD of Jean-Michel Jarre.

Sentences encode a logico-semantic proposition and an information-packaging instruction. The proposition in (4) is give (Sally, something, Andrew). The information-packaging instruction indicates what part of a given sentence carries information and where and how that information fits in the hearer’s information store. (Strictly speaking, it indicates where and how the speaker thinks the information fits in.) In (4b) the sequence She gave him is taken to be the ground, which is what is already established in the speaker’s information model. “Being established” is to be understood in terms of a file card having been opened for the relevant entities. A CD of Jean-Michel Jarre is the focus, which is information to be added to the hearer’s information model either by adding information to, or changing information on, a file card – here the card for the entity given to Andrew. In the appropriate circumstances the hearer may have to open a new file card. Tail elements are ground elements that are not links. The ground consists of a link element and a tail element. The link points to a specific file card in the hearer’s knowledge store – in (4b) she and him point to the cards for Sally and Andrew. Link elements are typically pronominal and in speech carry low pitch and amplitude. In spoken English the answer to (4a) would typically be a CD of Jean-Michel Jarre, with a zero link element. As will be shown in Section 5, spoken and written language have different devices for indicating what is new, and different devices for signaling new entities as opposed to new propositions, and they both have ways of defocusing or backgrounding entities that are given. The concepts of ground and tail will be important for the discussion of defocusing in Catalan.

Focus in the languages of Europe

133

4.6. Grosz and Sidner: Focus spaces and transition Grosz and Sidner (1986, 1990) appeal not just to linguistic structure, which pertains to the analysis and representation of utterances, but also to the notions of intentional and attentional structure. The former has to do with the intentions of the speaker in producing a particular discourse or discourse segment and the latter – with the salience of entities, properties, relations, and discourse purposes and intentions. We are concerned here with attentional structure, since it, like focus, has to do with salience. The important concepts are focus space and transition rule. A focus space is associated with each discourse segment, and in it are collected representations of entities salient in that discourse segment – that is, entities that are mentioned explicitly or become salient as speaker and hearer work to produce and interpret utterances.10 Within focus spaces all the different types of focus described by Dik are relevant, and also the devices mentioned above for introducing new entities and new propositions into focus spaces. Of course, speakers can choose one of these devices or simply use a neutral syntactic construction, as in much of the dialogue analyzed by Grosz and Sidner. As a discourse is built up, speakers pass from one focus space to another and it is here that the concept of transition rule becomes relevant; speakers and writers may choose to make a given transition salient, and the text schema in (5), taken from Levy (1979), shows how they can do this. (5) a. b.

c.

A text schema I had a lousy day at work. (i) Monday is a horrid day anyway. (ii) When I arrived : : : (iii) At the second year lecture : : : (iv) Right in the middle of my afternoon nap : : : (v) Just as I was leaving : : : (i) At the second year lecture (ii) the first thing was that the overhead projector was playing up (iii) then the students hadn’t done any reading (iv) but what bugged me most was : : :

Essentially, a time or place phrase is put into theme position and these phrases connect and contrast with the previous text. (As Halliday maintains, choice of theme is not random.) The text schema in (5) shows that there are major and minor transitions; the transition from the previous topic of conversation to the lousy day at work is very large, the transition from the topic set (5b) to topic set (5c) is less important, and the transitions within sets (5b) and (5c) are relatively minor. Transitions in texts go beyond the scope of this chapter but are relevant for typology. The English text

134

Jim Miller

schema in (5) and transitions in a spontaneous Russian text are discussed in more detail in Miller and Weinert (1998: 203–209), invoking the concept of macro-focus for the major transitions and micro-focus for the minor transitions. The transitions involve contrast but also cohesion, and they are rendered salient by various syntactic and lexical devices. It is in large part the richness and multiplicity of discourse purposes and devices that have made us adopt a data-driven approach.

5. New individuals and sets 5.1. Focus spaces, file cards and entities Halliday describes as new any information that the speaker presents as not recoverable from the situation; that is, any information which the speaker supposes that the listener will not be able to recover. This relates to Grosz and Sidner’s concept of focus spaces, since effective speakers and writers signal clearly when a new entity is proposed as the center of attention, frequently by means of some highlighting device. Highlighting is applied not just to split-new entities but also to entities that have been in the focus space at the center of attention but have been replaced by other entities. It is also applied to entities treated as inferable without previous mention. Vallduví, following Heim (1982), talks in terms of the listener having a knowledgestore which has to be updated. The knowledge-store is thought of as a set of file cards, each carrying the denotation of an entity. When speakers use an indefinite noun phrase, they treat the relevant entity as new and the indefinite NP can be treated as an instruction to the listener to open a new file card for a new entity. This interpretation of the role of indefinite NPs is unexceptionable (and indeed quite traditional), but, particularly in spoken language, more than an indefinite article is employed. For instance, speakers (and writers) have available two special constructions in English, the classic existential construction of English, as exemplified in (6), and the possessive existential construction, as exemplified in (7).11 (6)

There’s a Phillips screwdriver in the left-hand drawer.

(7)

We have an old Volvo. It keeps going but it drinks petrol.

Even if the speaker/writer uses the existential construction, the distinction between the definite and indefinite article is still crucial. Compare (8bi) and (8bii). (8)

a. b.

Who can we get to look after the dogs while we’re away? (i) There’s the girl two doors along. She loves dogs. (ii) There’s a girl at work who’s crazy about Labradors.

(8bi) introduces the girl into the conversation, but she is not treated as new. Rather, the listener is reminded of her existence. In Vallduví’s terms, the listener is not instructed to create a new file card but to go to an already existing card.12 In informal

Focus in the languages of Europe

135

speech, the use of the demonstrative this is typical for the introduction of new entities: There’s this girl at work, she loves Labradors. It may be that speakers use this in order to bring a new entity right into the foreground; for present purposes the central fact is that this in the existential construction signals a new entity, whereas the or that signals that the speaker is reminding the listener of a given entity.

5.2. English 5.2.1. New entities in written texts The above remarks about definite and indefinite articles in the existential construction apply to both spoken and written language, the this construction is spoken language only, or in representations of spoken language in written texts. Written narratives have their own conventions, including the use of definite NPs for completely new entities – the reader is expected to wait and find out, or to work out, who or what the entities are. Different genres have different ways of introducing new entities. Cook books, as observed by Brown and Yule (1983: 174–175), set out all the new entities at the beginning of each recipe in the form of a list of ingredients and possibly a list of required utensils, and a similar technique is used in do-it-yourself manuals. Expository prose has a range of fixed constructions in which new entities are introduced as direct or oblique objects: Let us consider now another problem connected with the phoneme; I wish to turn to the unit of analysis known as the morpheme; Consider the following example; It is worthwhile taking a brief look at the branch of phonology called autosegmental phonology; etc. 5.2.2. NP-clause A very common construction in spontaneous spoken language consists of a noun phrase followed by a complete clause containing a pronoun anaphor of the noun phrase. The construction is found in most of the languages in our sample, including Hungarian and Finnish, but it is not clear whether it is present in Irish and Turkish. The construction highlights given entities that are being introduced or reintroduced into a text. For English, Quirk et al. (1985: 1416–1417) give the examples below. (9)

a. b.

This man I was telling you about – well, he used to live next door to me. The book I lent you – have you finished it yet?

The left-detached noun phrase does not carry the intonation nucleus but is highlighted, being in first position and not being part of the following clause. Moreover, it may be separated from the clause by a pause. An example from conversation is given in (10).

136

Jim Miller

(10)

C1: what are the people like to work with the drivers and that. A3: the driver he’s really friendly you get a good laugh with him and the boy that’s full time he gives you a lot of laughs and that.

(10) is an important example. One view of this construction is that it is associated with planning problems; the speaker completes the noun phrase, runs into difficulties, and replans the remainder of the syntax as a separate clause. Quirk et al. (1985: 1417) observe that the construction is convenient for speakers in that long noun phrases can be kept separate instead of being incorporated into a clause as an awkward subject or object NP. Their observation is correct, but the fact is that in spontaneous speech, noun phrases are typically short and the construction typically occurs with NPs such as the one in (10) – note the two NP-clause structures in A3 in (10), the first with a simple NP, the driver, and the second with a complex NP, the boy that’s full time.13 (10) was uttered without any breaks in intonation, pauses, or hesitations symptomatic of planning problems; this, together with the frequent occurrence of the construction in a number of languages, indicates that we are dealing with a regular construction of spontaneous spoken language and not with some performance error. 5.2.3. see and know Two other constructions for presenting or re-presenting landmarks do not occur in written English (unless the written text is imitating spoken English). See occurs in a number of syntactic structures in the map task dialogues: do you see X, can you see X, you’ll see X, you see X, see X. The last construction is interrogative, as indicated by the intonation and by the fact that the listeners regularly respond with yes or no. One participant injects a note of certainty by using the formula you’ll see a diamond mine on your map, you’ll see a graveyard. A sample of spontaneous conversation is given in (11). (11) S5: K5: S6: K6: S7:

see Mr. Y he’s got rid of whatsername the old secretary hmm she’s not there any more I like her she she was horrible ah she’s not bad see the joggers it’s funny Mr. Y he’s reorganizing the junior school and all the girls in primary five they can go jogging.

The particular details of the presentative construction in (11) are typical of Scottish English, and possibly of other varieties of English, but the use of a construction built around the verb see is common to all. In standard British English (11) would be rendered you see Mr. Y and you see the joggers. Note that S uses the see construction to set the scene and introduce the major participants, Mr. Y and the joggers; S falls back on the NP-clause construction to make another predication of Mr. Y – Mr. Y

Focus in the languages of Europe

137

he’s reorganizing the junior school – and to bring the “primary five” girls into the set scene – all the girls in primary five they can go jogging. The parallel highlighting of Mr. Y and the girls in primary five can also be interpreted as highlighting the connection between them. Similar to (you) see X, are constructions with know: you know the windows of the flagship?, you know how you’ve got Green Bay?, know where it says “secret valley”? The construction is interrogative, the direct object of know is always definite, and the construction is used to reestablish already-mentioned landmarks or to refer to landmarks assumed to be shared.

5.3. German The NP-clause construction is not peculiar to English. Clear examples, (12) and (13), occur in the German map task dialogues, and examples such as (14) are found in spontaneous conversation. (12)

Also well die they

der gelbe und the yellow and mmm begrenzen mmm border

der the die the

rote Fluß red river oberste linke Ecke. top left-hand corner

(13)

Der the der that

Vulkan volcano ist ungefähr auf gleicher Höhe wie der rote Fluß. is about on same height as the red river

(14)

Aber auf der anderen Seite eh ein eh but on the other side eh a eh vernünftig erzogene Kinder sensibly brought-up children denen merkt man an to-them notices one to wenn sie etwas wissen wollen. when they something to-know want ‘On the other hand eh a eh you notice when sensibly brought-up children want to know something.’

In many such examples, including (14), there are no pauses between the NP and the clause (prima facie, therefore, the construction is not necessarily associated with planning problems) and where there are pauses, they occur in the clause between the pronoun and the rest of the clause, as shown by the position of mmm in (12). The NP in the NP-clause construction is in the nominative case in (12)–(13), but (15)

138

Jim Miller

shows that the NP can be in one case, here the accusative, diesen Flußlauf, while the anaphor in the clause can be in another case, here the dative – dem. In (14) vernünftig erzogene Kinder is in the nominative but denen (which has a phonetically reduced realization in (14)) is dative. (15) is interesting in that dem is dative for the very good reason that folgen ‘follow’ governs its object in the dative case. If the speaker had been aiming to produce a single clause in which Flußlauf was the complement of folgen but had been derouted by some planning problem which interrupted the clause, we might have expected diesem Flußlauf, that is, an NP in the dative case as object of folgen. The fact that the NP is in the accusative case suggests that the NP was produced as a separate NP, not as part of the folgen clause. (15)

Ja und dies-en Flußlauf dem folgen wir jetzt. yes and this-ACC rivercourse that-DAT follow we now ‘Yes and this rivercourse we now follow that.’

5.4. French The NP-clause construction is not only present in written French but occurs frequently in the spoken language. The NP is given; it functions as a link with the preceding discourse, or refers to an entity inferable from the context. (16) comes from Wehr (1994: 623) and the transcriptions in Blanche-Benveniste (1991) contain a very large number of examples. (16)

A: B:

Il y a un film que j’ai envie de voir. there-is a film that I-want to see Oui, j’ai vu qu’ on le joue au “Studio” yes I-saw that one it plays at-the “Studio” mais les acteurs, je ne les connais pas. but the actors I not them know not ‘Yes I saw that is showing at “The Studio” but I don’t know the actors.’

The anaphor in the clause can be an oblique object, such as y in (17a) and en in (17b). (17) a. b.

L’ école quand j’ y vais je suis pas ravi. the school when I there go I am not overjoyed Tu sais la cuisine nouvelle on en parle beaucoup. you know the cuisine nouvelle one of-it speaks a-lot ‘You know nouvelle cuisine – they talk a lot about it.’

Barnes (1985: 54) includes examples in which the anaphor is ce ‘that’, and the hostess in an elegant French salon, surveying ice cubes scattered over the table, uttered (19).

Focus in the languages of Europe

(18)

Les choux à la crème, c’est bien. the cabbages with the cream that-is good ‘Cabbages with cream is good.’

(19)

Les pinces c’est nul. The tongs it-is zero ‘The tongs are useless.’

139

For introducing completely new and uninferable entities French has an existential construction introduced by il y a ‘there is’ and a possessive-existential construction, as in j’ai un X qui ‘I have an X that : : :’. The possessive-existential is regularly used for the first mention of entities specific for the speaker: j’ai un enfant qui est malade ‘I have an child who is ill’.14

5.5. Russian Spoken Russian has entity-highlighting constructions similar to the English NPclause construction discussed above. (Data from Zemskaja [1973: 239–247].) Speakers use the construction to introduce or reintroduce an individual entity into the discourse. (NB: The entities being introduced are nonetheless present in the context of utterance or inferable.) The extra-clausal NP is in the nominative case, whereas the anaphor in the following clause can be in any appropriate case. (20a) has the nominative pronoun ona ‘she’, the subject of pridet, and (20b) has the dative pronoun emu, dative case being required by ravno. (20) a. b.

(21) a. b. c.

A Lena / ona skoro pridet. but Lena / she soon will-arrive Miša / emu vse ravno. Miša / to-him all equal ‘It’s all one to Miša.’ Vasja / Vasja / Cvety flowers Sobaka dog

Vasja ušel v školu. Vasja has-gone to school / cvetov ja ne pokupala. / flowers I not have-bought / poly vsegda grjaznye. / floors always dirty

In (21a) the second occurrence of Vasja is nominative case, as subject of ušel. Cvetov is genitive since it is the direct object of the negative verb ne pokupala. Zemskaja (1973: 244) gives as a typical context for (21b) žal’ cvetov u nas net ‘It’s a pity – we’ve got no flowers’. (21b) picks up the given entity with cvety and the clause conveys a proposition – ‘I haven’t bought flowers’. From Zemskaja’s comment it can

140

Jim Miller

be inferred that the first occurrence of Vasja in (21a) is also given. ((21c) is significant in that the clause poly vsegda grjaznye contains no noun or pronoun functioning as an anaphor to sobaka. This property makes it similar to topic-prominent constructions in languages such as Mandarin Chinese and Japanese, as discussed in Section 5.10 below.) Russian has flexible word order, but an examination of different word orders in Kapanadze and Zemskaja (1979) soon reveals that word order on its own is never used for highlighting; where highlighting occurs there is always a particle. This finding accords with the general thesis of Fernandez (1994), reported in Chapter 1, that particles play an important role in the organization of discourse. Particles will be discussed in Section 7 in connection with the German, Catalan, and Hungarian map task dialogues, but it will be useful here to take a look at the highlighting function of particles as exemplified in the Russian data. Russian has a large number of particles or micro-focusers whose purpose is to lend local salience to a constituent but without marking a major transition. Because of their highlighting function, they can be related to Grosz and Sidner’s concept of a transition from a focus space containing a given collection of entities to another focus space containing a (possibly overlapping but) different collection. Such microfocusers are že, kak ras, nepremenno, and vot. The first three are exemplified in (22), (23) and (24). (22)

Ivan cˇ asto pokidaet kabinet. Ivan often leaves study ‘Ivan often leaves his study.’ Petr že nikogda ne kidaet kabinet. Peter particle never not leaves study ‘Peter now, he never leaves his study.’

(23)

Kak raztab segodnja on mne govoril o vas. as once today he to-me was-speaking about you ‘It was just today that he was speaking to me about you.’

(24)

On nas zastavil izuˇcat’ arabskij nepremenno šrift-to//. he us made to-study Arabiˇctab absolutely script-that ‘He made us study Arabic script.’

(23) with kak raz is translated by means of an IT-cleft but the Russian example does not involve any major restructuring of the basic main clause construction. The time adverb segodnja is in theme position, whereby it automatically becomes salient, but, repeating the comment above, word order on its own rarely serves to highlight constituents. For instance, Bivon (1971) states that the first constituent in the SVO and OSV orders is occasionally emphatic, but his examples are those in (25a) and (25b).

Focus in the languages of Europe

(25) a.

b.

141

Imenno doˇc’, mysli o nej, dolžny byli zapolnit’ namely daughter thoughts about her supposed were to-fill pustotu. emptiness ‘It was his daughter, his thoughts about her that were supposed to fill the emptiness.’ K sožaleniju, imenno takoj xarakter nosit : : : ob"edinenie unfortunately namely such character bears union social-demokratov : : : of-social-democrats : : : ‘Unfortunately, this is exactly the character of the social democrats.’

In both examples above, the first NP is modified by the micro-focuser imenno. In any case, since SVO is the most frequent and the most unmarked order, it would be strange if occurrence in the S position alone conferred salience on constituents. Another very frequent focusing item is vot, which is a powerful deictic. It is used in contexts in which speakers of English could use see or look and is often accompanied by a pointing gesture, as in vot idet avtobus ‘look/see – the/a bus is coming’, or as in (26). (26)

Vot u mosta u ètogo /u mosta u ètogo idet granica. see at bridge at this /at bridge at this goes frontier ‘See, the frontier is at this one, at this bridge.’

The repetition of u mosta u ètogo is a second means of giving salience to the bridge but the initial salience comes from vot. Returning to (24), note that in addition to nepremenno preceding the highlighted NP there is a deictic item -to following šrift. Another example of this -to is in (27). This Russian example can be appropriately translated into English by means of an IT-cleft; the -to in Russian picks out Molˇcok from the set of relevant individuals in the given situation, namely the set of gravediggers working in the cemetery, and IT-clefts perform an analogous function in English. (27)

Štatnym zemlekopom byl odin Molˇcok, brigadir. Na nego-to state gravedigger was alone Molˇcok brigadier On him-that i pisalis’ narjady. and wrote-selves orders ‘Molˇcok, the brigade leader, was the only staff gravedigger. It was his name that was used for all the paperwork.’ (Sergei Kaledin The Humble Cemetery [Smirennoe Kladbišˇce])15

142

Jim Miller

5.6. Hungarian The Hungarian map task dialogues provide examples of the NP-clause construction, as in (28). The anaphor in the clause can have any grammatical function. (28) a.

b.

Igen de most az aranybánya ahogy haladok, az bal yes but now the gold-mine as I-go-along it left-hand kézröl maradt el, igaz? on stayed particle right? ‘Yes but now as I go on the gold mine has stayed on the left, right?’ de a vízesést azt alulról kerüld meg but the crest-falls that below-from go-round particle ‘But the Crest Falls, go round them from below.’

In (28a) az is subject, while in (28b), azt is direct object.

5.7. Finnish (29)

Tämä MTK, sehän ei ole ollenkaan voimakas järjestö. this MTK it-particle NEG it-is at-all strong organization ‘This MTK is not a powerful organization at all.’

5.8. Turkish The NP-clause construction appears to be absent from Turkish. When asked independently whether they had heard Turkish speakers using examples analogous to The driver you get a good laugh with him, the two Turkish informants declared that they had not come across such a construction. The closest example they offered was one with a nonsubject NP moved into clause-initial position. The neutral order in (30a) became the order in (30b), with bize functioning as link NP and carrying a rising pitch. (30) a.

b.

Oya biz-e resim-ler-i-ni göster-di. Oya we-DAT picture-PL-POSS.3-ACC show-PST(3SG) ‘Oya showed us her pictures.’ Bize, Oya resimlerini gösterdi.

5.9. Irish and Scottish Gaelic The Irish informant was unclear as to whether Irish has an exact equivalent of the driver, you get a good laugh with him and suggested that there is no equivalent of see

Focus in the languages of Europe

143

the driver? but speakers do use the equivalent of You know/remember the driver? One major difficulty in the investigation of spoken constructions is that speakers do not have sound intuitions about constructions that appear only in the spoken language. (This is relevant to the above discussion of Turkish.) Barbara Wehr (pers. comm.) observes that the Hartmann Corpus of spoken Irish contains the examples in (31). (31) a.

b.

Bhí mo mháthair mhór sa anis, bhí sí pósta. was my mother great this now was she married ‘My grandmother now, she was married.’ Bhí Micil Pháidín as a’ Meáll Rua, bhí sé lá amuigh i was Micil Pháidín from the Meáll Rua was he day out in Carraig a Meacan. Carraig a Meacan ‘Micil Páidín from Meall Rua, he was out one day in Carraig a Meacan.’16

(32) and (33) are examples of the NP-clause construction in Scottish Gaelic. (32) is from a television interview and (33) is from a private conversation. (Both examples were recorded during fieldwork.) (32)

Scottish Gaelic Am maighstir sgoil’ a bh’ agam DET master school REL was at-me ’s e Domhnall MacPhàil. is it Donald MacPhail ‘The school master that I had, it’s Donald MacPhail.’

(33)

I fhèin ’s e caoraich a th’ aice. she herself is it sheep REL be at-her ‘She herself, it is sheep that she has.’

Gillies (1993) notes the construction as common in speech and suggests that its development has been triggered by the attrition of the copula and/or the elimination of synthetic verb forms. The broader picture suggests quite strongly that the construction is typical of spoken language; it is certainly widespread in Europe, both in Indo-European and in other language families.

5.10. Bulgarian The Bulgarian map task dialogues have several examples of the NP-clause structure. Being in a rich, natural context these examples give us an insight into the discourse function of the Bulgarian construction. A typical example is reproduced in (34). (34)

Sinija okean mi se namira blue-the ocean to-me self finds

144

Jim Miller

veˇce vdjasno ot tova, koeto s˘um dostignala tuka. already to-right of that which I-am having-reached here Belija dom az go zaobikolix otljavo. white-the house I it have-gone-round on-the-left Ljavata mi r˘uka. left-the to-me hand ‘The blue ocean is to the right of where I’ve got to. The white house I’ve gone round on the left. My left hand.’ The “white house” and “blue ocean” are landmarks on the map. The utterances in (34) follow a discussion about the landmarks and whether the route is to the left or the right of this or that one. The speaker contrasts the position of the blue ocean on her right with the position of the white house. Note, however, that the focal accent falls on otljavo. A speaker of English might have said the white house now I went round that on the left or even as far as the white house is concerned I went round that on the left. (But the latter does sound too formal for the map task dialogues.)

5.11. Topic-prominent and subject-prominent languages Li and Thompson (1976) proposed that languages could be divided into topic-prominent, subject-prominent, both topic- and subject-prominent, and neither topic- nor subject-prominent. Typical of topic-prominent languages is the construction exemplified in (35). (35)

Mandarin Chinese Nè-chang hu˘o That-CLF fire xìngkiu xi¯aofang-dùi laí de kuái. fortunate fire-brigade come particle quick ‘With that fire/as for that fire, it was fortunate the fire brigade came quickly.’

Kiss (1998: 687–706) presents a critique of Li and Thompson’s hypothesis and develops more rigorous definitions of topic-prominent and subject-prominent languages which, for present purposes, we need not explore. Kiss (1998: 693) comments that very few of the languages of Europe she investigated were not topic-prominent; she takes the Celtic languages, especially Welsh, Irish, and Scottish Gaelic, to be neither topic- nor subject-prominent, and she tentatively suggests that the subject-prominent type may not be represented in Europe at all. Miller and Weinert (1998: 363–366) also discuss Li and Thompson’s hypothesis. They focus on Russian and the construction illustrated in (21), especially (21c).

Focus in the languages of Europe

145

(21a)–(21c) are typical of spontaneous spoken Russian but do not occur in written Russian; Miller and Weinert propose that some languages have topic-prominent constructions in their spoken varieties but subject-prominent constructions in their written varieties. Certainly the data from spoken Irish in Section 5.9 indicate that what applies to Russian may also apply to Irish, and that Kiss’s view of the Celtic languages might not be justified. It is now clear that this area of language structure and function requires detailed investigation of individual languages and solid crosslanguage comparison.

6. New propositions 6.1. English Other types of entities regularly introduced into discourses are properties and propositions. These may be brought in without fanfare but can be made salient both in writing and speech. The examples in (36) are typical of written English. (36) a. b. c. d. e.

X has one interesting/obvious/relevant/important property, namely : : : Another property of X that must be taken into account is : : : We turn now to a less/more attractive property of X. It is worthwhile remarking at this point that : : : Consider for a moment the fact that : : :

Spoken English has a variety of highlighting constructions that introduce properties and propositions into a discourse. The devices are not used in written English (unless the writer is devising dialogue) and are relatively short: like see, they serve to make a property or proposition salient, to concentrate the listener’s attention on it. 6.1.1. the thing is A very common construction is the thing is, which can be used to focus the listener’s attention either on properties or on propositions. Examples of properties are in (38)– (39), and an example of a proposition is in (37). (37)

A: B: C:

What about Edinburgh do the people go up there oh yeah a lot oh aye especially at night they go to the pictures but the thing is if you go to the pictures if you go to the late show you’re you’ve to run for buses

This example is interesting in several respects. It is typical of the data in that the thing is is never followed by a complementizer and there is regularly a pause between it

146

Jim Miller

and the next clause, which has its own pitch contour. These properties are compatible with a loose construction in which the thing is and you’ve to run for buses are juxtaposed rather than integrated. Variations on the the thing is construction are given in (38) and (39). (38)

A5: the only thing with Beth Beth’ll not spend money she’s mingy she’s really mean she wouldn’t give you two halfs for a one

Note the lack of a copula connecting the NP the only thing with Beth and the clause Beth’ll not spend money. (39)

A:

that’s the bad thing about the halls of residence there’s always people knocking on your door

In (39) the thing phrase is part of a separate clause. The construction is analogous to reverse WH-clefts which serve to halt the progress of an interaction and to sum up the section of dialogue that has just been uttered. (39) is preceded by one of the participants, a first-year university student, relating how she had tried to write an essay the previous evening, had sat down at her desk at 7pm but had not put pen to paper till 10pm because of people calling in to have a chat and drink coffee. The speaker in (39) sums up this aspect of life in the halls of residence, that pointing back to the interruptions narrated by the previous speaker. WH-clefts are discussed in Section 9.5.1, but (40) is appropriate to this section because, although containing the free relative clause what surprises you, it does not contain a WH-cleft construction. Instead the WH-clause is the direct object of a verb of telling. (40)

A:

I tell you what surprises you em//as I say : : : it’s marvelous how many people stayed in such a really small area

The classic WH-cleft is suitable for a relatively small amount of information that is to be highlighted in a well-planned piece of syntax. Not only is planning time very brief in spontaneous dialogue, but speakers regularly have large amounts of information to convey. I tell you what or I tell you what surprises you is an appropriate highlighting device, since it does not require, indeed cannot have, a complement clause. One clause does the highlighting and a separate clause carries the proposition to be communicated. 6.1.2. see introducing propositions See highlights propositions as well as individuals. In (41) it highlights propositions presented as a precondition for success in drawing the next section of route. The propositions are encoded in conditional clauses, which are not linked to a main

Focus in the languages of Europe

147

clause. Rather, each conditional clause is highlighted by See, and the punchline is encoded, not in a main clause, but in a noun phrase, that wee bit. This noun phrase is reminiscent of reverse WH-clefts, in which that points to the referent described in the preceding text. In fact this sequence could have been concluded with that’s where you go. Note that the final part of A’s instruction is encoded in a free relative, where the pilot would go, likewise highlighted by See. As in Section 5.2.3 we would expect you see in standard British English. (41)

A: B: A: B: A: B: A:

then you go down to the crashed aeroplane and turn a wee bit which hand side of the crashed aeroplane the the right wing if you look at it the it’ll be on your left wing it will be on the side that’s got the b the side that’s got the Burnt – written right see if you come down straight down right see if you come round the left of the hills right uh huh see if you go straight down but not go straight to the aeroplane right see where the see where the pilot would go that wee bit

Finally, one very common construction should be mentioned – entity-introducing, but also entity-changing: What about school? Do you enjoy the lessons? This example was used by the speaker to change the topic of conversation to a new topic which he thought might provoke more discussion. Also common in radio and TV discussions, but not in the HCRC databases, are as far as X is concerned, or as far as X. The much cited as for X is not in the databases and is not frequent in radio and TV discussions.

6.2. German German has proposition-introducing constructions analogous to the English ones listed above. Examples are listed in (42). (42) a. b. c. d.

das Problem ist ‘the problem is’ das Dumme ist ‘the stupid thing is’ die Schwierigkeit ist ‘the difficulty is’ das Gute ist ‘the good thing is’

These expressions either involve a straightforward NP, as in (42a) and (42c), followed by a complement clause, or an adjective functioning as a neuter noun, as in

148

Jim Miller

(42b) and (42d). They have an important property in common with WH-clefts – they point forward. In contrast, the clause das ist die Schwierigkeit in (43) points back, which is one function of reverse WH-clefts. (See Section 9.5.1 for a definition of the various cleft constructions.) Less integrated structures also occur – not as unintegrated as many of the constructions discussed in Chapter 1 but (43) is less integrated than Die Schwierigkeit ist daß, with a complement clause embedded in the main clause. (43) has a main clause (‘that is the difficulty’) followed by an adverbial clause of reason. The clause ich glaub ‘I believe’ is problematic, since it could be taken as a comment clause or as the main clause, with das ist die Schwierigkeit as complement of glaub. Whatever the syntactic analysis, das ist die Schwierigkeit points back, and the weil clause elaborates Schwierigkeit. In this respect, the construction is analogous to reverse WH-clefts in English, as in that’s the problem – he doesn’t understand the theory. (43)

Ja und ich glaub’ das ist die Schwierigkeit yes and I believe that is the difficulty d-ehm du du ehm bist ja jetzt nach y-ehm you you ehm are of-course now to

jetzt weil now because links gegangen. (the)left gone.

6.3. Greek Greek has constructions analogous to the English structures in Section 6.1 and to the German structures in Section 6.2. (44)

To thema me to Giani ine oti the thing with the John is that

(45)

To kalo tis olis istorias ine oti the good the-GEN whole-GEN story-GEN be-3SG.PRES that ‘The good part of the whole story is that : : :’

(46)

To kako me to Solen ine oti the bad with the Solen is that

6.4. Russian Spoken Russian has a number of constructions that allow speakers to focus on propositions. Examples are given in (47)–(48). (47)

– –

Leša, slušaj Slyšiš’? Leša listen do-you-hear Nu? what

Focus in the languages of Europe



(48) a.

b.

c.

d.

149

Takoe delo: zabud’, cˇ to bezxoz kopal. such thing forget that ownerless dug Ponjal? Normal’naja rodstvennaja mogila, ponjal? got-it normal relatives grave got-it ‘This is what: forget you dug that ownerless (grave). Got that? It’s an ordinary relatives’ grave, got that?’17 (Sergei Kaledin The Humble Cemetery [Smirennoe Kladbišˇce])

On teper’ ved’ u nas cˇ to? he now know at us what Každyj den’ po utram truscoj begaet. each day in mornings at-jog runs ‘Do you know what he’s up to now? He goes jogging every morning.’ On cˇ to? Na jug xoˇcet ètim letom? he what to south wants this summer ‘What does he want? To go to the south this summer?’ A vy tuda kak? Elektriˇckoj? but you there how by-electric-train ‘How did you get there? By electric train?’ Nu cˇ to že? OK what then Ja momental’no otpravil telegrammu Zorovu. I immediately sent telegram to-Zorov ‘OK, what did I do then? I immediately sent a telegram to Zorov.’

Takoe delo in (47) is reminiscent of I tell you what surprises you in (40); both takoe and what are deictics and point forward. Note that the English translation has this and not that. That’s what points back and drives home a point that the speaker has made; this is what points forward to something the speaker is about to say. In examples (48a)–(48d) the speakers ask a question which they immediately answer themselves. As mentioned in the discussion of WH-clefts in Section 9.5.1, the WH-words and their equivalents in other languages are indefinite deictics and the WH-words in the above examples, although traditionally classified as interrogative pronouns, are nonetheless nonspecific deictics pointing forward to an upcoming specification. Note the additional highlighting item ved’ in (48a). Vot, like see, is used to bring new propositions into focus, as in (49). (49)

Vse vremja/ splošnye povoroty // all time continuous bends // vot tebja kaˇcaet / iz storony v storonu. particle you it-rolls / from side to side

150

Jim Miller

A suitable English translation, one that preserves the simple syntax of the Russian, would be along the lines of like, you get rolled from side to side or you get rolled from side to side, see. (50)

Oni oˇcen’ blizko drug k drugu podxodjat // they very close one to other approach // i vot po kamennomu ložu teˇcet reka. and particle along stone bed flows river ‘They (the cliffs) are very close and there in a stone channel there’s a river.’

Vot in (50) could be interpreted as introducing a new entity, the river, but vot can also be taken as focusing the listeners’ attention on the entire proposition, which includes a mention of the river. The speaker is a young woman recounting her adventures with a walking group in the mountains of the Crimea and vot can be taken both metaphorically, as a highlighter, and literally, as a linguistic pointer. (Note the analogous use of voyez ‘see’ in the French example in (55).) The speaker invites her listeners to imagine themselves in this awesome canyon which looks as though it has been molded in one piece and says that in the stone were hollows in which floated autumn leaves of different hues. Her assertion of this last proposition is given below, the important feature being the occurrence of the phrase vot tak vot (tak = ‘thus’), which is described in the transcript as accompanied by a gesture. In this example vot tak vot follows the proposition but nonetheless highlights it.18 (51)

I tam v kamne takie mmm vpadiny/ and there in stone such mmm hollows/ i v nix voda and in them water os- list’ja raznocvetnye/ plavajut / vot tak vot. aut- leaves different-colored/ are-floating / particle ‘And in the stone there’s such–ehm–hollows and there’s water in them and there’s aut–different-colored leaves floating, just like this.’

There are examples in our data of vot alone following the proposition highlighted by it. One such instance is (52). (52)

Reka teˇcet / ona nazyvaetsja Kokkozka // river flow / it calls-self Kokkozka // Kokkozka // vot // gornaja reka. Kokkozka // particle // mountain river ‘There’s a river it’s called Kokkozka, Kokkozka that’s its name, a mountain river.’

Focus in the languages of Europe

151

The force of vot here is almost ‘yes, that’s right; that’s the name’. The name of the river is repeated too, probably because it is unfamiliar both to the speaker and her listeners. Highlighting can be achieved by simple repetition, as in (53). (53)

Ona ètogo ne xotela // ne xotela. she this not wanted // not wanted ‘She didn’t want this at all.’

6.5. French Blanche-Benveniste (1991: 124–125) supplies parallel examples from French, both for introducing new entities and new propositions. She gives an example of question and answer produced by different speakers but points out the very close relationship between questions and answers produced by one and the same speaker, especially when they contain the same verb. (54)

Et dans les protistes qu’ est-ce qu’ on trouve on trouve les and in the Protista what is-it that one finds one finds the bactéries les champignons et les algues. bacteria the fungi and the algae ‘What do you find in Protista? You find bacteria, fungi and algae.’

(55)

Voyez à Grenoble là dernièrement see-IMP at Grenoble there recently il y a combien il y a il y a huit jours there-is how-much there-is there-is eight days il y a une dame elle était : : : there’s a woman she was : : : ‘See at Grenoble there recently how long ago, eight days ago there’s a woman–she was : : :’

Similar devices can be found in spontaneous spoken English – cf. She must be what – sixty at least and So what did she do? She told him exactly what she thought of him – and it is probable that the technique whereby one and same speaker poses a question and answers it is widespread as a highlighting device in spontaneous speech.19

7. Russian word order 7.1. Introduction: Siewierska and Bivon It is worthwhile taking a brief look at word order in Russian both in connection with the function of word order change and with the difference between spoken and writ-

152

Jim Miller

ten language. It is also worthwhile reminding ourselves of two important points that recur throughout this chapter; that non-neutral word orders are often accompanied by particles and that the analysis of word order in discourse requires the careful distinction of text type and of the textual context in which a particular clause or sentence occurs. Given the flexible nature of Russian word order, we could reasonably suppose that changes in word order function as a highlighting device. The supposition would be reasonable but wrong, or at least partially wrong – see Section 5.5. We have already seen that in the Russian data from the oral questionnaire word order on its own does not highlight constituents; where there is a left-moved constituent, there too is a particle. One well-known analysis of Russian word order, Bivon (1971), states that S in the word order SVO and O in the order OVS can be emphatic, that is, what is here described as highlighted. As we have seen from Bivon’s examples, (25a) and (25b), the particle imenno plays the biggest part in the highlighting. Siewierska examines word order in a sample of written Polish – novels, expository texts, and biographies. The idea that word order might relate to highlighting is not entertained. She tests two major hypotheses: Hawkins’ early immediate constituent (EIC) recognition, according to which there is a preference for linearization patterns that allow the quickest recognition of constituent structure (cf. Hawkins 1990), and the classical hypothesis that constituents carrying given information precede constituents carrying new information.20 Siewierska (1993) concludes that it is only for SVO clauses that the EIC motivates a higher percentage of examples than the topic-comment hypothesis. For all other five orders, the latter motivates a significantly higher percentage than the EIC. The EIC, whose implementation rests on ratios, also fails to motivate SVO examples in which the object NP is very long and the S very short. Moreover, the work of Givón and his colleagues (cf. Givón 1983) has shown that subject NPs have a far higher topicality rating (in terms of referential distance, that is, the distance in words from the previous mention of their referent) than object NPs; that is, new entities are introduced by means of full NPs; if they become topical and are mentioned again, the subsequent mentions are made by means of pronouns or null NPs. Topical entities also tend to be referred to by means of subject NPs, and it follows that subject NPs tend to be short. The topic-comment principle by itself is not sufficient, however, and Siewierska invokes Dryer’s alternative weight principle, which states that long constituents prefer final position over other positions, and initial position over internal position.

7.2. Word order in spoken and written Russian and Polish The analysis of a small corpus of written Russian yielded results very different from Siewierska’s. The percentages of different word orders in the data examined

Focus in the languages of Europe Table 1. Siewierska – written

Table 2. Bivon – mainly written

Word order

Word order

SVO VOS OVS VSO SOV OSV

% of total 72.5 9.5 7.4 6.5 2.4 1.5

SVO OVS OSV VOS SOV VSO

153

% of total 79 11 4 2 1 1

Table 3. Miller – spoken

Word order SVO SOV OSV OVS

% of total 52 34 11 3

by Siewierska, Bivon, and this writer are shown in Tables 1–3. Siewierska’s data is exclusively from written texts, Bivon’s is mainly from written texts, and for this chapter spontaneous spoken data was analyzed, a sample of the transcripts published in Kapanadze and Zemskaja (1979). The main points here are: 1. In all three bodies of data the most frequent order is SVO. 2. The least frequent order in the Polish data, OSV, is the third most frequent in the written Russian data. 3. V-initial orders are more frequent in the written Polish than in the written Russian data. 4. SOV is second bottom in the tables for written Russian and Polish, but is very frequent in the spoken Russian corpus, so much so that SVO in the latter constitutes only 52 % of the total, compared with 72.5 % in written Polish and 79 % in mainly-written Russian. 5. There are no V-initial examples in the sample of spoken Russian. There is, however, a large number of clauses consisting of just a finite verb. 6. In the spoken Russian the large number of SOV examples can be largely explained by a preference for putting pronominal objects into preverbal position. Table 4 below indicates the high number of pronoun objects in the SOV order, contrasted with the high number of objects in the SVO order consisting of a noun or some other constituent plus a noun.21 However, even the NPs labelled “longer” are quite

154

Jim Miller

Table 4. Types of subject and object NP

SOV

Noun Pronoun Longer Total

OSV

OVS

SVO

S

O

S

O

S

O

S

O

2 16 0

1 15 2

0 8 1

3 6 0

0 1 1

0 2 0

2 17 1

9 4 7

18

9

2

20

simple compared with the NPs that occur in written language of the sort examined by Siewierska and Bivon. Examples are given in (56). (56)

vsja sem’ja tak nazyvaemyj skobar’ ètot oberkonduktor i mexanik kto-nibud’ iz vas èti pomešˇciki kakoj krupnejšij russkij poèt kto-nibud’ iz carstvujušˇcego doma list’ja raznocvetnye èta štuka takoj krepkij fundament on sam ego sestra

‘whole family’ ‘so called “skobar” ’ ‘that chief-guard and driver’ ‘somebody from you’ ‘these landowners’ ‘what most-important Russian poet’ ‘someone from royal house’ ‘leaves differently-colored’ ‘that thing’ ‘such strong-foundation’ ‘he himself’ ‘his sister’

7.3. Clause position in written Russian In most of the O-initial examples in the written Russian, O is given, though there are cases where a long object NP, not given, is placed first in the clause to allow a very short subject NP to come at the end. This jibes with comments by Zemskaja to the effect that in spoken Russian, first position in the clause is important, whereas in written Russian it is the last position. The following passage of written Russian illustrates the positioning of given constituents in clause-initial position in order to put new constituents at the end. (57)

[Early Greek schools of philosophy are listed, including the Megarans. The author goes on to give more details about the Megarans]

Focus in the languages of Europe

a. b.

c.

d. e.

155

Široko pol’zovalis’ u megarikov kosvennye dokazatel’stva. widely used-selves at Megarans indirect proofs To antonomii megarikov then antonymies of-the-Megarans imejut specifiˇceskuju logiˇceskuju prirodu. have specifiˇctab logical nature Osobenno proslavilsja svoimi paradoksami (“lžec”, “kuˇca”) especially was-famed for-his paradoxes liar heap Evbulid uˇcitel’ ne tol’ko Demokrita, Eubulides teacher not only of-Democritus no i Appolonija iz Kireny i istorika Eufanta. but also of-Appolonius of Cyrene and of-historian Eufantos Formulirovka paradoksa “lžec” takova : : : formulation of-the-paradox (the)liar thus vtoroj paradoks Evbulida pod nazvaniem “kuˇca” obyˇcno second paradox of-Eubulides under name heap usually formulirujut tak: they-formulate thus ‘Indirect proofs were widely used among the Megarans. The Megarans’ antonymies have their own particular logic. Particularly famous for his paradoxes “the liar” and “the heap” was Eubulides, the teacher not only of Democritus but of Appolonius of Cyrene and of the historian Eufantos. The paradox “the liar” is formulated thus : : : The second paradox of Eubulides, known as “the heap”, is usually formulated thus: : : :’

In (57a) megarikov is given and kosvennye dokazatel’stva is new. In (57b) antonomii is given and specifiˇceskuju logiˇceskuju prirodu is new. In (57c) svoimi paradoksami relates to the specific logical nature mentioned in (57b) and is thus given, and in any case the important information is the name of the famous philosopher and some of his properties. This information goes into the more prominent clause-final position. At the same time clause-final position contains a long noun phrase, with the relatively short predicate phrase in clause-initial position; clearly syntactic weight, and information weight, does have a part to play. In (57d) paradoksa “lžec” is given, referring back to svoimi paradoksami in (57iii) and tak is not only new but points forward to the next sentence. (57d) is followed by an account of the Liar paradox, which is followed by a paragraph break. The next paragraph begins with (57e), in which vtoroj paradoks Evbulida is given and acts as a bridge from the previous paragraph. The English translation reflects the Russian word order for (57c) and (57e), but only by using a passive in the latter.

156

Jim Miller

7.4. Word order as a highlighting device Russian word order alone does play a role in highlighting constituents, but only in clauses with one-place predicates. The S in VS clauses differs from the S in SV clauses as follows: it has greater referential distance, it is rarely pronominal, and it is typically longer. These properties reflect the typical function of VS clauses, namely, to introduce new entities into a discourse. However, VS order is also used for new episodes in an on-going narrative, as in (58), and for contrast, as in (59)–(61).22 (58)

A: B:

Skaži na kakoj say on which Nu vot particle particle

den’ vy sobralis’. day you got-together znaˇcit vyšli my utrom. that-is came-out we in-morning

(59)

Kto-to dolžen byl na cˇ to-to rešit’sja. Rešilas’ Šura. someone obliged was on something to-decide decided Šura ‘Someone had to decide on something. It was Šura who decided.’

(60)

Sam že on kopal redko. Kopali rebjata. self particle he dug rarely dug boys ‘But in fact he did hardly any digging. It was the boys who did the digging.’

(61)

Ne važen predmet oˇcarovanija, not important object of-fascination važna žažda byt’ oˇcarovannym [literary] important thirst to-be fascinated ‘It is not the object of fascination that is important. What is important is the longing to be fascinated.’

Finally, contrastive focus can be created by using either the fixed structure ne to cˇ to + infinitive followed by another negative clause which enjoys contrastive salience, or by using the fixed structure malo togo, cˇ to + clause. (To and togo are deictic, roughly equivalent to the English ‘that’, and cˇ to is a complementizer.) In both (62) and (63) a deictic pointing to one event is negated; the general structure is ‘not this but that’. Not vstat’ and not on sam rabotal but (only there is no overt adversative coordination) emu kazalos’, cˇ to : : : and on i materi s Nastej daval ukazanija. These two examples remind us how important a role deixis plays in language in general and in focusing constructions in particular, a role that is central to the English cleft constructions discussed in Section 9.5.1. (62)

Ne to cˇ to vstat’, not that that to-stand-up

Focus in the languages of Europe

157

emu kazalos’ cˇ to on ne možet otkryt’ glaz. to-him seemed that he not can open eyes ‘Far from standing up, it seemed he couldn’t even open his eyes.’ Malo togo, cˇ to on sam rabotal, little of-that that he self worked on i materi s Nastej daval ukazanija. he and to-mother with Nastja gave instructions ‘Quite apart from working himself, he even gave instructions to mother and Nastja.’

(63)

8. Reduction in salience 8.1. Overview Speakers and writers treat given information as recoverable by the listeners or readers. Since given information is supposedly recoverable and contrasts with information that is new and not recoverable (cf. the résumé of Vallduví’s framework in Section 4.5), the constituents that convey it are not highlighted and in fact are regularly reduced. Major phrasal categories are reduced to pro-elements or zero, as exemplified in (64). (64) a.

b. c.

The students went into the examination hall. They wrote their names on the script books and began to read the questions. Bethan went to the exhibition. Catriona couldn’t. I was ill yesterday. So was my wife.

In (64a) the NP the students is reduced to they in the second sentence and reduced further to zero in the second clause of that sentence, [: : :] began to read the questions. In (64b) went to the exhibition is replaced by could – the negative n’t adds new information – and in (64c) ill is replaced by so. (64a)–(64c) are presented as written language and organized into sentences, but in structure and vocabulary they are appropriate examples of spontaneous spoken English. This is an important comment, because ellipsis is far more frequent in spontaneous spoken language than in written language. In what follows, we concentrate on ellipsis of subject and object noun phrases.23 Quirk et al. (1985) provide numerous examples of situational ellipsis (ellipted items in bold) reproduced below in (65).24

158

Jim Miller

(65) a. b. c.

[I] told you so. [do you] want a drink? [it/the restaurant/the train] seems full.

In fact, the ellipsis of constituents carrying given information can extend over various turns in a conversation. (66a) is an excerpt from the corpus of English conversations and (66b) is part of the excerpt with the ellipted constituents in bold. The boy mentioned in M59 is one Richard, who has been under discussion. (66) a.

b.

M59 L59 A70 M60 L60 M61 L61 M59 L59 A70 M60 L60

what’s he going to do, that boy, anyway? play golf be a professional golfer is he? he would if he could I think he’s applied for a scholarship in this country? yeah what’s he going to do, that boy, anyway? [that boy is going to] play golf [that boy is going to] be a professional golfer is he [going to be a professional golfer] he would [be a professional golfer] if he could [be a professional golfer] I think he’s applied for a scholarship M61 [has he applied for a scholarship] in this country? L61 yeah

Categorizing the above ellipses is not straightforward. L59 and A70 certainly contain words that occur in the text, but the words are in a different order and in different constructions. M59 contains a WH-interrogative structure what’s he going to do with is and going separated by the subject NP, and the NP that boy is outside that clause. In L59 that boy and is and going to are gathered together in that order inside one clause. This is either a peripheral case of standard ellipsis or a case of quasi-ellipsis. Finding a solution to the problem is not a priority for present purposes, the main point being that situational ellipsis and large scale elision by means of standard or quasiellipsis are typical of spoken English, whereas elision in written English is typical of coordinate structures, nonfinite and verbless clauses, and comparative clauses (cf. Quirk et al. 1985: 910–913).

8.2. Ellipsis in Russian 8.2.1. Null subject NPs Like spoken and written English, spoken and written Russian differ with respect to types of ellipsis. Written Russian possesses various types of structural ellipsis, in-

Focus in the languages of Europe

159

cluding coordinate structures, nonfinite clauses, and one type of comparative clause. The details of these elliptical constructions are not relevant here; instead, we will concentrate on the occurrence of null subject and object NPs. Written and spoken Russian both allow null subject and direct object NPs on a much wider scale than in English, and spoken Russian offers far more instances of zero direct objects than written Russian. Consider the passage of English in (67). (67)

David came out of his room and shut the door. He walked quietly down the stairs and unlocked the front door.

There are subject NPs missing before shut the door and unlocked the front door but there is nothing unusual, since these are examples of structural ellipsis. Written Russian allows zero subject and object NPs outside coordinate structures; indeed such NPs can reach back across boundaries between subordinate and main clause, between sentences, and even across paragraph boundaries. (68) illustrates reaching back across sentence boundaries, as well as having coordinate clauses. The pairs of square brackets in bold indicate slots for subject NPs. (68)

A ona molˇcit. but she is-silent [: : :] Idet, ladoški v rukava svitera [: : :] sprjatala, i [: : :] walks hands into sleeves of-jumper [: : :] has-hidden and [: : :] molˇcit. [: : :] is-silent ‘She walks along; she has hidden her hands in the sleeves of her jumper and stays silent.’

The first sentence has a pronoun subject, ona. (The character has been previously introduced and named and is one of the main protagonists in this part of the story.) The second sentence has a null subject NP before idet. This verb form is third person singular and does not carry information about the sex of the agent referent, which has to be recovered from the previous sentence. In the above example the null NPs refer to what Prince (1981) calls textually evoked entities. In spontaneous spoken Russian, as in spoken English, instances of null NPs referring to situationally evoked entities (instances of situational ellipsis, in Quirk et al.’s terms) are frequent. An example is given in (69), which is an excerpt from the spoken texts recorded and transcribed in Kapanadze and Zemskaja (1979: 154). (69)

A:

B:

Zin/ èto ja Zin this I ‘It’s me.’ razmundirivajsja take-your-coat-off

160

Jim Miller

A: A:

B:

A:

[S] razmundirivajus’25 (I) am-taking-it-off ................................ Zin/ [S] vykljuˇcila [DO]/ ja prjam ne znaju Zin (I) have-switched-off (it) I really not know ‘Zin I’ve switched it off but I really don’t know.’ [S] Brenˇcit/ da? [it] is-’strumming’ yes ‘It’s making a strumming noise, isn’t it?’ Užašno // Ja tak ne mogu Terribly // I thus not am-able ‘A terrible noise, I can’t stand it.’

The null subject NP in (69) requiring comment is [S] brenˇcit. The radio is not referred to linguistically, although it is the referent of the null direct object NP in vykljuˇcila [DO]. Although not specified in the transcript, it is highly likely that speaker A drew attention to the radio by a hand gesture, by a nod, or by the direction of gaze. It is important to remember that in situated conversation, the central linguistic verbal system is supported by various paralinguistic systems which on their own are capable of achieving successful reference, and we suggest that one or more of the paralinguistic systems permits the situational ellipsis of the subject of brenˇcit and the direct object of vykljuˇcila.

8.3. Null object NPs Russian differs from English in the frequency of null direct object NPs. In fact, although zero direct object NPs do occur in English, they are limited to specific constructions. One is associated with sentences describing habitual actions, and especially actions that in a given culture are institutionalized. Cf. (70). (70)

Kate buys for Marks and Spencer.

Buy is listed in dictionaries of English as a transitive verb; (70) is highly marked and only possible in sentences describing someone’s occupation. It is not possible to have a zero direct object even in coordinate structures: *Mary didn’t want to pay so much for the painting so Kate bought. Another construction is associated with sentences describing actions that, within a given culture, apply to one particular class of entities. The cultural context enables the direct object to be omitted, as in (71); indeed, hunt, etc., can occur without a direct object even in sentences describing a single event. She’s hunting can only be used to refer to an event of hunting game, as opposed to hunting for her handbag, glasses, or car keys.

Focus in the languages of Europe

(71) a. b.

161

I’m reading, and I don’t want to be disturbed. They hunt in the autumn.

Eat, drink, cook are similar to hunt and read in the above respect. Cultural context also plays a role in the use and interpretation of these English verbs; He drinks is used not to state the obvious but to specify whether someone takes too many alcoholic drinks. (Hanks [1996: 106–115] has an excellent account of cultural context and the interpretation of the various Mayan verbs for eating and drinking.) Russian allows zero direct objects both in the above contexts and in others. We begin with examples in which zero direct objects are recoverable in the text. The examples in (72) show zero direct object NPs reaching back across clause and sentence boundaries.26 (72)

A on tol’ko smotrel na nee, but he only looked at her no tancevat’ ne [S] zval [O]. but to-dance not (he) invited (her) ‘but did not ask her to dance.’

(73)

Nakonec my konˇcaem, šef blagodarit [OO]. at-last we finish boss thanks (me)

(74)

Major prinjalsja osmatrivat’ vse karmany, major began examining all pockets, gnezda i zakoulki housings and crannies – [S] Našel [O]! – (I) have-found (it) I major podal v ljuk planšetku. and major put through hatch map-case ‘And the major passed a map-case through the hatch.’

(73) and (74) are complex examples in that the ellipsis can be seen as textual or as situational. For the reader of the stories the ellipsis is textual in that the entities referred to by the zero direct object NPs have been mentioned in the text. For the participants in the narrated events the ellipsis is situational. The excerpt in (74) follows a passage describing how the major had a hunch that the documents they were seeking had been hidden by their author in the tank the latter had commanded during the war. The map-case has not been mentioned before; the missing document has, but so far back in the text that the rules of coherent and cohesive text would require a full NP, not a zero NP.

162

Jim Miller

8.3.1. Null NPs in spoken and written Russian The incidence of null NPs in spontaneous spoken Russian and written Russian can be summarized as follows: null NPs are far more frequent in spoken than in written Russian. This applies to null subject NPs but significantly to null object NPs. Null direct object NPs are equally frequent in spoken and written narrative but the frequency increases significantly (by a factor of 3) in dialogue from plays. The greatest number of null subject and object NPs are to be found in transcripts of real-life dialogues (what Kapanadze and Zemskaja [1979] call “micro-dialogues”). Whereas in spoken and written narrative, and in dialogue from plays, the null direct object NPs mainly refer to textually evoked entities, in the real-life dialogues they refer to situationally evoked entities, as in (69).

8.4. Salience-reducing constructions In at least two languages in the sample, constituents are defocused by being extraposed rightwards out of the core of the clause. Vallduví (1994) gives the following examples from Catalan. (75) a.

b.

El Luc [s’estima el GOS]. the Luke loves the DOG ‘Luke loves his dog.’ El Lluc [se l’ESTIMA], el gos. the Luke self LOVES the dog ‘Luke LOVES his dog.’

In the unmarked example (75a) the nuclear accent falls on the rightmost constituent. In (75b) the focal constituent is the verb but the focal accent cannot move round the clause. The syntactic construction in (75b) enables the verb to be in the rightmost position in the core clause and to receive the focal accent. The verb is new; the participants Luke and the dog are given, and what is presented as new is the relationship between them. The rightmost position in the core clause is made available for the verb by the given noun phrase el gos being moved rightwards out of the core clause into a position where it is clearly signaled as background, given material. (76), also from Vallduví, shows an example in context. Treballs occurs in the first speaker’s contribution and is repeated, in the NP els treballs, in the second speaker’s contribution, in which it is extraposed from the core clause in order to allow macro to occur in rightmost position carrying the focal accent. (76)

A:

Tinc I-have aquest this

un ordinador que va molt bé té, mira a computer that works really well here look-at treball. paper

Focus in the languages of Europe

B:

163

Em pensava que els feies al macro, els I thought that them you-did on-the mainframe the treballs. papers

Erguvanli (1984: 43–71) discusses Turkish clauses with nonfinal V, in contradistinction to the canonical SOV order. Various types of constituents occur post-V: full and pronominal NPs, subordinate clauses, and adverbs. Excluded are indefinite NPs (except in imperative clauses), nonreferential direct object NPs, unstressed constituents, and WH-words. The post-V constituents are either treated by the speaker as discourse-predictable or are supplementary. (77) gives an example with the neutral word order, and (78) gives the same constituents but with the direct object in clause-final position. (Examples from Erguvanli [1984: 53].) (77)

Ben Nese Ablayı çok iyi buldum. I Nese Abla very well found ‘I found Nese Abla very well.’

(78)

A1: Nese Ablayı görüstuk. Nese Abla we-visited B: Nasıl buldun Nese Ablayı? ihtiyarlamı¸s mı? how you-found Nese Abla? she-aged-Q ‘How did you find Nese Abla? Has she aged?’ A2: Yoo, ben çok iyi buldum Nese Ablayı. no I very well I-found Nese Abla ‘No, I found Nese Abla very well.’

In (78), in A1 the direct object Nese Ablayı is new and precedes the verb. In B and A2 it is given and follows the verb. In Vallduví’s terms, it is no longer focal and is removed from the core of the clause to make way for the focal or new material, çok iyi. It could be argued that English too allows the rightward movement of NPs with reduced salience. L61 in (66a) is followed by M62: I mean there’s millions of them in America, sports scholarships. The problem is that an alternative analysis takes sports scholarships as an afterthought, and there are no criteria that permit a clear choice of analysis. The Catalan construction in (76) is different, because Vallduví (1994) is able to demonstrate that the Catalan construction cannot be analyzed as reflecting an afterthought. The key fact is that in certain contexts, the construction with the right-detached constituent is obligatory. In contrast, the English construction is never obligatory, and the main argument against the afterthought analysis is simply that many instances of right detachment have none of the expected properties of afterthought phrases, such as being separated from the core clause by a long pause, being accompanied by hesitations, or by phrases such as that is and I mean.

164

Jim Miller

Spoken Russian has a structure with right dislocation as exemplified in (79) from Zemskaja (1973: 245). (79) a. b.

On gde ležit it where is-lying Oni paˇckajutsja / they stain /

/ saxar? / sugar cˇ ernila / užasno èti / terribly these inks

It is not clear from Zemskaja’s discussion, which concentrates on the syntactic properties of the construction, what the exact function of the structure is, but it is a plausible hypothesis that it is a backgrounding or defocusing device. The speaker defocuses the noun phrases saxar and èti cˇ ernila, most probably because they refer to entities that are given; at the same time the speaker highlights gde ležit and paˇckajutsja, which convey the important part of the question encoded in (79a) and of the assertion encoded in (79b). Bossong (1981) proposes a similar analysis for modern spoken French. He points out that V(O)S order is practically the rule in modern spoken French, as in Ils sont fous ces Romains ‘They’re mad these Romans’. (A native speaker of French known to the author from the age of two days had this particular construction before the age of three. Speaking of her little brother, who was lying crying on the floor, she said il n’aime pas être par terre Ranald.) Bossong points out that the construction with clitic subject and object pronouns preceding the verb allows a full noun phrase to precede or follow the verb plus clitics, as in (80) and (81). (80) a.

b.

(81) a. b.

Frédéric il ne fume pas. Frédéric he not smokes not ‘Frédéric doesn’t smoke.’ Tout le Marais on l’ a battu. all the Marais one it has searched ‘We searched the whole Marais.’ Il savait tout ça mon papa. he knew all that my Dad Je ne leur avais rien fait aux Allemands. I not to-them had anything done to-the Germans ‘I hadn’t done anything to the Germans.’

The noun phrase in pre- or postverbal position can be subject, as in (80a) and (81b), or an object of some kind, direct in (80b) and oblique in (81b). Bossong proposes that the noun phrases in preverbal position signal a contrast; Frédéric and not someone else, the Marais and not some other quartier of Paris. The noun phrases in postverbal position are defocused (our term) and throw an “emphase émotive” on the verb phrase, savait tout ça and je ne leur avais rien fait. (Bossong treats the clitic pronouns as part of the verb.) Bossong is careful to point out that discourse effects can

Focus in the languages of Europe

165

vary and that constructions can lose their force through historical change, but his analysis is appropriate.) An analogous construction exists in some spoken varieties of English and the analyst is well advised to imitate Bossong’s cautious approach in dealing with them. Examples are in (82). (82) a. b.

Right outside the front gate it was, a huge tanker. There he was sitting in the bar, the guy I told you about.

(82a) certainly highlights the verb phrase carrying the information about where the tanker was, but the indefinite noun phrase in postverbal position highlights the tanker, which is being mentioned for the first time. The noun phrase in (82a) is likely to have its own intonation envelope and to carry emphatic stress. On the other hand, the postverbal noun phrase in (82b) is given and could carry low pitch or emphatic pitch. Much work remains to be done on the use of these constructions in particular languages and on cross-language comparison. This point is interestingly and worryingly demonstrated by example (83) from a piece of written Scottish Gaelic which appeared in the newspaper The Scotsman. It is interesting to see this instance of a clause-NP construction, but worrying because it appears to highlight the predication expressed by the clause and the entities of whom the predication is made and who are referred to by means of the NP. Another factor is the weight of the NP. (83)

Scottish Gaelic Nach iad a tha ladarna, NEG-INTERR-COP they REL BE impudent luchd an TBh [a] tha cho dall ’s gu bheil : : : people Det TV REL BE so blind that ‘Aren’t they impudent, the TV people who are so blind that : : :’

9. Highlighting devices (nonpresentational) This section and the next are based on the map task data, collected either from the corpora of spontaneous map task dialogues or from the questionnaires based (partly) on the map task. In this section, we discuss the general properties of the data, and in the next section we examine the spontaneous dialogues.

9.1. Discourse particles All the languages have discourse particles but not all the languages make equal use of them, at least from the evidence of the spontaneous map task dialogues. The English dialogues have no highlighting particles apart from like, whose status as a highlighter

166

Jim Miller

is controversial (but see Miller and Weinert 1995). The particles OK and right are very frequent and could be analyzed as having a highlighting function, but much more obvious is their role in managing the exchange of information between the participants in the map task. See and know, discussed in Sections 5.2.3 and 6.1.2 in connection with first mentions, have to be analyzed as verbs constituting the core of an interrogative clause. Hungarian, German, and Russian make heavy use of particles, and Catalan comes between the two groups. As forewarned in the introduction, for most of the languages in the sample these comments apply to the data obtained from the questionnaire and cannot be elevated into general statements about the languages in general. It is likely that the languages for which particles were not mentioned by the informants, such as Swedish and Italian, do in fact make use of particles – indeed, Fernandez (1994: 78–84) gives examples of Swedish particles. The Catalan data are instructive in this respect, because the informant who supplied data via the oral questionnaire produced cleft constructions, whereas in the spontaneous Catalan map task dialogues, we find only one cleft but many particles. Where informants supply particles but no clefts, as for German, Hungarian, Bulgarian, Finnish, and Turkish, we can be confident that we have a correct, though partial, view of the language. This confidence is based on the invisibility of particles when special syntactic highlighting constructions are available and is confirmed for the above-mentioned languages by grammars and by specialists in the languages.

9.2. Word order The English dialogues make no use of different word orders within one and the same construction, but German and Hungarian make heavy use of movement into clauseinitial position. Catalan, again on the evidence of the dialogues, falls in the middle. We might expect that a language such as Russian, rich in inflectional morphology and linkages of agreement and government, would use word order freely. There are no Russian map task dialogues, but on the basis of the answers to the oral questionnaire and of Kapanadze and Zemskaja’s corpus of spoken Russian, it is clear that Russian applies leftwards movement to given constituents, and that leftward-moved constituents are typically accompanied by a particle.

9.3. Contrastive focus position The concept of focus position is invoked by analysts of languages such as Turkish, Hungarian, and Finnish. (The literature talks only of focus position but it is specifically contrastive focus, as the discussion of Hungarian makes clear.) Turkish offers the clearest example of a focus position (see Erguvanli 1984: 34–37). First we must

Focus in the languages of Europe

167

distinguish between unmarked and marked order. The order S–DO–IO–V, as in (84) is unmarked. (84)

Murat parayı bir genç adama verdi. Murat the-money a young man-DAT gave ‘Murat gave the money to a young man.’

Erguvanli describes the preverbal position in unmarked word order as neutral focus; items in this position convey new information and are frequently indefinite and inanimate. An example with an indefinite subject NP in neutral focus is given in (85), which also has unmarked word order. (85)

Tası o˘glana bir adam attı. stone-DEF boy-DAT a man threw ‘A man threw the stone at the boy.’

(86) is an example of marked word order. The key fact is that the noun phrase immediately preceding the verb is definite. (86)

Parayı bir genç adama Murat verdi. money-DEF a young man-DAT Murat gave

(86) is appropriately translated into English as It was Murat who gave the money to a young man. That is, when the word order is marked, the constituent immediately preceding the verb carries contrastive focus.27 Highlighting in Finnish is achieved by means of intonation and particles, but if word order is used the highlighted constituent typically appears in clause-initial position. The highlighted constituent carries contrastive focus. Consider (87), which has neutral SVO order and the orders in (88) and (89). (87)

Marja tapasi Jussin Lontoossa. Marja-NOM met Jussi-ACC London-INESS ‘Marja met Jussi in London.’

(88)

JUSSIN Marja Lontoossa tapasi. Jussi-ACC Marja-NOM London-INESS met ‘It was Jussi that Marja met in London.’

(89)

LONTOOSSA Marja Jussin tapasi. London-INESS Marja-NOM Jussi-ACC met ‘It was in London that Marja met Jussi.’

Note that in the contrastive focus word order, the verb occurs at the end of the clause. Contrastive focus, as Chafe (1976) pointed out, has to do with the speaker or writer choosing one of a set of entities and excluding the rest. Thus, the speakers can use (utterances realizing) (88) to assert that out of the set of men that Marja might have met in London, she met Jussi and nobody else. Speakers can use (89) to assert that

168

Jim Miller

out of the set of places in which Marja could have met Jussi, she in fact met him in London and in no other place. Kiss (1987) analyzes Hungarian as having a preverbal focus slot, like Turkish, and as having free word order. Her 1987 analysis does not take account of the earlier paper by Kálmán et al. (1984) which distinguishes between neutral sentences with level prosody, as in (90) and what they call corrective sentences, as in (91). (Their corrective sentences correspond to Chafe’s contrastive focus construction.) (90)

Elment a moziba. he-went the cinema-to ‘He went to the cinema.’

(91)

elment a moziba. ‘He did go to the cinema.’

(92)

A moziba ment el. ‘(Actually) it’s the cinema he went to (as opposed to the theater, pub, etc.).’

(91) is an emphatic sentence. The verb carries the nuclear stress and the construction is used to contradict a statement. (92) is a contrastive sentence; in Chafe’s terms, the speaker reviews, as it were, the set of places that the person might have gone to and asserts that he in fact went to the cinema and not to any of the other possible destinations. The constituent being highlighted is a noun phrase, as opposed to the verb in (91). Both (91) and (92) are examples of contrastive focus constructions/corrective sentences,28 and it is they that have flexible word order and display complex interactions of topicalization, focusing, and scope relations. Kiss (1998: 707–708) offers an account of focus that continues the approach of Chafe and Kálmán. She distinguishes between an information focus and an identifying operator. Information focus has to do with the part of a sentence carrying new information; the identifying operator, as Kiss puts it, introduces a set and identifies a subset of it as being the exclusive domain of a particular predicate. She provides an example which illustrates the two types of focus in Hungarian but also clears up an apparent problem with (92): since (92) can be a reply to an ordinary WH-question, can it be an instance of a contrastive focus/corrective sentence? Kiss says that the question in (93) can be answered with (94) or (95). (93)

Kivel találkoztál a hangversenyen who you-met the concert-at ‘Who did you meet at the concert?’

(94)

Erzsivel találkoztam. Elizabeth I-met ‘I met Elizabeth [It was Elizabeth I met].’

Focus in the languages of Europe

(95)

169

Találkoztam Erzsivel. I-met Elizabeth ‘I met Elizabeth [among others].’

(94) has the identifying operator and is, in Kiss’s words, an exhaustive answer; the speaker states that he/she met Elizabeth and nobody else. (95) has information focus, which yields a nonexhaustive answer; the speaker met Elizabeth and possibly other people. Kiss works with an analysis in which the identifying or focus operator is preposed into immediately preverbal position; what is important for this discussion is that a clause with contrastive focus can function as a reply to an ordinary WH-question, and that (94) is parallel to (92). To sum up, focus position in Hungarian is associated with contrastive focus (contrastive highlighting in the terms of this chapter) or correction, to use Kálmán et al.’s terms. Focal stress is not confined to the preverbal slot. (96) is an example of a contrastive sentence with multiple contrast, with focal stress on both the preverbal and the postverbal constituents. (96)

Kati írt Péternek. Kati she-wrote Peter-to ‘As for Kati, it was Peter she wrote to.’

Focal stress can also fall on locational and directional verb prefixes, which Kálmán et al. (1984) analyze as being not in focus position but what they call “hocus position.” This phenomenon is exemplified in (97). (97)

A pék nem elfutott, hanem befutott. the baker not away-ran but in-ran ‘The baker didn’t run away, he ran in.’

Finally, focal stress may fall on phrases in sentence-initial position, as in (98). In view of the comments throughout this chapter about word order and particles, the fact that Kati is followed by the particle is is significant. Another topic for research is the cooccurrence of focal stress and particles (and word order). (98)

Kati is mindig Péternek írt. Kati too always Peter-to wrote

The complex situation in Hungarian can be summed up thus. The focus position invoked by Kiss and others is relevant only to the specific corrective construction, which is also the locus of flexible word order. In other constructions focal stress can fall on constituents in various clause positions. Given that English also has the IT-cleft syntactic structure, one of whose functions is to correct previous statements, the distance between English and Hungarian does not seem quite so great, though we will see in the discussion of the Hungarian map task dialogues that the use of particles separates the two languages quite clearly. On

170

Jim Miller

the assumption that further investigation will not reveal similar complexities in Turkish, we can accept for the moment that Turkish is the only language in the sample with a focus position, and interestingly, that the constituents in that position are interpreted as new or as contrastive depending on whether the word order is unmarked or marked.

9.4. Syntax and intonation Perhaps surprisingly, the functioning of intonation is not equally flexible in all languages. Vallduví (1994) contrasts English and Catalan. English sentences have a malleable prosodic contour which allows the nuclear accent to fall on any of the words in the sentence Luke loves his dog, depending on which word carries new information and is focal. In Catalan the nuclear accent is fixed over the right-hand core clause boundary and cannot be shifted to the left. This pattern causes no problems when the last lexical item is new information – the focus for Vallduví – but is awkward when the focus falls on one of the other words. Untypical focus has to be signaled in Catalan by means of detachment. The core clause contains focal information and the ground/old information is detached from the core clause and occurs to the left or right of it. The relevant Catalan examples are given in (99).29 The focal words are in capitals and the core of the clause is inside square brackets. (99) a.

b.

c.

El Lluc [s’ estima el the Luke self loves the ‘Luke loves his dog.’ El Lluc [se l’ estima], the Luke self it loves ‘Luke loves his dog.’ [Se l’ estima el lluc], self it loves the Luke, ‘luke loves his dog.’

gos]. dog el gos. the dog el gos. the dog

Irish faces the same restriction and also finds a syntactic way round it. Consider the neutral (or thetic or all-new) sentence in (100).30 (100)

Chuaigh Seán go Corcaigh inné. go-PST Sean to Cork yesterday ‘Sean went to Cork yesterday.’

In order to focus on a particular constituent, a two-clause structure is used analogous to the English IT-cleft. The first clause is introduced by is é (is-it), followed by the constituent to be focused, and the second clause is introduced by the relativizer a and contains the remainder of the constituents from the neutral clause. The examples in (101) illustrate the focusing of different constituents in (100).

Focus in the languages of Europe

(101) a.

b.

c.

171

Is é sean a chuaigh go Corcaigh inné. is it sean REL go-PST to Cork yesterday ‘It is sean who went to Cork yesterday.’ Is go corcaigh a chuaigh Seán inné. is to cork REL go-PST Sean yesterday ‘It was to Cork that Sean went yesterday.’ Is inné a chuaigh Seán go Corcaigh. is yesterday REL go-PST Sean to Cork ‘It was yesterday that Sean went to Cork.’31

The cleft construction cannot focus on minor constituents such as possessive pronouns. Instead an emphatic particle, san,32 is used, attached to the head noun, as in (102b). (102) a.

b.

Rug a-mhadra ar an bhfrancach. caught his-dog on the rat ‘His dog caught the rat.’ Rug a-mhadra-san ar an bhfrancach. caught HIS-dog-particle on the rat ‘It was his dog that caught the rat.’

9.5. Clefts The languages of the sample divide into those languages that have no cleft constructions and those that do. The latter subset divide into languages that have the full range of clefts, IT, WH, and reverse, those that do not have the full range, and those that have only a rudimentary cleft construction. 9.5.1. Definition of cleft constructions First let us define our terms, for convenience taking English as our exemplar. (103) illustrates the IT-cleft, (104) the WH-cleft (also termed the pseudo-cleft), and (105) the reverse WH-cleft (henceforth RWH). (103)

It’s that side you should be on.

(104)

What you want to do is curve round that wood.

(105)

That’s what I’ve done.

(IT-cleft) (WH-cleft) (reverse WH-cleft)

Following Hedberg (1990), the subject or complement of the copula will be termed the “cleft constituent” (that side, what you want to do, and that in (103)–(105), respectively). The label “cleft clause” is given to the reduced clauses in the WH-cleft (curve round that wood), the annex clause (to use the terminology of Quirk et al.

172

Jim Miller

1985) in IT-clefts (you should be on), and the WH-clause in the RWH-cleft (what I’ve done). The key properties of the cleft constructions are that they consist of two clauses (assuming that the sequence curve round the wood in (104) can be treated as a reduced clause) and contain either the personal pronoun it or a WH-pronoun. Since the term “WH-pronoun” is English-based, an explanation is in order if it is to be applied to other languages. We take the view that WH-words are indefinite deictics used for pointing at entities not in the immediate situation of utterance because not yet mentioned. Speakers either ask for a specification of an entity, as in Who is going to drive the car? or provide the specification, as in I tell you what – you drive the car and I’ll take the train or I remember what he bought for our first anniversary – a beautiful Chinese watercolor. In the first example, you drive the car specifies what and in the second example, a beautiful Chinese watercolor. In WH-clefts such as What you’re doing is you’re going down the side of the allotments the indefinite deictic what points forward to a specification about to be supplied by the instruction giver. That is, the indefinite deictics in the preceding examples are deictically similar to, for example, the WH-clefts used in the map task dialogues to signal that a section of discussion has reached its limit and that the instruction giver is going to specify the way forward. Reverse WH-clefts, as in (105), contain a demonstrative pronoun which signals proximity or its lack. The orientation point with respect to which proximity is established is typically the speaker. In the English reverse clefts this and that have a powerful pointing function: they point back at a previously mentioned entity and the WH-clause highlights some property of that entity. Reverse WH-clefts sum up and highlight but do not point forward. IT-clefts contain the pronoun IT, which is definite but does not involve proximity, unlike this and that, and is typically unstressed. It lacks the forward-pointing power of WH-clefts and the emphasizing and backward-pulling power of the RWHclefts with that or this. IT-clefts are perhaps best analyzed as unmarked with regard to the direction of the pointing, functioning as neutral attention markers with no referential function of their own. For cross-language investigations, an IT-cleft is one introduced by a deictic (whether demonstrative pronoun, personal pronoun, or an indefinite deictic) followed by a copula (whatever counts as a copula in a given language), followed by a predicate NP modified by a clause.33 The clause – Hedberg’s cleft clause – may be similar to a relative clause but not identical with it; Quirk et al. (1985: 1386–1387, 1407) label the English cleft clause an annex clause because of differences between it and relative clauses, but their evidence applies only to written English and indeed only to a strict canon of written English.34 A WH-cleft is one introduced by an indefinite deictic,35 followed by copula, followed by an NP or a clause. The clause may be full (what I’ve done is I’ve switched off the current) or reduced (what I’ve done

Focus in the languages of Europe

173

is switch off the current). The full clauses are typical of spontaneous spoken language. A reverse WH- cleft is one introduced by a demonstrative pronoun – that in (105), followed by a copula – ‘s in (105), followed by a free relative clause or the equivalent thereof – what I’ve done in (105). To count as a free relative, a given clause must at least contain an indefinite deictic and preferably also have the word order of relative clauses in the language concerned. The demonstrative pronoun in the English RWHcleft is typically that – that is, in all the RWH-clefts in the map task dialogues and in the corpus of spontaneous conversation. that is a powerful deictic in that it signals (relative) remoteness from the speaker and possibly also the hearer. In the map task dialogues, the remoteness is to be interpreted in terms of pointing back at a chunk of discourse and summing it up. In principle the demonstrative pronoun could be this, but the latter signals proximity to speaker and hearer and proximity in the map task dialogues would have to be interpreted in terms of a chunk of dialogue that is, so to speak, approaching the speaker and hearer. We would expect to find, for example, this is where you should be being used to point forward to information about to be provided; one such example is in (106). (106)

A1: I’m at the black house now at the left of the allotment B1: the wee black house, that’s brilliant, so have you got sort of a ushape a short of half u-shape? A2: yeah B2: right, that’s OK, that’s fine, right, this is where it gets difficult A3: ha-ha (laughs) B3: eh A4: right B4: cause you’re aiming into the middle of nowhere

In B2 the reverse WH-cleft with this does not point back. The location referred to is close to speaker and hearer because it is the starting point for the next, tricky section of route and because the cleft serves as a bridge from the previous set of instructions, relating to the black house, to the next set of instructions.36 We can usefully end this section on the English clefts by commenting on their intonation. Quirk et al. (1985: 1387) describe the characteristic intonation of IT-clefts, giving the example It was the DOG I gave the WATER to, in which the cleft constituent DOG carries a falling tone and WATER, a rising tone. This intonation pattern, it must be emphasized, is no more than characteristic or typical, and is certainly not necessary. Delin (1990), working with clefts from spontaneous speech, discovered that the nuclear accents fell on various constituents. One of her examples is It was John who WON the race, which is appropriate in a context where the speaker knows that John took part in a given race, but did not know that he had actually won it.

174

Jim Miller

9.5.2. Languages without clefts Finnish, Hungarian, and Turkish37 have no clefts; typically, pitch, left-movement, juxtaposition, or particles are used for highlighting. (107) a.

b.

c.

d.

e.

f.

Turkish ˘ Uçurumların d˙iger taraf-ın-dan git-meliy-di-m. cliff-PL-GEN other side-3SG.POSS-ABL go-ought-PST-1SG ‘It’s the other side of the cliff I’ve to go.’ I spoke to Fiona about grants. ˘ il ki. Bunlarla ilgilenen o deg˙ this-place-PL-COM deal-part she not particle ‘It’s not her that deals with them.’ ˘ il, kirsty (dir). Yardım eden Fiona deg˙ help do-PART Fiona not Kirsty (is) ‘It isn’t Fiona who’s helping but Kirsty.’ (i) Türkçe sözdizimi hakkında konu¸s-aca˘g-ım Turkish syntax about peak-FUT-1SG (ii) Yap-aca˘g-ım konu¸sma Türkçe šözdizimi make-FUT-1SG-PART lecture Turkish syntax hakkında. [preferred by the informants] about ‘What I’m going to talk about is Turkish syntax.’ Köprü-den geç-ip 2 cm çık-malı-sın. bridge-ABL pass-Prosuffix 2 cm climb-OBLG-2SG ‘What you’re going to do is go over the bridge and up 2cm.’ I¸ste, ora-dan geç-meli-sin. look, it-ABL pass-OBLG-2SG ‘That’s where you should be.’

˘ In (107a) the highlighting is done by pitch, with a nuclear accent on DI˙GER ‘other’. ˘ ˙ In (107b) the nuclear accent is on DEGIL, while in (107c) the highlighting involves ˙ and KIRsty and the juxtaposition of Fiona DEG˘ IL ˙ and a nuclear accent on DEG˘ IL kirsty. The particle ki in (107b) is used by speakers to reject presuppositions made by a previous speaker; here the first speaker announces that she has talked to Fiona, on the assumption that Fiona is the one who deals with grants. This assumption is incorrect, and the ki signals this. The Turkish informant gave no recognition to the WH-cleft in (107e), which in the English map task dialogues has the specific function of drawing a line under a discussion and at the same time pointing forward. With respect to cleft structures, the most interesting Turkish examples are (107b) and (107c). Turkish has interrogative pronouns – kim ‘who’, kaç ‘how many’, ne ‘what’, and hangi ‘which’ – but they do not form a phonologically cohesive set like

Focus in the languages of Europe

175

the TH- and WH-pronoun sets in English or the K-pronouns in Russian, and they do not have the range of functions attaching to the English WH-pronouns as described in Section 9.5.1. In particular, they do not occur in a Turkish equivalent of the English free relative clause construction – for example, what’s burning in what’s burning is the potatoes. (Note that WH-clefts with who are awkward: who’s telephoning is Fiona is distinctly peculiar, but if the construction is turned round it sounds better – Fiona is who’s telephoning. The unexceptionable construction, however, involves the one who; either the one who’s telephoning is Fiona or Fiona is the one who’s telephoning. Turkish does have an equivalent of what’s burning and the one who’s telephoning. It consists of a single word, derived by adding the suffix -an/-en to verb stems. In (107b) ilgilenen consists of the verb stem ilgilen- plus the participial suffix -en, while in (107c) eden derives from the verb stem et- plus -en, the d arising from a phonological process affecting intervocalic voiceless stops. These participial forms function as noun modifiers, yielding one Turkish equivalent of relative clauses in English, French, and so on; bunlarla ilgilenen is translated into English as ‘the one who deals with these things’. They also function as nouns; thus, the verb stem yaz- plus -an yields yazan, which can be translated as ‘the person writing’ and can function as subject, direct object, and so on. Consider the syntactic structure of (107c). It is a copula construction, de˘gil being the negative copula with a zero third person singular suffix – compare de˘gilim ‘I am not’, de˘gilsin ‘you are not’. The subject of the copula is yardım eden ‘the one helping’ and the complement is Fiona. The English IT-cleft is also a copula construction. In It isn’t Fiona who’s helping the (dummy) subject of is is it and the complement is Fiona. As mentioned above, the exact nature of the clause introduced here by who is unclear. While the Turkish and English constructions are both copular, they differ considerably in other respects. (107c) contains only one clause but the English IT-cleft contains two clauses. The Turkish construction contains a participial phrase, yard ım eden, whereas the English IT-cleft contains the deictic personal pronoun it. Furthermore, the Turkish construction, depending on the context, can also be equivalent to the English WH-cleft. It was pointed out above that English WH-clefts are not normally introduced by who but that an alternative construction is used, introduced by the one who. However, a closer parallel can be found with English WH-clefts introduced by what. As is shown by (107e), the Turkish informant chose to render the English cleft by a neutral Turkish main clause plus a clause containing a verb with a dummy suffix. An alternative Turkish rendering with an -an participle is shown in (107g) below. (107) g.

Senin yapaca˘gan köprüyü getmek(tir). you-GEN do-FUT-PART-2SG bridge-ACC cross-INF-(COP) ‘The thing/what you are going to do is cross the bridge.’

176

Jim Miller

In (107g) the Turkish participial phrase has specific reference for the speaker, whose task is to supply a specification of the referent for the hearer. In the English what you’re going to do is cross the bridge, what has the same function; as suggested above, it can be described as an indefinite deictic pointing forward to an upcoming specification. The upshot is that English WH-clefts introduced by what present more parallels with the Turkish participial construction than does the English ITcleft. Both are copular and both contain as subject a phrase with specific reference for the speaker/writer. (If the Turkish participial phrase were analyzed as a nonfinite clause, then both constructions would contain two clauses, and of course this parallel would carry over to the English IT-cleft.) Turkish has two sets of participles that correspond to relative clauses in the IndoEuropean languages of Europe. The -an participles that we have seen above are used (simplifying but not falsifying) where, in English say, the relative pronoun is the subject of the relative clause. Where the relative pronoun is an object, the Turkish participial construction has participle forms with the suffix -dik, as in (107h). (107) h.

Celia’nin istedi˘gi o yeni Türkçe sölüktür Celia-GEN want-PART-3SG.POSS that new Turkish dictionary ‘What Celia wants is that new Turkish dictionary.’

The Turkish participial construction can be glossed as ‘of-Celia her-wanting’. To conclude this discussion of English IT- and WH-clefts and their Turkish equivalents, here is the title of a newspaper article. (107) i.

Musluklardan akan su mu? tap-PL-ABL flow-PART water INT ‘Is it water that flows from the taps?’

Although the idiomatic English translation is an IT-cleft, a structure with a free relative clause is equally appropriate – Is what flows from the taps water? or, more colloquially, What flows from the taps – is it water. Finally, we should note that the Turkish equivalent of English RWH-clefts involves the same participial structure, as shown in (107j) and in (107k). (107) j.

k.

Celia’nin istedi˘gi o. Celia-GEN want-3SG.POSS that ‘What Celia wants is that/That’s what Celia wants.’ Senin okuman gereken bu i¸ste.38 you-GEN read-INF-2SG necessary this particle ‘What you should be reading is that indeed/That’s what you should be reading.’

A detailed comparison of English clefts and their Turkish equivalents requires a book to itself. What has been demonstrated here is that Turkish possesses a single

Focus in the languages of Europe

177

construction which corresponds roughly in structure to the English WH- and RWHstructures, in that it contains a substructure that could be labelled a headless or free relative “clause”. Moreover, the examples (107h)–(107k) not only have this “free relative clause” but are copular, like the WH- and RWH-clefts. With respect to reference, these “free relatives” are similar to the WH-pronouns in the English clefts in having referents that are specific-for-the-writer/speaker. These properties outweigh the fact that Turkish lacks a unitary set of WH-pronouns with the range of uses enjoyed by the English WH-pronouns and lacks the range of word orders in the English clefts. However, the exact discourse functions of the Turkish construction remain to be investigated. The Finnish structures supplied by the informant employ the same sort of devices. In the ‘on the other side of the cliffs’ example Finnish either puts emphatic pitch on toisella puolella ‘other side’ or uses both the emphatic pitch and a suffix -kin as in (108). (108)

Minun pitää-kin kulkea kallioiden toisella I-GEN must-1SG-particle go cliff-PL-GEN other-ABL puolella. side-ABL ‘It’s the other side of the cliffs I’ve to go.’

Carlson (1993) analyzes -kin and its vowel harmony alternant -kään as signaling the presence of some expectations. The expectations may be confirmed or confounded, possibly causing surprise to speaker or addressee. Carlson (1993: 82) provides the examples in (109). (109) a.

b.

Arvasin niin käyvän, ja niin kävikin. I-guessed that would-happen and that happened-particle ‘I guessed that would happen and happen it did.’ Arvasin ettei niin käy, mutta niin kävikin. I-guessed would-not that happen but it happened ‘I guessed that would not happen but happen it did.’

Carlson’s approach can be applied to (108) as follows. (108) relates to a situation in which the participants in the map task are working out one part of the route. The person drawing the route has thought from the instructions that it lay to one side of the cliffs, a landmark on the map. After discussion with the information-giver providing the instructions as to how to draw the route, it emerges that the route actually lies on the other side of the cliffs, contrary to the information-follower’s expectations. The English IT-cleft ties in directly with contrast/identification – X not Y, and indirectly with modality in the sense of one participant’s expectations. The Finnish particle ties in directly with modality and indirectly with contrast/identification. In spite of the

178

Jim Miller

difference in contribution to the discourse, both cleft and particle add extra coding and catch the addressee’s attention. In the IT-clefts with overt contrast, the Finnish construction, like the Turkish, employs juxtaposition – Ei Fiona, vaan Kirsti ‘not-is Fiona, but Kirsty’. The force of the reverse cleft examples is captured in Finnish by putting the highlighted item in clause initial position, as in (110). (110)

siella˘ sinun pitäisi olla. where/there you-GEN should be ‘That’s where you should be.’

(110) can be translated into English by means of an RWH-cleft, but clause initial position serves a general highlighting function that cannot always be rendered by means of an English RWH-cleft. Sometimes a particle must be used, as in (111), and sometimes an IT-cleft is appropriate, as in (112). (111)

Tapasi Marja Jussin Lontoossa. met Mary-NOM Jussi-ACC London-LOC ‘Mary did indeed meet Jussi in London.’

(112)

Marja Jussin Lontoossa tapasi. Mary-NOM Jussi-ACC London-LOC met ‘It was Mary who met Jussi in London.’

Examples of other particles are given in (113)–(115). (113)

Mikko-pa sen auton myi. Mikko-NOM-particle it-ACC car-ACC sold ‘It was mikko who sold the car.’

(114)

Mikko-ko sen auton siellä Lontoossa myi? Mikko-NOM-particle it-ACC car-ACC there London-ILL sold ‘Was it Mikko who sold the car in London?’

(115)

On-han Mikko luistelemaan oppinut. is-particle Mikko-NOM to-skate learned ‘Mikko has indeed learned how to skate.’

Carlson (1993: 88) proposes that the discourse rule for -pa is “Mark a contrastive element with -pa.” If Carlson is correct, it looks as though the IT-cleft and -pa correspond directly, unlike the IT-cleft and -kin/-kään. Hakulinen (1976), cited by Carlson, explains -han as being used when speakers choose to present something as common knowledge, even when it is previously known. Treating one person’s moot point as something generally known and accepted is a way of establishing superiority in a discussion. (115) is appropriate when on speaker has stated tentatively that

Focus in the languages of Europe

179

Mikko has learned to skate and another speaker wishes to present Mikko’s having learned to skate as generally known and accepted. Finally, let us note that spoken Finnish has a construction that looks like an ITcleft. (116)

Minä se olin joka sen ikkunan rikkoi. I-NOM it-NOM was who-NOM it-ACC window-ACC broke ‘It was me who broke the window.’

We can conclude this brief look at Finnish particles with an important typological point made by Hakulinen and Karlsson (1979: 300). Their analysis of -kin agrees in its broad outlines with that of Carlson. In addition, Hakulinen and Karlsson remark that in Finnish, clitic particles are closely linked to prosodic phenomena, especially intonation and stress, and that in Finnish a significant part of the pragmatic functions of prosody has been grammaticized (their term) into clitic particles. Languages can have both an IT-cleft construction and a construction involving word order change, but the two will not necessarily be interchangeable. For instance, Catalan has the two constructions as exemplified in (117). (117) a.

b.

El cotxe encara no he pagat. the car yet not I-have paid ‘I haven’t paid the car yet.’ És el cotxe que encara ne he pagat. it-is the car that yet not I-have paid ‘It’s the car that I haven’t paid yet.’

(117a) is appropriate where the hearer does not necessarily know that something is not paid for, whereas (117b) is appropriate where speaker and hearer are agreed that x is not paid for and the speaker is specifying the value of x. 9.5.3. Languages with clefts The languages that have clefts display great diversity. The relevant properties are: 1) the type of cleft: IT-cleft, WH-cleft, or reverse WH-cleft; 2) the exact structure of each construction; 3) the type of constituent that can function as the cleft head; 4) the discourse function of each construction. Nothing much will be said here about the fourth point. English not only has all three types of clefts but its WH-cleft construction contains a genuine free relative; in what I saw was a black panther loping across the moors the clause what I saw does not modify any overt noun or pronoun – in fact what functions as a relativizer in certain nonstandard varieties of English but not in the standard language. The corresponding French construction is ce que j’ai vu in which que j’ai vu modifies ce. What the two constructions have in common is that they are

180

Jim Miller

both seriously incomplete without a further clause specifying the entity referred to by the speaker. Italian, Spanish, and German have free relative constructions syntactically parallel to the English one. Sornicola (1988: 345) gives the Italian and Spanish examples in (118).39 (118) a.

b.

Italian Chi va a Roma sono io. who goes to Rome am I ‘It’s me that is going to Rome.’ Spanish Quien va a Roma soy yo. who goes to Rome am I

Weinert (1995) gives examples of German WH-clefts, one of which is reproduced in (119). Because this construction occurs infrequently, although available in principle, she had to take examples from spontaneous conversation since the map task dialogues had none.40 (119)

Ja was ich bei Bernstein yes what I with Bernstein immer besonders schwierig finde always especially difficult find ist die Tatsache, is the fact daß er nicht klar trennt zwischen that he not clearly distinguishes between linguistischen, soziologischen linguistic sociological und psychologischen Variablen. and psychological variables

The above example is a classic WH-cleft, with the free relative was ich bei Bernstein immer besonders schwierig finde connected by the copula ist with the noun phrase die Tatsache : : : The German map task dialogues offer examples of unintegrated WH-cleft structures, parallel to the spoken English unintegrated WH-cleft constructions discussed in Chapter 1. Consider (120). (120)

Und was wir and what we vom Start from-the start

für for aus out

die Route the route geh’n wir go we

brauchen ist need is über’s Raumschiff. over-the spaceship

The free relative clause is was wir für die Route brauchen. It is followed by the copula ist but the latter does not connect the free relative clause to some noun phrase.

Focus in the languages of Europe

181

The clause following the copula is independent both syntactically and with respect to the discourse; it contains no constituent linking up with was. Two informants supplied German data via the questionnaire. One declared that German has no clefts, or at least that they are not used in spontaneous speech, while the other informant did provide IT-clefts, as in (121)–(123). (121) a.

b.

c.

I spoke to Fiona. Oh, Fiona hat aber nichts mit Stipendien zu tun. oh Fiona has but nothing with grants to do ‘It’s not Fiona/her that deals with grants.’ Sie ist es aber nicht, she is it but not die mit Stipendien zu tun hat. who with grants to do has ‘It’s not her that deals with grants.’ Fiona? Die hat aber nichts Fiona she has but nothing mit den Stipendien zu tun. with the grants to do ‘But Fiona doesn’t deal with grants.’

The closest German equivalent to the English IT-cleft is (121b), which contains the relative clause die mit Stipendien zu tun hat. The order of es and sie is, however, the inverse of the order in the IT-cleft. All the German equivalents contain aber, and this appears to do the contrastive work that in English is given to the cleft construction. As observed above, one informant did produce cleft structures as the German equivalents of It isn’t Fiona who’s helping us, but Kirsty and It isn’t Fiona we’re helping, but Kirsty. These German IT-cleft structures are in (122b) and (123b). The informant classed them as “slightly formal” and one referee, a native speaker of German, describes them as highly artificial, theoretically possible but hardly ever used. These comments bring us back to the opposition described in Chapter 1 between the repertoire of construction available in a given language and the constructions actually used by the speakers and writers of the language. In the light of the referee’s comments, since the other informant excluded clefts from spontaneous speech, and since the informant who provided the clefts is an English specialist, it is safer to conclude that the data provided by the English specialist is contaminated by his knowledge of English and the time he spends working with English and that spontaneous spoken German has no IT-clefts. The appropriate German equivalents are (122a) and (123a), the former simply with two clauses contrasted in an adversative construction, the latter with an adversative construction reinforced by the particle aber. To drive the preceding point home, Regina Weinert (pers. comm.) points out that in any case clefts with das ‘that’ are far more common, as in (124). Note that (124),

182

Jim Miller

although a cleft structure introduced by das and with Fiona modified by the relative clause die das macht, also contains the particle doch. (122) a.

b.

(123) a.

b.

(124)

Kirsty hilft uns damit, nicht Fiona. Kirsty is-helping us it-with not Fiona ‘It isn’t Fiona who’s helping us, but Kirsty.’ Es ist aber nicht Fiona, die uns damit hilft, sondern Kirsty. it is but not Fiona who us with-it helps but Kirsty ‘It isn’t Fiona who’s helping us, but Kirsty.’ Wir helfen aber nicht Fiona, sondern Kirsty. we are-helping particle not Fiona but Kirsty ‘It isn’t Fiona we’re helping, but Kirsty.’ Es ist nicht Fiona, der wir helfen, sondern Kirsty. it is not Fiona who we are-helping but Kirsty ‘It’s not Fiona we’re helping, but Kirsty.’ Das ist doch die Fiona, die das macht. that is particle the Fiona who that does ‘It’s Fiona who deals with that.’

It seems that, although German has the cleft-like structure exemplified in (124), the central highlighting, contrastive role is played by the particles aber and doch and by the adversative structure. One final piece of evidence is that, when asked to place themselves in a particular map task situation in which English-speaking participants produced Oh it’s the other side of the cliffs I’ve to go, the German informants avoided an es or das cleft and used ja, as in (125). (125)

Ach / Mensch, ich muß ja oh / man I must particle zur anderen Seite der Klippen gehen. to-the other side of-the cliffs go

A type of WH-cleft also occurs which is not so much a cleft construction as a correlative, containing a was clause correlated with a das clause, as in (126).41 (126)

Was Sie hier nicht mitkriegen what you here not with-get das ist der Geruch unseres Phosphorwasserstoffes. that is the smell of-our phosphine ‘What you don’t get here is the smell of our phosphine.’

The Scandinavian languages – Icelandic, Danish, Norwegian, and Swedish – present a pattern quite similar to the German one. They all have IT-clefts, but the IT-clefts are most freely used in overtly contrastive examples such as (127). The IT-clefts were not

Focus in the languages of Europe

183

used by the Norwegian, Swedish, or Danish informants for the example in (125). The Norwegian and Swedish informants rejected the possibility of an IT-cleft, while the Danish informant did give an IT-cleft but deemed it only possible in spoken language and very “heavy,” not natural. The Icelandic informant likewise provided both an ITcleft and a basic declarative clause with focal accent on the relevant words, with the remark that the IT-cleft seemed more natural for the examples in (127a) and (127b). It seems safer to assume provisionally that Icelandic in fact patterns like the other Scandinavian languages. (127) a.

b. (128)

Danish Jeg talte med Fiona. I spoke to Fiona Åh, det er ikke hende, oh it is not her der/som har med legater at gœre. who has with grants to do ‘I spoke to Fiona. Oh, it’s not her that deals with grants.’ Det er ikke Fiona, der/som hjaelper, men Kirsty. it is not Fiona who helps but Kirsty Danish Jeg skull have gået I should have gone den anden vej rundt om klippen. the other way round about cliff-the ‘It’s the other side of the cliffs I should have gone.’

None of the Scandinavian languages have RWH-clefts – or at least none of the informants provided structures that could be analyzed as an RWH-cleft. The Danish, Swedish, and Norwegian informants used the IT-cleft construction, as in (129); the Icelandic informant used an IT-cleft for some examples but also a structure with a deictic constituent in clause-initial position, as in (130b). The Danish informant produced an example with a clause-initial deictic sådan for that’s how you should do it – see (131). This example with sådan points forward to (158) and the discussion of clause-initial so in German in place of the missing *das ist wie : : : WH-clefts are missing from the Scandinavian languages, too; at least, the Danish informants declared that there was a WH-cleft construction which was too unnatural to use either in speech or writing and the Icelandic informant thought the WH-cleft was better as a discourse opener – what I’m going to talk about today is X – than as specifying a subpart of a general task, as does what you’re going to do is go over the bridge and : : : Interestingly, the WH example that the Icelandic informant produced is not a genuine WH-construction; it contains not a WH-pronoun, but a T-pronoun

184

Jim Miller

modified by a relative clause – það sem þú átt að gera er ‘that which you are at doing is’. (129)

Danish Det er der, du skulle have været. it is there you should have been ‘That’s where you should have been.’

(130)

Icelandic Það er það sem þeir eru að gera. it is it that they are at do ‘That’s what they’re doing.’ Þar áttu að vera. there should-you to be ‘That’s where you should be.’

a.

b.

(131)

Danish Sådan her burde du gœre. so here should you do ‘This is how you should do it.’

As observed above, French lacks a free relative construction. Blanche-Benveniste (1991: 59–64) presents examples of IT-clefts (under the heading of “dispositif d’extraction”) and what she calls pseudo-clefts (“pseudo-clivé”). French IT-clefts allow a wide range of cleft constituents – noun phrases, adverbs, and prepositional phrases, as demonstrated in (132). (132) a.

b.

c.

d.

C’ est vraiment à la dernière minute qu’ il est mis au it is really at the last minute that it is put to-the point. point ‘It’s really at the last minute that it (draft) is finalized.’ C’ est ainsi qu’ il se comporte. it is thus that he himself behaves ‘This is how he behaves.’ C’ est que des transistors qu’ on fabrique. it is COMP ART.INDEF transistors that one makes ‘It is only transistors that we make.’ C’ est elle qui a passé l’ examen. C’ est pas moi it is she who has sat the exam it is not me ‘It was her that sat the exam, not me.’

The French construction is like the English IT-cleft in consisting of two clauses; the first is introduced by ce (it) followed by est, the third person singular form of the

Focus in the languages of Europe

185

copula. The second clause looks like a relative clause. The cleft constituent is picked out from other possible candidates and can be contrasted, as in (132d). According to Blanche-Benveniste (1991: 61) the contrastive function attaches more frequently to subject NPs, as in (132d), than to constituents with other grammatical functions. French does not have a reverse WH-cleft construction. The equivalents of the English examples that’s where you should be, that’s how you should do it, that’s what you should be reading are c’est là que vous devez être, c’est comme ça qu’on le fait, and c’est ça que vous devez lire. The English construction can be introduced by that or this, which, as discussed in Section 9.5.1, bring with them their locational meanings, distal vs. proximal. The French construction is introduced by ce, which has no locational meaning unless it is combined with la and ci, as in cela and ceci. The French examples are all parallel to English IT-clefts. The construction that Blanche-Benveniste calls a pseudo-clivé is not analogous to the English WH-cleft/pseudo-cleft. Consider the example in (133) and its English equivalent. (133)

Ce qui est dangereux pour le public c’est la petite that which is dangerous for the public it-is the petty délinquance. crime ‘What is dangerous for the public is petty crime.’

The English example has the free relative clause what is dangerous; the French has a head pronoun ce modified by the relative clause qui est dangereux. In the English example the free relative clause is followed directly by the copula is, and the free relative clause is the subject of is. In the French the relative clause is followed by a complete clause in which the copula est has its own subject, ce. That is, the French construction is more a correlative than a WH-cleft. Irish has only the IT-cleft construction, as in (134). (134)

Ní bhí Fíona ati ag cabhru ach Kirsty. NEG be-3F Fiona REL-BE at helping but Kirsty

Irish has a construction introduced by sin (é) (that is) which looks analogous to the English reverse WH-cleft but sin is followed not by a free relative clause but by the pronoun é modified by a relative clause. This construction is illustrated in (135). (135)

Sin é atá siad a dhéanamh. that-is it REL-BE they PROG do ‘That is what they are doing.’

The Irish equivalent of the English reverse WH-cleft typically has a pronoun é modified by a relative clause or a full noun phrase modified by a relative clause, as in (136a) and (136b). In (136c) there is a full noun phrase an chaoi which is modified

186

Jim Miller

by the prepositional phrase len a dhéanamh. This is not a cleft structure of any kind but a copula construction consisting of a single clause (at least in traditional terms). (136b) is an alternative to (135). Both contain a relative clause but it modifies a pronoun in (135) and a full noun in (136b). (136) a.

b.

c.

Sin (é) an áit that-is (it) the place ar cheart duit a bheith (ann). REL-BE right to-you to be (in-it) ‘That’s where you should be.’ Sin é an rud atá siad a dhéanamh. that-is it the thing REL-BE they PROG do ‘That is what they are doing.’ Sin (é) an chaoi len a dhéanamh. that-is (it) the way for its doing ‘That’s how you do it.’

The final Irish example, (137), shows the same cleft construction but introduced by is as opposed to sin. The English IT-cleft, from the map task dialogues, is highly contrastive. The force of the Irish equivalent is increased by the occurrence of the emphatic pronoun mise in the relative clause. (137)

(Is) ar thaobh eile na haille (it-is) on side other the-GEN-F.SG cliff a chaithfeas mise a dhul. REL must-REL-FUT I-EMPH GO-INF ‘It’s on the other side of the cliffs I’ve to go.’

Greek has a marginal WH-cleft construction but lacks IT-clefts and reverse WHclefts. The WH-cleft is exemplified in (138). (138)

Afto pou me enohli ine i adiaforia tu. this that me bothers is the indifference of-him ‘What bothers me is his indifference.’

As with French and Irish, what is missing is a free relative structure. Afto pou consists of a deictic, afto, modified by a relative clause with the complementizer pou. However, this WH structure was not used by the informants in their rendering of the discourse-opening WH example and of the WH example that is used to draw a line under one section of discussion and to point ahead to the next stage of the route-drawing. Consider (139) and (140). (139)

To thema pu tha mas apasxolti simera ine : : : the subject that will us occupy today is : : : ‘What we are going to discuss today is : : :’

Focus in the languages of Europe

(140)

187

Aku, ta pragmata ine apla, listen the things are simple tha perasis ti yefira : : : FUT you-cross the bridge ‘Listen it’s simple you will cross the bridge : : :’ = ‘what you’re going to do is cross the bridge : : :’

(140) could be rendered by the quasi-WH structure afto pou kanis ine to eksis – tha perasis ti yefira ‘that which you will do is the following – you will cross the bridge’. This second rendering brings out clearly the key property of the Greek quasi-WHcleft: it occurs where the copula is followed by another noun phrase, here to eksis. Another property is that it appears to be more frequent in writing; the informants preferred (140) for the situation involving speech. The IT-clefts received different treatment depending on whether there was an overt contrast. (141) is introduced by a particle diladhi and alli (other) carries focal accent. The informants were inclined to make the contrast overt, as indicated by the phrase in brackets. (141)

A! diladhi tha pao stin alli plevra (oxi s’afti). Ah that-is FUT go-1SG to-the other side (not to this) ‘Oh it’s the other side I have to go to.’

Where there was an overt contrast involving a negative main IT BE clause the force of the English IT-cleft was conveyed partly by an adversative structure and partly by word order. (142) has the adversative structure marked by alla and the order VS in the first clause, voithai i Fiona. (143) has an adversative structure marked by oxi and the direct object of the clause is in clause-initial position – tin Kirsty voithame. (142) a.

b.

(143)

Dhen mas voithai i Fiona, alla i Kirsty. not us helps the Fiona but the Kirsty ‘It’s not Fiona who’s helping us, but Kirsty.’ I Kirsty mas voithai, oxi i Fiona. the Kirsty us is-helping not the Fiona ‘It’s Kirsty who’s helping us, not Fiona.’ Tin Kirsti voithame, oxi ti Fiona. the Kirsty we-are-helping not the Fiona ‘It’s Kirsty we are helping, not Fiona.’

The force of the reverse WH-clefts is conveyed by putting the cleft constituent in clause-initial position, as in (144) and (145). (145) is taken to consist of the NP Syntactic Structures followed by the clause afto na dianasis. Afto is not at the beginning of the entire chunk of syntax but is in first position in the clause.

188

Jim Miller

(144)

Edo eprepe na ise (oxi eki). there it-behoves COMP be-2SG (not here) ‘That’s where you should be (not here).’

(145)

Syntactic Structures afto na diavasis. Syntactic Structures this COMP read-2SG ‘Syntactic Structures is what you should be reading.’

Other IT-clefts were also conveyed into Greek by means of word order and pitch. The Greek examples in (146)–(147) both contain a simple clause, but the constituent order changes to accommodate the force of the English IT-clefts. In (147) Mariana carries focal pitch, and pitch and order combine to achieve the highlighting, but in (146) Mariana can also carry focal accent. (146)

O Miltos ide ti Mariana. the Miltos saw the Mariana ‘Miltos saw Mariana.’

(147)

Ti Mariana ide o Miltos. the Mariana-DO saw the Miltos-S ‘It was Mariana that Miltos saw.’

We conclude this section by observing that, although no Slavic language has cleft constructions, spoken (but not yet written) Russian does have a rudimentary construction with the force of the English IT-cleft. That is, it identifies an entity and excludes all other possibilities. This construction consists of a demonstrative pronoun followed by a complete clause, as in (148). (148)

Èto ivan stuˇcit v dver’. that Ivan is-knocking at the-door ‘It’s ivan who is knocking at the door (and not Petr).’

The structure of (148) is not certain. Since the present-tense forms of the copula are zero, it may consist of two clauses: eto + copula and Ivan stuˇcit v dver’. Any constituent in the second clause can carry the focal accent.

10. Spontaneous map task dialogues 10.1. Introduction It is interesting to compare the cleft structures in the five sets of spontaneous map task dialogues with the cleft structures that are available in principle in the grammar of the languages involved. The biggest difference comes from the Catalan data; the questionnaire yielded a large number of cleft constructions, while the spontaneous

Focus in the languages of Europe

189

dialogues yielded only one cleft construction but many instances of particles – which term includes pieces of syntax such as o sigui ‘let it be’. The Hungarian dialogues offer a wider range of particles than were elicited by the questionnaire – Hungarian has no cleft constructions and the only attempts to construct a cleft construction came from an informant answering the questionnaire who was clearly calquing the English construction and immediately dismissed the result as “not Hungarian.” The German data from questionnaire and dialogues was consistent with respect to the lack of clefts but again the dialogues produced more particles. The Bulgarian map task dialogues yielded no cleft constructions, a result which matches the (meager) questionnaire data for Bulgarian. English cleft constructions were taken as the point of orientation for the questionnaire, but it should be pointed out that, while clefts are indeed used in spoken and written English, IT-clefts are very rare, WH- clefts are more frequent (with the proviso that the WH-cleft of spoken English has a less integrated structure than the WH-cleft of written English), and RWH- clefts, the typologically most unusual construction in our sample of languages, are most frequent of all. However, far more frequent than the clefts is the particle like, which is excluded from written academic discourse, is frowned on by arbiters of good literary style, but is very frequent in spontaneous speech and other less formal spoken genres. As a highlighting device particles are very frequent in all languages but where the written language makes use of other devices a questionnaire will not necessarily elicit all or any of the particles available.

10.2. English English, as will be evident from the preceding discussion, poses no problem; all three cleft constructions occur. Interestingly, as stated above, the IT-cleft is very rare in the map task dialogues (and in other texts). This contrasts with the distribution of cleft constructions across languages, with the IT-cleft being most widespread and the other types much rarer. IT-clefts are used to pick out a subset of entities from a set of available entities. RWH-clefts, in the map task dialogues, hold up the progress of the dialogue by pointing back at, and highlighting, some entity, including locations or sections of route, which has not only been mentioned but has been the subject of the immediately preceding discussion. The RWH-clefts drive home a piece of information. The WHclefts are used to draw a line under a piece of discussion and point forward to the next instruction and thereby to the next section of route. According to the standard analysis, these are focusing constructions which make particular constituents prominent. It is true that these constructions sometimes have a powerful micro-focus function, but as Delin (1990: 3–8) shows, the large majority

190

Jim Miller

of IT- and WH-clefts in a body of recorded conversation fail to make their allegedly focused item either the site for a tonic stress or the target of pronominal reference. Delin suggests that the function of the clefts is to be explained with reference to factors such as marking the distinction between presupposition/assertion for the content of the cleft, indicating a novel instantiation of the variable in the presupposition, and the maximality of that instantiation. Her proposals run parallel to components in Halliday’s treatment, as in Halliday (1985) when he glosses What the Duke gave to my aunt was this teapot as ‘the Duke gave this teapot and nothing else’. Sornicola (1988) draws attention to the important macro-focus function of the IT- and WH-clefts. IT-clefts regularly occur in the first sentence in written texts, at the beginning of major sections such as chapters or newspaper articles, or at the beginning of less major sections such as paragraphs. WH-clefts regularly create a climax at the end of major sections. Sornicola’s observations relate to written texts, but the same climax function (plus creating a bridge to the next section of dialogue) is performed by WH-clefts in the map task dialogues.

10.3. German IT-clefts are missing. The WH-clefts have a slightly different discourse function from that of English WH-clefts. Weinert (1995) suggests that the German WH-clefts serve not to introduce a proposition, but to expand on a proposition before it ceases to be the object of discussion. RWH-clefts are very infrequent (none in the German map task dialogues) and when they do occur, they have a very powerful highlighting effect. The initial deictic carries a strong focal accent, which can be reinforced by particles such as eigentlich, genau, and immer, while the force of the English ITclefts can be captured by particles such as ja and aber. (149a) and (149b) exemplify the construction with clause-initial deictic and particle. (149) a. b.

Das ist genau was ich meine. that is exactly what I mean Das ist doch genau was ich auch sage. that is particle exactly what I too say ‘That’s exactly what I say, too.’

Weinert (1995) observes that where we might expect an RWH in English, her German data offers clauses with das or da in clause-initial position. Consider (150)– (151) and A6 in (159). This construction in particular takes on the function of RWHclefts in English texts (as does an analogous construction in Bulgarian (152), SerboCroatian (153), Russian (154), and Polish (155)). Note the particles in the Russian and Polish examples. (150)

Ja und genau das wollte ich ihm nicht sagen yes and exactly that wanted I to-him not to-say

Focus in the languages of Europe

191

um ihn nicht zu kränken. so-as him not to hurt ‘Yes and that’s exactly what I didn’t want to tell him so as not to hurt him.’ (151)

Da endest dann. there finish-2SG then ‘That’s where you finish.’

(152)

Bulgarian Tam bi trjabvalo da si. there cond be-necessary-PST COMP you-are ‘That’s where you should be.’

(153) a. b.

Serbo-Croatian Tamo je tvoj cil. there is your goal Tamo je, gdje trebaš biti. there is where you-should be

(153b) brings us back to the problem of repertoires and usage. The informant declared that (153a) is the natural equivalent of that’s where you should be; that is, a simple copula construction. (153b) was produced by the informant and judged grammatical but not felicitous. Was the informant influenced by the English original or do grammars of Serbo-Croatian recognize an RWH construction which may be of limited occurrence? (154)

Russian Vot tuda vam nado prijti. particle to-there you-DAT necessary to-come ‘Look, that’s where you should be.’

(155)

Polish tam wła´snie powiniene´s sie˛ znale´zc´ there exactly you-should self find ‘That’s where you should be.’

Returning to German, it must be pointed out that not all German clauses with das or da in initial position correspond to RWH-clefts in English. For instance, the example in (156) points forward to the next section of route, whereas the RWH-clefts in the map task dialogues point back to reinforce information in a preceding section of dialogue. (157) points neither back nor forward, nor does it reinforce any information; the speaker is merely responding to the mention of the label ‘ruin’.

192

Jim Miller

(156)

Genau, und da gehst du jetzt waagerecht. exactly and there go you now horizontally ‘Exactly, and from there you go horizontally.’

(157) a.

A: B:

X ist an der unteren Grenze der Schrift ‘Ruine’. X is on the lower border of-the label ruin O Gott nee nee du das hab ich noch nicht ganz. o God no no you that have I yet not completely ‘Oh God–no, no–I haven’t got that at all.’

Although German has RWH-clefts, not only are they infrequent, but their structure is tightly restricted. English examples such as that’s why you go round it and that’s how you reach the goldmine have no direct counterparts in the German data. The status of das ist warum ‘that’s why’ is unclear; it does not occur in the German dialogues and some speakers of German (Regina Weinert pers. comm.) feel that it is unacceptable as it stands but might be rendered palatable by the addition of a particle – say, das ist genau warum ‘that is exactly why’. In contrast, *das ist wie ‘that is how’ is definitely not acceptable, even with particles. Weinert suggests that German speakers use instead the construction in (158) with clause-initial so. The map task dialogues offer no instances of das ist wo (‘that’s where’), though the corresponding English RWH-cleft is extremely common. It is not clear whether this is an accidental property of the dialogues or whether the construction is not acceptable, but in any case what occurs in the German dialogues is the construction exemplified in (151) with clause-initial da. (158)

Und und so ungefähr sieht sieht der Bogen aus. and and so roughly looks looks the curve out ‘And that is roughly how the curve looks/what the curve looks like.’

One more particle must be mentioned, and that is und zwar, which highlights information narrowing down the reference of a noun phrase. In this function zwar always combines with und. Consider (159). (159) a.

A:

Und du hast also rechts and you have therefore on-the-right keinen Aussichtspunkt mehr? no viewpoint more ‘And does that mean you have no other viewpoint on the right?’ B: Nee. no A2: Weil ich hab nämlich noch einen Aussichtspunkt because I have you-see yet one viewpoint ‘The thing is, I’ve got another viewpoint’

Focus in the languages of Europe

193

und zwar ist der direkt unter der Ostsee. and indeed is it directly under the Baltic ‘and where it is is right under the Baltic.’42 B2: Ja. yes A3: Und zwar eh and indeed eh ‘And like eh’ nicht unter dem ehm rechten unterem Zipfel not under the em right-hand bottom corner ‘not under the em right-hand bottom corner’ der Ostsee sondern of-the Baltic but ................................ A6: Ja da ist ungefähr der Aussichtspunkt. yes there is roughly the viewpoint ‘Yes that’s roughly where the viewpoint is.’

10.4. Catalan The Catalan data obtained via the oral questionnaire indicated that Catalan has ITand WH-clefts but no RWH-clefts. The equivalents offered for that’s where you should be are shown in (160a)–(160c). (160b) is a cleft, but an IT-cleft; (160a) has allà ‘there’ in clause-initial position and with focal stress; (160c) has unmarked word order but the clause is introduced by doncs ‘so’. (160) a.

b.

c.

Catalan Allà hauries de ser. there you-should to be ‘That’s where you should be.’ És allà que hauries de ser. is there that you-should to be ‘That’s where you should be.’ Doncs hauries de ser allà. so you-should to be there ‘That’s where you should be.’

However, in the Catalan spontaneous map task dialogues only one cleft occurred, the quasi-WH-cleft in (161). (161)

Allavorens lo que deus has de fer. then it that you-may you-have to do ‘Then what you may have to do’

194

Jim Miller

es vorejar, suposo, aquest sendero public. is to-border I suppose this path public ‘is to go along the edge of this public path.’ A major role is played by the particle o sigui. O is equivalent to or and sigui is the third person singular subjunctive form of the copula. However, ‘or let it be’ is hardly an adequate gloss in the various contexts, particularly where speakers are offering or requesting additional and not alternative information. On the basis of the English map task dialogues, we can imagine that an English-speaking participant might use a WH-cleft, an RWH-cleft, or like. For instance, o sigui introduces utterances in which the instruction giver adds to and clarifies an instruction or recapitulates or introduces a complex instruction, as in (164); and in which the instruction follower repeats the instruction giver’s words or adds to an instruction and asks for the addition to be verified as in (162) and (163). (162) a.

A1: I a l’ altura de la granja and at the level of the farm ‘And at the level of the farm’ B1: Si yes A2: Tires recte amunt you-go straight up ‘You go staight up’ B2: A quina banda de la granja, a la dreta to which side of the farm to the right ‘To which side of the farm to the right’ o a l’ esquerra? or to the left ‘or to the left?’ A2: A l’ esquerra to the left B3: O sigui a l’ esquerra de la granja that is to the left of the farm ‘That is to the left of the farm’ o sigui jo ara a-baixo. that is I now down ‘that is I go down.’

(163) a.

A:

Quan arribis al marge esquerra del salt d’aigua, when you-arrive at-the edge left of-the waterfall ‘When you arrive at the left edge of the waterfall’ tires per amunt. you-go towards up ‘you go upwards.’

Focus in the languages of Europe

B1: mhm A2: eh? ‘OK?’ B2: O sigui vorejant el salt d’aigua cap dalt that is bordering the waterfall towards up ‘So like going up along the edge of the waterfall.’ A3: Vorejant el salt d’aigua bordering the waterfall ‘Going along the edge of the waterfall.’ (164) a.

A1: D’acord doncs agafes, voreges aquests OK then take border this ‘OK take this border then’ que dius que tens per sota, suposo that you-say that you-have through below I-suppose ‘that you say you’ve got, underneath take it’ B1: Uns camps privats some fields private ‘private fields’ A2: que deu ser that might be ‘maybe’ si deu ser voreja ’ls per sota yes you-have-to-be bordering them through below ‘yes you have to go along the edge underneath’ i llavons recte amunt and then straight up B2: mhm A3: i voreges la pedrera de granit and border the quarry of granite ‘and you go along the edge of the granite quarry’ per per dalt by by the-top B3: per la dreta? on the right? A4: o sigui, comences per la dreta, el that is you-begin on the right it ‘so what you’re doing, you begin on the right’ B4: de la pedrera of the quarry A5: de la pedrera el passes per dalt of the quarry it you-pass by the-top ‘of the quarry, you pass by it at the top’

195

196

Jim Miller

i baixes una miqueta a l’ esquerra. and you-go-down a bit to the left ‘and go down a bit to the left.’ A number of other discourse particles occur: doncs in the proper context can be equivalent to ‘therefore’ or ‘then’, as in Doncs perq’e ho vas dir ‘so why did you say it then?’ but in many contexts this interpretation is inappropriate. However, in the map task dialogues it does count as extra coding which highlights a following phrase or clause and has to be included in the set of highlighters, and many of the occurrences can be interpreted as ‘therefore’. Consider (165) and (166). (165) a.

A1: Veure la sortida que la tens? let-me-see the start COMP it you-have ‘Let’s see, have you got the start?’ B1: Si yes A2: Ah, doncs sortim baixant pel paller ah then we-leave going-down by-the haystack ‘Ah, then we set off and go down past the haystack’ tens un paller? you-have a haystack ‘do you have a haystack?’

(166) a.

A1: Tens un castanyer? you-have a chestnut tree ‘Do you have a chestnut tree?’ B1: Si que tinc un castanyer. yes COMP I-have a chestnut tree ‘Yes, I’ve got a chestnut tree.’ A2: Doncs ja esta. well already is (done) ‘Well that’s it done.’

In (165) doncs can be interpreted as signaling that the speaker’s premise has been satisfied; the instruction follower’s map has the starting point and it follows that the speaker can issue the first instruction. Similarly in (166) the speaker’s premise is satisfied; the instruction follower’s map does have the chestnut tree and the speaker can specify the chestnut tree as the finishing point. A2 in (166) contains the clause ja esta. Ja is equivalent to ‘already’ or ‘finally’ and esta to ‘is’ or ‘is done’. This clause is used to signal the conclusion of the task; many of the English map task dialogues contain phrases such as that’s you finished or that’s it. There is one instance of es a dir ‘that is to say’ and various occurrences of llavors, allavors, and allavorens, which are variants of a particle with basically temporal

Focus in the languages of Europe

197

meaning, ‘then’. There are instances of the left-and right-dislocation discussed in Section 8.3 under the heading of defocusing, and instances of left dislocated constituents being highlighted, as in (167) and (168). (167)

Per aixo et deia que tens l’ arribada. for that and I-was-saying that you-have the arrival ‘That’s why I was saying that you have the finish.’

(168)

Mult aprop de la muntanya blanca has de passar. very close to the mountain white you-have to pass ‘You pass very close to the white mountain.’

The Catalan spontaneous map task dialogues seem short of highlighting devices in comparison with English, German, and Hungarian. One question that needs to be explored is whether the construction that was handled under salience-reduction, that is, left- or right-dislocation of given constituents out of the clause core in order to make way for new constituents, can also be seen as highlighting them. Consider (169) and (170). (169)

A vere me sembla que let-me-see me thinks that aixo de la granja tambe el tens tu, no? this of the farm big it have you no ‘Let me see I think – this big farm, you’ve got it, haven’t you?’

The big farm is treated as given, since it has already been mentioned. The speaker treats as new information el tens tu, which constitutes the clause core, but it is arguable that aixo de la granja tambe gains some salience simply by virtue of being in clause-initial position. The highlighting effect is stronger in (160) which is an instance of the same construction but which might correspond to an RWH-cleft in the same context in an English map task dialogue. Arguably, there is a contrast between previous instructions that B has understood and this one that he/she has not.43 (170)

A1: La voreges puges cap dalt it you-border you-go towards top ‘You go along its side towards the top’ pero diguessim mmmm agafant a la mitat but let’s-say mmmm starting in the middle ‘but let’s say ehm starting in the middle’ espera, espera, espera, a vere let-me-see wait wait wait ‘let me see, wait, wait, wait’ que t’ ho explico que t’ ho explico COMP to-you it I-explain COMP to-you it I-explain ‘till I explain it to you, till I explain it to you’

198

Jim Miller

B1: Si perque aixo no ho he entes. yes because this not it I-have understood ‘Yes because this is what I haven’t understood.’

10.5. Hungarian The data collected from informants by means of the oral questionnaire revealed that, like Finnish, Hungarian has no clefts and the highlighting achieved by clefts in English is achieved by means of the highlighting preverbal position, particles, and pitch. Consider the examples in (171). (171) a.

b.

c.

d.

e.

Ja, a sziklák másik oldalán kell mennem. yes, the cliffs’ other side-loc must I-go-INF ‘It’s the other side of the cliff I’ve to go.’ Nem Fiona segít, segít Kirsty. not Fiona helps helps Kirsty ‘It isn’t Fiona who’s helping, but Kirsty.’ A magyar mondattanról fogok beszélni. the Hungarian syntax-about will-I talk-inf ‘What I’m going to talk about is Hungarian syntax.’ [introduction to lecture] Most (pedig) akkor át-mész a híd-on, és fel-mesz now particle then across-you-go the bridge-on and up-you-go 2 cm-t 2 cm ‘What you’re going to do is cross the bridge and go up 2 cm.’ A következökben, the following-in át kell mened a híd-on, through have-to you-go the bridge-on és 2 cm-t kell menned fölfele. and 2 cms have-to you-go-inf upwards ‘What you’re going to do is go over the bridge and up 2 cm.’ Hát /Na ott kellene en-n-ed. particle there should be-you-inf ‘That’s where you should be.’

(171a), (171c), and (171e) show, respectively, an oblique object, a direct object, and a locational complement in preverbal position and thereby highlighted. (171d) and (171e) demonstrate the role of particles; pedig in (171d) signals that an instruction is coming. In (171e) hát is roughly equivalent to well and na in the spontaneous map task dialogues precedes utterances that slow down the discourse by pointing to

Focus in the languages of Europe

199

things. (171d) illustrates the use of noun phrases to provide extra coding, much as happens in English: for example, the next move is, you go over the bridge. Note, too, the adversative construction in (171b) with an SV-clause balancing a VS-clause. The Hungarian spontaneous map task dialogues confirm the above picture, add a correlative highlighting construction, draw attention to the highlighting power of deictics, and demonstrate the variety and frequency of highlighting particles. The correlative construction, in conjunction with word order, is exemplified in (172). (172)

És ami nincs meg, and what is-not particle kell rajzolnod valahová. azt majd föl that then particle must draw-INF-2S somewhere-to ‘And what you haven’t got you’ll have to mark somewhere (on the map).’

The set of entities not on the addressee’s map is made salient through being referred to by the independent NP és ami nincs meg; this NP is picked up by the anaphor direct object azt in the following clause. The highlighting power of deictics, particularly when at the front of a clause and accompanied by a particle, is shown by (171e) above. An example with different syntactic structure is in (173). (173)

És ott vagyunk a szép szikláknál, igaz? and there we-are the attractive cliffs-at true ‘And we’re there at the attractive cliffs, right?’

In this example, the target of és is spelled out by the noun phrase following the verb, a szép szikláknál. It is tempting to treat the example, out of context, as involving an afterthought, but it may be an example of given entities backgrounded by being put to the right of the clause. The particles that turn up in the spontaneous dialogues are akkor and na, which also turn up in the questionnaire data; tehát, with which the speaker consolidates – checks on a section of route drawn by the instruction follower or checks on an instruction just received. szóval, with which the speaker summarizes a piece of information – Szilvia Papp (pers. comm.) suggests in other words as an approximate gloss; hát which occurs when the speaker expresses impatience or incredulousness; is which is difficult to provide with a single English gloss; and e, which highlights constituents to which it is attached in interrogative sentences. Examples of the particles are given in (174)–(179). (174)

Tehát akkor ott menjek el derékszögben? particle particle there I-should-turn away at-a-right-angle ‘Like I should turn there at a right angle?’

(175)

Szóval neked nincsen majom? particle at-you not baboons ‘You mean/like you don’t have baboons?’

200

Jim Miller

(176)

Szóval visszafordulsz egy kicsit. particle you-turn-back a bit ‘What I mean is you turn back a bit/ that is, you turn back a bit.’

(177)

Hát akkor állítólag particle particle are-asserted eléggé hasonló a két lap, Zsolti. fairly similar the two map Zsolt ‘Well now the two maps are supposed to be fairly similar, Zsolt.’

(178) a.

Na akkor meg is érkeztél. particle then particle particle you-have-arrived ‘Well then you have arrived.’ Nincs is banánfád. NEG-be particle your-banana-tree ‘You haven’t even got a banana tree!’

b.

(179)

Piramis alatt van-e macskaköves utca? pyramid below is-particle cobbled street ‘Below the pyramid is there a cobbled street?’

The glosses of (174) and (175) contain like because it occurs frequently in similar contexts in the English map task dialogues.

10.6. Bulgarian As already remarked, the Bulgarian data collected via the questionnaire provided no cleft constructions. Where English has IT-clefts, the informant used particles or word order. (The denotation of ‘particle’ is stretched to take in items such as znaˇci in (180). This is the third person singular of a verb meaning ‘signify, mean’. In Atanasova et al. (1975) znaˇci is listed as impersonal and glossed as ‘so’ or ‘then’.) The questionnaire equivalents of English IT-clefts are given in (180)–(182). (180)

O, znaˇci ot drugata strana na skalata trjabva da oh it-means from other-the side of cliff-the is-necessary that mina. I-pass ‘Oh, it’s the other side of the cliffs I’ve to go.’

(181)

Ama ne se zanimava tja s˘us stipendiite. but not self occupies she with grants ‘But it’s not her that deals with grants.’

Focus in the languages of Europe

(182)

201

S nas ne se zanimava Fiona, a Kirsty. with us not self occupies Fiona but Kirsty ‘It’s not Fiona who deals with us, but Kirsty.’

In (180) the main proposition is highlighted by znaˇci and the information that contrasts with the speaker’s expectations is conveyed by the prepositional phrase ot drugata strana na skalata, which is in clause-initial position. The focal accent is on drugata. In (181) and (182) the cleft constituent in Hedberg’s terms (see Section 9.5.1) is the subject noun phrase, tja and Fiona; these follow the verb. In (182) there is also an adversative structure, ne : : : Fiona a Kirsty. That is, word order, particles, and adversative structures do the work of IT-clefts. Structures in which two clauses are balanced, either adversatively or disjunctively, occur in the spontaneous map task dialogues. Consider (183) and (184). (183)

Tuka mina njama, ima njakakvi malini. here mine is-not is some raspberries ‘There isn’t a mine here, there’s some raspberries.’

(184)

A:

B:

Otgore li trjabva da mina ili pod nego? above Q is-necessary that I-go or under it ‘Do I go above it or under it?’ Pod nego ne minavaš nad nego. under it not you-pass above it ‘Under it you don’t go above it.’ Ostavaš si pod nego. you-stay to-self under it ‘You keep below it.’

The map task dialogues do offer one example of a construction that is used where a speaker of English might well use an IT-cleft. The construction is interesting because it is introduced by tova (the deictic ‘that’); in this respect it is like the Russian construction in (148). The Bulgarian example is in (185). (185)

A:

B:

Da obaˇce po-k˘am sredata na lista. yes however close-to middle-the of page-the ‘Yes but closer to the middle of the page.’ Toˇcno taka. exactly so Tiva sa mote grešni malini tuka da. that are my-the incorrect raspberries here yes ‘That’s right it’s my wrong patch of raspberries that is here yes.’

The English WH-cleft in the questionnaire generated the Bulgarian example in (186). (186)

I taka, minavaš po mosta and so you-cross over bridge-the

202

Jim Miller

i otivaš 2 sm nagore. and you-go 2 cms up ‘What you’re going to do is cross the bridge and go up 2 cms.’ Where English speakers might have produced WH-clefts, the participants in the Bulgarian map task produced examples with various particles – znaˇci, faktiˇceski, and v smis˘ul. (NB: It is not being suggested that this account exhausts the meaning and usage of these Bulgarian words and phrases. We are merely pointing out that where English speakers might have used a cleft – to finish off a piece of discussion and point forward to the next stage of the task – the Bulgarian participants used these particles.) The relevant examples are in (187)–(189). (187)

Minavaš pokraj znaˇci pokraj tjax you-pass along so along these znaˇci se dvi’iš na-a-a-nagore so self you-move u-u-up i leko na-a-a-nadjasno. and slightly to-to-the-right ‘You go along what you’re doing is you go along the edge of these ones and you move up and slightly to the right.’

(188)

Da be da no v smis˘ul na kakva visoˇcina-a-a. yes OK yes but in sense to what height ‘Yes OK but what I want to know is how high.’

(189)

A pod nego taka stignala si pod nego and under it right having-climbed you-are under it sega prod˘ulžavaš now you-continue faktiˇceski nadjasno polukr˘ug izvivaš you-bend particle to-the-right semi-circle za da mineš nad nego. so-that you-pass above it ‘Under it right you’ve climbed up under it now you continue what you do is you curve to the right in a semi-circle so that you go underneath it.’

In (189) faktiˇceski is glossed as ‘particle’. It does not correspond to the English in fact or factually, although there is one example where it could be interpreted as in fact or in actual fact. But the force of this example, (190), can also be conveyed in English by a WH-cleft. (190)

I meždu-u-u i faktiˇceski and betwee-een and particle

Focus in the languages of Europe

203

tvojata mina se namira your-the mine self finds meždu poleto i malinite? between field-the and raspberries-the ‘And between the fact is that your mine is between the field and the raspberries.’ ‘And between what it is your mine is between the field and the raspberries.’ About the equivalent of RWH-clefts there is little to say except that, as in German, the Bulgarian construction involves putting a deictic in clause-initial position as in (191). (191)

Tam bi trjabvalo da si. there COND be-necessary-PST that you-are ‘That’s where you should be.’

One further particle must be exemplified, and that is imenno ‘namely’. It occurs regularly and is used where speakers of English would probably use an IT-cleft. It is exemplified in (192). (192)

A:

B: A:

Imaš li napisana zelena linija? you-have Q written green line ‘Have you got “green line” written?’ Da. yes Taka znaˇci OK so imenno s˘uas tozi ost˘ur zavoj namely with this sharp curve zaobikaljaš zelenata linija. you-go-round green-the line ‘It’s with this sharp curve that you go round the green line.’

11. Conclusions The major typological fact emerging from the investigation is that in Europe there is a west-east cline from languages with clefts to languages without clefts. In the east, the Finno-Ugric and Turkic languages do not have clefts at all. Indo-European languages in the east of Europe, such as Russian, Bulgarian, Serbo-Croatian, and Polish, do not have clefts, whereas Indo-European languages in the west of Europe do have clefts. Of the languages in the west of Europe, English is the most striking in

204

Jim Miller

having three cleft constructions which are not only described in grammars of English, but are in frequent use and occur in the English map task dialogues (and, of course, in conversation and in writing). Other languages in the west of Europe have one cleft construction or two, but not three. French has an IT- cleft and a WH-cleft – ce que nous allons faire, c’est rester ici, that which we are-going to-do is to-stay here = ‘what we’re going to do is stay here’. Norwegian, Swedish, and Icelandic have an IT-cleft but not separate RWH- and WH-clefts. Particles have a role to play in all languages but they are striking in the languages without clefts ( Indo-European and Finno-Ugric) and in German, which occupies a mid-point on the cline; German has clefts, which occur far less frequently than in English, but it has frequently-occurring particles. Of course, English also has particles – really, actually, in fact, namely come immediately to mind. The fact is, however, that in the English map task dialogues, particles do not occur except for like, which does have regular and analyzable patterns of occurrence (as described by Miller and Weinert 1995) but which is excluded from written language and from formal speech. In contrast, the German, Hungarian, and Catalan map task dialogues are well-supplied with particles and the questionnaire examples from Finnish and Russian also have particles. The use of word order requires further investigation. The Greek, Hungarian, and Finnish questionnaire data contain examples in which word order by itself highlights a constituent. There are Hungarian and Finnish examples of word order together with a particle. And there is the fact that for contrastive purposes, a special highlighting position is available in Finnish, Hungarian, and Turkish, preverbal in Hungarian and Turkish, and clause-initial in Finnish. Clause-initial position in Russian and German is for given constituents, and it is clear from the corpus and questionnaire data that Russian highlights constituents by means of word order plus particles or word order plus an overt contrast between constituents. Cleft constructions apart, two further major facts emerge. One is the importance of the distinction between spoken and written language. This is obviously reflected in the enormous role of intonation in speech and its complete absence from writing, and also in the role of particles. Particles do occur in written German, Russian, Finnish, etc., but many particles occur only in speech. (Cf. Fernandez [1994] for a detailed analysis of spoken and written French.) Where informants have provided examples of highlighting by word order alone, we may ask whether a sample of spontaneous conversation or map task dialogue would offer word order plus particles. This question is prompted by Zemskaja’s corpus of spontaneous spoken Russian and by the German spontaneous map task dialogues. Some informants explicitly commented on constructions as belonging to the written language. Some examples of clefts were produced by informants and then dismissed as not really Finnish, Italian, Hungarian, or Serbo-Croatian. These are probably examples that might occur in writing, but they also look like examples that are

Focus in the languages of Europe

205

calqued on English. It is quite possible that, just as in the past languages borrowed constructions from Latin, so languages will borrow constructions from English. The borrowings will be made by people who know English, particularly written English, and are likely to appear in writing. Typologists beware! The second major phenomenon is the role of particles. These have not been properly studied in contemporary linguistics, possibly because written English lacks them and possibly because where they are present they do not belong to core syntax and are difficult to analyze in terms of truth conditions. As a category, however, particles are present in abundance in the earliest European texts, namely, Homer (cf. Palmer 1962) and in Classical Greek, where their interpretation is far from straightforward (cf. Denniston 1934). In Homeric Greek, a number of particles are generally analyzed as deriving from deictics; this is an enduring property, obvious in the English clefts, but even more striking in the embryonic cleft structure in Polish and spoken Russian (cf. (148)). The data gathered here indicate that fresh research is needed, beginning with a typology of particles.

Notes 1.

I am indebted to the large number of friends and colleagues who helped in the preparation of this chapter. Barbara Wehr generously gave me her annotated Irish data and commented on the first draft. Rob Mullally, Elisabet Engdahl, Enric Vallduví, Lara Rjazanova, Jelena Meznari´c, Magda Stroinska, and Elina Rigler spent considerable time and effort supplying and discussing data from Irish, Swedish, Catalan, Russian, Serbo-Croatian, Polish, and Finnish. Szilvia Papp not only provided Hungarian data but rounded up other informants, taped and transcribed map task dialogues in Hungarian, and allowed me to use data from her MSc thesis. Regina Weinert recorded and analyzed German map task dialogues which she allowed me to use, shared her data from conversational German, and commented insightfully on an earlier version of this paper. Enric Vallduví and Merce Prat collected and transcribed Catalan map task dialogues, and the former dealt patiently and in detail with my inquiries about the grammatical structures in the data. Martine Grice gave me electronic transcriptions of ten Bulgarian map task dialogues. Celia Kerslake discussed Turkish at length and supplied some examples. At very short notice, Will Lamb selected examples from his corpus of spoken and written Scottish Gaelic and annotated them. Machtelt Bolkestein commented in detail on the first draft. Data was also supplied by Eranda Kabashi (Albanian), Luana Babini (Italian), Judit Szito (Hungarian), Harold Blaenes (Norwegian), Kristjan Arnason (Icelandic), Heike Stoldt and Hans Peters (German), Ulla Tuomarla (Finnish), and ˙ Zelal Güngördü and Yilmaz Kılıcaslan (Turkish). My colleagues in Theme Group 1 saved me from a number of errors, and comments from Georg Bossong and Bernard Comrie have led to various improvements. Finally, and especially, I must mention the late Johannes Bechert, who, in spite of serious illness, commented in valuable detail on an earlier draft of part of this chapter.

206 2.

3.

4. 5.

6.

7.

8. 9.

Jim Miller This concept of theme is compatible with the examples in Kiss (1998: 682) in which she demonstrates the necessity to distinguish between her topic and grammatical subject, a central point also made by Halliday (1967). Kiss provides Hungarian examples in which nouns in nominative, accusative, or dative case are in theme position. Halliday (1967) provides English examples in which subject or direct object noun phrases, or prepositional phrases, are in theme position. Why a particular constituent is in theme position in a particular clause or sentence is the interesting question. The theme constituent might be given and act as a bridge between the current chunk of text and previous text, or it might simply be made salient. The map task was first used by Brown et al. (1984) to elicit spoken language from school pupils. It has subsequently been used in a number of investigations, including the one drawn on here; 32 dialogues (out of a total of 128) produced by university undergraduates beginning their first year of study and coded for syntax, information structure, game structure, and certain phonetic phenomena such as assimilation and reduction. The dialogues were collected and analyzed in the Human Communication Research Centre at the University of Edinburgh. This is a problem noted by Dahl with respect to his tense-aspect questionnaire. See Dahl (1985). Our method of data collection has met with objections. John Sinclair (pers. comm., Conference on Spoken and Written Language, Prague, 1992) found the data unnatural in that it was collected under experimental conditions. This is a valid comment, to which the answer is that while the data may not be produced in natural conditions, the experimental control enables certain factors to be held constant over all the map task dialogues, which in turn enables reliable comparisons to be made. The method represents a trade-off between complete naturalness and experimental rigor which has yielded worthwhile results. The objection also loses some of its force because of the other types of data that have been employed in the preparation of this paper. The differences, as proposed in Chapter 1 of this volume and by many other linguists, come partly from the general properties of spoken as opposed to written language (no editing time, problems of short term memory, cooperation between participants in a dialogue) and partly from properties of the different mediums; spoken language has pitch, amplitude, and rhythm and written language does not. Equally, the properties of written language flow from the general properties of written language (time to plan, to read and reread texts, and to emend texts) and from properties of the medium (no pitch, amplitude, and rhythm). “Particles” here are not focus particles such as only or even but particles expressing impatience or surprise, particles signaling that the speaker is about to issue an instruction, particles signaling that a particular constituent is important. The concept of focus used by Kiss (1998: 707) appears to be exactly the concept developed by Chafe. One further point, made by Chafe (1976), is that contrast is not necessarily explicit, as in Dik’s examples in (3). The speaker or writer of a text that begins It was in 1966 that I first went to Moscow makes 1966 salient and implicitly contrasts it with all the other years that the audience might have in mind.

Focus in the languages of Europe 10.

11. 12.

13.

14.

15.

16.

For Grosz and Sidner focus spaces include representations of the discourse segment purpose, since they consider conversational participants to be focused not only on what they are talking about, but also on why they are talking about it. Cf. the discussion in Quirk et al. (1985: Ch. 18) and Givón (1990: Ch. 17). Of course, speaker A may have got things wrong. B can reply Oh yes – I’d forgotten about her, thereby indicating that he does indeed have a file card for the girl, or he can reply What girl two doors down?, which tells A that he has made incorrect assumptions about B’s knowledge store. Complexity is relative. Given that in spontaneous spoken English (in our map task dialogues and conversations, and in the data examined by Givón) approximately 65 % of NPs consist of pronouns, and very few NPs have more than one adjective, the second detached NP in A3 in (10) is indeed complex. Blanche-Benveniste (1991) exemplifies a range of uses of the existential construction: il y avait mon pote qui était accordéoniste ‘there was my pal who was accordeonist = my pal was on the accordeon’, il n’y a que la nuit qu’il dort ‘there is only the night that he sleeps = he only sleeps at night’. The translations of the passages from Kaledin are taken from the English version by Catriona Kelly. The value of this translation is that is was done by someone with no knowledge of this typological study. Barbara Wehr (not a native speaker but a specialist in Irish as well as Romance) further comments that she finds the following example quite natural. (i)

17. 18. 19. 20.

21.

207

An leabhar-seo a thug mé dhuit a bhfuil sí léite the book-this REL gave I to-you PTCL is it read agat faoi seo? by-you by now ‘The book I gave you – have you read it yet?’

Clearly further investigation is needed. This translation, too, is Kelly’s. We have no video recording of the conversation, but from other video material we can assume that vot tak vot is accompanied by physical gestures. The device has considerable rhetorical power but is easily overused, witness Mr. Chadband in Bleak House. This hypothesis was initially formulated with respect to Czech and then other IndoEuropean languages. Mithun (1987) argued that the given-precedes-new property does not apply to certain Amerindian languages. While there may be much truth in her claim, the problem of genre bedevils the discussion yet again. Mithun looks at spontaneous spoken language, but the earlier work on Czech, English, and other languages of Europe dealt with written work. See the comments in Miller (1995). The examples in (56) make it clear that the label “longer NP” is to be taken very much in a relative sense. Longer NPs contain one adjective and a noun, or a demonstrative and a noun, or a pronoun and a prepositional phrase. The bulk of the NPs consist of a pronoun or a single noun. They are very different from the NPs in the corpora exam-

208

22.

Jim Miller ined by Siewierska and Bivon but are quite typical of spontaneous spoken English and German (see Miller and Weinert 1998). With respect to Hawkins’ processing principles, the question arises as to how big a role the principles play in spontaneous spoken language where most NPs are only one constituent long. Another question is how important the principles are with respect to written texts, which can be quickly scanned by the reader and then reread for the details. Barbara Wehr (pers. comm.) points out that in Italian, highlighted noun phrases can appear in clause-final position. (i)

23.

24.

25.

26.

27.

For a general account of ellipsis, including criteria for recognizing ellipsis and numerous examples of various ellipted constituents, see Quirk et al. (1985: 883–913) and Matthews (1981: 38–49). These examples are all fine, but it is only in the final paragraph on categories of ellipsis, indeed in the section on structural ellipsis, that Quirk et al. (1985: 900) state that situational ellipsis is characteristic of familiar spoken English. The main thrust of the statement is that structural ellipsis is characteristic of written English. Note the translation of razmundirivajus’ as ‘I’m taking it off’ does not indicate a zero direct object. The verb is reflexive, -s’ being the reflexive suffix, and a gloss that conveys better the force of the Russian verb is ‘I am divesting myself (of my coat)’. In Quirk et al.’s (1985) terms these are cases of quasi-ellipsis, since the antecedent may be a subject or oblique object and therefore in a different case from the missing direct object pronoun, which would be in the accusative case. The Turkish construction, with its distinction between marked and unmarked word orders, is more complex than Mayan, another language with a position for focused constituents. The examples cited by Aissen (1992) all have to do with contrast, as in the following example. (i)

28.

Non trovi anche tu che Mario è bello? not find also you that Mario is handsome ‘Don’t you think also that Mario is handsome?’ Non lo trovo tanto bello– è più bello gino not him I-find so handsome is more handsome Gino ‘I don’t find him so handsome. gino’s more handsome.’

Something had landed at the foot of the tree, they went to look. There was a straw mat. “Hell, what could it be? Come on, let’s untie the straw mat!” the two men said to each other. They untied it. You know what? – Tzeb san-antrex la te staik un. girl San Andres clitic there they-found clitic ‘It was a girl from San Andres they found there.’

Papp (1994) observes that the most frequent Hungarian corrective sentence contains only one constituent. (i)

A:

Ki törte el a vázát? Péter? who broke 3SG the vase Peter

Focus in the languages of Europe

209

B:

29. 30. 31. 32. 33.

34.

35.

Nem, János no, John (It was John) A number of other orders are possible, depending on the interpretation of the nonfocal elements. For present purposes, these other orders are irrelevant. Data from Wehr (1991). In spoken Irish, the initial is is regularly elided, leaving, for example, SEAN a chuaigh go Corcaigh inné. Or -sean, depending on the phonological context. Givón (1979: 246–248) observes that the order of constituents in clefts is peculiar, with new and focused material preceding presupposed material. One would expect the converse order in the light of the principle whereby given material precedes new. (This order does not hold in American Indian languages.) Givón explains the order in historical terms, the cleft allegedly arising out of a copula construction followed by afterthought relative clause. This pragmatic discourse structure is subsequently grammaticalized into the type of integrated structure that is called an IT-cleft. Givón cites the following examples from a Bantu language called Kihung’an. (He also gives other examples which, he claims, illustrate the steps in the process from stretch of discourse to grammatical construction.) (i)

Kwe kit, kiim ki ka-swiimin Kipes zoon. be chair thing that he-bought Kipes yesterday ‘It’s a chair, the thing that Kipes bought yesterday.’

(ii)

Kwe kit Kipes ka-swiimin zoon. be chair Kipes he-bought yesterday (= [i])

These examples also demonstrate that outside Europe there are languages with cleft constructions. Quirk et al. (1985) point to the fact that whom and which are only marginally possible in clefts, and virtually impossible when preceded by a preposition. But these items are virtually absent from relative clauses in spontaneous spoken English. They observe that that can be omitted as the subject of a cleft clause but not as the subject of a relative clause: It was the President himself spoke to me vs. *The dog growled at me didn’t have a muzzle (= The dog that growled at me.). But the IT-cleft can be seen as having an existential-presentational strand, and other constructions of this type allow that to be omitted. Quirk et al. cite There’s a man lives in China but asterisk *I know a man lives in China. However, the latter construction is regular and normal in spontaneous spoken English, along with We had a French girl spent a month in our class (= We had a French girl that/who : : :) It is true, as Quirk et al. say, that relative clauses do not modify prepositional phrases and adverbial clauses – cf. It was in September that I first noticed it and It was because he was ill that we decided to return, but the distance between relative clauses and cleft clauses is not as great as they suppose in spontaneous spoken English. The rationale behind the use of WH-words in interrogatives can be reconstructed as the speaker saying, as it were “Someone is going to drive the car – specify the person.” The Russian equivalents of the English WH-words are instructive. Kto corresponds to who

210

36.

37.

38.

39.

40.

41.

Jim Miller in interrogatives but to anyone in conditional clauses: esli kto pridet = if anyone comes. To kto can be added the deictic to (= that ), yielding kto-to. A speaker saying kto-to zvonil (= Someone phoned) is conveying that he or she can specify the someone, even in vague terms such as It sounded like a woman of thirty or so. The addition of another affix -nibud’, as in kto-nibud’ yields a vaguer deictic: kto-nibud’ ukral mne portfel’ (= Someone has stolen my briefcase). The speaker cannot specify any properties of the someone. Reverse WH-clefts with this are more common in contexts such as guided tours. When the guides lead visitors from spot to spot on the tour, they say, for example, This is where the Archbishop was stabbed to death, This is where the cantankerous Corgi met the belligerent Borzoi. The SOV order of Turkish is apparently not a factor in the lack of clefts. Givón (1979: 248) cites Amharic examples furnished by Hetzron as demonstrating that Amharic, an SOV language, has clefts. The example of a cleft is: (i)

Maryam-in ya-gaddal-at yohannis naw. Mary-DO REL-killed-her John is ‘Who (the one who) killed Mary is John.’

(ii)

Yohannis naw (,), maryam-in ya-gaddala-w. John is Mary-DO REL-killed-DEF ‘It’s John the one who killed Mary/It’s John who killed Mary.’

The second example, as Givón comments, is not an integrated cleft construction but a stretch of discourse that could give rise to one. I¸ste appears to resemble vot in the Russian examples in (26) and (49)–(51). I¸ste is glossed by Lewis (1967: 217) as ‘behold!’ or ‘there!’ or ‘precisely!’. Vot is a powerful deictic, many of whose uses can be glossed with ‘behold!’. Speakers who comments nu vot after listening to a contribution to a conversation can be understood, in some contexts, as producing the Russian equivalent of ‘precisely!’. There are problems, discussed in some detail in Sornicola (1988: 348–356), with the agreement in number and person of the verbs in these constructions. The cleft constituent is also restricted in Spanish. Sornicola observes that excluded from the status of cleft constituent in IT-clefts are adverbial phrases, prepositional phrases, and temporal clauses. Weinert observes that such WH-clefts are rare in the German dialogues; in particular there are no examples with the German equivalent of do or happen. She suggests that the discourse function of such English WH-clefts attaches to German constructions with so. The informant who supplied German data via the oral questionnaire insisted that German speakers do not use IT-clefts or WH-clefts. The former claim is correct and the latter claim is correct insofar as WH-clefts are rare. There is a range of other constructions such as (i)

und das was inzwischen entdeckt worden ist ist X. and that what meanwhile discovered been has is X ‘and what has been discovered meanwhile is X.’

Focus in the languages of Europe

42.

43.

211

or even das was : : : das ist ‘that what : : : that is’ and was : : : das ist ‘what : : : that is’. The distribution and function of these constructions in the relevant databases has yet to be investigated. The original English map has East Lake as a landmark. The German map had Ostsee, with no article. Speaker A has interpreted it as die Ostsee ‘the Baltic’ instead of der Ostsee ‘the East Lake’. According to Enric Vallduví (pers. comm.) aixo in (169) has the same highlighted or contrastive feeling that topicalized elements have in English.

References Aissen, Judith L. 1992 Topic and focus in Mayan. Language 68: 43–80. Atanasova, T., M. Rankova, R. Rusev, D. Spasov, V. Filipov, and G. Chakalov 1975 Bulgarsko–Anglijski Rechnik. Sofia: Izdatelstvo Nauka i Izkustvo. Barnes, Betsy K. 1985 The Pragmatics of Left Detachment in Spoken Standard French. Amsterdam: John Benjamins. Bivon, Roy 1971 Element Order. Cambridge: Cambridge University Press. Blanche-Benveniste, Claire 1991 Le français parlé: études grammaticales. Paris: Éditions du Centre National de la Recherche Scientifique. Bossong, Georg 1981 Séquence et višée. Folia Linguistica XV: 237–252. Brown, Gillian, and George Yule 1983 Discourse Analysis. Cambridge: Cambridge University Press. Brown, Gillian, Anne Anderson, Richard Shillcock, and George Yule 1984 Teaching Talk. Cambridge: Cambridge University Press. Carlson, Lauri 1993 Dialogue games with Finnish clitics. In The 1993 Yearbook of the Linguistic Association of Finland, S. Shore and M. Vilkuna (eds.), 73–96. Helsinki: The Linguistic Association of Finland. Chafe, W. L. 1976 Givenness, contrastiveness, definiteness, subjects, topics and points of view. In Subject and Topic, Charles N. Li (ed.), 25–55. New York: Academic Press. Dahl, Östen 1985 Tense and Aspect Systems. Oxford: Blackwell. Delin, J. L. 1990 Accounting for cleft constructions in discourse: a multi-layered approach. Research paper HCRC/RP-5, University of Edinburgh, HCRC Publications. Denniston, J. D. 1934 The Greek Particles. Oxford: Clarendon Press. Dik, Simon C. 1980 Studies in Functional Grammar. London: Academic Press. Erguvanli, Eser 1984 The Function of Word Order in Turkish. Berkeley: University of California Press.

212

Jim Miller

Fernandez, M. M. Jocelyne 1994 Les particules énonciatives. Paris: Presses Universitaires de France. Gillies, William 1993 Scottish Gaelic. In The Celtic Languages, Martin Ball and James Fife (eds.), 145–227. London: Routledge. Givón, Talmy 1979 On Understanding Grammar. New York: Academic Press. 1990 Syntax. A Functional-typological Introduction II. Amsterdam: John Benjamins. Givón, Talmy (ed.) 1983 Topic Continuity in Discourse: Quantitative Cross-language Studies. Amsterdam: John Benjamins. Grosz, Barbara J. 1981 Focusing and description in natural language dialogues. In Elements of Discourse Understanding, Aravind Joshi, Bonnie Webber, and Ivan Sag (eds.), 84–105. Cambridge: Cambridge University Press. Grosz, Barbara J., and Candace L. Sidner 1986 Attention, intentions and the structure of discourse. Computational Linguistics 12: 175– 204. 1990 Plans for discourse. In Intentions in Communication, P. R. Cohen, J. Morgan, and M. E. Pollack (eds.), 417–444. Boston: MIT Press. Hakulinen, Auli 1976 Liitepartikkelin -han/hän syntaksia ja pragmatiikkaa. In Reports on Text Linguistics: Suomen kielen generatiivista lauseoppia 2. Åbo: Åbo Akademi Forskningsinstitut. Hakulinen, Auli, and Fred Karlsson 1979 Nykysuomen lauseoppia. Helsinki: SKS. Halliday, Michael A. K. 1967 Notes on transitivity and theme in English, Parts 1 and 2. Journal of Linguistics 3: 37– 81, 199–244. 1968 Notes on transitivity and theme in English, Part 3. Journal of Linguistics 4: 179–215. 1985 An Introduction to Functional Grammar. London: Edward Arnold. Hanks, William F. 1996 Language and Communicative Practices. Boulder, CO: Westview Press. Hawkins, John A. 1990 A parsing theory of word order universals. Linguistic Inquiry 21: 223–261. Hedberg, Nancy 1990 Discourse pragmatics and cleft sentences in English. Minneapolis: University of Minnesota PhD thesis. Heim, Irene 1982 The semantics of definite and indefinite noun phrases. Unpublished doctoral dissertation. Amherst: University of Massachusetts. Kálmán, L., G. Proszeky, A. Nadasdy, and C. Gy. Kálmán 1984 Hocus, focus, and verb types in Hungarian infinitive constructions. In Topic, Focus and Configurationality, W. Abraham and S. de Meij (eds.), 129–142. (Linguistik Aktuell Band 4.) Amsterdam: John Benjamins. Kapanadze, L. A., and E. A. Zemskaja 1979 Teksty. Moscow: Nauka. Kiss, Katalin É. 1987 Configurationality in Hungarian. Budapest: Akadémiai Kiadó/Dordrecht: D. Reidel. 1998 Discourse-configurationality in the languages of Europe. In Constituent Order in the Languages of Europe, A. Siewierska (ed.), 681–727. Berlin/New York: Mouton de Gruyter.

Focus in the languages of Europe

213

Kovtunova, I. I. 1969 Porjadok slov v russkom literaturnom jazyke XVIII-pervoj treti XIX v. Moscow: Nauka. Lapteva, O. A. 1976 Russkij razgovornyj sintaksis. Moscow: Nauka. Levy, D. M. 1979 Communicative goals and strategies. In Discourse and Syntax: Syntax and Semantics 12, T. Givón (ed.), 183–210. New York: Academic Press. Lewis, Geoffrey L. 1967 Turkish Grammar. Oxford: Oxford University Press. Li, Charles N., and Sandra A. Thompson 1976 Subject and topic: a new typology of language. In Subject and Topic, C. N. Li (ed.), 459–489. New York: Academic Press. Mackenzie, J. L., and M. E. Keizer 1990 On assigning pragmatic functions in English. In Pragmatic Functions: The View from the V. U. Working Papers in Functional Grammar 38, M. Hannay, J. L. Mackenzie, and M. E. Keizer (eds.), 1–33. Amsterdam: Free University of Amsterdam. Matthews, Peter 1981 Syntax. Cambridge: Cambridge University Press. Miller, Jim 1995 VS clauses in Russian: contrast, new episodes and theticity. Sprachtypologie und Universalienforschung 48: 125–141. Miller, Jim, and Regina Weinert 1995 The function of LIKE in dialogue. Journal of Pragmatics 23: 365–393. 1998 Spontaneous Spoken Language. Syntax and Discourse. Oxford: Clarendon Press. Mithun, Marianne 1987 Is basic word order universal? In Coherence and Grounding in Discourse, Russell S. Tomlin (ed.), 281–328. Amsterdam: John Benjamins. Palmer, L. R. 1962 The language of Homer. In A Companion to Homer, Alan J. B. Wace and Frank H. Stubbings (eds.), 75–178. London: Macmillan. Papp, Szilvia 1994 English and Hungarian focus constructions. Unpublished MSc thesis, Applied Linguistics, University of Edinburgh. Prince, Ellen 1981 Towards a taxonomy of given-new information. In Radical Pragmatics, Peter Cole (ed.), 223–256. New York: Academic Press. Quirk, Randolph, Sidney Greenbaum, Geoffrey Leech, and Jan Svartvik 1985 A Comprehensive Grammar of the English Language. London: Longman. Siewierska, Anna 1993 Syntactic weight vs. information structure and word order. Journal of Linguistics 29: 233–265. Sornicola, Rosanna 1988 It-clefts and Wh-clefts: two awkward sentence types. Journal of Linguistics 24: 343– 379. Vallduví, Enric 1993 Information packaging: a survey. Research Paper HCRC/RP, University of Edinburgh, Human Communication Research Centre. 1994 Catalan right-detachment and information packaging. Journal of Pragmatics 22: 573– 601. Wehr, Barbara 1991 Focus in Irish. Paper presented at the meeting of Eurotype Theme Group 1, Bremen.

214 1994

Jim Miller Topic und Focus-Konstruktionen im Französischen. In Sprachlicher Alltag. Festschrift für Wolf-Dieter Stempel, Annette Sabban and Christian Schmitt (eds.), 611–633. Tübingen: Max Niemeyer.

Weinert, Regina 1995 Focusing constructions in spoken language: clefts, Y-movement, thematization and deixis in English and German. Linguistische Berichte 159: 341–369. Weinert, Regina, and Jim Miller 1996 Cleft constructions in spoken language. Journal of Pragmatics 25: 173–202. Zemskaja, E. A 1973 Russkaja Razgovornaja Reˇc’. Moscow: Nauka.

Yaron Matras and A. Machtelt Bolkestein

Deixis and anaphora: Some case studies

1. Introduction The basic pragmatic distinction between “topical” and “focal,” or “thematic” and “rhematic,” parts of the message conveyed by a sentence-in-context, has, whatever terminology or details of definition are used, and whatever the problems with subcategorizing and operationalizing the discourse notions involved (for a survey, see Lambrecht 1994; Bolkestein 1996), undeniably proved its relevance for the explanation of various linguistic phenomena in many languages, such as prosodic contour, constituent order, the use of special constructions, and morphological marking (see, for example, Myhill 1992). One of the notions frequently cropping up in discussions concerning topicality or thematicity is that it is useful to distinguish different degrees of, or different causes of, “accessibility” of some referent for the addressee of the message (in the estimate of the speaker). Another formulation of basically the same insight is in terms of different degrees of “activation” of a referent at a particular point in the discourse: highly activated referents are easily accessible, semi-activated referents are less accessible and their identification requires more effort, and so on (see, e.g., Chafe 1994). There is supposedly some relation between degrees of activatedness or accessibility on the one hand, and the way in which referents are expressed on the other. While this insight is generally accepted in discourse studies, there is less agreement about which factors determine or influence such degrees of accessibility. In some approaches, degrees of accessibility are differentiated in terms of rather unproblematically applicable quantitative procedures, such as measuring linear distance between a referent and its antecedent in terms of the number of words or clauses or the number of competing referents (see Givón 1983 and other studies inspired by his approach). Other approaches stress the importance of semantic frame based processes resulting in inferability and evokedness of referents (see Prince 1981; Hannay 1985). Other approaches point to the relevance of a range of notions, from factors connected with hierarchical discourse structure (Fox 1987; Bolkestein 2001) or with other cognitive principles (Ariel 1990, 1991, 1994; Gundel et al. 1993), the degree of prominence of an entity, its relative importance in the discourse as a whole, and at the moment of introduction, the way in which it has been introduced

216

Yaron Matras and A. Machtelt Bolkestein

and its syntactic status (Cornish 1986; Berretta 1990; Bolkestein and Van de Grift 1994; Bolkestein 1996). All of these heterogeneous factors may influence how accessible some element will be later on in the discourse, or how long its accessibility will last, and, consequently, what expressions have to be used or tend to be preferred for referring to it. One of the strongest arguments in favor of making a distinction between referents of which the identity is somehow retrievable from the preceding discourse (or at least treated as such by the speaker) vs. referents which are not easily identifiable (or at least not so treated), is the fact that languages commonly have specific expression means to refer back to elements which have been mentioned earlier in the discourse and which are consequently sufficiently accessible not to require full noun phrase expression: demonstrative and other kinds of pronouns. The pronouns involved in extra- and intratextual deixis and anaphora often can be used both pro-nominally, that is, in the function of a noun phrase, and “pro-adjectivally,” that is, as determiners of a noun phrase (if so, as a rule this is possible for both their deictic and their anaphoric use). In the present chapter we will discuss the pronominal system of two different types of (Indo-)European languages (Classical Latin and modern Dutch) and two marginal European languages ( Romani, which is of Indo-Aryan origin, and Israeli Hebrew, a Semitic language), with special attention to the way in which the available pronouns are used in discourse for referring to elements of the preceding context, that is, outside their own sentence. The languages selected have quite different properties (such as rather rigid word order vs. flexible word order, or the presence vs. absence of case marking) and at the same time offer the additional advantage of the availability of sufficiently detailed data on discourse phenomena.

2.

Extratextual deixis, intratextual deixis and anaphora

When demonstrative pronouns are used for referring or pointing to some entity in the (extralinguistic) spatial and temporal situation, we speak of extratextual or situational deixis. Well known and common parameters for describing systems of deictic pronouns are spatial distance or proximity to the speaker or addressee, or to some third party. Other parameters may be (in)visibility of referents. Languages differ in how many of such distinctions are expressed in the deictic system (see Anderson and Keenan 1985 for a typological survey). Situational deixis is (usually) clearly distinguishable from referring to some element of the preceding discourse. Reference to elements of the discourse itself will be called intratextual deixis. Some linguists make a basic distinction between intratextual deixis (“pointing” or “putting entities in the universe of discourse,” to use Lyons’ [1977: 673] words) on the one hand, and anaphora (“presupposes that the referent

Deixis and anaphora: Some case studies

217

should already have its place in the universe-of-discourse”) on the other, and assume a close relation between the use of demonstratives and intratextual deixis and that of other expression means, such as zero or weak pronouns, or personal pronouns, and anaphora. Such a basic distinction is suggested by Lyons (1977: 668 f.), who, however, admits that it may be blurred in certain languages; see also Ehlich (1982), De Jong (1996), and Himmelmann (1996) for a critical note. For empirical investigation, it seems best to leave open the question whether any particular means of referring intratextually (such as demonstrative pronouns) perhaps fulfills a basically different function from that fulfilled by zero anaphora or by specific pronouns which cannot refer extratextually. If there is a clear-cut difference between performing intratextual deixis on the one hand and anaphora on the other, this will have to manifest itself in a thorough examination of the way in which the various expression means are actually employed in a specific language (see, for instance, the section on Romani below). For the moment, therefore, it seems best to look at the two notions as forming a continuum rather than a sharp distinction between basically different functions or actions of the speaker. Intratextual reference, including anaphora, may concern two semantically different types of entities. It may concern either items which are explicit constituents of some preceding clause (commonly first or second order entities: animate and inanimate participants, events, and so on), or the reference may be to the content of a preceding clause or of a larger discourse segment as a whole (that is, to second, third, or even higher order entities, such as predications, propositions, paragraphs and episodes, etc.). Such reference is sometimes alluded to as “indirect anaphora.”1 This semantic distinction between the types of entity referred back to may be relevant for describing the division of labor of pronominal expressions in a language, since languages may consistently select different pronominal expressions for the two types. Languages also differ in the range of expression means available for intratextual deixis and anaphora, and in the variability which they allow given the range of available expressions. Some languages, such as Classical Latin (see Section 3), may just leave out explicit mention of the element concerned altogether (zero expression) in certain circumstances, or they may possess specific pronouns especially reserved for reference to the most highly accessible entities, distinct from the demonstrative pronouns which may also serve an extratextual deictic function. This is what many linguists would call anaphoric pronouns in the strict sense, but they are sometimes alluded to by other terms, such as determinative pronouns. Even languages that do possess specific nondemonstrative pronouns reserved for anaphora in the strict sense may still employ demonstrative pronouns too for reference to elements which have been previously mentioned in the discourse and which therefore seem highly accessible. When a language exhibits this kind of choice between two classes of pronominal expressions, obviously a functional motivation may be involved, although it need not

218

Yaron Matras and A. Machtelt Bolkestein

be. When a language does not have specific anaphoric pronouns, there may still be a choice between different members of one class of demonstratives, for example the distal and the proximate one, as is the case in English or Dutch (see Section 4); again this choice may or may not be functionally motivated. In grammars (which frequently do not pay much attention to discourse phenomena) it is often explicitly or tacitly assumed that there is a rather straightforward relation between extratextual deictic use and the functions of the same class of expressions when used in monologue discourse, and that the latter is somehow derivable from the former, with some sort of metaphorical, sometimes rather farfetched extension of the parameters involved in situational deixis. Thus we find the assumption that distal pronouns are used when there is greater psychological or cognitive distance, etc. Upon closer observation, however, the way in which members of the class of demonstrative pronouns function in for example narrative discourse is not in all languages easily predictable from their individual situational functions, and deserves to be examined for each language separately. A survey of the frequency of distribution of the available pronominal expressions over pro-nominal and pro-adjectival anaphoric reference to “entities” (sentence constituents) and “discourse segments” (the content of clauses or paragraphs) in samples from different languages is given in Himmelmann (1996). Himmelmann distinguishes a third usage of demonstratives which he found to be common, but which has not received much attention. He labels this use “recognitional.” This use cannot be called anaphoric in the strict sense, nor can it be called deictic, since the pronoun neither refers back to an antecedent in the actual discourse context, nor does it point to an entity in space or time. Instead it refers to an entity which the speaker assumes the addressee can identify out of the blue on the basis of their shared knowledge (one could argue here that it is a form of situational deixis, in that it locates or points to a referent in some shared cognitive world). Himmelmann notes that in his sample, such “recognitional use” of a pronoun is exclusively fulfilled by pronouns which, when used for situational deixis, have a distal function. Below we will discuss an instance of this as a special case of intratextual deixis. While the speaker in performing extratextual deixis and intratextual deixis may be engaged in basically different types of activities, in actual discourse it is not always crystal clear whether we are dealing with the one or the other kind of reference. For example, in the case of monological sequences in which two referents are contrasted, as in the case of “the first mentioned” vs. “the last mentioned” item, the parameter “distance/proximity,” relevant in situational deixis, may lead to selection of a proximate pronoun for the latter and a distal pronoun for the former: this might be interpreted as an argument in favor of classifying such use as being in some sense situational (some would say “deictic” rather than “anaphoric”): one might maintain that the speaker has turned the preceding context into a component of the speech situation, parts of which are further away than others. However, the

Deixis and anaphora: Some case studies

219

majority of demonstratives in monological discourse are not explicitly contrastive in such a way, but are simply coreferent to some previously mentioned antecedent. The nature of the relation between their use in spatial deixis and in monologue discourse need not be straightforward at all, in spite of what grammars often assume, and is a matter for further investigation. The problem can be illustrated with the use of the Latin pronoun ille, which when used deictically refers to entities which are remote from both the speaker and the hearer. In intratextual reference, it is typically used to refer back to an antecedent which itself does not yet have the status of agent-subject. However, it is also the typical expression for “recognitional” use, and may in that case be paraphrased as ‘the/our well-known X’. While it is intuitively plausible to interpret the first usage as involving not linear, but mental or cognitive distance (a non-agent non-subject is perhaps semantically and syntactically less prominent than an agent-subject), it is rather hard to see the second use as involving mental distance, or if it does, it is a rather different kind of mental distance from the first. In the present chapter, we will discuss the systems of deictic and anaphoric pronouns of some European languages for extratextual and for intratextual reference, including anaphora in the strict sense, with special attention to how the available pronominal expressions from both domains function in actual discourse in each language. Our main attention will be given to intersentential (or: long distance) anaphora rather than intrasentential coreference, for which the rules may be partially purely syntactic. For the languages that we will discuss, we will give a survey of the expressions used in extratextual deixis, and of those used in intratextual deixis and anaphora (even if the set of expressions totally overlaps, which it normally will not). The two sets of factors which distinguish the extra- and the intratextual uses do not overlap: for the extratextual all (possibly more?) of the distinctions in Anderson and Keenan (1985) (distance/proximity, visibility, whatever) may in principle be relevant, whereas for the intratextual, both semantic type of referent (participant vs. text segment) and notions such as “degree of accessibility”: highly- or semi-active because of various reasons such as (i) (linear or hierarchical) distance, (ii) syntactic status of antecedent, (iii) the presence of competing referents, (iv) the information status of antecedent, (v) emphasis of identity of referent for whatever reason, (vi) whatever turns out to be relevant for the language in question. Note that in the case of extratextual deixis, the parameters are mutually exclusive. In the case of intratextual reference, on the other hand, this is often not the case. Moreover, in a certain contextual constellation, a particular means of expression may be preferred rather than obligatory, and more than one expression may be acceptable. Furthermore, languages may employ other means of expression deictically in certain cases, namely personal pronouns; and both personal pronouns and definite NPs (in languages which possess a definite article) may be used anaphorically as well, for referents which have been mentioned in the preceding discourse. Although the question of what determines whether the speaker uses one of these expressions rather than an anaphoric

220

Yaron Matras and A. Machtelt Bolkestein

or deictic pronoun is undoubtedly interesting, only occasional attention will be paid to personal pronouns and definiteness (cf. the section on Dutch).

3.

Extratextual deixis, intratextual deixis, and anaphora in Classical Latin

Classical Latin is a flexible word order language in which the linear order of constituents is sensitive but not easily predictable from their informational status. Constituents of which the identity is known from the context or situation may easily be left unexpressed. This phenomenon is not limited to syntactic subjects. Latin has no definite article.

3.1. Extratextual deixis The system of spatial deixis in Classical Latin is quite straightforward and well described in the handbooks (e.g., Kuehner and Stegmann 1912–1914; Hofmann and Szantyr 1965; and Touratier 1994). Apart from first and second person personal pronouns, which may perform one type of deixis, there are various pronominal expressions for third person referents. The first and second person personal pronouns refer to the speaker (ego ‘I’) or to a group of more than one person including the speaker (nos ‘we’, used both when including and excluding the addressee) and to the addressee (tu ‘you sg.’) or a group of more than one person including the addressee (vos ‘you pl.’) respectively. These personal pronouns do not convey information about gender, nor about social rank or relationship between the speaker and the addressee. When used in subject function, they are emphatic (see Pinkster 1987): as a rule, in Latin nonemphatic first, second, and third person, subjects are not expressed by an explicit pronoun, but identifiable by the verb inflection which conveys information about person and number. There are no weak or strong forms of these pronouns comparable to the difference between, for example, the strong forms jij/mij/zij ‘you/me/she or they’ and the weak forms je/me/ze ‘you/me/she or they’, in Dutch. For spatial deixis to third persons, Latin has a three-term system, like Turkish, Japanese, and Spanish (see Anderson and Keenan 1985: 282): the three demonstrative pronouns involved are hic (haec/hoc), iste (ista/istud), and ille (illa/illud). These pronouns may be used both pro-nominally and pro-adjectivally for both singular and plural entities.2 They convey information as to semantic or grammatical gender of the referent. There is no separate “impersonal” pronoun referring to nonspecified third person(s) comparable to Dutch men, German man, or French on (‘one’ or unspecific ‘they’). For unspecified third person(s) Latin may use third person plural inflection on the verb, or third person singular passive voice inflection, as in eunt ‘go:3PL’ or

Deixis and anaphora: Some case studies

221

itur ‘go:PASS:3SG’. There are also various indefinite pronouns which will be left out of account, since they do not function for spatial deixis or anaphorically. The demonstrative pronouns are specified for location with respect to speaker and addressee, that is, they are person-oriented rather than distance-oriented in Anderson and Keenan’s (1985) terms. The pronoun hic is used for referring to entities in the vicinity of the speaker (and as such, when pro-adjectival is sometimes used as an alternative to the first person possessive adjective meus ‘my’). The pronoun iste is used for referring to an entity in the vicinity of the addressee, and ille refers to entities which are remote, that is, neither close to the speaker nor to the addressee. Latin has one anaphoric pronoun, also called determinative in the grammars, namely is (ea/id). This is sometimes put on a par with first and second person personal pronouns as a “third person” pronoun. However, is does not take part in the system of person-oriented spatial deixis, and in fact does not refer spatially at all. It will be discussed below in connection with intratextual deixis and anaphora. The same holds for the possibility in Latin to connect clauses by means of the use of the so-called “free relative connection”, with a relative pronoun qui (quae/quod) in sentence-initial position.

3.2. The system of intratextual deixis and anaphora in Latin As far as intrasentential coreference is concerned, in the case of coreferents which fulfill the same syntactic function in two coordinated clauses or a main and subordinate clause of a complex sentence (that is, which are both subject or both object), one of them will normally remain unexpressed (either the second one or the one occurring in a subordinate clause).3 Under certain circumstances, a pronoun (usually is) occurs. A special case is the so-called “theme construction”: in this construction, a referent is introduced extraclausally, and resumed by a pronoun (again usually is) in the clause following it, as in (1): (1)

Cancer ater, is olet tumor:NOM.SG black:NOM.SG 3:SG.NOM smell:3SG ‘a black tumor, that smells’ (Cato Agr. 157.3)

Intrasentential anaphora will not be discussed in this chapter, nor will attention be paid to the conditions for the occurrence of reflexive pronouns (for further discussion of the theme construction in Latin, see Hoffmann 1989; Somers 1994). As far as intersentential coreference (or: long distance anaphora) is concerned, entities which have been mentioned before in the discourse may be referred to by a variety of expression means. No expression at all (“zero expression”) is common for referents which are fully accessible, not only when we are dealing with syntactic subjects. Understandably, such zero expression is most common in the case of subjects,

222

Yaron Matras and A. Machtelt Bolkestein

probably because for the properties of subjects, the verb ending provides information as to number and person anyway. However, it is by no means obligatory for subjects which have antecedents in the preceding clause, nor is zero expression limited to subjects: it also occurs when we are dealing with other main clause constituents, such as objects and indirect objects, and even with NP-level possessors. This phenomenon is often assumed to be characteristic of a looser, less careful style, but this claim deserves further investigation. An instance illustrating a number of zero anaphors with different syntactic status is (2) (in the free translation, we indicate zero anaphora by putting brackets around the referent): (2)

et tamen omnibus hominibus contra serpentes and yet all:DAT.PL man:DAT.PL against snake:ACC.PL inest venenum: ferunt ictum salivae ut be.in:3SG poison:NOM.SG say:3PL hit:ACC.SG spittle:GEN.SG as ferventis aquae contactum fugere; boiling:GEN.SG water:GEN.SG touch:ACC.SG flee:INF quod si in fauces penetraverit etiam mori, which:NOM.SG if in throat:ACC.PL penetrate:3SG too die:INF idque maxime humani ieiuni that:ACC.SG-and specially human:GEN.SG sober:GEN.SG oris mouth:GEN.SG ‘and yet all men have in them a poison against snakes: (they) say that (they) flee the touch of spittle (of men) as (if it were the touch) of boiling water; and if (it) has penetrated (their) throat (they) die, and especially (by spittle) of a sober mouth’ (Plin. Nat. 7.15)

Options other than just leaving the referent unexpressed are the three demonstrative pronouns mentioned above, and the nondemonstrative, anaphoric pronoun is. Furthermore, Latin allows a loose form of “relativization” known as the “free relative connection”: in this construction, a constituent of the preceding clause, or of an even earlier one, or some text segment, is referred to in a new (subordinate or main) clause by means of the relative pronoun qui (quae/quod). This latter pronoun can never carry emphasis. This type of construction is often alluded to as more or less equivalent to et/sed/nam is (‘and/but/for he’). It might be viewed as a manifestation of a nonrestrictive relative clause following a complete sentence. The boundary between nonrestrictive (expanding) relative clauses and this so-called free relative connection is hard to draw, since the two types of clause share the same syntactic behavior (see Touratier 1980; Lavency 1981; Bolkestein 1996): for example, they do not necessarily fall under the scope of the main clause sentence type, the verb may have its own illocutionary force, the relative pronoun may fulfill a syntactic and semantic function within another subordinate clause, as in (3a), and in indirect speech

Deixis and anaphora: Some case studies

223

passages, the verb of the relative clause may be treated as if it is a main verb. The most extreme (or most liberal) instances of this construction, as compared to what is allowed in other languages, are those where there is no antecedent in the immediately preceding clause at all, as in (3b) (note, however, that here hierarchical structure is clearly a relevant factor, in that the intervening clause is parenthetical): (3)

a.

b.

at ego basilicus sum: quem nisi but I king.like:NOM.SG be:1.SG REL:ACC.SG NEG-if oras, guttam non feres beg:2SG drop:ACC not carry:2SG ‘But I am king-like: (and) if you do not beg him, you will not get a drop’ (Plaut. Rud. 434) perspice : : : cum Caesare suav-issim-am see:IMP with Caesar:ABL pleasant-SUP-ACC.SG.F coniunctionem (haec enim me una : : : tabula relation:ACC.SG this:NOM for I:ACC one:NOM plank:NOM delectat). Qui quidam Quintum meum : : : quemadmodum delight:3SG REL PTL Quintus:ACC my:ACC how tractat honore! treat:3SG honor:ABL ‘see my most agreeable relation with Caesar (for that is the only piece of wood which delights me).(And) with what honor does he treat my Quintus!’ (Cic. Att. 4.19.2)

The terminological distinction between ordinary expanding relative clauses and the free relative connection depends on what the defining criteria are for classifying a clause as a relative clause.4 The conditions determining the choice between the various expression means available in Latin for intratextual deixis and anaphora have not been examined in detail in the handbooks, but have recently received attention in Bolkestein and Van de Grift (1994), Bolkestein (1996, 2001), Pennell Ross (1996), and De Jong (1996, 1998). These studies examine different samples of monologic and dialogic discourse. A variety of heterogeneous factors appear to be relevant, in that there is some correlation between each of them and preference for one or other of the expressions. In monologue discourse demonstrative iste hardly plays a role. However, in dialogue texts such as the comedies by Plautus (De Jong 1998) it occurs quite frequently for referring to discourse segments, although not in the same way as the other possible expressions for such reference may (see below). An important semantic factor distinguishing among the three demonstratives is whether the reference is to a constituent or to the content of a clause or larger discourse segment, such as an episode or paragraph. We will call this discourse or text reference. Zero pronoun is rarely found for the latter. Of the different pronouns, ille

224

Yaron Matras and A. Machtelt Bolkestein

is extremely rare for reference to preceding paragraph or clause content, whereas both is and hic (in their neuter forms, of course) are common. Partially connected with this, ille is also relatively more often animate, and fulfills the semantic function agent in the state of affairs in which it occurs more often than do both hic and is. However, hic and is are not fully interchangeable either, since is (or rather its neuter form id) tends to refer more frequently to the content of one single clause or proposition, whereas hic (or rather the plural neuter form haec) tends to refer to multiple events, that is, to the content of a series of clauses or a whole paragraph. The relative pronoun (singular neuter quod or plural quae) as well may be used for both purposes (with no clear tendency for single or multiple clauses). In monologue text, the pronoun iste is rare for clause reference, but in dialogue it occurs quite frequently to refer to the content of a preceding utterance of the addressee (De Jong 1998), referring to a piece of discourse. It may in such a text type also refer to a constituent within the addressee’s discourse (referring to an entity as ‘the X you just mentioned’). Difference or similarity in syntactic status between antecedent referent and the anaphoric expression is another factor playing a role, as an investigation of subject instances of these expressions shows (Bolkestein and Van de Grift 1994). Whereas zero pronoun tends to continue the same syntactic status, hic has some, less strong, preference for doing so, whereas is seems to be indifferent. The expression ille however, if itself subject (which it is in the majority of cases), exhibits a strong tendency to refer back to a nonsubject. The relative pronoun, on the other hand, rarely has a subject as its antecedent (and in fact itself tends to be a nonsubject too, as opposed to ille, see Bolkestein [1996]). The examples (4) are more or less representative for zero pronoun and ille:

(4)

a.

b.

Is coniurationem nobilitatis fecit : : : Id he:NOM conspiracy:ACC nobility:GEN make:3SG.PST it:ACC hoc facilius persuasit quod : : : DEM:ABL easily:COMP persuade:3SG because ‘He formed a conspiracy of the nobility. In this (he) persuaded (them) the more easily because : : :’ (Caes. Gal.1.2.1) Erat unus intus Nervius nomine Vertico, be:3SG one:NOM inside Nervian:NOM name:ABL V:ABL loco natus honesto, qui : : : Hic origin:ABL born:NOM honorable:ABL REL:NOM DEM:NOM servo : : : persuadet : : : ut litteras ad Caesarem slave:DAT persuade:3SG to letter:ACC.PL to Caesar:ACC deferat : : : Has ille in iaculo bring:3SG DEM:ACC.PL DEM:NOM on spear:ABL

Deixis and anaphora: Some case studies

225

illigatas effert. bound:ACC.PL bring.out:3SG ‘There was a Nervian in the camp, named V., born to a honorable estate, who : : : He (Hic referring to subj.) persuaded a slave to deliver as despatch to Caesar : : : That man (ille referring to non-subj.) carried forth the despatch bound on a javelin.’ (Caes. Gal. 5.45.3–4) In (5a) and (5b) the relevant tendencies in the case of zero pronoun and pronouns which are syntactically subject are summarized in two ways:5 (5)

a.

b.

Preferred expression for continued reference: continuation of subject: 0 > hic > is/ille continuation of nonsubject: ille > 0 > hic > is Relation subj./nonsubj. antecedent per pronoun and zero pronoun: % Antecedent subj. % Non-subj. anaph. subj. zero 68 32 hic 60 40 is 47 53 ille 17 83

In the case of ille, by far the majority of instances concern preceding contexts where two or more (animate) participants are interacting in changing semantic roles and syntactic functions (as is the case in (4b), where agent-subject ille refers back to an oblique antecedent). Consequently, this pronoun also scores high for the factor “ambiguity” as distinguished in Givón (1983), a factor which in itself does not reveal the syntactic and semantic function shift characteristic for ille. None of the other expressions manifests such a strong correlation to syntactic function switch, although zero subjects and hic-subjects both have subject antecedents more often than not, and is slightly more often has a non-subject as antecedent. Pragmatically, several tendencies are observable. As shown in Bolkestein and Van de Grift (1994), the discourse status of the antecedent in terms of degree of topicality and focality (cf. Dik 1989) is clearly relevant: antecedents which are already topical tend to be continued by zero pronoun, whereas referents which are focal more frequently are continued by one of the pronouns.6 (There is some correlation with the data regarding [dis]similarity of syntactic function here, because of the tendency [not a rule] for topicality and subject function to coincide.) However, the subtype of focality turns out to play a role as well: referents introduced in presentative and existential clauses as topics-to-be (often in subject function) are rarely continued by ille or by zero pronoun, and prefer hic or is (they may also be continued by a relative pronoun, which because of the rather frequent VS order in presentative clauses, may often be analyzed as a regular relative clause rather than a free relative connection). These tendencies are illustrated by (4b) above and also by (4a), which is the sequel

226

Yaron Matras and A. Machtelt Bolkestein

to (6), in which the referent Orcingetorix is introduced for the first time as a focal topic-to-be: (6)

(= the clause preceding (4a)) Apud Helvetios longe nobil-issim-us : : : fuit At H.:ACC.PL far noble-SUP-NOM.SG be:3SG Orcingetorix Orcingetorix:NOM.SG ‘With the Helvetians by far the most noble man was Orcingetorix’

In (7) the preferences are given: (7)

Discourse status antecedent and preferred continuation: topical (H 0 > ille > hic > is focal (H ille > hic > is > 0 focal topic-to-be (H hic > is > 0 > ille

Another factor that is discourse determined is connected with what the speaker is actually trying to do with the host clause containing the anaphoric expression, that is, the status or function of the discourse segment containing the anaphoric expression with respect to its surrounding context. A similar phenomenon has been observed to be relevant in Fox (1987) for the choice between personal pronoun, demonstrative pronoun, and full NP anaphora in English spoken and written texts, and by Vonk et al. (1992) for the choice between personal pronoun and full NP in Dutch. When the clause containing the anaphor either starts or closes a background passage, or when it is a return to the main storyline after some kind of digression, a pronoun will be preferred, even when the referent itself has also figured in the linearly preceding clauses, and could therefore be considered to be quite accessible/topical. For such “switches” in discourse level ille is very uncommon, but is and especially hic are selected rather than zero pronoun. The antecedent referent may quite well be more than one clause distant, especially if the intervening material is clearly background, as Pennell Ross (1996) points out. An illustration of hic after twice zero pronoun for the same referent on a discourse boundary is (8a). Skipping of intervening clausal material is not even excluded with what presumably is the weakest form of explicit coding, the free relative connection; an instance is (8b): (8)

a.

Eidem Alexandro et equi magna raritas same:DAT A.:DAT also horse:GEN big:NOM rarity:NOM contingit. Bucephalan eum vocaverunt sive ab : : : sive happen:3SG B.:ACC him:ACC call:3PL or from or ab : : : XVI talentis ferunt ex Philonici Pharsali from 16 talent:ABL.PL say:3PL from P.:GEN P.:GEN grege emptum etiam tum puero capto eius herd:ABL bought:ACC too then boy:ABL caught:ABL he:GEN

Deixis and anaphora: Some case studies

b.

227

decore. Neminem hic alium quam beauty:ABL nobody:ACC DEM:NOM other:ACC than Alexandrum regio instructus ornatu recepit in A.:ACC royal:ABL adorned:NOM outfit:ABL accept:3SG in sedem. saddle:ACC ‘The same Alexander also had the good fortune to own a great rarity in horseflesh. (They called him B., either because of : : : or because of : : : They say that [he=B.] was bought for sixteen talents from the herd of P., while [he=A.] even as a boy was captured by his beauty.) He (DEM) would not take anyone else on his back when adorned with the royal saddle.’ (Plin. Nat. 8.154) [sequence NP (is 0 0 ) hic] Decuriones conveniunt. Docent: : : non esse; neque: : :; decurions:NOM.PL gather:3PL explain:3PL not be:INF and.not Proinde habeat rationem posteritatis: : : Quorum PTL have:3SG account:ACC posterity:GEN.SG REL:GEN.PL oratione permotus. speech:ABL.SG moved:NOM.SG ‘The Decurions held a meeting; (they) explained that it was not the case : : : , nor : : :; he should take posterity into account : : : Moved by their (REL) speech : : :’ (Caes. Civ. 1.13) [sequence NP 0 (indirect speech) qui]

Whether there are systematic differences in the exact conditions under which the reference of various anaphoric expression means may skip clauses or intervening referents deserves further investigation, but they do not seem to be totally identical for the various expressions (see Bolkestein 1996). In the case of the relative pronoun, for example, the antecedent, even if linearly at some distance, still usually seems to remain relatively activated and accessible because of semantic reasons: for example the state of affairs conveyed by its host clause may be a natural and expected consequence of something referred to earlier (or part of a larger schema), or the referent of the relative pronoun has implicitly been present all along, because of the series of states of affairs involved. When the material intervening between anaphor and antecedent consists of indirect speech produced by that same referent, as in (8b), the speaker whose speech is recounted is such a clearly accessible referent. For hic and is no such limiting conditions on skipping clauses seem to hold, nor do they clearly differ in this respect. However, according to Pennell Ross’ (1996) findings statistically they differ at least in one aspect: whereas hic prefers a sentenceinitial position (72 % of its occurrences in her sample), is tends to occur mainly sentence-internally (84% of its occurrences in the corpus). The pronoun ille seems to be indifferent with respect to this parameter. Pennell Ross does not attempt to

228

Yaron Matras and A. Machtelt Bolkestein

give an explanation for this difference in preferred position, but suggests it may be connected with higher focality on referents of hic than of is. Indeed there are several circumstances in which we find hic, where it seems to carry some focality, and which favor sentence-initial position at the same time. One of them is illustrated by (9a), another by (9b): (9)

a.

b.

Praeerat : : : (summing up different commanders in different regions). chief.be:3SG Toti tamen officio maritimo M. Bibulus whole:DAT yet task:DAT of.sea:DAT M. B. praepositus cuncta adminstrabat. Ad hunc chief.made:NOM.SG all:ACC.PL organize:3SG at DEM:ACC.SG summa imperii respiciebat. peak:NOM.SG authority:GEN regard:3SG ‘The commander was: : : But. M. Bibulus, made commander of all activities at sea, was surveyor of the whole organization. Upon him (DEM) fell the main authority.’ (Caes. Civ. 3.5) His datis mandatis Brundisium : : : DEM:ABL.PL given:ABL.PL orders:ABL.PL Brundisi:ACC pervenit arrive:3SG ‘Having given these (DEM) instructions he arrived at Brundisium’ (Caes. Civ. 1.25)

In (9a) the host clause of sentence initial hic with a certain degree of emphasis summarizes and closes off the main point which the preceding segment has been leading up to, without actually offering much new information. In English or Dutch this would perhaps more naturally be done by means of a cleft construction like ‘so HE was the one who: : : / it was HIM who: : : (etc.)’. This construction is not frequent (although possible) in Classical Latin, and can be found in early Romance languages (see Wehr 1984). Clauses containing the pronoun is do not naturally allow of such a paraphrase. In (9b) the (sentence-initial) participial phrase containing hic in first position (here pro-adjectivally) is a sort of linking device: it is paragraph-initial, and starts a new episode while itself summarizing the content of the preceding clauses. So-called “ablativus absolutus” (loosely connected ablative) phrases of this type are relatively frequent in Caesar’s historiographical narrative and they contain either hic or the relative pronoun qui in the free relative connection, but very rarely is (the pronoun ille does not occur in phrases which function in this way at all, see our earlier remark about its characteristic usage). An alternative expression for fulfilling the same summarizing plus linking function in discourse at the beginning of a new episode are finite subordinate clauses: again hic and qui are more frequent in such clauses than is. It is our impression, but hard to prove conclusively, that when hic is

Deixis and anaphora: Some case studies

229

used, it is emphatic and implicitly contrastive to some other events occurring elsewhere, or alternative courses of action, whereas qui does not imply such contrast, nor does is in its less frequent occurrences in such linking clauses. The stronger identifying force may be due to the potential of hic as a deictic. A final use differentiating the available pronouns is the recognitional use distinguished in Himmelmann (1996), and well known from the handbooks on Latin. This use is neither situational in a concrete sense, nor intratextual, since it refers to a referent treated as belonging to the common knowledge of speaker and addressee. In some text types it is absent (e.g., in historiographical narratives), but in other, more interactive text types it is quite frequent, as in the letters of Cicero to his friend Atticus (see Bolkestein 2001). The “distal” pronoun ille is the only one used in this way, and extremely frequent in this use, especially pro-adjectivally, namely, as a determiner to a noun phrase. This use is illustrated in (10). Example (10) is actually the first sentence of a letter, showing that no anaphora in the strict sense is involved (unless it would be to an entity mentioned in the lost letter to which this one forms a reaction): (10)

Teucris illa lentum sane Teucris:NOM DEM:NOM.SG slow:NOM.SG rather negotium business:NOM.SG ‘That Teucris is rather a slow business’ (Cic. Att. 1.12.1)

The distribution of pronouns in tail head linking constructions in narratives, the specific function of iste in dialogue discourse, and the recognitional use of ille in interactive text types such as personal correspondence show that not every type of discourse manifests all functions of the pronouns available in the language. It also leads to differences in relative frequency of the individual expressions in different text types (for details, see Bolkestein 2001). While the picture for Latin is far from complete, at least the following factors are relevant to why the speaker tends to select a particular anaphoric expression for referring to an entity which is already to some extent accessible in the discourse itself or in the common knowledge of the speaker and hearer: semantic properties of the referent (discourse or constituent); recognitional use; the presence of competing referents; distance to antecedent; syntactic status of antecedent and of anaphoric element; informational status of antecedent; the discourse hierarchical relation between the containing clause and its discourse context; the degree of emphasis on the anaphoric entity itself. The tendencies observed are partially still impressionistic, and moreover, these factors are not mutually exclusive and some of them are interrelated. However, the existence of the correlations observed allows us to expect a particular expression to be most probably used in certain circumstances, and to predict the nonoccurrence of other expressions in others.

230

4.

Yaron Matras and A. Machtelt Bolkestein

Situational deixis, intratextual deixis, and anaphora in modern Dutch

Dutch is a fixed word order language to the extent that in declarative main clauses, the finite verb must occupy the second position, following either the subject or some other constituent (in the latter case, the subject will follow the verb). Although in spoken language certain kinds of contextually or situationally known referents (for example, contextually given sentence first objects) may be left unexpressed, pronouns cannot be freely “dropped.”

4.1. Situational deixis in Dutch The demonstrative system of Dutch is, like English, a two term system based on proximity to (deze/dit) and distance from (die/dat) the speaker. These pronouns (both having a non-neuter and a neuter form for natural and syntactic gender) are used both pro-nominally and pro-adjectivally. Especially in spoken language, they may also be doubled or reinforced by compatible deictic adverbials specifying proximity or distance (deze/dit : : : hier vs. die/dat : : : daar).7 Dutch furthermore has a definite article de/het ‘the non-neuter/neuter’ (in the singular but not in the plural specified for syntactic gender) and an (undifferentiated) indefinite one een ‘a’ for the singular. Indefinite plural NPs have no article. The articles are neutral as to location in space, and thus fall outside the domain of situational deixis. Specification as to proximity/distance with respect to the speaker is possible by combining the article with the locational adverbials as in de jongen hier/daar ‘the boy here/there’. The personal pronouns of Dutch are singular and plural first person (ik, oblique mij ‘I’, ‘me’ and wij, oblique ons ‘we’,’us’ the latter undifferentiated for the opposition inclusive/exclusive), second person (jij, oblique jou ‘you’ and jullie ‘you pl.’) and third person (singular hij/zij/het, oblique hem/haar/het ‘he/she/it’, differentiated for [natural or syntactic] gender, and plural zij, oblique hen, ze ‘they’, ‘them’). In the singular and plural, the second person is differentiated for social relation between speaker and hearer(s) (familiar jij/jullie vs. polite U). In some syntactic positions, some of the pronouns concerned have a “weak” variant: thus ’k, me ‘I, me’, je ‘you’, we ‘we’, ze ‘she’, ze ‘they, them’ and ‘em, ‘er ‘him’,’her’. These weak forms cannot be used for situational deixis or with contrastive focality as can the strong forms (apart from the neuter form het ‘it’). As in the case of the demonstrative pronouns, deictic reference of the third person pronoun may furthermore be reinforced as to proximity or distance from the speaker by means of the deictic locational adverbials as in hij/zij hier ‘he/she or they here’ vs. hij/zij daar ‘he/she or they there’ (obviously with first person pronouns only the proximate adverbial is compatible, as in ik/wij hier ‘ik/wij hier’ vs. ?? ik/wij daar ‘I/we there’). Weak forms of the personal pronouns cannot be so specified.

Deixis and anaphora: Some case studies

231

4.2. Intratextual deixis and anaphora in Dutch For reference to entities which are identifiable from the preceding discourse, Dutch uses strong and weak forms of personal pronouns (hij/zij/het ‘he/she/it’), definite NPs, and the same two demonstrative (proximate and distal) pronouns deze/dit en die/dat (both pro-nominally or pro-adjectivally) which are used for extratextual deixis (see Section 4.1). A recent corpus based study on factors which differentiate the use of definite NPs as opposed to demonstrative NPs is Maes (1994). According to Maes, definite NPs are used when the referent is identifiable (but not necessarily highly accessible in the context or situation), whereas both types of demonstratives presuppose high accessibility of the referent in the context or situation. This study is limited to instances of “adjectival” demonstrative pronouns, and includes both recognitional uses, and instances of so-called self-reference, such as (11), which are actually particular instances of extratextual deixis: (11)

Er zijn dit jaar in dit land vierhonderdtwintig there be DEM year in DEM country four.hundred.twenty moorden gepleegd. murders committed ‘This year, 420 murders have been committed in this country.’

Depending on whether the head noun is identical to the antecedent or lexically different and simple vs. complex, a further semantic distinction is made into three ways of anaphorically referring (or modificational type), labelled “classifying,” “attributive,” and “contextualizing” use. A classifying NP activates a class of elements of which the referent is a member (as in: the sparrow being referred back to as this/that bird). An attributive NP adds further information about the referent (as in: the report : : : this confidential report). Contextualizing NPs are a literal or partial repetition of their antecedents (the air pollution : : : this air pollution). This semantic distinction in itself does not differentiate between the proximate and the distal pronouns, since both can fulfill all three functions, the contextualizing function being the most frequent one for both (Maes 1994: 207). However, there is a certain correlation between classifying use and distal die and attributive use and deze. This correlation can be explained in terms of what Maes sees as the determining factor for the variation between proximate and distal pronouns (we will come back to this below). Maes confronts his findings with those of Kirsner (1979), Kirsner and Van Heuven (1988), and Kirsner et al. (1987). Their description of the use of Dutch demonstrative pronouns in discourse is formulated in terms of a difference in noteworthiness of the referent and in urgency or strength of deictic force. Maes also contrasts his findings with claims made for differences between the use of the English pronouns this and that in terms of degrees of accessibility in Ariel (1988, 1990, 1994) and in terms of “focus movement” in Sidner (1983). He rejects both approaches as unsatisfactory

232

Yaron Matras and A. Machtelt Bolkestein

for Dutch. Before turning to Maes’ description, we will consider some of the factors which have been shown to be relevant in other languages. With respect to the semantic parameter “type of entity referred to” (discourse segment vs. sentence constituent) both demonstrative pronouns are used for both purposes. In spoken Dutch, the distal pronoun die/dat is much more frequent for both semantic types of referents, however, and seems to be the unmarked choice: in spite of its “distal” meaning for situational deixis, it may refer both to a segment of the speaker’s own (recent) words, or to part of the addressee’s (recent) words, and even sentence internally, for example referring to a theme constituent. The proximate demonstrative pronoun deze/dit on the other hand, is statistically marked in spoken Dutch, and is mainly used for the reference to the speaker’s own words. In written Dutch, the frequency of deze vs. die strongly varies with text type, a fact confirmed in Maes’ samples. The handbook on Dutch (Geerts et al. 1984) states that the use of the proximate pronoun makes the impression of emphasizing the fact that the speaker is referring to the content of his own utterance. Other semantic characteristics are observed in Kirsner (1979). In a sample of written Dutch, Kirsner finds that deze NPs are significantly more frequently +human and singular, and refer to named individuals than die NPs. He and also Kirsner et al. (1987) and Kirsner and Van Heuven (1988) interpret these differences as connected with a presumed greater “noteworthiness” of deze NPs. As Maes points out (1994: 165), one might expect constituents with such semantic properties to be easier to identify and thus more accessible: “noteworthiness” probably meaning something like more important or mentally closer to the speaker. This seems to be at odds with another property, which Kirsner et al. (1987) ascribe to deze NPs, namely a higher deictic force or greater urgency (see below). Moreover, in the larger sample of Maes, which is more varied as to text type, the difference between deze and die NPs with respect to the property humanness is not found, although the difference with respect to number is. The parameter continuation vs. switch of syntactic and/or semantic function between antecedent and anaphoric referent does not clearly differentiate between the two pronouns. In Kirsner’s data, deze NPs are themselves somewhat more often subject than die NPs, but whether or not their antecedent has the same status is not investigated either by him or by other authors. Statistical data are lacking. In informal language, when compared to the use of personal pronouns, the distal pronoun die/dat appears regularly when the referent occurs in a semantic role and syntactic function different from its antecedent, as in (12a), where the anaphor (Die) has switched from fulfilling a non-agent/non-subject to an agent/subject function (Jan), as opposed to (12b), where the personal pronoun hij can be interpreted either as continuing the agent/subject (Jan) from the preceding clause or as coreferent to its patient (Piet):

Deixis and anaphora: Some case studies

(12) a.

b.

233

Gisteren sprak Jan Piet nog. Die heeft hem verteld. yesterday spoke Jan Piet still DEM has him told ‘Jan spoke with Piet just yesterday. He told him.’ (Die=Piet; non-subj. ( subj.) Gisteren sprak Jan Piet nog. Hij heeft hem verteld Pprn has Pprn told ‘Jan spoke with Piet just yesterday. He told him.’ (if unaccented, Hij=ambiguous, reading Hij=Jan preferred; subj. ( subj.)

Here it might perhaps still be argued that the existence of possible ambiguity causes a relatively low accessibility of the referent and, therefore, a larger mental distance, and that therefore the distal pronoun is to be expected. There are, however, many instances especially in spoken Dutch which allow distal die where no such ambiguity exists, nor a switch in syntactic function and/or semantic role between antecedent and anaphoric expression. In some handbooks, such as Geerts et al. (1984), it is suggested that in Dutch the choice between personal pronoun and the distal demonstrative is determined in favor of the former if the referent in question is topical, and already fully accessible, as in (13a), and of the distal demonstrative when the antecedent has just been introduced for the first time (13c)–(13d) below. However, it is not difficult to find attested instances where the referent has been quite recently mentioned and is therefore fully accessible, but where the distal pronoun is still chosen rather than the personal pronoun. Even in the case of intrasentential coreference, when the antecedent is mentioned in a loosely connected sentence first position, such as the theme constituent (13b), it will be continued by means of the distal pronoun. Especially in informal registers, a referent newly introduced in a presentative clause will also often be continued by means of the distal pronoun as well ((13c)–(13d)).8 In the latter case, there may or may not be a switch in syntactic function (see (13c)). Whether or not we are dealing with a syntactic and/or semantic function switch, it is the distal pronoun rather than the proximate one which is selected as an anaphoric expression, in spite of the proximity of the antecedent (note that this is different from what takes place in Latin: in instances such as (13b) the preferred pronoun would be anaphoric is, which has no deictic force whatsoever, and in (13c)–(13d) either is or the proximate one hic would be employed rather than the distal pronoun ille): (13) a.

De stoel stond in de hoek. Ik vond hem/?die/??deze niet DEF chair stood in DEF corner I found it/DIST/PROX not mooi. beautiful ‘The chair was standing in the corner. I did not like it/?that one/??this one.’

234

Yaron Matras and A. Machtelt Bolkestein

b.

c.

d.

Een hele week Bea, dat/??dit is teveel van het INDEF whole week B. DIST/PROX is too.much of DEF goede! good ‘A whole week of Bea, that/??this is too much!’ Van reder H. waren drie schepen op zee. Zouden of ship.owner H. were three ships at sea would die/??deze ooit terugkomen? DIST/PROX ever come-back ‘Of shipowner H. there were three ships at sea. Would they (lit. those)/ ??these ever return?’ Er stond een stoel in de hoek. Die/??deze vond ik there stood INDEF chair in DEF corner DIST/PROX found I niet mooi. not beautiful ‘There was a chair standing in the corner. I did not like it (lit. that one)/ ??this one.’

(When the referents carry contrastive focality, the pronouns receive extra emphasis in the prosodic contour: the pronouns in (13a)–(13d) classified as dispreferred are so only when they are nonfocal.) The following examples from less formal texts, such as children’s stories, give some more characteristic cases of a distal demonstrative referring to an already topical antecedent, with syntactic and semantic function shift (14a)–(14b) (the first expression is the one attested; in more formal written language, the expression selected would be the personal pronoun rather than the proximate one): (14) a.

b.

De wolf week terug. Samen moesten ze een zeer DEF wolf retreated back together must they INDEF very dreigende indruk maken op de wolf. Die/hij/??deze stond threatening impression make on DEF wolf DIST/he/PROX stood stokstijf. very-stiff ‘The wolf retreated. Together they had to make a very treathening impression on the wolf. He (distal DEM/personal prn./??proximate DEM) stood like a statue.’ Snel bukte ze zich en raapte de tak op. quickly bent she REFL and pick DEF branch up Die/hij/??deze woog zwaar. DIST/it/PROX weighed heavy ‘Quickly she bent and picked up the branch. It (distal DEM /personal Pprn./??proximate DEM) was heavy.’

Deixis and anaphora: Some case studies

235

Although these instances involve both syntactic and semantic function switches between antecedent and anaphor, as in (14a), there is no real ambiguity, in view of other factors (the difference in number between the successive agent/subjects in (14a), other semantic implications in (14b): no inaccessibility or mental distance seems to be involved). The dispreferred proximate pronoun would be acceptable if carrying contrastive focality (implying ‘as opposed to another wolf/branch’). One other parameter seems to play a role in the distribution of the distal demonstrative pronoun vs. that of personal pronouns. This is connected with the status of the discourse segments containing the referents with respect to each other. Examples (15a)–(15b) illustrate this point: (15) a.

b.

Anna is vroeg naar bed gegaan. Ze/Die voelde zich niet zo A. is early to bed gone she/DIST felt REFL not so lekker. good ‘Anna has gone to bed early. She/That-one did not feel very well.’ Anna ging vroeg naar bed. Ze/??Die werd om vijf uur A. went early to bed she/DIST became at five hour wakker. awake ‘Anna went to bed early. She woke up at five.’

In (15a) the second clause gives a background explanation (by the narrator to the addressee) of the event reported in the first clause. In (15b) the message conveyed by the two clauses is on the same level: two subsequent events on the storyline are narrated (note the same imperfect tense in (15b), whereas in (15a), the two clauses differ in tense). In the second clause of (15b), the distal demonstrative die (if not contrastive) would not be successful, whereas in (15a), it is fully acceptable, be it somewhat more informal sounding than the personal pronoun ze. Statistical data concerning this phenomenon are lacking. With respect to other conceivable discourse factors (such as the degree or type of topicality or focality of antecedent entity), these play a role in the choice between demonstrative pronoun on the one hand, and personal pronoun or definite NP on the other, rather than that they are relevant to the choice between the proximate and distal pronouns themselves. Maes (1994: 156–162) effectively argues against the applicability to Dutch of the description of demonstrative variation in English given in Sidner (1983). Without going into the details of Sidner’s “focus movement” account here, which actually is not quite in terms of differences in discourse status of either the antecedent referent or the anaphor, suffice it to say that the Dutch data in Maes (1994) do not reflect the properties which could be taken as proof for the relevance of Sidner’s focus movement, and that proximate vs. distal demonstrative variation does not seem to be motivated by such factors. An investigation which inde-

236

Yaron Matras and A. Machtelt Bolkestein

pendently determines degrees or types of either focality or topicality of antecedents and demonstrative pronouns in Dutch, and which investigates whether or not such distinctions are a relevant factor, is lacking. However, Kirsner (1979), Kirsner and Van Heuven (1988), and Kirsner et al. (1987) observe some statistical differences between the two demonstratives (apart from the “speaker’s own words” reference of deze/dit), which still lead them to suggest there is a functional, discourse-connected difference involved after all. Tendencies which might be interpreted as connected with differences in discourse status are the following: (i) differences in linear position in the sentence: proximate deze occurs significantly more frequently near the beginning of the sentence than distal die; (ii) referential distance to the antecedent: statistically, the distance between the two coreferent entities is actually larger in the case of deze than in the case of die, at least if intrasentential coreference is included in the count, as is the case in Kirsner’s data; in Maes’ (1994) larger sample of 837 adjectival demonstrative NPs, however, a slightly higher (nonsignificant) percentage of the distal pronoun occurs in sentences more than one clause earlier than the percentage of the proximate pronoun doing so; (iii) both in the data of Kirsner (1979) and in those of Maes (1994) the noun following deze is more often different from the antecedent noun than in the case of die; (iv) Kirsner (1979) furthermore states that deze occurs more often in main clauses and at the beginning of paragraphs than die. Kirsner (1979), Kirsner et al. (1987: 107), and Kirsner and Van Heuven (1988: 223) conclude from their observations that there is a qualitative difference in “reference type” between the two pronouns: they state that die as a rule simply just “re-mentions” a referent which is easy to identify (on the basis of [ii]), whereas deze is claimed to encourage the addressee to “reinterpret, re-analyze, summarize” (on the basis of [iii]) the referent, which may be more problematic to identify. In connection with the statistical differences found they take the proximate pronoun to fulfill a “more urgent” task, and have a “higher deictic force” than the distal one, and assume that the effort required from the addressee in the case of (proximate) deze is actually greater than in the case of (distal) die, in spite of the fact that, as mentioned above, deze NPs often refer to +human, singular and “named” entities, and are therefore presumably easy to identify. The notion of urgency and deictic force suggest problematic accessibility rather than unproblematic accessibility. Maes (1994: 170) shows on the basis of a comparison between his data and those discussed in Kirsner that there is insufficient support for this account of pronoun variation in Dutch. His general conclusion is that (a) “localistic” proposals which treat the variation in pronouns in Dutch in terms of a scale of either linear or some kind of mental distance are not satisfactory. Thus there is no straightforward relation in terms of the notion distance between on the one hand the situational use and on the other the intratextual use of the two demonstratives involved: as shown above, the distal one is often coreferent to the nearest, most recently mentioned and clearly

Deixis and anaphora: Some case studies

237

accessible item in the immediately preceding clause, especially in informal language; (b) secondly, “qualificational” proposals like Kirsner’s, which treats the variation in pronouns in terms of the qualitative relation of “noteworthiness” attributed by the speaker to the referent, assigning different degrees of urgency to the proximate and distal pronouns, are internally ambivalent and not intersubjectively supported by his data. Maes (1994: 171–212) proposes to describe the difference between the two (adjectival) demonstrative pronouns not in terms of differences in degree of accessibility, but in terms of a difference in selection of “reference domain.” He assumes that both demonstrative pronouns share the assumption that the referent is highly accessible. In his view, discourse always establishes at least one “reference domain” (with four coordination points, namely place and time of utterance, speaker, and context) and may establish other reference domains with other times/places and, for example, the addressee as coordination points. The proximate pronoun links the referent to the here and now and the producer of the discourse, whereas the distal pronoun links the referent to the not here and now and to the addressee of the discourse. Thus, both demonstrative pronouns have spatio-temporal and actor (speech participant) oriented, relational uses, which, as Maes (1994:187) points out may both be combined in particular occurrences. Support for this account of Dutch demonstratives comes not only from the existence of so-called self-referential uses of deze (where die would be less preferred) and uses of die in narratives (where deze would be less preferred), but also from the differences in distribution of the demonstratives over the various discourse types which form his sample. The difference between speakeroriented and addressee-oriented use of demonstratives is described as a difference in “relational mode,” namely “unequal” vs. “equal”. The plausibility of this distinction is supported by the co-occurrence in the sample of a range of explicit indications of the speaker’s choice of orientation, such as begrijp je ‘you see’ with the distal pronoun, in (16a), or zoals ik dat zou willen noemen ‘as I would like to call it’ with the proximate as in (16b) (Maes 1994: 179): (16) a.

b.

Die spanning van die twee werelden, begrijp je. DIST tension of DIST two worlds understand you ‘That tension of those two worlds, you see.’ Deze sprankelende wereld van passie, zoals ik dat zou PROX sparkling world of passion as I DIST would willen noemen, is eigen aan de schrijver. want call is proper to DEF author ‘This sparkling world of passion, as I would like to call it, is peculiar to the author.’

Other support for an account in terms of selection of reference domain as speakeroriented and unequal versus other (addressee-) oriented and equal comes from the

238

Yaron Matras and A. Machtelt Bolkestein

fact that the distribution of these modes varies with discourse type. Thus, for example, expository official instruction texts in the sample contain a relatively high number of deze NPs, whereas autobiographical narrative and opiniative personal comments texts contain a relatively high number of die NPs. The table in (17) gives the % of the adjectival demonstratives in two of the subsamples consisting of different types of discourse (Maes 1994: 189): (17)

Distribution of demonstratives per discourse type: Expository/instructional Autobiographical deze 85.9 10.3 die 14.1 89.7

Moreover, when we consider instances in which the type of relation between discourse segments seems to be mysteriously relevant for the acceptability versus nonacceptability of the distal pronoun instead of the personal pronoun, as in (15a)–(15b) above, one might argue that producing a clause which forms some kind of explanation of some preceding narrative clause may be viewed as an addressee-oriented action. There may well be a connection between the selection of a relational mode by the speaker and the type of relations between discourse segments. This seems to be a promising direction for future research. From the attested instances, it appears that among the attested cases of die NPs there are many not strictly anaphorical instances which would fall under what Himmelmann (1996) would label the “recognitional” use of demonstrative pronouns. In such cases, the relational aspect of awareness of the addressee is quite noticeably present. Obviously, such use is not present in all types of discourse. As we have seen, in a language like Latin, the recognitional use of the (distal) pronoun ille is also quite characteristic for addressee-oriented discourse types (Bolkestein 2001), which also results in a difference in frequency of the different pronouns in different discourse types. While the reference domain approach seems quite satisfactory for the description of Dutch demonstrative pronouns, it is not clear how universally relevant the notions involved are, that is, whether an analysis in terms of reference domain can be extended to other languages which have a two-term deictic system and use the same demonstratives for intratextual deixis. Where two two-term languages differ systematically, as do English and Dutch, the question arises what factors motivate English speakers to select the proximate pronoun in places where the Dutch speakers consistently select the distal one. Furthermore, it is not clear whether a parallel case could be made for the intratextual use of demonstrative pronouns from a three-term deictic system, and how the existence of anaphoric determinative pronouns would fit in. Summarizing, heterogeneous factors seem to correlate with the choice between personal pronoun and demonstrative pronoun on the one hand, and with the choice between distal and proximal pronoun in Dutch on the other. With respect to the lat-

Deixis and anaphora: Some case studies

239

ter, in informal spoken language the distal pronoun is the unmarked choice anyway. This means that in a spoken sample, an investigation of the parameters which influence the choice between personal pronoun and substantival die and between definite NPs and die NPs is perhaps more fruitful than the choice between the two terms of the demonstrative system. For the first choice, both the syntactic status of the antecedent and the hierarchical discourse relation between the containing clause and the surrounding discourse seem to play a role; for the second choice, position of the pronoun, type of referent, text type, and choice of referent domain seem to be relevant.

5. Deixis and anaphora in Romani and Modern Hebrew We have already addressed the problem of a continuum in the strategies of reference to space versus context above. In the following case studies, we examine more closely the relationship between two dimensions that are basic to the deictic system of European languages, namely, reference to space vs. reference to context, and the coding of distance or remoteness. We shall examine two cases of marginal European languages: Romani, which possesses a structural morphological system, the origin of which is Indo-Aryan, but which has been restructured in its pattern of sentence organization as a result of centuries of contact with European, especially Balkan languages, and which exhibits a distinction between contextual and situative deictics (see Matras 1998). Modern or Israeli Hebrew, a Semitic language, is strongly influenced by a European (mostly Slavic and Germanic, but also Romance) substratum owing to the origin of the dominant group of Jewish immigrants to Palestine who reintroduced Hebrew as a spoken language. While the deictics/anaphora of Biblical Hebrew reflect an opposition between reference to text and reference to spatial dimensions of the speech situation, in Israeli Hebrew the distribution of the forms in question is quite different, and much closer to the average European type.

5.1. Romani Romani has a complex system of deictics, showing a quadripartite set of forms. The two dimensions which figure in the opposition system of deictic expressions in Romani are the source of knowledge about the object of reference, and the accessibility vs. ambiguity of the object of reference with respect to other potential objects, that is, its exclusiveness or specificity. As for the source of knowledge, a distinction is made between reference to objects that are part of the spatial dimensions of the speech situation and are thus physically present and perceivable, and those which have been established in the course of the propositional structure of the discourse, that is part of

240

Yaron Matras and A. Machtelt Bolkestein

Table 1. Deictic expressions in Romani: ‘this’, ‘that’, and ‘here’, ‘there’ Spatial/situative

Masc. Fem. Pl. Place

Propositional/contextual

Open

Exclusive

Open

Exclusive

kado kadi kadala kathe

kako kaki kakala kadka

kodo kodi kodola kothe

kuko kuki kukola kudka

the linguistic context. The second dimension, the exclusiveness of the object of reference, is expressed through a distinction between “general” reference to an object that is easily accessible, being part of the immediate field of reference and unmistakenly identifiable to the hearer, and “specific” reference through a gesture which calls on the hearer to pick out a single object from a group of potential entities. While the actual forms differ among various dialects, these basic features appear across the different dialects of the language. Most dialects also share at least part of the structural characteristics of these functions: reference to the physical setting of the speech situation is typically represented by the vowel -a- in the stem of the deictic expression, while the vowels -o- or, more rarely, -u- represent the propositional context of the discourse as the referential space in which an object is located. Furthermore, specificity of reference is generally the more marked and structurally more complex form, exhibiting in most dialects either a preposed vowel or else a consonant, which often reduplicates the stem. Our examples are taken from the Kelderash/Lovari dialect, a dialect of northern Vlax branch of Romani, originating in the border areas of Romania, Hungary, and Slovakia. The two dialects, Kelderash and Lovari, are usually regarded as distinct by speakers, as well as in the literature. However, the variety under consideration here, recorded among an extended family of Vlax Romani speakers now based in northern Germany, and formerly in Poland and in Austria, shows features of both (for a more extensive discussion, see Matras 1994). Table 1 shows the system of demonstratives and place adverbs in this variety of Romani. Note that the same system of opposition occurs both with demonstratives and with place adverbs. Let us illustrate the Romani system by a number of examples: (18)

Dem tume kadala lila ando vast. gave:1SG you:PL these:PL papers in:DEF hand ‘I handed you these papers.’

In (18), the speaker and the audience of listeners share a common domain of reference, embedded in the perceptual space of the present speech situation; the speaker is

Deixis and anaphora: Some case studies

241

pointing to an object of this domain of reference, the papers which had been handed out. The a- vowel in the stem of the demonstrative indicates that the speech situation in its spatial dimension is the source of knowledge about the referent. The deictic expression here is actually accompanied by a physical gesture pointing to the referent. Thus, what the a-component in the deictic stem really does is convey a gesture in space by means of a linguistic expression, much in accordance with Bühler’s (1934) Zeigen as a gesticulated reference in real space. In addition to conveying a spatial gesture, the deixis in (18) is open, that is, there is no potential ambivalence with regard to the object to which reference is made. (19)

Samas ande ek hotelo, kodo sas ando : : : were:1PL in INDEF hotel that was:3 in ‘We were in a hotel, that was in : : :’

Here, the purpose of the deixis is to grant orientation to the hearer within the propositional context of the discourse. It is an entire event previously presented which is the object of the reference procedure. Note that reference is made to the propositional content of the first part of the utterance, rather than to the (physical) properties of the utterance as a speech event. The o-component in the stem of the deictic expression bears an instruction to the hearer to process propositional content conveyed in the linguistic dimension of the speech situation in search for the referent. Once again reference is immediate and therefore “open,” there being no ambivalence involved in identifying the object. kodo picks up the utterance which immediately precedes it. There is no evidence for it referring to a “remote” entity, nor is there reason to assume proximity to the hearer, rather than the speaker, or vice versa. While kodo in (19) refers in fact to the propositional content of an entire segment of speech, rather to a nominal entity, as does kadala in (18), the use of both forms is indeed comparable, and they form a structural paradigm: for example, the deixis in kado manuš ‘this man’, would be taken to mean ‘the man you see in front of you, next to you, etc., that is, the man who is present here and now, the man we can identify by pointing to with a physical gesture’. In kodo manuš the deixis would be interpreted as referring to ‘the man who has been mentioned, or will be mentioned henceforth, but is not visible or otherwise perceivable’. Turning to the second dimension in the deictic system, let us examine the function of the exclusive or specific deixis: (20)

Sas kako rom kaj kerdas kodo. was:3 that man REL did:3SG that ‘It was this/that man who did it.’

As indicated by the vowel -a- in the stem component of the demonstrative, reference is being made to a person who is present in the speech situation. However, this man is part of a crowd, a member of the group of some 25 persons who form the audience

242

Yaron Matras and A. Machtelt Bolkestein

of listeners to what the speaker is saying. The deixis makes the hearer aware of the ambivalence and instructs the hearer to take into consideration relevant information about the referent in trying to single him out from a group of potential reference objects. The reduplicated deixis in (20) thus excludes other potential members of a group by selecting a specific entity. (21)

našaven ame ande kukola thema kaj meras bokhatar expel:3PL us in those:PL countries REL die:1PL hunger:ABL ‘They expel us to those countries in which we starve to death.’

In (21), the demonstrative kukola qualifies the head noun as one which is to be further specified in the context of the discourse. It excludes it from the group of any other potential referents by calling the hearer’s attention to specifying propositional content that is to follow. Again, specification is achieved by means of the reduplicated -k- form in the stem of the deixis, while the contextual, linguistic or propositional reference is indicated by the vowel -u- in the first syllable. There is a further usage of the reduplicated, specific or exclusive deixis, which essentially shares the basic function already described. This type of usage may be called “the missing referent,” because it points to a nominal referent which the hearer is having difficulty retrieving lexically: (22)

Les mukhlja ando kako, ando hotelo. he:OBL left:3SG in that in hotel ‘He left him in what do you call it, at the hotel.’

(23)

A: B: A: A: B: A:

ˇ Kade kana sastilo, našle duj žene ando Cexo. so when recovered:3SG fled:3PL two people in Czech ašile? Aj kukola kaj and those:PL where stayed:3PL Ande Rumunija. in Romania So when he recovered, the two of them fled to Czech(oslovakia). And where did the other ones stay? In Romania.

In both examples (22) and (23), the deictics substitute for a nominal entity (a full noun in (22), probably a name or rather a set of names in (23)) which the speaker is either unable (22) or reluctant (23) to recall. Notice the differences, however. In (22), the problem is finding the right lexical expression for a concept. In other words, it is the speech event which is being blocked by the speaker’s failure to retrieve the correct word. The deixis is essentially a verbalization of a search for an expression, that is, for an ingredient of the speech event. The reference is therefore spatial, or one should rather say, perceptual, that is, one that captures sensory evidence rather than processed propositional elements. By contrast, in (23) the speaker is trying to

Deixis and anaphora: Some case studies

243

recall a conceptual entity, which is recovered from the preceding verbal context. Both expressions have reduplicated k-k- forms, instructing the hearer to single out a specific object from within the group of potential reference objects, or to construct a boundary between the actual reference object and other potential candidates. The distinction between spatial (extralinguistic) and contextual (intralinguistic) deixis, which in some European languages might reflect an exploitation of the proximate/remote opposition, is in Romani one of two primary grammaticalized functions of the system. There is, however, a secondary textual deixis that emerges in planned and prestructured discourse. It promotes pieces of the text to elements of the situation, capturing the speech event as a spatial/perceptual or extralinguistic phenomenon, rather than as a propositional unit, thus commenting on the extralinguistic properties of the discourse: (24) a.

b.

Ando 45to bers žanas ke but amare phurendar kana in 45th year know:1PL that many our:OBL old:ABL.PL when ankliste avri kaj sas len e baxt te came.out:3PL out REL was they:OBL DEF luck that anklen avri, xasarde but peske manušendar. come.out:3PL out lost:3PL many REFL people:ABL ‘In the year ‘45 we know that many of our elderly when they survived, who were lucky to survive, they had lost many of their people.’ Ame kado phenas sar ke šaj ginavas le gazeta: we this say:1PL as that can read:1PL DEF.OBL newspaper “štar manus mandar mule”. four people me:ABL died:3PL ‘We say this as though we were reading the newspaper: “Four of my people died.”’

Rules for participant tracking in the language generally draw on the combined devices of indefinite and definite marking, and the reference system involving contextual deixis (o-deixis), overt anaphora (vov/voj/von ‘he/she/they’, with l- in the oblique forms), and zero anaphora. Consider the following passage based on a written story rendered orally in Romani (Matras 1994: 44 ff.): (25) a.

b.

O jekh cinori historija Hebbel, jekh pisari, pisardas DEF H. INDEF writer wrote:3SG INDEF small:F story raklo, njamco, kaj bešelas ande-k cinoro pa jekh on INDEF boy German REL lived:IMP.3SG in-INDEF small:M gavoro. village:DIM E historija bišol “Kan niet verstaan.” DEF story is.called:3SG

244

Yaron Matras and A. Machtelt Bolkestein

c.

d.

e.

f.

g.

h.

i.

Kodo raklo anda Njamco anda cinoro gav kaj ande that boy from Germany from small:M village REL in pesko trajo cˇ i anklistas avri aresel ando baro REFL:M life not came.out:3SG out arrives:3SG in:DEF big:M foro Amsterdam, ando Holando town A. in Holland Dikhel o barvalimo ande kodo foro, dikhel e bolci, sees:3SG DEF wealth in that town sees:3SG DEF shops dikhel o porto. sees:3SG DEF port Pušel jekhe gažes po drom: “Kasko si asks:3SG INDEF:OBL man:OBL on:DEF way whose COP kado barvalimo sa kado?” this wealth all this o gažo pe leste, phenel Dikhel sees:3SG DEF man at him:LOC says:3SG leske: “Kan niet verstaan.” him:DAT Kodo bišol pe pe gažikani: cˇ i that means:3SG REFL on non.Gypsy.language not xaˇcarav tut. understand:1SG you:ACC raklo, “O gažo kasko si kado sa “Ah,” o DEF boy DEF man whose COP this all bišol ‘Kan niet verstaan’. ” is.called:3SG pa foro sa pušel pe “Kasko kako.” Kade vov nakhel he passes:3SG on town all asks:3SG REFL whose that so ‘Hebbel, a writer, wrote a short story about a boy, a German, who lived in a small town. The story is called “Kan niet verstaan.” This boy, who never in his life left his village, arrives in the big town of Amsterdam in Holland. He sees the wealth in that town, he sees the shops, he sees the port. He asks a man on the street: “Who does all this wealth belong to?” The man looks at him, and says to him: “Kan niet verstaan.” That means in the language of the Gaže [=non-Gypsies]: “I don’t understand you.” “Oh”, the boy, “The man to whom all this belongs is called ‘Kan niet verstaan’.”

Deixis and anaphora: Some case studies

245

So he passes through town and keeps asking himself, “Who does that belong to?” ’ Looking at a selection of the participating referents in the story, those underlined in example (25), the distribution of participant-tracking devices in the passage may be illustrated as follows: a. b. c. d. e. f. g. h. i.

story-indef. story-def. boy-cont.deix. boy-Ø.anaph. boy-Ø.anaph. man-def. man-Ø.anaph. quote-cont.deix. boy-def. boy-anaph.

boy-indef. town-def. town-cont.deix. man-indef. boy-anaph. boy-anaph.quote.

The letters correspond to the division into segments in the passage. The left column indicates prominent positions on the linear progression scheme of the sentence, while the right column represents referents which appear in less highlighted positions. We find basically three strategies involving reference devices for participants in the discourse: promotion, establishment, and recovery. Acquisition of definiteness figures as a general promotion device for newly-introduced referents, as seen the treatment of story across segments (25a)–(25b), as well as of man along (25e)–(25f), and of boy between (25f) and (25h). Deixis appears as a promotion device when a stronger referential force is required. This can be seen in the promotion of boy, the central participant, from a non-highlighted sentence position where it is introduced into the discourse in (25a), across an entire segment in which it disappears from the discourse in (25b), and on to the central position as the most prominent actor in (25c). Here, the contextual deixis kodo allows both retrieval of the referent from a marginal, distant position in the previous discourse, and its promotion into the most central position. The promotion of the quote in (25f)–(25g) from an entire proposition to a referent entity is also achieved by means of the contextual deixis. As opposed to promotion, where referential force is increased, establishment strategies show a gradual withdrawal of the referential force. Thus, zero anaphora is applied in (25d)–(25e) for boy, which continues to figure in the highlighted initial sentence position. Similarly, after the promotion of man from (25e)–(25f), definiteness makes way for zero anaphora in the progress of (25f). (Re)establishment of boy in the last two segments still involves degrading along the referential scale, but it is not zero anaphora, the minimal referential force, which is applied, but rather an overt anaphor. This apparently has to do with the somewhat less prominent position which the actor has in (25h), not being the theme of the sentence.

246

Yaron Matras and A. Machtelt Bolkestein

Referent recovery shows, much like promotion, an increase in referential force, though it co-occurs with backgrounding or “de-highlighting” on the sentential level. In (25c)–(25d), town is downgraded from the principal object of an action to an accompanying setting, but its recovery shows an increase of force from definiteness to the contextual deixis. In (25e)–(25f), boy is downgraded from the position of the central actor to the object/recipient in a backgrounded linear position, and its recovery involves an increase of force from zero- to overt anaphora. Throughout the passage, deixis proves to be the strongest force for referent recovery and promotion. Romani consistently makes use of the contextual o-deixis for the purpose of such participant tracking strategies. Notice that the situative a-deixis (kado, kako) only appears in quotes (segments (25e), (25h), (25i)), where it accompanies a pointing gesture in actual, extralinguistic space.

5.2. Hebrew Historically, Hebrew has two sets of pronouns, the demonstratives ze/zot/ele (m., f., pl. respectively), and the personal pronouns hu/hi/hem. Both sets can also be used as definite adjectival demonstratives. Like lexical adjectives, definite demonstrative adjectives are preceded by the definite article ha: ha-bayit ha-ze ‘this house’, habayit hahu ‘that house’. In addition, the set of demonstratives can also be used as “indefinite demonstrative adjectives”: bayit ze ‘this house’. This renders in effect a system in which demonstrative pronouns (ze) and personal pronouns (hu) are differentiated, and demonstrative adjectives may show a tripartite system: ze, haze, hahu. This older system survives in the literary forms of Modern (Israeli) Hebrew, while in the colloquial language the adjectival demonstrative set is reduced to just the two definite forms, ha-ze and hahu. The fact that third person pronouns derive from remote adjectival demonstratives (the set in hu) is not unusual in universal perspective. More noteworthy is the correlation between definiteness and remoteness. Ehlich (1979) had attempted a functionalpragmatic interpretation of the formal grouping of the expressions, arguing that the primary distinction is that between situative or spatial deixis (ze-forms), and contextual anaphora (hu-forms) (see also Ehlich 1982, 1983). He illustrates for Biblical Hebrew that ze-forms, whether definite or indefinite, invariantly point to elements of actual perception, while the hu-forms, whether pronouns or adjectival demonstratives, invariantly signal continuous reference to an entity which has already been established in the previous verbal discourse (or textual) context. This distribution may be illustrated for Biblical Hebrew by the following examples: (26)

(Kings I, 1:30) ki ka’ašer nišba‘ati lak : : : ki ken e‘ese ha-yyow m ha-ze. for as I.swore to.you for so I.do.FUT DEF-day DEF-this ‘For as I promised to you : : : for so I shall do on this day.’

Deixis and anaphora: Some case studies

(27)

247

(Kings I, 19:8) va-yaqom va-yo’kal va-yište va-yelek be-koah and-he.stood and-he.ate and-he.drank and-he.went in-force ha-’akila ha-hi ‘arba‘iy m yow m ve ‘arba‘im layla. day and forty night DEF-eating DEF-that forty ‘And he stood and ate and drank and walked by the force of (just) that food for forty days and forty nights.’

In (26), ha-ze refers to the time and place of the immediate speech situation. In (27), ha-hi continues the concept of ‘eating’ (’axila) already established by the root ’.x.l in the previous context by the verb yo’xal ‘he ate’. The same distribution is evident for the indefinite or pronominal forms: (28)

(Kings I, 18:7) va-yakirehuw va-yipol ‘al panav va-yo’mer and-he.recognized:him and-he.fell on face:his and-he-said ha-’ata ze adoni ‘eliyahu? QUE-you this lord:my E. ‘And he recongnized him and he bowed and said is it you my lord Elias?’

(29)

(Kings I, 18:39) va-yiplu ‘al pneyhem va-yomru’ ‘adonay hu ha-’elohim. and-they.fell on faces:their and-they.said Lord he DEF-God ‘And they bowed and said the Lord (he) is our God.’

Again, ze in (28) is situative, referring to an entity which is present in the immediate physical surroundings of the interaction, pointing, in fact, to a participant who is being addressed directly. By contrast, in (29) the referent is not present in the situation and is thus beyond direct perception; rather, the anaphor hu expresses direct continuity of a (previously mentioned) contextual referent. While basically retaining this morphological paradigm, Modern Hebrew tends to generalize the ze-deixis as a device for immediate reference, while the adjectivaldemonstrative form of the anaphor hu (ha-hu) is specialized for contextual reference. Consider the following examples from a Modern Hebrew novel (Gadi Taub, Ma haya kore im hayinu šoxaxim et dov [What would have happened had we forgotten Dov], Tel-Aviv, 1992), well known for its rendering of colloquial style: (30)

(p. 70) Hayiti kcat I.was a.bit haya lifney was before

lexuca miday. Biglal ze ze kara. Ze stressed:F too because this this happened this xaci šana. Im haya li kcat yoter nisayon half year if was to.me a.bit more experience

248

Yaron Matras and A. Machtelt Bolkestein

lo hayiti magi’a le-ze. Be-’emet. Ze rak ‘inyan šel NEG I.was arrive:F.SG to-this really this just matter POSS nisayon ha-dirot ha-’ele. ve-ze ‘od axarey se experience DEF-flat:PL DEF-these and-this even after REL araxti me-ha-šutaf še haya li ba-dira lifney ze. I.fled from-DEF-partner REL was to.me in.DEF-flat before this ‘I was a bit too uptight. That’s why it happened. It was half a year ago. If I had had a bit more experience this wouldn’t have happened to me. Really. It’s just a matter of experience, these flats. And that even after I ran away from the flatmate I had at the flat before that.’ (31)

(p. 75) Ha-davar ha-yaxid še hu ‘asa haya lehitkašer la-dira DEF-thing DEF-only REL he he.did was to.call to.DEF-flat be-Bney Brak, aval mišehu kvar saxar ota. Hu lo in-B. but someone already he.rented it.ACC.F he NEG hicta’er. Hu be-meyla lo xašav lagur ba-’ezor he.was.sorry he anyhow NEG he.thought to.live in.DEF-area ha-hu. DEF-that ‘The only thing he did was to call the flat in Bney Brak, but somebody had already rented it. He wasn’t sorry. He hadn’t thought of living in that area anyhow.’

In (30), ze generally reflects items which are obvious and immediately accessible to both speaker and hearer – the general topic of conversation, an event referred to previously, propositions in the immediate context, as well as the situative here-and-now. It is not, however, restricted to a spatial or perceptual reference dimension, but assumes a contextual, or even a broad conceptual scope as well. In (31), the anaphor he continues reference to the actor known to the reader. As an adjectival demonstrative, ha-hu singles out a particular referent from the previous context, drawing a boundary between it and other potential candidates, thus reminding us somewhat of Romani kuko, the exclusive deixis. There is still a difference between more rigid, literary style and colloquial Modern Hebrew. Consider the following contrasting examples:

Deixis and anaphora: Some case studies

249

(32)

(p. 68) “Ze ha-kol šel ba’al ha-bayit,” ha-pakid amar “dira this all POSS owner DEF-house DEF-clerk he.said flat meruhetet ze dira meruhetet.” furnished this flat furnished “‘It all belongs to the owner,” the clerk said. “A furnished flat is a furnished flat.”’

(33)

(p. 71) Ze male be-xol ha-šlatim ha-yerukim-levanim ha-’ele this full in-all DEF-sign:PL DEF-green:PL-white:PL DEF-these še yeš lahem sam : : : “Xayal hacda‘a hi amirat šalom”. REL EXIST to.them there soldier salute that saying greeting ‘It’s full of all those green-and-white signs that they have there: : : “Soldier, a salute is a greeting”.’

When rendering the spoken language, as in (32), the author employs ze as a copulatype expression referring to the theme of the predication. But when quoting the written, official language, such as that of military command posts in (33), the same type of construction makes use of hu, the anaphoric pronoun signaling continuous reference to a previously introduced contextual item.

6. Concluding remarks From the above brief discussion of the conditions of use of deictic and anaphoric expressions in various languages (Classical Latin, Dutch, Romani, and Modern Hebrew) it will be clear that not only do languages tend to have more than one expression available for referring to already identifiable entities, but also that the factors which influence the selection of one expression means rather than another differ considerably across languages. Pronouns that are used for situational deixis often can be used with a variety of textual functions as well, but the relation between the two types of usage are not always very straightforward. Factors which are observed to be relevant to the choice in one language need not be relevant in another, and often are a matter of statistical preference rather than of black-and-white (un)wellformedness anyway. The apparent variety of parameters certainly does not facilitate crosslinguistic comparison and typological generalization. Indeed, since very few investigations have been carried out so far, it is premature to guess what patterns might arise from a wide crosslinguistic coverage.

250

Yaron Matras and A. Machtelt Bolkestein

Appendix List of abbreviations of Classical Latin authors and works quoted Caes. Civ. Caes. Gal. Cato Agr. Cic. Att. Plaut. Rud. Plin. Nat.

C. Iulius Caesar De Bello Civili C. Iulius Caesar De Bello Gallico M. Porcius Cato De Agri Cultura M. Tullius Cicero Epistulae ad Atticum T. Maccius Plautus Rudens C. Plinius Secundus Naturalis Historia

Notes 1.

2. 3. 4.

5. 6.

7.

8.

When referring back to a segment of the discourse, the size of the unit referred to is not always unambiguously determinable. Conte (1996) uses the term “encapsulation” for instances in which a segment of the discourse (a proposition or a paragraph) is alluded to by a noun phrase (with or without an anaphoric determiner) such as ‘this idea,’ ‘this claim,’ etc. Languages differ in the degree to which they have abstract nouns available for doing this. The term “indirect anaphora” is also used for other types of circumstances where the anaphoric expression does not have one specific, uniquely identifiable antecedent referent, such as the much discussed “pronouns of laziness.” In what follows, we quote the nominative of the masculine forms of these pronouns rather than repeating the feminine and neuter forms each time. We will not go into reflexive pronouns here. Occasional observations made in this paper concerning the free relative connection are based on the criterion of (i) not being embraced by the clause containing the antecedent, and (ii) discontinuity between the antecedent and the relative pronoun by at least one other constituent, as in Bolkestein (1996). The corpus on which these tendencies are based is described in detail in Bolkestein and Van de Grift (1994). We use the terms topicality and focality as discourse determined notions, without preconceptions about possible systematic coding in the language under consideration (as in Bolkestein 1998), that is, as different from the notion “sentence topic” as used in Maslova and Bernini (this volume). Such combinations of a deictic pronoun with a deictic adverbial are also found in Swedish and Danish. Combinations of deictic pronoun and deictic adverbial also appear some Romance languages, cf., for example, (Northern) Italian questo (libro) qui ‘this (book) here’, quel (libro) là ‘that (book) there’. They appear to be lexicalized in French celui-ci, celui-là. In spoken Welsh (MacAuley 1992: 292) a postnominal deictic adverbial functions as a demonstrative, cf. y din yma ‘ART man here’ for ‘this man’, y dinion yna ‘ART men there’ for ‘these men’. In Dutch a combination of the distal pronoun with a proximate adverb die : : : hier seems to be more acceptable than a proximate pronoun with a distal adverb ?deze: : : daar. Although in Dutch the relative pronoun die/dat is morphologically identical to the

Deixis and anaphora: Some case studies

251

distal demonstrative, the main clause word order (Vf-2) makes it perfectly clear that we are dealing with a main clause.

References Anderson, Stephen R., and Edward L. Keenan 1985 Deixis. In Language Typology and Syntactic Description III, Timothy Shopen (ed.), 259–308. Cambridge: Cambridge University Press. Ariel, M. 1988 Referring and accessibility. Journal of Linguistics 24: 65–87. 1990 Accessing Noun-phrase Antecedents. London: Routledge. 1991 The function of accessibility in a theory of grammar. Journal of Pragmatics 16: 443– 463. 1994 Interpreting anaphoric expressions: a cognitive versus a pragmatic approach. Journal of Linguistics 30: 3–42. Berretta, Monica 1990 Catene anaforiche in prospettiva funzionale. Rivista di Linguistica 2: 91–120. Bolkestein, A. Machtelt 1996 Is qui et is? On the so called free relative connection in Latin. In Aspects of Latin, H. Rosén (ed.), 553–566. Innsbruck: Innsbrucker Beiträge zur Sprachwissenschaft. 1998 What to do with Topic and Focus? Evaluating pragmatic information. In Functional Grammar and Verbal Interaction, Mike Hannay and A. Machtelt Bolkestein (eds.), 193– 214. Amsterdam/Philadelphia: Benjamins. 2001 Discourse organization and anaphora in Latin. In Textual Parameters in Older Languages, S. Herring, Piet Van Reenen, and L. Schoesler (eds.), 107–138. Amsterdam: John Benjamins. Bolkestein, A. Machtelt, and Michel Van de Grift 1994 Participant tracking in Latin discourse. In Linguistic Studies in Latin, J. Herman (ed.), 283–302. Amsterdam: John Benjamins. Bühler, Karl 1934 Sprachtheorie. Stuttgart: Fischer. Chafe, Wallace 1994 Discourse, Consciousness and Time. Chicago: University of Chicago Press. Cornish, Francis 1986 Anaphoric Relations in English and French: A Discourse Perspective. London: Croom Helm. Conte, Maria-Elisabeth 1996 Anaphoric encapsulation. In Coherence and Anaphora, Walter de Mulder and Liliane Tasmowski (eds.). Monographic issue of the Belgian Journal of Linguistics 10: 1–10. De Jong, Jan R. 1996 The borderline between deixis and anaphora in Latin. In Aspects of Latin, H. Rosén (ed.), 499–510. Innsbruck: Innsbrucker Beiträge zur Sprachwissenschaft. 1998 Deictic and (pseudo-)anaphoric functions of the pronoun iste. In Latin in Use. Amsterdam Studies in the Pragmatics of Latin, R. Risselada (ed.), 19–35. Amsterdam: Gieben. Dik, Simon C. 1989 The Theory of Functional Grammar. Part 1: The Structure of the Clause. Dordrecht: Foris.

252

Yaron Matras and A. Machtelt Bolkestein

Ehlich, Konrad 1979 1982

Verwendungen der Deixis beim sprachlichen Handeln. Linguistisch-philologische Untersuchungen zum hebräischen deiktischen System. Frankfurt: Lang. Anaphora and deixis: same, similar, or different? In Speech, Place and Action, R. J. Jarvella and W. Klein (eds.), 315–338. Chichester: John Wiley & Sons. Deixis und anapher. In Essays on Deixis, G. Rauh (ed.), 79–97. Tübingen: Narr.

1983 Fox, Barbara A. 1987 Discourse Structure and Anaphora. Cambridge, MA: Cambridge University Press. Geerts, G., W. Haeseryn, J. De Rooij, and M. C. Van den Toorn (eds.) 1984 Algemene Nederlandse Spraakkunst. Groningen: Wolters-Noordhoff. Givón, Talmy 1983 Topic continuity in discourse: an introduction. In Topic Continuity in Discourse: A Quantitative Language Study, T. Givón (ed.), 1–42. Amsterdam: John Benjamins. Gundel, Jeanette K., Nancy Hedberg, and Ron Zacharski 1993 Cognitive status and the form of referring expressions in discourse. Language 69: 274– 307. Hannay, Mike 1985 Inferrability, discourse-boundness and sub-topics. In Syntax and Pragmatics in Functional Grammar, A. Machtelt Bolkestein, Casper De Groot, and J. Lachlan Mackenzie (eds.), 49–63. Dordrecht: Foris. Himmelmann, Nikolaus 1996 Demonstratives in narrative discourse: a taxonomy of universal uses. In Studies in Anaphora, Barbara Fox (ed.), 205–254. Amsterdam/Philadelphia: John Benjamins. Hoffmann, Maria E. 1989 A typology of Latin theme constituents. Cahiers de l’institut de linguistique de Louvain 15: 185–196. Hofmann, Johann, and Anton Szantyr 1965 Lateinische Syntax und Stilistik. Munich: Beck. Kirsner, Robert 1979 Deixis in discourse: an exploratory quantitative study of the modern Dutch demonstative adjectives. In Discourse and Syntax, T. Givón (ed.), 355–375. New York: Academic Press. Kirsner, Robert S., and Vincent J. Van Heuven 1988 The significance of demonstrative position in modern Dutch. Lingua 76: 209–248. Kirsner, Robert S., Vincent J. Van Heuven, and J. F. M. Vermeulen 1987 Text type, context and demonstrative choice in written Dutch: some experimental data. Text 7: 117–144. Kuehner, Raphael, and Carl Stegmann 1912–1914 Ausführliche Grammatik der lateinischen Sprache II: Satzlehre. Hannover: Hahnsche Buchhandlung. Lambrecht, Knud 1994 Information Structure and Sentence Form. Cambridge: Cambridge University Press. Lavency, Marius 1981 La proposition relative du latin classique. l’Antiquité Classique 50: 455–468. Lyons, John 1977 Semantics. Cambridge: Cambridge University Press. MacAuley, Donals (ed.) 1992 The Celtic Languages. Cambridge: Cambridge University Press. Maes, Alfons A. 1994 Nominal Anaphora and the Coherence of Discourse. Landegem: Nevelland.

Deixis and anaphora: Some case studies Matras, Yaron 1994 1998

253

Untersuchungen zu Grammatik und Diskurs des Romanes (Dialekt der Kelderasa/Lovara). Wiesbaden: Harrassowitz. Deixis and deictic oppositions in discourse: evidence from Romani. Journal of Pragmatics 29: 393–428.

Myhill, John 1992 Typological Discourse Analysis. Oxford: Blackwell. Pennell Ross, Deborah 1996 Anaphors and antecedents in narrative text. In Aspects of Latin, H. Rosén (ed.), 511– 525. Innsbruck: Innsbrucker Beiträge zur Sprachwissenschaft. Pinkster, Harm 1987 The use of subject pronouns in Latin. In Études de linguistique générale et de linguistique latine offertes en hommages a Guy Serbat, 369–379. Paris: SIG. Prince, Ellen F. 1981 Towards a taxonomy of given – new information. In Radical Pragmatics, Peter Cole (ed.), 223–255. New York: Academic Press. Sidner, C. L. 1983 Focussing in the comprehension of definite anaphora. In Computational Models of Discourse, M. Brady and R. C. Berwick (eds.), 267–330. Cambridge, MA: MIT Press. Somers, Maartje H. 1994 Theme and topic. The relation between discourse and constituent fronting in Latin. In Linguistic Studies in Latin, J. Herman (ed.), 151–163. Amsterdam: John Benjamins. Touratier, Christian 1980 La relative. Essai de théorie syntaxique. Paris: Klincksieck. 1994 Syntaxe Latine. Louvaine-la-Neuve: Peeters. Vonk, W., L. G. M. M. Hustinx, and H. G. Simons 1992 The use of referential expressions in structuring discourse. Language and Cognitive Processes 7: 301–333. Wehr, Barbara 1984 Diskurs-Strategien in Romanischen. Ein Beitrag zur romanischen Syntax. Tübingen: Gunter Narr Verlag.

Hans-Jürgen Sasse

Theticity

1. Introduction 1.1. The phenomenon The subject matter of this chapter is the semantic, syntactic, and discourse-pragmatic background as well as the crosslinguistic behavior of types of utterance exemplified by the following English sentences and their translation equivalents in other European (and some non-European) languages:1 (1)

(2)

My NECK hurts.2 Albanian: Basque: Bulgarian: French: German: Hungarian: Italian: Japanese: Modern Greek: Polish: Russian: Serbo-Croatian: Spanish: Irish:

Më dhemb GRYKA. Mina dut lepoan.3 ˘ Boli me GARLOTO. J’ai mon COU qui me fait MAL.4 Mein HALS tut weh/Mir tut der HALS weh.5 Fáj a TORKOM. Mi fa male il COLLO. KUBI ga ITAI. Ponai o LEMOS mu. GARDŁO mnie boli/Boli mnie GARDŁO. Bolet GORLO. Boli me VRAT/GRLO. Me duele el CUELLO. Tá pian i mo SCORNACH/Tá mo SCORNACH nimhneach.6

The PHONE’s ringing. Albanian: (Po) bie TELEFONI. Basque: TELEFONOAK jo du.7 ˘ zv˘ani. Bulgarian: TELEFONAT French: Y’a le TELEPHONE qui SONNE. German: Das TELEFON klingelt. Hungarian: Csöng a TELEFON. Italian: Squilla il TELEFONO.

256

Hans-Jürgen Sasse

Japanese: Modern Greek: Polish: Serbo-Croatian: Spanish: Irish:

DENWA ga NATTE iru yo. Xtipai to TILEFONO. TELEFON dzwoni. Zvoni TELEFON. Suena el TELÉFONO. Tá an FÓN ag ringáil.8

Sentences such as those listed in (1) and (2) are usually held to stand in opposition to sentences with a topical subject. The difference is said to be formally marked, for example, by VS order vs. topical SV order (as in Albanian po bie telefoni ‘the phone is ringing’ vs. telefoni po bie ‘the phone is ringing’), or by accent on the subject only vs. accent on both the subject and the verb (as in the English translations). The term “theticity” will be used in the following to label the specific phenomenological domain to which the sentences in (1) and (2) belong. It has long been commonplace that these and similar expressions occur at particular points in the discourse where “a new situation is presented as a whole.” We will try to depict and classify the various discourse situations in which these expressions have been found in the different languages, and we will try to trace out areas of crosslinguistic comparability. Finally, we will raise the question whether or not there is a common denominator which would justify a unified treatment of all these expressions in functional/semantic terms. For a number of reasons, the term “thetic” has not become exceedingly popular in certain parts of the scientific community. The main problem seems to lie in its nonlinguistic background, rooted in the philosophical tradition of the late nineteenth century, in particular, in the writings of Franz Brentano and Anton Marty (cf. Section 1.2 below). The term has thus received a “non-empirical smell” which many “ordinary working linguists” do not like (cf. Sornicola 1995). Moreover, the term is closely connected with the idea, repeatedly expressed in philosophical circles, that a “thetic judgment” is a cognitive operation sui generis, reflecting an act of predication sharply distinct from other types of predication such as the traditional bipartite Aristotelian subject-predicate judgment. This view is not shared by many empirical linguists. However, this is not the place to enter this controversial discussion. We will refrain from too far-reaching fundamental speculations and confine ourselves here to the description and crosslinguistic classification of certain utterance types and their usage. We have nevertheless decided – with certain provisions to be discussed in the sections below – to stick to the term “thetic” in favor of other terms that might suggest themselves (such as “all-new” or “ sentence focus”). We will therefore use this term throughout the present chapter. The justification for this decision lies mainly in the fact that it is a handy term, most widely known in many linguistic circles and thus, if stripped of its philosophical touch, most appropriate in serving as a neutral

Theticity

257

term for an integrative approach bringing together insights from different sources such as envisaged here.

1.2. Previous research We will not give a complete account of the history of research on the phenomenological domain called theticity here. For this purpose, the reader is referred to works such as Wehr (1984), Ulrich (1985), Sasse (1987), Haberland (1994), and Lambrecht (1994). In what follows, we will confine ourselves to a brief sketch of the main issues characterizing the research situation from which the present study proceeds. The linguistic analysis of sentences such as those cited in (1) and (2) above has followed a number of independent and quite different traditions. The point of departure for most data-oriented grammarians lies in the form of these utterances, which was considered to be “marked” vis-à-vis other types of clause structure in the respective languages. For instance, in the Italian, Spanish, Russian, Serbo-Croatian, Albanian, Modern Greek, and Hungarian examples, to mention just a few, the word order is predicate + subject, while the “normal” or “unmarked” word order for a declarative sentence in these languages is usually considered to be subject + predicate. In the English, German, Dutch, and Polish examples, the subject is markedly stressed and forms a closely-knit intonation unit with the predicate, in contrast to the “normal” intonation pattern found in the declarative sentences of these languages, which is said to have a stressed predicate. In French and Irish, a special construction is used which is optionally introduced by an existential expression followed by the subject and presents the predicate in a relative clause. In Irish, but not in French, this is also coupled with a difference in word order; the construction is called “abnormal order” by MacCana (1973) since the “normal” word order in Irish is verb-subject. Studies proceeding in this way – from the form of the utterances – have usually attempted to explain the formal differences in terms of one or the other functional or semantic framework. The relevant constructions have most frequently been explained in terms of “activation state” (given vs. new) as “all-new-utterances,” “news sentences,” “neutral descriptions,” or “entirely rhematic,” that is, roughly, as expressions containing no given element or, more precisely, as expressions in which both the subject and the predicate are new (Schmerling 1976; Kuno 1972; Wehr 1984; Krylova and Khavronina 1988; and many others). The problem with these terms (and the concepts usually hidden behind them) is, first of all, their commitment to a research tradition on information structure often operating with comparably illdefined concepts such as “given” and “new”. The functional explanations found in this kind of work are therefore often rather vague; it remains unclear what notions such as “rhematic,” “neutral,” etc., really mean and it has been disputed that “allnew” utterances really consist of entirely “new” elements given the fact that they

258

Hans-Jürgen Sasse

often contain pronouns and other elements readily discoverable from the preceding text. In summary, the notions employed in these circles have been either unsatisfactorily ambiguous (“neutral”) or confusing (“new” = “not aforementioned”?). Moreover, Chafe (1974) observed that, for simple subject-predicate sentences, two different intonation patterns are possible which do not seem to correspond to any differences in the activation state of the constituents involved.9 In all sentences in (3) and (4), both constituents are conceived of as conveying new information: (3)

a. b.

My SISTER is DYING. My SISTER died.

(4)

a. b.

The BUTTER MELTED. The BUTTER melted.

He explains the difference as follows: “It seems likely that the verb-noun combination of [the examples in b] form for the speaker a conceptual unity which is not present in the combinations of [the examples in a]” (Chafe 1974: 115). A similar approach was taken by Fuchs (1976, 1980), who coined the term “integration” to label a situation in which “the whole syntagm is introduced as one unit of information, ‘integrating’ its parts into one ‘globally new’ unit” (Fuchs 1980: 449).10 The term “integration” was recently taken up again by Jacobs (1992). The “integrative” nature of these utterances also constituted the point of departure for certain approaches stressing their “focal” character, a view expressed as early as in Müller-Hauser (1943) who uses the expression “mise en relief de la phrase entière.” “Presentational sentences,” “sentence focus,” and “event-reporting sentences” are further examples of terms used in this connection. Important studies along these lines are Lambrecht (1987) and, in particular, Lambrecht (1994); see also Drubig (1992). A different line of research was followed by Allerton and Cruttenden (1979), who investigated subject-accented sentences in English and claimed that they fall into three categories, characterized by the nature of the verb: (1) semantically empty, predictable predicates, (2) verbs of appearance and disappearance, and (3) verbs denoting a misfortune. Empty verbs are often of the kind which state the existence of their subjects by naming a typical state of affairs or activity which characterizes these subjects. Sentences like: (5)

a. b.

The SUN is shining. NIGHT is falling.

merely describe the existence or appearance of their subjects (‘there is sunshine’, ‘night appears’). For the most part, then, subject-accented sentences in English are existential sentences. Those which are not covered by “existence” are explained by Allerton and Cruttenden in terms of the semantic closeness of the verbs to the subjects they accompany:

Theticity

(6)

a. b. c.

259

The KETTLE’s boiling. The TELEPHONE’s ringing. The TAP’s leaking.

“In such cases we see that a kettle is capable of very few verbal activities other than boiling, that a telephone can do little except ring, etc.” (Allerton and Cruttenden 1979: 52). An approach which explains English and German subject-accented constructions exclusively in terms of the “existential” lexical semantics of the verbs involved has been rejected by several scholars, for example, Krifka (1984, for German), and Sasse (1987, for English), though similar statements have also been made for other constructions in other languages. In particular, VS sentences in Romance languages have often been analyzed in terms of the existential semantics of their verbs: “The verbs : : : tell us only or mainly that the subject exists or is present; is absent; begins; continues; is produced; occurs; appears; arrives” (Hatcher 1956: 7, on Spanish VS sentences). The existential character of Romance VS sentences had already been assumed by Blinkenberg (1928) for French. This was reaffirmed for French by Atkinson (1973) and Jonare (1976), and was confirmed for Italian by Lonzi (1974) and Wandruszka (1982). Quite independent of the approaches sketched so far, there had been a very early philosophical approach to the subject which did not proceed from linguistic considerations about the semantics of certain constructions but from logical considerations about the nature of predication. This is the approach from which the notion of “theticity” originated. The term was first introduced by Brentano (1874; to the English-speaking world best known as Brentano 1973) and was originally conceived of as part of a dichotomy “thetic vs. categorical,” which was held to reflect a “logical” distinction between two types of human judgment. Hence, Brentano and Marty do not speak of thetic and categorical utterances (implying that there could be some formal manifestation of the “logical” dichotomy11), but of thetic and categorical judgments. In this view, the “categorical judgment” is the traditional bipartite Aristotelian type of judgment consisting of a (logical) subject and a (logical) predicate, while the monolithic “thetic judgment” simply involves “the recognition or rejection” of the “material of a judgment”. As Ladusaw (1994, paraphrasing Kuroda 1972) puts it, the categorical judgment has a “presupposed” subject in the sense that a precondition for making the judgment is that “the mind of the judger must be directed first to an individual, before the predicate can be connected to it.” In other words, with a categorical judgment one says something about an entity, whereas the thetic judgment simply “poses” (hence “thetic” from Ancient Greek tith¯emi ‘to put, pose’) the existence of a certain state of affairs. This strong philosophical viewpoint, which regards the thetic/categorical distinction as a cognitive phenomenon sui generis, reflecting two radically different types of predication, is usually rejected today.12

260

Hans-Jürgen Sasse

Nonetheless, a number of less philosophically oriented linguistic adaptations of the Brentano-Marty approach have since appeared. The earliest I know of is Mathesius (1929), who claims that there are two basic types of “assertiveness,” one which is a simple presentation of an event (thetic), and one which has a topic-comment structure with an asserted comment set off from a presupposed topic (categorical). After a long period of silence, the issue was taken up again by Kuroda (1972), in particular, with respect to the behavior of the Japanese particles wa and ga, followed by Vattuone (1975) with respect to Genoese VS constructions, Ulrich (1985) with respect to VS clauses in Rumanian, and Sasse (1987) in a general typological account. The linguistic adaptations of the Brentano-Marty theory differ from the original philosophical approach in one important aspect. They all try to set up the thetic-categorical distinction as some sort of semantic universal and look for “manifestations” of it in the sentence structure of individual languages. Although this type of approach was rather characteristic of linguistic work in the 1970s and 1980s and has revealed a great number of important insights, it was ultimately doomed to failure because it tended to neglect the more subtle aspects of form-content relations in the individual languages. To sum up our brief historical account so far, we can distinguish one line of research proceeding from the form of utterances in individual languages and setting up “functions,” mostly in terms of information structure, and one line proceeding from the assumption that a rather abstract “universal” notion of monolithic, nonarticulated predication is somehow directly manifested in a certain sentence structure and, vice versa, that this sentence structure “expresses” the respective notion. Not surprisingly, the problem is not only a terminological one; it also pertains to the empirical basis from which the different approaches proceed. It is interesting to note that the sets of examples adduced in the different camps do not always seem to belong to identical phenomenological areas, which is evident from the differing types of sentences cited. Quite understandably, therefore, doubts have been raised as to whether there is really any empirical justification in assuming that theticity is a proper linguistic entity. In Brentano-Marty oriented circles, for instance, theticity has often been demonstrated by problematic examples such as the following: (7)

Inu ga hasitte iru. ‘A dog is running.’ (Kuroda 1972)

(8)

Nobody left. (Ladusaw 1994)

Sentences such as (7) are claimed to have both a “thetic form” (by virtue, for instance, of the particle ga in Japanese) and a “thetic logical interpretation.” However,

Theticity

261

such examples have always been cited in isolation; no indication is given as to where such expressions appear in actual discourse. Sentence (8), in turn, is claimed to be “thetic” on the basis of general logical considerations, although it does not have the form claimed by others to be typical of “thetic” sentences in English (subject accentuation or there is construction). Moreover, when translation equivalents of (8) in other languages are compared (for instance, Modern Greek, cf. Sasse 1995b), focus constructions come into play which render a “thetic” interpretation problematic. It remains doubtful whether sentences with “pronominal” subjects such as (8) are really interpretable as subject-predicate constructions in the same way as, say, in the examples (1) through (6) above. In most of the languages examined, they appear to follow conventionalized patterns and it seems wise to exclude such cases from the examination of productive mechanisms. Moreover, Brentano’s and Marty’s original German examples were mostly confined to existentials and weather verbs (such as those under (9)), whereas “all-new utterances,” “news sentences,” etc., are frequently exemplified by verbs of appearance, loss, or damage, as in (10): (9)

a. b.

(10) a. b.

Es regnet. ‘It is raining.’ Es findet ein Markt statt. ‘A market is being held.’ The British are coming. Johnson died.

It is readily accepted by many researchers that utterances such as those in (9) do not predicate a property of some entity (for instance, raining of it), but simply recognize or “pose” a situation. It is much more controversial, however, to maintain this for the utterances in (10) or for (7), given the fact that there is nothing to suggest that the syntactic and conceptual structure of such sentences fails to reflect a predicative (“aboutness”) relation. The latter seems to be counterintuitive to many people given the fact that the subject-predicate structure, coupled with a subject-predicate interpretation, is still there. What both groups of examples seem to have in common is that both of them are candidates to be used (inter alia) in discourse situations where the scope of the assertion extends across the entire proposition, and this is where the “judgment” approach and the “information structure” approach meet. The usual way of demonstrating broad scope of assertion is the application of tests, such as the wellknown question test with the frame “what’s the matter/what happened?”. All four utterances in (9) and (10) are equally good answers to these questions and as such turn out to be employable as utterances with a broad scope of assertion. Nevertheless, if “broad scope of assertion” is equated with “entirely new information,” there is still a considerable amount of sentences which do not show the expected form. Moreover,

262

Hans-Jürgen Sasse

the test rules out a number of expression types considered to be “thetic” by Brentano and Marty (such as universal statements, cf. Kuroda 1972 and Sasse 1987), but also a number of Romance, Modern Greek, etc., VS clauses which have been claimed to be “all-new utterances” but turn out to have verb focus (cf. Sasse 1995b), and a number subject-accented sentences in German, English, and other languages which turn out to be constructions with narrow focus on the subject. Finally, it has to be noted that different languages cover different sections of the entire phenomenological domain, which means that language-specific investigations which proceed from the form of utterances in single languages are not necessarily comparable from a functional point of view. In other words, if a “function” is generalized on the basis of a single language and then transferred to the analysis of another language, inadequate interpretations may arise. For instance, the description of German and English subject-accented sentences and the comparable constructions in French as “all-new” may not be entirely adequate, but is less inappropriate than a description of Hungarian VS constructions in the same terms, even if all of the constructions are used in superficially similar contexts (cf. Note 9). The crucial factor responsible for all this confusion seems to lie in the illegitimate mixture of formal and semantic considerations. This has led to two types of unfelicitous generalizations in two different directions: for the strong Brentano-Marty adherents, on the one hand, considerations of philosophical semantics (“judgment” structure, existence, quantifier semantics, etc.) usually had priority over form, though formal considerations were sometimes adduced to support the semantic analysis (e.g., by Kuroda 1972 in rejecting universal statements as an instance of “thetic” judgments: this was chiefly done by pointing out that they have a “categorical structure” in Japanese). For syntactic and discourse-oriented researchers, on the other hand, the center of attention had always been the form of the utterances, and the possibility of polysemous constructions, where one form has a number of distinct meanings and uses, was often neglected.

1.3. The present research strategy Given the state of affairs sketched in the preceding paragraph, it is necessary to develop a research strategy capable of coming to terms with two basic requirements: first, it should handle the form-function problem in a much more sophisticated manner, and second, it should guarantee comparability across languages. The two requirements are closely interconnected. One of the main faults of previous research on theticity (including my own) was failing to clearly distinguish between form and content as such, and, more specifically, between universal and language-specific aspects of both.13 Basically, nobody has ever tried to make explicit what kind of animal “theticity” really is and on what level of linguistic analysis it

Theticity

263

Table 1. Some prominent approaches to the thetic/categorical distinction Approach

Interpretation of distinction

Domain to which distinction is attributed

Original thetic/categorical approach Modified thetic/categorical approach

two cognitively different types of human “judgment” two different viewpoints manifested in different predicative structures different constellations of activation state

cognitive or “logical”

Activation state approach

Focus/background approach

difference in scope of assertion

Lexical approach

triggered by lexical properties

discourse/syntax; topic-comment articulation discourse/syntax; information value (“given”, “new”) discourse/syntax; focus-presupposition articulation lexicon

has to be dealt with. The explanations discussed in the previous section strongly disagree on the linguistic domain to which the different distinctions proposed should be attributed. Table 1 summarizes the interpretation of the distinctions and the domains to which the distinctions are attributed in the different approaches. Furthermore, the phenomena subsumed under the different labels are not necessarily always the same, as we have already seen. This is due to the fact that most authors have taken it for granted that “thetic,” “focus,” “all-new,” etc., are universally definable linguistic functions and that there are clearly identifiable (and crosslinguistically comparable) linguistic structures serving these functions in a straightforward way. This has led to considerable confusion of structural and semantic criteria, resulting in the neglect, to a large extent, of polyfunctionality and of subtle, language-specific constraints governing the use of the constructions in question. The methodology or, to put it more modestly, the heuristic strategy pursued here will therefore take previously accumulated knowledge about the phenomenological domain of theticity as a point of departure, without, however, postulating any general or “universal” functional concept in terms of which the entire domain would be defined. Instead, the domain will be tackled from all the different angles that have so far been subject of the relevant studies: the language-specific constructions involved, their place in the general language-specific pool of available constructions, ambiguities and oppositions, usualization and obligatorization, the use of the constructions in actual discourse, the language-specific constraints on the use of these constructions, and the interaction between lexical semantics and constructions. We will then

264

Hans-Jürgen Sasse

attempt to filter out areas of crosslinguistic comparability and, finally, try to arrive at proper typological generalizations. Preliminary research along these lines was done between 1992 and 1994 by members of Theme Group 1 of Eurotyp and resulted in Matras and Sasse (1995), henceforth called the “VS study.” In this study, a fine-grained analysis of verb-subject constructions in a number of European languages was presented. The approach was primarily semasiological and was based on the examination, in part quantificational, of the occurrence of verb-subject constructions in real texts of different text types. The VS study was extremely helpful in preparing the research for this paper since it yielded a number of important parameters along which other constructions which had played a role in the discussion of theticity could be analyzed. In the following, we will draw heavily on the results of this study; its findings will be the chief point of departure here.

2. Constructions 2.1. Formal types of “special” constructions in the domain of theticity In the literature on “thetic”/“all-new”/“presentational” etc. expressions, the following formal construction types are identified as relevant: 1. Highly accented subject + immediately following enclitic-like low-toned verb for “thetic”/“all-new”/“presentational” etc. utterances, as opposed to double accent on both subject and verb or accent on verb only for “categorical”/“topic-comment” etc. utterances. The alleged opposition can be exemplified by the English and German sentences under (11) and (12) below (cf. Lambrecht 1987, 1994). This type is called “prosodic inversion” by Lambrecht (1994: 318 ff.); following Allerton and Cruttenden (1979) we will use the term “subject accentuation” (henceforth, SAcc). (11)

Was ist los? What’s the matter?

Mein Auto ist kaputt. My car broke down.

(12)

What happened to your car? Was ist mit deinem Auto los?

My car broke down./My car broke down. Mein Auto ist kaputt./Mein Auto ist kaputt.14

2. Verb-subject order for “thetic” etc. utterances as opposed to subject-verb order for “categorical” etc. utterances. This opposition may be exemplified by the Italian sentences in (13). Similar oppositions have been claimed to occur in a great number of European as well as non-European languages; for a general survey, the reader

Theticity

265

is referred to Sasse (1987) and Ulrich (1985). This type will be referred to by the abbreviation VS. (13) a. b.

Mi si è rotta la macchina. ‘My car broke down.’ La macchina si è rotta. ‘The car broke down.’

3. The construction “subject + relative clause” (normally in the scope of an existential or deictic marker; if the subject is introduced by ‘have’ as an existential marker, it becomes an object) for “thetic” etc. utterances in opposition to various other structures found in “categorical” etc. utterances. This type was discussed under the label “split structure” in Sasse (1987: 538 ff.). In connection with Welsh, MacCana (1973) referred to it as “abnormal sentence.” In the following, we will use the abbreviation “Split” for “split structure.” The type is exemplified by French in examples (14)– (19), which demonstrate different variants depending on the presence and the nature of the existential/deictic markers. (14) a. b.

J’ai le COU qui me fait MAL. ‘My neck hurts.’ Mon cou il me fait MAL. ‘My neck hurts.’

(15)

Il y a un TUYAU qui fuit. ‘There’s a pipe leaking.’

(16)

Qu’est-ce qu’il y a? – C’est MAMAN qui me bat. ‘What’s the matter? - Mum’s hitting me.’

(17)

La MAISON qui brûle. ‘The house is burning.’

(18)

Voilà la SIRÈNE qui hurle. ‘The siren is wailing.’

(19)

Nous avons une invitation ce soir, et ma FEMME qui est malade! ‘We are invited out this evening; but my wife is sick!’

4. Subject incorporation (henceforth, SInc). In a “thetic” etc. utterance, the subject of the corresponding “categorical” etc. utterance is incorporated into the verb. Clear cases of morphological incorporation, that is, in the form of a real compound, have so far been found only in some Native American languages, for example, Iroquoian (cf. Sasse 1987: 548 ff.). They are rare in these languages and seem to be confined to very short existential statements.15 What is more frequent is some sort of “pseudo-incorporation” (Mithun’s incorporation type I, i.e., a closely-knit combination of noun and verb with no intervening material allowed, cf. Mithun 1984);

266

Hans-Jürgen Sasse

however, the noun is not fully incorporated, in the narrow sense, in that it does not entirely lose its word status. It is often very difficult to distinguish this from the SAcc cases. This type of SInc is exemplified in (20) below by a sentence from Boni (Eastern Cushitic), cf. also Sasse (1987: 544 ff.) for further examples. (20) a.

b.

c.

Რddi˛gé˛ e˛ -juudi father:1SG.POSS-die:3SG.M.PERF ‘My father died.’ Რddi˛gé˛ e˛ á juudi father:1SG.POSS VF-die:3SG.M.PERF ‘My father has died/is dead.’ (verb focus) Რddi˛gé˛ e˛ é juudi father:1SG.POSS NF die:3SG.M.PERF ‘It’s my father who died.’ (subject focus)

Three further types of relevant constructions may be mentioned in passing, since they do not seem to occur in European languages and are therefore not dealt with in the following: (1) verb nominalization, usually preceded by existential markers, for “event-central” types of utterances (cf. Sasse 1987: 552 ff.); (2) particles, as in Japanese (cf. Kuroda 1972); (3) special morphology for “thetic” predicates; cf. Sasse (1987: 553) on Tagalog; special verb morphology is also said to be attested in a number of lesser-known languages (Haiim Ben Rosén, pers. comm.). The occurrence of introductory existential markers with otherwise “unmarked” sentence types, which is reported for several languages, has also to be examined in the present connection. Something similar also occurs in English (cf. Lambrecht’s [1988b] There was a farmer had a dog).16 The four constructions exemplified in points 1–4 above regularly occur in European languages and are reported to be characteristic of utterances in the domain under consideration. SAcc seems to be by far the most frequent, closely followed by VS. As far as the latter is concerned, a distinction has to be made between languages which allow verbs in sentence-initial position (the “SV/VS alternating type” of Hopper 1985), and those in which the position before the verb has to be filled with something else when subject-verb inversion occurs (XVS). All languages of the SV/VS alternating type also allow XVS constructions. European languages of the SV/VS alternating type seem to be the following: Albanian, Bulgarian, Czech (marginal), Hungarian, Italian, Modern Greek, Polish (marginal), Portuguese, Rhaeto-Romance, Rumanian, Russian, Serbo-Croatian, and Spanish. Spoken German also allows VS with certain specific functions (cf. Section 4). XVS occurs in Dutch, English (marginal), French, German, and several other Germanic languages; it was also characteristic of earlier stages of several Romance languages. Latin was also of the SV/VS alternating type (cf. Bolkestein 1995).

Theticity

267

Several scholars have pointed out that the VS construction, especially when in the thetic function, is often characterized by lack of agreement between the initial verb and the subject, cf. French:17 (21)

Il est arrivé des bonnes nouvelles. ‘Good news have arrived.’

In Genoese, the initial predicate regularly has neither number nor gender agreement; it always stands in the third person singular masculine (Vattuone 1975). Lack of agreement is also characteristic, though not obligatory, in VS constructions in Vulgar Latin and Italian (cf. Wehr 1984, 1995). Split is said to be the most prominent construction in the thetic domain in French (Lambrecht 1987, 1988a, 1994; Wehr 1984), though it also marginally occurs in other Romance languages (Italian, Rhaeto-Romance, Spanish, and Catalan). The other European language of which it is highly characteristic is Welsh, cf. MacCana (1973) and Wehr (1984). As far as European languages are concerned, SInc constructions have so far only been described for Danish.18 The following is a Danish example of an SInc construction from Nedergaard Thomsen (1991): (22)

Der kom-nogle-fremmede til byen. there came-some-strangers to town ‘Some STRANGERS have come to town.’

Here, the sequence kom-nogle-fremmede is a phonological word. The type is strikingly reminiscent of the Boni case cited above: the NP is not really incorporated into the verb in the sense of a noun-verb compound since all the elements involved retain their morphological autonomy (i.e. they remain nominal, verbal, etc., constituents and do not become parts of compounds); nevertheless, it constitutes a formal amalgam with the verb both intonationally and structurally (no intervening elements).

2.2. Distribution of formal types across languages Almost all languages have more than one of the four types described in the preceding section. The distribution of types across selected European languages is given in Table 2 below. A note on Hungarian: in Hungarian, the “subject focus construction” (accented subject in the immediately preverbal position = focus position) participates in the domain of theticity in complementary distribution with VS. This could be seen as a case of SAcc: the general formal make-up of the construction (apart from the restrictions concerning definiteness and to be discussed below) is closely related to German, English, etc. SAcc constructions: accented subject, toneless verb. As in German and English, the Hungarian construction is ambiguous as to a “thetic” and a

268

Hans-Jürgen Sasse

Table 2. Distribution of “special” constructions across European languages Language Albanian Basque Bulgarian Catalan Czech Danish Dutch English French German Standard Vernacular Hungarian Italian Latin Modern Greek Polish Portuguese Rhaeto-Romance Rumanian Russian Serbo-Croatian Spanish Turkish Welsh/Irish a b

SAcc

VS

XVS

Split

SInc

marginal C C C C C C C C

C marginal C C marginal    Ca

C C C C C C C marginal C

   marginal     C

     C   

C C cf. note marginal ? marginal C C C marginal marginal marginal C C 

 C C C C C marginal C marginal C C C C  

C C C C C C C C C C C C C  

   marginal    marginal marginal    marginal  C

  cf. note            

Only in literary style. In insular Celtic, VS is the “unmarked” word order. It is not normally employed for “thetic”/“allnew” etc. utterances.

“subject focus” reading. It seems, however, that the combination of subject and verb is closer in Hungarian than in German and English SAcc constructions. Moreover, the Hungarian construction is part of the general focus mechanism of this language and the focus position is always the immediately preverbal position. In other words, the construction in Hungarian, though it superficially resembles SAcc, has a quite different status in the grammatical system. It could, however, very well be seen as a case of SInc in the loose sense used here and has in fact been so described in the literature.

Theticity

269

The general typological picture emerging from Table 2 is quite straightforward as far as the areal distribution of the constructions is concerned. A number of languages clearly stand out as “deviant”: Turkish is the only language which does not allow any type of verb-subject order, and Danish is the only language with SInc. Both are located on the margins of the area. Otherwise, the area is clearly divided into two larger sub-areas: roughly, a northwestern and a southeastern one. Although the SAcc construction is present in almost all of the languages, there are typical areal differences in how prominent it is in competition with other devices. It is only marginal in eastern and southeastern languages where VS is dominant. The farther to the West one proceeds, the more prominent it becomes. In Czech and Polish, it is clearly in competition with VS; still more so in (Standard) German where VS is absent and XVS is subject to the strongly grammaticalized “verb-second” strategy. In Dutch, it is in competition with the er-construction (a formal equivalent of the English there is construction), but the latter seems to be more prominent in this language. On the western and northwestern fringes of the area, on the other hand, it is absent (with the exception of English). The VS construction occupies a contiguous area covering Romance, Slavic, and Balkan languages with a strong concentration in the Balkans (with the exception of Bulgarian) where it has the broadest range of functions (cf. Section 3). The Hungarian case is a bit special due to the “focus configurationality” of this language (cf. Section 2.5.2); otherwise, it neatly fits into the eastern/southeastern pattern. It is interesting that Basque, in spite of its close contact with Spanish, is different in that it prefers the SAcc construction except for a number of well-defined lexical areas such as pain; it is very similar, in this respect, to Czech and Polish. The Split construction is confined to a smaller area in the West and is really prominent only in French and Celtic. However, it once may have had a wider distribution (cf. Wehr 1984).

2.3. Text frequency of “special” constructions It is often stated in the literature that the four “special” constructions under examination here are “special” or “marked” not only with respect to their functions (which is a problematic idea at any rate since we do not know what an “unmarked” function is) but also with respect to their text frequency in the respective languages. This is, in part, true; however, the picture is much more differentiated than usually assumed. In particular, the VS study has revealed that the frequency of VS constructions as opposed to (topical) SV constructions varies considerably according to the text type. Quantificational studies have been made for Russian, Italian, Spanish, Hungarian, and Modern Greek. A striking difference was found between spoken and written language. In Hungarian and Modern Greek, the percentage of VS clauses was signif-

270

Hans-Jürgen Sasse

Table 3. VS/SV proportion according to text type in Modern Greek

Advertisements Headlines Announcements (weekly magazine) Announcements (newspaper) Spoken narrative Crime story Modern novel

VS (%)

SV (%)

85.0 73.0 39.0 36.8 34.0 29.0 22.0

15.0 27.0 61.0 63.2 66.0 71.0 78.0

Table 4. VS/SV proportion in spoken vs. written text in Hungarian

Spoken conversation Written text (novel)

VS (%)

SV (%)

32.2 10.8

67.8 89.2

icantly higher in spoken than in written texts (Sasse 1995b: 145, 171). On the other hand, Miller (1995) reports that the highest figures were obtained in academic texts and in novels, the lowest in dialogues. Such differences are probably due to the fact that the different senses (cf. Section 2.4) of VS constructions were not distinguished in the quantificational studies, with the result that the different functional “range” covered by the individual languages yielded different figures of occurrence. Moreover, specific subtypes of VS constructions are restricted to or at least predominantly found in specific text types. If one proceeds from a finer-grained classification of text types, one will find tremendous differences even within the same “macro-genre.” For instance, in all of the modern languages with a literary tradition examined in the VS study, it was found that VS is used as a prominent strategy for headlines, advertisements, and announcements, but much lower figures were found in other sections of the newspapers (this pertains to Russian, Italian, Spanish, Hungarian, and Modern Greek, but also to Serbo-Croatian,19 which was not investigated in the VS study). The general situation is exemplified by figures for Modern Greek and Hungarian in Tables 3 and 4 (taken from Sasse 1995b, the Hungarian figures are adapted from Behrens 1982). The count excluded sentences with no overt subject. The difference between topical and focal subject was not taken into account. Similar figures can be found – mutatis mutandis – for Italian, Spanish, and SerboCroatian; for the situation in Russian, cf. the detailed tables in Miller (1995).

Theticity

271

No statistical investigations of this kind have been made so far for the other types of constructions. Impressionistically, SAcc constructions are rather frequent in German and in English. Nevertheless, this has not been corroborated by statistical data. During the preparatory phase of the VS study some years ago, I counted SAcc occurrences in a video recording of a German TV play with a duration of approximately 45 minutes. The result was a figure of less than 7 % SAcc sentences in comparison to 93 % of sentences covering all the other types of constructions occurring in German sentences. This is considerably lower than the figures obtained for VS constructions in the languages investigated in the VS study (except Russian). Of course, this figure has to be counterchecked against a larger corpus of data and, in particular, against a variety of text types.

2.4. Polysemy of constructions Before dealing with the question of how the different constructions identified above are tied to the domain of theticity, we will briefly examine their polysemy and exclude those of their functions and specific applications which do not seem relevant to the present subject. We will first deal with the VS construction, proceeding from the findings of the VS study. 2.4.1. VS constructions 2.4.1.1. Narrow focus There are at least two narrow focus constellations triggering VS order which have to be carefully separated from the thetic constellations: verb focus and contrastive subject focus. For most of the languages examined in the VS study, it has turned out that verb focus is one of the most prominent functions of VS order. As some of Schroeder’s (1995) examples suggest, this seems to include even Turkish, which is otherwise not a SV/VS alternating language. Other relevant languages of the sample are Russian, Latin, Italian, Spanish, Modern Greek, and Hungarian (Sasse 1995a: 10). To these, Rumanian, Albanian, and Serbo-Croatian may be added (for Rumanian, see Ulrich 1985; Albanian and Serbo-Croatian data come from the author’s own research). What kind of verb focus is expressed by VS is open to further investigation. It is clear that VS does not signal normal “completive” predicate focus found with verbs in a continuous chain of events predicated of a presupposed subject, but often involves a stronger degree of contrastive emphasis usually associated with “polar” or “verum” focus (see Dik et al. 1981 and Höhle 1992) which emphasizes the truth of the relevant predicate. On the other hand, there are many VS cases with the verb bearing a strong accent and therefore likely to represent instances of verb focus, which do not occur in the typical environments of “polar” focus. The exact presup-

272

Hans-Jürgen Sasse

positional/assertional conditions for verb focus thus remain unclear. The tendency for favoring or obligatorizing VS in sentence questions, noted for some languages, may be taken to represent one instance of verb focus. For a detailed analysis of verb focus in Rumanian VS clauses, the reader is referred to Ulrich (1985, in particular pp. 256–284), where the phenomenon is called “Abtönung.” It is often difficult to distinguish verb focus from cases usually called “right dislocation” or “afterthought”; arguments for a clear separation in Rumanian can be found in Ulrich (1985: 225 ff.). Contrastive subject focus was noted in the VS study for Russian, Modern Greek, Latin, Italian, Spanish, and Romani. The only language that does not express contrastive subject focus by VS seems to be Hungarian. There are good reasons for this due to the strongly grammaticalized preverbal focus position in this language.20 Contrasts may be oppositional or suppletive (e.g. with ‘also’); a special subtype of the contrastive focus VS construction, which occurs in several languages, is double contrast with an ‘as-for’ topic in front of the verb and the contrastive subject in postverbal focus position. For further information, the reader is referred to Matras and Sasse (1995). It has to be noted that verb focus VS and contrastive subject focus VS represent two different constructions once the intonational features are taken into account. Verb focus is represented by an intonational type which requires a strong accent on the verb, while subject focus requires the intonational peak on the subject. 2.4.1.2. Inversion It is necessary to distinguish “VS cases proper” from cases of “inversion” triggered by certain preverbal constituents (cf. also Myhill 1986). In Hungarian, this difference has a formal correlate in that certain cases identifiable as inversion allow subjects with zero articles in postverbal position while “VS cases proper” do not (Sasse 1995b: 174, with further references). The role of preverbal material in VS constructions in most of the SV/VS alternating languages is as yet poorly understood. In Modern Greek, Rumanian, and Albanian, at least, focus fronting triggers inversion. The situation in Hungarian is special due to the grammaticalized preverbal focus position (cf. Section 2.5). Bolkestein (1995) cites clear cases of focus fronting with inversion in Latin, but inversion seems to be optional here. In all languages of the VS sample, there seem to be preferences for focus fronting with certain emphatic modal adverbials (so much, very, etc.). A special instance of focus fronting is negation fronting, which occurs in all the languages examined in the VS study, but with differing degrees of grammaticalization. In all cases, the negation is procliticized to the verb. In Hungarian, it consequently occupies the focus position and may therefore be held responsible for the inversion of other material that might occur in this position. Whereas this peculiarity is absent from the other languages, all of them seem to have a tendency for putting negated predicates in front of their subjects, at

Theticity

273

least in short sentences (cf. the discussion for Modern Greek in Sasse 1995b). This is attested even for Latin (Bolkestein 1995: 153). One might speculate that fronting of negated verb forms is a special case of contrastive verb focus, or the negative counterpart thereof. Negative elements other than the negated verb form itself, viz., polar items, negative indefinite pronouns, or adverbials such as nothing, never, etc., negative quantifier phrases, are also often put in clause-initial position and, in this case, trigger inversion. Negation fronting with inversion also occurs in Western European languages which are not of the SV/VS alternating type; for example, it has been obligatorized in certain cases in English as an archaic relic of a former XV word order. A grammaticalized case of focus fronting can be seen in the obligatory preverbal position of interrogatives, which obligatorily triggers inversion in almost all languages investigated in the VS study and many more European languages (except, perhaps, Latin). Note again that the situation in Hungarian, though superficially the same, is different on closer inspection since interrogatives occupy the obligatory preverbal focus position; so, interrogative fronting has a different value in the overall grammatical system here. The inversion of subjects with verbs of saying after direct speech reported for almost all European languages independently of the SV/VS alternating type (even belletristic forms of English) may also best be considered as a fossilized case of inversion due to focus fronting. The fronted stretch of direct speech preceding this type of inversion may be regarded as the focal object of the clause. For further information see Sasse (1995a: 12, with further references). It is difficult to say whether or not adverbials of “setting,” that is, expressions indicating time, place, and circumstance and setting a frame for the following predication, also trigger inversion in some languages.21 In Modern Greek, Russian, Romani, Rumanian, and Albanian, at least, VS is strongly preferred after sentence-initial temporal adverbials including temporal clauses. In other languages of the VS sample, the role of adverbials of setting as triggers of inversion is doubtful (cf. Sasse 1995a: 12). The only language which does not have VS with “setting constituent fronting” at all is Latin (Bolkestein 1995). A final factor which seems to favor inversion is weight. Long subject constituents (so-called “heavy” constituents) have a tendency to be placed after the verb. Examples for such cases in Russian, Modern Greek, Hungarian, Italian, and Latin can be found in Matras and Sasse (1995), though some of the examples are controversial. 2.4.2. SAcc and Split constructions It is commonplace that SAcc constructions are often ambiguous for a narrow subject focus reading and a thetic reading. This is the case, at least, in those languages which use them as a prominent thetic construction. Other languages remain to be investi-

274

Hans-Jürgen Sasse

gated; it may turn out that there are formal differences in certain instances. By the way, the formal coincidence of subject focus and thetic is not uncommon and by no means restricted to SAcc constructions; it is also found in Japanese, which operates with particles (Kuroda 1972). Readings of SAcc constructions other than subject focus and thetic are not known to me. The construction is thus much less ambiguous than VS. As far as Split constructions are concerned, their ambiguity may depend on the subtype. Those introduced by deictic markers (such as French voilà : : :) and have are perhaps the only nonpolysemous thetic constructions found all over Europe. Things become more complicated with those types of Split constructions which have an initial existential or no presubjectal element at all (cf. French examples in Section 2.1), since the same sentence types are used for subject focus (cleft constructions). It has been repeatedly maintained that subject cleft constructions are intonationally different from thetic Split structures (Müller-Hauser 1943, cf. also Wehr 1984); nevertheless, the issue remains controversial since it is doubtful whether the intonational differences can be clearly identified in all cases.

2.5. Language-specific restrictions Each language imposes language-specific restrictions on the usability of its constructions. In the following, we will discuss a number of examples of such restrictions relevant to the four constructions dealt with in this chapter. The account given here is by no means intended as exhaustive. Detailed studies to this effect have only been made for VS constructions in some of the languages of the VS study; several other languages such as German, English, and French are well-investigated and a lot of material is available in the literature; however, the situation in most of the other European languages remains to be further investigated. The discussion will be given here under three headings: global syntactic patterns, monoargumentality, and definiteness and animacy. The role of lexical semantics is discussed in an extra section (Section 5). 2.5.1. Global syntactic patterns Languages differ considerably in the degree of rigidity found in their syntactic organization. This has a strong influence on the manipulability and discourse-functional exploitability of their constructions. There are relatively generous syntactic structures, which impose comparably few constraints on the application of certain constructions. The consequence is, among other things, that the constructions may be freely exploited for discourse-pragmatic purposes. This is probably what has traditionally been understood as “free” or “pragmatically-based” (cf. Mithun 1987) word

Theticity

275

order. Among the languages of Europe, Balkan languages such as Rumanian, Albanian, and Modern Greek may be cases in point. Apart from a small number of obligatorizations, most of the restrictions found in these languages seem to be lexical and/or pragmatically well-motivated. In the other extreme, there are languages with a rigid syntactic organization which imposes a considerable number of restrictions of a conventionalized, grammaticalized nature on their constructions. English basically has a rigid SVO order. Its potential of word order variation for discourse-pragmatic purposes is severely restricted. Mechanisms of exbraciation, left-dislocation, topic fronting are present, but limited. It therefore makes extensive use of cleft constructions and intonational devices. As far as other types of discourse-pragmatically exploitable word order constellations are concerned, VS is possible to a certain extent, provided that the preverbal position is filled with an element other than the subject (i.e., it is always XVS). Except for the fossilized cases of inversion alluded to in Section 2.4.1, there are only few types of VS constructions available. Some of them are relevant in the thetic domain. One is the well-known there is construction, which, in turn, is subject to lexical restrictions by virtue of its existential/deictic character. There is also a VS construction sometimes found in newspapers and recently described in detail by Birner and Ward (1992). It follows the pattern “participle (present or perfect) + auxiliary + subject” and is therefore possible only with periphrastic verb forms. For further restrictions on the occurrence of this construction, see Birner and Ward (1992, especially p. 8 ff.). Finally, there is a construction which is used in belletristic literature for scene-setting descriptions and requires an adverbial in preverbal position (cf. example (51) in Section 3.3 below). In the thetic domain, English also makes use of the SAcc construction but this, in turn, is restricted to intransitive clauses with a limited number of verbs. In all other cases, specifically “thetic” constructions are not available. English thus has to steer a middle course through a considerable number of “special” constructions (including the usual polyfunctional SV pattern), all of them in competition with each other and subject to heavy constraints. On thetic constructions in English, cf. also Drubig (1992). In German, which also uses SAcc constructions for thetic utterances, there seem to be far fewer restrictions on this type. But the VS pattern, which is also exploited in certain areas of the thetic domain, is likewise subject to the global organization of German syntax which is dominated by the verb-second principle. Similar to English, the VS pattern can thus only be employed in thetic utterances when the preverbal position is filled. This may be achieved, as in English, by expletives or adverbials. It is clear that the discourse-pragmatic exploitability of these constructions is limited. The most spectacular case of a global syntactic pattern bearing a strong impact on its thetic construction (in this case VS) is Hungarian. The most important grammatical factor influencing the employment of VS constructions in Hungarian is the

276

Hans-Jürgen Sasse

complex grammaticalized system of what É. Kiss calls “focus configurationality.”22 The central point is that the position immediately preceding the verb form plays a special role. It is filled by a single constituent called the “verb modifier” by Behrens (1982), a designation which I will adopt here. The verb modifier (henceforth, M) can be an adverb of any sort, a noun in any case form, an adjective, a preverb, or an infinitive, and forms a close intonational unit with the verb. In addition to the M position, there is a grammaticalized topic slot T in preverbal position. This can be preceded by adverbials of setting (S). If we symbolize optional constituents occurring between the T and the M positions by X and Y, the “configurational” make-up of the H sentence is, roughly, ((S)(T)(X)(M(V))(Y)) with all constituents except V being optional. There are certain rules governing the definiteness and referentiality features of those nominal constituents which may fill the M position. The general rule, often cited in the literature, is that nonreferential constituents occur in the M position, while referential constituents (either indefinite or definite) are placed after the verb: (23) a.

b.

c.

Újságot olvastam. newspaper I.read ‘I was reading a newspaper.’ Olvastam egy újságot. I.read a newspaper ‘I was reading a (specific) newspaper.’ Elolvastam az újságot. I.off.read the newspaper ‘I read the newspaper.’

Behrens (1982: 110) points out that indefinite referential constituents are not obligatorily restricted to the postverbal position; they may occupy the M position as well. In particular, this order is almost predictable when there are two arguments, one indefinite and one definite. Moreover, local adverbials may appear in the M position even when they are definite. Hungarian has a grammaticalized focus marking system distinguishing between verb focus and nonverbal constituent focus. The latter consists in placing the focus constituent in the M position. It is usually claimed in the literature that a focus constituent can be intonationally distinguished from a nonfocalized verb modifier by having higher pitch. What is even more important is the fact that the preverbal focus constituent ousts a possible nonfocal M from its preverbal position into a position after the verb (if M is not the focus constituent itself). This is usually called inversion.23

Theticity

277

There are certain elements which obligatorily occupy the M position. Among these are negations, interrogatives, and certain adverbs. One can say that these elements are obligatorily focalized. These elements may also have the effect of replacing possible candidates for the M slot and moving them into the postverbal position. Finally, it is important to note that both definite and indefinite noun phrases may occur in the focus position, but a definite constituent in preverbal position (with the exception of the local adverbials referred to above) is almost always readily identifiable as bearing narrow focus. It is clear that this complicated system has a strong bearing on possible occurrences of thetic VS clauses. In particular, the definiteness/referentiality constraint rules out VS clauses with articleless subjects. In such cases, the subject focus construction is used. The subject focus construction is thus ambiguous for a narrow focus and a thetic reading, if the preverbal position is filled by a noun phrase with zero article. As already mentioned, the subject focus construction resembles a SAcc or SInc construction, but due to the special conditions of focus configurationality it has a quite different status than the SAcc constructions of English or German. Furthermore, XV constructions must always be carefully examined as to whether or not the X is focal before they can be compared to similar constructions in other languages. 2.5.2. Monoargumentality Most widely discussed as a main restricting factor in the occurrence of thetic constructions is monoargumentality. This is a (perhaps misleading) term used to cover a number of formally heterogeneous relational constellations leading to a configuration of just one predicate plus one nominal argument. It should not be confused with intransitivity or monovalency, and it has to be taken as a surface structure criterion ruling out considerations on possible “underlying” direct, indirect, or oblique objects. Furthermore, the term “argument” is meant in the restrictive sense as an overt noun phrase (excluding pronominal clitics). The class of monoargumental predicates thus includes strict intransitive, monovalent predicates as well as medio-passives, reflexives, and support verb constructions, but also verbs which may have clausal arguments. In this sense, monoargumentality has turned out to be a strong restricting factor in Italian and an even stronger one in Spanish: irrespective of their discourse function, VS constructions are allowed in these languages only with monoargumental verbs (see Sornicola 1995 and Cennamo 1995 for further discussion). VS is allowed with transitive verbs in Italian, provided that they have a pronominalized object; otherwise, the so-called subject-indicating impersonals could not be properly described. In addition, it seems that different types of VS structures exhibit different degrees of sensitivity to the monoargumentality criterion. It is usually stronger in thetic functions of VS constructions than in other usages of these constructions, but the whole problem remains to be investigated in detail.

278

Hans-Jürgen Sasse

Table 5. Role of the monoargumentality factor in selected European languages Language Albanian English French German Hungarian Italian Modern Greek Rumanian Spanish

Construction VS SAcc Split SAcc VS VS VS VS VS

Monoargumentality factor low high low low low high low low high

Among the SV/VS alternating languages investigated in the VS study, monoargumentality can sometimes be observed as a statistical tendency, but not as a grammatical constraint. Some of the languages are not very sensitive to monoargumentality; in Hungarian and Modern Greek, transitive VS clauses are common and we have ample evidence that they occur in the “thetic” types. The same seems to be the case in Rumanian as far as can be judged from Ulrich (1985). Intransitives appear to be more frequent in these languages, at least with the “thetic” types, but we lack detailed statistical evidence. Perhaps for all other languages of the SV/VS alternating type, the characterization given for Latin by Bolkestein (1995: 36) can be applied: “The tendency to appear in VS order is strongest : : : in the case of intransitive, monovalent predicates with the verb to be prominently among them : : : However, bivalent predicates, transitive and intransitive, with both arguments expressed, are not incompatible with VS order.” Monoargumentality also seems to be a factor strongly influencing there is constructions and SAcc constructions in English (cf. above). It is less prominent in German SAcc constructions and in French Split constructions, though always observable as a statistical tendency. In summary, we can say that the monoargumentality parameter still remains open for further research, since the evidence is sometimes inconclusive and not exhaustive in this respect. In particular, we do not know exactly how the parameter works for different types of VS constructions. The monoargumentality constraint makes sense with clearly thetic constructions since, the more material a sentence contains, the more likely it becomes that it contains presupposed or focal material, which rules out a thetic construction. Table 5 gives an overview of the relevance of the monoargumentality factor as a strong grammatical constraint in selected European languages.

Theticity

279

2.5.3. Definiteness and animacy Further factors limiting the number of admitted thetic constructions may be the definiteness or animacy of the subjects involved. Interestingly enough, this constraint has so far only been posited for VS constructions given the fact that the well-investigated SAcc and Split constructions in English, German, French, etc., do not exhibit any restrictions of this sort. However, a clear case of a construction strongly resembling SAcc and heavily restricted by definiteness features may be found in Hungarian (see below). As far as SV/VS alternating languages are concerned, the definiteness features of the subject constitute a strong constraint on VS constructions in Hungarian and in Russian. Let us deal with Hungarian first, which is the more complicated case. Hungarian has a strict rule which permits only subjects with the definite or indefinite article to be placed after the verb (cf. Section 2.5.1). Correlated with this is the fact that the focus construction with the subject filling the focus slot is in complementary distribution with the VS construction insofar as nonreferential subjects with zero article obligatorily appear in subject focus position (i.e., in a nontopical SV order clearly distinct from SV with topical S). In other words, some of the thetic functions are distributed between VS and “subject focus” construction in that VS appears with definite subjects and “subject focus” with nonreferential subjects; indefinite specific subjects with the indefinite article may appear in both constructions. Examples can be found in Sasse (1995b: 176 f.), where further details are discussed. The “subject focus” construction is therefore ambiguous insofar as it has a narrow subject focus reading and a thetic reading. If we interpret the Hungarian subject focus construction as a case of SAcc (cf. above), this sensitivity to definiteness would contradict the claim, made in the first paragraph of this section, that, of the thetic constructions examined here, only VS is sensitive to definiteness. In Russian, the situation is nearly opposite. Postverbal subjects are per default interpreted as indefinite or nonreferential. Only verb focus constructions and episodeinitial24 VS constructions containing “given” subjects are exempted from this default interpretation. In other cases, position is a clear sign of definiteness and compensates for the lack of a definite article, so that indefinite subjects must be placed in postverbal position while definite subjects are excluded in this position. For further details see Maslova (1995) and Sasse (1995a: 22 f.). In sum, Hungarian and Russian handle features of definiteness and referentiality in an opposite way, and these things necessarily have a bearing on the language-specific restrictions of the actual use of VS and SV clauses in a text. Definiteness and animacy do not appear to play any important role as strictly grammaticalized restrictions on VS constructions in the other languages of the SV/ VS alternating type. What is often mentioned in the literature, however, is statistical preferences. For instance, Sornicola (1995) reports that in her Italian corpus, all sub-

280

Hans-Jürgen Sasse

jects in VS constructions of the thetic types (called “all-new” by her) have inanimate subjects, and in her Spanish corpus, at least the great majority of such subjects are inanimate. But at any rate, this does not seem to amount to a grammatical constraint.

2.5.4. Local restrictions In addition to the impact of language-specific global syntactic patterns and crosslinguistically observable tendencies, many languages have “local” obligatorizations of the relevant constructions. In Modern Greek, VS is obligatorized in certain types of relative clauses (cf. Sasse 1995b); similar conventionalizations in subordinate clauses are also discussed for Italian and Spanish (cf. Bernini 1995 and Sornicola 1995). Modern Greek and Albanian also obligatorize VS in non-factual clauses; the same restriction is reported for Romani by Matras (1995).

3. Theticity and discourse functions In the preceding section, we have examined four constructions which are relevant in the investigation of the phenomenological domain of theticity. We have investigated their formal structure; we have identified and filtered out a number of usages of these constructions which are clearly nonthetic and will be disregarded in the following sections; and we have described a number of important language-specific grammatical constraints which globally govern the use of these constructions. In the following, we will examine the various discourse situations in which thetic expressions have been said to typically occur. This will help us in arriving at a first approximation of crosslinguistic functional comparability. We will proceed from a list of functions recognized as relevant in the VS study and described in detail in Matras and Sasse (1995).

3.1. The relevant discourse-functional domains The following discourse functions involving the thetic complex are distinguished: annuntiative, introductive, interruptive, descriptive, and explanative. In addition, the VS study has revealed a “connective” function (with two subtypes, the “reactive/ consequential” and the “discontinuative” type, cf. Sasse 1995a: 16 and Matras 1995). Ranging the connective complex along with the other cases has turned out to be problematic, and an extra section will therefore be devoted to the discussion of these cases (Section 4).

Theticity

281

3.1.1. The annuntiative function One of the most prominent discourse configurations in which the thetic complex is found to be relevant is so-called “statements out of the blue,” such as monopropositional exclamations in spoken language and special text genres such as announcements and newspaper headlines in written language. We may call this the annuntiative function. The VS study clearly revealed that this function was served by verbsubject order as a productive strategy in most of the languages investigated (Italian, Spanish, Russian, Modern Greek, Hungarian; to these may be added Albanian, Rumanian, Serbo-Croatian, and several further languages not covered in the VS study). The same function is typically fulfilled by subject-accented sentences in Germanic languages (German, Dutch, English, etc.) and in certain other European languages (Basque, Polish, Czech, and others), sometimes in competition with VS. French and Welsh use Split constructions for this purpose. Some examples follow. Headlines: (24)

Cambia il governatore alla Bundesbank changes the governor to.the Bundesbank ‘At the Bundesbank, the governor changes’ (Italian; Bernini 1995)

(25)

Umer matematik Andrej Tixonov died mathematician Andrej Tixonov ‘(The) mathematician Andrej Tixonov died’ (Russian; Maslova 1995)

(26)

Anavlithike i dhiki ekdhoti efimeridhas was.deferred the lawsuit of.an.editor of.a.newspaper ‘Newspaper editor’s lawsuit deferred’ (Modern Greek; Sasse 1995b)

(27)

Poginuo britanski vojnik killed British soldier ‘British soldier killed’ (Serbo-Croatian; author’s corpus)

(28)

Schulbus verunglückt school bus crashed ‘School bus crashed’ (i.e., ‘School bus accident’) (German; author’s corpus)

282

Hans-Jürgen Sasse

Exclamations: (29)

Jönnek a szomszédok! come the neighbors ‘The neighbors are coming!’ (Hungarian; Ulrich 1985)

(30)

Tilefonise o Kostas! has.called the Kostas ‘Kostas has called!’ (Modern Greek; Sasse 1995b: 167)

(31)

A venit Antonescu! has come Antonescu ‘Antonescu has come!’ (Rumanian; Ulrich 1985: 163)

(32)

Einer Ihrer Klienten ist ermordet worden! one of.your clients is been murdered ‘One of your clients has been murdered!’ (German; author’s corpus)

(33)

J’ai ma femme qui est malade! I.have my wife who is sick ‘My wife is sick!’ (French; standard example)

Nevertheless, it was also found that there is no one-to-one correspondence between the annuntiative context and certain utterance types. First of all, the fact that each language imposes its own semantic and syntactic restrictions on the employability of certain constructions under certain circumstances leads to significant differences in the permitted scope of occurrence of the various constructions when compared across languages. These restrictions have already been dealt with in Section 2.5. Second, some languages seem to formally differentiate between annuntiative exclamations in spoken discourse and monopropositional news reports such as newspaper headlines. German, for instance, formally distinguishes exclamations in spoken language, which are typically full sentences of the SAcc type, from a headline strategy in newspapers, which operates with subject-accentuation, too, but makes use, in addition, of a predicate-dropping device characteristic of the telegraphic style of newspapers (cf. example (28) above). Split structures in French and Welsh and in the other languages in which they marginally occur seem to be restricted to exclamations but do not typically occur in headlines. Moreover, in none of the languages examined can it be said that the annuntiative discourse situation necessarily evokes one or several of the formal devices said

Theticity

283

to express theticity. Even if the different grammatical and lexical constraints imposed on the various constructions are filtered out, there still remains a considerable number of attested exclamations and headlines which, unexpectedly, do not follow the “thetic” pattern. Rather, constructions are used which involve “topical” subjects. This is clearly due to the trivial fact that news can always be reported about situations in which certain persons or entities are already in the center of attention (i.e. presupposed in the traditional sense). This constellation then evokes a topic construction. In the VS study, it was found that in nearly all cases where SV order (with “topical” S) or a verb-focus construction were chosen in such a context, the S’s were referring to discourse topics likely to be shared by the audience (cf. examples in Bernini [1995] for Italian, and Sasse [1995b] for Modern Greek and Hungarian; this has also been confirmed by a brief investigation of Serbo-Croatian and Albanian headlines). The usual precondition for the occurrence of topic constructions in headlines is that a certain person or entity, which had already been the subject of discussion in earlier articles of the newspaper, is involved in an “open” situation, that is, it is expected that he/she/it will do something, and this something is now being reported: (34)

Izrael elfogadta ez amerikai meghívást Israel accepted the American invitation ‘Israel accepts American invitation’ (Hungarian; Sasse 1995b: 183)

This is also the case in German predicate-dropped newspaper headlines, where subject-accentuation contrasts with double accentuation under precisely the same conditions which are responsible for the VS/topical SV alternation in the VS languages: (35)

Fulbright erkrankt Fulbright got.sick ‘Fulbright (got) sick’

(36)

Bonner Bau teurer Bonn’s construction more.expensive ‘Construction in Bonn more expensive’

Contrasts such as these are also encountered in spoken discourse and have been extensively discussed in the literature (cf. the summary of earlier writings and the discussion of further examples from different languages in Sasse [1987], where an interpretation in terms of the “background of expectation” was given; cf. also Lambrecht [1994] for a lucid discussion of the entire complex). Of course, this “expectability” of the subjects’ topical status can only be taken as a possibility of post festum interpretation and not as a predictive rule, given the fact that there is a considerable range of optionality to be expected here. This depends on the individual assumptions of speakers or writers with respect to what knowledge their addressees share in each special case.

284

Hans-Jürgen Sasse

Nevertheless, the high degree of consensus among speakers of different languages with respect to the interpretation of these cases clearly shows that the presuppositional “prehistory” in which news reports are embedded has a strong bearing on the choice of the construction employed and that, in turn, certain presuppositions are intimately connected with certain expression types. 3.1.2. The introductive function The next function to be discussed is the introductive function, by which we mean first mention subjects as a text-opening strategy. There are good reasons to restrict this function to the introduction of discourse participants. There is a second type of utterance often occurring in text-initial position with first mention subjects, namely, scene-setting background descriptions. These do not introduce participants into the discourse. Moreover, in the languages examined, they are frequently not expressed in exactly the same way as participant introductions. This frequent formal difference supports the assumption that scene-setting is an operation quite different from participant introduction. We will therefore posit a separate function of scene-setting which will be dealt with in the following paragraph. Interestingly, participant introduction is not a very frequent function of the “thetic” constructions discussed in the literature. The VS study has revealed the following results. In the Modern Greek corpus, VS scores comparably low as a text-initial word order in all genres investigated, while in Italian, there is a difference between the two corpuses on which Bernini’s investigation is based, though the frequency of text-initial VS does not seem to be exceedingly high in either case (Bernini 1995: 66). In the Hungarian newspaper texts, the first sentences are usually SV (with topical subject), though this may be, in part, explained by the fact that many of them repeat the VS title in the reverse order, the subject then becoming “topical.” Modern Hungarian narrative texts, however, do not very often begin with VS clauses either, and in none of the cases found does the VS construction introduce a new subject to be used as a topic in the subsequent text passage. For Russian, however, Miller (1995: 135) claims that “VS structures typically introduce the entity that is the discourse topic of the following piece of text, ranging from a sentence or two to a whole paragraph.” He thus explicitly describes the indication of a first mention subject as one of the three functions of VS order in Russian intransitive clauses (the other two being contrastive subjects and the introduction of new episodes with given subjects). This may be seen as a special device of Russian in connection with the absence of a definite article in this language and its compensation by word order (cf. Maslova 1995: 109). As for the other languages examined in the VS study, all we can say is that VS clauses may occur at the beginning of texts, but we lack exact figures on their frequency and their exact nature. One notable exception are clauses with the predicate ‘be’ and similar existentials for which we have ample evidence that they almost

Theticity

285

automatically trigger VS, at least in Italian, Modern Greek, Hungarian, Latin, and Russian. Outside of the languages covered by the VS study, the same situation was found in Albanian and Serbo-Croatian (unpublished research by this author), and in Rumanian (Ulrich 1985). This is confirmed by earlier studies on other Romance and non-Romance languages (cf. Wehr 1984: 15 ff., summarizing previous research). In most of the European languages, including those which otherwise do not make extensive use of verb-subject order, the existential (X)VS type figures prominently in introductory speech formulas of traditional narratives (“there was an X”). At least with existential predicates, then, we can confirm that VS clauses are preferably exploited for the introductive text function in many languages.25 The remainder of the constructions relevant in the “thetic domain” do not seem to be used in introductory phases, except for the Danish SInc construction which is, at any rate, coupled with VS. Neither Split constructions nor SAcc constructions ever play a role in participant-introducing contexts. Except for the comparably rare cases of participant introduction by means of the existential (X)VS construction mentioned in the preceding paragraph, the main introductive strategy for participants in modern language all over Europe is by first mentioning new participants in the form of grammatical objects of transitive verbs. Even more frequently, however, authors refrain from using a special introductive device altogether: the main hero of modern narratives is usually presented as a topic from the very first sentence on. 3.1.3. The interruptive function All authors of the VS study agreed that, in their languages, text-internal VS clauses (other than cases of grammaticalized inversion) never continue statements about a topical subject commented on in the preceding sentences. Typically, these VS clauses evoke a “sudden event” effect which we will call the interruptive function. They are usually preceded by a series of events in a topic chain which is disrupted by a sudden, unexpected new situation. Data from all our languages converge in their preference (in some cases even obligatoriness) of VS clauses in cases of sudden, interruptive events. The phone, the alarm clock, or the doorbell ringing, somebody knocking on the door, the door opening, the lights being turned on or off, etc., are notorious instances met over and over again. The same semantic areas are described in studies on other SV/VS alternating languages (see, for example, Wehr 1984 and Ulrich 1985), and were also found in the data from Albanian and Serbo-Croatian investigated by this author. Turning to thetic constructions other than VS structures, it has long been known that they figure prominently in interruptive contexts. The studies on SAcc in English and German, which present a wealth of relevant examples, have already been mentioned in Section 1, and the reader is again referred to the examples given there. The interruptive function is also amply documented for Split constructions in French and

286

Hans-Jürgen Sasse

Welsh; see again Wehr (1984) for both languages, MacCana (1973) on Welsh, and Ulrich (1985) and the various works by Lambrecht on French. 3.1.4. The descriptive function VS clauses with a scene-setting function, henceforth called “descriptive,” have likewise been found in all sample languages of the VS study, and, in addition, in Albanian and Serbo-Croatian. They usually (but not necessarily) involve existential verbs (often in a disguised form denoting a certain prototypical property of the subject such as shining of the sun) and describe environmental conditions presented as a background to the main story line. A lot of beautiful examples extracted from belletristic literature from Modern Greek and Hungarian can be found in Sasse (1995b); Ulrich (1985) also offers a considerable number of examples from Rumanian, for French see Atkinson (1973). German, English, Dutch, and other languages which use SAcc constructions regularly employ these for scene-setting descriptions: (37)

Ein Blizzard nahte. ‘A Blizzard was approaching.’

(38)

Silence descended.

In these languages, the SAcc construction in a descriptive function is usually restricted to very short sentences which do not involve much more material than the subject and the verb themselves. For more elaborate utterances, other strategies are used. In German, SAcc is more or less in complementary distribution with an XVS construction, with the preverbal position filled by a temporal, local or modal adverbial: (39)

In den Tälern löste sich der Nebel zögernd auf. ‘In the valleys, the fog hesitantly lifted.’

(40)

Schon senkte sich die Dunkelheit über das Moor. ‘Already had darkness fallen over the moor.’

(41)

Zu dieser Tageszeit waren nur wenige Menschen auf dem Marktplatz. ‘At this time of day, only a few people were on the market place.’

Many of the SV/VS alternating languages also prefer to employ similar XVS constructions for longer descriptive utterances. Examples from Modern Greek and Hungarian were given in Sasse (1995b). The following is an additional example from Rumanian, borrowed from Ulrich (1985: 117): (42)

La trei sute de metri spre stînga, în coasta ora¸sului [: : :] se leg˘anau, înghesuite, s¸lepuri de fier pîntecoase. ‘Three hundred meters to the left, [: : :] all crowded together, swayed the iron barges.’

Theticity

287

3.1.5. The explanative function The four constructions are prominently involved in expressing explanations of or elaborations on a given situation. This function is called the “explanative” function. It requires a presupposed event, that is, something which has already happened but remains to be identified. The explanative discourse setting thus always involves two parts: the first part, which establishes the presupposition, and the second part, which gives the explanation or elaboration. Depending on how the first part is established in the text, several subtypes may be distinguished: 1. The presupposition is established by the state of affairs referred to in the immediately preceding utterance. The sentence preceding the “thetic” construction indicates a state of affairs calling for an explanation in terms of a yet unknown event, cf. the following German example of an SAcc construction: (43)

Da trat ein jäher Wendepunkt in meinem Leben ein: meine Schwester kam zur Welt. ‘Then, there was a drastic change in my life: my sister was born.’

2. The presupposition is established by a question “what happened?” or quasisynonymous questions such as “why do you do X?” The following example of a French Split construction demonstrates this: (44) A: Qu’est-ce qu’il y a? B: C’est maman qui me bat! ‘A: What’s the matter? – B. Mum is hitting me!’ 3. The presupposition-establishing element is part of the extralinguistic setting, for example, a nonverbal element (gesture, mimics). Furthermore, the presupposition can be built up interactively by the questionanswer sequence (as in (44)), or monologically, as in (43), or in a combination of both, as in the following German example: (45) A: Wo will er hin? B: In die Klinik. Seine Frau wird operiert. ‘A: Where is he off to? – B: To the clinic. His wife is having an operation.’ The explanative function has been extensively discussed in the literature. For further information, the reader is referred to Wehr (1984), Ulrich (1985), Sasse (1987), Lambrecht (1994), and Matras and Sasse (1995), all of whom summarize and evaluate earlier scholarship on the subject.

288

Hans-Jürgen Sasse

3.2. Similarities and differences among the discourse functions The five discourse functions of thetic constructions described in the preceding section have been predominantly established on the basis of their environment, that is, the position in which they occur in the discourse. In the following, we will briefly point to certain similarities and differences among the various discourse functions. The most striking similarity is probably the fact that all thetic utterances which occur in these five discourse situations share the feature of the low presuppositionality of their subjects. In most of the cases, we have found that languages use a special construction when the subject does not refer to an entity which is presuppositionally established, that is, about which new information is expected, but rather, resort to topic constructions when this is the case. Disregarding language-specific restrictions for the moment, the following general tendencies have been observed: annuntiative, interruptive, and explanative utterances with non-presupposed subjects involve a thetic construction, while annuntiative, interruptive, and explanative utterances with presupposed subjects are expressed by a topical SV construction; clearly introductive utterances are always expressed by a special thetic construction (if there is an introduction at all), which may be attributed to the fact that their subject is always non-presupposed. The presuppositional status of descriptive utterances is less clear, and, not surprisingly, there is a certain fluctuation between thetic and topical constructions. Descriptive clauses share a number of important features with interruptive clauses, which set them apart from the other three types. Both interruptive and descriptive clauses may occur text- or paragraph-internally as well as text- or paragraph-initially. It is not the position in the text but the event as such that is responsible for the presuppositional interpretation. Interruption implies a change in reference to the presuppositional background, a break in referential continuity. The author may nevertheless begin a story with a sentence such as The telephone rang, assuming that this expression will evoke a frame in the reader’s mind in which it is conceived of as a sudden, interruptive event. The same is true of scene-setting descriptions. A scene-setting description does not require any specific textually determined presuppositional background for it to be understood (it is situated, so to speak, on a different level than the main story line) and may thus be placed anywhere in the text. Both interruptive and descriptive VS clauses may occur in chains, that is, there can be several interruptive events simultaneously or several states constituting a complex scene-description. There is, however, an important difference between interruptive and descriptive clauses with respect to the opposition between “thetic” constructions and constructions with topical S. In the SV/VS alternating languages, it was found that the opposition between VS and SV with topical S (where this is possible) is of a different nature in interruptive states of affairs than in descriptive ones. With interruptive states of affairs, VS is largely obligatory; the same proposition expressed by means of a

Theticity

289

construction with a topical subject loses its interruptive character. This means that certain noun-verb-combinations are quasi-lexicalized for their interruptive VS order. This is not the case with descriptive utterances. Here, VS order may sometimes be lexicalized, sometimes obligatorized for other reasons, but there are many cases where speakers say that word order doesn’t really matter. Moreover, “natural phenomena” (sun, moon, stars, fog, rain, etc.) may occur as presupposed topics, which makes sense in cases of previous mention, for example, if the topic is long-lasting fog, which is said to dissipate. This still may be held to be a descriptive utterance. The descriptive function can alternatively be signaled by VS in such cases as well. A similar situation was found in languages which use SAcc and XVS in descriptive contexts: topic constructions are often equally good, and it is not always possible to give plausible explanations for the difference. In summary, the point in the discourse where interruptions and scene-setting descriptions occur is not in any way communicatively predictable. This is in contrast to introductive, annuntiative, and explanative utterances. The former two are clearly characterized by not involving any previous text. They thus occur in a presuppositional environment in which no preestablished entity is available to which new information can be connected, except when the speaker/writer deliberately establishes such an entity on the spot. With explanative utterances, the relevant conditions of low presuppositionality are explicitly built up in the preceding discourse, and a lack of response to these explicitly established conditions would be a severe violation of Gricean maximes. In other words, if the language has a strategy at all for signaling the low presuppositionality of an entire proposition, it may be expected that this strategy is employed as an “unmarked” choice in situations where low presuppositionality is explicitly preestablished. There is a particular similarity between annuntiative and explanative thetic utterances. In principle, they can be seen as a continuum. On the extreme explanative end, a presupposition of an “open” situation (“something happened”) is explicitly built up in the discourse. However, we have also admitted cases in which explanative utterances are evoked implicitly, for instance, by means of interrogative gesture. From these, it is only a very small step to situations where a speaker just announces or exclaims a bit of information without caring whether or not the addressee is explicitly asking for it. In most cases, the speaker will automatically proceed from the assumption that the addressee is interested in the information (e.g. in the newspaper context); in a sense, then, annuntiative utterances could be regarded as implicitly evoked explanative utterances. Given the various similarities among the discourse situations in which thetic utterances occur, it comes as no surprise that there is a considerable amount of overlapping of the lexical material occurring in the different types. For instance, the occurrence of phone-ringing can be described as an interruptive event, but it may also be conveyed as an announcement or an explanation. All three of these are attested in

290

Hans-Jürgen Sasse

my German slip corpus from television plays; the only difference lies in the tense of the respective utterances: (46)

Er war in den Keller gegangen, um die Leitung zu reparieren. Das Telefon klingelte. Da muss er wohl raufgekommen sein. (Derrick) ‘He had gone down to the basement to fix the line. The telephone rang. It must have been at that point that he came back upstairs.’

(47)

Ich bin heute morgen schon um sechs aufgewacht. Das Telefon hatte geklingelt. (Derrick) ‘I have been awake since six o’clock this morning. The telephone had rung.’

(48)

Bea, das Telefon klingelt. Gehst du für mich ran, bitte? (Sterne des Südens) ‘Bea, the telephone’s ringing. Can you get it for me, please?’

3.3. Coverage of functions by construction types in the different languages From a typological point of view, it is interesting to compare the languages of our sample with respect to the discourse functions expressible by means of the various constructions. There are some striking points of agreement among the languages with respect to certain functions that are always or never expressed by a certain construction, regardless of the position of the construction in the overall grammatical system of the individual languages; nevertheless, significant differences among the languages may also be observed. We will first consider the similarities across the languages before we deal with the differences. All languages agree in the fact that participant introduction, if explicitly marked, is always done by means of (X)VS constructions, and never by means of any of the other thetic constructions. Even languages in which the major thetic construction is of a different type, such as English and French, use (X)VS constructions for this purpose (cf. English Once upon a time there was an X). In (Standard) German, English, and Dutch, where the preverbal position must be obligatorily filled, an XVS construction with an expletive element has to be used to fulfill the requirements for VS order (cf. German Es war einmal ein X). Slightly less absolute but still striking is the preference of (X)VS constructions for descriptive utterances. Not only do such constructions figure prominently in the SV/VS alternating languages, where they are of course expected, but also in languages which otherwise impose heavy restrictions on the occurrence of VS order, such as German or even English. German descriptive XVS sentences often bear a

Theticity

291

striking superficial resemblance to descriptive XVS sentences of the SV/VS alternating languages. More often than not, a one-to-one correspondence is possible: (49)

Modern Greek: Apo to parathalassio kendro akustike i melancholiki from the by.the.beach pub was.heard the melancholic melodhia enos saksofonu. melody of.a saxophone

(50)

German: Aus dem Lokal am Strand ertönte die melancholische form the pub on.the beach sounded the melancholic Melodie eines Saxophons. melody of.a saxophone ‘From the pub on the beach, the melancholic melody of a saxophone was heard.’

Even in languages with a strong preference for SV order such as English, XVS is a favorite construction for scene-setting descriptions in belletristic style: (51)

Naked towered the branches of the trees towards the sky.

SAcc constructions are only seldom used for descriptive utterances. In both German and English, they are restricted to very short utterances (cf. Section 2.1.4). In English, they are slightly more frequent than in German but, on the whole, SV sentences introduced by an adverbial are preferred. As far as I was able to ascertain, Split constructions are never used for descriptive utterances in any language. We will now turn to some general observations about differences between the languages. The first difference pertains to the role of the explanative function. In Europe, there seems to be a typological dividing line between languages in which the explanative setting plays a major role as a discourse strategy and languages in which it is of minor importance. Interestingly, this difference is closely connected with the expression type prominently used for “thetic” utterances. Generally speaking, many of the languages which use the VS strategy as a predominant device for all kinds of thetic expressions do not very frequently apply it in the typical explanative context. Not surprisingly, the explanative function had played an important role in the discussion of SAcc and Split constructions; it had been identified as one of the most important functions of subject-accented clauses in English and German and of the French and Welsh Split constructions (cf. Wehr 1984; Sasse 1987; and Lambrecht 1994 with further references). It has, however, been given much less attention in the literature on VS constructions. This is not to say that explanative VS constructions do not occur. The following sequences from Rumanian, Modern Greek, and Russian26 are good examples:

292

Hans-Jürgen Sasse

(52)

[: : :] dar pu¸sca n-a luat foc, a fost asudat praful [: : :]. ‘But the gun didn’t go off; the powder had gotten wet.’

(53)

Pao sto ipoyio. Kaike i asfalia. I.go to.the cellar burnt the fuse ‘I’m going to the cellar. The fuse has blown.’

(54)

Tri mesjaca nazad u nas v sem’e sluˇcilos’ gore. Umer naš three months ago at us in family happened grief died our otec. father ‘Three months ago we had grief in our family. Our father died.’

Similar examples are cited for Latin by Bolkestein (1995). Nevertheless, several authors report that the explanative context is less prominent in some of the SV/VS alternating languages, SV clauses with topical subjects frequently being a possible alternative, which is not the case with German SAcc constructions, for instance. For Modern Greek and particularly for Hungarian, the situation is discussed in Sasse (1995b). In the Modern Greek corpus, SV sentences with topical subject often occurred in explanative contexts when the subjects were animate. Further research is necessary, of course, but there seem to be good reasons for believing that this factor plays a role in other languages, too, so that the development of a proper explanative discourse strategy may sometimes be hampered by the semantic features of the constituents involved. In Hungarian, the situation is a bit different. As already indicated above (Section 2.5), VS constructions are, to a large extent, dependent on structural and lexical factors with the result that they are allowed to occur in explanative contexts when they typically occur elsewhere in this form; otherwise, they are simply not possible. At any rate, Hungarian seems to lack a proper explanative strategy comparable to that of German and English SAcc sentences and French Split constructions. A SV/VS alternating language in which VS constructions seem to be very frequently used for explanations is Rumanian, as described by Ulrich (1985); she calls such occurrences “faktumsetzend.” No restrictions on the animacy of subjects in these contexts seem to hold here. The second difference among the European languages is found in the distribution of discourse functions across the possible alternative constructions. In principle, it can be observed that in many languages, more than one of the two or three constructions available in the respective language can be freely used for one and the same function alternatively. The following is an example from my German corpus where the same statement is first uttered in the form of a SAcc construction, and then repeated in the form of XVS: (55)

(In an ICE train:) Meine Damen und Herren, dies ist eine Kundendurchsage. Ein KLEIDUNGSSTÜCK wurde im Bordrestaurant gefunden. Der Verlierer möch-

Theticity

293

Table 6. Range of functions of alternative constructions

Introductive Descriptive Interruptive Annuntiative Explanative

(X)VS

SAcc

Split

C C C C C

 (C) C C C

  C C C

te sich bitte im Dienstabteil, Wagen 9, melden. Ich wiederhole: Im Bordrestaurant wurde ein KLEIDUNGSSTÜCK gefunden. Der Verlierer : : : ‘Ladies and Gentlemen, this is a customers’ announcement. An article of clothing has been found in the restaurant. The owner of the lost object should please report to the Service Compartment in Wagon 9. I repeat: In the restaurant, an article of clothing has been found. The owner : : :’ In a similar (but opposite) way, many of the SV/VS alternating languages also allow SAcc constructions as a possible, though marginal, free alternative.27 Nevertheless, languages may also differentiate functions by means of different constructions. In languages with SAcc and Split constructions, these are never used for introductive utterances, (X)VS being employed instead (cf. above). Furthermore, Split constructions are never used for descriptive utterances. The (X)VS construction thus seems to have the broadest range of functions crosslinguistically, followed by the SAcc construction and the Split construction, as shown in Table 6. A number of language-specific distributional phenomena, often cutting across the five discourse functions, are also worth mentioning here. First and foremost, English has become famous for having the SAcc and the there is (i.e., a special subtype of the XVS) construction side by side, with certain distributional overlapping, but not being entirely equivalent. The same can be said of the proportion between Dutch SAcc and er-construction. French is similar in having VS (with expletive il, see Atkinson 1973) side-by-side with the Split construction. The range of application of the two structures is clearly differentiated (cf. Wehr 1984, summarizing earlier literature). In some Slavic languages, such as Polish, Czech, and Bulgarian, VS is obligatorized with a small number of well-defined expressions, such as pain and similar bodily affections; VS is possible outside these obligatorized areas in one or the other function, but SAcc seems to be preferred. This brings us to the third major difference among our languages: degrees of lexicalization and productivity. For English, Allerton and Cruttenden (1979) had already claimed that SAcc constructions in English only occur with a well-defined number

294

Hans-Jürgen Sasse

of verbs.28 To a still larger extent, this seems to be valid for there is constructions, which are often said to be restricted to so-called “unaccusative” verbs. Similar analyses have been offered for Romance VS constructions (cf. above). This is not the case in other languages. According to our investigations, VS constructions in languages such as Russian, Rumanian, and Modern Greek are much more open in this respect, and German allows a considerable freedom in the predicates of its SAcc construction. The role of lexical semantics will be discussed in Section 5 below.

4. The “connective” complex In addition to the five discourse functions described in the preceding section, some languages are claimed to have VS clauses with a special episode-opening function. Note that none of the other thetic constructions have ever been said to have such a function. This will be of some importance when it comes to the question, to be discussed at the end of this section, whether or not these cases belong to the thetic complex. Introduction of a new episode in narratives is mentioned by Miller (1995) as one of the three main functions of VS in Russian. It is also characteristic of Romani and Modern Greek, particularly in spoken narrative. These VS constructions are characterized by involving a given subject, usually a main character of the story, or even the speech participants themselves in the form of explicit personal pronouns (I, you). Note that in Russian (and the same holds true for Modern Greek and Romani), pronominal reference is normally included in the verb form, except when the referents of the pronouns are presented as contrastive (either contrastive focus or contrastive topic). These pronouns, however, are not contrastive in the usual sense and do not normally bear strong accent. In other words, these cases differ from both the narrow focus constructions considered in Section 2.4.1.1 and the lowpresuppositionality cases discussed in Section 3. The question arises as to their status and their affinity to the thetic complex discussed so far. First, it must be stated that a high degree of “givenness” of the subjects of noncontrastive VS clauses (other than those triggered by inversion and narrow focus, in which the situation is, at any rate, different) is not in itself a problem as long as “givenness” is equated with “definiteness” or “identifiability.” “Given” subjects in this sense are the rule rather than the exception with the annuntiative type of VS clauses, and both the interruptive and the descriptive type commonly involve “identifiable” subjects as well. For Italian, Bernini (1995: 52) explicitly states that VS clauses with definite subjects are common. In Hungarian, VS clauses must have definite or specific subjects (on the language-specific reasons for this see above), though pronominal subjects are rare. “Givenness” in this sense must be distinguished from referential distance, and this is what seems to be the relevant factor here. Unfortu-

Theticity

295

nately, a detailed count of referential distance of S in both VS and SV clauses is available only for Russian. Miller (1995: 134) indicates a significantly higher referential distance of S in VS than in SV clauses; nevertheless, VS clauses taking up a subject separated from its last mention by only a few sentences are common. As far as I know, statistical analyses of referential distance have not been made for any of the other SV/VS alternating languages, but the general impression is that the situation in Modern Greek and in Romani is very similar to that of Russian, while Italian, Spanish, and Hungarian may differ in this respect (Sasse 1995a: 16). In sum, the subject’s presuppositional status as such does not appear to play a significant role. Rather, what seems to be at issue here is the disruption of immediate topic continuity. New episodes typically involve a change in personnel and a new action at the same time, and it appears to be a function of VS clauses in some of the languages to signal this type of topic discontinuity. Note that this function does not seem to play a role in Italian, Spanish, and Hungarian, but has been reported for Latin (Bolkestein 1995) and Rumanian (Ulrich 1985). Episode-introducing VS sentences often do not only indicate a shift in personnel and action, but are presuppositionally somewhat more intimately connected to the preceding text. This presuppositional tie is not one of referent continuity, but a more complex one in terms of consequences of the preceding events. Matras (1995) calls this the “connective” function of VS clauses and describes it in detail for Romani narratives. This function is not only characteristic of Romani VS clauses, however, but is found in Russian, Rumanian, and Modern Greek as well. It even occurs in languages which are not of the SV/VS alternating type, such as Substandard German. The formula is X did a – thereupon Y did b, where the second part is signaled by VS. This type is particularly characteristic of spoken narrative and often involves highly “given” subjects. Again, this function seems to be absent in Hungarian, Spanish, and Italian, but occurs in Latin (Bolkestein 1995). Observations on episode-initial VS clauses in the different languages suggest the distinction between two subtypes of “connectiveness,” both of them highly prominent in some languages while totally lacking in the others: one that involves an explicit consequence or reaction to the immediately preceding state of affairs and one which does not, but rather, indicates a shift from one episode to the next. Instances of both can be seen in the following Romani examples taken from Matras (1995: 190 and 196, examples (1b) and (7c)): (56)

(They came and picked up my father too, they beat him) Taj gelas lesko káko taj poˇcindas vareso bare bare love and went:3SG his uncle and paid:3SG some big big money taj kindas les avri. and bought him out ‘And (so) his uncle went and paid a lot of money and bought him free.’

296 (57)

Hans-Jürgen Sasse

(And at ten o’clock, when we were all sleeping already, suddenly somebody knocked at the door. Well, I didn’t say anything, and my mother-inlaw, she didn’t say anything either.) (The next day, : : :) šutas pe e romni tele te sovel. threw:3SG refl def woman down comp sleep:3SG ‘: : : the woman went to sleep.’

While in (56) the uncle’s buying the father free clearly reacts to the father’s being arrested, in (57), the woman’s going to sleep is not a consequence whatsoever of the preceding states of affairs; it merely stands in a temporal sequence to them. We can therefore distinguish between a reactive/consequential and a simple discontinuative function of episode-initial VS clauses.29 As for the reactive type, Matras (1995) correctly points out that this is typical for (though not restricted to) specific semantic areas such as verbal reactions (thereupon she said/asked/answered. : : :) and emotions (thereupon she was pleased/laughed/started crying/was frightened : : :). It is perhaps significant that quite a number of examples of both “connective” types are XVS clauses, where V is preceded by adverbial material which includes entire adverbial clauses, and this should be taken into account for a reexamination of the inversion problem (cf. Section 2.4.1.2 above). The connective function has been discussed in great detail for Rumanian VS clauses by Ulrich (1985; cf., particularly, 284–302). She exempts the entire complex from the thetic domain and posits a special function (as part of the polysemy of Rumanian VS constructions), which she calls the “narrative” function. This is an attractive decision since it would solve the problem of the discrepancies in presuppositionality between these cases and the thetic cases examined in Section 3. Moreover, it was observed that this function is confined to VS constructions (in European languages at least), and it could be assumed that it is a specific function of VS order in these languages apart from the functions called thetic. An alternative analysis of these cases in terms of verb focus had already been proposed in Sasse (1995a: 17–18 and fn. 10). It could be argued that the clause-initial position of the verb in these instances signals a contrast to the state of affairs expressed in the preceding utterance. This would be in accordance with the fact that, in some of our languages at least, verb focus with pronominal referents as subjects usually requires an explicit personal pronoun in postverbal position. The connective VS clauses share this feature in all the languages examined. Episode-initial VS clauses could then be taken to represent an implicature of verb focus specifically conventionalized in certain languages but not in others. However, this is open for further research.

Theticity

5.

297

The impact of lexical semantics

The results of our research strongly support the traditional claim that certain semantic areas are destined for thetic constructions, such as existentials, verbs of appearance, psych-verbs, meteorological conditions, and the like. This is a core of existential semantics in a broader sense, which always seems to constitute the lexical semantic basis of the constructions in question. Exhaustive lists of verbs occurring in VS constructions of SV/VS alternating languages can be found in several contributions of Matras and Sasse (1995), in particular Bernini (1995) on Italian, Miller (1995) on Russian, and Sasse (1995b) on Modern Greek. In addition, Ulrich (1985) gives a detailed account of the situation in Rumanian, discussing other languages as well. In a German corpus of approximately 500 SAcc utterances recorded from television plays, the same semantic areas prevailed. On the other hand, all authors agree that VS constructions are not restricted to these areas and that no straightforward conditions in terms of semantic classes can be formulated. For Russian, Restan’s (1981) strong assumption about strict limitations on semantic classes was contradicted by Maslova (1995) for her corpus of headlines; Miller (1995), who examines a different corpus of text types, finds the semantic range of verbs in VS constructions even still more open. German SAcc constructions cover a considerable range of verbs; the statistical prevalence of the semantic areas referred to is only a tendency. No case was found where one of the thetic constructions dealt with in this paper is explicitly confined to or automatically triggered by a well-defined homogeneous semantic class of predicates. Even the alleged semantic restrictions of the English SAcc construction were found to be more flexible than hitherto assumed. There may be obligatorizations with single verbs, but the only case found so far was the verb ‘be’/‘exist’ in some languages. Of course, the semantic range is lowest in Italian and Spanish because of monoargumentality which a priori rules out all transitive verbs proper. Within the monoargumental domain, however, all types are found, though types involving less agentive subjects prevail. Agentivity and control properties of subjects are also said to be statistically relevant for the choice between SV and VS in studies of other languages, without, however, constituting strict conditioning factors. In all languages other than Italian and Spanish, the range of admissible semantic types largely depends on what subtypes of “theticity” can be expressed by VS order. It is clear that in those languages which have clear episode-opening functions of VS clauses, the semantic range must be much higher than in those which do not have them, since in the episode-initial position there are comparably much fewer restrictions on possible situations than, say, in the interruptive type. However, if connective functions are exempted from the domain of theticity, then it may well be that the number of verbs admissible in a thetic construction crosslinguistically decreases significantly. This brings us to the intricate problem of the interplay of lexical semantics and discourse strategies. In numerous expressions, thetic constructions are halfway lexi-

298

Hans-Jürgen Sasse

calized in the sense that a certain noun + verb combination is always unmarked in a certain construction, be it VS, SAcc, or Split, and highly marked (if ever possible) in a construction with a topical subject.30 This tendency for lexicalization is strongest in idiomatic expressions, to which certain existentials with abstract subjects, but also impersonal verbs, psych-verbs, and similar expressions indicating affection usually belong in our test languages. A further relatively large group of noun + verb combinations which appear in default VS order are those in which V and S are “in semantic agreement” (“lexical solidarities,” Coseriu 1967). This had already been claimed for English SAcc constructions by Allerton and Cruttenden (1979). The combination of the semantically affine lexical elements involved contains built-in default presuppositions which lead, in all of the languages examined, to a quasi-lexicalization of the entire expression in the form of an unmarked or default thetic construction. The notorious cases of phone-ringing and door-opening are good examples. Both states of affairs are predetermined for suddenness. It is significant in this connection that verbs of appearance occur more often in thetic expressions with interruptive character than verbs of disappearance, but the latter are frequently found in thetic utterances with explanative and annuntiative functions. Disappearance of someone or something often provokes an action which is then explained in terms of the loss (I have to go to the police station. Don Miller has escaped.). Typical annuntiative states of affairs referred to over and over again in the literature are mishaps, such as the loss of something, a gnat-bite, a pain, a dish burning in the oven, but positive events as well such as the suddenly flowering cactus. To what extent such cases are conventionalized and to what extent they can be exploited for creative processes in discourse is a languagespecific matter. Yet, there is a common core of quasi-lexicalized “theticity-relevant” states of affairs crosslinguistically associable with certain discourse positions and closely tied to the five discourse-pragmatic functions of thetic constructions set up in Section 3. Some of these correlations are exemplified in Table 7, which is a refined version of the table given in Sasse (1995a: 24). Without having been examined in detail, typological differences have been observed in the following areas: 1. Languages differ in the extent to which they allow generalizations out of the “existential” semantic core which is destined for thetic constructions. That is, they differ, so to speak, in the freedom of metaphorically extending existentials (including “dynamic” existentials such as appearance and disappearance) to other situations not originally being existential situations proper. 2. Languages differ in the semantic areas covered by a certain construction depending on the range of discourse functions served by the construction. 3. Languages differ in the degree of lexicalization. Lexicalizations always involve the existential core situations, such as bodily affections, but also sudden appearances, for example, core interruptives such as phone-ringing, etc. 4. Languages differ in their discourse strategies. In some languages, for example,

Theticity

299

Table 7. Examples of states of affairs typically associated with “thetic” VS and tendentially lexicalized Discourse functions

Associated semantic areas

Introductive Annuntiative

existentials + indefinite animate subjects appearance and disappearance, beginning, ending; expected results of actions (‘dinner is ready’), mishaps, gleeful news; pain, bodily conditions sudden events (phone ringing, door opening), appearance meteorological expressions, existentials with natural phenomena as subjects, existentials pertaining to habitual situations; beginning, lasting pertaining to habitual situations; beginning, lasting and ending of background scenery (in principle same as for annuntiative, but perhaps more open)

Interruptive Descriptive

Explanative/elaborative

certain constructions are predominantly lexicon-driven, while in others, comparable constructions are discourse-sensitive and form a characteristic pattern of text constitution independent of the lexical material occurring in the constructions.

6. Summary, conclusions, and perspectives for future research Through our typological research on theticity, a number of questions have been answered; others remain open for future research. Among the questions which have been answered are the following: 1. Is theticity a crosslinguistically comparable phenomenon? The answer is yes. We have been able to disclose a number of constructions which are used, in individual languages, for the expression of certain similar types of situations. These are comparable to the extent that they show very similar cumulations of functions when compared across languages. Moreover, the number of construction types found crosslinguistically is very small; that is, many languages employ at least superficially similar constructions in similar situations. 2. Is theticity a unitary phenomenon? The answer is clearly no. We have found five subtypes of theticity, each with its own phenomenological peculiarities. The fact that, in all of the languages examined, most or even all of the five subtypes are relevant for the use of the same formal device does not mean that they are all the same.31 3. Is theticity predictable in terms of a single, both necessary and sufficient criterion? The answer is no. We have found that the “all-new” criterion invoked by many scholars is not adequate. The criterion of low presuppositionality of the entire situation expressed was found to be a precondition for the use of a thetic utterance

300

Hans-Jürgen Sasse

in all of the five subtypes examined; however, it does not trigger thetic constructions even in languages in which the relevant constructions are not subject to strong grammatical or lexical constraints since topical constructions are always possible under the same conditions. Low presuppositionality is thus a necessary, but not a sufficient criterion. Rather, thetic constructions are connected with an additional act of assertion which explicitly signals the low presuppositionality of the state of affairs expressed, something like “look out, addressee, an assertion is being made that adds a new situation to your presuppositional fundus” (this is the idea of “sentence focus” recently favored by many researchers in the field). The permitted range of actual application of constructions with which this presuppositional/assertional signal is connected is determined by a variety of factors such as the language-specific polysemy of the construction, the synonymy of “competing” constructions, the nature of language-specific discourse strategies, the restrictions imposed by the global grammatical organization of the individual languages, etc. This makes the actual use of the relevant constructions rather difficult to generalize across languages. 4. Is theticity dependent on the lexical semantics of the constituents involved? To a certain extent, yes. The central lexical domain connected with thetic constructions is static or dynamic “existence.” Languages were found to differ in the degree of sensitivity to this factor. In some languages, obligatorization of thetic constructions was found in the most central areas of existential semantics. In all languages, thetic constructions show a statistical preponderance of verbs with clear existential semantics over verbs from other semantic domains. Further lexical semantic factors found to be relevant are semantic “solidarities,” animacy and control properties of the subjects involved, and others. 5. Is theticity a category? The answer is clearly no. It is a conglomeration of similar presuppositional/assertional conditions prevailing in similar semantic areas, which are frequently expressed by comparable constructions in different languages. 6. Is there a simple thetic-categorical distinction? Again, the answer is no. Thetic constructions always stand in opposition to a variety of other constructions which are not easily subsumed under a label of “categoricality.” For instance, narrow focus constructions are not categorical in the sense envisaged by earlier writers such as Brentano and Marty. It is a fact that thetic constructions are often opposed to constructions with a topical subject (which would probably be held to be categorical by these authors), but this is only one of the many possible syntactic oppositions in which they may be involved. There are two larger areas in which further research is clearly necessary. One is the problem of the origin and the nature of presuppositions. Which are the relevant factors which give rise to presuppositions and how can they be adequately represented? One of the main difficulties, which complicates research in this area enormously, resides in the fact that presuppositions come from two sides: from the discourse situation, and from the meaning of the utterance with all its components (i.e. both by the

Theticity

301

meaning of the construction and the lexical meaning of the constituents involved). It has not been possible so far to disentangle the complex interaction of these two sources of presuppositional conditions. The second problem open for investigation is the position of the phenomena examined here in the larger context of presuppositional/assertional phenomena generally. It is clear that the domain of theticity itself is situated within a larger domain which comprises the entire network of form-function relations pertaining to the distribution of presuppositional and assertional characteristics among the objects and situations of an utterance in actual discourse. The constructions dealt with in this chapter have been deliberately confined to subject-verb combinations since this was the traditional phenomenological domain from which we proceeded. However, in most of the languages examined, the subject-verb pattern is only one of several subpatterns of a more general pattern, which occurs under similar conditions with similar discourse-pragmatic effects. It has repeatedly been pointed out that SAcc constructions in German, Czech, Polish, for instance, show a striking parallelism to constructions involving datives in the preverbal position (cf. the striking similarity between German Seine Frau (nominative) ist krank ‘His wife is sick’ and Seiner Frau (dative) geht’s nicht gut ‘His wife is not feeling well’). Moreover, objects of transitive sentences behave intonationally in a very similar way. This opens a perspective for regarding SAcc as a subtype of a general pattern of noun-accentuation in closely-knit noun-verb combinations which represent a broad scope of assertion (broader than a narrow or constituent focus). A similar case can be made for Hungarian. The conditions under which VS is allowed in this language are exactly the same as those which allow VN combinations generally (cf. Section 2.5.1). It thus seems necessary to reexamine thetic constructions in this larger context in order to reveal the more general patterns by which languages are characterized. Such questions have already been addressed by several authors (e.g., Drubig 1992 for English) but much more work has to be done along these lines.

302

Hans-Jürgen Sasse

Appendix. Abbreviations abs art comp def exist nf poss perf refl vf vn 1SG 3SG

absolutive article complementizer definite existential verb or marker noun focus possessive perfective reflexive verb focus verbal noun first person singular third person singular

Notes 1.

2. 3. 4. 5.

6. 7. 8. 9.

Examples are mostly taken from the relevant literature, without specifying the source when standard examples occurring in many publications are cited. Otherwise, the sources are indicated in the form of statements such as “the following example is borrowed from : : :”. German examples of subject-accented sentences come from my own slip corpus of material transcribed from television plays, unless otherwise indicated. Serbo-Croatian and Irish material is likewise my own; I am indebted to Elvira Veselinovi´c for her kind help with both languages. VS sentences from Italian, Spanish, Latin, Romani, Russian, Modern Greek, and Hungarian are taken from the corpora examined in connection with the VS study undertaken by Eurotyp Thematic Group 1 between 1992 and 1994 (cf. Matras and Sasse 1995). I hereby collectively thank the members of the group for their excellent work. I am also indebted to Georg Bossong, Bernard Comrie, Dejan Mati´c, and Barbara Wehr for valuable suggestions. Throughout this paper, words carrying the intonation peak are set in capitals. Interlinear translation: pain-ART:ABS be-1SG neck-ART-INESSIVE. The French examples in (1) and (2) are colloquial (substandard in the case of (2)). This is an elaboration on a standard example often cited in the literature, see, in particular, Lambrecht (1987, 1994). The translations are taken from various sources, including own work with informants. Some of the translations really mean ‘my throat hurts’, but this is irrelevant in the present connection. Interlinear translation: EXIST pain in my throat/EXIST my throat sore. Interlinear translation: telephone-ART-ERG hit-PERF be-3SG. Interlinear translation: EXIST ART telephone at ring:VN. The term “all-new” is also a particularly bad choice with regard to languages such as

Theticity

10.

11.

12.

13.

14.

15. 16.

17. 18.

19. 20.

21.

303

Hungarian, in whose VS constructions the subjects are obligatorily definite and thus never “new” in one of the senses of this term. I avail myself of this opportunity to apologize to the late Anna Fuchs. In my 1987 paper, I listed her in the group of adherents of the “information structure” approach without mentioning that she had already expressed views very similar to the point I made in that article. On the contrary, Marty explicitly separated the notions of “psychological” (or “logical”) and grammatical subject-predicate structure and allowed for a discrepancy between the two which he explained in terms of his “theory of inner speech form.” In fact, one of his main points was the independence of Urteil (judgment) and Aussage (sentence form), cf. Marty (1897, 1918: 5–19). For brief historical accounts of the thetic/categorical debate, the reader is referred to Ulrich (1985), Sasse (1987), and, most recently, Haberland (1994) with further references. Note that I do not consider it illegitimate to look for correlations of form and content; on the contrary, this is what linguistics is all about. But when it comes to crosslinguistic comparison, the two have to be clearly separated. It is only in this way that an understanding of mismatches between universal hypotheses and language-specific phenomena – and, by implication, a proper typological generalization – becomes possible. The problem with these examples is that in actual discourse, the answer to questions of this type usually contains only the predicate, the subject being pronominalized. Discussion of these and similar problems of ellipsis is found in Lambrecht (1987, 1994); see also Sasse (1995a, fn. 4). The usual way of marking a thetic utterance in Cayuga is by means of the particle ne:’ ‘it is’ put in front of the otherwise unaltered sentence. A very special construction not to be dealt with in this article occurs in one European language: Finnish. In Finnish, the subject of a thetic utterance is in the partitive (cf. Belletti 1988). Even though this is too general a claim (and it wouldn’t typically apply to examples where the noun is a singular count noun) this case is interesting as it involves case morphology as a partial indicator of theticity. Since no detailed study of the behavior of thetic expressions in Finnish is as yet available, we will not comment further on this issue. This raises the question of whether the “subject” of such constructions is really a subject, cf. Lazard (1994). It could perhaps be argued hat VS in present-day Spanish and former historical stages of Portuguese, Italian, and other Romance languages represents a kind of subject incorporation in an embryonic stage; see Bossong (1998). I am indebted to Elvira Topalovi´c and Elvira Veselinovi´c for help with the SerboCroatian data. This has not always been the case. Contrastive subject focus with VS is attested in earlier stages of the language, cf. Sasse (1995: 179 f.); good examples can be found in Behrens (1989). For the role of adverbials of setting as a factor triggering VS in languages other than

304

22.

23.

24. 25. 26. 27. 28.

Hans-Jürgen Sasse those of the VS sample, see Myhill (1986). Exhaustive descriptions of this system are presented, from a generative point of view, in É. Kiss (1987) and numerous other publications by the same author, and in Behrens (1982), in a nongenerative framework. A historical account can be found in Behrens (1989). For present purposes, we will confine ourselves to a brief summary of those facts which are relevant for the subsequent discussion of VS order as a thetic device. Note that this is a different sense of inversion than that understood in Section 2.4.1.2 (where it was confined to inverted subjects), though the effects may sometimes look superficially similar. The episode-opening VS type will be dealt with in Section 4. A detailed account of traditional and modern strategies for introducing protagonists in French narratives can be found in Bossong (1984). The Rumanian example is adapted from Ulrich (1985: 152), the Greek and Russian examples come from the VS study (cf. Sasse 1995a: 18 and Miller 1995: 140). Research has yet to be done to ascertain whether these are true alternatives or have a different semantics, for example, subject focus. Comrie (pers. comm.) points out that this claim is possibly incorrect: “I suspect that the set of verbs is productive, as in the following, where I make up words that might conceivably apply to a time machine: (i)

29.

30.

31.

A: B:

What’s wrong with the time machine? The INTEROCITER’s devabulated.”

The terms “connective” and “discontinuative” may seem contradictory at first sight, but are not since they refer to different layers of presuppositonal depth: with respect to the overall network of text structure, these utterances “connect” text segments, while with respect to topic continuity, they “disconnect” continuous chains. In these cases, a topical subject is usually possible only with verum focus on the verb. In addition to the fact that this is a rare situation in most of the cases under discussion, in those languages which express “verum focus” (i.e. the positive counterpart of polar verb focus) by fronting the verb with a strong accent, the order would still be VS! Of course, as Bernard Comrie (pers. comm.) points out to me, the existence of subtypes does not necessarily contradict the unity of a phenomenon. Nevertheless, the main point here is that it proves difficult to define a set of necessary and sufficient criteria in terms of which the domain under discussion could be described in a straightforward way; consequently, it looks more like a case of “family resemblance” rather than anything else. See points 3–6 below.

References Allerton, David J., and Alan Cruttenden 1979 Three reasons for accenting a definite subject. Journal of Linguistics 15: 49–53. Atkinson, J. C. 1973 The Two Forms of Subject Inversion in Modern French. The Hague: Mouton.

Theticity Behrens, Leila 1982

1989

305

Zur funktionalen Motivation der Wortstellung: Untersuchungen anhand des Ungarischen. Munich: Veröffentlichungen des Finnisch-Ugrischen Seminars an der Universität München. Textanalyse und Sprachwandel. Historische Syntax des Ungarischen: ein empirisches Modell. Frankfurt am Main: Peter Lang.

Belletti, Adriana 1988 The case of unaccusatives. Linguistic Inquiry19: 1–34. Bernini, Giuliano 1995 Verb-subject order in Italian: an investigation of short announcements and telecast news. In Verb-Subject Order and Theticity in European Languages, (Sprachtypologie und Universalienforschung 48 [1/2]), Y. Matras and H.-J. Sasse (eds.), 44–71. Berlin: Akademie Verlag. Birner, Betty, and Gregory L. Ward 1992 On the interpretation of VP inversion in American English. Journal of Linguistics 28: 1–12. Blinkenberg, Andreas 1928 L’ordre des mots en français moderne. Copenhagen: Høst. Bolkestein, Machtelt 1995 Functions of verb-subject order in Latin. In Verb-Subject Order and Theticity in European Languages, (Sprachtypologie und Universalienforschung 48 [1/2]), Y. Matras and H.-J. Sasse (eds.), 32–43. Berlin: Akademie Verlag. Bossong, Georg 1984 Zur Linguistik des Textanfangs in der französischen Erzählliteratur. Zeitschrift für französische Sprache und Literatur 94: 1–24. 1998 La inversión narrativa y la tipología del español. In Estudios de lingüística textual, José Perona, Herminia Provencio Garrigós, Estanislao Ramón Trives, and Agustín Vera Lujan (eds.), 79–88. Murcia: Universidad de Murcia. Brentano, Franz 1973 Psychology from an Empirical Point of View. [Psychologie vom empirischen Standpunkt (1874, 1924), transl. by Antos C. Rancurello, D. B. Terrell, and Linda L. McAlister]. Sections V–IX. London: Routledge and Kegan Paul. Cennamo, Michela 1995 Transitivity and VS order in Italian reflexives. In Verb-Subject Order and Theticity in European Languages, (Sprachtypologie und Universalienforschung 48 [1/2]), Y. Matras and H.-J. Sasse (eds.), 84–105. Berlin: Akademie Verlag. Chafe, Wallace L. 1974 Language and consciousness. Language 50: 111–133. Coseriu, Eugenio 1967 Lexikalische Solidaritäten. Poetica 1: 293–303. Dik, Simon, Maria E. Hoffman, Jan R. de Jong, Sie Ing Djiang, Harry Stroomer, and Lourens DeVries 1981 On the typology of focus phenomena. In Perspectives on Functional Grammar, Teun Hoekstra, Harry van der Hulst, and Michael Moortgat (eds.), 41–74. Dordrecht: Foris. Drubig, H. Bernhard 1992 Zur Frage der grammatischen Repräsentation thetischer und kategorischer Sätze. In Informationsstruktur und Grammatik, (Linguistische Berichte, Sonderheft 4/1991/92), Joachim Jacobs (ed.), 142–195. Opladen: Westdeutscher Verlag. É. Kiss, Katalin 1987 Configurationality in Hungarian. Budapest: Akadémiai Kiadó/ Dordrecht: Reidel. Fuchs, Anna 1976 “Normaler” und “kontrastiver” Akzent. Lingua 38: 293–312.

306 1980

Hans-Jürgen Sasse

Accented subjects in “all-new” sentences. In Wege zur Universalienforschung (Festschrift für Hansjakob Seiler), Gunter Brettschneider and Christian Lehmann (eds.), 449–461. Tübingen: Narr. Haberland, Hartmut 1994 The thetic/categorical distinction. In The Encyclopedia of Language and Linguistics, Vol. 9: Tab to Zor, R. E. Asher and J. M. Y. Simpson (eds.), 4605–4606. Oxford: Pergamon Press. Hatcher, A. G. 1956 Theme and underlying questions. Two studies of Spanish word order. Supplement to Word 12. Höhle, Tilman N. 1992 Über Verum-Fokus im Deutschen. In Informationsstruktur und Grammatik, (Linguistische Berichte, Sonderheft 4/1991/92), Joachim Jacobs (ed.), 112–141. Opladen: Westdeutscher Verlag. Hopper, Paul 1985 Discourse functions and word order shift: a typological study of the VS/SV alternation. In Language Typology 1985, Winfred P. Lehmann (ed.), 123–140. Amsterdam: John Benjamins. Jacobs, Joachim 1992 Integration. (Arbeiten des SFB 282 “Theorie des Lexikons” 13). Düsseldorf: Heinrich Heine-Universität. Jonare, Brigitta 1976 L’inversion dans la principale non-interrogative an français contemporain. Stockholm: Almqvist & Weksell. Krifka, Manfred 1984 Fokus, Topik, syntaktische Struktur und semantische Interpretation. Unpublished ms. Krylova, Olga A., and S. A. Khavronina 1988 Word Order in Russian. Moscow: Nauka. Kuno, Susumu 1972 Functional sentence perspective. A case study from Japanese and English. Linguistic Inquiry 3: 269–320. Kuroda, S.-Y. 1972 The categorical and the thetic judgment. Evidence from Japanese syntax. Foundations of Language 9: 153–185. Ladusaw, William A. 1994 Thetic and categorical, stage and individual, weak and strong. Unpublished ms. Santa Cruz: Linguistics Research Center. Lambrecht, Knud 1987 Sentence focus, information structure, and the thetic-categorical distinction. In Berkeley Linguistics Society: Proceedings of the Thirteenth Annual Meeting February 14–16, 1987: General Session and Parasession on Grammar and Cognition, Jon Aske, Natasha Beery, Laura Michaelis, and Hana Filip (eds.), 366–382. Berkeley, CA: Berkeley Linguistics Society. 1988a Presentational cleft constructions in spoken French. In Clause Combining in Grammar and Discourse, John Haiman and Sandra A. Thompson (eds.), 135–180. Amsterdam: John Benjamins. 1988b There was a farmer had a dog. Syntactic amalgams revisited. In Proceedings of the Fourteenth Annual Meeting of the Berkeley Linguistics Society, Shelly Axmaker, Annie Jaisser, and Helen Singmaster (eds.), 319–339. Berkeley, CA: Berkeley Linguistics Society. 1994 Information Structure and Sentence Form. Cambridge: Cambridge University Press.

Theticity Lazard, Gilbert 1994 Lonzi, Lidia 1974

307

L’actant H: sujet ou objet? Bulletin de la Société de Linguistique 89: 1–28.

L’articolazione presupposizione-asserzione e l’ordine V-S in italiano. In Fenomeni morfologici e sintattici nell’italiano contemporaneo, Vol. 1, Mario Medici and Antonella Sangregorio (eds.), 197–215. Rome: Bulzoni. MacCana, Proinsias 1973 On Celtic word order and the Welsh “abnormal” sentence. Ériu 24: 90–120. Marty, Anton 1897 Über die Scheidung von grammatischem, logischem und psychologischem Subjekt resp. Prädikat. Archiv für systematische Philosophie 3: 174–190, 294–333. 1918 Gesammelte Schriften. Vol. 2. 1. Abteilung: Schriften zur deskriptiven Psychologie und Sprachphilosophie. Halle: Niemeyer. Maslova, Elena 1995 VS/SV opposition in Russian and the thetic/categorical distinction. In Verb-Subject Order and Theticity in European Languages, (Sprachtypologie und Universalienforschung 48 [1/2]), Y. Matras and H.-J. Sasse (eds.), 106–124. Berlin: Akademie Verlag. Mathesius, Vilém 1983 [1929] Functional linguistics. In Praguiana, J. Vachek (ed.), 121–142. Amsterdam: John Benjamins. Matras, Yaron 1995 Connective (VS) order in Romani. In Verb-Subject Order and Theticity in European Languages, (Sprachtypologie und Universalienforschung 48 [1/2]), Y. Matras and H.-J. Sasse (eds.), 189–203. Berlin: Akademie Verlag. Matras, Yaron, and Hans-Jürgen Sasse (eds.) 1995 Verb-Subject Order and Theticity in European Languages. (Sprachtypologie und Universalienforschung 48 [1/2]). Berlin: Akademie Verlag. Miller, Jim 1995 VS clauses in Russian: contrast, new episodes and theticity. In Verb-Subject Order and Theticity in European Languages, (Sprachtypologie und Universalienforschung 48 [1/2]), Y. Matras and H.-J. Sasse (eds.), 125–141. Berlin: Akademie Verlag. Mithun, Marianne 1984 The evolution of noun incorporation. Language 60: 847–894. 1987 Is basic word order universal? In Coherence and Grounding in Discourse, R. S. Tomlin (ed.), 281–328. Amsterdam: John Benjamins. Müller-Hauser, M. L. 1943 La mise en relief d’une idée en français moderne. Geneva: Droz. Myhill, John 1986 The two VS constructions in Rumanian. Linguistics 24: 331–350. Nedergaard Thomsen, Ole 1991 Unit accentuation as an expression device for predicate formation in Danish? Acta Linguistics Hafniensia 23: 145–196. Restan, Per 1981 The position of the finite verb in some elementary declarative sentences in modern Russia. In The Slavic Verb: An Anthology Presented to Hans Christian Sørenson, Per Jacobsen and Helen L. Krag (eds.), 147–160. Copenhagen: Rosenkilde and Bagger. Sasse, Hans-Jürgen 1987 The thetic/categorical distinction revisited. Linguistics 25: 511–580. 1995a “Theticity” and VS order: a case study. In Verb-Subject Order and Theticity in European Languages, (Sprachtypologie und Universalienforschung 48 [1/2]), Y. Matras and H.-J. Sasse (eds.), 3–31. Berlin: Akademie Verlag.

308 1995b

Hans-Jürgen Sasse

A contrastive study of VS clauses in Modern Greek and Hungarian. In Verb-Subject Order and Theticity in European Languages, (Sprachtypologie und Universalienforschung 48 [1/2]), Y. Matras and H.-J. Sasse (eds.), 142–188. Berlin: Akademie Verlag. Schmerling, Susan F. 1976 Aspects of English Sentence Stress. Austin: University of Texas Press. Schroeder, Christoph 1995 Post-predicate arguments in Turkish. In Verb-Subject Order and Theticity in European Languages, (Sprachtypologie und Universalienforschung 48 [1/2]), Y. Matras and H.-J. Sasse (eds.), 204–218. Berlin: Akademie Verlag. Sornicola, Rosanna 1995 Theticity, VS order and the interplay of syntax, semantics and pragmatics. In VerbSubject Order and Theticity in European Languages, (Sprachtypologie und Universalienforschung 48 [1/2]), Y. Matras and H.-J. Sasse (eds.), 72–83. Berlin: Akademie Verlag. Ulrich, Miori¸ta 1985 Thetisch und kategorisch. Tübingen: Narr. Vattuone, Bart 1975 Notes on Genoese syntax. Kernel “VOS” strings and theme-rheme structures. Studi italiani di linguistica teorica ed applicata 4: 335–378. Wandruszka, Ulrich 1982 Studien zur italienischen Wortstellung. Tübingen: Narr. Wehr, Barbara 1984 Diskursstrategien im Romanischen. Tübingen: Narr. 1995 SE-Diathese im Italienischen. Tübingen: Narr.

III.

From form to function

Alan Cruttenden

The de-accenting of given information: A cognitive universal?

1. Introduction Prosody is relevant to continuity in spoken discourse in a number of ways. For example, variations in pitch height may be used for discourse topic delineation. Geluykens (1993: 181) defined such topics operationally: they are (i) “non-recoverable from previous contexts, and (ii) persistent in the following context.” Such discourse topics can be introduced, in English for example, by wh- and yes/no-questions, by existential there, or even explicitly, for example, I’ll tell you something else. A new discourse topic regularly produces a higher pitch on the first accented syllable after the change of topic compared with the pitch of the last accented syllable of the previous topic (Brown et al. 1980; Sinclair and Brazil 1982; Brazil 1985, who refers to higher “key”). Additionally, the first intonational phrase of a new discourse topic can be marked by a jump to a higher mean fundamental frequency and the use of a wider pitch range (where the mean is being measured within intonational phrases, and range involves extent of deviation from the mean) (see Lehiste 1975, 1982; Brown et al. 1980; Hirschberg 1994 for English; Couper-Kuhlen 1983 for German; Swerts and Collier 1992 for Dutch). An extra-low fall, often accompanied by a longer pause and lower amplitude, may mark the end of an intonational phrase which corresponds with the end of a discourse topic (see Lehiste 1979, 1982). These uses of pitch height for topic delineation are putatively universal. The span of these features for topic delineation is often referred to as a paratone (e.g., Yule 1980). A second way in which prosody is relevant to discourse continuity is in the use of rising tones. All crosslinguistic evidence from the world’s languages suggests that rises are used for a set of meanings which can be labelled “open” (Cruttenden 1981, 1986). This set of meanings includes polar question, nonfinality, implication, doubt, and request. The meaning that is particularly relevant here is nonfinality. Hirst and Di Cristo (1998: 27), in a survey of the intonation of twenty languages, state: “Like questions, unfinished utterances or continuatives are commonly pronounced with rising intonation in many languages. In fact a raised final pitch is perhaps even more common for continuatives than for questions.” Recently, in a strictly controlled acoustic and perceptual experiment, Swerts and Collier (1992) used an elicitation

312

Alan Cruttenden

technique by asking speakers of Dutch to describe a sequence in which a series of shapes (brown circle, red triangle, yellow square, etc.) occurred in a network; they then extracted the description of a similar shape (i) in the middle of a string of shapes, (ii) at the end of a nonterminal branch of the network, and (iii) at the end of the whole network. The results from fifteen speakers showed eleven out of fifteen using a rise at (i) and (ii), and fourteen out of fifteen using a fall at (iii). The notion of nonfinality is usually taken to mean sentence nonfinal because much intonation analysis has been done on written texts read aloud, and in this case, sentence-nonfinal intonational phrases commonly have a rise and sentence-final intonational phrases have a fall. But in conversational exchanges, such rises will often occur at the ends of sentences – insofar as sentences can be easily defined in such a register. So the use of a rise for nonfinality is really more relevant to discourse than to sentences. Thus we might say that one use of rising tones is to indicate discourse nonfinality and it might therefore be argued that this means topic continuity (and, hence, that use of a falling tone indicates topic completion). But clearly this is overstating the case. Firstly, the use of falls and rises is not wholly a question of topic nonfinality. In English, for example, that type of rise which is immediately preceded by a fall (and, hence, is known as a fall-rise) is often used for various meanings involving “implications,” for example, Be ˇcareful (where ˇ indicates the beginning of a fallrise on an accented syllable), where the meaning is “: : : or else X might happen.” Of course there is an unspoken “continuation rise” here, but this would extend the idea of continuation rise far further than is usual. Secondly, there is no evidence that the chunks of utterance delimited by each occurrence of a fall are regularly coextensive with those delimited by pitch height. Looking at matters in the reverse way, topics can with reasonable certainty be said to be delineated by pitch height and within such topics, rises may indicate continuity. But beyond this we are very much into a speaker-optional area; different speakers may actually choose to use different numbers of continuity-indicating rises. Alternatively, differences in proportion of rise usage can be related to crosslinguistic or dialectal variation (see Cruttenden 1995). There is a much discussed tendency among young speakers of Australian and New Zealand English to use a very high proportion of rises (see, for example, Guy et al. 1986; Britain and Newman 1992). Such rises appear to be used as a “check” on speaker-attention (Guy and Vonwiller 1984). While there is obviously some relationship between rises and topic continuity, there are at the same time so many other factors governing rise usage that it is difficult to get confirmation of any crosslinguistic universals or differences. A third way in which prosody contributes to discourse cohesion concerns the indication of the topic or the setting of a sentence. We are, of course, here specifically using the term “sentence topic” as opposed to the term “discourse topic,” which was dealt with above (also see, in particular, Bernini and Maslova, this volume). Where a sentence topic is marked by left dislocation (see Sornicola, this volume), it is com-

The de-accenting of given information: A cognitive universal?

313

monly set apart by being given a separate intonational phrase (with or without later pronominal recapitulation), for example: (1)

Soccer/ I love/ but as for rugby/ I hate it.

(2)

Quelle carte/ le abbiamo trovate noi/ dem.pl paper.pl 3pl.f.acc.clit have.1pl find.part.pst we probabilmente. probably ‘Probably we are the ones who found the papers.’ (Bernini 1992: 33)

Similarly “settings” (see Bernini and Maslova, this volume), by which we refer chiefly to sentence, time, and place adverbials, are optionally set apart by being given their own intonational phrase, for example: (3)

Unfortunately/ it didn’t work. Usually/ they’re not of a high-enough quality. In France/ they do it this way.

(4)

Quella notte/ passai molte ore a leggere il diario. dem night/ pass.1sg.pst many hour.pl to read.inf def diary ‘That night I spent many hours reading the diary.’ (Bernini 1992: 29)

(For more detail on the phrasing of such initial topics and settings, see Cruttenden 1986: 75–80, 145–147.) Fourthly, prosody is also related to discourse cohesion by the de-accenting of old or “given” information; a summary of some of the previous work in this area is given in Section 2, and the rest of this chapter investigates this phenomenon in a number of European languages. Before proceeding to Section 2, however, it must be made clear that none of the above relationships between discourse and intonation has ever been investigated crosslinguistically. There has always been an unwritten assumption that such things are in some way, or at least to some degree, common to all languages, that they represent some sort of cognitive universal. While this has never been stated explicitly, it has regularly been implied by the fact that writers rarely feel the need to say that of course their findings are limited to English or Dutch or whatever. It does indeed look from the limited number of languages separately studied for each of these phenomena that there is an element of universality, but at the same time there are clearly variations. So the use of rising tones for nonfinality looks to be a universal tendency; yet the need to mark nonfinality or finality clearly varies across languages and dialects. A higher mean pitch for topic-beginnings may well be universal, but the actual extra height of such extra mean pitch may vary from language to language. The delineation of a sentence topic by the use of a separate intonational phrase may

314

Alan Cruttenden

occur more regularly in some languages than in others. The de-accenting of old or “given” information may not be as obligatory in some languages as it is in English. The remainder of this chapter tries to establish a methodology for investigating this last phenomenon crosslinguistically and tentatively presents some first results.

2. De-accenting and given information The notion of de-accenting previously-mentioned information has been around in studies of English intonation for a long time. For example, Walker (1781: 15): “Those things with which we suppose our hearers to be pre-acquainted we express by such a subordination of stress as is suitable to the small importance of things already understood.” It continued to be mentioned by various writers throughout the nineteenth century, for example, Bell and Bell (1879: 11): “In phrases and sentences all words that express ideas new to the context are distinguished by accent; and all words that have been previously stated or implied are unaccented.” In the last thirty years, the notion has received a further boost of interest, principally through the influence of Halliday and his use of the terms “given” and “new” (see, in particular, Halliday 1967). In Halliday’s system of tonicity (i.e., tonic placement, where the “tonic” [alias “nucleus”] is the primary accent in an intonational phrase), the unmarked condition for tonic placement, that is, when the whole intonational phrase is new, is on the final lexical item; on the other hand, “marked tonicity [= nonfinal tonic] occurs, in general, under either (or both) of two conditions. Either some element other than the one just specified [= the last lexical item in the tone group] is ‘contrastive’; or the element just specified (and possibly others before it) is ‘given’ – has been mentioned before [= textual givenness] or is present in the situation [= situational givenness]” (Halliday 1967: 23). Whether situational givenness leads to de-accenting as certainly as textual givenness has always been unclear; Brown (1983) found that it did lead to de-accenting (although it did not involve lexical attenuation). Chafe (1974, 1976, 1994) emphasizes the role of the hearer and sees givenness as “fundamentally a matter of the speaker’s belief that the item is in the addressee’s consciousness” (Chafe 1976: 32). As Chafe (1976: 30) points out, we are not dealing just with what is known and what is not known (as Clark and Haviland [1977] interpret givenness in their notion of a given-new contract) but with known information which is activated into consciousness. Prince (1981, 1982) distinguished between discourse-new and hearer-new in devising a familiarity scale ranging from “evoked” (both hearer- and discourse-old) and “unused” (hearer-old but discoursenew) through “inferrables” to “brand-new” (hearer-new and discourse-new). Birner (1994) surveyed inversions in these terms (an inversion was defined as involving the appearance of the logical subject in postverbal position while some other constituent replaced it in clause-initial position). She found that 76% of inversions contained

The de-accenting of given information: A cognitive universal?

315

preposed discourse-old information followed by postposed discourse-new information. This suggests that accenting new information is frequently achieved, even in English, by changing the word order rather than changing the linear position of the accent. There is a general problem with unmarked tonicity (i.e. accent placement in “all-new” intonational phrases) in that there are a number of types of structures which regularly have the primary accent in nonfinal position even though all-new information is involved (Cruttenden 1990: 82–87), the most celebrated being sentences of the thetic type (see Sasse, this volume) like The telephone rang. There is also a problem with the occurrence of contrast on something which has been previously just mentioned (e.g. It’s not what you said but the way you SAID it [Maidment 1990: 22]); this must be regarded as a case of accented given information rather than new information; in this regard it is worth noting that an analysis of one set of data found that 23.5 % of stress (=accented) assignments were on given items in English (Lehman 1977: 218). Because there are problems with the independent assessment of givenness (i.e., one which does not become circular by deciding on givenness by reference to deaccenting) in particular intonational phrases, it is important to use an unarguable case of givenness in any crosslinguistic investigation. Repetition would seem to be such a case (see Nooteboom and Kruyt [1987: 1521] for the unacceptability of accented repeated items in Dutch). An archetypal repetition involves a lexical item repeated in an immediately following intonational phrase, for example, The stadium where Manchester United play is in the west of Manchester. In this example Manchester, as a repeated and hence given item, is de-accented and the accent is thrown back onto the word west. Repetition is used in this chapter to explore the correlation of givenness and de-accenting crosslinguistically.

3. De-accenting and re-accenting De-accenting of repeated items is certainly obligatory for the type of English which Halliday was describing (roughly RP) and for most other dialects of English. And it is, as has already been noted, easy to think of de-accenting as some sort of cognitive universal: “The principal linguistic effects of the given-new distinction in English, and perhaps all languages, reduce to the fact that given information is conveyed in a weaker and more attenuated manner than new” (Chafe 1976: 31, my italics). Yet such attenuation or de-accenting (and certainly its obligatory use) may not be as universal as first thought. There have been hints in the literature to this effect, for example: (i) Crystal (1975: 44) on Brazilian Portuguese: “This tendency [to keep the tonic syllable on the last item in the tone unit] applies even when one has repeated items in coordinate constructions, for example:

316

Alan Cruttenden

Esti livro custa cinco dolares1 e esti aqui tres dem book cost.3sg.prs five dollar.pl and dem here three dolares. dollar.pl ‘This book costs five dollars and this other one three dollars.’ ”

(5)

(ii) Ladd (1990: 813) gives examples from Rumanian, for example: (6)

[o s˘a vedem] ce ave¸ti s¸i ce nu ave¸ti. : : : subj see.1pl what have.2sg.prs and what not have.2sg.prs ‘: : : we see what we have and what we don’t have.’

(iii) Ortiz-Lira (1994: 441–444) gives many relevant examples from Spanish (principally Chilean, although his examples span most national and regional varieties), for example: (7)

a.

b.

c.

d.

¿Vienes/ o no vienes? come.2sg.prs or neg come.2sg.prs ‘Can you come or not?’ ¿Con leche o sin leche? with milk or without milk ‘With or without milk?’ ¿Si tú vas/ yo voy? if you.2sg.nom go.2sg I.sg.nom go.1sg.prs ‘If you go, can I go?’ Escribo canciones porque escucho cosas/ veo write.1sg.prs song.pl because listen.1sg.prs thing.pl see.1sg.prs cosas/ huelo cosas. thing.pl smell.1sg.prs thing.pl ‘I write songs because I listen to things, see things, smell things.’

(iv) Vallduví (1993) (referred to by Miller, this volume), reports that in Catalan the nuclear accent is fixed over the rightmost core clause boundary. Miller also suggests that Irish has the same restriction. (v) MacWhinney and Bates (1978), interested primarily in the developmental aspects of the given/new distinction, investigated the use of “emphatic stress” (= nonfinal tonic or nucleus) as one of the linguistic devices which might be expected to correlate with such use. There were 120 subjects in their experiment, 40 Americans, 40 Hungarians, and 40 Italians; within each language there were ten three-year-olds, ten four-year olds, ten five-year-olds, and ten adults. As expected, there was overall an increase in the use of “emphatic stress” with age, but, more interestingly for this chapter, English subjects of all ages, including the adults, used more emphatic stress than the Hungarian and Italian subjects. In fact the Hungarian subjects used it virtually not at all. The suggestion made was that the Hungarian and Italian subjects

The de-accenting of given information: A cognitive universal?

317

made a compensating use of word order variation (such a negative correlation between varying nucleus placement and varying word order has often been remarked on in intonational studies, e.g., Stockwell and Bowen [1965: 33] on Spanish, Agard and Di Pietro [1965: 72] on Italian, and Schubiger [1958: 82] on French). (vi) In more phonetically- rather than informationally-oriented research, Gårding (1981) elicited Greek, Swedish, and French versions of the sentence Madame Marianne Mallarmé has a mandolin from Madrid in a context which called for a focus on Mallarmé. She reported that “in Greek and French the pitch contour is flattened after focus. For Swedish, on the other hand, the situation is different. Here the accents still have their pitch configurations after focus” (Gårding 1981: 152). The reason for this was suggested to lie in the contrastive lexical status of the accents in Swedish. The hints in the literature above suggest that it might not be only in languages with lexical accents that post-focal accents are at least sometimes unflattened. (vii) Some varieties of English have also been reported to use lexical re-accenting. Vanderslice and Pierson (1967) give examples from Hawaiian English, for example: (8)

Forty t’ree per cent is gavment owned, and fifty seven per cent is privately owned.

Gumperz (1982: 125) and Bansal (1990: 228) both report a tendency in Indian English to keep stress on a final item, regardless of whether it involves given information, for example: (9)

A: B:

You just weren’t listening. I was listening.

Low (1994), from F0 measurements of the sentence I went to the shop to buy sweets but they had totally run out of sweets and a second similar sentence, found three speakers of Singapore English consistently re-accenting. (However, three British speakers also re-accented in half the cases, which may reflect a problem with read speech.) Similar suggestions have been made about Nigerian English and London Jamaican English in conference presentations by Egbe (1979) and Wells (1992). With such comments in mind, I constructed an intonational test in English involving ten different setting-response dialogue types (three different exemplars were used in setting-response 10 because of translational difficulties, which will become evident and be explained below) where the response involved a lexical item repeated from the setting. The pairs were loosely translated into eleven languages, and pairs of native speakers of the languages (varying in number according to availability) were recorded reading the settings and responses. The location of the intonational nucleus (= last pitch accent) was marked auditorily by me and by a trained intonationist working independently; there was 97 % agreement at the first analysis and the remaining 3 % of sentences were resolved by discussion. Besides English, for

318

Alan Cruttenden

which seven speakers were recorded as a “control” (and in which all seven speakers de-accented in every response with the exception of 10a – for discussion of this, see Section 4.10 below), the languages and number of speakers were as follows: French (principally Provence) 14, Italian (principally Sicily) 11, Lithuanian (Vilnius) 4, Arabic (Tunisia) 4, Albanian (Tirana) 4, Russian (Moscow 2, Dushambe 2) 4, Spanish (Chile 2, Castilian 1, Catalonian Spanish 1) 4, Greek (Athens) 4, Macedonian (Skopje) 2, Swedish (Skåne 3, Östersund 1) 4, and German (Westphalia 2, Munich 1, Tübingen/Hamburg 1) 4. It must, of course, be obvious that the test is in the nature of a prototype study of intonational typology; I was as much concerned to open up a new methodology for crosslinguistic intonational research as I was in the actual results. Thus obviously no statistical validity can yet be claimed. The English version of the ten settingresponse dialogues is given in Appendix 1. The response versions in the eleven other languages, together with the results in terms of nucleus placement, are given in Appendix 2. In Appendix 2 and in the body of the chapter, Russian is transliterated according to the type of transliteration given in Appendix 3 of Comrie (1981), Macedonian is transliterated according to Friedman (1993), Greek is transcribed phonemically as in Joseph and Philippaki-Warburton (1987), and the Tunisian Arabic (not of course being generally written at all) is transcribed according to a relatively narrow system devised by Giuliano Bernini. The extent of de-accenting in each language is shown in Table 1. I now discuss the results for each setting-response separately in Section 4 (paying particular attention to cases of re-accenting) before discussing the results in more general terms in Section 5. Intermorphemic transcriptions for all languages are given in Appendix 2 and are only repeated in the text where a particular syntactic point is at issue.

4. Settings and responses 4.1. Setting-response 1 The English version of the first example is as follows: A: B:

If you don’t hurry up, you’ll be late. I don’t care if we are late.

In this example late, a repeated item, and we, with identical, though shifted, reference to you, are treated as old information in English and are hence de-accented. There are two alternatives for the placement of the accent, depending on the reduction or nonreduction of we’re or we are. With reduction the accent will be on care hence I don’t care if we’re late; without reduction the accent may be on are, hence I don’t care if we are late. The latter accentuation emphasizes the positive polarity of the

The de-accenting of given information: A cognitive universal?

319

Table 1. Re-accenting in twelve languages

Albanian (4) Arabic (4) English (7) French (14) German (4) Greek (4) Italian (11) Lithuanian (4) Macedonian (2) Russian (4) Spanish (4) Swedish (4)

1

2

3

4

5

6

7

8

9

10a

10b

10c

0 1 7 3 0 0 0 0 1 1 3 3

1 0 7 6 1 1 0 0 1 0 1 4

0 0 7 13 0 1 0 0 1 0 4 3

0 4 7 3 0 1 5b 0 2 0 4b 0

3 3 7 11 0 3 — 0 0 1 4 3

2 1 7 3 0 4 3 0 2 0 4 2

0 3 7 2 0 0 1 1 0 0 3 2

0 4a 7 14a 0 0 — 0 0 — 2 3

3 3 7 14 — — 10 — — 1c 4 —

3 3 3 14 — — 10 — 0 — 3 —

2 4 7 10 0 — 8 4 0 3 — 2

— 3 7 11 0 0 7 2 1 3 — 1

(Total number of informants shown in brackets) a Involves the change of gender b Involves a change to a causative verb c Only one out of two (other two responses irrelevant)

outcome. When we look at other languages, we find that most languages behave similarly to English. Indeed almost all my varying numbers of informants de-accented. In French, where I had fourteen informants for the sentence Je m’en fiche d’être en retard, eleven put the last accent on fiche and three re-accented retard. In the Spanish sentence No me importa si llegamos tarde, three of the four speakers re-accented tarde. This accords with the regular re-accenting in other Spanish setting-responses and the evidence from Ortiz-Lira (1995). Three of the four Swedish speakers reaccented, which accords with the evidence from Gårding (1981). One speaker out of four re-accented in Arabic; in the light of varying response from other settingresponses, this may or may not represent a genuine option in Arabic. (It might be considered as an occasional vagary from an inexpert reader.) One of the four speakers of Russian re-accented, as did one of the two speakers of Macedonian, although evidence from other sentences firmly suggests that no significance should be attached to these readings. The overall conclusion for this setting-response pair is that it produces obligatory de-accenting in all the languages tested, with the exceptions of optional re-accenting in French, Spanish, and Swedish.

320

Alan Cruttenden

4.2. Setting-response 2 This sequence is as follows: A: B:

Would you like to come to dinner tonight. I’m afraid it’s only chicken. I don’t like chicken.

In this example chicken is repeated, the repetition involving only the one word where in setting-response 1, a whole clause was repeated. In English, the accent is thrown back onto like, leaving chicken as an unaccented “tail” (although it may realize the rising part of a falling-rising tone on like). The situation for the languages tested is much the same as for setting-response 1 (it is evidently not significant whether a word or a clause is repeated). Six of the fourteen French speakers re-accented café in the sentence Merci, je bois pas de café, whereas four preferred to accent pas and four bois (the possibility of accenting the negative particle in counter-assertative or counter-presuppositional clauses is a further type of accenting needing crosslinguistic investigation). All four speakers of Swedish re-accented. In other languages, one speaker re-accented in Arabic, German, Greek, Macedonian, and Spanish. The speaker of Greek had a particularly wooden reading style and also re-accented in setting-responses 3 and 4; I judge all these readings not to be significant and therefore do not discuss them further under Sections 4.3 and 4.4. The same applies to the speaker of Macedonian. One German speaker divided the utterance into two intonational phrases and the additional phrase and nucleus on Hühnchen was evidently an emphatic device. In the light of results from other setting-responses, re-accenting in setting-response 2 may or may not represent a genuine option in Arabic but almost certainly does represent such an option in Spanish (indeed this sentence is definitely the exception in presenting three out of four cases of de-accenting).

4.3. Setting-response 3 This sequence is as follows: A: B:

You need a pair of black shoes for the wedding. I’ve already got a pair of black shoes.

In this case a phrase, a pair of black shoes, rather than a word or a clause is involved. In English the phrase would be de-accented, throwing the accent back onto got. Once again the pattern of results is very similar for other languages, except that now we have all but one of the French speakers re-accenting paire in the sentence J’en ai déjà une paire and all four speakers of Spanish re-accenting negros in Yo ya tengo un par de zapatos negros. Like the earlier responses, three out of four of the Swedish

The de-accenting of given information: A cognitive universal?

321

informants re-accented. In the first three setting-responses we have no cases of reaccenting in Italian; this was unexpected in the light of the comments of Ladd (1990) and other informal personal communications. In this setting-response it may be that the use of the anticipatory le in Ce le ho già un paio di scarpe nere reinforces the de-accenting of the final phrase. One speaker of Macedonian re-accented; as in the previous pairs this is probably not significant.

4.4. Setting-response 4 This sequence is as follows: A: B:

Why do you keep getting angry? Because John makes me angry.

In English the accent would be moved forward to makes and angry would be deaccented. However, the non-English versions of this pair varied somewhat in structure from language to language, partly because informants in some languages made predicates the same (e.g., A: : : : be angry, B: : : : be angry; or A: : : : make angry, or B: : : : make angry), and partly in the varying use of a causative verb in the response where the setting used a copula plus an adjective. Those languages where a change from adjective to causative was involved (in final position) accented the causative (Italian arrabbiare, Macedonian luti, Spanish enoje – a causative verb was also involved in Greek but this was simply repeated from the setting and was not accented by any of the four speakers). Although I have shown these in Table 1 as examples of re-accenting (as opposed to de-accenting), it is probably more correct to regard them simply as examples of “accenting.” The decision to regard the accent on the causative in these cases as accentuation or re-accentuation relates to the decision to treat “given” information as notional or linguistic, that is, is it the notion of “angry” or the word angry which is given? The accenting of the causative verb can either be regarded as an example of the re-accenting of the repeated piece of notional information of being angry (notionally given) or as the accentuation of the word representing the causative verb not previously mentioned (linguistically new) or, lastly, as the accentuation of a new piece of notional information conveyed by the grammatical change (notionally new). No certain choice between the three can be made, although intuition suggests that the last of the three alternatives is the correct one (and this is partly confirmed by the results under Section 4.8 below). This is again an area where further work could be revealing theoretically, both for intonation and lexical structure. In those languages which did not involve a change from adjective to causative, the two languages where some speakers used re-accenting were French, where a minority of three speakers re-accented colère, and Arabic, where all four speakers re-accented the adjective:

322

Alan Cruttenden

(10)

A:

B:

Qale:š be:qi: metGašiš? ˙ why CONT (= continuously PART) angry ‘Why are you getting angry with me?’ Qala: xa:ter Yahya: be:qi: metGašiš. ˙ ˙ because John CONT (= continuously PART) angry ‘Because John gets angry.’

The unambiguous findings in the case of Arabic make clear that at least in certain structural relationships (between that of the setting and the response) re-accenting is a possibility in Arabic. The case of Swedish is interesting. The construction of the response almost exactly parallels that of English: (11)

A:

B:

Varför blir du alltid så arg på mig? why become.prs you.sg.nom always so angry on I.obj ‘Why do you keep getting angry with me?’ För att du gör mig arg. for that you.sg.nom make I.obj angry ‘Because you make me angry.’

and the movement of the nucleus from the adjective to the auxiliary causative also follows the English pattern. The interest lies in the fact that this is the only settingresponse pair in Swedish (apart from morphological de-accenting under (10)) where a majority of the Swedes de-accented. I have no explanation why a causative auxiliary should have this unique effect. The significance of the Macedonian response is not altogether clear. The setting-response was: (12) a.

A:

B:

Zošto postojano se lutiš? why always rfl anger.2sg ‘Why are you always getting angry?’ Zašto Džon me luti. because John I.obj anger.3sg ‘Because John makes me angry.’

In the setting the verb is reflexive, whereas in the response it is transitive. Both informants accentuated the transitive verb of the response. This may parallel the Italian and Spanish accenting of the causative verb and hence more correctly involve accented new rather re-accented given.

4.5. Setting-response 5 This sequence is as follows: A: B:

He earns at least thirty thousand pounds. I think he earns forty thousand.

The de-accenting of given information: A cognitive universal?

323

In English the accent in the response is on forty. In this sequence the notion of “contrast” plays a part which it has not played before: thus besides thousand being repeated, forty is now contrasted with thirty (amounts were of course adjusted to the local currency). It would seem a reasonable hypothesis that a combination of repeated and contrasted items would lead more strongly to de-accenting than repeatedness alone. Once again, re-accenting occurred in the expected languages. In Swedish, three out of four re-accented -tus-. In French, eleven speakers re-accented francs in quatre vingt francs in reply to soixante mille francs; in Spanish, all four speakers re-accented mil in cuarenta mil in response to treinta mil dólares; in Arabic, three out of four speakers re-accented milyu:n in tisQai:n milyu:n in reply to Tmeni:n milyu:n. Additionally, three of the four speakers˙ of Albanian also re-accented mijë in dyzet mijë in response to tridhjetë mijë lekë; and three of the four Greek speakers re-accented xiliádhes. One speaker of Russian re-accented tysjaˇc. While the single re-accenting in Russian can be regarded as a rogue result, the three out of four for Albanian and Greek look to be significant. This actually contradicts the hypothesis of the last paragraph, contrast producing a somewhat wider use of re-accenting.

4.6. Setting-response 6 This sequence is as follows: A: B:

I make the answer sixteen point one. Well, I make it twenty-six point one.

As in setting-response 5, here a positive accentuation for contrast as well as the de-accenting of old information is involved. In this example the point one is deaccented in English and the accent in the response is on six. In French, the sentence Eh bien ma réponse est vingt six virgule un produced de-accenting with the accent on six from eleven of the fourteen informants, and re-accenting with the accent on un from three informants. The four Spanish speakers re-accented uno, two of the four Swedish speakers re-accented ett, one of the four Arabic speakers re-accented wahd, and, as in the previous setting-response, a majority of the Albanian speakers ˙ re-accented. Additionally and for the first time (if we exclude the causative example) a minority of Italian speakers also re-accented, A me invece ventisei virgola uno producing de-accenting on ventisei from eight informants, but re-accenting with the accent on uno from three informants. Of the four Greek speakers, one re-accented ena; in this language the contrast was between íkosi-éksi and triánta-éksi and the other three speakers re-accented the éksi. The two Macedonian speakers also reaccented eden; and again, as in setting-responses 1–3, this is taken as a result of the wooden reading style of one of the two Macedonian speakers.

324

Alan Cruttenden

As in setting-response 5, the presence of contrast increases the number of languages using re-accenting (although it does decrease the numbers using it in French). Hence the hypothesis that contrast increases the likelihood of de-accenting (and hence increases re-accenting) is again contradicted. The sort of contrast in Sections 4.5. and 4.6 involves numerals. It may be that numeral contrast is rather special and generally allows re-accenting more easily than other types of contrast which are investigated in Sections 4.7 and 4.8.

4.7. Setting-response 7 The sequence here was: A: B:

I did all the work. You mean your sister did all the work.

Once again, contrast is present, but this sequence involves person contrast rather than number contrast as in setting-responses 5 and 6. Re-accenting is less widespread for this sequence than for sequences 5 and 6. Only two speakers of French re-accented travail, although all four Spanish speakers re-accented trabajo, three of the four Arabic speakers re-accented the l-kul of l-xedma l-kul (‘work all’), and two of the four Swedish informants re-accented jobbet. One of the four Lithuanian speakers also re-accented but, in the light of the other results for Lithuanian, this looks like a rogue result. There is overall somewhat less re-accenting in this case where person contrast rather than number contrast is involved and the language spread of re-accenting is now similar to that for those setting-responses where given information but not contrast was involved. The one exception is French, where, although there is less re-accenting when compared with number contrast, the two speakers re-accenting compared with larger numbers re-accenting in setting-responses 1–4. In general, the findings from this setting-response may indicate the expected effect of the hypothesis that contrast will decrease the possibility of re-accentuation.

4.8. Setting-response 8 The sequence here was: A: B:

Do you like your new house? Yes, I’m happy, and my wife is very happy.

This example is somewhat different from those preceding: firstly, the repetition is contained within the response, and secondly, a special type of contrast is involved

The de-accenting of given information: A cognitive universal?

325

produced by the use of the intensifier very. In English this intensifier is itself accented along with the de-accentuation of happy. The results here are clear-cut. Reaccentuation is present in only four languages. All fourteen French informants, three of the four Arabic speakers, two of the four Swedish speakers, and two of the four Spanish informants re-accented (the re-accentuation in French and Arabic involved a change from masculine to feminine in the form of the adjective [content/contente, ferha:n/ferha:na]). Otherwise there was no re-accenting. ˙ ˙ are clear from the results of this setting-response: Two points 1. A change of gender produces accenting. Like the accenting of causative verbs under Section 4.4 such a result could be regarded as notional re-accenting of the given notion of “happy,” or linguistic accenting of the feminine form of the word, or, finally, notional accenting of “feminine.” Since there are almost no other examples of re-accentuation for this setting-response, it appears that the second or third solution is the correct one and hence, we are here dealing with cases of accented new information rather than the re-accenting of old information. 2. If responses of the type under 1. are excluded, this is the clearest case yet of widespread de-accentuation. Evidently the use of an intensifier leads more regularly to de-accenting than other cases involving given information, including those involving given information plus contrast. In addition to the languages (French and Spanish) where gender change was involved, the exceptions to de-accenting are in Swedish and Arabic, as would be expected from other setting-responses.

4.9. Setting-response 9 This sequence concerns the reading of football results (for a detailed exposition of this topic in English, see Cruttenden 1974). The English setting-response is as follows: A: B:

What was the score? Liverpool 1 (one), Manchester United 1 (one).

In English, the repetition of the word one produces de-accenting and hence the last accent is thrown back onto the preceding word United. It might be objected that this is a very specialized type of intonational usage but, in fact, it is merely an example of a more widespread use of the pattern. The same sort of correlated construction produces sentences like John has two and Mary has two and many similar. This test does not work (as I found out too late) in all languages, because a different way of reading results on radio and television is often used corresponding to Liverpool–Manchester United 1–1. This applied in German, Greek, Lithuanian, Macedonian, and Swedish, where the informants said that they could not possibly

326

Alan Cruttenden

read results in the same way as in English; in addition two of the four Russian informants responded in this nonrelevant way. Like setting-response 8, the repetition in this sequence is contained within the response. However, in this case we are not dealing with an intensifier but with contrasted teams in the context of the same number, which produces the same strong tendency towards re-accenting as in the two previous setting-responses involving contrast and numbers (setting-responses 5 and 6). For those languages where the test was legitimate, the findings were quite unambiguous: only two informants (one Italian and one Russian) out of the whole total used de-accenting. What this means is that de-accenting appears to be impossible with this setting-response for French, Italian, Spanish, and Arabic. For example, the Arabic production was always: (13)

l-Pafri:qi: wahd u-t-tara:jˇ wah-d. ART-Afriqi one˙ and-ART-Taraji one˙ ‘Afriqi 1, Taradgi 1.’

and the Italian (except for one speaker) was: (14)

Inter uno Roma uno. Inter one Roma one ‘Inter 1, Roma 1.’

This sequence provides further evidence for de-accenting being more limited where numbers are involved and, hence, re-accenting being more common.

4.10. Setting-response 10 The sequences discussed here involved the following pairs (three pairs were used so that at least one of the three pairs in all the languages tested would use the same root together with an affix): (a)

A: B:

I’m very interested in home news. It’s necessary to keep a balance between national and international affairs.

(b)

A: B:

That reply is correct. You mean it’s incorrect.

(c)

A: B:

I think the locals are very friendly. I think they’re very unfriendly.

I call this type of de-accenting morphological de-accenting (since a word is subsequently de-accented when becoming one morpheme of a complex word). In English de-accenting was predicted in such cases: international affairs, incorrect, unfriendly. The seven English responders did so de-accent in setting-response 10b and 10c but

The de-accenting of given information: A cognitive universal?

327

the responses were not entirely as expected in 10a. It was assumed that “affairs” would be treated loosely as a synonym of “news” and, hence, as given information and consequently de-accented. But in fact, four of the seven speakers accented “affairs”; I analyze these in Table 1 as examples of the absence of de-accenting although it would probably be more correct to view the responders as not accepting “affairs” as a synonym of “news” and hence treating it as new rather than given information. The other three responders in 10a did de-accent “affairs” and, more relevantly to morphological de-accenting, moved the accent from -na- to in-. In French and Italian, a large majority favored re-accenting in all three subsequences, for example, French internationales, incorrecte, impolis, and Italian internazionali, dissimili, antipatica. In Arabic, results were similar although only in one case was proper morphological re-accenting present, namely, u-l-xarijˇeyy-a; in the other two cases the negative muš was used giving muš sehi:h and muš metrobbi:n ˙ ˙ (and hence we are dealing with re-accentuation following˙a negative rather than morphological re-accentuation). In Albanian, Lithuanian, Russian, and Spanish, where smaller numbers of responses were recorded, the results nevertheless favored reaccenting, that is, Albanian ndërkombëtare, nuk drejtë (this latter a NEG example like those for Arabic), Lithuanian neteisingas, nedraugiški, and Russian nepravil’ny, neprivetlivy and Spanish internacionales. German, Greek, and Swedish (only marginally) favored the English model of de-accenting (while one speaker of Macedonian put the accent on mnogu ‘very’). In those languages which have a widespread use of re-accenting, at least in some types of repetition, it appears that morphological de-accenting is generally an unlikely option. This is true of Arabic, French, Italian, and Spanish. It also appears true of some languages which elsewhere do de-accent: Lithuanian and Russian (both of which regularly de-accent elsewhere), and Albanian (which de-accents apart from cases of numeral contrast).

5. Conclusions 5.1. Re-accenting in Romance and other languages Some languages clearly favor de-accenting or re-accenting more than others, as can be seen from Table 2 and Figure 1. Re-accenting is in general common in the Romance languages. The figure for Spanish re-accenting is 32/40 (see also Ortiz-Lira 1995), for French it is 104/168, although for Italian, while the total figure is 44/110, most of the re-accenting is in the morphological cases, and it is clear that re-accenting is rather less prevalent than in Spanish and French. Re-accenting also appears to be common in Arabic (28/48), in Swedish (23/40) and, to a lesser extent, in Albanian (14/33). On the other hand,

328

Alan Cruttenden

Table 2. De-accenting and re-accenting totals

Albanian Arabic English French German Greek Italian Lithuanian Macedonian Russian Spanish Swedish

Total

Re-accenting

De-accenting

33 48 84 168 40 36 110 40 22 38 40 40

14 28 4 104 1 9 44 7 8 8 32 23

19 20 80 64 39 27 66 33 14 30 8 17

in Greek (9/40), Lithuanian (7/40), and the Germanic and Slavonic languages tested (German 1/40, Macedonian 8/22, Russian 8/38 [but including six cases of morphological de-accenting]), re-accenting is much less common. But it is by no means a clear-cut difference: while de-accenting is obligatory in English (and perhaps some other non-Romance languages), in many cases, de-accenting and re-accenting both appear to be options (probably in the Romance languages). There are a number of possible factors which might govern the choice between the options: 1. Dialectal variation. The only relevant evidence here is from Spanish. Of the four Spanish speakers transcribed, two were from Santiago, Chile, one was from Madrid, and one was from Barcelona; there was no significant difference in their responses. 16/20 were re-accentings in the Santiago data, while the Madrid and Barcelona speakers re-accented in 8/10 responses each. Also Ortiz-Lira (1995) found many examples of re-accenting in Mexican and Argentinean Spanish. This suggests that re-accenting is not limited to a particular dialect or dialects of Spanish. 2. Interspeaker variation. I have no evidence that some speakers systematically use re-accenting while others do not, apart from a small number of cases, in particular one speaker of Greek, who seemed unable to read except in a wooden style. Incompetence in reading may account for a number of other exceptional responses in the data; in particular, it may account for the re-accenting in the Macedonian responses in setting-responses 1, 2, 3, and 6. 3. Stylistic variation. Re-accenting might be used as a discourse device (emphasizing cohesion) rather like the accenting of prepositions, auxiliaries and the verb be in English (for example, the possible accenting of are in setting-response 1:

The de-accenting of given information: A cognitive universal? 120

329

Re-accenting De-accenting

100

Percentage

80 60 40 20 0 GER

ENG

LITH

RUS

GRK

MAC

ALB

ITAL

SWED

ARAB

FREN

SPAN

Language

Figure 1. Re-accenting and de-accenting

I don’t care if you are late; also see Cruttenden and Faber 1991). Indeed reaccenting itself appears to be sometimes used as a discourse device even in English. For example, Maidment (1990: 22) gives the following example: (15)

A: B:

Why don’t you go to London. I don’t want to go to London.

As Maidment (1990: 22) points out: “At first sight [this example] seems a little odd, but if we assume that the two speakers have been through this argument before, then its oddity disappears.” The theoretical oddity with such an example and others like it in English is that de-accenting, by marking the end of the new information and indicating the extent of the following given information, is itself a discourse device; re-accenting then appears to be an overlaid discourse device which depends on the expected use of de-accenting. Some such interaction between de-accenting and reaccenting may be operating more regularly in those languages where re-accenting occurs more frequently.

5.2. Contrast Phrases like “contrastive accent,” “contrastive stress,” and “contrastive focus” are frequently used in linguistics. One implication in these phrases is that there is some special sort of accentual pattern present; but, pace Chafe’s (1976: 35) implication of high pitch, there is certainly no unique pitch movement associated with contrast in

330

Alan Cruttenden

English. Contrast does often involve a nonfinal accent placement, though this sort of placement is not unique to contrast (see, for example, setting-responses 1–3 above). Another implication in phrases like “contrastive accent” is that the notion of “contrastive” is well-defined, but in practice, it is not. The occurrence of any word in any sentence is contrastive in one way in that it contrasts with all other words which might fill the same slot. But “contrastive” must obviously be given a more limited definition if it is to be of use in discourse analysis. Taglicht (1982) proposes restricting “contrastive” in various ways involving (i) regularly recurring contextual notions like red and green for traffic lights, (ii) a pair of “primary opposites” (the word “opposites” here has itself to be widely understood – it seems to mean the implicit or explicit opposing of one term to another), and (iii) two pairs of primary and secondary opposites, for example, He votes Conservative and they vote Labour. Chafe (1976: 34) interprets Taglicht’s point (ii) somewhat more widely: “The only consistent factor seems to be that the speaker assumes that a limited number of candidates is available in the addressee’s mind (whether or not the addressee could list all of them). Often the number is one, often it is larger, but when it is unlimited the sentence fails to be contrastive.” A number of the setting-responses described in this chapter involve a contrast of the sort in (ii) and (iii) above. In Section 4.4 the original version of the settingresponse involved a pair of primary and secondary opposites “John” and “you,” and “getting angry” and “making angry.” In Sections 4.5 and 4.6 there are numeral contrasts between two pairs of numbers. In Section 4.7 there is a person contrast between “I” and “your sister.” In Section 4.8 there are a pair of primary and secondary opposites “I” and “my wife” and “happy” and “very happy” (it is not clear what sort of contrast is involved in the secondary opposite here – perhaps a contrast between “unmarked normal level” and “very”). In Section 4.10 contrast is also involved, but it is at the morphological rather than the word level and this is discussed in Section 5.3 below. It was previously suggested in Section 4.5 that the presence of contrast might produce additional pressure to de-accenting. The results from Sections 4.4 to 4.8 are relevant to this hypothesis. The results from Section 4.4 are difficult to interpret for the hypothesis because of the alterations introduced by the informants: some changed the setting-response into a primary only contrast, namely, A: you BE angry, B: John BE angry, others changed the response to a causative verb. However, Sections 4.5– 4.8 are interpretable. In Sections 4.5 and 4.6, where a numeral contrast is used, there is a slightly wider use of re-accenting when compared with Sections 4.1–4.3, which does not support the hypothesis; whereas in Sections 4.7 and 4.8 involving person contrast (in Section 4.8 persons are involved in the primary contrast even though the contrastive accent is on the secondary contrast of “very”), there is lesser use of re-accenting compared with Sections 4.1–4.3, which supports the hypothesis. There is evidently something particular to numerals which does not favor de-accenting;

The de-accenting of given information: A cognitive universal?

331

whereas other types of contrast, on the basis of the slim evidence here, appear to encourage de-accentuation.

5.3. Discourse and grammar There is obviously no clear-cut distinction between discourse and grammar. Often particular grammatical structures will be the means whereby a discoursal connection is established, as in clefting or left dislocation. In the context of this chapter it has been pointed out that accent placement can affect discourse or be constrained in its effects on discourse because of interactions between grammatical factors and potential accent placements. Most obviously, as shown in setting-response 10, morphological de-accenting is more restricted in some languages than in others and may also be more restricted than word de-accenting in a particular language. Thus, as expected from the more general results on re-accenting, Arabic, French, Italian, and Spanish disfavor morphological de-accenting. Also as expected, German and Greek follow the English pattern of favoring de-accenting. But there is a group of languages in the middle – Albanian, Lithuanian, and Russian – which, although favoring de-accenting at the word level, favor re-accenting at the morphological level. Morpheme de-accenting has a different potential crosslinguistically than word de-accenting. A much stronger candidate for a linguistic universal is grammatical newness (the usual discussions of new and given information refer to lexical newness). In settingresponse 4, a change to a causative verb or to an auxiliary causative produces accenting on that verb or auxiliary. In Spanish, the change from llevas enojando to enoje produces accenting on the causative from all speakers; in Italian, a change from arrabbi to fai arrabbiare produces accenting on arrabbiare (5/11) and on fai (3/11); in Russian, a change from serdit’sja to zastavljaeš’ serdit’sja produces the accent on zastavljaeš’ (3/4); and in Swedish, blir arg becoming gör arg produces an accent on gör. Only in French is there an ambiguity in the results: the change from es en colère to mets en colère resulting in the accent on colère (3/14), mets (3/14) but on preceding toi (8/14). Similarly, a change of gender also leads to accenting of the new. In setting-response 8, a change from content to contente produced accenting in French (14/14), and a change from ferha:n to ferha:na accenting in Arabic (3/4). Grammatical new˙ as much˙ as lexical newness. ness produces accenting Probably only some of the factors which govern de-accenting and re-accenting have here been touched on, yet it is clear that de-accenting of given information is not a simple cognitive universal. On the contrary, various discoursal and grammatical factors interact with language-specific variation to produce both optional and obligatory de-accenting and re-accenting in particular examples. The study of the varied

332

Alan Cruttenden

contribution of accent and intonation to discourse cohesion in different languages is only just beginning. This chapter has represented an attempt at a new methodology and some new findings.

Appendix 1. De-accentuation test (general) Guidelines to informants: (a) Please do a rough translation of setting and response keeping repeated words if possible. In (10a), (10b) and (10c) morphological identity is crucial; if this is not possible with the given examples, please construct alternatives. More generally there will be considerable cultural problems with languages outside European cultures; in these cases, the nearest cultural equivalents are of course acceptable (the key factor is that there should be repeated words or morphemes towards the end of the response). (b) One speaker to say the settings; a second speaker the responses. Record the setting-response pairs. (c) Leave an interval of approximately 10 secs between one setting-response pair and the next (so that there is no interference between pairs). (d) Let me have the recording together with a copy of the translations, if possible with a literal translation back into English underneath the non-English versions.

Setting-response pairs: (1)

A: B:

If you don’t hurry up, we’ll be late. I don’t care if we are late.

(2)

A: B:

Would you like to come to dinner tonight? I’m afraid it’s only chicken. I don’t like chicken.

(3)

A: B:

You need a pair of black shoes for the wedding. I’ve already got a pair of black shoes.

(4)

A: B:

Why do you keep getting angry? Because John makes me angry.

(5)

A: B:

He earns at least thirty thousand pounds. I think he earns nearly forty thousand. (Adjust amounts to local currency).

The de-accenting of given information: A cognitive universal?

333

(6)

A: B:

I make the answer twenty-six point one. Well I make it thirty-six point one. (An earlier version for French, Italian, Macedonian, Russian, and Swedish used the contrast between sixteen and twenty-six, which had a lesser possibility of re-accenting).

(7)

A: B:

I did all the work. You mean your sister did all the work.

(8)

A: B:

Do you like your new house? Yes, I’m happy, and my wife is very happy.

(9)

A: B:

What was the score? Inter-Milan 1 Roma 1. (Use names of football teams in relevant country.)

(10) a.

A: B:

b.

A: B: A: B:

I’m always more interested in home news. I think it’s important to keep a balance between national and international affairs. That reply is correct. You mean that reply is incorrect. I think the locals are very friendly. I think they’re very unfriendly.

c.

Appendix 2. Translated setting-responses and focal accent location A Albanian (1)

S’ ka gjë nëse vonohemi. NEG have.3SG.PRS thing if be late.1PL.PRS.PROG ‘I don’t care if we are late.’

(2)

Nuk më pëlqen pula. NEG I.DAT please.3SG.PRS chicken.DEF ‘I don’t like chicken.’

(3)

Kam këpucë të zeza. have.1SG.PRS shoe.PL PART black.F.PL ‘I’ve got black shoes.’

(4)

Sepse Gjoni është i zemëruar. because John.DEF COP.3SG.PRS PART angry ‘Because John is angry.’

334

Alan Cruttenden

(5)

Mendoj se fiton rreth dyzet mijë. think.1SG.PRS that earn.3SG.PRS about forty thousand ‘I think he earns about forty thousand.’

(6)

Kurse unë iu përgjigja while I.NOM 3SG.DAT.3.SG.RFL answer.PST.DEF pyetjes tridhjetë e gjashtë një. question.DEF.DAT thirty and six one ‘I make it thirty-six point one.’

(7)

Deshe të thuash motra want.2SG.PST.DEF COMP say.2SG.PRS.SUBJ sister.DEF jote e bëri të gjithë punën. your.2SG.F PART do.3SG.PST.DEF all work.DEF.ACC ‘You mean your sister did all the work.’

(8)

Po, jam i lumtur kurse ime shoqe është yes COP.1SG.PRS PART happy while my.F wife COP.3SG.PRS shumë e lumtur. very PART happy ‘Yes, I am happy and my wife is very happy.’

(9)

Flamurtari dy, Besa dy. Flamurtari.DEF two Besa.DEF two ‘Flamurtari 2, Besa 2.’

(10) a.

Eshtë e domosdoshme të ruhet baraspesha COP.3SG.PRS PART important that keep.3SG.PRS balance mes çështjeve kombëtare dhe ndërkombëtare. between affairs national.PL and international.PL ‘It is important to keep a balance between national and international affairs.’ Ju mendoni se kjo përgjigje nuk you.NOM.2SG mean.2SG COMP DEM.F answer NEG është e drejtë. COP.3SG.PRS PART correct ‘You mean that that answer is incorrect.’

b.

Results from three Albanian informants (1)

ka (2), gjë (1)

(2)

nuk (2), pula (1)

(3)

kam (3)

The de-accenting of given information: A cognitive universal?

(4)

Gjoni (2), është (1)

(5)

mije (3)

(6)

tridhjetë (1), gjashtë (2)

(7)

jote (3)

(8)

shumë (3)

(9)

dy (second occurrence)(3)

(10) a. b.

nderkombetare (3) drejtë (2), nuk (1)

335

B Arabic (Tunisian) (Relatively narrow transcription; [z] = pharyngealized interdental fricative, i.e., the ˙ emphatic counterpart to [ð]). (1)

ma yi-hemn-i:-š iðeke:n n-usl-u maxr. ˙ NEG 3SG.M-concern.1SG-NEG if 1-arrive.IPFV-PL late ‘I’m not worried if we are late.’

(2)

ma n-akel-iš ˇjejˇ. NEG 1SG-eat.IPFV-NEG chicken ‘I don’t eat chicken.’

(3)

le:, Qand-i: sabba:t ekhel. ˙ ˙ black no at-1SG shoe.PL ‘I’ve got black shoes.’

(4)

Qala: xa:ter Yahya: be:qi: metGašiš. ˙ ˙ because John CONT (= continuously PART) angry ‘Because John is still angry.’

(5)

taqri:b tisQai:n n-zon Pilli yi-rbah ˙ ˙ ˙ almost ninety 1SG-think.IPFV COMP 3SG.M-earn.IPFV milyu:n. million.PL ‘I think he earns almost ninety million.’

(6)

wa Pana n-sammi:-h sitta-u-TlaTi:n wahd. and I 1SG-name.IPFV-3SG.M six-and-thirty one˙ ‘And I make it thirty-six point one.’

336

Alan Cruttenden

(7)

yi-thebb t-qu:l Pilli uxt-ek heyye ˙ 2SG.M-want.IPFV 2SG.M-say.IPFV COMP sister-2SG.M she Pilli Qaml-at l-xedma l-kul. really do.PFV-3SG.F ART-work ART-all ‘You mean your sister really did all the work.’

(8)

Pey Pana ferha:n u-mart-i: yeser ferha:n-a. ˙ ˙ yes I happy and-wife-1SG very happy-F ‘Yes, I’m happy and my wife is very happy.’

(9)

l-Pafri:qi: wahd u-t-Tara:jˇi: wahd. ART-Afriqi one˙ and-ART-Taraji one˙ ‘Afriqi one, Taraji one.’

(10) a.

yi-lzem mitwezn-i:n l-Pumu:r l-dexileyy-a 3SG.M-COP.must.IPFV balanced-PL ART-affair.PL ART-internal-F u-l-xarijˇeyy-a. and-ART-external-F ‘It is necessary to keep a balance between internal and external affairs.’ yi-thebb t-qu:l Pilli l-jˇawa:b ha:dha ˙ 3SG.M-want.IPFV 2SG.M-say.IPFV COMP ART-answer this muš sehi:h. ˙ ˙ ˙ NEG correct ‘You mean this answer is not correct.’ n-zon Pilli-humma yeser muš metrobbi:n. ˙ 1SG-think.IPFV COMP-3PL.M very NEG friendly ‘I think the people are not very friendly.’

b.

c.

Results from four Arabic informants (1)

yi-hemn-i:-š (4)

(2)

n-akel-iš (4)

(3)

Qand-i: (4)

(4)

metGašiš (4) ˙ milyu:n (3), tisQai:n (1) ˙ sitta-u-TlaTi:n (3), wahd (1) ˙ l-kul (2), uxt-ek (1), Qaml-at (1)

(5) (6) (7) (8) (9)

ferha:n-a (3), yeser (1), ˙ wahd (second occurrence)(4) ˙

The de-accenting of given information: A cognitive universal?

(10) a. b. c.

337

u-l-xarijˇeyy-a (3), u-l-xarijˇeyy (1) sehi:h (4) ˙ ˙ ˙ metrobbi:n (3), muš (1)

C French (1)

Je m’ en fiche d’ être en I.NOM RFL it.GEN be indifferent.1SG.PRS of COP.INF in retard. lateness ‘I myself am indifferent about being late.’

(2)

Merci, je bois pas de café. thanks I.NOM drink.1SG.PRS NEG of coffee ‘Thank you, I don’t drink coffee.’

(3)

J’ en ai déjà une paire. I.NOM of.them have.1SG.PRS already INDEF.F pair ‘I already have a pair of them.’

(4)

C’ est toi qui me DEM COP.3SG.PRS you.EMPH REL.HUM.NOM me.OBJ mets en colère. put.3SG.PRS in anger ‘It is you who makes me angry.’

(5)

Moi je pense qu’ il gagne I.EMPH I.NOM think.1SG.PRS COMP he.NOM. earn.3SG.PRS au moins quatre-vingt mille francs. at least eighty thousand francs ‘I think he earns at least eighty thousand francs.’

(6)

D’après moi c’ est vingt six virgule un. according I.EMPH DEM COP.3SG.PRS twenty six comma one ‘According to me, it is twenty-six point one.’

(7)

Tu veux dire que ta soeur a you.NOM want.2SG.PRS say.INF COMP your.F.SG sister has.SG fait tout le travail. do.PART.PST all DEF work ‘You mean your sister did all the work.’

338

Alan Cruttenden

(8)

Oui, j’ en suis très content et mon yes I.NOM it.GEN COP.1SG.PRS very happy and my.M.SG. fils en est très content. son it.GEN COP.3SG.PRS very happy ‘Yes, I am very happy and my son is very happy.’

(9)

Paris St. Germain un, St. Etienne un. Paris St Germain one, St Etienne one ‘Paris St. Germain 1, St. Etienne 1.’

(10) a.

Il faut garder un equilibre entre it be.necessary.3SG.PRS keep.INF INDEF equilibrium between les affaires nationales et les affaires internationales. DEF affair.PL national.PL and DEF affair.PL international.PL ‘It is necessary to maintain a balance between national and international affairs.’ Non, cette réponse est incorrecte. No DEM.F.SG reply COP.3SG.PRS incorrect.F.SG ‘No, this reply is incorrect.’ Moi, je les trouve très impolis. I.EMPH I.NOM them.OBJ find.1SG.PRS very impolite ‘I find them very impolite.’

b.

c.

Results from fourteen French informants (1)

fiche (11), retard (3)

(2)

café (6), pas (4), bois (4)

(3)

paire (13), déjà (1)

(4)

toi (8), mets (3), colère (3)

(5)

francs (11), vingt (2), quatre (1)

(6)

six (11), un (3)

(7)

soeur (12), travail (2)

(8)

content(e) (14)

(9)

un (14)

(10) a. b. c.

-nales (11), inter- (3) -recte (10), in- (4) -polis (11), très (3)

The de-accenting of given information: A cognitive universal?

339

D German (1)

Es ist mir egal, wenn wir zu spät it COP.3SG.PRS I.DAT equal COND we too late kommen. come.1PL.PRS ‘It’s all the same to me if we arrive too late.’

(2)

Ich hasse Hühnchen. I hate.1SG.PRS chicken ‘I hate chicken.’

(3)

Ich habe schon ein Paar schwarze Schuhe. I have.1SG.PRS already INDEF pair black shoe.PL ‘I already have a pair of black shoes.’

(4)

Weil Johannes dauernd wütend wird. because John constantly angry become.3SG.PRS ‘Because John is constantly angry.’

(5)

Ich glaube, er verdient an die I think.1SG.PRS he.NOM earn.3SG.PRS to the hundertzwanzigtausend. hundred.twenty.thousand ‘I think he earns a hundred and twenty thousand.’

(6)

Bei mir sechsunddreissig Komma eins. by I.DAT six.and.thirty comma one ‘I think it is thirty-six point one.’

(7)

deine Schwester hat die ganze Arbeit Du meinst you mean.2SG.PRS your sister have.3SG DEF whole work gemacht. make.PART.PST ‘You mean your sister has done all the work.’

(8)

Ja, ich bin glücklich, und meine Frau yes I.NOM COP.1SG.PRS lucky and my wife ist sehr glücklich. COP.3SG.PRS very lucky ‘Yes I am lucky and my wife is very lucky.’

(9)

Inter-Mailand–Rom eins zu eins. Inter-Milan–Rome one to one (not relevant)

340

Alan Cruttenden

(10) a. b.

c.

(not relevant) Du meinst, diese Antwort ist unrichtig. you.NOM mean.2SG.PRS this answer COP.3SG.PRS incorrect ‘You mean this answer is incorrect.’ Ich finde sie überhaupt nicht nett. I.NOM find.1SG.PRS them especially NEG friendly ‘I find them particularly unfriendly.’

Results from four German informants (1)

egal (4)

(2)

hasse (3), Hühnchen (1)

(3)

habe (4)

(4)

Johannes (4)

(5)

hundertzwanzigtausend (4)

(6)

sechsunddreissig (4)

(7)

Schwester (4)

(8)

sehr (4)

(9)

(not relevant)

(10) a. b. c.

(not relevant) unrichtig (4) überhaupt (4)

E Greek (1)

Dhen me niázi an arghísume. NEG me.ACC care.3SG.PRS if be late.1PL.PRS ‘I do not care if we are late.’

(2)

Dhen mu arési to kotópulo. NEG me.GEN like.3SG.PRS DEF.SG.NT.ACC chicken.SG.NT.ACC ‘I don’t like chicken.’

(3)

Éxo ídi éna zevghári have.1SG.PRS already INDEF.SG.NT.ACC pair.SG.NT.ACC mávra papútsia. black.PL.NT.ACC shoes.PL.NT.ACC ‘I already have a pair of black shoes.’

The de-accenting of given information: A cognitive universal?

341

(4)

Epidhí o ghiánnis thimóni sinéxia. because DEF.SG.M.NOM John get angry.1SG.M continuously ‘Because John is always getting angry.’

(5)

Nomízo óti xérdhizi saránta xiliádhes. think.1SG.PRS COMP earn.3SG.PRS forty thousand ‘I think she earns forty thousand.’

(6)

Eghó tóto apantó triánta-éksi kómma éna. I.NOM then answer.1SG.PRS thirty-six comma one ‘But I make the answer thirty-six point one.’

(7)

Thélis na pis i want.2SG.PRS COMP say.2SG.AOR.SUBJ DEF.SG.F.NOM adhelfí su ékane óli sister.SG.F.NOM you.GEN do.3SG.IMPF all.SG.F.ACC ti dhuliá. DEF.SG.F.ACC work.SG.F.ACC ‘You mean your sister did all the work.’

(8)

Ne, mu arési polí ke arési ke yes me.GEN like.3SG.PRS much and like.3SG.PRS and sti ghinéka mu tu ídhio. to.DEF.SG.F.ACC wife.SG.F.ACC me.GEN DEF.SG.GEN same ‘Yes I like it very much and my wife likes too.’ (not relevant)

(9)

Olimpiakós-Panathanaíkos 1–1. (not relevant)

(10) c.

Nómizo, óti íne áfilokseni. think.1SG.PRS COMP COP.3PL.PRS unfriendly.PL.M.NOM ‘I think they are unfriendly.’

Results from four Greek informants (1)

niazi (4)

(2)

aresi (3), kotopulo (1)

(3)

idi (3), zevghari (1)

(4)

ghiannis (3), sinexia (1)

(5)

saranta (1), xiliadhes (3)

(6)

eksi (3), ena (1)

342

Alan Cruttenden

(7)

adhelfi (4)

(8)

ghineka (4)

(9)

(not relevant)

(10) c.

afilokseni (4)

F Italian (1)

Me ne frego del ritardo. me.ACC it.GEN care.1SG.PRS of.DEF delay ‘I don’t care if we are late.’

(2)

Non mi piace il pollo. NEG me.ACC please.3SG.PRS the chicken.NOM ‘I don’t like chicken.’

(3)

Ce le ho già un paio di it.LOC them.3PL.ACC have.1SG.PRS already a pair of scarpe nere. shoe.F.PL black.F.PL ‘I already have a pair of black shoes.’

(4)

Perché tu mi fai arrabbiare. because you.NOM me.ACC make.2SG get angry.INF ‘Because you make me angry.’

(5)

Penso che ne guadagni sugli think.1SG.PRS COMP it.GEN earn.3SG.PRS.SUBJ on the.M.PL ottanta. eighty ‘I think that he earns about eighty.’

(6)

A me invece ventisei virgola uno. to me.ACC instead twenty.six comma one ‘But I make it twenty-six point one.’

(7)

Vuoi dire che tua sorella want.2SG.PRS say.INF COMP your.F.SG sister.F.SG ha fatto tutto. have.3SG.PRS do.M.SG.PST.PART everything.SG.M ‘You mean your sister has done everything.’

The de-accenting of given information: A cognitive universal?

343

(8)

Si, mi piace, a mia moglie piace yes me.DAT please.3SG.PRS to my.F.SG wife please.3SG.PRS molto. a lot ‘Yes, I like it and my wife likes it a lot.’

(9)

Inter uno, Roma uno. Inter one, Roma one ‘Inter 1, Roma 1’

(10) a.

Erano bloccati per sciopero i COP.3PL.IMPF. block.PST.PART.M.PL for strike.M.SG the.M.PL voli nazionali e internazionali. flight.M.PL national.ADJ.M.PL and international.ADJ.M.PL ‘National and international flights are blocked by the strike.’ A me invece sembrano molto dissimili. to me.ACC instead seem.3PL.PRS very dissimilar.M.PL ‘But they seem very dissimilar to me.’ Io trovo che sia molto I.NOM find.1SG.PRS COMP COP.3SG.PRS.SUBJ very antipatica. unfriendly.F.SG ‘I find her very unfriendly.’

b.

c.

Results from eleven Italian informants (1)

frego (11)

(2)

piace (11)

(3)

già (11)

(4)

arrabiare (5), tu (3), fai (3)

(5)

(not relevant)

(6)

ventisei (8), uno (3)

(7)

sorella (10), tutto (1)

(8)

(not relevant)

(9)

uno (second occurrence)(10), Roma (1)

(10) a. b. c.

-nal- (10), int- (1) -sim- (8), dis- (3) -pat- (8), an- (3)

344

Alan Cruttenden

G Lithuanian (1)

Man nesvarbu ar mes pav˙esluosim. I.DAT NEG.important if we PREF.late.1PL.FUT ‘I don’t care if we are late.’

(2)

Aš nem˙egstu višˇciuko. I.NOM NEG.like.1SG.PRS chicken.GEN ‘I don’t like chicken.’

(3)

batu˛ Aš jau turiu juodu˛ I.NOM already have.1SG.PRS black.PL.GEN shoe.PL.GEN pora˛ . pair.SG.ACC ‘I have already got a pair of black shoes.’

(4)

Tod˙el, kad Jonas pyksta. because COMP John angry.3SG.PRS ‘Because John is angry.’

(5)

Man atrodo jis gauna beveik dvidešimts me.DAT seem.3SG.PRS he.NOM get.3SG.PRS nearly twenty t¯ukstanˇciu˛ talonu˛. thousand.GEN coupon.PL ‘It seems he earns nearly twenty thousand coupons.’

(6)

O aš gavau trisdešimts šeši kablelis vienas. and I.NOM get.1SG.PST thirty six point one ‘Well, I make it thirty-six point one.’

(7)

Tu nori pasakyti, kad tavo sesuo you.NOM want.2SG.PRS pref.say.INF COMP your sister darba˛ . atliko visa˛ do.3SG.PST all.ACC work.ACC ‘You mean your sister did all the work.’

(8)

Taip, aš laimingas ir mano žmona labai laiminga. yes I.NOM happy.M.SG and my wife very happy.F.SG ‘Yes, I am happy and my wife is very happy.’

(9)

(not relevant)

(10) a.

(not relevant)

The de-accenting of given information: A cognitive universal?

b.

c.

Tu manai, tas you.NOM think.2SG.PRS COMP neteisingas. NEG.correct ‘You find that reply is incorrect.’ Aš nau, jie I.NOM think.1SG.PRS they.NOM nedraugiški. NEG.friendly.PL.M ‘I think they are very unfriendly.’

atsakymas yra reply COP.3SG.PRS

yra labai COP.3PL.PRS very

Results from four Lithuanian informants (1)

nesvarbu (4)

(2)

nem˙egstu (4)

(3)

turiu (4)

(4)

Jonas (4)

(5)

dvidešimts (4)

(6)

trisdešimts (3), aš (1)

(7)

sesuo (3), darba˛ (1)

(8)

žmona (4)

(9)

(not relevant)

(10) a. b. c.

(not relevant) neteisingas (4) labai (1), nedraugiški (1), nedraugiški (2)

H Macedonian (1)

Ne mi e gajle ako zadocnime. NEG I.DAT COP.3SG care if be late.1PL ‘I don’t care if we are late.’

(2)

Ne jadam pileško. NEG eat.1SG chicken ‘I don’t eat chicken.’

345

346

Alan Cruttenden

(3)

Vek’e imam eden par crni konduri. already have.1SG one pair black.PL shoe.PL ‘I’ve already got a pair of black shoes.’

(4)

Zašto Džon me luti. because John me.ACC anger.3SG.PRS ‘Because John make me angry.’

(5)

Mislam deka zarabotuva skoro cˇ etirieset iljadi think.1SG.PRS COMP earn.3SG nearly forty thousand denari. denari ‘I think he earns nearly forty thousand dinari.’

(6)

Togaš jas k’e go odgovoram trieset i šest toˇcka then I.NOM FUT it.OBJ answer.1SG thirty and six point eden. one ‘I make it thirty-six point one.’

(7)

Misliš sestra ti ja završi celata rabota. think.2SG sister you.DAT it.3SG.F do.3SG.PST all work ‘You mean your sister did all the work.’

(8)

Da zadovolen sum i žena mi e mnogu yes happy COP.1SG. and wife I.DAT COP.3SG.PRS very zadovolna. happy.F.SG ‘Yes, I’m happy and my wife is very happy.’

(9)

Vardar–Teteks eden eden. (not relevant)

(10) a.

Mislam deka e važno da se think.1SG COMP COP.3SG important COMP RFL održuva ramnoteža meg’u nacionalnite i maintain.3SG.PRS balance.NOM between national.PL.DEF and internacionalnite odvivanja. international.PL.DEF affair.PL ‘I think it is important to maintain a balance between national and international affairs.’ Misliš toj odgovor e netoˇcen. think.1SG DEM reply COP.3PL incorrect ‘I think that reply is incorrect.’

b.

The de-accenting of given information: A cognitive universal?

c.

347

Jas mislam deka se mnogu neljubezni. I think.1SG COMP COP.3PL very NEG.friendly.PL ‘I think they are very unfriendly.’

Results from two Macedonian informants (1)

gajle (1), zadocnime (1)

(2)

ne (1), pileško (1)

(3)

konduri (2)

(4)

luti (2)

(5)

cˇ etirieset (2)

(6)

eden (2)

(7)

sestra (2)

(8)

žena (2)

(9)

(not relevant)

(10) a. b. c.

internacionalnite (2) netoˇcen (2) mnogu (1), neljubezni (1)

I

Russian

(1)

A mne vse ravno, opozdaem but me.DAT everything.NOM same.NOM.N.SG late.1PL.PFV.FUT li my. if we.NOM ‘I don’t care whether we’re late or not.’

(2)

Ja ne ljublju kuricu. I.NOM NEG like.1SG.PRS chicken.SG.ACC ‘I don’t like chicken.’

(3)

cˇ ernye tufli. U menja uže est’ at me.GEN already COP.PRS black.PL.NOM shoes.PL.NOM ‘I’ve already got black shoes.’

(4)

ty zastavljaeš’ menja serdit’sja. Potomu cˇ to because COMP you.NOM force.2SG.PRS me.ACC angry.INF ‘Because you force me to be angry.’

348

Alan Cruttenden

(5)

Ja dumaju, on zarabatyvaet poˇcti sorok I think.1SG.PRS he.NOM earn.3SG.PRS almost forty.ACC tysjaˇc. thousand.GEN.PL ‘I think he earns almost forty thousand.’

(6)

A u menja poluˇcaetsja dvadcat’ šest’ i but at me.GEN come.out.3SG.PRS twenty six.NOM and odin. one.NOM ‘I get twenty-six point one.’

(7)

Ty imeeš’ v vidu, cˇ to tvoja you.NOM have.2SG.PRS in view COMP your.F.SG.NOM sestra vse sdelala. sister.SG.NOM everything.ACC do.F.SG.PFV.PST ‘You mean that your sister has done everything.’

(8)

Da ja dovolen i moja žena tože. yes I.NOM satisfied.M.SG and my.F.SG.NOM wife.SG.NOM too ‘Yes, I am happy and my wife is too.’ (not relevant)

(9)

Spartak odin, Dinamo odin. Spartak one Dinamo one ‘Spartak 1, Dinamo 1’

(10) a.

Nužno cˇ to byl balans meždu necessary COMP COP.M.SG.PF balance.SG.NOM between vnutrennimi i meždunarodnymi otnošenijami. internal.INST.PL and international.INST.PL affairs.INST.PL ‘There must be a balance between internal and international affairs.’ cˇ to oni oba A ja dumaju and I.NOM think.1SG.PRS COMP they.PL.NOM both.NOM.M nepravil’ny. NEG.correct.PL ‘I think they are both incorrect.’ cˇ to oni neprivetlivy. A mne kažetsja, but me.DAT seem.3SG.PRS COMP they.PL.NOM NEG.friendly.PL ‘They seem unfriendly to me.’

b.

c.

Results from two Russian informants (1)

opazdaem (1), ravno (3)

The de-accenting of given information: A cognitive universal?

349

(2)

ljublju (3), ja (1)

(3)

est’(4)

(4)

zastavljaeš’ (3), ty (1)

(5)

sorok (3), tysjaˇc (1)

(6)

šest’ (4)

(7)

sestra (4)

(8)

tože (4) (not relevant)

(9)

odin (1), Dinamo (1) (two not relevant, given as Spartak–Dinamo 1–1)

(10) a. b. c.

meždunarodnymi (2), otnošenijami (2) (not relevant) nepravil’ny (3), nepravil’ny (1) neprivetlivy (3), neprivetlivy (1)

J

Spanish

(1)

No me importa si llegamos tarde. NEG me.DAT matter.3SG if arrive.1PL late ‘It doesn’t matter to me if we are late.’

(2)

No me gusta el pollo. NEG me.DAT please.3SG DEF.M chicken ‘I don’t like chicken.’

(3)

Yo ya tengo un par de zapatos negros. I.NOM already have.1SG.PRS INDEF.M pair of shoes black.PL ‘I already have a pair of black shoes.’

(4)

Porque Juan hace que me enoje. because John make.3SG.PRS COMP me.ACC annoy.3SG.PRS ‘Because John makes me annoyed.’

(5)

Yo creo que gana por lo menos I.NOM believe.1SG.PRS COMP earn.3SG.PRS at the least cuarenta mil. forty thousand ‘I believe that he earns at least forty thousand.’

(6)

Y a mí treinta y seis coma uno. and to me thirty and six comma one ‘And I make it thirty-six point one.’

350

Alan Cruttenden

(7)

Lo que quieres decir es que tu it.ACC COMP want.2SG.PRS say.INF COP.3SG COMP your.SG hermana hizo todo el trabajo. sister do.3SG.PST all DEF.M work ‘You mean your sister did all the work.’

(8)

Sí, estoy feliz, y mi esposo(a) también está feliz. yes COP.1SG happy and my wife also COP.3SG.F happy ‘Yes, I am happy and my wife is also happy.’

(9)

Real uno, Atlético uno. Real one Atletico one

(10) a.

Yo creo que es importante I.NOM believe.1SG.PRS COMP COP.3SG.PRS important mantener el equilibrio entre los acontecimientos maintain.INF DEF.M balance between DEF.PL event.PL nacionales e internacionales. national.PL and international.PL ‘I believe that it is important to maintain a balance between national and international affairs.’

Results from four Spanish informants (1)

importa (1), tarde (3)

(2)

gusta (3), pollo (1)

(3)

negros (4)

(4)

enoje (4)

(5)

mil (4)

(6)

uno (4)

(7)

hermana (1), trabajo (3)

(8)

también (2), feliz (2)

(9)

uno (4)

(10) a.

internationales (1), internacionales (3)

The de-accenting of given information: A cognitive universal?

351

K Swedish (1)

Jag struntar i om vi kommer för sent. I.NOM NEG.bother.PRS in if we.NOM come.PRS too late ‘I don’t care if we are late.’

(2)

Jag tycker inte om kyckling. I.NOM like.PRS NEG about chicken ‘I don’t like chicken.’

(3)

Jag har redan ett par svarta skor. I.NOM have.PRS already a pair black shoe.PL ‘I’ve already got a pair of black shoes.’

(4)

För att du gör mig arg. for that you.SG.NOM make.PRS I.OBJ angry ‘Because you make me angry.’

(5)

Jag tror att han tjänar nästan I.NOM believe.PRS COMP he earn.PRS almost hundrafemtiotusen. one.hundred.and.fifty.thousand ‘I think he earns nearly one hundred and fifty thousand.’

(6)

Jag får det till tjugosex komma ett. I.NOM get.PRS it to twenty-six point one ‘Well I make it twenty-six point one.’

(7)

Du menar att din syster gjorde you.SG.NOM mean.PRS COMP you.SG.POSS sister do.PST jobbet. job.DEF ‘You mean your sister did the job.’

(8)

Ja, jag är nöjd, och min fru är mycket yes I.NOM COP.PRS pleased and I.POSS wife COP.PRS very nöjd. pleased ‘Yes, I’m happy and my wife is very happy.’

(9)

(not relevant)

352

Alan Cruttenden

(10) a.

b.

c.

Det är viktigt att hålla balans mellan nationella it COP.PRS important to keep.INF balance between national och internationella angelägenheter. and international concerns ‘It is important to keep a balance between national and international concerns.’ Du menar att han uppträder inkorrekt. you.SG.NOM mean.PRS COMP he behave.PRS incorrectly ‘You mean that he behaves incorrectly.’ Jag tycker att de är mycket ovänliga. I.NOM think.PRS COMP they COP.PRS very unfriendly ‘I think that they are very unfriendly.’

Results from four Swedish informants (1)

struntar (1), sent (3)

(2)

kyckling (4)

(3)

har (1), skor (3)

(4)

gör (4)

(5)

hundrafemtiotusen (1), hundrafemtiotusen (3)

(6)

seks (2), ett (2)

(7)

syster (2), jobbet (2)

(8)

mycket (1), nöjd (3)

(9)

(not relevant)

(10) a. b. c.

(not relevant) inkorrekt (2), inkorrekt (2) ovänliga (3), ovänliga (1)

Notes . 1.

I wish to thank all the informants who gave generously of their time, and in particular Kathleen Newton for her editorial help. Bold type indicates accented syllables and underlining indicates re-accented words.

The de-accenting of given information: A cognitive universal?

353

References Agard, Frederick B., and Robert J. Di Pietro 1965 The Sounds of English and Italian. Chicago: University of Chicago Press. Bansal, R. K. 1990 The pronunciation of English in India. In Studies in the Pronunciation of English, Susan Ramsaran (ed.), 219–230. London: Routledge. Bell, David Charles, and Alexander Melville Bell 1879 Bell’s Standard Elocutionist. London: Hodder & Stoughton. Bernini, Giuliano 1992 Establishing sentence themes in Italian written discourse. In EUROTYP Working Papers I/2, Theme Group 1: Pragmatic Organization of Discourse, Giuliano Bernini and Davide Ricca (eds.), 21–56. Programme in Language Typology: European Science Foundation. Birner, Betty J. 1994 Information status and word order: an analysis of English inversion. Language 70: 233– 259. Brazil, David 1985 The Communicative Value of Intonation in English. Cambridge: Cambridge University Press. Britain, David, and John J. Newman 1992 High rising terminals in New Zealand English. Journal of the International Phonetic Association 22: 1–11. Brown, Gillian 1983 Intonation, the categories given/new, and other sorts of knowledge. In Prosody: Models and Measurements, Anne Cutler and D. Robert Ladd (eds), 67–78. Berlin: Springer. Brown, Gillian, Karen L. Currie, and Joanne Kenworthy 1980 Questions of Intonation. London: Croom Helm. Chafe, Wallace L. 1974 Language and consciousness. Language 50: 111–133. 1976 Givenness, contrastiveness, definiteness, subjects, topics, and point of view. In Subject and Topic, Charles N. Li (ed.), 25–55. New York: Academic Press. 1994 Discourse, Consciousness, and Time: The Flow and Displacement of Conscious Experience in Speaking and Writing. Chicago: University of Chicago Press. Clark, Herbert H., and Susan E. Haviland 1977 Comprehension and the given-new contract. In Discourse Production and Comprehension, Roy O. Freedle (ed.), 1–40. Norwood, NJ: Ablex Publishing. Comrie, Bernard 1981 The Languages of the Soviet Union. Cambridge: Cambridge University Press. Couper-Kuhlen, Elizabeth 1983 Intonatorische Kohäsion: Eine macroprosodisches Untersuchung. Zeitschrift für Literaturwissenschaft und Linguistik 49: 79–100. Cruttenden, Alan 1974 An experiment involving comprehension of intonation in children from 7 to 10. Journal of Child Language 1: 221–232. 1981 Falls and rises: meanings and universals. Journal of Linguistics 17: 77–91. 1986 Intonation. Cambridge: Cambridge University Press. 1990 Nucleus placement and three classes of exception. In Studies in the Pronunciation of English, Susan Ramsaran (ed.), 9–18. London: Routledge. 1995 Rises in English. In Studies in General and English Phonetics: Essays in Honour of Professor J. D. O’Connor, Jack Windsor Lewis (ed.), 155–173. London: Routledge.

354

Alan Cruttenden

Cruttenden, Alan, and David Faber 1991 The accentuation of prepositions. Journal of Pragmatics 15: 165–186. Crystal, David 1975 Prosodic features and linguistic theory. In The English Tone of Voice, D. Crystal (ed.), 1–46. London: Arnold. Egbe, D. I. 1979 Intonation and meaning in Nigerian English: a case study. Paper presented at the Annual General Conference of the Modern Languages Association of Nigeria, University of Ife, Ife-Ife, Nigeria. Friedman, Victor. A. 1993 Macedonian. In The Slavonic Languages, Bernard Comrie and Greville G. Corbett (eds.), 249–305. London: Routledge. Gårding, Eva 1981 Contrastive prosody: a model and its applications. Studia Linguistica 35: 146–165. Geluykens, Ronald 1993 Topic introduction in English conversation. Transactions of the Philological Society 91: 181–214. Gumperz, John. J. 1982 Discourse Strategies. Cambridge: Cambridge University Press. Guy, Gregory, Barbara Horvath, Julia Vonwiller, Elaine Disley, and Inge Rogers 1986 An intonational change in progress in Australian English. Language in Society 15: 23– 52. Guy, Gregory, and Julia Vonwiller 1984 The meaning of an intonation in Australian English. Australian Journal of Linguistics 4: 1–17. Halliday, Michael A. K. 1967 Intonation and Grammar in British English. The Hague: Mouton. Hirschberg, Julia 1994 Studies of intonation and discourse. Paper presented to the ESCA workshop on prosody, Lund University, Sweden. Hirst, Daniel, and Albert Di Cristo 1998 Intonation Systems. A Survey of Twenty Languages. Cambridge: Cambridge University Press. Joseph, Brian, and Irene Philippaki-Warburton 1987 Modern Greek. London: Routledge. Ladd, D. Robert 1990 Intonation vs. grammar [review of Dwight L. Bolinger (1989), Intonation and Its Uses]. Language 66: 806–816. Lehiste, Ilse 1975 The phonetic structure of paragraphs. In Structure and Process in Speech Perception, Antonie Cohen and Sidney Nooteboom (eds.), 195–203. Berlin: Springer. 1979 Perception of sentence and paragraph boundaries. In Frontiers of Speech Perception, Björn Lindblom and S. Öhman (eds.), 191–201. London: Academic Press. 1982 Some phonetic characteristics of discourse. Studia Linguistica 36: 117–130. Lehman, Christina 1977 A re-analysis of given-ness: stress in discourse. Proceedings of the Chicago Linguistic Society 13: 316–324. Low, Ee-Ling 1994 Intonation patterns in Singapore English. Unpublished MA thesis, University of Cambridge.

The de-accenting of given information: A cognitive universal?

355

MacWhinney, Brian, and Elizabeth Bates 1978 Sentential devices for conveying givenness and newness: a cross-cultural developmental study. Journal of Verbal Learning and Verbal Behavior 17: 539–558. Maidment, John A. 1990 Focus and tone in English intonation. In Studies in the Pronunciation of English, Susan Ramsaran (ed.), 19–26. London: Routledge. Nooteboom, Sieb G., and J. G. Kruyt 1987 Accents, focus distribution, and the perceived distribution of given and new information: an experiment. Journal of the Acoustical Society of America 82 (5): 1512–1524. Ortiz-Lira, Hector 1994 A contrastive analysis of English and Spanish sentence accentuation. Unpublished doctoral dissertation, University of Manchester. 1995 Nucleus placement in English and Spanish: a pilot study of patterns of interference. In Studies in English and General Phonetics: Essays in Honour of Professor J. D. O’Connor, Jack Windsor Lewis (ed.), 255–265. London: Routledge. Prince, Ellen. F. 1981 Towards a taxonomy of given/new information. In Radical Pragmatics, Peter Cole (ed.), 223–256. New York: Academic Press. 1982 The ZPG letter: subjects, definiteness, and information-status. In Discourse Description: Diverse Linguistic Analysis of a Fund-raising Text, William C. Mann and Sandra A. Thompson (eds.), 295–325. Philadelphia: John Benjamins. Schubiger, Maria 1958 English Intonation: Its Form and Function. Tübingen: Niemeyer. Sinclair, John McH., and David Brazil 1982 Teacher Talk. Oxford: Oxford University Press. Stockwell, Robert P., and J. Donald Bowen 1965 The Sounds of English and Spanish. Chicago: University of Chicago Press. Swerts, Marc, and René Collier 1992 On the controlled elicitation of spontaneous speech. Speech Communication 11: 463– 468. Taglicht, Josef 1982 Intonation and the assessment of information. Journal of Linguistics 18: 213–230. Vallduví, Enric 1993 Information packaging: a survey. Research Paper HCRC/RP-44, University of Edinburgh, Human Communication Research Centre. Vanderslice, Ralph, and Laura S. Pierson 1967 Prosodic features of Hawaiian English. Quarterly Journal of Speech 53: 156–166. Walker, John 1781 Elements of Elocution. 2 vols. London: printed for the author. [Reprinted by the Scolar Press, Menston, Yorkshire, England, English Linguistics 1500–1800. No. 266]. Wells, William H. G. 1992 Phonetic aspects of focus in London Jamaican. Paper presented at the biennial meeting of the British Association of Academic Phoneticians, University of Cambridge. Yule, George 1980 Speaker topics and major paratones. Lingua 52: 38–47.

Rosanna Sornicola

Interaction of syntactic and pragmatic factors on basic word order in the languages of Europe

1. Theoretical and methodological foundations1 1.1. A pragmatic study of word order The study which is the subject of the present work is an attempt to arrive at an integrated view of the interplay of the various levels of analysis – especially the syntactic and pragmatic ones – on the typology of basic constituent order.2 Adopting such a perspective represents a move away from traditional typological research. A preliminary explanation is required so that, on the one hand, the concepts and terminology of a “pragmatic study” of word order (henceforth, WO) may be clarified and, on the other, that attention may be drawn to the consequences they have for the typological approach followed here. In a more restricted and technical sense, the study offers a pragmatic perspective because the effects of pragmatic functions on the “dynamics” of rules of order is considered crucial (cf. Section 1.8). But the term “pragmatic” is also used here with a less technical and wider meaning, one which is linked to the first but which goes beyond it in many respects. In this sense the term “pragmatic” would be better understood as equivalent to “functional.” Three concepts may be defined that characterize the wider meaning of the term “pragmatic” used here. They are in turn interrelated: the multifactoriality, the function, and the actual conditions of WO. The present study has a multifactorial viewpoint. Constituents are seen as carriers of grammatical functions (GF), semantic functions (SF), and pragmatic functions (PF). This representation goes back to an integrated conception of grammatical structures, the properties of which are considered with reference to the interplay of the (syntactic), semantic, and pragmatic grammatical levels. Such a conception is all the more indispensable in the study of order structures, in that these may be subject to alteration of a greater or lesser degree,3 according to variation in the functional values associated with the representation of the constituents. For example, an NP with the GF S may occur in very different structural positions according to whether it has the PF of TOPIC or FOCUS. On the other hand, an NP with the PF of FOCUS may have

358

Rosanna Sornicola

a different distribution according to whether it is syntactically encoded as S or as O and/or whether it has the SF of agent (= A) or of patient (= P). The options relating to a particular distribution are, of course, dependent on the typological characteristics of the various languages. The second key concept is that of the function of WO within individual languages and within types of languages. This regards the need to study the interaction of syntactic, semantic, and pragmatic properties and of other factors, rhythmic and prosodic for example, in determining neutral and non-neutral orders. The attempt to analyze the function of WO may therefore be understood as an attempt to ascertain not only what may be defined as the “static” dimension of the WO, in other words, neutral orders, but also its dynamics, that is, the non-neutral or marked orders.4 As will be seen, the latter have essentially to do with a pragmatic conception. In a certain sense, it may be said that while neutral orders are considered “static,” marked orders are “dynamic.” Corresponding to this intuition is the fact that, in some approaches, nonneutral or marked WO phenomena have been represented as the effect of syntactic “processes” on basic structures.5 In its multifactorial consideration of constituent order and constituent function, the present study is linked to the European functionalist tradition6 as well as to more recent issues concerning WO “flexibility.”7 The third concept which defines our approach concerns the central importance given to the actual conditions under which WO phenomena arise. In concrete terms, it has involved constant attention to contextual factors, the type of text, variation in the written/spoken register, and, wherever possible, what is traditionally described as “stylistics,” in other words, idiosyncratic variation in individual writers/speakers. Attention to actual conditions has meant attention also to the diachrony of WO mechanisms: from time to time, where documentation has been available, it has seemed appropriate to try and take into account the permanence or discontinuity of a WO property. In fact, variability or continuity of WO patterns over time seems to constitute an important typological parameter.

1.2. Word order between competence and performance Some of the factors just mentioned concern pragmatic competence, others are connected to discourse planning or perceptual strategies more closely related to performance. In fact, a good part of what has been called the “dynamics” of WO phenomena, for example the characteristic pragmatic processes of emphasizing topicalization or focalization by means of WO, seems to concern the vast intermediate area between competence and performance. On the methodological level, it has meant the use of data elicited by means of grammaticality judgements and, more generally, idealization techniques, as well as data gathered from actual texts.

Interaction of syntactic and pragmatic factors

359

Although attention has been paid to the intermediate area between competence and performance and to structures actually produced, this has not, however, meant that cognitive or perceptual factors have been given pre-eminence in the study of WO. While this seems to be a legitimate area of research (psycholinguistic factors are indeed of paramount importance in empirical issues concerning WO),8 it was decided that preference should be given to linguistic examination of the strategies which determine WO structures in the various languages. Of course, concepts such as TOPIC and FOCUS have a cognitive content which is not easily modeled. This is perhaps one of the reasons which have made it difficult to arrive at unanimously accepted definitions of PFs (cf. Section 1.8 and thereafter). Nonetheless, a linguistic modeling may and should be sought. With view to a typological comparison in particular, it seemed necessary in fact not to overestimate so-called “natural” or “iconic” factors. Such overestimation seems to be the modern form of an old idea concerning the existence of a relationship between word order and the order of thoughts. But the choice of a truly linguistic modeling arises from another necessity. The exploration of psycholinguistic factors takes us closer to the mechanisms of performance and these, in turn, characterize “stylistic” (individual) variation rather than the typological properties of the language.9 It is true, in any case, that the problem of linearization of constituents, of “how to get words into line”10 remains a crucial one for WO. The question regarding the relationship between linearization in performance, always and inevitably text-oriented and individual, and linearization as a characteristic of a language, or more especially of a linguistic type, appears here in all its complexity. It is an aspect of a broader problem concerning the relationship between performance and competence, rendered more difficult by the inherent variability of WO patterns. This variability, which, as will be seen, also characterizes languages with so-called “rigid” constituent order, often makes definition of a “type” problematic. The concept of variability may be understood in differing ways. It may be understood as sensitivity to the combination of interacting factors, or as the co-existence of sometimes very different patterns in the language, which may be explained only by the contribution made by historical and sociolinguistic conditions. In considering the structural properties that determine a type, one must not forget to take into account the extent to which it has been formed by “educated” normativization, from above, how far it relates to spontaneous developments in the spoken language (see Sornicola 1981; Miller and Weinert 1998)), and also the extent to which it is the continuation or persistence of old inherited patterns and how much it is due to the external influence of prestigious languages, such as Latin or Greek.11 This is all the more important in an investigation into the typology of the languages of Europe, which over the span of several millennia have been exposed to very complex and multi-layered cultural influences. But it would be wrong to confine the problem of interference to the languages of Europe that have been exposed to the Latin and/or

360

Rosanna Sornicola

Greek superstratum. It also arises, in a different but no less important way, in the languages of Inner Asia, for which, in absence of written documents prior to the eighth century AD, it is impossible to carry out a reconstruction of their linguistic history. As Denis Sinor (1990) observes: The nature of the relationship between the Uralic (Finno-Ugric and Samoyed) and Altaic (Turkic, Mongol and Tunguz) languages which, as far back as the available data allow us to go, have constituted the dominant linguistic group in the forest and tundra zones of Inner Asia, cannot be established with any degree of certainty. While there are those scholars who aver that some or even all of them are genetically related – that is, that they descend from a common, ancestral Ursprache – others, including myself, believe that the elements which they unquestionably have in common are due to constant interaction over the centuries if not millennia, and that they result from convergent rather than divergent development. (Sinor 1990: 16) With respect to the problem of WO, in addition to the above-mentioned historical circumstances, we should also note the observations of Thomason and Kaufman (1988), according to whom: The evidence we have collected does not support the often implicit assumption, in the literature on WO change, that WO patterns constitute a fundamental “deep” structural feature relatively impervious to foreign influence. On the contrary, WO seems to be the easiest sort of syntactic feature to borrow or to acquire via language shift. (Thomason and Kaufman 1988: 55) For all these reasons, examination of WO in a “pragmatic” perspective should not be confined only to perceptual or cognitive (performance) factors, but should also take into account the interweaving of the structural, stylistic, sociolinguistic, and historical factors which have contributed to establishing certain patterns. It is the very interweaving of these factors that allows for the passing from the level of performance, which is stylistic and individual, to the level of the characteristic structures of the language and more especially of the “type.” This is to be understood as a general principle superordinate to processes whose crystallization is visible in the structures of the language. For this reason, it seems to us that we need to reassess the view, common to certain traditions of typological study, that the preliminary condition for research is the security that “we are dealing with a typological pattern that is a result of general principles, not of historical accidents (common descent or contact)” (Croft 1990: 22).

Interaction of syntactic and pragmatic factors

361

1.3. Microscopy and macroscopy But to what extent can the typologist take into account the complexity of this interweaving? The comparative dimension in fact makes it more difficult to check the multiplicity of the factors mentioned in every language. In short, in our attempt to establish a pragmatic typology of WO, we intend to explore patterns of WO in individual languages on the basis of the particular conditions which determine them and, by this means, to arrive at a typological comparison, inevitably problematic and provisional. As will be seen, it is possible to establish “geometric” rules relative to WO. But such rules are the linguist’s precarious constructs. They need to be continuously reexamined with respect to historical, sociolinguistic, and stylistic variability. In confronting these questions, it was necessary to make a distinction between two different but complementary points of view. The viewpoint of classical typology is the “macroperspective.” Here the differences between languages are established “from above,” as in an aerial photograph. The pragmatic view, especially in the meaning proposed here, is the “microperspective,” which requires assessment of the particular conditions in which WO patterns are determined, as in a ground survey. In this sense it is a “fine grained” investigation, as under the microscope. In the study of WO typology, as in other areas of contemporary scientific research, it is evident that rules appearing at the macroscopic level may disappear like an optical illusion when looked at from the microscopic level, where other rules of a quite different nature may appear. What may seem a clear and well-defined structure from the macroscopic viewpoint breaks up into many different phenomena, for which the same description is uncertain. The uncertainty of “historical” descriptions12 takes over geometric clarity. On the other hand, an eagle’s perspective may reveal connections and interrelations invisible from the ground. However, the two points of view are not irreconcilable. On the contrary, the results from microscopic examination may be crucial to the vast macroscopic constructions, and, vice versa, the latter may offer a perspective, albeit provisional and precarious, on the difficult task of orienting oneself in the multiplicity of pragmatic phenomena. In the present work we have tried to put together arguments from both perspectives.

1.4. The languages of the corpus The sample assembled for the study consists of fourteen languages: Basque, Turkish, Georgian, Hungarian, Finnish, Estonian, Russian, French, Italian, Spanish, English, German, Irish, and Welsh. In choosing the languages we have tried to represent the various types of basic order found in the languages of Europe, also taking account of

362

Rosanna Sornicola

different genetic affiliations and areal distributions. Space limitations of the present work have prevented including in the corpus other languages worthy to be examined as, for example, Modern Greek. The criterion of the representation of basic order, however, has constituted only a very general point of departure for the study of the pragmatic properties of WO. Within each basic order, numerous subrealizations may in fact be identified. Furthermore, basic order, although fundamental, constitutes only one of the parameters to be taken into account when studying the pragmatic properties of WO. It seems to act in a restrictive way, defining the options which are not structurally possible, rather than in a positive way, that is allowing the specification of pragmatically significant processes (topicalization, focalization, etc.). The sample shows a bias in favor of certain genetic groupings and one areal block: Indo-European languages have the greatest weight in terms of the number of languages examined (eight out of fourteen). The part of Europe to the west of a line drawn from Finland to Hungary and as far as Turkey is well represented, the area to the east much less so. This may in part be explained by the chosen historical and cultural conception of the European space. While, geographically speaking, the extreme eastern boundary of Europe has been recognized, at least since the eighteenth century, as part of the Urals, it is nonetheless true that historically and culturally, the Mediterranean region to the south, and the region beyond the Volga plains and of the Dnepr to the north have for millennia continued to attract Euro-Asiatic populations. As far as the latter and their languages are concerned, it is extremely difficult to reconstruct their history to any great depth due to the scarcity of direct documentary sources. But, as Sinor (1990) observes, it is possible to make out historical processes over a long period that contrast the two areas in terms of sedentary vs. nomadic worlds and their differing economic and cultural standards.13 As often happens in the presence of such differences, definition of the societies which may be placed in the lower economic level is made in negative terms with respect to the hegemonic societies and by those hegemonic societies themselves.14 Above all, it hardly needs mentioning that not only is “European” not a linguistic concept, it is not even a geographical one. The geographical boundaries of Europe have themselves undergone changes over the course of time, in line with different cultural conceptions.15 One of the most controversial questions has been the demarcation of the borders between “Europe” and “Asia”.16 In Europe, as a historically defined cultural space, the Indo-European languages have prevailed over genetically different languages which have been confined to the eastern margins (or, in the case of Basque, to a residual zone in the far west). However, the choice of languages in the sample reflects, in part, the limits imposed by the methodology of gathering data (cf. Section 1.5). Within certain families, such as Romance or Germanic, more languages were considered in order to represent typological variation within a particular family.

Interaction of syntactic and pragmatic factors

363

There is good reason to suspect that some of the languages of the corpus are not independent (or that they converge due to historical as well as areal considerations). But, as mentioned in Section 1.2, in an area with a high rate of contact and exchange, as in the European space, it would be unthinkable to construct a sample of languages that did not present such a bias (cf. method for reducing the bias in Section 1.3). On the other hand, processes of convergence are highly visible not only within the Indo-European languages, but also between the Ugro-Finnic and Indo-European languages (see Comrie 1981a: 124).

1.5. The method of gathering data The approach taken in the research described in Section 1.1 has required the use of several methods for gathering data. The pragmatic description of WO structures and, in particular, the dynamic analysis of the pragmatic processes associated with them, requires in the first place direct examination of written texts or spoken language produced by native speakers, or of judgements made by native speakers on the acceptability and the context of usage of appropriately preconstructed structures presented to them, thus allowing for the collection of different types of data. Both techniques are more characteristic of “in-depth” research on individual languages than of crosslinguistic studies. As far as the former is concerned, simple elicitation of a text would not have sufficed for the specific aim of the study: certain marked WO structures occur rarely, individual variability may be high, etc. On the other hand, the type of questionnaire which needs to be set up for a study of speakers’ judgements would require a direct and thorough knowledge of all the languages in the sample. Other types of questionnaires, of an “explorative” nature (questionnaires, that is, in which the linguist manipulates “structures” which are not lexically realized), present problems for a pragmatic examination: the actual use of a structure and its association with a particular context or register lie below the threshold of consciousness, often even amongst native speakers who are linguists. For these reasons, preference was given to the use of reliable descriptions (grammars and other secondary sources), to which, in some cases, additional primary sources were added (texts produced by native speakers and/or native speakers’ judgements).17 At times, this has in some ways influenced the choice of individual languages in the sample: once the typological requirements were established and a balanced representation of the various genetic groupings guaranteed, languages were chosen for which there were comprehensive WO descriptions which accommodated the complexities of what has here been defined as “pragmatic.” Wherever possible, the properties of languages belonging to different branches of the same genetic family were considered in order to establish the diachrony of a given phenomenon.

364

Rosanna Sornicola

Of course, the priority assigned to secondary sources has brought with it problems of another sort, such as the bias of the individual monographs or treatises towards one or another linguistic theory, or at least towards a particular descriptive approach.18 Recognition of this has in fact highlighted the considerable differences in grammatical traditions on WO, which are often associated with particular language families. These differences are interesting in themselves, and they show that WO is an area of linguistic phenomena where descriptive objectivity must take into account not only the complexity of the phenomena themselves, but also the metalanguage which has developed over time to describe it. For this reason, although an attempt has been made to transcodify individual descriptions according to a single homogeneous model, some consideration has been given to the specific properties of the theoretical models describing the phenomena under examination (cf. Sections 1.6 and 1.7).

1.6. Topology and syntax: Towards a geometry of the sentence 1.6.1. Notions of order The analytical procedure followed for each language in the sample has in the first place provided a description of the dominant order and thus a description of its possible variants and factors associated with it. Such a method seemed appropriate because examination of pragmatically determined variations can be carried out more effectively on the basis of neutral structures, without contextual conditions; in other words, an examination of the statics suitably precedes an examination of the dynamics of the WO structures.19 The basic order which, as is well known, does not necessarily coincide with the dominant order,20 has been considered as an initial macrocondition, on the basis of which finer/more subtle articulations may be studied, in other words microarticulations, such as dominant (neutral) orders, non-neutral orders, marked orders. The macrocondition constitutes a more abstract representation than the microconditions. This conceptual distinction seems to be important, even though from an empirical point of view the macrocondition of the basic order may in many cases coincide with one of the microconditions, that is, with that of the dominant order. Traditional typological parameters of correlation between the basic order of major constituents and other serial structures have been taken into account, especially in problematic cases where the scarcity of accessible data or differences of opinion in the existing literature or even the very complexity of the data make it difficult to determine the microarticulations. On the other hand, the importance of such a correlation must not be overestimated, since there are valid reasons to doubt the existence of an implicational relationship between basic constituent order at the sentence level and constituent order at a lower level.21

Interaction of syntactic and pragmatic factors

365

Where the available data allow it, rules of a statistical nature (frequency) have also been taken into account, such as that relating to the prevalent order.22 As the work progressed, the distinction between dominant or neutral orders and non-neutral orders has acquired particular importance. The former are characterized by the fact that they can always occur, while non-neutral orders may occur only under particular textual or contextual conditions. In one section of the literature on word order, non-neutral orders are also called “unmarked,”23 but here preference has been given to use of the term “marked” as a synonym of “emphatic” and, thus, to a distinction between non-neutral and marked orders. In our terms, the concept of non-neutrality includes that of markedness, but not vice versa. This seemed appropriate since in some languages, non-neutral contextually conditioned orders are not necessarily emphatic (cf. Russian and the discussion in Section 2.4.3.2), while in other languages, in particular the so-called “configurational” languages, most variations on the dominant order are associated with processes of emphatic focalization (cf. the Romance languages, English, and in part, German). As has been observed by Abraham (1986) from a generative perspective, in configurational languages “the underlying WO as defined by grammatical criteria is subject to linear variation as required by textual organization. Movement according to textual parameters results in marked linearization”; on the other hand, in non-configurational languages “any variety of WO is base-generated. Surface order will be obtained only by way of the needs of the discourse organization” (Abraham 1986: 15–16). 1.6.2. Models of order as a result of movement In generative models, the idea that languages allow basic orders and derived orders has been formalized in terms of movement rules which capture the correspondence between configurations with constituents in situ and configurations with constituents that are not in situ. Although highly prevalent, models which generalize movement rules raise serious problems of a theoretical nature.24 Representation by means of constituent movement has nonetheless a heuristic validity, perhaps for descriptive rather than explanatory purposes, and for some languages (configurational ones) more than others. In fact, the markedness of an order generally seems to be the result of a process which moves constituents from their “canonical” position in neutral orders. It is not, that is, an absolute but rather a differential parameter. It constitutes a sort of “segmental” correlate to what happens with contrastive stress at the suprasegmental level. As is well known, there is no single physical correlate to this type of stress pattern which functions in an absolute manner within the phonic sequence; in other words, what is perceived as contrastive stress is a series of differential parameters. In this sense, both contrastive stress and marked WO conform to a more general principle, according to which everything in the language is differential or relational. As will be seen, it is possible to arrive at a typology of differentiality for WO (cf. Section 6).

366

Rosanna Sornicola

Of particular importance has been the treatment of non-neutral “marked” orders, characteristic of structures with contrastive FOCUS (the so-called “second instance sentences” of the Prague School). The patterns associated with these structures have very particular properties in the various languages examined (and, it may be assumed, in all languages) and need to be dealt with separately. The discussion which has arisen in recent years in much of the pragmatic typology literature, concerning the nature of languages which have FOCUS in P1 or Pn (i.e., initial or final) position seems very confused, and does not distinguish the concepts of marked FOCUS and unmarked FOCUS. In many of the languages of Europe, FOCUS in P1 is only marked FOCUS: both SVO with weak WO flexibility and VSO languages show such behavior (cf. Sections 2.4.4 and 2.4.6). Account must equally be taken of Daneš’ observations that there are semantic types of sentences which may occur with only one pattern of order, even in languages with so-called free constituent order (consider, for example, a pattern such as that in Czech Lev je selma ‘Lion is a beast’).25 Daneš makes a distinction between “strong rules,” which relate to grammaticalized and fixed orders, “weak rules,” which relate to “usual” orders, and “free rules,” which relate to so-called “variable orders.”26 For this reason, it was considered appropriate to make a distinction in the discussion between the typology of structures with two arguments and that of structures with one argument (cf. Sections 2 and 3). As will be seen, it is possible to establish interesting correlations between basic two-argument patterns and one-argument patterns, and between basic two-argument patterns and certain distributional topological properties of the constituents which carry PF (cf. Section 3). 1.6.3. The concept of position Any study of WO presupposes a model of the sentence space organized according to “position.” It is possible to have a purely linear or sequential representation of positions (serial topological models), or a representation which takes positions to be projections of structural configurations, as in the generative models (complex topological models; but cf. Sections 1.7.2.1 and 1.7.2.2 for a critical examination of this). The problem that immediately arises is how positions should be counted. In fact, the concept of “position” is closely related to that of “constituent.” In other words, every computation of position refers to a particular level of representation by constituents. It is also clear that simple recourse to a linearly ordered sequence of positions P1 : : : Pn is not sufficiently informative/distinctive for describing WO properties in natural languages. If concepts such as “sentence initial position” and “sentence final position” could, at first glance, disregard structural configurational properties

Interaction of syntactic and pragmatic factors

367

(initial position = P1 , final position = Pn ), as far as other crucial positions are concerned, such as the immediately preverbal position or the immediately postverbal position, reference must continuously be made to categorial notions. The objective of studying the pragmatic typology of basic constituents obviously goes hand in hand with consideration of the constituents NP (S), V, NP (O) which, as is well known, are not all constituents on the same hierarchical level. The constituent NP (S) is on the same level as the VP projection which contains V as its head and the NP (O) branch governed by V. If the criterion of counting positions on the basis of the major constituents were adopted, NP (S) would be unproblematic as a position, but it would be more difficult to establish a correspondence between V and NP (O). If the same criterion applied to NP (= S) were followed, the entire VP would be associated with a single position. In this work, the convention of counting each of the phrases NP (S), NP (O) as a position, and of counting V as a position of its own, has been adopted, even though this creates a dishomogeneity in the criteria followed. However, such an assumption is indispensable for a typological study of WO, in which the special relational properties of the head constituent of VP play a crucial role. In SVO languages, V serves as a demarcation of the two GFs S and O. In SOV languages, V acts as the right-hand boundary of the nuclear sentence (all that follows is extrasentential and backgrounded: cf. Section 2.2.1). How important and how precarious the correspondence between position and constituent is may be seen in some of the problems concerning the representation of adverbial constituents in position P1 in the Germanic languages. But other difficulties must also be taken into consideration. The very notions of initial and final position cannot be straightforwardly defined without reference to theoretical models. For example, it is disputable whether in structures with left (or right) dislocation, or with hanging topics, very widespread as “natural” phenomena among the world’s languages, the dislocated or hanging element may be considered as being in P1 (Pn ), or otherwise. As is well known, some researchers have claimed it necessary to restrict the computation of positions P1 : : : Pn to constituents of the sentence nucleus only, reserving a different positional denotation (X1 : : : Xn ) for extranuclear elements.27 Now, a typological perspective may make the setting up of a single theoretical grid for the computation of positions more complicated. Extranuclear positions at the extreme left or right periphery have important textual properties concerning anchoring to the preceding and succeeding discourse, as seen in configurational languages such as the Romance and Germanic languages. In Russian, the equivalent of these pragmatic properties are the positions P1 , Pn which are not necessarily extrasentential (cf. Section 2.4.3.2). A further difficulty is posed by differences between the written and spoken language. The intrinsic structural segmentation of the spoken language complicates the establishment of correspondences between the set of positions and the set of categories/structural functions.

368

Rosanna Sornicola

Finally, special mention should be made of some structures that cannot be analyzed in a straightforward way, where the relationship between constituent and position is problematic. This is the case, for example, with cleft sentences in the Celtic languages (or in French), where establishing which constituent occurs in P1 is at the least controversial (cf. Sections 2.4.4.1.2 and 2.4.6.2). 1.6.4. Representation by fields The model of representation by “fields” determined by the verb is much used in works of functional typology. Thus a “preverbal field” is distinguished from a “postverbal field.”28 This model is influenced by the Germanic languages, where the verb has effectively a strong property of topological demarcation of the sentence. However, it is questionable whether the fields model is absolute. It is not, in fact, a representation which can lay claims to universality. For example, what may be considered the postverbal field in SOV languages has structural and pragmatic properties which differ from those of what may be considered the postverbal field in an SVO language. Structurally, if X is the constituent occupying the postverbal field, in SOV languages this will yield a representation: SOVX while in SVO languages it will yield a representation: SVX As far as pragmatic properties are concerned, in SOV languages the postverbal field always has a backgrounding function, while in SVO languages it may have a focalizing function (in neutral structures)29 or a thematizing/backgrounding function (in marked structures). The conclusion, therefore, may be that it is difficult to arrive at a consistent topology of the sentence of typological importance. Yet on an empirical basis it is worthwhile making attempts in this direction. Some spatial properties can be captured and can be described in a fairly analytical way, even if the task may not be considered exhausted, but merely started. It is as well to bear in mind, in any case, that topology is only one of the aspects of a typology of WO.

1.7. The descriptive traditions 1.7.1. Diversity in traditions of word order study A review of the reference literature on WO in individual languages (or groups of languages genetically interrelated) shows quite clearly to what extent assessment of the

Interaction of syntactic and pragmatic factors

369

actual phenomena, a prerequisite of any typological comparison, is mediated by the grammatical representations and models with which they are described. In this respect, there are different descriptive and explanatory traditions; some coincide with the study of specific linguistic families: the functionalistic models of Praguean inspiration dominate in the literature on Slavonic languages, generative models prevail in the literature of the last twenty years on the Germanic languages. At times, as in the case of the Romance or Celtic languages, the co-presence of descriptive traditions is noted. It is difficult to get away from the conclusion that these metalinguistic differences reflect more complex cultural differences. With respect to the comparison a no lesser difficulty is constituted by: (a) the different empirical data which are examined; (b) the different structure of the descriptions. For example, in the functionalist literature great importance has been given to the correlation between structural factors and prosodic factors, of which there are detailed and very interesting studies (cf. Cruttenden 1986, and this volume). On the other hand, in generative literature interest is focused on structural factors, while prosodic correlates such as stress (FOCUS) are looked upon as dependent variables with respect to them. Some ideas already present in the literature in the nineteenth century seem to have had and continue to have considerable importance, and have been taken up now and then in recent studies. Among them, those which may be worth of a mention are “WO freedom,” the influence of “case,” and the distinction between grammatical WO and pragmatic WO. The first two concepts are already found in strict relationship with each other in Weil (1879) and in other later works;30 the third is found in Meillet (1912: 147–148) and also in Mathesius (1939, 1941–1942), and more recently in Thompson (1978). In contemporary literature these ideas seem to have been given a somewhat more rigorous treatment compared to earlier formulations. A general problem in comparative typology, and one that resurfaces in a conspicuous way, is the relative nature of the parameters used, with respect to the language adopted (more or less implicitly) as the element of comparison. In fact, neither the concept of freedom (or rigidity) of WO, nor that of the pervasive influence of case, nor the distinction between languages with grammatical WO and languages with pragmatic WO are absolute principles. It is all the more important to bear this in mind in works on pragmatic phenomena which, as has been said, require a viewpoint which takes multiplicity and flexibility into account. Similar caution applies to other parameters of pragmatic typology which have been widely used in recent years, such as the distinction between topic-prominent languages and subject-prominent languages. The problem is complicated by the fact that both the generative and “neo-empiricist” typological31 literatures adopt English as the preferred language of reference for comparison.

370

Rosanna Sornicola

1.7.2. The d-configurationality model Partly in the light of this, the recent development of models such as d-configurationality will be considered, where the problems which arise in generative theories (and even more so in the EST phase) and in some functional theories seem to become more acute. The basic problem is that such theories do not distinguish between the two different dimensions of syntactic and topological representation. The notions of S, O, and categorial notions (N, V, NP, VP) belong to specifically syntactic representations, while concepts such as “position” and “field” belong to topological representations. In generative models, functional notions such as S, O are defined on the basis of positional notions. A similar albeit implicit assumption can also be made out in Dik’s functional models, which axiomatize GFs and associate them with linearization rules: “S and O are typically tied to specific positions, such that the S position precedes the O position. If S/O are not relevant in a language, constituent ordering will mainly be determined by pragmatic factors” (Dik 1989: 365). Elsewhere it is observed that languages such as Hungarian seem not to have specified GFs.32 Now, position is only one parameter amongst others for codification of GFs,33 but here it becomes a basic parameter: if a language does not display it, it follows that it does not have specified GFs. This conceptualization is biased by the restriction of the analysis to languages, such as English, in which GFs are codified with position as the element of comparison. 1.7.2.1. Heltoft’s criticism of generative models The observations of Heltoft, who analyzes generative models on the basis of Diderichsen’s traditional Danish model, seem to be no less important with respect to the confusion between syntactic and topological representations. According to Heltoft, the development of trace theory within EST appears – seen from the point of view of Diderichsen’s paradigm – to be an attempt to solve the problems involved in a topological description, but without introducing any particular topological level of analysis. No particularly topological notions are introduced into the theory (like fields or positions, for example), but the notion of “empty position” is replaced by two different complex notions: 1) empty categories, i.e. nodes where no lexical material is generated in the base, and 2) traces, i.e. nodes from which base-generated lexical material is moved transformationally. This leads to a peculiar kind of ambiguity with regard to the interpretation of the syntactic trees, which, consequently, must function simultaneously as syntactic and topological models. The branching in the trees (and the corresponding PS-rules) become ambiguous, and their significance either as syntactic categories or positions is determined by the origin of

Interaction of syntactic and pragmatic factors

371

the lexical material: base-generated lexical material is dominated by a category, whereas moved lexical material occupies a position. This, in turn, implies several things, partly of a general theoretical nature : : : The theoretical problem lies in the fact that it is impossible to interpret the tree diagrams as constituent structures, i.e. we are no longer dealing with a division of the sentence into its component parts. (Heltoft 1986: 61) Heltoft’s observations make reference to crucial questions, such as the relationship between positional representation and representation by constituent. We will see why forthwith. In topological models, actual representations and virtual representations may be distinguished. The former are distinguished by the presence of actual positions, the second by the presence of virtual positions. A string of actual positions must always be associated with a configuration by constituent, in such a way that every Pi corresponds to one and only one constituent Xi (this implies that there are as many positional representations as there are constituent levels). The total topological representation of a language is the set of the sequences P1 P2 : : : Pn which are associated in such a way to structural configurations. For example, in the Celtic languages V-initial and S-initial structures give rise, respectively, to the following correspondences: P1 V

P2 S

:::

Pn O

P1 S

P2 V

:::

Pn O

A syntactic representation (categorial or functional), in its constituent or functional relational reality, is by its very nature always associated with a model of actual positions. In fact, every constituent of a given syntactic structure is always found in a particular actual position. In such a syntactic representation, therefore, there is never an empty position (as in structuralist models and the first generative models). The syntactic representations are more real than positional representations by fields. But topological representations may, as has been said, contain virtual positions. The Felder (fields) model drawn up for the Germanic languages, such as those of Diderichsen or others (Section 2.4.5.3) are models of virtual position. They express an abstract or virtual maximal schema,34 independent of any particular structure. In these virtual representations, particular positions may be empty. In this case, an empty position expresses the fact that the structure does not saturate all of the cells of the virtual schema. Therefore, as Heltoft observes, in the EST/GB generative models there are positions which are not categories (as for example TOP). The EST models thus mix topological and syntactic representations. Correspondences between categories (GF)

372

Rosanna Sornicola

and positions are established aprioristically. A second point of confusion is the fact that both actual and virtual positions are used. A not very different conceptual mix also arises in a functional model such as Dik’s, where positional symbols co-occur with categorial/functional symbols in representations of the type: P1

V

S

O

where P1 merely expresses the potential occurrence of a constituent in such a position. These mixes may be considered the consequence of not having decided which representation is the independent variable and which the dependent. In a topological model, the independent variable is always position and the dependent variables are categorial or functional structures. In the functionalist models, the independent variable is always the abstract relation, in itself nonlinear, between constituents which carry a given function, while positional properties are the dependent variable. In other words, they are a spatial (linear) realization of logico-semantic relations. In this work representations have been adopted in which correspondences are established between topological (positional) models and categorial/functional configurations. In fact, the positions of a string were considered first and, hence, every position has been associated with a category/function pair. The WO patterns of a language are therefore the set of the correspondences thus defined. 1.7.2.2. Configurationality and d-configurationality The concept of d-configurational languages has developed in contrast to that of configurational languages. With regard to the latter, traditionally defined as languages in which the constituents with the GF of S and O have preestablished positions, dconfigurational languages are defined by the fact that the PFs of TOPIC and FOCUS may occupy fixed positions in them. The specificity of the conditions for d-configurationality has often been pointed out in the literature. The fact that WO is conditioned by discourse factors is not in fact sufficient for this to be determined. As Vilkuna (1989: 18) observes, “the additional requirement is that an essential part of the sentence structure be considered by positions for constituents with specific discourse roles. Provided that such functions have an independent status in the syntactic description of a language, one can claim that the language has syntactic functions with a discourse basis.” More recently, Kiss (1995) has defined d-configurationality thus: The (discourse-)semantic function “topic,” serving to foreground a specific individual that something will be predicated about (not necessarily identical with the grammatical subject), is expressed through a particular structural relation (in other words, it is associated with a particular structural position) : : : The (discourse-)semantic function “focus,”

Interaction of syntactic and pragmatic factors

373

expressing identification, is realized through a particular structural relation (that is, by movement into a particular structural position). (Kiss 1995: 6) Kiss observes that although these two properties often co-occur, they are independent. This in a certain sense is obvious, since it concerns the positional realization of quite different semantic (pragmatic) functions. Furthermore, at the level of realization, there are also clear differences between the range of positions which are characteristic of the TOPIC and those which are characteristic of the FOCUS. The d-configurational model may be criticized in various respects. Some of the difficulties are of a general nature. Although in Kiss’ (1995) more recent formulation, PFs are defined independently on the semantico-pragmatic level and, hence, positional properties are described as realizational properties, the model strongly favors the latter in a way that gives scant (or no) account of the complexity of interrelationships on the topological, syntactic, and pragmatic levels. The independent definition of PFs, on the semantico-pragmatic level, is not developed to its full potential. In fact, PFs are assimilated tout court to constituents which occupy a given position in the sentence, in conformity with the generative models critically examined in Section 1.7.2.1. In fact, it may be observed that if we assign the correct definition to PFs such as FOCUS in pragmatic terms, it will be seen that such a function cannot coincide entirely with a nominal or verbal constituent, etc., but with the whole sentence structure. These difficulties are also linked to more specific problems. For example, in much research dealing with characteristic FOCUS positions, no distinction is made between nonmarked and marked orders. As far as determining the positions of PFs is concerned, account must be taken of the fact that in the vast majority of languages, at least for the FOCUS function, there is no unique characteristic position, but a range of characteristic positions, according to variation in the syntactic and pragmatic parameters. Assessment of the position is further complicated by the fact that there is not even a biunique correspondence between position and the constituent which carries a given PF, in languages in which the range of positional options is very limited, as Turkish or Hungarian. In fact, in these languages as in others, the categorial nature of the constituent which carries the FOCUS function is a basic factor in determining the position: when FOCUS is the verb itself, the position of the FOCUS is different from that which the function would have if it were carried by a nominal constituent (cf. Sections 2.2.1 and 2.3.2). A more interesting version of d-configurationality has recently been elaborated by Abraham (1995), who critically examines the relationship between theme-rheme structure (T-R) and constituent structure.35 He observes that while other languages

374

Rosanna Sornicola

such as Hungarian may be considered [configurational] and [Corganized according to T-R], languages such as German are both [+configurational] and [Corganized according to T-R],36 which constitutes an anomaly in the traditional formulation of d-configurationality. Regarding this, it may be observed that the same situation may be found in languages other than German, such as the Romance languages. Of the three logical possibilities: (a) the T-R structure determines the constituent structure; (b) the T-R structure may be derived from the constituent structure; (c) the T-R organization and the sentence grammar are autonomous; Abraham claims that possibility (b) is methodologically the more interesting one.37 The claim that the T-R structure may be derived from the constituent structure in such languages is to an extent possible (cf. Section 2.4.5).

1.8. Form and function in the study of word order Finally, it is necessary to specify some theoretical assumptions of the work, such as the inventory of terms and concepts adopted, relating to the functions of the various levels (syntactic, semantic, pragmatic). It was felt appropriate to keep the dimension of constituency distinct from that of the functions of the various levels, as indeed the latter between themselves, and thus to study the interrelationships between these properties. This corresponds to the methodological principle of identifying variables and then studying their interrelationships. The topological (positional) structure has been assumed to be the independent variable, while the other variables (constituent properties, functions of the various levels) have been considered dependent variables. Definition of the structural domain within which WO phenomena are studied has necessitated a choice. Obviously a pragmatic study requires that the context of analysis be a whole portion of text. We shall call this the “maximal domain,” distinct from the “proper domain,” constituted by the sentence structure. In fact, although it is necessary to refer to a preceding or subsequent context in order for some WO phenomena to be understood in their textual dynamics (on the semantic/pragmatic level), it is always within a sentence structure that the topological properties of WO must be studied. The maximal domain is therefore connected to semantics and pragmatics, while the proper domain is connected to topological and syntactic structure (for an examination of the topological and syntactic properties of the proper domain, as determined by the preceding and subsequent maximal domain, cf. Section 5). It must be made clear that it is the simple declarative sentence that is examined as the proper domain here.38 For these reasons, examination of topological and syntactic structure is considered methodologically more important than examination of context. It reflects an orienta-

Interaction of syntactic and pragmatic factors

375

tion where, at least from a descriptive point of view, first the structure and then the function must be determined, even though the two directions of analysis, from form to function and from function to form, must continually overlap. In short, the chosen point of view is that the structures must be established first and then the dynamics of the WO. This has various methodological consequences: (a) neutral structures must be analyzed first, without necessarily referring to context, and then the marked structures, where reference to the context is more or less indispensable; (b) examination of the position-constituent-GF/SF correlations39 precedes examination of the correlation between these and PFs. 1.8.1. Grammatical functions, semantic functions, pragmatic functions On the basis of what has been observed, it is clear that each basic constituent has been considered as having three possible functions: X = (GF, SF, PF) GFs range over the set S, O, indirect O (IO); SFs over the set agent, patient, beneficiary, etc.: PFs over the set TOPIC, FOCUS, TAIL. As far as GFs are concerned, some problems are caused by the cases crucial to their determination, as in the ergative-absolutive languages in the sample (in Basque and Georgian, in fact, a problem arises in determining S with respect to the constituents with ergative or absolutive case) or in the Balto-Finnic languages (in Finnish and Estonian a problem arises in determining the functions S and O with respect to the partitive case). A further question is constituted by the treatment of constituents marked for morphological dative case but corresponding to a direct O (as in German: cf. Section 2.4.5.5.4, examples (70e) and (70f)). The biggest problem in defining functions is, however, posed by the PFs TOPIC and FOCUS. As is well known, they are terms with strong polysemy due to the varying uses made of them in the different linguistic traditions.40 In line with the principle just stated, that the functions of each level must first be defined in terms of the notions of that level (which, of course, does not exclude them being correlated with properties defined on other levels), TOPIC and FOCUS have been defined respectively as the function of “aboutness” and the function of “highlighting.” The PFs are therefore represented as primitive concepts of a semantico-pragmatic nature, whose structural manifestation is distributed over more than one level, namely, at the prosodic and the syntactic levels. What seems to be interesting for a theory of PFs is an examination of the interrelationships between the correlates on different levels. An example of an outcome of this type of examination may be given by the generalization which emerges in Section 2.4, according to which a relationship exists in configurational languages between the movement of a constituent into a position in

376

Rosanna Sornicola

which it is not in situ and the fact that it attracts nuclear stress with contrastive function. Examination of such interrelationships seems to be a prerequisite for a study of the pragmatic typology of WO. The notions of “aboutness” and “highlighting” involve a sort of “solidarity” with other concepts, which may be represented thus: Aboutness – GIVENNESS Highlighting – foregrounding – NEWNESS Note that as far as the first set of concepts is concerned, they do not form a necessary relationship: the function of “aboutness” does not automatically assign the feature [Cgiven] (think of TOPICS which introduce a new discourse referent). For this reason it is distinct from the textual properties traditionally used to define the concept of “theme” in textual linguistics.41 It is, furthermore, neutral with respect to the backgrounding/foregrounding dichotomy. A further characterization of the concept of “aboutness” concerns its relationship with referentiality and argument status. The function of “aboutness” is in fact more typically conveyed by referential elements than by predicative elements. Furthermore, it may not be associated with expletive elements (for example “dummy” pronouns). But the concept of “aboutness” is also related to that of argument status. As is well known, this last property is, in turn, strictly interrelated with referentiality, in the sense that if an element is an argument, it must be referential.42 Now, the function of “aboutness” is typical of elements with values high up in the argument hierarchy. The relationship between the function of “highlighting” and that of “foregrounding” is biconditional: the one implies the other and vice versa; neither of the two functions, however, implies the feature [Cnew], nor does the feature [Cnew] imply either of the other two functions. Highlighting is then a different pragmatic property from rhematization. The latter in fact is a typically textual property, connected to the progression of information in the sentence and the text, while this is not necessarily the case for highlighting. The definition of FOCUS in terms of highlighting could be more subtly articulated in terms of the subfunctions that FOCUS covers at the semantic-pragmatic level. Dik et al. (1981: 60 ff.) identify at least six subtypes: Completive FOCUS Selective FOCUS Expanding FOCUS Restricting FOCUS Replacing FOCUS Parallel FOCUS

Interaction of syntactic and pragmatic factors

377

Of course, even here, as in all semantic classifications, the number of subtypes could be expanded. However, such finer subclassifications are not very easy to use in a systematic way in large-scale typological surveys. We shall not resort to them here. A separate problem is that of the distinction between FOCUS in neutral sentences and FOCUS in marked sentences. In the first case, a characterization of FOCUS is in itself problematic (cf. Dik et al. 1981: 51). But even the relationship between the function FOCUS and that of “contrast” is not indisputable (cf. Dik et al. 1981: 57 ff.). In this work, the function of contrast is considered a special subtype of FOCUS. A recurring question in much of the literature on different languages (and particularly crucial in the study of WO in the Celtic languages: cf. Sections 2.4.6.2 and 2.4.6.4) is the confusion between the concept of TOPIC and that of FOCUS at the semantic-pragmatic level. In fact, it has been noted in many places that the function of establishing center of attention may be associated not only with “highlighting,” and thus with FOCUS, but also with “aboutness,” and thus TOPIC. However, this does not seem to be an irreconcilable opposition. In a certain sense, it could be said that both functions belong to a single, more general pragmatic function, which is that of “centering attention.” However, compared to FOCUS, TOPIC takes the lowest place in such a general function, while FOCUS takes the highest. This definition, which has its basis in the linearity of the discourse, finds its raison d’être in the sentence. In this sense, the neutrality of the TOPIC with respect to the backgrounding/ foregrounding dichotomy may be reinterpreted as the ability of the TOPIC to assume one or the other function according to the textual features specified in it: topics which introduce [Cnew] referents have something of a foregrounding value, although inferior to that of the constituent in FOCUS (for this hypothesis, cf. Sornicola 1993a). This representation allows, on the one hand, the functions TOPIC and FOCUS to be assimilated in terms of the increasing (or decreasing) hierarchy of “centering of attention” and on the other hand, the function TAIL (afterthought) to be distinguished from the first two functions. Thus, the following schema may be arrived at:

Table 1. Distribution of features of “centering attention” among PFs

TOPIC FOCUS TAIL

Backgrounding

Foregrounding

C  C

C C 

1.8.1.1. The syntactic and prosodic realization of pragmatic functions At the level of syntactic structure TOPIC and FOCUS always coincide with a particular constituent, the length and structural complexity of which are variable. The

378

Rosanna Sornicola

methodological principles outlined with respect to the relationship between the maximal domain and the proper domain in the analysis of WO (cf. Section 1.8) are also valid here. At the maximal domain level (text), TOPIC and FOCUS may each coincide with a whole sentence (cf. Sornicola 1981, and this chapter Section 4), while at the minimal domain level (sentence), TOPIC and FOCUS each coincide with a constituent or with a sequence of constituents.43 Of course, identification of the constituent structure which realizes the PFs, and in particular FOCUS, depends in part on the presuppositions associated with the structure. Although definition of the concepts of TOPIC and FOCUS are independent of their projection onto a syntactic structure, from a descriptive point of view it seems essential to take the syntactic structures as the point of departure from which to arrive at the pragmatic functions (cf. Sornicola 1993b). However, even from a descriptive point of view, it is important for the analysis to be multifactorial and bidirectional; in other words, it should proceed from GFs to PFs and vice versa. As has been seen in the previous paragraph, there are basic semantico-pragmatic differences between the function TOPIC and the function FOCUS. Similarly, they show considerable asymmetries on the level of realization. Despite what has been said about the relationship between TOPIC, aboutness, and referentiality/argument status, at the minimal domain the function TOPIC always coincides with constituents with the feature [V].44 From a topological point of view, it may be observed that the tendency to encode the function TOPIC in one of the sentence initial positions is a widespread phenomenon amongst the world’s languages. The function FOCUS has a wider range of realizational options. It may in fact be realized by means of suprasegmental (accentual) and/or positional devices, but may equally be realized by means of processes which isolate the constituent, as in French for example (cf. Section 2.4.4.1.2) or, again, by inserting into the structure lexical material which has a characteristic function of calling to attention: particles as in the case of Lapp (cf. Fernandez 1986, 1994) or pieces of more complex configurations, as in the case of the Celtic languages (cf. Section 2.4.6.2) or English (cf. Miller, this volume). Categorially as well, FOCUS may coincide with a wider range of constituents than TOPIC; in fact it also allows predicative constituents as its categorial domain, and in this sense it may be said that it ranges over any type of constituent.45 Finally, FOCUS is realized in a wider and different set of characteristic positions than that of TOPIC; this set varies according to the type of language and to whether it concerns an unmarked or marked FOCUS. In the generative literature, FOCUS has often been made to coincide with prosodic correlates of the phrase structure.46 It is in fact defined as the constituent that receives the main stress. From this point of view, differentiation between unmarked and marked FOCUS may be made on the basis of “default stress” and “contrastive stress,”

Interaction of syntactic and pragmatic factors

379

which, however, involve interesting empirical problems in the prosodic analysis.47 On the other hand, in much of the functionalist literature, stress is considered to be a prosodic correlate of pragmatic FOCUS.48 In this work, the latter perspective is adopted.

2. Word order in sentences with two arguments: The relationship between position, syntactic function, and pragmatic function 2.1. TOPIC position The range of TOPIC positions in the languages of the corpus is much more restricted than that of FOCUS positions. In both unmarked and marked orders, SOV languages have the constituent carrying the PF TOPIC in one of the positions preceding Pk (i.e., the immediately preverbal position [see Figures 1–2]). In the SVO languages in the unmarked sentences the constituent carrying the PF TOPIC is in one of the positions preceding V, while in the marked sentences it may occupy either any preverbal or any postverbal position (see Section 2.4). The lack of one-to-one mapping between position and TOPIC function is also typical of VSO languages, where the TOPIC position varies according to the unmarked or marked character of the structure of order (see Section 2.4.6.6).

2.2. Languages with O in preverbal position in the basic order and FOCUS in preverbal position In a vast linguistic area, encompassing Altaic languages of the Turkic (Turkish, Tatar) and Mongolian branch, Uralic languages such as the Ugric (Ob-Ugric and Hungarian) and Volgaic (Mari) languages, as well as Basque, the constituent which bears the function FOCUS is obligatorily placed in the position immediately preceding the verb.49 Specification of the nature of the constituent in this position is necessary because, as will be seen in Section 4.1, different configurations arise when it is V that carries the function FOCUS. This has interesting repercussions on the theoretical side because it challenges attempts to express the function FOCUS in unequivocal terms of position (configuration) (cf. also the discussion on Hungarian). This characteristic of argument FOCUS, that it occurs in immediately preverbal position, co-occurs with a collection of other syntactic and morphological properties: (1) all the languages mentioned have basic SOV order (except Hungarian, the assessment of which is problematic: cf. Section 2.3.2); (2) all are agglutinating; (3) in addition, all occurrences of WH-words in interrogative sentences are obligato-

380

Rosanna Sornicola

rily placed in the immediately preverbal position.50 The areal distribution might lead to the belief that these characteristics are part of the Uralo-Altaic family inheritance, reinforced perhaps by processes of areal convergence (as in the Mari-Turkic relationship), but a broader study, extended to other Euro-Asiatic areas, shows that a number of the features in question are present in most of the modern Indo-Aryan languages as well (Hindi, Nepalese, Bengali, etc.51): with the exception of Kashmiri,52 they all have basic SOV order, and have FOCUS and WH-words positioned immediately before V,53 although defining their morphology as inflectional or agglutinating is more problematic.54 Moreover, the situation with respect to the oldest phase of the Indo-European languages, the basic pattern of which was SOV,55 shows that word order as such cannot be correlated with the WH-word position parameter. In fact, in these languages WHwords in interrogative sentences occurred in a position preceding the verb, often in P1 , but were in no way specifically associated with the immediately preverbal position.56 In fact, none of these three apparently connected parameters – SOV order, agglutination, and WH-words in immediately preverbal position – seems to form an implicational relationship with the obligatory placing of FOCUS in immediately preverbal position. For convenience, we shall indicate the placing of FOCUS in immediately preverbal position as A, SOV order as B, agglutination as C, WH-words in immediately preverbal position as D. It can be said that A does not entail B (cf. the situation in Hungarian described in Section 2.3.2), and, conversely, B does not entail A (cf. Japanese, an SOV language which, however, does not exhibit parameter A). Also worth mentioning is the fact that some SOV languages, such as Somali and Quechua, do not have property A, but have a property that may be called A’, defined as: If X is the constituent with the feature [Cfocus], X may occur in one of the preverbal positions.57 A0 is obviously a quite different property from A: the only focal position defined by A is a proper subset of the range of possible focal positions defined by A0 . As far as C is concerned, this parameter seems to be only indirectly related to A. It is, however, well known that agglutination exhibits a strong correlation with parameter B.58 More promising instead is the relationship between A and D. Making crosslinguistic generalizations, Horvath proposed the existence of a close relationship between the position of FOCUS and the position of WH-words in the following terms: The syntactic position(s) in which non-echo interrogative WH-phrases can appear in a language L will be identical to or be a proper subset of the positions in which FOCUS constituents can appear in a language L. (Horvath 1986: 122)59

Interaction of syntactic and pragmatic factors

381

In languages in which X with the feature [Cfocus] occurs in immediately preverbal position, WH-words are expected to occur in the same position, and therefore it is expected that A entails D. In fact, there does not seem to be a single case in which A but not D is found.60 In conclusion, it seems, as far as is known, that the immediately preverbal position for X [Cfocus] is indeed a property which has developed in SOV languages; it is, however, genetically and areally too localized to be considered a characteristic of substantial typological significance (the only clear evidence of it comes from some Ural-Altaic languages and Basque). Although the relationship between association of the FOCUS constituent with immediately preverbal position and SOV type is not implicational, it nonetheless seems to be far from accidental: in Section 6, the hypothesis is advanced that it arises from a syntactic property, that is, the occurrence of O immediately preceding V. 2.2.1. Basque and Turkish Basque and Turkish are two languages in which obligatory placement of the argument constituent in FOCUS in preverbal position has been well studied.61 Consider the following examples from Basque: (1)

Aita-k untzia aurdik-i d-u. father-erg vase(abs) throw-pst.part 3sg.obj-have(3sg.sbj) ‘The father has thrown the vase.’

(2)

Untzia aitak aurdiki du. ‘The vase, it is the father who has thrown it.’62

Sentence (1) shows “normal” word order in Basque. The pragmatic function attributable to the O constituent is that of unmarked FOCUS: it is clearly analogous to what occurs in SVO languages where in sentences with a “neutral” prosodic contour, the pragmatic function of unmarked FOCUS may be assigned to the constituent with the syntactic function O. On the one hand, as is clear from example (2), untzia in P1 is the TOPIC, while whichever constituent other than O occurs in the immediately preverbal position constitutes a marked (contrastive or emphatic) FOCUS. If, on the other hand, an O occupying the immediately preverbal position (i.e., its canonical position in Basque) is also associated with a prosodic contour which carries heavy stress, it may be considered a marked FOCUS.63 Note that O occurring in its “normal” (i.e., immediately preverbal) position may carry what Lafitte (1962: 17) defines “affective stress.” It is not clear whether in this case it can be considered a marked FOCUS (e.g., a constituent with a contrastive value) or merely a constituent with a prosodic intensification of affective (emotional) nature. Similar considerations apply to the examples from the Turkic languages. In Turkish, a structure such as (3a) has a neutral constituent order:

382 (3)

Rosanna Sornicola

a.

Murat para-yı bu adam-a ver-di. Murat money-acc this man-dat give-pst (3sg) ‘Murat gave the money to this man.’

Structures (3b)–(3f) are also possible, each of which has a marked order, in the sense of being both nonbasic and pragmatically marked: (3)

b. c. d. e. f.

Murat bu adama parayı verdi. ‘As for Murat, it is the money that he gave to this man.’ Parayı Murat bu adama verdi. ‘As for the money, it is to this man that Murat gave it.’ Parayı bu adama Murat verdi. ‘As for the money, it is Murat who gave it to this man.’ Bu adama Murat parayı verdi. ‘As to this man, it is the money that Murat gave to him.’ Bu adama parayı Murat verdi. ‘As to this man, it is Murat who gave the money to him.’64

As can be seen, the FOCUS constituent must be placed in the immediately preverbal position. It should be noted that, as in Basque, unmarked or neutral FOCUS (that is “the element that conveys the new information in a sentence with an unmarked order”) is that “immediately to the left of the verb.”65 The situation with respect to WO in Turkish is further complicated by the effects of semantic and pragmatic factors such as features of definiteness, animacy, and referentiality (cf. Section 5.1). The structure with FOCUS (whether neutral or unmarked) in immediately preverbal position is also found in Tatar: (4)

a.

b.

Kolxozˇc@-lar uzgan jel bu traktor-n@ sat@p al-gan-nar collective.farmer-pl last year this tractor-acc buy-pst.part-pl i-de. be-pst ‘The collective farmers bought this tractor last year.’ Bu traktorn@ uzgan jel kolkozˇc@lar sat@p algannar ide. ‘As for this tractor, the ones who bought it last year were the collective farmers.’66

In those languages which follow this pattern, the following linear ordering of the pragmatic functions TOPIC and FOCUS is clearly discernable: P1 : : : Pk V, where P1 = TOPIC; Pk = FOCUS This characteristic is related to another one which is also found in the languages examined: the demarcation of two distinct fields with respect to V, the “preverbal field” and the “postverbal field,” the first reserved for codification of the functions TOPIC and FOCUS, the second for linear deployment of informationally given or of

Interaction of syntactic and pragmatic factors

383

“backgrounded” material, in accordance with conditions partly linked to afterthought processes (cf. Schroeder 1995). As far as the pre-field is concerned, on the basis of what was said in Section 1, the sequence P1 : : : Pk may be represented as a scale of highlighting increasing in value from P1 to Pk . This is a reformulation of the interpretation laid down by classic functional sentence perspective theory, according to which the “normal” distribution of information in a sentence would go from the minimum in P1 to the maximum in Pn , where V = Pj and the constituent in Pn follows V.67 It confirms that it is not the position of V with respect to the argument constituents S and O that is crucial for information flow, but only that of S and O.68 It may therefore be claimed that when FOCUS is not distributed over the entire sentence, but instead coincides with a particular argument of V, its position is determined by the position (or range of positions) that O may assume in a language (cf. Section 6). However, when it is V that carries FOCUS, the languages in question show additional important properties. A different analysis is offered by Rebuschi (1984: 77) who observes that in the Navarro-Labordino variety of Basque, if a sentence has the order O + participle + Aux, the focal element may only be the participle or the O + participle phrase.69 In Navarro-Labordino, disambiguation of this double possibility may be achieved by means of the Aux – V participle inversion structure which unambiguously identifies the constituent preceding Aux as FOCUS. In any case, these properties force a reappraisal of the d-configurationality model, which establishes a biunique relation between position and PF. Bearing in mind the limitations just mentioned, it seems that the following generalizations with respect to the languages under examination may be put forward: – Biuniqueness of the relationship between position and FOCUS (Pk in the configuration Pk V is the FOCUS); this property holds for constituents with referential value; – Isomorphism of the relationship position/FOCUS between structures with marked FOCUS and those with unmarked FOCUS; – Lack of influence of grammatical functions with respect to the allocation of FOCUS. On the basis of what has been said, the representation in Figures 1–2 may be arrived at; this shows a specific correspondence between position, pragmatic function, and grammatical function.

384 P1

Rosanna Sornicola ...

T

Pk

V

F

unmarked order

O Figure 1. Alignment of Positions, PFs and GFs in unmarked orders of SOV languages

P1 T

...

Pk F 8 9

ˆ ˆ > ˆV> =
ˆ > ˆ ˆ ; : > O

Figure 29. Relation between position P1 and GFs in the possible space of VS0 languages (in marked and unmarked sentences)

possible movement of S is to the preverbal position. On the other hand, VSO languages present some interesting structural affinities with SVO languages. Summarizing what has emerged in Section 2.4.6, the following model may be formulated: Principle X The VSO languages of the corpus show different positional properties for codification of PFs in neutral and in marked sentences; Principle XI They do not have a unique FOCUS position for neutral sentences and marked sentences; Principle XII In most varieties of these languages, in marked sentences, P1 is the only position for FOCUS. Furthermore, as was stated in Section 2.4.6.4, in VSO languages, P1 is a position in which either S or O may occur with the PF of TOPIC in neutral sentences. As a consequence, there is reason to believe that in the languages in question P1 is a multifunctional pragmatic position. In VSO languages, therefore, as in SVO languages, GF and PF movement is possible under pragmatic conditions. On the other hand, these languages have a characteristic conspicuous position, valid for both TOPIC and marked FOCUS (cf. Section 2.4.6.4). There would, however, be no justification in assimilating this latter property to that of PF codification, operating in SOV languages such as Basque and Turkish. In fact, the “movement” of constituents to P1 under pragmatic conditions is a widespread phenomenon among the world’s languages. In other words, P1 is a generally conspicuous position in many languages, presumably for reasons related to the serial organization of the utterance. On the basis of what has been said, the representations in Figures 29 and 30 apply in VSO languages. Figure 29 represents the fact that P1 is a possible position for every basic constituent in both neutral and marked structures, while Figure 30 represents the fact that such a position may be occupied by the PF TOPIC or the PF FOCUS.352

Interaction of syntactic and pragmatic factors

485

P1 8 9 < TOPIC = :FOCUS; Figure 30. Relation between position P1 and PFs in the possible space of VS0 languages (in marked and unmarked sentences)

Extrasentential domain

V

S

O

Extrasentential domain

Figure 31. Proper sentence domain and extra-sentential domain in VSO languages (Model 1)

Extrasentential domain

P1 . . . 8 9 ˆ > ˆV> ˆ = < > S > ˆ > ˆ ˆ ; : > O

...

Pn

Extrasentential domain

Figure 32. Proper sentence domain and extra-sentential domain in VSO languages (Model 2)

The representation of possible space is more difficult. If one accepts the idea that preverbal S/O occupy an extrasentential position, it may be claimed that V signals the left-hand edge of the sentence domain, according to the representation in Figure 31. An alternative representation would be that in Figure 32, which shows that position P1 is not typically reserved for V but is “flexible” in that it may be exploited by any one of the basic constituents under different pragmatic conditions. This representation would, in fact, appear to be preferable, since in Section 3.3 it was claimed that in the VSO languages of the sample labile rules of order may be postulated for both intransitive and transitive structures. It suggests that V is not the only constituent which delimits the leftmost edge of the proper sentence domain, but rather that this boundary is signaled by a variable. This would give rise to a further property differentiating VSO and SOV languages. In the latter, the relationship between position and constituent is rigidly defined, while in VSO languages it is not. The topological properties which have emerged in this work lead one to think that WO phenomena may be considered as more than just the realization of grammatical relations. The topological influences on WO are presumably stronger than some of the regularities which have been traditionally postulated, such as the distinction between languages with pragmatic WO and those with grammatical WO or the in-

486

Rosanna Sornicola

cidence of “morphological case.” On the other hand, what has emerged in this work is that there is a relationship between the topological principles mentioned and other means of PF codification, such as the use of clefting processes (as in French and the Celtic languages) and intonation (as in English). These are, as was stated in Section 2.4.4, demarcation phenomena typical of languages in which the linear axis is comparatively less exploited for the codification of PFs. Such phenomena appear to be language-specific. Another result which has emerged from this work is that the topological properties relevant to bi-argumental structures are different from those which are relevant to monoargumental structures, but that they seem to be correlated in an interesting way (cf. Section 3.6.2).

6.2. Word order and the asymmetry of pragmatic functions In Section 2.1, it was noted that grammatical codification of TOPIC and FOCUS by means of WO reveals these functions to be typically asymmetrical. The first is in fact realized by a narrower range of possible positions than the second. This characterization concerns neutral rather than marked sentences. In fact, in all the languages of the corpus, the property of being “what is being spoken about” in neutral sentences is manifested by assigning TOPIC to P1 or in any case to a preverbal position.353 Furthermore, in all the languages of the corpus, TOPIC tends to coincide with the GF S in neutral sentences. Minor differences have emerged regarding the possible patterns in which TOPIC coincides with another grammatical function, as in Russian, for example (cf. Section 2.4.3.2). In fact, it has been seen that in this case, TOPIC in P1 may be a constituent other than S. The question regarding the linear properties of TOPIC and its association with a given GF in marked sentences is more controversial. As will be recalled, doubts have, in fact, been raised about whether one can always identify TOPIC in marked sentences, since “aboutness” may be – so to speak – “absorbed” by FOCUS (cf. Section 1.8.1). With the exception of Basque and Turkish, the range of possible positions for FOCUS is wider in marked sentences than in neutral sentences. One may well ask whether TOPIC, like FOCUS, is definable in terms of a differential strategy (cf. Sections 6.1.3 and 6.2.2). It is plausible to claim that, among the world’s languages, TOPIC is less characterized than FOCUS in terms of differential movement strategies, a conclusion which seems consistent with the more general asymmetry between the two PFs. With respect to the codification of FOCUS in particular, the languages of the corpus may be grouped according to two types: (a) languages with no difference at all between neutral (unmarked) and marked sentences;

Interaction of syntactic and pragmatic factors

487

(b) languages in which PF codification is differentiated according to whether the sentence is neutral or marked. Basque and Turkish belong to the first group, as do languages in which TOPIC is always in the same P1 position and FOCUS is always in the same immediately preverbal Pk position. The Indo-European languages, as well as Finnish and Estonian, belong to the second group; in these languages, FOCUS has a wider range of possible positional codifications in marked sentences (cf. Figures 3–7 in Section 2.4.1). Before examining this typological bifurcation more closely, it would be appropriate to look in greater depth at the relationship between position, constituency, GF, and PF, and in light of this, to reconsider the possible codifications of FOCUS which have emerged in the course of this study. 6.2.1. The relationship between position, grammatical function, and pragmatic function It should first of all be pointed out once again (cf. Section 1.6.2) that the concept of “movement” is not without theoretical problems354 and that it should rather be used as a handy metaphor for heuristic purposes. Furthermore, the expression “function movement” must be understood as an abbreviation for a conceptually more complex expression. In fact, the primary manifestation of movement is always of constituents and only in second place may one speak of “GF or PF movement.” In the present work, a four-level model has been employed, in which a different representation corresponds to each level: Representation by position This is the set of n-ple (P1 , P2 , : : :, Pn ), where each Pi is a position. Representation by constituent This is the set of n-ple (X1 , X2 , : : :, Xn ), where each Xi is a constituent. Representation by GF This is the set of n-ple (1 , 2 , : : :, n ), where each i is a GF. Representation by PF This is the set of n-ple (1 , 2 , : : :, n ), where each i is a PF. In light of what has been said so far, it may be claimed that a direct relationship generally only arises between positions and constituents, and in addition between constituents and GFs on the one hand, and between constituents and PFs on the other. The representation of Figure 33 may therefore be arrived at. This expresses the fact that the relationship between position and GF and between position and PF is mediated by the representation by constituent. However, this model may be modified on the basis of properties of individual languages. In

488

Rosanna Sornicola Representation by position

Representation by constituent

Representation by GF

Representation by PF

Figure 33. Relationships between positions, constituents, GFs, and PFs

fact, the relationship between position and GF and between position and PF is differently determined according to the type of language. The same may be said of the relationship between GF and PF. In theoretical and typological research, the nature of the relationship between GF and PF has been a controversial matter for a while.355 In particular, the current debate may be summed up in the question of whether it is the GFs which determine PFs or vice versa. The view held in the present work is that any relationship between GF and PF may only be considered in relation to the level of positional codification of the two sets of functions, since, as was said in Section 1, it is assumed that they have independent abstract representations. For this reason, the conclusions arrived at here may be considered to be fundamentally different from those obtained in the framework of the d-configurational model. In languages where PFs have a conspicuous canonical position, such as Basque and Turkish, there are several GF configurations but only one PF configuration. Variation in the basic order is found only with respect to the GFs (cf. Section 2.2.1). In other words, the GFs but not the PFs may be moved. Furthermore, the relationship between GF and PF is mutually independent and is not subject to any conditions: independently of whatever its GF may be, a constituent acquires a given PF by virtue of occupying a given position, which is unique. On the other hand, in the SVO languages of the corpus, both GFs and PFs may be moved and both types of function are interdependent, at least as far as FOCUS is concerned. In fact, as will be mentioned in the next section, in order for a constituent to carry (marked) FOCUS, it must occur in a position which is not the canonical one for its GF. This means that a constituent may be positionally determined as FOCUS only after it has been assigned its GF. In neutral sentences, instead, although there is a more or less regular correspondence between GF and PF, it cannot be claimed that PFs are determined from GFs or vice versa.

Interaction of syntactic and pragmatic factors

489

6.2.2. Codification of FOCUS by means of word order We may now turn to an examination of the two different strategies for syntactically codifying FOCUS that have emerged from the languages of the corpus. In the first, FOCUS, whether unmarked or marked, is assigned to the immediately preverbal position Pk . We shall call this “Strategy 1.” The second strategy has two closely interrelated properties: (a) FOCUS is associated with different structural positions in unmarked and marked sentences; in particular, the position of FOCUS tends to coincide with that of O in unmarked sentences; (b) in marked sentences, FOCUS is not associated with a specific unique position, but with a set of characteristic positions which must be differentially defined as the set of positions that the constituents with the GFs S or O cannot occupy in unmarked sentences. Regarding this, it should be noted that in marked sentences, both the functions S and O may coincide with FOCUS. In this strategy it is not an absolute positional property, but a relative or differential property which characterizes FOCUS, that is, it is the difference between the canonical position that the constituent has in neutral sentences (generally unique356) and the position which it occupies in marked sentences (a position which is not unique, but varies according to the marked pattern). We shall call this “Strategy 2.” The difference between the types of structure resulting from one or the other strategy may be formulated in terms of absence of movement vs. movement of the PFs (for further details on this, cf. Section 6.2.3 below). It is clear that Strategy 1 is without PF movement, that is, these functions have fixed positions. Strategy 2 is instead characterized by PF movement along the linear axis, that is, these functions have moveable positions. An important and consistent characteristic of the second strategy is that constituent movement to a noncanonical position resulting in its bearing the function of marked FOCUS, always co-occurs with the constituent in question receiving nuclear stress. It should be pointed out, however, that in the languages of the corpus with PF movement, principle XIII generally applies, according to which: Principle XIII Movement of a constituent from its canonical position to the anterior (preverbal) sentence space is preferred to movement to the postverbal position. This principle has sometimes been characterized as “movement to P1 ” and has been considered a sort of universal. But it is doubtful whether it is anything other than a fairly widespread tendency among the world’s languages. VSO languages, which behave only in part like SVO languages (cf. Section 3.1.2), require separate discussion. Although WO is used in the codification of PFs, it cannot

490

Rosanna Sornicola

be said that on its own it allows unmarked structures and marked structures to be differentiated. Context and suprasegmental features are instead used for this purpose (cf. Section 2.4.6.6). The two strategies mentioned constitute a typological parameter which differentiates the SOV from the SVO languages. This typological difference seems of interest; discussion of matters arising from it will be developed in the following paragraphs. They concern the correlation between types of basic order and types of PF codification by WO, the nature of the relationship between the GF O and the PF FOCUS, the potential relationship between morphological type and type of PF codification and, finally, the areal distribution of types of PF codification by WO. In SOV languages, the position Pk is conspicuous. In SVO languages, instead, the crucial WO property is not the position, but the “change” or difference in position of the constituents which carry GFs. In SVO languages, therefore, every noncanonical position may be said to be a conspicuous position. This property seems to be determined by a principle of differentiation; its importance in the functioning of the languages has already been expressed by the well-known structuralist principle which states that “dans la langue il n’y a que des differences.”357 This is the same principle which can be seen operating in prosodic structures. Phenomena of “intonation” and “accent” cannot, as is well known, be determined on the basis of the absolute or intrinsic characteristics of an element, but rather on the basis of the relative or differential value that they assume with respect to the corresponding value of other elements in the same structure. Similarly, the pragmatic functions of WO may also be realized by characteristics of relative prominence; they may, that is, depend on comparison with another term. As will be seen, this type of codification is more widespread than codification by means of absolute position. The differential character of WO can be interpreted in two ways. Firstly, in the wider and more common meaning, “order” is in itself a relative concept independent of the type of language (cf. Section 1.6.1). The second meaning is more specific and may be associated with the typological property which has now emerged and which in short may be reformulated thus:

Principle XIV (principle of “differentiation of the codification of FOCUS”) In languages which conform to Strategy 2, the PF FOCUS in marked sentences may be represented positionally as the set of pairs: X (Pi , Pj ) such that for each pair, Pi is the canonical position which a constituent X with a given GF  occupies in a given neutral sentence, and Pj is the position which it occupies in a corresponding marked sentence structure.

Interaction of syntactic and pragmatic factors

491

6.2.3. The relationship between the grammatical function O and FOCUS A point which appears particularly interesting is the relationship between the GF O and the PF FOCUS. In many languages, the constituent with the function O coincides with FOCUS in unmarked sentences, according to a principle of constituent structure which may be stated thus: Principle XV The constituent with the greater depth of embedding is that which bears FOCUS.358 As is clear, this constituency property holds independently of the serial ordering of O with respect to V. In other words, it applies not only in SVO languages, but also in SOV languages. We shall call the relationship between the constituent structure and FOCUS the “constituency property of FOCUS.” Note that this is not necessarily connected with the property which allocates prosodic FOCUS, since the latter may coincide with a constituent which is not the most embedded. This is not surprising if one considers that the prosodic structure is not, in principle, isomorphic with the syntactic structure. Principle XV may be reformulated in generative terms, by assigning to the PF FOCUS a syntactic structure representation in which the constituent X = FOCUS is a projection of V (a feature assigned by V). In other words, X = FOCUS may be considered a constituent governed by V (for discussion of this possibility, cf. Kiss 1995; Kiss [ed.] 1995; Jo 1995). It is immediately obvious that in languages with a differential strategy, principle XV may apply only to neutral or unmarked structures; only in such structures, in fact, do the most deeply embedded constituents belong to the maximal projection of V (that is, they are governed by V). This principle does not, however, generally apply to structures with marked FOCUS. In fact, these have derived representations in which a given constituent carries FOCUS not because of its original constituency property, but because it occupies a noncanonical position. This may apply not only to O, but also to constituents with the GF S, that is, constituents which occupy a node high in the tree structure and are not governed by V. Therefore, in these cases the fact that the constituent which bears FOCUS is positionally defined as the most embedded constituent is neither a necessary nor sufficient condition. If an element is moved from a less embedded to a more embedded position, the definition in terms of “most embedded constituent” will be trivially true (in the sense that it is not the position which defines the constituent). If instead an element bearing neutral FOCUS is moved from a more embedded position to a less embedded one, the definition is simply not true. In marked sentences, then, the constituent which carries FOCUS is not necessarily the one with the greatest degree of embedding. Furthermore, in languages with PF

492

Rosanna Sornicola

movement (that is, languages with a differential strategy) assignment of FOCUS to a given GF does not automatically and exclusively depend on a phrase structure property that can be represented by a single phrase structure tree. In fact, to take account of the differential character of the property in question, at least two structures are needed; the first comprising information concerning the canonical positions of the different constituents, the second comprising information relative to the “derived” positions. In any event, in these cases formulation of the relationship between FOCUS and GF in terms of assignment of the feature [Cfocus] to a constituent by V may be held as unsatisfactory. The problem is no less thorny for languages without PF movement. Given that they have a fixed position Pk for FOCUS (both marked and unmarked FOCUS), and furthermore, that it coincides with the canonical position for O, a plausible hypothesis would be that the property whereby V assigns the feature [Cfocus] holds not only in neutral or unmarked sentences, but also in marked sentences. However, it would be interesting to establish whether the constituent with the GF S which turns up in position Pk becomes “deeply embedded” only because it occupies the canonical position of an O constituent. This is not, however, an easy problem to solve out of a theoretical framework. The hypothesis that in languages such as Basque and Turkish the position of O determines the position of FOCUS shall be called “Hypothesis 1.” Note that such hypothesis implies that Principle XV holds not only across languages but across the distinction of unmarked and marked sentences as well. But to what extent can this be maintained? The problem just raised requires consideration of the relationship between V and the preverbal position in languages characterized as agglutinating. Is it really because Pk is a conspicuous position in terms of constituency that it has been generalized as FOCUS position? After all, principle XV – which should explain why O-position determines position of FOCUS constituent – is not so general. We have seen that it does not hold in marked sentences of languages with PF movement. It may be irrelevant – for other reasons – also in languages without PF movement. In fact, in Turkish specific semantic factors – like animacy and definiteness – are at work in determining what constituent must fill the immediately preverbal position (cf. Section 5.1). The possibility that nonconfigurational, semantic factors may determine a fixed FOCUS position Pk will be called “Hypothesis 2.” 6.2.4. Some possible typological correlations with the parameter “FOCUS in Pk ” 6.2.4.1. Hypothesis 1 Two groups of indications, of a typological and a diachronic nature respectively, seem to favor Hypothesis 1. They offer, in particular, a unitary explanation of the situation in Basque and Turkish as well as Aghem, an African language of the Bantu

Interaction of syntactic and pragmatic factors

493

family. The latter language, in fact, has a basic SVO order and a fixed position for FOCUS coinciding with that of O.359 A second group of indications is provided by Hungarian. As was said in Section 2.3.2, there is reason to believe that grammaticalization of the immediately preverbal position for FOCUS is relatively recent in this language. Furthermore, and more interesting for the present study, it seems to go hand in hand with the increase in the frequency of occurrence of O in immediately preverbal position. Acceptance of Hypothesis 1 would lead to the conclusion that the FOCUS position is determined by the GF O. There are, however, certain empirical and theoretical difficulties that such a conclusion must be reconciled with. The question that, above all, must be asked concerns the range of the empirical domain of the property that FOCUS is determined by O. Is it a generalization supported by other SOV languages,360 or is the property in question restricted to Basque and Turkish? Examination of the wider empirical domain is very important for confirmation of Hypothesis 1. In fact, one would expect that in other languages in which O precedes V in the basic/dominant order the property of codification of FOCUS in immediately preverbal position Pk has developed. As will be seen below, results from examination of data from various SOV languages do not appear to conclusively confirm Hypothesis 1 (cf. Section 6.2.4.1.2). From a descriptive point of view, it may be claimed that Hypothesis 1 is restricted to the few languages in which FOCUS occupies position Pk . However, the scope of the potential correlation between O and FOCUS is of a more general nature. This lies more properly in the theoretical domain. For this reason, a conclusion such as that just advanced would not be satisfactory since it would create more problems than it would solve. In particular, it would raise the question: why has this property developed in a tiny number of SOV languages and not in others? A secondary, but not negligible, empirical problem is that a model of representation such as that arising from Hypothesis 1 would not be compatible with some of the results which have emerged from studies in pragmatics and language acquisition over the last thirty years. It has in fact been shown that the GF S has ontogenetically developed from the PF TOPIC. Furthermore, over the last decades interesting evidence has been gathered that grammaticalization of PF TOPIC into GF S is also a path in the diachronic development of many languages.361 6.2.4.1.1. Some typological properties of SOV languages. Examination of the pragmatic properties of WO in a wider corpus of SOV languages offers some food for thought. Japanese, Korean, Kannada, Burushaski, Somali, and Quechua have been examined. In these languages, the basic SOV order has degrees of “dominance” varying according to the basic constituents in the sentence. Although all these languages clearly favor V in the final position of the proper sentence domain, they allow scrambling phenomena of various types within this domain.362

494

Rosanna Sornicola

Two parameters have been considered: the properties of the postverbal field and the position of FOCUS. It turns out that, with the possible exception of only Burushaski,363 in all these languages the postverbal field is given over mainly to afterthoughts or backgrounded material, just as in Basque and Turkish.364 As far as FOCUS position is concerned, only Kannada and Quechua appear to exhibit a property linking the FOCUS constituent to a given region in the proper sentence domain.365 However, in neither case is it the same property as that found in Basque and Turkish. In Kannada the immediately preverbal position for the constituent in FOCUS is a fairly common strategy, although position P1 is accessible to some constituents in FOCUS.366 Quechua in its turn offers a different picture which will now be considered in more detail, along with Japanese and Somali. These three languages may be considered representative of three different typologies with respect to codification of FOCUS order in SOV languages. 6.2.4.1.2. Japanese, Somali, and Quechua. In Japanese, the function FOCUS is codified by many processes acting together. The particles wa, ga, yo, ne, and se appear to be of particular importance; these mark the constituent that immediately follows them as FOCUS.367 As far as WO is concerned, Hinds (1986: 150) observes that “any perturbation of neutral sentence order will create emphasis” and that “scrambling or postposing will result in a display of emphasis.” Although interesting, these remarks must be treated with caution in typological comparison: the concept of “emphasis” to which Hinds refers is not homogeneous with that of FOCUS used in the present study, nor with the descriptive tools exploited by those studying other SOV languages. This shows clearly, then, that typological comparison is difficult even when based on reliable studies of individual languages. In particular, examination of “postposing” as a means of emphasis is not very convincing368 or at least it suffers the same pitfalls as Kuno’s examination of Turkish criticized by Erguvanli. It is unclear – from the available data – whether Japanese behaves in a different way from Basque or Turkish as far as the structuring and function of the postverbal space is concerned. With due caution, then, Hinds’ observations lead one to think that Japanese has free constituent movement in the preverbal space and PF movement. In fact, the examples reported by Hinds (1986: 150) show that the constituent in FOCUS, marked by a particle, may occur anywhere in the preverbal space. In any case, what appears to be important is that in Japanese, the canonical space within which “scrambling” is defined is the preverbal space, a situation similar to that found in Basque or Turkish. In Somali, which does not have a conspicuous position Pk for FOCUS, the entire preverbal space appears to be given over to constituent movement and to movement of the PF FOCUS. P1 has particular importance and may be considered a conspicuous position for FOCUS. It should be pointed out, however, that – as in Japanese and Quechua – PFs are identified by the occurrence of particles with topicalizing or focalizing value (cf. Svolacchia et al. 1995).

Interaction of syntactic and pragmatic factors

495

The situation in Quechua offers interesting, but problematic, comparisons with that in Basque and Turkish. In a study on Huallaga variety, Weber (1989: 15) has pointed out that of the characteristic properties of an SOV language, the less consistently displayed is actually the SOV pattern in main sentences. Weber describes an order which is “fairly free, particularly in nonsubordinate clauses” with considerable variation according to speaker and to speakers’ styles within the same dialect (Weber 1989: 15, 402–403). This description agrees with that of Calvo Pérez (1993: 41), according to whom in Cuzco Quechua all six orders SOV, S0 OV, S00 VO, S00 VO0 , O0 VS, and O0 S00 V should be possible with varying pragmatic values.369 The structural and pragmatic analysis is complicated by the presence of particles, such as -mi/-n, qa, the semantic values of which are not easy to analyze satisfactorily. These particles have been described in the literature as “evidential.” According to some, -mi and -n are probably focalizing particles, and -qa is probably a thematizing particle;370 others, while claiming that this distinction is problematic, nonetheless recognize that -mi occurs more naturally with NEW or rhematic information,371 and that -qa often occurs with thematic material. In fact, in many contexts the first particle has the characteristic evidential value of ‘learned by direct experience’; the second may indicate contrast in some contexts, while in others it is affixed to the focal element in yes/no-questions.372 In any case, Weber’s analysis allows the following formula for canonical phrase structure (i.e., that to which most sentences in Huallaga Quechua conform) to be drawn up: (176)

(X-qa)0n

(Y-EVD)01

V {O-qa, S-qa}

The formula states that there may be any number of elements with -qa, followed by the element with -mi/-shi/-chi (i.e., the evidentials), which must precede the main verb or be the main verb. The verb in its turn must be followed by one or two elements with -qa, which must necessarily be S or O (Weber 1989: 428). Weber observes that the evidential suffix must not be identified with a particular element, for example, the last thematic or the first rhematic element. He claims that the only conclusion that may be reached is that the thematic material occurs to the right of the evidential suffix and the rhematic material follows the last preverbal -qa.373 This observation would seem to indicate that Quechua behaves like Basque or Turkish as far as closure of the sentence space after V is concerned. Whether this means that a position Pk which is salient may be postulated for Quechua seems more difficult to say. 6.2.4.2. Hypothesis 2 Hypothesis 2 is also not without problems. According to this hypothesis, it would be semantic features – like animacy, definiteness, and referentiality – which determine the position of preverbal FOCUS. The typological correlation with languages such as Aghem would be lost, but it would be possible to obtain a more finely-grained

496

Rosanna Sornicola

description of the properties of a particular language. We therefore find ourselves up against a typical problem concerning the relationship between the microscopic and the macroscopic study: the more an analysis takes the peculiarities of individual languages into account, the more difficult it becomes to make typological comparison, to determine uniform types (for example, is it really possible to classify Turkish along with Basque as far as the property of FOCUS position is concerned? In particular, do the features animacy and definiteness have the same function in Basque as in Turkish?), and to establish universal tendencies. It has to be admitted that the more detail one goes into from the microscopic perspective, the more difficult it becomes to determine uniform and homogeneous types. A very general problem is that “sameness of properties” across languages is a problematic notion in itself: what prima facie appears as an identity may turn out to be – on closer inspection – a difference. Regarding the positional properties of German, for example, comparison between German and Hungarian has been seen to be illusory (cf. Section 2.4.5.5.5). The influence of semantic factors can be seen in another possible correlation, that between FOCUS position and the position of WH-words. As is well known, WHconstituents typically have the value [CFOCUS]. In fact, all the languages which have the property of FOCUS fixed in Pk , such as Basque, Turkish, Mongolian, and some Indo-Aryan languages, also have the property of WH-words canonically occurring in Pk .374 What appears of some interest is the fact that languages such as Japanese, Korean, and Somali, which do not have a conspicuous position Pk for FOCUS, equally do not confine WH-words in such position.375 Kannada and Quechua, which, albeit in different ways, have typological affinities with Basque and Turkish, diverge in an interesting way with respect to these languages and between themselves. In Kannada WH-words may occur in P1 , in Pk , or in another position in the proper sentence domain,376 which is of some interest since these are the same positions which a non-WH-constituent with FOCUS may occupy. In Quechua, instead the typical position for WH-words is P1 , a position which does not coincide with that reserved for non-WH- elements in FOCUS.377 The correlation between fixed position for FOCUS in Pk and fixed position for WH-words is therefore important, but as far as current knowledge goes, it is difficult to establish which of the two factors has acted upon the other. Hungarian, which has fixed position for FOCUS in Pk , but not for WH-words, offers an argument in support of the claim that WH-words may not be the motivating force in the process. 6.2.4.3. A third hypothesis A third hypothesis (we shall call it “Hypothesis 3”) would be to postulate a relationship between the various properties of PF codification and morphological properties such as agglutination and inflection. In fact, the type without PF movement occurs in SOV languages which are typically agglutinating, such as Basque and Turkish.

Interaction of syntactic and pragmatic factors

497

Agglutination could be responsible for the linear property of (left) adjacency to V of the FOCUS constituent. As is well known, agglutinating languages generally have a lesser degree of freedom of constituent movement than inflecting languages, and this seems to be all the more true in languages such as Basque and Turkish, in which agglutination phenomena favor the occurrence of well-known processes of phrasal association of morphemes. These processes are, among other things, clearly evident within headmodifier structures, such as for example V + subordinate clause, in which the modifier is normally attracted to preverbal position.378 The position of inflectional case morphemes in NPs such as: (177)

Ankara ve Izmire gid-ece˘g-im. Ankara and Smyrna.dat go-fut-1sg ‘I am going to Ankara and Smyrna.’

also confirms the impression that agglutinating morphology, which favors phrasal compounds of elements such that a unique inflectional morpheme applies to two lexical morphemes at the same time,379 may also be a decisive factor in determining phenomena like the conspicuous FOCUS position Pk in Turkish. Inflecting languages, on the other hand, show a wide range of constituent movement due to the phenomenon described by Meillet (1912) as “autonomy of the word.” In fact, in these languages the word is the typical locus of grammatical relations, while in canonically agglutinating languages such as Turkish, this is not the case. This may well account for the differences which have been pointed out between Turkish and Hungarian, with respect to, for example, the influence of the features animacy and definiteness, and the behavior of WH-words. Although distinguished by agglutinating processes, Hungarian is a less typically agglutinating language than Turkish and therefore has a freer constituent order, in conformity with what is postulated by Hypothesis 3. However, not even the correlation between morphological type and type of PF codification by position can offer an explanation on its own. Agglutinating SOV languages like Japanese, Korean, and Somali have not developed the property under examination,380 which has an areal distribution confined to a few regions of the globe.

Notes 1.

I am indebted to Giuliano Bernini, Georg Bossong, Bernard Comrie, Peter Matthews, Gabriella Mazzon, and Erich Poppe for comments and observations that have helped me to improve the work. To Giorgio Banti, Marcello Chierchi, and Giampaolo Salvi I owe information on data from Semitic languages, Georgian, and Hungarian, respectively. Mistakes are my own and I take full responsibility for this chapter. I also wish

498

2.

3. 4. 5. 6.

7. 8.

9. 10. 11.

12.

13. 14.

Rosanna Sornicola to thank the colleagues in the group of research on “Pragmatic Organisation of Discourse” of the EUROTYP Project of the European Science Foundation. The fruitful discussions with them during the years of the EUROTYP research gave me interesting opportunities for reflection. Giulia Perretti and Giovanni Palumbo assisted me in editing the text at various stages of its preparation. To Giovanni Palumbo I also owe valuable comments and observations. Finally, I want to acknowledge financial support from the Department of Modern Philology of the University of Naples Federico II, without which this piece of research would not have been possible. By basic constituent order, the relative order of S, V, and O is meant, as in the standard meaning taken from Greenberg (1963). The term “constituent” is of course to denote both elements of a categorical nature and those of a functional nature, such as S, O. Although this confusion is often pointed out in the typological literature, the expression has fallen into common use. For terminological and conceptual clarification with respect to the general organization of the present work, cf. Section 1.8. These alterations depend in the first place on the language type, but the factor of “style” also has a role to play. Regarding the importance of this distinction, see Daneš’ (1967: 217) observations. In the generative literature this has been represented in terms of movement rules. For an assessment of this point of view, cf. Section 1.6.2. This tradition extends from the work of theorists of the Prague School (cf. Sornicola and Svoboda 1992) to the more recent works of Dik and Comrie (see, e.g., Dik 1978, 1983, 1989; Comrie 1981b). For studies concerned more specifically with WO, cf. Siewierska (1988, 1998) and Connolly (1991). Cf. Payne (1992), Mohanan and Mohanan (1994). For an assessment of the role of psycholinguistic (or cognitive) factors in typological research on WO, see Tomlin (1986: 3–5, 133–139). Chafe (1986) and Givón (1983) have especially emphasized the cognitive constraints on the organization of sentential or textual information (cf. also the various contributions in Givón [ed.] [1983]). For a different line of research, see Vennemann (1972) and Hawkins (1983, 1994). Cf. Holden and Krupp (1987: 269). Cf. the papers in Nuyts and de Schutter (1987). For the influence of Latin and Greek on the syntax of modern languages of Europe, see Blatt (1957), Nykrog (1957), Sørensen (1957), Stender-Petersen and Jordal (1957). Of particular importance were the linguistic effects of Bible translation from either Greek or Latin on European vernaculars during the Middle Ages (see the contributions in Lampe [1969: 338–491]). For problems concerning the relationship between history and typology, cf. Croft (1990: 18–25), in particular p. 24: “This discussion of the sampling problem should make clear to the reader how diachronic considerations enter typology even at its methodological foundations”; Bell (1978: 146): “I think that a case can be made that such research [on language universals] can properly be conceived as sampling language changes, not languages themselves.” Cf. Sinor (1990: 16). Cf. Sinor (1990: 16–18).

Interaction of syntactic and pragmatic factors 15.

16. 17. 18. 19.

20.

21. 22.

23. 24.

25. 26. 27. 28. 29. 30.

31. 32. 33. 34. 35.

499

Cf. PET (1995–1997: IV, 437b). The question of an exact boundary for Europe arises only in the sixteenth century and acquires truly scientific form only with the birth of modern geography. Interesting observations regarding this question may be found in Sinor (1990). Primary sources were gathered for Russian, German, English, French, Italian, and Spanish. For a discussion of these problems, cf. Croft (1990: 25–26). The methodological problems that this option involves have so far not received sufficient critical attention in the literature on WO: for a defense of this point of view, cf. Abraham (1995). Cf. Greenberg (1963: 76), Siewierska (1988: 8 ff.), and Connolly (1991: 16–19). For a criticism of the notion of “basic WO” as traditionally used in contemporary typology, see Brody (1984). Connolly (1991: 37) convincingly criticizes the notion of “underlying order” of earlier generative models. Cf. Comrie (1981b: Ch. 4) and Croft (1990: 24–25). Until recent years, studies on variation of constituent order in a typological perspective have not been many: see Steele (1978), Holden and Krupp (1987), Connolly (1991), and Hawkins (1994). Cf. Daneš (1967: 218), who compares the concept of “neutral order” to that of “usual order.” Cf. also Dryer (1995). Cf. the excellent observations made by de Meij and Marácz (1986: 268 ff., and especially 273) and Comrie (1996). An analysis of the problem may also be found in Kiss (1995). This problem will be readdressed in Section 6. Cf. Daneš (1967: 221). Cf. Daneš (1967: 223). On the problem of variable orders, cf. this chapter, Section 3. See, for example, Dik (1978: 21, 1989: 359 ff.). This idea can also be found in recent generative models. Cf. Dik (1989: 342–370) with a lengthy discussion of this. Or rhematizing if neutral structures are considered as being without FOCUS. Cf. Weil (1879) and for the history of ideas on WO, cf. also Holland (1980: especially Ch. 1). The “prehistory” of ideas on WO goes back to the Middle Ages and even beyond until Classical Antiquity: cf. Scaglione (1972). Of particular interest is the extensive and philosophically based discussion on “ordre naturel” in Enlightenment France: for an overview of the debate, cf. Bossong (1990: 229–269). An interesting examination of the post-Greenbergian movements can be found in P. H. Matthews (1994: 44–45). Cf. de Groot (1989: 10–11 and Ch. IV) and Dik (1989: 363–365). See, for example, the treatment of the GF S in Keenan (1976). On the concept of maximality, cf. Heltoft (1986). Abraham claims that the concepts of theme and rheme should not be considered only in their textual dimension, but should be determined positionally and with respect to the structure in constituents (Abraham 1995: 605). While theme is here synonymous with TOPIC, rheme is not synonymous with FOCUS; the latter in fact is considered to be the stress-bearing constituent and as such is a subconcept of rheme.

500 36. 37. 38. 39. 40.

41. 42. 43.

44. 45.

46. 47.

48.

49.

50.

Rosanna Sornicola Note that the expression “organized according to T-R” is equivalent here to “d-configurational.” On the question, cf. Abraham (1995: 606–607). In the present work, it is not possible to study the properties of WO in interrogative sentences since they require a detailed examination in themselves. In some cases (cf. the section on Russian) SFs have also been important. Cf. Daneš (1967: 217 and Note 2) for the history of the terminological and conceptual pair TOPIC/COMMENT (FOCUS). See also van Oosten (1985), Yokoyama (1986: Part 1), Holden and Krupp (1987: 258–259), Dik (1978: 87 ff., 141 ff., 1989: 266 ff., 277 ff.), Sgall et al. (1986: 175 ff.), Sornicola (1993a), Lambrecht (1994: 117 ff.), Maslova and Bernini (this volume), Miller (this volume). For an examination of the definitions of “theme” in textual linguistics, cf. Halliday (1967a, 1967b, 1967c, 1968); Dik (1978: 132 ff.). The reverse relationship is of course not valid. Cf. Höhle (1982: 99), who observes that FOCUS does not necessarily coincide with a constituent, but may, in particular in presuppositional conditions, be distributed over a string of constituents. Note that V may well be thematic, but as far as the distinction made between TOPIC and theme is concerned, this is a different property. Note that in many of the languages of Europe, the category “preposition” is also an element in the set of possible constituents over which the function FOCUS ranges: cf. Italian Sono stato IN casa ‘I was AT home’ (lit. [I] am been IN house) with corrective FOCUS. The typological distribution of this property would, however, deserve further investigation. Cf. von Stechow and Uhmann (1986) for an overview of the relationships between prosodic and phrasal FOCUS in the generative literature. For a recent proposal regarding differentiation in this sense, cf. Abraham (1995: 615 ff.). For problems concerning the prosodic analysis, cf. Ladd (1996: 160 ff., 198 ff.) and Cruttenden (this volume). Cf. Dik (1978: 131 ff.), who opposes identification of FOCUS with the constituent which receives main stress, and Comrie (1984), with a clear formulation of this point of view. For Turkic languages, cf. Comrie (1981a: 79–80); of particular importance for Turkish is Erguvanli’s (1984) study; for Mongolian, see Poppe (1954). For the Uralic languages, see Comrie (1981a: 92, 122–124), who emphasizes how important a property the SOV/SVO feature is in differentiating the Eastern Uralic languages – some of which, such as Mari (Cheremis), have been exposed to a strong Turkish influence – from the Western ones. Some of the latter in fact (especially the Balto-Finnic languages) have come under the influence of Indo-European languages (cf. Sauvageot 1973: 9–94). For Basque, cf. de Rijk (1978, especially 109) and Ortiz de Urbina (1995). The literature on Hungarian is extensive: cf. Section 2.3.2. For the general problem of the preverbal position as locus of FOCUS placement, see Kim (1988). For a comprehensive study of these characteristics in the Uralic and the Altaic languages, see Bese et al. (1970). For the correlation between the position of WH-words

Interaction of syntactic and pragmatic factors

51.

52. 53. 54.

55.

56. 57.

58.

59. 60.

61.

62. 63.

501

and the position of FOCUS in the Altaic languages, cf. Comrie (1981a: 79–80, 124); for Turkish in particular, see Erguvanli (1984: 34–35). On the position of WH-words in Basque, cf. de Rijk (1978: 103) and Ortiz de Urbina (1995: 100). For Hungarian see, among others, Kiss (1987: 56), Horvath (1986: 44 ff.), de Groot (1989: 25). The agglutinating characteristic is more problematic for the Uralic languages (cf. Comrie 1981a: 177–220), which move to a greater or lesser extent away from the type in question, “while still being in general closer to this type than to the fusional type represented by the older Indo-European languages” (Comrie 1981a: 117). For the general view of the relationship between SOV order and agglutination, see Lehmann (1973). For an introduction to Indo-Aryan languages and their numerous dialectal varieties, cf. Masica (1991: 8–23); for some problems in identifying such languages, cf. Masica (1991: 27 ff.). Cf. Masica (1991: 335), Bhatt (1994). Cf. Masica (1991: 395, especially ex. (463)), altogether similar to the Turkic structures with respect to the positional properties of the WH-constituent. Cf. Masica (1991: 332–333, and 395 for WO); for the complex question of morphological properties, cf. Masica (1991: 212–216). On WO in Indian languages, see also Swarajya Lakshmi and Mukherjee (1996). Cf. Delbrück (1878: 14, 1888: 16, 1900: 82–83) and more recently Lehmann (1974: 30 ff.), C. Watkins (1976), Verpoorten (1977). For a historical overview of the problem, cf. Holland (1980: Ch. 1). Cf. Delbrück (1900: 259–260) and Lehmann (1974: 121). The parameter in question is therefore presumably an innovation in modern Indo-Aryan languages. The situation in Somali and Quechua is further complicated by the fact that in both languages, the constituent with the feature [Cfocus] is classified according to its combination with certain particles as well as by its position and prosodic properties. For Somali, cf. Svolacchia et al. (1995); for Quechua, cf. Muysken (1995). Cf. Austerlitz (1970). Moreover, not even this relationship is implicational: Cushitic languages, which are SOV, are highly synthetic (cf. AWL 1994: 274). Modern IndoAryan languages, which are SOV, as previously mentioned, are not uncontroversially considered agglutinating. Horvath adds that the principle spells out a necessary but not sufficient condition for the position of WH-phrases. Georgian presents an interesting case; D is found (cf. Aronson 1990: 184 [sentence no. 16], 186 [sentence no. 44]) but not A (cf. Section 2.3.1). The situation in Armenian is the same (cf. Comrie 1984). However, as is well known, this does not deny the validity of the relationship A entails D. The considerable degree of dialectal fragmentation found in Basque must be taken into consideration. The data made use of here are from literary Navarro-Labordino, a variety that enjoys considerable prestige and that served as the base upon which the linguistic standard was established. Lafitte (1962: 47) translates (2) as ‘le vase, c’est le père qui l’a jeté.’ According to Lafitte (1962: 48), in exceptional cases, inversion of Aux and V participle may be a device for focalizing the constituent which precedes the verbal block: cf.

502

64. 65. 66. 67.

68. 69.

70.

71. 72. 73. 74. 75. 76. 77. 78.

79.

Rosanna Sornicola Aitak du aurdiki ‘c’est le père qui l’a jeté,’ Aitak untzia du aurdiki ‘c’est le vase que le père a jeté.’ Lafitte observes that “ici le relief est très accusé, même violent.” See, however, the observations of Rebuschi (1984: 77) on this point; he claims that inversion of Aux and participle V is anything but exceptional. On the basis of an analysis conducted on actual texts, he maintains that the phenomenon has no contrastive value at all. Examples (3a)–(3f) are taken from Erguvanli (1984: 33–34). For a discussion of Turkish WO, see also Banfi (1998). As observed by Erguvanli (1984: 37). Cf. Comrie (1981a: 79). In other words, word order in a sentence would conform to an increase in the so-called communicative dynamism: cf. Firbas (1965) and for a more recent formulation, Firbas (1992). This theory has been adapted in the framework of American functionalism by Chafe (cf. latterly Chafe 1994: Ch. 13). Regarding this suggestion, see for example Desz˝o (1973: 254). This description does not seem to take into account the difference between broad and narrow FOCUS: O + participle is broad FOCUS, while O must be narrow FOCUS. On the other hand, there are other ways of focalizing V: in Vizcayno and Guipuzcoano, the verb egin ‘to do’ is inserted immediately after V and before Aux (cf. Rebuschi 1984: 71). Indirect O raises specific questions and the problems associated with it which arise in Basque and Turkish are not considered here. Indirect O, in any case, has not been held to be a basic constituent. Cf. Comrie (1981a: 222). For an analysis of WO problems in Georgian from a generative perspective, cf. Boeder (1989). The problem in general, as well as the results reported here, are discussed by Vogt (1971, 1974), where a critical examination of other literature may also be found. The examples are taken from Vogt (1974: 52–53). The linguistic remarks and translation into English are due to Marcello Cherchi. I quote this word in the form given by Vogt (1974: 72) (note that in Standard Georgian it has the form /gamarZvebuls/). In Georgian folklore a dev is a type of demon. I would like to thank Marcello Cherchi for supplying me with this clarification. Note that Vogt exemplifies each of the two structures with only a single occurrence from his corpus. Cf. Harris (1981: 14–18). The two different types of prosodic contour are described by Kálmán et al. (1986: 130– 131). “Eradicating stress” is defined as “a main stress that is not necessarily stronger than a normal main stress, but which ‘eradicates’ all subsequent stresses, and thus, cannot be followed by any more [sic!] main stress (except for multiple contrast)” (Kálmán et al. 1986: 132). On the other hand, in sentences which have a focused constituent, the postverbal position of O is strongly preferred if the sentence is imperative or contains a negation (Behrens 1989: 142).

Interaction of syntactic and pragmatic factors 80.

81.

82.

83.

These sentences are reported by Kenesei (1986: 144), without a detailed pragmatic analysis. I integrated the pragmatic interpretation which I put forward here on the basis of a native speaker’s judgements. This characteristic has been ably described by de Groot (1989), according to whom Hungarian has two special positions in the clause, P1 , or the position of the constituent with TOPIC function, and P0 or the position of the constituent with FOCUS function. The examples are taken from Kálmán et al. (1986: 133, 140). The HOCUS position may also be occupied by a pre-verb, cf. Kálmán et al. (1986: 133, 140), Tompa (1985: 227) (with different terminology and descriptive devices). Cf. also Behrens (1989: 148–149), according to whom the position in front of the verb in nineteenth century texts is to be analyzed as being without FOCUS for [referential] O; cf. also, however, what Behrens (1989) has to say in Note 182. Komlósy (1986: 218) claims that in a sentence such as: (i)

84.

85.

86.

87.

503

Mari lát-t-a János-t a kert-ben. Mary see:pst:3sg.sbj.3sg.def.obj John-acc Art garden-iness ‘Mary saw John in the garden.’

(with main stress on látta) the FOCUS position is completely empty. Komlósy (1986: 220). Hungarian, however, has structural properties which differ from those found in Turkish with respect to the placement of the FOCUS constituent and this can be clearly seen from two further characteristics: (a) differences in the effects of semantic features in determining the constituent occupying the immediately preverbal position (cf. Section 5.1); (b) differences in the distribution of subordinate clauses. De Groot observes that when the FOCUS function coincides with a finite subordinate clause, the latter may not occur in front of the verb, but after it; in this case the immediately preverbal position will be filled by a “dummy element” which represents the subordinate clause and which indicates the focal property of the latter element (de Groot 1989: 25–26). It should be emphasized that only structures in which the positions P1 : : : Pn constitute the proper domain of the sentence, and not structures with extrasentential constituents (right- or left-dislocated, as in the schema of Fig. 7), are considered here. From a historical point of view, the suggestion, defended by Meillet in many works, that there was originally a close relationship between “autonomy” of the word and freedom of WO in Indo-European is important; later, the historically documented IndoEuropean languages probably developed specific patterns of WO. For an interesting examination of hypotheses concerning WO in Indo-European languages between the nineteenth and twentieth centuries, cf. Holland (1980: Ch. 1). For the development of the WO patterns from Latin to the Romance languages, cf. Meillet (1906: 8–9) who, in conformity with his more general ideas, saw in these languages a loss of word autonomy. The influence of the hypothesis that word autonomy correlates with freedom of constituents in the sentence is also found in Wartburg (1950: 104) with respect to free word order in Old French (constrained, however, by the rules of the everyday language). For the diachronic development of French, cf. MarchelloNizia (1995). For the diachronic development of English, cf. Jespersen ([1922] 1969: 344–345, 355–357).

504 88.

89.

90. 91.

92. 93. 94. 95. 96. 97. 98.

Rosanna Sornicola For German, cf. Section 2.4.5.5; for Icelandic, see Pétursson (1978: 138 ff.), Kress (1982: 263 ff.), and in a generative framework, Sigurðsson (1989: 5–6). The situation in Modern Greek is also interesting: despite conservation of a case morphology, the modern language has moved away from the old language (cf. Dover 1960; Schwyzer 1950: 691 mentions a “habitual” or usual order, which limited the structural freedom of Classical Greek), in reducing the range of structural possibilities and in choosing SVO as the predominant pattern (cf. De Simone Brouwer 1921: 217). Thumb (1912 : 200–201) maintains that “modern Greek vernacular has, on the whole, maintained the ancient Greek freedom in the order of words, i.e. all kinds of combination are possible in the sequence of the composite parts of the clauses.” However, the counterpart of this potential freedom is “a recognized normal sequence of words, so that any deviation from the same lends a special emphasis on the irregular member” (Thumb 1912: 201). Cf. also Kalitsunakis (1963: 183 ff.). For a recent account in a generative grammar framework, cf. Tsimpli (1995). In a corpus constituted of a sample of 2254 sentences taken from Ælfric’s Homilies (the last phase of Old English, circa 1000), Kohonen (1982) found 54 % SVO order and only 13 % SOV order. Cf. furthermore Mitchell (1985: II, 963 ff.). Dik (1989: 351) is of the same opinion. On the diachronic and typological problems relating to the basic SVO order in these languages, cf. Comrie (1981a: 92–93, 121 ff.); the literature on WO in Finnish is quite large: cf. Hakulinen (1961, 1979: especially 499 ff.) and van Steenbergen (1989) who, on the basis of experiments, shows the configurational nature of the language. Vilkuna (1989: 9) observes that in a simple sentence consisting of S, V, O and some further complement or adjunct “all permutations are grammatically possible.” In both Estonian and Finnish, specific grammatical restrictions on the complex form of the verb may be observed (cf. Kippasto 1989: 86 with a comparative study of Estonian and Votian; Vilkuna 1989: 9). Cf. Sauvageot (1973: 193–194). Cf. Tauli (1983: 24–25), although he uses the term “emphasis” and not focalization. Both examples are taken from Tauli (1983: 54). All the examples are taken from Hakulinen (1961: 315). I have indicated with an apostrophe in front of the constituent the fact that this is the one that is stressed. The suggestion that all deviations from basic SVO order (“default”) are marked has also been put forward more recently by Vilkuna (1989: 42). Hakulinen (1961: 316) observes, in fact, that the beginning and the end are “the most stressed positions in the sentence.” Vilkuna has adapted to the d-configurationality framework an idea put forward by Karttunen and Kay (1985), according to whom the discourse functions contrast and topic probably have the following fixed positions: (i)

99.

Contrast, Topic, : : :

Vilkuna, however, proposes specific descriptive differences with respect to the d-configurationality model put forward by Kiss for Hungarian. Cf. Vilkuna (1989: 37).

Interaction of syntactic and pragmatic factors 100.

101. 102. 103. 104.

105. 106. 107.

108.

505

Vilkuna (1989: 42). The pragmatic analysis is, however, perplexing: if a T constituent, defined as the intersection of the concept of old information and aboutness (cf. Vilkuna 1989: 38, 79–84), appears in these configurations, how is it possible that the examples quoted are “all-new sentences”? On p. 42 of Vilkuna (1989) “default T” is defined as “the grammatical subject, if present.” Vilkuna also mentions “a little K,” which occurs in the V-field, but it refers to something else. Vilkuna (1989), however, often refers to this author (on p. 41, his 1976 work is quoted). I refer to the suggestion that because a constituent is a TOPIC, it must have referential value. In fact, this concept is only indirectly alluded to on p. 40 (Vilkuna 1989), in the wake of work by Hakulinen and Karlsson, which talks of T in terms of a “nominal constituent.” Vilkuna herself observes that T is a function “normally” realized by the constituents NP and Adv (Vilkuna 1989: 38). For an extensive and lucid account of FOCUS function values, cf. Dik (1989: 288 ff. and in particular 282–285). On the way such a model may apply to Finnish, see Tarvainen’s (1985: 363) observations. Vilkuna herself has reservations over this. For Bulgarian, cf. Beaulieux (1933: 359–360) and Guentchéva (1994); for Polish, Meillet and De Willman-Grabowska (1921: 203–204); for Serbo-Croatian, Meillet and Vaillant (1924: 288). The literature on Czech from the Prague School is vast (see the fundamental work by Mathesius 1941–1942; Daneš 1967; Havránek 1968; Uhlíˇrová 1969; Svoboda 1984; Sgall et al. 1986: 175 ff., 194 ff.); the literature on Russian is also extensive (cf. Adamec 1966 and Krylova and Khavronina 1976 for an approach in terms of Praguean functionalism; see also Benoist 1979, 1988 and the works of Nikolaeva, Kovtunova, Sirotinina, Lapteva, quoted in Yokoyama 1986 and Fougeron 1989, who may be referred to for both the bibliography and a critical examination). In a different perspective, studies based on spoken texts have recently shown that basic SVO order may be postulated for various Slavonic languages. For Russian, Holden and Krupp (1987: 267) claim that “there is strong support : : : for the notion that there is a ‘basic’ SVO sequencing of constituents which affects native speaker’s acceptability judgements of sentences regardless of their contextual associations, i.e. independent of the distribution of given/new information, or theme-rheme, etc.” In a similar vein, Uroševi´c et al. (1986: 178) point out that in Serbo-Croatian “OSV is the next most frequent order in speech after SVO, while V-initial orders are the least frequent.” In the generative literature, Babby (1978), making the case for Russian, and Tajsner (1990) for Polish, both oppose the view that SVO is the basic order (Babby claims that there is no independent basic WO in Russian other than that determined by pragmatic conditions). The two studies use terms and concepts which are variously drawn from the Slavonic linguistic tradition (theme and rheme, etc.). I will endeavor as far as possible to adapt the data and the analyses presented in them to a different theoretical framework, such as the one presented here. Cf. Yokoyama (1986: 173–175) for the history of the problem of WO in Russian and the impact of the Praguean literature has had on it.

506 109.

110.

111. 112.

113.

114. 115. 116. 117.

Rosanna Sornicola Cf. Meillet and Vaillant (1924: 288). Dik (1989: 364) claims that Serbo-Croatian has a P1 CLT P0 V X structure (where CLT is a “strongly defined clause-second position used for a variety of clitics”). This model has, however, been drawn up according to the view that the GFs S and O are nonexistent in Serbo-Croatian; the general substance of this view has been criticized here in Section 1. Cf. Gvozdanovi´c (1981: 127), who reports the following orders as having marked FOCUS: Petar knjigu cˇ ita, Knjigu cˇ ita Petar, Knjigu Petar cˇ ita ‘Peter reads a book.’ On the position of interrogative pronouns, cf. Gvozdanovi´c (1981: 125). Cf. Fougeron (1989: 206–207) and the numerous references reported there. Fougeron’s monograph is basically rooted in the framework of the Praguean School. The concept of “theme” is defined as “known” or “GIVEN”, that of rheme as “NEW”. The limitations of these concepts are pointed out (cf. Fougeron 1989: 4–5), but no alternative formulation is drawn up. A key concept in the work is that of “noyau d’information,” borrowed from Perrot and defined as “l’élément qui est reconnu par nos auditeurs comme le porteur de l’information principale du message” (Fougeron 1989: 108); on the prosodic level, this element always carries the sentential stress. The concept of “noyau d’information” partially overlaps with that of FOCUS, an expression never used in the monograph. It all depends on the textual and the prosodic dimensions: although the author examines in detail the various positions and constituents in which the core of information falls, this is done from an entirely empirical perspective. Amongst the various consequences of this approach, I will merely point out the lack of distinction between all-in-FOCUS structures and structures with FOCUS on a single constituent. The work is, nonetheless, of considerable value for the richness of its experimental and contextual analyses. The theoretical categories with which Yokoyama (1986) has worked seem to be more sophisticated; Yokoyama’s work shows the complexity of the relationship among suprasegmental factors, constituent structure, and semantic and pragmatic factors. Cf. Yokoyama (1986: 198). The author observes furthermore that the neutral, contextfree variants correspond to what she defines as Type I utterances. Nichols (1985: 190– 191) proposes to give up the terminological and conceptual distinction of neutral vs. non-neutral intonation in favor of level vs. nonlevel prosody. Fougeron (1989: 207). For the textual distribution of these structures, see Section 5.2. Fougeron (1989: 305–308). The example is taken from Fougeron (1989: 288). Fougeron’s pragmatic interpretation of sentence (27) does not seem entirely convincing. According to the author, the prosodic characteristics (that is, the lack of the two prosodic indicators of theme: a rise in the fundamental frequency of the stressed syllable and the possibility of a virtual pause between the constituents) allow the structure to be interpreted as entirely rhematic. This analysis, which pays attention mainly to prosodic factors, is furthermore based on a factor to which too much emphasis has perhaps been given: replies given by a group of hearers, who saw a strong cohesiveness in such a structure (cf. Fougeron 1989: 294). The impression of cohesiveness, however, may have nothing to do with the rhematic property being distributed over the entire sentence.

Interaction of syntactic and pragmatic factors 118. 119.

120. 121. 122. 123.

124.

125. 126.

127.

128.

129. 130. 131.

507

The example is taken from Fougeron (1989: 291). Fougeron (1989: 291–292) observes that its role in the message and its prosodic characteristics “permettent de considérer que le composant c’est le thème.” Fougeron notes furthermore that in structures of this sort, the complement of V “introduit un élément nouveau, qui situe l’information contenue dans la séquence SV” (Fougeron 1989: 289). Fougeron (1989: 292). Examples (29) and (30) are taken from Fougeron (1989: 298). The example is taken from Fougeron (1989: 419). Note that in Russian, as in other languages of Europe with high WO flexibility, movement of a constituent from its canonical position does not in itself constitute a precondition for focalization of that constituent, but may result in an “all-in-FOCUS” structure. This property seems to regard V-initial structures in particular: cf. Section 4.2. Cf. Srpová (1990: 408), who in the contrastive analysis of Czech and French WO observes that a WO variation in Czech may correspond to such a variation in French as well, but this often results in prominence either by prosodic means or by means of a grammatical construction of the type ‘It is X who : : :’. Examples (32)–(36) are taken from Fougeron (1989: 427–437). Fougeron herself (1989: 440–441) observes that sentences such as (32)–(36) may encode two types of opposition: a phrasal opposition, in which the constituent which carries the nuclear stress determines a contrast within an adversative construction; or a paradigmatic opposition, for which the element in question determines a contrast with respect to other potential elements. This characteristic seems to coincide with a property mentioned by Fougeron (1989: 362), that it is the position of the stress and not the type of constituent on which it falls that determines the overall melodic pattern. The data and the interpretations offered by Fougeron are not always consistent. This could make the conclusion arrived at here problematic. For example, regarding C S V0 structures (where C stands generically for “complement”), Fougeron observes that C, separated from V (it normally would be postposed) and isolated from the rest of the sentence, is obligatorily “assigned prominence” at the prosodic level (Fougeron 1989: 425). In Fougeron’s terminology, such structures are neither entirely rhematic (that is, all-in-FOCUS sentences) nor are they sentences segmented into unmarked theme and rheme. In any case, it is not clear why in other cases (cf. the example quoted here as (28)) a complement separated from V and isolated from the sentences should be described as “theme” and the whole structure considered divided into theme and rheme. These inconsistencies are probably due to the nature of the conceptual tools with which Fougeron has worked. Yokoyama’s data reported as (25)–(26c) are crucial to this. There exist a few limited structural exceptions as well: see the discussion of example (42) further on. It may be that the difference between Italian and Spanish on the one hand (where an alternative possibility is to mark the moved constituent with an identifying construction) and French on the other lies in the different prosodic properties of these languages.

508 132.

133.

134. 135. 136.

137. 138.

139. 140.

141.

142.

143.

144.

Rosanna Sornicola The interpretation in which the constituent Mario in (42) coincides with a TAIL is not considered here. Out of context, (39) would be ambiguous (that is, it would allow both the interpretation determined by the OVS functional structure as well as the interpretation determined by the SVO functional structure); (40) instead could not in any circumstances ever be interpreted as a SOV structure. As would be the case in LUCIA, ha aiutato Mario; LUCIA, Mario ha aiutato; Ha aiutato MARIO, Lucia. If the fact that the nuclear stress falls on Mario is taken into account, no virtual pause could occur in (42) (Ha aiutato LUCIA, Mario would be possible, but this is another type altogether). On the semantic interpretation of structures with leftward movement of O without a resumptive pronoun in Italian, cf. Berretta (1998). Cf. Sornicola (1988); Weinert and Miller (1996). Parentheses denote the optionality of the pause. On dislocations in Italian, see Cinque (1977, 1990: Ch. 2), Sornicola (1981), Berruto (1986), Benincà (1988), Maslova and Bernini (this volume). The only exception is (43), which allows a contrastive interpretation of ‘Mario’ in a prosodic contour in which there is no pause before Mario. On the problem of how to compute the position of the postcopular constituents in FOCUS in cleft sentences of this kind, see the observations on Celtic languages in Section 2.4.6.2. This, of course, applies to structures containing verbs with two arguments; as to oneargument structures, see Section 3. In Italian an isomorphic type Mario è LUCIA che ha aiutato [lit. ‘Mario it-is LUCIA that he-has helped’] is possible, although rare. The Romance types are also interesting because they add weight to the hypothesis put forward in Section 2.4.6.2 concerning the relationship between the identifying structure and WO in Celtic languages. Note that French has the usual S topicalization possibilities: Son père, il m’a donné le livre. The types in question have been dealt with in different ways by Lambrecht (1981), Barnes (1985), and Blanche-Benveniste (1990). The French syntactic types with isolation of constituents have been dealt with also, among others, by Tesnière (1959: 175), Bossong (1981), Heger (1982), Jacob (1990: 78 ff.). The typological characteristics of French with respect to these types become clear when compared with those of the Slavonic languages: cf. Srpová (1990). Structures such as those found in the Romance languages with postverbal S in FOCUS (cf. (42)) may be rendered in English by assigning the nuclear stress to S placed in typical preverbal position (cf. Daneš 1967). O in P1 may instead sometimes have a “highlighting” value, as in the following example from literary prose, where O is NEW: “This morning the village school opened. I had twenty scholars. But three of the number can read: none can write a cypher. Several knit and a few sew a little. They speak with the broadest accent of the district. At present, they and I have a difficulty in understanding each other’s language. Some of them are unmannered, rough, intractable, as well as ignorant; but others are docile, have a wish to learn, and evince a disposition that pleases me. I must not forget that these coarsely-clad little peasants are of flesh and blood as good as the scions of

Interaction of syntactic and pragmatic factors

145. 146.

147. 148.

149.

150.

151. 152.

509

gentlest genealogy; and that the germs of native excellence, kind feeling, are as likely to exist in their hearts as in those of the best-born. My duty will be to develop these germs: surely I shall find some happiness in discharging that office. Much enjoyment I do not expect in the life opening before me: yet it will, doubtless, if I regulate my mind, and exert my powers as I ought, yield me enough to live on from day to day” (Jane Eyre [ed. Nal Penguin, New York, 1960: 361]). Italian and Spanish may also make use of simple WO variation or clefting in order to encode narrow FOCUS. Note that a cleft construction such as It is MARY that I love does not have the same semantic representation as structure (48): some speakers in fact claim that the cleft structure expresses a stronger contrast than (48). However, cleft constructions in English are more frequent in spoken than in written registers (cf. Breivik 1986: 817). The data gathered for Russian by Holden and Krupp (1987), as well as the data gathered for Serbo-Croatian by Uroševi´c et al. (1986) show this unequivocably. Both Russian and Serbo-Croatian data were obtained from comprehension tests. However, Fougeron’s (1989) data for Russian, drawn from a corpus of spontaneous spoken language, also lead to similar conclusions. For the origin of Diderichsen’s ideas, cf. Heltoft (1986) and in particular Henriksen (1986: 75–76). A summary of Drach’s ideas and a picture of the scientific influences on him may be found in Etzensperger (1979: 28–51). The influence of Ries and also of various structuralist models on both Diderichsen and Drach clearly emerges from the works of Etzensperger and Henriksen (cf. Etzensperger 1979: 26–27; Henriksen 1986: 75–76). Moreover, regarding the idea of a sentence schema comprising empty positions, Etzensperger puts forward the hypothesis that Drach may have been influenced by some of Bühler’s psychological experiments (Etzensperger 1979: 33 and n. 20 for further observations). For Diderichsen, the verb was “the element which indicated the relationship of the conceptual content of the sentence to the reality” (Henriksen 1986: 76), while for Drach “ein geschlossener Satz kommt erst zustande, wenn ein Verbum finitum ihn trägt und seine Vorstellungsinhalte als wirklich und gültig beglaubigt” (cited in Etzensperger 1979: 35). These traditions are well represented in the rich bibliography cited by Behaghel (1932: 1–3, 10–11). This position and the properties which defined it were identified for the old IndoEuropean languages by Wackernagel (1892). However, the unstressed nature of the verb in the main clause had already previously been recognized for Vedic (cf. Delbrück [1888] 1976: 35–36 with bibliography; Delbrück 1900: 41–42, where the main V of independent sentences is ascribed to a category of words that may or may not have stress; furthermore it is assigned an intermediate stress value between typically unstressed elements like particles and fully stressed words). Wackernagel’s idea also had an important impact on the study of WO in the Germanic languages (cf. Wackernagel 1892; Scaglione 1981: 110 ff.). It provided an explanatory hypothesis for the syntactic change from the order with the verb in main clause final position, docu-

510

153. 154.

155. 156.

157.

Rosanna Sornicola mented in the oldest phases of the Indo-European languages, to the order with the verb in medial position, according to many already characteristic of the oldest phases of the Germanic languages (for an overview of the various opinions between the end of the nineteenth century and the first decades of this century, cf. Scaglione 1981: 108–126). Cf. Behaghel cited in Etzensperger (1979: 165), Drach cited in Etzensperger (1979: 35). Note that in some grammatical treatments, such as the recent generative one, the V-2 phenomenon is defined on the basis of declarative sentences and WH-interrogatives: cf. Haider (1986), Holmberg (1986), Platzack (1986). The examples are taken from Platzack (1986: 27). Cf. also Dooley Collberg (1988: 132–133). The German examples are taken from Helbig and Buscha (1989: 564–565). Cf. furthermore Haegeman (1991: 522–523) for Dutch examples, and Pétursson (1978: 138) for Icelandic examples. In Danish and Swedish only some types of adverbs may intervene between Aux and nonfinite V in the structure Aux + nonfinite V in main clauses; in both languages the characteristic structure of main clauses may be represented as: (i)

Initial position Finite V (S) Sentential adverb O/complement/S Other adverbs

Nonfinite V

(cf. Allan et al. 1995: 492; Holmes and Hinchliffe 1994: 503). In German and Dutch instead the characteristic structure with Satzklammer is found (cf. Engel 1991: 304–305; Haegeman 1991: 523). The presence vs. absence of a structure with Satzklammer is therefore a phenomenon which differentiates contemporary Germanic languages. The asymmetry between the position of finite V in main clauses (P2 ) and in subordinate clauses (Pn ) also marks a distinction between the modern Germanic languages: again, German and Dutch have in common alternation between V-2 order in main clauses and orders with finite V in final position in subordinate clauses (cf. Scherpenisse 1986), whereas the Scandinavian and Icelandic languages do not have such alternation (cf. Pétursson 1978: 139; Platzack 1986: 28; Allan et al. 1995: 492–493, 497; Holmes and Hinchliffe 1994: 503, 509). Moreover, Danish and Swedish (but not Icelandic: cf. Sigurðsson 1986) show a different asymmetry between the order of constituents in main clauses and subordinate clauses: in the latter the following structure is found: (ii)

158. 159. 160.

Conj S Sentential adverb Other adverbs

Finite V

Nonfinite V

O/complement

It is hardly worth nothing that the structural similarity between German and Dutch on the one hand and Swedish and Danish on the other is related to historical factors. The example is taken from Haegeman (1991: 522). The example is taken from Platzack (1986: 27). Cf. Pétursson (1978: 138) for contemporary Icelandic examples. For this formula, cf. Haegeman (1991: 522). Cf. Dik (1980: 158) for a representation within the framework of his functionalist model.

Interaction of syntactic and pragmatic factors 161.

162. 163.

164. 165.

166. 167.

168.

169.

511

These are not the only definitions of Mittelfeld, Vorfeld, Nachfeld in German grammatical tradition. For an overview of the history of these notions, see Etzensperger (1979). A concise presentation of Engel’s model, its deviations from other topological models, and the problems it poses can be seen in Etzensperger (1979: 194–203). For an examination of the type of element which may occur in the Vorfeld or in the Mittelfeld, cf. Engel (1991: 310 ff., 320 ff. respectively). Some constituents may not occur in the Vorfeld, for example the Abtönungspartikeln and adverbs, such as sehr ‘very’, außerordentlich ‘extraordinarily’, which regularly occur as adjective modifiers (cf. Engel 1991: 312, 313). The expletive es, whose distribution is confined to the Vorfeld, may not occur in the Mittelfeld (cf. Engel 1991: 309). Cf. Engel (1982: 223). It has been observed that this constraint has an apparent exception. Sentences such as Mit dem Ball ins Gesicht hat er mir geworfen lit. ‘with the ball into the face has he me thrown’, ‘Throw the ball in my face he did’, Gestern am Strand hat er mir ein Geheimnis gesagt lit. ‘Yesterday at the beach has he me a secret said’, ‘Yesterday on the beach, he told me a secret’ seem to have two constituents in P1 . Scherpenisse (1986: 53–55) explains this structure as the effect of two conditions that do not violate the constraint of the single constituent in P1 : (a) the string in P1 must be continuous; (b) it must be dominated by a unique node (which is not necessarily a maximal projection). Such structure is described by Scherpenisse as an “adverb cluster”; it is associated to a characteristic prosodic property, that is, the fall of a heavy stress on the rightmost constituent of the cluster. Other syntactic types with more than one constituent before V are less easy to analyze in that they cannot be reduced to constituents which are dominated by a single node. They are characterized by the occurrence of particular adverbs before V, namely, allerdings ‘certainly’, sogar ‘even’, nur ‘only’, nicht ‘not’: Graf Pocci allerdings hat amüsante Sachen geschrieben ‘Earl Pocci certainly wrote amusing things’, Nicht das wollte ich eigentlich erzählen ‘It was not that I really wanted to tell’ (cf. Tarvainen 1985: 382); however, these two examples seem to be different, as in the first NP and Adv are in fact discontinuous constituents, while in the latter, negation is merely an operator with scope on the constituent in P1 . Cf. Sigurðsson (1989: 5, n. 1) for the exclusion of structures with ja from these more problematic cases. Engel (1991: 316) observes, however, than any “complement” which is a heavy constituent may occur in the Nachfeld. For the constituents which may occur in the Nachfeld, cf. Engel (1991: 316–318). Pronominal complements and modal adverbials have a distribution which excludes the Nachfeld (cf. Engel 1991: 316). This function is ably described by Weinrich (1993: 83). For frequency in the spoken language of structures with circumstantials of time and place, cf. also Engel (1991: 316). For structures which require the Rahmenbildung, cf. Helbig and Buscha (1989: 567). See also the more recent description of German WO patterns in Zifonun et al. (1997: 1495–1680).

512 170. 171.

172.

173. 174. 175. 176.

177. 178.

179. 180.

181. 182.

183.

Rosanna Sornicola Cf. here Section 3. Cf. Haider (1986: 56), who observes that in colloquial German nominative and accusative pronouns may be omitted in A position. Haider shares Huang’s (1984) opinion according to which an empty operator appears in position A in such cases, which implies that V is always in P2 . Cf. also Cardinaletti (1994). An opinion held by researchers in various frameworks is that the development of obligatory pronouns goes hand in hand with the V-2 type (cf. Haiman 1974; Dik 1980: 157), but this view seems to suffer from an excessive teleologism. Note, however, that according to Platzack (1986: 46, n. 2) in Swedish the phenomenon under examination is less common than in Icelandic. Especially in coordinate clauses introduced by und (cf. Behaghel 1932: 30–37). Cf. Platzack (1986: 47, n. 6). Abraham (1986: 16) defines the configurational option as that in which “the underlying word-order as defined by grammatical criteria is subject to linear variation as required by textual organization” and “movement according to textual parameters results in marked linearization.” Cf. Ich gab dem Mädchen ein Buch ‘I gave to the girl [Cdefinite] a book [definite]’, Ich gab das Buch einem Mädchen ‘I gave the book [Cdefinite] to a girl [definite]’. According to Abraham, the PO argument, with basically locative value, “contributes the most to the meaning of the predicate and as such is the verb’s closest argument” (Abraham 1986: 16; cf. also the summary on pp. 29–30). This description is not, however, very convincing: it cannot be upheld with strictly semantic criteria and may be regarded as circular. Cf. Daneš (1967: 226, 228) and the discussion here in Sections 2.4.4.2 and 2.4.4.3. Cf. Karl hat dem Kind das Buch geschenkt, Karl hat dem KIND das Buch geschenkt, Karl hat dem Kind das BUCH geschenkt, Karl hat dem Kind das Buch GESCHENKT (cf. Höhle 1982: 86, with examination of the various pragmatic conditions associated with such sequences). For the distinction between “lexical items” (N, Adj, V, Adv) and “grammatical items” (Art, Prep) with respect to stress, cf. Cruttenden (1986: 82). In any case, in semantic-pragmatic conditions of corrective contrast, constituents such as articles and prepositions may also attract nuclear stress. For the relationship between “normal” order and allocation of nuclear stress, cf. Höhle (1982: 85–86, 91–92), who examines the phenomenon within a textually-oriented approach. Cf. Höhle (1982: 125–126). This is the so-called intonation with “normal stress,” which presents problems for a definition in prosodic terms: cf. Cruttenden (1986: 81, 94–95), who prefers to use the concept of “broad focus.” Abraham (1995: 615) makes a distinction between “grammatikalischer (default) Fokusakzent” and “markierter oder kontrastiver (semantischer) Fokusakzent.” Cf. also Jacobs’ (1988: 120) distinction between “neutrale Akzentuierung” and “normale Akzentuierung.” Cf. Höhle (1982: 85–92), Abraham (1986: 17–18), and Jacobs (1988: 107), who assumes a principle of “FOCUS-last”: “Wenn das MF (= Mittelfeld) genau ein Fi -Satzglied K enthält, so folgt K auf alle anderen Satzglieder.” Counterexamples may be explained by the interaction of other principles (cf. Jacobs 1988: 107–108).

Interaction of syntactic and pragmatic factors 184.

185. 186.

187. 188. 189.

190. 191. 192. 193. 194.

195.

196.

513

Cf. Abraham (1995: 616, 624 n. 58). Semantic and textual factors such as definiteness vs. indefiniteness of the head of NP also play a role in determining the FOCUS. For the intonation pattern with stress on O, see Cruttenden (1986: 148). Note that in the written language, the neutral position of the temporal adverb would be immediately after V: Karl liest heute ein Buch. This situation is found in other languages (for English, cf. Cruttenden 1986: 84; for Italian, cf. Sornicola 1993a), and seems to be connected with a general property of SVO languages. Cf. Engel (1982: 229, 1991: 330, 332). Cf. Engel (1991: 31). On the relationship between constituent movement and stress in German, Engel (1991: 332) observes: “Meist ist die stellungsbedingte Hervorhebung zugleich von stärkerer Betonung begleitet; in solchen Fällen erweist sich die Intonation als Konsequenz der Stellung.” Of course, highlighting may also occur with purely prosodic means, that is, a given constituent may keep its canonical position and convey the nuclear stress. The example is taken from Scherpenisse (1986: 45). On the problem of left dislocation, cf. Altmann (1981, especially chs. 4–7) and Scherpenisse (1986: 55). In neutral intonation, the sentential stress coincides with the verb. Cf. Engel (1991: 316 ff., 332). Cf. Scherpenisse (1986: 64–65), Abraham (1986, 1995: 606–607). Jacobs (1988: 108– 109) has an interesting problematic formulation of the question. Note that when the Mittelfeld contains a PrepP and a DO, the order PrepP + DO (with DO occupying position Xi ) is possible only when DO is [C definite]: cf. Abraham (1986: 18). In sentences with Satzklammer which have non-neutral order, nuclear stress is linked to the immediately preverbal position. This conclusion is reached by Abraham for sentences with Satzklammer which have one of the following WO patterns: (i)

S IO DO V

(ii)

S DO IO V

(iii)

S DO PrepP V

(iv)

S PrepP DO V

The first and third patterns show an unmarked (more frequent) order, while the second and fourth show a marked (more rare) order. Now, precisely the first and third patterns also have a wider range of accentual options than that of the second and fourth patterns. The latter only admit a single option in which the nuclear stress falls on the constituent that is not in situ, which occurs immediately in front of V, as can be seen from examples (71a)–(71d). Being so defined, such a property does not seem to be related to the determination of a characteristic position for FOCUS. For an overview of the problem, cf. Scaglione (1981). One of the most debated questions has been that of the influence of Latin patterns on German WO (a clear and succinct examination can be found in Wells 1987: 205–259). Rather than thinking of an influence of Classical Latin on German WO, it seems more convincing to accept

514

197.

198. 199.

200.

201.

202. 203. 204. 205. 206. 207. 208.

Rosanna Sornicola Burdach’s (1914) claim of an influence of Chancery Latin based on Humanistic Italian models (Fleischmann 1973 opposes this idea). In any case, as far as main clauses are concerned, studies carried out on texts show that in the oldest documents of the Germanic languages the dominant order was already SVO (cf. Braunmüller 1982: 138– 139), which was, however, in competition with other orders. These fluctuations persist throughout the course of many centuries. The sixteenth century texts still exhibit WO structures and, more generally, syntactic structures different from those grammaticalized in the present-day phase (cf. Wells 1987: 253–254). The structure with Satzklammer is already found in Old High German (in Notker), but is not grammaticalized. This situation still prevails in the sixteenth century (cf. Wells 1987: 258). Ebert’s (1980) investigation of the Nuremberg texts has shown the importance of the role of the Chancery: while up to the fifteenth century, chancery texts differ only slightly from those of individual writers of the region with respect to the extent of Satzklammer, from the sixteenth century the divergence becomes greater. Ebert has shown a distinct influence of administrative usage on individual writers in Nuremberg. It is interesting that less educated writers tend to use the frame less frequently. Ebert hypothesizes that the Satzklammer had to a greater or lesser extent caught on in the spoken language of the educated circle comprising the city’s administrators, without necessarily reaching the dialectal or colloquial spoken registers of Nuremberg. For the influence of the press on grammaticalization of the structure with Satzklammer, cf. Wells (1987: 259). The situation with respect to Breton is more controversial, cf. Timm (1989, 1991). For Irish, cf. Stenson (1981: 40–41), McCloskey (1983), Ó Siadhail (1989: 205–206). For Welsh, cf. MacCana (1973, 1991), Williams (1980), Rouveret (1990), Tallerman (1991). See also the contributions in Hendrik (1990) and Borsley and Roberts (1996). For a critique of such models, cf. Anderson and Chung (1977) and the papers in Gazdar et al. (1983), in particular, on the possibility to postulate a VP phrase in VSO languages. On the other hand, recent proposals to eliminate the linear representation in the phrase structure rules of generative grammar may overcome such difficulties. The problem of the VP constituent in VSO languages has been recently re-examined by various contributions in Carnie and Guilfoyle (2000). It is clear that these difficulties do not arise in functional models, in which linearization rules are simply the realization of nonlinear functional representations (cf. Dik 1989, especially 333 ff.). In Irish, no constituent may be placed between V and S (cf. Stenson 1981: 41) except for a small number of parenthetic expressions, such as muise ‘indeed’, mh’anam ‘certainly’ (cf. Ó Siadhail 1989: 217). The example is taken from Ó Siadhail (1989: 205). Cf. the statistics in Tomlin (1986: 17–22). In this sense, Dik’s (1980: 155 ff.) observations are to be treated with caution. Cf. Timm (1989, 1991: 278), Delanoy (1990) for Breton. The example is taken from Stenson (1981: 40–41). Cf. Stenson (1981: 40–41). In fact V is embedded in a relative clause (cf. Stenson 1981: 40–41). In both Irish and Welsh, the relative clause is generally introduced by a particle a

Interaction of syntactic and pragmatic factors

209. 210. 211. 212. 213. 214. 215.

216.

217. 218.

515

(of different origin in the two languages); in the oldest phases of Irish this particle was not present, but the V morphology had special properties: cf. Pedersen ([1909– 1913] 1976: vol. II, 217–223), Lewis and Pedersen (1937: 236–245). The difficulties of analyzing the Celtic structures in question have been pointed out in different ways by various scholars: see, for example, MacCoisdealbha (1998: especially chs. 2 and 3), Shisha-Halevy (1995: 149–156, 2000: 80). Note that the pronoun is optional in type (73). For this structural type, cf. MacCana (1991: 73), Evans ([1964] 1989: 140–141). For Middle Welsh, see also Shisha-Halevy (1995: 153–157). On the problem regarding the structural representation of cleft sentences, cf. Sornicola (1988) and the bibliography included therein. For the Irish example, cf. Ó Siadhail (1989: 305); for the Welsh example, Watkins (1991: 345). This choice, which is not without potential criticism, could be justified by the fact that what counts here is the relative order of V and its arguments. For an examination of this problem, see the discussion in MacCana (1991: 60). For the V-initial type in ancient monuments of Indo-European languages, see Delbrück (1900: 58–61), Marouzeau (1938: 49 ff.), Schwyzer (1950: 689 ff.), Gonda (1952: 26– 33), Dressler (1969), Verpoorten (1977: 38–46), Holland (1980: Ch. I). Gonda (1952: 72 ff.) maintains that V-initial position in Sanskrit has the function of a “mise en relief” as well as an affective value; in other terms, V-initial sentences are – in his opinion – structures with marked FOCUS. On the other hand, Dressler (1969: 3 ff.) considers V-initial position “normal” in most ancient Indo-European languages for sentences in narrative textual progressions. Dressler’s hypothesis allows a consideration of V-initial sentences as a marked pattern in that it cannot be found in all textual environments (see here Section 1.6.1). As to the diachronic explanation of the Irish type, see Watkins (1963), Wagner (1959: 152 ff., 205 ff.) (who believes in the influence of a HamitoSemitic substratum), and Wagner (1967). The V-initial structure of the noncontinental Celtic languages is unanimously considered an innovation (see MacCoisdealbha 1998: 1–2). The two types show structural similarities (in both the “fronted” constituent is followed by the relativized verb) and differences (in the AS the verb of the relative sentence agrees with the head of the “fronted” NP, while this is not the case in MS). Besides, MS is a marked sentence (in fact, its fronted constituent is a marked FOCUS), while AS is not. For a summary of the differences between the two types, see Fife and King (1991: 84–85). Note that MS and AS also show a different diachronic development: the first has always been documented as a marked order, while the latter was the prevailing unmarked structure of Middle Welsh (see MacCana 1991). On the problem of WO in Middle Welsh, see E. Poppe (1989, 1990, 1991a, 1991b: 14–34), Willis (1998: 7 ff., 51 ff.). For examples of constituents with a different GF in the preverbal position, see Willis (1998: 4). Cf. Willis (1998: 168); see also Williams (1980: 4–5) for a discussion of other syntactic differences concerning the form and the distribution of the negative marker.

516 219.

220. 221.

222.

223. 224. 225. 226.

Rosanna Sornicola Cf. Williams (1980: 168), Willis (1998: 4). See also Shisha-Halevy (1995) for a different pragmatic interpretation based on Middle Welsh data. Cf. Watkins (1991: 342), who points out that fronting of sentential adjuncts with the function of setting the scene is a recurring phenomenon in fairytales. Position X of the post-field in the configuration VSOX is indeed typical of adjuncts in both Welsh (cf. Watkins 1991: 334) and in Irish (cf. Stenson 1981: 41–42; Ó Siadhail 1989: 205–206). Most specialists consider the MS to be always emphatic. Contrary to this view, Rowlands (1980: 220) considers that “emphasis is not a matter of degrees that vary with context” and that “MS need not be emphatic at all, though it is grammatically an emphatic structure” (quoted in Fife and King 1991: 87). The phenomenon has been described in Old Irish by MacCoisdealbha (1998: Ch. 4); cf. furthermore MacCana (1973: 100). For modern phases, see Stenson (1981: 47). The example is taken from MacCana (1973: 98). The example is reported by Watkins (1991: 334). For the form mae, cf. Pedersen (1976: vol. 2, 426). The analysis in the literature of the syntactic and pragmatic properties of the types mentioned so far is not very clear. The distinction between the cleft type (76b) and noncleft types (cf. (78)) is defended by Watkins (1991: 332 ff.) for Welsh. In the second group he places both the AS and the MS. According to Watkins, if one leaves aside certain structural differences and differences in the semantic/pragmatic value, the two types may be grouped together by important properties, relating to subordination, negation, and some aspects of agreement (Watkins 1991: 332). Watkins maintains that the most characteristic property of the cleft type is the relativization marker which precedes V (Watkins 1991: 341), even though a multiplicity of factors may make recognition of this problematic: for example, there is a difference between spoken and written registers with respect to the relative pronoun a, which is regularly cancelled in spoken Welsh, present in literary prose and only occasionally cancelled in informal written language (Watkins 1991: 334). Moreover, even when deletion has been ascertained the soft mutation of the initial V segment is unequivocable proof of the underlying presence of the pronoun (Watkins 1991: 334, n. 8): for discussion of further problems, cf. Watkins (1991: 335–341). Similar to Watkins’ proposal is that of MacCana (1991: 62 ff.) who arrives at similar descriptive results, although his examination of the semantic and pragmatic functions of fronting phenomena seems to be more articulated. Contrary to Watkins and MacCana, however, Fife and King claim that there is no single cleft type in Welsh, but that a distinction should be made between AS and MS on the basis of their respective syntactic and distributional properties (Fife and King 1991: 84–85): in short, AS may represent a topicalization structure, in which the fronted constituent has all the characteristic functions of a TOPIC, both sentential (those of adding salience to the TOPIC, of establishing the theme of the sentence, of promoting informative progression) and discoursive (those of giving cohesion to the discourse, of TOPIC shift, and the stylistic function of suspense building devices, anticlimax strategies, etc.; cf. Fife and King 1991: 132). The MS instead is

Interaction of syntactic and pragmatic factors

227. 228. 229. 230.

231. 232. 233.

234. 235.

236.

237.

517

probably structurally and functionally a cleft sentence, its primary pragmatic function being in fact that of contrast. However, the two authors’ hypothesis seems to suffer from two kinds of problems: direct resort to universal pragmatic processes to explain two historically determined types; a none too clear definition of the very pragmatic processes resorted to. If fronting in the MS is associated with a phenomenon of contrastive FOCUS, fronting for topicalization as in the AS refers just as equally to a “focused structure” (cf. Fife and King 1991: 122: “topicality is one sort of focusing device”). The two authors had started out with the observation that “the information structuring concept most relevant to the Welsh data is what we call focus. This term refers to the degree to which an item is made more prominent in the communicative structure of the sentence” and add “one very common form of marked focus is what the literature calls topicalization” (Fife and King 1991: 95). For a criticism to Fife’s views, cf. Shisha-Halevy (1995: 152). The example is taken from MacCana (1973: 106). The example is taken from MacCana (1973: 106). The example is taken from Lewis (1942: 20). The hypothesis that it may be a survivor of the AS had been put forward by Lewis (1942). This idea has been convincingly refuted by Watkins (1991: 331) and by MacCana (1991: 47). For Irish, cf. MacCana (1973: 106–110, 1991: 70–71). Sy’n is the contracted form of sydd ‘who is (are)’+ yn, the particle preceding the verbal noun in the progressive construction. MacCana (1991: 71) speaks of “a common functional element that exists between the several subcategories of relativized V-second sentence,” but does not make the nature of such an element any clearer. This last point has quite rightly been emphasized by Fife and King (1991: in particular 142–143). MacCoisdealbha (1998: 111) had held that topics constitute an intermediate category between proper contrast and simple intraclausal thematic structuration, such as passivization. For a view of TOPIC and FOCUS as a graded scale, cf. Sornicola (1993a). Jongeling (1991: 107) considers VSO the basic WO of Classical Hebrew, while SVO is the order conveying emphasis or contrast. However, Modern Hebrew has SVO as basic order (see Glinert 1989: 413). In fact, the change from VSO to SVO can already be observed in the internal diachrony of Biblical Hebrew: see Givón (1977b). Modern Standard Arabic has more than one dominant order, though V S O can be considered basic (cf. Agius 1991; Ingham 1991 for dialectal varieties). On basic VSO order of Polynesian languages and their competing non-V-initial orders which are due to topicalization processes, cf. Chapin (1978) (for the Easter Island language), Anderson and Chung (1977) (for Samoan and Tongan). Jongeling’s work, which also typologically compares Semitic languages and Welsh, has sound critical observations on Tomlin’s ideas: “The most important objection is that Tomlin does not explain why there are still V S O languages left” (Jongeling 1991: 110, n. 29). Regarding this, see the observations of other authors who have dealt with VSO lan-

518

238. 239. 240.

241. 242.

243.

244.

245. 246.

247. 248. 249.

Rosanna Sornicola guages: cf. Anderson and Chung (1977: 23). Jongeling (1991: 106) and Ingham (1991: 727) are of the same opinion. The morphology of the verb exhibits a first person singular suffix, the etymology of which is a first person singular pronominal clitic. Cf. MacCana (1973: 117). Except for the present and imperfect of bod ‘be’, which are instead preceded by y (i) (at least in literary Welsh): cf. Watkins (1991: 340 and n. 19). On the use of ef (e) as a referential pronoun and as an expletive, cf. Willis (1998: 175). For the forms fe, fo see ibidem. Brockelmann (1913: 171) makes explicit reference to this phenomenon, pointing out that contemporary speakers of Semitic languages are no longer aware of it. In all the Hamito-Semitic languages, with the exception of Egyptian, an extremely archaic stem *ya is attested in the third person masculine singular prefix of the prefix conjugation, having a personal and determinative value (cf. Garbini 1984: 81 and the bibliography ibidem n. 1). Similarly, the prefix t- represents a reduced form of the stem of the second singular personal pronoun: cf. Garbini (1984: 224). Cf. Anderson and Chung (1977: 7–9) for a more detailed discussion. The “clitic placement” rule which they postulate is a typical synchronic rule formulated in terms of descriptive economy: “First, a clitic pronoun copy of a pronominal S is created and positioned before the verb. Then, if the pronoun subject is unemphatic, it is deleted, leaving the clitic copy as the only residue of the original subject.” Such a rule may have no significance for the formulation of diachronic hypotheses. In the Hebrew of Early Biblical poetry, the conjugation with suffixes is already well established as a means of expressing the past; it is the form most used in narrative progressions and usually expresses punctual aspect, connected to an action already completed, while the conjugation with prefixes usually expresses durative aspect, connected to an action which the speaker considers to be incomplete (cf. Sáenz-Badillos 1993: 58–59). In Late Biblical Hebrew this distinction has already been grammaticalized in the verbal system, with the suffixal conjugation used only for the past and the prefixal conjugation only for the future (cf. Sáenz-Badillos 1993: 129). For the extent to which aspectual and pragmatic factors affect the VS order in the diachronic development of Hebrew, cf. also Givón (1977a). See also further on, Daneš (1967: 218–219, 223–224). I refer to the fact that even verbs in passive or pseudo-intransitive constructions may belong to the situation being described. Now, while an intransitive verb is inherently so (the property is represented in the subcategorization frame of the lexeme), a passive, reflexive, or pseudo-intransitive is intransitive by derivation (the property has nothing to do with the subcategorization frame of the lexeme). Cf. for Russian vernulsja otec, lit. ‘is returned father’ ‘father has returned’; Italian se ne è andato il gatto, lit. ‘Refl.Prn.2sg from there is gone the cat’, ‘the cat has left’. For an examination of this typological resemblance between the Slavonic and Romance languages, cf. Cennamo (1993). The literature is vast and I will limit myself here to reference to Ulrich (1985: ch. 1); for an examination of the history of the concept “thetic” in the German linguistic

Interaction of syntactic and pragmatic factors

250.

251.

252. 253. 254.

255. 256. 257.

258. 259. 260.

261. 262. 263. 264.

519

tradition, cf. also Haberland (1993) and Sasse (1987). For the concept of “presentative”, cf. Hetzron (1971, 1975). There are numerous papers dedicated to examination of the structures in questions: cf. Lambrecht (1994: 39–43, 177–181) with particular reference to the Romance languages and English; Bernini (1995) for Italian; Abraham (1995: 626–627) for German; Erdmann (1990) for English; Kirsner (1979) for Dutch; Fougeron (1989: ch. 5 and 6) for Russian. The literature on this structure is also vast. For a general view of the problem, cf. Lyons (1977: 723) and especially Sasse (1987, 1995a). See Euler (1982) for a comparative analysis of this structure in ancient Indo-European languages; for modern Indo-European languages, see Ulrich (1985: 82–114). Interesting observations on the function of this structure in spoken Italian can be found in Berruto (1986) and Bernini (1991, 1995). In many Indo-European languages, locative-existential structures may be used as temporal circumstantials that function as introducers of referents in narrative texts, for example fables: cf. Yokoyama (1986: 228–229) for Russian. It is not unusual for presentative and locative-existential structures to be interchangeable: cf. Gadet (1992: 71, 76–77), Lambrecht (1994: 169–170). Of course a separate discussion is required for languages in which the structural elements are ((Adv loc), ‘have’, O), such as in the Spanish structures hay + NP and French structures il y a + NP, where O appears in postverbal position. The example is quoted from Yokoyama (1986: 282). The idea of different types of intransitive verbs was originally presented by Perlmutter (1978). Verbs like arrivare require ‘be’-auxiliary and their single NP can undergo ne-cliticization, while verbs like parlare require ‘have’-auxiliary and their single NP cannot undergo ne-cliticization (cf. Burzio 1986: 16 ff.; and more recently, Graffi 1994: 235– 237, 275–277). For the terminological problems, see Haegeman (1991: 298–312) and Graffi (1994: 236 ff.). For a discussion of this problem, cf. Sornicola (1994). Such a probability applies in spoken rather than written texts: cf. Sornicola (1995b) for an examination relating to Italian and Spanish. In some verbs in Italian, such as bastare ‘to suffice’, servire ‘to serve’, interessare ‘to interest’, etc., the NP which controls agreement occurs consistently in O position: cf. Sornicola (1990). Cf. Sornicola (1994: 36–37). Cf. Comrie (1981b: Sect. 6.2). Morphological incorporation of the nominal root with the feature [animate] to the verbal root in functionally eventive structures is a phenomenon which is found in some native American languages (cf. Sasse 1987). Cf. Gonda (1952: 73) and the bibliography quoted therein. Cf. Fougeron (1989: 260, 305 ff.). Note that in such a case, the verb is always in the past tense. Cf. W. K. Matthews (1960: 51) and Yokoyama (1986: 284 ff.). See also here Section 4.1. Cf. Hakulinen (1961: 318–319).

520 265. 266.

267.

268.

269. 270.

271. 272.

273. 274. 275. 276. 277.

278. 279. 280.

281. 282.

Rosanna Sornicola Note, however, that structures with one-argument verbs are in the majority in MacCana’s (1973) data. Cf. Fougeron (1989: 106–140), from where all Russian examples quoted here are taken. The apostrophe indicates that the constituent carries nuclear stress. The VS pattern has a characteristic intonational profile: basic stress is strongest when it falls on the S constituent (cf. Fougeron 1989: 150). Furthermore “le palier mélodique, situé dans la zone des fréquences élevées et accompagné d’une quasi-stabilité de l’intensité (relativement forte), crée une sorte de ‘retardement de resolution’ et maintient une certaine tension” (Fougeron 1989: 151). An interesting subclass of eventive structures characteristically occurring with the VS pattern is that of “meteorological verbs”; for further examples from other languages, see here Section 3.2.1.1. Cf. Fougeron (1989: 138). Fougeron observes that the combination of the syntactic pattern and the prosodic pattern contribute to the informative balance of the two constituents, that is, neither prevails in the structure of the sentence. In these styles the variants V0 / S, with S in TAIL position (afterthought), are allowed. In this context, a structure with the same WO but with neutral prosodic contour, that is, Papà è tornato, would also be possible. In any case, as far as the comparison between Russian and Italian is concerned, it should be observed that all the structures described by Fougeron are from her corpus of spoken language, which doesn’t exclude the fact that in Russian, there may also be oscillation between SV and VS. Yokoyama’s data, in fact, give reason to think that such an oscillation exists (cf. Yokoyama 1986: 184–194). Cf. Lerch (1934: 258). In the Germanic languages, VS is allowed only if V does not occupy position P1 (cf. Section 2.4.5). But cf. Section 3.6.3 for examples of VS order with intransitive V in P1 in colloquial and dialectal Modern German. The structures belonging to this group have been described by Bossong (1984, 1987) under the label of “paradigmatic inversion.” The example is taken from a fable text, reported by Yokoyama (1986: 285–286). It is a narrative sequence. The example is taken from Yokoyama (1986: 184). Cf. Sornicola (1995b). The example is taken from the French translation of H. F. Ellenberger’s, History of the Discovery of Unconscious (Ellenberger 1994: 187). There is often described as the “grammatical” S in present-day grammars. Obviously, however, the whole structure has a split of syntactic properties of subjecthood between there and the postverbal NP. The examples are from Quirk et al. (1985: 1408). Quirk et al. (1985: 1408). For Estonian, cf. Tauli (1983: 159), who observes that when the structure comprises the constituents S, V, and Adv, “all possible combinations occur, but with different stress and frequency.” Cf. Karlsson (1987: 77 ff.). Examples (38) and (39) are from Fromm (1982: 143).

Interaction of syntactic and pragmatic factors 283.

284.

285. 286. 287. 288. 289. 290. 291. 292. 293. 294. 295. 296. 297. 298.

299. 300. 301.

302. 303. 304.

521

This example is from Hakulinen (1961: 320). These types of inversion are probably inherited in Finnish (the example with the verb of saying is also found in dialectal varieties) and probably differ from the type of inversion due to interference from Swedish (cf. Hakulinen 1961: 318–319). As to the stress pattern, nuclear stress is on S (cf. Hakulinen 1961: 319). On the problem of patterns of order in existential sentences, cf. also Vähämäki (1984: 446), who claims that statistically existential sentences in Finnish have the pattern XVS. Vilkuna (1989: 10) records cases of V-initial with intransitive verbs of movement in the translation of Russian fables and considers them to be exceptional, although possible in particular styles. Both examples are taken from Fife and King (1991: 145–146). Cf. also MacCana (1991: 67) who claims that SV order appears in the presence of “expressive force” distributed over the entire sentence. Cf. Watkins (1991: 341); the locative adjunct occurs in P1 in the structure. Cf. Ó Siadhail (1989: 217). All the Basque examples are taken from Lafitte (1962: 47–48). Cf. Erguvanli (1984: 44–45). Cf. Erguvanli (1984: 47). Cf. Erguvanli (1984: 17). Cf. Erguvanli (1984: 17). Examples (140)–(143) are taken from Erguvanli (1984: 15–16). Cf. Erguvanli (1984: 6 and 69, n. 3). Cf. Erguvanli (1984: 44, 68–69). Erguvanli (1984: 129). Cf. Erguvanli (1984: 130), with an interesting discussion on the relationship between stress and definiteness. Cf. also Sasse (1995b) and Sasse (this volume). Cf. Behrens (1989: 117), who does not include these verbs in the same group as the preceding ones, since she claims that SV/VS oscillation in this case is essentially due to textual factors of dialogue progression. In VS order S is, however, always rhematic. Furthermore, Behrens observes that in cataphoric uses of S in modern Hungarian, S goes in front of V, while it is postposed to V when preceded by quotation of a direct discourse. Cf. Behrens’ (1989: 116–142) examination. In Behrens’ data, the percentage of structures with focalization (of the entire sentence or one of its constituents) varies according to the period and to texts. Cf. Behrens (1989: 116). However, in Old Hungarian S [definite] and [referential] could also occur in postverbal position. The picture is further complicated by the fact that in these old diachronic phases, preverbal S [referential] always occurs in the characteristic immediately preverbal position. This is very clear from the fact that, in these languages, it is equally possible to identify a “dominant” order: cf. the discussion on Russian in Section 2.4.3.2. Cf. Sornicola et al. (1994). This work, however, does not present a wide statistical base like Uhlíˇrová’s. The choice of the auxiliary ‘to be’ or ‘to have’ is a parameter which may undergo

522

305. 306. 307. 308. 309.

310.

311.

312.

313. 314.

315.

316.

Rosanna Sornicola considerable alteration in the diachronic development of a language (for English, cf. Rydén and Brorström 1987); it is, furthermore, too closely linked to specific characteristics of individual groups of languages (languages such as Chinese which do not have auxiliaries of this type come to mind). Even more restricted typologically is the diffusion of structures with ne. Cf. Keenan (1976); Plank (1984); and the papers in Plank (ed.) (1984). The situation of VSO languages is, in a way, uninteresting in this respect: see further on in this chapter. Albeit with differences between languages with more flexible orders and languages with more rigid orders, as has been said in Section 2. Maan is a Thuringian form for Mann. Examples (149) and (150) are from Behaghel (1932: 38), who quotes (149) as possible in his own speech. Interestingly, he also quotes examples with VS order for structures with a transitive verb, from both colloquial and dialectal German varieties. The two structures may convey a value of surprise (I owe this interpretation to Erich Poppe; his judgement is that they are both non-neutral sentences). Example (151) is a structure described by Russ (1989: 258) for the dialect of Palatinate. Cf. Hakulinen (1961: 319–320). The inversion due to the Swedish influence is structurally similar to that of Modern Standard Germanic languages, that is, it respects the V-2 constraint. This is presumably a universal tendency, since in many of the world’s languages unmarked narrow FOCUS tends to coincide with a referential constituent. This seems to be true not only of FOCUS as a PF, but also of its prosodic correlates (see Ladd 1996: 188–189). However, as far as the latter are concerned, it does seem that determining the canonical allocation of FOCUS as stress is more complex: see Cruttenden (1986: 147–150). Ladd (1996: 167 ff.) has an interesting discussion on the independence of prosodic structure from the structure of PFs and on its typological variation across languages. Proof of this is supplied by numerous data: cf. Hakulinen (1961: 315) for Finnish, Kenesei (1986: 144) for Hungarian, and Erguvanli (1984: 60) for Turkish. For Italian, see Section 2.4.4.1.1. Cf. Vilkuna (1989: 10). Similar considerations apply to Italian (cf. the percentages on initial V reported in Section 2.4.4.1.1). Behrens’ (1989: 118–142) data on VS structures in Hungarian of the fifteenth and nineteenth centuries may also serve as indirect proof of this. VS structures with V in FOCUS form less than 10 % of the total VS structures of the corpus. These data are to be treated with caution since they concern the relative order of S and V, not the absolute position of V in the sentence. More rarely, in certain contexts S in the post-field may have a value other than that of adding information, a similar situation to that described by Erguvanli (1984: 60–62) for Turkish. Regarding these structures, as well as those described in Section 2.4.4.1.1, it should be pointed out that structures with pronominal objects proclitic to V are preferable to these; that is, respectively Lo AMA Mario / Lucia and Lo AMA / Lucia / Mario.

Interaction of syntactic and pragmatic factors 317. 318.

319. 320.

321. 322. 323. 324. 325. 326. 327.

328. 329. 330. 331. 332. 333. 334.

335.

336. 337. 338. 339. 340. 341.

523

This function can be seen in Russian since the earliest documentation (cf. Gonda 1952: 73; W. K. Matthews 1960: 51). The fact that speakers may perceive prosodic prominence on constituents which is not confirmed by experimental analysis is a well-known phenomenon in experimental phonetics. Cf. Erguvanli (1984: 61), who observes that the unmarked order would be I¸ste ben bu i¸sin böyle sıkı¸sıklı˘ga gelece˘gini bildim. The example is taken from Watkins (1991: 345). The neutral (unmarked) structure would be Roeddwn i yn edrych ar y gannwyll ‘I was looking at the candle’ (I owe this example to Erich Poppe, pers. comm.). Watkins (1991: 344) observes that the type is the usual option in colloquial Welsh for explanation of the action carried out. The example and the description of the structural and pragmatic characteristics of this type are from MacCana (1973: 110). According to a sample of native Russian speakers, it would be impossible to introduce a pause between the constituents (cf. Fougeron 1989: 310). The example is taken from Yokoyama (1986: 272). Yokoyama (1986: 273). For a study of the influence of the feature animacy on WO in Russian, cf. Maslova (1995). Erguvanli (1984: 44) criticizes the structural conditions set down by Hankamer (1971) for the extraposition rule which should generate structures with the verb in nonfinal position in Turkish. Cf. Erguvanli (1984: 33–34). On the definition of “definite,” cf. Erguvanli (1984: 17–18). Erguvanli (1984: 33). Cf. Erguvanli (1984: 15–16). Cf. Erguvanli (1984: 28–29). The prosodic properties of the second sentence are also interesting: after bu sürpriz there must be a pause (intonation break): cf. Erguvanli (1984: 29). Whether this may be considered a real phenomenon of incorporation in Turkish has been discussed but seems to be highly controversial: for a summary of the arguments for and against this idea, cf. Erguvanli (1984: 24–26). See the examples reported on p. 22 and also the discussion on p. 28 (Erguvanli 1984). On the whole question of the effect of semantic features on word order in Turkish, cf. also Erguvanli-Taylan (1987) with discussion of additional data. Cf. Kenesei (1986: 145). Behrens (1989: 142). Cf. Behrens (1989: 142). Behrens (1989: 14). Behrens (1989: 142). I shall limit myself to mention of the Praguean tradition (cf. Firbas 1992; Sornicola and Svoboda 1992) and the European and American traditions influenced by it: cf. Dik (1989); Chafe (1994).

524 342. 343.

344. 345.

346.

347. 348. 349. 350. 351.

352.

Rosanna Sornicola These concepts are taken from Fougeron’s (1989: 300–301) study of Russian. Fougeron provides data from which it can be inferred that the final or nonfinal position of the constituent bearing the nuclear stress (which, note, may have any GF) is decisive in the orientation of the sentence, respectively rightwards or leftwards (cf. Fougeron 1989: 300–301, 362–367). Fougeron also observes that sentences such as Juru Serëža cˇ itaet are linked to the left context. The description of these phenomena in Russian given by Yokoyama (1986: especially 190–191) is different. The author, in fact, establishes a relationship between the linear parameters of WO, those concerned with stress, and the informational parameters GIVEN/NEW. The distribution GIVEN/NEW is in effect related to the type of intonation: with intonation type I, without sentence stress, there is a progression from GIVEN to NEW (cf. example (26a) reported in Section 2.4.3.2), while with intonation type II, in which there is sentence stress, there may be a progression from NEW to GIVEN (cf. example (26b) in Section 2.4.3.2). The situation in which the NEW constituent occurs between two GIVEN constituents with intonation type II (cf. example (26c) in Section 2.4.3.2) is worth mentioning. Such a case would be more problematic in the Romance and Germanic languages. In Figure 21, the dotted line represents the right-hand context of the sentence. In Figure 22, the dotted line represents the left-hand context of the sentence. This is the so-called “inverse” or “artificial” order variously examined in the functionalist tradition. The inverse order is always accompanied by nuclear stress falling on the NEW constituent in P1 : for Russian, cf. Yokoyama (1986: 190–191). Structures with full NPs (which would, of course, be left dislocations of O) such as French Le café, c’est maman qui le veut and Italian Il caffè lo vuole la mamma would not be impossible in spontaneous speech with its redundant features; however, with respect to the context quoted for the Russian example sentences (171)–(172) are far more acceptable. The same property is encountered in Basque (cf. Lafitte [1944] 1978: 46–48) and in Cheremis (Mari) (cf. Lewy 1922: 168). For a discussion of the change of order from Latin to Romance languages, cf. Sornicola (1995a, 2000). Leaving aside, for simplicity, situations such as those in the Germanic languages in which the sentence is closed by a particle. Due to their greater freedom of constituent order, Finnish and Russian show certain peculiarities in this respect, which have been discussed in Section 2.4.3. The weak leftward delimitation of the sentence domain in these languages is related to the fact that no true dislocation structures (i.e., structures with the dislocated constituent anaphorically resumed in the “body” of the sentence) may occur. Thus, it is perhaps possible to envisage an “open” preverbal field. Note that such a representation has been claimed for similar S/O initial structures in Semitic languages such as Arabic (cf. Holes 1995: 203 ff.) and Hebrew (cf. Jongeling 1991: 107 ff.). However, the difficulties discussed in Section 2.4.6.1 must be borne in mind about the determination of position P1 in VSO languages. It would, in fact, be quite reasonable to believe that the constituents with the GF S/O that occur before V occupy not P1 but rather an extrasentential position.

Interaction of syntactic and pragmatic factors 353.

354. 355. 356. 357. 358. 359.

360.

361. 362.

363.

364.

365.

366.

367. 368. 369.

525

This generalization does not apply to the Celtic languages, although – as has been seen – even in these languages topicalization processes may exist which place S or O in P1 in neutral sentences. In addition to the literature quoted in Section 1.6.2, cf. also the interesting criticism by Marantz (1984) within the generative framework. Cf. Abraham (1995: 606–607). This, of course, does not apply to Russian which, as was said in Section 2.4.3.2, does not have a unique position for FOCUS in neutral sentences. Cf. CLG (1916: 172). Cf. Sornicola (1986), Cinque (1993), Ladd (1996). Cf. Watters (1979), Horvath (1986: 124 ff.). Note that Aghem is described as a language with rigid constituent order, which leads one to think that such a parameter contributes to determining the property under examination. Although Basque and Turkish are not languages with rigid constituent order, like many agglutinating languages, however, they present phenomena with relative rigidity of constituents. The structure under examination has, in fact, very rare counterparts in SVO languages. The situation in Aghem, an SVO language with FOCUS fixed in the immediately postverbal position, appears to be rather peculiar. For a new start of these old trends of research, see, for example, Gruber (1967) and the contributions in Li (1976). For Korean, cf. Sohn (1994: 86, 209–210), who notes, however, that the “complement must always immediately precede the verb.” For Kannada, cf. Sridhar (1990: 137). The case of Japanese and Quechua will be examined in more detail in Section 6.2.4.1.2. If the description given by Tiffou and Pesot (1989: 70–71) can be relied upon – and it is not at all clear that it can be – the two authors report examples of the constituent with the GF ERG in sentence-final position with the value of FOCUS. Cf. Sridhar (1990: 138) for Kannada. For Somali, examples (10c) and (10f) from Svolacchia et al. (1995) show the same property. Sohn’s (1994: 195) description of Korean, while on the one hand confirming that the postverbal position is reserved for afterthoughts, on the other hand shows that such a position may also be the locus of “emphatic” constituents; the relevant examples, however, seem dubious and the same reservations may be had as with respect to certain descriptions of Japanese (cf. below). Cf. Sohn (1994: 232) for Korean; he describes a multiplicity of possible positions for FOCUS in the proper sentence domain, a situation similar to that in Japanese. For Burushaski, cf. Tiffou and Pesot (1989: 70–71). Cf. Sridhar (1990: 138–139). He notes, moreover, that “the initial placement of the object constituent is only for discourse continuity (old information) and not for emphasis” (Sridhar 1990: 138). However, the examples quoted by him seem to point at a scenario in which position P1 could be filled by an IO with FOCUS function (cf. ex. (510)), but not by an O with such PF. Cf. Hinds (1986: 148–150). Cf. Hinds’ (1986: 150, 166) examples. Calvo Pérez denotes with “0 ” the TOPIC and with “00 ” the FOCUS (in his terminology, “el validador”).

526 370. 371.

372. 373. 374. 375.

376. 377. 378. 379. 380.

Rosanna Sornicola Cf. Calvo Pérez (1993: 41). Weber quite rightly distinguishes between the concepts of NEW or rhematic element and FOCUS. He observes that in some studies on other varieties of Quechua, as opposed to that which he investigated, the particles -mi and -shi have been identified as markers of the FOCUS element; according to him this is not always true in Huallaga Quechua (Weber 1989: 427). Cf. Weber (1989: 408–409, 419). Weber (1989: 429). Cf. also Muysken (1995: 385), who claims that Weber’s formula expresses more than a frequency pattern. The problem has already been mentioned in Section 2.2 (see also the bibliography cited therein and Poppe [1954: 173] for Mongolian). In Japanese, the interrogative element generally occupies the same position in the sentence as it does in the corresponding declarative (with the only exception of interrogative sentences that have undergone a process of clefting): cf. Hinds (1986: 39–40). In Quechua, the interrogative element is usually (but not always) fronted to sentenceinitial position (cf. Weber 1989: 19). In Somali, as far as may be deduced from Svolacchia et al.’s (1995: 73–75) examples, it is in one of the positions in the preverbal space. Cf. Sridhar (1990: 8). Cf. Weber (1989: 327–328). For an examination of this situation and the few exceptions to it due to models borrowed from Persian, cf. Erguvanli (1984: 72–117). Matthews (1991: 216), who cites the example reported here, has called attention to the significance of such a process in a typological framework. For Altaic languages, cf. Comrie (1981a: 77), who moreover observes that within this family, the validity of the parameter under examination has been verified in detail only for certain Turkic languages. For Indo-Aryan languages, cf. Section 2.2 and the references therein.

References Abraham, Werner 1986 Word-order in the middle field of the German sentence. In Topic, Focus and Configurationality, Werner Abraham and Sjaak de Meij (eds.), 15–38. Amsterdam and Philadelphia: John Benjamins. 1995

Deutsche Syntax im Sprachvergleich. Grundlegung einer typologischen Syntax des Deutschen. Tübingen: Narr.

Adamec, Pˇremysl 1966 Porjadok slov v sovremennom russkom jazyke. Prague: Academia. Adams, Jim N. 1976

A typological approach to Latin word-order. Indogermanische Forschungen 81: 70–99.

Agius, Dionisius A. 1991 Precedence of VOS over VSO in Modern Standard Arabic. In Semitic Studies in Honour of W. Leslau on the Occasion of his 85th Birthday, Vol. 1, Alan S. Kaye (ed.), 39–55. Wiesbaden: Harrassowitz.

Interaction of syntactic and pragmatic factors

527

Allan, Robin, Philip Holmes, and Thomas Lundskaer-Nielsen 1995 Danish: A Comprehensive Grammar. London: Routledge. Altmann, Hans 1981 Formen der “Herausstellung” im Deutschen. Tübingen: Niemeyer. Anderson, Stephen R., and Sandra Chung 1977 On grammatical relations and clause structure in verb-initial languages. In Grammatical Relations, (Syntax and Semantics 8), Peter Cole and Jerrold M. Sadock (eds.), 1–25. New York: Academic Press. Aronson, Howard I. 1990 Georgian. A Reading Grammar. Columbus, OH: Slavica Publishers. Austerlitz, Robert 1970 Agglutination in Northern Eurasia in perspective. In Studies in General and Oriental Linguistics Presented to Shirô Hattori on the Occasion of his Sixtieth Birthday, Roman Jakobson and Shigeo Kawamoto (eds.), 1–5. Tokyo: TEC. AWL = Moseley, Christopher, and R. E. Asher (eds.) 1994 Atlas of the World’s Languages. London: Routledge. Babby, Leonard H. 1978 Word-order, case and grammatical relations in Russian. Folia Slavica 2: 35–45. Banfi, Emanuele 1998 Tra pragmatica e sintassi: alcuni aspetti della frase interrogativa nel turco (di Turchia). In Atti del XXX Congresso Internazionale della Società di Linguistica Italiana (Pavia 26–28 settembre 1996), Paolo Ramat and Elisa Roma (eds.), 385–395. Rome: Bulzoni. Barnes, Betsy K. 1985 The Pragmatics of Left Detachment in Spoken Standard French. Amsterdam: John Benjamins. Beaulieux, Léon 1933 Grammaire de la langue bulgare. Paris: Champion. Behaghel, Otto 1932 Deutsche Syntax, Vol. IV: Wortstellung, Periodenbau. Heidelberg: Winters. Behrens, Leila 1989 Textanalyse und Sprachwandel. Historische Syntax des Ungarischen: ein empirisches Modell. Frankfurt a. M.: Peter Lang. Bell, Alan 1978 Language samples. In Universals of Human Language, Vol. I: Method and Theory, Joseph H. Greenberg, Charles A. Ferguson, and Edith Moravcsik (eds.), 123–156. Stanford, CA: Stanford University Press. Belletti, Adriana 1988 The case of unaccusatives. Linguistic Inquiry 19 (1): 1–34. Benincà, Paola 1988 L’ordine degli elementi nella frase e le costruzioni marcate. In Grande grammatica italiana di consultazione, Vol. I: La frase. I sintagmi nominale e preposizionale, Lorenzo Renzi (ed.), 115–225. Bologna: Il Mulino. Benoist, Jean-Pierre 1979 Les fonctions de l’ordre des mots en russe moderne. Romans et nouvelles de Gorki. Paris: Institut d’Études Slaves. 1988 Remarques sur l’ordre des mots “grammatical” en russe moderne. Place du determinant accordé. Étude diachronique et synchronique. Revue d’Études Slaves 60 (3): 649–656. Bernini, Giuliano 1991 Frasi relative nel parlato colloquiale. In La lingua degli studenti universitari, Cristina Lavinio and Alberto A. Sobrero (eds.), 165–187. Florence: La Nuova Italia.

528 1995

Rosanna Sornicola Verb-subject order in Italian: an investigation of short announcements and telecast news. Sprachtypologie und Universalienforschung 48 (1/2): 44–71.

Berretta, Monica 1998 Valori pragmatici diversi dell’ordine O V (O V S / O S V) nell’italiano contemporaneo. In Atti del XXI Congresso Internazionale di Linguistica e Filologia Romanza (Palermo, 18–24 settembre 1995), Vol. 2, Giovanni Ruffino (ed.), 81–90. Tübingen: Niemeyer. Berruto, Gaetano 1986 Le dislocazioni a destra in italiano. In Tema-rema in italiano, Harro Stammerjohann (ed.), 55–69. Tübingen: Narr. Bese, L., L. Dezs˝o, and J. Gulya 1970 On the syntactic typology of the Uralic and Altaic languages. In Theoretical Problems of Typology and the Northern Eurasian Languages, L. Dezs˝o and P. Haidú (eds.), 113– 128. Amsterdam: Grüner. Bhatt, Rakesh Mohan 1994 Structure of the Kashmiri clause. In Theoretical Perspectives on Word-order in South Asian Languages, Miriam Butt, Tracy Holloway King, and Gillian Ramchand (eds.), 31–65. Stanford, CA: Center for the Study of Language and Information. Blanche-Benveniste, Claire 1990 Le français parlé. Études grammaticales. Paris: Éditions du CNRS. Blatt, Franz 1957 Latin influence on European syntax. In The Classical Pattern of Modern Western Civilization: Language, (Acta Congressus Madvigiani, Travaux du Cercle Linguistique de Copenhagen 11), Franz Blatt (ed.), 33–69. Copenhagen: E. Munksgaard. Boeder, Winfried 1989 Verbal person marking, noun phrase and word-order in Georgian. In Configurationality. The Typology of Asymmetries, László Marácz and Pieter Muysken (eds.), 159–184. Dordrecht: Foris. Borsley, Robert D., and Ian Roberts (eds.) 1996 The Syntax of the Celtic Languages. Cambridge: Cambridge University Press. Bossong, Georg 1981 Séquence et visée. L’expression positionnelle du thème et du rhème en français parlé. Folia Linguistica 15: 237–252. 1984 Diachronie und Pragmatik der spanischen Wortstellung. Zeitschrift für romanische Philologie 100: 92–111. 1987 Wortstellung, Satzperspektive und Textkonstitution im Iberoromanischen, dargestellt am Beispiel eines Textes von Juan Rulfo. Iberoromania 19: 1–16. 1990 Sprachwissenschaft und Sprachphilosophie in der Romania. Tübingen: Narr. Braunmüller, Kurt 1982 Syntaxtypologische Studien zum Germanischen. Tübingen: Narr. Breivik, Leiv E. 1986 Some remarks on cleft sentences in present-day English. In Linguistics across Historical and Geographical Boundaries. In Honour of Jacek Fisiak, Vol. 2, Dieter Kastovsky and Aleksander Szwedek (eds.), 815–826. Berlin and New York: Mouton de Gruyter. Brockelmann, Carl 1913 Grundriss der vergleichenden Grammatik der semitischen Sprachen, Bd. 2: Syntax, Berlin: Verlag von Reuther & Reichard. Brody, Jill 1984 Some problems with the concept of basic word-order. Linguistics 22: 711–736. Burdach, Konrad 1914 Über deutsche Erziehung. Zeitschrift für den deutschen Unterricht 28: 657–678.

Interaction of syntactic and pragmatic factors

529

Burzio, Luigi 1986 Italian Syntax. Dordrecht: Foris. Calvo Pérez, Julio 1993 Pragmática y gramática del Quechua Cuzqueño. (Monumenta Lingüística Andina 2). Cuzco: Centro de Estudios Regionales Andinos Bartolomé de Las Casas. Cardinaletti, Anna 1994 La sintassi dei pronomi. Uno studio comparativo delle lingue germaniche e delle lingue romanze. Bologna: il Mulino. Carnie, Andrew and Guilfoyle, Eithne (eds.) 2000 The Syntax of Verb Initial Languages. Oxford: Oxford University Press. Cennamo, Michela 1993 The Reanalysis of Reflexives: A Diachronic Perspective. Naples: Liguori. 1995 Transitivity and V S order in Italian reflexives. Sprachtypologie und Universalienforschung 48 (1/2): 84–105. Chafe, Wallace 1986 Cognitive constraints on information flow. In Coherence and Grounding in Discourse, Russel S. Tomlin (ed.), 21–51. Amsterdam: John Benjamins. 1994 Discourse, Consciousness and Time. Chicago: Chicago University Press. Chapin, Paul G. 1978 Easter Island: a characteristic V S O language. In Syntactic Typology, Winfred P. Lehmann (ed.), 139–168. Austin, TX: University of Texas Press. Cinque, Guglielmo 1977 The movement nature of left-dislocation. Linguistic Inquiry 8: 397–411. ¯ 1990 Types of A-Dependencies. Cambridge, MA: MIT Press. 1993 A null theory of phrase and compound stress. Linguistic Inquiry 24: 239–297. CLG = Saussure, Ferdinand de 1916 Cours de linguistique générale. Charles Bally and Alfred Sechehaye (eds.). Paris: Payot. Como, Paola 1995 Fumetto, giallo e romanzo d’autore: un’indagine sull’ordine delle parole. Unpublished M.A. thesis in sociolinguistics, Università di Napoli “Federico II”. Comrie, Bernard 1981a The Languages of the Soviet Union. Cambridge: Cambridge University Press. 1981b Language Universals and Linguistic Typology. Syntax and Morphology. Oxford: Blackwell. 1984 Some formal properties of focus in Modern Eastern Armenian. Annual of Armenian Linguistics 5: 1–21. 1996 Sprache und Sprachen: Universalien und Typologie. In Deutsch-typologisch, Ewald Land and Gisela Zifonun (eds.), 16–30. Berlin and New York: Mouton de Gruyter. Connolly, J. H. 1991 Constituent Order in Functional Grammar: Synchronic and Diachronic Perspectives. Berlin and New York: Foris. Croft, William 1990 Typology and Universals. Cambridge: Cambridge University Press. Cruttenden, Alan 1986 Intonation. Cambridge: Cambridge University Press. Currie, Oliver 2000 Word order stability and change from a sociolinguistic perspective: the case of Early Modern Welsh. In Stability, Variation and Change of Word Order Patterns over Time, Rosanna Sornicola, Erich Poppe, and Ariel Shisha-Halevy (eds.), 203–230. Amsterdam: John Benjamins.

530

Rosanna Sornicola

Daneš, František 1967 Order of elements and sentence intonation. In Intonation, Dwight Bolinger (ed.), 216– 232. Harmondsworth: Penguin. de Groot, Casper 1989 Predicate Structure in a Functional Grammar of Hungarian. Dordrecht: Foris. Delanoy, Arnaud 1990 Breton and the word-order concept. Word 42 (2): 219–221. Delbrück, Berthold 1878 Die Altindische Wortfolge aus dem Çatapathabr¯ahmana. Halle: Verlag der Buchhand˙ lung des Waisenhauses. 1900 Vergleichende Syntax der indogermanischen Sprachen. Vol. III. Strassburg: Trübner. 1976 Altindische Syntax. Darmstadt: Wissenschaftliche Buchgesellschaft. Original edition Halle 1988. de Meij, Sjaak, and László K. Marácz 1986 On question sentences in Hungarian. In Topic, Focus and Configurationality, Werner Abraham and Sjaak de Meij (eds.), 253–276. Amsterdam and Philadelphia: John Benjamins. de Rijk, R. P. G. 1978 Topic fronting, focus positioning and the nature of the verb phrase in Basque. In Studies on Fronting, F. Jansen (ed.), 81–112. Lisse: Peter de Ridder Press. De Simone Brouwer, Francesco 1921 Grammatica della lingua greca. Naples: Tipografia della Reale Università. Dezs˝o, László 1973 Topics on syntactic typology. Paper presented at the Symposium on Typology, Prague 14–17 August 1973. [Also translated in Italian as: Problemi di tipologia sintattica. In La tipologia linguistica, Paolo Ramat (ed.), 239–261. Bologna: Il Mulino, 1976.] Diderichsen, Paul 1936a Prolegomena til en metodisk dansk Syntax. Forhandlinger paa det ottende nordiske Filologmode i København den 12–14 August 1935, Copenhagen: Schultz. [Reprinted in Paul Diderichsen. Helhed og Struktur. Udvalgte Sprogvidenskabelige Afhandlinger, Anders Bjerrum et al. (eds.), Copenhagen: Gad.] 1936b Om den danske Sætnings Struktur. Selskab for nordisk filologi, Aarsberetning for 1936: 5–6. Dik, Simon C. 1978 Functional Grammar. Amsterdam: North Holland. 1980 Studies in Functional Grammar. London and New York: Academic Press. 1983 Advances in Functional Grammar. Dordrecht: Foris. 1989 The Theory of Functional Grammar. Dordrecht: Foris. Dik, Simon C., Maria E. Hoffmann, Jan R. de Jong, Sie J. Dijang, Harry Stroomer, and Lourens de Vries 1981 On the typology of focus phenomena. In Perspectives on Functional Grammar, Teun Hoekstra, Harry van der Hulst, and Michael Moortgat (eds.), 41–74. Dordrecht: Foris. Dooley Collberg, Sheila 1988 GPSG and the verb-second phenomenon in Swedish. In Papers from the Tenth Scandinavian Conference of Linguistics, Vol. 1, Victoria Rosén (ed.), 132–146. Bergen: University of Bergen. Dover, K. J. 1960 Greek Word-order. Cambridge: Cambridge University Press. Drach, Erich 1963 Grundgedanken der deutschen Satzlehre. Frankfurt a.M.: Diesterweg. Original edition Frankfurt 1937.

Interaction of syntactic and pragmatic factors

531

Dressler, Wolfgang 1969 Eine textsyntaktische Regel der indogermanischen Wortstellung. Zeitschrift für vergleichende Sprachforschung (KZ) 83 (1): 1–25. Dryer, Matthew S. 1995 Frequency and pragmatically unmarked word order. In Word Order in Discourse, Pamela Downing and Michael Noonan (eds.), 105–135. Amsterdam: John Benjamins. Ebert, Robert P. 1980 Social and stylistic variation in Early New High German word-order: the sentence frame (Satzrahmen). Beiträge zur Geschichte der deutschen Sprache und Literatur (PBB) 102: 357–398. Ellenberger, Henry F. 1994 Histoire de la découverte de l’inconscient. Paris: Fayard. Engel, Ulrich 1982 Syntax der deutschen Gegenwartssprache. Berlin: Schmidt. 1991 Deutsche Grammatik. 2nd revised edition, Heidelberg: Julius Groos Verlag. Erdmann, Peter 1990 Discourse and Grammar. Focussing and Defocussing in English. Tübingen: Niemeyer. Erguvanli, Eser 1984 The Function of Word-Order in Turkish Grammar. (University of California Publications in Linguistics 106). Berkeley and Los Angeles: University of California Press. Erguvanli-Taylan, Eser 1987 The role of semantic features in Turkish word-order. Folia Linguistica 21 (2–4): 215– 227. Etzensperger, Jürg 1979 Die Wortstellung der deutschen Gegenwartssprache als Forschungsobjekt. Berlin and New York: Walter de Gruyter. Euler, Wolfram 1982 Es war ein König. Eine Einleitungsformel mit indogermanischer Grundlage. In Sprachwissenschaft in Innsbruck: Arbeiten von Mitgliedern und Freunden des Instituts für Sprachwissenschaft der Universität Innsbruck aus Anlass des fünfzigjährigen Bestehens des Instituts im Jahre 1978 und zum Gedenken an die 25. Wiederkehr des Todestages von Hermann Ammann am 12. September 1981, Wolfgang Meid, Hermann Ölberg, and Hans Schmeja (eds.), 53–68. Innsbruck: Das Institut. Evans, Simon D. 1989 A Grammar of Middle Welsh. Dublin: Dublin Institute for Advanced Studies. Original edition 1964. Fernandez, M. M. Jocelyne 1986 Word-order and enunciative particles in the Sami (“Lappish”) dialogue. A contrastive and panchronic approach. Finnisch-Ugrische Mitteilungen 10: 113–125. 1994 Les particules énonciatives dans la construction du discours. Paris: Presses Universitaires de France. Fife, James, and Gareth King 1991 Focus and the Welsh “abnormal sentence”: a cross-linguistic perspective. In Studies in Brythonic Word-order, James Fife and Erich Poppe (eds.), 81–153. Amsterdam: John Benjamins. Fillmore, Charles 1968 The case for case. In Universals in Linguistics, Emmon Bach and Robert T. Harms (eds.), 1–88. New York: Holt, Rinehart & Winston. Firbas, Ian 1964 On defining the theme in functional sentence analysis. Travaux Linguistiques de Prague 1: 267–280.

532 1965 1992

Rosanna Sornicola

A note on transition proper in functional sentence perspective. PhP 8: 170–176. Functional Sentence Perspective in Written and Spoken Communication. Cambridge: Cambridge University Press. Fleischmann, Klaus 1973 Verbstellung und Relieftheorie. Ein Versuch zur Geschichte des deutschen Nebensatzes. Munich: Fink. Fougeron, Irina 1989 Prosodie et organisation du message. Paris: Klincksieck. Freeman, R. A. 1937 Syntax of Old Icelandic. Leeds: Chorley & Pickersgill. Fromm, H. 1982 Finnische Grammatik. Heidelberg: Winter. Gadet, Françoise 1992 Le français populaire. Paris: Presses Universitaires de France. Garbini, Giovanni 1984 Le lingue semitiche. Studi di storia linguistica. Naples: Istituto Universitario Orientale, Dipartimento di Studi Asiatici. Gazdar, Gerald, Ewan Klein, and Geoffrey K. Pullum (eds.) 1983 Order, Concord and Constituency. Dordrecht: Foris. Givón, Talmy 1977a On the V S word-order in Israeli Hebrew: pragmatics and typological change. In Studies in Modern Hebrew Syntax and Semantics: The Transformational-Generative Approach, (North-Holland Linguistic Series 32), Peter Cole (ed.), 153–181. Amsterdam: NorthHolland. 1977b The drift from VSO to SVO in Biblical Hebrew. In Mechanisms of Syntactic Change, Charles Li (ed.), 181–254. Austin, TX: Texas University Press. 1983 Topic continuity in discourse: an introduction. In Topic Continuity in Discourse. A Quantitative Cross-language Study, Talmy Givón (ed.), 1–41. Amsterdam and Philadelphia: John Benjamins. Givón, Talmy (ed.) 1983 Topic Continuity in Discourse. A Quantitative Cross-language Study. Amsterdam and Philadelphia: John Benjamins. Glinert, Lewis 1989 The Grammar of Modern Hebrew. Cambridge: Cambridge University Press. Goldenberg, Gideon 1998 Studies in Semitic Linguistics. Selected Writings. Jerusalem: The Magnes Press & The Hebrew University. Gonda, J. 1952 Remarques sur la place du verbe dans la phrase active et moyenne en langue sanscrite. Utrecht: Oosthoek’s Uitgevers. Graffi, Giorgio 1994 Sintassi. Bologna: Il Mulino. Greenberg, Joseph H. 1963 Some universals of grammar with particular reference to the order of meaningful elements. In Universals of Language, Joseph H. Greenberg (ed.), 73–113. Cambridge, MA: MIT Press. Gruber, Jeffrey 1967 Topicalization in child language. Foundations of Language 3: 37–65. Guentchéva, Zlatka 1994 Thématisation de l’objet en bulgare. Bern: Lang.

Interaction of syntactic and pragmatic factors

533

Gvozdanovi´c, Jadranka 1981 Word-order and displacement in Serbo-Croatian. In Predication and Expression in Functional Grammar, Machtelt Bolkestein et al. (eds.), 125–141. London: Academic Press. Haberland, Hartmut 1993 Thetic/categorical distinction. In Encyclopedia of Language and Linguistics, Vol. 9: Syntax, R. E. Asher and J. M. Y. Simpson (eds.), 4605–4606. Oxford and New York: Pergamon Press. Haegeman, Liliane 1991 Introduction to Government and Binding Theory. Oxford: Blackwell. Haider, Hubert 1986 Verb-second in German. In Verb-second Phenomena in Germanic Languages, Hubert Haider and Martin Prinzhorn (eds.), 49–76. Dordrecht: Foris. Haiman, John 1974 Targets and Syntactic Change. The Hague: Mouton. Hakulinen, Auli, Fred Karlsson, and Maria Vilkuna 1980 Suomen tekstilauseiden piirteitä: kvantitatiivinen tutkimus. (Publication 6) Helsinki: University of Helsinki, Department of General Linguistics. Hakulinen, Lauri 1961 The Structure and Development of the Finnish Language. (Indiana University Publications, Uralic and Altaic Series 3). Bloomington, IN: Indiana University Press. 1979 Suomen kielen rakenne ja kehitys. Helsinki: Otava. Halliday, Michael A. K. 1967a Theme and information in the English clause. In System and Function in Language. Selected Papers, Gunther Kress (ed.), 174–188. Oxford: Oxford University Press. 1967b Notes on transitivity and theme in English, Part 1. Journal of Linguistics 3 (1): 37–81. 1967c Notes on transitivity and theme in English, Part 2. Journal of Linguistics 3 (2): 199–244. 1968 Notes on transitivity and theme in English, Part 3. Journal of Linguistics 4 (2): 179–215. Hankamer, Jorge 1971 Constraints on deletion in syntax. Yale university dissertation. Harris, Alice C. 1981 Georgian Syntax. A Study in Relational Grammar. Cambridge: Cambridge University Press. Havránek, Bohuslav 1968 Quelques problèmes de l’étude diachronique de la structure syntactique, surtout en slave. Travaux Linguistiques de Prague 3: 9–16. Hawkins, John A. 1983 Word-order Universals. New York: Academic Press. 1994 A Performance Theory of Order and Constituency. Cambridge: Cambridge University Press. Heger, Klaus 1982 “Il la lui a donné, à Jean, son père, sa moto.” Neue Überlegungen zu einem alten Beispiel. In Festschrift für Johannes Hubschmid zum 65. Geburtstag: Beiträge zur allgemeinen, indogermanischen und romanischen Sprachwissenschaft, Otto Winkelmann and Mario Braisch (eds.), 53–66. Bern: Francke. Helbig, Gerhard, and Joachim Buscha 1989 Deutsche Grammatik. Leipzig: Enzyklopädie. Heltoft, Lars 1986 The V-2 analysis: a reply from Diderichsen-tradition. In Scandinavian Syntax, Östen Dahl and Anders Holmberg (eds.), 50–67. Stockholm: Institute of Linguistics.

534

Rosanna Sornicola

Hemon, Roparz 1984 A Historical Morphology and Syntax of Breton. Dublin: The Dublin Institute for Advanced Studies. Hendrik, Randall (ed.) 1990 The Syntax of the Modern Celtic Languages. (Syntax and Semantics 23). San Diego, CA: Academic Press. Henriksen, Carol 1986 Historical Danish syntax, within the framework of Paul Diderichsen’s syntactic theory. In Scandinavian Syntax, Östen Dahl and Anders Holmberg (eds.), 75–87. Stockholm: Institute of Linguistics. Hetzron, Robert 1971 Presentative function and presentative movement. Studies in African Linguistics (Supplement 2): 79–105. 1975 The presentative movement, or why the ideal word-order is V S O P. In Word-order and Word-order Change, Charles Li (ed.), 345–388. Austin, TX: University of Texas Press. Hinds, John 1986 Japanese. London: Routledge. Höhle, Tilman N. 1982 Explikationen für “normale Betonung” und “normale Wortstellung”. In Satzglieder im Deutschen, Werner Abraham (ed.), 75–153. Tübingen: Narr. Holden, Kyril, and Maria Krupp 1987 Word-order in Russian transitive sentences. Folia Slavica 8 (2–3): 254–271. Holes, Clive 1995 Modern Arabic. Structures, Functions and Varieties. London: Longman. Holland, Gary B. 1980 Problems of word-order change in selected Indo-European languages. Unpublished doctoral dissertation, University of California at Berkeley. Holloway King, Tracy 1995 Configuring Topic and Focus in Russian. Stanford, CA: Center for the Study of Language and Information. Holmberg, Anders 1986 Word-order and Syntactic Features in the Scandinavian Languages and English. Stockholm: Department of General Linguistics. 1989 Verb movement in Finnish. In Papers from the Fifth Nordic Conference on Bilingualism, Vol. 1, (Studies in Language 14), Jussi Niemi (ed.), 112–130. Joensuu: Joensuun yliopisto, Humanistinen tiedekunta. Holmes, Philip, and Ian Hinchliffe 1994 Swedish. A Comprehensive Grammar. London: Routledge. Horvath, Julia 1986 Focus in the Theory of Grammar and the Syntax of Hungarian. Dordrecht: Foris. Huang, C.-T. James 1984 On the distribution and reference of empty pronouns. Linguistic Inquiry 15: 531–574. Ingham, B. 1991 Sentence structure in Khuzistani Arabic. In Semitic Studies in Honour of W. Leslau on the Occasion of his 85th Birthday, Vol. 1, Alan S. Kaye (ed.), 714–728. Wiesbaden: Harrassowitz. Jacob, Daniel 1990 Markierung von Aktantenfunktionen und “Prädetermination” im Französischen. Tübingen: Niemeyer.

Interaction of syntactic and pragmatic factors

535

Jacobs, Joachim 1988 Fokus-Hintergrund-Gliederung und Grammatik. In Intonationsforschungen, Hans Altmann (ed.), 98–134. Tübingen: Niemeyer. Jakobson, Roman 1963 Implications of language universals for linguistics. In Universals of Language, Joseph H. Greenberg (ed.), 208–219. Cambridge, MA: MIT Press. Jespersen, Otto 1969 Language. Its Nature, Development and Origin. London: Allen & Unwin. Original edition 1922. Jo, Mi-Jeung 1995 The theory of syntactic focalization on a subcategorization feature of verbs. In Discourse Configurational Languages, Katalin É. Kiss (ed.), 335–374. Oxford: Oxford University Press. Jongeling, K. 1991 On the V S O character of Hebrew. In Studies in Hebrew and Aramaic Syntax Presented to Professor J. Hoftijzer, K. Jongeling, H. L. Murre-Van den Berg, and L. Van Rompay (eds.), 103–111. Leiden: Brill. Kalitsunakis, Johannes 1963 Grammatik der neugriechischen Volkssprache. Berlin: Walter de Gruyter. Kálmán, László, Gábor Prószéky, Ádám Nádasdy, and György Kálmán C. 1986 Hocus, focus and verb types in Hungarian infinitive constructions. In Topic, Focus and Configurationality, Werner Abraham and Sjaak de Meij (eds.), 129–142. Amsterdam and Philadelphia: John Benjamins. Karlsson, Fred 1987 Finnish Grammar. Porvoo-Helsinki-Juva: Werner Söderström Osakeyhtiö. Karttunen, Lauri, and Martin Kay 1985 Parsing in a free word-order language. In Natural Language Parsing, D. Dowty, L. Karttunen, and A. Zwicky (eds.), 279–306. Cambridge: Cambridge University Press. Keenan, Edward 1976 Towards a universal definition of subject. In Subject and Topic, Charles Li (ed.), 303– 333. New York: Academic Press. Kenesei, István 1986 On the logic of word-order in Hungarian. In Topic, Focus and Configurationality, Werner Abraham and Sjaak de Meij (eds.), 143–160. Amsterdam and Philadelphia: John Benjamins. Kim, Alan Hyun-Oak 1988 Pre-verbal focussing and type XXIII language. In Studies in Syntactic Typology, Michael Hammond, Edith Moravcsik, and Jessica R. Wirth (eds.), 147–169. Amsterdam: John Benjamins. Kippasto, Ann 1989 Word-order in Votian. In Proceedings of the Fifth International Finno-Ugrist Students’ Conference, Riho Grünthal, Sirpa Penttinen, and Tapani Salminen (eds.), 84–89. Helsinki: Finno-Ugrian Society. Kirsner, Robert S. 1979 The Problem of Presentative Sentences in Modern Dutch. Amsterdam: North-Holland. Kiss, Katalin É. 1987 Configurationality in Hungarian. Dordrecht: Reidel. 1995 Introduction. In Discourse Configurational Languages, Katalin É. Kiss (ed.), 3–27. Oxford: Oxford University Press. Kiss, Katalin É. (ed.) 1995 Discourse Configurational Languages. Oxford: Oxford University Press.

536

Rosanna Sornicola

Kohonen, Viljo 1982 A note on factors affecting the position of accusative objects and complements in Ælfric’s Catholic Homilies I. In Approaches to Word-order: Reports on Text Linguistics, Nils Erik Enkvist and Viljo Kohonen (eds.), 175–196. Åbo: Åbo Akademi. Komlósy, András 1986 Focussing on focus in Hungarian. In Topic, Focus and Configurationality, Werner Abraham and Sjaak de Meij (eds.), 215–226. Amsterdam and Philadelphia: John Benjamins. Kovtunova, I. I. 1976 Sovremennyj russkij jazyk. Porjadok slov i aktual’noe cˇ lenenie predloženija. Moscow: Prosvešˇcenie. Kress, Bruno 1982 Isländische Grammatik. Leipzig: Enzyklopädie. Krylova, O., and S. Khavronina 1976 Word-order in Russian Sentences. Moscow: Russian Language Publishers. Ladd, Robert D. 1996 Intonational Phonology. Cambridge: Cambridge University Press. Lafitte, Pierre 1978 Grammaire Basque (Navarro-Labourdin Littéraire). Donostia: Elkar. Original edition 1944. Lambrecht, Knud 1981 Topic, Antitopic and Verb-agreement in Non-standard French. Amsterdam: John Benjamins. 1994 Information Structure and Sentence Form. Cambridge: Cambridge University Press. Lampe, G. W. H. (ed.) 1969 The Cambridge History of the Bible, Vol. 2: The West from the Fathers to the Reformation. Cambridge: Cambridge University Press. Lehmann, Winfred P. 1973 A structural principle of language and its implications. Language 49: 47–66. 1974 Proto-Indo-European Syntax. Austin, TX: University of Texas Press. Lerch, Eugen 1934 Historische Französische Syntax, Bd. III: Modalität. Leipzig: Reisland. Levelt, William, J. M. 1989 Speaking. From Intention to Articulation. Cambridge, MA: MIT Press. Lewis, Henry 1942 The Sentence in Welsh. London: British Academy. Lewis, Henry, and Holger Pedersen 1937 A Concise Comparative Celtic Grammar. Göttingen: Vandenhoeck & Ruprecht. Lewy, E. 1922 Tscheremissische Grammatik. Leipzig: Haessel Verlag. Li, Charles (ed.) 1976 Subject and Topic. New York: Academic Press. Linde, P. 1923 Die Stellung des Verbs in der lateinischen Prosa. Glotta 12: 153–178. Lyons, John 1977 Semantics. 2 Vols. Cambridge: Cambridge University Press. MacCana, Proinsias 1973 On Celtic word-order and the Welsh abnormal sentence. Ériu 24: 90–120. 1991 Further notes on constituent order in Welsh. In Studies in Brythonic Word-order, James Fife and Erich Poppe (eds.), 45–80. Amsterdam: John Benjamins.

Interaction of syntactic and pragmatic factors

537

MacCoisdealbha, Pádraig 1998 The Syntax of the Sentence in Old Irish. Selected Studies from a Descriptive, Historical and Comparative Point of View. Tübingen: Niemeyer. Marantz, Alec P. 1984 On the Nature of Grammatical Relations. Cambridge, MA: MIT Press. Marchello-Nizia, Christiane 1995 L’évolution du français. Ordre des mots, démonstratifs, accent tonique. Paris: Colin. Marouzeau, Jules 1938 L’ordre des mots dans la phrase latine, Vol. 2: Le verbe. Paris: Les Belles Lettres. Masica, Colin P. 1991 The Indo-Aryan Languages. Cambridge: Cambridge University Press. Maslova, Elena 1995 V S / S V opposition in Russian and the thetic/categorical distinction. Sprachtypologie und Universalienforschung 48 (1/2): 106–124. Mathesius, Vilém 1939 O tak zvaném aktuálním cˇ lenˇení vˇetném. Slovo a slovesnost 5: 171–174. ˇ 1941–1942 Ze srovnávacích studií slovosledných. Casopis pro moderní filologii 28: 181–190; 302– 307. Matthews, Peter H. 1991 Morphology. Cambridge: Cambridge University Press. 1994 Grammatical Theory in the United States from Bloomfield to Chomsky. Cambridge: Cambridge University Press. Matthews, W. K. 1960 Russian Historical Grammar. London: The Athlone Press. McCloskey, James 1983 A VP in a VSO language? In Order, Concord and Constituency, Gerald Gazdar, Ewan Klein, and Geoffrey K. Pullum (eds.), 9–55. Dordrecht: Foris. Meillet, Antoine 1906 L’état actuel des études de linguistique générale, Leçon d’ouverture du cours de Grammaire comparée au Collège de France, lue le mardi 13 février 1906. [Also in 1965 Linguistique historique et linguistique générale, Antoine Meillet (ed.), 1–18. Paris: Champion.] 1912 L’évolution des formes grammaticales. Scientia 12 (26): 6. [Also in 1965 Linguistique historique et linguistique générale, Antoine Meillet (ed.), 130–148. Paris: Champion.] Meillet, A., and H. De Willman-Grabowska 1921 Grammaire de la langue polonaise. Paris: Champion. Meillet, A., and A. Vaillant 1924 Grammaire de la langue serbo-croate. Paris: Champion. Milano, Emma 1994 Sul parlato: problemi di ordine delle parole. Unpublished M.A. thesis in sociolinguistics, Università di Napoli “Federico II”. Miller, Jim 1995 VS clauses in Russian: contrast, new episodes and theticity. Sprachtypologie und Universalienforschung 48 (1/2): 125–141. Miller, Jim, and Regina Weinert 1998 Spontaneous Spoken Language. Oxford: Clarendon Press. Mitchell, Bruce 1985 Old English Syntax. 2 Vols. Oxford: Clarendon Press.

538

Rosanna Sornicola

Mohanan, Karuvannur Puthanveettil, and Tara Mohanan 1994 Issues in word-order. In Theoretical Perspectives on Word-order in South Asian Languages, Miriam Butt, Tracy Holloway King, and Gillian Ramchand (eds.), 153–216. Stanford, CA: Center for the Study of Language and Information. Muysken, Pieter 1995 Focus in Quechua. In Discourse Configurational Languages, Katalin É. Kiss (ed.), 375– 393. Oxford: Oxford University Press. Nichols, Johanna 1985 The grammatical marking of theme in literary Russian. In Issues in Russian Morphosyntax, Michael S. Flier and Richard D. Brecht (eds.), 170–208. Columbus, OH: Slavica Publishers. Nuyts, Jan, and George de Schutter (eds.) 1987 Getting One’s Words into Line. On Word Order and Functional Grammar. Dordrecht: Foris. Nykrog, Per 1957 L’influence latine savante sur la syntaxe du français. In The Classical Pattern of Modern Western Civilization: Language, (Acta Congressus Madvigiani, Travaux du Cercle Linguistique de Copenhagen 11), Franz Blatt (ed.), 89–114. Copenhagen: E. Munksgaard. Ortiz de Urbina, John 1995 Residual verb-second and verb-first in Basque. In Discourse Configurational Languages, Katalin É. Kiss (ed.), 99–121. Oxford: Oxford University Press. Ó Siadhail, Mícheál 1989 Modern Irish. Grammatical Structure and Dialectal Variation. Cambridge: Cambridge University Press. Payne, Dorothy (ed.) 1992 Pragmatics of Word-order Flexibility. Amsterdam and Philadelphia: John Benjamins. Pedersen, Holger 1976 Vergleichende Grammatik der keltischen Sprachen. 2 Vols. Göttingen: Vandenhoeck & Ruprecht. Original edition 1909–1913. Perlmutter, David 1978 Impersonal passives and the unaccusative hypothesis. Berkeley Linguistic Society 4: 157–189. PET 1995–1997 Piccola Enciclopedia Treccani, 12 Vols. Rome: Istituto dell’Enciclopedia Italiana. Pétursson, Magnus 1978 Isländisch. Hamburg: Helmut Buske Verlag. Plank, Frans 1984 Ces obscurs objets du désir. Introduction. In Objects. Towards a Theory of Grammatical Relations, Frans Plank (ed.), 1–8. London: Academic Press. Plank, Frans (ed.) 1984 Objects. Towards a Theory of Grammatical Relations. London: Academic Press. Platzack, Christer 1986 The position of the finite verb in Swedish. In Verb-second Phenomena in Germanic Languages, Hubert Haider and Martin Prinzhorn (eds.), 27–48. Dordrecht: Foris. Poppe, Erich 1989 Constituent ordering in Breudwyt Maxen Wledic. Bulletin of the Board of Celtic Studies 36: 43–63. 1990 Word-order patterns in Breudwyt Ronabwy. In Celtic Linguistics: Ieithyddiaeth Geltaidd. Readings in the Brythonic Languages. Festschrift for T.A. Watkins, Martin J. Ball, James Fife, Erich Poppe, and Jenny Rowland (eds.), 445–460. Amsterdam: John Benjamins.

Interaction of syntactic and pragmatic factors 1991a

1991b 2000

539

Word-order in Cyfranc Lludd a Llefelys: notes on the pragmatics of constituent-ordering in MW narrative prose. In Studies in Brythonic Word-order, James Fife and Erich Poppe (eds.), 155–204. Amsterdam: John Benjamins. Untersuchungen zur Worstellung im Mittelkymrischen. Hamburg: Helmut Buske. Constituent order in Middle Welsh: the stability of the pragmatic principle. In Stability, Variation and Change of Word Order Patterns over Time, Rosanna Sornicola, Erich Poppe, and Ariel Shisha-Halevy (eds.), 41–51. Amsterdam: John Benjamins.

Poppe, Nicholas 1954 Grammar of Written Mongolian. Wiesbaden: Harrassowitz. Quirk, Randolph, Sidney Greenbaum, Geoffrey Leech, and Jan Svartvik 1985 A Comprehensive Grammar of the English Language. London: Longman. Rebuschi, Georges 1984 Structure de l’énoncé en basque. Paris: Société d’études linguistiques et anthropologiques de France. Renzi, Lorenzo 1984 La tipologia dell’ordine delle parole e le lingue romanze. Linguistica 24: 27–59. Rouveret, Alain 1990 X-bar theory, minimality and barrierhood in Welsh. In The Syntax of the Modern Celtic Languages, (Syntax and Semantics 23), Randall Hendrik (ed.), 27–79. San Diego, CA: Academic Press. Rowlands, Eurys I. 1980 Y frawddeg gymysg a’r frawddeg dro. Bulletin of the Board of Celtic Studies 28: 218– 222. Russ, Charles V. J. (ed.) 1989 The Dialects of Modern German. A Linguistic Survey. Stanford, CA: Stanford University Press. Rydén, Mats, and Sverker Brorström 1987 The Be/Have Variation with Intransitives in English. Stockholm: Almqvist & Wiksell. Sáenz-Badillos, Angel 1993 A History of the Hebrew Language. Cambridge: Cambridge University Press. Salvi, Giampaolo 1993 Ordine delle parole e struttura della frase nelle lingue romanze antiche. In Grammatikographie der romanischen Sprachen, (Akten der gleichnamigen Sektion des Bamberger Romanistentages [23–29.9.1991]), Christian Schmitt (ed.), 455–477. Bonn: Romanistischer Verlag. 1997 La posizione tipologica dell’italiano tra le lingue romanze. Italienische Studien 18: 25– 38. Sasse, Hans-Jürgen 1987 The thetic/categorical distinction revisited. Linguistics 25: 511–580. 1995a Theticity and V S order: a case study. Sprachtypologie und Universalienforschung 48 (1/2): 3–31. 1995b V S clauses in Modern Greek and Hungarian. Sprachtypologie und Universalienforschung 48 (1/2): 142–188. Sauvageot, Aurélien 1973 L’élaboration de la langue finnoise. Paris: Klincksieck. Scaglione, Aldo 1972 The Classical Theory of Composition from its Origins to the Present. A Historical Survey. Chapel Hill, NC: University of North Carolina Press. 1981 The Theory of German Word-order from the Renaissance to the Present. Minneapolis, MN: University of Minnesota Press.

540

Rosanna Sornicola

Scherpenisse, Wim 1986 The Connection between Base Structure and Linearization Restrictions in German and Dutch. Frankfurt a. M.: Peter Lang. Schroeder, Christoph 1995 Post-predicate argument in Turkish. Sprachtypologie und Universalienforschung 48 (1/2): 204–218. Schwyzer, Eduard 1950 Griechische Grammatik, Band 2: Syntax und syntaktische Stilistik. Munich: Beck’sche Verlagsbuchhandlung. Sgall, Peter, Eva Hajiˇcová, and Jarmila Panevová 1986 The Meaning of the Sentence in its Semantic and Pragmatic Aspects. Dordrecht: Reidel. Shisha-Halevy, Ariel 1995 Structural sketches of Middle Welsh syntax. I: The converter systems. Studia Celtica 29: 127–223. 2000 Stability in clausal/phrasal pattern constituent sequencing: 4000 years of Egyptian (with some theoretical reflections, also on Celtic). In Stability, Variation and Change of Word Order Patterns over Time, Rosanna Sornicola, Erich Poppe, and Ariel Shisha-Halevy (eds.), 71–100. Amsterdam: John Benjamins. Siewierska, Anna 1988 Word-Order Rules. Beckenham: Croom Helm. Siewierska, Anna (ed.) 1998 Constituent Order in the Languages of Europe. Berlin and New York: Mouton de Gruyter. Sigurðsson, Halldór Á. 1986 Verb post-second in a V2 language. In Scandinavian Syntax, Östen Dahl and Anders Holmberg (eds.), 138–149. Stockholm: Institute of Linguistics. 1989 Verbal Syntax and Case in Icelandic in a Comparative GB Approach. Lund: Department of Scandinavian Languages. Sinor, Denis 1990 Introduction. In The Cambridge History of Early Inner Asia, Denis Sinor (ed.), 1–18. Cambridge: Cambridge University Press. Sohn, Ho-min 1994 Korean. London: Routledge. Sørensen, Knud 1957 Latin influence on English syntax. In The Classical Pattern of Modern Western Civilization: Language, (Acta Congressus Madvigiani, Travaux du Cercle Linguistique de Copenhagen 11), Franz Blatt (ed.), 131–155. Copenhagen: E. Munksgaard. Sornicola, Rosanna 1981 Sul parlato. Bologna: Il Mulino. 1986 Costituenza, dipendenza e la struttura tema-rema in Italiano. In Tema-rema in Italiano, Harro Stammerjohann (ed.), 121–139. Tübingen: Narr. 1988 It-clefts and wh-clefts: two awkward sentence types. Journal of Linguistics 24: 343–379. 1990 Soggetti prototipici e soggetti non prototipici: l’italiano a confronto con le altre lingue europee. Lingua e Stile 25 (1): 3–24. 1993a Topic, focus and word-order. In Encyclopedia of Language and Linguistics, Vol. 9: Syntax, R. E. Asher and J. M. Y. Simpson (eds.), 4633b–4640b. Oxford and New York: Pergamon Press. 1993b The many routes of functionalism. Rivista di Linguistica 5 (1): 157–176. 1994 On word-order variability: a study from a corpus of Italian. Lingua e Stile 29 (1): 25–57.

Interaction of syntactic and pragmatic factors 1995a

541

Mutamenti di prospettiva culturale nelle lingue europee moderne: l’influenza del latino sulla sintassi. In Kulturwandel im Spiegel des Sprachwandels, Karl-Egon Lönne (ed.), 41–58. Tübingen and Basel: Francke. 1995b Theticity, V S order and the interplay of syntax, semantics and pragmatics. Sprachtypologie und Universalienforschung 48 (1/2): 72–83. 1998 The interpretation of historical sources as a problem for diachronic typology: wordorder in English rhetorical and grammatical treatises of XVIth and early XVIIth centuries. In The Virtues of Language. History in Language, Linguistics and Text in Memory of Thomas Frank, Dieter Stein and Rosanna Sornicola (eds.), 81–128. Amsterdam: John Benjamins. 1999 Labile orders and grammatical functions. The functional representation of one-argument structures. In Boundaries of Morphology and Syntax, Lunella Mereu (ed.), 291–305. Amsterdam: John Benjamins. 2000 Stability, variation and change in word order: some evidence from the Romance languages. In Stability, Variation and Change of Word Order Patterns over Time, Rosanna Sornicola, Erich Poppe, and Ariel Shisha-Halevy (eds.), 101–115. Amsterdam: John Benjamins. Sornicola, Rosanna, and Aleš Svoboda (eds.) 1992 Il campo di tensione. La sintassi della Scuola di Praga. Naples: Liguori. Sornicola, Rosanna, Marcello Barbato, Maria Cesarano, and Daniela Puolato 1994 Word order in old Romance languages. Paper presented at the EUROTYP project meeting of the research group on Pragmatic Organization of Discourse, Naples, 11–13 November 1994. Sridhar, S. N. 1990 Kannada. London: Routledge. Srpová, Milena 1990 Quelques remarques sur la dynamique énonciative en tchèque et en français. Revue d’Études Slaves 62 (1/2): 403–416. Stanchev, Svillen 1987 Some observations on the order of constituents in Bulgarian. In Getting One’s Words into Line. On Word Order and Functional Grammar, Jan Nuyts and George de Schutter (eds.), 33–43. Dordrecht: Foris. Steele, Susan 1978 Word-order variation: a typological study. In Universals of Human Languages, Vol. 4: Syntax, Joseph H. Greenberg, Charles A. Ferguson, and Edith A. Moravcsik (eds.), 585– 623. Stanford, CA: Stanford University Press. Stender-Petersen, Adolf, and Knud Jordal 1957 Das griechisch-byzantinische Erbe im Russischen. In The Classical Pattern of Modern Western Civilization: Language, (Acta Congressus Madvigiani, Travaux du Cercle Linguistique de Copenhagen 11), Franz Blatt (ed.), 163–218. Copenhagen: E. Munksgaard. Stenson, Nancy 1981 Studies in Irish Syntax. Tübingen: Narr. Svoboda, Aleš ˇ 1984 Ceské slovosledné pozice z pohledu aktuálního cˇ lenˇení [Czech word-order from the viewpoint of functional sentence perspective]. Slovo a slovesnost 45 (1): 22–34; 45 (2): 88–103. Svolacchia, Marco, Lunella Mereu, and Annarita Puglielli 1995 Aspects of discourse configurationality in Somali. In Discourse Configurational Languages, Katalin É. Kiss (ed.), 65–98. Oxford: Oxford University Press.

542

Rosanna Sornicola

Swarajya Lakshmi, V., and Aditi Mukherjee (eds.) 1996 Word-order in Indian Languages. Hyderabad: Centre of Advanced Study in Linguistics Osmania University & Brooklins Corporation. Tajsner, Przemyslaw 1990 Scrambling and the Polish word-order. An alternative hypothesis. Papers and Studies in Contrastive Linguistics 25: 43–57. Tallerman, Maggie 1991 The directionality of the head subcategorization in Welsh. In Studies in Brythonic Wordorder, James Fife and Erich Poppe (eds.), 311–328. Amsterdam: John Benjamins. Tarvainen, Kalevi 1985 Kontrastive Syntax Deutsch-Finnisch. Heidelberg: Julius Groos Verlag. Tauli, Valter 1983 Standard Estonian Grammar, Part 2: Syntax. Uppsala: Borgströms Tryckeri. Tesnière, Lucien 1959 Éléments de syntaxe structurale. Paris: Klincksieck. Thomason, Sarah, and Terrence Kaufman 1988 Language Contact, Creolization and Genetic Linguistics. Berkeley and Los Angeles: California University Press. Thompson, Sandra 1978 Modern English from a typological point of view: some implications on the function of word-order. Linguistische Berichte 54: 19–35. Thráinsson, Höskuldeer 1986 V1, V2, V3 in Icelandic. In Verb-second Phenomena in Germanic Languages, Huber Haider and Martin Prinzhorn (eds.), 169–194. Dordrecht: Foris. Thumb, Albert 1912 Handbook of the Modern Greek Vernacular. Grammar, Texts, Glossary. Edinburgh: Clark. Tiffou, Étienne, and Jürgen Pesot 1989 Contes du Yasin. Introduction au bourouchaski du Yasin avec grammaire et dictionnaire analytique. Paris: Peeters. Timm, Lenora 1989 Word-order in 20th century Breton. Natural Languages and Linguistic Theory 7 (3): 361–378. 1991 Discourse pragmatics of NP-initial sentences in Breton. In Studies in Brythonic Wordorder, James Fife and Erich Poppe (eds.), 275–310. Amsterdam: John Benjamins. Tomlin, Russell S. 1986 Basic Word-order. Functional Principles. London: Croom Helm. Tompa, József 1985 Kleine ungarische Grammatik. Budapest: Akadémiai Kiado. Tsimpli, Ianthi Maria 1995 Focusing in Modern Greek. In Discourse Configurational Languages, Katalin É. Kiss (ed.), 176–206. Oxford: Oxford University Press. Uhlíˇrová, Ludmila 1969 Vztah syntaktické funkce vˇetného cˇ lenu a jeho místa ve vˇetˇe [The relationship between syntactic function and linear position of sentence elements. Methods and results of statistical research]. Slovo a slovesnost 30: 358–370 [Italian translation in Rosanna Sornicola and Aleš Svoboda (eds.)]. Ulrich, Miori¸ta 1985 Thetisch und Kategorisch. Tübingen: Narr. Uroševi´c, Zoran M., C. Carello, M. Saviic, G. Lukatela, and M. T. Turvey 1986 Some word-order effects in Serbo-Croat. Language and Speech 29 (2): 177–195.

Interaction of syntactic and pragmatic factors

543

Vähämäki, Börje 1984 Existence and Identity. A Study of the Semantics and Syntax of Existential Sentences in Finnish. Åbo: Åbo Akademi. van Oosten, Jeanne 1985 On the Nature of Topics, Agents, and Subjects: A Cognitive Explanation. Bloomington, IN: Indiana University Linguistic Club. van Steenbergen, Marlies 1989 Finnish: configurational or not? In Configurationality. The Typology of Asymmetries, László Marácz and Pieter Muysken (eds.), 143–158. Dordrecht: Foris. Vennemann, Theo 1972 Analogy in generative grammar: the origin of word-order. In Proceedings of the XIIth Congress of Linguists, Vol. 2, Luigi Heilmann (ed.), 79–83. Bologna: Il Mulino. Verpoorten, J.-M. 1977 L’ordre des mots dans l’Aitareya-Br¯ahmana. Paris: Les Belles Lettres. ˙ Vilkuna, Maria 1989 Free Word-order in Finnish. Its Syntax and Discourse Functions. Helsinki: Soumalaisen Kirjallisuunden Seura. Vogt, Hans 1971 Grammaire de la langue géorgienne. Oslo: Universitets-forlaget. 1974 L’ordre des mots en géorgien moderne. Bedi Kartlisa 32: 48–56. von Stechow, Armin, and Susanne Uhmann 1986 Some remarks on focus projection. In Topic, Focus and Configurationality, Werner Abraham and Sjaak de Meij (eds.), 295–320. Amsterdam and Philadelphia: John Benjamins. Wackernagel, Jacob 1892 Über ein Gesetz der indogermanischen Wortstellung. Indogermanische Forschungen 1: 333–436. [Also in Wackernagel, Jacob (1953). Kleine Schriften 1. 1–104. Göttingen: Vandenhoeck & Ruprecht]. Wagner, Heinrich 1959 Das Verbum in den Sprachen der britischen Inseln. Ein Beitrag zur geographischen Typologie des Verbums. Tübingen: Buchreihe der Zeitschrift für Celtische Philologie 1. 1967 Zur unregelmässigen Wortstellung in der altirischen Allitterationsdichtung. Beiträge zur Indogermanistik und Keltologie: 289–314. Wartburg, Walter von 1950 Évolution et structure de la langue française. Bern: Francke. Watkins, Calvert 1963 Preliminaries to a historical and comparative analysis of the syntax of the Old Irish verb. Celtica 6: 1–49. 1976 Towards Proto-Indo-European syntax: problems and pseudo-problems. In Papers from the Parasession on Diachronic Syntax (April 22, 1976), B. Steever, Carol A. Walker, and Salikoko S. Mufwene (eds.), 305–326. Chicago: Chicago Linguistic Society. Watkins, T. Arwyn 1991 The function of the cleft and non-cleft constituents orders in Modern Welsh. In Studies in Brythonic Word-order, James Fife and Erich Poppe (eds.), 329–351. Amsterdam: John Benjamins. Watters, John Robert 1979 Focus in Aghem: a study of its formal correlates and typology. In Aghem Grammatical Structure, (Southern California Occasional Papers in Linguistics 7), Larry M. Hyman (ed.), 137–197. Los Angeles: University of Southern California.

544

Rosanna Sornicola

Weber, David J. 1989 A Grammar of Huallaga (Huánuco) Quechua. Berkeley and London: University of California Press. Weil, Henri 1879 De l’ordre des mots dans les langues anciennes comparées aux langues modernes. Paris: Vieweg. Weinert, Regina, and Jim Miller 1996 Cleft constructions in spoken language. Journal of Pragmatics 25: 173–206. Weinrich, Harald 1993 Textgrammatik der deutschen Sprache. Mannheim and Leipzig: Duden Verlag. Wells, Christopher J. 1987 German: A Linguistic History to 1945. Oxford: Oxford University Press. Williams, Stephen J. 1980 A Welsh Grammar. Cardiff: University of Wales Press. Willis, David W. E. 1998 Syntactic Change in Welsh. A Study of the Loss of Verb-second. Oxford: Clarendon Press. Yokoyama, Olga 1986 Discourse and Word-order. Amsterdam and Philadelphia: John Benjamins. Zifonun, Gisela, Ludger Hoffmann, and Bruno Strecker 1997 Grammatik der deutschen Sprache. 3 Vols. Berlin: Walter de Gruyter.

Christoph Schroeder

Articles and article systems in some areas of Europe

1. Introduction This study tries to present an overview of article systems in the languages in some geographical areas of Europe. It concentrates on Northern Europe, Southeast Europe, and the Eastern margins of Europe and on some interjacent areas. These areas were chosen because article usage in the languages of these areas differs significantly from the common Western European pattern of established definite and indefinite articles. We will investigate areas with two definite articles (see Sections 4.1 and 4.2), areas where definite articles originate from possessive suffixes (see Section 4.6), areas with emergent definite and indefinite articles, and transitions between these areas.1 The overview includes both definite as well as indefinite articles, though when investigating transitions within and between the areas the focus is on definite article systems. Definite and indefinite articles are morphological devices which express grammatical categories of the same family. The co-occurrence of definite and indefinite articles is a pervasive feature in the languages of Europe (cf. Haspelmath 2001). However, cros-linguistically speaking, there is no systematic connection between the occurrence of definite and indefinite articles. Languages may have both (e.g. English), they may have none (e.g. Georgian), and they may equally express one of the two grammatical categories of “indefiniteness” or “definiteness” by means of one (or even more) article(s) and provide other means to express the other. For example, most varieties of Kirmanji (Northern Kurdish) have an indefinite article but not a definite one (but see Section 3.3), and Macedonian has definite articles but not an indefinite one (see Sections 4.4.1 and 4.4.6). Thus, the notion of “article system” does not pertain to a system including both definite and indefinite articles, but to a system of the usage of the article or of articles of one category in one language. In order to investigate article systems, first the grammatical categories they express, namely, “definiteness,” “indefiniteness,” and “specificity,” and the notions related to these categories have to be described. This is done in Section 2, based on Schroeder (1999). Section 3 presents and discusses the notions of “definite article” and “indefinite article.” Section 4 comprises a survey on article systems. It is subdivided according to areas, namely, “The North,” “The Balkans,” etc. Furthermore,

546

Christoph Schroeder

two excursuses – Sections 4.2 and 4.7 – deal in more detail with specific phenomena in unplanned spoken discourse. While Section 4.2 describes the system of definite articles after prepositions in the Low German dialects of the North Sea area, Section 4.7 aims at finding out whether the possessive suffix of the third person singular in Turkish can be called an “article in statu nascendi.” Section 5 concludes the findings of Section 4 and combines them with the discussions of Sections 2 and 3. A few words on the history of this chapter are in order: The central ideas which form the base of this study originate from discussions with my academic teacher Johannes Bechert, one of the founding fathers of EUROTYP. His untimely death in July 1994 was an irreplaceable loss for all of us who had the privilege to work together with this warm-hearted person and fine scholar. When the members of the EUROTYP Working Group I “Pragmatic Organization of Discourse” asked whether I could take over the task of writing the chapter for the final volume which Johannes Bechert had intended to write. I agreed, provided that I could use as a starting point two papers by Johannes Bechert on the progress of our research project (Bechert 1993). The papers form the basis for the main parts of the subsections 4.1, 4.3, 4.5 and 4.6 of this chapter, as well as for the conclusions in section 5. Of course, I accept full responsibility for misinterpretations, misunderstandings, and mistakes.

2.

The categories

2.1. Definiteness 2.1.1. Identifiability “Definiteness” will be defined as a grammatical category expressing the “identifiability” of a referent of a noun phrase. “Identifiability” is not a category of objective truth, but a pragmatic category based on the “mental object” or the “imaginary object” that emerges as a cognitive process in the ongoing discourse (cf. Sasse 1992: 180). When marking a noun phrase as “definite,” the speaker expects the listener to be able to classify the referent of the noun phrase as “identifiable” and locate it accordingly in his/her permanent register or in the register he/she keeps of the actual discourse. “Identifiability,” then, is a mutual category, shared by both speaker and hearer. We may roughly distinguish between three grounds on which identifiability can be based: i) identifiability based on deixis; ii) identifiability based on anaphora (see Sections 2.1.2 and 2.1.3); iii) identifiability based on the (nondeictical) situational context, or, in a wider sense, identifiability based on nonanaphora (and nondeixis) (see Sections 2.1.4 and 2.1.5). With regard to the first basis of identifiability, more will be said when I discuss the expression side of definiteness (see Section 3.3). At this point, it suffices to say

Articles and article systems in some areas of Europe

547

that in the case of “identifiability based on deixis,” the referent of the definite noun phrase is present in the speech situation and can be pointed out. In the following four paragraphs, I present, discuss and – where necessary – further distinguish the other two bases. I shall endeavor to be as concise as possible and use examples from Standard English. 2.1.2. Identifiability based on anaphora The term “anaphoric noun phrases” will be restricted to refer exclusively to definite noun phrases of which the identifiability of the referent is established by linguistic means. Usually, an anaphoric noun phrase requires an antecedent in the verbal pre-text. As to the semantic relationship between an anaphoric noun phrase and its antecedent, further classification is necessary. Compare: (1)

Yesterday we went to a new pub round the corner. The beer was good.

with (2)

Yesterday we wanted to have a beer in a new pub round the corner. But the pub was closed.

Both the beer in (1) and the pub in (2) are definite noun phrases. This allows the hearer or reader to assume the identifiability of the referents. In order to establish a basis for their identifiability, he/she locates the referents in the text-based register of the discourse and links them to possible antecedents. In (1) and (2), the phrase a new pub will be interpreted as the antecedent to both concepts. Thus, both the beer in (1) as well as the pub in (2) are anaphoric noun phrases. But the anaphoric quality differs: the beer in (1) is anaphoric since in certain social or cultural communities, the referent of “beer” is related to the referent of “pub”. Following Hawkins (1984), we may say that “beer” belongs to the “pragmatic set” of “pub”. That is, the introduction of a certain referent into the discourse establishes the identifiability of a whole range of referents that are part of this referent in a cultural or material sense. We may also speak of a “whole-part relationship in a wider sense.” The apposition “in a wider sense” is necessary, since “part” not only refers to inclusive or associated parts of the referent of the antecedent (e.g., page as an inclusive part of book or the color as an associated part of car). “Part” also refers to exclusive parts, that is, parts of which the referent of the antecedent noun phrase establishes a larger set to which in turn the referent of the anaphoric noun phrase belongs. This type of anaphora is discussed in Erkü and Gundel (1987). Consider the following example: (3)

The ant daubs parts of her burden onto a cocoon and passes the rest to a thirsty larva. (Erkü and Gundel 1987: 535)

548

Christoph Schroeder

Here, the rest can successfully be interpreted as having anaphoric reference because the antecedent set parts of her burden establishes a larger set, namely, her burden, of which the rest is a part. Exclusive anaphora also appear to be responsible for the anaphoric interpretation of definite noun phrases denoting time and space concepts: (4)

I never drink vodka. The morning after is said to be horrible.

(5)

I enjoy hiking. The first couple of miles your feet might hurt but after a while you don’t care.

In these examples, the “larger sets” established by the antecedent phrases (drink vodka, enjoy hiking) are the dimensions of time and space of which the anaphoric phrases denote parts. 2.1.3. Relationship of identity As opposed to the anaphoric relationships discussed above, the anaphoric relationship between a new pub and the pub in (2) is not a “whole-part relationship in a wider sense” but a “relationship of identity,” that is, the hearer will interpret the referent of the pub as identical with the referent of a new pub. Note that if I speak of “relationships of identity,” it is not to say that the anaphoric noun phrase has to show complete semantic or even lexical identity with the phrase that is interpreted as its antecedent. It may also be another description for the antecedent which is metaphorically, metonymically, or synonymously connected to the semantics of the antecedent. Thus, even if we replace the pub in (2) with the place – as in (6) below – the relationship between a new pub and the place will still be interpreted as an anaphoric relationship of identity: (6)

Yesterday we wanted to have a beer in a new pub round the corner. But the place was closed.

Similar to whole-part relationships, these other descriptions are dependent on the implications and presuppositions of the specific discourse as well as on cultural conventions. However, the distinction between “whole-part relationships” and “relationships of identity” remains untouched by these considerations: in part-whole relations, the referent of the anaphoric noun phrase is a different (mental) object to the referent of the antecedent. On the other hand, in relationships of identity the (mental) object remains the same, even though the anaphoric noun phrase may only denote aspects of the referent of the antecedent. I will subsequently follow Erkü and Gundel (1987) and call anaphoric part-whole relationships in the wider sense “indirect anaphora,” while anaphoric relationships of identity will be called “direct anaphora.”2

Articles and article systems in some areas of Europe

549

2.1.4. Identifiability and situational context By thus drawing a line between “anaphora” and “identifiability,” we acknowledge that in principle, besides definite noun phrases of which the identifiability of the referent is interpreted as based on anaphora, there must be other bases for identifiability, namely, nonanaphoric – and nondeictic – bases. Consider: (7)

[A married couple is having their evening meal. The man remarks:] The car needs repairing.

The car in (7) is not established by linguistic means, thus it has nonanaphoric reference. Neither is the referent present in the speech situation. The reason why the woman will still locate the referent in her register of identifiable (mental) objects is the fact that at the time and place of speech, only one car can qualify as the referent, namely, the car the two people own. The car, then, has unique reference within a certain situational context – in the example above, the context of the married couple – and its identifiability does not require an establishment by linguistic means, nor does it require the presence of the referent. The referent of the car in (7) is an individual object. In a similar way, the referent of the sun in (8) below is an individual object, and the noun phrase has unique reference. The difference between the car in (7) and the sun in (8) is the fact that the car requires a restricted situational context within which it may have unique reference, while the sun is not very sensitive as to the context in which it may have unique reference, in fact, it has unique reference in almost all times and places of speech – except, probably, science fiction and certain professional contexts of astronomers. (8)

The sun is shining.

Also, noun phrases which refer to institutions may have unique reference, but their unique interpretation may be as restricted as the car in (7) or as wide as the sun in (8). In uttering (9) below to somebody who does not know that I always go to the same post office, I may still feel understood, since my intention is to refer to the institution and not to the individual post office around the corner from my house: (9)

I’m off to the post office.

As an institution, the post office may refer uniquely, even if I and the person I am speaking to merely share the context of living in a country with an institutionalized postal service. As an individual object, the post office has unique reference if we live in the same village and the village only has one post office. The reference properties of the post office in (9) parallel the reference properties of unique natural phenomena like the snow, the forest, the hay, the beach, etc., and instruments in adpositional phrases such as by car, by bike, etc. Apart from the anaphoric interpretation, and also depending on the situational context within which

550

Christoph Schroeder

it is uttered, the beach in (10) below may have two unique interpretations. Within a restricted situational context, it may be interpreted as referring to a beach the speaker and I know because it is the only beach around, or because it is the beach we usually go to. Within a wider situational context, no individual object is referred to but the “natural institution” of beaches. (10)

The best place to spend the day in this weather is the beach.

Thus, unique reference is not limited by the kind of object referred to, but by situational context or the time and place of speech within which it is interpreted. I will use the situational context to distinguish between two types of referents to which unique reference may be made: i) unique referents in a restricted situational context – which are always individual objects; and ii) unique referents in a wider situational context – which may be individual objects, for example, the sun, but also institutions and natural phenomena. 2.1.5. Generic NPs and proper names Two problems arise from our discussions in the previous subsections. The first is whether to interpret generic noun phrases as having unique reference. Consider the dodo: (11)

The dodo is an animal which used to live on Mauritius.

The identifiability of the noun phrase the dodo parallels the identifiability of the noun phrases with unique reference in the examples (7), (8), (9), and (10) in that it does not need to be established by linguistic means in order to be interpreted according to the intentions of the speaker, that is, in the case of example (11), to be interpreted as generic, namely, referring to the class of dodos. Thus, generic noun phrases and unique noun phrases share the (negative) criterion that they have nonanaphoric reference. But there are two important differences between generic noun phrases and unique noun phrases. First of all, generic noun phrases do not require a situational context to be interpreted as such. Secondly, unique noun phrases are interpreted as such on the basis of the combination of the definite coding plus the absence of an antecedent, just as anaphoric noun phrases are interpreted as such on the basis of the combination of the definite coding plus the presence of an antecedent. On the other hand, generic noun phrases do not need to be definite. Instead of (11), we could also have A dodo is an animal : : : and would still receive the generic interpretation. We could even have an indefinite plural Dodos are animals : : : without the generic interpretation being lost. Thus, although it is a necessary requirement of a generic noun phrase to be nonanaphoric at first mention, definiteness is not a necessary requirement for generic interpretation. It seems that for generic noun phrases to be interpreted as such, beside their nonanaphoric status, criteria like word order, the

Articles and article systems in some areas of Europe

551

syntactic role of the phrase, the semantics of the verb, as well as modal and aspectual categories of the verb also play an important role (see Hawkins 1978: 214 ff.; Dik 1989: 143 ff.; and Chesterman 1991: 32 ff. for discussions). The second problem is whether to interpret proper names as uniquely referring. As with uniquely referring noun phrases and generic noun phrases, their identifiability does not require them to be established by linguistic means, thus they are nonanaphoric. However, they differ from unique noun phrases and from generic noun phrases in that the identifiability of their referents is inherent, that is, it does not require a combination of criteria. As has become evident from the discussion, unique noun phrases, generic noun phrases, and proper names share the (negative) property of having nonanaphoric reference. In the following, the notion of “nonanaphoric reference” will be used to cover all three of these referential categories, where applicable.

2.2. Indefiniteness, specificity, referentiality, nonreferentiality Next to “definiteness,” the four other notions which also belong to the “family” of grammatical categories expressing the referential properties of noun phrases, namely, “indefiniteness,” “specificity,” “referentiality,” and “nonreferentiality” have to be defined. “Indefiniteness” will be defined as a grammatical category expressing the unmarked status of a referent of a noun phrase with regard to its identifiability. Note that “unmarked identifiability” does not correspond to “nonreferring,” since indefinite noun phrases refer to a (mental) object which is not (yet) established as identifiable. Also, in the case of “specificity,” the referent is not (yet) established in the (permanent or actual) discourse register, thus identifiability as a mutual category, shared by both speaker and listener, is not at issue. But it is indicated as a category on behalf of the speaker. That is, by marking a noun phrase “specific,” the speaker openly indicates that for him/her the referent is in fact identifiable. When a noun phrase is nonreferring, that is, when it ascribes a quality or a status, it will be called “nonreferential.” This approach to nonreferentiality is fairly strict. It does not include nouns like Eng. church in (12) which, in a language with definite and indefinite articles, do not require an article when used in certain adpositional phrases and which are often analyzed as “nonreferential” (see also Section 4.2 and the discussion therein): (12)

John never goes to church.

Here, these kinds of phrases are understood as definite noun phrases with nonanaphoric reference. Church in (12) may have an “institutional” reading as well as an “object” reading in the same way as the post office in (9) above.

552

Christoph Schroeder

Having thus defined nonreferentiality, it is clear that the notion of “referentiality” embraces definiteness, specificity, and indefiniteness, that is, the three grammatical categories which express that a noun phrase has a referent.

3. Articles 3.1. Means of expression of (in)definiteness In European languages, the contrast between definiteness and indefiniteness and related categories can be expressed by a wide variety of formal devices, both intraas well as crosslinguistically: it may be expressed by syntactic means such as word order and sentence stress (e.g., Russian, cf. Birkenmaier 1979). It may further be expressed by case oppositions, for example, the opposition between nominative/ accusative and partitive case marking in the Balto-Finnic languages, likewise the opposition between nominative/accusative and genitive in Baltic, Slavic, and earlier stages of Germanic. In addition to other functions, these oppositions may also express the contrast between definiteness and indefiniteness (cf. Larsson 1983). Similarly, Armenian, Iranian, and Turkish use the presence or the absence of an accusative case marker in order to distinguish between definite and indefinite direct objects respectively (cf. Bossong 1985), and Hungarian has two different series of verbal agreement suffixes which denote the contrast between definite and indefinite direct objects. A contrast may also be expressed by different series of morphemes which link adjectives or attributes to their head nouns, for example, the Baltic languages use two different series of adjective endings (cf. Section 4.3.1) and Kirmanji has two different series of attributors (the so-called “ezafet”) to express the contrast. Moreover, all languages use demonstrative adjectives to express the category of definiteness with noun phrases which have deictic reference. Furthermore, many European languages have a special morpheme called a “definite article.” Some languages – for example, Fering and Jutish, as well as the Low German dialects of the North Sea area and some of the Rhineland dialects – have two of these morphemes, and at least one language in Europe (Macedonian) has three definite articles to express certain aspects of the category of definiteness. Some languages contrast their definite article(s) partly with an indefinite article, some with the absence of an article (e.g. Macedonian, as already mentioned in Section 1, but also Icelandic, Irish, and Basque). And we already named Kirmanji as an example of a European language which has established an indefinite article, but not a definite one (but see Section 3.3).

Articles and article systems in some areas of Europe

553

3.2. Definite articles Having presented some aspects of the variety of formal devices to express the categories of definiteness and indefiniteness, it is now necessary to define the main subject of this chapter, namely, articles. I will first introduce the notion of definite articles (this section, and Sections 3.3 and 3.4) and subsequently, the notion of indefinite articles (Sections 3.5 to 3.10). Definite articles will be defined as: – a morphological device (free morpheme, enclitic morpheme, or affix) with the primary function to denote the definiteness of noun phrases with anaphoric and/or nonanaphoric reference. The requirement that the definite article be a “morphological device” exempts the term “article” from syntactic means which may express the category of definiteness. Secondly, the requirement that the “primary function” of the definite article be to denote the definiteness of a noun phrase exempts the term “definite article” from case-marking morphemes in languages where case marking is sensitive to the category of definiteness, as well as from attributors, adjectival and verbal agreement morphemes when these have special “definite” series. Thirdly, the “primary function” requirement as well as the functional requirement on the article that it denote the definiteness of noun phrases “with anaphoric and/or nonanaphoric reference” exclude the term “definite article” from demonstratives – at least in those instances where the demonstratives are used with nouns which have deictic reference. But the definition of definite articles does not specify whether article usage should reflect both the anaphoric and nonanaphoric base of identifiability as introduced in Section 2.1. Thus, neither are those morphemes excluded which are restricted to noun phrases with nonanaphoric reference, nor are those morphemes excluded which are restricted only to noun phrases with anaphoric reference. Furthermore, a language may exclude certain noun phrases from the use of articles, for example, articles may not be used with proper names and certain other noun phrases with nonanaphoric reference, as in the English example (12) above. Also the use of the definite article(s) may be further motivated by pragmatic factors (cf., amongst others, the excursion on Low German, Section 4.2). I nevertheless continue to call the relevant morphemes in these cases “articles.”

3.3. Demonstratives and possessives as articles There are, however, certain “borderline cases” that reflect the dynamics within which languages acquire definite articles. One of these cases is demonstrative adjectives that are used to express the definiteness of noun phrases with anaphoric reference – a

554

Christoph Schroeder

common phenomenon in languages both with and without definite articles. When the use of demonstratives is restricted to noun phrases with direct anaphoric reference, and when the respective demonstratives are also used for deictic reference, one does not usually call them “articles” but speaks of an anaphoric use of demonstratives – motivated, as Ehlich (1982) suggests, by the opposing pragmatic “procedures” fulfilled by anaphoric and deictic expressions respectively. While the deictic procedure performed by deictic expressions is “a linguistic instrument for achieving focusing of the hearer’s attention towards a specific item which is part of the respective deictic space” (Ehlich 1982: 325), the anaphoric procedure, fulfilled by anaphoric expressions such as pronouns and (partly) articles, is “a linguistic instrument for having the hearer continue (sustain) a previously established focus towards a specific item on which he had oriented his attention earlier” (Ehlich 1982: 330). The anaphoric use of demonstratives, then, is still “pointing,” but “pointing back within texts” (Ehlich 1982: 333), while direct anaphoric expressions continue their referents. However, while Ehlich’s consideration might help to understand the problem on a theoretical level, it seems that in the process of the emergence of definite articles from demonstratives as well as in the anaphoric use of demonstratives, no clear-cut distinction can be drawn between deixis and anaphora. The emergence of articles from demonstratives starts with “pointing back within texts,” and the anaphoric use of demonstratives is best described as a continuum from deixis to anaphora (cf. Bechert 1993: 21 ff.; Croft 1990: 220; Himmelmann 1997: 23 ff.). Furthermore, there are cases which reflect intermediate stages between demonstratives and articles. These may be cases where the demonstrative-derived, nascent article has no deictic counterpart but may only be used with direct anaphoric noun phrases (cf. the “anaphorical demonstratives” mentioned in Croft 1990: 219). Similarly, the new article may have a deictic counterpart and is only used with direct anaphoric noun phrases but morphologically is treated differently from its demonstrative counterpart. An example for this is Tamian, a Latvian dialect (cf. Section 4.3.4). Another example is provided by dialects of Kirmanji. They may use a postpositive form ha(n)/-a to mark direct anaphoric noun phrases. The same form is also used in combination with preposed demonstratives, where it intensifies the deictic aspect (similar to English “this one here”), cf. Schroeder (2002: 200). In the literary language, only the latter use of the postpositive form is found. Another “borderline case” are possessive markers. When they are used to express the definiteness of noun phrases with both anaphoric and nonanaphoric reference, there is no reason why we should not call them “articles.” Also, when they are only used to express the definiteness of noun phrases with anaphoric reference, we may call them articles, just as the “anaphoric article” in Jutish, Fering, and the Rhineland dialects of Low German is called “article” (see Sections 4.1 and 4.2). The emergence of articles from possessives seems to start with the extensive use of the possessive with noun phrases which have indirect anaphoric reference. Whether the use of the

Articles and article systems in some areas of Europe

555

possessive suffix in Turkish reflects such a “borderline case” will be investigated in more detail in Section 4.6.

3.4. Syntacticization trends of definite articles The use and establishment of definite articles, however, does not end with their anaphoric and nonanaphoric uses as described in the previous sections. Hawkins (1978: 130–149) discusses several uses of definite articles in English under the heading of “unfamiliar uses,” such as first-mention uses in noun phrases with establishing relative clauses, as in (13), and the obligatory use of the definite article with NP complements, as in (14), as well as with genitive and nominal modifiers (the front page of the Guardian, the name Algernon), with superlatives (the first person to sail to America), ordinal numbers (the first), and the co-occurrence of articles with certain identifying adjectives, such as the same : : :, the only : : : (13)

What’s wrong with Bill? Oh, the woman he went out with last night was nasty to him.

(14)

Bill is amazed by the fact that there is so much life on earth.

Himmelmann (1997: 101–108) analyzes these examples within the grammaticalization of the definite article after the establishment of the definitizing use. English has arrived at “the next stage,” so to speak, which is reflected by a stronger syntactization or formalization of article usage. This may further develop into a “specific article,” as described by Greenberg (1978, 1981).

3.5. Indefinite articles “Indefinite article” will be defined as: – a morphological device (free morpheme, enclitic morpheme, or affix) with the primary function to denote the indefiniteness of noun phrases. Again, the requirement that the indefinite article be a “morphological device” excludes the term “indefinite article” from syntactic devices that may express the category of indefiniteness, and the requirement that the indefinite article’s “primary function” be to denote the indefiniteness of a noun phrase excludes the term “indefinite article” from case-marking devices, attributors, and adjectival and verbal agreement morphemes when these have special “indefinite” series. The “primary function” requirement also excludes the term “indefinite article” from quantifiers – and this is where the problem starts. How are we to distinguish the numeral “one” from the indefinite article when in all European languages that express the category of indefiniteness by means of an article, this article is derived from the cardinal numeral

556

Christoph Schroeder

“one” and still has a close connection to the numeral since it may only be used with nouns in the singular which denote countable objects? The most important distinction, of course, is the function: the numeral “one” expresses the fact that the set to which the referent of the noun phrase belongs consists of exactly one entity, while the indefinite article expresses the fact that the noun phrase has a referent – at least when it is used with referential noun phrases. Thus, the numeral denotes an extension of a set, while the indefinite article individuates.3 Most European languages that have an indefinite article have turned the functional difference between the numeral “one” and the indefinite article into formal means of distinction.4 When the article and the numeral are phonologically the same and behave in the same way morphosyntactically, the distinction may be maintained by stress, that is, while the article is unstressed, the numeral is stressed. But morphological, syntactical, and phonological differences also occur. For example in English, the numeral one and the indefinite article a are different morphemes, though of common origin (cf. Christophersen 1939). Furthermore in Turkish, the distinction between the numeral “one” bir and the indefinite article bir is maintained by stress on the numeral as well as by different positions in the noun phrase: while the numeral always precedes the adjective, the unmarked position of the indefinite article is in pre-position of the head of the noun phrase, for example, bìr güzel kız ‘one beautiful girl’ as opposed to güzel bir kız ‘a beautiful girl’. Similarly in Kirmanji, the numeral “one” yek is a preposed free morpheme like all other numerals, and the indefinite article is a suffix -ek.5

3.6. From numeral to indefinite article As is well known from works like Christophersen (1939), Givón (1981), Hopper (1987), and Heine (1997), languages may be classified as to the way they use the indefinite article, or, to be more precise, the way they distinguish in function between the numeral “one” and the indefinite article. I will introduce the classification in the following subsections in order to be able to refer to it in Section 4 whenever the data at hand allows a conclusion. It is possible to distinguish between four stages of a gradual functional split between the numeral “one” and the indefinite article which may occur in a given language. Note that when I say that the indefinite article is used “with indefinite noun phrases,” this only pertains to indefinite noun phrases which fulfill the grammatical and semantic requirements for indefinite article usage, that is, they are singular nouns denoting countable objects.

Articles and article systems in some areas of Europe

557

3.7. Emergent indefinite article In stage 1, the numeral “one” is not used to express indefiniteness. But it may be used as a lexical device to express the specificity of the referent of a noun phrase. Since it appears that this stage is a predecessor for the development of an indefinite article, we will call this type of use of the numeral “one” an “emergent indefinite article.” Russian is a language with an “emergent indefinite article”: the numeral “one” odin (NOM.M) / odna (NOM.F) / odno (NOM.N) may be used to express the “specificity” of a noun phrase and in this use competes with certain adjectives (e.g. kakoj-to ‘a certain’). This function of the numeral “one” is frequently found in headlines of scientific articles, for example, ob odnom sluˇcae palatalizacii ‘of a (certain) occurrence of palatalization’.6

3.8. Pragmatic indefinite article In stage 2, the indefinite article is used with specific noun phrases and also with indefinite noun phrases which have certain pragmatic functions, that is, they are constituents of “neutral descriptions,” that is, sentences representing nothing but new information (cf. Kuno 1972: 298, and the discussion in Wehr 1984: 49 ff.) or they introduce pragmatically salient referents into the discourse for further reference (Boeder 2002). When in a language an indefinite article is exclusively used this way, I will call it a “pragmatic indefinite article.” Bulgarian is a case in point: according to Mayer (1988), the Bulgarian indefinite article edno (NOM.N) / edna (NOM.F) / edin (NOM.M) is obligatory when an indefinite noun phrase is part of a neutral description (examples (15) to (17) from Mayer 1988: 117 f.): (15)

Edn-a žena vervi po ulica-ta. a-NOM.F woman walk(PRES.3SG) along street-the ‘A woman is walking along the street.’

The sentence above could be uttered in reply to the question “What is happening?”. The same sentence in reply to the question “Who is walking along the street?” does not necessarily require the indefinite article with the sentence-initial indefinite noun phrase, since in that case the noun phrase in question represents the comment and not the topic of the utterance (cf. Mayer 1988: 117). Also, the indefinite article in Bulgarian is used in constructions which serve to introduce new referents into the discourse. In these constructions, the noun which represents the new referent typically is in postverbal position:

558 (16)

Christoph Schroeder

kruša-ta ja e risuva-lo edn-o dete. pear-the she AUX sketch-PRET.3SG.N a-NOM.N child ‘A child sketched the pear.’

Again, the indefinite article may be omitted if the noun phrase in question does not introduce a referent but is contrastive and stressed accordingly: (17)

krušata ja e risuvalo DETE. ‘A child (not an adult) sketched the pear.’

3.9. Referential indefinite article In stage 3, the indefinite article is used with all indefinite noun phrases, irrespective of the pragmatic functions they have. When in a language an indefinite article is exclusively used this way, I will call it a “referential indefinite article.” Sasse (1992) adduces numerous examples from Albanian dialects to show that Albanian maintains a clear distinction between referential and nonreferential nouns: the indefinite article një in Albanian is generally used with referential (indefinite) noun phrases as in (18). Also the Albanian equivalents of (16) and (17) would have an indefinite article. But the indefinite article is never combined with nouns in predicative use in the strict sense, cf. (19), nor with complements to verbs like ‘become’ or ‘turn out to be’, cf. (20), with nouns in equative constructions like (21), or with nouns after phrases equivalent to the English ‘a kind of’, cf. (22) (examples (18) to (22) from Sasse [1992: 182 f.], Arnavite, Southern Toskian): (18)

Erdh një djaljë. come(PRET.3SG) a boy ‘A boy came.’

(19)

Koça ja djaljë. Koça is boy ‘Koça is a boy.’

(20)

Duall paljonjeri. come.out(PRET.3SG) rascal ‘He turned out to be a rascal.’

(21)

Ha si ljugat. eat(PRET.3SG) like vampire ‘He eats like a vampire.’

(22)

Çë njeri ja aí. what person is that(M) ‘What kind of a person is that!’

Articles and article systems in some areas of Europe

559

3.10. Extended indefinite article In stage 4, the indefinite article is used with all indefinite noun phrases as well as with nonreferential noun phrases. When in a language an indefinite article is used this way, I will call it an “extended indefinite article.” English is a clear example of a language with an “extended indefinite article,” since the indefinite article is obligatory with all indefinite noun phrases as well as with all nonreferential nouns. Thus, nouns in predicative constructions always have the indefinite article – except, of course, nouns in identificational predications, for example, I am the doctor. Also, German has an “extended indefinite article,” though it behaves differently from English with regard to the use of the article in predicative constructions. That is, in certain status-assigning predications, the indefinite article only appears when the noun is further modified by an adjective, cf. (23) vs. (24): (23)

Er ist Dichter. he is poet ‘He is a poet.’

(24)

Er ist ein bekannter Dichter. he is a well-known poet ‘He is a well-known poet.’

Note that the use of the indefinite article in (23) would not lead to the ungrammaticality of the construction but to a different, evaluative meaning: an utterance like er ist ein Dichter could question the validity of the predication in the sense of “he behaves like a poet but he isn’t really one”, or it could express that the person’s status of being a poet is merely a short-term one or differs from a normal poet’s status in some other way (Kolde 1995: 144 f.).7 This type of opposition between a neutral meaning in the regular use of the indefinite article as a pragmatic, referential, or extended article and an evaluative implication of a deviating predicative use of the indefinite article or the numeral “one” in predications is also found in Finnish (Chesterman 1991: 153, see Section 4.3.4) and Gagauz (see Section 4.6.8), and it is reported for English, Greek, Turkish, Spanish, and Hungarian by Harlig (1986), as well as for Bulgarian by Mayer (1988: 120). The fact that German regularly requires the indefinite article in all types of affective predications (du bist ein Schatz ‘you are a sweetheart’, du bist ein Dummkopf ‘you are a thickhead’) may be seen as having its offspring from this opposition.

560

Christoph Schroeder

4. The survey 4.1. The north: Two definite articles 4.1.1. North Frisian and Low German The North Frisian dialect Western Fering, spoken on two German North Sea islands, Föhr and Amrum, has two definite articles, di (M) / det (F, N), dön (PL), and a (M) / at (F, N), a (PL): (25)

Di hünj hee tus-wark. ‘The dog has toothache.’ (Ebert 1971: 82 f.)

(26)

A hünj hee tus-wark. ‘The dog has toothache.’ (Ebert 1971: 82 f.)

Ebert (1971) describes the functional difference between the two articles as shaped by the opposition between anaphoric and nonanaphoric reference: Both utterances presuppose that the addressee already knows which dog the speaker has in view. But the prerequisites are different in the two cases. [(25)] is an appropriate utterance if the dog has been specified in the preceding verbal context; in this case, the article di has an anaphoric function, it is pointing to the referent within the text. [(26)] presupposes that the dog does not have to be specified because at the time and place of the speech act only one dog is qualified as a referent: the uniqueness of reference is warranted by the speech situation. (Ebert 1971: 83) (transl.: Johannes Bechert) According to Ebert, the nonanaphoric article is used with all types of noun phrases with nonanaphoric reference, including proper names. A parallel opposition can be found in the distribution of the two definite articles in the Low German dialect of Cologne (cf. Himmelmann 1997: 54 f.) and in the dialects of the Rhineland described by Heinrichs (Amern near Krefeld, Heinrichs 1954: 85–103) and Hartmann (Mönchengladbach, Hartmann 1982: 187–207). Dialect of Mönchengladbach (Hartmann 1982: 196 f.):8 (27)

[@t kENk Es am ’jôi:@n@] ‘The child is crying.’ (also: ‘my, our : : : child : : :’)

Articles and article systems in some areas of Europe

(28)

561

[dat kENk Es am ’jôi:@n@] ‘The child is crying (now).’ (also: ‘that child : : :’)

According to Hartmann (1982), the nonanaphoric article [d@ô] (M), /[d@] (F) /[@t]/[t] (N) /[d@] (PL) is used with all types of noun phrases with unique reference as identified in Section 2.1.4. It is also commonly used with noun phrases denoting inalienables as well as with generic noun phrases. And in combination with proper names it “communicates that the person referred to is known in relation to the presupposed context of a village, a family, or a social group, etc.” (Hartmann 1982: 196). When using the other kind of definite article, the unstressed [dE] (M), /[dı] (F) /[dat] (N) /[dı:] (PL) as in example (28), the speaker “refers to a certain child known and identifiable to the hearer by preceding verbal context (anaphoric use) or by a situation common to speaker and hearer (deictic use)” (Hartmann 1982: 196). The anaphoric article may also refer deictically, and it has a pejorative connotation when used with proper names. The opposition between two definite articles is also prevalent in spoken Standard German (and to a lesser extent in written German), though only with (primary) prepositions. Compare (29) with (30): (29)

Hans geht ins Kino. Hans goes in:the cinema ‘Hans goes/is going to the cinema.’

(30)

Hans geht in das Kino. ‘Hans goes/is.going to the cinema.’ (also with a stressed article: ‘: : : to that cinema.’)

In (29), the definite article and the preposition are contracted into one portmanteau form, whereas in (30), they are two distinct forms. As for the opposition between the two forms of the article, Hartmann (1982) writes: Two interpretations can be given for [(29)]: ins Kino may be understood either in a generic or in a specific sense. If used specifically it refers to a special object, presupposing that there is only one cinema within a presupposed context like a small town. Ins Kino is interpreted then as a unicum in relation to a specific context. In opposition to [(29)], [(30)] cannot be understood in a generic interpretation but only in a specific one. [(30)] is anaphorically used if there are corresponding conditions of the [verbal] co-text. (Hartmann 1982: 196) We may add that spoken varieties of German may also use the definite article with proper nouns and prefer the portmanteau forms in combinations of preposition, article, and proper name.

562

Christoph Schroeder

Also, the Low German dialects of the North Sea area distinguish between two forms of the definite article after prepositions. This will be investigated in more detail in Section 4.2 below. 4.1.2. Danish dialects of Jutland Most of the Danish dialects of Jutland seem to have roughly the same distribution of functions between their two definite articles as the North Frisian dialects of Amrum and Föhr, and the Cologne and Rhineland dialects have. Ebert (1971: 194) compares Jutish: (31)

æ smaa kartofler ‘the small potatoes’

vs. (32)

dem smaa kartofler ‘the small potatoes’

with Fering (dialect of Föhr): (33)

a letj eerdaapler ‘the small potatoes’

vs. (34)

dön letj eerdaapler ‘the small potatoes’

(31) and (33) mean ‘the potatoes, which are small’, where the relative clause in the paraphrase is nonrestrictive (nonspecific use, as Ebert puts it), while (32) and (34) can be circumscribed by ‘the small ones of the potatoes’ or ‘those potatoes which are small’ (re-identifying use, in Ebert’s terminology). The definite article in (32) and (34) is the anaphoric variety; the definite article used in (31) and (33) is nonanaphoric. 4.1.3. Standard Danish Other Danish dialects and Standard Danish also have two definite articles, but the principle of their distribution is different, at least in the standard language, and one of the two articles is postpositive, that is, suffixed to the noun. In Standard Danish, the preposed article (the so-called “adjective article”) is used before nouns with adjectival attributes, while the postpositive article occurs in other noun phrases: (35)

kartofl-er-ne potato-PL-the ‘the potatoes’

Articles and article systems in some areas of Europe

563

vs. (36)

de små kartofl-er the small potato-PL ‘the small potatoes’

(36) corresponds to (31) and (32) indiscriminately, while the Jutish counterpart of (35) would be either (37)

æ kartofler ‘the potatoes’ (nonanaphoric, e.g., generic)

or (38)

dem kartofler ‘the potatoes’ (anaphoric)

So we seem to have a complete shift of the criterion of distribution from Jutish to Standard Danish. But if we go into details, we can still detect the anaphoric/ nonanaphoric opposition in the background, as it were, for example, before relative clauses (examples from Ebert 1971: 193): (39)

Kender du pige-n, som sad ved side-n af dig? know you girl-the who sat at side-the of you ‘Do you know the girl who was sitting beside you?’

and also (40)

Kender du den pige, som sad ved side-n af dig? know you the girl who sat at side-the of you ‘Do you know the girl who was sitting beside you?’

but only (41)

Pige-n – som ikke ville med – gik ind i hus-et. girl-the who not wanted with went in into house-the ‘The girl – who did not want to come along – went into the house.’

(39) and (40) include restrictive relative clauses (re-identifying, in Ebert’s terminology); the relative clause in (41) is nonrestrictive, Ebert calls it “appositive.” den in (40) is anaphoric – and perhaps a borderline case between definite article and demonstrative pronoun (which would be stressed: dén pige). The preposed article can also express contrastiveness (Ebert 1971: 193): (42)

Brød-et jeg købte igår var ikke så godt som det brød bread-the I bought yesterday was not so good as the bread jeg købte idag. I bought today. ‘The bread I bought yesterday was not as good as the bread I bought today.’

564

Christoph Schroeder

4.1.4. Swedish, Norwegian, Faroese In Swedish, as well as in both Norwegian languages, Bokmål and Nynorsk, and in Faroese, the following opposition is characteristic (example from Swedish): (43)

stad-en city-the ‘the city/the town’

vs. (44)

den stora stad-en the big city-the ‘the big city’

The principle of distribution is again presence vs. absence of adjectival attributes within the noun phrase, but normally both articles appear in nominals with adjectives, the preposed one before the adjective and the postpositive one suffixed to the noun. The anaphoric/nonanaphoric criterion shows up in various cases. In Faroese, nouns denoting unique concepts require neither post- nor prepositive article (cf. Barnes and Weyhe 1994: 207). And in Swedish, the postpositive article is not used if a noun with an adjective and a prepositive article is followed by a restrictive relative clause – while on the other hand, frozen adjective-noun combinations which refer to unique concepts do not require the prepositive article, for example, vita hus-et ‘the White House’ (in Washington) as opposed to det vita hus-et ‘the white house’ (Perridon 1994: 327). Similar examples are found in Bokmål and Nynorsk, cf. Ebert (1971) (following Lundeby [1965]) and Perridon [1994]). We have seen that the clear distinction between the anaphoric and the nonanaphoric article which shows up in Western Fering is levelled out in the Scandinavian languages to a certain extent. However, dialects exhibit alternating patterns. Perridon notes: [I]n most dialects of Norway and North Sweden, and to a much lesser extent even in some Danish dialects (cf. Reinhammar 1975: 82), singular (proper) names, as well as common names that function as proper names, such as far ‘father’, are provided with a prepositive article which derives from, or is identical with, the personal pronoun han ‘he’ ho ‘she’: ho Kari, han far, han gamle Ola ‘old Ola’. (Perridon 1994: 328) This seems to be a typical phenomenon in areal linguistics: the shift from one organizing principle of a linguistic subsystem to another in geographical space is effective in broad outline but not in each and every detail. Remnants of the “losing” principle can be found in the corners and recesses of the “new” system. They may even be reinforced by way of applying to other resources of the language, such as, in this case, personal pronouns, which develop into (restricted) nonanaphoric articles.

Articles and article systems in some areas of Europe

565

All languages cited in this section have an “extended indefinite article” which is derived from the numeral “one” and parallels German with regard to restrictions in the use with predicative nouns (see Section 3.10).

4.2. Excursus 1: Two definite articles in the Low German dialects of the North Sea region Low German varieties9 in the North Sea region have a free article morpheme (henceforth called the “full form of the definite article”) which is used to express the definiteness of noun phrases with both anaphoric and nonanaphoric reference. In prepositional phrases, however, the dialects distinguish between two series of definite articles:10 (45)

von=’ne Hóchtied / Ja wat schall ik da sonns’ noch votäln? from=the wedding yes what shall I there further else tell Von de Drácht könn’=ik noch wat säig’n, : : : from the costume could=I yet something say ‘about the wedding : : : What else could I say about it? I could say something about the costume.’

In (45), von de Dracht ‘about the costume’ has the full form of the definite article. It is indirectly anaphoric in that it builds up a relationship to the pragmatic set of the antecedent Hochtied. The antecedent, however, is a prepositional phrase which carries the enclitic definite article and has a nominal complement with unique, nonanaphoric reference. The different Low German forms of the enclitic definite article in the North Sea region are: (46)

Dialects of Eastern Frisia: COMM./N.SG ’t PL ’e/’@

(47)

Other dialects: COMM.SG ’e/’@ N.SG ’t PL ’e/’@ A slash (/) indicates individual speaker’s variations. In prepositions with a final /n/, the article form of the feminine singular and of all plural forms is ’ne. The Holstein dialect (north of Hamburg) has kept an oblique masculine ’n (cf. Wahrig-Burfeind 1989: 51).

With nonanaphoric noun phrases after prepositions, the use of the enclitic definite article appears to be the established form of reference, cf. von’ne Hóchtied ‘from the wedding’ in (45).

566

Christoph Schroeder

Nouns without any determiner in prepositional phrases are a lot more frequent than in Standard German. In these phrases, the noun always has unique reference as a location, institution, instrument, or time (i.e., ut Dörp ‘from the village’, to Hochtied ‘to the wedding’, in Köök ‘in the kitchen’, in Kaark ‘in church’, mid Rad ‘by bicycle’, in Winner ‘in winter’). It seems that in prepositional phrases, the distribution between the non-use and the use of the enclitic article with noun phrases which have unique reference is a reflection of the distinction between “wider situational context” and “restricted situational context” (cf. Section 2.1.4). That is, the non-use of the article always coincides with unique reference in a wider situational context. But this point remains to be investigated more thoroughly.11 The forms of the full definite article in prepositional phrases are:12 (48)

COMM.SG N.SG PL

däi, de, d@ dat, de, d@ de, d@

In the following, we will investigate the motivations for the use of the full form in prepositional phrases.13 It will become clear that next to the anaphoric/nonanaphoric distinction and related matters (i.e., deictic use and semantic specification of the noun phrase in question), there is also a pragmatic motivation at work, in that the distribution between the full form and the enclitic form of the article contributes to the hierarchical organization of the information conveyed in the sentence. As stated above, the full form is the means to establish the direct or indirect anaphoric reference of the noun phrase in question. We have seen (45) as an example for indirect anaphoric reference. The following is an example of direct anaphora; a referent (bicycle) is introduced in the first phrase and is anaphorically referred to in the second: (49)

häp mi mien érst aigent Rad doar+för köaft (: : :) mid de have me my first own bicycle for+that bought with the Rád bün ik do’ immer nou’ Wainer nou’=’t Reol+schkil bicycle am I there always to Wainer to=the middle+school ‘I bought my first own bicycle with that [money]. (: : :). With the bicycle I used to go to Wäiner to the middle school.’

Also, the full form is always used when the definite article is used to refer deictically. See (50), where the speaker describes a drawing of traditional wedding costumes which she has displayed in front of her: (50)

Und híer kann man dit uk seer schön säin: Düsse/Düsse and here can one that also very nice see this/this Stráif’n up de Schört. stripes on the shirt ‘And here one can see that very nicely: these stripes on the shirt.’

Articles and article systems in some areas of Europe

567

Furthermore, the full form is used when the noun is modified by an adjective, as in (51), when it is the head of a relative clause, as in (52), or when other appositional specification is added, as in (53): (51)

van dat geáu Land from the good country ‘from the good country’

(52)

doar arbeit’n nu’ häil völ Lü hier up däi Meierwerft, there work now very many people here at the Meierwerft däi däi groude Lúxus+schipp’n baut. which which big luxury+liners builds ‘There are very many people around here working at the Meierdocks, which build big luxury liners.’

(53)

De Kinnä von de Schaul, the children from the school von de K-/ ierst Klass bit no’ de taist Klass from the first class up to the last class ‘the children from school, from the first class to the last class’

Next to the distinction between anaphoric and nonanaphoric reference and related matters, we also find examples which allow us to assume that a pragmatic motivation is at work in the choice of the form of the definite article. There appear to be three pragmatic environments in which the full form of the definite article is preferred to the enclitic form: 1) when the phrase has thematic properties, 2) when the referent is established for further reference, and 3) when the phrase is the informational peak of the sentence, that is, when it is the focus. My notion of “theme” is equivalent to what Chafe (1976) labels “topics, the Chinese style,” that is, the “spatial, temporal or individual framework within which the main predication holds” (Chafe 1976: 50 f.). Like Standard German, Low German has verb-second position and reserves the sentence-initial position for the theme or the topic respectively. When prepositional phrases with definite complements appear in this position, the full form of the definite article is always used. We will interpret this as a correlation between the full form of the definite article and the thematic property of the phrase in question – a correlation which may override the anaphoric/nonanaphoric distinction. Example (54) below provides an example in which both the phrase with the enclitic definite article and the phrase with the full definite article refer uniquely and the use of the different forms of articles can clearly be attributed to the different pragmatic roles of the phrases: in the first sentence, op ’e Wérft (‘at the docks’) is part of the information about the topic mien Vádder (‘my father’). In the sentence following, however, op de Wárft’n provides the thematic frame within which the new referent ’n besonnder Oot von Platt (‘a special variety of patois’) is introduced:

568 (54)

Christoph Schroeder

Mie(n) Vádder häd op=’e Wérft a’beit, he=’s Timmermann my father has on=the dock work he=is carpenter wess’n. – Äh – op de Wárft’n givt dat doch=’n besónnder been at the docks gives that however=a special Oot von Platt. variety of patois ‘My father used to work on the dock, he was a carpenter. – Äh – On the docks there exists a special variety of patois.’

Secondly, we note that definite noun phrases receive the full form of the definite article when the referent is established for further reference, as in (55): (55)

Und óff-/ óffmals ouk, käim ouk vör, and often also happened also (preverb) dat man mid de Butschers, that one with the tramps wee Se weet’n, wat Butschers sind, Bútschers sind : : : if you know what tramps are tramps are ‘And often also, it also happened that with the tramps – if you know what tramps are, well, tramps are : : :’ (Butscher is a special ideolectal expression for ‘tramps’.)

Third, we note that the focal role of the prepositional phrase motivates the use of the full form of the definite article. In order to be able to distinguish “focus” from “comment,” I follow Wehr (1994: 623 ff.) and reserve the notion of focus for nontopical information which is either contrastive in a narrow or a wider sense (“X, not Y”, resp. “X and nobody/nothing else”) or emphatic. Low German follows the pattern of stressing contrastive and emphatic information and, looking at the data, we find that the full form is always used when a constituent of the phrase (preposition, article, or noun) receives contrastive or emphatic stress: (56)

Emphatic stress: doar is=’e MID de Emmer wäx+loup’n there is=she with the bucket away+run ‘there she ran away with the bucket’

(57)

Contrastive stress: ob=’@ Krank’n+kaass’ oder bi d@ INNUNGS+krank’n+kaass up=the sick+fund or at the guild’s+sick+fund ’with the sick fund or with the guild’s sick fund’

Similarly, we find the full form of the definite article in emphatic exclamations like (58). Note that in the second part of (58), the same phrase turns up again, this time with the enclitic definite article and emphatic stress on the modal particle uk ‘also’:

Articles and article systems in some areas of Europe

(58)

569

sowat Ähnliches wie uns plá’dütschen Ob’nd something similar as our patois evening is ouk in Ré@n imma ein groud’n Éltern+Ob’nd west. is also in Ré@n always a big parents+evening been von de SCHAUL ut! (: : :) from the school out Jede Klass häd ehr Schtück bröcht. (: : :) every class has her play brought Und do is’ mindestens tein Prozent von Pla’dütsch and there is at.least ten percent from patois schprickt -äh / spoken pla’dütsches -äh-/-äh- Sketche mockt. (: : :) patois sketch made UK von=’ne Schaul ut! also from=the school out ‘something similar to our patois evening has always been a big parents evening in Rehna. (Organized) by the school! (: : :) Every class used to perform its play. And at least ten percent of that used to be in patois – made sketches in patois. Also (organized) by the school!’

The following is an example of contrast in a wider sense (“X and nobody/nothing else”) motivating the use of the full form: in the first part, an de Kinner ‘to the children’ is part of a sentential predication, a “sentential comment,” so to speak, on the discussion previous to the uttering. In the second part, the phrase turns up again. This time it has the enclitic definite article and is part of the comment in a sentence which has a topic-comment structure with the topic ik (‘I’): (59)

[The speaker talks about her techniques as a schoolteacher of Low German and resumes:] Nur so koomt man ran an de Kinner. only this.way comes one near to the children ‘Only in this way does one get through to the children.’ [A little while later, the speaker talks about the indifference of the parents who do not share her enthusiasm for teaching Low German. Nevertheless, she is determined to go on with her work:] Und kloar, dat is’ trauriech, desweg’n./ and certainly that is sad because.of.that Ja, ik mutt weller an=‘ne Kinner ran, nä? yes I must further to=the children near no ‘And, of course, it is sad, because of that. Yes, I must go on trying to get through to the children, mustn’t I?’

570

Christoph Schroeder

We can conclude this excursion by stating that the distribution of the two definite articles in the Low German dialects of the North Sea area is a means to distinguish between the anaphoric and the nonanaphoric reference of definite nouns. But it goes beyond this in that it also contributes to a hierarchical organization of the information conveyed in the sentence. Thus, the anaphoric article is used with nouns which convey the central information of the proposition, that is, nouns which are going to be topics, which are themes, or which are the informational peak of the sentence. On the other hand, the nonanaphoric article is used with nouns which convey noncentral information, that is, information which forms the background to the main proposition.

4.3. Transitions I 4.3.1. Adjectival endings The standard varieties of the Baltic languages do not have definite articles. To a certain extent, case oppositions may express definiteness and indefiniteness, just as in the Balto-Finnic and the Slavic languages (cf. the introductory remark in Section 3.1). But the Baltic languages employ an additional method for contrasting definiteness and indefiniteness, that is, the use of two different series of adjective endings. This device is not available to the Balto-Finnic languages. In this respect, the Baltic languages belong to an area that includes Germanic languages on the one hand and Slavic languages on the other, again with changing principles of organization of the relevant subsystems. Compare the following examples: Lithuanian: (60)

g˜er-as profèsorius good-NOM.SG.M.INDEF professor ‘a good professor’

vs. (61)

ger-àsis profèsorius good-NOM.SG.M.DEF professor ‘the good professor’

and Latvian: (62)

vs.

lab-s profesors good-NOM.SG.M.INDEF professor ‘a good professor’

Articles and article systems in some areas of Europe

(63)

571

lab-ais profesors good-NOM.SG.M.DEF professor ‘the good professor’

as against German: (64)

ein gut-er Professor INDEF.NOM.SG.M good-NOM.SG.M.INDEF professor ‘a good professor’

vs. (65)

der gut-e Professor DEF.NOM.SG.M good-NOM.SG.M.DEF professor ‘the good professor’

and Danish/Swedish: (66)

en god-Ø professor INDEF good-COMM.SG.INDEF professor ‘a good professor’

vs. (67) a.

b.

Danish: den god-e DEF.COMM.SG good-COMM.SG.DEF ‘the good professor’ Swedish: den god-a DEF.COMM.SG good-COMM.SG.DEF ‘the good professor’

professor professor

professor-n professor-DEF.COMM.SG

In these (and other) Germanic languages, the different series of adjectival endings function as additional marks of definiteness vs. indefiniteness, together with the articles; in the Baltic languages, the oppositions of the adjectival suffixes have to bear the functional load by themselves. 4.3.2. Attributive and predicative forms of adjectives The contrast between “long” and “short” forms of adjectives in Russian seems to be a different matter altogether – synchronically speaking. On the whole, the “long” forms are attributive, the “short” forms are predicative (with certain provisos, e.g., temporariness of the predicated property, occurrence of adjuncts in the predicative construction); that is, the formal opposition denotes a different contrast. Examples:

572

Christoph Schroeder

attributive adjective: (68)

bol’n-ój rebënok ill-NOM.SG.M.ATTR child ‘an ill child / the ill child’

vs. predicative adjective: (69) a.

b.

rebënok bólen child ill(NOM.SG.M.PRED) ‘the child is (temporarily) ill’ rebënok bol’n-ój child ill-NOM.SG.M.ATTR ‘the child is (chronically) ill’

attributive adjective: (70)

úzk-ij plašˇc tight-NOM.SG.M.ATTR raincoat ‘a tight raincoat / the tight raincoat’

vs. predicative adjective: (71) a.

b.

plašˇc mne úzok raincoat to.me tight(NOM.SG.M.PRED) ‘the raincoat is too tight for me’ plašˇc úzk-ij raincoat tight-NOM.SG.M.ATTR ‘the raincoat is tight’

The predicative forms of the adjectives in Baltic and some Germanic languages are as follows: (72)

Lithuanian: profesorius yra ger-as professor is good-NOM.SG.M.INDEF ‘the professor is good’

(73)

Latvian: profesors ir lab-s professor is good-NOM.SG.M.INDEF ‘the professor is good’

(74)

Danish: professor-en er god-Ø professor-DEF.COMM.SG is good-COMM.SG.INDEF ‘the professor is good’

Articles and article systems in some areas of Europe

(75)

Swedish: professor-n är god-Ø professor-DEF.COMM.SG is good-COMM.SG.INDEF ‘the professor is good’

(76)

German: der Professor ist gut-Ø DEF.NOM.SG.M professor is good-PRED ‘the professor is good’

573

The predicative forms of the adjectives coincide with the indefinite forms in Lithuanian, Latvian, and the Scandinavian languages; only in German is the predicative form different from the indefinite forms in that the predicative adjective has a zero suffix (or no suffix at all). So a certain formal agreement seems to hold between predicativity and indefiniteness. Since predicativity implies nonreferentiality, this means that indefiniteness and nonreferentiality can be denoted by the same morphological devices (in certain subsystems of some languages). Furthermore, since referentiality and nonreferentiality belong to the same “family” of categories as definiteness and indefiniteness, there is indeed a functional connection between the two series of adjectival forms in Russian and the pairs (or, in the case of German, triplets) of adjectival forms in Baltic and Germanic: Russian adjectives join in the “chain” of adjectival subsystems of Baltic and Germanic languages, even though definiteness vs. indefiniteness is not expressed in the Russian adjective. Thus, we notice a shifting continuum of organizing principles in the subsystems of articles and adjectives: – definiteness vs. indefiniteness/nonreferentiality in Germanic languages expressed by articles, and – following this – by adjectives (nonreferentiality only by adjectives); – definiteness vs. indefiniteness/nonreferentiality in Baltic expressed by adjectives; – referentiality vs. nonreferentiality in Russian expressed by adjectives. 4.3.3. Baltic languages Thus, the Baltic languages Latvian and Lithuanian constitute a transitional zone between the West European area with articles and an Eastern one without them. The standard languages of this zone have no articles, but they do show an opposition between definite and indefinite adjectives – just as the Germanic languages in Northern and Central Europe do. Moreover, the Western dialect of Latvian, Tamian, shows a kind of preposed definite article (tas/t¯a) before adjectives: the dative and locative singular forms and the dative-instrumental and locative plural forms of the definite adjective in the standard language are replaced, in Tamian, by tas/t¯a plus the adjective – while the adjective

574

Christoph Schroeder

has the form which would have to be called the “indefinite form” in the standard language, in both the corresponding case forms (G¯aters 1977: 86 f.). This combination of a special preposed article (tas/t¯a) with adjectives is reminiscent of the definite article in Scandinavian languages. Compare Standard Latvian with Tamian: (77)

Standard Latvian: uz gald-a ir div-i š¸k¯ıvj-i. on table-GEN.SG.M are two-NOM.PL.M plate-NOM.PL.M Liel-aj¯a š¸k¯ıv-¯ı ir vir-a, big-LOC.SG.M.DEF plate-LOC.SG.M is soup-NOM.SG.F bet maz-aj¯a – ga¸l-a. but small-LOC.SG.M.DEF meat-NOM.SG.F ‘Two plates are on the table. There is soup on the big plate, but meat on the small one.’

(78)

Tamian (Ance): uz gald i dìv š¸k¯ıv. Tài li˜ele sˆ k¸ ˜ıv ve i zu¯p, bet tài maze – gà¸l.

Note that tas as a demonstrative can only combine with definite forms of adjectives, not with indefinite ones, in the standard language (Budi¸na-Lazdi¸na 1968: 47).14 Also, Latvian dialects seem to have an emergent indefinite article viens/viena (which is also the cardinal numeral “one”). See the beginning of a fairytale in nine Latvian dialects and in the standard language. The first sentence reads in Standard Latvian: (79)

Sen-os laik-os k¯ad-am old-LOC.PL.M.INDEF time-LOC.PL.M some-DAT.SG.M vec¯ıt-im bij-is ¸loti old.man-DAT.SG.M been-NOM.SG.M15 very gudr-s clever-NOM.SG.M.INDEF un skaist-s d¯el-s and handsome-NOM.SG.M.INDEF son-NOM.SG.M ‘In the old days, an old man had a very clever and handsome son.’

Instead of the indefinite pronoun k¯ads ‘some’, which presents the specific ‘old man’, all nine dialects show viens ‘one’: (80) a. b. c. d. e. f.

Va´cijûs làikûs vînàm v´ e´ceîšàm b’eis´ c´ îži gudrys i smuks dâls. Vacûs làikûs vînàm vec’eîšàm beji“s dyžàn gudrys i skàisc dâłs. Va´cejûs làíkûs vînàm v´ e´cîšàm biijs c´ îši gud´ris i smuks dâls. Vá:cuós làikuós viénàm vé:cišàm“ biis sti:pri gudris un šmú:ks dáls. Ve˛ c¯ uôs la˜ıkuôs viênàm ve¯c˜ıšàm bìš “vare˛ n gudris un ska˜ısc dê˛ ls. ˆ bijs vare˛ n gudris un smuks dê˛ ls. Ve˛ cuôs la˜ıkuôs viênam ˆ vecîtim

Articles and article systems in some areas of Europe

g. h. i.

575

Ve˛ cuôs la˜ıkuôs viênàm vecišàm bîs dikt gûdrs ùn smuks de˛ ˆls. Ve˛ ces la˜ıkes viênåm ve˛ cåm v˜ıråm bìis varen gudrs un smuks de˛ ˆls. “ m ˚n smuks dê˛ ls. ˆ bîš br˜ınu ˆ gu:drs uˆ Ve˛ c¯ es la˜ıkas viênam ˆ vecišam ˚ (a to d: High Latvian dialects of R¯ezna, Zvirgzdine, Vi¸laka, M¯arciena; e and f: Middle Latvian dialects of Kauguri and Dž¯ukste; g to i: Tamian dialects of Stende, Ance, Vainiži.)

Endzelin (1922: 357) writes: “if we compare the High Latvian fairy tales with those written in Low Latvian, we see that the Low Latvian texts make more frequent use of the indefinite pronoun viêns.16 This may (but must not) be due to German influence. However, viêns has never developed into a real article. German ‘die Eiche ist ein schöner Baum’ [‘the oak is a beautiful kind of tree’] is translated into Latvian as ozols ir skaists kùoks” (transl.: CS). 4.3.4. Balto-Finnic languages Also the neighboring Balto-Finnic languages Estonian and Finnish have “emergent indefinite articles,” üks/yksi (colloquially yks), which also mean “one” as a cardinal numeral (cf. Saareste as quoted by Turunen [1988: 62] for Estonian, and Chesterman [1991: 152 f.] and Dasinger and Küntay [1998] for Finnish). Both compete with adjectives expressing the specificity of the nouns they modify (e.g. Finnish eräs, muuan ‘a certain’). Chesterman also cites an example from unplanned spoken discourse, in which the “emergent indefinite article” is used as an emphatic intensifier in a predicative construction (see Section 3.10 for a discussion): (81)

Sä oot kylla yks idiooti. you(NOM) are yes one(NOM) idiot(NOM) ‘You really are an idiot!’ (Chesterman 1991: 153)

Ultan (1978) and Vogel (1993) cite colloquial Finnish as being on the verge of developing a definite article se (NOM) or plural ne (NOM). The source of this form is exceptional since it is neither a proximate nor a distal demonstrative, but one which “stands for referents which the speaker considers to be in the addressee’s sphere” (Laury 1997: 77).17 However, according to the description Chesterman (1991) gives of se/ne, its attributive use parallels that of a demonstrative, since in determining function it only occurs in noun phrases with direct anaphoric or deictic reference. In the occasional uses in noun phrases with nonanaphoric reference, it rather has to be described as a focus particle: (82)

Kylläpäs se kuu paistaa kauniisti. yes(EMPH) it(NOM) moon(NOM) shines beautifully ‘The moon really does shine beautifully.’ (Chesterman 1991: 150)

576

Christoph Schroeder

4.3.5. Slovene, Serbian, Croatian To the north of the Balkans, Slovene, Serbian, and Croatian constitute a second transitional zone, this time between two areas with articles, the Western area and the Balkans. Standard Slovene and standard Serbian and Croatian have no definite articles. However, Vincenot (1975: 150) notes that in Slovene dialects and in colloquial speech, a preposed definite article tâ/tâ/tô is in use. In the standard language, tâ ‘this’ is a demonstrative. All three languages show an opposition between the long and the short form of the adjective – the long form is used with definite noun phrases, the short form with indefinites, cf. Serbian/Croatian mlâd¯ı bràt ‘the young brother’ as opposed to mlâd bràt ‘a young brother’ (Meillet and Vaillant 1980: 100) and Slovenian nòv klobúk ‘a new hat’ vs. nôvi klobúk ‘the new hat’ (Svane 1958: 51–56). Predicative adjectives have the short form. We find a mention of a Serbo-Croatian indefinite article in Meillet and Vaillant (1980: 100): “Le numératif jèdan ‘un’ soit employé couramment en fonction de l’article indéfini.” The examples provided by Reichenkron (1966: 345) suggest that the preposed jèdan is consistently used to introduce new referents for further reference, thus we may speak of a pragmatic indefinite article. As for Slovene, Vincenot (1975: 150) notes that in the dialects and in colloquial speech, an indefinite article èn/éna/éno is used. Whether it should be classified as “emergent” or “pragmatic” must be left open. All other Slavic languages either show a different opposition of adjectival forms, viz. attributive (long) vs. predicative (short), with the short forms generally dying out, or no opposition at all: Sorbian only has long forms, Bulgarian and Macedonian only have short forms (with some residues of the long forms in the masculine singular). Since attributivity vs. predicativity is connected with referentiality vs. nonreferentiality, there is a functional link between the two varieties of adjectival oppositions within the Balto-Slavic area. On the other hand, the definite/indefinite variety of Slovene, Serbian, and Croatian is geographically connected with the same variety in the Germanic languages of Central and Northern Europe and with Baltic. Serbian and Croatian share the indefinite article with Central and Western Europe, and substandard Slovene shares the preposed definite and indefinite articles with Central and Western Europe – just as Hungarian does with a definite article a/az and an indefinite article egy. The definite article is derived from a distal demonstrative and the indefinite article is a “referential indefinite article” since, according to Stephanides (1974), it exactly follows the patterns of referential indefinite article usage exemplified in Section 3.9 above.

577

Articles and article systems in some areas of Europe

4.4. The Balkan Sprachbund and the south of it 4.4.1. Macedonian To the south of the area of Serbian and Croatian, Macedonian marks the entrance into the area of the Balkan Sprachbund.18 The postpositive definite article shows up, but there are three of them. We will call them “t-form,” “v-form,” and “n-form”: (83) t-form v-form n-form

M.SG -ot -ov -on

F.SG -ta -va -na

N.SG -to -vo -no

M.PL/F.PL -te -ve -ne

N.PL19 -ta -va -na

The forms are suffixed to the first constituent of the noun phrase that is capable of taking these suffixes. Nouns, adjectives, possessive and intensive pronouns, as well as numerals may take an article form. When a first constituent of a noun phrase, say, a demonstrative pronoun, cannot take the article, it is attached to the following member of the noun phrase, cf. óva détovo ‘this child (here)’. The opposition between the three articles is generally described as exhibiting the well-known spatial oppositions of the demonstratives: the t-form is described as “neutral” with respect to deixis, but it may also denote the presence of the object near to the hearer, cf. kniga ‘a book’ – kniga-ta ‘the book’ / ‘that book with you’. The v-form denotes proximity, cf. kniga-va ‘this book here (within my reach)’. The n-form denotes distance, cf. kniga-na ‘that book over there (within our sight)’ (Boji´c and Oschlies 1984: 38). The articles parallel the demonstratives not only in the spatial opposition they denote, but also in the phonological form, cf. cˇ ovek-ot ‘the man’/‘that man near you’ with toj cˇ ovek ‘id’, cˇ ovek-ov ‘this man here (near me/near us)’ with ovoj cˇ ovek ‘id’, cˇ ovek-on ‘that man over there’ with onoj cˇ ovek ‘id’ (Boji´c and Oschlies 1984: 41). None of the articles is used with proper names. Again, we find that grammars of the standard language are not very helpful with regard to the use of the different articles in discourse. Looking at dialects, a clearer, though possibly different, picture emerges. In an investigation into the west Macedonian dialect of Radožda-Vevˇcani, Hendriks (1976) writes: There can be said to be a semantic opposition between the t-form on the one hand and the v- and n-forms on the other, insofar as the former has a strictly identificational use whereas the latter forms always have additional functions besides the identificational function : : : The t-forms occur more frequently than the v- and n-forms taken together. Also the functions of the v- and the n-form may be neutralized in the t-form. It is noted in passing that the use of the v- and n-forms differs distinctly from that in the standard language in which notably the use of the n-form appears to be quite rare. (Hendriks 1976: 210)

578

Christoph Schroeder

The “additional functions” of the v-form and the n-form in the way Hendriks (1976) describes them reveal an article system with certain parallels to the system of two definite articles found in some of the Germanic languages (cf. Section 4.1). In Macedonian dialect texts, the v-form of the article is mainly used in anaphoric and – to a lesser extent – in cataphoric function. Furthermore, the function of denoting spatial proximity is extended to include proximity in the temporal sense (cf. vëtšerov ‘tonight, this evening’)20 as well as a functional equivalent to the possessive pronoun of the first person(s). With nouns designating parts of the (speaker’s) body, as well as with nouns designating locations to which the speaker stands in an (subjectively) inalienable relationship, that is, the home land or the village he/she belongs to, the v-form is consistently used, cf. plëštive ‘my shoulders’, makedoniáva ‘this/our Macedonia’. Hendriks (1976: 216) notes: “the sentence ja tja i ízmia zöbive ’I’ll brush my teeth’, the use of the v- form is obligatory : : : since the reference is to one’s own teeth. The form zöbite would mean (the brushing of) someone else’s teeth.” The n-form on the other hand, besides denoting spatial distance, is used in noun phrases with referents which are unique institutions (as opposed to unique objects, cf. Section 2.1.4). For example, in the sentence toj sákat da šétat po dóždon ‘He likes to walk in the rain’ (Hendriks 1976: 216), dóždon ‘the rain’ refers to the “natural phenomenon” of rain and not to any specific rainfall. Also, the n-form is used in noun phrases with generic reference, cf. Hendriks (1976: 216): “For example, the informant explained to me in the following way for what animals the verb vréskat ‘scream, yell’ can be used: vréskat krávana, kútšeno ’cows, dogs scream’.” 4.4.2. Bulgarian To the west of Macedonian, Bulgarian uses the t-form as the genuine definite article. The morphosyntactic rules of its usage are the same as in Macedonian with the difference that certain phonemic variations of the article occur according to the stem-final phoneme of the constituent that takes the article. Furthermore, masculine nouns can take long and short forms of the article. In written standard Bulgarian, the long and short forms of the masculine article are distributed according to syntactic patterns: while the long form is used with subject noun phrases, predicate nouns, and attributive nouns, the short form is used with noun phrases in all other syntactic roles. The spoken language lacks this distinction (cf. Mayer [1988: 57 ff.] for a detailed analysis). It is a well-known fact that Macedonian and Bulgarian exhibit a number of important differences when it comes to a comparison of the standard languages. In the dialects, however, we are faced with a different situation. The distribution of the article system between Macedonian and Bulgarian is best described not with reference to the standard languages, but by saying that the simple system of the definite article

Articles and article systems in some areas of Europe

579

with only one definite article is found in most of the Bulgarian dialects as well as in the north Macedonian dialect of Kriva Palanka and the Macedonian dialects to the east of the river Vardar. The three-fold system of articles, on the other hand, is found in the central Macedonian dialects to the west of the river Vardar, which formed the basis for the standard language, as well as in the western Macedonian dialects towards the Albanian border. It is also found in Torlakian, a Serbian dialect spoken in the Serbian-Bulgarian border region to the north of Kriva Palanka, as well as in a number of Bulgarian dialects spoken in the Rhodopes, that is, the mountain area in the south of Bulgaria near the Greek border (cf. Stölting 1970: 53).

4.4.3. Albanian, Rumanian In all languages of the Balkan Sprachbund, the definite article is suffixed to the first constituent of the noun phrase, and it is etymologically based on a demonstrative or, in the case of more than one definite article, on demonstratives. Whenever there is only one definite article, this is used to express the definiteness of noun phrases with both anaphoric and nonanaphoric reference. Certain exceptions can be noted in the case of Rumanian and Bulgarian. Both languages do not use their definite article with proper names and in prepositional phrases with nouns that refer to unique institutions like school and work (cf. Mayer [1988: 102] for Bulgarian, in Rumanian also with natural phenomena like forest, cf. Beyrer et al. [1987: 91]). Albanian is noted by Boretzky (1968) to be more consistent than Rumanian in the use of the article in prepositional phrases.21 Also, the Albanian definite article is used with proper names. In a noun phrase consisting of only one element, Albanian, Rumanian, Bulgarian, and Macedonian appear to display more or less the same pattern, though with differences regarding the secondary grammatical information encoded in the article morpheme: (84) a.

b.

(85) a.

Albanian: gur ‘stone’ vajzë ‘girl’

Rumanian: munte ‘mountain’

gur-i stone-DEF.M.SG(NOM) ‘the stone’ vajz-a girl-DEF.F.SG(NOM) ‘the girl’

munte-le mountain-DEF.NOM/ACC.M.SG ‘the mountain’

580

Christoph Schroeder

b.

cod ‘code’

c.

fat˘a ‘girl’

(86) a.

Bulgarian:22 sestra ‘sister’

b.

more ‘sea’

c.

kon ‘horse’

cod-ul code-DEF.NOM/ACC.N.SG ‘the code’ fat-a girl-DEF.NOM/ACC.F.SG ‘the girl’ sestra-ta sister-DEF.F.SG ‘the sister’ more-to sea-DEF.N.SG ‘the sea’ kon-jat (long form) / kon-ja (short form) horse-DEF.M.SG ‘the horse’

The morphosyntax of more complex noun phrases reveals a different situation: while in Bulgarian and Macedonian the word order in the noun phrase is the common Slavic type adjective-noun, Rumanian and Albanian have noun-adjective order, though both languages allow the preposition of adjectives as a device of contrast and emphasis. Another difference lies in the internal organization of the noun phrase. Again, there are more parallels between Rumanian and Albanian than there are between Bulgarian and either of the two other languages. Bulgarian consistently adheres to the adjective-noun order and uses prepositions to indicate the dependencies between the head of a noun phrase and (postposed) nominal attributes. Albanian, on the other hand, uses a linker morpheme in the preposed position of attributive adjectives whatever their position, cf. (87) and (88), as well as in the preposed position of nominal attributes and in the preposed position of adjectives and numerals that are independent nouns, cf. (89). The nominative form of the linker is formally identical with the nominative form of the Albanian definite article: (87)

qytet-i i madh town-DEF.M.SG(NOM) LNK.NOM.M.SG big ‘the big town’ (Hetzer 1996: 72 f.)

(88)

më i madh-i qytet very LNK.NOM.M.SG big-DEF.M.SG town(NOM) ‘the biggest town’ (Hetzer 1996: 72 f.)

Articles and article systems in some areas of Europe

(89)

581

të shtatë LNK.NOM.PL seven(NOM) ‘the seven’ (Hetzer 1996: 72 f.)

Rumanian also makes use of linker morphemes, though they do not have the same shape as the article. With adjectival attributes, the morpheme cel (“articolul adjectival,” I only give the NOM/ACC.M/N.SG form), which etymologically is a contraction of the Latin ecce ille, may turn up as an intensifier (see (90) versus (91)). Furthermore, cel is obligatory in the preposed position of adjectives which are separated from the head noun, as in (92), or which function as independent nouns, cf. (93): (90)

om-ul bun man-DEF.NOM.M.SG good ‘the good man’

(91)

om-ul cel bun man-DEF.NOM.M.SG LNK.NOM.M.SG good ‘the good man’ / ‘the man who is good’

(92)

cod-ul civil s¸i cel code-DEF.NOM/ACC.N.SG civil and LNK.NOM/ACC.N.SG penal penalty ‘the code of civil and penal law’

(93)

cel bun LNK.NOM.M.SG good ‘the good one’

A further linker morpheme, the “articolul posesival-genitival,” with the forms al (M/N.SG), a (FEM.SG), ai (M.PL), and ale (F/N.PL), is obligatory in the preposed position of nominal attributes, cf. (94), as well as with ordinal numbers which function as independent nouns, cf. (95) and (96): (94)

un cal al prieten-ului INDEF.M horse LNK.M.SG friend-DEF.M.GEN.SG ‘a horse of the friend’

(95)

al doilea M/N.SG second ‘the second’

(96)

un al doilea INDEF M/N.SG second ‘a second’

582

Christoph Schroeder

Thus, al and cel have the linking function in common with the respective Albanian morpheme. The difference between Rumanian and Albanian with regard to the linking system is that Albanian requires the linker whenever an attribute turns up, while Rumanian distinguishes between nominal attributes that must be accompanied by a linker and adjectival attributes that only require a linker when they are separated from the head noun. Furthermore, Albanian uses a morpheme in linker function which is partially homonymous with the article morpheme, while Rumanian uses different morphemes for the function of expressing definiteness and the function of linking. The question remains whether the Rumanian and Albanian linker morphemes can be called preposed articles. Stölting (1970: 97 ff.) and Hetzer (1996) clearly show that this is not possible. The linker morphemes have primarily phrase-constituting functions, though certain anaphoric functions (in the case of the combination of the linker and an independent or separated adjective) as well as pragmatic functions (in the case of Rumanian cel) of the linker morphemes can be identified. 4.4.4. Greek To the south of the Balkan Sprachbund, Greek adheres to the continuum of demonstrative-derived articles, this time with a preposed article. The article is used to express the definiteness of noun phrases with both anaphoric and nonanaphoric reference, including proper names – a pattern Greek shares with Albanian (cf. Galton 1973). As to the position of the article, Greek appears to be rather isolated from the languages to the north of it in that it follows the adjective-noun order and places the article in front of the first constituent of the noun phrase: (97)

i ellinik-í ghlóssa DEF.NOM.F.SG Greek-NOM.F.SG language ‘the Greek language’

When the adjective is contrasted, the preferred word order in the Greek noun phrase appears to be noun-adjective. Here, Hamp (1975) and Bechert (1990: 134) observe a parallel between Albanian and Greek with regard to the phrase-constituting function the article can have. That is, in the noun-adjective order, if there is a definite article in the noun phrase, it must be reiterated in the position between the head noun and the adjective: (98)

i kóri i meghál-i DEF.NOM.F.SG daughter DEF.NOM.F.SG elder-NOM.F.SG ‘the elder daughter’ (as opposed to the younger daughter)

This type of construction is also preferred when the adjective forms a more complex phrase (cf. Ruge 1986: 106), for example:

Articles and article systems in some areas of Europe

(99)

583

to spíti to pio oré-o DEF.NOM.N.SG. house DEF.NOM.N.SG more nice-NOM.N.SG ap’ ól-a than all-ACC.N.PL ‘the nicest house of all’

The reiterated article does not contribute to the definiteness of the noun phrase but to its constitution. In this respect, it resembles the Albanian linker morpheme. The parallels are limited, though, since the occurrence of the respective morpheme in Albanian is not dependent on the existence of a definite article in the noun phrase, as is the case in Greek. Furthermore, the Greek article only has a linker function when a marked order of elements in the noun phrase is adhered to, whereas in Albanian the linker has to appear whenever there is an attribute in the noun phrase. But we are faced with a shifting continuum regarding the relationship between articles and linker morphemes from Rumanian, where article and linker are different morphemes, to Albanian, where the linker is homonymous with the article but may appear in definite as well as in indefinite noun phrases, and then to Greek, where the linker is again homonymous with the article but only appears in definite noun phrases with a marked order of constituents.

4.4.5. Article reiteration in Greek and Maltese On the one hand, article reiteration in Greek may be regarded as influenced by the linking function of noun-phrase internal morphemes in the Sprachbund languages. On the other hand, it may be seen as reinforced by Semitic influences, where the noun-adjective order is most common. In Biblical and Modern Hebrew, as well as in Arabic, the preposed article, which is derived from a demonstrative (cf. Anghelescu 1974: 45 for Arabic), is obligatorily reiterated in the position between the head and every following adjective when the noun phrase is definite. Maltese, the most Indo-Europeanized Semitic language, adheres to a slightly different principle, which seems to bridge the gap between Semitic at the other end of the Mediterranean and Greek. First, reiteration of the article is not obligatory. It is rather a matter of restrictive vs. nonrestrictive attribution whether the article is reiterated in front of the adjective. Plank and Moravcsik (1996: 188) state: “Thus, in il-mara t-twila [DEF-woman DEF-tall, ‘the tall woman’] a contrast is likely to be implied to a woman that is not tall, while in il-mara twila [DEF-woman tall, ‘the tall woman’] the tallness of the woman is likely to be part of the addressee’s advance knowledge. Accordingly, adjectives that are inherently contrastive demand re-articulation.” Also, Maltese allows the adjective to precede the noun, in which case the article only appears before the adjective (cf. the discussion of “definite article doubling” in Rijkhoff [1998: 344 f.]).

584

Christoph Schroeder

4.4.6. Indefinite articles As to indefinite articles, all four stages of the distinction between the numeral “one” and the indefinite article can be found across the Balkans. In Rumanian, the indefinite article, un (NOM/ACC.M/N) / o (NOM/ACC.F), is an “extended indefinite article,” since besides expressing indefiniteness, it is also used with nonreferential noun phrases in predicative constructions, for example, castorul este un roz˘ator ‘the bear is a rodent’. Bulgarian, on the other hand, exhibits a “pragmatic indefinite article” (cf. the examples in Section 3.8), while in the Macedonian dialect cited above, the numeral “one” may express specificity in the same way the Russian cardinal numeral “one” does, cf. Hendriks (1976: 208).23 Greek, Albanian, and the Albanian dialects spoken in Greece are clear examples of languages using a “referential indefinite article.” The indefinite article is consistently used with referential nouns in the singular with countable referents, but never in predicative constructions, cf. Sasse (1992) and the examples given in Section 3.9 for Albanian, and Ruge (1986: 123) for Greek. Maltese appears to have an emergent indefinite article. According to Plank and Moravcsik (1996: 184), the use of the numeral wieèed/waèda (M/F) ‘one’ in the introduction of animate referents shows a certain degree of grammaticalization.

4.5. Offshoots The continuum between demonstratives and definite articles is valid for the Western and Balkan area with their offshoots (see below): diachronically as well as synchronically, definite articles and demonstrative pronouns in those areas belong together, much in the same way as indefinite articles seem to be connected with the cardinal number “one” throughout the whole of Europe. Offshoots of the Western and Balkan continua are found in Mordvin and supposedly in Russian dialects. Mordvin is the only Finno-Ugric language which shows a suffixed definite article -s/-(n)t’/-ne. It is derived from a demonstrative (Zaicz 1998: 191; cf. also Collinder 1957: 230). Russian has no definite article, though Russian dialects show suffixed deictic elements (-to and its variants -tu, -ta, -ti, -ot, -et, -m – the vowels tend to harmonize with the preceding word, cf. Leinonen [1998: 74] and the reference given therein). However, they function rather as focus particles than as definite articles.24 Cf. Standard Russian -to in: (100)

zaˇcém ty na krýš-u-to what-for you(2SG.NOM) on roof-ACC.SG-FOC po-léz? PFV-climb:PRET.SG.M

Articles and article systems in some areas of Europe

585

‘(But) why did you start to climb the roof ?’ (Kuznecov 1973: 196) (101)

mné-to oní mog-lí by to.me(DAT.SG)-FOC they can-PRET.PL MODAL PARTICLE vsë rasskazát’. everything tell(PFV.INF) ‘To me they could have told everything.’ (Kuznecov 1973: 196)

The forms are remnants of an article-like use of the (distal) demonstrative tot" which is documented in Russian texts of the seventeenth century. When tot" was used anaphorically, the form occurred in postposition and agreed with the head noun in case, number, and gender. As to the formal and areal distribution of the postpositive forms today, Panzer (1984) summarizes Kuz’mina and Nemˇcenko (1962) as follows: A complete postpositive article which is correctly inflected throughout and used congruently, is today no longer present anywhere in Russian. At most we find a differentiation between gender and number in nominative and accusative singular and plural, while the casus obliquii have at most “harmonized” and neutralized forms. When we consider the regional distribution, it becomes clear that the most varied structures occur in the north-east of Russia, approximately on the line Leningrad – Vladimir – Saratov : : :; the harmonized endings occur in an area to the west of this one, which lies to the east and south-east of Novgorod as far as Kalinin and in an area of south-east Russia between Ryazan-Tula and Saratov : : :; the remaining areas, as far as the line Velikie-Luki – Orël – Kursk, have only the unvaried to. (Panzer 1984: 114) While Kiparsky (1967: 151) suggests Mordvinian influences in the first occurrence of these forms, Leinonen (1998), drawing on the vitality of -to in the Northern dialects, suggests an influence of the determining possessive suffixes of some of the Uralic languages, especially Komi (see the following section).

4.6. The eastern margins of Europe: Possessive suffixes 4.6.1. Uralic The eastern area without articles includes West Slavic, East Slavic, Baltic, and BaltoFinnic standard languages, as well as Standard Slovene and Standard Serbian and Croatian. To the east of that area, we find an area with definite articles or emergent

586

Christoph Schroeder

definite articles, respectively, which have nothing in common with demonstratives. This area comprises most other Uralic languages and, to a certain extent, also Turkic languages. Let us begin with Collinder’s (1957: 251) observation on Mari (Cheremis): “Cheremis has no article, but the Px3sg (or the Px2sg) may have this function” (Px = possessive suffix(es). Similar laconic remarks on a possible article function of possessive suffixes are common in grammatical sketches of Kamas (Kamassian) (Collinder 1957: 494), Khanty (Ostyak) (Collinder 1957: 349), Komi (Rédei 1978), Mansi (Vogul) (Collinder 1957: 322; Keresztes 1998: 411), Nenets (Yurak) (Collinder 1957: 426), Nganasan (Helimski 1998a: 496), Selkup (Ostyak Samoyed) (Collinder 1957: 459; Helimski 1998b: 562; Kuznecova et al. 1980: 187), and Udmurt (Votyak) (Collinder 1957: 276; Csúcs 1998: 285). 4.6.2. Komi Comparatively well documented is the use of the possessive suffix, both second and third person singular, in Komi. From the examples given in Schlachter (1960), Prokuševa (1984), and Leinonen (1998) it appears that the possessive suffixes are common devices for expressing the definiteness of nouns – anaphoric (both direct and indirect) as well as nonanaphoric. As Leinonen (1998) notes, unique referents are usually possessive-marked, cf. šondi-ys (sun-POSS.3SG) ‘the sun’. Indirect anaphoric reference is exemplified in (102), in which the referents of the possessivemarked nouns are members of a previously introduced set: (102)

Ol-isny vyl-isny kyk vok. live-IMPERF.3PL live-IMPERF.3PL two brother Iˇcöt vok-ys semej-nöj, cˇ eljadj-a. small brother-POSS.3SG family-ADJ child-ADJ A ydžyd vok-ys sem’ja-töm, ozyr. but big brother-POSS.3SG family-PRV rich ‘There lived two brothers. The little brother had a family, children. But the big brother was without family, rich.’ (Leinonen 1998: 85)

An example for direct anaphoric reference established by means of the possessive suffix is the following. Note that both possessive suffixes (second and third person) are used. We will discuss their distribution in Section 4.6.3.

Articles and article systems in some areas of Europe

(103)

587

[The scene takes place in a resting-house between the receptionist and a kolchos woman:] Sy-lys’ registratorša-ys kuim-ys’ jual-öma: s/he-ELAT receptionist-POSS.3SG three-ELAT ask-PERF.3SG “F.I.O.?” F.I.O. Kolxoznica-yd vesig kivyl’ oz vart. kolchos.woman-POSS.2SG even reaction NEG.3SG.PRES make Sèki registratorša-yd stöˇcmöd-as: then receptionist-POSS.2SG clarify-FUT.3SG “Familija, imja, otˇcestvo” pö surname first.name patronym QUOT ‘The receptionist had asked her three times: “F.I.O?” The kolchos woman did not even react. Then the receptionist clarified: “Familija, imja, otcˇ estvo?” ’ (Leinonen 1998: 86)

4.6.3. 2nd person and 3rd person possessive suffixes The passages quoted from Collinder mostly mention the possessive suffix of the 3.SG only; the possessive suffix of the 2.SG as a definite article appears in addition to that of the 3.SG in the description of Mari (see above, also Alhoniemi 1993: 75), in Décsy’s (1966: 64, § 5.11) description of Nenets, in descriptions of Nganasan ( Terešˇcenko 1979; Helimski 1998a: 496), and in descriptions of Komi (Rédei 1978; Leinonen 1998; Prokuševa 1984; Schlachter 1960). With regard to Selkup, Collinder (1957: 459) says: “the Px3sg (sometimes even other Px) may function as a definite article.” Where both possessive suffixes have the determining function, their distribution has certain parallels to the two definite articles and their distribution in the Cologne and Rhineland dialects, as well as in North Frisian and Jutish (see Section 4.1), that is, they express aspects of the opposition between anaphoric and nonanaphoric reference. For example, in Nganasan the possessive suffix of the 2.SG has the anaphoric function while the possessive suffix of the 3.SG denotes unique referents of wider situational contexts: (104)

Nuo-zu ‘the sky’ (-zu = POSS.3SG.NOM.SG), m@u-zu ‘the earth’ / kouzu ‘the sun’ / sürü-zu ‘the snow’ (Terešˇcenko 1979: 95)

On the other hand, the short text in Terešˇcenko (1979: 95) demonstrates the anaphoric use of the second person possessive suffix: “The landlord has a daughter (koptua,

588

Christoph Schroeder

literally ’girl’). The girl (koptua-r@ ; r@ =POSS.2SG.NOM.SG, literally: ’your girl’) is very beautiful.” With Komi, the situation appears to be slightly different. Rédei (1978) writes: The 3rd person denotes a certain (emotional) distance between the speaker and/or the hearer and the topic of the speech. For example: ve˛ ras li˛mji˛s si˛le˛ ma nín ’in the forest the snow has already melted’ [lit.: ’in his forest his snow has already melted’ CS]. The sentence simply states a fact. On the other hand ve˛ rad li˛mji˛d si˛le˛ ma nín ’in the forest the snow has already melted’ [lit.: ’in your forest your snow has already melted’ CS], besides stating a fact, also indicates that the speaker and/or the hearer have previously talked about this snow in this forest, or that the speaker and/or the hearer (or both) had been in that forest before they talked about it. Thus they are more related to this forest and this snow than to any other forest or snow. (Rédei 1978: 77) (transl.: CS). On the other hand, judging from the narrative texts cited in Schlachter’s detailed investigation into the possessive suffixes in Komi (Schlachter 1960), it seems that a distribution based on reference – anaphoric (POSS.2SG) vs. nonanaphoric (POSS.3SG), as suggested by Rédei – is only maintained in certain moments. While the possessive suffix of the third person is the unmarked device expressing (anaphoric and nonanaphoric) definiteness, the possessive suffix of the second person takes over whenever there is some sort of a turning point or climax in the narrative. Schlachter (1960: 109) speaks of “subjectivization” (“Subjektive Nuancierung”), that is, in certain moments the speaker transforms the narrative from the sphere of the participants of the narrative into the sphere of the participants of the speech situation. Thus in (103) above, the switch from third to second person possessive suffix prepares the listener for the catch line of the narrative. 4.6.4. Possessive suffixes as emerging articles One difference between Komi and, as we will see in this section, also Mari, on the one hand and, for example, German or English on the other seems to be that possessive suffixes as markers of definiteness are optional in Komi and Mari, but articles as markers of definiteness are obligatory in German and English – if the conditions of their use are met by the text passage in question. In other words, while definite articles are a full-blown category in German and English, they are an emergent category in Komi and Mari. Accordingly, the degree of grammaticalization of the possessive suffix(es) in the coding of definiteness can be expected to vary. See Collinder’s Mari text no. 1 (Collinder 1957: 263–265; the source is given on p. xiv). In English the story runs as follows:

Articles and article systems in some areas of Europe

(105) a.

589

The angel and the devil 1. Once (long ago) the angel and the devil disputed. 2. The devil says to the angel: 3. “Everybody calls me hither and thither. 4. I cannot go to all places. 5. A great many people used to call me. 6. When a baby is born, the mother gives the baby to me.” 7. The angel says to the devil: 8. “How do they call you?” 9. “When the father and the mother get angry at the baby, they say, speaking of the baby: 10. ‘This baby is born by the devil.’ ” 11. The angel says to the devil: 12. “I do not believe you.” 13. – “If you do not believe, I shall show you my book. 14. With me (= in the book) everybody is recorded who has called me.” 15. The angel says to the devil: 16. “I do not give the sinner to you. 17. I shall redeem him myself (and) take his sins away.”

Sentences 1, 2, 6, 7, 9, 10, 11, 15, 16 read in Mari:

(105) b.

1.

2.

6.

7.

ožno godõ-m suksõ dön ia ücaše-n formerly time-ACC angel with devil dispute-PART õl-t. be-3PL.PRS25 ‘Once the angel and the devil disputed.’ ia ojla suksõ-lan: devil say(3SG.PRS) angel-ALL ‘The devil says to the angel:’ “aza šoce-š,” mane-š, “awa-¸se baby be.born-3SG.PRS speak-3SG.PRS mother-POSS.3SG mõ-lam pua” – mane-š – “aza-m.” I-ALL give(3SG.PRS) speak-3SG.PRS baby-ACC ‘When a baby is born, says he, the mother gives the baby to me, says he.’ suksõ ojla ia-lan: angel say(3SG.PRS) devil-ALL ‘The angel says to the devil:’

590

Christoph Schroeder

9.

“awa-že šõdeška,” mane-š, mother-POSS.3SG get.angry(3SG.PRS) speak-3SG.PRS “at’a-že father-POSS.3SG õrwezõ-ž-lan, õrwezõ-m ojla” – boy/girl-POSS.3SG-ALL boy/girl-ACC say(3SG.PRS) mane-š speak-3SG.PRS ‘When the father and the mother get angry at the baby, they say, speaking of the baby:’ 10. “ti õrweze ia-gõc šocõ-n, ojla,” this boy/girl devil-ELAT be.born-PART say(3SG.PRS) mane-š speak-3SG.PRS “‘This baby is born by the devil.”’ 11. and 15. (identical) suksõ-ž ojla ia-lan angel-POSS.3SG say(3SG.PRS) devil-ALL ‘The angel says to the devil:’ 16. “mõ´n” – mane-š – “tõ-lan-et I speak-3SG.PRS thou-ALL-POSS.2SG om NEG:verb(1SG.PRS) pu” – mane-š – sulõkan jeNõ-m give speak-3SG.PRS sinner man-ACC ‘ “I” – says he – “do not give the sinner to you” – says he.’ In sentences 1 and 10, the English definite articles lack the anaphoric function; in 16, the anaphoric function is not very clear. In all other sentences, there is no doubt in this respect. In 1, 2, 7, 10, 16, nothing in the Mari text corresponds to the English articles. In 6, the translation ‘its mother’ is perfectly possible. In 9, such a translation would seem more difficult: ‘ “When its mother gets angry,” says he, “(and) its father, at his/her baby : : :’ ”. But this kind of cross-reference has to be expected; it is also normal in Turkish. The crucial case is suksõ-ž ‘the angel’ in 11 and 15: an interpretation ‘his angel’, that is, the devil’s angel, seems to be out of question. Sentence 11 and 15 both introduce statements in which the angel opposes what the devil says. In sentence 7 on the other hand, ‘the angel’ has no possessive suffix. The sentence introduces a statement in which the angel asks the devil for further explanation of his standpoint. Judging from the example, in Mari, then, the occurrence of the possessive suffix in direct anaphora seems to be combined with the function of contrasting. According to Prokuševa (1984) and Leinonen (1998), this is also the situation in

Articles and article systems in some areas of Europe

591

Komi. However, there appears to be a stronger degree of grammaticalization of the possessive suffixes as a marker of definiteness in Komi than in Mari. 4.6.5. Indefinite articles in Uralic Collinder mentions indefinite articles deriving from the numeral “one” for Nenets (Collinder 1957: 426) and Selkup (Collinder 1957: 459). For Udmurt, the same can be concluded from the texts in Collinder’s survey as well as from Csúcs (1998: 297). The respective examples indicate that the indefinite article function of the numeral “one” in Nenets, Selkup, and Udmurt is restricted to express the specificity of the noun in question, since the indefinite article is not consistently used when new referents are introduced into the discourse for further reference. But consider the second paragraph of a fairytale in Udmurt where the hero of the story is introduced: (106)

odig udmurt syˇcE nuwes-ky myn-em. one Udmurt such forest-ILL go-NARR(3SG) ‘An Udmurt (Votyak) went into such a forest.’ (Collinder 1957: 291)

This seems to be typical for an “emergent indefinite article” on the way to becoming a “pragmatic indefinite article”: only the most prominent referents, namely, those which are going to have a consistent topical role, are introduced by means of this form. 4.6.6. Turkic For Turkic languages, which also have no demonstrative-based definite article,26 it seems less clear whether the possessive suffix of the third person – or any other possessive suffix – is used in that function. In his characterization of the structure of the Turkic languages, Johanson (1998: 51) notes: “In addition to its normal possessive function, a third-person singular possessive suffix may refer to a known entity or, anaphorically, to something preceding it in the discourse.” An article-like function of the possessive suffix is, however, denied categorically: “This limited function does not make it a definite article in the proper sense” (Johanson 1998: 51, cf. also Johanson 1977). Grønbech (1936: 92 ff.), on the other hand, used to call the possessive suffix of the third person an “article.” He gives some examples of the anaphoric function in continuous texts from various Turkic languages (Grønbech 1936: 98 f., § 136) – mostly indirect anaphora, though in some examples the possessive suffix can even be interpreted as establishing a direct anaphoric relationship, cf. the example from Tuvan (Grønbech 1936: 98): pir küskä t¯yp alGaš , : : : küskäzin qara taš arazyNGa köp salGan, ’he found one/a mouse (pir küskä) : : : he buried the mouse (küskä-zin – mouse-POSS.3SG.ACC, ’his mouse’) between black stones’.

592

Christoph Schroeder

Benzing (1940: 255) writes on Chuvash: “Words which are related to each other in a certain way take the possessive suffix [3SG]” (transl.: CS). He presumes an influence from Mari which is geographically connected with Chuvash. The Chuvash examples given in Benzing (1940: 255) again illustrate an extensive use of the possessive suffix of the third person singular in the establishment of indirect anaphora, for example, s´@rtan@ sikseX purat’, a6@l@ š@nzaX purat’ ‘the pike (´s@rtan@ – pikeˇ pike’) springs, the water ˇ (a6@ ˇ l-@ – water-POSS.3SG, ˇ ’his water’) POSS.3SG, ’his ˇ freezes’. For Dolgan, the neighboring Turkic language of Nganasan, also the use of the possessive suffix of the second person in anaphoric function is noted by Stachowski (1998).

4.6.7. Adverbs and adjectives of place and time A common pattern in all Turkic languages is the use of the possessive suffix of the third person with adverbs and adjectives of place and time, for example, Uzbek k˙ecˇ a˙ -si (night-POSS.3SG) ‘at night’, as well as with quantifiers and adjectives in (pro)nominal function, for example, Kirghiz b¯ar-ï. (all-POSS.3SG) ‘all (of it, them)’, Turkish ba¸skası (other-POSS.3SG), ‘the other (one)’.27 Benzing (1940: 266) calls this phenomenon “relativische Anknüpfung” and points out the productivity of this operation in the establishment of anaphoric relationships in the text as well as in partitive relations.28 So it seems that cases in favor of Grønbech’s individual choice of the name “article” for the possessive suffix exist. But this does not mean that he has demonstrated the definite article as a full-blown category (“systematische Rolle” as Johanson [1977: 1199] puts it). We will discuss this question in more detail with regard to the use of the Turkish possessive suffix of the third person singular in Section 4.7.

4.6.8. Indefinite articles in Turkic The use of the cardinal numeral “one” in the function of an indefinite article in the Turkic languages may very roughly be described in the form of a geographical continuum from north or northeast to south or southwest. Yakut does not have an indefinite article, neither do the Turkic languages in the neighborhood of the Uralic languages, namely, Chuvash and Tatar (to speak of the ones of which descriptions and texts were at hand). Kazakh appears to make use of a “pragmatic indefinite article,” that is, the numeral “one” is a means to introduce new referents. See (107), the beginning of a short text about a hungry boy, Qoirat, who finds somebody with a piece of bread and subsequently talks with this person about the bread. In the second line, only the referent which is subsequently going to play an active role in the discourse is marked by means of bir ‘a/one’:

Articles and article systems in some areas of Europe

(107)

593

Qoirat qatty ašyG-yp, šarša-p kel-e Qoirat very hungry-CONV tired-CONV come-CONV žat-sa bir adam tandyr-dan nan šyGar-yp AUX-CONJ(3SG) a man bakery-ABL bread bring.out-CONV žat-yr eken. AUX-PRS COP ‘While Qoirat, hungry and tired, was approaching, a man came out of a bakery with a loaf of bread.’

Karalkapak, Turkmen, Uzbek, Uighur, and, to the west of the Kaspian Sea, Azeri, also use the cardinal “one” in indefinite article function, though I am not sure whether to describe it as a “pragmatic indefinite article” or a “referential indefinite article.” However, the indefinite article in these languages is never used with nouns in predicative constructions. Gagauz makes use of a “referential indefinite article.” All referential (indefinite) singular nouns have the indefinite article, but it is not used in predicative constructions and complements to verbs roughly equivalent to English ‘become’, ‘turn into’, ‘be similar to’, etc., cf. (108), except in cases of emphasis, as in (109) – a phenomenon discussed in Section 3.10 above: (108)

O benzär-di xajyrsyz adam-a. he be.similar-PST(3SG) ill-natured man-DAT ‘He was an ill-natured man.’

(109)

O bir gözel japy-da, zabun insan-žyk-ty. she a nice shape-LOC fine person-DIMIN-PST(3SG) ‘She was a fine person in (such) good shape.’

Turkish systematically distinguishes between the numeral ‘one’ bir and the indefinite article by means of different positions in the noun phrase, cf. the introductory remark in Section 3.5.29 The indefinite article is on the verge of becoming an “extended indefinite article,” that is, apart from its consistent use with referential noun phrases, it may also be used in predications (cf. Schroeder 1999: ch. 3). Furthermore, Turkish has also established a grammatical means to express the category of “specificity” with direct objects, see (110) vs. (111) below. In the first example, the direct object does not receive the accusative suffix and is indefinite, while in (111), the indefinite article in combination with the accusative suffix expresses the specificity of the noun in the role of the direct object (cf. Comrie 1978; Schroeder 1999: 72–75 for discussions): (110)

indefinite direct object: Bir adam gör-dü-m. a man see-PST-1SG ‘I saw a man.’

594 (111)

Christoph Schroeder

specific direct object: Bir adam-ı gör-dü-m. a man-ACC see-PST-1SG ‘I saw a (certain) man.’

4.7. Excursus 2: The possessive suffix 3 SG in Turkish – a definite article?30 To start with, we will distinguish between three domains in which the possessive suffix of the third person singular is used with nouns.31 4.7.1. Agreement First of all, the possessive suffix may be an agreement suffix (APOSS = “agreement possessive”) which is part of the full personal possessive paradigm. We can call it “agreement possessive” since it is obligatorily attached to the head of a noun phrase with a genitive attribute or a possessive pronoun. But the genitive attribute or possessive pronoun may also be left unrepresented. In this case, the possessive can be said to establish an indirect anaphoric relationship to an antecedent. See egzos ‘exhaust’ in (112). It has indirect anaphoric reference on the basis of the agreement possessive, while arabam (‘my car’) serves as the antecedent: (112)

Araba-m bozul-du car-APOSS.1SG break.down-PST(3SG) dün yol-da gider-ken egzos-u yesterday way-LOC go-CONV exhaust.pipe-APOSS.3SG dü¸s-tü. fall-PST(3SG) ‘My car had a breakdown. Yesterday on the road the exhaust pipe fell off.’

It is important to note that the possessive-marked noun may also receive the indefinite article bir: (113)

Araba-m bozul-du car-APOSS.1SG break.down-PST(3SG) dün yol-da gider-ken bir lastı˘g-ı yesterday way-LOC go-CONV a tyre-APOSS.3SG patla-dı. burst-PST(3SG) ‘My car had a breakdown. Yesterday on the road a tire burst.’

Articles and article systems in some areas of Europe

595

a tire in (113) is included in the set of tires which form a part of car and are established as such. As we can see, the establishment of a set by means of indirect anaphoric reference does not lead to the establishment of each member of this set, but only to the establishment of the set as such. The same holds for partitive reference. But recall that in Section 2.2 we defined “specificity” as a grammatical category expressing that the speaker classifies the referent of the noun phrase as “identifiable,” though, since the referent is not established in the register of the hearer, this identifiability is not expressed as such. Similarly, the referents of nouns which are possessive-marked and have the indefinite article can be interpreted as not established but nevertheless identifiable on the basis of their inclusive relation to an established set. Possessive-marked nouns with the indefinite article are, then, “specific.” 4.7.2. Relational possessive The second domain of the use of the possessive suffix 3SG in Turkish has already been pointed out with regard to other Turkic languages (cf. Section 4.6.7). Only the possessive suffix of the third person singular may be suffixed to quantifiers, adjectives, and adverbs of place and time. Since in Turkish the attributes of these forms – if represented – may be ablative-marked and in certain cases also bare, we cannot generally speak of an “agreeing possessive” in this case. I will call this use of the possessive suffix “relational possessive” (RPOSS) – a term which is partly adopted from Benzing’s (1940) term “relativische Anknüpfung” (cf. Section 4.6.7). Adverbs of place and time must receive the relational possessive when they are subject or object arguments (see the example of dı¸sarısı ‘the outside’ in (118) below). Furthermore, adverbs of place and time in adverbial function may optionally receive the relational possessive in order to relate the setting of the proposition which they modify to an antecedent. Cf. (114), where possessive-marking on the adverb önce ‘before’ relates the action modified by the adverb to the time of the action expressed in the complement noterde tasdik etmemizi: (114)

Ö˘grenci ka˘gıt-lar-ımız-ı noter-de student document-PL-APOSS.1PL-ACC notary-LOC tasdik et-me-miz-i ve daha confirmation do-INF-POSS.1PL-ACC and more önce-sin-de before-RPOSS-LOC tercüme et-tir-me-miz-i söyle-di. translation do-CAUS-INF-POSS.1PL-ACC say-PST(3SG) ‘She said that we should have our student documents certified by a notary and that we should have them translated beforehand.’

596

Christoph Schroeder

Adjectives and quantifiers obligatorily receive the relational possessive when they are heads of partitive phrases: (115) a.

elma-lar-dan güzel-ler-i apple-PL-ABL nice-PL-RPOSS ‘the nice ones of the apples’ b. *elmalardan güzeller

(116) a.

arkada¸s-lar-dan be¸s-i friend-PL-ABL five-RPOSS ‘five of the friends’ b. *arkada¸slardan be¸s

4.7.3. Compound marker Thirdly, the possessive suffix of the third person singular is used as a compound marker (CP) when two or more nouns form a compound (see van Schaaik 1992 for details): (117)

çevre örgüt-ü environment organization-CP ‘environmental organization’

This latter use of the possessive suffix has to be distinguished sharply from the other two uses since the compound-marking function of the possessive 3SG does not add anything to the anaphoric properties of the compound noun phrase. It is a pure phrase-constituting device. This can be shown by the behavior of the accusative suffix in Turkish in combination with possessive-marked nouns. As already mentioned, Turkish is a language with “differential object marking,” a term adopted from Bossong (1985). That is, the use of the accusative suffix depends upon the referential properties of the direct object. Definite and specific direct objects always have the accusative suffix, while indefinite objects do not (cf. Johanson 1977 and Nilsson 1985 for details). Direct objects which have the compound possessive are not accusativemarked as long as their definiteness or specificity is not ensured by other means. On the other hand, direct objects which have the possessive suffix on the basis of one of the other two uses (APOSS, RPOSS) always require the accusative suffix. 4.7.4. Domains of use of possessive suffixes In the first two uses of the possessive suffix, however, the extent to which the possessive suffix is used appears to encompass all domains of “indirect anaphoric reference” as defined in Section 2.1.2. See the following example from spoken discourse: (118)

1:

s¸imdilik içer-de be¸s ki¸si çalı¸s-ıyor-uz at.present inside-LOC five person work-PROGR-1PL

Articles and article systems in some areas of Europe

2:

1:

597

atölye-ler-e da˘gıt-ıyor-uz workshop-PL-DAT distribute-PRS-1PL biz yalnız kesim bölüm-ü-yüz we only cutting department-CP-1PL dikim-in-i de s¸ey-de sewing-APOSS.3SG-ACC also thing-LOC yap-tır-ıyor-sunuz make-CAUS-PROGR-2PL idare ed-iyor-uz tabii organization do-PROGR-1PL certainly dı¸sarı-sı da bizim outside-RPOSS also our 1: ‘At present there are five of us working inside. We distribute (the work) to the workshops. We are only the cutting department.’ 2: ‘So you also have the sewing done in the whatsit.’ 1: ‘We organize that, certainly. What is outside (the outside) also belongs to us.’

The antecedent of the APOSS and RPOSS-marked constituents is the discourse topic, which is not always easy to define but appears to be rather “fuzzy” in unplanned discourse. In the example above, both dı¸sarısı ‘the outside’ as well as dikimini ‘the sewing’ relate to the work/the workplace of the speaker, a tailor shop. dikimini is an example of an indirect anaphoric noun phrase where the referent stands in an inclusive relationship to the antecedent, that is, forms part of the “pragmatic set” of the antecedent. With dı¸sarısı in (118), as well as with öncesinde ‘beforehand/before (that)’ in (114), the antecedent serves as the location of which the possessive-marked constituent denotes a part. In Section 2.1.2, this type of indirect anaphoric relationships is called “exclusive (indirect) anaphora.” 4.7.5. Categorial status of possessive suffixes Examples of the extensive use of the possessive suffix in creating indirect anaphoric relationships are numerous in spoken Turkish discourse. But the anaphoric relationship created by means of the possessive suffix may never be direct, neither are nouns with unique reference marked with the possessive suffix, as we have seen in some Uralic languages in Section 4.6. Could we then, perhaps, call the possessive suffix in Turkish an “indirect article,” that is, a marker of definiteness reserved for nominals which are definite on the basis of their indirect anaphoric reference? This, too, has to be denied. First, as we have seen, possessive-marking on a noun may be combined with the indefinite article, thereby rendering the possessive-marked noun specific. Second, there remains a fundamental semantic difference between an article – be it a

598

Christoph Schroeder

“specific” or a definite article – and the possessive suffix in Turkish: by means of an article, a certain status of referentiality – definiteness, specificity, or indefiniteness – is indicated. The type of anaphoric relationship, however, is left to the hearer – though the article often gives some “hints,” cf. the article systems in Macedonian, Jutish, etc. The possessive suffix in Turkish, on the other hand, establishes a certain type of anaphoric relationship and leaves the status of referentiality – definite or specific – to the interpretation of the hearer, this interpretation being based on the hearer’s world and discourse knowledge. Thus, it is not the grammatical categories of “definiteness” and “specificity” as such which are expressed by the possessive suffix. Rather, it establishes the indirect anaphoric relationships which lead to their interpretation. In the case of Turkish, then, Johanson (1977, 1998) has to be given credit. The possessive suffix is not an article. More empirical work on the Turkic languages in the neighborhood of Uralic languages is needed in the search for possible transitions between the use of the possessive suffix in definitizing function, as in Komi and Mari, and its use as a marker of indirect anaphoric reference, as in Turkish.

5. Conclusions 5.1. Functions of definite articles The primary function of definite articles is to express the grammatical category of definiteness. But definite articles not only indicate the identifiability of the referent of a noun phrase. In many languages that have definite articles, either the articles themselves or the distribution of their use also indicate where the referent is to be located. Some languages contrast their definite article with no article, for example, English, Bulgarian, and Rumanian. Here, the contrast between the use of the article and the non-use of the article merely indicates the contrast between one type of nonanaphoric reference, namely, proper names, and all other types of anaphoric and nonanaphoric reference. In these languages, sometimes also another type of nonanaphoric reference, namely, unique reference, may be indicated by the non-use of the article, mostly in prepositional phrases and typically with nouns which refer to unique institutions or natural phenomena (cf. Section 4.4.3). Also, a typical phenomenon in languages which have a demonstrative-derived definite article is the non-use of the article in frozen phrases, for example, binominals (e.g. face to face), which are also nonanaphoric (cf. Himmelmann 1998 and the respective remarks in Note 11). The distribution of the two definite articles in Jutish, Fering, and the Rhineland and Cologne dialects of Low German maintains a distinction between anaphoric and

Articles and article systems in some areas of Europe

599

nonanaphoric reference. Remnants of this distinction can be found in Standard German and the Low German dialects of the North Sea coastal area, where the same opposition of two articles only exists with nouns after prepositions; also, some remnants can be found in Danish and the Germanic languages of Scandinavia, though these languages primarily distribute their two definite articles between different types of noun phrases and more or less follow the English, Bulgarian, and Rumanian pattern of article usage. A different case is the distribution of three definite articles in Macedonian. One of the articles, namely, the “neutral form,” may be used to express the category of definiteness regardless of the location or the identifiability of the referent – except, again, that this article is not used with proper names. The other two articles maintain a more specific distinction with regard to the location of the referent, except that here the line is not drawn between anaphoric and nonanaphoric reference, but between anaphoric reference and unique reference of a restricted situational context on the one hand, and unique reference of a wider situational context as well as generic reference on the other. The formal means by which this pattern is construed reflects the path from which it emerged. While the “proximate article” indicates that the identifiability of the referent of the noun phrase in question is actually established between speaker and hearer, that is, established “here,” regardless of a distinction between text and situational context, the “distal article” indicates identifiability on the basis of a situational context which is wider than the context of the actual speech situation. The use of the two possessive suffixes in Komi seems to follow the Macedonian pattern (if, for a moment, we exclude the “neutral” article in Macedonian). The formal means by which the pattern in Komi is construed again reflects the path from which it emerged: while the possessive suffix of the second person singular indicates that the identifiability of the referent of the noun phrase in question is actually established between speaker and hearer, regardless of a distinction between text and situational context, the possessive suffix of the third person, that is, “the person who is not present,” indicates identifiability on the basis of a situational context which is wider than the context of the actual speech situation. As for Nganasan and the other Uralic languages which use both possessive suffixes to express the category of definiteness, it is not quite clear whether they adhere to the Jutish/Fering pattern of anaphoric vs. nonanaphoric reference or to the Komi/Macedonian pattern.

5.2. Continua of development of articles We have observed three continua across European languages: from the numeral “one” to the indefinite article, from a demonstrative pronoun to a definite article (or

600

Christoph Schroeder

to a focus particle), and from the possessive suffixes of the second and third persons to a definite article. It has not been the aim of this study to discuss the etymological development of these continua and the path of grammaticalization inherent in them (cf. Greenberg 1978, 1981, 1985; Ultan 1978; Vogel 1993; and Himmelmann 1997 on the development of definite articles, and Givón 1981; Christophersen 1939 on indefinite articles), but I will try to summarize them and add some hypothetical considerations as to how they can be explained. 5.2.1. Continuum 1: Indefinite articles The starting point for the indefinite article is the pragmatic function of introducing a referent. In order to be introduced, the element in question has to be characterized as an individual since the prototypical referent, in the European languages, seems to be a living being or a concrete countable thing, and a single element at that. Countability and singleness are properties of noun phrases which are characterized by the cardinal number “one”.32 The prerequisite for the pragmatic function is the use of the numeral “one” to express the category of “specificity,” which indicates that the speaker has previous knowledge about the referent which he is introducing. In the languages which have an “extended indefinite article,” that is, the languages of Northern and Western Europe as well as Rumanian, these functions are combined with others. There, the indefinite article has evolved into a marker of all nondefinite, singular count nouns. In the languages which have a “referential indefinite article,” namely, Hungarian, Albanian, Greek, Gagauz, and Turkish, the indefinite article contributes to a distinction between referential and nonreferential nouns, thus the pragmatic function is combined with a generalized means of indicating the nonidentifiability of the referent of a noun phrase. Turkish, however, seems to be developing an “extended indefinite article.” The variety of uses of the numeral “one” in the other languages ranges from neither function, for example, Komi and Yakut, to the specificity-marking function of the numeral without the pragmatic function, for example, Russian, further to the use of that specificity-marker in the introduction of the most prominent entities, that is, discourse topics, as in Udmurt and, finally, to a systematization of the pragmatic function, as in Bulgarian, which allows us to speak of an “indefinite article.” Also Kazakh seems to be progressing towards the use of a “pragmatic indefinite article.” 5.2.2. Continuum 2: Definite articles The second and the third continua can be explained by the hypothesis that the prototypical function of the definite article is anaphoric, that is, it has either deictic

Articles and article systems in some areas of Europe

601

function within the text (direct anaphoric use) or it expresses part-whole relationships within the text (indirect anaphoric use) and from there, the use of the article may spread to nonanaphoric uses with the distinctions described above. Demonstrative pronouns are deictic, so they can develop into definite articles – or into focus particles. The development of demonstratives into focus particles, as in Russian, is only one reflection of the fact that deixis always seems to play a role in the hierarchization of the information conveyed in the sentence. Another reflection is the observation made with regard to the distribution of articles in the Low German dialects of the North Sea area (cf. Section 4.2): the anaphoric article, that is, the article which is more deictic in contrast to the other, is used with nouns which convey central information of the proposition, while the nonanaphoric article is used with nouns which convey noncentral information. Thus, there is a general relationship between “pointing” and “importance” which may override the distribution of articles on the basis of reference properties. 5.2.3. Continuum 3: Possessive suffixes The third continuum exploits the fact that part-whole relationships constitute the basis for indirect anaphora. The prototypical function of the possessive suffix in the languages which use it in the establishment of indirect anaphora, and also in the generalized function of expressing definiteness, seems to be the establishment of partitive relationships, part-whole relationships, and relationships of association rather than the expression of ownership (cf. Schroeder 1999: 186); so the possessive suffix can be used as a marker of anaphora. Thus, a special, rather frequently used way of relating two terms to each other develops into a generalized means of relating terms within the text to each other immediately (third person) or by relating them to the partner in the speech situation (second person). However, in this continuum, the end only seems to be reached in some of the Uralic languages of the Eastern margins of Europe, while in the Turkic languages the possessive-derived definite article remains a category in statu nascendi or, as shown with regard to Turkish in Section 4.7, the possessive suffix remains restricted to the function of establishing indirect anaphora.

Notes 1.

I am grateful for the help of Wolfgang Wildgen, Thomas Stolz, Nikolaus Himmelmann, Heinz Vater, Winfried Boeder, Wilfried Stölting, Armin Hetzer, Marja Leinonen, Rogier Nieuweboer, Vicki May, Rajna Damjanova, Lars Johanson, Bernard Comrie, Georg Bossong, Giuliano Bernini, and the other members of our lively group. Thomas Stolz, Nikolaus Himmelmann, Heinz Vater, Winfried Boeder and Kari Fraurud, as well as Bernard Comrie, Georg Bossong, and Giuliano Bernini, went through the whole

602

2.

3. 4.

5.

6. 7.

8. 9.

Christoph Schroeder manuscript and pointed out some important shortcomings. Rogier Nieuweboer advised me with regard to Section 4.3, and Wilfried Stölting as well as Armin Hetzer with regard to Section 4.4; Marja Leinonen shared her ideas about Komi with me, Rajna Damjanova provided Bulgarian data, and to Lars Johanson I owe Note 26. Of course the full responsibility for shortcomings and errors rests on me. Also, I would like to thank the German Research Association (Deutsche Forschungsgemeinschaft) for a three-year grant for the research project “Pragmatik und Syntax” which enabled Magdalene Döscher, Renate Wahrig-Burfeind, and myself to “scratch the surface” of a fascinating aspect of human language. It was Johannes Bechert who started the project. After his death, his role was taken over by Wolfgang Wildgen, who always found the time to discuss my ideas with me. Note, though, that the approach pursued here differs from the approach in Erkü and Gundel (1987) in that here, the term “indirect anaphora” is restricted to “inclusive,” “exclusive,” and “associated” part-whole relationships, while in Erkü and Gundel (1987), also “created anaphora,” that is, anaphora with a non-nominal antecedent, is classified as indirect anaphora (cf. Schroeder 1999: 12 for further details). See Hawkins (1991) for elaborate discussions. But see Vater (2002), who defends thye standpoint that the indefinite article (at least in German) should be called a ‘quantifier’. Of course we also have distributional properties at hand to distinguish between the indefinite article and the numeral: while indefinite articles may not be combined with definite articles and demonstratives in one noun phrase, the numeral “one” may occur together with means to express the definiteness of noun phrases. Distributional properties, however, are secondary, since they can only be investigated once a clear formal distinction between the indefinite article and the numeral “one” is maintained in a language. Bedir Khan and Lescott (1986: 191) claim that the suffix may also function as a numeral, even in compound numbers, but the Kirmanji speakers I asked rejected the examples given therein. Similarly in Czech: O jednom typu jazykových kontakt˚u distanˇcních ’of a (certain) type of language contacts from a distance’. Eichhoff (1993) points out that in this respect, the use of the indefinite article in Ich bin ein Berliner (‘I am a Berliner’), the famous utterance of John F. Kennedy during his visit to Berlin in 1961, is not an interferential error. Rather, Kennedy expressed that his status of being a Berliner was not one acquired through birth and residence, that is, that it differed from the “normal” status in that it was acquired through solidarity and affection. The examples are given in Hartmann’s own phonetic transcriptions. The dialects at issue are the northern Low German dialects (Nordniederdeutsch) spoken in the region of Dithmarschen in the state of Schleswig-Holstein (north of Hamburg), and the northwest of the state of Lower Saxony (Holsteinisches Niederdeutsch, Nordniedersächsisch, Oldenburgisches Niederdeutsch, Ostfriesisches Niederdeutsch, Emsländisch). The dialects of Eastern Frisia (Ostfriesisches Niederdeutsch) show significant lexical, phonological, and morphological differences from the other dialects –

Articles and article systems in some areas of Europe

10.

11.

12.

13.

14.

15. 16.

17.

18.

19.

603

presumably a Frisian substratum, cf. Foerste (1954). The data on which this section is based was tape-recorded and transcribed by Magdalene Döscher and preselected jointly by Magdalene Döscher, Renate Wahrig-Burfeind, and myself. The transcription of the examples is based on Ehlich and Rehbein’s “Halbinterpretative Arbeitstranskription” (HIAT; Ehlich and Rehbein 1976: 21–41). A slash (“/”) marks a turn-internal interruption. Standardized dialect orthographies are used where available, though strong variations are noted. “=” represents a clitic construction. Within the framework of our research project “Pragmatics and Syntax,” Renate WahrigBurfeind also prepared an investigation into the relationship between pronominalization, the continuum of deixis, and topicality in Low German (cf. Wahrig-Burfeind 1998). To a great extent, her paper shaped my ideas as to how to interpret the data. Errors and misinterpretations are, of course, to be blamed on me. But see Himmelmann (1998), where the openness of a language for permitting articles in the adpositional phrase is seen as a criterion for the degree of the grammaticalization of the article in the language in question. Adpositional phrases, as well frozen phrases like, for example, binominals (e.g. face to face), form “remnants,” so to speak, of less grammaticalized periods of article usage. In the course of the grammaticalization of the article, its use “intrudes” to varying degrees into adpositional phrases. According to Wahrig-Burfeind (1998), the forms represent a deictic continuum in the text which runs from the left (stronger deixis) to the right (weaker deixis). But WahrigBurfeind fails to show what motivates the choice of forms. It should be mentioned, though, that the regular occurrence of the full article form in frozen phrases (cf. op de Räich kräich’n [up the row to get] ‘to get along’) is neglected here. Incidentally, tas/t¯a, pl. tie/t¯as, compete with the personal pronouns of the 3rd person, vi¸ns¸ /vi¸na, pl. vi¸ni/vi¸nas (Veksler and Jurik 1975: 172); cf. German der/die/das vs. er/sie/es (pl. die vs. sie) as personal pronouns (Bethke 1990) and the system of personal pronouns in Danish with its suppletive forms den/det/de (sg. of common gender/ntr. sg./pl.). beside han/hun (m.sg./f.sg.) and so on. “Modus relativus”; characterizes an account given from hearsay. “High Latvian (“Infläntisch”) = the High Latvian dialects of Rositten, Ludsen, and Dünaburg, namely, R¯ezekne, Ludza, Daugavpils (Endzelin 1922: 4); “Low Latvian” (“Niederlettisch”) comprises the Middle Latvian and Tamian dialects. Note, however, that also as a demonstrative, se may be called the neutral form as opposed to the other two – and the most often used demonstrative, at this (Nieuweboer pers. comm.). As for the concept of the Sprachbund, I follow the traditional approach expressed by Schaller (1975: 100). It acknowledges that the core languages of the Sprachbund are Albanian, Macedonian, Bulgarian, and Rumanian. While it is clear that no sharp lines can be drawn, and that Serbian, Croatian, Romani, Greek, and Turkish exhibit a number of “Balkanisms,” it remains a fact that the first-mentioned languages still share the most features (cf., among others, the overview given in Joseph 1992). The neuter form of the plural is also used for collective plurality (cf. Boji´c and Oschlies

604

20.

21. 22. 23. 24.

25. 26.

27. 28.

29.

30. 31.

32.

Christoph Schroeder 1984: 25). Here, the -v-form of the article has taken over the deictic function from an older deictic element -si (short /i/), cf. Serbo-Croatian veˇceras ‘tonight’, danas ‘today’ (Stölting pers. comm.). See also Himmelmann (1998: 326 ff.) for a contrastive investigation of article use in Rumanian and Albanian prepositional phrases. Since the system of Bulgarian parallels the system of Macedonian with regard to the morphosyntax of the noun phrase, I do not add examples from Macedonian. Standard grammars of Macedonian do not mention an indefinite article. While Kiparsky (1967: 151) calls them “adversative markers,” Leinonen (1998: 83) subsumes the functions of the forms under “prominence through accessibility,” that is, contrastiveness, emphasis, and the anchoring of possessed items to the possessor. ücašen õlt forms a periphrastic preterite. Calques from Hebrew h¯a in older translations of religious texts from Hebrew into Turkic languages are, of course, a different matter. In such texts, Karaim uses a preposed definite article (Trakai ol, Haliˇc osol), cf. Pritsak (1959: 31). The Codex Comanicus, a Kipchak text from the fourteenth century, shows similar uses of the personal pronoun of the third person singular ol, cf. Drimba (1973). For further examples, see Benzing (1940) on Chuvash, Menges (1959) on Tuvinian (Soyut), and von Gabain (1945: 84) as well as Boeschoten (1998: 369) on Uzbek. A direct outcome of this use of the possessive suffix is, for example, the development of the relative pronoun xangisi ‘which’ in written Gagauz. It has developed from xangi (hangi) plus the possessive suffix. In Turkish as well as in other Southern Turkic languages, hangi is the unspecific interrogative ‘which’, while in hangi-si (whichPOSS.3SG) the possessive suffix relates the reference to an antecedent superset ‘which of them’. Uzbek also allows the word order adjective–indefinite article–noun, but according to Sjoberg (1963: 40) this is rather used in order to contrast the meaning of the adjective and not in order to distinguish between the numeral “one” and the indefinite article. But the phenomenon in Uzbek may indicate where the change of position starts. This section is based on chapter 6 in Schroeder (1999). The possessive suffix is also used as the personal suffix on embedded nonfinite verbal constructions where it agrees with the genitive-marked subject of the embedded phrase. I will not be concerned with this aspect of possessive marking in Turkish. How all this works in an area with numeral classifiers is a different question.

References Alhoniemi, A. 1993

Grammatik der tscheremissischen Sprache. Hamburg: Buske.

Anghelescu, Nadia 1974 Sur le système de l’article en arabe. Revue roumaine des linguistique 19: 45–52.

Articles and article systems in some areas of Europe

605

Barnes, Michel P., and Eivind Weyhe 1994 Faroese. In The Germanic Languages, (Routledge Language Family Descriptions), Ekkehard König and Johan van der Auwera (eds.), 190–218. London: Routledge. Bechert, Johannes 1990 The structure of the noun in European languages. In Toward a Typology of European Languages, (Empirical Approaches to Language Typology 8), Johannes Bechert, Giuliano Bernini, and Claude Buridant (eds.), 115–140. Berlin and New York: Mouton de Gruyter. [Reprinted in Sprache in Raum und Zeit. In memoriam Johannes Bechert. Band 1: Johannes Bechert: Kleine Schriften, Winfried Boeder, Christoph Schroeder, Karl Heinz Wagner, and Wolfgang Wildgen (eds.), 151–171. Tübingen: Narr.] 1993 Definiteness and article systems. In EUROTYP Working Papers I/4. Strassbourg: European Science Foundation. Bedir Khan, Emir Djeladet, and Roger Lescott 1986 Kurdische Grammatik. Kurmancî-Dialekt. (German transl. by Helga Shahidi und D. N. MacKenzie.) Bonn: Kurdisches Institut, Deutsche Sektion/Verlag für Kultur und Wissenschaft. [First published: (1970). Grammaire kurde (Dialecte kurmandji). Paris: Adrien Maisonneuve.] Benzing, Johannes 1940 Tschuwaschische Forschungen (I). Zeitschrift der deutschen Morgenländischen Gesellschaft 94 (New Series 19): 251–267. Bethke, Inge 1990 der die das als Pronomen. (Studien Deutsch 11). Munich: Iudicium. Beyrer, Arthur, Klaus Bochmann, and Siegfried Bronsert 1987 Grammatik der rumänischen Sprache der Gegenwart. Leipzig: Enzyklopädie. Birkenmaier, Willy 1979 Artikelfunktionen in einer artikellosen Sprache. (Forum Slavicum 34). Munich: Fink. Boeder, Winfried 2002 Variations on one. In Language: Context and Cognition. Papers in Honour of WolfDietrich Bald’s 60th Birthday, Sybil Scholz, Monika Klages, Evelyn Hantson, and Ute Römer (eds.), 29–34. Munich: Langenscheidt. Boeschoten, Hendrik 1998 Uzbek. In The Turkic Languages, (Routledge Language Family Descriptions), Lars Johanson and Éva Ágnes Csató (eds.), 357–378. London: Routledge. Boji´c, Vera, and Wolf Oschlies 1984 Lehrbuch der mazedonischen Sprache. (Slavistische Beiträge 175). Munich: Otto Sagner. Boretzky, Norbert 1968 Der bestimmte Artikel nach Präpositionen im Albanischen. Zeitschrift für Vergleichende Sprachforschung 82: 129–152. Bossong, Georg 1985 Empirische Universalienforschung. Differentielle Objektmarkierung in den neuiranischen Sprachen. (Ars Linguistica 14). Tübingen: Narr. Budi¸na-Lazdi¸na, Ter¯esa 1968 Lehrbuch der lettischen Sprache. Västerås: Latvieˆsu Centr¯al¯as Komitejas Izdevums. Chafe, Wallace 1976 Givenness, contrastiveness, definiteness, subjects and point of view. In Subject and Topic, Charles Li (ed.), 25–57. New York: Academic Press. Chesterman, Andrew 1991 On Definiteness. A Study with Special Reference to English and Finnish. (Cambridge Studies in Linguistics 56). Cambridge: Cambridge University Press.

606

Christoph Schroeder

Christophersen, Paul 1939 The Articles. A Study of Their Theory and Use in English. Copenhagen: Munksgaard. Collinder, Björn 1957 Survey of the Uralic Languages. Compiled by Björn Collinder in collaboration with other scholars. Stockholm: Almqvist & Wiksell. Comrie, Bernard 1978 Definite direct objects and referent idenitification. Pragmatics Microfiche 3.1.D3. Croft, William 1990 Typology and Universals. Cambridge: Cambridge University Press. Csúcs, Sándor 1998 Udmurt. In The Uralic Languages, (Routledge Language Family Descriptions), Daniel Abondolo (ed.), 276–304. London: Routledge. Dasinger, Lisa, and Aylin Küntay 1998 Introducing referents in elicited discourse: Finnish vs. Turkish. In Perspectives on Language Acquisition. Selected Papers from the VIIIth International Congress for the Study of Child Language, Ayhan Aksu-Koç, Eser Erguvanlı-Taylan, Sumru Özsoy, and Aylin Küntay (eds.), 263–276. Istanbul: Bosphorus University. Décsy, Gyula 1966 Yurak chrestomathy. (Indiana University Publications, Uralic and Altaic Series 50). Bloomington, IN: Indiana University Press/The Hague: Mouton. Dik, Simon 1989 Theory of Functional Grammar. Part I: The Structure of the Clause. Dordrecht: Foris. Drimba, Vladimir 1973 Syntaxe comane. Bucarest: Editura Academiei/Leiden: Brill. Ebert, Karen Heide 1971 Referenz, Sprechsituation und die bestimmten Artikel in einem nordfriesischen Dialekt. Braist/Bredstedt: Nordfriisk Instituut. Ehlich, Konrad 1982 Anaphora and deixis: same, similar, or different? In Speech, Place, and Action. Studies in Deixis and Related Topics, Robert J. Jarvella and Wolfgang Klein (eds.), 315–338. Chichester: Wiley. Ehlich, Konrad, and Jochen Rehbein 1976 Halbinterpretative Arbeitstranskriptionen (HIAT). Linguistische Berichte 45: 21–41. Eichhoff, Jürgen 1993 “Ich bin ein Berliner”: a history and a linguistic clarification. Monatshefte 85: 71–80. Endzelin, Jan 1922 Lettische Grammatik. Riga: Kommissionsverlag A. Gulbis. Erkü, Feride, and Jeanette Gundel 1987 The pragmatics of indirect anaphors. In The Pragmatic Perspective. Selected Papers from the 1985 International Pragmatics Conference, (Pragmatics & Beyond, Companion Series 5), Jef Verschueren and Marcella Bertuccelli-Papi (eds.), 533–546. Amsterdam and Philadelphia: John Benjamins. Foerste, William 1954 Geschichte der niederdeutschen Mundarten. In Deutsche Philologie im Aufriß, Bd. 2. , W. Stammler (ed.), 1905–2038. Berlin: Schmidt. Galton, Herbert 1973 The function of the definite article in some Indo-European languages: the grammatical category of determinacy. Linguistics 107: 5–13. G¯aters, Alfr¯eds 1977 Die lettische Sprache und ihre Dialekte. (Trends in Linguistics, State-of-the-Art Reports 9). The Hague: Mouton.

Articles and article systems in some areas of Europe Givón, Talmy 1981

607

On the development of the numeral ‘one’ as an indefinite marker. Folia Linguistica Historica 2 (1): 35–53. Greenberg, Joseph H. 1978 How does a language aquire gender markers? In Universals of Human Language. Vol. 3: Word Structure, Joseph H. Greenberg, Charles A. Ferguson, and Edith A. Moravcsik (eds.), 47–82. Stanford, CA: Stanford University Press. 1981 Nilo-Saharan moveable -k as a stage III article. Journal of African Languages and Linguistics 3: 103–112. 1985 Some iconic relationships among place, time, and discourse deixis. In Iconicity in Syntax. Proceedings of a Symposium on Iconicity in Syntax, Stanford, June 24–26, 1983. John Haiman (ed.), 271–287. Amsterdam and Philadelphia: John Benjamins. Grønbech, Kaare 1936 Der türkische Sprachbau I. Copenhagen: Levin & Munksgaard. Hamp, Eric P. 1975 Current tasks in grammar and phonology. Revue Romaine de Linguistique 20 (2): 93– 104. Harlig, Jeff 1986 One little word that does so much. Papers of the Chicago Linguistic Society 22 (2): 91–104. Hartmann, Dietrich 1982 Deixis and anaphora in German dialects: the semantics and pragmatics of two definite articles in dialectal varieties. In Here and There. Cross-linguistic Studies on Deixis and Demonstration, Jürgen Weissenborn and Wolfgang Klein (eds.), 187–207. Amsterdam and Philidelphia: John Benjamins. Haspelmath, Martin 2001 The European linguistic area: Standard Average European. In Language Typology and Language Universals. An International Handbook (Handbooks of Linguistics and Communication Science HSK 20.1), Martin Haspelmath, Ekkehard König, Wulf Österreicher, and Wolfgang Raible (eds.), 1492–1510. Berlin/New York: Mouton de Gruyter. Hawkins, John A. 1978 Definiteness and Indefiniteness. A Study in Reference and Grammaticality Prediction. London: Croom Helm. 1984 A note on referent identifiability and co-presence. Journal of Pragmatics 8: 649–659. 1991 On (in)definite articles: implications and (un)grammaticality prediction. Journal of Linguistics 27: 405–442. Heine, Bernd 1997 Cognitive Foundations of Grammar. Oxford: Oxford University Press. Heinrichs, Heinrich Matthias 1954 Studien zum bestimmten Artikel in den germanischen Sprachen. Giessen: Wilhelm Schmitz. Helimski (Xelimskij), Eugene 1998a Nganasan. In The Uralic Languages, (Routledge Language Family Descriptions), Daniel Abondolo (ed.), 480–515. London: Routledge. 1998b Selkup. In The Uralic Languages, (Routledge Language Family Descriptions), Daniel Abondolo (ed.), 548–578. London: Routledge. Hendriks, P. 1976 The Radožda-Vevˇcani dialect of Macedonian. (Publications on Macedonian 1). Lisse: Peter de Ridde Press. Hetzer, Armin 1996 Zu den albanisch-rumänischen Parallelen. Balkanarchiv (New Series) 21: 71–80.

608

Christoph Schroeder

Himmelmann, Nikolaus 1997 Deiktikon, Artikel, Nominalphrase. Zur Emergenz syntaktischer Struktur. (Linguistische Arbeiten 362). Tübingen: Niemeyer. 1998 Regularity in irregularity: article use in adpositional phrases. Linguistic Typology 2 (3): 315–354. Hopper, Paul 1987 Emergent grammar. Berkeley Linguistics Society 13: 139–157. Johanson, Lars 1977 Bestimmtheit und Mitteilungsperspektive im türkischen Satz. Zeitschrift der deutschen Morgenländischen Gesellschaft, Suppl. III/2: 1186–1203. [Reprinted in: Johanson, Lars (1991). Linguistische Beiträge zur Gesamtturkologie, p. 225–242. Budapest: Akadémiai Kiadó.] 1998 The structure of Turkic. In The Turkic Languages, (Routledge Language Family Descriptions), Lars Johanson and Éva Ágnes Csató (eds.), 30–66. London: Routledge. Joseph, B. D. 1992 The Balkan languages. In International Encyclopedia of Linguistics. Vol. 1, W. Bright (ed.), 153–155. Oxford: Pergamon. Keresztes, László 1998 Mansi. In The Uralic Languages, (Routledge Language Family Descriptions), Daniel Abondolo (ed.), 387–427. London: Routledge. Kiparsky, Valentin 1967 Russische historische Grammatik. Band II: Die Entwicklung des Formensystems. Heidelberg: Carl Winter. Kolde, Gottfried 1995 Indefinitheit und Zählbarkeit. Sprachkritik des Indefinitartikels im Deutschen und in anderen europäischen Sprachen: ein Indiz für gegenwärtigen Sprach(norm)wandel? In Sprachnormen und Sprachormwandel in gegenwärtigen europäischen Sprachen: Beiträge zur gleichnamigen Fachkonferenz November 1994 am Fachbereich Sprach- und Literaturwissenschaften der Universität Rostock, (Rostocker Beiträge zur Sprachwissenschaft 1), Oskar Müller, Dieter Nerius, Jürgen Schmidt-Radefeldt, and Friedrich Ungerer (eds.), 141–152. Rostock: Universität Rostock, Fachbereich Sprach- und Literaturwissenschaften. Kuno, Susumo 1972 Functional sentence perspective: a case study from Japanese and English. Linguistic Inquiry 3: 269–320. Kuz’mina, I. B., and E. V. Nemˇcenko 1962 K voprosu o postpozitivnyx cˇ asticax v russkix govorax. Materialy i issledovanija po russkoj dialektologii (New Series): 3–32. Kuznecov, Pëtr Savviˇc 1973 Russkaja dialektologija. Moscow: Prosvešˇcenie. Kuznecova, A. I., E. A. Xelimskij (Helimski), and E. V. Gruškina 1980 Oˇcerki po sel’kupskomu yazyku. Tazovskij dialekt. Tom 1. (Publicacii otdelenija strukturnoj i prikladnoj Lingvistiki. Serija monografij: materialy polevyx issledovanij 8). Moscow: Izdatel’stvo Moskovskogo universiteta. Larsson, Lars-Gunnar 1983 Studien zum Partitivgebrauch in den ostseefinnischen Sprachen. (Acta Universitatis Upsaliensis, Studia Uralica et Altaica Upsaliensa 15). Uppsala: Almqvist & Wiksell. Laury, Ritva 1997 Demonstratives in Interaction. The Emergence of a Definite Article in Finnish. (Studies in Discourse and Grammar 7). Amsterdam and Philidelphia: John Benjamins.

Articles and article systems in some areas of Europe

609

Leinonen, Marja 1998 The postpositive particle -to of Northern Russian dialects, compared with Permic languages (Komi Zyrian). In Studia slavica finlandensia in congressu XII slavistarum internationali cracoviae anno MCMXCVIII oblate, Jyrki Papinniemi, Jouko Lindstedt, and Pekka Pesonen (eds.), 74–90. Helsinki: Institute for Russian and East European Studies. Lundeby, Einar 1965 Overbestemt substantiv i norsk og de andre nordiske språk. Oslo: Universitetsforlaget. Mayer, Gerald L. 1988 The Definite Article in Contemporary Standard Bulgarian. (Osteuropa-Institut an der Freien Universität Berlin. Balkanologische Veröffentlichungen 14). Wiesbaden: Harrassowitz. Meillet, Antoine, and André Vaillant 1980 Grammaire de la langue serbo-croate. 2nd ed. Paris: Champion. Menges, Karl Heinrich 1959 Das Sojonische und das Karagassische. In Philologiae Turcicae Fundamenta, Vol. 1, Jean Deny, Kaare Grønbech, Helmuth Scheel, and Zeki Velidi Togan (eds.), 640–670. Stuttgart: Steiner. Nilsson, Birgit 1985 Case Marking Semantics in Turkish. Stockholm: Stockholms Universitet, Department of Linguistics. Panzer, Baldur 1984 Parallel internal development or interference? Post-positive article and possessive perfect in North Germanic and Northern Russian dialects. In Scandinavian Language Contacts, P. Sture Ureland and Iain Clarkson (eds.), 111–132. Cambridge: Cambridge University Press. Perridon, Harry 1994 The Scandinavian article systems: unity in diversity. In Proceedings of the VIIIth Conference of Nordic and General Linguistics, August 16–21 1993. General Session. Vol. II, (Gothenburg Papers in Theoretical Linguistics 69), Jens Allwood, Bo Ralph, Paula Andersson, Dora Kos-Dienes, and Åsa Wengelin (eds.), 323–333. Gothenburg: Göteborgs Universitet. Plank, Frans, and Edith Moravcsik 1996 The Maltese article: language-particulars and universals. Rivista di Linguistica 8 (1): 183–212. Pritsak, Omeljan 1959 Das Karaimische. In Philologiae Turcicae Fundamenta, Vol. 1, Jean Deny, Kaare Grønbech, Helmuth Scheel, and Zeki Velidi Togan (eds.), 318–340. Stuttgart: Steiner. Prokuševa, T. I. 1984 Kategorija opredelennosti-neopredelennosti v Komi jazyke. In Vzaimodelejstvie FinnoUgorskix i Russkogo jazykov, (Trudy Instituta Jazyka, Literatura i Istorii), 43–50. Syktyvkar: Akademija Nauk SSSR, Komi Filal. Rédei, Károly 1978 Syrjänische Chrestomathie. Mit Grammatik und Glossar. (German transl. of Hungarian original by Attila Dabó.) Vienna: Verband der wissenschaftlichen Gesellschaften Österreichs. Reichenkron, Günter 1966 Anfänge einer Artikelausbildung im Serbokroatischen? Die Welt der Slaven 11: 337– 352. Reinhammar, Vidar 1975 Pronomenstudier. Uppsala: Almqvist & Wiksell.

610

Christoph Schroeder

Rijkhoff, Jan 1998

Ruge, Hans 1986

Order in the noun phrase of the languages of Europe. In Constituent Order in the Languages of Europe, (Empirical Approaches to Language Typology, EUROTYP 20–1), Anna Siewierska (ed.), 321–381. Berlin and New York: Mouton de Gruyter. Grammatik des Neugriechischen. Lautlehre, Formenlehre, Syntax. Cologne: Romiosini.

Sasse, Hans-Jürgen 1992 Zum Ausdruck von Referentialität und Nichtreferentialität im Arvanitischen. In Beiträge zur sprachlichen, literarischen und kulturellen Vielfalt in den Philologien. Fs. Rupprecht Rohr, Gabriele Birken-Silverman and Gerda Rössler (eds.), 179–191. Stuttgart: Steiner. Schaller, H. W. 1975

Die Balkansprachen. Heidelberg: Carl Winter.

Schlachter, Wolfgang 1960 Studien zum Possessivsuffix des Syrjänischen. (Finnisch-Ugrische Studien 111). Berlin: Akademie-Verlag. Schroeder, Christoph 1999 The Turkish Nominal Phrase in Spoken Discourse. (Turcologica 40). Wiesbaden: Harrassowitz. 2002

Zur Nominalphrasenstruktur des Kurmancî. In Philology, Typology and Language Structure. Festschrift for Winfried Boeder on the Occasion of his 65th Birthday, Wolfram Bublitz, Manfred von Roncador, and Heinz Vater (eds.), 191–210. Frankfurt/Main: Ewald Lang.

Sjoberg, Andrée F. 1963 Uzbek Structural Grammar. (Uralic and Altaic Series 18). Bloomington, IN: Indiana University/The Hague: Mouton. Stachowski, Marek 1998 An example of Nganasan-Dolgan linguistic contact. Turkic Languages 2: 126–129. Stephanides, Éva 1974 A Contrastive Study of the English and Hungarian Article. (The Hungarian-English Contrastive Linguistics Project, Working Papers 5). Budapest: Linguistics Institute of the Hungarian Academy of Sciences/Washington, DC: Center for Applied Linguistics. Stölting, Wilfried 1970 Beiträge zur Geschichte des Artikels im Bulgarischen. (Slavistische Beiträge 44). Munich: Otto Sagner. Svane, Gunnar Olaf 1958 Grammatik der slowenischen Schriftsprache. Copenhagen: Rosenkilde & Bagger. Terešˇcenko, Natal’ja Mitrofanovna 1979 Nganasanskij jazyk. (Akademija Nauk SSSR, Institut jazykoznanija, Leningradskoe otdelenie). Leningrad: Nauka. Turunen, Aimo 1988 The Balto-Finnic languages. In The Uralic Languages. Description, History and Foreign Influences, (Handbuch der Orientalistik, VIII. Abtlg., Band 1), Denis Sinor (ed.), 58–83. Leiden: Brill. Ultan, Russell 1978

On the development of a definite article. In Language Universals. Papers from the Conference Held at Gummersbach/Cologne, Germany, October 3–8, 1976, Hansjakob Seiler (ed.), 249–265. Tübingen: Narr.

Articles and article systems in some areas of Europe

611

van Schaaik, Gerjan 1992 The treatment of Turkish nominal compounds in FG. In Layered Structure and Reference in a Functional Perspective, Michael Fortescue, Peter Harder, and Lars Kristoffersen (eds.), 231–252. Amsterdam: John Benjamins. Vater, Heinz 2002 The word class ‘Article’. In Lexicology. An International Handbook on the Nature and Structure of Words and Vocabularies (Handbooks of Linguistics and Communication Science HSK 21.1), Cruse, D. Alan, Franz Hundsnurscher, Michael Job, and Peter Rolf Lutzeier (eds.), 621–628. Berlin/New York: Mouton de Gruyter. Veksler, Bunim Xlavnoviˇc and Vladimir Avgustoviˇc Jurik 1975 Latyšskij jazyk (Samouˇcitel’). 2nd ed. Riga: Zvajgzne. Vincenot, Claude 1975 Essai de grammaire slovène. Ljubljana: Mladinska knjiga. Vogel, Petra 1993 Über den Zusammenhang von definitem Artikel und Ferndeixis. Zeitschrift für Sprachtypologie und Universalienforschung (STUF) 3: 222–233. von Gabain, Annemarie 1945 Özbekische Grammatik. Leipzig: Harrassowitz. Wahrig-Burfeind, Renate 1989 Nominales und pronominales Genus im südlichen Nordseegebiet: eine areallinguistische Untersuchung. Munich: tuduv. 1998 Bestimmtheit bei Artikelwörtern und Personalpronomina im Niederdeutschen. In Sprache in Raum und Zeit. In memoriam Johannes Bechert. Band 2: Beiträge zur empirischen Sprachwissenschaft, Winfried Boeder, Christoph Schroeder, Karl Heinz Wagner and Wolfgang Wildgen (eds.), 319–334. Tübingen: Narr. Wehr, Barbara 1984 Diskurs-Strategien im Romanischen. Ein Beitrag zur romanischen Syntax. (Romanica Monacensia 22). Tübingen: Narr. 1994 Topic- und Focus-Konstruktionen im Französischen. In Sprachlicher Alltag. Linguistik – Rhetorik – Literaturwissenschaft. Festschrift für Wolf-Dieter Stempel, 7. Juli 1994, Annette Sabban and Christian Schmitt (eds.), 611–633. Tübingen: Niemeyer. Zaicz, Gábor 1998 Mordva. In The Uralic Languages, (Routledge Language Family Descriptions), Daniel Abondolo (ed.), 184–218. London: Routledge.

Indexes

Subject index aboutness, 261, 375, 377, 486 accent, 144, 162, 329, 378, 401, 490 accessibility, 229 accessibility of reference, 217, 219, 239 activation, 70, 101, 107 adjective attributive, 571 definite, 570 indefinite, 570 predicative, 571 all-new, 256, 261, 299, 315 anaphor(a), 217, 246, 249 direct, 601 indirect, 217, 601 zero, 112 animacy, 279, 445, 466, 495 antecedent, 547 focality of, 235 syntactic status of, 224, 232 topicality of, 235 article definite, 552, 553, 560, 601 emergent, 586 enclitic, 565, 567 indefinite, 555 extended, 559, 600 pragmatic, 557, 600 referential, 558, 600 article reiteration, 583 backgrounding, 376, 377, 445 categorical, 259 clause, 19 main, 420 subordinate, 420 cleft, 125, 128, 171, 204, 228, 274, 331, 368, 408, 410, 429, 434, 437 IT-, 43, 172, 189 reverse WH-, 173, 189 WH-, 42, 189

WH-word, 172 coding of distance, 239 comment, 77, 81, 83, 97, 109, 568 competence, 31, 358 configurationality, 421 contrast(iveness), 129, 156, 166, 229, 324, 329, 377, 568 coreference, 221 d-configurationality, 370, 373, 383, 463 de-accenting, 313, 314, 318, 330 morphological, 326, 331 definiteness, 279, 294, 387, 466, 495, 546, 550 defocusing, 123, 164 deixis, 128, 241, 249, 577 contextual, 243 exclusive, 241 extratextual, 218, 231 intratextual, 216, 218 situational, 230, 249 spatial, 219, 220, 243, 246 textual, 243 demonstrative, 217, 219, 221, 553, 601 distal, 231, 238 proximate, 231, 238 dependency, 24 discourse, 329, 331 discourse function, 123, 280, 288, 292, 298 annuntiative, 281 connective, 295 descriptive, 286 episode-opening, 294 explanative, 287, 291 interruptive, 285 introductive, 284 discourse goal, 123 discourse particle, 32, 165 discourse strategy, 298 discourse topic, 103

616

Subject index

dislocation, 367 left, 70, 331, 390, 405, 424, 433 right, 102, 272 distance, 577 distribution, 24 ellipsis, 157 eventive, 446 existential construction, 139 focalization, 392, 408, 416, 445, 483 focus, 86, 122, 268, 359, 372, 376, 379, 383, 402, 486, 489–491, 497, 568 contrastive, 366, 405 contrastive subject, 271 marked, 387, 401, 413, 423, 432, 438 sentence, 256, 300, 464 unmarked, 387, 389, 411, 422 verb, 271, 296, 461 foregrounding, 376, 377, 445 fronting, 272, 437 gapping, 21 given, 128, 154, 157, 294, 313, 315, 331, 392, 470–472 grammatical function, 375, 383, 390, 392, 421, 457, 459, 466, 487 grammaticalization, 273, 428, 441, 447, 466, 493, 555, 588 highlighting, 39, 122, 128, 375, 377 identifiability, 546, 550 indefiniteness, 551 information packaging, 32, 70 intonation, 19, 170, 301, 317 introduction of episode, 436 introduction of new entities, 156 introduction of new propositions, 145, 147, 151 introduction of participants, 290 introduction of referents, 600 inversion, 272, 415 lexical semantics, 300

lexicalization, 289, 293, 298 linker morpheme, 580 Mittelfeld, 416, 426 monoargumentality, 277, 442, 444, 456, 460 morphology agglutinating, 379, 496 case, 391, 413 Nachfeld, 416 new, 128, 154, 315, 331, 392, 470–472 (non)-configurational languages, 365 nonfinality, 311 nonreferentiality, 551 NP-clause construction, 135, 137–139, 142, 143 nuclear stress, 402, 403 numeral “one”, 555, 600 object, 491 null, 159, 160 packaging, 68 particle(s), 124, 128, 204, 378, 494 pause, 19, 22, 24, 36 performance, 31, 358 pitch, 24, 129, 311, 329 position, 367, 371, 383, 415, 479, 481, 487 clause initial, 154 empty, 371 initial, 437 postverbal, 164, 469, 475, 489 preverbal, 153, 164, 379, 387, 468, 489, 493 sentence-final, 28, 102, 387 sentence-initial, 28, 102 SVO, 477 possessive, 554 relational, 595 possessive suffix, 586, 588, 591, 601 predication, 256, 259 presentative construction, 136 presuppositionality, 300 primary participant, 85, 98 productivity, 293

Subject index pronoun anaphoric, 217, 221 demonstrative, 216 prosody, 311–313, 378, 387, 491 proximity, 577 re-accenting, 318, 330 reference, 217, 227, 547, 561, 565 discourse, 223 nonanaphoric, 549, 550 unique, 549 reference domain, 237, 238, 240 reference point, 90 reference to context, 239 reference to entities, 232 reference to space, 239 referentiality, 279, 387, 466, 495 repetition, 315 right-detachment, see dislocation, right scene-setting, 284, 286 sentence marked, 486, 489, 491 system, 18 text, 18 unmarked, 486, 489 sentence field, 368 sentence space, 366, 479, 480, 483, 489, 495 postverbal, 482 preverbal, 482 sentence structure, 374 sentential stress, 401 space postverbal, 483 preverbal, 483 spatial/temporal frame, 95, 99 specificity, 551 split structure, 265 spoken language, 11, 13, 124, 128, 204, 249, 269, 282, 363, 367, 392, 419, 428, 460, 476, 478 planned, 36 spontaneous, 12, 14, 15, 127, 416 subject, 85

617

null, 159, 222 subject accentuation, 264, 271 subject incorporation, 265 subject-prominence, 144 subordinate clause, 28 syntax, 124 fragmented, 29, 30 tail, 377, 416, 462, 475 thematicization, 445 theme, 122, 221, 376, 567 theme-rheme structure, 373 thetic, 256, 260, 315, 443 theticity, 86, 257, 259, 263, 299, 301 tone, 311 topic, 122, 283, 312, 359, 372, 431, 432, 486 ˛-topic, 70, 87, 91, 92, 95, 96 contrastive, 409 fronting, 83 fronting (f -topic), 79 hanging, 73, 83, 89, 97, 367 hanging (h-topic), 79 non-nominal, 80 packaging, 71 s-topic, 83 s-topic construction, 79 sentence, 68, 69, 100, 312 topic-comment, 68, 70, 100, 569 topic continuity, 295 topic discontinuity, 113 topic encoding, 69, 75 double, 112 merged, 109 topic prominence, 73, 97, 103, 107, 144 topicalization, 392, 403, 431, 483 variation dialectal, 328 interspeaker, 328 stylistic, 328 Vorfeld, 416 WH-pronoun, 48, 49, 54, 55 WH-word, 149, 379, 392, 496 whole-part relationship, 547

618

Subject index

word order, 124, 128, 153, 166, 204, 315, 459 dominant, 365 free, 274, 392 marked, 466 neutral, 365 non-neutral, 365 pragmatic, 363 pragmatic typology of, 361 SOV, 379, 481, 482, 490

SV, 442, 459 SVO, 391, 407, 442, 481, 490 V-2, 391, 414, 478 VS, 264, 270, 277, 442, 459 VSO, 429, 440, 484 written language, 11, 13, 35, 124, 127, 128, 204, 249, 269, 363, 367, 392, 419, 428, 452, 460, 476, 478 zero pronoun, 223

Language index Aghem, 492, 495, 525 Albanian, 255–257, 266, 271–273, 275, 280, 281, 283, 285, 286, 318, 323, 327, 331, 333, 334, 558, 579, 580, 582–584, 600, 603, 604 Altaic, 360, 379, 466, 500, 526 American Indian, 209 Amerindian, 207 Amharic, 210 Arabic, 318–327, 331, 336, 435, 524 Modern Standard, 517 Tunisian, 318, 335 Armenian, 386, 501, 552 Azeri, 593 Balkan, 239, 269, 275 Baltic, 552, 570–573, 576, 585 Balto-Finnic, 375, 391, 393, 396, 403, 406, 446, 500, 552, 570, 575, 585 Bambara, 47, 48 Bantu, 209, 492 Basque, 255, 269, 281, 361, 362, 375, 379, 381–383, 386, 391, 397, 424, 445, 453, 455, 459, 461, 462, 482, 484, 486–488, 492, 494, 496, 500–502, 521, 524, 525, 552 Bengali, 47, 48, 54, 56, 380 Bokmål, 564 Boni, 266, 267 Brazilian Portuguese, 315 Breton, 441, 514 Bulgarian, 123, 126, 128, 143, 166, 189– 191, 200–203, 255, 266, 269, 293, 397, 505, 557, 559, 576, 578–580, 584, 598–600, 603, 604 East, 57 spoken, 578 West, 57 written standard, 578 Burushaski, 493

Castilian, 318 Catalan, 59, 123, 126, 132, 140, 162, 166, 179, 188, 193, 197, 204, 316, 318 Caucasian, 384 Cayuga, 303 Celtic, 144, 145, 269, 369, 377, 378, 438, 440, 452, 463, 464, 466, 486, 508, 515, 525 Cheremis, 524, 586 Chilean, 316 Chinese, 46, 74, 522 Chukchee, 115 Chuvash, 592, 604 Croatian, 576, 585, 603 Cushitic, 501 Eastern, 266 Cuzco Quechua, 495 Czech, 51, 207, 266, 269, 281, 293, 397, 460, 505, 507, 602

163, 267,

429, 478,

301,

Danish, 182, 184, 250, 267, 269, 285, 414, 510, 562, 564, 571, 572, 599, 603 Standard, 562, 563 Dinka, 111 Dolgan, 592 Dutch, 75, 218, 220, 226, 228, 230, 232, 235–238, 250, 257, 266, 269, 281, 286, 290, 293, 311, 313, 315, 415, 417, 466, 510, 519 modern, 216 spoken, 232, 233 written, 232 Easter Island language, 517 Egyptian, 518 English, 10, 12, 14, 17, 26, 31, 34, 46, 49, 50, 59, 67, 72, 74, 75, 78, 82, 89, 92, 93, 95, 105, 107, 121, 123, 124, 126, 127, 133–135, 141, 156, 158, 160,

620

Language index

161, 163, 165, 166, 169, 170, 172– 179, 181, 184–188, 190, 192, 194, 196, 197, 199–207, 209, 210, 218, 226, 228, 231, 235, 238, 255, 257– 259, 261, 262, 264, 266–269, 271, 273–275, 277–279, 281, 285, 286, 290–293, 297, 298, 301, 311–318, 320–323, 325, 326, 328, 361, 365, 369, 370, 391, 393, 409–411, 413, 414, 422, 424, 426, 447, 449–451, 460, 464, 466, 478, 486, 499, 503, 509, 513, 519, 522, 545, 553, 555, 556, 559, 588, 590, 598, 599 American, 45 Australian, 312 Hawaiian, 317 Indian, 317 London Jamaican , 317 Middle, 51 Modern, 18 modern written, 17 New Zealand, 312 Nigerian, 317 nonstandard, 48, 50, 57, 58 Old, 392, 461, 504 Scottish, 42, 49–53, 136 spoken, 28, 40, 42, 43, 49, 50, 56, 132, 145, 158, 159, 165, 189, 208, 209 spoken nonstandard, 52 spoken standard, 56 spontaneous, 124 spontaneous spoken, 29, 31, 44, 45, 50, 53, 151, 157, 207, 208 Standard, 547 standard British, 136, 147 standard written, 54 written, 11, 18, 20, 28, 29, 42, 44, 49, 50, 53, 56, 58, 129, 136, 145, 158, 189, 205, 208, 452 Estonian, 361, 375, 391, 393, 396, 452, 470, 473, 487, 504, 520, 575 Faroese, 564 Fering, 552, 554, 562, 598, 599

Western, 560, 564 Finnish, 14, 125–127, 135, 166, 167, 174, 177–179, 198, 204, 303, 361, 375, 391, 393, 395, 396, 452, 458, 460, 461, 470, 473, 487, 504, 505, 521, 522, 524, 559, 575 dialects, 461 spontaneous spoken, 61 Finno-Ugric, 203, 204, 360, 584 French, 18, 34, 41, 59, 78, 83, 93, 94, 102, 139, 151, 175, 179, 184–186, 204, 220, 250, 255, 257, 259, 265–267, 269, 274, 278, 279, 281, 282, 285– 287, 290–293, 304, 317–321, 323– 327, 331, 337, 338, 361, 368, 378, 403, 408–411, 413, 437, 447, 450, 460, 464, 486, 499, 503, 507, 508, 524 modern spoken, 164 Old, 23, 461, 503 spoken, 43, 51, 60, 403 written, 24, 60, 138 written standard, 403 Frisian North, 562, 587 Gagauz, 559, 593, 600 written, 604 Genoese, 260, 267 Georgian, 361, 375, 384, 386, 501, 502, 545 German, 20, 34, 72, 75, 82, 89, 93–95, 101, 102, 110, 121, 123, 126, 127, 140, 147, 166, 180–183, 189–192, 197, 203, 204, 208, 210, 220, 255, 257, 259, 261, 262, 264, 267–269, 271, 274, 275, 277–279, 281–283, 285– 287, 290–292, 294, 301, 311, 318, 320, 325, 327, 331, 339, 340, 361, 365, 374, 375, 393, 414, 415, 417– 423, 426–428, 447, 450, 459, 464, 466, 477, 496, 499, 504, 510, 513, 518, 519, 522, 559, 561, 565, 571, 573, 575, 588 colloquial, 460 Low, 546, 552–554, 560, 562, 565, 567–

Language index 570, 598, 599, 601–603 Old High, 514 spoken, 266, 418, 460 spoken Standard, 561 Standard, 566, 567, 599 Substandard, 295 German dialects, 460 Germanic, 52, 239, 266, 281, 328, 362, 367– 369, 371, 391, 396, 409, 413, 414, 418, 419, 447, 460, 470, 471, 473, 474, 478, 509, 514, 520, 524, 552, 570–573, 576, 578, 599 Modern Standard, 522 Greek, 26, 128, 148, 186–188, 204, 304, 317, 318, 320, 321, 323, 325, 327, 331, 340, 341, 359, 477, 478, 498, 559, 582–584, 600, 603 Classical, 25, 26, 205, 504 Homeric, 205 Modern, 255–257, 261, 262, 266, 269– 273, 275, 278, 280–286, 291, 292, 294, 295, 297, 362, 391, 406, 413, 504 Guipuzcoano, 502 Hamito-Semitic, 515, 518 Hebrew, 50, 56, 246, 518, 524, 604 Biblical, 239, 246, 517, 583 Classical, 517 Israeli, 216, 239 Modern, 239, 247, 248, 517, 583 Modern (Israeli), 246 Hindi, 48, 56, 380 Huallaga, 495 Huallaga Quechua, 526 Hungarian, 72, 75, 83, 123, 125–127, 135, 140, 142, 166, 168, 169, 174, 189, 197–199, 204, 208, 255, 257, 266– 273, 275, 276, 278, 279, 281–286, 292, 294, 295, 301, 303, 316, 361, 370, 379, 380, 387, 389, 398, 426, 427, 429, 455, 461, 466, 469, 470, 493, 496, 497, 500, 501, 503, 504, 521, 522, 552, 559, 576, 600 Modern, 389

621

Old, 389, 521 Icelandic, 182–184, 204, 391, 418, 504, 510, 512, 552 Old, 418 Indian, 501 Indo-Aryan, 216, 239, 380, 496, 501, 526 Indo-European, 48, 49, 53, 55, 56, 143, 176, 203, 204, 207, 362, 363, 380, 391, 414, 432, 442, 446, 449, 450, 452, 477, 478, 487, 501, 503, 509, 515, 519 Iranian, 552 Irish, 77, 135, 142, 144, 170, 185, 186, 207, 255–257, 316, 361, 429–434, 437, 440, 452, 464, 514–517, 552 Old, 516 spoken, 143, 145, 209 Iroquoian, 265 Italian, 51, 58, 74, 75, 77, 78, 81, 94, 101, 102, 107, 109, 110, 115, 166, 180, 208, 250, 255, 257, 259, 266, 267, 269–273, 277, 279–281, 283– 285, 294, 295, 297, 303, 316–318, 321–323, 326, 327, 331, 342, 343, 361, 401, 403, 406, 409–411, 413, 444, 445, 447–450, 459, 461, 462, 466, 470, 499, 507–509, 513, 519, 520, 522 Neapolitan, 16 spoken, 403, 519 Japanese, 74, 76, 140, 220, 255, 256, 260, 262, 266, 274, 380, 493, 494, 496, 497, 525, 526 Jutish, 552, 554, 563, 587, 598, 599 Kamas(sian), 586 Kannada, 493, 496, 525 Karaim, 604 Karalkapak, 593 Kashmiri, 380 Kazakh, 592, 600 Kelderash, 240 Khanty, 586 Kihung’an, 209

622

Language index

Kipchak, 604 Kirghiz, 592 Kirmanji, 545, 552, 554, 556, 602 Komi, 585–588, 591, 598–600 Korean, 74, 91, 493, 496, 497, 525 Lapp, 378 Latin, 57, 205, 219, 221–223, 229, 233, 271–273, 278, 285, 292, 295, 359, 391, 420, 477, 478, 498, 503, 524 Classical, 18, 216, 217, 220, 228, 513 Vulgar, 267 written, 23, 24 Latvian, 554, 570, 572, 573 dialects, 574 High, 575, 603 Low, 603 Middle, 575, 603 Standard, 574 Lithuanian, 318, 324, 325, 327, 331, 344, 345, 391, 570, 572, 573 Lovari, 240 Macedonian, 318–323, 325, 327, 345, 347, 545, 552, 576–580, 584, 598, 599, 603, 604 Maltese, 103, 583, 584 Mandarin Chinese, 92, 140, 144 Mansi, 586 Mari, 379, 500, 524, 586–590, 592, 598 Mayan, 161, 208 Mongolian, 360, 379, 468, 496, 500, 526 Mordvin, 584, 585 Native American languages, 265 Navarro-Labordino, 501 Nenets, 586, 587, 591 Nepalese, 380 Nganasan, 586, 587, 592, 599 Norwegian, 182, 183, 204, 564 Nynorsk, 564 Ob-Ugric, 379 Occitan, 57 Ostyak, 586

Ostyak Samoyed, 586 Palatinate, 522 Persian, 46, 48, 50, 56, 526 Plattdeutsch, 57 Polish, 51, 153, 190, 191, 203, 205, 255– 257, 266, 269, 281, 293, 301, 505 written, 152, 153 Polynesian, 440 Portuguese, 266, 303 Provençal, 51 Quechua, 380, 493–496, 501, 526 Rhaeto-Romance, 266, 267 Romance, 228, 239, 259, 262, 267, 269, 285, 294, 303, 327, 328, 362, 365, 367, 369, 374, 391, 393, 396, 398, 401, 403, 405, 409, 411, 413, 417, 422, 424, 426, 443, 447, 457, 458, 470, 471, 473, 503, 508, 518, 519, 524 vernacular, 23 Romani, 216, 239, 240, 243, 246, 248, 272, 273, 280, 294, 295, 603 Rumanian, 260, 266, 271–273, 275, 278, 281, 282, 285, 286, 291, 292, 294– 297, 304, 316, 579–584, 598–600, 603, 604 Russian, 21, 25, 27, 31, 41, 48, 56, 71, 72, 74, 75, 77, 78, 80, 82, 87, 89, 93–96, 99, 100, 104, 107, 108, 110, 121, 123, 127–129, 134, 140, 141, 144, 150– 152, 155, 156, 160, 161, 166, 175, 190, 191, 201, 203, 204, 209, 210, 255, 257, 266, 269–273, 279, 281, 284, 285, 291, 294, 295, 297, 304, 318, 319, 323, 326, 327, 331, 347, 348, 361, 365, 367, 391, 397–399, 401, 402, 411, 413, 446–448, 450, 459, 461, 462, 465, 466, 470, 472, 473, 482, 486, 499, 500, 505, 507, 509, 518–521, 523–525, 552, 557, 573, 584, 585, 600, 601 spoken, 9, 31, 43, 45, 54, 58, 127, 148, 153, 154, 158, 164, 188, 205, 447, 462

Language index spoken standard, 12, 56 spontaneous spoken, 30, 34, 55, 124, 126, 145, 162, 204 spontaneous spoken standard, 59 Standard, 584 standard written, 54 written, 31, 36, 43, 55, 56, 59, 145, 153, 154, 158, 162 Sami, 32, 33 Northern, 32 Samoan, 517 Samoyed, 360 Sanskrit, 515 Scandinavian, 182, 183, 510, 564, 573, 574 Scots, 57 Broad, 60 Scottish Gaelic, 143, 144 written, 165 Selkup, 586, 587, 591 Semitic, 216, 239, 438, 440, 441, 517, 518, 524, 583 Serbian, 576, 579, 603 Standard, 585 Serbo-Croatian, 126, 128, 190, 191, 203, 255–257, 266, 270, 271, 281, 283, 285, 286, 397, 505, 506, 509, 576 Singapore English, 317 Sinor, 362 Slavic, 51, 188, 239, 269, 293, 552, 570, 576, 580 East, 585 West, 585 Slavonic, 328, 369, 391, 393, 396, 397, 403, 406, 411, 413, 443, 447, 458, 466, 470, 471, 505, 508, 518 Slovene, 576 Standard, 585 Somali, 380, 493, 494, 496, 497, 501, 525, 526 Sorbian, 576 Lower, 57 Upper, 57 Soyut, 604

623

Spanish, 78, 82, 180, 210, 220, 255–257, 259, 266, 267, 269–272, 277, 280, 281, 295, 297, 303, 316–327, 331, 349, 350, 361, 403, 406, 409–411, 413, 445, 447, 459, 461, 466, 499, 507, 509, 519, 559 spoken, 403 Standard Average European, 92 Swedish, 75, 166, 182, 183, 204, 250, 317– 320, 322–325, 327, 331, 351, 352, 391, 414, 415, 418, 419, 461, 510, 512, 521, 522, 564, 571, 573 Old, 59 Swiss German, 57 Tagalog, 266 Tamian, 554, 573–575, 603 Tamil, 49 Tatar, 379, 382, 592 Tongan, 441, 517 Torlakian, 579 Tunguz, 360 Turkic, 203, 360, 379, 381, 500, 526, 592, 595, 601 Southern, 604 Turkish, 45, 121, 127, 135, 142, 143, 166, 168, 170, 174–178, 208, 220, 269, 271, 361, 379, 381, 386, 391, 397, 424, 426, 427, 445, 454, 455, 461, 462, 466, 477, 480, 482, 484, 486–488, 494, 500–502, 522, 523, 525, 552, 555, 556, 559, 592–596, 600, 601, 603, 604 spoken, 597 Turkmen, 593 Tuvan, 591, 604 Tz’utujil, 91 Udmurt, 586, 591, 600 Ugric, 379 Ugro-Finnic, 363, 477 Uighur, 593 Ural-Altaic, 381, 468

591,

163, 210, 382, 429, 470, 492, 546, 598,

624

Language index

Uralic, 360, 466, 500, 585, 586, 592, 597– 599, 601 Uzbek, 592, 593, 604 Vedic, 509 Vizcayno, 502 Vogul, 586 Volgaic, 379 Votian, 504 Votyak, 586

Welsh, 144, 265, 267, 281, 282, 286, 291, 361, 429–434, 436, 437, 440, 441, 452, 463, 464, 514–516, 518, 523 Middle, 441, 515, 516 Modern, 441 spoken, 250, 437 Yakut, 592, 600 Yiddish, 83 Yukag(h)ir, 61, 90

Author index Abraham, 365, 373, 374, 421, 422, 428, 499, 500, 512, 513, 519, 525 Adamec, 505 Adams, 392 Agard, 317 Agius, 517 Aijmer, 115 Aissen, 91, 92, 208 Alhoniemi, 587 Allan, 510 Allerton, 258, 259, 264, 293, 298 Altmann, 513 Andersen, 111 Anderson, 216, 219–221, 514, 517, 518 Anghelescu, 583 Ariel, 215, 231 Aronson, 501 Atanasova, 200 Atkinson, 259, 286, 293 Austerlitz, 501 Babby, 505 Banfi, 502 Bansal, 317 Barnes, 138, 508, 564 Bates, 316 Beaman, 15 Beaulieux, 505 Bechert, 546, 554, 582 Bedir Khan, 602 Behaghel, 417, 446, 509, 510, 512, 522 Behrens, 270, 276, 303, 304, 389, 455, 469, 502, 503, 521–523 Bell, 314, 498 Belletti, 303, 458 Benincà, 78, 81, 508 Benoist, 505 Benzing, 591, 592, 595, 604 Bernini, 94, 280, 281, 284, 294, 297, 313, 445, 464, 500, 508, 519 Berretta, 101, 103, 115, 216, 508

Berruto, 508, 519 Bese, 500 Bethke, 603 Beyrer, 579 Bhatt, 501 Biber, 10–12, 15 Birkenmaier, 552 Birner, 275, 314 Bivon, 140, 152–154, 208 Blanche-Benveniste, 57, 60, 138, 151, 184, 185, 207, 508 Blatt, 498 Blinkenberg, 259 Boeder, 502, 557 Boeschoten, 604 Boji´c, 577, 604 Bolkestein, 69, 215, 216, 222–225, 227, 229, 238, 250, 266, 272, 273, 278, 292, 295 Boretzky, 579 Borsley, 514 Bossong, 164, 165, 303, 304, 499, 508, 520, 552, 596 Bowen, 317 Braunmüller, 514 Brazil, 311 Breivik, 509 Brentano, 256, 261, 262, 300 Brentano-Marty, 260 Britain, 312 Brockelmann, 518 Brody, 499 Brorström, 522 Brown, 19, 135, 206, 311, 314 Budi¸na-Lazdi¸na, 574 Bühler, 241, 509 Büring, 99 Burdach, 514 Burton, 18 Burzio, 444, 519

626

Author index

Buscha, 422, 510, 511 Cadiot, 78, 94 Calvo Pérez, 495, 525, 526 Campbell, 79 Cardinaletti, 512 Carlson, 177–179 Carnie, 514 Cennamo, 277, 443, 518 Chafe, 19, 28, 46, 57, 58, 68, 70, 71, 80, 85, 99, 129, 130, 132, 167, 206, 215, 258, 314, 315, 329, 396, 498, 502, 523, 567 Chapin, 517 Chen, 74 Cheshire, 52 Chesterman, 551, 559, 575 Christophersen, 556, 600 Chung, 514, 517, 518 Cinque, 508, 525 Clark, 314 Collier, 311 Collinder, 584, 586–588, 591 Como, 407 Comrie, 46–48, 50, 85, 87, 318, 363, 386, 473, 477, 498–502, 504, 519, 526, 593 Connolly, 498, 499 Conte, 250 Cornish, 216 Coseriu, 298 Couper-Kuhlen, 311 Cowan, 114 Croft, 112, 360, 498, 499, 554 Crookes, 58 Cruttenden, 258, 259, 264, 293, 298, 311– 313, 325, 329, 369, 410, 464, 500, 512, 513, 522 Crystal, 315 Csúcs, 586, 591 Currie, 442 Dahl, 206 Daneš, 70, 366, 410, 442, 457, 498–500, 505, 508, 512, 518 Danielewicz, 19, 46, 58

Danon-Boileau, 59 Dasinger, 575 Davison, 87 de Groot, 499, 501, 503 De Jong, 223, 224 de Meij, 499 de Rijk, 500, 501 de Schutter, 498 De Simone Brouwer, 392, 406, 504 De Willman-Grabowska, 505 Décsy, 587 Delanoy, 514 Delbrück, 501, 509, 515 Delin, 173, 189 Denniston, 205 Desz˝o, 502 Di Cristo, 311 Di Pietro, 317 Diderichsen, 370, 414, 509 Dik, 69, 129, 131, 133, 206, 225, 271, 370, 372, 498–500, 504, 505, 510, 512, 514, 523, 551 Dooley Collberg, 510 Dover, 504 Downing, 78 Drach, 414, 509, 510 Dressler, 442, 446, 515 Drimba, 604 Drubig, 258, 275, 301 Dryer, 112, 152, 499 Ducrot, 59 Ebert, 514, 560, 562–564 Egbe, 317 Ehlich, 246, 554, 603 Eichhoff, 602 Endzelin, 575, 603 Engel, 424, 510, 511, 513 Enkvist, 29 Erdmann, 519 Erguvanli, 163, 166, 167, 463, 467, 468, 475, 494, 500–502, 521–523, 526 Erguvanli-Taylan, 523 Erkü, 547, 548, 602

Author index Eroms, 115 Etzensperger, 509–511 Euler, 519 Evans, 515 Faber, 329 Fernandez, 14, 61, 140, 166, 204, 378, 477 Fernandez-Vest, 9, 14, 32, 34, 37–39, 41, 57, 59 Fici Giusti, 78 Field, 15, 16 Fife, 515–517, 521 Fillmore, 396 Firbas, 396, 502, 523 Fischer, 103 Fleischmann, 514 Foerste, 603 Foley, 72, 87, 100, 106 Fougeron, 397–400, 413, 447, 462, 465, 471, 472, 505–507, 509, 519, 520, 523, 524 Fox, 215, 226 Freeman, 418 Friedman, 318 Fromm, 470, 520 Fuchs, 258 Gadet, 519 Galton, 582 Garbini, 518 Gazdar, 514 Geerts, 232, 233 Geluykens, 113 Gillies, 143 Givón, 88, 93, 103, 122, 152, 207, 209, 210, 215, 225, 498, 517, 518, 556, 600 Glinert, 517 Goldenberg, 441 Gonda, 442, 515, 519, 523 Graffi, 519 Greenberg, 431, 438, 498, 499, 555, 600 Grosz, 129, 133, 134, 140, 207 Gruber, 525 Grønbech, 591, 592

627

Guentchéva, 505 Guilfoyle, 514 Guiraud, 23, 24 Gumperz, 317 Gundel, 68, 70, 74, 80, 87, 103, 107, 112, 215, 547, 548, 602 Guy, 312 Gvozdanovi´c, 506 Gårding, 317, 319 Haberland, 257, 303, 519 Haegeman, 420, 510, 519 Hagège, 60, 93 Haider, 510, 512 Haiman, 107, 512 Hakulinen, 178, 179, 394, 396, 407, 470, 504, 505, 519, 521, 522 Halliday, 15, 16, 18, 19, 57, 123, 129–131, 133, 134, 190, 206, 314, 315, 500 Hamp, 582 Hankamer, 523 Hanks, 161 Hannay, 215 Harlig, 559 Harris, 79, 385, 502 Hartmann, 560, 561, 602 Hatcher, 259 Haviland, 314 Havránek, 505 Hawkins, 152, 208, 392, 498, 499, 547, 551, 555, 602 Hawkinson, 85 Heath, 33 Hedberg, 171, 172, 201 Heger, 508 Heim, 134 Heine, 556 Heinrichs, 560 Helbig, 422, 510, 511 Helimski, 586, 587 Heltoft, 370, 371, 499, 509 Hemon, 441 Hendrik, 514 Hendriks, 577, 578, 584

628

Author index

Henne, 47 Henriksen, 509 Hetzer, 580–582 Hetzron, 519 Himmelmann, 218, 229, 238, 554, 555, 560, 598, 600, 603, 604 Hinchliffe, 510 Hinds, 525, 526 Hirschberg, 311 Hirst, 311 Hockett, 68 Höhle, 271, 500, 512 Hoffmann, 221 Hofmann, 220 Holden, 413, 470, 498–500, 505, 509 Holes, 435, 524 Holland, 499, 503, 515 Holloway King, 402 Holmberg, 397, 510 Holmes, 510 Hopper, 266, 556 Horvath, 380, 387, 525 Hovy, 58 Huang, 512 Hunt, 58 Hyman, 85 Ingham, 517, 518 Jacob, 508 Jacobs, 68–70, 72, 75, 93, 95, 97, 99, 114, 258, 512, 513 Jakobson, 397 Jastrow, 103 Jespersen, 503 Jo, 491 Johanson, 591, 592, 596, 598 Jonare, 259 Jongeling, 517, 518, 524 Jordal, 498 Joseph, 318, 603 Jurik, 603 Kalitsunakis, 504 Kálmán, 168, 169, 387, 388, 502, 503

Kapanadze, 58, 124, 126, 140, 153, 162, 166 Karlsson, 179, 458, 505, 520 Karttunen, 396, 504 Kaseviˇc, 76 Kaufman, 360 Kay, 396, 504 Keenan, 87, 106, 115, 216, 219–221, 522 Keizer, 131 Kenesei, 387, 503, 522, 523 Keresztes, 586 Khavronina, 257, 505 Kim, 397, 500 King, 515–517, 521 Kiparsky, 585, 604 Kippasto, 504 Kirsner, 231, 232, 236, 237, 519 Kiss, 72, 86, 114, 116, 144, 145, 168, 206, 276, 304, 372, 373, 387, 491, 499, 501, 504 Kohonen, 504 Kolde, 559 Komlósy, 387, 388, 503 Kovtunova, 127, 462, 505 Kozinskij, 85 Kratzer, 99 Kress, 504 Krifka, 259 Kroll, 57, 58 Krupp, 413, 470, 498–500, 505, 509 Kruyt, 315 Krylova, 257, 505 Kuehner, 220 Küntay, 575 Kumashiro, 91 Kuno, 108, 257, 494, 557 Kuroda, 259, 260, 262, 266, 274 Kuz’mina, 585 Kuznecova, 586

159,

499,

169, 388,

Ladd, 316, 321, 464, 466, 500, 522, 525 Ladusaw, 259 Lafitte, 381, 463, 501, 502, 521, 524

Author index Lambrecht, 74, 81, 85, 102, 108, 215, 257, 258, 264, 266, 267, 283, 286, 287, 291, 302, 303, 464, 500, 508, 519 Lampe, 498 Langacker, 90–92 Lapteva, 9, 34, 54, 55, 57, 59, 124, 505 Larsson, 552 Laury, 575 Lavency, 222 Lazard, 303 Lehiste, 311 Lehman, Christine, 315 Lehmann, Christian, 46 Lehmann, Winfred P., 501 Leinonen, 584–587, 590, 604 Lerch, 520 Lescott, 602 Levelt, 477 Levy, 129, 133 Lewis, 210, 434, 515, 517 Lewy, 524 Li, 67, 70, 74, 91, 92, 97, 107, 112, 113, 144, 525 Linde, 392 Linell, 19, 58 Lockwood, 51, 83 Lötscher, 115 Lonzi, 259 Low, 317 Lundeby, 564 Lyons, 17, 519 Macaulay, 9, 57 MacCana, 257, 265, 267, 286, 430, 431, 433, 435–437, 441, 446, 514–518, 520, 521, 523 MacCoisdealbha, 515–517 Mackenzie, 131 MacWhinney, 316 Maes, 231, 235–238 Maidment, 315, 329 Maienborn, 95 Mann, 58 Marácz, 499

629

Marantz, 525 Marchello-Nizia, 503 Marouzeau, 515 Marty, 256, 261, 262, 300, 303 Masica, 501 Maslova, 279, 281, 284, 297, 500, 508, 523 Mathesius, 70, 260, 369, 393, 505 Matras, 240, 243, 264, 272, 273, 280, 287, 296, 297, 302 Matthews, 59, 208, 499, 519, 523, 526 Matthiessen, 58 Mayer, 557, 559, 578, 579 McCloskey, 514 Meillet, 369, 497, 503, 505, 506, 576 Menges, 604 Meyer-Lübke, 46, 51 Milano, 460 Miller, 32, 46, 57, 59, 60, 74, 78, 123, 125, 134, 144, 145, 166, 207, 208, 270, 284, 295, 297, 304, 477, 500, 508 Milroy, 9, 57 Mitchell, 446, 504 Mithun, 207, 265, 274 Mohanan, 498 Moravcsik, 583, 584 Morel, 59 Müller-Hauser, 258 Mukherjee, 501 Murray, 46 Muysken, 501, 526 Myhill, 215, 272, 304 Nedergaard Thomsen, 267 Nedjalkov, 115 Nemˇcenko, 585 Newman, 312 Nichols, 506 Nikolaeva, 398, 505 Nilsson, 596 Nooteboom, 315 Nuyts, 498 Nykrog, 498 Ó Siadhail, 77, 514–516, 521 Ong, 24

630

Author index

Ortiz de Urbina, 500, 501 Ortiz-Lira, 316, 319, 327, 328 Oschlies, 577, 604 Palmer, 23, 24, 205 Panzer, 585 Papp, 208 Paršin, 95, 99, 106 Payne, 498 Pedersen, 515, 516 Pennell Ross, 223, 226, 227 Perlmutter, 519 Perridon, 564 Pétursson, 504, 510 Philippaki-Warburton, 318 Pierson, 317 Pinkster, 220 Plank, 522, 583, 584 Platzack, 419, 510, 512 Polinsky, 68 Poole, 15, 16 Poppe, 442, 500, 515, 522, 526 Prince, 70, 159, 215, 314 Pritsak, 604 Prokuševa, 586, 587, 590 Quirk, 19, 45, 57–60, 135, 136, 157, 159, 171–173, 207–209, 520 Radford, 27 Rebuschi, 383, 463, 502 Rédei, 586–588 Rehbein, 603 Reichenkron, 576 Reinhammar, 564 Reinhart, 114, 116 Renzi, 409 Restan, 297 Ries, 509 Rijkhoff, 583 Roberts, 514 Rouveret, 514 Rowlands, 516 Ruge, 582, 584 Rulon, 58

Russ, 522 Rydén, 522 Salvi, 409 Sáenz-Badillos, 518 Sasse, 86, 257, 259–262, 264–266, 270– 273, 279–283, 286, 287, 291, 292, 295–298, 302–304, 519, 521, 546, 558, 584 Sauvageot, 500, 504 Scaglione, 413, 478, 499, 509, 510, 513 Schaller, 603 Scherpenisse, 510, 511, 513 Schlachter, 586–588 Schmerling, 257 Schroeder, 383, 545, 554, 593, 601, 602, 604 Schubiger, 317 Schulz, 33, 34 Schwyzer, 504, 515 Sgall, 500, 505 Shibatani, 76, 88, 115 Shisha-Halevy, 408, 433, 515–517 Sidner, 129, 133, 134, 140, 207, 231, 235 Siewierska, 152–154, 208, 397, 498, 499 Sigurðsson, 415, 419, 504, 510, 511 Sinclair, 311 Sinor, 498 Sirotinina, 45, 505 Sjoberg, 604 Sohn, 525 Sokolovskaja, 85 Somers, 221 Sornicola, 16, 29, 57, 115, 180, 190, 210, 256, 277, 279, 280, 377, 378, 403, 407, 443–445, 448, 452, 459–461, 464, 470, 477, 478, 498, 500, 508, 513, 515, 517, 519–521, 523–525 Sridhar, 525, 526 Srpová, 507, 508 Stachowski, 592 Stanchev, 397 Steele, 499 Stegmann, 220

Author index Stender-Petersen, 498 Stenson, 514, 516 Stephanides, 576 Stockwell, 317 Stölting, 579, 582 Svane, 576 Svartvik, 59 Svoboda, 115, 498, 505, 523 Svolacchia, 494, 501, 525, 526 Swarajya Lakshmi, 501 Swerts, 311 Szantyr, 220 Sørensen, 498 Taglicht, 330 Tajsner, 505 Tallerman, 514 Tarvainen, 470, 505, 511 Tauli, 393, 470, 504, 520 Terešˇcenko, 587 Tesnière, 508 Thomason, 360 Thompson, 58, 67, 70, 74, 91, 92, 97, 107, 112, 113, 144, 369, 393 Thráinsson, 419 Thumb, 504 Timm, 514 Tomlin, 114, 396, 438, 498, 514, 517 Tompa, 503 Touratier, 220, 222 Trudgill, 52 Tsimpli, 406, 504 Turunen, 575 Uhlíˇrová, 456, 460, 505, 521 Uhmann, 500 Ulrich, 257, 260, 265, 271, 272, 278, 282, 285–287, 292, 295–297, 303, 304, 518, 519 Ultan, 575, 600 Uroševi´c, 505, 509 Vähämäki, 521 Vaillant, 505, 506, 576

631

Vallduví, 32, 59, 76, 114, 116, 129, 130, 132, 134, 157, 162, 163, 170, 316 Van de Grift, 216, 223–225, 250 Van Heuven, 231, 232, 236 van Oosten, 500 van Schaaik, 596 Van Valin, 72, 87, 100, 106 Vanderslice, 317 Vater, 602 Vattuone, 260, 267 Veksler, 603 Vennemann, 498 Verpoorten, 515 Vilkuna, 372, 395, 396, 504, 505, 521, 522 Vincenot, 576 Vogel, 575, 600 Vogt, 384, 502 von Gabain, 604 von Stechow, 500 Vondrák, 51 Vonk, 226 Wackernagel, 509 Wackernagel-Jolles, 20, 59 Wagner, 515 Wahrig-Burfeind, 565, 603 Walker, 314 Wandruszka, 259 Ward, 275 Wartburg, 503 Watkins, 433, 434, 436, 515–518, 523 Watters, 525 Weber, 495, 526 Wehr, 138, 209, 228, 257, 267, 269, 274, 285–287, 291, 293, 557, 568 Weil, 369, 391, 499 Weinert, 32, 46, 57, 59, 60, 123, 125, 134, 144, 145, 166, 180, 190, 208, 210, 477, 508 Weinrich, 511 Wells, 317, 513, 514 Weyhe, 564 Williams, 432, 514–516 Willis, 515, 518

632

Author index

Yokoyama, 397, 398, 467, 500, 505–507, 519, 520, 523, 524 Yule, 135, 311

Zemskaja, 9, 12, 30, 31, 36, 41, 57, 58, 124, 126, 127, 139, 140, 153, 154, 159, 162, 164, 166, 204

Zaicz, 584

Zifonun, 511