Coordination Relations in the Languages of Europe and Beyond (Empirical Approaches to Languages Typology)

Coordination Relations in the Languages of Europe and Beyond Caterina Mauri Mouton de Gruyter Coordination Relations...

Author: Mauri | Caterina

43 downloads 783 Views 6MB Size Report

This content was uploaded by our users and we assume good faith they have the permission to share this book. If you own the copyright to this book and it is wrongfully on our website, we offer a simple DMCA procedure to remove your content from our site. Start by pressing the button below!

Report copyright / DMCA form

DOWNLOAD PDF

Coordination Relations in the Languages of Europe and Beyond

Caterina Mauri

Mouton de Gruyter

Coordination Relations in the Languages of Europe and Beyond

≥

Empirical Approaches to Language Typology 42

Editors Georg Bossong Bernard Comrie Yaron Matras

Mouton de Gruyter Berlin · New York

Coordination Relations in the Languages of Europe and Beyond by Caterina Mauri

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 Mauri, Caterina, 1981⫺ Coordination relations in the languages of Europe and beyond / by Caterina Mauri. p. cm. ⫺ (Empirical approaches to language typology ; 42) Includes bibliographical references and index. ISBN 978-3-11-020439-1 (cloth : alk. paper) 1. Europe ⫺ Languages ⫺ Coordinate constructions. I. Title. P380.M28 2008 4151.7⫺dc22 2008032765

Bibliographic information published by the Deutsche Nationalbibliothek The Deutsche Nationalbibliothek lists this publication in the Deutsche Nationalbibliografie; detailed bibliographic data is available in the Internet at http://dnb.d-nb.de.

ISBN 978-3-11-020439-1 ISSN 0933-761X © Copyright 2008 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. Printed in Germany.

Acknowledgements

This book is a revised version of my doctoral dissertation (University of Pavia 2007). It is not easy to mention all the people who played a role in structuring the ideas and the pages of this work. First of all, I would like to thank people at the Dipartimento di Linguistica Teorica e Applicata of the University of Pavia for forming such an inspiring working environment and for giving me the opportunity to experiment what it really means to do linguistics, in Italy and abroad. I am especially grateful to those who followed the development of this research at its various stages. I would like to express my deepest gratitude to Sonia Cristofaro, who had a crucial impact in my growing up as a linguist, since the day I decided to leave Italo Calvino for typology. She oversaw this work from the very beginning and in all its phases, spending a lot of time and energies in carefully reading drafts of this book and providing precious advice. I am and will always be indebted to her for all the support and inspiration, linguistically and otherwise. I would like her to know that I never felt alone along the way. Thank you to Anna Giacalone and Paolo Ramat, who believed in me, supported me, involved me in exciting research projects. They always found the time to listen to my ideas, sharing with me their extensive and deep knowledge of linguistics. I learned a lot from them, and this has been a real privilege for me. I sincerely thank Michele Prandi, for his precious philosophical perspective, Silvia Luraghi, for the valuable bibliographical hints and for her ability to bring problems into focus, and Gianguido Manzelli, for having introduced me to the fascinating history of Slavic languages. Andrea Sansò has been an essential source of emotional support and intellectual exchange, and I am sincerely grateful to him for his friendship and trust. I discussed parts of this research, at different stages and in different circumstances, with Cecilia Andorno, Michela Cennamo, Pierluigi Cuzzolin, Tullio De Mauro, Maurizio Gnerre, John Hawkins, Paolo Poccetti, Davide Ricca, Edoardo Lombardi Vallauri, Bernhard Wälchli, Björn Wiemer. I would like to thank them because their words have been, more or less consciously, very useful hints on the direction(s) to follow. Part of my Ph.D thesis has been carried out in Germany, at the University of Erfurt and at the Freie Universität of Berlin. The time spent there has been incredibly important for my personal growth, both from a professional and a

vi Acknowledgements human point of view. Besides learning a beautiful language, I met a number of extremely valuable people. I would like to thank Christian Lehmann for the opportunity to work at the Sprachwissenschaft Institut of the University of Erfurt and for the time he took to discuss my unripe ideas, helping me to ﬁnd a direction along which they could grow. I also would like to express my gratitude to Yoko Nishina and especially Amani Bohoussou. He has taught me a lot of interesting things about NànáfwˆE, has widely discussed with me the idea of the irreality of alternatives and has ultimately made my winter in Erfurt much happier with his generous smile. I have thoroughly enjoyed and beneﬁted from my stay in Berlin. There I had extremely interesting talks with Norbert Dittmar, Kristina Kotcheva and Marcela Adamíková, whose information on Slavic languages have been crucial to ﬁgure out the fuzzy meaning of a. I am especially indebted to Ewald Lang, who got interested in my research and participated with his curiosity and his personal library, a small part of which is now with me in Milan. In August 2004, I spent one month working at the Max Planck Institute für evolutionäre Anthropologie in Leipzig, where I had the opportunity to meet and talk with brilliant linguists, whose method and thought have deeply inspired my work. Talking with Martin Haspelmath has been a crucial moment in the elaboration of my dissertation and I would like to express my deep gratitude to him for his wise as well as truly inspiring advice. His work on coordinating constructions was the point of departure of my research and a constant reference model during its development. Furthermore, I received valuable comments from the audiences at the University of Bergamo, the Max Planck Institute für evolutionäre Anthropologie at Leipzig, the Colloquia at the Freie Universität, the Humboldt Universität of Berlin and the University of Erfurt, the Leipzig Students’ Conference in Linguistics, the Workshop on Semantic Maps during the ALT 7 Conference in Paris, the MORBO / CONF meeting at the University of Bologna and the Ph.D. students conference at the SITLEC Department of Forlì. I would also like to thank Yaron Matras for the comments he provided on the manuscript as an editor of the EALT series, Ursula Kleinhenz and Wolfgang Konwitschny, for their constant help at the very ﬁnal stage of this path. A special thank goes to all the people who kindly ﬁlled out the questionnaire, providing precious information about their native languages. In particular, I would like to thank Agnieszka Latos, Najla Rebej, Manana Topadze, Amani Bohoussou, Yoko Nishina and Kristina Kotcheva for discussing data with me.

Acknowledgements

vii

There are also a number of people I would like to thank, because they shared time, ideas and laughters with me during these years of research: Valentina Albino, Giorgio Arcodia, Irene Cordovani, Federica Da Milano, Martina Fanari, Francesca Frontini, Amira Lakhdar, Cristina Mariotti, Francesca Mazzariello, Nicoletta Puddu, Najla Rebej, and Lorenzo Spreaﬁco. A special thank goes to Chiara Mauri and Maicol Formentelli, who corrected and revised my English. I am extremely grateful to all the non-linguists who patiently listened to my reasoning about coordination and gave me their precious feedback and help. I know it might have been a sort of sacriﬁce sometimes and I really appreciate it. Thanks to those with whom I share everything but linguistics: it’s there I ﬁnd the energies to dive back into research. The last thank goes to my family, who always encouraged me and to whom this book is dedicated. È tra le vostre parole che ho iniziato a cercare. . . di imparare a parlare.

Contents Acknowledgements . List of ﬁgures . . . . List of tables . . . . . List of abbreviations 1

2

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. v . xiii . xiv . xvii

Theoretical foundations: the notion of coordination 1.1 Introduction to this research . . . . . . . . . . . . . . . . . 1.1.1 Overview . . . . . . . . . . . . . . . . . . . . . . . 1.1.2 A functional-typological approach to coordination . 1.1.3 Outline of the work . . . . . . . . . . . . . . . . . . 1.2 The language data: a focus on Europe . . . . . . . . . . . . 1.2.1 The European sample and the Comparison sample . 1.2.1.1 The European sample . . . . . . . . . . . 1.2.1.2 The Comparison sample . . . . . . . . . . 1.2.2 The questionnaire . . . . . . . . . . . . . . . . . . . 1.2.3 Two levels of analysis and the twofold organization of the study . . . . . . . . . . . . . . . . . . . . . . 1.3 The notion of coordination . . . . . . . . . . . . . . . . . . 1.3.1 Traditional deﬁnitions of coordination . . . . . . . . 1.3.2 Alternative approaches to the deﬁnition of coordination . . . . . . . . . . . . . . . . . . . . . 1.3.3 The functional ‘parallelisms’ of coordination . . . . 1.3.3.1 Semantic parallelism . . . . . . . . . . . 1.3.3.2 Conceptual parallelism . . . . . . . . . . 1.3.3.3 Pragmatic parallelism . . . . . . . . . . . 1.3.4 Coordination relation: a cross-linguistically valid deﬁnition . . . . . . . . . . . . . . . . . . . . . . . 1.3.5 Types of coordination relations . . . . . . . . . . . . Parameters of analysis 2.1 Object of analysis and methodology . . . . . . . . . . . . . 2.1.1 Delimiting the scope . . . . . . . . . . . . . . . . . 2.1.2 Methods for cross-linguistic comparison: selecting the parameters . . . . . . . . . . . . . . . . . . . .

1 1 1 6 10 11 14 14 17 19 22 23 23 28 31 32 35 37 41 44 49 49 49 53

x Contents 2.2 2.3

2.4 2.5 3

4

Semantic parameters: temporality, conﬂict and aim . . . . Morphosyntactic parameters . . . . . . . . . . . . . . . . 2.3.1 The coding of the relation . . . . . . . . . . . . . 2.3.1.1 Presence of an overt coordinating marker 2.3.1.2 Morphophonological complexity . . . . 2.3.2 The coding of the SoAs . . . . . . . . . . . . . . . 2.3.2.1 Syntactic parallelism of the construction Degree of coding . . . . . . . . . . . . . . . . . . . . . . Cross-linguistic variation: an overview . . . . . . . . . . .

Combination relations 3.1 The notion of combination relation . . . . . . . . . . . 3.1.1 Combination as cooccurrence . . . . . . . . . 3.1.2 Combination types . . . . . . . . . . . . . . . 3.1.2.1 Temporal combination . . . . . . . . 3.1.2.2 Atemporal combination . . . . . . . 3.2 The coding of combination relations . . . . . . . . . . 3.2.1 Presence of overt coordinating markers . . . . 3.2.2 Morphophonological complexity of the attested overt markers . . . . . . . . . . . . . . . . . . 3.3 The coding of combined SoAs . . . . . . . . . . . . . 3.4 Functional motivations in the coding of combination . 3.5 Language data . . . . . . . . . . . . . . . . . . . . . . Contrast relations 4.1 The notion of contrast relation . . . . . . . . . . 4.1.1 Contrast as conﬂicting cooccurrence . . . 4.1.2 Contrast types . . . . . . . . . . . . . . 4.1.2.1 Oppositive contrast . . . . . . 4.1.2.2 Corrective contrast . . . . . . . 4.1.2.3 Counterexpectative contrast . . 4.2 The coding of contrast relations . . . . . . . . . 4.2.1 The contrast coding implication . . . . . 4.2.2 The contrast conceptual space . . . . . . 4.2.3 The contrast coding complexity hierarchy 4.3 The coding of conﬂicting SoAs . . . . . . . . . . 4.4 Functional motivations in the coding of contrast . 4.5 Language data . . . . . . . . . . . . . . . . . . .

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

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

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

. . . . . . . . .

54 57 57 60 63 65 66 70 77 80 80 80 83 84 85 86 86

. . . . . . .

. . . . . . .

. . . . . . .

. . . .

. . . .

. 97 . 100 . 103 . 110

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

119 120 120 121 122 123 124 126 126 130 136 139 141 147

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

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

Contents

5 Alternative relations 5.1 The notion of alternative relation . . . . . . . . . . 5.1.1 Alternative as non-cooccurrence . . . . . . 5.1.2 Alternative types . . . . . . . . . . . . . . 5.1.2.1 Simple alternative . . . . . . . . 5.1.2.2 Choice-aimed alternative . . . . 5.2 The coding of alternative relations . . . . . . . . . 5.2.1 The alternative coding implication . . . . . 5.2.2 The alternative irreality implication . . . . 5.3 The coding of alternative SoAs . . . . . . . . . . . 5.4 Functional motivations in the coding of alternative . 5.5 Language data . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

6 Combination, contrast and alternative relations in comparison 6.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . 6.2 Combination and contrast . . . . . . . . . . . . . . . . . . . 6.2.1 The degree of coding . . . . . . . . . . . . . . . . . 6.2.1.1 The combination-contrast coding implication . . . . . . . . . . . . . . . . . 6.2.1.2 The combination-contrast conceptual space . . . . . . . . . . . . . . . . . . . . 6.2.1.3 The coding complexity hierarchy of combination and contrast relations . . . . 6.2.2 The combination-contrast parallelism implication . . 6.2.3 Functional motivations . . . . . . . . . . . . . . . . 6.2.4 Language data . . . . . . . . . . . . . . . . . . . . 6.3 Combination and alternative . . . . . . . . . . . . . . . . . 6.3.1 The degree of coding . . . . . . . . . . . . . . . . . 6.3.1.1 The combination-alternative coding implication . . . . . . . . . . . . . . . . . 6.3.1.2 Semantic domains: combination, alternative and irreality . . . . . . . . . . 6.3.1.3 The combination-alternative coding complexity implication . . . . . . . . . . 6.3.2 The combination-alternative parallelism implication 6.3.3 Functional motivations . . . . . . . . . . . . . . . . 6.3.4 Language data . . . . . . . . . . . . . . . . . . . . 6.4 Combination, contrast and alternative: concluding remarks .

xi 154 155 155 157 159 160 161 161 170 183 184 187 194 194 196 196 196 199 206 209 211 216 230 230 230 231 234 237 239 242 256

L== >=C4=CB

CBHF5GH 5B8 5@H9FB5H=J9 =B 7CAD5F=GCB -<9 7CCF8=B5H=CB 7C8=B; =AD@=75H=CBG CBHF5GH 5B8 5@H9FB5H=J9 HKC 8=::9F9BH G9A5BH=7 5L9G

#! # "BHFC8I7H=CB CA6=B5H=CB F9@5H=CBG CA6=B5H=CB =B IFCD9 GMBH<9G=G 5B8 7CB7@IG=CBG CBHF5GH F9@5H=CBG CBHF5GH =B IFCD9 GMBH<9G=G 5B8 7CB7@IG=CBG @H9FB5H=J9 F9@5H=CBG @H9FB5H=J9 =B IFCD9 GMBH<9G=G 5B8 7CB7@IG=CBG CB7@IG=CBG aB8IH(Fb @5B;I5;9G

#! !"!

DD9B8=L DD9B8=L DD9B8=L DD9B8=L DD9B8=L

-<9 @5B;I5;9G =B H<9 G5AD@9G ,CIF79G C: =B:CFA5H=CB 9C;F5D<=75@ @C75H=CB C: H<9 @5B;I5;9G -<9 EI9GH=CBB5=F9 -<9 7CBGHFI7H=CBG 9L5A=B98 =B H<9 GHI8M

'CH9G +9:9F9B79G "B89L C: @5B;I5;9G "B89L C: 5IH97HG

List ofigures

1

The cooccurrence vs. non-cooccurrence dimension . . . . . .

2 3

The non-cooccurrence dimension . . . . . . . . . . . . . . . . 155 Alternative: elaboration of the non-cooccurrence dimension . . 156

4 5 6 7 8 9 10

Malchukov’s Semantic Map . . . . . . . . . . . Comparison between contrast semantic maps . . Coordination relations: summary . . . . . . . . . Coordination relations in comparison: summary . The combination-contrast semantic dimension . . The combination-alternative semantic dimension Semantic dimensions of coordination . . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

204 205 263 265 269 271 273

11 12 13 14 15

The languages of the SAE . . . . . . . . . . . . . . . . . Constructions for combination in the languages of Europe Constructions for contrast in the languages of Europe . . Constructions for alternative in the languages of Europe . And-But-Or languages . . . . . . . . . . . . . . . . . .

. . . . .

. . . . .

. . . . .

275 280 284 288 290

16 17

Location of the examined European languages . . . . . . . . . 307 Location of the examined non-European languages . . . . . . 308

. . . . . . .

. . . . . . .

. . . . . . .

81

!" "!

+9DF9G9BH98 D<M@5 =B H<9 IFCD95B G5AD@9 +9DF9G9BH98D<M@5 =B H<9 CAD5F=GCB G5AD@9 %5B;I5;9G 5B5@MN98 K=H< EI9GH=CBB5=F9G

&CFD

C8=B; C: 7CBHF5GH F9@5H=CBG 5HH9GH98 HMD9G HH9GH98 7CBHF5GH G9A5BH=7 A5DG -<9 7CBHF5GH 7C8=B; 7CAD@9L=HM <=9F5F7<M IFCD95B ,5AD@9 (J9FH 7CCF8=B5H=B; A5F?9FG 5B8 G9A5BH=7 8CA5=B CAD5F=GCB ,5AD@9 (J9FH 7CCF8=B5H=B; A5F?9FG 5B8 G9A5BH=7 8CA5=B IFCD95B ,5AD@9 &CFD

C8=B; C: 7

List of ables

xv

26 27

Comparison Sample. Morphophonological complexity . . . . 191 Comparison Sample. Syntactic parallelism . . . . . . . . . . . 192

28 29 30 31 32

47

Overt markers for combination and contrast relations . . . . . 196 The combination-contrast semantic maps . . . . . . . . . . . . 200 The combination-contrast coding complexity hierarchy . . . . 207 The combination-contrast parallelism implication . . . . . . . 210 European Sample. Combination and contrast in comparison: morphophonological complexity . . . . . . . . . . . . . . . . 216 Comparison Sample. Combination and contrast in comparison: morphophonological complexity . . . . . . . . . . . . . . . . 219 European Sample. Combination and contrast in comparison: semantic domains . . . . . . . . . . . . . . . . . . . . . . . . 222 Comparison Sample. Coding of combination and contrast in comparison: semantic domain . . . . . . . . . . . . . . . . . . 224 European Sample. Combination and contrast in comparison: syntactic parallelism . . . . . . . . . . . . . . . . . . . . . . . 226 Comparison Sample. Combination and contrast in comparison: syntactic parallelism . . . . . . . . . . . . . . . . . . . . . . . 228 Overt markers for combination and alternative relations . . . . 230 The combination-alternative coding complexity hierarchy . . . 235 The combination-alternative parallelism implication . . . . . . 238 European Sample. Combination and alternative in comparison: morphophonological complexity . . . . . . . . . . . . . . . . 242 Comparison Sample. Combination and alternative in comparison: morphophonological complexity . . . . . . . . . . . . . . . . 245 European sample. Combination and alternative in comparison: semantic domains . . . . . . . . . . . . . . . . . . . . . . . . 248 Comparison Sample. Combination and alternative in comparison: semantic domains . . . . . . . . . . . . . . . . . . . . . . . . 250 European Sample. Combination and alternative in comparison: syntactic parallelism . . . . . . . . . . . . . . . . . . . . . . . 252 Comparison Sample. Combination and alternative in comparison: syntactic parallelism . . . . . . . . . . . . . . . . . . . . . . . 254 Coding of contrast and alternative relations: attested types. . . 257

48 49

Languages of the European sample . . . . . . . . . . . . . . . 298 Languages of the Comparison sample . . . . . . . . . . . . . 300

33 34 35 36 37 38 39 40 41 42 43 44 45 46

xvi

50 51 52 53 54

Sources for the European sample . . . . . . . Sources for the Comparison sample . . . . . The questionnaire used in this research . . . . Constructions examined: European sample . . Constructions examined: Comparison sample

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

302 304 309 312 317

List of abbreviations The following abbreviations are used in the interlinear glosses of language examples. In examples taken from a descriptive grammar, the glosses are generally the same as those used in the grammar. In examples taken from questionnaires, the glosses have been made either by the author or by the informant, and in both cases at least one reference grammar has been consulted. In examples taken from questionnaires, the relevant coordinating markers have been glossed with the general label COORD, without deﬁning the speciﬁc coordination relation expressed. The reason for this choice is to keep the presentation of data as separate as possible from its interpretation. 1sg 2sg 3sg 1pl 2pl 3pl ABS ACC ADESS ADVB ALIEN ALL AOC AOR ART AUG AUX CA COMP COND CONJ CONT CONV COORD COP CORE D : PVG DAT

ﬁrst person singular second person singular third person singular ﬁrst person plural second person plural third person plural absolutive accusative adessive adverbial alienable allative aorist converb aorist article augmentative auxiliary connective adverbial complementizer conditional conjunctive contemporary mood converb coordinating marker copula non-nominative core article distal extension: point of view of goal dative

IMPF IMPFV INCOMP INF INFER INFL IND INDEF INESS INSTR INT INTENS INTERJ INTR IRR LOC M MOD N NEG NOM NR

nrp OBJ OBL OPT POL POSS

imperfective imperfective past incompletive inﬁnitive inferential inﬂectional particle indicative indeﬁnite inessive instrumental interrogative intensiﬁer interjection intransitive irrealis locative masculine modal neuter negation nominative nominalizer non-realis past object general oblique article optative polite possessive

xviii

List of abbreviations

DEIC DECL DEF DIREC DIM DS DU DUB DUR EMPH ERG EXCL

f FOC FUT GEN HON HUM HORT ILL IMP

deictic declarative deﬁnite directive diminutive different subject dual dubitative durative emphatic ergative exclusive feminine focus future genitive honoriﬁc human hortative illative imperative

PRF PROG PRON PRS PST PT PTCP Q QUOT R REF

rp SEQ SBST SBJ

sf SS SUP TOP TR WP

perfective marker progressive pronoun present past particle participle question marker quotation marker realis referential realis past sequential substantivizer subject singular feminine same subject superessive topic transitive witnessed past

Other symbols and abbreviations → > d G Gx SoA

implies separation marker between the cut-off points of an implicational hierarchy dedicated marker or construction general marker or general construction x identiﬁes a given general construction attested in a speciﬁc language state of affair

Chapter 1 Theoretical foundations: the notion of coordination 1.1

Introduction to this research

1.1.1 Overview This work investigates the cross-linguistic coding of the three conceptual relations of combination (1.1), contrast (1.2) and alternative (1.3) between functionally equivalent states of affairs, that is, states of affairs characterized by an autonomous cognitive proﬁle (neither is presented in the perspective of the other) and by the presence of some illocutionary force (see section 1.3). (1.1) The summer ends and everybody goes back to work. (1.2) The summer ends but many people are still on holiday. (1.3) Are you coming to the cinema tonight or do you relax at home? These conceptual relations are usually discussed within the traditional approach to coordination under the labels ‘conjunction’, ‘adversativity’ and ‘disjunction’ respectively, by virtue of their being often coded by means of similar structurally symmetric constructions. In this work however, the notion of coordination will not be deﬁned in the traditional way, i.e. in structural terms, but it will be identiﬁed on the basis of functional criteria. A given relation will be deﬁned as a coordination relation when it is established between two conceptually and functionallyparallel states of affairs (see section 1.3 for a detailed deﬁnition of coordination in functional terms). A given construction will be thus deﬁned as coordinating when it is used to establish a coordination relation, independently of its morphosyntactic properties. It is well known that coordination relations may be established at various syntactic levels (see Dik 1968 and Payne 1985), but in this study only sentential coordination and coordination between verbal phrases will be examined, since object of this analysis are the relations of combination, alternative and contrast between states of affairs. As can be seen in examples (1.4), (1.5) and (1.6), the morphosyntactic properties of the constructions expressing combination, contrast and alternative may vary. In example (1.4) from Japanese, the ﬁrst state of affair is coded

2 Theoretical foundations: the notion of coordination by a verb form sufﬁxed by the combination marker -i, which cannot receive markers of tense, aspect or mood and depends on the last verb for the speciﬁcations of such categories . The same happens in example (1.5) from Korean, in which the ﬁrst verb is sufﬁxed by the alternative marker -kena and depends on the last verb for further speciﬁcations of tense, aspect and mood. (1.4) Japanese watashi-wa hatarak-i piitaa-wa neru 1sg-TOP work-COORD Peter-TOP sleep ‘I work and Peter sleeps.’1 (Y.N., from questionnaire) (1.5) Korean Minsu-ka o-kena nae-ka ka-n-ta. Minsu-NOM come-COORD 1sg-NOM go-INCOMP-DECL ‘Minsu comes here or I go there.’ (Y.M.S., p.c.) Such a syntactic dependency is normally associated with subordinating constructions, whereas coordination is traditionally characterized in terms of syntactic independence and symmetry of the coordinated clauses (as can be seen in (1.6)). However, since this study compares all the strategies which express the three coordination relations of combination, contrast and alternative, without restrictions on the morphosyntactic properties they may show, cases like (1.4) and (1.5) will also be included in the analysis. (1.6) Hungarian, Finno-Ugric, Uralic Nagyon szomjas vagyok, de nem szeretem a very thirsty be.1sg COORD NEG love.1sg DEF narancslét. orange.juice ‘I’m very thirsty but I don’t like orange juice.’ (F.M., from questionnaire) The starting point of this research is thus located at the functional level, i.e. the level which deals with the aspects of language related to its communicative function, like semantics, information structure and the mechanisms of

Introduction to this research

3

interpersonal communication. Once the notion of coordination relation is deﬁned at this level (see section 1.3), the various strategies attested to express such relations are examined in a broad language sample, with no restrictions on the morphosyntactic properties they may show. Coordination has been widely examined in more formal approaches, but with a different purpose from the one pursued here. Instead of describing the cross-linguistic variation in the coding of certain conceptual relations, such studies focus on the identiﬁcation of an abstract formal structure that deﬁnes coordination universally, regardless of the conceptual relations expressed and usually based on one language or on a restricted sample of languages. Scholars like Johannessen (1998), Camacho (2003) and Rebuschi (2005) consider coordinate structures as headed constructions. In particular, coordinate structures are treated like conjunction phrases, in which the coordinator is the head, the ﬁrst conjunct the speciﬁer, and the second conjunct the complement. As Borsley (2005) pointed out, this conception is widely accepted within Principles and Parameters theories, but it is rejected within other frameworks. Borsley (2005) himself rejects the idea that coordinate structures are conjunction phrases and argues that they are rather to be analyzed as adjunct phrases. This different conception of coordination is held by Bresnan (2000), Yuasa and Sadock (2002) and Cormack and Smith (2005). The generative approaches focus on the formal structure characterizing coordination, regardless of the conceptual relations expressed and of the crosslinguistic variation in the coding of such relations. For this reason the debate within the generative framework and the relevant literature will not be addressed here (see Borsley 2005 and Cormack and Smith 2005 for a detailed discussion on the generative approaches to coordination). In the typological literature, on the other hand, many cross-linguistic studies have been conducted on subordinating constructions (cf. Cristofaro 2003 and Givón 1990 and 2001), but the coding of coordination relations has received less attention. A broad cross-linguistic perspective has been adopted by Payne (1985) and Haspelmath (2004 and 2007), in which an overall analysis of coordination and a description of its various facets are provided. Haspelmath’s work, in particular, takes into account the problematic deﬁnition of coordination, its morphosyntactic properties, the various syntactic levels on which it occurs, the ellipsis phenomena, and the semantic distinctions between conjunction, disjunction and adversativity. Two other major typological works, both focusing on nominal coordination, have been made by Stassen (2001) and Wälchli (2005). Stassen’s

+74>A4C820; 5>D=30C8>=B C74 =>C8>= >5 2>>A38=0C8>= GHI8M =89BH=:=9G H<9 K=89GDF958 8=GH=B7H=CB 69HK99B =3%5B;I5;9G 5B8 .8C7 %5B;I5;9G -<9 :CFA9F IG9 8=::9F9BH GHF5H9;=9G :CF 7CA=H5H=J9 5B8 7CB>IB7 H=CB 9AD@CM=B; 5 GD97=:=7 GHF5H9;M :CF 7CB>IB7H=CB K<=@9 H<9 @5HH9F IG9 H<9 G5A9 ACFDIB7H=J9 7CBGHFI7H=CB K<5H <9 75@@G a7C7CADCIB8Gb K<=7< 5F9 GD97=5@ H=;IB7H=J9 7CCF8=B5H=CB =G 5B5@MN98 &CGH C: H<9 HMDC@C;=75@ GHI8=9G CB 7CCF8=B5H=CB IB7H=J9 7CBGHFI7H=CBG 5B8 @95J9 7CBGHFI7H=CBG 7C8=B; 7CBHF5GH 5B8 5@H9FB5H=J9 F9@5H=CBG 5 6=H =B H<9 657?;FCIB8 -<=G GHI8M HA?7>BH=C02C82 D5F5A9H9FG K=@@ 69 H<9B 58CDH98 =B H<9 5B5@MG=G C: 5HH9GH98 7CBGHFI7H=CBG 9LDF9GG=B; H<9 J5F=CIG 7CCF8=B5H=CB F9@5H=CB HMD9G H<9 7C8=B; C: H<9 F9@5H=CB 5B8 H<9 7C8=B; C: H<9 GH5H9G C: 5::5=FG -<9 7C8=B; C: H<9 F9@5H=CB 7CB79FBG H<9 DF9G9B79 5B8 H<9 ACFD<08= C: 957< 5HH9GH98 7CBGHFI7H=CB -<9 G9A5BH=7 8CA5=B C: 5 7CBGHFI7H=CB =G H<9 G9H C: 7CCF8=B5H=CB F9@5H=CBG H<5H =H A5M 9LDF9GG 8=GH=B;I=G<=B; 69HK99B ;9B9F5@ GHF5H9;=9G IG98

Introduction to this research

5

for more than one relation, and dedicated strategies, speciﬁcally used for just one coordination relation. The analysis of the attested coordinating constructions on the basis of semantic and morphosyntactic parameters reveals implicational constraints on cross-linguistic variation, both in the coding of each coordination relation and in the comparison between them. The degree of semantic speciﬁcity of a construction will turn out to be directly proportional to the morphophonological complexity of the coordinating marker used: the higher the number of relations expressed, the simpler the marker’s morphophonology (cf. chapters 4 and 6). Moreover, markers that code combination relations, either general or dedicated, are morphophonologically simpler than those expressing contrast and alternative (cf. section 6.4). Furthermore, some coordination relations are more likely to be expressed without any overt markers, as a result of their being more easily inferable from the context. In particular, if in a given language a contrast relation generated by the denial of an expectation is expressed by simple juxtaposition, this strategy will also be available for contrast relations generated by opposition and correction. In alternative relations, on the other hand, if no marker is used in the expression of a simple alternative, then no marker will be used for an alternative where a choice is required. The coding of the states of affairs will turn out to be affected by the internal semantics of the established relation. In general, coordination relations show a cross-linguistic tendency for syntactically parallel constructions, but if a language uses a non-parallel strategy for the expression of a coordination relation at all, it will be certainly used for the sequential combination of two states of affairs. Moreover, the exam of the attested semantic domains allows for a thorough analysis of how the three coordination relations at issue are related to each other. In particular, through the identiﬁcation of semantic maps, the attested semantic domains highlight a conceptual hierarchical organization of coordination, pointing to the unmarked role of combination with respect to contrast and alternative, by which it turns out to be implied (see chapter 6). The three coordination relations are thus not simply equivalent semantic types of coordination, as they are traditionally treated, and do not stand on the same hierarchical level. On the contrary, not only combination results to be the basic relation coded by the most basic constructions, but it also constitutes the origin of two separate axes of progressive semantic speciﬁcity.

6 Theoretical foundations: the notion of coordination The combination-contrast conceptual space goes from combination to contrast through the progressive speciﬁcation of the discontinuity characterizing the linked states of affairs (see Fig. 8). The combination-alternative conceptual space, instead, goes from combination to alternative through the progressive speciﬁcation of the irreality of the states of affairs, which are identiﬁed as replaceable possibilities (see Fig. 9). As will be explained in section 1.2, this work has a strong focus on Europe. The availability of ﬁrst hand data and good descriptions for European languages allows to draw a detailed picture of the European variation in the coding of coordination relations. In chapter 7, a number of maps will be presented for the constructions attested in Europe, showing that some features pertain to the most peripheral areas, whereas others are displayed by the languages spoken in the nuclear part of Europe. Moreover, a particular language type will be identiﬁed which shows a set of morphosyntactic features in the coding of coordination relations that is rarely attested outside Europe. Since such languages are characterized, among other features, by the presence of a general overt marker for each of the three relations of combination, alternative and contrast, they will be called ‘And-But-Or languages’. As brieﬂy stated in the course of this introductory section, this work is based on a functional deﬁnition of the object of analysis, aims at identifying regular patterns within the cross-linguistic variation and explains these patterns in terms of functional and semantic factors. These methodological and theoretical aspects are characteristic of the approach to the study of language adopted in this research, namely the functional-typological framework (see Croft 1990 and Comrie 1981).

1.1.2

A functional-typological approach to coordination

The functional-typological approach to the study of language is characterized by two major assumptions: the importance of the cross-linguistic comparison as the main source of data and the crucial role of functional factors, both in deﬁning the research domain and in explaining the results. The main goal of the functional-typological approach is to analyze the grammatical organization of the languages of the world and to identify possible regular patterns of variation, which are assumed to mirror the organization of experience in the human mind and are thus explained in terms of functional principles, such as semantic and pragmatic motivations (Cristofaro and Ramat 1999: 15-32).

Introduction to this research

7

Given the central role played by linguistic data in typological research, the selection of a representative language sample is a key methodological aspect, as pointed out by Rijkhoff et al. (1993: 171). Depending on the goal of the research, the ideal typological sample may be a probability sample, which consists of geographically and genetically independent languages, or a variety sample. In the latter case languages are selected in such a way that all linguistic families are represented and every family is represented in proportion to the degree of its internal differentiation, i.e. the number of branches and subgroups it shows. As will be made clear in section 1.2, no typologically balanced sample will be used in this work. Rather, a sample of 37 European languages will be compared to a sample of 37 non-European languages. Even though the two samples together amount to 74 languages, covering more than twenty language families, the availability of good descriptions and ﬁrst hand data has biased the selection of the examined languages. Hence the sample cannot be considered balanced, but rather a convenience one. This study fully aligns to the functional-typological perspective in assigning great importance to functional factors in the deﬁnition of the research domain. According to this approach, functional situations are assumed to be universal across languages, unlike morphosyntactic constructions. Human beings share a common set of cognitive mechanisms and intentions, a common system of perception and thought, which deals with basic categories such as time, space and motion, and a common set of communicative needs, such as the establishment of given/new information, the communication of certain relations among concepts and not others, and the creation of cohesion within the discourse. Since every human language has to deal with all these aspects in order to fulﬁll its main function of means of communication, the deﬁnition of the object of comparison in terms of such functional factors - and in any case external to the language - allows for a universal applicability. Accordingly, the deﬁnition of the object of this analysis, namely what a coordination relation is, will be based on functional criteria, and not on morphosyntactic ones (see section 1.3). The structural means through which languages code a given conceptual situation vary signiﬁcantly, even though not randomly. Within the functionaltypological approach, constraints on cross-linguistic variation deﬁne what a possible language is, and are thus named linguistic universals (Croft 2003: chapter 3). Linguistic universals may be unrestricted or implicational, de-

8 Theoretical foundations: the notion of coordination pending on the number of parameters involved and on the number of types they identify. Unrestricted universals identify only one type on the basis of one parameter: all languages of the world behave in the same way with respect to that given parameter. Implicational universals, on the other hand, identify several types on the basis of at least two parameters. The typical formulation of such universals is in terms of logical implications, according to which in a given language the presence of a feature X implies the presence of another feature Y (Croft 2003: 54). As will be argued in chapters from 3 to 6, the cross-linguistic coding of coordination shows a regular variation, which can be formulated both in terms of implications and in terms of implicational hierarchies. Yet, no unrestricted generalization will be made. A particular type of implicational cross-linguistic pattern of variation, semantic maps (Haspelmath 2003), will play a crucial role in the analysis of coordination relations. The aim of a semantic map is the identiﬁcation and representation of the conceptual situations that are expressed by means of the same construction, within and across languages. Croft (2003: 144-52) makes a distinction between semantic map, which represents the multifunctionality of a given construction in a given language, and conceptual space, the overall representation of which conceptual situations may be expressed by the same construction across languages. Other labels have been used in the typological literature to denote this type of generalization. Kortmann (1997: 177), for instance, talks about ‘cognitive map’, emphasizing the presumed connections between the functions mapped in the representation and the cognitive organization of the encoded meanings. In this work, the labels ‘semantic map’ and ‘conceptual space’ will be used accordingly to Croft’s assessment. The direction of analysis followed in the creation and interpretation of a semantic map is different from the one adopted in the identiﬁcation of implicational universals. In the latter, the research goes from a functional deﬁnition to the attested morphosyntactic coding strategies. In the former, on the other hand, the research goes from the attested morphosyntactic constructions - already identiﬁed as expressing a given function - to the conceptual situations they speciﬁcally encode. The basic idea of a semantic map is to highlight the fact that a multifunctional construction does not express random meanings, but that these meanings are organized in an orderly way. Given a map of the type ‘function X function Y - function Z - function W’, it has to be read in the following way: in a given language, if a single construction expresses function X and function

Introduction to this research

9

W, it will also express the intermediate functions on the map. For instance, given a map ‘direction - recipient - beneﬁciary’, if the same construction is used to express beneﬁciary and direction it will also be used to denote the recipient role. In this study, an interesting conceptual space will be identiﬁed by comparing the general constructions expressing combination and contrast relations, as will be explained in chapters 4 and 6. Combination and contrast relations follow each other within the combination-contrast conceptual space along the order ‘sequential combination - simultaneous combination - atemporal combination - oppositive contrast - corrective contrast - counterexpectative contrast’ (cf. Table 29). If in a given language a marker can be used to express more than one combination or contrast relation, it will be used for relations which stand close to each other on the conceptual space. The interpretation of a conceptual space is based on the principle that recurrent similarity in form reﬂects similarity in meaning (Haiman 1985: 26). The recurrent use of the same marker for two different relations is thus taken as an indicator of the conceptual or pragmatic proximity of the two relations. The organization of the speciﬁc conceptual situations on the map represents universal relations among constructions coding the examined concepts, thus allowing not only for the identiﬁcation of further restrictions on the crosslinguistic variation, but also for a deeper analysis and comprehension of the conceptual level itself. Constraints on cross-linguistic variation are correlated to the functional factors that may motivate their regularity, since, as already said, one of the basic assumptions of this approach is that the structure of linguistic expressions is at least partially motivated by their function. In general, the principle of economy, i.e. the tendency to reduce the phonetic substance and the coded information of linguistic expressions as much as possible, explains a good number of restrictions to the cross-linguistic variation and is thus considered one of the basic functional motivations of linguistic structures (Croft 1990: 156-60). Economy will be identiﬁed as the major principle at work in the coding of coordination relations, both with respect to the presence of overt coordinating markers and to the syntactic parallelism of the attested coordinating constructions. The other basic functional motivation recognized by the functional-typological approach is iconicity, i.e. the correspondence between the form and the function of linguistic expressions (Croft 2003: 102-10; 205-19). There are two types of iconicity: isomorphisms and diagrammatic iconicity. Isomor-

10 Theoretical foundations: the notion of coordination phism is the one-to-one correspondence between parts of linguistic structure and parts of conceptual structure, so that the same form always encodes the same meaning and vice versa. Diagrammatic iconicity, on the other hand, is the correspondence between relations among parts of the linguistic structure and relations among the concepts that they encode. Diagrammatic iconicity will play an important role in the analysis of the coordination conceptual spaces. The recurrent use of the same construction for different relations iconically mirrors some conceptual or pragmatic contiguity between them. Hence coordination relations which are often expressed by the same constructions are conceptually close or frequently pragmatically connected to each other.

1.1.3

Outline of the work

The work is organized as follows. Section 1.2 describes the language samples analyzed, their inﬂuence on the structure of this work and the composite methodology adopted in data collection. In section 1.3 the traditional approach to coordination is discussed and the notion of coordination relation is deﬁned in functional terms, pointing out which coordination relations will be examined throughout the work. In chapter 2 the object of this study will be delimited and the parameters taken into account will be described. First, semantic and morphosyntactic criteria will be examined, then the methodology followed in the cross-linguistic comparison will be discussed. Chapters 3, 4 and 5 provide the data concerning the cross-linguistic coding of combination, contrast and alternative relations respectively. Each chapter starts with the functional deﬁnition of the examined relation, describing the semantic subtypes identiﬁed on the basis of the semantic parameters. Then the analysis of data will highlight implicational constraints on cross-linguistic variation, which will be discussed in relation to each morphosyntactic parameter taken into account. Finally, the implicational patterns attested in the coding of each coordination relation will be connected to the semantic properties of the speciﬁc relations expressed and to the different manifestations of the overall functional principle of syntagmatic economy. In chapter 6 the coding patterns attested for the expression of combination, contrast and alternative will be compared, highlighting further implicational constraints on cross-linguistic variation. In particular, on the basis of the attested regularities in the coding of coordination relations, a re-organization of the conceptual space internal to coordination itself will be proposed, so that

The language data: a focus on Europe

11

contrast and alternative are identiﬁed as separate semantic speciﬁcation of the very basic relation of combination. Chapter 7 focuses on European data, pointing out the geographical distribution of the various coding patterns within Europe. Moreover, the comparison with non-European data will identify a set of features which characterize the coding of coordination attested in at least a subset of European languages. Chapter 8 summarizes the major results of this research and brieﬂy indicates some possible developments which may further explain and enrich the attested patterns of coordination coding. Finally, ﬁve appendixes follow: the ﬁrst appendix presents and classiﬁes the languages examined in the two samples, the second one and the third one indicate the sources of data and the geographical location for each language examined, the fourth appendix presents the questionnaire used to gather data and the last one shows all the constructions taken into account in this research.

1.2

The language data: a focus on Europe

This section deals with the languages analyzed and their inﬂuence on the structure of this work. This aspect is crucial in two respects. On the one hand, the language sample strongly inﬂuences the scope of any research, in that it delimits the kind of results that may be achieved. A language sample consisting of ﬁve languages, for instance, allows for an in-depth analysis of data but could hardly identify general (if not universal) patterns of variation. A language sample of 100 languages, by contrast, could hardly reach a deep level of analysis for every language examined, but could reveal regular patterns of cross-linguistic variation which might turn out to be universal. On the other hand, the language sample is a crucial aspect for this work in that it determines not only the aims, but also the overall structure of the analysis, as will be made clear in the following sections. The aim of this research is to uncover the mechanisms underlying the coding of the coordination relations established between two states of affairs. Speciﬁcally, what is at issue is the coding of the various sub-types of combination, contrast and alternative which have been identiﬁed in this chapter. Data on these sub-relations is not easy to ﬁnd across languages, either because grammatical descriptions often look for equivalents of the English connectives and, or and but, without taking into account different and more subtle distinctions, or, which is even worse, associate the term coordination

12 Theoretical foundations: the notion of coordination solely to the combination relation, thus restricting the analysis to conjunction. In those cases, it is necessary to search for revealing examples throughout the whole grammar by using an inductive method. A possible solution to this difﬁculty is to gather ﬁrst hand data by means of tools speciﬁcally designed for the identiﬁcation of the phenomena under exam, like questionnaires. The main limit of this methodology is that native speakers who can ﬁll out a questionnaire are not available for a large number of languages. Restricting the sample to data gathered by means of questionnaires could strongly limit the scope of the research, especially if the analysis has a typological perspective and aims at investigating the attested variety in the coding of a given functional situation. In order to collect comparable ﬁrst-hand data about the various sub-types of coordination maintaining a broad cross-linguistic perspective, a composite methodology and two different language samples have been adopted in this work. One sample consists of 37 European languages (European sample), for which native speakers have been interviewed using questionnaires and detailed grammatical descriptions have been consulted. The other sample (Comparison sample) contains 37 languages from the rest of the world. In this case, data shows different levels of detail, going from ﬁrst hand data gathered through questionnaires to data inductively found in examples throughout grammars. The difference between data sources has led to a twofold level of analysis for the two samples and, consequently, to a twofold organization of the whole research, which necessarily shows a strong focus on Europe. As Rijkhoff et al. (1993: 171) point out, in order to investigate the possible variation in the coding of a given phenomenon, the internal diversity of the sample should be maximized and languages should be selected in such a way that all the most important linguistic families are represented.2 The twofold level of analysis adopted in this work aims at building a sample which meets the requirements of both a typological survey, in which the cross-linguistic variation can be represented, and a ﬁne grained research, in which subtle conceptual distinctions are addressed and a deeper level of analysis is required. Nonetheless, it must be born in mind that no speciﬁc algorithm (like the one suggested by Rijkhoff et al. 1993) was followed in the choice of the languages and the availability of native speakers and good descriptive grammars has played a major role. Therefore the two samples are essentially convenience samples (cf. Cysouw 2005: 555) and they consequently allow for cross-linguistic predictions that are not as far-reaching as if they were based on a balanced variety sample.

The language data: a focus on Europe

13

The availability of high quality data for the European languages supports an analysis of the attested phenomena from an areal point of view. As Ramat (1998: 227) puts it, the basic assumption of Areal Typology is that typologically relevant features may be borrowed by and from genetically unrelated languages through contacts within the same geographical area. However, in order to state that a given shared typological feature is the result of an areal convergence, it is necessary to prove the reciprocal inﬂuences that have led to the linguistic uniformity within the area, be they of a historical, economical, social or cultural nature. A typical example of linguistic area is the Sprachbund in the Balkans, where Albanian, Romani, Bulgarian and Macedonian primarily, but also Serbo-Croatian, Turkish and Greek to a lesser extent, have converged to a series of similar features, such as the merge of dative and genitive case, the loss of inﬁnitives and the postposition of clitic deﬁnite articles (Van der Auwera 1998: 259-263). An important factor in establishing whether the linguistic uniformity within a given geographic area constitutes an instance of linguistic area or not is the presence/absence of the relevant bunch of features in the neighboring languages. It is indeed necessary to delimit the area with respect to the surrounding linguistic contexts and to verify whether that bunch of features is widely attested in the world’s languages or if it is characteristic of the speciﬁc languages spoken in the area. However, within a linguistic area the linguistic phenomena are not equally shown by all languages, but we can identify a nucleus, a periphery and a cline of gradual transition from the core to the peripheral members. Such a geographical continuity is the consequence of language contact phenomena and of the progressive spreading of features characterizing the speciﬁc area. Traditionally, the examined areas have been relatively small, but recently there has been increased attention on larger areas, such as Europe. The languages spoken in the European area have indeed been in close contact for centuries and have shared much of their history, through an uninterrupted melting pot of dominations and migrations. Among typologists many efforts have been made to understand whether we can talk about a European linguistic area, the so-called Standard Average European (SAE, term coined by Benjamin Lee Whorf in 1939), or not (see for instance the EUROTYPproject funded by the European Science Foundation).3 In the next section and in chapter 7, it will be shown how this study may give a contribute to the areal picture of Europe.

+74>A4C820; 5>D=30C8>=B C74 =>C8>= >5 2>>A38=0C8>=

-<9 IFCD95B G5AD@9 5B8 H<9 CAD5F=GCB G5AD@9

+74 DA>?40= B0?40= B0=E4=84=24 B5HIF9 C: H<=G G5AD@9 29H =H F9DF9G9BHG 5@@ H<9 :5A=@=9G 9L5A=B98 =B H<9 HKC A5>CF =BH9FB5H=CB5@ DFC>97HG CB IFCD95B @5B;I5;9G 89J9@CD98 GC :5F H<9 &-2) 5B8 .+(-2) DFC>97HG G 75B 69 C6G9FJ98 =B -56@9 H<9 .+(-2) ;I=89@=B9G <5J9 699B :C@ @CK98 K=H< F9GD97H HC H<9 =B7@IG=CB C: H<9 '5?<5;<9GH5B=5B 'CFH< 5I 75G=5B H<9 $5FHJ9@=5B 5B8 H<9 @H5=7 @5B;I5;9G =B H<9 IFCD95B G5AD@9 7: 5@GC $CFHA5BB 5B8 9FB=B= 5B8 +5A5H C@@CK=B; 9FB=B= 5B8 +5A5H H<9 @H5=7 :5A=@M =G CB@M F9DF9G9BH98 6M -IF?=G< 5B8 BC .F5@=7 @5B;I5;9G CH<9F H<5B H<9 .;F=7 5B8 =BBC.;F=7 CB9G <5J9 699B 7CBG=89F98 &CF9CJ9F 5G -56@95B8=;IF9 =B H<9 DD9B8=L G
The language data: a focus on Europe

15

Table 1: Representation of individual phyla in the European sample INDO - EUROPEAN Celtic Romance Germanic Slavic Baltic Isolates

27 1 7 8 7 2 2

URALIC

Ugro-Finnic Ugric

3 2 1

BASQUE

1

ALTAIC

1

Turkic NAKH - DAGHESTANIAN Nakh Daghestanian

3 1 2

KARTVELIAN

1

AFRO - ASIATIC Semitic

1

Moreover, 31 languages out of 37 have also been investigated by means of questionnaires. As will be explained in detail in section 1.2.2 below, the questionnaire consisted of 27 English or Italian sentences (depending on the version) representing all the sub-relations of coordination at issue, and native speakers were asked to render these sentences in their own language in all the possible ways they could think of. In this way a new, systematic set of comparable data has been gathered for a good number of languages, allowing to verify and, at the same time, to integrate the information in descriptive grammars. For the languages that have been examined only through descriptive grammars and dictionaries, many different sources have been usually compared, in order to maintain the level of analysis as deep as possible. The availability of many high-quality descriptions and the possibility to interview native speakers are the main reasons to build a European sample in a cross-linguistic study like this, where relatively subtle conceptual distinctions are under exam. However, there is also a more practical explanation underlying this choice. In fact, this research started as part of a project on Europe and the Mediterranean from a linguistic point of view4 and this has obviously caused a displacement of the language selection towards Europe, especially at the ﬁrst stage of the study. Besides the possibility to reach a deep level of analysis and the participation in a European project, there are at least two other good reasons for focusing on Europe. First of all, no comparative studies have been made which examine the three main coordination relations at the same time, comparing

16 Theoretical foundations: the notion of coordination their coding strategies. Within the studies made on European languages, especially those adopting an areal perspective, many phenomena common to the SAE languages have been found. The features characterizing the SAE are not yet incontrovertibly deﬁned. However, there are a few properties for which good evidence has been provided: the presence of both a deﬁnite and an indeﬁnite article, postnominal relative clauses with inﬂecting and resumptive relative pronouns, the use of the verb ‘to have’ as an auxiliary for the transitive perfect verb form, the nominative form for the expression of the experiencer argument of perception and cognition verbs, the expression of passive by means of a passive participle and a copula verb, and others (see Haspelmath 2001 for a large list of the SAE features). These phenomena are identiﬁed as characterizing this area because they are consistently attested in a good number of contiguous languages, are not shown by the neighboring languages and are not as widespread in the world as they are within Europe (Haspelmath 1998: 274). Haspelmath (2001: 1502) suggests that the pattern A and-B could be as well considered a typical European feature, in that the presence of a speciﬁc conjunctive marker preposed to the second conjunct is not widespread in the languages neighboring the SAE area and, in general, is not widespread in the world. The peculiarity of this pattern is not the presence of a speciﬁc conjunctive marker, which occurs in two thirds of the 260 languages analyzed by Stassen (2001), but its position, which is only one of the possibilities attested in the world’s languages. However, no further studies on coordinating constructions have been made on the languages of Europe. In this research, the identiﬁcation of characteristic European features will be achieved by comparing the European sample with the Comparison sample (see below), whose label indicates its main function as comparison tool between what happens in Europe and what happens outside the European area, even though it is not statistically representative. As will be shown in chapter 7, it is possible to identify some features that seem to be typical of the SAE area, the most important of which is the presence of a general free marker for each of the three coordination relations, determining a system of the type ‘and-but-or’. It may sound normal to some European speakers, especially coming from western Europe, but a cross-linguistic glance shows that this system is not very widespread outside Europe and even within the SAE there is a lot of variation. Quite a few languages, mainly belonging to the core and the periphery, do not have it.

+74 ;0=6D064 30C0 0 5>2DB >= DA>?4

&CF9CJ9F H<9 F=7< 5ACIBH C: 89H5=@98 85H5 5J5=@56@9 :CF IFCD95B @5B ;I5;9G 5@@CKG HC 89H9FA=B9 BCH CB@M F9;I@5F=H=9G 6IH 5@GC H<9 J5F=5H=CB 5H H9GH98 K=H<=B H<9 5F95 K<=7< K=@@ HIFB CIH HC 69 BCH GC =BH9FB5@@M ILH5DC G=H=CB G $CFHA5BB <=;<@=;IGH C: H<9 KCF@8bG @5B;I5;9G IFCD95B @5B;I5;9G 7CBGH=HIH9 5 <=;< DFCDCFH=CB C: H<9 @5B;I5;9G K<=7< <5J9 5 89J9@CD98 KF=HH9B F9;=GH9F 5B8 5 @CB; @=H9F5FM HF58=H=CB ,=B79 H<9 HKC D5F5A9H9FG C: G9A5BH=7 8CA5=B 5B8 89;F99 C: 7C8=B; 65G=75@@M D9F H5=B HC GMB89H=7 7CBGHFI7H=CBG 5B8 :C7IG CB H<9 F9@5H=CBG 7C8=:=98 6M A95BG C: CJ9FH 7CCF8=B5H=B; A5F?9FG @5B;I5;9G K=H< 5 7CAD@9H9 G9H C: 7CCF8=B5H=B; A5F?9FG @=?9 H<9 IFCD95B CB9G 7CBGH=HIH9 5 :5JCF98 5B8 =895@ G5AD@9 +74 >= B0
18 Theoretical foundations: the notion of coordination of the generalizations made. Secondly, the constructions attested in Europe could not be compared with what happens in the rest of the world and therefore it would not be possible to identify speciﬁc European features. In order to state that something is peculiar, it is necessary to state, at the same time, with respect to what it is peculiar. Therefore another sample has been built, consisting of 37 non-European languages from all over the world (see Table 49). This sample has been labeled Comparison sample, in order to highlight that its main function is to provide a background of data against which European coordinating constructions can be analyzed. Even though it gives the sample its name, it is not the only function it has. The inclusion of data from extra-European languages is indeed necessary in order to truly investigate the mechanisms whereby coordination relations are coded across languages. The Comparison sample is not balanced, but it is rather a convenience sample, since the choice of the languages has been strongly inﬂuenced by the availability of good grammatical descriptions. Yet, it represents twenty different linguistic families spoken all over the world, trying not to exclude large parts of the globe and not to consider only adjacent languages (see Fig. 17 in Appendix). Table 2 shows the individual phyla represented in the sample. Table 2: Representation of individual phyla in the Comparison sample INDO - EUROPEAN Indo-Iranian

2 2

AFRO - ASIATIC Semitic Chadic Cushitic

7 2 2 3

Malayo-Polynesian

4 4

NADENE

1

AUSTRONESIAN

Nuclear AUSTRALIAN

1

Gunwingguan NIGER - CONGO Volta-Congo Atlantic

4 3 1

Mississippi

MOSETENAN

1

TRANS NEWGUINEA

1

ISOLATES

2

CHAPACURA

1

SIOUAN

Wanham DRAVIDIC

1

1

The language data: a focus on Europe

19

Table 2: Continued Tamil-Kannada

HOKAN

1

Yuman SINO - TIBETAN Tibeto-Burman

4 4

ESKIMO

1

Aleut AUSTRO - ASIATIC Viet-Muong

1

YUKAGHIR

1

NILO - SAHARAN Nilotic

1

CREOLE

1

English based

Only 7 of these languages have been looked into by means of questionnaires. The quality of the descriptive grammars varied from comprehensive and detailed ones to grammars with many revealing examples but a vague semantic description. The reliability of data has anyway been judged satisfactory, since this was the pre-requisite for including a language in the sample. Yet, there are cases where some data is missing, especially data concerning the coding of the sub-types of contrast. The depth of analysis reached in this sample is consequently not homogeneous, since it depends on the kind of information found for each language. This sample can nonetheless point to the constructions attested outside Europe, widening the range of coordinating strategies taken into account and thus allowing for a greater scope of the generalizations identiﬁed in this work.

1.2.2

The questionnaire

The questionnaire used to gather new data is an elicitation questionnaire, whereby native speakers have been asked to render in their own language a set of sentences representing the various sub-types of combination, alternative and contrast at issue (see the questionnaire in Table 52, in the Appendix). The semantic context has been made clear in almost all sentences, so that the various sub-relations could be clearly identiﬁable. Where the context has been left unspeciﬁed, this was made on purpose, in order to examine the constructions used in case of semantic underspeciﬁcation. For each sentence, the speaker has been asked to provide all the possible ways in which the given conceptual situation could be expressed in his/her native language.

20 Theoretical foundations: the notion of coordination The questionnaire has been ﬁlled out by speakers with a high education level (where possible, linguists have been interviewed), whose proﬁciency in the language of the questionnaire was very good. Two versions of the questionnaire have been prepared, one in English and one in Italian, since the research has been mostly conducted in Italy. In several occasions, the questionnaire has been the starting point for further discussions, which have led to the creation and analysis of sentences other than the one attested in the questionnaire, following the speakers’ indications. Moreover, as it often happens in this type of research (cf. Bernini and Ramat 1996), the questionnaire has been modiﬁed as the various feedback has been received, in order to make it more complete and coherent with the aims of this study. At the beginning, the questionnaire also contained sentences in which the coordination relations were established between subordinate states of affairs and between entities (i.e. noun phrases). The decision to restrict the study to the coordination of independent states of affairs has been made only later on in the work and the questionnaire has been consequently changed. The questionnaire has mostly been used in the exam of the European sample, where 31 languages out of 37 have been investigated with this methodology. In the Comparison sample, instead, it has been difﬁcult to ﬁnd native speakers available for interviews and only 7 languages out of 37 have been examined with a questionnaire. In Table 3 the languages analyzed by means of questionnaires are listed and the number of questionnaires ﬁlled out for each language is indicated. Table 3: Languages analyzed with questionnaires Sample

Languages examined with questionnaires

E UROPEAN SAMPLE: 31 languages, 54 questionnaires

Albanian (1), Basque (1), Bulgarian (1), Catalan (1), Czech (1), Danish (2), Dutch (1), English (1), Estonian (1), Finnish (2), French (3), Georgian (1), German (2), Greek (2), Hungarian (2), Icelandic (1), Irish (1), Italian (5), Latvian (1), Lithuanian (3), Luxembourgish (1), Maltese (1), Norwegian (1), Polish (2), Portuguese (2), Romanian (3), Russian (4), Sardinian (2), Serbo-Croatian (1), Spanish (3), Swedish (1)

N ON -E UROPEAN SAM PLE: 7 languages, 7 questionnaires

Arabic (1), Hausa (1), Hebrew (1), Japanese (1), Korean (1), NànáfwˆE (1), Persian (1)

+74 ;0=6D064 30C0 0 5>2DB >= DA>?4

G 9FB=B= 5B8 +5A5H 5F;I9 H<9 IG9 C: 5 EI9GH=CBB5=F9 9B H5=@G HKC :57HCFG H<5H A5M 75IG9 GCA9 A9H
22 Theoretical foundations: the notion of coordination two methodologies are not competing but have been jointly used for the same aim, namely to build a set of data as reliable and comprehensive as possible.

1.2.3

Two levels of analysis and the twofold organization of the study

As should have become clear from what has been said in the last two sections, the two samples have been analyzed with different levels of depth, crucially depending on the availability of detailed sources of information. This has determined a twofold structure of the whole research. On the ﬁrst level of analysis, the two samples are jointly examined for a total of 74 European and extra-European languages. The depth of investigation on this level is internally non-homogeneous, but still allows to highlight interesting patterns of coding. At this level, the research has focused on the whole range of possible constructions attested to code the coordination relations across languages, in order to identify regular patterns of variation on the basis of the parameters pointed out in chapter 2. The second level of analysis goes more in depth with the exam of the European sample. On the one hand, European data is compared with extraEuropean data, in order to highlight the patterns of coding that characterize SAE languages. On the other hand, the cross-linguistic variation attested inside the European region is itself analyzed, and a non-random areal distribution of particular phenomena is outlined. In chapters 3, 4, 5 and 6 the two samples will be jointly examined on the ﬁrst level of analysis and all data will be presented by means of separate tables for European and non-European languages, in order to highlight in the course of the entire work how European patterns of coding differ, if they do, from the non-European ones. In chapter 7, then, the scope of the research will concentrate on Europe and on the characterization of this area in terms of both the coding regularities and the cross-linguistic variation attested. These two scopes have led to two different kinds of results. First of all, the joint analysis of the two samples has revealed implicational patterns of variation, in terms of both hierarchies and semantic maps. In chapters 3, 4 and 5 the coding of each coordination relation is individually examined and constraints on cross-linguistic variation are pointed out within the coding of each of the three relations. In chapter 6, on the other hand, the constructions expressing the three relations are contrasted and compared, revealing new implicational patterns of coding within the overall domain of coordination.

The notion of coordination

23

Even though such patterns are generalizations made on the basis of an internally diverse sample, they will not be claimed to be universal, because the two samples, as already remarked, are not balanced. However, although the identiﬁed patterns cannot be said to be universals, they reveal interesting results which both conﬁrm previous analysis of the coding mechanisms and suggest some new hypothesis. As widely argued, the availability of comprehensive data for the European languages determines a strong focus on Europe in this work, leading to two main results presented in chapter 7. First, the geographical distribution of the various constructions within Europe highlights well deﬁned geographical clusterings of particular phenomena. Second, the comparison of the two samples allows to identify those patterns of coding that strongly characterize European languages, either because they are very frequently attested in Europe and only very rarely in the Comparison sample, or because they are attested in the latter, but never occur in Europe. These two types of results will lead to the identiﬁcation of an area within Europe characterized by ‘And-But-Or’ languages, i.e. languages that code the three coordination relations by means of the same set of constructions. This speciﬁc clustering of coding strategies will be argued to be not widely attested in the two samples and therefore possibly conceivable as typically European.

1.3 1.3.1

The notion of coordination Traditional deﬁnitions of coordination

The term coordination traditionally deﬁnes a grammatical relation characterized by the reciprocal syntactic independence of the related phrases and by the identity of their syntactic category with the syntactic category of the complex phrase they constitute (Lehmann 1988: 182, Culicover and Jackendoff 1997: 196-197). Such a deﬁnition identiﬁes constructions at all syntactic levels, thus subsuming both phrasal and clausal coordination, and establishes the syntactic symmetry as the characterizing property of coordination. The deﬁnition of coordination has traditionally been formulated in opposition to the deﬁnition of dependency,6 originating a binary distinction mainly in terms of syntactic dependence vs. independence. In purely formal terms, a relation of dependency identiﬁes a head, i.e. the phrase which determines the syntactic

24 Theoretical foundations: the notion of coordination category of the whole construction, and a dependent, whose grammatical category does not have any inﬂuence on the category of the construction. Such a sharp formal division between coordination and dependency has however turned out to cause problems both within and across languages, because it lumps completely different phenomena under the same label only on the basis of their sharing syntactic dependence or syntactic independence. Narrowing our scope to the clause level, which is the object of this research, the purely structural deﬁnition of coordination has been traditionally characterized as missing the morphosyntactic phenomena that are traditionally associated to subordination.7 Formal deﬁnitions of subordination specify it as a grammatical category identiﬁable on the basis of morphosyntactic criteria such as syntactic dependence, syntactic embedding and nonﬁnite verb forms (Haspelmath 1995: 8, Van Valin 2006). According to the view which opposes coordination to subordination in a binary distinction, it follows that coordination is a grammatical category showing syntactic independence, no embedding and ﬁnite verb forms. Moreover, the morphosyntactic properties deﬁning subordination are traditionally associated with non-asserted information, whereas the morphosyntactic properties deﬁning coordination are traditionally associated to asserted information (cf. Lehmann 1988: 193-200). Yet, these deﬁnitions entail problems, mainly regarding the internal consistency of the status of coordination and subordination as universal, thus cross-linguistically valid, grammatical categories. Following Dryer’s (Dryer 1996) broad deﬁnition of grammatical relation as any ‘grouping of relationships . . . that plays a role in the grammar of a particular language’,8 Cristofaro (2003: chapter 1) shows that the morphosyntactic criteria associated with subordination do not identify internally consistent classes of phenomena, but rather delimit language speciﬁc grammatical categories whose elements share the given deﬁning feature and are nonhomogeneous under many other respects. Each deﬁning parameter taken individually does not indeed combine with the same other parameters. Embedded clauses, for instance, may show ﬁnite and nonﬁnite verb forms, as is evident in example (1.7), where the relevant verb forms are underlined. The embedded relative clause in (1.7a) has a ﬁnite verb form, whereas the adverbial embedded clause in (example 1.7b) has an inﬁnitive verb form.

The notion of coordination

25

(1.7) Italian, Romance, Indo-European hai (a) la ragazza [che mi the girl that 1sg.DAT AUX.PRS:2sg presentato] è simpatica. introduce:PST.PTCP be.PRS.3sg nice ‘The girl that you have introduced to me is nice.’ (b) Lui ha [girando la testa], che visto, He AUX.PRS:3sg see.PST.PTCP turning the head that qualcuno scappava. somebody escape:IMPFV:3sg ‘He has seen, turning his head, that somebody was escaping.’ Furthermore, nonﬁnite verb forms may convey asserted or non-asserted information (1.8) and may be classiﬁed as nonﬁnite on the basis of different criteria (the loss of all the verbal categories, the loss of just some of them, the loss of person agreement, and so on). The sentence in (1.8), where the ﬁrst verb is sufﬁxed by -Ip and is nonﬁnite, may indeed be translated either with ‘Hasan went to work and came back home’ or with ‘Hasan, having gone to work, came back home’, depending on the context. This means that the same construction may assert both states of affairs, leading to a coordinate translation as the one in the example, or may simply assert one of them, leading to an embedding reading. (1.8) Turkish, Altaic gid-ip ev-e dön-dü Hasan i¸s-e Hasan work-DAT go-and home-DAT come-PST ‘Hasan went to work and came back home/ Hasan, having gone to work, came back home’ (Kornﬁlt 1997: 110) Not only the traditional parameters for subordination do not identify a crosslinguistically consistent set of constructions, but some of them are attested in constructions which are commonly used as counterparts to coordinating constructions, as traditionally deﬁned. Speciﬁcally, some languages show constructions characterized by the lack of embedding (which should be typical of coordination) and the presence of nonﬁnite, syntactically dependent verb forms (which should instead be typical of subordination). We have already

26 Theoretical foundations: the notion of coordination seen an example from Turkish in (1.8), whose translation into English may require the use of coordination. Let us see another example from Korean. (1.9) Korean Minca-nun mikwuk-ey na-nun ilpon-ey ka-(ss)-ko 1sg-TOP Japan-to go-(PST)-and Minca-TOP America-to ka-ss-ta. go-PST-DECL ‘I went to Japan and Minca to America.’ (Sohn 1994: 118) The construction exempliﬁed in (1.9) is constituted by a nonﬁnite verb form, which can receive tense but not modal speciﬁcations, followed by a fully inﬂected form, on which the ﬁrst nonﬁnite verb depends. Both verb forms in (1.9) convey asserted information, since both states of affairs are open to challenge and denial. The opposite may also happen and a coordinate-like construction may be used to convey a traditionally subordinate relation. Culicover and Jackendoff (1997) examine sentences in English that apparently display coordinate-like properties, but whose semantics is rather typical of a standard conditional clause (example 1.10). (1.10) You drink another can of beer and I’m leaving. In sentences like (1.10), the order of the two clauses cannot be inverted, extraction from either clause is allowed and the typical coordinate connective and is used, which cannot occur in a clause-initial position. These three features are identiﬁed by Culicover and Jackendoff as typical of coordinate clauses. Yet, with respect to other features, these sentences resemble embedded clauses, since they allow cataphoric reference and across-the-boundary extraction. Culicover and Jackendoff’s conclusion is that in such cases we actually deal with coordinate constructions and the deviations from the typical coordinate-like behaviors are due to the more subordinate-like semantics. There is, however, no independent evidence for such a bipartite analysis, as they acknowledge (Culicover and Jackendoff 1997: 215). The cases illustrated so far signiﬁcantly defy the bipartite morphosyntactic distinction between coordination and subordination, highlighting areas of overlapping where the single formal parameters combine across the rigid

The notion of coordination

27

borders of the traditional deﬁnitions of coordination and subordination. Gil’s analysis (Gil 1991) of syntactic categories in Riau Indonesian illustrates the most extreme consequences of a rigid morphosyntactic distinction between coordination and subordination. Starting from a purely syntactic deﬁnition of coordination, Gil labels as coordinate a construction like (1.11). The two words makan ‘eat’ and ayam ‘chicken’ are identiﬁed as belonging to the same grammatical category S, since Riau Indonesian is a strongly isolating language in which there is no distinction between the syntactic categories of noun, adjective, verb and sentence. Every word, as Gil (2004: 376) argues, belongs to the general syntactic category S roughly corresponding to the notion of Sentence. Therefore, considering only the two formal parameters of reciprocal syntactic independence and identity of syntactic category, the simple sentence in (1.11) meaning ‘the chicken is eating’ would result to be an instance of coordination, just like the English complex syntagms ‘me and you’ or ‘I cook and you eat’, whose semantics is completely different. (1.11) Riau Indonesian (Gil 1991) [[makan]S [ayam]S ]S eat chicken ‘The chicken eats/the chicken has eaten etc.’ (‘an association of eating and chicken’ Gil 2004: 374) Deﬁning the object in this analysis in the traditional morphosyntactic terms would therefore lead to considering very different phenomena and to leaving out of the analysis cases like (1.8) and in (1.9). The exclusion of these constructions on the basis of formal parameters would prevent the analysis from grasping their conceptual equivalence to English coordinate andconstructions and, as we have already seen, it is speciﬁcally this conceptual equivalence, together with the morphosyntactic variation expressing it, that constitutes the main object of interest in typological research. From what has been said so far, it is evident that a morphosyntactic deﬁnition of coordination would restrict the analysis to phenomena that are identical with respect to the deﬁnitional set of features, but may differ with respect to other morphosyntactic behaviors and do not necessarily express the same conceptual relation (see example (1.11)). Such a morphosyntactic delimitation of the object of analysis does not lead to identify a internally coherent

28 Theoretical foundations: the notion of coordination set of constructions. It could be useful in studies that adopt a form-to-function perspective, to examine the possible functions expressed by constructions that show a given array of morphosyntactic properties. However, this is not the case here, since this work adopts a function-to-form perspective, focusing on the different constructions used to express the three main coordination relations. In fact, in this survey, cases like (1.8) and (1.9) would need to be included in the analysis. Consequently, the traditional deﬁnition of coordination cannot constitute a valid tertium comparationis for this research.

1.3.2 Alternative approaches to the deﬁnition of coordination Two alternative approaches have been suggested in order to take into account the attested cross-linguistic variation and the constructions bridging across the two categories. The ﬁrst approach relativizes the concepts of coordination and subordination as positions on a scale while the second one identiﬁes a third syntactic category accounting for the untypical cases. According to the ﬁrst solution, coordination and subordination are not deﬁned on the basis of a set of features, but are rather identiﬁed as poles of a continuum. This continuum is in turn constituted by various parallel continua determined by the values of speciﬁc morphosyntactic features, that need not necessarily be all present nor combine with the same other features in all languages (Lehmann 1988). Lehmann (1988: 217) identiﬁes six features on the basis of which clause linkage phenomena may be analyzed: the degree of integration of the subordinate clause into the main one, the main clause syntactic level of the subordinate clause, its desententialization, the grammaticalization of the main verb, the interlacing of the two clauses, and the explicitness of the linking. Lehmann identiﬁes two opposing forces underlying the continuum on which the various values of these features distribute: the one which pushes towards the elaboration of information and the other towards its compression. Thus, at the left hand pole of the continuum there is the highest degree of elaboration. Here we ﬁnd instances of the traditionally called coordinate constructions (parataxis in Lehmann’s terminology),9 characterized by a high syntactic level, reciprocal independence of the clauses, lack of embedding, lack of overlapping and explicit linking. Conversely, at the right hand pole of the continuum we ﬁnd the highest degree of compression, exempliﬁed by embedded nominalized clauses without explicit linking. In between, the various

The notion of coordination

29

constructions attested in the world’s languages, like (1.8) and (1.9) in Turkish and Korean for instance, ﬁnd a position in the continuum on the basis of the values they show for each single parameter (Lehmann 1988: 184-185). Such a continuum approach constitutes a valid tool for cross-linguistic research on complex sentences, insofar as it catches the impossibility of deﬁning internal consistent categories for coordination and subordination and highlights the need for a set of independent formal parameters separately characterizing each attested construction. However, if one uses a given point on the continuum (i.e. a given sentence type) as a deﬁnition for cross-linguistic research, the same restrictions highlighted for the traditional deﬁnition would arise. Not all languages are expected to show the sentence type identiﬁed as tertium comparationis, and languages not presenting it would need to be excluded from the sample. On the other hand, if one does not choose a sentence type on the continuum as a departure point for their analysis, this approach does not offer any other unambiguous parameter to establish what constructions one should look for and include in their research. This means that the various morphosyntactic features building the continuum work well as parameters, and as such they will be used in this work (see chapter 2), but not as deﬁning criteria for a cross-linguistic valid concept of coordination. The second alternative approach establishes instead a new category which speciﬁcally identiﬁes cases like (1.8) and (1.9) (Van Valin 2006). Van Valin (2006) makes a tripartite distinction between coordination, subordination and co-subordination, identifying each of the three categories on the basis of the two parameters of embedding and dependency. Coordination lacks both of them, co-subordination shows syntactic dependency but no embedding and subordination shows both of them. The term co-subordination suggests the bridging status of this construction type between the other two, capturing an important, as well as ﬂeeting, property of this type of strategy. Longacre (1985: 263-284) refers to cases of co-subordination as chaining structures, following a terminology widely attested to describe Papuan languages. The term ‘chain’ derives from the long sequences attested in languages that make use of this strategy, where many nonﬁnite verbs follow one another and all depend on the ﬁnal ﬁnite verb. Nonﬁnite verb forms are often sufﬁxed by switch-reference markers, which indicate whether the subject of the following clause is different or not from the preceding one. In (1.12) an example of chaining structure is given.

30 Theoretical foundations: the notion of coordination (1.12) Wojokeso, Papua New Guinea sukwo’miyomo hofantiso toho s1kuno nome-hon1ngk-i mosquitoes bite darkness came-SEQ-3sg(DS) night.in nelof-ahon1ngk-i nakwo mempo saho kokoko u us.bit-SEQ-3sg(DS) INTENS EXCL we outside sleep mafosyawosofo not.sleep ‘Darkness came and at night mosquitoes bit us an awful lot so (being) outside we couldn’t sleep’ (Longacre 1985: 270) Longacre (1985: 237-238) further states that the distinction between coordination and subordination (as deﬁned in morphosyntactic terms) does not pertain to the languages where the chaining strategy is dominant, because it is often not easy to decide whether a given chain should be translated by means of an English coordinate or subordinate construction. He thus makes an overall distinction between ‘co-ranking’ languages, for which the bipartition between coordination and subordination is relevant, and ‘chaining’ languages, in which it makes no sense to talk about coordination or subordination. That is to say, the distinction between coordination and subordination is neutralized in such languages, thus conﬁrming the limited applicability of these notions as traditionally deﬁned. All the approaches identifying a third category between coordination and subordination suffer the same problems highlighted for the traditional view, since they simply point out a new combination of morphosyntactic parameters. This analysis allows for accurate formal descriptions, avoiding terminological ambiguities regarding the morphosyntactic properties of the attested dependency constructions. However, once again, the dependency may be caused by different phenomena across languages (lack of person marking, lack of tense, mood and aspect markers, lack of just one of them) and the internal consistency of the new category is thus no more straightforward than that of the other two. Moreover, whatever category one may choose as a starting point for a cross-linguistic research, the languages missing it will necessarily be excluded from the sample, thus reducing the scope of the analysis. The deﬁnition of the object of analysis on the basis of morphosyntactic parameters, whatever they may be, would reduce the sample to a set of constructions showing the same values with respect to those parameters, without

+74 =>C8>= >5 2>>A38=0C8>=

B979GG5F=@M <5J=B; H<9 G5A9 :IB7H=CB BCF H<9 G5A9 J5@I9G K=H< F9GD97H HC CH<9F ACFD97H C: 5B5@MG=G H<9 89:=B=B; D5F5A9H9FG AIGH 69 G9H CB H<9 :IB7H=CB5@ @9J9@ "B H<=G K5M H<9 C4AC8D< 2>=8B C: H<9 F9G95F7< =G 5 7CB79DHI5@ G=HI5H=CB 5B8 K<5H =G 5H =GGI9 =G =HG 9LDF9GG=CB 57FCGG @5B;I5;9G 7CB79DHI5@ G=HI5H=CB =G IB=J9FG5@ 5B8 75B 69 GCA9 2>=9D=2CB 70E4 C> 14 /?0A0;;4; 8= 24AC08= A4B?42CB =>C H4C 5D;;H D=34ABC>>3 b

"B H<9 DF9J=CIG G97H=CB =H <5G 699B A589 7@95F H<5H H<9 ACFD=8B :CF 5 7FCGG@=B;I=GH=7 F9 G95F7< @=?9 H<9 CB9 5H =GGI9 CCF8=B5H=CB K=@@ HC=B98 GH5H9G C: 5::5=FG 5F9 7CB79DHI5@=N98 D9F79=J98 5B8 7CAAIB=75H98 =B H<9 8=G7CIFG9 7CBH9LH -<=G :IB7H=CB5@ D5F5@@9@=GA =G GHF=7H@M 7CBB97H98 HC H<9 GMBH57H=7 CB9 5B8 H<9 7CB GHFI7H=CBG 9LDF9GG=B; =H @5F;9@M 7C=B7=89 K=H< H<9 7CBGHFI7H=CBG K<=7< 5F9 HF5 8=H=CB5@@M =89BH=:=98 5G 7CCF8=B5H9 CB H<9 65G=G C: ACFD
32 Theoretical foundations: the notion of coordination tion are ‘equivalent as to grammatical function’, but do not necessarily have the same grammatical structure nor do they necessarily belong to the same syntactic category. The functional parallelism of two joined states of affairs is the conceptual situation that constitutes the invariant term of this cross-linguistic comparison, so that all the constructions expressing it will be included in the analysis, independently of their morphosyntactic (non-)parallelism. The morphosyntactic parallelism as such, instead, will be one of the parameters used in the analysis of the examined constructions (see chapter 2). By state of affairs will be meant here the concept of something that can be the case in some world, and can be evaluated in terms of its existence (Siewierska 1991, Dik 1997). The term ‘state of affairs’ will be understood as a hyperonym for the words ‘situation’, ‘event’, ‘process’ and ‘action’ (see Van Valin 2006: 82-89 for detailed deﬁnitions). The term ‘state of affairs’ is preferred because it does not characterize the entity in any particular sense, whereas ‘situation’ or ‘process’ may convey a static vs. dynamic connotation. Henceforth, states of affair will be abbreviated with the acronym SoA. Let us ﬁrst of all analyze the notion of functional parallelism, by exploring the plurality of parallelisms that may be described on the semantic, the cognitive and the pragmatic level. Such a tripartite distinction is not always easy to draw and may somehow sound as forcing the complex net of interrelations characterizing what is here called functional level. However, different scholars have highlighted different respects under which two SoAs can be considered ‘functionally’ parallel, as many sides of the same overall phenomenon. The intent of treating the functional parallelism of coordination in terms of a plurality of parallelisms is thus to systematize into a unitary account the different facets that the main functional approaches to coordination have pinpointed, mirroring the complexity underlying the general ‘functional’ label. 1.3.3.1 Semantic parallelism One of the ﬁrst approaches to coordination that has highlighted the semantic parallelism underlying a coordination relation is the one proposed by Schachter (1977), who states that ‘(. . . ) whether or not identity of syntactic category is a necessary condition on conjoinability, it is not a sufﬁcient one’ (Schachter 1977: 89). He deals with English sentences like the ones in (1.13) and (1.14) (Schachter 1977: 89).

The notion of coordination

33

(1.13) *John ate with his mother and with good appetite. (1.14) *John probably and unwillingly went to bed. In both cases, the constituents have the same syntactic category but have different semantic functions: in (1.13) an accompaniment phrase is joined with a manner phrase, in (1.14) a speaker-oriented adverb is joined with a subjectoriented adverb (Schachter 1977: 89). Assuming a clear distinction between syntactic categories and semantic functions, Schachter (1977: 90) establishes the Coordinate Constituent Constraint (CCC), according to which the constituents of a coordinate construction must belong to the same syntactic category and have the same semantic function in the sentence. A similar approach, even though not in terms of constraint, is adopted by Haspelmath (2004 and 2007). He deﬁnes as coordinate construction any syntactic construction ‘in which two or more units of the same type are combined into a larger unit and still have the same semantic relations with other surrounding elements’ (Haspelmath 2004: 34). Haspelmath does not refer to any syntactic category parallelism, but means by ‘type’ a semantic deﬁnition of head and dependent10 and takes the identity of semantic roles played by the linked elements as the identifying criterion for coordination (Haspelmath 2007: 42). He proposes this deﬁnition in order to account for those morphosyntactic asymmetric constructions whose meaning is equivalent to that of English symmetric ones. According to him, a semantic deﬁnition of coordination may also include syntactically dependent constructions like the one in (1.15). (1.15) Japanese yama-de hatarai-te, obaasan-ga Ojii.san.ga old.man-NOM mountain-at work-COORD old.woman-NOM mise-no ban-o shi-ta store-GEN sitting-ACC do-PST ‘The old man worked at the mountain, and the old woman tended the store.’ (Yuasa and Sadock 2002: 92, quoted in Haspelmath 2004: 34) In (1.15) the verb form of the ﬁrst clause is sufﬁxed by the converb -te and cannot receive further speciﬁcations of time, depending for their expression on the verb in the last clause. We have already met this kind of syntactic dependence in what we called the ‘bridging’ cases between coordination and

34 Theoretical foundations: the notion of coordination subordination. Haspelmath, however, suggests that a construction like the one in (1.15) should be considered an instance of coordination, by virtue of its semantic parallelism. The identiﬁcation of coordination on the basis of the identity of semantic functions is especially adequate when the coordinated entities are not SoAs. In a sentence like (1.16), ‘hard’ and ‘with conviction’ are combined, both standing in a relation of manner with the main action. In this case, the two elements have the same external semantic relations, that is, they stand in the same semantic relations with the rest of the sentence. (1.16) I am working [hard and with conviction.]manner In a sentence like (1.15), on the other hand, the combined elements are SoAs and their combination exhausts the whole sentence, therefore there are no external elements with which the two SoAs stand in a semantic relation. In particular, it is possible to identify the external semantic role of two coordinated SoAs in case they are subordinate to a main clause, as in (1.17), where both SoAs are necessary conditions for the realization of the SoA described by the main clause. (1.17) I would leave, [if I had money and if I had luck.]condition However, in a sentence where no subordination is established, but two SoAs are simply coordinated to each other (like in (1.15)), the identity of function of the linked SoAs is better described in terms other than the semantic parallelism. Even if he does not mention it in his deﬁnition of coordination, Haspelmath also refers to a symmetry of prominence between the involved SoAs, arguing that they have the same status and neither of them is more salient or important than the other (Haspelmath 2004: 3). This characterization of coordination is more based on conceptual and pragmatic terms, rather than on semantic ones, and may better describe the functional parallelism of two independent coordinated SoAs. Yet, Haspelmath gives no further elaboration of the functional parallelism of coordination along these two dimensions. On the contrary, other scholars have identiﬁed conceptual and pragmatic parallelism as deﬁning properties of the coordination relation itself. Let us now examine their approaches.

The notion of coordination

35

1.3.3.2 Conceptual parallelism A number of scholars have characterized two coordinated SoAs mainly on the basis of their conceptual equivalence, rather than focusing on the speciﬁc semantic functions they fulﬁl within the complex sentence. Haiman (1985: 99), for instance, states that ‘the relation of coordination is itself one of conceptual symmetry’. He claims that some phenomena of asymmetry which are proper of coordination (like the temporal interpretation in (1.19) below) are a consequence of the conceptual symmetry itself (Haiman 1985: 100). As Haiman argues, tense iconicity distinguishes coordinate constructions from subordinate ones. Subordinate clauses are indeed traditionally identiﬁed, among other criteria, also on the basis of their ability of being fronted without affecting the general meaning of the sentence, as can be seen in example (1.18).11 What enables a subordinate clause to be fronted is the fact that it is somehow backgrounded to the main one and the respective temporal collocation is not affected by the order in which the two clauses are presented (Haiman 1985: 100). (1.18)

(a) Although I don’t like it that much, I am going to dance tonight. (b) I am going to dance tonight, although I don’t like it that much.

On the contrary, if the order in which two coordinate clauses are presented is reversed, the meaning of the whole construction varies, as can be seen in example (1.19), giving rise to what Haiman calls linear asymmetry. According to Haiman, the clauses can be juxtaposed along the time line as long as they are of ‘equal rank’ (Haiman 1985: 100), that is, if they are equal in prominence. (1.19)

(a) Spike left and got up early the next day (Haiman 1985: 100). (b) Spike got up early the next day and left (Haiman 1985: 100).

The absence of this conceptual parallelism causes the absence of tense iconicity, because the joined clauses are not necessarily located within the same time line, nor are they necessarily sequential. Rather, the two denoted SoAs may be simultaneous or they may even occur in an order which is opposite to the one in which they are presented (example 1.18). Haiman (1985: 101) argues that the linear nature of language has to deal with two dimensions along which a parallelism may be established: the temporal dimension on the one hand, along which two SoAs may either be si-

36 Theoretical foundations: the notion of coordination multaneous (parallel) or sequential (non-parallel), and the prominence12 dimension on the other hand, along which the two SoAs may be presented as equivalent or asymmetric. Haiman characterizes coordination on the basis of the symmetry of prominence, which may in turn coexist either with temporal parallelism, if the two SoAs are simultaneous, or with temporal asymmetry, in case one SoA follows the other. Subordination entails instead an asymmetry of prominence between the two joined clauses, which may as well combine with both temporal succession and simultaneity. The concept of ‘equal rank’ addressed by Haiman can be intuitively clear, but still remains rather abstract and lacks the support of independent evidence in Haiman’s discussion itself. Let us now examine another analysis of coordination in terms of conceptual parallelism, conducted by Langacker within the frame of Cognitive Grammar (Langacker 1987a). According to Langacker (1987a: 483), ‘the essence of coordination is the conceptual juxtaposition of co-equal structures’ and all the main properties of coordination follow from this conceptual parallelism. In order to understand what he means by co-equal structures, some basic notions of Langacker’s theory are necessary. Within the Cognitive Grammar, any linguistic structure is analyzed as having a semantic and a phonological pole. Each semantic pole consists in turn of a cognitive base and a cognitive proﬁle. The base consists of the scope of the predication, that is, all the aspects of a scene which are recalled and implied in the predication. The proﬁle denotes the speciﬁc substructure of the base which is designated by the predication and receives special cognitive prominence. For instance, the base for island is a conﬁguration made of land and sea, while its proﬁle identiﬁes the piece of land. Coast shares the same cognitive base of island, but has a different proﬁle, in that it assigns a special prominence to the perimeter of land which borders the water (Langacker 1987b). Given two joined SoAs, they may have autonomous proﬁles or they may show an asymmetry, so that the proﬁle of one of them is overridden by that of the other. The former conﬁguration corresponds to a coordination relation between SoAs, the latter to a subordination relation, in which the SoA with an autonomous proﬁle is denoted by the main clause (Langacker 1987a: 436). For instance, (1.20) Alarms ringing, the burglar ﬂed. designates the process of ﬂeeing, not that of ringing, and the proﬁle of the subordinate clause is thus overridden by that of the main one. In a sentence

The notion of coordination

37

like the one in (1.21) instead, each of the two clauses has an autonomous proﬁle and there is thus no reciprocal inclusion between the semantic poles of the two SoAs. (1.21) The Cubs won and the Padres lost. Langacker’s analysis focuses on the cognitive organization of coordination relations, which are ‘analyzed as having not just one proﬁle, but two or more separate and co-equals proﬁles representing the same basic class’ (Langacker 1987a: 472). This means that a given complex sentence, consisting of two joined clauses, may designate either two processes, in case each SoA has an autonomous cognitive proﬁle (1.21), or only one process, in case one of the SoA lacks an autonomous proﬁle (1.20). He builds his theory starting from the English language and does not attempt any cross-linguistic comparison. Furthermore, he does not give any independent evidence for the identiﬁcation of co-equal autonomous proﬁles independently of the morphosyntactic level. Yet, his analysis of coordination and subordination relations has been adopted by Cristofaro (2003) in her typological study on subordination. As will be made clear in the next section, Cristofaro has identiﬁed a correspondence between the conceptual (non-)parallelism of two given SoAs and the assertive (non-)parallelism of the clauses expressing them, postulating what she labels Asymmetry Assumption (Cristofaro 2003: 29-50). Such a correspondence allows for independent criteria in the identiﬁcation of a conceptual parallelism between two SoAs, regardless of the speciﬁc morphosyntactic properties of the languages analyzed. 1.3.3.3 Pragmatic parallelism From what has been said so far, two SoAs joined in a coordination relation are characterized by having equal conceptual prominence, i.e. autonomous cognitive proﬁles. There is a third functional dimension on which the two SoAs have been analyzed as parallel, namely the level of pragmatics, where information is inserted into the net of communicative assumptions of speaker and hearer, and may thus be asserted, asked or ordered. As brieﬂy mentioned, Cristofaro (2003) has established an equivalence between the conceptual and the assertive symmetry of two joined clauses. Speciﬁcally, given two SoAs, she equates the conceptual distinction between

38 Theoretical foundations: the notion of coordination autonomy and non-autonomy of their cognitive proﬁles, as examined by Langacker, with the pragmatic distinction between asserted and non-asserted information (Cristofaro 2003: 30). Assertion is a controversial notion, which has been object of several debates that will not be addressed here. However, two major senses can be distinguished: a semantic sense and a pragmatic sense. Semantic assertion deals with the truth-conditions established by speciﬁc semantic relations, whereby the presupposed part of a sentence must be true in order for the asserted part of the sentence to be either true or false (cf. McCawley 1981: 326-30). Assertion is however here meant in its pragmatic sense. As Lambrecht (1994: 52) deﬁnes it, it consists of what ‘the hearer is expected to know or take for granted as a result of hearing the sentence uttered’. In uttering the sentence (1.22) I ﬁnally met the woman who moved in downstairs. (Lambrecht 1994: 51) the speaker wants to communicate that he ﬁnally met the woman, not that a woman moved in downstairs. Two things are assumed. First, that the hearer already knows, or is ready to take for granted, that a woman moved in downstairs. Secondly that the hearer has not been acquainted yet with the fact that the speaker has ﬁnally met that woman. According to the Asymmetry Assumption, the asymmetry in the communicative organization of the sentence in (1.22) reﬂects the asymmetry in the cognitive organization of the two SoAs. Indeed, only one of them is really designated and has an autonomous proﬁle, namely the one which corresponds to what is asserted, i.e. meeting the woman. In establishing this equivalence, Cristofaro (2003: 31) plausibly postulates that a SoA with an autonomous cognitive proﬁle designates what the speaker intends to communicate, i.e. a pragmatic assertion, whereas a SoA lacking an autonomous proﬁle designates non-asserted information. She then deﬁnes subordination as a situation of functional asymmetry in which the proﬁle of one of two linked SoAs is overridden by that of the other. What makes this deﬁnition applicable to all languages, regardless of their grammatical features, is the Asymmetry Assumption, because pragmatic situations are assumed to be universal and it is reasonable to hypothesize that every language distinguishes between asserted and non-asserted information. Hence, the use of tests for assertiveness, like tag questions or explicit denial, allows for the identiﬁcation of subordinate clauses across languages, denoting a SoA lacking an autonomous proﬁle (Cristofaro 2003: 32-48).

The notion of coordination

39

Examples (1.23) and (1.24) are used by Cristofaro (2003: 32) to give instances of the tests. In (1.23) the explicit negation challenges only the fact that the burglar ﬂed, since it would make no sense to challenge what the hearer is expected to take for granted. The same type of phenomenon can be observed in (1.24), where the tag question aims at asking conﬁrmation of what has been asserted by the speaker, whereas a tag question referred to the alarms ringing would sound rather odd. (1.23) It is not the case that, alarms ringing, the burglar ﬂed. (1.24) Alarms ringing, the burglar ﬂed, didn’t he? (*didn’t they?) Cristofaro’s Asymmetry Assumption has obvious reﬂections on the deﬁnition of the coordination relation. In the preceding section, coordination has been characterized on the basis of the conceptual parallelism of the SoAs involved, in that both have autonomous cognitive proﬁles. Following the Asymmetry Assumption, the autonomy of each proﬁle corresponds to the presence of assertive force in both SoAs, thus yielding to a symmetric communicative organization of the relation. We may then conclude that coordination is characterized by a pragmatic parallelism, in which the SoAs have the same communicative status. If the assertiveness tests are applied to a sentence consisting of two conceptually parallel SoAs, both will be challenged, thus revealing their assertive force. This can be observed in examples (1.25) and (1.26). The sentence negation in (1.25) and the tag questions in (1.26) have scope over both SoAs, thus equally challenging the two assertions. (1.25) It is not the case that the alarms rang and the burglars ﬂed. (1.26) The alarms rang and the burglars ﬂed, didn’t they? There are two possible conﬁgurations where two linked SoAs are pragmatically parallel. If each SoA has an autonomous proﬁle, the two clauses expressing them both constitute pragmatic assertions. On the contrary, if neither of the SoAs has an autonomous proﬁle, and still neither is overridden by the proﬁle of the other, none of the clauses expressing them constitute a pragmatic assertion. In the former case, we are dealing with what has traditionally been called a coordination of main clauses (example (1.27)). In the latter, the coordination relation is instead established between two subordinate clauses (example (1.28)).

40 Theoretical foundations: the notion of coordination (1.27) My parents are in Greece but my sister remained at home. (1.28) I am very concentrated, because I want to ﬁnish this work as soon as possible and reach him by the sea. However, this work focuses on the coordination between SoAs which are not identiﬁable as subordinate, in order to avoid the interpretation of phenomena due to their subordinate status as phenomena due to the coding of coordination (see more discussion about this point in section 2.1). Hence, only clauses denoting asserted SoAs will be considered. Yet, it is plausible to state that coordination is generally characterized by a pragmatic parallelism, being the symmetry in the communicative organization of a sentence the consequence of its conceptual symmetric structure. Two coordinate clauses require the same assertive force, be it null or be it full, because any asymmetric communicative organization could not co-exist with a symmetric conceptual one. An analysis in terms of assertiveness is perfectly applicable to declarative coordinate clauses, but it does not work with the coordination of interrogative or imperative clauses, since they do not assert, but rather ask or order something. What characterizes two coordinate SoAs, then, is not their assertiveness, but more generally the presence of some illocutionary force, be it declarative, imperative or interrogative (see examples (1.29) and (1.30)). (1.29) Tidy up your room and make your homework! (1.30) Do you like lamb or do you prefer a vegetarian dish? Coordination has been distinguished on the basis of its illocutionary organization by Foley and Van Valin (1984: 239-44) and Verstraete (2005: 613), who argue that each coordinate clause is characterized by the presence of illocutionary force, either separately or shared, while subordinate clauses lack any illocutionary force. The illocutionary force of an utterance is the speaker’s intention in producing that utterance and this intention distinguishes a speciﬁc illocutionary act. In turn, an illocutionary act is an instance of a culturallydeﬁned speech act. In other words, in order to function as a speech act, a sentence must be characterized by some illocutionary force, like asking, declaring, promising and ordering. As Givón (1979: 780-81) points out, there are no ﬁxed speech act types across languages, but rather we ﬁnd a continuum along which declarative (or assertive), interrogative and imperative speech acts emerge as ‘categorial

The notion of coordination

41

peaks’, identifying three basic sentence types (König and Siemund 2007). Declarative sentences are characterized by an assertive illocutionary force, whereby some information is imparted. Imperative sentences are characterized by an imperative illocutionary force, whereby some action is elicited. Finally, interrogative sentences are characterized by an interrogative illocutionary force, whereby some information is elicited. Generalizing Cristofaro’s assumption, it is plausible to equate the presence of an autonomous cognitive proﬁle with the presence of illocutionary force. The proﬁle of a given SoA designates what the speaker intends to communicate, either the assertion or the elicitation of something. A SoA which does not have an autonomous cognitive proﬁle is not at issue in the utterance and therefore does not have any illocutionary force.

1.3.4

Coordination relation: a cross-linguistically valid deﬁnition

It should be now clear that the label ‘functional parallelism’ covers a complex net of different dimensions, which all participate in establishing the symmetry of status of the involved SoAs. By ‘status’, we mean at the same time semantic function, cognitive proﬁle and illocutionary force. As just said, all these dimensions are interrelated and play an equivalent role in establishing a situation of coordination. A coordination relation between two SoAs can be thus deﬁned as a relation established between functionally equivalent SoAs, that is, SoAs which have the same semantic function, autonomous cognitive proﬁles, and are both coded by utterances characterized by the presence of some illocutionary force.

Yet, of the various functional parallelisms characterizing the coordination relation, only the pragmatic parallelism provides us with concrete means for identifying coordination relations across languages, since pragmatic communicative conﬁgurations are universal and are universally identiﬁable, independently of the grammatical structure of the various languages. It is possible to pinpoint the presence of illocutionary force by means of tests. The tests for assertiveness have already been introduced and mainly consist of tag questions and sentence negation. As for the identiﬁcation of interrogative and imperative illocutionary force, the fact that these two speech acts are somehow always overtly signalled across languages (by means of

42 Theoretical foundations: the notion of coordination some interrogative or imperative marker, which may be a particle, a verbal mood, word order inversion or simple intonation, Givón 1990: 786) makes the task easier. The sentence in (1.31) from Finnish, for instance, consists of two clauses each of which has the question marker -kö, signalling the presence of an interrogative illocutionary force in both SoAs. (1.31) Finnish, Finnic, Uralic jää-mme-kö mene-mme-kö koulu-un huomenna vai school-ILL tomorrow COORD stay-1pl-Q go-1pl-Q koti-in home-ILL ‘Do we go to school tomorrow or do we stay at home?’ (J.Y., questionnaire) If no explicit indicator of illocutionary force is given, we are able to identify its presence by means of other criteria. Being both acts of elicitation, it is possible to test the presence of interrogative and imperative illocutionary force by asserting the given SoAs as possible felicitous responses to the elicitation. We have already seen that only what is asserted may be challenged, but also the reverse is true: only what is challenged (i.e. what is at issue) can be asserted. This symmetry reveals the fact that the different types of illocutionary force stand in a paradigmatic relation and constitute different ways of relating to what is at issue in the speciﬁc communicative act. Two instances are given in (1.32) and (1.33). In example (1.32) the information that the speaker wants to elicit is the effective occurrence of the ﬁrst SoA, that is, that somebody said something, not what this person has said. The assertion of the ﬁrst SoA may indeed be a felicitous response to the elicitation, while the assertion of the second one could not. The same happens in (1.33), in which what is elicited is not some information but rather an action. The assertion of the ﬁrst SoA does not make any sense, because it is not what the speaker wants to elicit, while the assertion of the other two SoAs constitute a felicitous response to the speaker’s elicitation. (1.32) Did he say that it’s raining? He said it. *it’s raining.

The notion of coordination

43

(1.33) After having sat down, tell me the truth and don’t lie again! I tell you the truth and I don’t lie again. *I sit down. When native speakers are available, it is possible to directly ask for any clariﬁcation in case of doubt (see section 1.2 for the use of questionnaires). In case the only source of information is a descriptive grammar, there are two possible conditions. On the one hand, there may be explicit information regarding the illocutionary organization of the sentence under exam, on the other hand such information may lack. In case some overt information is available, it could consist of an explicit discussion of the author of the grammar or it could be signalled by the presence of markers indicating speciﬁc speech acts (see example (1.31) above). In Lezgian, for instance, the aorist converb -na may be used to express a sequence of actions sharing the same agent, as in (1.34). The author of the grammar explicitly states that he translates the Lezgian construction in (1.34) with an English coordinate construction because the SoA expressed by the converb clause has the same prominence of the other one (Haspelmath 1993: 376). The intention underlying the translation is thus to maintain the conceptual organization of the original sentence. (1.34) Lezgian, Nakh-Daghestanian cˇ a-z wiri-da-z Naˇcal’nik.di, sehne.di-z e˜qeˇc-na director.ERG [stage-DAT go.out-AOC] we-DAT all-SBST-DAT cˇ uxsaˇgul laha-na thanks say-AOR ‘The director came onto the stage and thanked all of us.’ (Haspelmath 1993: 376) On the other hand, however, there are also cases where no information is available on the communicative organization of the sentence, and this mainly happens in case of declarative sentences. In such cases, as Cristofaro (2003: 41) argues, one has to rely on translations. Namely, it is to be assumed that the translation preserves the conceptual and communicative organization of the original sentence, following the thought made explicit by Haspelmath (1993: 376) above. If the author chooses an English coordinate construction to render a given sentence, it will presumably be because this construction better mirrors the conceptual organization of the two linked SoAs. This procedure

+74>A4C820; 5>D=30C8>=B C74 =>C8>= >5 2>>A38=0C8>= =G C: 7CIFG9 =B8=F97H 5B8 H<9 H9GHG AIGH 69 5DD@=98 HC H<9 HF5BG@5H98 B;@=G< G9BH9B79G K<=7< A5M C6J=CIG@M69 >I8;98 ACF9 CF @9GG 577IF5H9 5B8 ACF9 CF @9GG 7CBG=GH9BH K=H< H<9 CF=;=B5@ CB9G !CK9J9F HF5BG@5H=CBG G=8B C: H<=G F9G95F7< =G 5 F9@5H=CB C: G9A5BH=7 7CB79DHI5@ 5B8 DF5;A5H=7 D5F5@@9@=GA 69HK99B HKC ,CG ": HKC ,CG 5F9 G9A5BH=75@@M 7CB79DHI5@@M 5B8 DF5;A5H=75@@M D5F5@@9@ H<9M 5F9 G5=8 HC GH5B8 =B 5 7CCF8=B5H=CB F9@5H=CB ,=B79 H<9 5=A C: H<=G F9G95F7< =G H<9 5B5@MG=G C: H<9 ACFD97H C: H<=G F9G95F7<

-MD9G C: 7CCF8=B5H=CB F9@5H=CBG

=J9B HKC ,CG K<=7< 75B 69 =89BH=:=98 5G :IB7H=CB5@@M D5F5@@9@ H<9M 75B GH5B8 =B 8=::9F9BH 7CB79DHI5@ F9@5H=CBG -IB7H=CB 8=G>IB7H=CB 5B8 58J9FG5H=J=HM FCI;<@M 7CFF9GDCB8=B; HC H<9 A95B=B;G C: a5B8b aCFb 5B8 a6IHb =? %CB;57F9 !5GD9@A5H< 5B8 -<9G9 8=GH=B7H=CBG <5J9 HF58=H=CB5@@M 699B =89BH=:=98 H
The notion of coordination

45

If we compare the two sentences in (1.35) from Japanese, we can see that, within the same language, two different types of coordination relations are coded by constructions that have different morphosyntactic features. The construction used in (1.35a) to express a combination relation (what is traditionally called ‘conjunctive’) is not syntactically symmetric, since the ﬁrst verb form is sufﬁxed by the converb -i and depends on the second verb. The construction used in (1.35b) to express a contrast relation, instead, is syntactically symmetric, since the ﬁrst verb form is sufﬁxed by the marker -ga but neither of the two verbs depends on the other. (1.35) Japanese Hanako-ga Furansu (a) Taroo-ga Amerika ni ik-i, Taro-NOM America to go-CONV Hanako-NOM France ni it-ta to go-PST ‘Taro went to America and Hanako went to France’ (adapted from Alpatov and Podlesskaya 1995: 472) (b) nodo-ga kawaki-masi-ta-ga orenzizyuusu-wa throat-NOM dry-HON-PRF-COORD orange.juice-TOP suki-dewaarimasen like-NEG ‘I am thirsty but I don’t like orange juice’ (Y.N., questionnaire) Furthermore, there are languages which use overt connectives to code one type of coordination relation and simple juxtaposition of two clauses to express another type of coordination relation. Koromfe, for instance, uses an overt connective to express an alternative relation (‘disjunctive’, in traditional terms) between two SoAs (1.36a) but does not use any marker to signal simple combination (1.36b). (1.36) Koromfe, Gur, Niger-Congo la a m˜u˜ı maa bεllaa (a) maa b2dini or (proper name) come.PROG with ART rice or kemde bεllaa la a fã˜ı (proper name) come.PROG with ART milletporridge ‘Badini will bring some rice or Kemde will bring some millet porridge.’ (Rennison 1997: 91)

46 Theoretical foundations: the notion of coordination (b) d@ PRON .3sg. HUM

sa zoe dãN nε d@ enter.PST house.sg in PRON.1sg father.sg

tuko sit ‘He entered the house and my father sat down.’ (Rennison 1997: 88) Examples (1.35) and (1.36) highlight the cross-linguistic variation attested in the coding of the various coordination relations, which are not always conveyed by an overt dedicated marker, nor are they necessarily expressed by a morphosyntactically parallel construction. Nonetheless, the cross-linguistic variation is not arbitrary, as will become clear from the results of this work. The fact that the coordination relations are mostly expressed by syntactically symmetric constructions with explicit markers is itself a signiﬁcant datum, which is certainly worth of attention and plausibly mirrors the very nature of such conceptual relations. Moreover, cross-linguistic comparison points to strong connections between the morphosyntactic properties of a given construction and the speciﬁc type of coordination relation expressed. This means that speciﬁc formal properties tend to correlate with particular conceptual situations. In order to understand the mechanisms which link the form to the speciﬁc function it expresses, a thorough analysis of the conceptual structures at issue is necessary. Two functionally parallel SoAs constitute the object of this analysis to the extent that they are linked in one of the following three coordination relations: combination, contrast or alternative. Any morphosyntactic construction used to convey one of these relations will be taken into account, regardless of its morphosyntactic properties. These three types of conceptual relations do not exhaust the list of possible coordination relations. Dik (1968: 277-79), for instance, includes certain types of causal relations within the set (the ones conveyed by constructions with for in English, car in French and denn in German). The conclusive relation (‘I was tired, so I decided to go to bed’) and the correlative relation (‘The more I travel, the more I want to do it’) can be coordination relations too. This research is however restricted to the three relations mentioned above. One objection that may be raised is that there is no use in keeping this threefold distinction as the starting point of this work, if the three relation types are not identiﬁed as possible functions of the same construction type.

The notion of coordination

47

It could sound strange to criticize the form-to-function approach and, at the same time, to take advantage of what has been discovered through it. However, if the aim of linguistics is to uncover the mechanisms underlying the structuring of languages, the two perspectives should not be seen as antagonist, but as present at the same time (cf. Prandi 2004: chapters 1 and 2, and Ramat 1999). Therefore, the results achieved by researching in one direction should become the ground for the next research in the opposite direction. The fact that combination, contrast and alternative are the conceptual relations that languages most frequently express by means of syntactically symmetric structures, for instance, is not devoid of meaning. On the contrary, such a result is a strong invitation for an analysis which goes in the opposite direction. The fact that a speciﬁc morphosyntactic feature may be frequently associated with certain conceptual relations does not mean that these relations are necessarily always expressed by constructions showing that feature, neither within nor across languages. The comparison of the strategies used to express a given set of relations may (and does) reveal a complex picture of interconnections between form and function. So more light can be cast both on the conceptual structure of coordination relations, which become the center of the analysis and not just a listing of possible meanings, and on the mechanisms underlying their formal coding in language. In this research, the coordination relations of combination, contrast and alternative are identiﬁed as the basic types of symmetric links that may be established between two independent SoAs.13 At the basis of any further semantic speciﬁcation or argumentation, two linked SoAs must be basically presented as equally asserted (or elicited) co-existent parts of a unitary frame, or as equally supposed possibilities for a single slot within the given frame. Two co-existent parts of a frame may be presented as non-conﬂicting, thus standing in a relation of simple combination, or as conﬂicting in one or more respects, thus standing in a contrast relation. Two equally supposed possibilities are instead not asserted as co-existent parts, but are rather presented as potential substitutes for each other within the frame, thus standing in a relation of alternative. Some scholars, trying to systematize the coordination relations, have captured this fundamental opposition by means of logic formalizations (cf. Payne 1985). Boolean logic has indeed two operators which show some similarities with the meaning of the connectives ‘and’ and ‘or’ in natural language, to the point that these connectives are commonly used to refer to the logic operators. The symbol ∧ roughly corresponds to ‘and’, in that a proposition p ∧

48 Theoretical foundations: the notion of coordination q is true only if both p and q are true. The symbol ∨ roughly corresponds to ‘or’, in that a proposition p ∨ q is true if only one, either p or q, is true.14 Even though conjunction and disjunction are objects of analysis of Boolean logic, logic can hardly be the basis for a semantic analysis of conjunctive and disjunctive constructions in natural language.15 The relation between the semantics of Boolean disjunction and the semantics of conjunctive and disjunctive constructions in natural language is an interesting and complicated topic, but it will not be addressed here. The opposition between equi-asserted and equi-supposed SoAs is better represented by the basic distinction between cooccurring and non-cooccurring SoAs, which recalls another well-known dichotomy, namely the distinction between syntagmatic and paradigmatic axis. The three basic coordination relations considered in this study will be thus analyzed along these two dimensions. The distinction between cooccurrence and non-cooccurrence will turn out to be useful in the interpretation of data as well, because it clearly shows that combination and contrast on the one hand and alternative on the other do not stand one after the other along the same line, but rather develop along two separate, though related, dimensions (see chapter 6). Speciﬁcally, combination and contrast are the basic relations that may be established in praesentia and elaborate the very concept of cooccurrence, which may give rise to a conﬂict or not. Alternative is the basic relation that may be established in absentia and elaborates the notion of substitution and selection among equivalent possibilities. The notions of combination, contrast and alternative will be deeply examined in the ﬁrst sections of chapters 3, 4 and 5, before discussing their respective cross-linguistic coding.

Chapter 2 Parameters of analysis In this chapter some main methodological issues will be addressed. First, a brief introduction will be made, in which the object of analysis will be delimited and the selection of the parameters for a cross-linguistic research will be discussed (section 2.1). Then the individual parameters will be examined. Section 2.2 focuses on the criteria used at the semantic level, temporality, conﬂict and aim, on the basis of which the three coordination relations will be distinguished in different subtypes. In section 2.3 the discussion deals with the morphosyntactic parameters whereby the attested coordinating constructions are analyzed. The coding of coordination relations is examined in two respects: the coding of the relation and the coding of the coordinated SoAs. In section 2.4 the semantic level and the morphosyntactic one are linked along two complementary directions. Each construction is considered in its semantic domain, i.e. the set of coordination relations it may be used for, and each coordination relation is in turn considered from the point of view of its degree of coding, i.e. the degree to which it may be expressed by a dedicated construction. Finally, section 2.5 presents the methodology adopted in the analysis of the attested constraints on cross-linguistic variation, distinguishing between implicational patterns and non-implicational generalizations. Moreover, an overview of the different impacts that the various parameters have in the analysis of the various relations is provided, highlighting that certain parameters reveal implicational constraints in the coding of certain coordination relations, but not in the coding of others.

2.1 2.1.1

Object of analysis and methodology Delimiting the scope

The object of this research is the morphosyntactic coding of the coordination relations of combination, contrast and alternative between SoAs. In fact, the analysis is restricted to a subset of the possible constructions expressing combination, contrast and alternative.

50 Parameters of analysis First of all, only binary relations are considered, i.e. relations established between two SoAs. Secondly, the research is restricted to the basic forms attested in each language, namely to those cases where the relation is not emphasized (as in ‘both. . . and . . . ’, Payne 1985: 4). Thirdly, only the positive relations are considered and their negated counterparts have been left out of the analysis. Finally, as already mentioned in the previous chapter (section 1.3.3.3), only SoAs which are not identiﬁable as subordinate will be examined, in order not to interpret phenomena which are due to their subordinate status as phenomena due to the coding of coordination. Let us now brieﬂy discuss each restriction, in order to clearly delimit the scope of the research. It is well known that combination and alternative are characterized by their ability to be recursively established between a potentially inﬁnite number of entities, as can be seen in example (2.1). Contrast, on the other hand, occurs only between couples of SoAs. (2.1) Mary reads a book, John writes a letter, Andrew watches tv . . . and Susy sleeps. Some phenomena pertain only to multiple coordination, like the obligatory omission of the connective in all but the last clause or its obligatory repetition in each coordinated element. However, the coding of the coordination relation and the coding of the SoAs involved do not change from binary to multiple coordination, thus leaving the basic structure of the construction unaltered. Given that the exclusion of multiple coordination from the analysis should not change its main results, only binary combinations and alternatives will be examined in this work, since they constitute the minimal sets within which it is possible to establish a coordination relation. The second restriction concerns the instances of emphatic coordination, as Payne (1985: 4) calls it. Constructions like ‘both . . . and . . . ’ in English or ‘sia . . . sia . . . ’ in Italian, where emphasis is given on the fact that each element of the combination is separately considered, are excluded from the analysis. Languages behave differently with respect to the expression of separate consideration of the coordinated elements. Some languages repeat the basic connective (cf. Russian i. . . i. . . for the expression of separate combination, or Italian o. . . o. . . for the expression of emphatic alternative), others use the basic connective associated with a different form (English both. . . and . . . ) or a speciﬁc construction completely different from the basic one (German sowohl. . . als auch. . . , Italian sia. . . sia/ che. . . ) (cf. Haspelmath 2007). However, emphatic coordination normally

Object of analysis and methodology

51

links entities or properties, only rarely SoAs and, in case of emphatic coordination between SoAs, they are generally subordinate, as shown in example (2.2). For these reasons, instances of emphatic coordination will not be examined, even though a cross-linguistic research on the strategies used to express emphatic coordination could be an interesting development of this work. (2.2) Italian, Romance, Indo-European viaggiare sia riposarmi a casa sia Amo love:1sg COORD travel.INF COORD rest.INF:myself at home ‘I love both traveling and resting at home.’ This restriction concerns the cases of emphatic coordination and should not be taken as an exclusion of those constructions where two connectives are obligatory. A sentence like the one from Chechen in (2.3), for instance, is included in the analysis, even if it shows two connectives (. . .’a . . . ’a), because it expresses the basic combination relation, without emphatic separate consideration of the two SoAs. (2.3) Chechen, Nakh-Daghestanian chai’a mer dara, [. . . ] peetar-ie juxa-vaxaniehw inn-ALL back-go.PST.COND tea:and drink.FUT be.IMPF byysa’a joaqq-ur jara [. . . ] night:and spend-FUT be.IMPF ‘[. . . ] if we had returned to the inn, we could have drunk tea and spent the night [. . . ]’ (Jeschull 2004: 253) The third limit of this work is the exclusion of negated relations (called rejection by Payne 1985: 4), expressed by constructions like neither. . . nor. . . in English, né. . . né. . . in Italian, neque. . . neque. . . in Latin or weder. . . noch. . . in German. The reason for excluding them is that their coding is obviously strictly connected with the strategies that languages use to express negation, which is a different, albeit related, notion from the ones at issue in this work. Such constructions may originate from both basic conjunctive and disjunctive constructions (as instantiated by Latin and English, respectively), or they may have completely different forms. As Haspelmath (2007) points out, this is connected with the logical equivalence between disjunction with wide scope

52 Parameters of analysis negation and conjunction with narrow scope negation. In other words, ‘not (X or Y)’ is equivalent to ‘not X and not Y’ and languages seem to mirror this functional equivalence at the morphosyntactic level. By virtue of their connection with the coding of the notion of negation, the constructions used to express the negative coordination have been widely described by Bernini and Ramat (1996) in a comparative research on the negative sentence in the European languages. Their occurrence in non-European languages is not well attested, but this may simply be due to the scarcity of comprehensive descriptions of coordinating constructions in languages other than the European ones. Finally, this research focuses on non-subordinate SoAs, in order to isolate the morphosyntactic phenomena which characterize the coding of the coordination relations. Example (2.4) shows an alternative between subordinate SoAs. The verb forms encoding them are nonﬁnite (to play and to sing) and one may think that their non-ﬁniteness may be a consequence of their being part of a coordination relation. Yet, the fact that these two verbs are nonﬁnite is a consequence of them jointly expressing a subordination relation of purpose (cf. Cristofaro 2003). (2.4) She applied for the Music School in order to play an instrument or to sing in the choir. The analysis of how a coordination relation between subordinate SoAs is coded would reveal further levels of interaction between the two conceptual situations of functional parallelism and non-parallelism. Such research would be even more interesting if compared to the strategies used to express coordination between entities (nominal coordination), in order to see whether those languages which use different constructions for nominal and clausal coordination (cf. Haspelmath 2005) use the nominal strategy to code the coordination of subordinate SoAs.16 However, a cross-linguistic study on the coding of the coordination relation between subordinate SoAs goes beyond the scope of this research and requires further studies. The aim of this section was twofold. First of all, it was aimed to clearly deﬁne the object of analysis, in pointing out which combination, contrast and alternative constructions are examined. Secondly, it was also meant to highlight the limits of this work: it constitutes an attempt to systematically account for the basic coordination relations and their cross-linguistic coding, but the issues addressed are much fewer than the questions still open. Before the parameters of analysis are presented in detail, a brief introduction to the overall

&1942C >5 0=0;HB8B 0=3 <4C7>3>;>6H

A9H

&9H
-<9 :IB7H=CB5@HMDC@C;=75@ 5DDFC57< 58CDH98 =B H<=G GHI8M GHFCB;@M =B:@I9B79G H<9 7IGH=:=98 -<9 DFC79GG C: =BJ9GH=;5H=CB 58CDH98 H=8B 'CH CB@M H<9 7C8=B; C: 7CA6=B5H=CB 7CBHF5GH 5B8 5@H9FB5H=J9 =G 5B5@MN98 6IH 5@GC H<9 7C8=B; C: H<9 J5F=CIG GI6HMD9G C: 7CA6=B5H=CB H<9 J5F=CIG GI6HMD9G C: 7CBHF5GH 5B8 H<9 J5F=CIG GI6HMD9G C: 5@H9FB5H=J9 -<9 7CBGHFI7H=CBG 7C8=B; H<9 F9@5H=CBG 5B8 GI6F9@5H=CBG =89BH=:=98 5F9 H<9B 9L5A=B98 IB89F HKC F9GD97HG K<9H<9F 5B8 38=6 >5 C74 A4;0C8>= 5@ @CKG HC 7CAD5F9 H<9 A95BG 6M K<=7< H<9 J5F=CIG 7CB79DHI5@ G=HI5H=CBG 5F9 CJ9FH@M =B8=75H98 = 9 @9L=75@=N98 57FCGG @5B;I5;9G -KC EI9GH=CBG 5F9 5H =G GI9 <9F9 H<9 DF9G9B79 C: 5B CJ9FH A5F?9F K<=7< 9LD@=7=H@M G=;B5@G H<9 GD97=:=7 GI6F9@5H=CB 5B8 =B 75G9 =H =G DF9G9BH H<9 =BH9FB5@ 7CAD@9L=HM C: H<9 A5F?9F -<9 G97CB8 D5F5A9H9F =BGH958 895@G K=H< H<9 BH=C02C82 ?0A0;;4;8B< C: H<9

54 Parameters of analysis construction with respect to the verb forms encoding the SoAs. The parallel vs. non-parallel coding of two linked SoAs is inﬂuenced by the semantic properties and structure of the speciﬁc relation linking them. This parameter allows to compare the functional parallelism to the syntactic one, highlighting how the speciﬁc semantic properties of the various coordination relations determine different degrees of syntactic parallelism. The analysis of the attested coordinating constructions reveals that there is no one-to-one correspondence between the sub-relations identiﬁed and the attested morphosyntactic constructions, but rather a single construction may be used to express a variety of conceptual situations (section 2.4). This polysemy is however not random. The analysis and comparison of the semantic domains of each construction allows to build a semantic map of the subrelations of coordination, thus enabling to represent the conceptual space of functional parallelism in a more structured way than a simple list. Two parameters are traditionally considered central in the analysis of coordination, but they will not be analyzed here: ellipsis phenomena and the position of connectives with respect to the linked SoAs. Ellipsis phenomena characterizing coordinating constructions have been studied, among others, by Van Oirsouw (1987), Sanders (1997) and Harries-Delisle (1978). In this research, however, no speciﬁc inquiry on ellipsis or gapping has been made, since data collected through grammars and questionnaires did not allow a cross-linguistic comparison of such phenomena.17 On the other hand, for every attested coordinating connective the pre-positive or post-positive position is known (cf. Haspelmath 2007), but it has simply not been put at the center of the analysis and no patterns of variation have been looked for in this respect.

2.2

Semantic parameters: temporality, conﬂict and aim

Within the literature on coordination, many scholars have identiﬁed a number of semantic subtypes of coordination (cf. Haspelmath 2004 and 2007, Payne 1985, Longacre 1985, Lang 1984). Yet, these classiﬁcations do not identify the same subtypes, depending crucially on the languages taken into account by the various authors, nor do the parameters used to identify them correspond. The most structured classiﬁcations are those made by Payne (1985) and Lang (1984), which are based on formal abstractions of coordination relations and are essentially formulated in logical terms.

Semantic parameters: temporality, conﬂict and aim

55

As discussed in the preceding chapter however, in this study coordination relations are not deﬁned in terms of logic, but rather with respect to the cooccurrence and non-cooccurrence dimensions, along which two SoAs may be asserted (or elicited) as jointly real or as replaceable possibilities. Therefore, a new systematization is proposed here, based on criteria that are consistent with this distinction. The results of this classiﬁcation correspond to the ones reported by Haspelmath (2004 and 2007). Nonetheless, whereas he identiﬁes the various subtypes of coordination on the basis of the lexicalized constructions which express them across languages, here an independent functional analysis of each subtype is also provided. The three parameters at issue are temporality, conﬂict and aim. Temporality pertains to the way in which the linked SoAs relate to each other and thus concerns the predication level of the sentence. Conﬂict pertains to the way in which the linked SoAs are perceived by the speaker and is thus located on the propositional level. The third parameter, aim, pertains instead to the intention of the speaker in establishing the given relation and thus concerns the discourse level, that is, the speech act itself. These three parameters do not exhaust the list of semantic criteria that could be used and other parameters would of course lead to identify further sub-types of coordination relations. The three levels just mentioned belong to the four-layer subdivision of the clause proposed within the Functional Grammar (Dik 1997, Siewierska 1991). Each level encompasses the lower ones and designates a different entity. The most basic layer is that of predicates, which denote properties or relations, and terms, which denote things. On the second layer we ﬁnd predications designating SoAs, which in turn consist of relations between things and properties. The relation of combination and the parameter of temporality are located on this layer. The third layer is that of proposition and concerns what is thought about the SoAs. The relations of contrast and alternative are established at this level, as well as the parameter of conﬂict. The uppermost level is ﬁnally that of the clause, which encompasses the speech act as a whole, in the various facets of its communicative dimension. The parameter of aim is located on this layer. Let us now brieﬂy analyze the three parameters. The parameter of temporality identiﬁes the location of the linked SoAs on the temporal axis. It basically distinguishes between temporal (2.5a) and atemporal relations (2.5b). A temporal relation may be sequential, if it is characterized by the location of the SoAs along the same time axis at successive points, or simultaneous, if the SoAs are located at the same point along the same time axis. An atempo-

'0A0<4C4AB >5 0=0;HB8B F5@ F9@5H=CB =G 7<5F57H9F=N98 6M H<9 @C75H=CB C: H<9 ,CG CIHG=89 H<9 H=A9 5L=G CF 6M H<9 =FF9@9J5B79 C: H<9=F H9ADCF5@ 7C@@C75H=CB K<=7< A5M 69 9=H<9F G9 EI9BH=5@ CF G=AI@H5B9CIG K=H?4=43 C74 3>>A 0=3 A0= 0F0H 6 8A3B H 0=3 B74B BF8< -<9 D5F5A9H9F C: 2>=82C DC=BHG CIH H<9 DF9G9B79 5B8 9GD97=5@@M H<9 B5HIF9 C: 5 7CB:@=7H 69HK99B H<9 HKC ,CG "H A5?9G 5B CJ9F5@@ 8=GH=B7H=CB 69HK99B 7CB:@=7H=B; 5B8 BCB7CB:@=7H=B; F9@5H=CBG 5B8 7@5GG=:=9G H<9 7CB:@=7H=B; F9@5 H=CBG CB H<9 65G=G C: H<9 CF=;=B C: H<9 7CB:@=7H -<9F9 A5M 69 G9J9F5@ F95GCBG IB89F@M=B; 5 7CB:@=7H 5B8 C6J=CIG@M CB@M H<9 A5=B CB9G K=@@ 69 7CBG=89F98 7CB:@=7H 7CI@8 H4B =>C ?;0H 10B:4C10;; 6 '4C4A 8B =>C BCD3H8=6 1DC 74 8B ?;0H8=6 8= C74 60A34= 7 '4C4A 8B A4;0G8=6 F74A40B 70E4 144= F>A:8=6 C74 F7>;4 30H -<9 D5F5A9H9F C: 08< DC=BHG CIH H<9 7CAAIB=75H=J9 =BH9BH=CB K<=7< A5?9G H<9 GD95?9F 9GH56@=G< 5 7CCF8=B5H=CB F9@5H=CB 69HK99B HKC ,CG "H =89BH=:=9G HKC DCGG=6=@=H=9G H<9 GD95?9F 9GH56@=G<9G H<9 F9@5H=CB =B CF89F HC 5GG9FH CF 9@=7=H =B:CFA5H=CB 56CIH H<9 B4C C: @=B?98 ,CG 5G 5 6C8M 5 CF H<9M 9GH56@=G< H<9 F9@5H=CB =B CF89F HC 5GG9FH CF 9@=7=H =B:CFA5H=CB 56CIH H<9 8=38E83D0; @=B?98 ,CG 6 -<=G D5F5A9H9F K=@@ 69 :IB85A9BH5@ =B H<9 =89BH=:=75H=CB C: H<9 GI6HMD9G C: 5@H9FB5H=J9 F9@5H=CB 5 ,BD0;;H A403 >A F0C27 +- D=C8; ;0C4 6 > F4 6> C> B27>>; >A 3> F4 BC0H 0C 7><4 C><>AA>F 57< C: H<9 HA0; 5 5B8 0C4A0; 6 7CA6=B5H=CB 89D9B8=B; CB H<9 DF9G9B79 C: 5 H9ADCF5@ G9EI9B79 =B K<=7< H<9 HKC GH5H9G C: 5::5=FG H5?9 D5FH %CB;57F9 -<9 F95GCB C: H<9 7CB:@=7H 8=GH=B;I=G<9G =BGH958 69HK99B

Morphosyntactic parameters

57

counterexpectative, corrective and oppositive contrast, depending on whether the conﬂict originates in the denial of an expectation (2.6a), in the correction of an overtly negated element (2.6b) or in the semantic opposition of the two states of affairs (2.6c) respectively (Lang 2000). Finally, the aim of an alternative distinguishes between choice-aimed (2.7a) and simple (2.7b) alternative, depending on whether the alternative is established to deﬁne a set of options or to make a choice between them. A more detailed discussion on each sub-type will be provided in the next chapters (sections 3.1, 4.1 and 5.1).

2.3

Morphosyntactic parameters

Any construction expressing one or more coordination relations or one or more sub-types of coordination relation will be included in this research. Each attested construction will be then examined on the basis of two major morphosyntactic parameters: one concerning the presence, the complexity and the semantics of markers encoding the speciﬁc coordination relations, and the other concerning the syntactic parallelism of the construction with respect to the verb forms encoding the two SoAs.

2.3.1

The coding of the relation

The coding of the relation primarily concerns the presence of an overt marker encoding one of the coordination relations. It may be further distinguished in two aspects: on the one hand, the presence of the marker itself is contrasted with its absence; on the other hand, the morphophonological complexity of the attested coordinating markers is analyzed in terms of number of syllables, number of morphemes and free or bound status of the marker. Each of these aspects will be examined and exempliﬁed in the next sections. Before proceeding any further with the discussion, however, the concept of coordinating marker needs to be deﬁned, in order to clarify what markers, among all the particles which to some extent participate in delimiting the semantic content of a relation, will be considered to be coordinating markers. By coordinating marker, or connective (the two terms will be interchangeable throughout the work), is here meant any ‘overt relational marker’ (Wälchli 2005: 37) which encodes at least one coordination relation or a sub-type of it.

58 Parameters of analysis These markers have been referred to as ‘conjunctions’ in the European tradition, mainly denoting free and invariable morphemes. The term conjunction however, is here restricted to the expression of the combination relation and will thus not be adopted in the European traditional sense. Dik (1968: 34-41) and Haspelmath (2007) propose the label ‘coordinator’, which is indeed clear and clearly contrasting with its antonym ‘subordinator’ (Kortmann 1997). Yet, in this research the term ‘coordinating marker’ is preferred because it is more transparent with respect to the two notions underlying it: the presence of an overt marker, whatever its morphological nature may be, and its function of establishing a coordination relation between the SoAs it connects. The expression ‘whatever its morphological nature may be’ refers to the fact that a coordinating marker needs not necessarily be a free morpheme nor its position must be necessarily initial, as it happens in most European languages. Different strategies are attested to overtly indicate a coordination relation and it would be difﬁcult to include all these strategies under one and the same morphosyntactic delimitation. In the following sentence from Turkish for instance, the marker establishing the combination relation is the sufﬁx -Ip, which is a bound form usually analyzed as belonging to the verbal paradigm (see Johanson 1995). In other contexts, this verb form may also be used to express subordinate relations and it is here considered as a coordinating marker by virtue of its coordinating function. (2.8) Turkish, Altaic (Kornﬁlt 1997: 110) Hasan i¸s-e gid-ip ev-e dön-dü Hasan work-DAT go-and home-DAT come-PST ‘Hasan went to work and came back home.’ Different morphosyntactic characterizations of coordinating markers have been proposed, even though the identiﬁcation of a speciﬁc category is still a matter of dispute. Lang (2002: 636), for instance, deﬁnes what he calls ‘Konjunktion’ as a particle connecting two syntactic elements into a complex one, which has an invariable morphology and explicitly indicates elements with the same function as interconnected. Dik (1968: 34-37), on the other hand, assumes that ‘two members can never be coordinated by more than one coordinator’, and this has afterwards become one of the main criteria for distinguishing coordinating markers from adverbial connectives.

Morphosyntactic parameters

59

Dik (1968: 34) states that, given a particle which might be a coordinator, this should be juxtaposed to another particle, whose coordinator-status has already been established. In case the two particles can be added to each other, the one at issue must not be identiﬁed as a coordinator. If their juxtaposition is not possible, then it can be considered a coordinator. In English, this test identiﬁes for as a coordinator but excludes yet. In Italian it includes mentre ‘while’ (in its non-subordinative meaning, e.g. (2.9a)), but excludes invece ‘instead’ (e.g. (2.9b)). Such a method however has some cross-linguistic limits, since the attested coordinating markers are not all on the same level of grammaticalization. In fact, there could be markers which are on the way to become coordinators but that do not show all the speciﬁc properties of the category yet and can cooccur with other coordinators (cf. Mithun 1988). (2.9) Italian, Romance, Indo-European già. (a) Io studio ancora (*e) mentre lui lavora COORD he work:3sg already I study:1sg still ‘I am still a student while he already works.’ (b) Voleva attore e invece fare l’ want:IMPFV:3sg do DEF actor COORD instead è diventato avvocato. AUX . PRS:3sg become: PST. PTCP lawyer ‘He wanted to be an actor and became a lawyer, instead.’ Both Kortmann (1997) and Haspelmath (2007) highlight the impossibility to assign sharp boundaries to the category of coordinators (the term is here used to maintain their terminology), because of the different grammatical properties of languages and because of the different levels of grammaticalization shown by the various coordinating markers. Following them, it seems best to deal with this issue in terms of a cline or a continuum. Along this continuum, the markers included in the analysis as coordinating markers are the most grammaticalized ones. All the instances identiﬁed by means of the cooccurrence test will be encompassed by this criterion, since the impossibility to cooccur is a consequence of the high grammaticalization18 of the coordinating markers.

60 Parameters of analysis 2.3.1.1

Presence of an overt coordinating marker

This parameter distinguishes between syndetic and asyndetic coordinating constructions, the former characterized by the presence of an overt coordinating marker and the latter characterized by its absence. Asyndetic constructions consist of the simple juxtaposition of the two SoAs, and the speciﬁc coordination relation existing between them is inferred from the context of communication and from their semantic properties. Asyndesis is always possible and occurs in every language as a more or less stylistically marked strategy. English, for instance, normally uses overt markers to signal the various coordination relations (and, or and but), but may resort to juxtaposition in particular contexts, like listings of many SoAs or narrative chains (2.10). Moreover, as Longacre (1985: 251) points out, certain antithetical relations may be conveyed either by simply setting the two SoAs against each other by means of a juxtaposition or by using a marker between them. An example of such double option is given in (2.11), in which the corrective contrast is equally conveyed with an asyndetic construction in (2.11a) and a syndetic one in (2.11b). (2.10) I came, I saw, I conquered. (2.11) (a) It’s not black, it’s white. (b) It’s not black, but white. Mithun (1988: 331-36) argues that there are different degrees to which the expression of coordination relations is grammaticalized across languages and there are several languages which ‘contain no grammaticized markers of coordination at all’ (Mithun 1988: 331). In such languages, the coordination relations are expressed by asyndetic constructions characterized by different intonation patterns. Mithun identiﬁes two main ways in which the relations are signalled by intonation: there could be no intonation break or they may be separated by a pause, roughly corresponding to a comma, with a non-ﬁnal pitch contour. She has remarked a regular use of such strategies across languages. On the one hand, the absence of intonation break is associated to the expression of subparts of a single event, roughly corresponding to sequential combination. On the other hand, a comma pause is associated with the expression of conceptually distinct aspects of a scene, roughly corresponding to atemporal combination.

Morphosyntactic parameters

61

Since the juxtapositive strategy is always available across languages, this research will consider as asyndetic coordinating constructions only those cases where a given coordination relation is normally expressed without any coordinating marker, whereby ‘normally’ means that the asyndetic strategy is the most stylistically unmarked or is the only strategy available for the expression of the given relation. In Chechen, for instance, the asyndetic strategy is very widespread for the combination of SoAs, even in cases where the linked SoAs are only two (Jeschull 2004: 252), and thus constitutes a stylistically unmarked strategy (2.12). As we have already seen in example (2.3) Chechen also has a syndetic strategy, but none of the two options seems to be more prominent than the other. In such a cases, both the syndetic and the asyndetic construction will be examined. (2.12) Chechen, Nakh-Daghestanian. byylira darc hwovziira Mox c’iiza wind howl.INF start:WP blizzard turn.around:WP ‘The wind started to howl and the blizzard turned around.’ (Jeschull 2004: 252) Asyndetic constructions are also attested for the expression of the alternative relation, as can be seen in example (2.13). As Merlan (1982: 33) points out, in Mangarayi there is no overt expression which directly corresponds to English or, either at the phrase level or at the clause level. Consideration of alternatives is often expressed by sequences of juxtaposed clauses containing the dubitative adverb maNaya ‘perhaps’. This adverb, although necessary to infer an alternative relation, cannot be considered a coordinating connective, since it also occurs when only one SoA is presented. (2.13) Mangarayi, Gunwingguan, Australian maNaya d.ayi maNaya ja-0-n / . iNa-n perhaps 3-3sg-come-PRS perhaps NEG ‘Perhaps he’ll come, perhaps not.’, i.e. ‘it is possible that he may or may not come’ (Merlan 1982: 39) On the contrary, when an overt coordinating marker is present, the coordination relation is not to be inferred, but is lexicalized in the marker itself, as can be seen in example (2.14) from Polish, in which the alternative relation is overtly coded by the marker lub.

62 Parameters of analysis (2.14) Polish, Slavic, Indo-European. lub az˙ do pó´zna czytam Zazwyczaj pisze˛ Usually write.PRS.1sg COORD read.PRS.1sg until to late ‘Usually I write or I read until late.’ (A.L., questionnaire) As will be made clear in section 2.4, the presence of a coordinating marker does not exclude the necessity of inferential enrichment, because it does not necessarily code the speciﬁc relation type at issue. In case the marker is used to express many relation types, further speciﬁcations regarding the semantic parameters of temporality, conﬂict and aim are left to inference. This is not the case with the Polish example in (2.14), in which lub lexicalizes the simple alternative, but it is indeed the case with the English connective or, which may express both a simple and a choice-aimed alternative. Before moving on to the morphophonological complexity of coordinating markers, the case of markers attested in chain languages (cf. Longacre 1985) is discussed. In these languages, the ﬁrst verb form is often sufﬁxed by a marker indicating whether the subject of that verb is the same as the one of the following verb, and such markers are commonly known as switchreference markers. They will be considered markers of coordination as long as they also encode, besides switch-reference, a coordination relation. Let us consider, for instance, the case of Tauya. (2.15) Tauya, Brahman, Trans-New Guinea 0-tu-e-ĳa / (a) ne fofe-a-te 3sg come-3sg-DS 3sg-give-1/2-IND ‘He came and I gave (it) to him.’ (Mac Donald 1990: 224) (b) wate ya-pi-ĳai yate-pa ni-e-ĳa house 1sg-GEN-ADESS go-SS eat-1/2-IND ‘I went home and ate.’(Mac Donald 1990: 224) In Tauya, the medial sufﬁxes -te and -pa do not only denote different-subject and same-subject respectively, but they also indicate that the relation established between the two SoAs is one of coordination. Mac Donald (1990: 224) calls them same-subject and different-subject coordinate medial verbs, as opposed to the subordinate ones. The use of the two terms in the grammar refers to a basically functional distinction, shown by the systematic use of English

Morphosyntactic parameters

63

coordinate constructions to translate coordinate medial verbs (cf. (2.15)) and English subordinate constructions to translate subordinate medial verbs. Therefore, the medial sufﬁxes -te and -pa will be considered coordinating markers, even though they also mark the identity vs. diversity of subject from one clause to the other. On the contrary, switch-reference markers that do not make a distinction between the types of relation established, and can thus be used both to express coordination and subordination relations, will not be considered coordination markers, because they would only mark switchreference and the relation itself would not be coded anywhere. In that case the construction would be considered to be asyndetic. 2.3.1.2

Morphophonological complexity

All the attested syndetic constructions are examined on the basis of the (strictly synchronic) morphophonological internal complexity of the coordinating markers. Every coordinating marker has been analyzed in three respects: the free or bound nature of the marker, the number of its syllables and the number of the morphemes of which it consists. This parameter has been chosen following the assumption made by Kortmann in his work on the European subordinators. He argues that ‘detailed cross-linguistic information on the morphosemantics of one particular lexical category can be used for testing possibly universal tendencies in the domain of form-meaning and form-function mappings. ’ (Kortmann 1997: 113)

In his study, the morphophonological complexity of subordinators resulted to be indirectly proportional to their semantic and functional versatility. In other words, the more general is the function of a given subordinator, the simpler is its internal structure. Kortmann calls this result ‘Inverse Relation Hypothesis’. Bearing in mind that the aim of this research is ultimately the analysis of the form-function mapping in the expression of coordination relations, Kortmann’s Inverse Relation Hypothesis constitutes a good precedent, because it shows that the analysis of the morphophonological complexity of connective markers correlates with their semantics. The exam of the internal complexity of the attested coordinating markers not only may reveal the mechanisms of interaction between the speciﬁc coordination relations and their coding, but

64 Parameters of analysis may also highlight markedness phenomena indicating that some coordination relations are more basic than others. Data resulting from this parameter are closely connected to the grammaticalization level reached by the marker and to its frequency of use. As will be made clear throughout the work, and especially in sections 6.2.3 and 6.3.3, a high frequency of use determines the morphophonological erosion of a marker (Croft 2003: 110-16, 256), thus determining its progressive simpliﬁcation and the loss of internal transparency. The same phenomena also occur in consequence of a process of grammaticalization, whereby lexical items tend to develop grammatical functions, progressively losing their formal and semantic transparency, morphological complexity, phonological substance and syntactic freedom (Croft 2003: chapter 8, Hopper and Traugott 2003). Let us now see in detail the dimensions along which the internal complexity of the markers is analyzed. As for the criteria adopted in establishing the bound or free nature of a marker, given the central role that prosody plays in this respect and the scarcity of prosodic data, I had to rely on descriptive grammars, when no native speakers were available for clariﬁcations. Normally, a morpheme is analyzed as bound when it is not phonologically or syntactically independent, but is always attached to another morpheme. However, criteria for establishing the clitic or afﬁx status of a morpheme are largely language speciﬁc. An example of bound overt marker is given in (2.16) from Hebrew, where the combination marker ve= is proclitic and bound to the following word. Examples of free markers are found in the great majority of the European connectives, like the English connectives and, but and or, or the Finnish ones, vai and tai meaning ‘or’, ja meaning ‘and’, mutta and vaan meaning ‘but’. (2.16) Hebrew, Semitic, Afro-Asiatic Harbè studentìm lomdìm bemèshech hayòm many student:pl study:3pl during day ve=ovdìm baèrev COORD=work:3pl at.night ‘Many students study during the day and work at night.’ (H.M., questionnaire)

Morphosyntactic parameters

65

As for the number of syllables, the attested coordinating markers vary from monosyllabic (like the Italian e, o and ma), to polysyllabic (like German oder, Spanish pero and sino or Lezgian taˆxajt’a, meaning ‘or’). Furthermore, polysyllabic overt markers can be monomorphemic or polymorphemic. Examples of bisyllabic monomorphemic marker are German corrective marker sondern (2.17), or Danish marker eller, whereas examples of polysyllabic and polymorphemic markers are French oppositive marker tandis que, Italian corrective marker bensì or English oppositive marker whereas (cf. Kortmann 1997: 78). (2.17) German, Germanic, Indo-European nicht in seinem Zimmer, sondern Peter studiert Peter study:IND.PRS.3sg NEG in his:DAT room.DAT COORD im Moment spielt Garten er im in:DEF.DAT moment play:IND.PRS.3sg 3sg in:DEF.DAT garden ‘Peter is not studying in his room, but he’s playing in the garden.’ (C.W., questionnaire)

2.3.2

The coding of the SoAs

The coding of the SoAs concerns the syntactic parallelism of the construction in relation to the verb forms encoding the two SoAs. The question at issue is whether the functional parallelism characterizing the coordination relations is mirrored at the syntactic level by a reciprocal syntactic independence of the two verb forms. As stated in the introduction to this chapter, the internal parallelism of a construction could also be tested by means of other parameters, like the analysis of ellipsis phenomena or the position of the coordinating marker (cf. Haspelmath 2007). However, the parallelism that will be investigated here concerns only the verb forms coding the linked SoAs, which may be deranked or balanced. Before examining the notion of syntactic parallelism in detail, let us ﬁrst deﬁne the two types of verb forms which will be considered. Following Stassen (1985) and Cristofaro (2003), we can distinguish between two main types of verbal strategy: balancing and deranking (Stassen 1985: 76-83). A verb form is deﬁned as balanced when it can occur as it is also in independent declarative clauses taken in isolation, whereas it is analyzed

66 Parameters of analysis as deranked if it cannot be found in such clauses, because it lacks certain distinctions (such as tense, mood, aspect or person agreement) or it is a special form not allowed in independent clauses (Cristofaro 2003: 50-60). In other words, a deranked verb form is structurally different from the corresponding one that would be used in independent isolate clauses. This distinction is similar to that established by Haspelmath (1995: 26) between dependent and independent verb forms, in that the parameter distinguishing between them is their ability to occur in declarative independent clauses, without restrictions on the reasons why this may occur. Let us now discuss some examples of balanced and deranked verbs in relation to the coordinating constructions at issue. 2.3.2.1 Syntactic parallelism of the construction The syntactic parallelism distinguishes between parallel and non-parallel coordinating constructions. A parallel coordinating construction is characterized by the same coding strategy for both SoAs, either balancing or deranking, while a non-parallel coordinating construction is characterized by different coding strategies for the two SoAs, one balancing and the other deranking. Parallel constructions are widespread in the languages that Longacre (1985: 237-238) calls co-ranking (see section 1.3.2). In (2.18) and (2.19) two examples of parallel constructions are given from Portuguese and Icelandic. In both cases the syntactic parallelism is determined by the balanced status of the verb forms encoding both SoAs, none of which depends on the other, since they are both able to receive the speciﬁcations they would show in an isolated declarative clause. (2.18) Portuguese, Romance, Indo-European vou para Veneza Roma e viver Deixei Venice leave:PST.3sg Roma COORD go:PRS.3sg live.INF to ‘I left Rome and I will move to Venice.’ (A.R., questionnaire) (2.19) Icelandic, Germanic, Indo-European Ég heyrði þrumu og það er 1sg hear.PST thunder:GEN COORD 3sg be(AUX).PRS.3sg farið að rigna start:PST.PTCP INF rain

Morphosyntactic parameters

67

‘I heard a thunder and it started to rain.’ (A.H., questionnaire) A parallel construction does not necessarily consist of balanced verb forms. The two sentences in (2.20) from Malayalam, for instance, are examples of syntactically parallel combination and alternative constructions, whose parallelism is determined by the use of the same deranked form to code both SoAs. In both cases, the verbs are nonﬁnite and receive sufﬁxes: in (2.20a) they are sufﬁxed with the combination marker -um, while in (2.20b) they are sufﬁxed with the alternative marker -oo. These markers cannot be sufﬁxed to ﬁnite balanced verb forms (Asher and Kumari 1997: 135) but only to phrases and nonﬁnite verbs. The deranked verbs are then followed in both sentences by the ‘dummy verb’ ceyyuka, used to convey tense and mood (Asher and Kumari 1997: 135). If one wants to establish a combination relation between balanced verbs, no overt marker can be used and the only possibility is to juxtapose them in two separate sentences. On the contrary, if two balanced forms are to be linked in an alternative relation, there is the possibility of adding a free marker in between. (2.20) Malayalam, Tamil-Kannada, Dravidic kôùïan pookukay-um ceytu (a) raaman varikay-um Raman come:INF-COORD Krishnan go:INF-COORD do:PST ‘Raman came and Krishnan went’ (Asher and Kumari 1997: 135) (b) ñaan coor@ uïïukay-oo cappaatti tinnukay-oo ¯ ¯¯ 1sg rice eat:INF-COORD chapati eat:INF-COORD ceyyaam do:FUT:MOD ‘I shall eat rice or eat chapati.’ (Asher and Kumari 1997: 140) In the same language, the construction expressing contrast is parallel in a different way, since both verb forms coding the SoAs are balanced, as can be seen in example (2.21), and can receive tense and mood markers. In this case, the relation is not coded by bound morphemes, but by the free connective pakùe located between the two clauses.

68 Parameters of analysis (2.21) Malayalam, Tamil-Kannada, Dravidic avan daridran, aaï@, pakùe sahaayam sviikarikkilla be:PRS COORD help accept:FUT:NEG 3sg poor:M ‘He is poor but won’t accept help.’ (Asher and Kumari 1997: 138) Let us now examine some instances of non-parallel coordinating constructions. Many examples have already been provided in the previous chapter, in section 1.3.1 (cf. examples (1.8) from Turkish, (1.9) from Korean, and (1.4) from Japanese). Let us now examine the case of Lezgian, where the aorist converb -na is sufﬁxed to the ﬁrst verb and cannot receive further speciﬁcations of mood.19 (2.22) Lezgian, Nakh-Daghestanian (Haspelmath 1993: 377) cˇ na tadi-z fe-na am De ša, PT come: IMPF we: ERG quick- ADVB go- AOC he: ABS kuˇcuk-in bury-HORT ‘Come on, let us go quickly and bury him.’ Almost all the examples provided so far are instances of narrative coordinative converbs (Nedjalkov 1995: 109). They occur in Turkic languages and in Japanese among others, but are not attested in Slavic and Finno-Ugric languages, which are the European languages where the use of converbs is most common. Narrative converbs may express a variety of functions and result in being less semantically speciﬁed than the corresponding parallel constructions. Besides their narrative function, they are also normally used to express causal, conditional and manner functions. What is relevant for this research is that they may constitute the normal strategy employed to express some coordination relations. Converbal forms, however, are not the only type of non-parallel constructions attested in the sample. In example (2.23) we can see a sentence from West Greenlandic, in which a non-parallel combinative strategy is used. This language has also the possibility to use a parallel strategy, and the non-parallel one is only allowed when the subjects of the two clauses are co-referential. As can be observed from the example, the two SoAs are coded by different strategies, since the second

Morphosyntactic parameters

69

verb form is in the so-called contemporative mood followed by the coordinating marker =lu, while the ﬁrst one is in the superordinate mood, which is the one attested in independent isolated clauses. (2.23) West Greenlandic, Eskimo-Aleut ingil-luni=lu isir-puq come.in-3sg:IND sit.down-4sg:CONT=and ‘She came in and sat down.’ (Fortescue 1984: 120) Non-parallel coordinating constructions are widely attested in chaining languages, where the medial verbs are in most cases sufﬁxed by switch-reference markers and cannot receive person agreement, nor can they show mood speciﬁcations. In the preceding sections two examples from Tauya were provided in order to exemplify the merging of switch-reference with the coding of the combination relation (cf. example (2.15)). One of the two sentences is repeated here, in order to point out the syntactic non-parallelism of the construction. The ﬁrst SoA is indeed coded by a verb form followed by the different-subject medial coordinative sufﬁx -te and cannot receive any mood indications. For mood speciﬁcations, it depends on the last ﬁnite verb, which shows an indicative mood marker. (2.24) Tauya, Brahman, Trans-New Guinea 0-tu-e-ĳa / ne fofe-a-te 3sg come-3sg-DS 3sg-give-1/2-IND ‘He came and I gave (it) to him.’ (Mac Donald 1990: 224) The syntactic parallelism, which has traditionally been considered a deﬁning property of coordination, is thus here adopted as a parameter. The reasons determining the deranked status of a verb form may vary (e.g. it may not be able to receive tense, aspect, mood or person markers or it may indicate all these categories by means of special forms that could not occur in independent clauses). Each time a non-parallel construction is described, the speciﬁc features determining the deranked status of the verb will be discussed.

70 Parameters of analysis 2.4

Degree of coding

In the previous sections, the discussion focused on the various sub-types of coordination relations taken into account in this research and on the morphosyntactic parameters used to examine every attested construction expressing them. The main concern of this research is the analysis of how the semantic and the morphosyntactic level interact. The set of coordination sub-relations expressed by each construction, along a form-to-function direction of analysis, will be called semantic domain. From the opposite perspective, i.e. along the function-to-form direction, the degree to which each sub-type of coordination relation is explicitly coded across languages will be called degree of coding of each sub-relation of coordination. Let us now see in detail what these deﬁnitions mean. As already remarked, there is not always a one-to-one correspondence between the set of coordinating constructions available in a language and the various sub-types of coordination they express. On the contrary, coordinating constructions are often polysemous. Before discussing how this polysemy will be dealt with in this research, some examples are given in order to make the issue clear. Examples (2.25), (2.26) and (2.27) show some instances of coordinating markers which are not polysemous, but express just one relation of coordination and may thus be said to be dedicated coordinating markers. Example (2.25) from Georgian illustrates the two markers attested in the language for the expression of alternative. In (2.25a) the alternative relation expressed is choice-aimed, hence the answer to the question must consist of one of the two SoAs. In this case the two alternatives are linked by means of the dedicated marker tu, which could not be used to establish a simple alternative. On the other hand, (2.25b) exempliﬁes the simple alternative construction, in which no choice is requested and an answer like ‘yes’ or ‘ok’ would be perfectly acceptable. Here the linked SoAs are connected by means of the marker an, which could not be used to convey a choice-aimed alternative relation. As can be observed in the translations, the two different Georgian markers are both rendered in English with or.

Degree of coding

71

(2.25) Georgian, Kartvelian c’avidet tu saxlši (a) Xval sk’olaši tomorrow school:LOC go:1pl:OPT COORD home:LOC davrtˇcet? stay:1pl:OPT ‘Tomorrow do we go to school or do we stay at home?’ (M.T., questionnaire) (b) Ginda gaviseirnot ertad, an cˇ ai davliot, want:2sg walk:1pl together COORD tea drink:1pl an k’inos vuq’urot, an . . . rac ginda? want:2sg COORD movie watch:1pl COORD what ‘Would you go for a walk with me, or have a cup of tea, or watch a movie, or . . . whatever?’ (M.T., questionnaire) Example (2.26) from Serbo-Croatian instantiates a construction dedicated to the expression of sequential combination. The general combination marker i is used in (2.26a). This construction is not speciﬁed in terms of temporality and may therefore be used to express both temporal and atemporal combination. The construction attested in (2.26b), instead, is dedicated to the expression of the sequential temporal combination. The two SoAs are located in different subsequent points along the same time axis and are linked by the coordinating marker pa, which could not be used in (2.26a). (2.26) Serbo-Croatian, Slavic, Indo-European mogu više slušati (a) Zauzet sam i ne busy be:1sg and NEG can:1sg anymore listen ‘I’m busy and I can’t listen anymore.’ (Brown and Alt 2004: 70) ˇ (b) Cuo grmljavinu pa sam je hear:PTCP.PST AUX:1sg thunder:ACC COORD AUX:3sg poˇcela kiša begin:PTCP.PST rain ‘I heard a thunder and it started to rain.’ (M.C., questionnaire) Finally, two instances of dedicated contrast markers can be observed in (2.27) from Finnish. Example (2.27a) shows two SoAs linked in a counterexpectative contrast relation, which is coded by means of the marker mutta. This

72 Parameters of analysis coordinating marker can only be found in contexts where an expectation is denied and this denial determines a contrast. It could not be used to express the corrective contrast relation in (2.27b), in which the other marker vaan is employed. Vaan is very similar to the German connective sondern and requires the ﬁrst SoA to be overtly disputed (cf. section 4.1.2). (2.27) Finnish, Balto-Finnic, Uralic hyvin janoinen, mutta en (a) Olen COORD NEG.1sg be.IND.PRS:1sg very thirsty pidä appelsiinimehusta. like:IND.PRS.sg orange.juice ‘I am very thirsty, but I don’t like orange juice.’ (S.O., questionnaire) (b) Peter ei lukemassa ole huoneessaan, Peter NEG be.IND.PRS:3sg study:INF3 :INESS room:INESS vaan leikkimässä puutarhassa. COORD play: INF 3 : INESS garden:INESS ‘ Peter is not studying in his room, but (he is) playing in the garden.’ (S.O., questionnaire) Besides dedicated constructions, there are many polysemous constructions too. In the previous examples, English translations by means of or, and and but provide three instances where one and the same marker is used to express what in other languages requires speciﬁc constructions. Coordinating connectives like the ones attested in English are said to be more general, in that they are used to express more than one relation. Within the same language, general and dedicated markers may co-exist. Example (2.28) shows the case of Italian, where the general marker ma may be used to convey both a counterexpectative and a corrective contrast, whereas the other markers però and bensì are dedicated. Speciﬁcally, però can only be used if the contrast relation results from the denial of an expectation (2.28a), while bensì, which is rarely used nowadays, only encodes the corrective contrast relation.

Degree of coding

73

(2.28) Italian, Romance, Indo-European volentieri ma/però/*bensì non (a) Stasera verrei NEG tonight come:COND:1sg with.pleasure COORD posso can:PRS:1sg ‘I would love to come tonight, but I can’t.’ (b) Giovanni mi ha detto non di raggiungerlo in John 1sg.DAT has told NEG of reach:him in ufﬁcio ma/*però/bensì di aspettarlo qui ofﬁce COORD of wait.for:him here ‘John didn’t tell me to reach him in his ofﬁce, but to wait for him here.’ Another instance of polysemy is attested in Lithuanian (2.29). In this case, the same construction with the marker o is used to express both the oppositive contrast (2.29a) and the corrective one (2.29b), but it is not used in case the contrast is determined by the denial of an expectation, where the marker bèt is employed. The connective o is functionally very similar to the Slavic marker a, in that both convey opposition and correction. (2.29) Lithuanian, Baltic, Indo-European Petras miega o (a) Aš dirbu I work:PRS.1sg COORD Peter sleep:PRS.3sg ‘I work and Peter sleeps.’ (L.R., questionnaire) (b) Peter nesimoko savo kambaryje o Peter NEG:study:PRS.3sg his room:LOC COORD žaidžia sode play:PRS.3sg garden:LOC ‘Peter is not studying in his room but he’s playing in the garden.’ (V.Ž., questionnaire) As can be observed in the translation, the English equivalent for the sentence in (2.29a) is the combination construction with and, because a speciﬁc marker like whereas would result in being much stronger than the Lithuanian o. It will be argued in section 4.1.2 that oppositive contrast and atemporal combination

74 Parameters of analysis are not easy to distinguish. It is more a matter of degree and emphasis either on the combination itself or on the symmetric opposition of the SoAs. The Lithuanian connective o is indeed used to express the combination of somewhat discontinuous SoAs (cf. Kotcheva 2005, Givón 1990: 849, see discussion in section 3.2), when a change in the participants or in the actions occurs. Two discontinuous SoAs may then simply stand in a combination relation or the speaker may want to point out an opposition between them. More will be said about the conceptual proximity of discontinuous atemporal combination and opposition in the course of the work, especially in chapters 4 and 6. For now, it sufﬁces to highlight the polysemy of the Lithuanian marker, which also includes the expression of the corrective relation. Based on the semantic domain of the attested coordinating constructions, it is thus possible to distinguish between dedicated constructions, specialized in the expression of only one sub-relation of coordination, and more general ones. The concept of speciﬁc marker is absolute, since it applies to those cases where a given construction can be used only in one speciﬁc situation. On the contrary, the concept of general marker is scalar, since the degree of generality of the markers may vary, depending on the number of coordination relations that they cover. The composite semantic domain of general markers may be treated in different ways and an enormous amount of literature exists on this topic. Haspelmath (2003) uses the term multifunctionality to neutrally talk about linguistic entities which express more than one function, without committing himself to rigorous claims of monosemy, polysemy or homonymy. Monosemist positions claim that, in case of multifunctionality, a unique abstract meaning should be identiﬁed and the various functions should be considered as different uses which originate in relation to the context. A monosemist position is held by Gil (2004), who labels the vague abstract meaning as ‘macrofunction’ and analyzes the various meanings as context-driven speciﬁcations (cf. also Coseriu’s discussion on the very general meaning of mit, Coseriu 1970). On the other hand, the polysemist and homonymist approaches recognize the various functions as conventional meanings and not only as context derived notions, but explain them in two different ways. The polysemists identify a motivation behind the association of various functions into one grammatical item, which may be due to historical, pragmatic or cognitive reasons. The homonymists, instead, consider the different functions as belonging to different lexemes, thus establishing a one-to-one relationship between the functions and the ways they are expressed.

Degree of coding

75

In his study on subordinators, Kortmann (1997: 89) uses the term ‘polyfunctionality’ to refer to the range of adverbial functions shown by each subordinator and to identify its primary and secondary meanings. He thus identiﬁes particular criteria to discriminate between the different senses a subordinator may have, like divergent syntactic constraints and semantic uniqueness, and further distinguishes between primary and secondary meanings on the basis of whether or not a given reading will necessarily occur in a speciﬁc context (Kortmann 1997: 94). However, a polysemous analysis in terms of polyfunctionality sounds a bit forced in cases like the English connective and. Would we really want to say that and has two different (though related) functions, namely sequential and non-sequential combination, and that one of them is primary with respect to the other? It sounds more promising to consider such cases from a monosemist point of view, as markers which simply encode the general combination relation, without any further speciﬁcations. The same arguments are valid for the alternative connective or, which encodes the alternative relation, without specifying its aim. Yet, one should not exclude that cases of polysemy may also occur, for instance through an historical semantic shift which widens the functional scope of a given connective by including conceptually neighboring functions. The term ‘semantic domain’ wants to be neutral with respect to the possible hierarchical organizations of the various functions fulﬁlled and leaves the possibility to choose in each case whether we are dealing with a macrofunction or with a set of related but different senses. A semantic domain simply delimits the set of coordinating functions that a coordinating marker or construction may have. The range of functions other than the coordinating ones will not be considered.20 As will be exempliﬁed in the next chapter for instance, a semantically vague sufﬁx like the Turkish narrative converb -Ip, which only codes the successive advancement of a narration (Nedjalkov 1995: 109), will be considered dedicated, because it may express a sequential combination but could not be used for simultaneous or atemporal combination. The dedicated vs. general characterization of a construction in this study only refers to the coordination relations it may express, without taking into account possible functions other than the coordinating ones, since the aim of this parameter is to depict the patterns of multi/macro-functionality that are internal to coordination and are thus established between the (sub-)relations of combination, contrast and alternative.

76 Parameters of analysis As remarked at the beginning of this section, the interaction between semantic and morphosyntactic level will be examined from two symmetric perspectives. On the one hand, the attested constructions are examined with respect to their semantic domain, on the other hand, each sub-type of coordination relation is examined with respect to the degree to which it is explicitly coded across languages. The degree of coding identiﬁes the extent to which a given relation is overtly coded and the extent to which it is inferred from the semantics of the SoAs involved. Namely, following the terminology proposed by Prandi (2004), a given coordination relation may be punctually coded, if it is expressed by a dedicated construction, or it may be undercoded, if it is expressed by a general construction used to express other relations too, which leaves further speciﬁcations to inferential enrichment (Prandi 2004: 297-98). The more speciﬁc is the encoded relation, the less is left to inference. In case of juxtaposition, i.e. absence of coding, Prandi (2004: 299-302) talks about inferential bridging, meaning the process whereby the conceptual relation existing between two SoAs is completely built through inference. Therefore, asserting that the Italian marker ma is general means that its semantic domain consists of at least two sub-types of contrast, the corrective and the counterexpectative one. The relation of corrective contrast is thus undercoded when it is expressed by ma in Italian, because part of its expression is provided by the connective and part comes from inferential enrichment. On the contrary, the German connective aber is dedicated to the expression of counterexpectative contrast and sondern is dedicated to the expression of the corrective sub-type. In German, the two sub-relations are thus punctually coded by dedicated constructions. The comparison of the various semantic domains will be better captured by means of semantic maps. In section 6.4.3 a special type of bidimensional representation will be proposed, where semantic domain and degree of coding will be jointly delineated for the coordination relations examined. Anyway, the interaction between morphosyntactic and semantic level will be constantly at issue throughout the analysis of the attested coordinating constructions.

Cross-linguistic variation: an overview

2.5

77

Cross-linguistic variation: an overview

As argued in section 1.1.2, the functional-typological approach identiﬁes two types of constraints on cross-linguistic variation: unrestricted and implicational universals. In this research, the study of the attested coordinating constructions only reveals patterns of the second type, that is, analyzable in terms of implications. Moreover, as already argued in section 1.2, no universality ambition is pursued here, because the two samples are not balanced and data have a strong displacement on Europe. However, with the due amount of caution, the cross-linguistic variation attested in the two samples gives a sufﬁciently rich image of the coordinating constructions possibly found across languages, for two major reasons. First, a good number of different families is represented, both within and outside Europe. Secondly, some of the phenomena taken into account, such as the degree of coding and the semantic domain of the attested constructions, show a great variation even among geographically close and genetically related languages. Therefore, it is possible to examine a diversiﬁed set of coordinating constructions even within genetically and geographically related languages. Data has no statistical value, of course, but allows to establish signiﬁcant generalizations. It is possible to identify a number of constraints on the cross-linguistic variation attested in the coding of coordination relations that are sufﬁciently well supported to postulate regular interconnections between particular coding phenomena and particular functional motivations. The exam of the attested coordinating constructions gives rise to two types of generalizations: generalizations identifying implicational patterns of variation, on the one hand, and generalizations describing tendencies of variation, on the other hand. The generalizations identifying implicational patterns of variation of the form ‘A → B’ will be considered well supported if the ratio between the two asymmetric types ‘B and not A’ and ‘A and not B’ is at least ﬁve to zero, that is, if no asymmetric type ‘A and not B’ is attested and if the attested asymmetric type ‘B and not A’ is supported by at least ﬁve languages. Given two parameters and the four possible combinations of their values, an implicational pattern identiﬁes three combinations as attested and excludes the fourth one. The implicational patterns of variation may be stated in terms of simple implications or in terms of implicational hierarchies (see section 1.1.2). In the latter case, in order to have a well supported hierarchy, the minimum number

78 Parameters of analysis of languages for the attested asymmetric type is three, if no counterexample is attested, while it remains ﬁve in case some counterexample is present. Furthermore, an implicational pattern of variation may as well be formulated in terms of a conceptual space. There are also generalizations that identify regular patterns of coding, but these patterns are supported by a very small number of languages and the asymmetric attested type is represented by less than ﬁve languages. In such situations, even though no systematic schemas are found, the various values of the parameters under exam are not equally distributed and attested, but some coding strategies are preferred to others and this may not be random. Instances of implicational patterns which are supported by a too small number of cases are the patterns identiﬁed with respect to the coding of two conﬂicting or alternative SoAs (see chapters 4 and 5). In these cases, one type is widely attested, namely the use of syntactically parallel constructions, two types are attested by a small number of languages, namely the use of non-parallel constructions for all relations and the asymmetric use of such constructions just for one relation, and the fourth type is not attested at all. Such implicational generalizations, even though they are not well supported, may still depict interesting distributions of data. The analysis of speciﬁc parameters may also reveal no pattern at all, as is the case for the coding of combination relations, that is, the attested constructions may show a tendency for particular values, but no implicational distribution. Both in weak implicational patterns and in non-implicational distributions, the generalizations are more of a descriptive kind, since they have a weak (or absent) predictive force and mainly provide a picture of the attested cross-linguistic variation. For these reasons, weak implicational patterns and non-implicational distributions will be referred to as tendencies. The coding of the three coordination relations is not equally sensible to the various parameters of analysis. As will become clear in the next chapters, some parameters reveal implicational patterns of variation in the coding of certain relations, but only non-implicational tendencies in the coding of other relations. This means that certain aspects of coding are cross-linguistically more salient in the expression of particular conceptual relations. The presence of overt markers, for instance, will be particularly relevant in the analysis of constructions expressing contrast and alternative relations, while the parameter of syntactic parallelism will be especially salient in the analysis of constructions that express combination.

A>BB;8=6D8BC82 E0A80C8>= 0= >E4AE84F

"B H<9 B9LH HC=BH@M 5B5@MN98 5B8 H<9 7C8=B; C: 957< C: H<9 H38=6 >5 C74 BD1A4;0C8>=B K=@@ 69 5B5@MN98 @CC?=B; :CF GMB89H=7 JG 5GMB89H=7 7CBGHFI7 H=CBG 5B8 ACFD38=6 >5 C74 *>B H5?=B; =BHC 577CIBH H<9 GMBH57H=7 D5F5@@9@=GA C: H<9 5HH9GH98 7CBGHFI7H=CBG G97H=CBG 5B8 -<9 =BH9F57H=CB 69HK99B G9 A5BH=7 5B8 ACFD=0; 4G?;0=0C8>=B IB89F@M=B; H<9 5HH9GH98 7CBGHF5=BHG CB 7FCGG@=B;I=GH=7 J5F=5H=CB G97H=CBG 5B8 5B8 K=H< H<9 DF9G9BH5H=CB C: H<9 F9@9J5BH 85H5 G97H=CBG 5B8

Chapter 3 Combination relations The object of this chapter is the analysis of the cross-linguistic coding of combination relations. Before discussing the attested constructions, a deﬁnition of the notion of combination relation is provided in section 3.1. The combination types identiﬁed on the basis of the semantic parameter of temporality will then be deﬁned and exempliﬁed in section 3.1.2. The attested constructions will ﬁrst be analyzed with respect to the coding of the relation, highlighting the presence of overt coordinating markers and their morphophonological complexity. This parameter will point out some cross-linguistic tendencies, but no implicational pattern (section 3.2). The coding of the SoAs will instead reveal a neat implicational pattern of coding, which will be the object of section 3.3. The functional motivations underlying the non-random variation in the coding of combination will be discussed in section 3.4, in which the principle of economy will be argued to play a decisive role in the parallel vs. nonparallel coding of two combined SoAs. Data supporting the outlined typology of combination constructions will be presented in section 3.5, in which information about all the parameters examined is provided.

3.1 3.1.1

The notion of combination relation Combination as cooccurrence

In section 1.3.5, the three coordination relations have been characterized with respect to two different dimensions. Combination and contrast have been located on a cooccurrence dimension, where the linked SoAs are jointly asserted or elicited as cooccurring. Alternative, on the other hand, has been located on a non-cooccurrence dimension, where the SoAs constitute replaceable non-cooccurring possibilities. In Fig. 1 this conceptual opposition is represented along a vertical and a horizontal dimension. Cooccurring SoAs are highlighted by the dashed box and are located one after the other. Conversely, non-cooccurring SoAs are located one above the other, in order to make their status of replaceable pos-

The notion of combination relation

81

sibilities clear. The English connectives and, but and or between the SoAs signal the relations of combination, contrast and alternative that may be established between cooccurrent and non-cooccurrent SoAs respectively.

Figure 1: Cooccurrence vs. non-cooccurrence opposition

Two SoAs linked on the horizontal dimension in a combination or a contrast relation are assumed to cooccur. They are therefore to be jointly asserted or elicited (in a question or an order), since the speaker wants the hearer to assume that the linked SoAs are jointly real or not real, present or not present. The two SoAs may be simply considered in their cooccurrence or they may be considered under those respects that differentiate them, thus regarding their cooccurrence as conﬂicting. In the ﬁrst case, they are said to stand in a combination relation, whereas in the latter they are said to stand in a contrast relation. The notion of contrast will be discussed in detail in section 4.1 and the notion of alternative will be the object of section 5.1. Let us now deﬁne what we mean by combination relation. The two SoAs connected by a combination relation in Fig. 1 originate the following sentence: (3.1) He took the car and got stuck in the trafﬁc. where the only conceptual link connecting the SoAs establishes their cooccurrence. Any further speciﬁcation that may arise from combination is the outcome of inferential enrichment. The combination relation can indeed be

><18=0C8>= A4;0C8>=B 89:=B98 5G H<9 65G=7 7CC77IFF9B79 F9@5H=CB 9L=GH=B; 69HK99B 9@9A9BHG C: 5 A95B=B;:I@ 7CAD@9L KC=B=B; CD9F5H=CB ,=A=@5F 5F;IA9BHG 5F9 GI;;9GH98 6M =? K C78B <>A= 8=6b a4 C>>: C74 20Ab 5B8 a74 6>C BCD2: 8= C74 CA052b 5F9 =BGH5BH=5H=CBG C: H<9 7CB79DHI5@ 9BH=HM aH<=B;G H<5H <9 8=8 H<=G ACFB=B;b K<=7< 7CFF9GDCB8G HC H<9 F95GCB K<M H<9 HKC ,CG 5F9 D9FH=B9BH@M 7CA6=B98 -<9 DF9G9B79 C: 5 7CAACB =BH9;F5HCF 5@GC 7<5F57H9F=N9G 7CBHF5GH 5B8 5@H9FB5H=J9 G=B79 5BM H=A9 HKC ,CG 5F9 F9@5H98 H<=G <5DD9BG 6975IG9 H<9F9 =G 5 7CAACB HCD=7 K=H< F9 GD97H HC K<=7< H<9M 5F9 D9FH=B9BH !CK9J9F 5 G=AD@9 7CA6=B5H=CB 8C9G BCH :IFH<9F GD97=:M H<9 F95GCB :CF 9GH56@=G<=B; H<9 F9@5H=CB -<9 D9FH=B9B79 89H9F A=B98 6M H<9 7CAACB =BH9;F5HCF =G =HG9@: H<9 75IG9 5B8 H<9 5=A :CF 7CA6=B=B; H<9 ,CG ,=B79 =H =G H<9 ACGH 65G=7 5B8 G9A5BH=75@@M IBGD97=:=98 7CC77IFF9B79 F9@5 H=CB 7CA6=B5H=CB A5M 69 G99B 5G H<9 DFC79GG H
The notion of combination relation

83

be speciﬁed in various ways. As will be argued in section 6.4.3, combination constitutes the origin for a cooccurrence dimension along which all the other relations entailing the co-existence of the SoAs develop. To sum up, a combination relation is established between functionally parallel SoAs on the basis of a common integrator, without further speciﬁcations on the nature of the association.

3.1.2

Combination types

The combination relation can be further subdivided into two types on the basis of the parameter of temporality. Before discussing each type by providing examples, let us ﬁrst examine how combination interacts with the other two parameters. The combination relation is essentially ‘vague’ with respect to the presence of a conﬂict between the SoAs, as argued by Payne (1985: 4). A conﬂict may be inferred from the context, but the relation as such does not entail it and can be thus classiﬁed as a non-conﬂicting one. As will be made clear throughout the work, combination constructions may sometimes be used also to convey conﬂicting situations because, even if they do not code it, they simply do not forbid to infer a conﬂict within the combined SoAs. As for the parameter of aim, the intention of a speaker who establishes a combination between two SoAs is always to assert or elicit information about the whole block of combined SoAs (example (3.2)), since the relation itself implies that they be jointly at issue. It would simply make no sense to establish a combination between two SoAs in order to give or elicit information about just one of them. (3.2) Mary is sick and Peter had a setback. Is it the case that Mary is sick and Peter had a setback? The location of the linked SoAs along the time axis, instead, makes an overall distinction between temporal and atemporal combination, that has been recognized by a number of scholars (cf. Longacre 1985, Haiman 1985, Halliday and Hasan 1976 and Langacker 1987a). Let us now provide more details and examples for the two types of combination.

84 Combination relations 3.1.2.1 Temporal combination Temporal combination is characterized by the fact that the SoAs are located on a temporal axis. Temporal combination can be further subdivided in sequential and simultaneous combination. Sequential combination is characterized by the location of the linked SoAs along the same time axis at successive points, so that they are interconnected as part of the same overall sequence of events. This type may be lexicalized by a dedicated connective in some languages, or it may be inferred from a simple combination. Simultaneous combination is instead characterized by the temporal overlap of the SoAs, which are located at the same point along the time axis. As pointed out by Haiman (1985: 100), two sequential SoAs may be presented in a different order than the one in which they occur only if the temporal relation is overtly coded and one of the two SoAs is subordinate to the other (as in ‘He came to me, after you met him’). In this case, the functional parallelism fails and only one SoA is at issue in the speech act. Longacre calls the sequential combination succession (Longacre 1985: 244), deﬁning it as an underlying ‘and then’ relation in which the SoAs are temporally ordered. The main consequence of this order along the time axis is a linear asymmetry. In other words, if the order in which the SoAs are presented is reversed, the meaning of the combination changes, as can be observed in examples (3.3) and (3.4) (‘asymmetric and’, Lakoff 1971: 126). (3.3) The police came into the room and everyone swallowed their cigarettes. (Lakoff 1971: 127) (3.4) Everyone swallowed their cigarettes and the police came into the room. (Lakoff 1971: 127) Simultaneous combination (Longacre 1985: 243 calls it overlap), on the contrary, is characterized by the absence of a reciprocal order in which the SoAs occur, since they happen to occupy the same point on the time axis. Two examples of simultaneous combination are provided in (3.5) and (3.6): (3.5) He is dancing and clapping his hands. (3.6) At noon, I was eating at the park and Mary was sitting next to me. Simultaneous combination is often established between durative events which cover the same lapse of time. However, if the point along the time axis where

The notion of combination relation

85

the SoAs cooccur is not made explicit, the combination may easily be interpreted as atemporal, i.e. a combination in which the linked SoAs are jointly considered as cooccurring within a certain lapse of time, but the order in which they occur or their possible simultaneity are not at issue. 3.1.2.2 Atemporal combination In case the location of combined SoAs along the time axis is simply not relevant to the combination itself, the relation is said to be an atemporal one (Longacre 1985: 241 calls it coupling or non-temporal underlying ‘and’ relation). An atemporal combination may either combine SoAs outside the time axis, establishing a relation that is expected to be always valid, or it may combine two SoAs within the time axis regardless of their respective location. An instance of atemporal combination established outside the time axis is provided in (3.7), in which the block of SoAs linked through the combination is asserted as occurring in general, without relating it to any temporal reference. Examples (3.8) and (3.9) instead provide two instances of atemporal combination established within the time axis, regardless of the speciﬁc order in which the SoAs occur. In (3.9) for instance, it is not relevant whether the speaker took the shower before Paul checked his e-mail or the two actions took place at the same time. What is at issue is simply the cooccurrence of the SoAs within the same lapse of time. (3.7) Doctors are rich and lawyers marry pretty girls. (Lakoff 1971: 129) (3.8) Mary is sick and Peter had a setback (that’s why they are not here). (3.9) When we came back home, I took a shower and Paul checked his e-mail. Lakoff labels the instances of combination where the SoAs are not located in a successive order ‘symmetric and’ (Lakoff 1971: 115-126), without further distinguishing between simultaneous and atemporal combination. In fact, in many cases it is not easy to decide whether the two SoAs are combined in an overlapping or in a atemporal sense. I suggest this is due to the two senses being intimately similar, because both simultaneous SoAs and atemporally combined SoAs do not have to be

86 Combination relations presented in a special order. Simultaneous SoAs simply do not have an intrinsic order, whereas two SoAs combined in an atemporal combination are linked in such a way that their respective order is irrelevant. As will become clear in the next section and in the course of the work, simultaneous and atemporal combination may always be expressed by the same construction, and this is plausibly due to their similar internal structure. Moreover, when atemporal combination is established between SoAs having different participants, it is normally used to convey a comparison of the type ‘on the one hand. . . on the other hand. . . ’. This property of atemporal combination makes it very similar to a type of contrast, namely the oppositive contrast relation (see deﬁnition in section 4.1.2), which is characterized by the symmetric opposition of somewhat antonymic SoAs. As a result of this conceptual closeness, it is often not simple to clearly discriminate between the two. In many languages, like English for instance, the same construction may be ambiguously read as both an opposition and an atemporal combination, like in (3.10). (3.10) John works in a store and Mary is still a student. In (3.10), the speaker may simply want to say what John and Mary do for living or he may want to highlight an opposition between the fact that John has a salary whereas Mary does not gain any money, because she is still a student. The contrast is not overtly signalled in that case, but is rather inferred from the parallelism of the SoAs and from the context. If the speaker wanted to make the opposition explicit, the SoAs would be linked by a connective like while or whereas. More on this topic will be said in the next section and in chapter 4.

3.2

The coding of combination relations

3.2.1 Presence of overt coordinating markers All data regarding the presence of overt combination markers is reported in Tables 6 and 7, with information from the European sample and from the Comparison sample respectively. The analysis of these data takes simultaneously into account the presence vs. absence of overt coordinating markers and the semantic domain of the attested markers.

The coding of combination relations

87

No implicational pattern is attested in the coding of combination relations. However, it can be observed that all the languages examined may express simultaneous combination and atemporal combination by means of the same general construction. As already stated in the previous section, the conceptual closeness between the two relations is plausibly the reason for their cross-linguistic joint coding. The only dedicated markers attested are used to express sequential combination, whereas there are no dedicated connectives for atemporal combination or for simultaneous combination. To the contrary, there are general markers for the expression of just atemporal combination and opposition, especially in the Slavic languages. The distribution of the attested types is different in and outside Europe. All European languages show at least one general marker for all combination types (equivalent to English and), and in some cases also dedicated markers for sequential combination. Furthermore some European languages may express certain instances of atemporal combination (i.e. those in which the linked SoAs are discontinuous) by means of a general construction which is used also to express opposition, but which does not express temporal combination, either sequential or simultaneous. In the Comparison sample, on the contrary, there are both languages with a general marker for all combination types and languages without a general marker for all combination types. In particular, languages without an overall general combination marker may use asyndetic constructions or different markers for sequential combination on the one hand and non-sequential (atemporal and simultaneous) combination on the other hand. Let us now examine each attested language type in detail. Languages with an overt general marker for the temporal and atemporal combination. As already stated, all European languages have at least one general marker used for both temporal and atemporal combination. However, such general connectives are also attested in the Comparison sample. Examples of languages with one general combination marker and no dedicated marker are English (and), Italian (e), French (et), Albanian (e/dhe) in Europe and Hocak ˛ (ánaga), ˛ NànáfwˆE (kp´Ekún), West Greenlandic (=lu) outside Europe (other examples can be observed in Tables 6 and 7 in this chapter, and in Tables 53 and 54 in the Appendix). In these languages, the temporal or atemporal nature of the relation is reconstructed by means of inferential enrichment. This means that it is the semantic content of the two SoAs as well

88 Combination relations as the context of communication which give rise to a temporal or atemporal reading. There are three further types in which the use of a general marker for all combination relations co-exists with other markers. Some languages have a general marker used for all combination relations and a dedicated sequential one, like Turkish and Lezgian (see examples (1.8) and (1.34)). In both languages the marker dedicated to sequential combination is part of the verbal paradigm and occurs in syntactically non-parallel constructions. The verb forms like the ones attested in Lezgian and Turkish (3.11a) are traditionally referred to as narrative converbs.21 Furthermore, the dedicated sequential marker co-exists with a general combination marker which occurs in a syntactically parallel construction (=ni, wa in Lezgian, =dA or ve in Turkish, cf. (3.11b)). (3.11) Turkish, Altaic gid-ip ev-e dön-dü (a) Hasan i¸s-e Hasan work-DAT go-and home-DAT come-PST ‘Hasan went to work and came back home’ (Kornﬁlt 1997: 110) (b) Hasan i¸s-in-e git-ti, Ali ev-in-e Hasan work-3sg-DAT go-PST Ali house-3sg-DAT dön-dü ve ben park-ta kal-dı-m return-PST and I park-LOC stay-PST-1sg ‘Hasan went to work, Ali returned home and I stayed in the park.’ (Kornﬁlt 1997: 109) The general combination marker may also be accompanied by another marker which does not convey a temporal relation, but only atemporal combination. The languages showing this conﬁguration are all part of the European sample and employ this marker also to express the oppositive contrast relation. Example (3.12) shows an instance of this construction type from Serbo-Croatian. (3.12) Serbo-Croatian, Slavic, Indo-European (a) Ivan putuje a i Marija putuje. Ivan travel:3sg and too Mary travel:3sg ‘Ivan is traveling and(/*while) Marija is traveling too.’ (Brown and Alt 2004: 70)

The coding of combination relations

89

Petar spava (b) Ja radim a I work:1sg COORD Peter sleep:3sg ‘I work and(/while) Peter sleeps.’ (M.C., questionnaire) In this example the atemporal combination of the two SoAs in (3.12a) is not conveyed by i, which is the general combination marker attested in SerboCroatian (see example (2.26)), but by means of the connective a. This connective is usually characterized in terms of weak contrast, since it is commonly employed to convey oppositive relations like the one in (3.12b) and is only used when the exact temporality of the linked SoAs is irrelevant. Descriptive grammars usually gloss these markers with an and/but label, which leaves their speciﬁc function quite obscure. Yet, in (3.12a) the cooccurrence of a with i, which in this context means ‘also, too’ and thus denotes an addition without opposition, suggests that its semantic domain is wider than the simple oppositive contrast, and encompasses an atemporal additive meaning, too (cf. Brown and Alt 2004: 70). The connective a is present in other Slavic languages, like Russian, Polish, Belorussian and Bulgarian. Roughly equivalent markers are Ukrainian ta, Lithuanian o and Chechen tq’a. The additive value of a in Russian is pointed out by Malchukov (2004: 183), who exempliﬁes it with the following sentence: (3.13) Russian, Slavic, Indo-European Vremja uxodit bystro, a s nim uxodjat ljudi time pass:3sg quickly and with it pass:3pl people ‘Time passes quickly and(/*while) with it people pass (away).’ (adapted from Malchukov 2004: 183) These markers are not general combination ones, but they are not dedicated either, since they are used to convey at least two types of coordination relation: atemporal combination and opposition. What the two relations in (3.12a) and (3.13) have in common with (3.12b) is an interruption of continuity between the two SoAs, determined by a change of participants or actions (cf. Kotcheva 2005, Givón 1990: 849). Markers like a are thus used to link discontinuous SoAs,22 whose discontinuity can be due to a switch of subjects (referential contrast), to a switch of verbs, a switch of the object, or a switch of subject, verb and object (topic contrast) (Givón 1990: 850-51).

><18=0C8>= A4;0C8>=B -<9G9 A5F?9FG K=@@ 69 :IFH<9F 8=G7IGG98 =B G97H=CB 9GD97=5@@M K=H< F9 GD97H HC H<9 9LDF9GG=CB C: CDDCG=H=J9 7CBHF5GH 5B8 =HG 8=GH=B7H=CB :FCA G=AD@9 5H9ADCF5@ 7CA6=B5H=CB 77CF8=B; HC H<9 89:=B=H=CB C: 7CA6=B5H=CB =B B9; 5H=J9 H9FAG 5G 5 7CC77IFF9B79 F9@5H=CB K<=7< =G BCH :IFH<9F GD97=:=98 9J9FM 7CA6=B5H=CB =B K<=7< HKC 8=G7CBH=BICIG ,CG 5F9 @=B?98 F8C7>DC 4G?;828C;H 8=3820C8=6 C74 4G8BC4=24 >5 0 2>=82C =G 7CBG=89F98 5B =BGH5B79 C: 5H9ADCF5@ 7CA6=B5H=CB BM :IFH<9F CDDCG=H=J9 F958=B; 89F=J9G :FCA =B:9F9BH=5@ 9BF=7< A9BH CBJ9FG9@M =: H<9 7CB:@=7H =G GCA9= CA0382C8>= b H<9F9@5H=CB =G 7CBG=89F98 5B =BGH5B79 C: CDDCG=H=J9 7CBHF5GH (IHG=89 IFCD9 ;9B9F5@ A5F?9FG :CF 8=G7CBH=BICIG 5H9ADCF5@ 7CA6=B5 H=CB 5B8 CDDCG=H=CB 5F9 5HH9GH98 CB@M K<9B BC ;9B9F5@ 7CA6=B5H=CB A5F?9F =G DF9G9BH 9:CF9 8=G7IGG=B; H<9 @5B;I5;9G @57?=B; 5 ;9B9F5@ A5F?9F :CF H9A DCF5@ 5B8 5H9ADCF5@ 7CA6=B5H=CB @9H IG BCK 7CBG=89F HKC 75G9G =B K<=7< H<9 ;9B9F5@ 7CA6=B5H=CB A5F?9F 7C9L=GHG K=H< 6CH< 5 ;9B9F5@ A5F?9F :CF 5H9A DCF5@ 7CA6=B5H=CB 5B8 CDDCG=H=CB 5B8 5 898=75H98 A5F?9F 9LDF9GG=B; CB@M G9EI9BH=5@ 7CA6=B5H=CB -<9G9 HKC @5B;I5;9G 5F9 ,9F6CFC5H=5B 5B8 I@;5F =5B L5AD@9G C: ;9B9F5@ A5F?9FG :CF 5H9ADCF5@ 7CA6=B5H=CB 5B8 CDDCG=H=CB <5J9 5@F958M 699B ;=J9B 56CJ9 5B8 5B =BGH5B79 C: 898=75H98 A5F?9F :CF H<9 G9EI9BH=5@ 7CA6=B5H=CB <58 699B G B0< 94 6A<;90E8=D ?0 <95F)-) ),- .1 G; H24;0 :8O0 69;=B)-) ),- F5=B a" <95F8 5 H
The coding of combination relations

91

sequential markers are attested both with and without a general marker for atemporal combination and opposition, but almost always when a general combination marker is absent. The only language having both a general combination marker (wa) and a dedicated sequential marker (fa) is Arabic.23 However, the presence of a dedicated sequential marker seems less frequent than the presence of a general marker for discontinuous SoAs linked in an atemporal combination or in an oppositive contrast. In the European sample, 10 languages out of 37 have such a general marker and only two of them (Serbo-Croatian and Bulgarian) also show a dedicated sequential connective. On the other hand, only two languages out of 37 (Turkish and Lezgian) use a dedicated construction expressing sequential combination and no general marker for atemporal combination and opposition. Furthermore, in both cases the sequential construction is a narrative converbal form. As already said, all European languages have at least one general marker for all combination types, hence the four remaining language types, which lack such a general marker, are attested only in the Comparison sample. Let us now examine how the combination relation is expressed in those languages that lack a general connective equivalent to English and. Languages with no overt general marker for temporal and atemporal combination. The use of simple juxtaposition for the expression of temporal and atemporal combination is especially widespread in languages with a mostly spoken tradition. As Mithun (1988) points out, asyndetic constructions may render the conceptual closeness vs. separateness of the SoAs by means of different intonation patterns. The absence of a pause usually denotes two SoAs which take part in the same overall sequence, whereas the presence of a pause means a separation between the events. As can be observed in Table 7, asyndesis is a frequent alternative option also for those languages which have overt coordinating markers. Many languages seem to have some structural device signalling the combination of two SoAs, but this device is not obligatory and in some cases it is at the very beginning of a grammaticalization process, or consists of clause chaining lacking any structural differentiation between coordinating and subordinating constructions. Maricopa, for instance, has no overt marker coding a combination relation, either general or dedicated, and this is what made Gil title his paper ‘Aristotle goes to Arizona and ﬁnds a language without And’ (Gil 1991), highlighting the non-universality of connectives equivalent to and. In Mari-

><18=0C8>= A4;0C8>=B 7CD5 H<9 7CA6=B5H=CB C: HKC ,CG =G 9LDF9GG98 6M A95BG C: H<9 >ILH5DCG=H=CB C: HKC 7@5IG9G H<9 :=FGH C: K<=7< 7CBH5=BG 5 89F5B?98 J9F6 :CFA GI::=L98 K=H< 5 GK=H7<F9:9F9B79 A5F?9F 5G 75B 69 C6G9FJ98 =B 9L5AD@9 &5F=7CD5 2IA5B !C?5B 5 :054 B8B7< ?0BC4; <0B7: 7C::99 8F=B?., D=9 95H.F95@ a-<9M 8F5B? 7C::99 5B8 H<9M 5H9 D=9 b -<9M -<9M CF8CB =H:F8 C00E: C?DHB7 6 E88B7 FC7?,# IB7@9 D9@H,, ?=@@)+ a FC7? <=H AM IB7@9 5B8 ?=@@98 <=A b CF8CB -<9 G9BH9B79 =B 5 9LDF9GG9G 5B 5H9ADCF5@ 7CA6=B5H=CB K<=@9 H<9 G9B H9B79 =B 6 7CBJ9MG 5 G9EI9B79 C: GI779GG=J9 ,CG "B 6CH< 75G9G H<9 J9F6 :CFA GI::=L98 6M H<9 GK=H7<F9:9F9B79 A5F?9F 75BBCH 69 GD97=:=98 :CF H9BG9 5B8 ACC8 89D9B8=B; CB H<9 :=B5@ J9F6 :CF H<9 9LDF9GG=CB C: H<9G9 75H 9;CF=9G "B H<9 DF9J=CIG 7<5DH9F HKC 9L5AD@9G :FCA -5IM5 K9F9 G97HG C: H<9 @=B?98 7@5IG9G 5B8 8C BCH G5M 5BMH<=B; 56CIH H<9 HMD9 C: F9@5H=CB 69=B; 9GH56 @=G<98 -<9F9:CF9 H<9M 75BBCH 69 7CBG=89F98 CJ9FH 7CCF8=B5H=B; A5F?9FG 5B8 H<9 7CA6=B5H=CB F9@5H=CB =G G=AD@M =B:9FF98 6M H<9=F @=B?5;9 8=::9F9BH G=HI5H=CB =G 5HH9GH98 =B $=G= K<9F9 5 ;9B9F5@ 7CA6=B5H=CB A5F?9F =G @=?9K=G9 7CAD@9H9@M 56G9BH 6IH BCH 9J9B 5 ACFDILH5DCG98 5B8 B9=H<9F F979=J9G A5F?9FG CF ACFDILH5DCG=H=CB CF H<9M A5M <5J9 HKC 8=::9F9BH A5F?9FG CB9 :CF G9EI9B H=5@ 7CA6=B5H=CB 5B8 H<9 CH<9F CB9 :CF BCBG9EI9BH=5@ 7CA6=B5H=CB 5H9ADCF5@

The coding of combination relations

93

and simultaneous). In case only one overt marker is present, both possibilities are attested, with no marker expressing the sequential combination and an overt marker for the other two combination types, or with an overt sequential marker and no marker for atemporal and simultaneous combination. Hdi and Lango, for instance, use simple juxtaposition as the main strategy for the expression of both temporal and atemporal relations, but they also have the possibility to employ an overt connective for the expression of a sequential combination. In both cases, the connectives derive from verbs and are not fully grammaticalized. In Lango tˆE, roughly meaning ‘and then’, is always conjugated in the habitual and takes inﬁnitive complements (Noonan 1992: 193). In Hdi the verb lá, meaning ‘depart, go’, is used in its nominalized form, marking person agreement with pronouns belonging to the possessive set, to indicate separation and temporal sequentiality of the SoAs it links. As Frajzyngier and Shay (2002: 428-31) argue, the verb lá has entered a process of grammaticalization, whereby it has developed a purely coordinating sequential function which still synchronically co-exists with its original lexical meaning. In the intermediate stage, during which the combination function has arisen, the verb ‘to go’ worked as a bridge between two events, in which a subject had to move to another location in order to continue its action or to begin a new one there. Example (3.16) shows the original lexical meaning ‘to go’ together with the new coordinating function of lá. The original meaning can be observed in lá-b-ì ’went away’, whereas the following lá-ghà has a coordinating function, which is proved by the fact that its subject (Hyena) that does not move.24 (3.16) Hdi, Chadic, Afro-Asiatic krì kà lá-b-ì lá-ghà pákáw ghúvì mbàâ ká then COMP dog SEQ go-out-REF go-D : PVG hyena kà mná-n-tá krì SEQ tell-3sg- REF dog ‘Then dog went away and Hyena said to Dog . . . ’ (Frajzyngier and Shay 2002: 429-30) On the contrary, Tuvaluan, Koromfe and Mosetén have only an overt connective used for atemporal and simultaneous combination, whereas sequential combination is not explicitly signalled. Tuvaluan and Koromfe show a

94 Combination relations general marker which is used also to express contrast relations. In these languages, sequential combination is indeed opposed to all the other cooccurrence relations, in that two SoAs linked within a temporal sequence are simply juxtaposed, while any kind of absence or interruption of sequentiality is overtly signalled by kae in Tuvaluan and by la in Koromfe. Hence, these languages make an overall distinction between relations established between continuous and non-continuous SoAs, expressing the former by means of simple juxtaposition and the latter either by means of juxtaposition or with an overt marker. Mosetén shows instead a general marker for atemporal and simultaneous combination which is not used for the expression of contrast relations. As Sakel (2004: 403) argues, in Mosetén the juxtaposition of clauses expresses both succession in time and simultaneous events, while the marker jö’dyë’yä’ can only be used to combine SoAs occurring at the same time or outside the temporal dimension. This connective is polymorphemic, as jö’dyë’ denotes a ‘coordination with the context’ and -yä’ is the adessive marker (Sakel 2004: 405). This marker is not used to express opposition, since the oppositive relation is coded by means of another construction in Mosetén, where the ﬁrst element of the second clause is sufﬁxed with the marker -ki (Sakel 2004: 407). Finally, four languages in the Comparison sample have no general combination markers, but express both temporal and atemporal combination by means of overt coordinating markers: Supyire, Harar Oromo, Tukang Besi and Maori. Supyire has two dedicated markers expressing the sequential combination, namely the narrative connectives kà and mà, which indicate at the same time a sequence meaning and the switch or continuity of subjects. On the other hand, the general marker sí is used to express atemporal and simultaneous combination and all contrast relations. Carlson (1994) explicitly points out that in Supyire the latent contrast, like the one normally inferred from the combination of discontinuous SoAs, is always made overt by marking the interruption of continuity. By contrast, Tukang Besi and Maori do not have any general marker used also to express contrast, but use two connectives which only express combination. In Maori aa is used for the sequential combination and hoki for the atemporal and simultaneous one. In Tukang Besi kene is used when the two SoAs are simultaneous or atemporally linked, and maka conveys a sequential order, normally in narrative contexts. The last two markers are exempliﬁed in (3.17):

The coding of combination relations

95

(3.17) Tukang Besi, Malayo-Polynesian, Austronesian mia no-rato kene no-ganta-’e na uwe CORE person 3 R -arrive and 3 R -scoop-3 OBJ NOM water ‘. . . people keep coming and fetching water . . . ’ (Donohue 1993: 427)

(a) Te

(b) Jari, sa-rato-no i umbu na Ndokendoke so when-arrive-3POSS OBL edge NOM Monkey o-sampi-’e-mo a loka iso maka o-manga 3R-peel-3OBJ-PRF NOM banana yon and.then 3R-eat ‘So when Monkey arrived at the top he peeled the bananas and then ate them.’ (Donohue 1993: 427) Finally, the strategies attested in Harar Oromo are similar to those just seen in Tukang Besi (in that no general marker common to contrast relations is used), but they differ in the nature of the dedicated sequential marker attested. In Tukang Besi, maka is an independent marker which does not take part in any paradigm, whereas in Harar Oromo the sequential combination of two SoAs is expressed by sufﬁxing the ﬁrst verb form with the non-negative past marker, even if the SoA occurs in the future, so that only the last verb form has a free choice of verb form. In order to make this construction clear, an example of sequential combination is given in (3.18a), while in (3.18b) the non-sequential marker ﬁ is shown. (3.18) Harar Oromo, Cushitic, Afro-Asiatic yaa’-a (a) borúu ás d’uf-é tomorrow here come-PST stay-IMPF ‘Tomorrow he will come and stay here.’ (Owens 1997: 215) (b) heddúu matáa afáan orómóo dubbata ﬁ tíkk’áshó mostly language Oromo speak:IMPF and little afáan adáarée dubbata ﬁ arabá xéesa jira language Harari speak:IMPF and Arabic in exist ‘Mostly Oromo is spoken, and also a little Harari, and there is also Arabic spoken in it.’ (Owens 1997: 219) The non-negative past marker is grammaticalized in this construction, losing its original tense meaning and simply signalling a sequential combination

><18=0C8>= A4;0C8>=B 69HK99B H<9 HKC ,CG -<9 56G9B79 C: 5 H9ADCF5@ A95B=B; 75B 69 C6G9FJ98 =B 5 K<9F9 H<9 KILH5DCG=H=J9 GHF5H9;M =B B5FF5H=J9 7CBH9LHG K<=7< 5F9 IGI5@@M =B H<9 D5GH H9BG9 "BH9F9GH=B;@M =B -I?5B; 9G= =B &5CF= 5B8 =B !5F5F (FCAC H<9 BCB G9EI9BH=5@ A5F?9F =G H<9 G5A9 A5F?9F IG98 HC 7CA6=B9 9BH=H=9G 9B7C898 6M BCIB D
The coding of combination relations

97

examined shows a general connective which is only used for atemporal and simultaneous combination. In the Comparison sample, instead, 5 languages have fully grammaticalized markers used to express only atemporal and simultaneous combination and 3 languages use a general marker for atemporal combination and some contrast relations. The presence of overt markers conveying non-sequential combination, i.e. atemporal and simultaneous combination, is generally more widespread and such markers are more grammaticalized than the dedicated sequential ones. Finally, different markers for sequential combination on the one hand and simultaneous and atemporal combination on the other hand tend to occur when no general combination marker is available and, in that case, they normally co-exist with asyndetic constructions. Only four languages have no general combination marker but express both types of combination with overt markers (Supyire, Harar Oromo, Tukang Besi and Maori). The motivations underlying these cross-linguistic coding tendencies will be investigated in section 3.4. Before moving on to the coding of the SoAs, let us examine the morphophonological complexity of the attested markers used to express combination relations.

3.2.2 Morphophonological complexity of the attested overt markers Data on the morphophonological complexity of all the attested combination markers is reported in Tables 8 and 9 in this chapter and in Tables 53 and 54 in the Appendix. The analysis of the attested markers does not reveal implicational patterns of variation, but shows the cross-linguistic widespread tendency to code combination relations by means of morphophonologically simple connectives, mainly monosyllabic and monomorphemic. European data differ from non-European ones, in that most European languages (all except for Irish, Basque and Chechen) have a free monosyllabic general combination marker. In the Comparison sample, on the other hand, more variation is attested, even though the overall tendency to morphophonological simplicity is conﬁrmed. Speciﬁcally, as can be observed in Table 9, outside Europe bound markers seem to be more widespread and there are some bisyllabic markers. Two languages (Wari’ and Mosetén) even show polymorphemic connectives. As can be observed in data from Japanese, Chechen, Hungarian and SerboCroatian in Table 4, in languages with both a general marker for all combina-

98 Combination relations tion relations and a general marker for discontinuous atemporal combination and opposition, the two markers tend to have the same degree of complexity. Moreover, general combination markers and sequential markers which cooccur in the same language also show the same degree of morphophonological simplicity, as can be observed in the Lezgian type in the table (cf. also Arabic wa and fa). The most signiﬁcant generalization suggested by data on the morphophonological complexity of the attested combination markers is the widespread tendency towards monomorphemic, mainly monosyllabic markers. In Europe, most of the markers used for combination are free morphemes, while outside Europe several bound morphemes are also attested (cf. discussion in chapter 7). Before moving on to the discussion on the coding of two combined SoAs, let us consider the two cases of polymorphemic markers. The composition of the Mosetén connective jö’dyë’yä’ has been already made clear. The other polymorphemic combination marker attested in the Comparison sample is ca’ na, characterizing the resumptive construction of Wari’. Wari’ normally expresses sequential and non-sequential combination by means of simple juxtaposition. Besides the asyndetic strategy, it also has the possibility of putting the verbal phrase ca’ na after the last SoA, meaning ‘thus it was’. (3.19) Wari’, Wanham, Chapacura ’Om ca ca wijimi caca, ’om jin’ not:exist INFL:nrp/p stubborn 3pl.M not:exist INFL:nrp/p afraid caca-in ma’ quem, ca’ na. 3pl.M-3.N that:prox:hearer ref this:N 3s:rp/p ‘They were not stubborn (and) they were not afraid of it, thus it was.’ (Everett and Kern 1997: 160) Example (3.19) shows the use of ca’ na. Its morphology is completely transparent and is composed by the neuter deictic ca’ followed by the realis past, third person inﬂectional particle. The juxtaposition of the two elements gives rise to an existential predication meaning ‘thus it was’. As Everett and Kern (1997: 160) argue, this construction is commonly used to express the combination relation.

Tukang Besi

Maori

Harar Oromo

Supyire

Serbo-Croatian

Hungarian

Chechen

Lezgian

Japanese

S EMANTIC DOMAIN

all combination types atemp. combination/opposition all combination types all combination types sequential combination all combination types atemp. combination/opposition all combination types atemp. combination/opposition all combination types atemp. combination/opposition sequential combination atemp., simult. comb./opposition sequential combination sequential combination atemporal, simultaneous combination sequential combination atemporal, simultaneous combination sequential combination atemporal, simultaneous combination

M ARKER

-te -i -ni wa -na ’a tq’a és meg i a pa sì làa -é ﬁ aa hoki maka kene

bound, monosyllabic, monomorphemic bound, monosyllabic, monomorphemic bound, monosyllabic, monomorphemic free, monosyllabic, monomorphemic bound, monosyllabic, monomorphemic bound, monosyllabic, monomorphemic free, monosyllabic, monomorphemic free, monosyllabic, monomorphemic free, monosyllabic, monomorphemic free, monosyllabic, monomorphemic free, monosyllabic, monomorphemic free, monosyllabic, monomorphemic free, monosyllabic, monomorphemic free, monosyllabic, monomorphemic bound, monosyllabic, monomorphemic free, monosyllabic, monomorphemic free, monosyllabic, monomorphemic free, bisyllabic, monomorphemic free, bisyllabic, monomorphemic free, bisyllabic, monomorphemic

M ORPHOPHONOLOGICAL PROPERTIES

Table 4: Morphophonological complexity of the attested combination markers: examples of languages with more than one marker

The coding of combination relations

99

100 Combination relations 3.3

The coding of combined SoAs

In the preceding section, it has been argued that the coding of combination relations shows some common tendencies across languages, but does not reveal implicational patterns of variation. The analysis now takes into account the same constructions from another point of view and examines the coding of the combined SoAs with respect to the syntactic parallelism vs. non-parallelism of the attested strategies. Data on the syntactic parallelism of the constructions used for combination relations are reported in Tables 10 and 11 in section 3.5, and in the general tables in the Appendix (Tables 53 and 54). Syntactic parallelism has been deﬁned with respect to the verb forms encoding the combined SoAs. If they are both balanced or both deranked, the construction is parallel, whereas the presence of just one deranked verb determines the syntactic non-parallelism of the construction. The exam of the syntactic parallelism of constructions expressing temporal and atemporal combination has revealed a non-random distribution of parallel and non-parallel strategies. The identiﬁed pattern of variation can be formulated as follows: (3.20) The combination parallelism implication: Non-parallel construction for simultaneous and atemporal combination → non-parallel construction for sequential combination. In a given language, the use of a non-parallel construction to express a nonsequential combination, either simultaneous or atemporal, implies the use of a non-parallel construction to express the sequential combination. The combination parallelism implication excludes a language in which a non-sequential combination may be expressed by means of a non-parallel construction, while the sequential combination is only expressed by a parallel one. Table 5 shows the attested language types. Besides the syntactic parallelism of the constructions, the table also indicates their semantic domain. G indicates a general construction, thus used for more than one relation, and d indicates a dedicated construction. In order to distinguish the various general constructions attested within a language, every general construction is univocally identiﬁed by a Greek letter (Gα , Gβ , Gγ . . . ). In the table, if the expression of two relations is indicated by G followed by the same letter, it means

The coding of combined SoAs

101

that these relations are expressed by means of the same general construction. For instance, in Table 5 the Portuguese construction with the connective e is univocally identiﬁed by Gα , which is used to express atemporal, simultaneous and sequential combination, and is characterized by syntactic parallelism (indicated by the symbol +). The notation adopted in this table will recur in the rest of the work, as it allows to represent the semantic domain of the attested constructions together with the relevant parameter under exam. Table 5: syntactic parallelism of constructions expressing combination: attested types. + = syntactically parallel construction; - = syntactically non-parallel construction; d = presence of a dedicated marker; Gx = general construction, with x identifying a given construction attested in the language

sequential c. Kolyma Yukaghir Japanese Korean Harar Oromo Lezgian Maori Portuguese Polish Serbo-Croatian

simultaneous c.

atemporal c.

Gα -

Gα -

Gα -

Gα +/Gβ Gα +/Gβ d-

Gα +/Gβ Gα +/Gβ -

Gα +/Gβ -/Gγ Gα +/Gβ -

Gα + Gα + Gα + Gα + Gα + Gα +

Gα + Gα + Gα + Gα + Gα +, Gβ + Gα +, Gβ +

Gα +, d -

d+ Gα + Gα + Gα +, d+

The lines in Table 5 indicate the cut-off points of the implication and identify the three attested language types. The ﬁrst three languages (Kolyma Yukaghir, Japanese and Korean) exemplify three coding conﬁgurations characterized by the presence of syntactic non-parallelism in the expression of both temporal and atemporal combination. The use of non-parallel constructions to convey all combination types is only attested in the Comparison sample. In (3.21) an example from Kolyma Yukaghir is given, where a sequential (3.21a) and a simultaneous (3.21b) combination are conveyed by non-parallel constructions. In both cases, the verb coding the ﬁrst SoA is deranked and receives a switch-reference marker. (3.21) Kolyma Yukaghir, Yukaghir omos’ (a) loˇcil arqa jaqa-de-ge lejtej-m ﬁre near arrive-3sg-DS learn-TR:3sg well

102 Combination relations ‘He came closer to the ﬁreplace and she recognized him.’ (Maslova 2003: 380) (b) terikie-die iNd’e-t modo-j old.woman-DIM sew-SS:IMPF sit-INTR:3sg ‘The old woman was sitting and sewing.’ (Maslova 2003: 383) In the table, Harar Oromo and Lezgian exemplify two coding conﬁgurations where a non-parallel construction is available only for the expression of sequential combination, while non-sequential SoAs may only be linked by means of a parallel construction. This type is attested both in the European and in the Comparison sample. In Europe, the two languages belonging to this type are Lezgian and Turkish (see examples 1.8 and 1.34). In both cases, the deranked verb receives a sufﬁx belonging to the verbal paradigm and characterized by a narrative function. In the Comparison sample, languages belonging to this type are, besides Harar Oromo, Lango, Marathi and Supyire. The parallel and the non-parallel constructions attested in Turkish are exempliﬁed in (3.22). In (3.22a) the two SoAs are linked in a non-sequential combination and are both encoded by balanced verb forms, whereas in (3.22b) the ﬁrst SoA is coded by a deranked form, sufﬁxed by the narrative converb -Ip. (3.22) Turkish, Turkic, Altaic (a) Hasan i¸s-in-e git-ti, Ali ev-in-e Hasan work-3sg-DAT go-PST Ali house-3sg-DAT dön-dü ve ben park-ta kal-dı-m return-PST and I park-LOC stay-PST-1sg ‘Hasan went to work, Ali returned home and I stayed in the park.’ (Kornﬁlt 1997: 109) (b) çocuk bir kasık çorba al-ıp içer child one spoon soup take-and eat.PRS ‘The child takes a spoon of soup and eats.’ (Haspelmath 2007: 9) In the last four languages in Table 5, the coding conﬁgurations are characterized by a syntactic parallelism in the expression of both temporal and atemporal combination. This third type is the most widespread in both samples, thus conﬁrming the traditional association between a combination relation

Functional motivations in the coding of combination

103

between two SoAs and a syntactically parallel construction. In particular, all the European languages examined, except for Turkish and Lezgian, have only parallel combination constructions. The semantic domains of the attested non-parallel constructions are mainly of two types: these constructions may either be dedicated for sequential combination or they may be used for all combination types. All the constructions used to express atemporal combination and some contrast relations are syntactically parallel, both in the European and in the Comparison sample, except for the Japanese construction with -i, employed to express discontinuous atemporal combination and opposition (e.g. (3.23)). (3.23) Japanese yom-i Hanako-wa tegami-o Taroo-wa hon-o Taro-TOP book-ACC read-COORD Hanako-TOP letter-ACC kai-ta write-PST ‘Taro read a book and Hanako wrote a letter.’ (Nishina 2004: 2) However, as Nishina (2004: 3, 10) points out, the converb -i is not precluded from the expression of temporal combination, but its prototypic use is to convey atemporal combination and to contrast symmetric situations. In this study, it is treated as a general marker restricted to the expression of discontinuous atemporal combination and opposition, because no instances of its sequential use have been found either in the questionnaire or in the bibliographical sources. On the contrary, the converb -te is considered a general combination marker, because many scholars have given and discussed instances of temporal and atemporal uses of this converb (cf. Hasegawa 1996). Its general use is also veriﬁed in the questionnaire.

3.4

Functional motivations in the coding of combination

The presence of overt markers in clause linkage can be accounted for in terms of two different principles: the principle of syntagmatic economy and the principle of iconicity of distance. Haiman (1985: 159) deﬁnes as syntagmatic or discourse economy ‘the tendency to economize on the length or complexity of any utterance or mes-

104 Combination relations sage’, as opposed to paradigmatic economy, which ‘economizes on the inventory of signs within a system’. Two major manifestations of the principle of syntagmatic economy are the morphophonological reduction of the most frequent expressions (known as Zipf’s law) and the omission of information that is recoverable from the context. Both these manifestations of the principle of syntagmatic economy play a role in the coding of coordination relations. As for the presence of overt markers in clause linkage, the action of this principle has two effects: the most used connectives will tend to be phonologically reduced and the relations which are more easily inferable from the context will not need to receive overt marking. However, as shown in the preceding sections, the coding of combination does not reveal any asymmetry in the presence of overt markers for the expression of temporal and atemporal combination. There are languages which do not have any overt combination connective at all, languages which use overt connectives for both combination types and languages which use an overt connective only in the expression of one relation, some just for the sequential combination and others just for the simultaneous and atemporal combination. This distribution may be accounted for in terms of syntagmatic economy: none of the sub-relations of combination is more easily inferable than the others or is radically more frequent in discourse than the others. In other words, the parameter of temporality identiﬁes two types of combination, whose differentiation in discourse is not signiﬁcant enough to determine cross-linguistic coding asymmetries. The other principle at work in the use of overt markers is the principle of iconicity of distance, according to which the formal distance between linguistic expressions corresponds to the conceptual distance between their meanings (Haiman 1985: 102-147, Croft 1990: 174-83). It is a sub-type of what Haiman (1985: 11) calls iconic motivation, which establishes a correspondence between linguistic structure and conceptual structure, so that the relations between conceptual entities are diagrammatically reﬂected by the relations between the linguistic structures expressing them. Iconic motivation is opposed to isomorphism, i.e. the one-to-one correspondence between parts of linguistic structure and parts of conceptual experience (Haiman 1985: 11). As Haiman (1985: 102) argues, ‘if a linguistic utterance were deﬁnable phonetically, the “linguistic distance” between two expressions would be equivalent to the time interval between them.’ This metaphor emphasizes the correspondence between the distance of two linguistic elements and the temporal separation of the elements they denote. If these elements are clauses, the con-

Functional motivations in the coding of combination

105

ceptual distance of the linked SoAs will be mirrored by the presence of linguistic material separating them, like a connective, whereas their conceptual closeness will be mirrored by the omission or reduction of material between them. Haiman deﬁnes the conceptual distance between two SoAs on the basis of four factors: the semantic features, properties or parts they share, the degree to which they affect each other, the extent to which they are factually inseparable and their perception as a unit (Haiman 1985: 107). Causative relations instantiate SoAs which are conceptually very close. Coordination relations, on the other hand, are typically identiﬁed as instances of conceptually unrelated SoAs. However, as Haiman (1985: 111) and Givón (1979: 875) argue, it is also possible to distinguish between different degrees of conceptual closeness within combined SoAs, depending on whether the SoAs have different subjects or topics, whether they denote different events or simply different aspects of a single event and whether they represent events occurring at different times or simultaneously. According to Haiman, the use of an overt marker often denotes ‘disassociation’ (Haiman 1985: 114) between the SoAs it links, that is, overt connectives increase the conceptual distance between the clauses. This disassociation may denote a long temporal interval between the SoAs, a difference in the participants or a difference in polarity. What Haiman deﬁnes as disassociation closely resembles what Givón calls discontinuity. The two agree in remarking that a greater conceptual distance, as the one determined by what they call disassociation and discontinuity, is cross-linguistically characterized by overt marking. The parameter of temporality, however, distinguishes between two relations in which the linked SoAs may have the same degree of conceptual closeness. Two sequential or simultaneous SoAs are not necessarily conceptually closer than SoAs linked in an atemporal combination. Sequential and simultaneous SoAs may either be spatially contiguous or occur in distant places, they may have the same or different subjects and, in case of sequential SoAs, they may occur one immediately after the other or they may be separated by a long time interval. Likewise, atemporality itself does not imply that the two SoAs are conceptually close or distant, since they may be simultaneous or sequential, spatially contiguous or distant, and the subjects may either be the same or different. What matters is that their collocation on the temporal axis is not relevant to the combination itself. In other words, sequential, simultaneous and atemporal combination relations may link SoAs which are

106 Combination relations spatio-temporally contiguous and thus conceptually close, as well as SoAs which are not contiguous and conceptually more distant. However, as argued in the preceding sections, three major coding tendencies can be identiﬁed. First of all, simultaneous and atemporal combination may always be expressed by means of the same constructions. Secondly, discontinuous atemporal combination is often expressed by general markers used also for the expression of oppositive contrast. Finally, dedicated sequential markers are mostly part of the verbal paradigm or of the switch-reference system. Different kinds of dedicated sequential markers are quite rare or at a low stage of grammaticalization. Let us now go through the possible explanations for such cross-linguistic tendencies. As already argued in section 3.1.2, simultaneous and atemporal combination share an important conceptual feature, which differentiates them from sequential combination. Two SoAs linked in a simultaneous or atemporal combination are not presented in a sequential order. In case of simultaneity, the order is intrinsically absent, in case of atemporality, the possible order is irrelevant to the relation itself. This regularity may be accounted for in terms of iconicity. As Haiman (1985: 26) puts it, the recurrent similarity in form reﬂects similarity in meaning. Therefore, the recurrent expression of simultaneous and atemporal combination by means of the same construction may be explained on the basis of their sharing the lack of an obligatory sequential order in which the linked SoAs must be presented. The expression of discontinuous atemporal combination by means of general markers used also to convey an oppositive contrast relation is due to both the conceptual closeness of the two relations and to the fact that opposition is easily, and frequently, inferable from a non-sequential combination. Both opposition and atemporal combination are indeed characterized by the irrelevance of the temporal location of the SoAs. What differentiates them is the emphasis put on the simple cooccurrence or on the contraposition of the SoAs. Atemporal combination establishes only the cooccurrence of two SoAs, without further specifying the nature of the relation, whereas opposition speciﬁes the atemporal cooccurrence as originating a conﬂict. The third cross-linguistic trend concerns dedicated sequential markers, which are frequently part of the verbal paradigm or of the switch-reference system. The reason underlying this widespread property of dedicated sequential markers is to be found in their use within narrative contexts. In narrative contexts, strategies indicating the continuity of the narration are very common, in order to create internal cohesion and to spell out the

Functional motivations in the coding of combination

107

rhythm of the plot. Therefore, especially in verb-ﬁnal languages, the verbal paradigm often presents speciﬁc narrative forms, which are normally called narrative converbs (cf. Haspelmath and König 1995) and are basically used to link successive SoAs without further specifying the relation. A similar function is fulﬁlled by switch-reference markers, which indicate the identity or diversity of the subject of the subsequent clause in the narration. There are two languages where the only syndetic strategy consists of a sequential marker, which is not a narrative converb nor a switch-reference marker: Hdi and Lango. In both cases the two markers derive from verbs and are at a low stage of grammaticalization. What is interesting is that these verbs have developed their combination function in narrative contexts. As pointed out by Mithun (1988: 345-349), the devices for narrative cohesion like adverbial forms connecting to the previous or the subsequent discourse may be the source for combination markers. In these two languages, however, the situation is a bit different, because the original lexical item does not have any intrinsic linking function. The case of Hdi is especially clear. As already pointed out, the verb lá means ‘depart, go’ and has developed its sequential function in contexts where a subject had to move to another location in order to continue its action or to begin a new one there. As a coordinating marker, it occurs in its nominalized form, marking person agreement with pronouns belonging to the possessive set, and denotes separation (thus conﬁrming Haiman’s hypothesis about disassociation) and temporal sequentiality of the SoAs it links. Example (3.16) is repeated here, since it shows the original lexical meaning ‘to go’ which co-exists with the combinative function. (3.24) Hdi, Chadic, Afro-Asiatic krì kà lá-b-ì lá-ghà pákáw ghúvì mbàâ ká then COMP dog SEQ go-out-REF go-D : PVG hyena kà mná-n-tá krì SEQ tell-3sg- REF dog ‘Then dog went away and Hyena said to Dog . . . ’ (Frajzyngier and Shay 2002: 429-30) The source of grammaticalization crucially inﬂuences the semantics of the connective. If a lexical item develops its combinative function in narrative contexts, it becomes a sequential marker and then may expand its scope to include other types of combination.

108 Combination relations Proceeding now with the analysis of how two combined SoAs are coded, the parameter of syntactic parallelism reveals a regular pattern of cross-linguistic variation. The combination parallelism implication states that if a language has a non-parallel construction to express a non-sequential combination, either simultaneous or atemporal, it will also have a non-parallel construction to express the combination between sequential SoAs. This asymmetry may be explained in terms of syntagmatic economy. As pointed out in section 3.3, in all languages showing a non-parallel construction for the expression of sequential combination and a parallel strategy for atemporal and simultaneous combination, the non-parallel construction is dedicated for the sequential combination and is characterized by a deranked verb form lacking tense and mood speciﬁcations. This is the case for -Ip in Turkish (see example (1.8)), -na in Lezgian (see example (2.22)), -é in Harar Oromo (example (3.18)), tˆE in Lango, -¯un in Marathi (see example (3.25)) and kà/mà in Supyire. Example (3.25) shows the non-parallel strategy used in Marathi for the expression of sequential combination. As Pandharipande (1997: 164) argues, only the last verb form is ﬁnite, while the preceding ones are conjunctive participial forms, i.e. deranked forms which cannot receive tense and mood speciﬁcations. (3.25) Marathi, Indo-Iranian, Indo-European an¯u mumba¯ıl¯a dz¯au¯ n Madh¯ul¯a bhet.u¯ n Anu Bombay:to go:CONJ.PTCP Madhu:ACC meet:CONJ.PTCP paratl¯ı return:PST:3sf ‘Anu went to Bombay, met with Madhu and returned.’ (Pandharipande 1997: 164) According to the economic principle of information recoverability already discussed above, information that is recoverable from the context is not further speciﬁed. Given the temporal collocation of one of the linked SoAs, a sequential temporal relation predetermines the temporal collocation of the other one. In the attested non-parallel constructions, the overt marker is dedicated for sequential combination and explicitly lexicalizes the successive location of the SoAs on the temporal axis. Therefore, if one of the SoAs is fully localized,

Functional motivations in the coding of combination

109

showing all the categories of tense, aspect and mood, the temporal location of the other one is recoverable from the context. Speciﬁcally, it is recoverable from the temporal location indicated on the ﬁnite verb. As argued by Chung and Timberlake (1985: 256), the temporal collocation of the linked SoAs inﬂuences the aspect and mood values. Chung and Timberlake deﬁne aspect as the characterization of the SoAs with respect to a point or interval of time, and mood as the characterization of the SoAs with respect to alternative worlds that might exist at some point in time. The direct consequence of the conceptual similarity of tense, aspect and mood is that the temporal location of the SoAs crucially determines the range of possible aspect and mood values that the SoAs may have. In particular, as Chung and Timberlake (1985: 257) point out, whereas a simultaneity relation tends to correlate with an imperfective aspect, a sequential relation correlates with a perfective aspectual value. This is due to the fact that two sequential SoAs constitute successive phases of the same macro-event and, especially in narrative contexts, they denote actions that have been carried out at successive moments. Finally, two SoAs which are combined as parts of the same overall sequence of events are jointly asserted as realis or jointly elicited, as widely argued in section 3.1, and thus tend to share the same mood. Therefore, the temporal, aspectual and modal characterization of the linked SoAs is predetermined by the sequential relation itself. In case the relation is overtly coded by a dedicated marker, information relative to the tense, aspect and mood of the linked SoAs may be made explicit only once. If one of the SoAs is coded by a ﬁnite verb showing all the relevant information, and a dedicated sequential connective is present, the temporal, aspectual and modal properties of the other SoA will follow. What has been said so far may hold for the expression of simultaneous combination too, because the temporal location of two simultaneous SoAs is predetermined by the relation, like in sequential combinations, and this directly inﬂuences the aspect values. Yet, simultaneous combination is always expressed by the same construction which is used also for atemporal combination and there are no dedicated markers coding the simultaneity of two coordinated SoAs (cf. discussion in sections 3.2). Hence, the overt markers used to express simultaneous combination are always ambiguous between a simultaneous and an atemporal meaning and do not speciﬁcally code the temporality of the SoAs. Therefore, even though the modal properties of the SoAs are the same, as a result of their being jointly at issue (either in an assertion,

110 Combination relations in a question or in an order), the temporal and aspectual characterization of the SoAs is not recoverable from the context and needs to be overtly marked on each verb form. On the contrary, as already argued, in all languages where sequential combination is the only relation expressed by a non-parallel construction, a dedicated overt marker coding the sequentiality of the SoAs is present.

3.5

Language data

Table 6: European Sample. Coding of combination relations: presence of overt coordinating markers and their semantic domain. - = no overt coordinating marker; d= dedicated; Gx = general, with x identifying a given construction attested in the language; blank= no information available Language

Albanian Basque Belorussian Bulgarian Catalan Chechen Czech Danish Dargi Dutch English Estonian Finnish French Georgian German Greek Hungarian Icelandic Irish Italian Latvian Lezgian Lithuanian

Temporal combination Sequential

Simultaneous

Gα Gα Gα Gα /d Gα -/Gα Gα Gα Gα Gα Gα Gα Gα Gα Gα Gα Gα Gα Gα Gα Gα Gα Gα /d Gα

Gα Gα Gα Gα Gα -/Gα Gα Gα Gα Gα Gα Gα Gα Gα Gα Gα Gα Gα Gα Gα Gα Gα Gα Gα

Atemporal combination Gα Gα Gα /Gβ Gα /Gβ Gα -/Gα /Gβ Gα Gα Gα Gα Gα Gα Gα Gα Gα Gα Gα Gα /Gβ Gα Gα Gα Gα Gα Gα /Gβ

Language data

111

(Continued) Language

Luxembourgish Maltese Norwegian Polish Portuguese Romanian Russian Sardinian Serbo-Croatian Spanish Swedish Turkish Ukrainian

Temporal combination Sequential

Simultaneous

Gα Gα Gα Gα Gα Gα Gα Gα Gα /d Gα Gα -/Gα /d Gα

Gα Gα Gα Gα Gα Gα Gα Gα Gα Gα Gα -/Gα Gα

Atemporal combination Gα Gα Gα Gα /Gβ Gα Gα /Gβ Gα /Gβ Gα Gα /Gβ Gα Gα -/Gα Gα /Gβ

Table 7: Comparison Sample. Coding of combination relations: presence of overt coordinating markers and their semantic domain. - = no overt coordinating marker; d= dedicated; Gx = general, with x identifying a given construction attested in the language; blank= no information available Language

Arabic Dumi Hakha Lai Harar Oromo Hausa Hdi Hebrew Hocak ˛ Iraqw Jamul Tiipay Japanese Kisi Kol. Yukaghir Korean Koromfe Kuskokwim A.

Temporal combination Sequential

Simultaneous

d/Gα Gα Gα d -/Gα -/d Gα Gα Gα Gα Gα Gα Gα

Gα -/Gα Gα Gα -/Gα Gα Gα Gα Gα Gα Gα -/Gα Gα

Atemporal combination Gα -/Gα Gα Gα -/Gα Gα Gα Gα Gα Gα /Gβ Gα -/Gα Gα

112 Combination relations (Continued) Language

Lango Limbu Malayalam Mangarayi Maori Marathi Maricopa Meithei Mosetén NànáfwˆE Ndyuka Persian Rapanui Somali Supyire Tauya Tukang Besi Tuvaluan Vietnamese Wari’ W.Greenlandic

Temporal combination Sequential

Simultaneous

-/d -/Gα -/Gα -/d -/Gα /d -/Gα -/Gα -/Gα Gα -/Gα Gα d Gα -/d -/Gα -/Gα -/Gα

-/Gα -/Gα -/Gα -/Gα -/Gα -/Gα -/Gα -/Gα Gα -/Gα Gα Gα Gα -/Gα -/Gα -/Gα -/Gα Gα

Atemporal combination -/Gα -/Gα -/Gα -/Gα -/Gα -/Gα -/Gα -/Gα Gα -/Gα Gα Gα Gα -/Gα -/Gα -/Gα -/Gα Gα

Table 8: European Sample. Coding of combination relations: morphophonological complexity of the attested markers. + = presence of the given feature; - = absence of the given feature; opp.= opposition; atemp.= atemporal combination; comb.= combination; blank= no information available Language

Marker

Semantic domain

Free

Polysyllabic

Polymorphemic

Albanian Basque Belorussian

e/dhe eta i a i a ta/da i ’a

all comb. types all comb. types all comb. types atemp.comb., opp. all comb. types atemp.comb., opp. sequential comb. all comb. types all comb. types

+ + + + + + + + -

+ -

-

Bulgarian

Catalan Chechen

Language data

113

(Continued) Language

Czech Danish Dargi Dutch English Estonian Finnish French Georgian German Greek Hungarian Icelandic Irish Italian Latvian Lezgian Lithuanian Luxembourgish Maltese Norwegian Polish Portuguese Romanian Russian Sardinian Serbo-Croatian

Spanish Swedish Turkish

Marker

Semantic domain

Free

Polysyllabic

Polymorphemic

tq’a a og wa -ra en and ja ja et da und kai és meg og agus e un -ni, wa -na ir o ann u og i a e s¸i iar i a e i a pa y och =dA ve

atemp.comb., opp. all comb. types all comb. types all comb. types all comb. types all comb. types all comb. types all comb. types all comb. types all comb. types all comb. types all comb. types all comb. types all comb. types atemp.comb., opp. all comb. types all comb. types all comb. types all comb. types all comb. types sequential comb. all comb. types atemp.comb., opp. all comb. types all comb. types all comb. types all comb. types atemp.comb., opp. all comb. types all comb. types atemp.comb., opp. all comb. types atemp.comb., opp. all comb. types all comb. types atemp.comb., opp. sequential comb. all comb. types all comb. types all comb. types all comb. types

+ + + + + + + + + + + + + + + + + + -/+ + + + + + + + + + + + + + + + + + + +

+ -

-

114 Combination relations (Continued) Language

Marker

Semantic domain

Free

Polysyllabic

Polymorphemic

Ukrainian

-Ip i ta

sequential comb. all comb. types atemp.comb., opp.

+ +

-

-

Table 9: Comparison Sample. Coding of combination relations: morphophonological complexity of the attested markers. + = presence of the given feature; = absence of the given feature; opp.= opposition; sim.= simultaneous combination; atemp.= atemporal combination; comb.= combination; blank= no information available Language

Marker

Semantic domain

Free

Polysyllabic

Polymorphemic

Arabic

wa fa -k@ =ĳií -é ﬁ kuma lá veánaga ˛ nee switch-r -te si -i

all comb. types sequential comb. all comb. types all comb. types sequential comb. sim., atemp. comb. all comb. types sequential comb. all comb. types all comb. types all comb. types all comb. types all comb. types all comb. types atemp.comb., opp. none none all comb. types all comb. types none all comb. types sequential comb. all comb. types all comb. types none sequential comb. sim., atemp. comb.

+ + + + + + + + -

+ + -

-

+

+

-

+ + -

-

-

+ +

+

-

Dumi Hakha Lai Harar Oromo Hausa Hdi Hebrew Hocak ˛ Iraqw Jamul Tiipay Japanese

Kisi Kol. Yukaghir Korean Koromfe Kuskokwim A. Lango Limbu Malayalam Mangarayi Maori

-ko kuliko ts’eĳ tˆE -an -um aa hoki

Language data

115

(Continued) Language

Marker

Semantic domain

Free

Polysyllabic

Polymorphemic

Marathi

a¯ n.i wa

all comb. types all comb. types none all comb. types sim., atemp. comb. all comb. types all comb. types all comb. types all comb. types all comb. types all comb. types sequential comb. atemp.comb., opp. all comb. types sequential comb. sim., atemp. comb. atemp.comb., opp. all comb. types all comb. types all comb. types

+ +

+ -

-

+ + + + + + + + + + + + + + -

+ + + + + + -

+ + -

Maricopa Meithei Mosetén NànáfwˆE Ndyuka Persian Rapanui Somali Supyire Tauya Tukang Besi Tuvaluan Vietnamese Wari’ W.Greenlandic

aduga jö’dyë’yä kp´Ekún da va e -na oo kà/mà sì switch-r maka kene kae và ca’ na =lu

Table 10: European Sample. Coding of combined SoAs: syntactic parallelism of the attested constructions. + = syntactically parallel; - = syntactically nonparallel; d= dedicated; Gx = general, with x identifying a given construction attested in the language; blank= no information available Language

Albanian Basque Belorussian Bulgarian Catalan Chechen Czech Danish Dargi Dutch

Temporal combination Sequential

Simultaneous

Gα + Gα + Gα + Gα +/d Gα + Gα + Gα + Gα + Gα + Gα +

Gα + Gα + Gα + Gα + Gα + Gα + Gα + Gα + Gα + Gα +

Atemporal combination Gα + Gα + Gα +/Gβ + Gα +/Gβ + Gα + Gα +/Gβ + Gα + Gα + Gα + Gα +

116 Combination relations (Continued) Language

English Estonian Finnish French Georgian German Greek Hungarian Icelandic Irish Italian Latvian Lezgian Lithuanian Luxembourgish Maltese Norwegian Polish Portuguese Romanian Russian Sardinian Serbo-Croatian Spanish Swedish Turkish Ukrainian

Temporal combination Sequential

Simultaneous

Gα + Gα + Gα + Gα + Gα + Gα + Gα + Gα + Gα + Gα + Gα + Gα + Gα +/d Gα + Gα + Gα + Gα + Gα + Gα + Gα + Gα + Gα + Gα +/d+ Gα + Gα + Gα +/d Gα +

Gα + Gα + Gα + Gα + Gα + Gα + Gα + Gα + Gα + Gα + Gα + Gα + Gα + Gα + Gα + Gα + Gα + Gα + Gα + Gα + Gα + Gα + Gα + Gα + Gα + Gα + Gα +

Atemporal combination Gα + Gα + Gα + Gα + Gα + Gα + Gα + Gα +/Gβ + Gα + Gα + Gα + Gα + Gα + Gα /Gβ + Gα + Gα + Gα + Gα +/Gβ + Gα + Gα +/Gβ + Gα +/Gβ + Gα + Gα +/Gβ + Gα + Gα + Gα + Gα +/Gβ +

Language data

117

Table 11: Comparison Sample. Coding of combined SoAs: syntactic parallelism of the attested constructions. + = syntactically parallel; - = syntactically non-parallel; d= dedicated; Gx = general, with x identifying a given construction attested in the language; blank= no information available Language

Arabic Dumi Hakha Lai Harar Oromo Hausa Hdi Hebrew Hocak ˛ Iraqw Jamul Tiipay Japanese Kisi Kol. Yukaghir Korean Koromfe Kuskokwim A. Lango Limbu Malayalam Mangarayi Maori Marathi Maricopa Meithei Mosetén NànáfwˆE Ndyuka Persian Rapanui Somali Supyire Tauya Tukang Besi Tuvaluan Vietnamese

Temporal combination Sequential combination

Simultaneous combination

Gα +/d+ Gα + Gα + dGα + Gα +/d + Gα + Gα + Gα + Gα Gα +/Gβ Gα + Gα Gα +/Gβ Gα + Gα + Gα +/d Gα + Gα + Gα + d+ Gα +/d Gα +/Gβ Gα + Gα + Gα + Gα + Gα + Gα + Gα + dGα d+ Gα + Gα +

Gα + Gα + Gα + Gα + Gα + Gα + Gα + Gα + Gα + Gα Gα +/Gβ Gα + Gα Gα +/Gβ Gα + Gα + Gα + Gα + Gα + Gα + Gα + Gα + Gα +/Gβ Gα + Gα +/Gβ + Gα + Gα + Gα + Gα + Gα + Gα + Gα Gα + Gα +/Gβ + Gα +

Atemporal combination

Gα + Gα + Gα + Gα + Gα + Gα + Gα + Gα + Gα + Gα Gα +/Gβ -/Gγ Gα + Gα Gα +/Gβ Gα +, Gβ + Gα + Gα + Gα + Gα + Gα + Gα + Gα + Gα +/Gβ Gα + Gα +/Gβ + Gα + Gα + Gα + Gα + Gα + Gα + Gα Gα + Gα +/Gβ + Gα +

118 Combination relations (Continued) Language

Wari’ W.Greenlandic

Temporal combination Sequential

Simultaneous

Gα + Gα +/Gβ -

Gα + Gα +/Gβ -

Atemporal combination Gα + Gα +/Gβ -

Chapter 4 Contrast relations This chapter discusses the attested constructions expressing contrast with respect to the parameters of coding of the relation (section 4.2) and coding of the SoAs (4.3). Before proceeding to the exam of the attested constructions, a deﬁnition of the notion of contrast relation is provided in section 4.1. Then the contrast types identiﬁed on the basis of the semantic parameter of conﬂict will be discussed in section 4.1.2. As will become clear in the course of this chapter, the presence and the nature of overt markers coding contrast obey implicational patterns of crosslinguistic variation. In section 4.2 two implicational patterns concerning the overt contrast markers will be described: the contrast coding implication, concerning the presence of overt markers for the three types of contrast, and the contrast coding complexity hierarchy, concerning the morphophonological complexity of the attested connectives. Moreover, a conceptual space will be built concerning the distribution of the various constructions across different types of contrast relation. Information on the coding of corrective and oppositive contrast relations is not often easy to ﬁnd in descriptive grammars, mainly because these distinctions have not been considered in traditional semantic approaches to coordination until recently. The counterexpectative relation has instead been regarded as adversative par excellence and therefore is more often described. Hence, questionnaires played a fundamental role in the identiﬁcation of corrective and oppositive strategies, because they made it possible to gather detailed data even when the relevant information was missing from the grammar. As a result, the study of the attested contrast markers is mainly based on data from the European sample, because the high number of questionnaires collected made it possible to obtain a complete set of data, which was not possible for the extra-European languages. The descriptive grammars examined for the non-European languages often lacked information about the various types of contrast and the number of questionnaires collected is not high enough to integrate all the missing information. This does not mean that no extra-European data have been taken into account. On the contrary, for the majority of the Comparison sample languages it has been possible to ﬁg-

120 Contrast relations ure out what the relevant constructions were. However, non-European data is much less complete than European data. In section 4.3 the coding of the SoAs will be examined and a description of the attested non-parallel constructions will be made. In section 4.4 a functional account for the cross-linguistic patterns found in the coding of contrast relations will be proposed, based on the principle of syntagmatic economy. As in the preceding section, the reasons why one parameter is crosslinguistically more relevant than the other will be discussed. Finally, all the data concerning the attested constructions used to express contrast relations are presented in section 4.5, where different tables are provided for European and non-European languages. Each time an implicational pattern or a cross-linguistic tendency is stated, however, smaller tables showing the attested language types will be used, in order to make the generalizations clear. 4.1 4.1.1

The notion of contrast relation Contrast as conﬂicting cooccurrence

In section 3.1 it has been stated that contrast and combination can be characterized as cooccurrence relations, because in both cases the linked SoAs are jointly at issue as co-existent facets of an overall scene. Combination was deﬁned in negative terms, as a relation which only establishes the cooccurrence of two SoAs within a common frame without further specifying the nature of the relation. However, a cooccurrence relation established between two SoAs can be further characterized as somehow conﬂicting. In this case, the two SoAs are not simply combined for some reason, but their combination gives rise to a contrast. In section 3.1, Fig. 1 represented the two dimensions of cooccurrent and non-cooccurrent SoAs. In that ﬁgure, the two SoAs associated in a contrast relation originate the following sentence: (4.1) He got stuck in the trafﬁc but managed to arrive in time. The contrast relation implies the combination of two SoAs and further characterizes it in terms of a more or less speciﬁc conﬂict. The two SoAs ‘he got stuck in the trafﬁc’ and ‘he managed to arrive in time’ are asserted as cooccurring, i.e. the speaker wants the hearer to assume that they are jointly

The notion of contrast relation

121

real. Furthermore, their co-assertion is the cause of some surprise, in that they were not expected to occur together. As Malchukov (2004: 183) points out, following Mann and Thompson (1988), two SoAs standing in a contrast relation (i) have some aspects in common (at least the common integrator), (ii) are different under some respects (otherwise they would be identical) and, most importantly, (iii) they are compared with respect to these differences. In other words, what distinguishes the relations of combination and contrast is the presence of the speaker’s perspective, which not only combines but also compares the two cooccurring SoAs, conceiving them in their conﬂicting properties. Contrast as such does not exist in the world, but it is established by the speakers when they compare two cooccurring entities (cf. Blakemore and Carston 1999 and 2005, and Sweetser 1990). Rudolph (1996: 20) describes the connection of contrast as the speaker’s opinion that two SoAs are valid simultaneously and that the second SoA marks a conﬂict with some information given in the ﬁrst one (cf. also Dik 1968: 277). To sum up, a contrast relation is established between functionally parallel SoAs on the basis of a common integrator, just like combination, but unlike combination it further speciﬁes the association in terms of some conﬂict (see also Mauri 2006). 4.1.2

Contrast types

The contrast relation can be further divided into three types on the basis of the conﬂict parameter, which speciﬁes the nature of the conﬂict existing between the linked SoAs. As for the parameter of aim, the intention of a speaker who establishes a contrast between two SoAs is always to assert or elicit information about the whole block of contrasted SoAs, since the relation itself implies that they be jointly at issue, just like in combination. There would be no point in combining two SoAs into a contrast, if the aim of the speaker was to give or elicit information about the individual SoAs. Moving on to the parameter of temporality, the location of the contrasted SoAs along the time axis is not decisive in the identiﬁcation of the types of contrast, because two SoAs may be linked in a conﬂicting relation both if they are sequential (4.2) or simultaneous (4.3). (4.2) It’s unbelievable, I arrived this morning and you are coming now!!

122 Contrast relations (4.3) I am very tired but I have to work. In particular, the identiﬁcation of a contrast puts the temporality of the linked SoAs in the background, because the reason of the conﬂict is to be looked for in the semantics of the SoAs or in the expectations they originate, regardless of their respective temporal order. In case the temporal location of the SoAs participates in creating a conﬂict, as in (4.2), this happens because the temporality is overtly expressed by antonymic temporal references (cf. this morning vs. now). The point is that the conﬂict is established by the speaker on a different time axis than the one where the SoAs occur, and this makes the temporality value irrelevant. As a result of the irrelevance of the temporal location of the SoAs, the parameter of temporality can be regarded as having the ﬁxed atemporal value. The decisive parameter is instead the nature of the conﬂict, on the basis of which three main contrast types are identiﬁed, depending on whether the conﬂicting relation is due to a semantic opposition, a correction or the denial of some expectation (cf. Haspelmath 2007, Lang 1984, Kotcheva 2005, Rudolph 1996). 4.1.2.1 Oppositive contrast The relation of opposition is exempliﬁed in sentences (4.4) and (4.5), where two SoAs are opposed to each other by virtue of their being parallel but different. Haspelmath (2007) deﬁnes this relation as a contrast between two SoAs with no conﬂicting expectations. (4.4) I am working and she is sleeping! (4.5) I bought a pair of shoes whereas Sue found a nice skirt. Some scholars call this type of contrast weak adversativity (cf. Luraghi 1990 about Hittite, Brown 1993 about Serbo-Croatian, Rothstein 1993 about Polish), highlighting the fact that the two SoAs are set against each other simply because they are somehow antithetic with respect to the common integrator (cf. section 3.1 in the previous chapter). Moreover, two SoAs linked in an oppositive contrast relation have usually different participants. Lakoff (1971: 131-136) talks about this relation in terms of ‘semantic opposition’ and describes it as characterized by a component of symmetric ‘and’

The notion of contrast relation

123

(i.e. atemporal combination) together with a presupposition of difference in meaning, whereby the two SoAs become or contain antonyms. Moreover, the order in which the two SoAs are presented may be reversed without affecting the general meaning of the assertion, exactly like in the relation of atemporal combination. (4.6) Dogs are very attached to their masters while cats love their independence. (4.7) Cats love their independence while dogs are very attached to their masters. The conceptual border between atemporal combination of discontinuous SoAs and oppositive contrast is not easy to draw. An opposition relation is often inferred from an atemporal combination established between discontinuous SoAs and, in turn, an atemporal combination is often established in order to point out a parallelism between two SoAs, highlighting their difference in meaning (cf. examples (3.7), (3.8) and (3.10)). Languages tend to convey the two relations of atemporal combination and opposition by means of the same constructions (see for instance the construction with and in English, in (4.4)). However, the existence of dedicated connectives encoding the oppositive contrast, such as mentre in Italian, or while in English (cf. (4.6) and (4.7)), leads to conclude that the two relations are distinguishable, albeit very close to each other. The relationship between atemporal combination and opposition is better represented along a continuum, as will be proposed in section 6.2.1.2, rather than by means of a neat contraposition. An opposition is almost always inferable from a discontinuous atemporal combination and, in that case, the choice between the two readings is a matter of degree. More emphasis could be given on the simple cooccurrence of the two SoAs or on their antonymic aspects. 4.1.2.2 Corrective contrast If the conﬂict existing between two SoAs is generated by the negation of the ﬁrst SoA and its successive substitution with the second SoA, the contrast relation is said to be corrective (cf. Rudolph 1996 and Lang 1984). This notion is quite uncontroversial in the literature on coordination, because this type of contrast is clearly distinguishable from all others and can be easily identiﬁed

>=CA0BC A4;0C8>=B 57FCGG @5B;I5;9G "B 5B8 HKC =BGH5B79G C: H<=G 7CBHF5GH HMD9 5F9 ;=J9B '4C4A 8B =>C BCD3H8=6 8= 78B A>>< 1DC 74 8B ?;0H8=6 8= C74 60A34= 4 383 =>C AD= D?>= C74 78;; 1DC B8F;H 0=3 ;0I8;H 5>;;>F8=6 C74 A4BC >5 C74 6A>D? =J9B HKC ,CG ? 5B8 @ H<9 7CFF97H=CB F9@5H=CB 75B 69 7<5F57H9F=N98 =B HIGH=:M=B; H<9 8=GDIH9 BG7CA 6F9 5B8 I7FCH ": H<9 GI6GH=HIH=CB 7CB79FBG 5 G=B;@9 5GD97H CF 9@9A9BH C: H<9 HKC ,CG H<9 F9GH C: H<9 7@5IG9 =G C:H9B CA=HH98 =B CF89F HC :C7IG CB H<9 9@9A9BH 6FCI;C ?DC H>DA 106 >= C74 :8C274= C01;4 1DC >= C74 B>50 -<9 7CFF97H=J9 7CBHF5GH G<5F9G GCA9 DFCD9FH=9G K=H< 6CH< CDDCG=H=CB 5B8 5@ H9FB5H=J9 (B H<9 CB9 <5B8 =H =G 5@K5MG 9GH56@=G<98 69HK99B 5 DCG=H=J9 5B8 5 B9;5H=J9 ,C HD=C4A4G?42C0C8E4 2>=CA0BC -<9 H<=F8 HMD9 C: 7CBHF5GH =G 7<5F57H9F=N98 6M 5 7CB:@=7H CF=;=B5H98 6M H<9 89 B=5@ C: GCA9 9LD97H5H=CB 5B8 =H K=@@ 69 75@@98 2>D=C4A4G?42C0C8E4 :C@@CK=B;

The notion of contrast relation

125

the terminology proposed by Scorretti (1988) for Italian (avversatività controaspettativa). It is the type of contrast which has been discussed the most. As Lang (2000: 245) puts it, it involves relating two combined SoAs to an assumption which, if not available in explicit terms, has to be inferred from the context. Speciﬁcally, the assertion of the second SoA is in contrast to ‘an assumption that may be either read or inferred from previous information’ (Lang 2000: 246), in what Lang calls a source/target relation. The second SoA constitutes the source from which to look for a pertinent assumption that meets the condition of contrast, and this assumption constitutes the target. In examples (4.11) and (4.12), two different instances of counterexpectative contrast are given and the relation source-target is made explicit for both. In (4.11) the assumption is originated in the ﬁrst SoA, since the fact that John is tall would lead to think that he is also good at basketball, and the assertion of the second SoA contrasts with this assumption. (4.11) John is tall but he’s no good at basketball. (Lakoff 1971: 133) - John is tall → ASSUMPTION: ‘he is good at basketball’. - ‘he is good at basketball’ → CONTRASTS with he is not good at basketball. The sentence given in (4.12), instead, is often considered as an oppositive relation (cf. Malchukov 2004 and Rudolph 1996) following Lakoff’s deﬁnition (Lakoff 1971: 131-136). As can be seen from the common integrators indicated under the various possibilities in (4.13), the contrast in (4.12) is not an opposition but implies the denial of an assumption, only the assumption is not generated by the ﬁrst SoA, but lies in the context of discourse. Specifically, the assumption negated in (4.13a) is that both John and Bill are rich, whereas the assumption negated in (4.13c) is that only one of them is rich. A semantic opposition as deﬁned by Lakoff could be inferred from (4.13b), where no assumption is denied. (4.12) John is rich but Bill is poor. (4.13)

(a) John is rich but Bill is poor. //As for John and Bill, are they both rich?// (b) John is rich and Bill is poor. //As for John and Bill, what about their income?// (c) John is rich but Bill is rich, too. //As for John and Bill, they aren’t both rich, are they?//

126 Contrast relations (d) *John is rich but Bill is rich. //???// In order to code an oppositive relation, the sentence in (4.12) should be changed into ‘John is rich while Bill is poor’. In the literature on adversativity, one of the reasons that has led some scholars (cf. Lakoff 1971 and Malchukov 2004) to state that a sentence like (4.12) is not a case of counterexpectative contrast is that it is not replaceable by a concessive subordinating construction, whereas a sentence like the one in (4.11) is. (4.14) Although John is tall, he’s not good at basketball. (4.15) *Although John is rich, Bill is poor. However, what the difference highlighted in (4.14) and (4.15) shows is simply that in one case (4.14) the denied assumption is generated by the ﬁrst SoA, so there is a frustration of causality like the one characterizing the notion of concessivity. By contrast, in the second case (4.15) the assumption lies in the context and it is thus not possible to reformulate the relation in terms of concessive frustrated causality. The structure SoA1 causes assumption X which is denied by SoA2 −−−−→ −−−−−−−−−−−−−→

can reword only one type of counterexpectative contrast, namely the one where the origin of the assumption is the ﬁrst SoA. However, it cannot be used to reword the case in which the origin of the assumption is outside the sentence. In this work it will be assumed that counterexpectative contrast encompasses both cases, the case in which the origin of the denied assumption lies in the context outside the sentence and the case in which it lies within the sentence.

4.2 4.2.1

The coding of contrast relations The contrast coding implication

Data concerning the presence of overt markers coding the three types of contrast relations can be found in Tables 15 and 16 in this chapter, or in the general Tables 53 and 54 in the Appendix. This section only deals with the

+74 2>38=6 >5 2>=CA0BC A4;0C8>=B

DF9G9B79 JG 56G9B79 C: CJ9FH 7CCF8=B5H=B; A5F?9FG K=H97H C: H<9 B9LH G97H=CB K<9F9 5 7CB79DHI5@ GD579 65G98 CB H<9 5HH9GH98 G9A5BH=7 8CA5=BG K=@@ 69 89G7F=698 -<9 9L5A C: H<9 CJ9FH A5F?9FG IG98 HC 7C89 H<9 7CBHF5GH F9@5H=CBG <=;< @=;A 2>D=C4A4G?42C0C8E4 2>=CA0BC BH=34B8B 5>A >??>B8C8E4 2>=CA0BC 0BH=34B8B 5>A 2>AA42C8E4 2>=CA0BC ": =B 5 ;=J9B @5B;I5;9 7CIBH9F9LD97H5H=J9 7CBHF5GH =G BCFA5@@M 9LDF9GG98 6M A95BG C: 5B 5GMB89H=7 7CBGHFI7H=CB CDDCG=H=J9 7CBHF5GH 7CFF97H=J9 7CBHF5GH CF 6CH< K=@@ 5@GC 69 9LDF9GG98 6M A95BG C: 5B 5GMB89H=7 7CBGHFI7H=CB CB J9FG9@M =: 5 @5B;I5;9 A5M IG9 5B CJ9FH 7CCF8=B5H=B; A5F?9F HC 9LDF9GG CDDCG =H=J9 CF 7CFF97H=J9 7CBHF5GH =H A5M 5@GC IG9 5B CJ9FH A5F?9F =B H<9 9LDF9GG=CB C: 7CIBH9F9LD97H5H=J9 7CBHF5GH 'CB9 C: H<9 @5B;I5;9G 9L5A=B98 9LDF9GG9G 7CFF97H=CB 5B8 CDDCG=H=CB 6M A95BG C: CJ9FH A5F?9FG 5B8 7CIBH9F9LD97H5H=J9 7CBHF5GH HILH5DCG=H=CB -56@9 G
(DDCG=H=J9 7CBHF5GH

CFF97H=J9 7CBHF5GH

CIBH9F9LD97H 7CBHF5GH

-<9 :=FGH HKC @5B;I5;9G 05F=b 5B8 %5B;C =BGH5BH=5H9 H<9 75G9 =B K<=7< H<9 HILH5DCG=H=J9 GHF5H9;M =G J9FM K=89GDF958 5@GC =B H<9 7C8=B; C: 7CA6=B5H=CB F9@5H=CBG G99 G97H=CB :CF 5 89H5=@98 7CAD5F=GCB 69HK99B 7CA6=B5H=CB 5B8 7CBHF5GH 7C8=B; GHF5H9;=9G

128 Contrast relations Wari’ is the only language in which no contrast marker at all is attested, whereas in Lango and Limbu the juxtapositive strategy is still the most common one, but it is also possible to use an overt connective to signal counterexpectative contrast (´EntˆO in Lango and kere in Limbu). A slightly different case is that of Kolyma Yukaghir, where the asyndetic construction used to express the three contrast relations is the same as the one used for combination (see example (3.21)). Namely, it is a non-parallel construction in which the ﬁrst verb form receives a switch-reference marker, which however cannot be considered an overt coordinating marker because it does not distinguish between the expression of subordinate and coordinate relations. An example of this strategy is given in (4.17) (4.17) Kolyma Yukaghir, Yukaghir cˇ olha-j-de-ge el-¯už¯u t¯at š¯al-e CA tree- INSTR push- PRF-3sg- DS NEG-move(NEG:3sg) ‘She pushed it with a stick, but it did not move.’ (Maslova 2003: 373) The sentence in (4.17) expresses two SoAs linked by a contrast relation, where the contrast is generated by negating the expectation that something which has been pushed with a stick normally will move. However, the contrast relation is not overtly signalled, but only inferred from the combination of the two SoAs. The second group of languages in Table 12 - Maori, Georgian, Tukang Besi and Hausa - instantiates the case in which the expression of counterexpectative contrast requires the use of an overt marker, while the other two relations may be expressed by simple juxtaposition. There is no implication between the use of overt markers for corrective contrast and for the oppositive one, since there are languages which use a syndetic construction just to express correction and not to express opposition as well as languages showing an overt marker for opposition and simple juxtaposition for correction (in that case, opposition is normally expressed by means of the construction used also for combination relations). This second language type is attested in both samples, although among European languages only Georgian and Turkish systematically use juxtaposition to express oppositive or corrective contrast, while all other languages normally use overt markers for the three contrast relations. In the Comparison sample, on the other hand, asyndetic strategies are much more common.

The coding of contrast relations

129

Speciﬁcally, as can be observed in Table 15, Turkish and Georgian have the possibility to use or not to use an overt marker in the expression of oppositive and corrective relations, while the expression of counterexpectative contrast normally requires some overt connective. Georgian corrective and counterexpectative constructions are shown in (4.18): (4.18) Georgian, Kartvelian k’i ar mecadineobs, (a) Petre tavis otaxši Peter his room:LOC EMPH NEG study:3sg baGši tamašobs. garden:LOC play:3sg ‘Peter is not studying in his room, but he is playing in the garden.’ (M.T., questionnaire) (b) bavshvma icis ara kartuli, aramed it’aliuri child know:3sg NEG Georgian, COORD Italian ‘The child does not speak Georgian, but Italian.’ (Manana Topadze, p.c.) (c) dzalian mc’q’uria, magram portoxlis c’veni Very be.thirsty.1sg COORD orange:GEN juice:NOM ar momc’ons. NEG please.me:3sg ‘I’m very thirsty but I don’t like orange juice.’ (M.T., questionnaire) The sentence in (4.18a) exempliﬁes the corrective strategy that is more common in Georgian. The ﬁrst clause contains the emphatic marker k’i, which strengthens the negation of the ﬁrst SoA, and the second clause follows without any marker signalling the corrective relation. As can be observed in (4.18b), the same corrective relation might also be signalled by the overt marker aramed, which is anyway not obligatory. Yet, the use of aramed is characterized by a very formal connotation and is not commonly used in spoken language (Manana Topadze, p.c.). Finally, as can be observed in (4.18c), the connective magram explicitly codes the contrast determined by the denial of an expectation. It must be remarked that the use of a juxtapositive strategy for the expression of correction is widespread across languages at the spoken level (cf.

130 Contrast relations in English ‘Peter is not studying in his room, he is playing in the garden’). In such constructions, the opposite polarity and a speciﬁc intonation help in identifying the juxtaposition as an instance of correction. However, juxtaposition is often limited to the colloquial level and both in written and in spoken language the use of overt markers is preferred (cf. Italian, French, English, German among others). Finally, again referring to Table 12, the third attested type is characterized by the use of overt markers for the expression of all contrast relations. As can be observed in Table 16, in the Comparison sample the use of syndetic constructions for the expression of the three contrast relations is just one of the possibilities and does not stand out as the preferred conﬁguration. Among the non-European languages under exam, it is attested in Arabic, Hebrew, Japanese and Persian. Within the European sample, on the other hand, the use of syndetic strategies is attested in all the languages examined. In languages which use some overt coordinating marker for the expression of the three relations, these markers are normally functionally differentiated. In what follows the semantic domains of the attested contrast markers will be analyzed.

4.2.2

The contrast conceptual space

The semantic domains of the attested contrast markers are dealt with in a separate section because of the wide and complex range of combinations. Data on the semantic domains of the attested contrast markers are represented in Table 15 and 16 and in the general tables at the end of the Appendix. The attested cross-linguistic variation in the degree of coding of the three contrast relations lends itself to be represented by means of a conceptual space, reported in (4.19) and exempliﬁed in Table 13. (4.19) The contrast conceptual space: O PPOSITION · · · C ORRECTION · · · D ENIAL OF AN EXPECTATION If a coordinating marker is used to express more than one contrast relation, it will convey relations which stand close to each other in the conceptual space. Therefore, if a marker is used to express both opposition and counterexpectative contrast, it will also be used to express the corrective contrast relation.

The coding of contrast relations

131

Table 13 shows the attested semantic domain conﬁgurations, i.e. the attested semantic map types. Each attested type is exempliﬁed by one language in the table, even though at least three languages per type are found in the samples (see Tables 15 and 16). Table 13: Attested contrast semantic map types OPPOSITION

German

Italian

French

und

sondern

e mentre

bensì

a

Polish

a

Russian

Hausa

Tuvaluan

aber

ma

et tandis que

S-Croatian

DENIAL OF EX PECTATION

CORRECTION

però

mais

v‚ec´

ali

ale

a

no

kuma

amman

kae

As can be observed in Table 13, languages that do not have a general marker used for discontinuous atemporal combination and opposition, like Polish a, commonly express opposition by means of the general construction used for combination, thus leaving further conceptual speciﬁcations of op-

132 Contrast relations position to inferential enrichment. However, these languages normally also develop a dedicated oppositive marker, which is often grammaticalized from subordinators denoting contemporaneity or collocation of the two events in the same place (Traugott 1986: 143-44), like while and whereas in English. In (4.20) and (4.21) two examples from Italian and French are given. (4.20) Italian, Romance, Indo-European una cartolina e/mentre comprato Maria ha Mary AUX.PRS:3sg buy:PST.PTCP INDEF postcard COORD Giovanni un poster. INDEF poster John ‘Mary bought a postcard and/whereas John a poster.’ (4.21) French, Romance, Indo-European une carte postale acheté Marie a Mary AUX.PRS:3sg buy:PST.PTCP INDEF card postal et/tandis que John un poster. COORD John INDEF poster ‘Mary bought a postcard and/whereas John a poster.’ (M.V., questionnaire) In both languages, the general combination construction is commonly used to express an opposition between two SoAs, but a dedicated option is also available. In Italian, the oppositive connective is mentre, which has developed a contrastive function from an original meaning of simultaneity. Just like English while, mentre ﬁrstly denoted a short lapse of time, from which a contemporaneity meaning has arisen. Then, the inference of opposition, which frequently originates from contemporaneity, has grammaticalized into a new coordinating function. The three stages of the process of grammaticalization co-exist nowadays. Moreover, this pattern of grammaticalization of dedicated oppositive connectives seems to be recursive across languages, since the meaning of contemporaneity is also the basic meaning of tandis que (cf. Heine and Kuteva 2002: 291). Languages with a general marker for discontinuous atemporal combination and opposition cannot use the general combination marker to convey an opposition. The use of the general combination connective i in (4.22), instead of the marker a, would prevent the oppositive reading.

The coding of contrast relations

133

(4.22) Polish, Slavic, Indo-European my jemy! Oni płaca˛ a 3pl pay.3pl COORD 1pl eat.1pl ‘They pay and we eat!’ (A.L., questionnaire) Oppositive contrast may always be expressed by means of the same general construction used for atemporal combination, as will be more widely argued in section 6.2. This general construction may be the one also used to express all combination relations (like French et, English and, German und) or it may be restricted to discontinuous atemporal combination and opposition (like Polish a). In Russian, Bulgarian, Chechen and Lithuanian the general marker expressing discontinuous atemporal combination and opposition is used also to express corrective contrast, as can be seen from the Lithuanian example in (4.23). In these cases the relations of opposition and correction are undercoded. In (4.23a) the SoAs are linked in an oppositive relation, whereas in (4.23b) the relation established is one of correction. In these languages, the counterexpectative contrast is coded by a dedicated marker. (4.23) Lithuanian, Baltic, Indo-European Petras miega o (a) Aš dirbu I work:PRS.1sg COORD Peter sleep:PRS.3sg ‘I work and Peter sleeps.’ (L.R., questionnaire) (b) Peter nesimoko savo kambaryje o Peter NEG:study:PRS.3sg his room:LOC COORD žaidžia sode play:PRS.3sg garden:LOC ‘Peter is not studying in his room but he’s playing in the garden.’ (V.Ž., questionnaire) The same strategy for opposition and correction is also attested in Hausa and Japanese, but in this case the construction is the same also for all combination relations. What happens in these cases is that the same strategy is employed to express all combination and contrast relations, except for the counterexpectative one. In (4.24) from Japanese, the use of the general converb -te for the expression of the corrective contrast is shown.

134 Contrast relations (4.24) Japanese tyuumonsi-ta-no-wa kootya-de-naku-te koohii-desu order-PRF-NR-TOP tea-COP-NEG-COORD coffee-COP ‘What I ordered is not tea, but it’s coffee.’ (Y.N., questionnaire) More will be said about such cases in section 6.2, where the constructions coding combination and contrast relations will be compared. For now, sufﬁce it to remark that corrective and oppositive relations are strongly undercoded in these languages, since the same connective is used for a variety of cooccurrence relations, and semantic speciﬁcations identifying the particular relation at issue are left to inferential enrichment. As shown by the ﬁrst ﬁve languages in Table 13, whatever strategy is used to code opposition, corrective and counterexpectative contrast may be coded either by dedicated markers, as in German, Serbo-Croatian, Hungarian and Spanish among others, or by a general marker used for both, as in French and Polish on the Table. In (4.25) the contrast markers attested in Hungarian are exempliﬁed. No general marker for correction and denial of an expectation is available, but rather two dedicated markers are used. (4.25) Hungarian, Finno-Ugric, Uralic (a) Nagyon szomjas vagyok, de/*hanem nem szeretem a very thirsty be.1sg COORD NEG love.1sg DEF narancslét. orange.juice ‘I’m very thirsty but I don’t like orange juice.’ (F.M., questionnaire) (b) Péter nem a szobában tanul, hanem/*de a Péter NEG DEF room.INESS study.3sg, COORD DEF kertben játszik. garden.INESS play.3sg ‘Peter is not studying in his room but he’s playing in the garden.’ (F.M., questionnaire) In (4.26), on the other hand, the use of a general marker is exempliﬁed from Portuguese. As can be observed in (4.27), in Italian the two possibilities coexist: ma is a general marker for corrective and counterexpectative contrast,

The coding of contrast relations

135

però is a dedicated counterexpectative marker and bensì is a dedicated corrective one. (4.26) Portuguese, Romance, Indo-European sede, mas de sumo não gosto (a) Tenho have.PRS.1sg thirst COORD NEG like.PRS.1sg of juice de laranja of orange ‘I’m thirsty but I don’t like orange juice.’ (A.R., questionnaire) (b) O

no quarto a estudar, Pedro não está Pedro NEG stay.PRS.3sg in room to study mas a brincar no jardim. COORD to play in garden ‘Peter is not studying in his room but he’s playing in the garden.’ (A.R., questionnaire) DEF

(4.27) Italian, Romance, Indo-European volentieri ma/però/*bensì non a. Stasera verrei NEG tonight come.COND.1sg with.pleasure COORD posso can.PRS.1sg ‘I would love to come tonight, but I can’t.’ b. Giovanni ha non di raggiungerlo detto John AUX . PRS:3sg tell: PST. PTCP NEG of reach:him in ufﬁcio ma/*però/bensì di aspettarlo qui in ofﬁce COORD of wait.for:him here ‘John didn’t tell me to reach him in his ofﬁce, but to wait for him here’ As is shown by the Russian and Hausa types in Table 13, if a general marker expressing only contrast relations is attested at all, it will be used for corrective and counterexpectative contrast. The use of general markers expressing opposition, as already discussed, may be extended to correction and denial of expectation, but they always express also discontinuous atemporal combination.

136 Contrast relations The last type in Table 13 is instantiated by languages like Tuvaluan, Supyire and Koromfe, which use the same construction for all types of contrast relations and for discontinuous atemporal combination. Thus, all the speciﬁc relations expressed by such general markers are undercoded and need to be inferred from the context. There is also one language in the Comparison sample, Tauya, in which all the contrast relations are expressed by means of the same strategy used for the two combination relations. In other words, the underspeciﬁcation in that case is even greater, since all coordination relations involving the cooccurrence of two SoAs are expressed by means of the same general construction and may thus be identiﬁed only through inferential enrichment. A wider semantic map comprising both combination and contrast coding strategies will be proposed in section 6.2, in order to illustrate the cases of overlap between the expression of combination and the expression of contrast. For now, the analysis of the semantic domains attested within the coding of contrast has revealed the ordered conceptual contiguity between the three contrast sub-relations. How a conceptual space can be interpreted has already been pointed out in chapter 1, but it will be further discussed in section 4.4 with reference to the contrast conceptual space identiﬁed so far. The conceptual space illustrated here is slightly different from the map proposed by Malchukov (2004), in that corrective contrast is located between oppositive and counterexpectative contrast; in Malchukov’s map, on the other hand, correction is only linked to opposition. Since Malchukov considers contrast relations together with combination relations, his model will be discussed in chapter 6, where a map resulting from the comparison of combination and contrast constructions will be proposed. Sufﬁce it to say, for now, that the differences existing between the two maps are due to a different deﬁnition of counterexpectative contrast.

4.2.3 The contrast coding complexity hierarchy Data concerning the morphophonological complexity of the attested contrast markers are reported in Tables 17 and 18 in this chapter. All attested markers are listed and classiﬁed on the basis of their free vs. bound nature, mono- vs. polysyllabic and mono- vs. polymorphemic structure. The exam of these data reveals another implicational pattern of variation characterizing the coding of contrast relations. This hierarchy conﬁrms

The coding of contrast relations

137

the asymmetry between form and function already pointed out by Kortmann (1997), according to which the more general a connective is, the simpler is its morphophonology. The contrast coding complexity hierarchy can be formulated as follows: (4.28) The contrast coding complexity hierarchy: general marker used also for a combination relation > general marker used only for contrast relations > dedicated marker for counterexpectative contrast, dedicated marker for corrective contrast, dedicated marker for oppositive contrast. If a language has one of the markers indicated on the hierarchy, it will be at least as morphophonologically complex as the markers to its left: - the use of a monomorphemic marker for the expression of a function on the hierarchy implies the use of a monomorphemic marker for the expression of the functions on the left; - the use of a monosyllabic marker for the expression of a function on the hierarchy implies the use of monosyllabic marker for the expression of the functions on the left; - the use of a bound marker for the expression of a function on the hierarchy implies the use of a bound marker for the expression of the functions on the left. In other words, dedicated markers encoding counterexpectative, oppositive and corrective contrast tend to be more complex than general markers used for more than one coordination relation. Moreover, in languages having both a general marker used also to express a combination relation (like Polish a) and a general marker employed only for contrast relations, the latter will express correction and denial of an expectation (like Italian ma, or Polish ale) and will be morphophonologically more complex than the connective employed also for combination. In Table 14 the attested complexity conﬁgurations are arranged in the same order as they are in the hierarchy, in order to highlight the increasing morphophonological complexity along the hierarchy. Each of the attested conﬁgurations is illustrated by one language. Examples from the European languages are more numerous, because of the greater comprehensiveness of the information collected. However, as can be observed in Table 18, nonEuropean languages conﬁrm this hierarchy and no counterexamples have been found.

138 Contrast relations This hierarchy reveals a major difference in complexity between dedicated markers and general ones. No attested general contrast marker is polymorphemic, for instance, while some dedicated markers, especially the oppositive ones, consist of more than one morpheme (cf. French tandis que, English whereas). Moreover, dedicated markers are more transparent and it is often possible to reconstruct their internal morphology. The two corrective markers sino in Spanish and hanem in Hungarian, for instance, have the same internal structure ‘if + not’ (si and ha mean ‘if’, no and nem mean ‘not’). Table 14: The contrast coding complexity hierarchy: attested complexity patterns.– = absence of the relevant marker

German: Basque: Finnish: Georgian: Italian: Albanian: French: Hungarian: Chechen: Russian: Japanese: Polish: Koromfe:

General also for comb. – – – – – – – men tq’a a -te, -i a kae

> > > > > > > > > > > > >

General only for contrast – – – – ma por mais – – – – ale –

> > > > > > > > > > > > >

Dedicated counterx.

Dedicated corr.

Dedicated opp.

aber baina mutta magram però – – de amma no -ga – –

sondern baizik vaan aramed bensì – – hanem – – – – –

während – – k’i mentre kurse tandis que pedig – – – – –

Finally, dedicated markers tend to be polysyllabic, while general markers, especially the ones used also for some combination relation, are monosyllabic. In particular, all attested general markers used also to express discontinuous atemporal combination consist of just one syllable (Slavic a, Lithuanian o, Supyire sì), while the general markers for corrective and counterexpectative contrast may either be monosyllabic - like English but - or polysyllabic like Hocak ˛ nu˛ni˛ge. As for the bound or free nature of contrast markers, the picture is different within and outside Europe. European languages only have free connectives, except for Dargi which has the bound counterexpectative marker -gu co-existing with the free form amma. Outside Europe, as Table 18 shows,

The coding of conﬂicting SoAs

139

bound markers are instead attested and their distribution follows the pattern pointed out in the hierarchy in (4.28).

4.3

The coding of conﬂicting SoAs

In the analysis of the constructions used to express combination relations, the coding of the SoAs has turned out to follow a regular implicational pattern. In the coding of contrast, instead, this parameter does not reveal well supported patterns of variation. The syntactic parallelism of constructions expressing contrast is not random, but data are not enough to establish any implicational pattern. Data regarding the syntactic parallelism of the constructions coding contrast are reported in Table 19, in which information concerning the Comparison sample is presented. There is no table for the European sample, because all the constructions attested in Europe to express contrast are syntactically parallel (but see the general Table 53 in the Appendix). Given the internal homogeneity of European languages, this parameter is only applied to nonEuropean languages. Syntactically parallel contrast constructions are the majority also in the Comparison sample. However, there are some languages which use nonparallel constructions for the expression of particular types of contrast. Japanese and Jamul Tiipay may express opposition and correction with a non-parallel construction, but must use a parallel strategy in the expression of counterexpectative contrast. Tauya, Korean and West Greenlandic, on the other hand, are able to code all contrast relations by means of a syntactically non-parallel construction. Finally, no examined language uses a non-parallel construction for counterexpectative contrast without using a non-parallel construction also for the expression of corrective and oppositive relation. Even though three language types are attested and the fourth possible one is not, the distribution of parallel and non-parallel constructions in the expression of contrast only highlights the clear cross-linguistic tendency to code two conﬂicting SoAs by means of balancing strategies. The number of languages using a non-parallel construction at all is too limited to establish a valid implication. Let us now examine the few cases in which non-parallel constructions are used to code two conﬂicting SoAs. As already said, the three languages having the possibility to use nonparallel constructions in the expression of all contrast types are Tauya, Korean

140 Contrast relations and West Greenlandic. Tauya expresses all contrast types by means of the general construction used also for all combination relations, which is syntactically non-parallel. The speciﬁc relations expressed are thus identiﬁed each time through inferential enrichments. Korean employs the general combination construction only in the expression of opposition, and has a dedicated construction (the sufﬁx -(u)na) to convey a contrast relation generated by the denial of an expectation. Both the strategy used for combination and the strategy used for contrast are syntactically non-parallel, even though they have correspondent parallel constructions which make use of free connectives (kuliko for combination and kulena for contrast). Finally, West Greenlandic shows a non-parallel general construction for corrective and counterexpectative contrast. As can be observed in (4.29), the ﬁrst SoA is coded by a balanced verb form, while the second is in the contemporative mood and the contrastive enclitic li is attached to the ﬁrst constituent of the second clause. This non-parallel construction is only possible when the subjects of the two clauses are co-referential, otherwise another strategy must be used, with the free marker kisianni occurring between two balancing clauses. (4.29) West Greenlandic, Aleut, Eskimo qimmiq taanna nakuarsuu-vuq saamasuu-lluni=li dog that be.strong-3sg.IND be.gentle-4sg.CONT=COORD ‘That dog is strong but it’s gentle’ (Fortescue 1984: 123) Two languages use non-parallel constructions only for the expression of opposition and correction: Japanese and Jamul Tiipay. Let us examine an example from Japanese in (4.30). As can be observed in (4.30a), Japanese expresses the oppositive contrast by means of the general verbal sufﬁx -i, which is used also to express discontinuous atemporal combination and is attached to a deranked verb form. Corrective contrast is expressed by means of the verbal sufﬁx -te, as illustrated in example (4.24), which is also employed to express all relations of combination and opposition. On the other hand, counterexpectative contrast in (4.30b) is coded by the overt marker -ga, which is a bound morpheme, but is attached to a ﬁnite verb form and thus occurs in a parallel construction, even if the subjects of the SoAs are co-referential.

Functional motivations in the coding of contrast

141

(4.30) Japanese (a) watashi-wa hatarak-i piitaa-wa neru 1sg-TOP work-COORD Peter-TOP sleep ‘I work and Peter sleeps!’ (Y.N., questionnaire) kawaki-masi-ta-ga orenzizyuusu-wa (b) nodo-ga Throat-NOM dry-HON-PRF-COORD orange.juice-TOP suki-dewaarimasen like-NEG ‘I’m very thirsty, but I don’t like orange juice.’ (Y.N., questionnaire) A similar situation is attested in Jamul Tiipay, where the expression of oppositive and corrective contrast requires the use of the same switch-reference markers attested in constructions expressing combination, while the counterexpectative relation is coded by the sufﬁx -pe, attached to a ﬁnite verb form. Languages that do not use any non-parallel construction in the expression of contrast are the most widespread in both samples. The functional motivations underlying the strong cross-linguistic tendency for parallel contrast constructions will be discussed in the next section. 4.4

Functional motivations in the coding of contrast

The three implicational patterns identiﬁed in the coding of contrast relations, as well as the cross-linguistic tendencies in the syntactic parallelism of contrastive constructions, may be explained in terms of the same functional principle invoked in chapter 3 to explain the coding patterns of combination, namely the principle of syntagmatic economy. The contrast coding implication in (4.16) states that if an asyndetic strategy is used to express the counterexpectative contrast, an asyndetic strategy will also be available for the expression of corrective and oppositive contrast. This implication may be paraphrased in the following terms: if an overt marker is available for the expression of correction or opposition, the contrast generated by the denial of an expectation can be conveyed by means of an overt marker, too. This implication can be accounted for in terms of syntagmatic economy, according to which there is no need to overtly signal information that is recoverable form the context (Haiman 1985). This principle may be invoked

142 Contrast relations to account for the contrast coding implication insofar as the conﬂict underlying the three contrast relations is inferable from the simple juxtaposition of two SoAs. Some contrast relations are less easily inferable from the context, while others are easier to infer. Overt marking will be used for the former rather than for the latter. The two relations of opposition and correction are both characterized by a conﬂict inherent in their semantics. In the case of opposition, the conﬂict depends on the somehow antonymic relation existing between the two SoAs, which are presented as different, if not opposite, facets of the same scene. In sentence (4.31) for instance, the two actions of working and relaxing denote the two poles of a hypothetical scale which goes from activity to non-activity. The conﬂict existing between them is thus easily inferable from a context of discourse where the common integrator (cf. discussion in section 3.1) sets the background for their antonymic relation. (4.31) (Tomorrow we have a conference...) I am WORKING, YOU are RELAXING in front of the TV! In the case of correction, the conﬂict is determined by the opposite polarity of the two clauses. The ﬁrst SoA is overtly negated, while the second is positively asserted as a substitute of the ﬁrst one. Given a common integrator available in the context of communication, that is, a common topic, the juxtaposition of a negative and a positive SoA easily gives rise to an inference of conﬂict, especially if it is accompanied by some emphatic intonation or marker (cf. the Georgian example in (4.18a)). Moreover, the two SoAs normally denote situations that are antithetic with respect to the common integrator. A sentence like (4.32) easily generates an inference of substitution: given the common integrator ‘reasons why he came to London’, the ﬁrst SoA is overtly denied and the second is presented as the correct situation. (4.32) He did not come here to visit London; he came for a conference. Counterexpectative contrast, on the other hand, is less easy to infer from the simple juxtaposition of two SoAs. In this case the conﬂict is not inherent in the semantics of two somehow antithetic SoAs, but originates from a contradiction between the semantics of one SoA and some expectation activated by the other SoA or by the context of communication, identiﬁed through an inferential process.

Functional motivations in the coding of contrast

143

As Lang (2000: 248-249) points out, the presence of an overt marker makes the presence of a conﬂict explicit and furthermore tells the hearer that he has to look for an assumption contrasting with the second SoA. The absence of a device indicating the counterexpectative contrast may lead the hearer to establish a relation between the two SoAs as such, and not between one of them and the expectations accompanying the other SoA. Let us consider for instance the sentence in (4.33): (4.33) The UN forces have arrived to Lebanon, ten civilians died this morning. Even if the speaker’s intent is to establish a counterexpectative contrast between the SoAs in (4.33), the hearer may happen to establish a relation between the SoAs as such and this relation may involve no conﬂict at all. The sentence in (4.33) may be interpreted in a completely different way. For instance, following the iconic order in which the events are presented, the death of the civilians could be interpreted as caused by the UN forces after their arrival. In that case, the inferred relation would be one of sequential combination, or one of causal chain. The presence of the connective but between the two SoAs, on the other hand, makes the presence of a conﬂict explicit and tells the hearer that the second SoA is to be taken as a source for a contrastive assumption. The hearer may then establish a relation between the second SoA and some expectation generated by the ﬁrst one, namely that the arrival of UN forces in Lebanon interrupted the virulence of the war. (4.34) The UN forces have arrived to Lebanon, but ten civilians died this morning. From this discussion it should be clear why languages that have only one dedicated contrast marker will employ it to encode the denial of an expectation, rather than an opposition or a correction. The counterexpectative contrast is indeed more difﬁcult to infer from simple juxtaposition and is thus more likely to receive overt indicators. Let us now examine the contrast conceptual space identiﬁed on the basis of the attested semantic domains. At the basis of the interpretation of any semantic map or space is the iconic principle according to which ‘recurrent similarity in form reﬂects similarity in meaning’ (Haiman 1985: 26). The

144 Contrast relations recurrent use of the same marker for two different relations is thus taken as an indicator of the functional proximity of the two relations. The location of the various types of contrast along the semantic maps reported in Table 13 thus mirrors their functional contiguity. This correspondence is also proved by the fact that if a marker widens its function and encompasses new relations, the ﬁrst relations to be included will be the ones contiguous on the map. Consequently, if two non-contiguous relations are expressed by the same construction, this construction will also be able to express the intermediate one.25 Two coordination relations are close to each other or distant on the contrast conceptual space depending on whether they share some functional features, are often associated in discourse or are easily inferred one from the other. Let us then examine what features determine the actual location of the relations on the conceptual space. Opposition is followed by correction because the two relations have two aspects in common: they imply an atemporal combination of discontinuous SoAs and they are characterized by a conﬂict the origin of which lies in the antonymic semantics of the linked SoAs. The antonymic conﬂict typical of correction is determined by the opposite polarity of the linked SoAs, since the ﬁrst one is overtly denied and the second one is positively asserted as the substitute. Moreover, a corrective relation is established between two SoAs when the ﬁrst one has been previously asserted in the context of communication or if there is a shared expectation for it to be true. The corrective relation thus consists of the explicit denial of that expectation, with the substitution of the wrong SoA with the correct one. The presence of a denial of an expectation in the discourse is also shared by the counterexpectative relation, and this common functional property motivates the close location of corrective and counterexpectative contrast in the conceptual space. These two relations indeed share the presence of an expectation to deny, while opposition is the only contrast relation lacking it. In corrective contrast the expectation is explicitly denied, while in counterexpectative contrast the expectation is implicit in the sentence itself or in the context. The conﬂict characterizing counterexpectative contrast does not depend on the antonymic semantics of the linked SoAs, but only on the denial of an implicit assumption, and this is what motivates the distance between counterexpectative and oppositive contrast on the conceptual space (cf. further discussion on this topic in chapter 6).

&5 64 /08 130$&&% 8*5) 5)& .05*7"5*0/4 6/%&3-:*/( 5)& $0/53"45 $0%*/( $0.1-&9*5: )*&3"3$): "$$03%*/( 50 8)*$) %&%*$"5&% $0/53"45 ."3,&34 "3& "5 -&"45 "4 $0.1-&9 "4 5)& (&/&3"- ."3,&34 &913&44*/( .03& 5)"/ 0/& $0/53"45 3&-"5*0/ 635)&3.03& (&/&3"- ."3,&34 &/$0%*/( $033&$5*7& "/% $06/5&3&9 1&$5"5*7& $0/53"45 "3& "5 -&"45 "4 $0.1-&9 "4 5)& (&/&3"- ."3,&34 64&% "-40 50 &913&44 40.& $0.#*/"5*0/ 3&-"5*0/ )*4 1"55&3/ 0' $3044-*/(6*45*$ 7"3*"5*0/ $0/'*3.4 035."//<4 "44&44.&/5 0' 5)& '03.'6/$5*0/ "4:..&53: 035."// */ "$$03%"/$& 50 8)*$) 5)& .03& (&/&3"- " $0//&$5*7& *4 *& 5)& .03& 3&-"5*0/4 *5 .": &9 13&44 5)& -08&3 *4 *54 %&(3&& 0' .031)01)0/0-0(*$"- $0.1-&9*5: 035."// )"4 7&3*'*&% 5)*4 */%*3&$5 1301035*0/"- $033&410/%&/$& '03 5)& "55&45&% 630 1&"/ 46#03%*/"5034 / 5)*4 456%: 5)& &9". 0' 5)& "55&45&% $0/53"45 ."3,&34 1307*%&4 '635)&3 &7*%&/$& '03 5)*4 (&/&3"-*;"5*0/ )& 13*/$*1-& "5 803, */ 5)& $0/53"45 $0%*/( $0.1-&9*5: )*&3"3$): *4 !*1'<4 "8 0' ##3&7*"5*0/ 0' 03%4 8)*$) *4 */ 563/ " ."/*'&45"5*0/ 0' 5)& .03& (&/&3"- 13*/$*1-& 0' 4:/5"(."5*$ &$0/0.: 4 !*1' "3(6&4 5)& '3&26&/$: 0' 64& 0' " -*/(6*45*$ &913&44*0/ $033&-"5&4 8*5) *54 '03."- $0.1-&9*5: 8*5) *54 4&."/5*$ 7&34"5*-*5: 8*5) 5)& /6.#&3 0' $0/5&954 */ 8)*$) *5 .": 0$$63 "/% 8*5) *54 "(& $' 035."// / 05)&3 803%4 5)& .03& (&/&3"- " $0//&$5*7&*4 5)& )*()&3 "3& 5)& /6.#&3 0' $0/5&954 */ 8)*$) *5 .": 0$$63 "/% 5)& /6.#&3 0' 3&-"5*0/4 *5 .": &913&44 0/4&26&/5-: 5)& .03& (&/&3"- " ."3,&3 *4 5)& .03& '3&26&/5 *5 8*-- #& */ %*4$0634& 54 1)0/0-0(*$"- 46#45"/$& 8*-- 5)64 #& &30%&% -&"%*/( 50 .03 1)01)0/0-0(*$"--: 4*.1-& '03.4 #7*064-: 5)& .03& "/$*&/5 " ."3,&3 *4 5)& .03& 130#"#*-*5*&4 *5 )"4 50 %&7&-01 '635)&3 '6/$5*0/4 "/% 5)64 50 8*%&/ *54 .&"/*/( 45"35*/( 5)& $)"*/ 0' 1)&/0.&/" +645 -*45&% ' 5)& "55&45&% $0/53"45 ."3,&34 5)& (&/&3"- 0/&4 64&% "-40 '03 40.& $0. #*/"5*0/ 3&-"5*0/4 4)08 5)& )*()&45 4&."/5*$ 7&34"5*-*5: "/% "3& 5)& .045 '3& 26&/5 */ %*4$0634& 4*/$& 5)&: "3& 64&% 50 &913&44 5)& 7&3: #"4*$ "/% '3&26&/5 3&-"5*0/ 0' $0.#*/"5*0/ )&4& ."3,&34 "3& .031)01)0/0-0(*$"--: 5)& 4*. 1-&45 ."3,&34 64&% 50 &913&44 $0/53"45 $0/4*45*/( 0' 0/& 4:--"#-& "/% 0/& .031)&.& )& 4*.1-*$*5: $)"3"$5&3*;*/( 5)&4& $0//&$5*7&4 8*-- #& %*4$644&% */ %&15) */ 4&$5*0/ 8)&3& 5)& $0.1"3*40/ 0' $0.#*/"5*0/ "/% $0/53"45 $0%*/( 8*-- #& %&"-5 8*5) )& .03& " ."3,&3 3&%6$&4 5)& 3"/(& 0' 3&-"5*0/4 *5 .": &913&44 "/% 5)& -08&3 5)& */53*/4*$ 7&34"5*-*5: 0' 5)& 3&-"5*0/4 &913&44&% *4 5)& .03& *54 .03 1)01)0/0-0(*$"- $0.1-&9*5: */$3&"4&4 "4 " 3&46-5 0' *54 -08&3 '3&26&/$: 0'

146 Contrast relations occurrence. Markers used for the expression of the corrective and the counterexpectative contrast denote semantic relations that are more speciﬁc than the basic atemporal combination and have a lower intrinsic versatility. They are therefore less frequent than general markers used also for combination, but at the same time they are simpler than dedicated contrast connectives. Dedicated markers are only used for one relation and are thus less frequent and less semantically versatile. To conclude, as pointed out in section 4.3, there is a cross-linguistic tendency for the use of parallel constructions in the coding of two conﬂicting SoAs. Bearing in mind what has been said about the syntactic parallelism of constructions used for combination (section 3.4), it is plausible to relate the widespread syntactic parallelism of the attested contrast constructions to the same economic explanation. Basically, as can be observed in (4.35) and (4.36), contrast relations do not predetermine any semantic properties of the SoAs they link, which may be either simultaneous or sequential and may have the same or different aspect. Moreover, conﬂicting SoAs may share participants, as in (4.35) or may have different subjects, as in (4.36). (4.35) I would love to come, but I just received the order to remain here. (4.36) Mike always works a lot but the Professor has given him a bad note. Therefore, even if dedicated markers were used, no information relative to the semantic properties of the SoAs would be recoverable from the context, simply because contrast relations do not predetermine any aspect of the SoAs they link. As a result, constructions used to express contrast are most frequently parallel. The only cases in which no parallel construction is available at all are those languages which use the same very general non-parallel construction for the expression of both combination and contrast relations, like Tauya.

Language data

4.5

147

Language data

Table 15: European Sample. Coding of contrast relations: Presence of overt coordinating markers and their semantic domain. – = absence of an overt marker; d= dedicated; Gx = general, with x identifying a given construction attested in the language; blank= no information available Language

Oppositive contrast

Corrective contrast

Counterex contrast

Albanian Basque Belorussian Bulgarian Catalan Chechen Czech Danish Dargi Dutch English Estonian Finnish French Georgian German Greek Hungarian Icelandic Irish Italian Latvian Lezgian Lithuanian Luxembourgish Maltese Norwegian Polish Portuguese Romanian Russian Sardinian Serbo-Croatian Spanish

Gα /d Gα Gα Gα Gα Gα Gα Gα

Gβ d Gβ Gα d Gα Gβ Gβ

Gα Gα /d Gα Gα Gα /d -/d Gα /d Gα Gα Gα Gα Gα /d Gα

Gβ Gβ d d Gβ -/d d Gβ d d Gβ Gβ /d Gβ

Gα Gα Gα Gα Gα Gα Gα Gα Gα Gα Gα

Gα Gβ Gβ Gβ Gβ Gβ d Gα Gβ d d

Gβ d Gβ d d d Gβ Gβ d Gβ Gβ d d Gβ d d Gβ d d Gβ Gβ /d Gβ d d Gβ Gβ Gβ Gβ Gβ d d Gβ d d

148 Contrast relations (Continued) Language

Oppositive contrast

Corrective contrast

Counterex contrast

Swedish Turkish Ukrainian

Gα Gα Gα

d Gβ Gβ

d Gβ Gβ

Table 16: Comparison Sample. Coding of contrast relations: Presence of overt coordinating markers and their semantic domain. – = absence of an overt marker; d= dedicated; Gx = general, with x identifying a given construction attested in the language; blank= no information available Language

Oppositive contrast

Corrective contrast

Counterex contrast

Arabic Dumi Hakha Lai Harar Oromo Hausa Hdi Hebrew Hocak ˛ Iraqw Jamul Tiipay Japanese Kisi Kol. Yukaghir Korean Koromfe Kuskokwim A. Lango Limbu Malayalam Mangarayi Maori Marathi Maricopa Meithei Mosetén NànáfwˆE Ndyuka Persian Rapanui Somali

Gα /d Gα

d Gα

d

Gα Gα

Gβ d Gα

Gα Gα /Gβ Gα Gα

Gα Gα -

-

d Gα Gα -

Gα Gα Gα Gα

Gβ d -

d Gβ d d Gα d d d d Gα Gα d d d d Gα Gα d d Gβ d d d d

Language data

149

(Continued) Language

Oppositive contrast

Corrective contrast

Counterex contrast

Supyire Tauya Tukang Besi Tuvaluan Vietnamese Wari’ W.Greenlandic

Gα Gα Gα d -

Gα Gα Gα Gα Gα Gα

Gα /d Gα Gα Gα Gα /d Gα

Table 17: European Sample. Coding of contrast relations: morphophonological complexity of the attested coordinating markers. + = presence of the given feature; - = absence of the given feature; opp.= opposition; atemp.= atemporal combination; comb.= combination; cor= correction; coun= counterexpectative contrast; blank= no information available Language

Marker

Semantic domain

Free

Polysyllabic

Polymorphem.

Albanian

por kurse baina baizik a alĲe a

coun., cor. contrast opp. contrast coun. contrast cor. contrast atemp.comb., opp. coun., cor. contrast atemp.comb., opp., cor. coun. contrast coun. contrast cor. contrast atemp.comb., opp., cor. coun. contrast coun., cor. contrast coun., cor. contrast coun. contrast coun. contrast coun., cor. contrast coun., cor. contrast opp. contrast opp. contrast coun. contrast coun. contrast cor. contrast

+ + + + + + +

+ + + -

-

+ + + +

+ + -

-

+ + + + + + + + + + +

+ + + + + -

+ -

Basque Belorussian Bulgarian

Catalan Chechen

Czech Danish Dargi Dutch English

Estonian

no pero sino tq’a amma ale men -gu amma maar but while whereas kuid aga vaid

150 Contrast relations (Continued) Language

Marker

Semantic domain

Free

Polysyllabic

Polymorphem.

Finnish

mutta vaan mais tandis que aramed magram ki aber sondern während alla meg de hanem en heldur ach ma però bensì bet amma o

coun. contrast cor. contrast coun., cor. contrast opp. contrast

+ + + +

+ +

+

cor. contrast coun. contrast opp. contrast coun. contrast cor. contrast opp. contrast coun., cor. contrast atemp.comb., opp. coun. contrast cor. contrast coun. contrast cor. contrast coun., cor. contrast coun., cor. contrast coun. contrast cor. contrast coun., cor. contrast coun. contrast atemp.comb., opp., cor. coun. contrast coun., cor. contrast coun., cor. contrast coun., cor. contrast atemp.comb., opp. coun., cor. contrast coun., cor. contrast atemp.comb., opp. coun. contrast cor. contrast atemp.comb., opp., cor. coun. contrast coun., cor. contrast atemp.comb., opp. coun. contrast

+ + + + + + + + + + + + + + + + + + +

+ + + + + + + + + + + -

+ -

+ + + + + + + + + + +

+ + -

-

+ + + +

+

-

French

Georgian

German

Greek Hungarian

Icelandic Irish Italian

Latvian Lezgian Lithuanian

Luxembourgish Maltese Norwegian Polish Portuguese Romanian

Russian

Sardinian Serbo-Croatian

bèt mee imma men a ale mas iar dar ci a no ma a ali

Language data

151

(Continued) Language

Spanish Swedish Turkish Ukrainian

Marker

Semantic domain

Free

Polysyllabic

Polymorphem.

n‚ego v‚ec´ pero sino men utan fakat ama a alĲe

cor. contrast cor. contrast coun. contrast cor. contrast coun. contrast cor. contrast coun., cor. contrast coun., cor. contrast atemp.comb., opp. coun., cor. contrast

+ + + + + + + + + +

+ + + + + + +

-

Table 18: Comparison Sample. Coding of contrast relations: morphophonological complexity of the attested coordinating markers. + = presence of the given feature; - = absence of the given feature; opp.= opposition; atemp.= atemporal combination; comb.= combination; cor= correction; coun= counterexpectative contrast; blank= no information available Language

Marker

Semantic domain

Free

Polysyllabic

Polymorphem.

Arabic

laakin ﬁè¯i bal

coun. contrast opp. contrast cor. contrast

+ + +

+ + -

-

k’ófáamô immó àmmaa àmá aval éla nu˛ni˛ge

coun. contrast coun. contrast coun., cor. contrast coun. contrast coun. contrast cor. contrast coun., cor. contrast

+ + + + + + +

+ + + + + + +

+ -

-pes -i -ga k´E (mi)

coun. contrast atemp.comb., opp. coun. contrast coun. contrast none coun. contrast coun. contrast coun. contrast

+

-/+

-

+ -

+ + -

-

Dumi Hakha Lai Harar Oromo Hausa Hdi Hebrew Hocak ˛ Iraqw Jamul Tiipay Japanese Kisi Kol. Yukaghir Korean

kulena -manun -(u)na

152 Contrast relations (Continued) Language

Marker

Semantic domain

Free

Polysyllabic

Polymorphem.

Koromfe

la

+

-

-

Kuskokwim A. Lango Limbu Malayalam

ĳedinh E´ ntˆO kere pakùe ennaal eNkilum gana engari pan. parãntu

atemp.comb., opp., cor., coun. coun., cor. contrast coun. contrast coun. contrast coun. contrast coun. contrast coun. contrast coun. contrast coun., cor. contrast coun., cor. contrast coun., cor. contrast

+ + + + + + + + + +

+ + + + + + + + +

+ -

+ + + + + + + +

+ + + + + + -

+ -

+

+

-

+ +

+ -

-

+ + +

-

-

+

+

-

Mangarayi Maori Marathi Maricopa Meithei Mosetén NànáfwˆE Ndyuka Persian Rapanui Somali Supyire

adupu -tsa’ -ki sàng´E ma ´ vœli ´ bœlke pero -se laakìin sì ï` kàà

Tauya Tukang Besi Tuvaluan Vietnamese

Wari’ W.Greenlandic

toka kae mà nhung còn =li kisianni

coun. contrast coun. contrast opp. contrast coun., cor. contrast coun. contrast coun. contrast cor. contrast coun. contrast coun. contrast coun. contrast atemp.comb., opp., cor., coun. coun. contrast none coun., cor. contrast atemp.comb., opp., cor., coun. coun., cor. contrast coun. contrast opp. contrast none coun., cor. contrast coun., cor. contrast

Language data

153

Table 19: Comparison Sample. Coding of two contrasted SoAs: syntactic parallelism of the attested constructions. + = syntactically parallel; - = syntactically non-parallel; d= dedicated; Gx = general, with x identifying a given construction attested in the language; jx= juxtaposition; blank= no information available Language

Oppositive contrast

Corrective contrast

Counterex. contrast

Arabic Dumi Hakha Lai Harar Oromo Hausa Hdi Hebrew Hocak ˛ Iraqw Jamul Tiipay Japanese Kisi Kol. Yukaghir Korean Koromfe Kuskokwim A. Lango Limbu Malayalam Mangarayi Maori Marathi Maricopa Meithei Mosetén NànáfwˆE Ndyuka Persian Rapanui Somali Supyire Tauya Tukang Besi Tuvaluan Vietnamese Wari’ W.Greenlandic

Gα +/d+ jx+

d+ jx+

d+

Gα + Gα + Gα +

Gβ + Gα + d+ Gα +

Gα Gα -/+ /Gβ jx+

Gα Gα jx+

Gα +/Gα + jx+ jx+ jx+ jx+

d+ jx+/ Gβ + jx+ Gα + jx+ Gα + Gα Gα + d+ jx+

Gα + Gα + jx+ jx+ jx+ jx+/Gα + Gα +

Gα + jx+ d+ jx+ Gα + Gα Gα + Gα + Gα + jx+ Gα +/-

d+ Gβ + d+ d+ Gα + d+ d+ d+ jx d+/Gα + Gα + jx+/d+ jx+/d+ d+ d+ Gα + Gα + d+ d+ Gα + d+ d+ jx+/d+ d+ Gα +/d+ Gα Gα + Gα + Gα +/d+ jx+ Gα +/-

Chapter 5 Alternative relations This chapter deals with the coding of alternative relations, taking into account both the presence and nature of overt markers and the syntactic parallelism of the attested constructions. In section 5.1 a deﬁnition of the notion of alternative relation is ﬁrst of all provided and section 5.1.2 discusses the types of alternative that can be identiﬁed on the basis of the parameter of aim. The exam of the two parameters of coding of the relation and coding of the SoAs mirrors the picture outlined for the coding of contrast relations. The analysis of the attested coordinating markers will reveal implicational patterns of variation, whereas the coding of the SoAs will only highlight some cross-linguistic tendencies. The two implications identiﬁed with respect to the coding of the alternative relation are discussed in section 5.2. The comparison of syndesis and asyndesis in the expression of the choice-aimed and simple alternative will reveal the alternative coding implication. A detailed exam of how alternative is expressed in absence of an overt connective will furthermore highlight the close connection between the alternative relation as such and the notion of irreality, giving rise to a second implicational pattern (see also Mauri 2008). Since the analysis of the morphophonological complexity of the attested coordinating markers does not reveal any implicational pattern, this aspect will be covered within the discussion of the other two implicational patterns. Section 5.3 examines the syntactic parallelism of the attested constructions and points out the widespread tendency to employ parallel strategies to express alternative relations. Finally, the functional explanations underlying the attested implicational patterns of variation will be discussed in section 5.4 and it will be argued that syntagmatic economy is the major principle at work in the cross-linguistic coding of alternative relations. All data on the coding of alternative relation are presented in section 5.5, even though smaller tables will also be provided in the discussion of the implicational patterns, in order to make the regularities more evident.

+74 =>C8>= >5 0;C4A=0C8E4 A4;0C8>=

" " "$ "

@H9FB5H=J9 5G BCB7CC77IFF9B79 "B =; H<9 HKC 8=A9BG=CBG C: 7CC77IFF9B79 5B8 BCB7CC77IFF9B79 HC BCD2: 8= C74 CA052 74 6>C ;>BC 0;>=6 C74 F0H 74 A0= >DC >5 60B 6IH <9 A5B5;98 HC 5FF=J9 =B H=A9 -<9 85G<98 6CL 8=A9BG=CB CB K<=7< H<9 J5F=CIG DCGG=6=@=H=9G K<=7< 8C BCH 7CC77IF GH5B8 5G F9D@57956@9 ,CG

86DA4 -<9 7CC77IFF9B79 JG BCB7CC77IFF9B79 CDDCG=H=CB H<9 BCB7CC77IFF9B79 8=A9BG=CB

-<9 9@9A9BHG @=B?98 =B 5 F9@5H=CB C: BCB7CC77IFF9B79 69@CB; HC H<9 G5A9 G9H 6M J=FHI9 C: H<9 7CAACB :IB7H=CB H<9M G<5F9 = 9 H<9M 75B 5@@ C77IF =B H<9 G5A9 7CBH9LH 6IH BCH 5H H<9 G5A9 H=A9 "B 5 ;=J9B 7CBH9LH CB9 A9A69F C: H<9 D5F58=;A G9H =G GHFI7HIF5@@M 5B8 :IB7H=CB5@@M F9D@57956@9 K=H< 5BCH<9F ": H<9 9@9A9BHG @=B?98 5F9 ,CG H<9M 7CBGH=HIH9 5 G9H C: 9EI5@@M GIDDCG98 DCGG=6=@=H=9G 5B8 H<9 GD95?9F 8C9G BCH ?BCK K<=7< CB9 <5G C77IFF98 CF K=@@ C77IF =J9B H<9 K9@@ ?BCKB @=B95F=HM C: H<9 @=B;I=GH=7 G=;B !5=A5B @=B;I=GH=7 7CAAIB=75H=CB =G CB@M F95@=N98 6M >ILH5DCG=B; @=B;I=GH=7 9LDF9G

;C4A=0C8E4 A4;0C8>=B G=CBG CB9 5:H9F H<9 CH<9F =B 5 7CC77IFF9B79 8=A9BG=CB G=B79 KCF8G 5B8 G9B H9B79G 7CA6=B9 :C@@CK=B; 957< CH<9F 5B8 =H =G BCH DCGG=6@9 HC DFC8I79 ACF9 H<5B CB9 G=;B G=AI@H5B9CIG@M -<9 F9DF9G9BH5H=CB C: H<9 BCB7CC77IFF9B79 8=A9BG=CB =B H<9 @5B;I5;9 H<9B KCI@8 G99A HC 69 DF97@I898 6M a@=A=H5H=CBG C: H<9 A98=IAb !5=A5B G=B79 H<=G 8=A9BG=CB 65G=75@@M 7CB79FBG F9@5H=CBG 8= 01B4=C80 = 9 69HK99B 56G9BH 9@9A9BHG CF =FF95@=G ,CG " DFC DCG9 <9F9 HC 7CBG=89F H<9 5@H9FB5H=J9 F9@5H=CB 5G H<9 9@56CF5H=CB C: H<9 BCB 7CC77IFF9B79 8=A9BG=CB CB H<9 7CC77IFF9B79 CB9 B 5@H9FB5H=J9 F9@5H=CB 69HK99B HKC ,CG 7CA6=B9G H<9 9@9A9BHG C: 5 D5F58=;A5H=7 G9H 5G 7CC77IFF=B; 9@9A9BHG C: 5 @=B;I=GH=7 57H "B CH<9F KCF8G H<9 CB@M K5M @5B;I5;9 <5G HC 9LDF9GG 5 F9@5H=CB 8= 01B4=C80 =G 6M 7CA6=B=B; H<9 =BJC@J98 9@9A9BHG 8= ?A04B4=C80 5B8 G=;B5@@=B; H<5H H<9M 5F9 BCH HC 69 7CB G=89F98 5G 7CC77IFF=B; 6IH H<9M 5F9 F5H<9F DCH9BH=5@ GI6GH=HIH9G :CF 957< CH<9F 5B8 5F9 H
86DA4 @H9FB5H=J9 F9@5H=CB H<9 9@56CF5H=CB C: H<9 BCB7CC77IFF9B79 8=A9BG=CB CB H<9 7CC77IFF9B79 CB9

"B H<9 DF9798=B; 7<5DH9FG =H <5G 699B 5F;I98 H<5H HKC ,CG @=B?98 =B 5 7CC77IFF9B79 F9@5H=CB 9=H<9F 7CA6=B5H=CB CF 7CBHF5GH 5F9 5GG9FH98 CF 9@=7=H98 HC;9H<9F G=B79 K<5H =G 9GH56@=G<98 6M H<9 F9@5H=CB =G H<9=F 69=B; >C=BH@M J5@=8 -<=G A95BG H<5H H<9 5GG9FH=CB C: 5 7CA6=B5H=CB CF 7CBHF5GH F9@5H=CB 9BH5=@G H<5H 957< ,C =G 5@GC =B8=J=8I5@@M 5GG9FH98 (B H<9 7CBHF5FM H<9 5GG9FH=CB C: 5B 5@H9FB5H=J9 F9@5H=CB CB@M G5MG H<5H 5 G9H C: 9EI=J5@9BH 5B8 7CBHF5GH=B;

The notion of alternative relation

157

possibilities occurs, but it does not say anything about the occurrence of the individual possibilities. The existence of the set does not imply the occurrence of the SoAs that take part in it. The relation between the illocutionary force and alternative will be thoroughly discussed in the next section, in which two basic types of alternative relation will be identiﬁed. (5.1) He got stuck in the trafﬁc or he got lost along the way. The sentence in (5.1) is the assertion of an alternative relation between the two SoAs ‘He got stuck in the trafﬁc’ and ‘he got lost along the way’. This means that the speaker wants the hearer to assume that there is a set of equivalent possibilities and these possibilities consist of the two SoAs at issue. The function shared by the alternative SoAs, by virtue of which the paradigmatic set is established, is the common integrator of the relation, that is, the answer to the question ‘what are the possible things that might have happened to him?’. The two SoAs are possible instances of the common integrator ‘set of possible things that happened to him’. To sum up, an alternative relation is established between functionally parallel SoAs on the basis of a common integrator, just like combination and contrast, but unlike them it further speciﬁes the association in terms of noncooccurrence of replaceable possibilities (cf. Mauri 2008).

5.1.2 Alternative types Two alternative SoAs are equivalent possibilities and as such they are not located along the time axis. Therefore the parameter of temporality has a ﬁxed atemporal value in alternative relations. Moreover, two alternative SoAs stand in a paradigmatic contrast with each other. In other words, their condition of equivalent and replaceable possibilities determines the conﬂicting nature of the relation linking them (cf. Saussure 1983). If the two SoAs were not somehow conﬂicting, they would not constitute potential substitutes to each other. However, the parameter of conﬂict does not identify speciﬁc alternative subtypes. The alternative relation can be instead further distinguished into two types on the basis of the parameter of aim. As already stated, it distinguishes between two possible intentions on the part of the speaker. Speakers may establish the given relation in order to assert or elicit information about the block

158 Alternative relations of linked SoAs (i.e. the whole set that the relation delimits) or they may establish the relation in order to give or receive information about the single SoAs. If the relation established is an alternative one, speakers may pursue both aims: they may be interested in asserting or eliciting information about the set of options delimited by the alternative relation, or they may establish the alternative relation in order to assert or elicit information about the single SoAs. Yet, if they establish an alternative relation between two SoAs it means that they have no information about the reality status of the two SoAs, otherwise, if they had some, there would be no point in establishing an alternative relation between them. If the speaker’s aim in establishing the alternative is to put the single SoAs at issue, the only way they can do it is by asking for information about their reality value27 by means of interrogative sentences. They simply would have no data to assert something about the single SoAs, except for their standing in an alternative relation with each other. To sum up, a speaker who establishes an alternative relation may have two aims: assert or elicit information about the whole set of possibilities (examples (5.2) and (5.3)), or elicit information about the single alternatives (example (5.4)). (5.2) This summer we will make a trip to Russia or we will reach our friends in Sicily. (5.3) Is it the case that this summer you’ll make a trip to Russia or you’ll reach your friends in Sicily? (5.4) (What are you doing this summer?) Will you make a trip to Russia or will you reach your friends in Sicily? In the ﬁrst case ((5.2) and (5.3)), the SoAs are jointly at issue, and the speaker’s aim is to talk about a set of options. This type will be called simple alternative. In the second case (example (5.4)), the SoAs are considered separately, and the speaker’s aim is to ask for information about the single SoAs. Being the SoAs in alternative, the easiest way to elicit information about them is by asking for a choice, whereby one SoA is selected as realis and the other as irrealis. This type will be called choice-aimed alternative. The same two types have been identiﬁed by Dik (1968: 276) and Haspelmath (2007). Dik establishes the distinction in terms of manner. He argues

The notion of alternative relation

159

that the manner in which the alternative is presented determines a basic distinction that languages seem to encode: namely, the alternative relation can be ‘either A or B’ or ‘either A or B, which one?’. In the ﬁrst case the two SoAs are simply presented as alternatives, whereas in the second case the two SoAs are presented as alternatives and there is a need or request for a choice between the two. Haspelmath, instead, talks about standard and interrogative disjunction for the simple and the choice-aimed alternative, respectively. In a previous work I used this terminology too (Mauri 2008), but it turned out to be slightly confusing. First of all, it might suggest that all alternative relations occurring in interrogative sentences are choice-aimed, but this is not true, as can be observed in (5.3) and as will be further exempliﬁed below. Moreover, the label ‘standard’ suggests that the simple alternative is the normal one, whereas it is not more normal than the choice-aimed one. It is just simpler, lacking the necessity of a choice. It is true that choice-aimed alternative implies the setting of options and in this respect the simple alternative may be viewed as more basic. Yet, as will be explained in the course of this chapter, data does not support this hypothesis. The terms ‘choice-aimed’ and ‘simple’ are preferred to ‘interrogative’ and ‘standard’ because they are neutral with respect to how basic the relation is, and more transparent with respect to their distinguishing features. 5.1.2.1

Simple alternative

If the speaker’s intention is to talk about the whole set of alternatives, the relation is called simple alternative. In (5.5) and (5.6) two examples of this type are given. (5.5) Usually, I watch tv or I read until late at night. (5.6) (I don’t know why he is so late) He got stuck in the trafﬁc or he got lost along the way. In these sentences, the speaker does not commit to the reality value of the single SoAs, but only depicts them as equivalent possibilities. In (5.5) the two possibilities both occur, but not at the same time. The speaker’s interest is to assert the set of possibilities, not the speciﬁc occurrence of the single SoAs. In (5.6) instead, the speaker makes a conjecture about what has actually

160 Alternative relations happened and he imagines two possibilities, without knowing or caring about which one actually occurred or is going to occur. Once again, their aim is to hypothesize a set of options, not to ﬁnd out which one is a fact. It has been argued that the simple alternative can be established in questions as well, even though a set of options is more commonly asserted than elicited. Three different instances of interrogative sentences containing sets of options are given in (5.7), (5.8) and (5.9). (5.7) (Did I get it right?) Do you either make a trip to Russia or reach our friends in Sicily, this summer? - Yes, these are the two possibilities. (5.8) (So you have ﬁnally restricted the range of possibilities to two...) You either buy a book to your brother or you print a poster of his friends, right? - Exactly. - No: I either buy a book or I give him a nice picture of the two of us. (5.9) (I am tired, I don’t want to study this afternoon . . . ) Would you go for a walk with me, or have a cup of tea, or watch a movie, or . . . whatever? - Ok. In (5.7) and (5.8), the speaker’s intention is to get some information conﬁrmed, that is, they expect that the answer will be positive and that the set of options is asserted as real. These cases are therefore not neutral with respect to the speaker’s assumptions, although they both challenge the set of alternatives as a whole. In (5.9) instead, the speaker has no assumption regarding the reality value of the set of possibilities. They suggest the set of options and ask for its acceptance on the part of the hearer. They are not eliciting information about the single occurrence of one of them, that is, they are not interested in knowing which one of the SoAs will actually take place, but they just want the hearer to consider the whole range of alternatives and say whether it is real or not. 5.1.2.2

Choice-aimed alternative

If the speaker establishes an alternative relation between two SoAs with the aim of eliciting information about their individual occurrence, the relation is

The coding of alternative relations

161

called choice-aimed alternative, because they pursue their aim by asking for a choice between the possibilities. In (5.10) and (5.11) two examples of this relation are given. In both cases speakers simultaneously establish an alternative relation between SoAs and ask about which SoA will or did take place. (5.10) Do you come with us or do you stay here? - I will stay here, thanks. - I’ll come with you. (5.11) Did he come or was it a dream? - He came. - It was a dream! The assertion of either SoA can be a felicitous answer to the elicitation, and an answer which does not make a choice between the options but simply makes an assertion about the alternative relation (like ‘Ok’, or ‘No’) would not make sense. As argued in the preceding sections, a SoA-aimed communication is not possible if the established relation is one of combination or contrast. This fundamental difference between the two cooccurrence relations on the one hand and the alternative relation on the other hand is due to the assumed epistemic knowledge of the speaker. When they establish one of the two cooccurrence relations, they normally know the reality values of the SoAs and this allows them to assert their cooccurrence. If they establish an alternative relation, instead, it means that the speaker either does not have any knowledge about the reality value of the SoAs or they are not interested in making it explicit. They may therefore be interested in getting information about the single SoAs, in order to retrieve the knowledge they are missing.

5.2 The coding of alternative relations 5.2.1 The alternative coding implication Data concerning the presence of general and dedicated alternative markers are reported in Table 23 and 24 at the end of this chapter and in the general tables 53 and 54 in the Appendix. As already said, the morphophonological complexity of the attested markers is not discussed in a separate section, because the relevant information can

162 Alternative relations be discussed together with the two implicational patterns presented in this section. However, two tables with all the data collected on the morphophonological complexity of the markers have been provided in Tables 25 and 26, in order to make information accessible, despite the absence of implicational regularities. The possible combinations of presence vs. absence of an overt connective coding simple alternative and presence vs. absence of an overt connective coding choice-aimed alternative determine four different language types, which can be observed in Table 20. Table 20: Possible coding of choice-aimed and simple alternative within the same language. + = attested, – = not attested Presence of overt connective for simple alternative

Absence of

overt connective for simple alternative

Presence of

overt connective for c-aimed alternative Absence of overt connective for c-aimed alternative

+

–

+

+

Of these four language types, one is not attested. There is no language in the two samples that uses an overt connective to convey the choice-aimed alternative relation and no marker to convey the simple alternative relation. It is thus possible to identify the alternative coding implication, which can be formulated as follows: (5.12) The alternative coding implication: Asyndesis for simple alternative → Asyndesis for choice-aimed alternative In a given language, if the simple alternative relation can be expressed without using an overt connective, then also the choice-aimed alternative relation can be expressed without using an overt connective. This means that, in a given language, if it is possible to use a marker for choice-aimed alternative it will also be possible to use a marker for the expression of simple alternative. The attested types are shown in Table 21, where each type is instantiated by one language.

The coding of alternative relations

163

The ﬁrst attested type is characterized by the presence of a connective in both choice-aimed and simple alternative constructions, and is instantiated by Dutch, Basque, Albanian and Japanese in Table 21. All of the European languages examined belong to this group. However, an overt connective for both alternative relations is also found in the Comparison sample. Table 21: Coding of the alternative relation: attested types. - = absence of an overt coordinating marker; d = presence of a dedicated marker; Gx = general construction, with x identifying a given construction attested in the language; blank= no information available Dutch Basque Albanian Japanese Meithei Korean Wari’ Mangarayi

Choice-aimed alternative

Simple alternative

Gα Gα /d

Gα Gα

d -/d -

d d -/d d -

There are languages that only display a general marker for both alternative relations, like Italian o, French ou (5.13), German oder and the majority of European languages. Basque, Dargi and Arabic, besides the general alternative marker, also have a dedicated marker for the expression of a choiceaimed alternative. (5.13) French, Romance, Indo-European (a) Il a le bus ou il attendu 3sg AUX.PRS.3sg wait.PTCP:PST the bus COORD 3sg a un taxi appelé taxi AUX . PRS.3sg call. PTCP: PST a ‘He waited for the bus or he called a taxi.’ (b) Est-ce le que il a attendu be.PRS.3sg-it that 3sg AUX.PRS.3sg wait.PTCP:PST the bus ou il a un taxi? appelé bus COORD 3sg AUX.PRS.3sg call.PTCP:PST a taxi ‘Did he wait for the bus or did he call a taxi?’ Other languages express both alternative relations by means of an overt marker but do not have any general connective that could be used for both alterna-

164 Alternative relations tive types. In these languages, two dedicated markers are attested. In (5.14) the case of Polish is exempliﬁed, where lub is the simple alternative connective and czy the choice-aimed one. Two dedicated markers are also attested in other European languages, like Finnish (tai ∼ vai, cf. (5.15)), Albanian (ose ∼ apo), Belorussian (abĲo ∼ ci), Georgian (an ∼ tu, cf. (5.16), Ukrainian (abĲo ∼ cˇ y) and Lezgian (ja/wa ja ∼ taˆxajt’a), (see Tables 23 and 53). (5.14) Polish, Slavic, Indo-European lub az˙ do czytam (a) Zazwyczaj pisze˛ Usually write.PRS.1sg COORD read.PRS.1sg until to pó´zna late ‘Usually I write or I read until late.’ (A.L., questionnaire) (b) Idziemy jutro do szkoły czy w zostajemy go.PRS.1pl tomorrow to school COORD stay.PRS.1pl at domu? home ‘Do we go to school tomorrow or do we stay at home?’ (A.L., questionnaire) (5.15) Finnish, Ugro-Finnic, Uralic (a) joskus ovi avaut-ui tai ikkuna sometimes door open-3sg.PRF COORD window paiskaut-ui kiinni become-3sg.PRF closed ‘Sometimes a door opened or a window slammed.’ (J.Y., questionnaire) (b) mene-mme-kö koulu-un huomenna vai jää-mme-kö go-1pl-Q school-ILL tomorrow COORD stay-1pl-Q koti-in home-ILL ‘Do we go to school tomorrow or do we stay at home?’ (J.Y., questionnaire)

The coding of alternative relations

165

(5.16) Georgian, Kartvelian cˇ ai davliot, (a) Ginda gaviseirnot ertad, an want:2sg walk:1pl together COORD tea drink:1pl an k’inos vuq’urot, an . . . rac ginda? COORD movie watch:1pl COORD . . . what want:2sg ’Would you go for a walk with me, or have a cup of tea, or watch a movie, or ... whatever?’ (M. T., questionnaire) (b) Xval sk’olaši c’avidet tu saxlši tomorrow school:LOC go:1pl:OPT COORD home:LOC davrˇcet? stay:1pl:OPT ’Tomorrow do we go to school or do we stay at home?’ (M. T., questionnaire) The use of dedicated alternative connectives is widespread also outside Europe. Example (5.17) from Marathi illustrates the expression of the two possible types of alternative relation. In (5.17a) no choice is needed. A set of options is simply presented and the dedicated connective used to convey this type of relation is k˜ımw¯a. On the other hand, in (5.17b) the alternative relation is established in order to ask for a choice and the obligatory connective is k¯ı. Two different connectives for choice-aimed and simple alternative are also attested in Somali, where amá conveys the simple alternative relation (5.18a) and misé conveys the choice-aimed alternative relation (5.18b): (5.17) Marathi, Indo-Iranian, Indo-European s´u´srus.es¯a.th¯ı sut..t¯ı ghe¯ıl (a) madh¯u a¯ ¯ıtSy¯a Madhu mother:GEN looking.after.for leave take:FUT:3sg k˜ımw¯a/*k¯ı til¯a hspit.almadhe .tthew¯ıl 3sg.ACC hospital:in keep:FUT:3sg COORD ‘Madhu will leave to take care of his mother or keep her in the hospital.’ (Pandharipande 1997: 162) (b) to b¯adz¯ar¯at gel¯a k¯ı/*k˜ımw¯a ghar¯ı 3sg market.LOC go:PST:3sg.M COORD home:LOC gel¯a? go:PST:3sg.M

166 Alternative relations ‘Did he go to the market or did he go home?’ (Pandharipande 1997: 163) (5.18) Somali, Cushitic, Afro-Asiatic (a) Amá

wuu kéeni doonaa amá wuu sóo.díri doonaa 3sg bring that that COORD 3sg send ‘Either he will bring it or he will send it.’ (Saeed 1993: 275) COORD

(b) ma tégaysaa misé waad jóogaysaa? INT go:2sg COORD here stay:2sg ‘Are you going or are you staying?’ (Saeed 1993: 275) The second attested type in Tables 20 and 21 is characterized by the presence of a connective in the expression of simple alternative and by the absence of a marker for choice-aimed alternative. Languages in which the expression of choice-aimed alternative involves an optional marker also belong to this type. This is, for example, the case in Japanese, as shown in (5.19). The dedicated choice-aimed connective soretomo is often omitted (5.19a), while the dedicated simple alternative marker -ka is always present and is sufﬁxed to the ﬁrst verb form (5.19b). The sufﬁx -ka in (5.19a) is not to be confused with the alternative marker, although they are homophonous, because -ka in (5.19a) is an interrogative sufﬁx and is repeated after each verb form. (5.19) Japanese (a) gakkoo-e iki-masu-ka (soretomo) ie-ni school-ALL go-HON-INT COORD house-LOC i-masu-ka be-HON-INT ‘Do we go to school or do we stay at home?’ (Y.N., questionnaire) (b) tokidoki doa-ga aku-ka mado-ga simaru sometimes door-NOM open-COORD window-NOM slam ‘Sometimes a door opened or a window slammed.’ (Y.N., questionnaire) Languages of this type are only attested in the Comparison sample, because all European languages express both alternative relations by means of overt

The coding of alternative relations

167

markers. Other instances of this coding conﬁguration are Hakha Lai, Koromfe, Lango, Malayalam, Meithei and Korean (see 5.20). Example (5.20a) shows a construction used to express a choice-aimed alternative and it consists of the simple juxtaposition of two interrogative clauses. (5.20b), on the other hand, displays the marker -kena ‘or’, which is sufﬁxed to a deranked verb form and could never occur in questions. (5.20) Korean salam-ul ponay-l-kka-yo? (a) wuli-ka ka-l-kka-yo? 1pl-NOM go-PRS-Q-POL person-ACC send-PRS-Q-POL ‘Shall we go or shall we send a person?’ (Sohn 1994: 122) o-kena nae-ka ka-n-ta. (b) Minsu-ka Minsu-NOM come-COORD 1sg-NOM go-INCOMP-DECL ‘Minsu comes here or I go there.’ (Y.M.S., p.c.) There is then a third language type, which is characterized by the absence of any overt connective in the expression of both simple and choice-aimed alternative. Languages of this type are not frequent and are only attested in the Comparison sample. In Wari’, for instance, there is no connective coding alternative and the meaning of alternative can be conveyed in two ways. The ﬁrst possible strategy is shown in example (5.21a), in which the two states of affairs are juxtaposed and each of them is introduced by a conditional particle mo, whose meaning is similar to that of English if. Example (5.21b) shows the second strategy. Here again the two states of affairs are juxtaposed but, instead of a conditional particle, in each of the two states of affairs we can see a dubitative adverb ’am, whose meaning is ‘perhaps’. The choiceaimed alternative is conveyed by the simple juxtaposition of two interrogative clauses. (5.21) Wari’, Chapacura-Wanam ta pa’ ta’ hwam ca, realis.future kill 1sg:realis.future ﬁsh 3sg.M mo ta pa’ ta’ carawa ca. COND realis.future kill 1sg:realis.future animal 3sg.M ‘Either he will ﬁsh or he will hunt.’ (lit. ‘if he (says) “I will kill ﬁsh”, if he (says) “I will kill animals”.’) (Everett and Kern 1997: 162)

(a) Mo

COND

168 Alternative relations (b) ’Am ’e’ ca ’am mi’ pin ca perhaps live 3sg.M perhaps give complete 3sg.M ‘Either he will live or he will die.’ (lit.‘perhaps he will live perhaps he will die’) (Everett and Kern 1997: 162) A similar strategy is also attested in Mangarayi. As Merlan (1982: 33) points out, in Mangarayi there is no overt expression that directly corresponds to English or, either at the phrase level or at the clause level. Consideration of alternatives is often expressed by sequences with maNaya ‘perhaps’. This adverb, though necessary to infer an alternative relation, cannot be considered an alternative marker, since it also occurs when only one possibility is presented. Like in Wari’, the choice-aimed alternative is expressed by juxtaposing two interrogative clauses. (5.22) Mangarayi, Gunwingguan, Australian (Merlan 1982: 39) maNaya d.ayi maNaya ja-0-n / . iNa-n perhaps 3-3sg-come-PRS perhaps NEG ‘Perhaps he’ll come, perhaps not.’, i.e. ‘it is possible that he may or may not come’ This option seems to be available in every language, as can be seen from the English translation in (5.22). However, in English it is also possible to use the general alternative connective or, which is indeed the most widespread and normal strategy for the coding of alternative relations. On the other hand, in Mangarayi the juxtaposing construction is the only strategy available to express alternative. Before proceeding to examine more in depth the strategies used to express an alternative when no overt marker is available, let us now outline the crosslinguistic tendencies with respect to the morphophonological complexity of the attested markers. In the European sample, the attested alternative markers are morphophonologically more complex than the ones used to express combination (cf. section 6.3.1). 19 languages out of 37 have at least one bisyllabic alternative marker. Examples of bisyllabic markers are German oder, Swedish eller, Russian ili, and Polish albo. Examples of monosyllabic markers are English or, French ou, Italian o, Finnish vai and tai, Romanian sau and Maltese jew. Moreover, Lezgian, Dargi and Turkish show polymorphemic markers in the expression of alternative, which consist of a morpheme meaning ‘and’

The coding of alternative relations

169

combined with a morpheme meaning ‘or’, that is, alternative: Dargi yara (yara ‘or-and’, see example (5.23)), Lezgian waya (wa-ya ‘and-or’) and Turkish veya (ve-ya ‘and-or’). In example (5.23) yara is glossed as ‘or-and’, in order to show the transparency of its internal morphology. However, the alternative relation is encoded by the marker as a whole and ya-, although its meaning is that of alternative, occurs on its own only in negated sentences. (5.23) Dargi, Nakh-Daghestanian džäς äl ya-ra ya-ra tuxtur.li-ˇci arq’-ad, èu you(ABS) tomorrow or-and doctor-SUP leave-FUT.2sg or-and èänˇci-la dura.ulq-ad work-LOC go.out:M-FUT.2sg ‘Tomorrow you’ll either go to the doctor or you’ll go to work.’ (Van den Berg 2004: 204) Finally, in the European sample no bound markers are attested in the expression of alternative relations. All European alternative markers are free forms. The greater morphophonological complexity of alternative markers is probably connected with the fact that they are less frequent than the ones coding combination. However, this topic will be discussed in the next chapter, where the constructions used to code combination and alternative relations will be compared. In the Comparison sample, the picture is similar and the only signiﬁcant difference between European and non-European languages is that in the Comparison sample 5 languages out of 37 show bound markers expressing alternative, whereas no such cases are attested in Europe. However, the great majority of alternative markers consist of free morphemes also outside Europe. Instances of polymorphemic markers are attested in many non-European languages, like Hakha Lai, Hdi, Upper Kuskokwim Athabaskan. In two languages the attested markers are similar in structure to the ones described for European languages: Hausa, where the connective coding alternative kokuma is composed by ko- plus the combination marker kuma an the simple alternative marker is k˜ımw¯a and is composed by ki plus the combination morpheme wa. Finally, Hebrew shows a slightly different structure, in that the alternative marker o can stand alone, but is usually employed together with she- meaning ‘that’, as shown in (5.24).

170 Alternative relations (5.24) Hebrew, Semitic, Afro-Asiatic she-anàchnu Anàchnu holchìm levèt-hasèfer machàr o we go:1pl to-school tomorrow COORD that-we nisharìm babait? stay:1pl at.home ‘Do we go to school tomorrow or do we stay at home?’ (lit. ‘. . . or that we stay at home?’)(H.M., questionnaire)

5.2.2 The alternative irreality implication The examined asyndetic constructions used to express alternative are characterized by the presence of some marker indicating the non-realized nature of the involved SoAs (cf. examples (5.20), (5.21) and (5.22)). Such markers may either code the irreality of the linked SoAs or they may code notions which imply the non-realization of the SoAs, such as imagination, possibility, wish, interrogation, necessity, obligation and so on, where a given SoA is presented as not yet realized, or in which there is no certainty about its occurrence. Examples of constructions in which the linked SoAs show markers indicating their non-realized nature have already been discussed in the previous section (cf. (5.22) above for the presence of dubitative adverbs). A further example is provided here from Hakha Lai, where the question particle indicates that the SoAs are questioned and the speaker does not know whether they have taken place or not. In other words, the interrogative markers in (5.25) denote the notion of interrogation, which implies the irreality of the relevant SoAs. (5.25) Hakha Lai, Tibeto-Burman (Peterson and VanBik 2004: 352) ĳaa

ĳa-siì=maá ĳa-sií=ĳaĳ tsùn suy hlayhlaak 3sg.SBJ-be1 =Q 3sg.SBJ-be2 =LOC DEIC gold ladder thlaak=maá na-duĳ Nuùn hlayhlaak thlaak=maá drop2 =Q 2sg.SBJ-want silver ladder drop2 =Q thiàr hlayhlaak thlaak=daĳ tiaĳ na-duĳ khàn drop2 =INT QUOT DEIC 2sg.SBJ-want iron ladder làNĳaak-piì=niĳ tsùn ĳa-vón-hàl ĳàn-tií raven-AUG=ERG DEIC 3sg.SBJ-DIREC-ask2 3pl.SBJ-say INTERJ

The coding of alternative relations

171

‘ “Ah, is that so? If that’s the case, do you want me to drop a gold ladder, a silver ladder or an iron ladder?” the great raven asked her, they say.’ (lit. ‘Ah, is that so? If that’s the case, do you want me to drop a gold ladder? Do you want me to drop a silver ladder? Do you want me to drop an iron ladder?’ the great raven asked her, they say.) Before moving on to further examples, the concepts of realis vs. irrealis value and realis vs. irrealis markers will be deﬁned. The concept of reality value is usually discussed within the complex domain of modality, whose delimitation is still a matter of dispute (see Bybee et al. 1994).28 I deﬁne as reality value of a given proposition the actuality status of the SoA it describes, that is to say whether it is realized or not. Following Elliot (2000: 66-67), it is possible to distinguish between realis and irrealis propositions: (5.26)

- A proposition is said to be REALIS when it asserts that a SoA is an ‘actualized and certain fact of reality’ (Elliot 2000: 66). - A proposition is said to be IRREALIS when ‘it implies that a SoA belongs to the realm of the imagined or hypothetical, and as such it constitutes a potential or possible event but it is not an observable fact of reality’ (Elliot 2000: 67).

Irrealis propositions belong to the domains of imagination, possibility, wish, interrogation, necessity, obligation and so on, in which a given SoA is presented as not having taken place, or where the speaker is not sure about its occurrence. Other terms have been used to label similar distinctions. Lyons (1977: 796), for instance, talks about ‘factive’ and ‘non-factive’ utterances, whereas Palmer (1986: 17-18) uses the labels ‘factual’ and ‘non-factual’. All these terms, just like the ‘realis’ and ‘irrealis’ distinction just identiﬁed, refer to the functional level, and not to the structural one.29 On the structural level we deﬁne as: (5.27)

- R EALIS MARKERS all the morphosyntactic means (adverbs, sentence particles, verb forms) which speciﬁcally encode the realis value of a given SoA or which encode notions that imply the reality of the relevant SoA within a given clause. - I RREALIS MARKERS all the morphosyntactic means (adverbs, sentence particles, verb forms) which speciﬁcally encode the

172 Alternative relations irrealis value of a given SoA or which encode notions that imply the irreality of the relevant SoA within a given clause. The reality value does not have an homogeneous formal realization across languages. As Bybee et al. (1994: 238) show, if this distinction is formulated with reference to a binary morphosyntactic correspondence, it can hardly be argued to be universal. It is in fact very common to ﬁnd languages that have many different ways of marking both the realis and the irrealis domain. Elliot (2000: 80) talks about the reality status in terms of a grammatical category which is ‘realized differently in different languages’. There are languages that obligatorily mark the reality status of a proposition by means of a comprehensive and dedicated mood system, others in which the system is partial and needs to be indicated only in speciﬁc syntactic contexts, and ﬁnally there are languages like English, in which the reality status is not expressed by dedicated syntactic constructions, but it is ‘realized periphrastically’, by means of markers which imply a given reality status (Elliot 2000: 80). For instance, markers indicating a notion that implies the irreality of the relevant SoAs are interrogative markers. Since the content of a question is by deﬁnition not asserted as a certain fact of reality, any marker of interrogative form, either intonation or explicit particles, implies the irreality of the questioned SoA and therefore falls under the deﬁnition of irrealis marker provided in (5.27). The reality value of a given proposition may also remain not indicated explicitly, as in the case in which no marker directly or indirectly indicates the reality status of the SoAs and the construction can be used in both realis and irrealis contexts. In example (5.28a) and (5.28b) two uses of the present tense of the indicative mood in Italian are shown. (5.28a) shows a realis context, where the SoA expressed by means of a present tense of the indicative mood is taking place. On the other hand, (5.28b) illustrates the same verb form used to express a wish and the SoA is thus not stated as occurred or occurring, but rather as something that the speaker intends to do. In these two cases, the presence of some lexical temporal speciﬁcation is crucial in determining the realis vs. irrealis context. (5.28)

(a) In questo momento cucino, poi in this moment cook:IND.PRS.1sg afterwards mi riposo 1sg.ACC rest:IND.PRS.1sg

The coding of alternative relations

173

‘At this moment I am cooking, then I rest.’ → REALIS CONTEXT. (b) Domani dormo tutto il giorno tomorrow sleep:IND.PRS.1sg all DEF day ‘Tomorrow I going to sleep the whole day.’ → IRREALIS CONTEXT. Depending (i) on the presence vs. absence of explicit markers coding (or implying) the reality value of the SoAs and (ii) on the contexts in which the speciﬁc verb forms may occur, it is possible to distinguish between constructions which are overtly marked as realis (5.29a), constructions which are overtly marked as irrealis (5.29b) and constructions where the reality value is not overtly indicated (see (5.28) above and (5.32a) below). (5.29) Bukiyip (Conrad and Wogiga 1991: 282-102, cited in Elliot 2000: 63) (a) m-a-lpok 1pl-R-ﬁght ‘We are ﬁghting/we fought.’ (b) m-u-lpok 1pl-IRR-ﬁght ‘we will ﬁght.’ Yet none of the constructions expressing alternative attested in the samples is overtly marked as realis. Also in English the overt assertion of two alternative SoAs as realis would sound rather odd. A sentence like ‘Tonight I will certainly go to the cinema or I will certainly stay at home’ does not seem to make sense. As will be argued in section 5.4, this regularity reveals a close connection between the alternative relation and the irrealis value of the involved propositions. What is relevant for this survey is thus the presence vs. absence of markers expressing the irreality of the linked SoAs within a construction used for alternative. Two construction types are identiﬁed by this parameter: on the one hand alternative constructions with irrealis markers (overtly marked as irrealis), on the other hand constructions without irrealis markers (where the reality value is left unspeciﬁed). As already pointed out, by irrealis marker is meant here any marker which directly encodes the notion of irreality (cf. example

174 Alternative relations (5.29b)) or encodes a notion which implies that of irreality (cf. interrogative, dubitative and hypothetical markers). Let us now see an example of a construction with an irrealis marker. In Tauya, one possible way of marking the irrealis value of a proposition is by means of a verbal sufﬁx. In (5.30) we see a construction expressing a simple alternative relation. (5.30) Tauya, Brahman, Trans-New Guinea tei-sa yate-amu-rafu-ĳa pe tufuma-sa Teri-ADESS go-1sg:FUT-DUB-IND COORD Tuguma-ADESS yate-amu-rafu-ĳa go-1sg.FUT-DUB-IND ‘Maybe I’ll go to Teri or maybe I’ll go to Tuguma.’ (Mac Donald 1990: 139) The construction in (5.30) contains the connective pe, which encodes the alternative relation, and each SoA must be overtly marked as irrealis by the verbal dubitative sufﬁx -rafu, whose main function is to express uncertainty. Example (5.31) shows the Tauya construction used to convey a choice-aimed alternative relation. (5.31) Tauya, Brahman, Trans-New Guinea ne-ra ĳini-a-nae pe ni-a-nae 3sg-TOP sleep-3sg-Q COORD eat-3sg-Q ‘Did he sleep or did he eat?’ (Mac Donald 1990: 139) Here, too, the construction has overt markers signalling the irrealis value of the SoAs, now indicated not by the verbal sufﬁx -rafu, but by the interrogative mood sufﬁx -nae. What is questioned is by deﬁnition not asserted as occurring or having occurred, and is therefore irrealis. Tauya constructions coding alternative require in fact each SoA to be marked with the dubitative sufﬁx or with an interrogative marker, in order to convey a simple or a choice-aimed alternative relation respectively. The connective pe cannot link overtly realis propositions or propositions where the reality value is not indicated overtly (Mac Donald 1990: 139-142). Before proceeding to constructions without irrealis markers, some remarks on the coding of choice-aimed alternative are necessary. Not every language

The coding of alternative relations

175

has speciﬁc question particles or interrogative sufﬁxes like Tauya (ex. 5.31) and Hakha Lai (ex. 5.25), but it can be argued that every language distinguishes between interrogative and declarative by means of some interrogative marker, which may be a particle, interrogative mood, word order inversion or simply intonation (which is probably universal, Givón 1990: 786). As already argued in section 5.1.2, the choice-aimed alternative relation is always expressed by an interrogative construction. Furthermore, interrogative constructions are always characterized by some markers indicating that the SoAs are questioned and thus have an irrealis status. Therefore, we can conclude that constructions used to express a choice-aimed alternative are always overtly marked as irrealis, because interrogative markers imply the irrealis status of the relevant SoAs. On the other hand, constructions expressing the simple alternative relation do not show restrictions concerning the explicit coding of the reality value for each SoA. It can be either overtly marked or left unspeciﬁed, without compromising the general alternative reading. Let us now consider two further examples from English (5.32) and Italian (5.33), where the alternatives are expressed by clauses in the indicative mood, which is the most usual way of reporting SoAs in both languages and can be used to describe both realis and irrealis SoAs (cf. discussion on example (5.28)). In (5.32a) and (5.33a) the constructions do not have irrealis markers, whereas in (5.32b) and (5.33b) the irrealis value of the alternatives is overtly marked by a modal verb (may) and by two dubitative adverbs (forse and magari, both meaning ‘perhaps’) respectively. In these cases, the speaker makes a hypothesis about what is going to happen, or about what happened, and imagines two possibilities, without knowing or caring about which one actually occurred or is going to occur. (5.32) English (a) I take the bus or I go by bike. → the SoAs are not overtly marked as irrealis (unspeciﬁed reality value). (b) I may take the bus or I may go by bike. → the SoAs are overtly marked as irrealis

176 Alternative relations (5.33) Italian, Romance, Indo-European è (a) Dopo il lavoro o After the work COORD be(AUX).IND.PRS.3sg andato subito a casa o go.PTCP.PST straight ALL home COORD fatto ha prima un po’ di have(AUX).IND.PRS.3sg do.PTCP.PST before a little of spesa. shopping ‘After the work, he went straight home or he did some shopping ﬁrst.’ (b) E’ tutto bagnato in terra. Forse be.IND.PRS.3sg all wet in ground perhaps è o magari piovuto be(AUX).IND.PRS.3sg rain:PTCP.PST COORD maybe hanno le strade lavato have(AUX).IND.PRS.3pl clean:PTCP.PST the streets ‘It’s all wet on the ground. Perhaps it has rained or maybe they have cleaned the streets.’ In the sample there are no cases of constructions expressing alternative in which each SoA is overtly marked as realis. Conversely, constructions with overt markers indicating the irreality of the linked SoAs occur frequently. Reality markers are indeed incompatible with the notion of alternative, which is established between possible, not realized, SoAs. Likewise, the presence of irrealis markers in the expression of this relation directly follows from the semantics of the alternative relation itself, which links non-cooccurring SoAs, seen as replaceable possibilities (cf. deﬁnition of alternative in section 5.1). Based on the four possible combinations of presence vs. absence of connectives coding alternative and presence vs. absence of irrealis markers, there are four possible types of constructions for the expression of alternative relations: constructions with both markers coding alternative and markers expressing irreality, constructions with neither alternative nor irrealis markers, constructions with a connective coding alternative but without any irrealis marker, and constructions without any alternative connective but with some irrealis marker. The four possibilities are shown in Table 22.

The coding of alternative relations

177

Table 22: Presence of irrealis markers and alternative markers: possible combinations. + = attested, – = not attested Presence

Absence

of alternative markers

of alternative markers

+

+

+

–

Presence

of irrealis markers Absence

of irrealis markers

Of these four types of constructions, one is not attested. There are no constructions used to express alternative with neither connectives coding the alternative relation nor irrealis markers. This means that, if a construction expressing alternative does not have any connective coding the relation, it will always show some overt marker indicating the irreality of the linked SoAs. The alternative irreality implication can thus be stated as follows: (5.34) The alternative irreality implication: Absence of a connective coding alternative → Presence of some irrealis marker. More speciﬁcally, what is implied is that each SoA must display an irrealis marker, so it is presented as possible, rather than occurring or realized. Either a connective coding the alternative relation or some overt irrealis marker is necessary to convey an alternative relation. If neither of the two occurs, it is difﬁcult to infer an alternative reading and the construction fails to fulﬁll an alternative function. This implication proves that the irrealis value of the linked SoAs is a crucial aspect of the alternative relation. In particular, the irreality of the SoAs is implied by markers coding the alternative meaning, which in this respect may be considered irrealis markers just like the interrogative ones. In (5.27), an irrealis marker has been deﬁned as a marker which directly codes the irrealis status of the relevant SoA or which codes a notion implying that of irreality. The latter characterization of irrealis marker includes also connectives coding alternative, since they code a notion which implies the non-realized status of

178 Alternative relations the linked SoAs. Hence, when no alternative marker is present, the irreality of the linked SoAs needs to be explicitly signalled. Let us now examine the attested construction types. As shown in Table 22, the ﬁrst attested construction type is characterized by the presence of both a connective coding alternative and some irrealis marker. Some examples of this construction type have already been shown from Tauya (examples (5.30) and (5.31)), which has the alternative connective pe and requires, in order for the construction to be grammatical, that each SoA be individually marked with an interrogative or a dubitative sufﬁx. This construction type is attested, at least to express choice-aimed alternative, in all those languages that have an overt alternative marker. We have already seen some examples from English (5.32b) and Italian (5.33b), where the coexistence of some alternative connective and irrealis marker always occurs in the expression of choice-aimed alternative and is optional for the simple one. Let us now see one more example from NànáfwˆE, in which the choice-aimed construction is characterized by an obligatory interrogative intonation (Amani Bohoussou, p.c.) and by the presence of the speciﬁc choiceaimed connective ánzˆE, which could not be used to convey a simple alternative relation. (5.35) NànáfwˆE, Kwa, Niger-Congo é k´O sùklú áñímán ánzˆE é kà àwló wá? 1pl go school tomorrow COORD 1pl stay home here ‘Do we go to school tomorrow or do we stay home?’ (A.B., questionnaire) The second attested construction type is characterized by the presence of a connective coding alternative and the absence of any irrealis marker. Such constructions are only attested to convey simple alternative relations, since the expression of the choice-aimed alternative relation always requires some interrogative (and thus irrealis) marker. In examples (5.33a) and (5.32a) we already saw some examples of constructions expressing simple alternative with an unspeciﬁed reality value. Let us now consider one more example from West Greenlandic:

The coding of alternative relations

179

(5.36) West Greenlandic, Aleut, Eskimo imaluunniit aningaasa-ati-qa-nngil-aq money-ALIEN-have-NEG-3sg.IND COORD piqa-nngit-su-usaar-puq have-NEG-INTR.PTCP-pretend.to-3sg.IND ‘He has no money or pretends not to.’ (Fortescue 1984: 123) In this case, like in English and Italian, the simple alternative relation is conveyed by a construction which has an overt connective (imaluunniit) but no overt indications of reality value, since the basic indicative mood form can be used both in realis and in irrealis contexts, and no other explicit irrealis markers are used. The third attested construction type is characterized by the absence of a connective coding alternative and by the presence of some overt irrealis marker in each SoA. It is mainly attested to convey choice-aimed alternative relations, but it can also be found for simple alternative, as we have already seen in the examples from Hakha Lai (5.25) and Mangarayi (5.22). An example from Korean has already been given in (5.20). Let us now examine the case of Dargi in (5.37), where the juxtaposition of two interrogative clauses conveys the choice-aimed alternative relation. There is the possibility to use an overt marker coding alternative (aèi), but it is often omitted. The sentence in (5.37) would be better rendered in English by means of NP disjunction, but for Dargi speakers NP disjunction in interrogative sentences is problematic and clausal disjunction is preferred (Van den Berg 2004: 203).30 (5.37) Dargi, East Caucasian, North Caucasian (aèi) nerˇg b-ir-eèe-w b-ir-eèe-w? pilaw pilaf(ABS) N-do-FUT.1pl-INT (or) soup(ABS) N-do-FUT.1pl-INT ‘Shall we make pilaf or (shall we make) soup?’ (Van den Berg 2004: 203) All the constructions in the sample that lack a connective coding alternative show some overt marking, which indicates that each of the two SoAs is irrealis (the SoAs can additionally both fall under the scope of a unique irrealis marker, such as a modal auxiliary). Each SoA taken individually is thus presented as possible, rather than as actual. It can be concluded that the relation

180 Alternative relations of alternative may be conveyed in two ways, (i) either by a means of a connective coding alternative (ii) or by means of an underspeciﬁed construction, where a contextual inference based on the irrealis nature of the two juxtaposed SoAs gives rise to the alternative reading. The major role played by irrealis markers in the coding of the alternative relation is linked to the semantic nature of the relation itself. In section 5.1, the alternative relation has been characterized as established on a non-cooccurrence dimension, since it links a set of paradigmatic possibilities. Given a slot ‘X’ in a possible world, it can be occupied by only one of the two alternative SoAs at a time. In other words, two alternative SoAs are conceptualized as equivalent possibilities, only one of which will or did actually take place at the speciﬁc moment which constitutes the free slot ‘X’. Until a choice is made or the speaker gets to know which hypothesis is realized at that given time, either SoA could be the non-occurring one and is therefore conceptualized as irrealis. If no connective explicitly indicates that the linked SoAs are not occurring (or have occurred), but are only possible, some other marker needs to be used to code their nature of possibilities, otherwise they would be interpreted as facts. By contrast, if one or both SoAs are overtly marked as realis, i.e. asserted as realized, this means that the speaker knows which SoA will or did occur at that given time and there is no alternative to posit. The central role of irreality in the concept of alternative has already been pointed out by Ohori (2004: 56-57), who cites cases of what he calls ‘underdifferentiation’ between conjunction and disjunction. Ohori examines the following Maricopa examples: (5.38) Maricopa, Hokan, Yuman Bill-š vĳaawuum (a.) John-š John-NOM Bill-NOM come.3pl.FUT ‘John and Bill will come.’ (Gil 1991: 99) (b.) John-š Bill-š vĳaawuumšaa John-NOM Bill-NOM come.3pl.FUT.INFER ‘John or Bill will come.’ (Gil 1991: 99) In (5.38a), the verb is marked as future only and the SoA being described “is securely believed by the speaker with higher certainty”.31 In (5.38b), on the other hand, the verb has an inferential sufﬁx, which means that the speaker is

The coding of alternative relations

181

not certain about what is going to happen. ‘The AND-OR distinction is thus dependent on the choice of a modal-like element on the verb. If certain, the resulting interpretation is AND. If uncertain, it is OR’ (Ohori 2004: 57). As Ohori (2004: 64) puts it, the realis-irrealis distinction is a crucial factor in forming the concepts of AND and OR. A list of entities, he says, is likely to be interpreted conjunctively (=AND) in some languages when the predication’s modality is realis. On the other hand, when the predication’s modality is irrealis, this favors the acknowledgement of alternative possibilities (=OR). A quick diachronic glance seems to conﬁrm the close connection between irreality and the alternative relation. Many connectives coding alternative indeed originate from or evolve into irrealis markers, such as interrogative particles (like Polish czy, for instance, see Heine and Kuteva 2002: 226-227) or hypothetical forms. Speciﬁcally, there are a few cases where a hypothetical construction with a negated protasis has developed into an alternative connective. For instance, the Lezgian choice-aimed connective taˆxajt’a was originally the conditional form of the negated aorist participle of xˆ un ‘be’, thus meaning ‘if it is not’ (Haspelmath 1993: 332). Also Hakha Lai displays a recent alternative marker which is still transparent in its components. As seen from example (5.39), -làw-leè is the combination of the negation -làw and the ancient conditional sufﬁx -leè. At present, Hakha Lai uses a new form for the conditional construction, and this quite complicated way of expressing an alternative relation is on the way to grammaticalization as an alternative connective. (5.39) Hakha Lai, Tibeto-Burman haàkhaà-ĳaĳ làwthlawpaa falaám ĳa-kal-làw-leè farmer Falam 3sg.SBJ-go-NEG-COND Hakha-LOC ĳa-ĳùm 3sg.SBJ-exist ‘The farmer goes to Falam or he stays in Hakha.’ (lit. ‘The farmer, if he doesn’t go to Falam, he stays in Hakha’)(Peterson and Van Bik 2004: 339) A similar construction is normally used in NànáfwˆE to encode a simple alternative relation:

182 Alternative relations (5.40) NànáfwˆE, Kwa, Niger-Congo fùndr´Eti O´ c´En wjélj´E s´E nán ánwán j´E O´ tíké that it open.PRF FOC window day some if NEG door j´E ñín O´ n that.it slam.PRF FOC ‘Sometimes a door opened or a window slammed.’ (lit. ‘Sometimes, if it wasn’t a door that opened, it was a window that slammed’)(A.B., questionnaire) In all these cases, the construction as a whole is built as a hypothetical one. In particular, the hypothesis of the non-occurrence of the ﬁrst SoA is the condition for the second SoA to occur, and both SoAs are then presented as irrealis. Also in colloquial Italian it is possible to use ‘sennò’ (< se no) ‘if not’, instead of ‘o’, to convey a simple alternative relation (5.41). This construction, however, cannot be used to convey choice-aimed alternative relations, the result would be ungrammatical. (5.41) Italian, Indo-European, Romance cinema, sennò stiamo al a Andiamo go:IND.PRS.1pl ALL.DEF cinema if.not stay:IND.PRS.1pl LOC casa, sennò facciamo passeggiata una home if.not make:IND.PRS.1pl INDEF walk ..dimmi tu! tell.IMP.2sg:1sg.DAT you ‘We can go to the cinema, or we can stay at home, or we can go for a walk...it’s up to you!’ The conceptual closeness between the meaning of alternative and hypothetical/irrealis modality is conﬁrmed in diachronic perspective as well, since these data point to possible paths of grammaticalization along which alternative connectives arise from (or develop into) irrealis constructions. However, a survey on the grammaticalization phenomena leading to connectives coding alternative goes beyond the scope of this research.

The coding of alternative SoAs

183

5.3 The coding of alternative SoAs The parameter of coding of the SoAs does not reveal any cross-linguistic implicational pattern of variation, in that almost all the attested alternative constructions in the two samples consist of two balanced verb forms. Speciﬁcally, all European languages only use syntactically parallel constructions to code alternative relations, therefore no table has been built for the European sample for this parameter. Outside Europe, on the other hand, three of the 37 analyzed languages have non-parallel constructions expressing alternative. All data relative to the Comparison sample can be observed in Table 27. Moreover, the two general tables at the end of the Appendix report information about all the examined constructions in European and nonEuropean languages. Non-parallel constructions for alternative are only attested in Jamul Tiipay, West Greenlandic and Korean. This homogeneity reveals a strong crosslinguistic tendency for syntactically parallel alternative constructions, which itself constitutes a noteworthy regularity in the coding of two alternative SoAs. The non-parallel construction attested in Jamul Tiipay consists of two clauses linked by the connective nyamaaw, whose morphological composition is ’nya ‘when’ plus maaw ‘not be’ (Miller 2001: 257-258), mirroring the negative conditional construction exempliﬁed in the last section. In all the attested examples the mood of the second verb form is overtly marked by the sufﬁx -x ‘irrealis’ (5.42) or by an imperative sufﬁx, while the mood of the ﬁrst verb form cannot be overtly signalled. The presence of the connective nyamaaw implies that the two clauses have the same irrealis value. (5.42) Jamul Tiipay, Yuman, Hokan (Miller 2001: 257) ‘yúxwik m-achuumuch nyamaaw nyaap ny-achu’much-x-s’ maybe 3/2-kill.pl or.else me.ABS 3/1-kill.pl-IRR-EMPH we’-i 3-say “‘Maybe they will kill you or else they will kill me”, he said.’ In West Greenlandic, the second verb form may be in the contemporary mood, that is the same mood we have examined in non-parallel constructions used to express combination and contrast relations, receiving the alternative

184 Alternative relations sufﬁx =luunniit, as shown in example (5.43). However, this non-parallel strategy co-exists with a parallel one, where two balanced verb forms are linked by means of the connective imaluunniit. (5.43) West Greenlandic, Aleut, Eskimo aningaasa-ati-qa-nngil-aq money-ALIEN-have-NEG-3sg.IND piqa-nngit-su-usaar-luni=luunniit have-NEG-INTR-pretend.to-4sg.CONT=COORD ‘He has no money or he pretends not to.’ (Fortescue 1984: 123) In West Greenlandic the same construction is used to express choice-aimed and simple alternative relations. The third language showing a non-parallel construction, namely Korean, uses it just for the expression of the simple alternative, whereas the choice-aimed one is conveyed by simply juxtaposing two balanced interrogative sentences (see example (5.20)). Since Korean is the only language displaying an asymmetry between the coding of choice-aimed and simple alternative, it is not possible to identify any implicational pattern in the cross-linguistic coding of two alternative SoAs. Given the scarcity of data supporting it, this asymmetry may also be due to pure casuality. What emerges as strong regularity is instead the use of syntactically parallel constructions to code alternative relations. The reasons underlying this tendency will be explained in the next section. 5.4 Functional motivations in the coding of alternative The possibility of using an overt connective to code a choice-aimed alternative relation in a given language implies that also a simple alternative relation can be conveyed by means of an overt alternative connective. In order to understand the motivations underlying this implication, ﬁrst of all we need to analyze in what respects connectives coding alternative differ from irrealis markers. As already said, they overlap in the expression of the irreality of each SoA, since a connective coding alternative implies the irreality of each alternative. However, an alternative relation is not characterized by the irreality of the involved SoAs only, even though this is a crucial point. There is at least one further semantic property that is necessary in order for two possibilities to be conceived as alternatives. Two SoAs can be conceived

Functional motivations in the coding of alternative

185

as alternatives if they constitute possible and non-cooccurring substitutes for each other. Alternative markers and irrealis markers differ exactly in this respect: the function of an alternative connective is to convey an alternative relation as a whole, including reciprocal replaceability and non-cooccurrence of SoAs, whereas an irrealis marker only conveys the irreality of a given SoA. Therefore, if a construction expressing alternative lacks an overt connective, the irreality of the SoAs will be expressed by means of irrealis markers and their status of non-cooccurring substitutes will be inferred from the context. By contrast, if an alternative connective is used, neither aspect (irreality and mutual replaceability) needs to be inferred from the context, but both are implicit in the alternative meaning conveyed by the alternative marker. Going back to the implicational pattern of variation, we can now try to give an answer to the question of why, if a language has a construction lacking an overt alternative marker, this will be used at least for choice-aimed alternative relations. The use of an overt marker in the expression of simple alternative allows for the absence of overt irrealis markers, since the irreality of the relevant SoAs is part of the meaning of the connective and does not need to be further speciﬁed. Each clause may then occur with a basic unmarked verb form (for instance a present indicative), and explicit irrealis markers are no longer needed. The principle at work here is that of syntagmatic economy. The notion of irreality, being an implication of the meaning of the connective coding alternative, is not further speciﬁed. On the other hand, even if an alternative marker is used in the expression of choice-aimed alternative, the irrealis status of the SoAs is still overtly marked by interrogative markers. In this case, the use of a connective coding alternative would lead to redundancy in the expression of irreality, since irreality would be implied both by the interrogative markers and by the alternative connective. Furthermore, when there is no connective coding alternative, the alternative reading can only be inferred from the context or assumed on the basis of the expected consistency of the speaker’s intentions. Data show that if a language has this kind of juxtaposing strategy, it will ﬁrst of all be used to express choice-aimed alternative. Therefore, it is plausible to suggest that it is easier to infer an alternative relation from the juxtaposition of two interrogative clauses than from the juxtaposition of two declarative clauses (once again, the principle at work is syntagmatic economy).

186 Alternative relations The reason why two juxtaposed interrogative clauses are more easily interpreted as alternatives is presumably related to the nature of the relation that can be postulated between the relevant SoAs. Two alternative SoAs stand in semantic contrast to each other. ‘He loves her’ and ‘He wants to marry her’, for instance, would not stand in an alternative relation because they lack this contrast. When two SoAs that stand in a semantic contrast are juxtaposed in a declarative sentence however, this may easily be for reasons other than the existence of an alternative relation between the two. For example, such a juxtaposition usually occurs because of temporal/causal sequentiality (the two SoAs are juxtaposed because they occur one after the other), simultaneity (the two SoAs are juxtaposed because they occur at the same time) or some contrast (the two SoAs may be opposed to each other), thus assuming the cooccurrence of the SoA. On the other hand, if the two SoAs are encoded by two juxtaposed interrogative clauses, this means that they are questioned and that the speaker does not know whether they actually occur. Since they are not presented as actually occurring, the reason for presenting the two SoAs together will hardly be that they are linked by a relation of temporal/causal sequentiality or simultaneity. In fact, there seems to be no particular communicative reason to question two SoAs that stand in a semantic contrast, if they are not alternatives. Consequently, if two SoAs standing in a semantic contrast are juxtaposed in an interrogative form, they will be most easily interpreted as alternatives. As stated in (5.34), in the sample there are no constructions expressing alternative lacking both irrealis markers and alternative markers. Two alternative SoAs are not overtly indicated as irrealis only if their irrealis status is already implicit in the alternative meaning expressed by the connective itself. Once again, the principle of syntagmatic economy is at work. Information that is recoverable from the context (in this case, recoverable from the alternative connective) needs no further speciﬁcation (Haiman 1985). If there is no connective coding alternative, on the other hand, languages obligatorily mark the irreality of SoAs by means of already available strategies, such as dubitative particles or moods, hypothetical or interrogative forms. In order to be presented as alternatives, the SoAs need to be characterized in terms of possibility. A construction without an alternative connective and without irrealis markers is a construction which is not able to encode the potential status of the SoAs, and hence fails to capture a basic aspect of their nature of alternatives. This construction type is not attested because it

Language data

187

would lead to a loss of communicative value, since not only is the concept of alternative not encoded, but it is also not recoverable from the context. Finally, the great majority of the examined languages in both samples uses parallel strategies to code both choice-aimed and simple alternative. This arguably depends on the fact that the alternative relation as such only predetermines the irrealis status of the SoAs it links, but does not restrict their tense or aspectual features. Two alternative SoAs are indeed completely independent and do not even take part in the same overall macro-event, because there is no certainty as to which one actually occurs. Moreover, the participants involved may coincide or be different. Therefore, even if an overt connective coding alternative is present, information concerning the linked SoAs could not be recoverable from the context, but would need to be explicitly indicated for each SoA by means of balanced verb forms.

5.5

Language data

Table 23: European Sample. Coding of alternative relations: presence of overt coordinating markers and their semantic domain. d= dedicated; Gx = general, with x identifying a given construction attested in the language; blank= no information available Language

Choice-aimedalternative

Simple alternative

Albanian Basque Belorussian Bulgarian Catalan Chechen Czech Danish Dargi Dutch English Estonian Finnish French Georgian German Greek

d Gα /d d Gα Gα Gα Gα Gα Gα /d Gα Gα Gα d Gα d Gα Gα

d Gα d Gα Gα Gα Gα Gα Gα Gα Gα Gα d Gα d Gα Gα

188 Alternative relations (Continued) Language

Choice-aimedalternative

Simple alternative

Hungarian Icelandic Irish Italian Latvian Lezgian Lithuanian Luxembourgish Maltese Norwegian Polish Portuguese Romanian Russian Sardinian Serbo-Croatian Spanish Swedish Turkish Ukrainian

Gα Gα Gα Gα Gα d Gα Gα Gα Gα d Gα Gα Gα Gα Gα Gα Gα Gα d

Gα Gα Gα Gα Gα d Gα Gα Gα Gα d Gα Gα Gα Gα Gα Gα Gα Gα d

Table 24: Comparison Sample. Coding of alternative relations: presence of overt coordinating markers and their semantic domain. - = absence of an overt marker; d= dedicated; Gx = general, with x identifying a given construction attested in the language; blank= no information available Language

Choice-aimedalternative

Simple alternative

Arabic Dumi Hakha Lai Harar Oromo Hausa Hdi Hebrew Hocak ˛ Iraqw Jamul Tiipay Japanese Kisi Kol. Yukaghir

Gα /d Gα -

Gα Gα d

Gα Gα Gα Gα Gα

Gα Gα Gα Gα Gα d d Gα

-/d Gα -

Language data

189

(Continued) Language

Choice-aimedalternative

Simple alternative

Korean Koromfe Kuskokwim A. Lango Limbu Malayalam Mangarayi Maori Marathi Maricopa Meithei Mosetén NànáfwˆE Ndyuka Persian Rapanui Somali Supyire Tauya Tukang Besi Tuvaluan Vietnamese Wari’ W.Greenlandic

-

d d d d

Gα d

d Gα d

-

-/d

d Gα Gα -/Gα d d Gα Gα Gα d Gα

d Gα Gα -/Gα d d Gα Gα Gα d Gα

Table 25: European Sample. Coding of alternative relations: morphophonological complexity of the attested coordinating markers. + = presence of the given feature; - = absence of the given feature; alt.= alternative; choice-a= choice-aimed; blank= no information available Language

Marker

Semantic domain

Free

Polysyllabic

Polymorphem.

Albanian

ose/o apo/a edu ala abĲo ci ili o

simple alt. choice-a alt. both alt. types choice-a alt. simple alt. choice-a alt. both alt. types both alt. types

+ + + + + + + +

-/+ -/+ + + + + + -

-

Basque Belorussian Bulgarian Catalan

190 Alternative relations (Continued) Language

Marker

Semantic domain

Free

Polysyllabic

Polymorphem.

Chechen Czech

ja nebo ci ˇ eller yara aèi of or või tai vai ou an tu oder i vagy eða nó o/oppure vai ja wa ja taˆxajt’a ar/arbà oder jew eller lub albo czy ou ori sau ili o ili o eller veya abĲo

both alt. types both alt. types both alt. types both alt. types both alt. types choice-a alt. both alt. types both alt. types both alt. types simple alt. choice-a alt. both alt. types simple alt. choice-a alt. both alt. types both alt. types both alt. types both alt. types both alt. types both alt. types both alt. types simple alt. simple alt. choice-a alt. both alt. types both alt. types both alt. types both alt. types simple alt. simple alt. choice-a alt. both alt. types both alt. types both alt. types both alt. types both alt. types both alt. types both alt. types both alt. types both alt. types simple alt.

+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + +

+ + + + + + + -/+ + + -/+ + + + + + + + -/+ -/+

+ + + -/+ -

Danish Dargi Dutch English Estonian Finnish French Georgian German Greek Hungarian Icelandic Irish Italian Latvian Lezgian

Lithuanian Luxembourgish Maltese Norwegian Polish

Portuguese Romanian Russian Sardinian Serbo-Croatian Spanish Swedish Turkish Ukrainian

Language data

191

(Continued) Language

Marker

Semantic domain

Free

Polysyllabic

Polymorphem.

cˇ y

choice-a alt.

+

-

-

Table 26: Comparison Sample. Coding of alternative relations: morphophonological complexity of the attested coordinating markers. + = presence of the given feature; - = absence of the given feature; alt.= alternative; choice-a= choice-aimed; blank= no information available Language

Marker

Semantic domain

Free

Polysyllabic

Polymorphem.

Arabic

aw am ye -làw=leè

both alt. types choice-a alt. both alt. types simple alt.

+ + + +

+

+

kó kokuma á nà o (she) ni˛˛igéšge laqáa nyamaaw -ka soretomo bàà

both alt. types both alt. types both alt. types both alt. types both alt. types both alt. types simple alt. simple alt. choice-a alt. both alt. types none simple alt. simple alt. simple alt. simple alt. simple alt. none simple alt. simple alt. none both alt. types simple alt.

+ + + + + + + + +

+ + -/+ + + + + -

+ + -/+ + -

+ + + +

+ + + +

+ -

+

+

-

+ +

+ +

+

choice-a alt.

+

-

-

Dumi Hakha Lai Harar Oromo Hausa Hdi Hebrew Hocak ˛ Iraqw Jamul Tiipay Japanese Kisi Kol. Yukaghir Korean Koromfe Kuskokwim A. Lango Limbu Malayalam Mangarayi Maori Marathi

Maricopa

-kena hokun maa h0nek’è òñò -oo alleNkil raanei k˜ımw¯a/ athw¯a k¯ı

192 Alternative relations (Continued) Language

Marker

Semantic domain

Free

Polysyllabic

Polymorphem.

Meithei Mosetén NanafwˆE

nattrega

simple alt.

+

+

-

+ + + + + + + + + + + + + +

+ + + + + + + -/+

-

+ -

+ +

-

Ndyuka Persian Rapanui Somali Supyire Tauya Tukang Besi Tuvaluan Vietnamese Wari’ W.Greenlandic

wj´Elj´E ánzˆE efu yâ o ama misé yô làa pe tawa io me/pe ho˘a. hay(là)

simple alt. choice-a alt. both alt. types both alt. types both alt. types simple alt. choice-a alt. simple alt. choice-a alt. both alt. types both alt. types both alt. types simple alt. choice-a alt. none imaluunniit both alt. types =luunniit both alt. types

Table 27: Comparison Sample. Coding of two alternative SoAs: syntactic parallelism of the attested constructions. + = syntactically parallel; - = syntactically non-parallel; d= dedicated; Gx = general, with x identifying a given construction attested in the language; blank= no information available Language

Simple alternative

Choice-aimedalternative

Arabic Dumi Hakha Lai Harar Oromo Hausa Hdi Hebrew Hocak ˛ Iraqw Jamul Tiipay Japanese Kisi Kol. Yukaghir

Gα + Gα + d+

Gα +/d+ Gα + d+

Gα + Gα + Gα + Gα + Gα +

Gα + Gα + Gα + Gα + Gα /J+ dd+ Gα + d+

d+ Gα +

Language data (Continued) Language

Simple alternative

Choice-aimedalternative

Korean Koromfe Kuskokwim A. Lango Limbu Malayalam Mangarayi Maori Marathi Maricopa Meithei Mosetén NànáfwˆE Ndyuka Persian Rapanui Somali Supyire Tauya Tukang Besi Tuvaluan Vietnamese Wari’ West Greenlandic

d-/+ d+ d+ d+

d+ d+

d+ d+ Gα + d+

d+ d+ d+ Gα + d+

d+

d+

d+ Gα + Gα + Gα + d+ d+ Gα + Gα + Gα + d+ d+ Gα /-

d+ Gα + Gα + Gα + d+ d+ Gα + Gα + Gα + d+ d+ Gα +/-

193

Chapter 6 Combination, contrast and alternative relations in comparison 6.1

Introduction

In this chapter the strategies coding combination, contrast and alternative relations are compared, in order to point out the asymmetries of coding between the various coordination relations and to highlight the patterns internal to coordination itself. As brieﬂy mentioned in the introduction to this study, conjunction, disjunction and adversativity are traditionally considered three equivalent semantic sub-types of coordination, as deﬁned in structural terms. In this research however, the perspective has been reversed, and conjunctive, disjunctive and adversative constructions are treated as the morphosyntactic constructions encoding the three coordination relations of combination, contrast and alternative, with no expectations or assumptions regarding their morphosyntactic properties. As widely argued in the previous chapters, the cross-linguistic patterns for the coding of these three relations reveal a more complex picture than the simple tripartite subdivision proﬁled within the traditional approach to coordination. What is at issue in this chapter is the possibility to actually consider combination, contrast and alternative as equivalent types of coordination relation. In other words, we are going to examine if these relations stand on the same level or if they show some internal hierarchical organization. The comparison develops in two phases. First, combination is separately compared to contrast (section 6.2) and to alternative (section 6.3). Second, the three coordination relations are examined together, within a general discussion on the cross-linguistic coding of coordination (section 6.4). Since the comparison of contrast and alternative does not reveal particular coding patterns, it will not be discussed in a dedicated section, but it will be encompassed by the overall comparison of the three relations. All data relating to the coding of the relation and the coding of the SoAs have been presented in the preceding chapters for each of the relations under exam. However, at the end of sections 6.2 and 6.3, tables reporting the

Introduction

195

relevant data will be provided, to make the comparison visible for all the examined languages. In section 6.2.1 combination and contrast are compared with respect to the presence and nature of coordinating markers. Two implicational patterns of variation will be identiﬁed, one concerning the presence and the other concerning the complexity of the attested overt connectives. The comparison of the semantic domains will be represented in the combination-contrast conceptual space, where all the combination and contrast types will be ordered along a functional proximity dimension. Section 6.2.2 will then compare the coding of combined and conﬂicting SoAs, highlighting the combination-contrast parallelism implication. Section 6.2 will end with the functional explanations of the identiﬁed cross-linguistic patterns of variation and with the tables comparing the relevant data in comparison. A similar structure of analysis is maintained in section 6.3, where the constructions coding combination relations are compared to the ones coding alternative. First, the comparison takes into account the presence of overt coordinating markers and their morphophonological complexity, providing the basis for two cross-linguistic implications. Then, the attested combination and alternative semantic domains are compared, pointing out the neat cross-linguistic tendency to code the two coordination relations by means of completely different constructions. The parameter of syntactic parallelism will then be considered and the combination-alternative parallelism implication will be identiﬁed in section 6.3.2. Functional explanations and tables with the relevant data will then follow. This chapter ends with the conclusive remarks on the cross-linguistic coding of coordination relations, summarized in section 6.4, where the three relations are jointly discussed. In comparison to both contrast and alternative, combination turns out to be cross-linguistically coded by simpler forms, more easily inferable from asyndetic constructions and more frequently coded by non-parallel constructions. This asymmetry of coding depends on the fact that combination is more basic than the other two coordination relations. The cross-linguistic coding of combination, alternative and contrast highlights that the three relations are not three equivalent subtypes of coordination. Rather, contrast and alternative develop along two different and separate semantic axes, having their origin in combination (cf. Figures 8 and 9).

196 Combination, contrast and alternative relations in comparison 6.2

Combination and contrast

6.2.1 The degree of coding 6.2.1.1 The combination-contrast coding implication The comparison of the attested overt markers expressing combination and contrast relations can be found in Tables 34 and 35 at the end of this section. The analysis of these data reveals the following implication: (6.1) The combination-contrast coding implication: sequential, simultaneous, atemporal combination, oppositive contrast, corrective contrast → counterexpectative contrast In a given language, the presence of an (optional) overt marker expressing one of the three combination relations, oppositive and corrective contrast implies the presence of an overt marker also for the expression of counterexpectative contrast. In other words, if a syndetic strategy is found for combination, opposition and correction, a syndetic strategy is also found for counterexpectative contrast. In section 6.2.3 the motivations underlying this implicational pattern will be highlighted. Table 28 shows the attested language types. Table 28: Overt markers for combination and contrast relations: cut-off points in the combination-contrast coding hierarchy. + = presence of an overt marker; – = absence of an overt marker Seq Italian + Georgian + Chechen –/+ Tuvaluan – Hausa –/+ Maori –/+ Hdi –/+ Limbu –/+ Lango –/+ Wari’ –

Temporal Sim + + –/+ –/+ –/+ –/+ – –/+ – –

Atemp

Opp

Correc

Count

+ + –/+ –/+ –/+ –/+ – –/+ – –

+ –/+ + + –/+ – – – – –

+ –/+ + + –/+ –/+ – – – –

+ + + + + + + –/+ –/+ –

As can be observed in Table 28, no implicational pattern is attested in the coding of combination relations, oppositive and corrective contrast. On

Combination and contrast

197

the one hand, there are languages like Maori, Hdi, Limbu and Lango, which have the possibility to express combination relations, or at least one of them, by means of overt coordinating markers, while this possibility is not attested for oppositive or corrective contrast relations. On the other hand, we also ﬁnd languages like Tuvaluan and Koromfe, which show overt markers in the expression of opposition and correction, but this possibility is not available for sequential combination. The cross-linguistic generalization in (6.1) does not have a high predictive power, because it only excludes languages in which combination, opposition or correction may be expressed by a syndetic strategy whereas this possibility is not available for counterexpectative contrast. Cristofaro (2003: 84-90) calls this type of cross-linguistic patterns quantiﬁed implications, i.e. implications involving the logic quantiﬁers ∃ and ∀.32 Quantiﬁed implications do not account for whether a given phenomenon is possible or impossible (cf. chapter 1), but rather examine and constrain the set of different possible ways in which the phenomenon may occur across languages and indicate the number of items in a given set (every item ∼ at least one item) for which the statement in the implication holds (for a detailed discussion on quantiﬁed implications, see Cristofaro 2003: chapter 4). Georgian and Persian constitute two interesting examples. Combination is commonly expressed by means of overt markers in both languages, whereas correction, despite the possibility to use overt connectives, is normally conveyed by a juxtapositive strategy. Three examples from Persian are shown in (6.2). In (6.2a) non-sequential combination is coded by the marker va, in (6.2b), instead, a corrective contrast relation is conveyed by simply juxtaposing the two clauses, while in (6.2c) an overt marker is employed to code a counterexpectative contrast. (6.2) Persian, Indo-Iranian, Indo-European John yek poster (a) Mary yek kartpostal khar-id va Mary one postcard buy-3sg COORD John one poster ‘Mary bought a postcard and John a Poster.’ (A.A, questionnaire) (b) Peter dar otaq-ash dars ne-mi-xân-ad, dar Peter in room-3sg lesson NEG-DUR-study-3sg in baq bâzi mi-kon-ad garden play DUR-do-3sg

198 Combination, contrast and alternative relations in comparison ‘Peter is not studying in his room, but he’s playing in the garden.’ (A.A, questionnaire) (c) Man xeili teˆsne am vœli âb portaqal dust ´ I very thirsty be:1sg COORD water orange like nadâram NEG:have:1sg ‘I am very thirsty but I don’t like orange juice.’ (A.A, questionnaire) A similar situation is attested in Georgian, where the general combination marker da is commonly used to convey the two combination relations, but corrective contrast is normally rendered by juxtaposing an emphatically negated clause to a positive one (see example (4.18)). As already pointed out, in both languages there is a marker that could be used to convey correction: aramed in Georgian (Hillery 1996) and bœlke in ´ Persian (Stilo 2004: 305-306). Yet, these syndetic strategies pertain to written language and are restricted to formal contexts. Georgian and Persian native speakers who were asked to ﬁll out the questionnaire did not use any overt marker to code corrective contrast, as the examples have shown. As argued in section 4.1.2, this may be explained by the discourse function that is typically associated to correction. The conﬂict underlying correction requires the ﬁrst SoA to be overtly disputed, normally in response to a wrong assertion made by the interlocutor, and the second SoA is then brought in as the right substitute. The context thus makes a corrective contrast easily inferable even if no overt marker is used. Juxtaposition in the expression of correction also occurs in other languages as an alternative to the syndetic strategy. As can be observed in (6.3), also in English it is possible to express correction without any overt connective. While in Georgian and Persian juxtaposition is the normal way of expressing this contrast type, in English the omission of an overt marker is usually aimed at emphasizing the parallelism between the wrong and the correct answer, thus expressing the conﬂict only by means of the structural parallelism and intonation. (6.3) I don’t want to do a Ph.D, I want to ﬁnd a job in a publishing company.

Combination and contrast

199

6.2.1.2 The combination-contrast conceptual space The semantic domains of the attested constructions coding combination and contrast relations are not casual, as widely argued in section 4.2.2 in the discussion about the contrast conceptual space. General constructions do not select the expressed situations in a random way, but rather convey functionally contiguous relations. The comparison of data relating to the attested semantic domains can be observed in Tables 34 and 35. The results of this comparison are shown in (6.4) and in Table 29, where general and dedicated constructions are jointly examined. The examined combination and contrast types can be sorted along the following combination-contrast conceptual space: (6.4) The combination-contrast conceptual space: S EQ COMB .

· · · S IM COMB . · · · ATEMP. COMB· · · O PP · · · C ORR · · · D ENIAL EXP.

If a coordination marker is used to express more than one combination or contrast relation, it will convey relations that are contiguous on the conceptual space in (6.4). Therefore, if a general marker is used to express relations that do not stand next to each other on the space, it will also be able to express the relations located in between. For instance, if the same connective may be employed for sequential combination and opposition, it will also be used to express simultaneous and atemporal combination. Likewise, if a marker can be used to express an atemporal combination relation and a corrective contrast, it will also be able to convey an oppositive contrast relation. Table 29 shows the attested semantic maps. Each semantic map is exempliﬁed by one language, whose connectives are explicitly indicated. In Mangarayi, asyndesis is the only strategy employed in the expression of combination, opposition and correction, while the contrast generated by a denial of an expectation is overtly signalled by means of the marker gana. The counterexpectative contrast is thus formally opposed to all other cooccurrence relations. As Merlan (1982: 31) argues, this language has no connectives similar to the English and, since a combination between two SoAs is simply conveyed by a rising-falling intonation pattern with a brief pause between the two clauses. The overt marker gana is instead employed to express

200 Combination, contrast and alternative relations in comparison Table 29: The combination-contrast semantic maps: combination and contrast semantic domains. ATP = atemporal combination; – = no overt marker SEQ

SIM

ATP

OPP

CORR

COUN

–

Mangarayi

Tuvaluan

–

Hdi

lá

gana

kae

–

Maori

aa

àmá

hoki –

engari

kuma

Hausa

amman

i

Russian

a

i

S-Croatian pa

Polish

no

a

v‚ec´

i a

German

Italian

ali

ale

und

sondern

e

aber

ma mentre

bensì

però

Combination and contrast

201

that something is contrary to expectation or wish (Merlan 1982: 32), either in reference to the context of communication (as can be observed in (6.5)) or to the ﬁrst SoA. (6.5) Mangarayi, Gunwingguan, Australian n.awugu 0-d / . aymingan gana n.awuwa 0-bal / . añ this side ABS-non.sacred COORD other side ABS-sacred ‘This side is non-sacred, but the other side is sacred.’ (Merlan 1982: 32) The case of Tuvaluan is different, since the only marker shown by the language is a general marker expressing all cooccurrence relations except for sequential combination. In this case, a formal opposition is thus established between sequential combination on the one hand and all the other cooccurrence relations on the other hand. Whereas the connective gana in Mangarayi is a dedicated marker, kae in Tuvaluan is a general one. Further speciﬁcations deﬁning the exact nature of the relevant relation are left to inferential enrichment. As already pointed out in section 3.2, there are no dedicated simultaneous markers, as simultaneous and atemporal combination may always be expressed by the same constructions across languages. What distinguishes sequential combination from simultaneous and atemporal combination is the presence of a reciprocal order of the linked SoAs. The order in which two SoAs linked in a sequential combination are presented cannot be inverted without changing the overall meaning of the combination, in consequence of their successive occurrence in the real world. Two simultaneous SoAs may instead be combined in any order without affecting the overall meaning of the relation, because there is no intrinsic order in their occurrence. Finally, in atemporal combinations the SoAs may either occur at the same point along the time axis or at successive points, but the speaker is simply not interested in specifying their temporality. The relation is established to convey the cooccurrence of two SoAs within a certain lapse of time, regardless of the speciﬁc order in which they actually occur. Hence, atemporal and simultaneous combination are both characterized by the lack of a sequential order between the SoAs they link. The order in which the SoAs occur is only relevant in a sequential combination, and this is plausibly the reason why sequential combination is often coded by dedicated constructions.

202 Combination, contrast and alternative relations in comparison The types represented by Hdi and Maori in Table 29 are characterized by the presence of overt markers for the two extremes of the conceptual continuum, that is, sequential combination on the one hand and counterexpectative contrast on the other hand. In Hdi asyndesis is used to express all the relations but the counterexpectative contrast, for which the Arabic connective àmá is employed. However, besides juxtaposition, sequential combination may also be conveyed by means of the dedicated marker lá, which is on the way to grammaticalization and still shows many verbal features of its ‘t already pointed out in section 3.2.1. The intermediate relations are all expressed by juxtaposition. Maori, on the other hand, employs syndetic strategies to express the three combination relations, corrective and counterexpectative contrast. It shows the dedicated sequential marker aa, the general connective hoki for simultaneous and atemporal combination, and the general connective engari for correction and denial of an expectation, at the right hand pole of the map. Juxtaposition, however, remains a basic strategy for the expression of all combination relations and oppositive contrast, thus co-existing with the use of overt markers. Comparing the types examined so far with the case of Hausa, the ﬁfth semantic map on Table 29, the tendency for counterexpectative contrast to be marked differently from sequential combination relation becomes evident. In Hausa, all combination and contrast relations require syndetic constructions. Temporal and atemporal combination, opposition and correction are conveyed by the general construction with kuma, and only the expression of counterexpectative contrast employs a different, dedicated connective (amman). Japanese shows a similar situation, since the converb -te is normally found in combinative, oppositive and corrective constructions, but in the expression of counterexpectative contrast the sufﬁx -ga is employed. The remaining semantic maps instantiated in Table 29 basically show the attested distributions of general and dedicated markers along the conceptual domains of combination and contrast already discussed in chapter 4. Russian, Serbo-Croatian and Polish have the general connective i used for temporal and atemporal combination, and the general connective a used for discontinuous atemporal combination (see discussion in sections 3.2 and 4.2) and opposition. The latter marker is thus employed across the border between combination and contrast. In languages having such a marker, opposition is always conveyed through it. In languages like German or Italian, which do not

Combination and contrast

203

have such a bridging connective, opposition is instead expressed by means of the general marker also employed for temporal and atemporal combination (like and in English) and in certain cases, by means of a dedicated marker like while, which develops an oppositive meaning from an original subordinate simultaneity one. However, in either case, there is a general construction expressing both simultaneous and atemporal combination (like i in the Slavic languages) and a general construction expressing both discontinuous atemporal combination and opposition (like a in the Slavic languages). In languages like English or Italian, the same general construction is used for all these relations. In Russian (and also in Chechen, Lithuanian and Bulgarian), as can be observed in Table 29, the connective a may be used not only for discontinuous atemporal combination and opposition, but also for correction. A slightly different semantic map has been proposed by Malchukov (2004), within a general cross-linguistic analysis of the recurrent polysemy patterns shown by coordinating connectives. The set of relations examined by Malchukov is wider than the one considered in this study, and encompasses mirative, concessive and comitative relations, which have not been taken into account in this research. The semantic map proposed by Malchukov (2004: 178) is reported in Fig. 4. Alternative relations will not be considered here, since they constitute the object of the next section. Restricting the comparison between Malchukov’s map and the map proposed in (6.4) to the relations they have in common, let us consider the location of what Malchukov labels consecutive, additive, contrastive, correction and adversative. These labels are used by Malchukov to indicate the relation types that are called here, respectively, sequential combination, atemporal combination, opposition, correction and counterexpectative contrast. The comparison of the two maps in Fig. 5 shows a major difference, namely the position of corrective contrast with respect to oppositive (i.e. contrastive in the map) and counterexpectative (i.e. adversative in the map) contrast. In Malchukov’s assessment, as in the one proposed here, the oppositive relation is located between the combinative and the counterexpectative one. However, in the combination-contrast conceptual space proposed in the present study, corrective contrast is placed between the oppositive and the counterexpectative relation, while in Malchukov’s map it is only linked to opposition.

204 Combination, contrast and alternative relations in comparison

Figure 4: Malchukov’s Semantic Map for coordinating connectives (Malchukov 2004: 178)

According to Malchukov’s representation, if an adversative marker is used to express a corrective and a counterexpectative contrast it will also be able to convey an oppositive relation. Likewise, it is possible to have markers expressing the counterexpectative and the oppositive contrast without them being used for correction. In this research, however, many connectives only used for correction and denial of expectation are attested (like ale in Polish, but in English and mais in French, for instance) and there is instead no marker expressing only opposition and denial of expectation. The difference between Malkuchov’s assessment and the combinationcontrast conceptual space proposed in (6.4) is a direct consequence of Malkuchov’s different treatment of sentences like (6.6). (6.6) Paul is rich but Mike is poor. In the approach adopted in this work, such sentences are regarded as examples of counterexpectative contrast. In particular, following Lang (2000), the location of the expectation identiﬁes two types of counterexpectative contrast. In

Combination and contrast

205

Figure 5: Comparison between Malchukov’s Semantic Map for coordinating connectives and the combination-contrast conceptual space proposed in (6.4)

the ﬁst type, the expectation is generated by the ﬁrst SoA, while in the second type it originates outside the sentence, within the context of communication. A sentence like (6.6) is thus examined as a counterexpectative relation of the second type, since the denied expectation is that Paul and Mike are both rich (cf. discussion in section 4.1.2). Opposition has instead been deﬁned as a symmetric contraposition of two somehow antonymic situations, without the negation of any expectation. In English, opposition is normally rendered by means of the general construction expressing temporal and atemporal combination together with some special intonation or by means of the dedicated marker while/whereas. Malchukov (2004: 179-84), on the other hand, regards sentences such as (6.6) as instances of oppositive contrast, following Lakoff’s (Lakoff 1971) deﬁnition of ‘semantic opposition’ (cf. discussion in chapter 4). He analyzes the sentence in (6.6) as involving no denial of expectation at all, but only a symmetric opposition. Therefore, in his account of contrast markers, the English connective but is used to express denial of an expectation, correction

206 Combination, contrast and alternative relations in comparison and also opposition. In particular, he states that this marker is able to express counterexpectative and corrective contrast, which are distant on the map he proposes, insofar as it may be also used to express opposition, which is located between the corrective and the counterexpectative contrast (Malchukov 2004: 184, 193). This would hold for all the general markers that have been identiﬁed here as expressing the relations of counterexpectative and corrective contrast, like Italian ma, French mais, Polish ale, etc. (cf. Table 29). 6.2.1.3 The coding complexity hierarchy of combination and contrast relations The morphophonological complexity of the markers used to express combination and contrast relations reveals another cross-linguistic implicational pattern, conﬁrming the form-function asymmetry already discussed in section 4.2.3 within the analysis of contrast connectives. The relevant data have been reported in Table 32 and 33. The regular pattern of variation revealed by their comparison can be stated as follows: (6.7) The combination-contrast coding complexity hierarchy: dedicated marker for sequential combination, general marker expressing at least one combination relation > general marker only expressing contrast relations > dedicated marker for a contrast relation. If a language has one of the markers indicated on the hierarchy, it will be at least as morphophonologically complex as the markers to its left. This hierarchy includes the one identiﬁed in (4.28) within the coding of contrast relations. The only counterexample found in the samples is Mosetén, in which the combination marker jö’dyë’yä is more complex than the contrast markers -tsa’ and -ki. The hierarchy states that dedicated markers encoding counterexpectative, oppositive and corrective contrast are at least as complex as the general markers used only for contrast relations, i.e. markers employed for corrective and counterexpectative contrast. These general contrast markers are in turn at least as complex as dedicated sequential markers and general markers used for at least one combination relation.

German: Basque: Finnish: Georgian: Italian: Albanian: French: Hungarian: S-Croatian: Chechen: Russian: Polish: Maori: Supyire:

– – – – – – – – pa – – – aa kà

TIAL

DEDICATED SEQUEN -

und eta ja da e e/dhe et és, meg i, a ’a, tq’a i, a i, a hoki sì

GENERAL ALSO FOR COMBINA TION

> > > > > > > > > > > > > > – – – – ma por mais – – – – ale engari –

GENERAL ONLY FOR CONTRAST

> > > > > > > > > > > > > > aber baina mutta magram però – – de ali amma no – – ï` kàà

DEDICATED COUNTER .

sondern baizik vaan – bensì – – hanem n‚ego/v‚ec´ – – – – –

DEDICATED CORR .

während – – k’i mentre kurse tandis que pedig – – – – – –

DEDICATED OPPOS .

Table 30: The combination-contrast coding complexity hierarchy: attested complexity patterns. – = absence of the given marker

Combination and contrast

207

208 Combination, contrast and alternative relations in comparison Whenever the oppositive contrast is expressed by a general construction, this construction will express opposition and atemporal combination. The label ‘general marker expressing at least one combination relation’ encompasses many different semantic domains. To this set belong markers like Maori hoki, used for simultaneous and atemporal combination, Polish and Russian i, employed for temporal and atemporal combination, English and and Italian e, employed in the expression of temporal and atemporal combination and opposition, and markers like -te in Japanese, used for the expression of all combination and contrast relations except for the counterexpectative one. Likewise, general markers like Polish a, used for atemporal combination and opposition (and in case of Russian, also for correction), or connectives like Tuvaluan kae, expressing atemporal combination and all contrast relations, are also included in this set, since they express at least one combination relation, namely the atemporal one. On the other hand, markers like ale in Polish, ma in Italian and por in Albanian, are general markers, but they are used only to convey contrast relations, namely corrective and counterexpectative contrast. Instead, connectives like German aber for counterexpectative contrast and sondern for correction, French tandis que for opposition and Russian no for the counterexpectative contrast are dedicated markers for the three contrast types. The attested language types are represented in Table 30, where the general and dedicated markers are ranged in the same order as in the hierarchy, to highlight the increasing morphophonological complexity. Each of the attested conﬁgurations is illustrated by one language. This hierarchy conﬁrms the correspondence between the functional versatility of a marker and its morphophonological simplicity. General markers are located at the left hand of the hierarchy, while dedicated ones are more complex, and thus stand at the right hand of the hierarchy. The fact that sequential markers are located at the left hand of the hierarchy, even though they are dedicated connectives, depends on the basic nature of the combination relation they convey. Combination has been deﬁned in negative terms, that is, as a relation establishing the cooccurrence of two SoAs without further speciﬁcation of the reason or the nature of this relation (see section 3.1). The location of two combined SoAs on the temporal axis distinguishes between sequential and non-sequential combination. However, this distinction only gives information about the temporality of the SoAs and does not provide any further speciﬁca-

Combination and contrast

209

tion regarding the nature of the relation itself, which simply remains one of cooccurrence. Hence, dedicated sequential markers still convey a basic and semantically simple relation, which is very frequently established in discourse, especially in narrative contexts. In other words, as will be discussed in section 6.2.3, combination relations, be they temporal or atemporal, constitute a basic way of linking two SoAs in discourse, since they establish their cooccurrence in a very general and unspeciﬁed relation. The basic nature of combination determines the high frequency of use of markers, which therefore tend to show phenomena of phonological reduction typically associated with frequency of use. The morphophonological complexity of dedicated markers expressing contrast relations has been already discussed in section 4.2.3, where some cases of polymorphemic markers have been examined. The functional motivations underlying this hierarchy will be further explained in detail in section 6.2.3.

6.2.2

The combination-contrast parallelism implication

Data on the coding of combined and conﬂicting SoAs are reported in Tables 36 and 37, where information about European and non-European languages is provided. The comparison of these data reveals a cross-linguistic implicational pattern that can be stated as follows: (6.8) The combination-contrast parallelism implication: non-parallel construction for simultaneous and atemporal combination, opposition, correction, denial of expectation → non-parallel construction for sequential combination In a given language, if a syntactically non-parallel construction may be used to convey a contrast relation or a non-sequential combination, either simultaneous or atemporal, a non-parallel construction will also be available to express a sequential combination. In other words, if a non-parallel construction is at all available in a given language, it will certainly be used to express sequential combinations. The implicational pattern presented in (6.8) is the synthesis of the combination parallelism implication presented in (3.20) together with the data

210 Combination, contrast and alternative relations in comparison on contrast constructions presented in section 4.3. Two languages, Japanese and Jamul Tiipay, have a non-parallel construction for combination relations, opposition and correction, but use a parallel construction in the expression of counterexpectative contrast. In both cases, the non-parallel construction is the same as the one used for the expression of sequential combination (cf. discussion in section 4.3). However, two languages are not enough to constitute a cut-off point along a hierarchy and it is not possible to exclude that the implication is due to chance. In Table 31 all the attested conﬁgurations of parallel and non-parallel combination and contrast constructions are summarized. Table 31: The combination-contrast parallelism implication: attested types. + = syntactically parallel construction; - = syntactically non-parallel construction

Tauya Korean West Greenlandic Jamul Tiipay Japanese Supyire Turkish Lango Hebrew Vietnamese

seq +/+/+/ +/ + +

Temporal sim +/+/+ + + + +

Atemp

Oppos

Correct

+/+/+ + + + +

+/-

-

+ + + +

+/+ + + + +

Denial expect +/+/+ + + + + + +

West Greenlandic, Korean, Kolyma Yukaghir and Tauya are the only languages having non-parallel constructions for both combination and contrast relations. Examples from these languages have already been discussed in detail in section 4.3, and they will not be repeated here. Languages having a non-parallel construction only for sequential combination are attested in both the European and the Comparison sample. Among the European languages, only Turkish and Lezgian make use of deranked verb forms, while outside Europe sequential non-parallel constructions are attested in a number of languages, like Harar Oromo, Lango, Marathi and Supyire.

Combination and contrast

6.2.3

211

Functional motivations

The coding implication of combination and contrast relations in (6.1) states that if one of the three relations of combination, oppositive or corrective contrast may be expressed by means of a syndetic construction, such a construction will also be available for the expression of counterexpectative contrast. In other words, there are languages in which combination, opposition and correction may be expressed by a juxtaposition whereas counterexpectative contrast requires an overt connective, but there are no languages in which counterexpectative contrast may be expressed by juxtaposition and all the other cooccurrence relations necessarily require overt markers. Relations of combination, opposition and correction are more easily inferable than counterexpectative contrast and are thus more likely to be expressed without any overt marker. In order to infer a combination relation, the hearer is simply required to identify the two SoAs as cooccurring within a common frame. Contrast relations instead are more complex to infer, because they require the identiﬁcation of a conﬂict within the cooccurrence. Nonetheless, the conﬂict characterizing oppositive and corrective contrast is easily inferable from the parallel semantics of the SoAs even when no overt marker is present (as argued in section 4.4). In correction and opposition the conﬂict lies in the very semantics of the linked SoAs. Two opposed SoAs must be recognized as somehow antonymic with respect to a common topic, while two SoAs linked in a corrective contrast must be recognized as antonymic and must show opposite polarity - the ﬁrst is overtly negated and the second is positively asserted. On the other hand, the conﬂict characterizing the counterexpectative contrast does not reside in the semantics of the two SoAs, but in the contradiction between one SoA and the expectations which usually accompany the other one. The identiﬁcation of the counterexpectative conﬂict thus requires the hearer to establish a relation between one of the SoAs and the expectations generated by the other. The absence of any overt connective may lead to relate the two SoAs, without considering the assumptions underlying them (cf. discussion of example (4.34)). The conﬂict generated by the denial of an expectation is thus the most complex to infer from the simple juxtaposition of two SoAs and is the most likely to be expressed by an overt connective. The way in which combination and contrast relations are distributed on the two sides of the coding implication partly mirrors their location within the combination-contrast conceptual space depicted in Table 29. The coding

212 Combination, contrast and alternative relations in comparison implication has been explained with respect to the increasing difﬁculty of inferring the various relations when no overt marker is present. The conceptual space, on the other hand, is analyzed on the basis of their functional contiguity. Two coordination relations are close to each other on the conceptual space by virtue of their sharing conceptual features or being easily inferred one from the other. The contiguity of temporal and atemporal combination can be accounted for by the fact that they convey the same basic cooccurrence relation, whose nature is not further speciﬁed (the parameter of temporality is independent of the nature of the relation as such, and it only pertains to the linear ordering of the linked SoAs along the temporal axis). Simultaneous and atemporal combinations are close to each other on the combination-contrast conceptual space because they share the absence of an order between the linked SoAs and they are often associated in discourse. While two SoAs linked in a sequential combination occur in a certain order (6.9), whose reversal causes a change in the overall meaning of the relation, two simultaneous SoAs have no intrinsic order (6.10). Likewise, two SoAs linked in an atemporal combination have no order because the presence of any intrinsic order is irrelevant to the relation, which is simply one of cooccurrence within a certain lapse of time. In (6.11) the two SoAs may either have occurred at the same time or one after the other, but this is simply not relevant. (6.9) We went to the cinema and met my sister. (6.10) It was a scary movie: we held our hands and shivered the whole time. (6.11) When we came home, I prepared the dinner and my sister looked for a nice movie to watch on TV. What seems to be mostly salient in discourse is not the precise collocation of the SoAs on the time axis, but rather the presence vs. absence of a sequential order between them (cf. symmetric vs. asymmetric ’and’, Lakoff 1971: 155126). Since simultaneous and atemporal combination both lack an intrinsic order, speakers often do not resort to different constructions to code them. Moving rightwards on the conceptual space we ﬁnd opposition. The relation of opposition implies a discontinuous atemporal combination of two SoAs, that is, a combination of two SoAs which do not affect each other

Combination and contrast

213

and show a change in their participants and actions, regardless of their temporality. In an oppositive relation the discontinuous atemporal combination is further speciﬁed as generating a conﬂict, due to the somehow antonymic nature of the linked SoAs (cf. section 4.1). In discourse, an oppositive relation is frequently inferred from an atemporal combination, since contrast does not reside in the world as such but is projected onto things by the speaker’s perspective. Therefore, given an atemporal combination of two discontinuous SoAs, the hearer may project his own perspective and identify a conﬂict within the combination, even if the speaker does not explicitly indicate it. The two relations of atemporal combination (in its discontinuous instances) and opposition thus only differ with respect to the degree of attention paid to the differences existing between the linked SoAs. This in turn depends on whether emphasis is placed on the cooccurrence as such or on the contrast between the linked SoAs. Such a subtle border is mirrored by the cross-linguistic coding of the two relations, which may always be expressed by means of the same construction (see discussion in section 6.2.1.2). Along the combination-contrast conceptual space, opposition is followed by correction and counterexpectative contrast. The reasons underlying this disposition have been widely discussed in section 4.4, therefore they will be only brieﬂy summarized here. Opposition and correction share the atemporal combination of discontinuous SoAs and the presence of a conﬂict generated by the antonymic nature of the linked SoAs. Furthermore, a corrective contrast implies that either the ﬁrst SoA has been previously asserted or there is an expectation for the ﬁrst SoA to be valid. Hence, it consists of the explicit denial of that expectation, followed by the positive assertion of the SoA substituting it. The latter property of correction determines its conceptual contiguity with the counterexpectative relation, in that the two contrast relations share the denial of some expectation. In corrective contrast the expectation is explicitly disputed and constitutes the positive correspondent of the ﬁrst SoA, while in counterexpectative contrast the expectation remains implicit. Let us now examine the motivations for the coding complexity hierarchy. In (6.7) it has been stated that dedicated markers coding a contrast relation are at least as morphophonologically complex as general markers used for corrective and counterexpectative contrast, which in turn are at least as morphophonologically complex as general and dedicated markers used for at least one combination relation.

214 Combination, contrast and alternative relations in comparison The more general a coordinating marker is, the more frequently it will be used and will undergo phenomena of phonological erosion, which are typically associated with a high frequency of use. This may account for the reason why general markers tend to be morphophonologically simpler than the dedicated ones. Moreover, it is plausible to assume that the more basic and semantically unspeciﬁed a conceptual relation is, the more it tends to correlate with a high frequency of use. Hence, since combination is the simplest coordination relation, in that it only establishes the cooccurrence of two SoAs, it is the most frequently used in discourse (Ohori 2004: 61). Therefore, markers used to express at least one combination relation, either general or dedicated, tend to be simpler than markers only used to express contrast. The reasons identiﬁed to explain the combination-contrast coding implication and the coding complexity are both based on the principle of economy. On the one hand, the absence of an overt coordinating marker correlates with the degree to which a speciﬁc conceptual relation is inferable from the context, since what is already available from the context does not need to be further indicated. On the other hand, the morphophonological complexity of a given marker indirectly correlates with its frequency of use, because most frequent forms tend to undergo a phonological reduction due to the tendency to economize on the length of the most used expressions. The principle of economy plays a major role also in the explanation of the combination-contrast parallelism hierarchy. As stated in (6.8), the presence of a non-parallel construction in the expression of a contrast relation or a nonsequential combination, either simultaneous or atemporal, implies the use of a non-parallel construction also in the expression of sequential combination. The four languages showing a non-parallel construction in the expression of all combination and contrast relation have been already analyzed in sections 4.3 and 6.2.2, where the nature of the attested non-parallel constructions has been examined. The phenomenon that requires an explanation is the asymmetric distribution of non-parallel coordinating constructions in languages like Turkish or Marathi, where it is only attested for the expression of sequential combination. As pointed out in chapter 3 (sections 3.3 and 3.4), in all languages showing this conﬁguration, the dedicated non-parallel sequential construction is characterized by a marker coding the sequential temporal meaning and by a deranked verb form lacking tense and mood speciﬁcations. This is the case for -Ip in Turkish (see example (1.8)), -na in Lezgian (see example (2.22)), -é

Combination and contrast

215

in Harar Oromo (example (3.18)), tˆE in Lango, -¯un in Marathi (see example (3.25)) and kà/mà in Supyire. The presence of an overt marker coding the temporal relation of the SoAs makes the successive temporal location of the SoAs explicit. According to the economic principle of information recoverability, already discussed in the explanation of the combination-contrast coding hierarchy, information that is recoverable from the context is not further speciﬁed. Given the temporal collocation of one SoA, a sequential temporal relation predetermines the collocation of the other one, which will precede or follow, depending on which SoA is fully temporally characterized. Given the presence of a dedicated sequential marker, if one of the SoAs is located in time, i.e. coded by means of a balanced verb form, the temporal location of the other one is recoverable from the context. Together with the temporal collocation, an overt sequential marker also predetermines the aspect and the mood of the sequential SoAs. As argued in section 3.4, two sequentially combined SoAs are jointly asserted or elicited and share the same mood. Furthermore, SoAs linked in a sequential combination are also likely to have the same aspectual features. Particularly in narrative contexts, sequential SoAs are often coded by perfective verb forms, which indicate actions realized at successive moments. The presence of dedicated sequential non-parallel constructions is thus motivated in economic terms. The temporal collocation of the linked SoAs and their aspect and mood features are predetermined by the sequential relation itself, which is overtly coded by a dedicated marker. Hence, there is no need to specify tense, aspect and mood explicitly for both SoAs. If one SoA is coded by a ﬁnite verb, showing all the relevant information, the temporal location of the other will follow. The same phenomenon may occur in the expression of simultaneous combination, since also in this case the time reference of two simultaneous SoAs is predetermined by the relation. However, simultaneous combination is always expressed by the same construction used also for the expression of atemporal combination, so there are no dedicated markers coding the simultaneity of two coordinated SoAs (cf. discussion in sections 3.2 and 3.4). Hence, the attested overt markers are always ambiguous between a simultaneous and an atemporal meaning. As a result, even though the modal and aspectual properties of the SoAs are plausibly the same, the temporal location of the SoAs is not recoverable from the context and thus needs to be overtly marked on each verb form.

216 Combination, contrast and alternative relations in comparison 6.2.4

Language data

Table 32: European Sample. Coding of combination and contrast relations in comparison: morphophonological complexity of the attested coordinating markers. + = presence of the given feature; - = absence of the given feature; opp.= opposition; atemp.= atemporal combination; comb.= combination; cor= correction; coun= counterexpectative contrast; blank= no information available Language

Marker

Semantic domain

Free

Polysyllabic

Polymorphem.

Albanian

e/dhe por kurse eta baina baizik i a alĲe i a

all comb. types coun., cor. contrast opp. contrast all comb. types coun. contrast cor. contrast all comb. types atemp.comb., opp. coun., cor. contrast all comb. types atemp.comb., opp., corr. sequential comb. coun. contrast all comb. types coun. contrast cor. contrast all comb. types atemp.comb., opp., corr. coun. contrast all comb. types coun., cor. contrast all comb. types coun., cor. contrast all comb. types all comb. types coun. contrast coun. contrast all comb. types coun., cor. contrast all comb. types coun., cor. contrast

+ + + + + + + + + + +

+ + + + -

-

+ + + + + +

+ + -

-

+ + + + + + + + + + +

+ + + -

-

Basque

Belorussian

Bulgarian

Catalan

Chechen

Czech Danish Dargi

Dutch English

ta/da no i pero sino ’a tq’a amma a ale og men wa -ra -gu amma en maar and but

Combination and contrast

217

(Continued) Language

Estonian

Finnish

French

Georgian

German

Greek Hungarian

Icelandic

Irish Italian

Latvian Lezgian

Lithuanian

Marker

Semantic domain

Free

Polysyllabic

Polymorphem.

whereas ja aga kuid vaid ja mutta vaan et mais tandis que da aramed magram ki und aber sondern während kai alla és meg de hanem og en heldur agus ach e ma però bensì un bet -ni, wa -na amma ir

opp. contrast all comb. types coun. contrast coun. contrast cor. contrast all comb. types coun. contrast cor. contrast all comb. types coun., cor. contrast opp. contrast

+ + + + + + + + + + +

-/+ + + +

-/+ +

all comb. types cor. contrast coun. contrast opp. contrast all comb. types coun. contrast cor. contrast opp. contrast all comb. types coun., cor. contrast all comb. types atemp.comb., opp. coun. contrast cor. contrast all comb. types coun. contrast cor. contrast all comb. types coun., cor. contrast all comb. types coun., cor. contrast coun. contrast cor. contrast all comb. types coun., cor. contrast all comb. types sequential comb. coun. contrast all comb. types

+ + + + + + + + + + + + + + + + + + + + + + + + + -/+ + +

+ + + + + + + + + + + + -

+ -

218 Combination, contrast and alternative relations in comparison (Continued) Language

Luxembourgish Maltese Norwegian Polish

Portuguese Romanian

Russian

Sardinian Serbo-Croatian

Spanish

Swedish

Turkish

Marker

Semantic domain

Free

Polysyllabic

Polymorphem.

o

atemp.comb., opp., cor. coun. contrast all comb. types coun., cor. contrast all comb. types coun., cor. contrast all comb. types coun., cor. contrast all comb. types atemp.comb., opp. coun., cor. contrast all comb. types coun., cor. contrast all comb. types atemp.comb., opp. coun. contrast cor. contrast all comb. types atemp.comb., opp., cor. coun. contrast all comb. types coun., cor. contrast all comb. types sequential comb. atemp.comb., opp. coun. contrast cor. contrast cor. contrast all comb. types coun. contrast cor. contrast all comb. types coun. contrast cor. contrast all comb. types all comb. types sequential comb. coun., cor. contrast coun., cor. contrast

+

-

-

+ + + + + + + + + + + + + + + + + +

+ + -

-

+ + + + + + + + + + + + + + + + + +

+ + + + + + +

-

bèt ann mee u imma og men i a ale e mas s¸i iar dar ci i a no e ma i pa a ali n‚ego v‚ec´ y pero sino och men utan =dA ve -Ip fakat ama

Combination and contrast

219

(Continued) Language

Marker

Semantic domain

Free

Polysyllabic

Polymorphem.

Ukrainian

i ta a alĲe

all comb. types atemp.comb., opp. atemp.comb., opp. coun., cor. contrast

+ + + +

+

-

Table 33: Comparison Sample. Coding of combination and contrast relations in comparison: morphophonological complexity of the attested coordinating markers.+ = presence of the given feature; - = absence of the given feature; opp.= opposition; sim.= simultaneous combination; atemp.= atemporal combination; comb.= combination; cor= correction; coun= counterexpectative contrast; blank= no information available Language

Marker

Semantic domain

Free

Polysyllabic

Polymorphem.

Arabic

Iraqw Jamul Tiipay

wa fa laakin ﬁè¯i bal -k@ =ĳií -é ﬁ k’ófáamô immó kuma àmmaa lá àmá veaval éla ánaga ˛ nu˛ni˛ge nee switch-r

+ + + + + + + + + + + + + + + + + -

+ + + + + + + + + + + -

+ + -

Japanese

-pes -te

all comb. types sequential comb. coun. contrast opp. contrast cor. contrast all comb. types all comb. types sequential comb. sim., atemp. comb. coun. contrast coun. contrast all comb. types coun., cor. contrast sequential comb. coun. contrast all comb. types coun. contrast cor. contrast all comb. types coun., cor. contrast all comb. types all comb. types, opp., corr. coun. contrast all comb. types, opp., corr.

-

-

-

Dumi Hakha Lai Harar Oromo

Hausa Hdi Hebrew

Hocak ˛

220 Combination, contrast and alternative relations in comparison (Continued) Language

Kisi Kol. Yukaghir Korean

Koromfe Kuskokwim A. Lango Limbu Malayalam

Mangarayi Maori

Marathi

Maricopa Meithei Mosetén

NànáfwˆE Ndyuka Persian

Marker

Semantic domain

Free

Polysyllabic

Polymorphem.

si -i -ga k´E (mi)

all comb. types atemp.comb., opp. coun. contrast coun. contrast

+ +

-/+

-

-ko kuliko kulena -manun -(u)na la

+ + +

+ + + -

-

ts’eĳ ĳedinh tˆE E´ ntˆO -an kere -um pakùe ennaal eNkilum gana aa hoki engari a¯ n.i wa pan. parãntu

all comb. types all comb. types coun. contrast coun. contrast coun. contrast atemp.comb., opp., cor., coun. all comb. types coun., cor. contrast sequential comb. coun. contrast all comb. types coun. contrast all comb. types coun. contrast coun. contrast coun. contrast coun. contrast sequential comb. sim., atemp. comb. coun., cor. contrast all comb. types all comb. types coun., cor. contrast coun., cor. contrast

+ + + + + + + + + + + + + + + +

+ + + + + + + + + + +

+ -

aduga adupu jö’dyë’yä -tsa’ -ki kp´Ekún sàng´E da ma va

all comb. types coun. contrast sim., atemp. comb. coun. contrast opp. contrast all comb. types coun., cor. contrast all comb. types coun. contrast all comb. types

+ + + + + + + +

+ + + + + -

+ -

Combination and contrast

221

(Continued) Language

Rapanui Somali

Supyire

Tauya Tukang Besi

Tuvaluan Vietnamese

Wari’ W.Greenlandic

Marker

Semantic domain

Free

Polysyllabic

Polymorphem.

´ vœli ´ bœlke e pero -na oo -se laakìin kà/mà sì

coun. contrast cor. contrast all comb. types coun. contrast all comb. types all comb. types coun. contrast coun. contrast sequential comb. atemp.comb., opp., cor., coun. coun. contrast all comb.and contrast types sequential comb. sim., atemp. comb. coun., cor. contrast atemp.comb., opp., cor., coun. all comb. types coun., cor. contrast coun. contrast opp. contrast all comb. types none all comb. types coun., cor. contrast coun., cor. contrast

+ + + + + + + +

+ + + + -

+ -

+ -

+ -

-

+ + + +

+ + + -

-

+ + + + +

+

+

+

+

-

ï` kàà switch-r maka kene toka kae và mà nhung còn ca’ na =lu =li kisianni

Albanian Basque Belorussian Bulgarian Catalan Chechen Czech Danish Dargi Dutch English Estonian Finnish French Georgian German Greek Hungarian Icelandic Irish Italian

Language

Gα Gα Gα Gα /d Gα -/Gα Gα Gα Gα Gα Gα Gα Gα Gα Gα Gα Gα Gα Gα Gα Gα

Sequential Gα Gα Gα Gα Gα -/Gα Gα Gα Gα Gα Gα Gα Gα Gα Gα Gα Gα Gα Gα Gα Gα

Simultaneous

Temporal combination

Gα Gα Gα /Gβ Gα /Gβ Gα -/Gα /Gβ Gα Gα Gα Gα Gα Gα Gα Gα Gα Gα Gα Gα /Gβ Gα Gα Gα

Atemporal comb Gβ d Gγ Gβ d Gβ Gβ Gβ Gβ Gβ d d Gβ -/d d Gβ d d Gβ Gβ /d

Gα Gα /d Gα Gα Gα /d -/d Gα /d Gα Gα Gα Gα Gα /d

Corrective contrast

Gα /d Gα Gβ Gβ Gα Gβ Gα Gα

Oppositive contrast Gβ d Gγ d d d Gβ Gβ d Gβ Gβ d d Gβ d d Gβ d d Gβ Gβ /d

Counterex contrast

Table 34: European Sample. Coding of combination and contrast relations in comparison: presence of overt coordinating markers and their semantic domain. - = absence of an overt marker; d= dedicated; Gx = general, with x identifying a given construction attested in the language; blank= no information available

222 Combination, contrast and alternative relations in comparison

Latvian Lezgian Lithuanian Luxembourgish Maltese Norwegian Polish Portuguese Romanian Russian Sardinian Serbo-Croatian Spanish Swedish Turkish Ukrainian

Language

(Continued)

Gα Gα /d Gα Gα Gα Gα Gα Gα Gα Gα Gα Gα /d Gα Gα -/Gα /d Gα

Sequential Gα Gα Gα Gα Gα Gα Gα Gα Gα Gα Gα Gα Gα Gα -/Gα Gα

Simultaneous

Temporal combination

Gα Gα Gα /Gβ Gα Gα Gα Gα /Gβ Gα Gα /Gβ G α / Gβ Gα Gα /Gβ Gα Gα -/Gα Gα /Gβ

Atemporal comb Gβ Gβ Gβ Gβ Gβ Gγ Gβ d Gβ Gβ d d d -/Gβ Gγ

Gβ Gα Gα Gα Gβ Gα Gβ Gβ Gα Gβ Gα Gα -/Gα Gβ

Corrective contrast

Gα

Oppositive contrast Gβ d d Gβ Gβ Gβ Gγ Gβ d d Gβ d d d Gβ Gγ

Counterex contrast

Combination and contrast

223

Arabic Dumi Hakha Lai Harar Oromo Hausa Hdi Hebrew Hocak ˛ Iraqw Jamul Tiipay Japanese Kisi Kol. Yukaghir Korean Koromfe Kuskokwim A. Lango Limbu Malayalam Mangarayi Maori

Language

Gα /d Gα Gα d -/Gα -/d Gα Gα Gα Gα Gα Gα Gα -/d -/Gα -/Gα -/d

Sequential Gα -/Gα Gα Gα -/Gα Gα Gα Gα Gα Gα Gα -/Gα Gα -/Gα -/Gα -/Gα

Simultaneous

Temporal combination

Gα -/Gα Gα Gα -/Gα Gα J Gα Gα Gα Gα /Gβ Gα -/Gα Gα -/Gα -/Gα -/Gα

Atemporal comb

Gα Gα -

Gα Gα /Gβ Gα Gα

-

-

G2 d Gβ

Gα Gα

Gβ

Gα Gβ -

d Gα

Corrective contrast

Gα /d Gα

Oppositive contrast

d d d d Gα Gβ d d d d Gβ

d Gβ d d Gβ

d

Counterex contrast

Table 35: Comparison Sample. Coding of combination and contrast relations in comparison: presence of overt coordinating markers and their semantic domain. - = absence of an overt marker; d= dedicated; Gx = general, with x identifying a given construction attested in the language; blank= no information available

224 Combination, contrast and alternative relations in comparison

Marathi Maricopa Meithei Mosetén NànáfwˆE Ndyuka Persian Rapanui Somali Supyire Tauya Tukang Besi Tuvaluan Vietnamese Wari’ W.Greenlandic

Language

(Continued)

-/Gα -/Gα -/Gα -/Gα Gα -/Gα Gα d Gα -/d -/Gα -/Gα -/Gα

Sequential -/Gα -/Gα -/Gα -/Gα -/Gα Gα -/Gα Gα Gα Gα -/Gα -/Gα -/Gα -/Gα Gα

Simultaneous

Temporal combination

-/Gα -/Gα -/Gα -/Gα -/Gα Gα -/Gα Gα Gα Gα -/Gα -/Gα -/Gα -/Gα Gα

Atemporal comb

Gα Gα Gα d -

d Gα Gα -

Oppositive contrast

Gα Gα Gβ Gα Gβ Gβ

Gβ d -

Gβ

Corrective contrast

d d Gβ d d d d Gα /d Gα Gβ Gα Gβ /d Gβ

Gβ

Counterex

Combination and contrast

225

Albanian Basque Belorussian Bulgarian Catalan Chechen Czech Danish Dargi Dutch English Estonian Finnish French Georgian German Greek Hungarian Icelandic Irish Italian

Language

Gα + Gα + Gα + Gα +/d+ Gα + Gα + Gα + Gα + Gα + Gα + Gα + Gα + Gα + Gα + Gα + Gα + Gα + Gα + Gα + Gα + Gα +

Sequential Gα + Gα + Gα + Gα + Gα + Gα + Gα + Gα + Gα + Gα + Gα + Gα + Gα + Gα + Gα + Gα + Gα + Gα + Gα + Gα + Gα +

Simultaneous

Temporal combination

Gα + Gα + Gα +/Gβ + Gα +/Gβ + Gα + Gα +/Gβ + Gα + Gα + Gα + Gα + Gα + Gα + Gα + Gα + Gα + Gα + Gα + Gα +/Gβ + Gα + Gα + Gα +

Atemporal comb Gβ + d+ Gγ + Gβ + d+ Gβ + Gβ + Gβ + Gβ + Gβ + d+ d+ Gβ + d+ d+ Gβ + d+ d+ Gβ + Gβ +/d+

Gα + Gα +/d+ Gα + Gα + Gα +/d+ d+ Gα +/d+ Gα + Gα + Gα + Gα + Gα +/d+

Corrective contrast

Gα /d+ Gα + Gβ + Gβ + Gα + Gβ + Gα + Gα +

Oppositive contrast

Gβ + d+ Gγ + d+ d+ d+ Gβ + Gβ + d+ Gβ + Gβ + d+ d+ Gβ + d+ d+ Gβ + d+ d+ Gβ + Gβ +/d+

Counterex contrast

Table 36: European Sample. Combination and contrast in comparison: syntactic parallelism of the attested constructions. + = syntactically parallel; - = syntactically non parallel; d= dedicated; Gx = general, with x identifying a given construction attested in the language; blank= no information available

226 Combination, contrast and alternative relations in comparison

Latvian Lezgian Lithuanian Luxembourgish Maltese Norwegian Polish Portuguese Romanian Russian Sardinian Serbo-Croatian Spanish Swedish Turkish Ukrainian

Language

(Continued)

Gα + Gα +/d Gα + Gα + Gα + Gα + Gα + Gα + Gα + Gα + Gα + Gα +/d+ Gα + Gα + Gα +/d Gα +

Sequential Gα + Gα + Gα + Gα + Gα + Gα + Gα + Gα + Gα + Gα + Gα + Gα + Gα + Gα + Gα + Gα +

Simultaneous

Temporal combination

Gα + Gα + Gα +/Gβ + Gα + Gα + Gα + Gα +/Gβ + Gα + Gα +/Gβ + Gα +/Gβ + Gα + Gα +/Gβ + Gα + Gα + Gα + Gα +/Gβ +

Atemporal comb Gβ + Gβ + Gβ + Gβ + Gβ + Gγ + Gβ + d+ Gβ + Gβ + d+ d+ d+ Gβ + Gγ +

Gβ + Gα + Gα + Gα + Gβ + Gα + Gβ + Gβ + Gα + Gβ + Gα + Gα + Gα + Gβ +

Corrective contrast

Gα +

Oppositive contrast Gβ + d+ d+ Gβ + Gβ + Gβ + Gγ + Gβ + d+ d+ Gβ + d+ d+ d+ Gβ + Gγ +

Counterex contrast

Combination and contrast

227

Arabic Dumi Hakha Lai Harar Oromo Hausa Hdi Hebrew Hocak ˛ Iraqw Jamul Tiipay Japanese Kisi Kol. Yukaghir Korean Koromfe Kuskokwim A. Lango Limbu Malayalam Mangarayi Maori

Language

Gα +/d+ Gα +/Gβ + Gα + dGα + Gα +/d+ Gα + Gα + Gα + Gα Gα -/+ Gα + Gα Gα -/+ Gα + Gα + Gα +/dGα + Gα + Gα + d+

Sequential Gα + Gα +/Gβ + Gα + Gα + Gα + Gα + Gα + Gα + Gα + Gα Gα -/+ Gα + Gα Gα -/+ Gα +/Gβ + Gα + Gα + Gα + Gα + Gα + Gα +

Simultaneous

Temporal combination

Gα + Gα +/Gβ + Gα + Gα + Gα + Gα + Gα + Gα + Gα + Gα Gα -/+ /Gβ Gα + Gα Gα -/+ Gα + Gα + Gα + Gα + Gα + Gα + Gα +

Atemporal comb

Gα Gα Gα +

Gα Gα -/+ /Gβ Gα +

Gβ +

Gα + Gα + Gα +

Gβ +/Gγ +

Gβ + Gβ + Gα + Gα + Gα +

Gβ + Gα + d+ Gβ +

Gα + Gα + Gα +

Gα +/Gβ +

d+ Gβ +

Corrective contrast

Gα +/d+ Gβ +

Oppositive contrast

d+ d+ d+ Gα d+/Gβ + Gβ + Gα +/d+ Gα +/d+ d+ d+ Gγ +

d+ Gβ + d+ d+ Gβ +

d+

Counterex contrast

Table 37: Comparison Sample. Combination and contrast in comparison: syntactic parallelism of the attested constructions. + = syntactically parallel; - = syntactically non parallel; d= dedicated; Gx = general, with x identifying a given construction attested in the language; blank= no information available

228 Combination, contrast and alternative relations in comparison

Marathi Maricopa Meithei Mosetén NànáfwˆE Ndyuka Persian Rapanui Somali Supyire Tauya Tukang Besi Tuvaluan Vietnamese Wari’ W.Greenlandic

Language

(Continued)

Gα +/dGα -/+ Gα + Gα + Gα + Gα + Gα + Gα + Gα + dGα Gα +/d+ Gα + Gα + Gα + Gα -/+

Sequential Gα + Gα -/+ Gα + Gα +/Gβ + Gα + Gα + Gα + Gα + Gα + Gα + Gα Gα +/Gβ + Gα + Gα + Gα + Gα -/+

Simultaneous

Temporal combination

Gα + Gα -/+ Gα + Gα +/Gβ + Gα + Gα + Gα + Gα + Gα + Gα + Gα Gα +/Gβ + Gα +/Gβ + Gα + Gα + Gα -/+

Atemporal comb

Gβ + d+ Gα +

Gα + Gα -

d+ Gα + Gα + Gα + Gα +

Oppositive contrast

Gα + Gα Gγ + Gβ + Gβ + Gα + Gβ +/-

Gβ + Gα + d+ Gα +

Gβ +

Corrective contrast

d+ d+ Gβ + d+ d+ Gα +/d+ d+ Gα +/d+ Gα Gγ + Gβ + Gβ +/d+ Gα + Gβ +/-

Gβ +

Counterex contrast

Combination and contrast

229

230 Combination, contrast and alternative relations in comparison 6.3 Combination and alternative 6.3.1 The degree of coding 6.3.1.1

The combination-alternative coding implication

The comparison of the attested markers expressing combination and alternative relations is reported in Tables 43 and 44. The exam of these data reveals the following implicational pattern: (6.12) The combination-alternative coding implication: asyndesis for simple alternative → asyndesis for temporal and atemporal combination, asyndesis for choice-aimed alternative In a given language, if a simple alternative relation may normally be expressed by means of an asyndetic construction, such a strategy will be also available for the expression of temporal and atemporal combination relations and for the choice-aimed alternative. Likewise, if combination relations and the choice-aimed alternative are normally expressed by an overt marker, simple alternative will also be expressed by means of an overt connective. The cut-off point identiﬁed by the implication in (6.12) are shown in Table 38. Table 38: Overt markers for combination and alternative relations: cut-off point in the combination-alternative coding implication. + = presence of an overt marker; – = absence of an overt marker

Mangarayi Wari’ Hausa Malayalam Korean Latvian

Temporal c. Seq Sim – – –/+ –/+ –/+ –/+ –/+ –/+ + + + +

Atemporal c.

Choice-aimed a.

Simple a.

– – –/+ –/+ + +

– – + – – +

– – + + + +

As can be observed in the table, when simple alternative is expressed by means of an overt marker, combination relations and choice-aimed alternative may either be expressed by juxtaposition or may show overt connectives. There are languages like Mangarayi which make an extensive use of juxtaposition in the expression of all coordination relations and do not employ

Combination and alternative

231

any overt connective neither in the expression of combination nor in that of alternative. As shown in example (5.22), two alternative SoAs are usually juxtaposed to each other and their irreality is indicated by the dubitative adverbs maNaya, meaning ‘perhaps’. Similar strategies are also attested in Wari’ (cf. example (5.21)), where asyndesis is the most common coordinating construction. In this language, however, combination relations may also be conveyed by means of the resumptive construction showing the marker ca’ na ‘thus it was’ at the end of the combination. There are a number of languages in the Comparison sample expressing simple alternative by means of an overt marker and choice-aimed alternative or combination relations by means of juxtaposition. In Table 38 the cases of Hausa, Malayalam and Korean are reported. Hausa uses the same connective kó/kokuma for both alternative types and allows for both syndetic and asyndetic strategies in the expression of combination. Malayalam, on the other hand, shows the twofold possibility of syndesis and asyndesis in the expression of combination, but only employs juxtaposition in the expression of choice-aimed alternative. Finally, Korean makes use of an overt marker in the expression of combination relations (-ko or kuliko, cf. example (1.9)), but conveys choice-aimed alternatives through the juxtaposition of two interrogative clauses (cf. example (5.20)). 6.3.1.2

Semantic domains: combination, alternative and irreality

The comparison of the semantic domains characterizing the constructions used for combination and alternative reveals a neat separation in the coding of the two relations, as can be observed in Tables 43 and 44. As pointed out in section 6.2.1, combination and contrast relations may be expressed by means of the same construction and even by means of the same overt marker. Such overlapping phenomena are not attested in the coding of combination and alternative. All languages in the two samples make use of different constructions to convey these two coordination relations. In the European sample, all languages employ distinct overt markers both for combination and alternative. Outside Europe some languages express one or both relations by means of juxtaposition, even if the employment of distinct constructions and different markers is the most common situation.

232 Combination, contrast and alternative relations in comparison Combination and alternative show indeed different conceptual properties,

maguably reason for their different cross-linguistic coding. Combination has no implications as to the reality value of the linked SoAs, because it only implies that they are jointly asserted or questioned, that is, that they are jointly at issue. Alternative, on the other hand, implies that the linked SoAs are not real because only one of them is expected to occur and the speaker does not know which one. If one of the two SoAs was asserted as realis, this requirement would fail and there would simply be no reason to establish an alternative relation. The conceptual distance between the two relations is signiﬁcant, because they differ in two major respects. Combination makes no restriction on the reality value of the SoAs it links, but predicts their cooccurrence. Alternative, instead, implies the irreality of the SoAs and presents them as replaceable, non-cooccurring possibilities. As repeatedly argued, general markers do not code relations at random, but select functionally close notions. Consequently, the more conceptually distant two relations are, the less likely they are to be expressed by means of the same construction. The fact that combination and alternative imply antithetic conceptual properties is at the basis of their being cross-linguistically coded by means of different constructions. Cases like Mangarayi and Wari’ (cf. examples (5.22) and (5.21)), where both alternative and combination relations are expressed by juxtaposition, may be treated as instances of a general asyndetic construction used for the expression of both relations. As argued in section 5.2.2, asyndetic constructions expressing alternative are always characterized by the presence of some irreality markers, that is, markers coding the irrealis status of the linked SoAs or markers coding interrogative, dubitative or hypothetical notions, which in turn imply irreality. Constructions like (6.13) basically consist of the combination of two irrealis SoAs, thus mirroring the semantics of the notion of alternative itself, which has been deﬁned as the combination of two replaceable possibilities (see section 5.1). (6.13) Wari’, Chapacura-Wanam ’Am ’e’ ca ’am mi’ pin ca perhaps live 3sg.M perhaps give complete 3sg.M

Combination and alternative

233

‘Either he will live or he will die.’ (lit.‘perhaps he will live perhaps he will die’) (Everett and Kern 1997: 162) Nonetheless, even in these marginal constructions, combination and alternative are formally distinguished, since asyndetic combinative constructions do not have any restrictions on the reality status of the linked SoAs, while the expression of alternative always requires the juxtaposed clauses to be internally - directly or indirectly - marked for irreality. Another case in which the same construction is used for both combination and alternative is when the alternative relation is made explicit by means of expressions such as ‘what do you choose between X and Y?’ or ‘the possibilities are X and Y’, where the alternative SoAs are combined in a set of possibilities. This is what happens in the following example from Thai (6.14). Ohori (2004: 58-59) gives this example as an instance of ‘underdifferentiation’ between combination and alternative. However, as can be observed in (6.14b), choice-aimed alternative is made explicit by the lexical verb lyâk ‘choose’. Once the relation has been made clear, the two alternatives are combined into the set of possibilities among which the choice can be done. (6.14) Thai (a) COOn kàp mEErîi paj duu nˇaN John with/and Mary go see movie ‘John and Mary went to see a movie.’ (Ohori 2004: 58) (b) Th@@ càĳ lyâk tham ĳaraj ráĳwàaN paj duu nˇaN you will choose do what between go see movie kàp paj sýy khˇOON and/or go buy thing ‘Which would you like to do, go to see a movie or go shopping?’ (Ohori 2004: 58) To sum up, in case the same construction may be used for the expression of alternative and combination, irrealis contexts tend to be associated with an alternative interpretation, even though they do not exclude a combination reading, and realis contexts only allow for combination relations. In order to univocally interpret a relation as alternative, the relation may be paraphrased and made explicit, as in (6.14b).

234 Combination, contrast and alternative relations in comparison 6.3.1.3

The combination-alternative coding complexity implication

The morphophonological complexity of the attested markers expressing combination and alternative relations is compared in Tables 41 and 42. The comparison of these data highlights the following coding complexity implication: (6.15) The combination-alternative coding complexity implication: marker used for at least one alternative relation → marker used for at least one combination relation Overt markers used to express alternative relations, either general or dedicated, are at least as morphophonologically complex as the markers used to express at least one combination relation. As already pointed out in section 6.2.1.3, the label ‘marker used for at least one combination relation’ encompasses both dedicated and general markers. Some languages exemplifying this implication are shown in Table 39. As already mentioned in section 5.2.1, languages like Turkish, Lezgian and Dargi in the European sample, and Hausa, Marathi in the Comparison sample show polymorphemic connectives containing the marker used to express atemporal combination together with a marker expressing the alternative (cf. example (5.23) from Dargi). An example showing the two general markers used for combination and alternative in Turkish is given in (6.16), in which ve conveys combination and veya a simple alternative. (6.16) Turkish, Turkic, Altaic (a) Hasan i¸s-in-e git-ti, Ali ev-in-e Hasan work-3sg-DAT go-PST Ali house-3sg-DAT dön-dü ve ben park-ta kal-dı-m return-PST COORD I park-LOC stay-PST-1sg ‘Hasan went to work, Ali returned home and I stayed in the park.’ (Kornﬁlt 1997: 109) git-ti veya Ali çar¸sı-ya (b) Hasan i¸s-e Hasan work-DAT go-PST COORD Ali market-DAT çık-ti go.out-PST ‘Hasan went to work or Ali went shopping.’ (Kornﬁlt 1997: 111)

German: Basque: Dargi: Italian: Hausa: Hungarian: West Greenlandic: SerboCroatian: Maori: Iraqw: Finnish: Polish: Lezgian: Supyire: Japanese: i, a hoki nee ja i, a -ni, wa sì -te, -i

– – – –

pa

aa – – – -na kà –

TION

GENERAL ALSO FOR COMBINA -

und eta wa, -ra e kuma és, men =lu

– –

DEDICATED SEQUENTIAL

raanei laqáa – – – – –

> > > > > > >

>

oder edu yara o, oppure kokuma vagy imaluunniit, =luunniit ili

> > > > > > >

GENERAL ONLY FOR ALTERNA TIVE

– – vai czy taˆxajt’a làa –

–

– ala aèi – – – –

DEDICATED CHOICE AIMED ALT.

– – tai lub/albo ja, wa ja yô -ka

–

– – – – – – –

DEDICATED SIMPLE ALT.

Table 39: The combination-alternative coding complexity hierarchy: attested complexity patterns. – = absence of the given marker

Combination and alternative

235

236 Combination, contrast and alternative relations in comparison Other examples of polymorphemic markers expressing alternative have been given in section 5.2.1. Polymorphemic markers expressing combination are instead very rare and not fully grammaticalized, like ca’ na in Wari’, whose meaning ‘thus it was’ is still transparent and affects the position of the marker, which occurs at the end of the combination as a resumptive device. The trend for alternative markers to be more complex than markers coding combination has been remarked by Haspelmath (2004: 27) and Ohori (2004: 60-62). As Ohori argues, alternative markers tend to be coded by free words rather than by bound morphemes like clitics or afﬁxes. In particular, from the combination-alternative coding complexity implication, it follows that if a language has a bound marker expressing alternative relations, it will also have a bound marker expressing at least one combination relation. This implication is exempliﬁed by Lezgian, as can be observed in Table 39, in which bound markers are used for combination (=ni and -na), but only free connectives occur in the expression of alternative (taˆxajt’a, ja, wa ja), by Dargi (comb. -ra, alt. yara) and by Hebrew (comb. ve-, alt. o, o she-). Other languages have only bound markers for combination and show the twofold option in the expression of alternative. Malayalam, for instance, only has the bound marker -um for combination and has both a bound and a free marker for alternative, -oo and alleNkil respectively. A similar situation is attested in West Greenlandic, as shown in the table. In (6.17) the two constructions expressing alternative in Malayalam are shown, while the strategy expressing combination has already been discussed in section 2.3.2.1 (example (2.20) is repeated here for the reader’s convenience). The bound markers are sufﬁxed to nonﬁnite verb forms and the speciﬁcations of tense, aspect and mood are carried by the dummy verb ceyyuka (cf. discussion in section 2.3.2.1). (6.17) Malayalam, Tamil-Kannada, Dravidic (a) raaman varikay-um kôùïan pookukay-um ceytu Raman come:INF-COORD Krishnan go:INF-COORD do:PST ‘Raman came and Krishnan went’ (Asher and Kumari 1997: 135) (b) avan varikay-oo aareyeNkilum ayakkukay-oo he come:INF-COORD someone:ACC send:INF-COORD ceyyum do:FUT

Combination and alternative

237

‘Either he will come or he will send someone.’ (Asher and Kumari 1997: 140) alleNkil cappaatti (c) ñaan coor@ uïïaam ¯ I rice eat:FUT:MOD otherwise chapati tinnaam ¯¯ eat:FUT:MOD ‘I’ll eat rice or I’ll eat chapati.’ (Asher and Kumari 1997: 139) As thoroughly discussed in chapter 5 and as Haspelmath (2004: 27) points out, there are languages where there seems to be no grammaticalized means of expressing an alternative between SoAs. Either the relation is conveyed by the juxtaposition of irrealis clauses or there may be some very complex and recent structures, like -làw-leè in Hakha Lai, which is the combination of the negation -làw and the ancient conditional sufﬁx -leè (see example (5.39)). Ohori (2004: 61-62) further highlights the higher textual frequency of markers expressing combination with respect to those used for alternative relations. He takes into account the connectives attested in English and German, examining their occurrence in three English formal conversations and in ten German storytelling texts. The results of his frequency study reveals that markers expressing combination are more than 6 times more frequent than the ones coding alternative and in narrative texts the ratio is even higher, with und being around 17 times more frequent than oder. In section 6.3.3 the connection between frequency and the attested simpler morphophonological complexity of combination markers will be discussed, conﬁrming the already remarked relation between the attested constraints on cross-linguistic coding of coordination and the functional principle of economy. 6.3.2 The combination-alternative parallelism implication The comparison of the coding of combined and alternative SoAs can be observed in Tables 45 and 46. The analysis of these data reveals the following cross-linguistic implication: (6.18) The combination-alternative parallelism implication: simultaneous combination, atemporal combination, simple alternative and choice-aimed alternative → sequential combination

238 Combination, contrast and alternative relations in comparison In a given language, if a syntactically non-parallel construction may be used to convey an alternative relation or a non-sequential combination, be it simultaneous or atemporal, a non-parallel construction will also be available for the expression of sequential combination. To sum up, if a non-parallel construction is at all available in a given language, it will certainly be used to express sequential combination. The attested language types are exempliﬁed in Table 40. Table 40: The combination-alternative parallelism implication: attested types. + = syntactically parallel construction; - = syntactically non-parallel construction; blank= no information available Temporal

Jamul Tiipay West Greenlandic Korean Tauya Japanese Kolyma Yukaghir Supyire Turkish Lango Hebrew Vietnamese

Seq +/+/+/+/+ +

Sim +/+/+ + + + +

Atemp

Simple alt

Choiceaimed alt

+/+/+ + + + +

+/+/+ +

+/+ + + + + +

+ + + + +

+ +

Examples of non-parallel sequential constructions have been already discussed in section 6.2.2, where the coding of two combined SoAs has been examined. A further cut-off point may be identiﬁed between simultaneous and atemporal combination on the one hand and alternative relations on the other, since there are languages having non-parallel constructions for all the combination relations but none for alternative, while the reverse is not attested. However, the number of languages supporting this further asymmetry are not enough to make the combination-alternative parallelism implication a hierarchy. There are only three languages showing a non-parallel construction for the expression of all combination relations and a parallel construction for the expression of alternative: Japanese, Tauya and Kolyma Yukaghir. In these three languages the non-parallel construction is the same used for sequential combination. Moreover, as pointed out in section 5.3, only three languages ex-

Combination and alternative

239

press alternative by means of a non-parallel construction: West Greenlandic, Jamul Tiipay and Korean. These three languages use a non-parallel construction also to express combination. The individual cases of Jamul Tiipay, West Greenlandic and Korean have already been discussed in detail in chapter 5. In the European sample all languages use parallel syntactic constructions to code alternative and two languages, Turkish and Lezgian, show a nonparallel construction for sequential combination. In the Comparison sample more cross-linguistic variation is attested, even though the strong tendency for parallel constructions in the expression of alternative relations is conﬁrmed.

6.3.3

Functional motivations

In the preceding sections on the comparison between combination and contrast, the attested constraints on cross-linguistic variation have been explained on the basis of the functional principle of economy, both in relation to frequency of use and to recoverability of information. In particular, combination relations are more easily inferable from context and more frequent in discourse, as a result of their basic conceptual structure. A similar result emerges from the comparison between combination and alternative. Let us examine how the principle of economy motivates the attested coding patterns. According to the combination-alternative coding implication stated in (6.12), if a simple alternative relation may normally be expressed by means of an asyndetic construction, such a strategy will be also available for the expression of temporal and atemporal combination relations and for the choiceaimed alternative. We have already seen in chapters 3 and 5 that the main motivation underlying the use of syndetic or asyndetic constructions is the recoverability of the speciﬁc relation from the context. In other words, the more easily a relation is inferable from the context or the semantics of the SoAs, the more likely it is to be expressed by juxtaposition. Combination has already been argued to be a basic relation which is easily inferable from the juxtaposition of two SoAs, for it only establishes their cooccurrence and does not specify further the nature of the relation. The implication furthermore suggests that also choice-aimed alternative is more easily inferable than simple alternative. As pointed out in section 5.4, the motivation underlying this asymmetry internal to alternative relation is connected

240 Combination, contrast and alternative relations in comparison to the fact that choice-aimed alternative relations are always coded through interrogative form. In particular, it is easier to infer an alternative relation from the juxtaposition of two interrogative clauses than from the juxtaposition of two declarative clauses. Two alternative SoAs stand in a paradigmatic contrast to each other. When two SoAs which stand in a semantic contrast are juxtaposed in a declarative sentence, this may well be because the speaker wants to establish a combination or a contrast relation, and the possibility that they are juxtaposed because they stand in an alternative relation is only one of the options, and plausibly not the ﬁrst. By contrast, if the linked SoAs are coded by two juxtaposed interrogative clauses, it means that they are questioned and that the speaker is not able to assert their occurrence, but he rather wants to elicit some information about them. Since they are not presented as actually occurring, it is more difﬁcult that the reason for presenting the two SoAs together is to establish a combination or a contrast relation between them, because these two relations imply that the speaker is at least aware of the cooccurrence of the SoAs. In fact, if two SoAs standing in a semantic contrast are juxtaposed in an interrogative form, they will be most easily interpreted as alternatives. As a result, choice-aimed alternative is more likely to be expressed with an asyndetic strategy, that is, without a marker coding the alternative relation. The principle of economy also motivates the higher morphophonological complexity of the attested markers expressing alternative relations. The combination-alternative coding complexity in (6.15) states that markers used for at least one combination relation are morphophonologically simpler than those employed for alternative. In section 6.2.3 it has thoroughly been argued that the morphophonological simplicity of combination markers, either general or dedicated, is related to their frequency of use, for the most frequent linguistic expressions tend to be minimized and undergo phonological erosion. Ohori (2004: 60-62) proves that combination is much more frequently established in discourse than alternative. This is mainly due to the fact that combination is the most basic interclausal relation establishing the cooccurrence of two SoAs. The alternative relation implies that of combination, as shown by the attested patterns of polymorphism. In order to be presented as alternatives, two SoAs must indeed be ﬁrst of all combined and then the combination must be speciﬁed in terms of irreality and replaceability of the SoAs. Being a more complex relation, alternative is less frequently established in

Combination and alternative

241

discourse and this motivates the higher morphophonological complexity of markers expressing alternative relations. The alternative relation is a further speciﬁcation of a generic combination. First, two SoAs are put together for some reason and presented as possibilities; ﬁnally, their potential replaceability is pointed out. The opposition between the cooccurrence notion characterizing combination and the non-cooccurrence notion characterizing alternative is at the origin of their strongly differentiated cross-linguistic coding. No such opposition exists between combination and contrast, because they both require the cooccurrence of the linked SoAs. Such a conceptual distance is mirrored in the cross-linguistic coding of combination and alternative, since no general marker used for both relations is attested in the examined samples (cf. discussion in section 6.3.1). The conceptual structure of the notion of alternative has consequences also on the coding of two alternative SoAs. In section 6.3.2 use of a non-parallel construction in the expression of alternative or nonsequential combination implies the use of such a strategy also in the expression of sequential combination. The reasons for the use of non-parallel strategies for sequential combination have been already pointed out in section 6.2.3. Basically, dedicated sequential constructions code the sequential temporality of the linked SoAs, which predetermines the relative temporal location of the linked SoAs. Tense, aspect and mood speciﬁcations may be marked only once for just one SoA, because information on the temporal location of the other follows. Alternative relations, instead, do not predetermine any semantic property of the linked SoAs, except for their being irrealis. Therefore, tense and aspect characterizations must be overtly speciﬁed for each SoA.

242 Combination, contrast and alternative relations in comparison 6.3.4

Language data

Table 41: European Sample. Coding of combination and alternative relations in comparison: morphophonological complexity of the attested coordinating markers. + = presence of the given feature; - = absence of the given feature; alt.= alternative; choice-a= choice-aimed; opp.= opposition; atemp.= atemporal combination; comb.= combination; blank= no information available Language

Marker

Semantic domain

Free

Polysyllabic

Polymorphem.

Albanian

e/dhe ose/o apo/a eta edu ala i a abĲo ci i a

all comb. types simple alt. choice-a alt. all comb. types both alt. types choice-a alt. all comb. types atemp.comb., opp. simple alt. choice-a alt. all comb. types atemp.comb., opp., corr. sequential comb. both alt. types all comb. types both alt. types all comb. types atemp.comb., opp., corr. both alt. types all comb. types both alt. types both alt. types all comb. types both alt. types all comb. types all comb. types both alt. types choice-a alt. all comb. types both alt. types all comb. types

+ + + + + + + + + + + +

-/+ -/+ + + + + + -

-

+ + + + +

+ -

-

+ + + + + + + + + + + +

+ + + + -

+ -

Basque

Belorussian

Bulgarian

Catalan Chechen

Czech

Danish Dargi

Dutch English

ta/da ili i o ’a tq’a ja a nebo ci ˇ og eller wa -ra yara aèi en of and

Combination and alternative

243

(Continued) Language

Estonian Finnish

French Georgian

German Greek Hungarian

Icelandic Irish Italian Latvian Lezgian

Lithuanian

Luxembourgish Maltese Norwegian Polish

Marker

Semantic domain

Free

Polysyllabic

Polymorphem.

or ja või ja tai vai et ou da an tu und oder kai i és meg vagy og eða agus nó e o/oppure un vai -ni, wa -na ja wa ja taˆxajt’a ir o ar/arbà ann oder u jew og eller i

both alt. types all comb. types both alt. types all comb. types simple alt. choice-a alt. all comb. types both alt. types all comb. types simple alt. choice-a alt. all comb. types both alt. types all comb. types both alt. types all comb. types atemp.comb., opp. both alt. types all comb. types both alt. types all comb. types both alt. types all comb. types both alt. types all comb. types both alt. types all comb. types sequential comb. simple alt. simple alt. choice-a alt. all comb. types atemp.comb., opp. both alt. types all comb. types both alt. types all comb. types both alt. types all comb. types both alt. types all comb. types

+ + + + + + + + + + + + + + + + + + + + + + + + + + -/+ + + + + + + + + + + + + +

+ + + + -/+ + + -/+ + + -

+ + -

244 Combination, contrast and alternative relations in comparison (Continued) Language

Portuguese Romanian

Russian

Sardinian Serbo-Croatian

Spanish Swedish Turkish

Ukrainian

Marker

Semantic domain

Free

Polysyllabic

Polymorphem.

a lub albo czy e ou s¸i iar ori sau i a

atemp.comb., opp. simple alt. simple alt. choice-a alt. all comb. types both alt. types all comb. types atemp.comb., opp. both alt. types both alt. types all comb. types atemp.comb., opp., corr. both alt. types all comb. types both alt. types all comb. types atemp.comb., opp. sequential comb. both alt. types all comb. types both alt. types all comb. types both alt. types all comb. types all comb. types sequential comb. both alt. types all comb. types atemp.comb., opp. simple alt. choice-a alt.

+ + + + + + + + + + + +

+ + -

-

+ + + + + + + + + + + + + + + + +

+ + + -/+ -/+ -

-/+ -

ili e o i a pa ili y o och eller =dA ve -Ip veya i ta abĲo cˇ y

Combination and alternative

245

Table 42: Comparison Sample. Coding of combination and alternative relations in comparison: morphophonological complexity of the attested coordinating markers.+ = presence of the given feature; - = absence of the given feature; alt.= alternative; choice-a= choice-aimed; opp.= opposition; atemp.= atemporal combination; sim.= simultaneous combination; comb.= combination; blank= no information available Language

Marker

Semantic domain

Free

Polysyllabic

Polymorphem.

Arabic

wa fa aw am -k@ ye =ĳií -làw=leè -é ﬁ

all comb. types sequential comb. both alt. types choice-a alt. all comb. types both alt. types all comb. types simple alt. sequential comb. sim., atemp. comb.

+ + + + + + +

+ -

+ -

Jamul Tiipay

kuma kó kokuma lá á nà veo (she) ánaga ˛ ni˛˛igéšge nee laqáa switch-r

+ + + + + + + + + + -

+ + + -/+ + + + -

+ + -/+ -

Japanese

nyamaaw -te

+ -

+ -

-

si -i -ka soretomo bàà

all comb. types both alt. types both alt. types sequential comb. both alt. types all comb. types both alt. types all comb. types both alt. types all comb. types both alt. types all comb. types, opp.,corr. simple alt. all comb. types, opp., corr. all comb. types atemp.comb., opp. simple alt. choice-a alt. both alt. types

+ + +

+ -

-

-ko kuliko -kena

all comb. types all comb. types simple alt.

+ -

+ +

-

Dumi Hakha Lai Harar Oromo

Hausa

Hdi Hebrew Hocak ˛ Iraqw

Kisi Kol. Yukaghir Korean

246 Combination, contrast and alternative relations in comparison (Continued) Language

Koromfe Kuskokwim A. Lango Limbu Malayalam

Mangarayi Maori

Marathi

Maricopa Meithei Mosetén NànáfwˆE

Ndyuka Persian Rapanui Somali

Supyire

Marker

Semantic domain

Free

Polysyllabic

Polymorphem.

hokun maa ts’eĳ h0nek’è tˆE òñò -an -um -oo alleNkil

simple alt. simple alt. all comb. types simple alt. sequential comb. simple alt. all comb. types all comb. types simple alt. simple alt.

+ + + + + + +

+ + + +

+ -

aa hoki raanei a¯ n.i wa k˜ımw¯a/ athw¯a k¯ı

sequential comb. sim., atemp. comb. both alt. types all comb. types all comb. types simple alt.

+ + + + + +

+ + + +

+

choice-a alt.

+

-

-

aduga nattrega jö’dyë’yä

all comb. types simple alt. sim., atemp. comb.

+ + +

+ + +

+

kp´Ekún wj´Elj´E ánzˆE da efu va yâ e o -na oo ama misé kà/mà sì

all comb. types simple alt. choice-a alt. all comb. types both alt. types all comb. types both alt. types all comb. types both alt. types all comb. types all comb. types simple alt. choice-a alt. sequential comb. atemp.comb., opp., corr., coun. simple alt.

+ + + + + + + + + + + + + +

+ + + + + + -

-

+

-

-

yô

Combination and alternative

247

(Continued) Language

Marker

Semantic domain

Free

Polysyllabic

Polymorphem.

Tauya

làa switch-r

choice-a alt. all comb. and contrast types both alt. types sequential comb. sim., atemp. comb. both alt. types atemp.comb., opp. both alt. types all comb. types simple alt. choice-a alt. all comb. types all comb. types both alt. types both alt. types

+ -

-

-

+ + + + + + + + + + + -

+ + + -/+ + + +

+ -

Tukang Besi

Tuvaluan Vietnamese

Wari’ W.Greenlandic

pe maka kene tawa kae io me/pe và ho˘a. hay(là) ca’ na =lu imaluunniit =luunniit

Albanian Basque Belorussian Bulgarian Catalan Chechen Czech Danish Dargi Dutch English Estonian Finnish French Georgian German Greek Hungarian Icelandic Irish Italian

Language

Gα Gα Gβ Gα /d Gα -/Gα Gα Gα Gα Gα Gα Gα Gα Gα Gα Gα Gα Gα Gα Gα Gα

sequential Gα Gα Gα Gα Gα -/Gα Gα Gα Gα Gα Gα Gα Gα Gα Gα Gα Gα Gα Gα Gα Gα

simultaneous

Temporal combination

Gα Gα Gα /Gβ Gα /Gβ Gα -/Gα /Gβ Gα Gα Gα Gα Gα Gα Gα Gα Gα Gα Gα Gα /Gβ Gα Gα Gα

atemporal combination d Gβ /d d Gγ Gβ Gγ Gβ Gβ Gβ /d Gβ /d Gβ Gβ d Gβ d Gβ Gβ Gγ Gβ Gβ Gβ

choice-aimed alternative d Gβ d Gγ Gβ Gγ Gβ Gβ Gβ Gβ Gβ Gβ d Gβ d Gβ Gβ Gγ Gβ Gβ Gα

simple alternative

Table 43: European sample. Coding of the combination relation: presence of overt coordinating markers and their semantic domain. - = absence of an overt marker;d= dedicated; Gx = general, with x identifying a given construction attested in the language; blank= no information available

248 Combination, contrast and alternative relations in comparison

Latvian Lezgian Lithuanian Luxembourgish Maltese Norwegian Polish Portuguese Romanian Russian Sardinian Serbo-Croatian Spanish Swedish Turkish Ukrainian

Language

(Continued)

Gα Gα /d Gα Gα Gα Gα Gα Gα Gα Gα Gα Gα /d Gα Gα -/Gα /d Gα

Sequential Gα Gα Gα Gα Gα Gα Gα Gα Gα Gα Gα Gα Gα Gα -/Gα Gα

Simultaneous

Temporal combination

Gα Gα Gα /Gβ Gα Gα Gα Gα /Gβ Gα Gα /Gβ Gα /Gβ Gα Gα /Gβ Gα Gα -/Gα Gα /Gβ

Atemporal combination Gβ d Gγ Gβ Gβ Gβ d Gβ Gγ Gγ Gβ Gγ Gβ Gβ Gβ d

Choiceaimedalternative Gβ d Gγ Gβ Gβ Gβ d Gβ Gγ Gγ Gβ Gγ Gβ Gβ Gβ d

Simple alternative

Combination and alternative

249

Arabic Dumi Hakha Lai Harar Oromo Hausa Hdi Hebrew Hocak ˛ Iraqw Jamul Tiipay Japanese Kisi Kol. Yukaghir Korean Koromfe Kuskokwim A. Lango Limbu Malayalam Mangarayi Maori

Language

Gα /d -/Gα Gα d -/Gα -/d Gα Gα Gα Gα Gα Gα CJ -/d -/Gα -/Gα -/d

Sequential Gα Gα Gα Gα -/Gα Gα Gα Gα Gα Gα Gα CJ -/Gα -/Gα -/Gα

Simultaneous

Temporal combination

Gα Gα Gα Gα Gα /Gβ Gα -/Gα CJ -/Gα -/Gα -/Gα

Gα Gα Gα Gα

Atemporal combination

d Gβ

d d d d

Gβ Gα Gβ Gβ Gβ d d Gα

-/Gα Gβ Gα Gβ Gβ Gβ -/d Gα Gβ

Gβ Gβ d

Simple alternative

Gβ /d Gβ -

Choiceaimedalternative

Table 44: Comparison Sample. Coding of the combination relation: presence of overt coordinating markers and their semantic domain. - = absence of an overt marker;d= dedicated; Gx = general, with x identifying a given construction attested in the language; blank= no information available

250 Combination, contrast and alternative relations in comparison

Marathi Maricopa Meithei Mosetén NànáfwˆE Ndyuka Persian Rapanui Somali Supyire Tauya Tukang Besi Tuvaluan Vietnamese Wari’ W.Greenlandic

Language

(Continued)

-/Gα -/Gα -/Gα -/Gα Gα -/Gα Gα d Gα -/d -/Gα -/Gα -/Gα

Sequential -/Gα -/Gα -/Gα -/Gα -/Gα Gα -/Gα Gα Gα Gα -/Gα -/Gα -/Gα -/Gα -/Gα

Simultaneous

Temporal combination

-/Gα -/Gα -/Gα -/Gα -/Gα Gα -/Gα Gα Gα Gα -/Gα -/Gα -/Gα -/Gα -/Gα

Atemporal combination

-/d

d Gβ Gβ -/Gβ d d Gβ Gβ Gβ d Gβ d Gβ Gβ -/Gβ d d Gβ Gβ Gβ d Gβ

d

Simple alternative

d

Choiceaimedalternative

Combination and alternative

251

Albanian Basque Belorussian Bulgarian Catalan Chechen Czech Danish Dargi Dutch English Estonian Finnish French Georgian German Greek Hungarian Icelandic Irish Italian

Language

Gα + Gα + Gα + Gα +/d+ Gα + Gα + Gα + Gα + Gα + Gα + Gα + Gα + Gα + Gα + Gα + Gα + Gα + Gα + Gα + Gα + Gα +

Sequential Gα + Gα + Gα + Gα + Gα + Gα + Gα + Gα + Gα + Gα + Gα + Gα + Gα + Gα + Gα + Gα + Gα + Gα + Gα + Gα + Gα +

Simultaneous

Temporal combination

Gα + Gα + Gα +/Gβ + Gα +/Gβ + Gα + Gα +/Gβ + Gα + Gα + Gα + Gα + Gα + Gα + Gα + Gα + Gα + Gα + Gα + Gα +/Gβ + Gα + Gα + Gα +

Atemporal combination d+ Gβ + d+ Gγ + Gβ + Gγ + Gβ + Gβ + Gβ + Gβ + Gβ + Gβ + d+ Gβ + d+ Gβ + Gβ + Gβ + Gβ + Gβ + Gβ +

Simple alternative

d+ Gβ +/d+ d+ Gγ + Gβ + Gγ + Gβ + Gβ + Gβ +/d+ Gβ + Gβ + Gβ + d+ Gβ + d+ Gβ + Gβ + Gβ + Gβ + Gβ + Gβ +

Choiceaimedalternative

Table 45: European Sample. Combination and alternative in comparison: syntactic parallelism of the attested constructions. + = syntactically parallel; - = syntactically non parallel; d= dedicated; Gx = general, with x identifying a given construction attested in the language; blank= no information available

252 Combination, contrast and alternative relations in comparison

Latvian Lezgian Lithuanian Luxembourgish Maltese Norwegian Polish Portuguese Romanian Russian Sardinian Serbo-Croatian Spanish Swedish Turkish Ukrainian

Language

(Continued)

Gα + Gα +/d Gα + Gα + Gα + Gα + Gα + Gα + Gα + Gα + Gα + Gα +/d+ Gα + Gα + Gα +/d Gα +

Sequential Gα + Gα + Gα + Gα + Gα + Gα + Gα + Gα + Gα + Gα + Gα + Gα + Gα + Gα + Gα + Gα +

Simultaneous

Temporal combination

Gα + Gα + Gα +/Gβ + Gα + Gα + Gα + Gα +/Gβ + Gα + Gα +/Gβ + Gα +/Gβ + Gα + Gα +/Gβ + Gα + Gα + Gα + Gα +/Gβ +

Atemporal combination Gβ + d+ Gβ + Gβ + Gβ + Gβ + d+ Gβ + Gβ + Gβ + Gβ + Gβ + Gβ + Gβ + Gβ + d+

Simple alternative

Gβ + d+ Gβ + Gβ + Gβ + Gβ + d+ Gβ + Gβ + Gβ + Gβ + Gβ + Gβ + Gβ + Gβ + d+

Choiceaimedalternative

Combination and alternative

253

Arabic Dumi Hakha Lai Harar Oromo Hausa Hdi Hebrew Hocak ˛ Iraqw Jamul Tiipay Japanese Kisi Kol. Yukaghir Korean Koromfe Kuskokwim A. Lango Limbu Malayalam Mangarayi Maori

Language

Gα +/d+ Gα +/Gβ + Gα + dGα + Gα +/d+ Gα + Gα + Gα + Gα Gα -/+ Gα + Gα Gα -/+ Gα + Gα + Gα +/d Gα + Gα + Gα + d+

Sequential Gα + Gα +/Gβ + Gα + Gα + Gα + Gα + Gα + Gα + Gα + Gα Gα -/+ Gα + Gα Gα -/+ Gα + Gα + Gα + Gα + Gα + Gα + Gα +

Simultaneous

Temporal combination

Gα + Gα +/Gβ + Gα + Gα + Gα + Gα + Gα + Gα + Gα + Gα Gα -/+/Gβ Gα + Gα Gα -/+ Gα +/Gβ + Gα + Gα + Gα + Gα + Gα + Gα +

Atemporal combination

d+ d+ Gβ +

d-/+ d+ d+ d+

d+ d+ d+ Gβ +

Gβ + Gβ + Gβ + Gβ + Gβ + dd+ Gβ + d+ d+ d+

Gβ + Gβ + Gβ + Gβ + Gβ + d+ Gβ +

Gβ +/d+ Gγ + d+

Choiceaimedalternative

Gβ + Gγ +

Simple alternative

Table 46: Comparison Sample. Combination and alternative in comparison: syntactic parallelism of the attested constructions. + = syntactically parallel; - = syntactically non parallel; d= dedicated; Gx = general, with x identifying a given construction attested in the language; blank= no information available

254 Combination, contrast and alternative relations in comparison

Marathi Maricopa Meithei Mosetén NanafwˆE Ndyuka Persian Rapanui Somali Supyire Tauya Tukang Besi Tuvaluan Vietnamese Wari’ W.Greenlandic

Language

(Continued)

Gα +/d Gα -/+ Gα + Gα + Gα + Gα + Gα + Gα + Gα + dGα Gα +/d+ Gα + Gα + Gα + Gα -/+

Sequential Gα + Gα -/+ Gα + Gα + Gα + Gα + Gα + Gα + Gα + Gα + Gα Gα +/Gβ + Gα + Gα + Gα + Gα -/+

Simultaneous

Temporal combination

Gα + Gα -/+ Gα + Gα +/Gβ + Gα + Gα + Gα + Gα + Gα + Gα + Gα Gα +/Gβ + Gα +/Gβ + Gα + Gα + Gα -/+

Atemporal combination d+ d+ d+ Gβ + Gβ + Gβ + d+ d+ Gβ + Gγ + Gγ + d+ d+ Gβ +/-

d+ d+ Gβ + Gβ + Gβ + d+ d+ Gβ + Gγ + Gγ + d+ d+ Gβ +/-

Choiceaimedalternative

d+

Simple alternative

Combination and alternative

255

256 Combination, contrast and alternative relations in comparison 6.4 Combination, contrast and alternative: concluding remarks Now that the cross-linguistic coding of each coordination relation has been examined and combination has been separately compared to contrast and alternative, it is possible to draw some conclusions on the coding of the three basic coordination relations. Two general tables containing data on the attested constructions expressing combination, contrast and alternative are provided in the Appendix, Table 53 on European languages and Table 54 on non-European languages. As brieﬂy mentioned at the beginning of this chapter, while the separate comparison of combination with alternative and contrast gives rise to interesting coding patterns, the comparison of alternative with contrast does not reveal signiﬁcant coding regularities. In other words, the two coordination relations are both more difﬁcult to infer, less frequent and semantically more speciﬁed than combination, but there does not seem to be any considerable coding asymmetry between them. Before proceeding with the joint analysis of the three relations, let us ﬁrst of all discuss the comparison between the constructions expressing contrast and those expressing alternative relations.

6.4.1 Contrast and alternative in comparison No particular asymmetry emerges in the attested markers employed in the expression of contrast and alternative relations. The connectives are quite homogeneous in their showing both mono- and polysyllabic and mono- and polymorphemic structures. As can be observed in Tables 53 and 54, there are languages in which markers coding only contrast relations are at least as morphophonologically complex as those coding alternative - like Spanish, Arabic, Ukrainian and Persian - and languages where the opposite pattern occurs - like Hungarian, Jamul Tiipay, West Greenlandic and Norwegian. The degree of morphophonological complexity of the attested coordinating markers is connected with their frequency of use, which in turn depends on their semantic versatility and on the basicness of the relation established, as already argued in the course of this work. Hence, it follows that there must be no considerable difference between contrast and alternative relations under these respects. None of the two relations is signiﬁcantly more frequent

Combination, contrast and alternative: concluding remarks

257

than the other, or at least not enough to determine a cross-linguistic coding asymmetry. On the other hand, considering the use of syndetic vs. asyndetic constructions in the expression of contrast and alternative relations, the distribution is not random. However, it does not identify one of the two relations as unmarked with respect to the other, but it only conﬁrms that some contrast relations and some alternative relations are more likely to be expressed by overt markers than others. The attested constraint on cross-linguistic variation can be stated as follows: (6.19) The contrast-alternative coding implication: asyndesis for counterexpectative contrast, asyndesis for simple alternative → asyndesis for oppositive and corrective contrast, asyndesis for choice-aimed alternative If a language uses an asyndetic construction in the expression of counterexpectative contrast or simple alternative, it will also be able to express choiceaimed alternative, oppositive and corrective contrast by means of juxtaposition. The attested types are reported in Table 47. Table 47: Coding of contrast and alternative relation: attested types. - = absence of an overt coordinating marker; + = presence of an overt coordinating marker; blank= no information available

Wari’ Mangarayi Lango Kisi Ndyuka Koromfe Korean Maori Ndyuka Supyire

Oppositive c. +

Corrective c. +

+

-/+ +

Choice-aimed a. + + + + +

Simple a. + + + + + + + +

Counterexp c. + -/+ + + + + + + +

In the preceding chapters it was made clear that the coding of the relation is directly inﬂuenced by the principle of syntagmatic economy, according to which what is inferable from the context needs no further speciﬁcation. The coding of the various types of combination, contrast and alternative has

258 Combination, contrast and alternative relations in comparison turned out to be connected to the degree to which every speciﬁc relation can be inferred from the context. Speciﬁcally, the easier a relation is inferable, the less it needs to be overtly marked. Therefore, it is plausible that the principle of syntagmatic economy is at work also in the implication stated in (6.19). The fact that oppositive and corrective contrast are more easily inferable than counterexpectative contrast has been already widely argued in chapter 4. Likewise, the fact that choice-aimed alternative is more easily inferable than simple alternative was discussed in chapter 5. In other words, the contrast-alternative coding implication might be analyzed as a synthesis of the implicational patterns identiﬁed in (4.16) for contrast relations and in (5.12) for alternative relations. Yet, the conception of this implication as the bare synthesis of the contrast and alternative coding patterns does not account for three further phenomena highlighted by the implication: (i) the asymmetry in the coding of oppositive and corrective contrast on the one hand and simple alternative on the other, (ii) the asymmetry in the coding of choice-aimed alternative on the one hand and counterexpectative contrast on the other, (iii) the absence of asymmetry in the coding of the relations that stand on the same side in the implication, that is, oppositive contrast, corrective contrast, choice-aimed alternative on the one hand, counterexpectative contrast and simple alternative on the other hand. Therefore, it is necessary to explain why corrective and oppositive contrast are more likely to be expressed with juxtaposition than simple alternative, and why such an asymmetry does not hold between the two contrast relations and the choice-aimed alternative relation. Secondly, a reason should be identiﬁed for the fact that a choice-aimed alternative relation is more likely to be expressed by juxtaposition than a counterexpectative contrast, and why such an asymmetry does not hold between counterexpectative contrast and simple alternative. The principle of syntagmatic economy once again plays a major role in explaining these facts. Simple alternative, oppositive and corrective contrast relations typically occur in declarative sentences (see chapters 4 and 5). As König and Siemund (2007: 1-2) point out, declarative sentences are most frequently employed for speech acts stating the reality of the SoA involved, such as asserting and claiming. Therefore, two juxtaposed SoAs uttered in a declarative sentence are more easily interpreted as being both realis and cooccurring, rather than irrealis, equivalent possibilities (cf. section 5.4).

Combination, contrast and alternative: concluding remarks

259

According to the principle of syntagmatic economy, information that is recoverable from the context or from the assumptions of conversation coherence does not need to be further expressed. Therefore, it is plausible to assume that two contrasting SoAs juxtaposed in a declarative sentence will be assumed to be both realis and to cooccur in a contrast relation. If the speaker wants to eliminate the assumption that the SoAs cooccur, in order to link them as alternatives, an overt marker is likely to be used. On the other hand, two juxtaposed SoAs linked in an interrogative sentence are typically assumed to be irrealis, since the speech act of questioning is most frequently used to elicit some information about the reality status of each SoA. Consequently, two contrasting SoAs juxtaposed in an interrogative sentence are more easily interpreted as non-cooccurring alternatives, rather than as co-existing facts. Both in the case of asyndetic choice-aimed alternative and in the case of asyndetic oppositive and corrective contrast, the most likely interpretation of the juxtaposition corresponds to the speech act coherence assumptions, making the overt marking of the relation less urgent. This point explains the absence of coding asymmetry between the three relations. On the contrary, counterexpectative contrast shows a coding asymmetry with respect to choice-aimed alternative, even though it frequently occurs in declarative sentences and is characterized by the cooccurrence of contrasting SoAs, like correction and opposition. The fact that the counterexpectative relation is more likely to be expressed with an overt coordinating marker may be related to the higher difﬁculty in inferring a contrast originated from the denial of an expectation, rather than a paradigmatic contrast. Two alternative SoAs stand in a paradigmatic contrast, that is, they are antonymic with respect to their common possible context of occurrence. This contrast type is inherent in the semantics of the SoAs, in that they designate irrealis and equivalent possibilities, only one of which is allowed to occur. On the other hand, the counterexpectative contrast is not inherent in the semantics of the SoAs, but originates from a conﬂict between the semantics of one SoA and an expectation, either activated by the other SoA or by the context. As widely argued in chapter 4, if no overt marker is used, the hearer may establish a relation between the SoAs, without considering the expectations that one SoA or the context may activate. Instead, the presence of an overt counterexpectative connective explicitly indicates that some expectation is to be identiﬁed and then denied, in order to identify the speciﬁc contrast that the speaker wants to communicate. The counterexpectative contrast is thus more

260 Combination, contrast and alternative relations in comparison difﬁcult to infer than the paradigmatic one and is consequently more likely to be expressed by means of syndetic constructions. Finally, counterexpectative contrast and simple alternative do not show any cross-linguistic coding asymmetry. This is probably due to the fact that none of the two relations is signiﬁcantly easier to infer than the other. The properties that make these two relations more difﬁcult to infer are different and have been analyzed in the previous discussion. In the case of simple alternative, the difﬁculty lies in the identiﬁcation of irrealis and non-cooccurring SoAs within a declarative sentence, typically associated with the assertion of some SoAs as cooccurring and realis. In the case of counterexpectative contrast, the difﬁculty lies in the identiﬁcation of the contrast as originated by a contradiction involving an expectation and its denial, and not simply the inherent semantics of the linked SoAs. Moving now to the analysis of the attested semantic domains, there appears to be no overlap between the expression of contrast and alternative relations. In other words, these two coordination relations are cross-linguistically coded by means of different and dedicated strategies. However, as Bednarczuk (1971: 110) points out, there are cases of alternative and contrast markers sharing a similar etymology, like for instance Polish ale or Serbo-Croatian ali meaning ‘but’ and Russian ili or Slovak ali meaning ‘or’, both showing a combinative marker together with the Old Slavic restrictive particle li (cf. Adamíková 2005). Both contrast and alternative are characterized by the presence of some conﬂict between the SoAs, which are considered and linked on the basis of their differences, either as paradigmatic possibilities or as cooccurring facts. Therefore, in a diachronic analysis it should be possible to identify patterns connecting these two non-basic coordination relations. Yet, a diachronic exam of the attested markers goes beyond the scope of this study and is left for further research. An implicational hierarchy involving contrast and alternative markers has been pointed out by Matras (1998: 301-305) with regards to borrowing phenomena. Matras identiﬁes what he calls a contrast scale in coordination, according to which in bilingual contexts languages replacing combination markers also replace alternative markers, and languages replacing alternative markers also replace contrast markers. The borrowing hierarchy can be described as follows: (6.20) ’but’ > ’or’ > ’and’

Combination, contrast and alternative: concluding remarks

261

In other words, counterexpectative connectives are one of the ﬁrst discourse markers that are borrowed, followed by alternative markers and only at a later stage by combination markers. According to Matras, the order in which the markers follow each other on the borrowing hierarchy mirrors the different degrees of ‘intensity with which the speaker is required to intervene with hearer-sided mental processing activities’ (Matras 1998: 305) in establishing the relations of combination, alternative and contrast. The more the relation implies a contrast, the more the speaker has to maintain assertive authority despite the denial of the addressee’s expectations. To do so, bilingual speakers tend to adopt connectives of the pragmatically dominant language, because critical argumentations aimed at directing the addressee’s processing of the utterance are undertaken in the language in which a higher authority is achieved (Matras 1998: 321-325). The borrowing scale shows that contrast requires a higher level of interpretation and control over the addressee’s mental activity than alternative. Such a functional difference is mirrored by implicational patterns of borrowing, but seems to have no signiﬁcant reﬂections in the cross-linguistic coding of the two relations. Data in this research shows that contrast and alternative markers tend to be more or less equally complex and tend to have separate, non-overlapping semantic domains. Furthermore, as for the coding of two conﬂicting and alternative SoAs, both relations show a strong cross-linguistic tendency for parallel constructions, as pointed out in the preceding discussion, due to the fact that neither contrast nor alternative predetermine any semantic properties of the linked SoAs. Therefore, each verb form needs to receive tense, mood and aspect speciﬁcations (see discussion and examples in chapters 4 and 5). To conclude, the coding of contrast and alternative does not reveal particular interconnections between the two relations. Instead, they are coded by different and more or less equally complex constructions. Despite the fact that both relations involve a conﬂict between the linked SoAs, the two coordination relations are characterized by different conceptual structures. Contrast implies and originates from the cooccurrence of the linked SoAs, while alternative implies the irreality and the non-cooccurrence of the SoAs. As will be argued in the next sections, this conceptual distance, together with the consequent cross-linguistic coding differentiation, motivates the analysis of the examined coordination relations along two separate and independent semantic dimensions, both having their origin in combination.

><18=0C8>= 2>=CA0BC 0=3 0;C4A=0C8E4 A4;0C8>=B 8= 2>=

-<9 7CCF8=B5H=CB 7C8=B; =AD@=75H=CBG

-<9 FGH @9J9@ C: 5B5@MG=G C: H<=G KCF? =B K<=7< H<9 IFCD95B 5B8 H<9 CA D5F=GCB G5AD@9G <5J9 699B >C=BH@M 9L5A=B98 <5G F9J95@98 F9;I@5F 7FCGG@=B;I= GH=7D5HH9FBG C: J5F=5H=CB =B H<9 7C8=B; C: 7CCF8=B5H=CB F9@5H=CBG -<9 5B5@MG=G C: H<9 5HH9GH98 7CBGHFI7H=CBG G
:FCA H<9 7CBH9LH :CF H<9 CH<9F ,C !9B79 898=75H98 G9EI9BH=5@ 7CBGHFI7H=CBG 5F9 ACF9 @=?9@M HC 69 GMBH57H=75@@M BCBD5F5@@9@ 0=H<=B 7CBHF5GH CDDCG=H=CB 5B8 7CFF97H=CB 5F9 ACF9 95G=@M =B:9F56@9 H<5B H<9 7CBHF5GH CF=;=B5H98 6M H<9 89B=5@ C: 5B 9LD97H5H=CB !9B79 H<9 @5HH9F =G ACF9 @=?9@M HC 69 =B8=75H98 6M CJ9FH 7CBB97H=J9G -<9 F95GCB :CF H<=G 5GMAA9 HFM =G H<5H =B CDDCG=H=CB 5B8 7CFF97H=CB H<9 7CB CF=;=B5H9G =B H<9 5BHCBMA=7 G9A5BH=7G C: H<9 @=B?98 ,CG K<9F95G =B 5 7CIBH9F9LD97H5H=J9 7CBHF5GH H<9 7CB CF=;=B5H9G =B H<9 7CBHF58=7H=CB 69HK99B CB9 ,C 5B8 H<9 9LD97H5H=CBG 577CAD5BM=B; H<9 CH<9F ,C CF H<9 7CBH9LH C: 7CAAIB=75H=CB

F99 DC@MGM@@56=7 D<9A=7 A5F?9FG

&CGH@M D5F5@@9@ 7CBGHFI7H=CBG

9B9F5@ A5F?9F IG98 5@GC :CF 5 7CA 6=B5H=CB F9@5H=CB ;9B9F5@ A5F?9F CB@M IG98 :CF 7CBHF5GH F9@5H=CBG 898=75H98 A5F?9F :CF 7CIBH9F9LD97 H5H=J9 7CBHF5GH 898=75H98 A5F?9F :CF 7CFF97H=J9 7CBHF5GH 898=75H98 A5F?9F :CF CDDCG=H=J9 7CBHF5GH

&CGH@M D5F5@@9@ 7CBGHFI7H=CBG

&CGH@M ACBCGM@@56=7 5B8 ACBCACF D<9A=7 A5F?9FG

'CBD5F5@@9@ 7CBGHFI7H=CB :CF G=AI@ H5B9CIG 5B8 5H9ADCF5@ 7CA6=B5H=CB BCBD5F5@@9@ 7CBGHFI7H=CB :CF G9 EI9BH=5@ 7CA6=B5H=CB

(&)%1

,2'--" )+%%%",& /, '(' )+%%%",&

86DA4 CCF8=B5H=CB F9@5H=CBG GIAA5FM C: H<9 5HH9GH98 7C8=B; D5HH9FBG

&+$+, "-2

DC@MACF

6G9B79 C: 7CBB97H=J9 7C8=B; 5@H9F B5H=J9 DF9G9B79 C: GCA9 =FF95@=G A5F?9F

CBHF5GH 7CB79DHI5@ GD579 CDDCG= H=CB ^ 7CFF97H=CB ^ 89B=5@ 9LD97H5H=CB

G=AI@H5B9CIG 5B8 5H9ADCF5@ 7CA6= B5H=CB 5@K5MG 7C898 6M H<9 G5A9 7CB GHFI7H=CB

,&'-" (&"',

5B8

GMB89G=G :CF G=AD@9 5@H9FB5H=J9 GMB89G=G :CF 7
,",

GMB89G=G :CF 7CIBH9F9LD97H5H=J9 7CBHF5GH GMB89G=G :CF CDDCG=H=J9 5B8 7CFF97H=J9 7CBHF5GH

%-+'-"/

0=89GDF958 IG9 C: 5GMB89G=G CIHG=89 IFCD9

('-+,-

,2',", /, ,2'

(&"'-"('

><18=0C8>= 2>=CA0BC 0=3 0;C4A=0C8E4 2>=2;D38=6 A4<0A:B

264 Combination, contrast and alternative relations in comparison The analysis of the attested semantic domains reveals the internal organization of the contrast conceptual space, which highlights the functional proximity of opposition with correction on the one hand, and the proximity of correction with the denial of expectation contrast on the other hand. Finally, dedicated contrast markers are morphophonologically more complex than general ones, because the latter occur in more contexts and reduce their phonological substance in consequence of their frequency of use. Furthermore, there are markers that may be used for both combination and contrast relations. Such markers reveal that the internal conceptual structure of coordination is more complex than the simple threefold distinction between combination, contrast and alternative. There are indeed areas where the distinction between a relation and the other is rather a matter of emphasis. It is possible to identify an internal diversiﬁcation also in the expression of the alternative relation. First of all, alternative entails the irreality of the SoAs it links. As a result, if the alternative relation is not speciﬁcally coded by an overt marker, the linked SoAs must be somehow presented as irrealis anyway, in order to allow for an alternative interpretation. Moreover, choice-aimed alternative is more likely to be expressed by juxtaposition. This arguably depends on the fact that it is more easily inferable than simple alternative, even though the former implies the latter. It is indeed easier to infer an alternative relation from the juxtaposition of two interrogative clauses, than from the juxtaposition of two declarative ones. Further results have been achieved through the comparison of the strategies coding the three coordination relations. In particular, constructions expressing combination relations are more often asyndetic and the markers used to express at least one combination relations are morphophonologically simpler than those only used for alternative and contrast, either general or dedicated. In Fig. 7 the implicational patterns identiﬁed in the comparison of the three coordination relations under exam are summarized. As thoroughly argued in the course of this chapter, combination, contrast and alternative are not three equivalent sub-types of coordination. Combination only establishes the cooccurrence of two SoAs and the speaker only wants the linked SoAs to be considered as jointly at issue. Alternative and contrast, on the other hand, further specify the relation in terms of irreality and reciprocal replaceability or in terms of a conﬂict, respectively.

/,

EI=J5@9BH 7CAD@9L=HM

&CGH@M D5F5@@9@ 7CBGHFI7H=CBG

&5F?9F IG98 :CF 5H @95GH CB9 5@H9FB5 H=J9 F9@5H=CB A5F?9F IG98 :CF 5H @95GH CB9 7CA6=B5H=CB F9@5H=CB

'CB D5F5@@9@ 7CBGHFI7H=CB :CF G= AI@H5B9CIG 7CA6=B5H=CB 5H9ADCF5@ 7CA6=B5H=CB G=AD@9 5@H9FB5H=J9 5B8 7
'CBD5F5@@9@ 7CBGHFI7H=CB :CF G=AI@ H5B9CIG 5B8 5H9ADCF5@ 7CA6=B5H=CB CDDCG=H=CB 7CFF97H=CB 89B=5@ C: 9L D97H5H=CB BCBD5F5@@9@ 7CBGHFI7 H=CB :CF G9EI9BH=5@ 7CA6=B5H=CB

86DA4 CCF8=B5H=CB F9@5H=CBG =B 7CAD5F=GCB GIAA5FM C: H<9 5HH9GH98 7C8=B; D5HH9FBG

'(' )+%%%",&

,2'--" )+%%%",&

&+$+, "-2

98=75H98 A5F?9F :CF G9EI9BH=5@ 7CA6=B5H=CB ;9B9F5@ A5F?9F 9L DF9GG=B; 5H @95GH CB9 7CA6=B5H=CB F9@5H=CB ;9B9F5@ A5F?9F CB@M 9LDF9GG=B; 7CBHF5GH F9@5H=CBG 898=75H98 A5F?9F :CF 5 7CBHF5GH F9@5H=CB

=::9F9BH 7CBGHFI7H=CBG

=::9F9BH 7CBGHFI7H=CBG

G9EI9BH=5@ 7CA6 ^ G=AI@H 7CA6 ^ 5H9AD 7CA6 ^ CDDCG=H ^ 7CFF97H ^ 89B=5@ 9LD97H

,&'-" (&"',

(&)%1

GMB89G=G :CF 7CIBH9F9LD97H5H=J9 7CBHF5GH 5GMB89G=G :CF G=AD@9 5@H9F B5H=J9 5GMB89G=G :CF CDDCG=H=J9 5B8 7CFF97H=J9 7CBHF5GH 5GMB89G=G :CF 7
,",

GMB89G=G :CF G=AD@9 5@H9FB5H=J9 5GMB89G=G :CF H9ADCF5@ 5B8 5H9ADCF5@ 7CA6=B5H=CB 5GMB89G=G :CF 7
('-+,- %-+'-"/

,MB89G=G :CF G9EI9BH=5@ G=AI@H5B9 CIG 5H9ADCF5@ 7CA6=B5H=CB CDDCG =H=J9 7CBHF5GH 7CFF97H=J9 7CBHF5GH ,MB89G=G :CF 7CIBH9F9LD97H5H=J9 7CB HF5GH

(&"'-"(' %-+'-"/

,2',", /, ,2'

(&"'-"(' ('-+,-

><18=0C8>= 2>=CA0BC 0=3 0;C4A=0C8E4 2>=2;D38=6 A4<0A:B

266 Combination, contrast and alternative relations in comparison The basicness of combination is ﬁrst of all mirrored by the combinationcontrast and the combination-alternative coding implicational patterns, according to which combination relations are more likely to be expressed by means of simple juxtaposition, since they are more easily inferable than contrast and alternative ones. Moreover, the attested markers used for at least one combination relation are morphophonologically simpler than those coding contrast or alternative, as shown by the coding complexity implicational patterns. The basicness of the relation determines indeed a higher frequency of use of the markers expressing it, which therefore undergo processes of phonological erosion. The parameter of coding of the SoAs, instead, highlights a different property of coordination relations. Non-parallel constructions are traditionally explained with respect to two functional principles: iconicity of independence and syntagmatic economy (cf. Cristofaro 2003: chapter 9 and Haiman 1985: chapter 2 and 4). According to the principle of iconicity of independence, the less independent two concepts are, the less independent are the expressions coding them (Cristofaro 2003: chapter 9, Givón 1979: ch.13). A crucial notion of this principle is that of semantic integration of the linked SoAs, since it is one of the main reasons of interdependence between two SoAs. What characterizes two semantically integrated SoAs is the reduction of the boundaries between them, since their interconnection determines an erosion of their reciprocal limits (Cristofaro 2003: 120), so that it is difﬁcult to tell where the ﬁrst SoA ﬁnishes and where the second one starts. This notion characterizes a number of subordinate relations, like manipulative, phasals and perception predicates. According to the principle of syntagmatic economy, on the other hand, information that is available in the context does not need to be further speciﬁed, as widely discussed in the preceding sections. The presence of a deranked verb form may depend on the fact that the relation established between the two SoAs predetermines the relative temporal location of the SoAs. Hence, tense, aspect and mood speciﬁcations can be made explicit for just one SoA, because information on the other one are easily recoverable from the context. Coordination relations show a strong tendency to be coded by means of parallel constructions across languages. This is due to the fact that combination, alternative and contrast do not predetermine any semantic properties of the linked SoAs and are established between conceptually independent SoAs, which are not interconnected. The only exception, as widely argued in section 6.2.3, is sequential combination, which predetermines the temporal location

Combination, contrast and alternative: concluding remarks

267

o and, if expressed by a dedicate onstruction, it may eave the tense, aspect and mood speciﬁcations of one of the linked SoAs unspeciﬁed. The attested constraints on the cross-linguistic coding of coordination relations have all been explained with reference to a single overall functional principle, namely that of economy. Two main manifestations of this principle are at work in the coding of combination, contrast and alternative. On the one hand, what is already recoverable from the context does not need to be further speciﬁed, on the other hand, the most frequent linguistic expressions tend to be reduced. Both manifestations are instances of syntagmatic economy, that is, the tendency to economize on the length or complexity of any linguistic expression (Haiman 1985: 13). The principle of information recoverability is at work in the coding of the relation and in the coding of the SoAs. In the ﬁrst case, relations which are easily inferable from the juxtaposition of two SoAs are not further indicated by explicit markers. In the second case, tense, aspect and mood speciﬁcations of two SoAs linked in a sequential combination may be marked just for one SoA, because the relation itself predetermines the respective time reference of the other. The connection between frequency of use and economy, on the other hand, is evident in the morphophonological complexity of the attested markers: the more frequent ones tend to be morphophonologically simpler. The frequency of a marker may depend on the basicness of the relation it expresses or on its semantic domain: the more general is a marker, the higher is the number of relations it may be used for and, consequently, the more frequent is its occurrence in discourse.

6.4.3 Contrast and alternative: two different semantic axes The cross-linguistic coding of the three coordination relations has highlighted that both contrast and alternative separately show implicational coding asymmetries with respect to combination. However, the comparison of the two relations does not show particular coding patterns. Rather, contrast and alternative are expressed by different constructions characterized by a more or less equivalent degree of morphophonological complexity.

268 Combination, contrast and alternative relations in comparison The attested individual implications with respect to combination (see Figure 7) and their independent coding across languages suggest that contrast and alternative are independent relations. In particular, they may be analyzed as conceptual speciﬁcations of the basic combination relation along two separate and independent semantic dimensions. The exam of the attested semantic domains, the morphophonological complexity of coordinating markers and the different degrees to which the relations may be inferred by juxtapositive strategies suggest that the conceptual space of coordination relations is internally structured along two different and ultimately non-comparable directions of semantic speciﬁcation. Combination constitutes the most basic and simple relation that can be established between two SoAs. Then, this relation may be speciﬁed and enriched in many ways: the two SoAs may be identiﬁed as elements of a causal or temporal process, as conﬂicting facets of the same scene, as alternative possibilities, and so on and so forth. In Figures 8 and 9 the two dimensions of increasing semantic speciﬁcity characterizing contrast and alternative are represented. Fig. 8 shows the dimension along which a combination relation is further speciﬁed in terms of a conﬂict. The order in which the coordination relations occur from left to right mirrors their collocation along the combinationcontrast conceptual space described in Table 29, the explanation of which has been widely discussed in section 6.2.3. Both axes in Fig. 8 symbolize an increasing semantic speciﬁcation of the combination relation. Moving along the horizontal axis, the continuum goes from the simple combination of two SoAs to the establishment of a contrast between them. In order to establish a conﬂict between two SoAs, a discontinuity must be perceived between these SoAs, for example in terms of their participants, in terms of the type of action being described or in terms of the expectations they generate. Discontinuous SoAs may also be linked in a combination relation, if the speaker does not want to emphasize the conﬂict between them. However, as already argued, discontinuous SoAs may easily give rise to oppositive interpretations, because their combination may be interpreted with respect to the differences they show, rather than in terms of their pure cooccurrence. As can be observed in Fig. 8, discontinuity is the semantic dimension along which combination may develop into contrast. In order to present two SoAs as conﬂicting, it is indeed necessary that some discontinuity is per-

!!" #! # ! $#

may hardly be interpreted as standing in contrast (cf. Givón 1990: 850-51). "%! & '# # !(# ! t #) t ! !" #! i*&# #e d# !(!+ !) ! &,$ # $ !# !(" *&# ! ! !+ " -"#! - && $ ."! & '# ! !" $ # ! ! #) ! #*- !# !" & ! "$ ! "& ! !n can b #($.$$ ! !*" & $ !*" & $ !" #! &+%$ !"

270 Combination, contrast and alternative relations in comparison subtypes depending on the origin of the conflict originating it. The identiﬁcation of an increasing degree of semantic speciﬁcation is based on the cut-off points attested in two cross-linguistic implicational patterns of variation: the combination-contrast coding implication, concerning the presence of an overt coordinating marker (cf. (6.1)) and the combinationcontrast coding complexity hierarchy, concerning the morphophonological complexity of the attested markers (cf. (6.7)). These cross-linguistic patterns of variation have been explained with refe of the involved relations. On the one hand, been argued to be less easily inferable than hand, combination relations have been argued nating markers because they occur mor requen is due to their basic semantics. For some relations data did not show any particular functional asymmetry (like temporal and atemporal combination, or oppositive and corrective contrast). These relations are located one after the other on the basis of the combination-contrast conceptual space. In order to highlight the points along the horizontal axes where a coding cut-off point occurs, the symbol ">" has been used. The more rightwards and downwards a relation is located within the ﬁgure, the more semantically speciﬁed it is, along a hypothetical diagonal going from the origin of the axis towards the bottom right corner of the ﬁgure. Therefore, the representation in Fig. 8 predicts that, other things being equal, the closer a relation is to the bottom right corner of the ﬁgure, the more difﬁcult it will be to infer and the more complex dedicated markers will be. As shown in the preceding discussion, factors determining the morphosyntactic complexity of coordinating markers are diverse and cannot be restricted to the conceptual complexity of the relation they express. The same holds for the degree to which a relation is easily inferable from the context. However, the degree to which a basic cooccurrence relation is further semantically speciﬁed plays a decisive role in the cross-linguistic coding of coordination. Fig. 9 shows the semantic dimension along which a basic combination relation can be further speciﬁed in terms of non-cooccurrence. As pointed out in section 5.2.2, the notion of alternative implies the combination of two irrealis SoAs, i.e. the SoAs are jointly presented as a set of possibilities. hese sibilities may be further speciﬁed as being replaceable and non-cooccur ring.

><18=0C8>= 2>=CA0BC 0=3 0;C4A=0C8E4 2>=2;D38=6 A4<0A:B

" '+,"'

,&'-" ,)"""-2

(&"'-"(' " '+,"'

%-+'-"/

,&'-" ,)"""-2

B4@

B8<

0C4A0;

27>82408<43

B8
86DA4 -<9 7CA6=B5H=CB5@H9FB5H=J9 G9A5BH=7 8=A9BG=CB

"B H<9 5B5@MG=G C: 7CBHF5GH H<9 BCH=CB C: 8=G7CBH=BI=HM HIFB98 CIH HC D@5M 5 A5>CF FC@9 =B H<9 =B7F95G=B; GD97=:=75H=CB C: 7CA6=B5H=CB -<9 GD97=:=75H=CB C: 7CA6=B5H=CB =B H9FAG C: 5@H9FB5H=J9G =BGH958 ;C9G H
272 Combination, contrast and alternative relations in comparison axis in Fig. 9, combination and alternative are speciﬁed with respect to the parameters of temporality and aim. As pointed out for Fig. 8, the further a relation is located rightwards and downwards in the ﬁgure, the more difﬁcult it will be to infer and the more complex dedicated markers will be. Along with the conceptual complexity, the coding of coordination relations is inﬂuenced by numerous other factors. For instance, choice-aimed alternative appears to be more easily inferable from the context than simple alternative, because an alternative relation is more easily inferred from the juxtaposition of two interrogative clauses than from the juxtaposition of two declarative ones. On the other hand, the fact that alternative markers are at least as complex as combination markers is related to the semantic basicness of combination with respect to alternative (see the discussion in section 6.3.3). As shown by Ohori (2004), combination is more frequently established in discourse than alternative, as a result of its being more simple and basic, and is thus marked by simpler connectives. To sum up, combination, contrast and alternative are not equivalent subtypes of coordination and the cross-linguistic coding of these relations mirrors their different semantic structures. Combination is the most basic relation that may be established between two functionally parallel SoAs, while contrast and alternative are further speciﬁcations of this relation along two different and separate axes of increasing semantic speciﬁcity, as summarized in Fig. 10. A combination of SoAs may be speciﬁed either in terms of some discontinuity, originating a contrast, or in terms of the irreality of the SoAs it links, creating a set of alternative possibilities. Along the two axes in Fig. 10, the more rightwards and far from the horizontal axis a relation is located in the ﬁgure, the more semantically speciﬁed it is, along two hypothetical diagonals going from the origin of the axes towards the bottom right and the top right corners of the ﬁgure. The representation in Fig. 10 predicts that, other things being equal, the closer a relation is to the bottom right corner or to the top right corner of the ﬁgure, the more difﬁcult it will be to infer and the more dedicated markers coding it will be complex.

><18=0C8>= 2>=CA0BC 0=3 0;C4A=0C8E4 2>=2;D38=6 A4<0A:B

86DA4 -<9 G9A5BH=7 8=A9BG=CBG C: 7CCF8=B5H=CB F9@5H=CBG

Chapter 7 A focus on Europe 7.1

Introduction

As was pointed out in section 1.2, the uniformity and availability of European data makes it possible to examine the phenomena taken into account in this study at a deeper level of analysis. The European sample and the Comparison sample may be compared to identify the coding patterns that are not attested in the languages of Europe or, by contrast, that are attested in all European languages. Furthermore, the geographical distribution of these coding patterns can be examined. The differences between European and non-European data have been already highlighted in the preceding chapters. However, the comparison of the two samples was not the object of the ﬁrst level of analysis and has thus not been systematically examined. Now that the cross-linguistic variation has been presented and discussed, it is possible to systematically investigate how coordination relations are encoded in the languages of Europe, as opposed to non-European languages (cf. discussion in Mauri 2007). The aim of such a comparison is the identiﬁcation of the coding patterns which characterize European languages, to enrich the picture of phenomena shown by the SAE area. As discussed in section 1.2, increasing attention has been paid to the linguistic characterization of Europe in the last years (cf. projects EUROPTYP and MEDTYP, Ramat and Stolz 2002). This area shows at the same time a signiﬁcant internal diversiﬁcation, as will be pointed out in the next sections, and a consistent amount of similarities, as highlighted by Haspelmath (1998 and 2001). Given the availability of high quality descriptions and the detailed historical documentation, European languages thus constitutes a particularly advantageous research ﬁeld for the study of cross-linguistic variation and contact phenomena. Haspelmath (1998) identiﬁes a nucleus, a core and a periphery within the SAE linguistic area, as can be seen from Fig. 11. According to Haspelmath, the languages belonging to the nucleus are Dutch, German, French and Northern Italian, because they share the highest number of features characterizing the European Sprachbund. The other Romance languages, together with Germanic, Balkan and the West and South Slavic languages, belong to the core.

Introduction

275

Finally, in the periphery we have the East Slavic, the Baltic and the FinnoUgric languages, Maltese, Basque, Georgian and Armenian.

Figure 11: The languages of the Standard Average European (Haspelmath 1998: 273)

As can be observed by comparing Fig.11, Fig.16 and Table 48 in the Appendix, all languages indicated by Haspelmath as belonging to SAE have been examined in this study, except for Slovenian and Armenian. Furthermore, a number of languages spoken outside the SAE area have been also considered, in order to be able to delimit possible areal phenomena with respect to the surrounding languages. Turkish, Irish and three Nakh-Daghestanian languages, Dargi, Chechen and Lezgian, have been therefore included in the European sample. In sections 7.2, 7.3 and 7.4, the coding patterns attested in the European sample will be systematically analyzed with respect to combination, contrast and alternative respectively. In examining each coordination relation, the discussion will focus on the construction types that are not attested in Europe and on those that turn out to be widespread in this area. Finally, every section will show a synthetic overview of the geographical distribution of the attested coding patterns.

276 A focus on Europe Section 7.5 will sum up the results achieved in this second level of analysis and a restricted area in Central Western Europe will be identiﬁed, in which languages express the three coordination relations by means of similar constructions, showing the same values for each examined parameter.

7.2

Combination relations

All 37 European languages have at least one syndetic construction to express the combination of two SoAs. This means that all European languages have at least one overt marker used to express temporal and atemporal combination. However, in Turkish and Chechen asyndetic constructions are also used as the normal and stylistically unmarked way to express the combination of two SoAs. In chapter 2 two examples from Chechen (cf. (2.3) and (2.12)) were presented to illustrate both the syndetic and the asyndetic constructions attested in the language. The overwhelming predominance of overt markers in the languages of Europe could be related to the high degree of written language tradition that characterizes this area. In principle, it could also be a consequence of the high degree of contact among the languages, which could have caused the borrowing of overt markers in those languages that did not have any, given that combination markers are easily borrowed across languages (see Mithun 1988: 357 and Matras 1998: 301-305). However, the only two European instances of borrowing are attested in Dargi and Lezgian, which, besides their native combination markers, normally employ the Arabic connective wa. Borrowing phenomena of markers expressing combination are more widespread outside Europe, and the connection to the high degree of written language tradition seems to be the most likely explanation for the very restricted use of asyndesis (see Matras 1998: 301-305 for a detailed discussion on the borrowing of coordinating markers). All European languages except for Irish and Basque have at least one monosyllabic marker used for at least one combination relation. Irish and Basque use instead the bisyllabic connectives agus and eta respectively. Moreover, no polymorphemic marker is attested in European languages for the expression of combination relations. All European languages except for Chechen (see example (2.3), where the marker ’a is a clitic33 ) have at least one connective for the expression of combination which consists of a free morpheme. Dargi, Lezgian and Turkish

Combination relations

277

show both types and have the possibility to use either free morphemes or bound ones (Dargi -ra, Lezgian -na and Turkish -Ip). In any case, the use of free forms to combine two SoAs is certainly the most widespread strategy across Europe. Compared to the European picture, data from the Comparison sample is less uniform, but conﬁrms the tendency pointed out for European languages. Monosyllabic and monomorphemic markers are well attested, even if a number of languages employ bisyllabic markers (like Hausa kuma, Hocak ˛ ánaga, ˛ Meithei aduga, Mosetén jö’dyë’yä and NànáfwˆE kp´Ekún). Moreover, in the Comparison sample both free and bound markers are attested, even though the free ones outnumber clitic and afﬁxes. As far as the attested semantic domains are concerned, all 37 European languages have at least one general connective used for both temporal and atemporal combination, like English and, Italian e, Spanish y, Russian i, Finnish ja, and so on. These markers simply encode the general relation of combination, without further speciﬁcations on the temporal location of the linked SoAS. Besides the general marker, however, some languages also show more speciﬁc connectives. For instance, the non-parallel constructions attested in Turkish and Lezgian can only be used to express the relation of sequential combination. Dedicated sequential connectives can be found also in SerboCroatian (pa) and Bulgarian (ta/da). Other languages show general markers which are not employed for temporal combination, but are used for discontinuous atemporal combination and opposition, like a in Russian, Polish, Belorussian, Bulgarian and SerboCroatian (example 7.1), meg in Hungarian, iar in Rumanian, ta in Ukrainian, o in Lithuanian and tq’a in Chechen. As shown in the combination-contrast semantic space (Table 29 and Fig. 8), these connectives have a semantic domain which partly covers both combination and contrast, since they are basically used for the combination of discontinuous SoAs. As pointed out in section 3.2, languages with general markers for atemporal combination and opposition employ them in contexts where English would use the marker and. In (7.1) three examples from Serbo-Croatian are repeated, in order to illustrate the co-existence within the same language of a general marker employed for both combination types (7.1a), a general marker used for discontinuous atemporal combination and opposition (7.1b) and a dedicated sequential connective (7.1c).

278 A focus on Europe (7.1) Serbo-Croatian, Slavic, Indo-european (a.) Zauzet sam i ne mogu više slušati busy be:1sg COORD NEG can:1sg anymore listen ‘I’m busy and I can’t listen anymore.’ (Brown and Alt 2004: 70) Petar spava (b.) Ja radim a I work:1sg COORD Peter sleep:3sg ‘I work and Peter sleeps.’ (M.C., questionnaire) ˇ sam je grmljavinu pa (c.) Cuo hear:PTCP.PST AUX:1sg thunder:ACC COORD AUX:3sg poˇcela kiša begin:PTCP.PST rain ‘I heard a thunder and it started to rain.’ (M.C., questionnaire) While every European language has at least one general marker expressing both temporal and atemporal combination, in the Comparison sample there are languages which only show a dedicated construction for sequential combination and a general one for simultaneous and atemporal combination (cf. example (3.17) from Tukang Besi). In Tuvaluan, for instance, no general combination marker is available. The only attested marker kae is used to convey atemporal combination and all contrast relations, whereas sequential combination can only be expressed by means of a juxtapositive strategy. There also are non-European languages which have a general combination marker, like Somali oo, but this is attested in only 20 languages out of 37. All European languages have at least one syntactically parallel construction expressing combination relations, where both verb forms encoding the combined SoAs are balanced forms. Only two languages out of 37 (Turkish and Lezgian) show non-parallel constructions, where the ﬁrst verb form is deranked and syntactically depends on the last one. Examples from both languages have already been given in (1.8) and (1.34). Outside Europe, on the other hand, non-parallel constructions are more widespread and, most importantly, there are a few languages in the comparison sample which do not use parallel constructions to express combination. These languages are Tauya, Kolyma Yukaghir, Maricopa and Jamul Tiipay. Other languages show both possibilities, like Korean, West Greenlandic and Japanese. Lango, Harar Oromo and Supyire only use non-parallel constructions in the expression of sequential combination and could not express this relation with a parallel strategy.

Combination relations

7.2.1

279

Combination in Europe: synthesis and conclusions

In Fig. 12 a unitary picture of the cross-linguistic variation attested in Europe is given. As can be observed in the map, dedicated sequential constructions and markers used for at least discontinuous atemporal combination and opposition are attested in the Central Eastern part of Europe. By contrast, Central Western European languages only have a general marker expressing temporal and atemporal combination. The semantic domains of the attested markers used to express at least one combination relation thus highlight a clear difference between Central Western and Central Eastern European languages, since the latter show more cross-linguistic variation than the former. Furthermore, also non-parallel constructions, asyndetic constructions and bound markers are attested only in the Eastern periphery of Europe, in languages that Haspelmath identiﬁes either as non-belonging to the SAE area (cf. Fig. 11) or as belonging to its periphery. By contrast, the two languages which have only bisyllabic combination markers, Irish and Basque, are in the western periphery of Europe. However, both of them are not typical instances of SAE either, since Irish is not considered as being part of SAE and Basque is located in the periphery. SAE languages turn out to be very consistent with respect to the morphosyntactic properties of constructions expressing combination. Deviating behaviors are only attested in the periphery or in the languages surrounding the area. The exam of the semantic domains, instead, highlights a signiﬁcant variation within Europe. From what has been said throughout this section, on the basis of the variation attested in Europe and after the comparison with non-European data, it is now possible to characterize the coding of combination in European languages. The constructions expressing combination attested in the examined European languages are mainly syndetic and syntactically parallel. Overt markers used for at least one combination relation tend to be monosyllabic and free morphemes. Moreover, every European language has at least one general marker used to express both temporal and atemporal combination, and different markers (i.e. dedicated sequential connectives, and general markers used for atemporal combination and opposition) are only attested in Eastern Europe. Of the languages that Haspelmath (2001: 273) identiﬁes as belonging to the nucleus of the Standard Average European (SAE), only Polish, Serbo-Croatian and Bulgarian have markers other than the general one used for all combination relations.

G9EI9BH=5@ 7CA6=B5H=CB

///// ;9B9F5@ A5F?9F :CF 5H @95GH 5H9ADCF5@ 7CA6=B5H=CB 5B8 CDDCG=H=CB

86DA4 CBGHFI7H=CBG 9LDF9GG=B; 7CA6=B5H=CB =B H<9 @5B;I5;9G C: IFCD9 \\\\\\ 898=75H98 A5F?9F:CF

5>2DB >= DA>?4

Contrast relations

7.3

281

Contrast relations

From a purely morphosyntactic point of view, the constructions attested in Europe to code contrast relations are quite homogenous. The attested semantic domains, instead, reveal an interesting cross-linguistic variation, within which the Central Western part of Europe turns out to be once again more united in its behavior than the Eastern part. As far as the presence of overt contrast markers is concerned, all 37 European languages have at least one overt coordinating marker expressing at least one contrast relation. There is one contrast type, correction, that may normally be expressed in discourse by means of juxtaposition (cf. chapter 4). However, all European languages also show the possibility to mark this relation by means of an overt connective, even Turkish and Georgian, where asyndesis is the normal strategy to express correction (cf. example (4.18) from Georgian in chapter 4). The attested markers used only to express contrast are morphophonologically more complex than the ones used also to express a combination relation, as pointed out in section 6.2. All markers expressing contrast are free morphemes (only Dargi shows, besides amma borrowed from Arabic,34 the bound marker -gu). As for the number of syllables, 19 languages out of 37 show at least one polysyllabic marker expressing contrast. Among these languages we can count German (aber and sondern), Swedish (utan), Spanish (pero and sino) and Greek (alla). Examples of monosyllabic connectives are Italian ma, English but, French mais, Russian no, Danish and Norwegian men. Some of the attested polysyllabic markers are also polymorphemic, like the corrective markers in Hungarian, hanem, and in Spanish, sino, which have the same internal structure of a particle meaning ‘if’ (ha and si) combined with a negation (nem and no). Some complex markers are attested for the expression of opposition, where we ﬁnd connectives which are normally used to convey also other types of relations, mainly relations of simultaneity. Such markers have not been used by native speakers in the questionnaire, except for a few languages, and are not well described in grammars, because the semantic function of simultaneity is still felt to be the main one. Yet, there are situations in which such markers can be used to express oppositive contrast without implying that the two SoAs are simultaneous. Examples of these oppositive markers are French tandis que, which is also

282 A focus on Europe polymorphemic, Italian mentre, English while and German während. Because of the incompleteness of data, such markers are not represented in Fig. 13. This only means that more diachronic and typological research has to be done on this subject. The attested semantic domains reveal a signiﬁcant diversiﬁcation within Europe. There are languages with a general marker used to express at least atemporal combination and opposition (cf. section 7.2): for example Polish (example 4.22), which shows the marker a, also shared by Russian, Ukrainian, Bulgarian, Belorussian and Serbo-Croatian, Lithuanian (o), Hungarian (meg) and Chechen (tq’a). In Russian, Bulgarian, Chechen and Lithuanian, these general markers may also express the corrective contrast (see examples in chapter 4). Dedicated markers are attested for the expression of each contrast relation. Examples of dedicated oppositive connectives have been already provided (mentre in Italian or tandis que in French). Some languages have dedicated markers for correction and denial of an expectation, like German (aber ∼ sondern), Basque (baina ∼ baizik), Spanish (pero ∼ sino), Swedish (men ∼ utan), Finnish (mutta ∼ vaan), Estonian (aga ∼ vaid), Lithuanian (bèt ∼ o), Russian and Bulgarian (no ∼ a), Chechen (amma ∼ tq’a), Romanian (dar ∼ ci), Serbo-Croatian (ali ∼n‚ego/v‚ec´ ), Icelandic (en ∼ heldur) and Hungarian (de ∼ hanem). Finally, in the Central Western part of Europe, many languages also have a general marker used for both counterexpectative and corrective contrast: Italian (ma), French (mais), Dutch (maar), Luxembourgish (mee), English (but), Portuguese (mas), Czech (ale), Albanian (por), Greek (alla). The constructions expressing contrast attested in Europe are all syntactically parallel. The preference for parallel strategies in the expression of contrast is also conﬁrmed in the Comparison sample. Non-parallel constructions expressing contrast are only attested in Kolyma Yukaghir, Tauya, Korean and West Greenlandic, as discussed and exempliﬁed in section 4.3.

7.3.1

Contrast in Europe: synthesis and conclusions

The data presented throughout this section show that the constructions used for contrast relations in the languages of Europe have a high degree of uniformity in terms of morphosyntactic features: they are all parallel, mostly syndetic, with free mono or polysyllabic markers. By contrast, considering

Contrast relations

283

the semantic domain of the attested constructions, the languages of Europe encode the different subtypes of contrast in many different ways: with general markers only expressing contrast relations, general markers employed at least for atemporal combination and opposition, or dedicated markers. The attested cross-linguistic variation is shown in Fig. 13, where the areal distribution of dedicated markers clearly reveals a Central Western area in which languages behave in the same way, showing at least one general marker for corrective and counterexpectative contrast and no general markers of atemporal combination and opposition (the area left white). Dedicated markers and general markers used for atemporal combination and opposition are attested in the Eastern part of Europe and in Spain, at the Western periphery. To sum up, the expression of contrast in the languages of Europe is characterized by parallel and mostly syndetic constructions. Markers coding contrast relations consist of free, mono and polysyllabic morphemes. The exam of the semantic domains highlight instead three different areas: a central area with general markers for counterexpectative and corrective contrast; an eastern area with dedicated markers for counterexpectative and corrective contrast, and with general markers for discontinuous atemporal combination and opposition; a Basque-Hispanic area to the west with dedicated markers for counterexpectative and corrective contrast. Languages which are very similar under many respects differ in the semantic domains of markers coding contrast. For instance, Swedish has two dedicated markers for correction and denial of expectation, while Norwegian and Danish only show the general marker men. German has two dedicated connectives (aber and sondern), but Dutch and Luxembourgish do not (using maar and mee, respectively). The same difference is attested between Spanish and Portuguese. The former has the two different dedicated markers pero and sino, while the latter only uses the general connective mas. There may be contact phenomena at the origin of this interesting geographical distribution, as well as individual grammaticalization processes leading to the development of dedicated (or general) markers. More diachronic studies need to be done in order to shed some light on the reasons motivating the synchronic areal distribution.

86DA4 012345637012 89:482271; 60134<23 71 3=8 ><1;5<;82 0? 540:8 \\\\\ ><1;5<;82 @73= A8A76<38A B<4C842 ?04 605138489:863<37D8 <1A 60448637D8 60134<23 <1A 10 ;8184<> B<4C84 89:482271; 3=8B E03= ///// ;8184<> B<4C84 528A ?04 <3 >8<23 <38B:04<> 60BE71<3701 <1A 0::0273701

5>2DB >= DA>?4

Alternative relations

285

7.4 Alternative relations Let us now examine the constructions attested in Europe for the expression of alternative relations. As will be evident from the following discussion, these constructions show less cross-linguistic variation than constructions employed for combination and have more or less the same formal features across Europe. All 37 European languages only use syndetic constructions to express the alternative between two SoAs. This means that no European language uses an asyndetic alternative construction and that this relation is always coded by means of an overt coordinating marker. The fact that in the Comparison sample 12 languages out of 37 use at least one juxtapositive strategy for the expression of an alternative relation (cf. discussion in chapter 5) highlights that the complete absence of asyndesis in the expression of alternative strongly differentiates European languages from non-European languages. As already widely discussed, the attested asyndetic constructions are all characterized by some markers which overtly signal the irrealis potential status of the alternatives: interrogative markers in the expression of choiceaimed alternative and dubitative, conditional, hypothetical markers in the expression of simple alternative (example 6.5). In these constructions, the alternative relation is not fully encoded. What is encoded is the possible (or potential), rather than realis, status of the two SoAs, while the rest is left to inferential enrichment. The markers attested in the European languages that code alternative are morphophonologically more complex than the markers used for at least one combination relation. First of all, no bound markers are attested in the expression of alternative. All European connectives are free forms. While 35 European languages out of 37 have at least one monosyllabic combination marker, only 22 out of 37 have at least one monosyllabic marker for alternative, and 20 languages out of 37 show at least one bisyllabic connective for the expression of this relation. Examples of bisyllabic markers are German oder, Swedish eller, Russian ili and Polish albo. In Fig. 14, where a synthetic picture of European constructions expressing alternative is given, languages with a polysyllabic marker are indicated by a circle. Examples of monosyllabic connectives are instead English or, French ou, Italian o, Finnish vai and tai, Romanian sau and Maltese jew.

286 A focus on Europe Moreover, Lezgian, Dargi and Turkish show polymorphemic markers consisting of a morpheme meaning ‘and’ combined with a morpheme meaning ‘or’ (cf. section 5.2): Dargi yara (ya-ra ‘or-and’, see example 5.23), Lezgian waya (wa-ya ‘and-or’) and Turkish veya (ve-ya ‘and-or’). Turning now to the semantic domains of the attested constructions, the picture shows some cross-linguistic variation. 28 languages out of 37 have one general marker used for both choice-aimed and simple alternative. Examples of such general markers can be found in English or, French ou, Russian ili, Danish eller, Hungarian vagy and so on. There are, however, 9 European languages that do not have any general marker for alternative, but show two distinct dedicated connectives for choice-aimed alternative and for simple alternative relations. These languages are Polish (choice-aimed: czy, simple: lub/albo), Finnish (choice-aimed: vai, simple: tai), Belorussian (choice-aimed: ci, simple: abĲo), Albanian (choiceaimed: apo, simple: ose), Basque (choice-aimed: ala, simple: edo), Ukrainian (choice-aimed: cˇ y, simple: abĲo), Georgian (choice-aimed: tu, simple: an), Lezgian (choice-aimed: taˆxajt’a, simple: ja/wa-ja), Dargi (choice-aimed: aèi, simple: ya-ra). In (7.2) an example from Albanian is provided. In (7.2a) the choice-aimed alternative relation is coded by apo, while in (7.2b) the marker ose is used to convey a simple alternative. (7.2) Albanian, Indo-European (a) Nesër do ikiim ne shkollë apo do të rrimi tomorrow FUT go.1pl 1pl school COORD FUT stay.1pl ne shtëpi? 1pl home ‘Do we go to school tomorrow or do we stay at home?’ (L.S., questionnaire) (b) Zakonisht, shkruaj ose lexoj deri vonë usually write:1sg COORD read:1sg until late ‘Usually, I write or I read until late.’ (L.S., questionnaire) Nonetheless, languages having a general marker for both alternative relations are the majority and constitute a large interconnected area, as can be observed in Fig. 14. Dedicated markers are attested at the periphery of Europe and in the three Slavic languages Polish, Belorussian and Ukrainian, which

Alternative relations

287

developed this distinction from the same forms and probably inﬂuenced one another through contact phenomena. As far as syntactic parallelism is concerned, European languages behave in the same way and there is no cross-linguistic variation. All languages in the European sample use only syntactically parallel constructions to express choice-aimed and simple alternative relations. Although also in the Comparison sample there is a strong tendency for the use of parallel strategies in the expression of alternative relations, three languages employ non-parallel constructions: West Greenlandic, Korean and Jamul Tiipay (cf. discussion and examples in chapter 5). As widely discussed, the coding of two alternative SoAs tends to be parallel by virtue of the absence of any semantic predetermination intrinsic in the relation. In other words, the relation does not predetermine any semantic property of the linked SoAs, which thus require tense, aspect and mood characterizations.

7.4.1 Alternative in Europe: synthesis and conclusions The homogeneous behavior of European languages in the coding of alternative relations is shown in Fig. 14. Since they all have syndetic and parallel constructions, the only two parameters represented on the map are the morphophonological complexity of the attested markers (i.e. whether they are mono or bisyllabic) and their semantic domain, distinguishing between an area with dedicated choice-aimed and simple alternative markers and an area with a general connective used for both relations. As already said, languages with dedicated markers are mainly located in the Eastern part of Europe and, in any case, in the peripheral areas. The comparison with non-European languages highlights a set of features which can be said to characterize the constructions attested in Europe for the expression of alternative. The coding of alternative in Europe is characterized by the use of only syndetic and parallel constructions. All attested overt markers consist of free morphemes, both monosyllabic and bisyllabic. There is a large area with a general marker used for both choice-aimed and simple alternative relations, while dedicated markers are mainly attested in the periphery of Europe. Among the languages that Haspelmath (2001: 273) identiﬁes as belonging to the nucleus of SAE, only Polish and Albanian have dedicated alternative markers.

86DA4 012345637012 89:482271; <>3841<37D8 71 3=8 ><1;5<;82 0? 540:8 6=0768<7B8A <1A 27B:>8 <>3841<37D8

\\\\\ ;8184<> B<4C84 ?04

5>2DB >= DA>?4

Conclusions: ‘And-But-Or’ languages

7.5

289

Conclusions: ‘And-But-Or’ languages

In the course of this chapter, it has been shown that the European area shows both some internal cross-linguistic variation, mainly concerning the semantic domain of the attested constructions, and a high degree of structural homogeneity, especially if compared to the non-European data gathered in the Comparison sample. In this concluding section a further generalization will be made, based on the comparison of the attested constructions expressing combination, contrast and alternative in the European languages. Europe has turned out to be internally quite homogeneous in the coding of each of the three coordination relations analyzed, but an even more homogeneous area can be identiﬁed within the European borders, as represented in Fig. 15. I propose to call the languages located in this area ‘And-But-Or’ languages. The ‘And-But-Or’ languages are characterized by the following set of features: (7.3) - syndetic constructions for the three basic coordination relations; - internally parallel constructions for the three basic coordination relations; - free markers coding combination, contrast and alternative; - a general marker used for both temporal and atemporal combination (‘And’), a general marker used for both choice-aimed and simple alternative (‘Or’), a general marker used for the counterexpectative and the corrective contrast (‘But’) and no general markers used only for discontinuous atemporal combination and opposition. To this language type belong most Romance languages except for Rumanian, Spanish and Catalan (i.e. Italian, Sardinian, Portuguese and French), a Celtic language (Irish), all Germanic languages except for German, Icelandic, and Swedish (i.e. Norwegian, Danish, Dutch, Luxembourgish and English), only one Slavic language (Czech ), an Hellenic language (Greek) and Maltese, the only non-Indo-European language belonging to the area. The label ‘And-But-Or’ languages is proposed here for two main reasons. First of all, it is transparent with respect to at least two characterizing features: the presence of free markers and the use of three general connectives for three coordination relations examined in this research.

86DA4 1A53(4 ><1;5<;82

\\\\\ 48< @73= 3=8 1A53(4 ><1;5<;82

5>2DB >= DA>?4

Conclusions: ‘And-But-Or’ languages

291

Secondly, English is the most known language belonging to this type, therefore the use of the three English markers and, but and or allows for a straightforward identiﬁcation of the overall coding system, at least for people who can speak English. The three Maltese connective u, imma and jew could as well have been chosen to label the area, but the number of people speaking Maltese is rather low and the identiﬁcation of the coding system would have been less straightforward. Compared to non-European data, the set of features characterizing the ‘And-But-Or’ languages turns out to be not common at all. In the Comparison sample there is only one language showing the set of features pointed out in (7.3), namely Hocak, ˛ which expresses both combination relations by means of the free marker ánaga, ˛ both alternative relations by means of the marker ni˛˛igéšge, the counterexpectative and the corrective contrast by means of the marker nu˛ni˛ge. Furthermore, Hocak ˛ only uses parallel constructions and does not have a general marker for discontinuous atemporal combination and opposition. The series of features characterizing the‘And-But-Or’ languages is thus almost non-attested in the Comparison sample. Of course, many non-European languages show individual coding patterns which are also attested in the ‘And-But-Or’ languages, like the use of parallel constructions or the use of free coordinating markers. Yet, the co-existence of all features identiﬁed in (7.3) is not widespread at all and its occurrence in a number of adjacent languages of different families and sub-families is thus an interesting phenomenon, which is certainly worth further research in order to highlight the processes through which the actual areal distribution has developed. The explanations underlying the linguistic area where the ‘And-But-Or’ languages are spoken are probably connected to a variety of factors, that still need to be systematically inquired. The fact that many languages belonging to this area are genetically related to each other suggests that the coding patterns characterizing the area may derive from the patterns attested in the ancestor languages. However, even though it is plausible to hypothesize that the actual coding system has at least some root in the strategies used by the common ancestors, more research needs to be done in this direction. The actual geographical distribution might as well have been inﬂuenced by local historical and economic connections, that may have caused reciprocal inﬂuences among some languages within the area. For instance, Luxembourgish has borrowed the contrast marker mee from French (mais) and the marker expressing alternative from German oder, as a result of the close and

292 A focus on Europe lasting contacts with both languages. Yet, languages that have been in contact for centuries, like Spanish and Portuguese or Norwegian and Swedish, differ with respect to the expression of contrast: Portuguese and Norwegian use a general marker for the counterexpectative and the corrective contrast (mas and men respectively), whereas Spanish and Swedish use two different dedicated markers (pero and sino, men and utan). Contact phenomena may have occurred at a local level, but it is highly unlikely that all ‘And-But-Or’ languages had the same types of contacts with each other. Finally, a third factor may have played a role in the development of general and dedicated markers, namely the lexical origin of the grammaticalization processes through which the attested coordinating markers have developed. As Ducrot and Vogt (1979) point out, the Latin comparative construction of the type ‘A magis quam B’, has developed both a corrective and a counterexpectative function, as can be observed in Portuguese mas, French mais and Italian ma. On the other hand, markers developing from originally causal constructions, like the Latin construction per hoc, give rise to dedicated counterexpectative markers, like pero in Spanish and però in Italian. Markers developing from a simultaneity meaning, instead, tend to develop a dedicated oppositive function, like mentre in Italian, while in English, alors que in French. Likewise, alternative markers which may also function as interrogative particles, like czy in Polish or vai in Finnish, tend to be dedicated for the expression of choice-aimed alternative. Some grammaticalization patterns recur across languages and lead to hypothesize that, given the same lexical meaning at the origin of the grammaticalization process, there is a high probability that similar markers may develop. A comparative diachronic analysis of the attested coordinating constructions may thus highlight individual, even though parallel, patterns of grammaticalization. In any case, the explanations underlying the linguistic area where ‘And-But-Or’ languages are attested should be looked for in a variety of factors, among which genetic relatedness, contact phenomena and individual grammaticalization processes are likely to have played a role. There is one more aspect to point out, to conclude this chapter on European data. European languages are characterized by a strong morphosyntactic homogeneity in the coding of coordination relations, since they all employ parallel, syndetic constructions with free coordinating markers. Languages showing morphosyntactic properties deviating from the majority of European languages are located at the very Eastern periphery of Europe, in Turkey and

Conclusions: ‘And-But-Or’ languages

293

in the Caucasus. The inclusion of these two areas within Europe is indeed controversial. In this study, Altaic, Kartvelian and Nakh-Daghestanian languages have been included in the European sample following the EUROTYP guidelines. The aim of this inclusion was to consider the coding patterns attested in the languages surrounding the SAE area, in order to be able to clearly delimit the typical European coordination coding systems. On the basis of data presented in this chapter and throughout the whole work, it turns out that, at least concerning the coding of coordination relations, Altaic, Kartvelian and NakhDaghestanian languages deviate from the coding patterns attested in the other European languages, thus strengthening the linguistic border separating the Anatolian and Caucasian regions from the Western European area.

Chapter 8 Conclusion and prospects As stated in the introduction to this study, the aim of this research was to investigate the mechanisms underlying the coding of combination, contrast and alternative relations between SoAs. These relations have been identiﬁed as three basic instances of coordination, which has in turn been deﬁned in functional terms. The functional deﬁnition of coordination relation as a relation established between functionally parallel SoAs has made it possible to compare all the constructions attested to express combination, contrast and alternative, regardless of their speciﬁc morphosyntactic properties. The attested cross-linguistic variation reveals a number of constraints on the possible coding strategies, showing the mechanisms and the principles at work in the expression of coordination relations. The attested patterns demonstrate that the cross-linguistic coding of each coordination relation mirrors its internally complex and diversiﬁed conceptual structure. The different types identiﬁed within combination, contrast and alternative are indeed coded by different constructions across languages and the speciﬁc morphosyntactic properties shown by each strategy are directly connected to the speciﬁc relation type expressed. The principle of economy has turned out to be the basic functional motivation underlying the coding of coordination relations. The presence of overt markers signalling the speciﬁc relation and the syntactic parallelism of the construction depend on the recoverability of information from the context, either relative to the relation itself or to some properties of the linked SoAs. On the one hand, semantically more speciﬁed relations are more difﬁcult to infer from the simple juxtaposition of clauses and are more likely to be expressed by using overt connectives - namely counterexpectative contrast and simple alternative. On the other hand, if the relation expressed does not predetermine any semantic properties of the linked SoAs, they are both expressed by balancing strategies and parallel constructions - like in the expression of all coordination relations except for sequential combination. The morphophonological complexity of the attested overt markers is instead related to economy in a different way. More frequent markers tend to undergo processes of phonological and morphological erosion, thus becoming simpler. Hence, the degree of morphophonological complexity of a coor-

Conclusion and prospects

295

dinating marker is indirectly proportional to its frequency of use in discourse and its frequency in turn depends on the number of contexts where the given marker can be employed. As a result, markers coding basic relations like combination, either general or dedicated, are morphophonologically simpler than markers coding semantically more speciﬁed relations, like contrast or alternative, which are less frequently established in discourse. Moreover, general markers are in any case simpler than dedicated markers, which may express only one relation and thus occur in a low number of contexts. The comparison between the coding patterns attested for combination, contrast and alternative relations has thus highlighted a conceptual and coding complexity internal to coordination. The three relations examined are not equivalent instances of coordination relation, but are organized in a twofold hierarchical structure. Combination is the most basic and semantically unspeciﬁed linkage relation, for it is implied both by contrast and alternative. In order to identify a conﬂict or an alternative between two SoAs it is indeed necessary to combine them and consider them jointly. Contrast and alternative are regarded as further speciﬁcations of the combination relation along two different and separate semantic directions: contrast is a combination speciﬁed on the basis of a conﬂicting discontinuity between the SoAs, whereas alternative is a combination speciﬁed on the basis of the irreality and replaceability of the linked SoAs. The three coordination relations show different internal structures and these differences have consequences on their cross-linguistic coding. The basicness of combination is cross-linguistically mirrored by the use of morphophonologically simple markers and asyndetic constructions. By contrast, the other two coordination relations show more complex coding patterns, also employing polymorphemic connectives. Not only do combination, contrast and alternative not show an equivalent degree of conceptual complexity, but also the various types identiﬁed within each of the three relations are arranged along an orderly conceptual space. The analysis of the attested semantic domains has revealed recurring semantic maps across languages, whereby the combination, contrast and alternative types can be sorted along two axes of increasing semantic speciﬁcity. General constructions are used to express relations which stand close to each other in the combination-contrast and combination-alternative conceptual spaces. The two major results of this study can thus be described as follows. First of all, coordination is a very complex and internally diversiﬁed notion, since two functionally parallel SoAs may be linked in various relation types. These

296 Conclusion and prospects different relations, however, are not conceptually equivalent, nor are they randomly related to each other. Rather, the various coordination relations can be orderly analyzed along a twofold dimension of increasing semantic speciﬁcity, in a continuum going from bare combination to contrast and alternative. The increasing semantic speciﬁcity of contrast and alternative corresponds to an increasing complexity of their cross-linguistic coding. Secondly, the coding of the various coordination relations displays signiﬁcant patterns of cross-linguistic variation. The expression of combination, contrast and alternative is less predictable than it is commonly assumed and the use of the English ‘and-but-or’ system is widespread in Central-Western Europe, but it is rather rare across the world’s languages. The attested variation is not random, but is characterized by a set of implicational patterns, which are mainly determined by the interaction of economic principles with the degree of semantic speciﬁcity of the relation expressed. The focus on Europe and the comparison between European and nonEuropean data show that also within Europe the coding of coordination exhibits some cross-linguistic variation. However, this variation mainly concerns the semantic domains of the attested constructions, since the morphosyntactic properties of coordinating constructions attested in Europe show a high degree of homogeneity. Not surprisingly, the only languages having different morphosyntactic features are located at the very Eastern periphery of Europe, in Turkey and in the Caucasus, two areas whose inclusion within Europe is itself controversial (cf. section 1.2). The examination of the cross-linguistic variation attested in Europe identiﬁes smaller areas of languages showing coordinating markers with the same semantic domains. In particular, there is an area in Central-Western Europe where languages express combination, contrast and alternative by means of the same coding system, which is characterized by a general marker for temporal and atemporal combination, a general marker for corrective and counterexpectative contrast, a general marker for simple and choice-aimed alternative and no general marker for discontinuous atemporal combination and opposition. These languages have been named ‘And-But-Or’ languages. The geographical distribution of the attested semantic domains and the identiﬁcation of a restricted area with the ‘And-But-Or’ languages still need historical explanations. For now, they constitute a synchronic picture of a non-random dislocation of different coding patterns within Europe. Further diachronic research is needed, in order to highlight the various factors which may have led to the actual conﬁguration.

Conclusion and prospects

297

There are at least two directions along which this research could be continued. First, the attested implicational patterns may be veriﬁed on a typologically representative language sample, that is, on a balanced variety sample. If the number of non-European languages were increased in accordance to the requirements for a representative sample, this research could be improved in two ways. On the one hand, the attested constraints on cross-linguistic variation might be better supported and the generalizations might be more powerful. Moreover, new coding patterns may arise, which have not been found in this research due to the limited number of languages examined. On the other hand, the comparison of European data with a broader non-European sample may identify the internal homogeneity of European languages with greater evidence. The second direction in which this study may be continued is diachronic. As pointed out at various points of this work, the attested grammaticalization processes that give rise to coordinating markers seem to conﬁrm the coding patterns identiﬁed synchronically. In particular, they mirror the twofold conceptual organization of coordination along two axes of increasing semantic speciﬁcity. Polymorphemic markers coding contrast and alternative often consist of a combination marker with some further semantic adjunct specifying the conﬂicting or the alternative nature of the relation. Moreover, the development of markers coding alternative from hypothetical or interrogative constructions and the widespread use of switch-reference markers in the expression of contrast mirror the two dimensions of irreality and discontinuity respectively, along which alternative and contrast specify combination. A comparative diachronic exam of the evolution of coordinating markers would also be crucial to account for the actual geographical conﬁguration within Europe. Besides the possible inﬂuence of common genetic ancestors and local contact phenomena, the most likely motivation underlying the presence of general or dedicated markers is the development of these markers along individual grammaticalization patterns. In a comparative diachronic analysis, these patterns may then turn out to be recurrent across languages, because similar initial constructions have a high probability to develop along similar grammaticalization processes.

Appendix A – The languages in the samples

T HE E UROPEAN SAMPLE The genetic afﬁliation of the European languages is based on the Web site www.ethnologue.com. The classiﬁcation of Chechen, Dargi (Dargwa) and Lezgian is based on the World Atlas of Language Structures (WALS). Table 48: Languages of the European sample AFRO-ASIATIC(1)

Semitic(1)

Southern

36.Maltese

(Mlt)

ALTAIC(1)

Turkic(1)

Southern

31.Turkish

(Trk)

INDO-EUROPEAN(27)

Celtic(1) Romance(7)

Goidelic Western

1.Irish 2.Catalan 3.French 4.Portuguese 5.Spanish 6.Italian 7.Romanian 8.Sardinian 9.Danish 10.Swedish 11.Icelandic 12.Norwegian 13.Dutch 14.English 15.German 16.Luxembourgish 17.Czech 18.Polish 19.Belorussian 20.Russian 21.Ukrainian 22.Bulgarian 23.SerboCroatian 24.Lithuanian 25.Latvian

(Irl) (Cat) (Fr) (Prt) (Spn) (Itl) (Rum) (Srd) (Dan) (Swe) (Ice) (Nor) (Dut) (Eng) (Grm) (Lux)

Eastern

Germanic(8)

Northern

Western

Slavic(7)

Western Eastern

Southern

Baltic(2)

Eastern

(Czh) (Pol) (Bel) (Rus) (Ukr) (Blg) (SCr) (Lit) (Lat)

Appendix A – The languages in the samples

299

Table 48: (Continued) Isolates(2)

26.Greek 27.Albanian

(Grk) (Alb)

ISOLATE

Basque(1)

37.Basque

(Bsq)

KARTVELIAN

Georgian(1)

35.Georgian

(Grg)

NAKHDAGHESTANIAN(3)

Nakh(1)

32.Chechen

(Che)

Daghestanian(1) Lak-Dargwa Daghestanian(1) Lezgic

33.Dargi 34.Lezgian

(Drg) (Lzg)

Finnic(2)

28.Finnish 29.Estonian 30.Hungarian

(Fin) (Est) (Hng)

URALIC(3)

FinnoUgric(1)

Ugric

300 Appendix

T HE C OMPARISON SAMPLE The genetic afﬁliation of the non-European languages is based on the Web site www.ethnologue.com. Table 49: Languages of the Comparison sample AFRO-ASIATIC(7)

Semitic(2)

Chadic(2) Cushitic(3)

AUSTRALIAN(1)

Gunwingguan(1)

AUSTRO-ASIATIC(1)

Viet-Muong(1)

AUSTRONESIAN(4)

Malayo Polynesian(4)

Arabic Canaanit Western Biumandare Southern Eastern

Mangarayic

3.Modern Standard Arabic 4.Hebrew 5.Hausa 6.Hdi 7.Iraqw 8.Harar Oromo 9.Somali 28.Mangarayi 22.Vietnamese

Polynesian

23.Maori

Muna-Buton

24.Rapanui 25.Tuvaluan 26.Tukang Besi

CHAPACURA(1)

Wanham(1)

Madeira

32.Wari’

CREOLE(1)

English based(1)

Suriname

35.Ndyuka

DRAVIDIC(1)

TamilKannada(1)

Tamil

16.Malayalam

ESKIMO(1)

Aleut(1)

Inuit

21.W. landic

HOKAN(1)

Yuman(1)

Delta-California River-Yuman

36.Jamul Tiipay 37.Maricopa

INDO-EUROPEAN(2)

Indo-Iranian(2)

Iranian Indo-Aryan

1.Persian 2.Marathi

ISOLATES(2)

15.Korean

Green-

Appendix A – The languages in the samples

301

Table 49: (Continued) 14.Japanese MOSETENAN(1)

31.Mosetén

NADENE(1)

Nuclear(1)

Athapaskan

27.Kuskokwim

NIGER-CONGO(4)

Volta Congo(3)

Kwa Gur

Atlantic(1)

Southern

10. NànáfwˆE 11.Supyire 12.Koromfe 13.Kisi

NILO-SAHARAN(1)

Nilotic(1)

Southern

34.Lango

SINO-TIBETAN(4)

TibetoBurman(4) Kuki-Chin Meithei Kiranti

17.Hakha Lai 18.Meithei 19.Dumi 20.Limbu

Winnebago

29.Hocak ˛

Brahman

30.Tauya

SIOUAN(1) TRANS NewGUINEA(1) YUKAGHIR(1)

Mississippi(1)

33.Kolyma

Appendix B – Sources of information on the languages in the samples

T HE E UROPEAN SAMPLE Table 50: Sources for the European sample. The number of questionnaires collected for each language is reported in brackets. Where no brackets are present no questionnaires have been collected

Language

Questionnaires

Bibliographical ences

Albanian (1) Basque (1) Belorussian Bulgarian (1) Catalan (1) Chechen Czech (1) Danish (2) Dargi Dutch (1) English (1) Estonian (1) Finnish (2) French (4)

L.S. A.G.U.

Bucholz and Fiedler 1987

Georgian (1) German (2) Greek (2) Hungarian (2) Icelandic (1) Irish (1) Italian (5)

refer-

Saltarelli 1988 Mayo 1993

K.K. N.F

Feuillet 1996, Scatton 1993 Hualde 1992 Jeschull 2004

T.B. P.B; F.G.M

Short 1993 Allan et al. 1995 Van den Berg 2004

K.S. G.H. M.T. S.O.; J.Y. Z.B; M.V; A.B.; O.S. M.T.

Donaldson 1997

C.W; K.Z. A.G; G.C. A.H.; F.M. A.H. G.H. P.B.; M.B.; S.P.; A.C.; F.M.

Drosdowski 2005

Huddlestone 1988 Harms 1962 Karlsson 1987 Arrivé et al. 1986 Hewitt 1996, Aronson 1989, Hillery 1996 Holton et al. 1997 Kenesei et al. 1998 , Bánréti 1994 Maling and Zaenen 1990 Mcgonagle 1988 Scorretti 1988

Appendix B – Sources of information on the languages in the samples

303

Table 50: (Continued)

Language

Questionnaires

Bibliographical ences

Latvian (1) Lezgian Lithuanian (3) Luxembourgish (1) Maltese (1)

I.K.W.

Dini 1997

Norwegian (1) Polish (2) Portuguese (2) Romanian (3) Russian (4) Sardinian (1) Serbo-Croatian (1) Spanish (3) Swedish (1) Turkish Ukrainian

refer-

Haspelmath 1993

L.R.; J.P.; V.Ž. C.P. E.S.

Ambrazas 1997, Dini 1997

J.O. A.L.; W.B. L.M.; A.R. A.D.; D.G.; C.R. E.T; N.M. ; K.K. ; K.R. I.F. M.C.

Bratveit 1990

A.O.; M.D.; L.J. A.B.

Sarmiento 1989

Schanen 2004 Borg and Azzopardi-Alexander 1997 Fisiak et al. 1978, Rothstein 1993 Hutchinson 1996 Drago¸s 1995 Comtet 1997, Timberlake 1993 Corda 1994 Brown 1993, Brown and Alt 2004 Holmes and Hinchliffe 1997 Kornﬁlt 1997, Johanson 1995 Shevelov 1993

304 Appendix

T HE C OMPARISON SAMPLE Table 51: Sources for the Comparison sample. The number of questionnaires collected for each language is reported in brackets. Where no brackets are present no questionnaires have been collected

Language

Questionnaires

Bibliographical ences

Arabic (1) Dumi Hakha Lai Harar Oromo Hausa (1) Hdi Hebrew (1) Hocak ˛ Iraqw Jamul Tiipay Japanese (1)

N.R.

Caspari 1955, Clive 2004

Kisi Kolyma Yukaghir Korean (1) Koromfe Lango Limbu Malayalam Mangarayi Maori Marathi Maricopa Meithei Mosetén NànáfwˆE (1) Ndyuka Persian (1)

refer-

Van Driem 1993 Peterson and VanBik 2004 Owens 1997

S.M.

Smirnova 1982 Frajzyngier and Shay 2002

H.M.

Glinert 1989 Helmbrecht in prep. Mous 2004 Miller 2001

Y.N.

Hinds 1986, Alpatov and Podlesskaya 1995, Nishina 2004 Tucker Childs 1995 Maslova 2003

Y.M.S.

Sohn 1994 Rennison 1997 Noonan 1992 Van Driem 1987 Asher and Kumari 1997 Merlan 1982 Bauer 1993 Pandharipande 1997 Gordon 1986 Chelliah 1997 Sakel 2004

A.B.

Bohoussou 2008 Huttar and Huttar 1994

A.A.

Stilo 2004

Appendix B – Sources of information on the languages in the samples

305

Table 51: (Continued)

Language Rapanui Somali Supyire Tauya Tukang Besi Tuvaluan Upper Kuskokwim Athabaskan Vietnamese Wari’ West Greenlandic

Questionnaires

Bibliographical ences De Feu 1996 Saeed 1993 Carlson 1994 Mac Donald 1990 Donohue 1993 Besnier 2000 Kibrik 2004 ˜ NguyênÐ ình Hoà 1997 Everett and Kern 1997 Fortescue 1984

refer-

Appendix C – Geographical location of the languages in the samples

T HE E UROPEAN SAMPLE In Fig.16 the various European languages are referred to with the abbreviations listed in Table 48.

T HE C OMPARISON SAMPLE In Fig.17 the various languages are referred to with the numbers listed in Table 49.

86DA4 %06<3701 0? 3=8 89<1;5<;82

??4=38G J 4>6A0?7820; ;>20C8>= >5 C74 ;0=6D064B 8= C74 B0
Figure 17: Location of the examined non-European languages

308 Appendix

% & #!" ! $ ! +01;4 -=8 F58237011<748 528A 71 3=72 4828<46= 3=8 1;>72= D842701

G=<3 " 04A848A 72 =>C C40 1DC 2>5544

H05 6<1 40C >A 3A8=: <1I3=71; I05 @<13 -=<3 B<1 8B B78=6 0=3 B8=68=6 01 3=8 :784 0< F>A:8=6 0=3 '4C4A 8B B;44?8=6 H05 6<1 ;40E4 =>F 0=3 2><4 102: C78B 05C4A=>>= 740A3 0 C7D=34A 0=3 8C BC0AC43 C> A08= > F4 6> C> B27>>; C><>AA>F >A 3> F4 BC0H 0C 7><4 0< E4AH C78ABCH 1DC 3>=C ;8:4 >A0=64 9D824 '4C4A 8B =>C BCD3H8=6 8= 78B A>>< 1DC 74 8B ?;0H8=6 8= C74 60A34= *><4C8<4B 0 3>>A >?4=43 >A 0 F8=3>F B;0<<43 ;45C )><4 0=3 F8;; <>E4 C> -4=824 $0=H BCD34=CB BCD3H 3DA8=6 C74 30H 0=3 F>A: 8= C74 4E4=8=6 ,BD0;;H FA8C4 >A A403 D=C8; ;0C4 +74H ?0H 0=3 F4 40C +74 E0B4 54;; 3>F= 0=3 8C 383=C 1A40: G05>A I05 >7C8 20B8 C40 >A 2>5544 H82 :>8<28

310 Appendix

17. Mary bought a postcard and John a poster 18. Usually, Mary prepares and John eats the cake 19. I am not sure: Mary has just found a dog, or Jane a cat 20. Last night I heard a noise, probably he slammed or she broke the window 21. I should go to the Post Ofﬁce and then I’ll reach you. 22. I am tired, I don’t want to study this afternoon. . . Would you go for a walk with me, or have a cup of tea, or watch a movie, or. . . whatever? 23. Last night, I worked and Peter slept 24. I would love to go to the theater, but I have to stay at home tonight 25. My leaving and coming back after two days confused her. 26. I arrived in Cambridge and a friend saw me. 27. My parents arrived yesterday and my friends will arrive tomorrow.

Appendix E – The constructions examined in the study

Tables 53 and 54 show all the coordinating constructions examined in this study. The aim of the tables is not to make a list of all strategies available in the sample languages, but rather to illustrate the coordinating constructions that are attested in the questionnaires and in the descriptive grammars analyzed. It may thus be possible that existing coordinating constructions are not reported in the tables, as a consequence of their not being attested in the sources. The following notation and conventions are used in the tables: - When overt markers of coordination are present, they are explicitly indicated; - if a language has more than one possible construction to convey the same semantic relation, the attested constructions will be indicated one under the other; - when a structure of the type ‘B X-ra B’ is used, it means that the overt morpheme -ra is sufﬁxed to the ﬁrst constituent of the second clause; - when a structure of the type ‘Bint ’ or ‘Birr ’ is used, it means that the clause shows interrogative or generally irrealis markers, respectively; - B: SoA coded by a balanced verb form; - D: SoA coded by a deranked verb form; - switch-r: switch-reference marker which may be only used to link coordinated SoAs; - blank: no information available.

B apo B BaB

B edo B B ala B

Albanian

Basque

BoB

Catalan

Chechen

B ili B

B ili B

Bulgarian

B ja B

BoB

B abĲo B

Belorussian B ci B

B edo B

B ose B BoB

ALTERNATIVE c-aimed simple

Language

The European sample

Table 53:

T HE E UROPEAN SAMPLE

BB X’a BX’a B

BiB

BiB B ta/ da B

BiB

B eta B

BeB B dhe B

BB B X’a B X’a

BiB

BiB

BiB

B eta B

BeB B dhe B

BB B X’a B B tq’a B X’a

BiB

BiB BaB

BiB BaB

B eta B

BeB B dhe B

COMBINATION temporal atemporal sequential simult.

B tq’a B

BiB

BaB

BaB

B eta B

BeB B dhe B B B kurse B

oppositive

B tq’a B

B sino B

BaB

B alĲe B

B baizik B

B por B

CONTRAST corrective

B amma B

B pero B

B no B

B alĲe B

B baina B

B por B

counterexp.

312 Appendix

B eller B

B yara B

Danish

Dargi

B of B

B or B

B või B

B vai B

B ou B

Dutch

English

Estonian

Finnish

French

B aèi B

B nebo B B ci ˇ B

Czech

B ou B

B tai B

B või B

B or B

B of B

yara B yara B

B eller B

B nebo B B ci ˇ B

ALTERNATIVE c-aimed simple

Language

Table 53: (Continued)

B et B

B ja B

B ja B B ning B

B and B

B en B

B X-ra B

B wa B

B og B

BaB

B et B

B ja B

B ja B B ning B

B and B

B en B

B X-ra B

B wa B

B og B

BaB

B et B

B ja B

B ja B B ning B

B and B

B en B

B X-ra B

B wa B

B og B

BaB

COMBINATION temporal atemporal sequential simult.

B et B B tandis que B

B ja B

B ja B

B and B B while B

B en B

B og B

BaB

oppositive

B mais B

B vaan B

B vaid B

B but B

B maar B

B men B

B ale B

CONTRAST corrective

B mais B

B mutta B

B aga B B kuid B

B but B

B maar B

B amma B

B-gu B

B men B

B ale B

counterexp.

Appendix E – The constructions examined in the study

313

B tu B

B oder B

BiB

B vagy B

B eða B

B nó B

BoB B oppure B

Georgian

German

Greek

Hungarian

Icelandic

Irish

Italian

BoB B oppure B

B nó B

B eða B

B vagy B

BiB

B oder B

B an B

ALTERNATIVE c-aimed simple

Language

Table 53: (Continued)

BeB

B agus B

B og B

B és B

B kai B

B und B

B da B

BeB

B agus B

B og B

B és B

B kai B

B und B

B da B

BeB

B agus B

B og B

B és B B meg B

B kai B

B und B

B x k’i B

B da B

COMBINATION temporal atemporal sequential simult.

BeB B mentre B

B agus B

B og B

B és B B Xmeg B B Xpedig B

B kai B

B während B

B und B

B x k’i B

oppositive

B ma B B bensì B

B ach B

B heldur B

B hanem B

B alla B

B sondern B

B aramed B

BB

CONTRAST corrective

B ma B B però B

B ach B

B en B

B de B

B alla B

B aber B

B magram B BB

counterexp.

314 Appendix

B vai B

B taˆxajt’a B

B ar B B arbà B

B oder B

B jew B

B eller B

B czy B

Latvian

Lezgian

Lithuanian

Luxemb.

Maltese

Norwegian

Polish

B sau B B ori B

Romanian

B sau B B ori B

B ou B

Portuguese B ou B

B lub B B albo B

B eller B

jew B jew B

B oder B

B ar B B arbà B

ja B ja B B wa ja B

B vai B

ALTERNATIVE c-aimed simple

Language

Table 53: (Continued)

B s¸i B

BeB

BiB

B og B

BuB

B ann B

B ir B

B x-ni B B wa B D-na B

B un B

B s¸i B

BeB

BiB

B og B

BuB

B ann B

B ir B

B x-ni B B wa B

B un B

B s¸i B B iar B

BeB

BiB BaB

B og B

BuB

B ann B

B ir B BoB

B x-ni B B wa B

B un B

COMBINATION temporal atemporal sequential simult.

B iar B

BaB

B og B

BuB

B ann B

BoB

B un B

oppositive

B ci B B dar B

B mas B

B ale B

B men B

B imma B

B mee B

BoB

B bet B

CONTRAST corrective

B dar B B însˇa B

B mas B

B ale B

B men B

B imma B

B mee B

B bèt B

B amma B

B bet B

counterexp.

Appendix E – The constructions examined in the study

315

B ili B

BoB

B eller B

SerboCroatian

Spanish

Swedish

Ukrainian

B cˇ y B

BoB

Sardinian

Turkish

B ili B

Russian

B abĲo B

B veya B ya B ya B

B eller B

BoB

B ili B

BoB

B ili B

ALTERNATIVE c-aimed simple

Language

Table 53: (Continued)

BiB

BB B X=dA B B ve B D-Ip B

B och B

ByB

B pa B

BiB

BeB

BiB

BiB

BB B X=dA B B ve B

B och B

ByB

BiB

BeB

BiB

BiB B ta B

BB B X=dA B B ve B

B och B

ByB

BaB

BiB

BeB

BiB BaB

COMBINATION temporal atemporal sequential simult.

B ta B

B och B

ByB

BaB

BeB

BaB

oppositive

B alĲe B

BB B fakat B

B utan B

B sino B

B v‚ec´ B

B n‚ego B

B ma B

BaB

CONTRAST corrective

B alĲe B

B ama B B fakat B

B men B

B pero B

B ali B

B ma B

B no B

counterexp.

316 Appendix

B am B B aw B

B ye B ye

Bint Bint

Arabic

Dumi

Hakha Lai

BB B lá B

B á nà B

Hdi

B á nà B

BB B wa B B kuma B

B kó B B kokuma B

Hausa

B kó B B kokuma B

D-é B

B =ĳií B

B-k@ B BB

B wa B B fa B

BB

BB B wa B B kuma B

BﬁB

B =ĳií B

B(-k@) B BB

B wa B

BB

BB B wa B B kuma B

BﬁB

B =ĳií B

B(-k@) B BB

B wa B

COMBINATION temporal atemporal sequential simul.

Harar Oromo

B-làw=leè B

B ye B ye

B aw B

ALTERNATIVE c-aimed simple

Language

Table 54: The Comparison sample

T HE C OMPARISON SAMPLE

BB

BB B wa B B kuma B

BB

B wa B B ﬁè¯i B

oppositive

BB B àmmaa B

BB

B bal B

CONTRAST corrective

B àmá B

B àmmaa B

B k’ófáamô B

B immó B

B laakin B

counterexp.

Appendix E – The constructions examined in the study

317

BoB B o she-B

B ni˛˛igéšge B

B laqáa B

Hebrew

Hocak ˛

Iraqw

B-ka B(-ka)

Bint Bint Bint soretomo Bint

B bàà B

Bint Bint

Bint Bint

Japanese

Kisi

Kolyma Yukaghir

Korean

B hokun B D-kena B

B bàà B

D nyamaaw Birr

Jamul Tiipay

B laqáa B

B ni˛˛igéšge B

BoB B o she-B

ALTERNATIVE c-aimed simple

Language

Table 54: (Continued)

B kuliko B D-ko B

DB

BB

D-te B B si B

D-switch-r B

B nee B

B ánaga ˛ B

B ve-B

B kuliko B D-ko B

DB

BB

D-te B B si B

D-switch-r B

B nee B

B ánaga ˛ B

B ve-B

B kuliko B D-ko B

DB

BB

B si B

D-te B D-i B

D-switch-r B

B nee B

B ánaga ˛ B

B ve-B

COMBINATION temporal atemporal sequential simul.

DB

BB

D-te B D-i B

D-switch-r B

B ve-B

oppositive

DB

BB

D-te B

D-switch-r B

B nu˛ni˛ge B

B éla B

CONTRAST corrective

B kulena B B-man(un) B

DB

B k´E (mi) B

B-ga B

B-pes B

B nu˛ni˛ge B

B aval B

counterexp.

318 Appendix

B raanei B

B k¯ı B

Maori

Marathi

B k˜ımw¯a B

B raanei B BB

Birr Birr

Bint Bint

Mangarayi

B òñò B

B maa B

B alleNkil B D-oo D-oo (+dummy verb)

Bint Bint

Bint Bint

verb)

D-kena Dkena (+dummy

ALTERNATIVE c-aimed simple

Malayalam Bint Bint

Limbu

Lango

Koromfe

Language

Table 54: (Continued)

BB

BB B aa B

BB

BB D-um D-um (+dummy verb)

BB B-an B

BB B tˆE D

BB

BB

BB B hoki B

BB

BB D-um D-um (+dummy verb)

BB B B-an B-an B-an

BB

BB

BB

BB B hoki B

BB

BB D-um D-um (+dummy verb)

BB B B-an B-an B-an

BB

BB B la B

COMBINATION temporal atemporal sequential simul.

BB

BB

BB

BB

B la B

oppositive

B pan. B

BB B engari B

BB

BB

BB

B la B

CONTRAST corrective

B pan.B

B engari B

B gana B

eNkilum

B ennaal B B pakùe B

BB B kere B

BB B E´ ntˆO B

B la B

D-(u)na B

counterexp.

Appendix E – The constructions examined in the study

319

B yâ B

B yâ B

B va B

BB B da B

Persian

B efu B

B efu B

Ndyuka

B

B kp´Ekún B

wj´Elj´E wj´Elj´E B

B ánzˆE B

BB B aduga B

B va B

B va B

BB B da B

B kp´Ekún B

B kp´Ekún B BB B da B

BB

BB B jö’dyë’yä B

BB B aduga B

verb)

DB D D (+dummy

B a¯ n.i/ wa B B wa B

BB

BB B jö’dyë’yä B

BB B aduga B

verb)

verb)

NànáfwˆE

BB B nattrega B

DB D D (+dummy

B a¯ n.i/ wa B B wa B

DB D D (+dummy

B a¯ n.i/ wa B B wa B DB

COMBINATION temporal atemporal sequential simul.

BB

Bint Bint

B athw¯a B

ALTERNATIVE c-aimed simple

Mosetén

Meithei

Maricopa

Language

Table 54: (Continued)

BB

BB

B x-ki B

oppositive

BB

BB

B sàng´E B

B parãntu B

CONTRAST corrective

´ B B vœli

B ma B

B sàng´E B

Bx-tsa’ B

B adupu B

B parãntu B

counterexp.

320 Appendix

BB BoB

B misé B

B làa B

Bint pe Bint

B tawa B

B io me/pe B B me/pe B

Somali

Supyire

Tauya

Tukang Besi

Tuvaluan

pe

B io me/pe B B me/pe B

B tawa B

Bdub Bdub

B yô B yô

B ama B

BB BoB

BB

B kene B

B maka B BB

BB

D-switch-r B

B sì B

B oo B B x-na B

BB BeB

BB

D-switch-r B

B kà/ mà D

B oo B B x-na B

BB BeB

B =ò (Vafsi)

B

B kae B

BB

B kene B

BB

D-switch-r B

B sì B

B oo B B x-na B

BB BeB

B =ò (Vafsi)

B =ò (Vafsi)

Bint Bint

B

COMBINATION temporal atemporal sequential simul.

ALTERNATIVE c-aimed simple

Rapanui

Language

Table 54: (Continued)

B

B kae B

BB

D-switch-r B

B sì B

BB

B va B

oppositive

B kae B

B toka B

D-switch-r B

B sì B

B kae B

B toka B

D-switch-r B

B sì B B ï` kàà B

B x-se B B laakìin B

BB B pero B

´ B œmma B

´ B bœlke B

BB

counterexp.

CONTRAST corrective

Appendix E – The constructions examined in the study

321

very rare: B h0nek’è B

B ho˘a. B

Birr Birr

B imaluunniit B

B D=luunniit

Vietnamese B hay(là) B

Bint Bint

B imaluunniit B

B D=luunniit

Wari’

West Greenlandic

ALTERNATIVE c-aimed simple

none Upper Kuskokwim Athabaskan

Language

Table 54: (Continued)

B D=lu D(=lu) B

DB

B x=lu B

BB B B ca’ na

BB B và B

B ts’eĳ B

B D=lu

B x=lu B

BB B B ca’ na

BB B và B

B ts’eĳ B

D(=lu) B

B D=lu

B x=lu B

BB B B ca’ na

BB B và B

B ts’eĳ B

COMBINATION temporal atemporal sequential simul.

BB

B còn B

oppositive

B D=li

B x=li B

BB

B

B kisianni(li) B

B D=li

B x=li

BB

B mà B B nhung B

B ĳedinh B

B ĳedinh B

B mà B

counterexp.

CONTRAST corrective

322 Appendix

Notes

1. In Japanese -i is usually considered a converb (see Haspelmath and König 1995) and belongs to the verbal paradigm. The gloss COORD is thus a functional gloss, which refers to the semantic function of combination (see section 3.1.2) expressed here by -i. 2. The very notion of representative language sample and the criteria for selecting languages in a balanced way are still a matter of dispute in typology. A detailed discussion on the main approaches to language sampling can be found in Cysouw 2005. 3. Among the studies that have identiﬁed interesting geographic distributions of particular phenomena in European languages, I will recall here Kortmann’s work on the subordinators attested in the European languages (Kortmann 1997) and Bernini and Ramat’s study on the negative sentences in the European area (Bernini and Ramat 1996). 4. This work was part of the FIRB (Fondo per gli Investimenti della Ricerca di Base) project ‘Europa e Mediterraneo dal punto di vista linguistico: storia e prospettive’, coordinated by Paolo Ramat. 5. ‘Les conjonctions sont surtout utiles dans le discours solennel ou rituel, . . . ; elles sont indispensables dans la langue écrite qui, sans conjonctions, devient aisément inintelligible. De plus, le language solennel et la langue écrite recourent volontiers à des phrases composées de plusieurs membres pour exprimer une pensée complexe et nuancée’ (Meillet 1958b: 174). [The conjunctions are particularly useful in formal or ritual discourse, . . . ; they are indispensable in the written language, which would easily become unintelligible without conjunctions. Moreover, formal and written language prefer the use of compound sentences with several members for the expression of a complex idea full of nuances]. 6. As Lehmann (1988) and Haspelmath (2007) point out, the term subordination is traditionally used to refer to the clause level, whereas the more general label for syntactic asymmetric constructions is dependency. 7. In the course of this section, the terms ‘coordination, coordinate’ and ‘subordination, subordinate’ will be used to refer to morphosyntactic phenomena, accordingly with the terminology adopted in the traditional approaches described here. Throughout the rest of the book, however, the two terms will be used with reference to the functional level, accordingly with the functional deﬁnition of coordination that will be proposed in section 1.3.3. 8. For a discussion on the problematic concept of syntactic category in a crosslinguistic perspective see Ramat 1999, Croft 1991 and Dryer 1996.

324 Notes 9. Lehmann (1988: 181-182) distinguishes between relations of dependency and relations of sociation (non-dependency), restricting the terms subordination and parataxis to the relations of dependency and sociation between clauses, respectively. 10. As discussed by Croft (1996), the notion of ‘head’ can be deﬁned in semantic terms, as the entity which is speciﬁed and described by the dependent (see also Croft 2001). 11. This holds for adverbial subordinate clauses, but is not true for other types of subordinate clauses, like completive or relative clauses (cf. Cristofaro 2003). 12. Haiman (1985: 101) uses the term ‘prominence’ but does not give a clear definition of what it means. In this discussion, the word ‘prominence’ is repeated in order to reproduce Haiman’s arguments and is meant to denote a position of recognized importance both within the conceptual organization of the sentence and within the discourse. 13. Prandi (2004: 283-285) makes a crucial distinction between inherently complex processes and the linkage of independent ones. Considering two complex sentences like ‘I hope that it will rain’ and ‘The ﬁelds are green because it rained heavily’, he argues that the two SoAs linked in the former constitute an inherently complex event, since the act of hoping requires an object of hope in order to be complete, and this consists of the SoA ‘it will rain’. On the contrary, the two SoAs in the latter are independent, in that each one is complete on its own, and the causal relation establishes a link between them (Prandi 2004: 285). 14. In Boolean logic two types of disjunction are distinguished (Allwood, Andersson, and Dahl 1977, Ohori 2004, Dik 1968). The ﬁrst, called inclusive disjunction, is true iff, given two propositions p and q, at least one of them is true. The second type of disjunction is labeled exclusive. Exclusive disjunction is true iff only one of the disjoined propositions is true, and therefore iff the truth of the one excludes the truth of the other. 15. Dik (1968: 274-277) and Lakoff (1971: 142) have thoroughly and convincingly argued that there are two major problems with projecting the boolean distinction into natural language. First, truth-values cannot be assigned to expressions such as questions, wishes and hypothesis, since these cannot be evaluated in terms of their truth. Nonetheless, questions, wishes and hypothesis are frequently combined or contrasted in natural language, either as cooccurring SoAs or as alternatives. Secondly, there is a discrepancy between the semantic distinctions identiﬁed in Boolean logic and those actually coded by natural languages. For instance, the distinction between inclusive and exclusive disjunction appears to be only marginally relevant to natural language. Actually, languages do not seem to encode the distinction between inclusive and exclusive disjunction at all (Dik 1968: 275). Moreover, logic does not account for the differences between combination and contrast, since both are characterized by the truth of both linked propositions. Finally, as will become clear in the course of this

Notes

16.

17.

18.

19.

20.

21.

22.

23. 24.

325

work, different constructions are used in languages to code semantic distinctions which are not identiﬁed within logic, such as sequential and simultaneous combination, simple and choice-aimed alternative (see chapters 3 and 5). The use of the same coordinating strategy for noun phrases and subordinate, probably nonﬁnite, clauses would not be surprising, given that the acquisition of nominal morphology by subordinate verb forms is a widespread phenomenon (cf. Lehmann 1988). The use of the same strategy to code a relation between things and subordinate SoAs could be due to a similar cognitive conceptualization of the two, conﬁrming the hypothesis that subordinate SoAs, being characterized by the absence of an autonomous cognitive proﬁle, may be conceived as atemporal units just like things (cf. Cristofaro 2003: 262-270, Langacker 1987b: 63-72). The exam of ellipsis phenomena in coordinating constructions would rather require the exam of text-corpora from different languages, which would need a speciﬁc sample and therefore a speciﬁc research. Grammaticalization is a process whereby lexical items start to fulﬁll a grammatical function in certain contexts and, once they are grammaticalized, keep developing new grammatical functions (see Hopper and Traugott 2003, Dasher and Traugott 2002 and Mac Mahon 1994 for more detailed discussions). Haspelmath (1993: 376): “The Aorist converb is used to express chains of actions carried out by the same subject. Such sentences often have to be translated by means of coordinate clauses in English, because the action of the converb clause is not backgrounded to the same extent as the English participle would suggest”. Other attested non-coordinating functions of the constructions examined in this study are for instance the various subordinating functions of the converbal forms, the interrogative function of some alternative markers, like czy in Polish, or the subordinative functions of some adversative markers, like mentre in Italian or while in English. This type of multifunctionality will not be taken into account and these constructions will be only analyzed with regard to their coordinating functions. According to Nedjalkov (1995: 97), a converb is a verb form which is syntactically dependent on another verb form, but does not realize its semantic valencies. In particular, a narrative converb is used to link completed actions in a coordination relation which advances the narration (Nedjalkov 1995: 106-110). The notion of continuity is traditionally used in discourse analysis, where markers like and are commonly used as continuation indicating devices, whereas markers like but denote an interruption of continuity (Schiffrin 1987). Throughout this work, the term Arabic will be used referring to the variety of Modern Standard Arabic (cf. Clive 2004). Van Klinken (2000: 354-57) describes a similar pattern of grammaticalization in Tetun, a Central Malayo-Polynesian language. In this language, the verb hodi

326 Notes

25.

26. 27.

28.

29.

30.

’to bring/use’ acquires a coordinating function from an original purposive construction, in which the second clause describes the purpose for bringing an item mentioned in the ﬁrst clause. The original semantics of the verb is different in Hdi and in Tetun, but in both cases the coordinating function arises in purposive contexts implying some motion from the ﬁrst to the second SoA. Exceptions to this rule are not attested in the samples but are possible and would normally be explained in diachronic terms. As Mithun (1988) and Traugott (1986) have pointed out, the possible sources for the grammaticalization of coordinating markers are diverse and may come from narrative contexts, from spatial expressions or iterative meanings. If a given expression enters a grammaticalization process, develops a speciﬁc contrast function on the map and substitutes the old marker in the expression of that function, it may interrupt a previous situation of general undercoding, where all the contrast relations were expressed by means of the same general construction. Speciﬁcally, if the new connective expresses a relation located in the middle of the map, this would cause a discontinuity that apparently contradicts the iconic principle stated above. Yet, if the new marker develops further coordinating functions, its widening would proceed along the map, ﬁrst including the contiguous relations and then reaching the further ones. The simultaneous realization of multiple signs is possible in other semiotic systems, like music or the visual languages. By reality value of a given proposition is meant here the actuality status of the SoA it describes, that is to say whether the SoA is realized or not realized. The term reality value therefore refers to an overarching parameter that may have two values: realis and irrealis. Modality may refer to ‘objective’ circumstances that make the actuation of a SoA necessary or allowed, to the likelihood of a given SoA or to the degree of commitment of the speaker towards the truth of the proposition describing the SoA itself (Cristofaro 2003: 60). Generally speaking, modality can be considered as the level of the speaker’s (or agent’s) attitude or point of view on what is being said (the propositional content) (see Palmer 1986 and Bybee, Perkins, and Pagliuca 1994). This terminology is often used in the description of creole languages, Australian languages, languages of New Guinea and languages of North America, referring to the structural level, namely to particular verbal sufﬁxes or particles which encode the two values of realis and irrealis (see Bybee et al. 1994 and Elliot 2000). In this work, however, the terms ‘realis’ and ‘irrealis’ refer to a semantic distinction. In many languages (such as for instance Hakha Lai, Japanese, Koromfe, Tauya and Wari’), NP disjunction in interrogative sentences is not possible and clausal disjunction is used instead.

Notes

327

31. Future markers are usually considered as irrealis markers. Yet, as Ohori points out, the SoA in (5.38a) is ‘securely believed’ by the speaker. In this case the future SoA is treated as certain and shows some reality features. 32. As widely illustrated in a number of texts, such as Allwood et al. (1977), the meaning of the two basic logic quantiﬁers can be paraphrased as follows: ∃xA(x) means that ‘there is at least one x such that x has the property A’; ∀xA(x) means that ‘for every x it holds that x has the property A’ (cf. Cristofaro 2003: 87). 33. In Chechen ’a cliticizes onto the element immediately to the left of the lexical verb when independent clauses are conjoined, to the subject when the ﬁnite clauses show referential discontinuity, and to the verb itself when ﬁnite subordinate clauses are conjoined (Jeschull 2004: 252-253). 34. The function of Arabic amm¯a is to mark contrastive focus and it roughly means ’whereas’ (Matras 1998: 303).

References Abraham, Werner 1979 But. Studia Linguistica, XXXIII-II: 89–119. Adamíková, Marcela 2005 Das Inventar adversativer Konnektoren in den Slavinen. Polyslav, VIII: 45–58. Allan, Robin, Philip Holmes, and Tom Lundskær-Nielsen 1995 Danish: a comprehensive grammar of the modern language. London; New York: Routledge. Allwood, Jens, Lars-Gunnar Andersson, and Östen Dahl 1977 Logic in linguistics. Cambridge: Cambridge University Press. Alpatov, Vladimir M. and Vera I. Podlesskaya 1995 Converbs in Japanese. In Martin Haspelmath and Ekkehard König, (eds.), Converbs in Cross-linguistic Perspective, pp. 465–485. Berlin; New York: Mouton de Gruyter. Ambrazas, Vytautas 1997 Lithuanian grammar. Vilnius: Baltos Lancos. Anscombre, Jean-Claude and Oswald Ducrot 1977 Deux mais en Français? Lingua, 43: 23–40. Aronson, Howard I. 1989 Georgian: a reading grammar. Bloomington: Slavica. Arrivé, Michel, Françoise Gadet, and Michel Galmiche 1986 La grammaire d’aujourd’hui. Guide alphabetique de linguistique française. Paris: Flammarion. Asher, Ronald E. and T.C. Kumari 1997 Malayalam. London; New York: Routledge. Bánréti, Zoltán 1994 Coordination. In Ferenc Kiefer and Katalin É. Kiss, (eds.), The syntactic structure of Hungarian, 27 in Syntax and Semantics, pp. 355– 414. San Diego: Academic Press. Bauer, Winifred 1993 Maori. Descriptive Grammars. London; New York: Routledge. Bednarczuk, Leszek 1971 Indo-European Parataxis. Krakow: Wydawnictwo Naukowe. Bernini, Giuliano and Paolo Ramat 1996 Negative sentences in the languages of Europe. Berlin; New York: Mouton de Gruyter. Besnier, Niko 2000 Tuvaluan. A Polynesian Language of the Central Paciﬁc. London; New York: Routledge.

)454A4=24B @5?9ACF9 =5B9 5B8 +C69FH 5FGHCB -<9 DF5;A5H=7G C: 0=37CB>IB7H=CBG H<9 BCBB5FF5H=J9 75G9G "B CF=BB9 "H9B 5B8 8 '99@9A5B 98G ,#.'# .>A:8=6 '0?4AB 8= #8=6D8BC82B DD ^ %CB8CB .% @5?9ACF9 =5B9 5B8 +C6MB 5FGHCB

-<9 DF5;A5H=7G C: G9BH9BH=5@ 7CCF8=B5H=CB K=H< 0=3 #8=6D0 ^ C
#4B ?7A0B4B 2>
;5=BGH CB>) #8=6D0 ^ F5HJ9=H $5F@ %>AF4680= 382C8>=0AH %>AF4680==6;8B7 =6;8B7%>AF4680= %CB 8CB '9K 2CF? +CIH@98;9 F9GB5B #C5B

#4G820;5D=2C8>=0; BH=C0G (L:CF8 @57?K9@@ )I6@=G<9FG FCKB 05M@9G ,9F6CFC5H "B 9FB5F8 CAF=9 5B8 F9J=@@9 CF69HH 98G +74 *;0E>=82 #0=6D064B DD ^ %CB8CB '9K 2CF? +CIH@98;9 FCKB 05M@9G 5B8 -<9F9G5 @H

0=31>>: >5 *4A180= >B=80= 0=3 A>0C80= ,%+ I7;DC8>= >5 6A0<<0A +4=B4 0B?42C 0=3 <>30;8CH 8= C74 ;0= 6D064B >5 C74 F>A;3 <=75;C %CB8CB -<9 .B=J9FG=HM C: <=75;C )F9GG 5A57
+74 *CAD2CDA4 >5 >>A38=0C8>= >=9D=2C8>= 0=3 6A44<4=C '74=>< 4=0 8= *?0=8B7 0=3 &C74A #0=6D064B ,HI8=9G =B '5HIF5@ %5B;I5;9 5B8 %=B;I=GH=7 -<9CFM CF8F975 *D?H8A4 9F@=B '9K 2CF? &CIHCB 89 FIMH9F 5GD5F= 5F@ )5I@ 6A0<<0A >5 C74 A0182 ;0=6D064 5A6F=8;9 5A6F=8;9 .B=J9F G=HM )F9GG <9@@=5< ,5 $48C748 9F@=B '9K 2CF? &CIHCB 89 FIMH9F

References

331

Chung, Sandra and Alan Timberlake 1985 Tense, aspect, and mood. In Timothy Shopen, (ed.), Grammatical categories and the lexicon, volume III of Language typology and syntactic description, pp. 202–58. Cambridge: Cambridge University Press. Clive, Holes 2004 Modern Arabic: Structures, Functions, and Varieties. Georgetown: Georgetown University Press. Comrie, Bernard 1981 Language universals and linguistic typology. Oxford: Basil Blackwell. Comtet, Roger 1997 Grammaire du Russe Contemporaine. Toulouse: Presses Universitaires du Mirail. Corda, Francesco 1994 Grammatica moderna del sardo logudorese: con una proposta ortograﬁca, elementi di metrica e un glossario. Cagliari: Edizioni della Torre. Cormack, Annabel and Neil Smith 2005 What is coordination. Lingua, 115: 395–418. Coseriu, Eugenio 1970 Bedeutung und Bezeichnung im Lichte der strukturellen Semantik. In Peter Hartmann and Henry Vernay, (eds.), Sprachwissenschaft und Übersetzen, pp. 104–21. München: Hueber. Cristofaro, Sonia 2003 Subordination. Oxford: Oxford University Press. Cristofaro, Sonia and Paolo Ramat 1999 Introduzione alla tipologia linguistica. Roma: Carocci. Croft, William 1990 Typology and universals. Cambridge: Cambridge University Press. 1991 Syntactic categories and grammatical relations: the cognitive organisation of information. Chicago: University of Chicago Press. 1996 What’s a head? In Laurie Zaring and Johan Rooryck, (eds.), Phrase structure and the lexicon, pp. 35–75. Dordrecht: Kluwer. 2001 Radical Construction Grammar: Syntactic Theory in Typological Perspective. Cambridge: Cambridge University Press. 2003 Typology and universals. Cambridge: Cambridge University Press, 2nd edition. Culicover, Peter and Ray Jackendoff 1997 Semantic Subordination despite Syntactic Coordination. Linguistic Inquiry, 28: 195–217.

)454A4=24B MGCIK &=7<59@

*I5BH=H5H=J9 A9HAIB8 )=CHFCKG?= 98G (D0=C8C0C8E4 #8=6D8BC82B = =C4A=0C8>=0; 0=31>>: DD ^ 9F@=B '9K 2CF? &CIHCB 89 FIMH9F 5<@ OGH9B +4=B4 0=3 B?42C *HBC4
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f @9B5 ;4<4=C4 34 *8=C0G0 BC>A820 )><M=40B20 I7IF9GfH= 8=HIF5 8=857 H=75 Gf= D985;C;=75 FCG8CKG?= IBH<9F

=9 F5AA5H=? IB9BH69=B 8= "CD=0G0 "DC4=08 0=BB=D9; /C=79G C: +ID9FHbG %5B8 -<9 9@7CIFH %97HIF9 89@=J9F98 69:CF9 H<9 .B=J9FG=HM C: &5BHC65 CB 96FI5FM I7FCH (GK5@8 5B8 5F@CG /C;H 9 <068B Q <08B IB9 <MDCH<0=4 ^ @@=CH #9BB=:9F +

+95@=G 5B8 =FF95@=G :CFAG 5B8 7CB79DHG C: H<9 ;F5AA5H=75@=G5H=CB C: F95@=HM #8=6D8BC82 +H?>;>6H ^

References

333

Everett, Dan and Barbara Kern 1997 Wari’. Descriptive Grammars. London; New York: Routledge. Feuillet, Jack 1996 Grammaire synchronique du Bulgare. Paris: Institut Etudes Slaves. Fisiak, Jacek, Maria Lipi´nska-Grzegorek, and Tadeusz Zabrocki 1978 An introductory English-Polish contrastive grammar. Warszawa: Pa´nstwowe Wydawnictwo Naukowe. Foley, William and Robert Van Valin 1984 Functional Syntax and Universal Grammar. Cambridge: Cambridge University Press. Fortescue, Michael 1984 West Greenlandic. London: Croom Helm Descriptive Grammars. Frajzyngier, Zygmunt and Erin Shay 2002 A Grammar of Hdi. Berlin; New York: Mouton de Gruyter. Gil, David 1991 Aristotle goes to Arizona and ﬁnds a language without AND. In Dietmar Zaefferer, (ed.), Semantic Universals and Universal Semantic, pp. 96–130. Berlin: Foris. 2004 Riau Indonesian Sama. Explorations in macrofunctionality. In Martin Haspelmath, (ed.), Coordinating construnctions, pp. 371–424. Amsterdam; Philadelphia: John Benjamins. Givón, Talmy 1979 On understanding grammar. New York: Academic Press. 1990 Syntax: a functional-typological introduction, volume II. Amsterdam; Philadelphia: John Benjamins. 2001 Syntax. An introduction (2◦ edition), volume I e II. Amsterdam; Philadelphia: John Benjamins Publishing Co. Glinert, Lewis 1989 The Grammar of Modern Hebrew. Cambridge: Cambridge University Press. Gordon, Lynn 1986 Maricopa Morphology and Syntax. Berkeley: University of California. Haiman, John 1985 Natural syntax. Cambridge: Cambridge University Press. Halliday, M.A.K. and Ruqaiya Hasan 1976 Cohesion in English. London: Longman. Harms, Robert T. 1962 Estonian grammar. Bloomington: Indiana University. Harries-Delisle, Helga 1978 Coordination reduction. In Joseph Greenberg, (ed.), Universals of Human Language, volume 4: Syntax, pp. 515–83. Stanford: Stanford University Press.

)454A4=24B !5G9;5K5 2C?C -<9 BCBJ57ICIG G9A5BH=7G C: H9@=B?5;9 =B #5D5B9G9 !>DA=0; >5 'A06<0C82B ^ !5GD9@A5H< &5FH=B A0<<0A >5 #4I680= 9F@=B '9K 2CF? &CIHCB 89 FIMH9F -<9 7CBJ9F6 5G 5 7FCGG@=B;I=GH=75@@M J5@=8 75H9;CFM "B &5FH=B !5GD9@A5H< 5B8 ??9<5F8 $ZB=; 98G >=E4A1B 8= A>BB;8=6D8BC82 '4AB?42C8E4 DD ^ 9F@=B '9K 2CF? &CIHCB 89 FIMH9F !CK MCIB; =G ,H5B85F8 J9F5;9 IFCD95B #0=6D064 *284=24B

^ ,D97=5@ "GGI9 _F95@ -MDC@C;M`

-<9 IFCD95B %=B;I=GH=7 F95 ,H5B85F8 J9F5;9 IFCD95B "B &5F H=B !5GD9@A5H< ??9<5F8 $ZB=; 0I@: (9GH9FF9=7<9F 5B8 0C@:;5B; +5=6@9 98G #0=6D064 D=8E4AB0;B 0=3 ;0=6D064 CH?>;>6H 0= 8= C4A=0C8>=0; 70=31>>: DD ^ 9F@=B '9K 2CF? &CIHCB 89 FIMH9F

-<9 9CA9HFM C: F5AA5H=75@ &95B=B; ,9A5BH=7 &5DG 5B8 FCGG %=B;I=GH=7 CAD5F=GCB "B &=7<59@ -CA5G9@@C 98 +74 %4F 'B8 27>;>6H >5 #0=6D064 JC@IA9 DD ^ &5
CCF8=B5H=B; 7CBGHFI7H=CBG 5B CJ9FJ=9K "B &5FH=B !5GD9@A5H< 98 >>A38=0C8=6 2>=BCAD2C8>=B DD ^ AGH9F85A )<=@589@ D<=5 #C5A=BG

'CA=B5@ 5B8 /9F65@ CB>IB7H=CB "B !5GD9@A5H< &5FH=B &5HH<9K , FM9F 5J=8 =@ 5B8 9FB5F8 CAF=9 98G +74 .>A;3 C;0B >5 #0=6D064 *CAD2CDA4B DD ^ (L:CF8 (L:CF8 .B=J9FG=HM )F9GG

CCF8=B5H=CB "B -=ACH<M ,;>6H 0=3 ;8= 6D8BC82 34B2A8?C8>= DD ^ 5A6F=8;9 .) B8 98=H=CB !5GD9@A5H< &5FH=B 5B8 ??9<5F8 $ZB=; 98G >=E4A1B 8= A>BB#8=6D8BC82 '4A?42C8E4 9F@=B '9K 2CF? &CIHCB 89 FIMH9F !9=B9 9FB8 5B8 -5B=5 $IH9J5

.>A;3 ;4G82>= >5 6A0<<0C820;8I0C8>= 5A6F=8;9 5A6F=8;9 .B= J9FG=HM )F9GG !9@A6F975 >2 0=: .8==4106> &, !9K=HH 9CF;9 4>A680= 0 ;40A=4AB 6A0<<0A %CB8CB '9K 2CF? +CIH@98;9 !=@@9FM ) # +74 4>A680= #0=6D064 = >DC;8=4 6A0<<0C820; BD<<0AH
KKK 5FA5N= 7CA ;9CF;=5B !=B8G #C
References

335

Holmes, Philip and Ian Hinchliffe 1997 Swedish: an essential grammar. London; New York: Routledge. Holton, David, Peter Mackridge, and Irene Philippaki-Warburton 1997 Greek: a comprehensive grammar of the modern language. London; New York: Routledge. Hopper, Paul J. and Elizabeth Closs Traugott 2003 Grammaticalization (2nd edition). Cambridge: Cambridge University Press. Hualde, José Ignacio 1992 Catalan. Descriptive Grammars. London; New York: Routledge. Huddlestone, Rodney 1988 English grammar: an outline. Cambridge: Cambridge University Press. Hutchinson, Amelia P. 1996 Portuguese: an essential grammar. London; New York: Routledge. Huttar, George L. and Mary L. Huttar 1994 Ndyuka. London; New York: Routledge. Jeschull, Liane 2004 Coordination in Chechen. In Martin Haspelmath, (ed.), Coordinating construnctions, pp. 241–265. Amsterdam; Philadelphia: John Benjamins. Johannessen, Janne Bondi 1998 Coordination. Oxford: Oxford University Press. Johanson, Lars 1995 On Turkic converb clauses. In Martin Haspelmath and Ekkehard König, (eds.), Converbs in Cross-linguistic Perspective, pp. 313–347. Berlin; New York: Mouton de Gruyter. Karlsson, Fred 1987 Finnish Grammar. Porvoo; Helsinki; Juva: Werner Soderstrom Osakeyhtio. Kenesei, Istvan, Robert M. Vago, and Anna Fenyvesi 1998 Hungarian. Descriptive Grammars. London; New York: Routledge. Kibrik, Andrej A. 2004 Coordination in Upper Kuskokwim Athabaskan. In Martin Haspelmath, (ed.), Coordinating construnctions, pp. 537–554. Amsterdam; Philadelphia: John Benjamins. König, Ekkehard and Peter Siemund 2007 Speech act distinctions in grammar. In Timothy Shopen, (ed.), Language typology and linguistic description, pp. 276–324. Cambridge: CUP, 2nd edition. Kornﬁlt, Jaklin 1997 Turkish. Descriptive Grammars. London; New York: Routledge.

)454A4=24B $CFHA5BB 9FB8 3E4A180; BD1>A38=0C8>= CH?>;>6H 0=3 78BC>AH >5 03E4A180; BD1>A 38=0C>AB 10B43 >= DA>?40= ;0=6D064B 9F@=B '9K 2CF? &CIHCB 89 FIMH9F $CH7<9J5 $F=GH=B5

= 5G 5 A5F?9F C: 8=G7CBH=BI=HM 9J=89B79 C: FIB=7 =BG7F=DH=CBG $* 9F@=B %5?C:: +C6=B ":bG B8bG 5B8 IHbG 56CIH 7CB>IB7H=CB "B <5F@9G # =@@ACF9 5B8 -9FF9B79 %5;9B8C9B 98G *CD384B 8= #8=6D8BC82 *4<0=C82B DD ^ '9K 2CF? !C@H +=B9<5FH 5B8 0=BGHCB "B7 %5A6F97A<0C8>= BCAD2CDA4 0=3 B4=C4=24 5>A< 5A6F=8;9 ,HI8=9G =B %=B ;I=GH=7G 5A6F=8;9 5A6F=8;9 .B=J9FG=HM )F9GG %5B; K5@8 +74 B4<0=C82B >5 2>>A38=0C8>= AGH9F85A )<=@589@D<=5 #C5A=BG

8J9FG5H=J9 7CBB97HCFG CB 8=GH=B7H @9J9@G C: 8=G7CIFG9 5 F9 9L5A=B5H=CB C: J9 ,K99HG9FbG H=38C8>= >=24BB8>= >=CA0BC DD ^ 9F@=B '9K 2CF? &CIHCB 89 FIMH9F

=9 0CFH5FH a$CB>IB?H=CBb "B @5B FIG9 F5BN !IB8GBIFG7<9F &=7<59@ #C6 5B8 )9H9F +C@: %IHN9=9F 98G #4G8:>;>684 #4G82>;>6H 8= 8=C4A=0C8>=0;4B 0=31D27 IDA %0CDA D=3 *CAD:CDA E>= .NAC4A= D=3 .>ACB27LCI4= *" DD ^ 9F@=B '9K 2CF? 05@H9F 89 FIMH9F %5B;57?9F +CB5@8 0 5 >D=30C8>=B >5 >6=8C8E4 A0<<0A JC@IA9 " 5B8 "" ,H5B:CF8 5@= :CFB=5 ,H5B:CF8 .B=J9FG=HM )F9GG 6 'CIBG 5B8 /9F6G #0=6D064 ^ %9<18=8=6 8= 6A0<<0A 0=3 38B2>DAB4 DD ^

AGH9F85A )<=@589@D<=5 #C5A=BG %CB;57F9 +C69FH ,9BH9B79G 5G 7CA6=B5H=CBG C: 7@5IG9G "B -=ACH<M ,=BCAD2C8>=B JC@IA9 "" C: #0=6D064 CH?>;>6H 0=3 BH=C02 C82 34B2A8?C8>= DD ^ 5A6F=8;9 5A6F=8;9 .B=J9FG=HM )F9GG %IF5;<= ,=@J=5 &;3 8CC8C4 B4=C4=24 BCAD2CDA4 %CB8CB '9K 2CF? +CIH@98;9 %MCBG #C
References

337

Mac Donald, Lorna 1990 A Grammar of Tauya. Berlin; New York: Mouton de Gruyter. Mac Mahon, April S. 1994 Understanding language change. Cambridge: Cambridge University Press. Malchukov, Andrej 2004 Towards a semantic typology of adversative and contrast marking. Journal of Semantics, 21: 177–198. Maling, Joan and Annie Zaenen, (eds.) 1990 Modern Icelandic Syntax. San Diego: Academic Press. Mann, William and Sandra Thompson 1988 Rethorical structure theory: towards a functional theory of text organization. Text, 8: 243–281. Maslova, Elena 2003 A Grammar of Kolyma Yukaghir. Berlin; New York: Mouton de Gruyter. Matras, Yaron 1998 Utterance modiﬁers and universals of grammatical borrowing. Linguistics, 36-2: 281–331. Mauri, Caterina 2006 Combinazione e contrasto: i connettivi congiuntivi e avversativi nelle lingue d’Europa. Archivio Glottologico Italiano, 91(2): 166–202. 2007 Conjunctive, disjunctive and adversative constructions in Europe: Some areal considerations. In Paolo Ramat and Elisa Roma, (eds.), Europe and the Mediterranean as linguistic areas. Convergencies from a historical and typological perspective, pp. 183–214. Amsterdam; Philadelphia: John Benjamins. 2008 The irreality of alternatives: towards a typology of disjunction. Studies in Language, 32/1: 22–55. Mayo, Peter 1993 Belorussian. In Bernard Comrie and Greville G. Corbett, (eds.), The Slavonic Languages, pp. 887–946. London; New York: Routledge. McCawley, James D. 1981 Everything that linguists have always wanted to know about logic. Oxford: Basil Blackwell. Mcgonagle, Noel 1988 Irish Grammar. A Basic Handbook. Ofﬁcina Typographica. Meillet, Antoine 1958a Le renouvellement des conjonctions. In Linguistique historique et linguistique générale, pp. 159–174. Paris. Original edition in 191516: Champion. 1958b Linguistique historique et linguistique générale. Paris: Champion.

)454A4=24B &9F@5B F5B79G75 $0=60A0H8 %=B;I5 9G7F=DH=J9 ,HI8=9G AGH9F85A 'CFH< !C@@5B8 )I6@=G<=B; CAD5BM &=@@9F AM

A0<<0A >5 !0<18=8=6 8= 6A0<<0A 0=3 38B2>DAB4 DD ^ AGH9F85A )<=@589@D<=5 9B>5A=BG &CIG &55FH9B

-<9 ;F5AA5F C: 7CB>IB7H=J9 5B8 8=G>IB7H=J9 7CCF8=B5H=CB =B "F5EK "B &5FH=B !5GD9@A5H< 98 >>A38=0C8=6 2>=BCAD=2C8>=B DD ^

AGH9F85A )<=@589@D<=5 #C5A=BG '98>5@?CJ /@58=A=F ) ,CA9 HMDC@C;=75@ D5F5A9H9FG C: 7CBJ9F6G "B &5FH=B !5GD9@A5H< 5B8 ??9<5F8 $ZB=; 98G >=E4A1B 8= A>BB;8=6D8BC82 '4AB?42C8E4 DD ^ 9F@=B '9K 2CF? &CIHCB 89 FIMH9F e gXB< !CQ ';IMVB -84C=0<4B4 AGH9F85A )<=@589@D<=5 #C5A=BG '=G<=B5 2C?C

,5HNJ9F6=B8IB; IB8 ,5HNF98I?H=CB 5A 9=GD=9@ 89F >5D5B=G7<9B $CBJ9F6?CBGHFI?H=CB9B ** , A148CB?0?84A4 34B *4<8 =0AB 5PA *?A027F8BB4=B2705C 34A ,=8E4AB8CLC A5DAC 7CC?

FFF D=8 4A5DAC 34 B?A027F8BB4=B2705C **83, **83, ?35 ^ 'CCB5B &=7<59@ A0<<0A >5 #0=6> 9F@=B '9K 2CF? &CIHCB 89 FIMH9F (
CCF8=B5H=CB =B &9BH5@9G9 "B &5FH=B !5GD9@A5H< 98 >>A38=0C 8=6 2>=BCAD=2C8>=B DD ^ AGH9F85A )<=@589@D<=5 #C5A=BG (K9BG #CB5H<5B A0<<0A >5 0A0A &A><> !5A6IF; +CIH@98;9 )5@A9F F5B? +C69FH $>>3 0=3 <>30;8CH 5A6F=8;9 -9LH6CC?G =B %=B;I=GH=7G 5A 6F=8;9 5A6F=8;9 .B=J9FG=HM )F9GG )5B8<5F=D5B89 +5>9G=BCAD2C8>=B JC@IA9 "" C: #0=6D064 CH?>;>6H 0=3 BH=C02 C82 34B2A8?C8>= DD ^ 5A6F=8;9 5A6F=8;9 .B=J9FG=HM )F9GG

References

339

Peterson, David A. and Kenneth VanBik 2004 Coordination in Hakha Lai (Tibeto-Burman). In Martin Haspelmath, (ed.), Coordinating construnctions, pp. 333–354. Amsterdam; Philadelphia: John Benjamins. Prandi, Michele 2004 The building blocks of meaning. Amsterdam; Philadelphia: John Benjamins. Ramat, Paolo 1998 Typological comparison and linguistic areas: some introductory remarks. Language Sciences, 20(3): 227–240. Special Issue “Areal Typology”. 1999 Linguistic categories and linguists’ categorizations. Linguistics, 37: 157–180. Ramat, Paolo and Thomas Stolz, (eds.) 2002 Mediterranean Languages: Papers from the MEDTYP Workshop, Tirrenia, June 2000. Bochum: Brockmeyer. Rebuschi, Georges 2005 Generalizing the antisymmetric analysis of coordination to nominal modiﬁcation. Lingua, 115(4): 445–459. Rennison, John R. 1997 Koromfe. Descriptive Grammars. London; New York: Routledge. Rijkhoff, Jan, Dik Bakker, Kees Hengeveld, and Peter Kahrel 1993 A method in language sampling. Studies in Language, 17: 169–203. Rothstein, Robert 1993 Polish. In Bernard Comrie and Greville G. Corbett, (eds.), The Slavonic Languages, pp. 686–758. London; New York: Routledge. Rudolph, Elisabeth 1996 Contrast. Adversative and concessive expressions on sentence and text level. Berlin; New York: Walter de Gruyter. Saeed, John Ibrahim 1993 Somali reference grammar (Second revised edition). Kensington: Dunwoody Press. Sakel, Jeanette 2004 A Grammar of Mosetén. Berlin; New York: Mouton de Gruyter. Saltarelli, Mario 1988 Basque. Descriptive Grammars. London, New York: Croom Helm Ltd. Sanders, Gerald 1997 A functional typology of elliptical coordination. In Fred Eckman, (ed.), Current themes in linguistics. Bilingualism, experimental linguistics and language typologies, pp. 241–70. Washington, D.C: Hemisphere.

)454A4=24B ,5FA=9BHC +5AYB A0<0C820 10B820 34; B?06=>; %>A<0 H DB> &58F=8 )5\BGHKCK9 0M85KB=7HKC '5I?CK9 ,5IGGIF9 9F8=B5B8 89 >AB> 38 ;8=6D8BC820 64=4A0;4 HF5BG@5H98 6M -I@@=C 9 &5IFC +CA5 5F= %5H9FN5 (F=;=B5@ 98=H=CB =B ,75HHCB FB9GH I@;5F=5B "B 9FB5F8 CAF=9 5B8 F9J=@@9 CF69HH 98G +74 *;0E>=82 #0=6D064B DD ^ %CB8CB '9K 2CF? +CIH@98;9 ,7<57
'0A;>=B #DG4<1>DA64>8B #0=6D4 4C 2D;CDA4 ;8=6D8BC8@D4 3D= ?4C8C ?0HB 0D 2>4DA 34 ;DA>?4 )5F=G %b!5FA5HH5B ,7<=::F=B 96CF5< 8B2>DAB4 $0A:4AB *CD384B 8= =C4A02C8>=0; *>28>;8=6D8BC82B JC@ IA9 "" 5A6F=8;9 5A6F=8;9 .B=J9FG=HM )F9GG ,7CFF9HH= &5IFC %9 GHFIHHIF9 7CCF8=B5H9 "B %CF9BNC +9BN= 98 A0=34 6A0<<0C820 8C0;80=0 38 2>=BD;C0I8>=4 JC@IA9 " DD

^ C@C;B5 "@ &I@=BC ,<9J9@CJ 9CF;9 2 .?F5=B=5B "B 9FB5F8 CAF=9 5B8 F9J=@@9 CF69HH 98G +74 *;0E>=82 #0=6D064B DD ^ %CB8CB '9K 2CF? +CIH@98;9 ,=82 #0=6D064B DD ^ %CB8CB '9K 2CF? +CIH@98;9 ,=9K=9FG?5 BB5 D=2C8>=0; A0<<0A %CB8CB '9K 2CF? +CIH@98;9 ,A=FBCJ5 & +74 0DB0 ;0=6D064 0 34B2A8?C8E4 6A0<<0A %CB8CB +CIH@98;9 5B8 $9;5B ,CA40= 9G7F=DH=J9 F5AA5FG %CB8CB '9K 2CF? +CIH@98;9 ,H5GG9B %9CB >= 8= D=8E4AB0; 6A0<<0A (L:CF8 5G=@ @57?K9@@

'CIB D;>6H 0= 8=C4A=0C8>=0; 70=31>>: DD ^ 9F@=B '9K 2CF? &CIHCB 89 FIMH9F ,H=@C CB5@8

CCF8=B5H=CB =B H>A38=0C8=6 2>=BCAD=2C8>=B DD ^ AGH9F85A )<=@589@D<=5 #C5A=BG

)454A4=24B

,K99HG9F J9 A>< 4CH<>;>6H C> ?A06<0C82B $4C0?7>A820; 0=3 2D;CDA0; 0B?42CB >5 B4<0=C82 BCAD2CDA4B 5A6F=8;9 5A6F=8;9 .B=J9FG=HM )F9GG -=A69F@5?9 @5B +IGG=5B "B 9FB5F8 CAF=9 5B8 F9J=@@9 CF69HH 98G +74 *;0E>=82 #0=6D064B DD ^ %CB8CB '9K 2CF? +CIH@98;9 -F5I;CHH @=N569H< @CGG (B H<9 CF=;=BG C: a'b 5B8 a.-b 7CBB97H=J9G =B B;@=G< *CD384B 8= #0=6D064 ^ -I7?9F <=@8G 9CF;9 A0<<0A >5 "8B8 9F@=B '9K 2CF? &CIHCB 89 FIMH9F /5B 89B 9F; !9@A5

CCF8=B5H=B; 7CBGHFI7H=CBG =B 5;<9GH5B=5B @5B;I5;9G "B &5FH=B !5GD9@A5H< 98 >>A38=0C8=6 2>=BCAD=2C8>=B DD ^

A GH9F85A )<=@589@D<=5 #C5A=BG /5B 89F IK9F5 #C<5B +9J=G=H=B; H<9 5@?5B 5B8 &9GCA9F=75B @=B;I=GH=7 5F95G #0=6D064 *284=24B ^ ,D97=5@ "GGI9 _F95@ -MDC@C;M` /5B F=9A 9CF;9 A0<<0A >5 #8<1D 9F@=B '9K 2CF? &CIHCB 89 FIMH9F A0<<0A >5 D<8 9F@=B '9K 2CF? &CIHCB 89 FIMH9F /5B $@=B?9B 5H<5F=B5

FCA J9F6 HC 7CCF8=B5HCF =B -9HIB &240=82 #8=6D8BC82B ^ /5B (=FGCIK +C69FH + +74 BH=C0G >5 2>>A38=0C8>= %=B;I=GH=7G ,9F=9G '9K 2CF? FCCA !9@A /5B /5@=B +C69FH

G?;>A8=6 C74 *H=C0G*4<0=C82B =C4A5024 5A6F=8;9 5A6F=8;9 .B=J9FG=HM )F9GG /9FGHF59H9 #95B
-KC HMD9G C: 7CCF8=B5H=CB =B 7@5IG9 7CA6=B=B; #8=6D0 ^ 0R@7<@= 9FB<5F8

>2>D=3B 0=3 %0CDA0; >>A38=0C8>= (L:CF8 (L:CF8 .B=J9F G=HM )F9GG 2I5G5 HGIMC 5B8 #9FFM ,58C7?

)G9I8CGI6CF8=B5H=CB A=GA5H7< 69HK99B GMBH5L 5B8 G9A5BH=7G !>DA=0; >5 #8=6D8BC82B ^ 3=D: 9CF;9 $=B;G@9M D<0= 1470E8>A 0=3 C74 ?A8=28?;4 >5 ;40BC 455>AC 5A6F=8;9 & 88=GCB09G@9M

% #!

@65B=5B

F56=7

FA9B=5B 5GEI9

9@CFIGG=5B

I?=M=D I@;5F=5B ^

5H5@5B

<97<9B

^ N97<

5B=G<

5F;=

^

^ IA=

IH7<

B;@=G< ^ ^

GHCB=5B

=BB=G<

F9B7< ^

=34G>5;0=6D064B ^

9CF;=5B ^ ^

9FA5B

^ F99?

!5?<5 %5=

!5F5F (FCAC ^

!5IG5

!8=

!96F9K

!C75?

!IB;5F=5B

"79@5B8=7

"F5EK

"F=G<

"H5@=5B ^

^

#5AI@ -==D5M ^

#5D5B9G9 ^

344 Index of languages Kisi, 92, 111, 114, 117, 148, 151, 153, 188, 191, 192, 220, 224, 228, 245, 250, 254, 257, 301, 304, 318 Kolyma Yukaghir, 101, 111, 114, 117, 128, 148, 151, 153, 188, 191, 192, 210, 220, 224, 228, 238, 245, 250, 254, 278, 282, 301, 304, 318 Korean, 2, 20, 26, 29, 68, 101, 111, 114, 117, 139, 140, 148, 151, 153, 163, 167, 179, 183, 184, 189, 191, 193, 210, 220, 224, 228, 230, 231, 238, 239, 245, 250, 254, 257, 278, 282, 287, 300, 304, 318 Koromfe, 45, 93, 94, 111, 114, 117, 136, 138, 148, 152, 153, 167, 189, 191, 193, 197, 220, 224, 228, 246, 250, 254, 257, 301, 304, 319, 327 Kuskokwim Athapaskan, 111, 114, 117, 148, 152, 153, 189, 191, 193, 220, 224, 228, 246, 250, 254, 301 Lango, 93, 96, 102, 107, 108, 112, 114, 117, 127, 128, 148, 152, 153, 167, 189, 191, 193, 196, 197, 210, 215, 220, 224, 228, 238, 246, 250, 254, 257, 278, 301, 304, 319 Latin, 51, 292 Latvian, 20, 110, 113, 116, 147, 150, 188, 190, 217, 223, 227, 230, 243, 249, 253, 298, 303, 315 Lezgian, 43, 65, 68, 88, 90, 91, 98, 99, 101–103, 108, 110, 113, 116, 147, 150, 164, 168, 169, 181, 188, 190, 210, 214, 217, 223, 227, 234– 236, 239, 243, 249, 253, 275–278, 286, 298, 299, 303, 315 Limbu, 112, 114, 117, 128, 148, 152, 153, 189, 191, 193, 196, 197, 220, 224, 228, 246, 250, 254, 301, 304, 319 Lithuanian, 20, 73, 74, 89, 110, 113, 116, 133, 138, 147, 150, 188, 190,

203, 217, 223, 227, 243, 249, 253, 277, 282, 298, 303, 315 Luxembourgish, 20, 111, 113, 116, 147, 150, 188, 190, 218, 223, 227, 243, 249, 253, 282, 283, 289, 291, 298, 303, 315 Macedonian, 13 Malayalam, 67, 68, 112, 114, 117, 148, 152, 153, 167, 189, 191, 193, 220, 224, 228, 230, 231, 236, 246, 250, 254, 300, 304, 319 Maltese, 14, 20, 111, 113, 116, 147, 150, 168, 188, 190, 218, 223, 227, 243, 249, 253, 275, 285, 289, 291, 298, 303, 315 Mangarayi, 61, 112, 114, 117, 148, 152, 153, 163, 168, 179, 189, 191, 193, 199–201, 220, 224, 228, 230, 232, 246, 250, 254, 257, 300, 304, 319 Maori, 94, 96, 97, 99, 101, 112, 114, 117, 127, 128, 148, 152, 153, 189, 191, 193, 196, 197, 200, 202, 207, 208, 220, 224, 228, 235, 246, 250, 254, 257, 300, 304, 319 Marathi, 102, 108, 112, 115, 117, 148, 152, 153, 165, 169, 189, 191, 193, 210, 214, 215, 220, 225, 229, 234, 246, 251, 255, 300, 304, 319 Maricopa, 91, 92, 112, 115, 117, 148, 152, 153, 180, 189, 191, 193, 220, 225, 229, 246, 251, 255, 278, 300, 304, 320 Meithei, 112, 115, 117, 148, 152, 153, 163, 167, 189, 192, 193, 220, 225, 229, 246, 251, 255, 277, 301, 304, 320 Mosetén, 93, 94, 97, 98, 112, 115, 117, 148, 152, 153, 189, 192, 193, 206, 220, 225, 229, 246, 251, 255, 277, 301, 304, 320

=34G>5;0=6D064B

'QBP:K

'8MI?5

'CFK9;=5B

)9FG=5B

)C@=G<

^

^ )CFHI;I9G9

+5D5BI=

+CA5B= +CA5B=5B

+IGG=5B

,5F8=B=5B

,9F6CFC5H=5B ^

^ ,@CJ9B=5B ,CA5@=

,D5B=G<

^

,IDM=F9

,K98=G<

^

-5IM5

-9HIB -<5= -I?5B; 9G= ^

-IF?=G<

=34G >5 ;0=6D064B

^ -IJ5@I5B

.?F5=B=5B

/=9HB5A9G9

05F=b

^

09GH F99B@5B8=7

0C>C?9GC

% #" !

6F5<5A 09FB9F 85AW?CJP &5F79@5 @@5B +C6=B @@KCC8 #9BG @D5HCJ /@58=A=F @H -<9F9G5 A6F5N5G /MH5IH5G B89FGGCB %5FG IBB5F BG7CA6F9 #95B@5I89 FCBGCB !CK5F8 " FF=JT &=7<9@ G<9F +CB5@8 NNCD5F8=@9L5B89F &5F=9 PBFTH= 3C@HPB 58B5F7NI? %9GN9? 5??9F =? 5I9F 0=B=:F98 9FB=B= =I@=5BC 9GB=9F '=?C @5?9ACF9 =5B9 C
I7
CAF=9 9FB5F8 CAH9H +C;9F CF85 F5B79G7C CFA57? BB569@ CG9F=I I;9B=C F=GHC:5FC ,CB=5 ^ FC:H 0=@@=5A ^ I@=7CJ9F )9H9F MGCIK &=7<59@ 5<@ OGH9B 5G<9F +=7<5F8 9 9I /9FCB=75 =? ,=ACB =B= )=9HFC . CB5@8GCB FI79 CBCNMB;=9F 3M;AIBH 589H F5BSC=G9

348 Index of authors Galmiche, Michel, 302 Gil, David, 27, 74, 91, 180 Givón, Talmy, 3, 40, 74, 89, 105, 175, 266, 269 Glinert, Lewis, 304 Gordon, Lynn, 92, 304 Haiman, John, 9, 35, 36, 83, 84, 103– 107, 141, 143, 155, 156, 186, 266, 267, 324 Halliday, M.A.K., 83 Harms, Robert T., 302 Harries-Delisle, Helga, 54 Hasan, Ruqaiya, 83 Hasegawa, Yoko, 103 Haspelmath, Martin, 3, 8, 16, 24, 33, 34, 43, 44, 50–52, 54, 55, 58, 59, 65, 66, 68, 74, 102, 107, 122, 158, 159, 181, 236, 237, 274, 275, 279, 287, 303, 323, 325 Heine, Bernd, 132, 181 Helmbrecht, Johannes, 304 Hengeveld, Kees, 7, 12 Hewitt, George, 302 Hillery, P. J., 302 Hinchliffe, Ian, 303 Hinds, John, 304 Holmes, Philip, 302, 303 Holton, David, 302 Hopper, Paul J., 64, 325 Hualde, José Ignacio, 302 Huddlestone, Rodney, 302 Hutchinson, Amelia P. , 303 Huttar, George, 304 Huttar, Mary, 304 Jackendoff, Ray, 23, 26 Jeschull, Liane, 51, 61, 302, 328 Johannessen, Janne Bondi, 3 Johanson, Lars, 58, 303 König, Ekkehard, 41, 107, 258 Kahrel, Peter, 7, 12

Karlsson, Fred, 302 Kenesei, Istvan, 302 Kern, Barbara, 98, 167, 168, 233, 305 Kibrik, Andrej A., 305 Kornﬁlt, Jaklin, 25, 58, 88, 102, 234, 303 Kortmann, Bernd, 8, 14, 17, 58, 59, 63, 65, 75, 137, 145, 323 Kotcheva, Kristina, 74, 89, 122 Kumari, T.C., 67, 68, 236, 237, 304 Kuteva, Tania, 132, 181 Lakoff, Robin, 82, 84, 85, 122, 125, 126, 205, 212, 324 Lambrecht, Knud, 38 Lang, Ewald, 54, 57, 58, 82, 122, 123, 125, 143, 204 Langacker, Ronald W., 31, 36, 37, 83, 325 Lehmann, Christian, 23, 24, 28, 29, 323–325 Lipi´nska-Grzegorek, Maria, 303 Longacre, Robert E., 29, 30, 44, 54, 56, 60, 62, 66, 83–85 Lundskær-Nielsen, Tom, 302 Luraghi, Silvia, 122 Lyons, John, 171 Mac Donald, Lorna, 62, 69, 174, 305 Mac Mahon, April, 325 Mackridge, Peter, 302 Malchukov, Andrej, 89, 121, 125, 126, 136, 203–206 Maling, Joan, 302 Mann, William, 121 Maslova, Elena, 102, 128, 304 Matras, Yaron, 260, 261, 276 Mauri, Caterina, 121, 154, 157, 159, 274 Mayo, Peter, 302 McCawley, James D., 38 Mcgonagle, Noel, 302 Meillet, Antoine, 17, 323

=34G >5 0DC7>AB &9F@5B F5B79G75 &=@@9F AM &=H5@?CJ /@58=A=F e gXB< !CQ ';IMVB '=G<=B5 2C?C 'CCB5B &=7<59@ (
(K9BG #CB5H<5B )5;@=I75 0=@@=5A )5@A9F F5B? +C69FH )5B8<5F=D5B89 +5>9G??
+I8C@D< @=G569H< ^ ,58C7? #9FFM ,5998 #C

,7<57
% !#"!

58>IB7H D
58J9FG5H=J9 7CBGHFI7H=CB 58J9FG5H=J9 A5F?9F 58J9FG5H=J9 F9@5H=CB 5=A D5F5A9H9F ^ 5@H9FB5H=J9 A5F?9F F9@5H=CB ^ 7
^ ^

^ ^ ^

CDDCG=H=J9

^

G=AD@9 ^ ^ ^

^

^

^ JG 7CA6=B5H=CB B44 7CA6=B5H=CB JG 5@H9FB5H=J9 JG 7CBHF5GH B44 7CBHF5GH JG 5@ H9FB5H=J9 B8IH(F %5B;I5;9G ^

B8%5B;I5;9G

5GG9FH=CB ^

5GG9FH98 =B:CFA5H=CB ^ ^ BCB5GG9FH98 =B:CFA5H=CB DF5;A5H=7 G9A5BH=7 5GMAA9HFM GMAA9HFM GGIADH=CB ^ 7C;B=H=J9 :CFA:IB7H=CB :IB7H=CB5@ @=B95F C: DFCA=B9B79 H9ADCF5@ 5GMB89G=G B44 5GMB89H=7 7CBGHFI7H=CB 5GMB89H=7 7CBGHFI7H=CB ^ ^ ^ ^ ^ ^ ^

^ ^

65@5B798 J9F6 :CFA ^ 6=B5FM JG AI@H=D@9 7CCF8=B5H=CB CC@95B @C;=7 6CFFCK=B; C: 7CCF8=B5H=B; A5F?9FG

75IG5@ F9@5H=CB

7<5=B=B; @5B;I5;9G

=34G >5 BD1942CB 7@5IG9 7<5=B=B; 7CGI6CF8=B5H=CB 7C;B=H=J9 DFC @9 ^ 7CA6=B5H=CB A5F?9F ^ ^

F9@5H=CB ^

5H9ADCF5@ ^ ^ ^ ^ ^

^ ^

^

^ ^ ^

^

^ ^

^

G9EI9BH=5@ ^ ^

^ ^

^

^

^

G=AI@H5B9CIG ^ ^ ^ ^ ^ ^

^

^

JG 5@H9FB5H=J9 ^ JG 5@H9FB5H=J9 5B8 7CBHF5GH ^

JG 7CBHF5GH ^ 7CA=H5H=J9 7CAACB =BH9;F5HCF

7CB79DHI5@ GD579 B44 0;B> G9A5BH=7 A5DG ^ ^

^ 7CB79GG=J9 F9@5H=CB 7CB8=H=CB5@ 7CBGHFI7H=CB 7CB:@=7H D5F5A9H9F ^ 7CB>IB7H=CB 7CB>IB7H=J9 7CBGHFI7H=CB 7CB>IB7H=J9 7CCF8=B5H=CB 7CB>IB7H=J9 A5F?9F 7CB>IB7H=J9 D5FH=7=D=5@ :CFA D
7CBHF5GH JG 5@H9FB5H=J9 ^ A5F?9F

F9@5H=CB ^

7CFF97H=J9 ^

^ ^ ^ ^

^ ^

^

^ ^

^

^

7CIBH9F9LD97H5H=J9 ^

^

352 Index of subjects 133–146, 196, 197, 199, 202– 206, 208–211, 213, 257–260, 262–265, 270, 282–284, 289, 291, 292, 294, 296 oppositive, 4, 5, 9, 57, 65, 73, 74, 86–91, 94, 96, 98, 103, 106, 119, 122–144, 147–149, 153, 196, 197, 199, 202–206, 208–213, 257–259, 262–265, 268–270, 277, 279–284, 289, 291, 292, 296, 312–322 vs. combination, see combination vs. contrast converb, 33, 43, 45, 68, 75, 88, 91, 102, 103, 107, 133, 202, 323, 325, 326 cooccurrence dimension, 48, 55, 59, 80–83, 85, 89, 90, 94, 106, 120, 123, 134, 136, 155, 156, 161, 186, 199, 201, 208, 209, 211–214, 232, 239–241, 259, 261, 264, 268, 270 Coordinate Constituent Constraint, 33 coordinating construction (deﬁned), 1, 44 coordinating marker, xvii, 5, 9, 17, 57– 63, 65, 69–72, 75, 79, 80, 86, 91, 92, 94, 96, 107, 110, 111, 127, 128, 130, 147–149, 151, 154, 163, 187–189, 191, 195, 197, 214, 216, 219, 222, 224, 242, 245, 248, 250, 256, 259, 262, 268, 270, 276, 281, 285, 291, 292, 295–297, 326 dedicated, 70–72, 74–76, 79 general, 71, 72, 74–76, 79 coordination at various syntactic levels, 1, 23 functional deﬁnition of, 1, 31–44 morphosyntactic deﬁnition of, 23– 28 nominal, 3, 52, 325 of subordinate clauses, 39, 51, 52, 325

vs. subordination, 24, 26–30, 34, 35, 37, 63, 92 coordinator, see coordinating marker declarative sentence, 4, 40, 43, 65, 66, 175, 185, 186, 240, 258–260, 264, 272 degree of coding, 17, 49, 70, 76, 77, 130 denial of expectation, see contrast, counterexpectative dependency, 23, 24, 324 syntactic, 2, 23, 29, 30, 323 deranked verb form, 65–67, 69, 92, 100–102, 108, 140, 167, 210, 214, 266, 278 disjunction, 1, 3, 44, 48, 51, 179, 180, 194, 324, 325, 327 disjunctive construction, 48, 51, 194 exclusive, 324, 325 inclusive, 324, 325 interrogative, see alternative, choiceaimed standard, see alternative, simple dubitative form, 61, 167, 170, 174, 175, 178, 186, 231, 232, 285 economy (principle of), 9, 10, 80, 103, 104, 108, 120, 141, 145, 154, 185, 186, 214, 237, 239, 240, 257–259, 266, 267, 294 ellipsis, 54 embedding, 24–26, 28, 29 emphatic coordination, 50, 51 ﬁniteness ﬁnite verb form, 24, 29, 67, 69, 108, 109, 140, 141, 215, 327, 328 nonﬁnite verb form, 24–26, 29, 52, 67, 236, 325

=34G >5 BD1942CB :F9EI9B7M C: IG9

IB7H=CB5@ F5AA5F ;F5AA5H=75@=N5H=CB <MDCH<9H=75@ A5F?9F B44 0;B> 7CB8= H=CB5@ 7CBGHFI7H=CB

=7CB=7=HM DF=B7=D@9 C: =@@C7IH=CB5FM :CF79 ^ =AD9F5H=J9 A5F?9F G9BH9B79 ^ =BH9FFC;5H=J9 A5F?9F

ACC8 G9BH9B79 ^ ^ ^

=BHCB5H=CB D5HH9FB =BHCB5H=CB D5HH9FBG "BJ9FG9 +9@5H=CB !MDCH<9G=G =FF95@=G ^ ^

A5F?9F ^ ^ =FF95@=HM B44 =FF95@=G >ILH5DCG=H=CB B44 5GMB89H=7 7CBGHFI7 H=CB @=B;I=GH=7 5F95 A98=5@ J9F6G

ACFD
^

6CIB8 JG :F99 A5F?9F ACBC JG DC@MACFD<9A=7 A5F?9F ACBC JG DC@MGM@@56=7 A5F?9F AI@H=:IB7H=CB5@=HM B9;5H=J9 JG DCG=H=J9 7CCF8=B5H=CB BCB7CC77IFF9B79 8=A9BG=CB ^

BCBD5F5@@9@=GA GMBH57H=7 D5F5@@9@=GA 7CB79DHI5@ ^ 7CB79DHI5@@M D5F5@@9@ GH5H9G C: 5::5=FG :IB7H=CB5@ :IB7H=CB5@@M D5F5@@9@ GH5H9G C: 5::5=FG DF5;A5H=7 ^ G9A5BH=7 GMBH57H=7 ^

D5F5H5L=G DC@M:IB7H=CB5@=HM B44 AI@H=:IB7H=CB5@ =HM DC@MG9AM B44 AI@H=:IB7H=CB5@=HM

=34G >5 BD1942CB

EI5BH=:=98 =AD@=75H=CB F95@=G ^ ^

^ A5F?9F F95@=HM GH5HIG B44 F95@=HM J5@I9 F95@=HM J5@I9 ^ ^ , B44 ,H5B85F8 J9F5;9 IFCD95B G9A5BH=7 8CA5=B ^ ^ ^ ^ ^

^

^

^

^

G9A5BH=7 A5DG B44 0;B> 7CB79DHI5@ GD579

^ ,DF57<6IB8 ,H5B85F8 J9F5;9 IFCD95B

GH5H9 C: 5::5=FG 89 B98 GI6CF8=B5H=B; 7CBGHFI7H=CB GI6CF8=B5HCF GI6CF8=B5H=CB :IB7H=CB5@ 89:=B=H=CB C: ACFD

F9@5H=CB JG 7CCF8=B5H=CB B44 7CCF8=B5H=CB JG GI6CF8=B5H=CB GK=H7<F9:9F9B79 A5F?9FG GMAA9HFM B44 0;B> D5F5@@9@=GA ^ 5GG9FH=J9 7CB79DHI5@ C: DFCA=B9B79 GMBH57H=7 GMB89H=7 7CBGHFI7H=CB ^

GMBH57H=7 75H9;CFM GMBH57H=7 =B89D9B89B79 H9ADCF5@=HM D5F5A9H9F

0=H<%5B;I5;9G 3=D:bG %5K

Tense and Aspect in the Languages of Europe (Empirical Approaches to Language Typology : Eurotyp 20-6)

Read more

Coordination Models and Languages - COORDINATION 2011

Read more

Immigrant Languages in Europe

Read more

Immigrant languages in Europe

Read more

Alignment Change in Iranian Languages: A Construction Grammar Approach (Empirical Approaches to Language Typology)

Read more

Evidentials: Turkic, Iranian and Neighbouring Languages (Empirical Approaches to Language Typology, Vol 24)

Read more

The Ancient Languages of Europe

Read more

The Ancient Languages of Europe

Read more

The Changing Languages of Europe

Read more

Towards Otherland: Languages of Science and Languages Beyond

Read more

Coordination Models and Languages, 9 conf., COORDINATION 2007

Read more

On Comitatives and Related Categories: A Typological Study With Special Focus on the Languages of Europe (Emprical Approaches to Language Typology 33) (Emprical Approaches to Language Typology)

Read more

Quotative Indexes in African Languages. A Synchronic and Diachronic Survey (Empirical Approaches to Language Topology)

Read more

Grammatical Relations: A Cross-Linguistic Perspective on Their Syntax and Semantics (Empirical Approaches to Language Typology)

Read more

A Typology of Verbal Derivation in Ethiopian Afro-Asiatic Languages

Read more

Approaches to the Typology of Word Classes

Read more

The Dakotan Languages, and Their Relations to Other Languages (Dodo Press)

Read more

Minority Languages in Europe: Frameworks, Status, Prospects

Read more

The Internet and Languages

Read more

The Languages of Pao

Read more

The Languages of Pao

Read more

The Languages of Pao

Read more

The Languages of Pao

Read more

The Languages of Pao

Read more

The Evolution Of Languages

Read more

Languages of the Kimberley

Read more

Languages and Communities in Early Modern Europe (The Wiles Lectures)

Read more

The Use of Databases in Cross-Linguistic Studies (Empirical Approaches to Language Typology)

Read more

Variationstypologie Variation Typology: A Typological Handbook of European Languages

Read more

The Languages and Linguistics of Europe: A Comprehensive Guide

Read more

Recommend Documents

Tense and Aspect in the Languages of Europe (Empirical Approaches to Language Typology : Eurotyp 20-6)

Osten Dahl (Editor) Tense and Aspect in the Languages of Europe Empirical Approaches to Language Typology EUROTYP Ed...

Coordination Models and Languages - COORDINATION 2011

Lecture Notes in Computer Science Commenced Publication in 1973 Founding and Former Series Editors: Gerhard Goos, Juris...

Immigrant Languages in Europe

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

Immigrant languages in Europe

Alignment Change in Iranian Languages: A Construction Grammar Approach (Empirical Approaches to Language Typology)

Alignment Change in Iranian Languages: A Construction Grammar Approach Geoffrey L. J. Haig Mouton de Gruyter Alignme...

Evidentials: Turkic, Iranian and Neighbouring Languages (Empirical Approaches to Language Typology, Vol 24)

Lars Johanson Bo Utas (Editors) Evidenţiate Turkic, Iranian and Neighbouring Languages W DE G. Mouton de Gruyter Em...

The Ancient Languages of Europe

This page intentionally left blank THE ANCIENT LANGUAGES OF EUROPE This book, derived from the acclaimed Cambridge En...

The Ancient Languages of Europe

This page intentionally left blank THE ANCIENT LANGUAGES OF EUROPE This book, derived from the acclaimed Cambridge En...

The Changing Languages of Europe

The Changing Languages of Europe Dedicated to Wim Blockmans and the wonderful people of NIAS The Changing Languages...

Towards Otherland: Languages of Science and Languages Beyond

Rainer E. Zimmermann & Vladimir G. Budanov (eds.) Towards Otherland Languages of Science and Languages Beyond INTAS Vo...