Hypotetical modality 9027230544, 9789027230546
168 34 1MB
English Pages [310] Year 2000
Polecaj historie
Citation preview
HYPOTHETICAL MODALITY
STUDIES IN LANGUAGE COMPANION SERIES (SLCS) The SLCS series has been established as a companion series to STUDIES IN LANGUAGE, International Journal, sponsored by the Foundation “Foundations of language”. Series Editors Werner Abraham University of Groningen The Netherlands
Michael Noonan University of Wisconsin-Milwaukee USA Editorial Board
Joan Bybee (University of New Mexico) Ulrike Claudi (University of Cologne) Bernard Comrie (Max Planck Institute, Leipzig) William Croft (University of Manchester) Östen Dahl (University of Stockholm) Gerrit Dimmendaal (University of Leiden) Martin Haspelmath (Max Planck Institute, Leipzig) Ekkehard König (Free University of Berlin) Christian Lehmann (University of Erfurt) Robert Longacre (University of Texas, Arlington) Brian MacWhinney (Carnegie-Mellon University) Marianne Mithun (University of California, Santa Barbara) Edith Moravcsik (University of Wisconsin, Milwaukee) Masayoshi Shibatani (Kobe University) Russell Tomlin (University of Oregon) John Verhaar (The Hague)
Volume 51
Debra Ziegeler Hypothetical Modality Grammaticalisation in an L2 dialect
HYPOTHETICAL MODALITY Grammaticalisation in an L2 dialect
DEBRA ZIEGELER , Taiwan Monash University, Australia
JOHN BENJAMINS PUBLISHING COMPANY AMSTERDAM/PHILADELPHIA
8
TM
The paper used in this publication meets the minimum requirements of American National Standard for Information Sciences — Permanence of Paper for Printed Library Materials, ANSI Z39.48–1984.
Library of Congress Cataloging-in-Publication Data Ziegeler, Debra. Hypothetical modality : grammaticalisation in an L2 dialect / Debra Ziegeler. p. cm. -- (Studies in language companion series, ISSN 0165-7763 ; v. 51) Includes bibliographical references and indexes. 1. Modality (Linguistics) 2. Grammar, Comparative and general--Grammaticalization. 3. English language--Modality. 4. Second language acquisition. I. Title. II. Series. P299.M6 Z54 2000 415--dc21 99-462178 ISBN 90 272 3054 4 (Eur.) / 1 55619 937 6 (US) (alk. paper) CIP © 2000 – John Benjamins B.V. No part of this book may be reproduced in any form, by print, photoprint, microfilm, or any other means, without written permission from the publisher. John Benjamins Publishing Co. • P.O.Box 75577 • 1070 AN Amsterdam • The Netherlands John Benjamins North America • P.O.Box 27519 • Philadelphia PA 19118-0519 • USA
Table of Contents
Preamble . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
xi
Preface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xiii Abbreviations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xv List of Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xvii List of Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xix C 1 Introduction . . . . . . . . . . . . . . . . . . . . . 1.1 Preliminaries . . . . . . . . . . . . . . . . . 1.2 Main points . . . . . . . . . . . . . . . . . . 1.2.1 Defining hypothetical modality 1.2.2 The role of lexical retention . . 1.2.3 The hypothesis . . . . . . . . . . . 1.3 Overview . . . . . . . . . . . . . . . . . . . .
. . . . . . .
. . . . . . .
. . . . . . .
. . . . . . .
. . . . . . .
. . . . . . .
. . . . . . .
. . . . . . .
. . . . . . .
. . . . . . .
C 2 Hypothetical modality as a grammatical category . . . . 2.1 The grammatical derivation of hypothetical meaning 2.1.1 Temporality and hypotheticality . . . . . . . . . 2.1.1.1 The remoteness metaphor . . . . . . . . 2.1.2 Conditionals and conversational implicatures 2.1.3 The falsity of the antecedent clause . . . . . . . 2.1.4 Other features of the cluster . . . . . . . . . . . . 2.2 A principle for determining the grammatical basis of meaning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.3 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . .
. . . . . . .
. . . . . . .
. . . . . . .
. . . . . . .
. . . . . . .
. . . . . . .
. . . . . . .
. . . . . . .
. . . . . . .
. . . . . . .
. . . . . . .
1 1 5 5 5 6 7
......... ......... ......... ......... ......... ......... ......... hypothetical ......... .........
. . . . . . .
. . . . . . .
. . . . . . .
15 17 17 19 22 28 36
. . . 40 . . . 42
vi
TABLE OF CONTENTS
C 3 A diachronic corpus study of would . . . . . . . . . . . . . . . . 3.1 Grammaticalisation and grammaticality . . . . . . . . . . . 3.2 The diachronic grammaticalisation of WOULD . . . . . . 3.2.1 The conventionalisation of implicature in the grammaticalisation of modal verbs . . . . . . . . . 3.2.2 Hypotheticality . . . . . . . . . . . . . . . . . . . . . . . 3.2.3 WOULD HAVE . . . . . . . . . . . . . . . . . . . . . . . 3.2.4 Future-in-the-past . . . . . . . . . . . . . . . . . . . . . 3.3 The data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.3.1 Notes on the categories . . . . . . . . . . . . . . . . . 3.3.1.1 Environments . . . . . . . . . . . . . . . . . . . 3.3.2 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.4 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.4.1 The OE sample . . . . . . . . . . . . . . . . . . . . . . 3.4.2 The ME sample . . . . . . . . . . . . . . . . . . . . . . 3.4.3 The EME sample . . . . . . . . . . . . . . . . . . . . . 3.5 The context-dependent grammaticalisation of WOULD 3.6 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . 45 . . . . . . . . . . 46 . . . . . . . . . . 48 . . . . . . . . . . . . . .
. . . . . . . . . . . . . .
. . . . . . . . . . . . . .
. . . . . . . . . . . . . .
. . . . . . . . . . . . . .
. . . . . . . . . . . . . .
. . . . . . . . . . . . . .
. . . . . . . . . . . . . .
C 4 Singaporean English and substratum influences in the grammaticalisation of hypothetical modality . . . . . . . . . . . . . . . . . . 4.1 Singaporean English: a descriptive account . . . . . . . . . . . . . . . . 4.1.1 Substratum (L1) and contact languages . . . . . . . . . . . . . . 4.1.2 The status of Singaporean English . . . . . . . . . . . . . . . . . 4.1.3 Characteristic features of Singaporean English: Chinese L1 influence . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.2 Hypothetical modality . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.2.1 The expression of hypotheticality in Standard Chinese and substratum dialects . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.2.2 The expression of hypotheticality in Singaporean English . 4.2.2.1 An earlier study . . . . . . . . . . . . . . . . . . . . . . . . . 4.3 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . .
. . . . . . . . . . . . . .
48 52 54 56 59 60 61 63 66 66 70 73 79 81
. . . .
. . . .
85 86 86 88
. . 90 . . 94 . . . .
. . . .
94 104 106 110
vii
TABLE OF CONTENTS
C 5 The interaction of tense and aspect in the grammaticalisation of counterfactuality . . . . . . . . 5.1 Tense marking in Singaporean English . . . . . . . 5.2 Past modals: differences in counterfactual scope 5.3 The Study . . . . . . . . . . . . . . . . . . . . . . . . . . 5.3.1 Methodology . . . . . . . . . . . . . . . . . . . 5.3.2 The Questionnaire . . . . . . . . . . . . . . . . 5.3.3 Results . . . . . . . . . . . . . . . . . . . . . . . 5.4 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . 5.5 Concluding remarks . . . . . . . . . . . . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
111 112 116 119 119 120 121 124 133
C 6 Hypothetical WILL: A study in retention . . . . . . . . . . . . . . . . . . . . . 6.1 Australian English . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.1.1 Earlier discussions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.1.1.1 Grammaticalisation and the epistemic status of WILL 6.1.1.2 Stative verbs . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.1.1.3 The future progressive . . . . . . . . . . . . . . . . . . . . . 6.1.1.4 Subject sensitivity . . . . . . . . . . . . . . . . . . . . . . . . 6.1.1.5 Negation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.1.2 Methodology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.1.2.1 The Questionnaire . . . . . . . . . . . . . . . . . . . . . . . . 6.1.3 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.1.3.1 Results of Control Questionnaire . . . . . . . . . . . . . . 6.1.4 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.1.4.1 The control questionnaire . . . . . . . . . . . . . . . . . . . 6.1.5 A grammaticalisation chain . . . . . . . . . . . . . . . . . . . . . . . 6.1.6 Postscript: Variablity of retention in hypothetical clauses . . . 6.2 Singaporean English . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.2.1 Methodology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.2.2 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.2.3 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.2.4 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.3 British English — a control study . . . . . . . . . . . . . . . . . . . . . . . . 6.3.1 Methodology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.3.2 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.3.3 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.3.4 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . .
137 137 140 141 142 143 144 147 148 149 149 151 152 160 161 164 166 166 167 168 179 182 183 184 189 201
viii
TABLE OF CONTENTS
C 7 The Lexical Memory Traces Hypothesis . . . . . . . . . . . . . . . . . . . . . 7.1 The phenomenon of retention . . . . . . . . . . . . . . . . . . . . . . . . . . 7.1.1 Retention as illustrated in the results of grammaticalisation processes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7.1.1.1 French . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7.1.1.2 Mandarin . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7.1.1.3 Malay . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7.1.2 Defining the phenomenon . . . . . . . . . . . . . . . . . . . . . . . 7.1.3 L2 speakers’ overgeneralisations . . . . . . . . . . . . . . . . . . . 7.2 Towards an explanation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7.2.1 General considerations . . . . . . . . . . . . . . . . . . . . . . . . . . 7.2.1.1 Frequency of use . . . . . . . . . . . . . . . . . . . . . . . . 7.2.1.2 Analogical ‘matching’ . . . . . . . . . . . . . . . . . . . . . 7.2.1.3 Parallels in language development . . . . . . . . . . . . 7.3 Developing the hypothesis: parallels in historical and individual development . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7.3.1 Correlations in source concepts . . . . . . . . . . . . . . . . . . . 7.3.2 Actual instances of co-evolutionary parallels, and evidence from children’s errors . . . . . . . . . . . . . . . . . . . . . . . . . . 7.3.2.1 Modal verbs . . . . . . . . . . . . . . . . . . . . . . . . . . . 7.3.2.2 Mandarin ba . . . . . . . . . . . . . . . . . . . . . . . . . . . 7.3.2.3 Past tense . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7.3.2.4 Grammatical subjects . . . . . . . . . . . . . . . . . . . . . 7.3.2.5 Space-to-time mapping . . . . . . . . . . . . . . . . . . . . 7.3.3 Provisional conclusions . . . . . . . . . . . . . . . . . . . . . . . . . 7.4 A theoretical account of retention . . . . . . . . . . . . . . . . . . . . . . . 7.4.1 Retention as a product of acquisition . . . . . . . . . . . . . . . . 7.4.2 The Lexical Memory Traces Model . . . . . . . . . . . . . . . . . 7.4.3 Integrating the model into a theory: neurological order of development . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7.5 Applying the hypothesis to the results of the studies . . . . . . . . . . 7.5.1 Lexical Memory Traces in past tense stative verbs . . . . . . 7.5.2 By-passing the lexical stages of acquisition . . . . . . . . . . . 7.5.3 Reasons for the differences in rates of grammaticalisation . 7.5.4 LMTs as prototype extensions . . . . . . . . . . . . . . . . . . . . 7.6 Concluding remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . 205 . . 207 . . . . . . . . . . .
. . . . . . . . . . .
207 208 208 209 210 212 214 214 214 216 217
. . 218 . . 218 . . . . . . . . . .
. . . . . . . . . .
220 220 222 223 225 227 228 230 230 231
. . . . . . .
. . . . . . .
234 236 236 238 240 241 243
TABLE OF CONTENTS
ix
C 8 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 247 Appendices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 257 Appendix 1: Chapter 6 Questionnaire . . . . . . . . . . . . . . . . . . . . . . . . . . 257 Appendix 2: Chapter 6 Control Questionnaire . . . . . . . . . . . . . . . . . . . . 258 Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 259 Selected bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 259 Primary sources . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 277 Name Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 281 Subject Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 283
Preamble
In an earlier study, it was found that, amongst speakers of an L2 variety of English, there was no true correlation between the presence or absence of a morphological marking for expressing hypothetical meaning and the speaker’s ability to conceptualise hypothetical notions. The study had raised the question why, for many of the speakers concerned, the form of a past tense modal auxiliary would had been found to lack the salience perceived by native speakers to derive meanings of hypotheticality. The present work addresses the question using the principles of grammaticalisation theory, and considers the L2 variety within its historical context as a developing contact language, affected by more highly accelerated grammaticalisation processes than are usually associated with native speaker varieties. The grammaticalisation of hypothetical modality in the use of a past tense modal is explained with reference to the role of semantic retention, and the lower degree of retentive constraints in the L2 variety results in more rapid rates of grammaticalisation for speakers of such varieties. An empirical study investigates the presence of semantic retention in uses of would amongst native speakers, and the results are compared with a similar study undertaken with L2 speakers. The results reflect a generally stronger tendency for the effects of retention to be demonstrated in the linguistic behaviour of native-speaker informants than in those of an L2 variety. A hypothesis is proposed whereby the processes of the diachronic development in grammaticalisation are correlated with processes in individual acquisition, and the influence of retention as a constraint on grammaticalisation processes is considered from the viewpoint of early pre-grammatical development in the individual. The accelerated rates of the grammaticalisation of hypothetical would in the L2 variety are therefore considered in the light of two factors: (i) contact-induced constraints on the generalisation of past tense to include stative environments, and (ii) the absence for some speakers of the retention of volition senses in the modal, necessary to provide the implicatures of hypotheticality in native speaker varieties. Both these factors indicate the significance of continuity in transmission
xii
PREAMBLE
of grammaticalising features in a language, and suggest that the broader generalisation of grammaticalising features in an L2 variety is the result of a discontinuous historical development in the language itself.
Preface
This book is a revised version of my Ph. D. thesis: Aspects of the Grammaticalisation of Hypothetical Modality, which I completed at Monash University, Melbourne, Australia, in 1997, and has been updated with a new study (Chapter 3), undertaken after the original thesis was written. The primary goals in producing the book are to place a new focus on what is now an established field of investigation in grammaticalisation research, and to explore the possibilities of extending a theoretical base which is only now beginning to receive criticism from many angles (see, for example, Newmeyer 1998). In doing so, the exploration covers an area less frequently treated as the subject of grammaticalisation studies, the field of second language regional dialects which are becoming institutionalised in the locations where they are spoken, but are not strictly regarded as creoles for historical reasons. It is hoped that the studies presented will provide a sufficient cause to motivate further investigation in similar areas of grammaticalisation research. In the duration of my Ph. D. candidature, a number of people offered generous assistance in enabling the studies to be completed. I am especially grateful to my supervisor, Keith Allan, for his unselfish interest, patience, and encouragement during the three years of research, for granting me the freedom to branch out into foreign territory. Upon his suggestions (along with those of J.-C. Smith and Eve Sweetser), I decided to begin writing this book. I would also like to thank Mark Newbrook for his initial supervisory assistance in sociolinguistic methodology. For inspiration, welcome advice, and encouragement I am indebted to Bernd Heine and Elizabeth Traugott, whose introductory courses in grammaticalisation at the 1994 Australian Linguistics Institute gave me the impetus to launch into a new field and to see the relevance of grammaticalisation theory as an explanatory tool in analysing crossdialectal data. Amongst those who have offered helpful criticism, discussion, or comments on aspects of the studies presented I am grateful to Barbara Abbott, Maya Bradley, Hilary Chappell, Östen Dahl, Bernd Heine, Talmy Givón, J.-C. Smith and Elizabeth
xiv
PREFACE
Traugott, to numerous anonymous referees for Studies in Language, Language Sciences, Journal of Pragmatics and Cognitive Linguistics, as well as to lively audiences at the annual conferences of the Australian Linguistics Society between 1994 and 1996, the Pan-Asiatic Symposium 1996, the LACUS Forum 25, and the ‘New Reflections on Grammaticalization’ Symposium, Potsdam, 1999, at which various portions of the work have been presented. For funding support during my candidature I acknowledge the assistance of the Australian Department of Employment, Education, Training and Youth Affairs in the form of an Australian Postgraduate Award. I would also like to acknowledge the support of the Australian Vice-Chancellors’ Committee for the provision of an Australian Award for Research in Asia. I am especially indebted to the following people who assisted me in data collection: Mew Leng Hoh, Tom Estad, Jane Bishop, Ignatia Tsantarlis, numerous anonymous staff and students at Bukit Merah Secondary School, Singapore, and in London, Vivienne Heaton, Philip Blackwell, Michael Marland, and Jennifer Bryant, staff and students at the North Westminster Community School, and at the British Natural History Museum. Last but not least by any means I am eternally grateful to the constant support and advice of my husband, Dr. Tuck Choy, to my father, Lionel, and to the loyal feline companionship of Tiger Choy, whose ever-purring presence on my computer desk helped to sooth all the anxieties of producing and completing this work and the thesis out of which it grew. The book is dedicated to Phyllis May Ziegeler (1912–1992), for her lifelong support and inspiration.
Abbreviations
L1 L2
classifier classifier current relevant state inchoative first language second language Lexical Memory Trace(s) linking particle negative perfective aspect plural particle resultative complement singular
List of Figures
Figure 5.1 Graph showing percentages of counterfactual (CF) responses per group for each question . . . . . . . . . . . . . . . . . . . . . . . . . . . Figure 5.2 Graph showing the percentage of deviation of the counterfactual indices of the Singaporean groups from that of the Control . . . Figure 6.1 Degree of acceptability according to context found in the use of WILL in hypothetical predicates, indicating with a chain-like effect the predicted direction of grammaticalisation on a broader, diachronic scale . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Figure 6.2 Percentage of acceptable responses for both Singaporean and Australian groups, indicating the possible differences in rates of retention of former volitional meanings associated with the modal WOULD in hypothetical complements . . . . . . . . . . . . . . . . . Figure 6.3 Percentage of acceptable responses for LL1G, SG and AG, and indicating the possible differences in rates of retention and grammaticalisation rates associated with WOULD in hypothetical complements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Figure 7.1 Representational model of the development of retention as an ontogenetic/diachronic phenomenon in the compositional value of the grammaticalising expression . . . . . . . . . . . . . . . . . . . Figure 7.2 Representational model of the retention of volitional senses in the grammaticalising predictive meanings of the modal WOULD in affected environments . . . . . . . . . . . . . . . . . . . . . . . . . . . .
123 124
163
169
185
232
233
List of Tables
Table 2.1
Table 3.1 Table 3.2 Table 3.3 Table 3.4
Table 4.1 Table 4.2 Table 5.1 Table 5.2 Table 5.3 Table 6.1
Table 6.2 Table 6.3
Results in averages of two tasks eliciting counterfactual interpretations from native speaker and L2 speaker groups (Ziegeler 1993) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Distribution by function and environment of the modal forms wold-/uuold/nold- in the OE sample . . . . . . . . . . . . . . . . . . . Distribution by function and environment of the modal forms wold-/uuold/would/wald-/nold- in the ME sample . . . . . . . . . Distribution by function and environment of the modal forms wold-/would in the EME sample . . . . . . . . . . . . . . . . . . . . . Distribution of lexical and grammatical functions of WOULD in the Old English, Middle English, and Early Modern English data, as shown in main clauses only . . . . . . . . . . . . . . . . . . . . . . . Range of constructions used in Singaporean English to express a counterfactual conditional (Ziegeler 1994: 42) . . . . . . . . . . . Range of constructions used by the Australian English control group to express a counterfactual conditional . . . . . . . . . . . . Frequency of counterfactual implicatures for Singaporean and Australian groups, graded by educational level . . . . . . . . . . . Frequency of occurrence of the counterfactual index pattern, for Singaporean and Australian (control) groups . . . . . . . . . . . . . Proportion of deviation of the Singaporean counterfactual indices from the Control index . . . . . . . . . . . . . . . . . . . . . . . . . . . . Graded assessments of the main questionnaire by male and female Australian respondents, showing the acceptability of the distribution of WILL in hypothetical predicates by environment Graded assessments of the main questionnaire by male respondents only . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Graded assessments of the main questionnaire by female subjects only . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
37 63 64 65
77 107 108 122 123 122
150 150 151
xx Table 6.4
Table 6.5 Table 6.6 Table 6.7
Table 6.8 Table 6.9
Table 6.10 Table 6.11 Table 6.12 Table 6.13 Table 6.14 Table 7.1
LIST OF TABLES
Graded assessment of the example sentences in the control questionnaire (see Appendix) by both male and female respondents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Graded assessment of the control questionnaire by male respondents only . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Graded assessment of the control questionnaire by female respondents only . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Graded assessments of the questionnaire by male and female Singaporean respondents, showing the acceptability of the distribution of WILL in hypothetical complements by environment . . Comparative listing of the results of the Australian group (AG) with those of the Singaporean group (SG). . . . . . . . . . . . . . . . London L1 Group (LL1G): Graded assessments of the questionnaire by both male and female respondents from the London secondary school . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Distribution of responses for the London group of native speakers (older age museum workers group) . . . . . . . . . . . . . . . . . . . Distribution of responses for non-English L1 speakers (London secondary school) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Distribution of acceptances by all groups, arranged in order from highest to lowest scores . . . . . . . . . . . . . . . . . . . . . . . . . . . Variance of the scores for each question . . . . . . . . . . . . . . . . Correlation of average scores with functional description, illustrating the preferred acceptance patterns . . . . . . . . . . . . . . . Average acceptances of would in wish-complements in 3 enviroments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
151 152 152
168 168
184 186 186 187 188 200 213
C 1 Introduction
1.1 Preliminaries In the early eighties, an American psycholinguist, Alfred Bloom, undertook a series of investigations into the relationship between the absence of specific grammatical marking for counterfactuality in the Chinese language and a putative tendency (hypothesised by Bloom, from the basis of informal observations) for counterfactual reasoning to be exceptional, if not alien to Chinese culture. The publication of a book summarising these controversial investigations (Bloom 1981) led rapidly to a succession of counter-attacks from psycholinguistic areas, in particular from researchers of Chinese ethnicity (e.g. Au 1983; Liu 1985; Wu 1994), not only for the potentially prejudicial nature of its deterministic objectives, but also for the fact that it presented a methodologically deficient means of attempting to prove them, in the use of translated materials to elicit counterfactual interpretations from Chinese first language (L1) speakers. The hypothesis that the absence of grammatical markers for counterfactuality in one’s first language affects the ability to reason counterfactually was not adequately tested by such means, and this was the target for most of the attacks, which mainly attempted to demonstrate that such grammatical markers did exist in Chinese. However, by attempting to prove the existence of grammatical markers for counterfactuality, such studies continued to work on the initial assumption that counterfactuality could be linguistically determined, and were simply defending the Chinese language on those grounds. The studies which followed as a reaction to Bloom’s seminal investigations were mostly of a psycholinguistic and empirical nature. They were criticised by Lucy (1992: 246; 1996: 48) for their failure to adopt a typological and comparative framework, which would have allowed for evidence from other languages, particularly from non-Indo-European languages, to indicate whether or not Chinese was unique in lacking specific marking for counterfactuality. Such a
2
HYPOTHETICAL MODALITY
framework would also have demonstrated the ways in which other languages express similar concepts. Comparative, cross-linguistic studies reveal that languages carve up the conceptual sphere in different ways and in different directions according to the cognitive salience of a feature within a language. If Bloom had used such methods, he would have found that a particular category in one language may be absent from another due to the availability of alternative means to express the concepts it encodes. The absence of a grammatical category to specify a function in a language does not necessarily indicate the absence of the conceptual information it conveys. In the present study, it will be shown that hypotheticality or counterfactuality is not always explicitly stated in a particular form, but that languages may use existing forms at their disposal with other functions to imply such meanings, or, alternatively, may rely only on contextual evidence to obtain such meanings. The selection of means available to express hypothetical notions will thus vary from one language to another. Since the early psycholinguistic studies, investigations into counterfactual thinking in Chinese L1 speakers have largely been lain to rest. However, a considerable amount of dialectal variation was uncovered in the expression of counterfactual meanings amongst L1 Chinese speakers of Singaporean English in later studies (Ziegeler 1993, 1994). The aims of those studies were to show that investigations of the expression of counterfactuality amongst L1 Chinese speakers could be undertaken without the use of cumbersome translations as stimulus materials. If first language constraints were to result in the reduced salience of counterfactual concepts amongst Chinese speakers, then this could affect their willingness to seek out grammatical markers for counterfactuality in a second language (L2) — English. However, if the L2 provided the trigger for counterfactual reasoning, then such reasoning should be accessed whenever that language was used. The earlier studies therefore set the argument of an underlying absence of counterfactual concepts against the possibility that such concepts could be created anew, deterministically, by linguistic information whenever it was present in the discourse. The investigations also showed that it was possible to correlate evidence of the production of counterfactual expressions with evidence of their comprehension in the same study. The studies are summarised briefly in Chapter 4. Counterfactual interpretations of narrative sequences were, in fact, higher on average for the L2 speakers than for the L1 speakers of Australian English in the study (e.g. for one task 55% of the Singaporean subjects produced a counterfactual interpretation as against only 19% of the control L1 speakers). The evidence of higher rates of counterfactual interpretations from the Singaporean speakers from the presence
INTRODUCTION
3
of specific grammatical indicators may in fact reflect the possibility that such speakers were inclined to use the grammatical scaffolding more than the L1 speakers as a frame on which to establish form-function taxonomies that were less formally distinguishable in their L1, Chinese. It will be seen that the interpretation of counterfactual meaning in the Chinese language relies mainly on context and shared speaker-hearer knowledge. However, even in L1 English, grammatical devices (such as modal perfect conditional morphology) are not absolute in determining a counterfactual meaning, but merely contribute to a stronger evaluation by the hearer of the counterfactuality of the speaker’s intentions. The present study is not intended to contribute to the stream of attack which followed the controversy of Bloom’s initial speculations, nor to the proliferation of subsequent studies undertaken in defence of the Chinese counterfactual. However, with the development and recent advances in grammaticalisation studies, it is timely and important to re-examine the previous work in the expression of hypothetical modality, to discuss its relevance in an L2 context, and to extend it into the framework of a diachronic theory. Although much emphasis has been made in past research on the use of logical theories in explaining hypothetical modality (e.g. Lewis 1973; Lewis 1986; and McCawley 1981), and similar emphasis has focussed frequently on the use of pragmatics to explain (especially conditional) modality (e.g. Karttunen 1971; van der Auwera 1983; and Comrie 1986), it is only recently that studies have begun to consider modality of any kind from the viewpoint of an evolving historical phenomenon (e.g. Bybee and Pagliuca 1987; Traugott 1989; Bybee, Pagliuca and Perkins 1994). As will be seen in later chapters, the diachronic viewpoint incorporates both the pragmatic and the semantic aspects of the development of modality into a single explanation: that historical change is frequently the result of the gradual conventionalisation of conversational implicatures and pragmatic strengthening (Traugott 1988). Bybee, Pagliuca and Perkins (1994: 3) note the advantages of using a diachronic approach in the investigation of cross-linguistic grammatical categories, suggesting that it increases the explanatory power of linguistic theory to be able to attribute meanings associated with the synchronic forms of grammatical structures to factors of their development; that is, the specific lexical sources from which they were derived and the inferential mechanisms which were used to exploit those sources for an expanded range of functions. It is with respect to such diachronic developments that grammatical differences between languages can best be accounted for; similar approaches may therefore be applied to explain why a particular feature varies between L2 and L1 varieties of the same language. For this reason a panchronic approach based on grammaticalisation theory will be applied throughout the current study.
4
HYPOTHETICAL MODALITY
The present study proposes a novel approach to grammaticalisation theory in that it examines not the data from a single dialect of a language but the evidence from heterogeneous language sources; it is the study of grammaticalisation in a contact language which is affected by diverse influences from genetically-unrelated substrata but at the same time is not a true creole in the technical sense. In such a situation it will be seen that grammaticalisation progresses in ways similar to grammaticalisation in creoles, and that certain factors relating either to the grammaticalisation of L2 varieties generally or to the specific effects of the substrata will affect the rate and direction of grammaticalisation paths. The use of grammaticalisation theory for the analysis of contact language variation will achieve a number of clear goals. First, as noted above, diachronic approaches can be extremely powerful tools in explaining variational patterns present in the synchronic state of a language; and second, the data from variational studies as a basis for prediction will permit an empirical validation of such theories and an extension of the fundamental concepts associated with them. Apart from the advantage of using diachronic explanations to account for synchronic variation, another advantage of the present approach is that the synchronic variation exhibited in studies of L2 dialects can also reveal facts about grammaticalisation theory which would normally go unnoticed in studies based on only standard varieties of languages. Some of these facts will become evident throughout the study. The aims of the present study are therefore the following: in the first instance, it extends the recent studies of grammaticalisation in contact varieties (such as Baker and Syea 1996) by examining a situation of grammaticalisation in an L2 variety which has never had an official creole status; rather it was acquired principally through the education system. To this end, the study proposes a hypothesis to account for the more rapid rates of grammaticalisation found in such varieties, and to explain some of the constraints on grammatical usage found amongst L1 speakers. Second, it proposes to present an understanding of the meaning of hypothetical modality from the point of view of its diachronic development, and in terms of a phenomenon which is still grammaticalising at variable rates across different dialect groups of speakers. Such a phenomenon is shown to be motivated not by culture-related differences, but by the diversity of the language systems with which it is in contact. The study is not a contrastive one, but rather uses comparative crossdialectal resources to illuminate certain aspects of the nature of hypothetical modality which would be less easily perceived if examining only L1 material.
INTRODUCTION
5
1.2 Main points 1.2.1
Defining hypothetical modality
In Bybee, Pagliuca and Perkins (1994) and Palmer (1986), modality is defined as the semantic or conceptual notions expressed by formal features of mood. Since the distinguishing features of the subjunctive mood are almost obsolete in English now, the term ‘hypothetical modality’ will be used to invoke a range of concepts associated mainly with the use of a particular category of grammaticalising modal verb forms which largely have replaced the former subjunctive markers. In expressing hypotheticality, the modal verbs essentially express predictive meanings, but also have additional uses as deontic or root modals, related more closely to the meanings of their lexical sources. This class of verbs is well-known in the literature, and includes mainly forms such as should, could, may/might, and would. It is not within the scope of the present study to go into the finer explanations of the diverse uses of the class as a whole, but instead the study will focus more restrictively on the case of would as a marker of hypothetical modality.1 The essential nature of a predictive meaning as basic to the understanding of hypothetical modality is a functional explanation, and it is considered that different levels of factuality associated with linguistic meaning, such as open possibility, uncertainty, and impossibility, are all related to a single notion of prediction, but modified by the co-occurrence of contextual features. The complement of would contains a prediction, and as such, the entire clause containing the modal is represented as in a part-whole relation with factuality itself, since the forms which imply prediction, modals expressing ability, permission, obligation, and volition, are necessarily indexical to the actualisation of an event. It is such a representation that enables predictive modal structures, including conditional constructions, to be seen as candidates for Quantityimplicature scales, and also allows an explanation of the mechanism of Quantity implicatures in determining the meaning of would diachronically. 1.2.2
The role of lexical retention
The operation of grammaticalisation processes in determining the extent to which lexical retention is present is a factor of considerable importance to the present
1. For more explicit descriptive detail on the present-day functions of the other modals, the reader is advised to refer to Palmer (1986) or Coates (1983), amongst others.
6
HYPOTHETICAL MODALITY
study, as it will be shown that much of the L2 variation present as a synchronic phenomenon is directly related to the presence or absence of retention. Retention in grammaticalisation is usually seen to create constraints on the distribution of grammaticalising forms, or to evoke glimmers of vestigial lexical senses in areas of distribution otherwise considered to be semantically weakened, such as in the co-occurrence of the modal will with first person subjects, as demonstrated by Bybee and Pagliuca (1987). It will be shown that the development of hypothetical modal meanings is crucially affected by retention, and the differences occurring between L1 and L2 speakers in the expression of hypothetical modality can be found to correlate with the presence or absence of retention as a factor contributing to the creation of such meanings in their dialects. By referring to such processes, what may at first appear to be the product of random ‘learner’s errors’, or ‘overgeneralisations’ can be shown to have a much more far-reaching explanation in view of the differences between the diachronic routes of grammaticalisation respective to L2 varieties and those of L1 varieties: the overgeneralisations of L2 speakers, then, reveal the environments as yet unaffected by grammaticalisation in L1 varieties. Therefore, the use of L2 data contributes a highly significant perspective on the working processes of L1 grammaticalisation, by providing a comparative framework against which to observe its progress and to enable predictions to be made of the likely future directions that the grammaticalising L1 items will take. 1.2.3
The hypothesis
It should be emphasised that the present work is not intended as a descriptive account of dialectal variation, but that the Singaporean data presents a test case for a model on which future hypotheses referring to grammaticalisation in L2 situations could be based. One of the questions which arise from investigating such situations is how to explain L2 overgeneralisations in terms of comparative L1/L2 retention patterns. Central to the theoretical basis of the study is the proposal of the Lexical Memory Traces hypothesis, which is developed from comparisons of the L2 data in previous chapters, first language acquisition studies and the role of lexical sources in the development of grammatical categories. The hypothesis will use data from quantitative studies as a basis for a prediction regarding the L2 variation. The variation is seen as attributed to lower levels of lexical retention in the L2 speakers’ dialect, and it is questioned why the evidence of prior diachronic stages in the history of the language, reflected in lexical retention, is less apparent in the dialects of L2 speakers than in those of L1 speakers. This will be explained in terms of the parallels observed
INTRODUCTION
7
between the diachronic processes of grammaticalisation and the processes of grammaticalisation in the individual’s linguistic development. The hypothesis is supported by evidence from studies of children’s utterances, which, viewed in the context of such parallels, reveal close similarities in the nature of early acquisitional routes to the lexical sources of diachronic patterns. However, at the same time it must be emphasised that the parallels are not interdependent, and may not be remarkable: the role of diachrony does not in any way shape the course of ontogenetic grammaticalisation, and the parallels may be simply related by the fact that concrete concepts always form the basis for the extensions of meanings associated with the more complex processes of grammaticalisation, whatever the context. The facts about language development, manifested in two different domains, are presented in the hypothesis not as an end in themselves, but as a diagnostic tool for explaining the lower rates of lexical retention in the dialects of L2 speakers. The grammaticalisation of certain features in L2 dialects may be predicted to be accelerated in comparison with the processes experienced by a native speaker, since L2 grammaticalisation processes are shown as less likely to be affected by the retentive constraints of Lexical Memory Traces from early acquisitional stages.
1.3 Overview Diachronic explanations are based on observations of the use speakers make of a language at various times in its evolution. Grammaticalisation considers language evolution to be represented in a long series of overlapping pragmatic changes in the use of a grammaticalising item; for this reason, it will become clear that a pragmatic interpretation of hypothetical modality in its present-day usage is most admissible to the current hypotheses, rather than one which relies on the application of a logical apparatus. In Chapter 2, the role of hypothetical modality in conditional sentences is discussed as a prototypical model for most other environments in which it occurs. It is questioned whether there actually exists a metaphorical explanation of hypothetical modality in which the remoteness of fact is viewed in terms of the remoteness of time (as proposed by many accounts, including Hacking 1998; Fleischman 1989; and James 1982), when it is actually proximity to fact, not remoteness from it, which is inferred from reference to events occurring before the moment of speaking. The interpretation of a counterfactual conditional, defined at first by the presupposition of falsity in the antecedent clause (Lewis 1973), is considered in the light of the hearer’s inferences and the factors in the grammatical and/or contextual setting which
8
HYPOTHETICAL MODALITY
combine to contribute to the creation of that presupposition. Accounts which attribute the interpretations of counterfactuality to the operation of a conversational implicature (e.g. Comrie 1986) are also discussed, and the importance of pragmatic strengthening of a conditional to a bicondition (Geis and Zwicky 1971) is referred to in relation to conversational implicatures of causality relations between the two conditional clauses. Ultimately, it is seen that a principle can be isolated which analyses the range of grammatical morphology most often associated with hypothetical notions as related to a precondition of information specificity in the context of the counterfactual utterance, and this principle may be seen to have a crosslinguistic application. Traugott (1989) referred to the role of conversational implicatures in the development of epistemic modal meanings, and attributed this to an exploitation of Grice’s (1975) conversational Maxims. Following Sperber and Wilson (1986), Traugott uses a general term, Relevance, to account for the meaning extensions from deontic to weakly epistemic to strongly epistemic, but notes that this term is related to Grice’s lower-bounding second maxim of Quantity, and Horn’s (1984) R-Principle (which covered the maxims of Relevance and Manner, as well as Quantity 2). In the present study, the Gricean Maxim of Quantity 2 will be referred to, and in Chapter 3 it will be seen as instrumental in explaining the presence of conversational implicatures in the development of hypothetical modality. As noted earlier, the operation of the Quantity 2 maxim enables the inferences of prediction or future-projecting senses to be derived from the use of the modal verbs, which, since they developed originally from senses of power, permission, ability, and volition, refer to states which anticipate or are precursors to future action, and can metonymically imply the realisation or the future actuality of their predicated events (see Panther and Thornburg 1999). The mechanisms of Q2 implicatures can be seen also to affect the development of past tense forms of the modals, and as a diachronic process, in which the inferences obtained (prediction made relative to a past time point of reference), can be modified according to the type of clause in which the modal appears: factive or non-factive. Thus, it is not the modal itself which develops hypothetical meanings, but the modal in combination with its grammatical and discourse environment. Environmental factors therefore also contribute to the development of counterfactual meanings: however, in counterfactuals, these cannot contribute to the positive prediction of the modal complement, but predict instead a negation of the proposition contained in the complement. The resulting implicature is a Quantity 1 implicature, not a Quantity 2. The Quantity 1 implicature, though, is not considered as basic to the meaning of the modal, but rather derived from the presence of cancelling environmental elements such as contrastive clauses.
INTRODUCTION
9
As well as Quantity implicatures which influence the development of hypotheticality as a predictive sense in the modal predicate, the modal verb itself is seen to carry inferences of hypotheticality, as discussed by Bybee (1995). In this respect, the modals will and would can be differentiated in terms of modality, and not temporality, since would has long since lost any basic temporal meanings.2 The hypotheticality of modals such as would is claimed by Bybee to be related to the completion or non-completion of the action described in the modal predicate, since the modals are stative verbs. However, the same hypotheticality is found in non-modal stative verbs, as an inference deriving from the conflict between inherent lexical aspect (or Aktionsart) of stativity, and grammatical tense. If past tense developed from earlier perfective senses, as hypothesised in Bybee, Pagliuca and Perkins (1994), and applied then only to non-stative verbs, initially the meanings of perfectivity associated with non-stative uses may be retained, but may be represented as pragmatic inferences in stative situations, which cannot be inherently terminative. Non-modal statives do not necessarily predicate verbal complements, e.g., the copula was, and yet they still may produce inferences or implicatures of hypotheticality with regard to the continuation into the present of the past state they describe, e.g. he was an academic (and, surprisingly, he still is), in which the normal inferences of discontinuity of the past state are cancelled by the additional information that follows. The problem for this analysis is that in the dialects of L2 speakers whose first language does not mark stative verbs for past, i.e., has not yet grammaticalised a past tense category, there may be substratum influence which will affect the marking of past statives in the L2. This is exactly what has happened in the case of Singaporean English, and it is hypothesised that the less frequent occurrence of past tense marking on stative verbs and predicates in Singaporean English, as statistically quantified by Ho and Platt (1993), affects the development of hypothetical implicatures in the interpretation of past stative verbs by Singaporean speakers. This evidence is crucial in explaining why speakers of L1 Chinese may not always distinguish different levels of factuality when presented with only the linguistic indicators of its presence: a past modal verb, as shown in Chapter 5. The evidence of the substratum patterns demonstrates that there may be little differentiation for such speakers on the basis of modality between modal verbs such as will and would, and can and could, which appear to be used indiscriminately with regard to the factuality of the environment, as shown in previous studies.
2. Bybee (1995) notes that the only remaining temporal uses are in past habituals.
10
HYPOTHETICAL MODALITY
In Chapter 4, some of the earlier research on such features in Singaporean English will be reviewed, and a review of some of the past literature on the expression of hypothetical modality in the substrata will also be provided, since it may be suggested that substratum influences will affect the grammaticalisation of hypothetical modality in the L2. In addition, the chapter will include a brief introduction to the sociolinguistic status of Singaporean English. It is seen that even given the diversity of the background languages spoken in Singapore — Malay, Tamil, and a range of around 10 Chinese dialects are only the major language groups — there is nevertheless considerable uniformity in the L2 patterns for expressing hypotheticality. This is most likely because the background and contact languages generally do not have a fully grammaticalised category of past tense. Chapter 6 describes research into the presence of lexical retention using comparative data from L1 and L2 subjects. The subjects assess the acceptability of a number of sentences using the modal would in hypothetical complements of the verb wish. The sentences are varied according to co-occurrence features such as the presence of a stative predicate, animacy of the complement subject, negation, aspect, and cross-clause co-reference of the subject; for example: I wish the Porsche would belong to me (with an inanimate subject and a stative predicate). Acceptance rates are related to the presence or absence of meanings of volition in the modal; that is, to the retention of meanings relating to earlier stages in the diachronic development of the modal. It is shown that retention exhibited in this way as a synchronic phenomenon is less apparent in the intuitions of L2 speakers than in L1 speakers, regardless of regional differences (L1 data is taken from speakers of both Australian English and British English). Retention in modal meanings is therefore considered to be significant to the development of hypothetical interpretations amongst both L1 and L2 speakers, and to the generalisation of the modal’s distribution. In addition, the evidence of lower rates of retention perceived by L2 speakers in the meaning of the modal would can be correlated with the assumptions made in Ch. 5 that there will be less semantic salience for such speakers in the combination of past tense with a stative predicate. The differences in the retention rates between L1 and L2 speakers shown in the studies in Chapter 6 also illustrate, conversely, that the grammaticalisation rates for the L2 speakers are much higher and more rapid than for the L1 speakers. Since retention is as much a diachronic phenomenon as a synchronic one, such evidence may suggest that the L2 speakers have less access to the historical facts regarding the development of the language than L1 speakers, and, because of this, are disregarding the constraints imposed by historically earlier
INTRODUCTION
11
lexical co-occurrence restrictions. In order for retention represented in the individual’s intuitions to be described as a synchronic phenomenon, it must be seen as relevant to the diachronic stage at which the grammaticalising feature is currently situated. This presents a paradox, in that the individual speaker is unaware of the developments which lead to the present stage of grammaticalisation, and it is therefore questioned how a respondent can access the information regarding the diachronic development of the feature in his or her semantic intuitions. The question of accessing diachronic information is therefore implausible. The hypothesis offered is that the individual’s representations of retention instead are a product of earlier acquisitional stages in the grammaticalisation of the feature as an ontogenetic factor, which for independent reasons often run parallel to diachronic routes; and it is memory traces of such earlier stages (Lexical Memory Traces) which continue to influence later assessments of the distributional range of the grammaticalising feature. Earlier stages of acquisition are therefore seen to reflect earlier stages of diachronic development in the language across a community of speakers. The Lexical Memory Traces Hypothesis, while proposing that the parallel ontogenetic and diachronic routes of grammaticalisation are not causally related, does not indicate that the information from earlier diachronic stages is not of any significance at all. The level to which the language has grammaticalised synchronically will always be represented in a community of speakers as the same level of its diachronic realisation. A child whose acquisitional routes vaguely follow the same paths of development as the historical routes display is naturally unlikely to reach more advanced levels of grammaticalisation in her acquired dialect than in the previous generation. Innovations and extensions to new environments are more likely to appear in the adult language as slowly-developing inferences from more concrete meanings (as proposed by Slobin 1994). The diachronic paths of development may therefore not be important in discussing the psychological representation of lexical retention, but they are important in evaluating the level of the grammaticalisation of a certain feature retrospective to its past uses, and are useful in making predictions of future directions based on comparative evidence from other languages. In the present study, such predictions can be made on the basis of data from L2 speakers, which reveal in the nature of overgeneralisation the areas which have not yet been affected by the L1 grammaticalisation of the same feature. The lower rates of retention for the L2 speakers is hypothesised to be the result of differences in the acquisition process. In addition, though, for L2 speakers for whom the language is not a mother tongue, the need for greater efficiency of communication applies pressure to increase the distribution of certain grammaticalising features to generalise to
12
HYPOTHETICAL MODALITY
new environments at a faster rate than for L1 speakers. Whether this occurs in the current generation of speakers or is the fossilised result of overgeneralisations from earlier generations of speakers is not important; however, it is proposed in the present study that such overgeneralisations occur as the result of condensing or even ‘by-passing’ the earlier lexical stages of acquisition, and thus progressing more rapidly to the grammaticalised (bleached) stages in the need to make maximum use of a more limited inventory of available resources. It will be seen that the Lexical Memory Traces (LMT) Hypothesis cannot be applied in the same way to all cases of retention. The retention of senses of perfectivity in past tense stative predicates, discussed in Ch. 5, was found to be related to earlier stages of the acquisition of past tense in L1 dialects, but the lower rates of such retention cannot be attributed to a by-passing of such stages in Singaporean English. The data reflect the fact that the implicatures created by the retention of prototype meanings of perfectivity will not occur at all for some speakers: for such speakers the salient semantic information on the use of a verb form will be in the lexical aspect of the verb, not in the tense marking. Interference from substratum and contact languages (e.g. Chinese), in which stative predicates are not marked for perfectivity, affects the interpretation of past stative verbs in English for some speakers. The result is undergeneralisation, due to relative ‘hypo-grammaticalisation’ of past tense in such dialects, not overgeneralisation, represented as ‘hyper-grammaticalisation’. In the use of would, however, the lower incidence of retention for the L2 speakers results in overgeneralisation, not undergeneralisation of the grammaticalising feature, and it is such instances of overgeneralisation which may be explained by ‘bypassing’ or by more rapid progression through the earlier lexical stages of grammaticalisation, as outlined in the hypothesis. Thus, the study suggests in the first instance an explanation for hypothetical meaning as the result of pragmatic implicatures associated with verbs of volition, and illustrates the diachronic role of such implicatures in the developing modal meaning of would. It is shown that implicatures are also created by the conflict of grammatical tense with lexical (and grammatical) imperfective aspect, as tense appears to have developed diachronically from perfective aspect and still retains a pragmatic sense of perfectivity relating to earlier uses. Since volition verbs are lexically imperfective, being stative verbs, they are affected by the implicatures of hypotheticality that are created in such situations. Furthermore, there are additional implicatures created by the lexical senses of volition and intention that are retained in the meaning of the modal would, and which intensify the hypothetical implicatures resulting from the aspectual conflict. The use of L2 data illuminates this explanation, as it is seen that the retention of lexical senses of
INTRODUCTION
13
volition in the modal are less apparent for such speakers, and that past tense has not become generalised to stative or imperfective environments. The LMT Hypothesis, which explains distributional variation in synchronic stages of grammaticalisation, is extended to account for the variation in retention between L1 and L2 speakers. The study also presents a theoretical approach which encompasses two domains of research, and which, to present knowledge, have not yet been combined in such a way in earlier accounts. On the one hand, it offers a functional description of the representation of hypothetical meaning in the form of modality, drawing on the mechanical processes of grammaticalisation theory and the cognitive practices of pragmatics to explain hypotheticality from a synchronic as well as a diachronic perspective. While the meanings expressing hypotheticality remain relatively constant over time and across entire ranges of language families, the formal expression of such meanings is likely to exhibit shifting patterns of use in accordance with other diachronic changes occurring in the language at the same time. Therefore, the study also develops the description of hypothetical modality to locate it within a framework which has been so far relatively unexplored territory for grammaticalisation studies, that of creole-based and contact languages. The contrastive data from speakers of L2 varieties, which have a discontinuity in transmission, are invaluable in bringing to light the patterns of change contributing to the development of grammatical meanings in standard counterpart varieties with a continuous historical development. Thus, as well as presenting a new perspective on grammaticalisation processes in general, studies of the grammaticalisation of modality in L2 varieties may assist considerably in building a deeper understanding of hypothetical modal meanings based on diachronic factors. It will be revealed that a function-based, diachronic explanation can provide the most satisfactory means of explaining hypotheticality as part of a linguistic system in its present-day uses.
C 2 Hypothetical modality as a grammatical category
An investigation into the grammaticalisation of hypothetical modality in a secondlanguage English dialect cannot proceed without some definitions and concepts being initially clarified with respect to L1 varieties of English. Since a discussion of hypothetical modality cannot avoid reference to conditional constructions, the present chapter will examine such environments, looking at the extent to which hypothetical meaning in English is grammaticalised in the use of a particular structure, and, if it is not associated with a form-function assignment, how the grammatical meaning of hypotheticality may be otherwise determined. It will be argued that a collective ensemble of grammatical and/or discourse features all contribute to the derivation of hypothetical modal meaning, and that the operation of Gricean pragmatic principles underlies the basic interpretation of all hypothetical meanings. It is important to note that the term ‘hypotheticality’ has been adopted as a general cover term, following Comrie (1986), for describing meanings which are both counterfactual and those which are not counterfactual but share a common grammatical form with counterfactual expressions. That is, although it is generally accepted that in English the presence of a modal perfect form in a conditional apodosis (would have +V-ed), combined with a pluperfect form in the protasis, is most often indicative of a counterfactual conditional, such forms may also be used to express a past hypothesis about which the speaker is uncertain of supporting facts. This type of conditional can be described within the subcategory of ‘Hypothetical’ conditionals, which usually refer to those of non-past time reference, but use the simple past tense in the protasis and would + V in the apodosis. In other words, lower-case ‘hypothetical’ will refer to the general category of all ranges of hypothetical meaning, including counterfactuality, which is the highest level, while upper-case ‘Hypothetical’ will refer to a specific subcategory of hypothetical meaning in which the speaker is uncertain or unaware of the facts surrounding the utterance of a hypothesis. These conditionals can be shown to be independent of any particular time reference. The term ‘counterfactual’, nevetheless, is defined broadly for the moment as
16
HYPOTHETICAL MODALITY
meaning expressed when the speaker is aware, in making a counterfactual utterance, that the underlying proposition is contrary to known facts. Such a suggestion may lead one to believe that counterfactuality is equivalent to lying, but, while lying is a deliberate violation of Grice’s (1975) Quality Maxim, with the intention of concealing the contravened known facts from the hearer, the expression of counterfactuality permits the hearer to infer the negation of reality from the expression of only partial reality; the means by which this is inferred is from the structure of the utterance, from world knowledge about a situation, or from various discourse clues. Furthermore, lying involves false assertion, but the illocutionary intent of a counterfactual utterance is to express a (qualified) prediction, which is an irrealis statement and therefore not a true assertion (according to Bybee 1998, assertions only involve realis statements). Thus, only the speaker is, strictly speaking, aware of the counterfactuality of the utterance; the hearer must deduce it from the sum of the grammatical indicators and contextual clues. In the following review, therefore, the nature and form of the grammatical indicators usually associated with counterfactuality will be examined, in order to determine whether they can be assigned a predictable, one-toone relationship with a particular level of hypotheticality. The following account will also consider the correlation of temporality with levels of hypotheticality, and it will be seen that, although past temporality does increase the likelihood of a counterfactual interpretation, there is no reason to assume that the two notions are interdependent. There is, however, reason to believe that past and perfect morphology is inextricably linked to crosslinguistic notions of high hypotheticality, and that this is one of the most important determining factors of counterfactual inferences. Hypothetical modality will be seen then, at least when represented in conditional constructions, as having not so much a one-to-one relationship between a given form and an associated level of hypotheticality, but rather as made up of a cluster of features all contributing to the hearer’s evaluation of the level of hypotheticality; that is, as an inferred meaning. The more features that are present in the utterance, the more likely a counterfactual interpretation will ensue. The presence of such features will be discussed in 2.1: Section 2.1.1 will examine the feature of past time reference in conditionals and the question of the metaphorical representation of remoteness in past temporal morphology. In 2.1.2 the notion of a causal protasis will be reviewed as the result of conversational implicatures, and in 2.1.3 the features of biconditionality, negation and first person subjects appearing in hypothetical past conditionals will also be discussed. In 2.2 will be proposed a principle explaining the reason that the particular cluster of features are able to contribute to counterfactual interpretations in the utterances in which they appear.
HYPOTHETICAL MODALITY AS A GRAMMATICAL CATEGORY
17
2.1 The grammatical derivation of hypothetical meaning 2.1.1
Temporality and hypotheticality
For a traditional definition of counterfactuality in conditional constructions, we can assume that it will obtain when the truth value of the proposition contained in the protasis, or antecedent clause, is presupposed to be false. This is the benchmark used by most analysts to identify a counterfactual conditional, with the form of the protasis verb being in the pluperfect (had + V-ed) and the apodosis clause, or consequent, containing a modal (usually would) + have + Ved. Although, as suggested above, the concept of counterfactuality in the present study will be regarded as loosely associated with temporality, the majority of accounts stipulate that the usual temporal reference for a counterfactual is assumed to be past, since we are more capable of determining the truth or falsity of a proposition in the past than in the present or the future. Thus, the definitions proposed by Quirk, Leech and Svartvik (1987: 782) are not neutral to time distinctions, and furthermore, do not use the term ‘counterfactual’ but rather ‘hypothetical’ to indicate an implied rejection of the conditional premise, e.g. (1)
If you had listened to me, you wouldn’t have made mistakes (‘but you didn’t listen to me’).
They distinguish between hypothetical meaning in the past and in the present or future, where the meaning may simply be one of negative expectation, e.g. (2)
If you listened to me, you wouldn’t make mistakes. (‘but I don’t suppose you will listen to me’)1
With regard to present terminology, (1) is usually considered to be counterfactual, and (2) would be Hypothetical. However, it could be noted also in (2) that the conditional may also be suffixed with ‘… but you don’t listen to me’ (habitually), suggesting again an implied rejection of the conditional premise, and a time reference which is neither in the future nor in the past, but in the habitual present. Similarly (1) could be suffixed with ‘… but I don’t suppose you did listen to me’, indicating that the past hypothesis is just a surmise, and is not backed by any known fact. Thus (1) could be a Hypothetical conditional and (2) could well
1. The conditional types described in traditional grammars, of course, are only the prescribed forms for representing different levels of uncertainty. Actual data from native speakers shows a much greater variation in the types of conditional constructions selected (e.g. Werth 1997), and often there will be a ‘blending’ of the clause structures seen in (1) and (2).
18
HYPOTHETICAL MODALITY
be counterfactual. It should be noted here again that the definition offered by Quirk et al. states that there is an implied rejection of the antecedent clause proposition in hypothetical conditionals, not an asserted one. The present study considers Hypotheticality or counterfactuality to be a basic implicature in the understanding of such constructions, and proposes that all counterfactual meaning is pragmatically implied. Therefore, the implicature is that the speaker’s intent was to express an assertion about known facts, but the form of the speech act is only a predictive statement, as noted earlier. What is not so obvious is the nature of the linguistic devices that contribute to the creation of a Hypothetical or counterfactual implicature, and whether or not they can be isolated. The explanations offered by Quirk et al, although probably representative of most grammatically-based opinions, might not appear to adequately account for all the distinctions of temporality and modality which can be expressed by these constructions. The examples above indicate that the correlation of counterfactual or Hypothetical meaning with temporal reference is clearly underspecified in the grammatical morphology used, and appears to be based solely on matters of contextual interpretation. Counterfactuals which usually refer to past time, such as exemplified in (1), can even suggest the counterfactuality of future events, given prior knowledge of a contrary set of circumstances. This is illustrated by Dudman (1984: 149) with the (rather morbid) example illustrated in (3). The speaker could utter a conditional of this kind in full knowledge that ‘Grannie’ had just died, and the Friday being referred to was following the moment of speaking: (3)
If Grannie had missed the last bus on Friday, she would have walked home.
In (3), the counterfactuality of the statement is unrelated to the temporality which is in the future, suggesting that the hypothetical inferences originally conveyed in the past tense morphology have strengthened to the point at which past time reference is no longer required to convey them. We shall label this type of conditional Type A for further reference, as it indicates a special case in which the temporal meanings have been overshadowed by the hypothetical meanings. There is little doubt of the implied falsity of the protasis in such an example, given the background knowledge of the speaker, and also the strength of the assumption that Grannie would attempt to catch the last bus if she were still alive to do so. But the requirement is that this assumption accompanies the knowledge of her having died, as extra-linguistic contextual information. Dancygier (1998: 43) agrees that the use of the ‘back-shifted’ forms in conditional
HYPOTHETICAL MODALITY AS A GRAMMATICAL CATEGORY
19
constructions is related to degrees of assertability rather than temporal reference. (The term ‘back-shifting’ is avoided in the present study, as it seems to suggest that the diachronic facts regarding the evolution of hypothetical conditionals have been disregarded — see Ch. 3.) It appears then, that the past temporal meanings normally associated with counterfactual conditionals have been weakened (as also noted by Dahl 1997: 109)), and that an inference arising from the notion of prediction relating to past time, ‘counter-to-fact’ prediction, emerges to stand for the entire meaning of past + prediction. This represents, effectively, a metonymic extension, possibly indicating by the lifting of the normal constraints of past time reference a more advanced level of grammaticalisation. However, the relationship between past time and hypotheticality has often been assumed to be a metaphoric one, as discussed below. 2.1.1.1 The remoteness metaphor Given contexts such as that represented in (3), some accounts have, in fact, related the sense of factual remoteness derived from hypothetical expressions to the temporal remoteness expressed by past tense inflections in general. Huddleston (1989: 148) suggests that past tense in hypothetical conditionals indicates factual remoteness, and that features in the contextual setting of an utterance indicate whether this remoteness is temporal or factual. Lyons (1977: 809) raised the problem of truth or factuality usually being treated by logicians as timeless, or tenseless, and proposed that what may be regarded as tense distinctions (i.e. past versus non-past) may be better regarded as a distinction between remoteness and non-remoteness, so that tense would be classed as a kind of modality and associated with deixis; by this account he was suggesting that past tense is the product of remoteness and ‘factivity’ (1977: 819–20). Palmer (1986: 209–10) regards past tense as most closely related to modality, and in particular, to unreality. This assumption is based on the remoteness of past time from the moment of speaking and the exploitation of past tense for metaphorical analogies with factual remoteness; it raises the question whether or not future is equally remote from the speaker’s present. However, Palmer claims that the future could not be used in unreal conditions as it is already modalised and “… there would be a confusion of senses if it were used.” (1986: 213). Factual remoteness is already a feature of future meaning, but future suggests more uncertainty of meaning than the past or present about which it is possible to know facts. Accordingly, Lyons (1977: 820) maintains that the past is most often regarded as closed, so that statements about the past may have an inherent truth value, whereas the future is open and contingent, and statements about the future
20
HYPOTHETICAL MODALITY
are indeterminate as to truth value. For this reason, it is pragmatically more appropriate for a speaker to form hypotheses about the past, or hypotheses based upon past knowledge. The argument of ‘remoteness’ in time as a general indicator of factual distance has been proposed in a number of other accounts, including Hacking (1998), Fleischman (1989 and 1995) and James (1982). James in her study examines six Indo-European languages and six non-Indo-European languages and finds that in all the languages in the sample the past tense is used in at least the result clause of counterfactual conditionals (1982: 382). She maintains that the presence of past-tense marking in counterfactual sentences is to indicate the degree of distance from reality, and that this meaning derives from the fact that it is used when the time reference of the conditional is past, present, or gnomic, since it is with reference to such situations that we know sufficient factual information to present a contrary-to-fact statement (James 1982: 377). However, our factual knowledge about a situation is an indication of proximity to reality, not distance from reality, though the hypothetical utterance which is founded on such factual knowledge is certainly distant from reality. It appears, then, that past temporal marking is a requirement of expressions of counterfactuality, but the common link of remoteness between past and irrealis is not so clearly definable. The relation of temporal distance to modal distance is also described as a metaphorical device by Fleischman (1989), who suggests that the direction of the metaphor is from spatial and temporal distance or proximity to abstract conceptual or cognitive distance or proximity. Fleischman builds on James’ (1982) claims in suggesting that not only can factual distance and past tense be correlated, but that degrees of distance and pastness are also found to be linked (1989: 7), so that, as expected, higher levels of hypotheticality may be found in Romance conditional protases using the pluperfect. However, the problem with a metaphor explanation is that it does not allow for possible ambiguity between the source and the target meanings: a pluperfect conditional construction may be equally likely to express a past temporal hypothesis about which no facts are known at all (a Hypothetical conditional), and the only thing which is certain is the dependency relation between the two clauses. In this respect, a pluperfect conditional would not be any more remote than a conditional referring to a future possibility, e.g. (2) (also Hypothetical). Furthermore, examples like (3) appear on the surface to represent a metaphorical association between remoteness and past time reference, since the temporal meanings are bleached. But the most frequently implied meanings of counterfactuality are associated with pluperfect conditionals because past time reference enables us to make the strongest evaluations of the truth or falsity of a prediction; that is, the inferences of counterfactuality are
HYPOTHETICAL MODALITY AS A GRAMMATICAL CATEGORY
21
present and co-exist with past temporality right from the start, as discussed in 2.1.1, and therefore must be metonymical extensions of past predictive conditionals. As such, they would represent a gradual development, not typical of the sudden changes associated with metaphorical extensions. Fleischman also notes that the use of the temporal distance metaphor may appear crosslinguistically as a marker of evidentiality, speaker subjectivity, or a politeness marker, to express the social distance of the speaker from the utterance and to reduce the directness of requests; e.g. (4)
a. b.
Would/could you do me a favor? I thought/was thinking about asking you to dinner. (Fleischman 1989: 8)
Both of these examples use the marking for past to assert metaphorical social distance from the addressee. In a later work (1995), she notes that the use of the progressive in (4b) is more typical of such requests than the use of a simple verb form, and that there is a general preference in Romance languages for the imperfective aspect to express such utterances. The reason for this is that the use of the imperfective aspect does not predict the completion of the event being referred to, and implies to the addressee the continuation of the past state, giving the addressee the possibility of responding (Fleischman 1995: 52). This situation also cannot be regarded as metaphorical — a past state implying a present state is a hypothetical inference which may be seen to gradually evolve from the absence of particular lexical aspectual specifications of perfectivity in the verb. The implications of deriving modal senses in such ways are pragmatically quite complex, and involve a deeper explanation which will be discussed in greater detail in Ch. 5. Dahl (1997) rejects the temporal metaphor hypothesis, suggesting that it is rarely the case that past tense alone serves to mark hypotheticality in conditionals, but that the meaning is derived from a combination of the past tense and another feature, which in Romance languages is the imperfective aspect (in the protasis), and the conditional, or historical imperfect of the future (in the apodosis). Examples of such a combination appear in Fleischman (1995: 523): (5)
a. b.
Si j’avais le temps, je t’écrirais (non-past) ‘If I had time, I would write to you.’ Si j’avais eu le temps, je t’aurais écrit (past) ‘If I had had the time, I would have written to you.’
(5a) shows the imperfective in the protasis (avais) and in (5b) using the pluperfect (avais eu) ‘had had’, which is comprised of the (past) imperfective form of
22
HYPOTHETICAL MODALITY
avoir ‘have’ plus the past participle. Dahl notes that this use of the imperfective is often accompanied by a loss of aspectual senses, as in many cases the imperfective appears where a perfective past would normally appear. Dahl (1997: 101) also notes the presence of a modal denoting future time to be found frequently in a conditional apodosis, suggesting that it should be possible to compute the meaning of a counterfactual conditional from the meaning of its constituents, using the Tedeschi (1981) ‘branching futures’ model, which describes counterfactual meanings in conditionals as predictions made relative to a past point of reference. Givón (1994) also finds that crosslinguistically, counter-to-fact conditionals are a blend of realis (using past morphology) and irrealis (using modals, futures, or if-type conjunctions); Bybee (1995) discusses the frequent occurrence of past-futures in conditionals. It will be seen in Ch. 3 that, historically, Hypothetical and counterfactual conditionals appear to have evolved in the manner described by Tedeschi’s model. 2.1.2
Conditionals and conversational implicatures
As noted above, the usual definition of a counterfactual conditional is that the proposition contained in the protasis or antecedent clause is assumed to be false. This definition is basic to all accounts including those discussing negative epistemic stance (Fillmore 1990), and correlations between mental spaces (Fauconnier 1985; Sweetser 1996) or possible worlds (Lewis 1986). Such accounts are based on the construction of analogical conceptual domains in which a counterfactual prediction is possible, and are concerned mainly with the propositional content of a conditional. The present account instead attempts to explain and analyse the grammatical devices which contribute to the construction of an analogical counterfactual or hypothetical world. In doing so, it is assumed that the ability for the participants in a speech situation to construct such worlds is due to pragmatic inferencing procedures, and such procedures are arrived at by means of not only structural and discourse devices, such as the presence of past or perfect grammatical morphology and the use of conditional conjunctions such as if, but also clause-linking inferences. In many accounts such inferences have been considered to be conversational implicatures. Lewis (1973: 3) accepts the likelihood that the notion of the falsity of the proposition in the conditional protasis is a presupposition, but goes on to suggest that this may be merely a conversational implicature and therefore unrelated to any truth conditions at all. One characteristic of a conversational implicature is that it is an inference which can be cancelled without contradicting the utterance
HYPOTHETICAL MODALITY AS A GRAMMATICAL CATEGORY
23
from which it is derived.2 The hearer assumes a meaning over and above the meaning provided in the semantics of an utterance; thus, a cancellation of such an assumption does not affect the semantic content of the utterance itself. Lewis also claims that the truth value of a counterfactual may depend on contingent, empirical facts, not necessarily known to the participants in a conversation; furthermore, a counterfactual requires a factual premise or argument with which it should be backed (1973: 68–9). Thus, in the example (6)
If I had looked in my pocket, I would have found a penny
the truth value of this conditional, in order to be accurately determined, depends on whether there was a penny there in the first place. However, the use of such an example is a matter of the sincerity with which the speaker is expressing his or her own beliefs, as Lewis maintains when he suggests that it is a mistake to use a counterfactual conditional unless the antecedent is taken to be false by the speaker (1973: 3). That is, to do otherwise implies a contravention of the Cooperative Principle of Grice (1975). This also suggests that the meaning is determined by pragmatic means, and that there is no recourse to actual truth conditions in interpreting a counterfactual conditional, but only to the evaluation of truth as provided by contextual indicators. It will be shown later that the evaluation of the truth supporting a hypothetical utterance can in fact be taken as a measurable factor. Davies (1979: 158) suggests that the fact that a counterfactual makes a hypothesis about a past situation allows for knowledge of the facts surrounding the situation and this is why it is possible to argue from knowledge of the antecedent proposition to that of the consequent clause for the truth of the propositions concerned. However, he maintains that the grammatical construction is not the means by which a ‘contrary-to-fact’ sense is normally conveyed, but rather it is certain aspects of “common knowledge” relating to the content of the propositions in the conditional sentence. The speaker may use the grammatical form of a counterfactual conditional when neither the speaker nor the hearer know anything about the events presented in the protasis as well as when they do. He illustrates this point with the example (1979: 158): (7)
If John had been at the scene of the crime when the murder was committed, Mary would have seen him leaving. So we must get hold of her to find out if she did see him.
2. Presuppositions can also be cancelled, as noted in Gazdar (1979).
24
HYPOTHETICAL MODALITY
The ‘contrary-to-fact’ sense does not obtain in this conditional, since the sentence which follows presumes no factual knowledge about the propositions presented in either clause of the conditional, and acts to cancel the implicature of counterfactuality provided by the content of the protasis. It is merely a Hypothetical conditional, the epistemic judgement of the speaker on the truth value of either of the clauses does not affect the conditional dependence between them, which exists outside of the limitations of any truth values relating to the propositions contained in them (1979: 159–161). I have labelled this type of conditional Type B: this represents a case in which the temporality usually associated with counterfactual conditionals is preserved but the hypotheticality is weakened, in contrast to Type A (illustrated in (3)) in which the hypotheticality is strengthened but the temporality is weakened. Davies’ claim regarding the interdependence between the two clauses is supported by Palmer (1986: 189), who contends that the truth of the information in the protasis or the apodosis is not relevant to the meaning of the conditional, but that a conditional functions only to express a dependency of the truth of one proposition on another, whether its factual value is known or not. Palmer (1990: 170) regards this type of conditional as causal or predictive, and he also includes past time conditionals as predictive. Thus in the example (1990: 170): (8)
If John had come, Mary would have left
there is prediction about the hypothetical event of Mary’s leaving in the past, contingent on John’s coming. Palmer grants that the usual interpretation of such sentences is counterfactual, implying rejection of the information in the protasis, but he allows for the fact that not all past ‘unreal’ conditionals can be interpreted in this way. The following example: (9)
If John had come, Mary would have left. Therefore, since we now know that John came, we know that Mary left.
illustrates the point that in certain contexts, when the truth value of the information in the protasis is not previously known, a conditional such as this may be used to express uncertain past prediction. However, Palmer concedes that there is little point in making predictions about the past when the facts are already generally known (1990: 171).3 Past is thus an unlikely context for the generation
3. A similar analysis can be given to conditionals in the present (since we now know that John is coming, we know that Mary will leave) (Keith Allan, p.c.), showing that the truth or the falsity of the prediction is not associated with time reference. However, a prediction about a past event is most likely to yield a counter-to-fact interpretation because of the association of past and realis.
HYPOTHETICAL MODALITY AS A GRAMMATICAL CATEGORY
25
of predictions, but when predictions are made about the past, they are assumed to be made on a reliable information basis. A hypothetical conditional about the past such as (9), then, is less likely to be interpreted as non-counterfactual out of context. It is for this reason that a cancellation, such as in (9), is needed to reveal weaker, (non-counterfactual) implicatures. It might be argued at this stage that a form does exist in English for distinguishing past hypotheses of unknown truth value from those about which the facts are already known, and this conditional uses the past indicative in both clauses. However, Palmer defines a functional distinction between the past indicative conditional and the past counterfactual: the indicative does not seem to express a sense of causality and prediction, but the conditional dependence in this type seems to be based on mere inference alone (Palmer 1990: 175). This is shown in the examples: (10)
If John came (yesterday), Mary left; and If John comes tomorrow, Mary left.
Thus, in the first example, there is an inference derived from the knowledge that John came: it is that Mary left; in the second the fact that Mary left prior to John’s coming shows without doubt that there is no causality or prediction. This type of conditional has been described by Sweetser (1997: 125) as an epistemic conditional; that is, one that does not express conditional dependence between the events or states they denote, but merely the conditional relation between the speaker’s belief in the truth of the protasis and the speaker’s conclusions about the truth of the apodosis. Epistemic conditionals appear to be pragmatically stronger than predictive conditionals, and more highly subjective in the sense discussed by Traugott (1995a, and elsewhere) in that they describe the conclusions of the speaker rather than the predictions expressed. The function of a nonmodalised apodosis is to suggest via the absence of predictive forms a greater degree of certainty than would otherwise be the case (‘Mary would have left’ (prediction) is a weaker statement, truth-conditionally, than ‘Mary left’ (deduction, based on a given condition)). The inferences of prediction are still present due to the use of a modifying if-clause, but the prediction is taken to be strong enough to use the form normally used to make an assertive statement. What is at the centre of this argument is that it is the notion of a causal relation between the two clauses which appears to distinguish the subjunctive conditional from the indicative, not the degree of hypotheticality expressed, and it is the causal relation which determines whether a past counterfactual form of a conditional can be treated as predictive. But in the case of an indicative conditional, the implied causality between the two clauses is a further abstraction from the
26
HYPOTHETICAL MODALITY
causality suggested in the content of a modalised conditional, it demonstrates the causality of the act of believing p is to be the cause for the act of concluding q. Whether or not there are other constructions available to deal with the notion of a hypothetical protasis, the type of dependency between the two clauses is different in each case. The centrality of the causal relation between the two clauses of a hypothetical conditional is highlighted by Comrie (1986: 80), who discusses the differences between conditionals of material logic, and those of natural language, maintaining that the former can be used syntactically to convey an expression of the form, if p then q, producing an implicational relationship between any two propositions, regardless of whether they are semantically related. Thus a conditional of the type (1986: 80): (11)
If Paris is the capital of Spain, two is an odd number
could be said to be true under a logical interpretation, because both propositions, p and q, are false. The distinction Comrie makes between the material logic conditional and that of natural language is that in the latter, the content of the proposition in the protasis must be causally linked to the content of the information in the apodosis, and so determinable by it. The causal relation between the two clauses is hypothesised by Comrie to be the basis for a conversational implicature, which, as noted above by Lewis (1973) (and also Karttunen and Peters 1977) had been earlier postulated as an explanation for the falsity of the protasis clause. It is from this implicature that the meaning of the conditional is presumed to be derived. Comrie’s proposals for the classification of conditionals are typologicallybased but somewhat controversial in their claims. He suggests that the distinctions normally made between ‘counterfactual’, and ‘hypothetical’ conditionals (as exemplified by Quirk et al. 1987 in (1) and (2), respectively) do not exist, and that the subjective evaluation of the participants in a conversation determines the degree of hypotheticality along a continuum of probability/hypotheticality correlations. The sentence with the lowest degree of hypotheticality would have the highest degree of probability — a factual sentence — and the ‘counterfactual’ clause would have the highest degree of hypotheticality, although not implying counterfactuality (Comrie 1986: 88). In addition, he claims that conditionals are incapable of expressing counterfactuality and that English lacks counterfactual conditionals, if by this is meant that the falsity of the protasis or the apodosis can be deduced logically from the construction of the sentence (1986: 89). To illustrate his point, Comrie shows that the content of the protasis may not be necessarily false (1986: 89):
HYPOTHETICAL MODALITY AS A GRAMMATICAL CATEGORY
(12)
27
If the butler had done it, we would have found just the clues that we did in fact find.
If the meaning in this conditional is derived by conversational implicature, the relative clause appended to the apodosis of this conditional acts to cancel the implicature that the apodosis is false, and the truth or the falsity of the protasis cannot be determined. Comrie uses such an example to show the possibility of the truth of the proposition in the protasis, given the presupposition that there were indeed clues found, this information being contained in the relative clause. The truth or the falsity of the proposition in the protasis then becomes a matter of individual judgement, and the meaning of the construction becomes an artifact of the dependency relations expressed between the clauses and inferrable by the hearer on the basis of contextual information. Dancygier (1993: 411, and 1998: Ch. 2) argues that examples similar to (12) such as, If Mary were allergic to penicillin, she would have exactly the symptoms she is showing (Karttunen and Peters 1977: 365), do not act to cancel any implicature, but that they actually mention contradictory pieces of evidence, and can be continued with further such contradictions, e.g. but in fact we know that Mary is not allergic to penicillin, suggesting that the sentence cannot be interpreted either factually or counterfactually. She assumes that the strong hypothetical verb forms are signalling actual ‘knowledge to the contrary’, and that this is part of what is being communicated and therefore cannot be cancelled. Dancygier’s arguments indicate the degree to which the conversational implicature may have become conventionalised at least for some speakers, but it is not knowledge which is conveyed in hypothetical verb forms, but predictions. Dancygier also raises the matter of causality, suggesting that it is inseparable from sequentiality (1993: 413; see also Dancygier 1998: Ch. 3). Sequentiality is well-known in coordinated constructions to be an iconic signal of the chronological ordering of events, and indeed, the suggestion of a temporal ordering of events in a conditional construction, marked by the protasis preceding the apodosis, appears to contribute a great deal to the relation of causality between the two clauses. This could explain also why some languages, such as Mandarin Chinese, have an obligatory order of protasis before apodosis in conditional sentences; furthermore, if clause order is the only means of deriving the causal link and the counterfactual implicature, then it is vouched for in a language such as Mandarin Chinese, which otherwise has no verbal morphology indicating degrees of modal remoteness to supply the same meaning. Comrie (1986: 86) rejects the possibility of the linear order of the clauses necessarily contributing to the causality, since other causal subordinate constructions may occur without any particular order or sequence
28
HYPOTHETICAL MODALITY
contributing to the causality (because-clauses can follow the clause indicating the resulting event), and clause order can be reversed at least in English without altering the level of hypotheticality. It is also possible to use a counterfactual construction with only a remotely causal relation between the protasis and the apodosis, as shown above in (12). Comrie instead proposes that it is the function of topicalising the conditions of argumentation in discourse that results in the preposing of the protasis. This may well be the case for Chinese, since it is known for its topic-prominent sentence structure, a feature which distinguishes it typologically from many other languages (Li and Thompson 1989: 85). 2.1.3
The falsity of the antecedent clause
Comrie’s example, however, may be analysed from a different point of view. As will be discussed in Ch. 3, the Gricean Maxim of Quantity has been previously invoked to explain the grammaticalisation of epistemic modals in English. The Quantity Maxim is subdivided into two tenets: (i) “Make your contribution as informative as is required for the current purposes of the exchange”; and (ii) “Do not make your contribution more informative than is required” (Grice 1975: 45). These two principles, therefore, polarise the pragmatic control on the information quantity of the utterance resulting in a negative extension by the hearer in (i) and a positive extension in (ii). The pragmatic control is therefore divided between that of the speaker and that of the hearer, as the first tenet (henceforth Q1, as termed in Levinson 1995) is hearer-based, the hearer economising on the interpretation of an utterance (essentially Zipf’s Auditor’s Economy according to Horn 1984) and the second (Q2) is speaker-based, in that it is the speaker who is using economy of expression (Horn 1984: 13), operating by the Principle of Least Effort (Zipf 1949, cited in Horn 1984). Thus, Q1, which is central to the ‘Q(uantity)-Principle’ in Horn (1984), maximises the coding of expressive material in an utterance, inducing an interpretation that anything left unsaid was not true, or not known to be true or false (Atlas and Levinson 1981: 37–8), while Q2 (subsumed into the ‘R(elation)-Principle’ of Horn 1984), reduces the coding of information and invites inferences to include information not already supplied, because of the minimalisation of the information already provided by the speaker.4
4. Horn (1984: 12) notes that Q2 is akin to the Maxim of Relation in that the principles of Relevance determine the type of extension of informativeness (Traugott 1989 also notes this.) However, for the purposes of the present account, Relevance will not be invoked as essential to the explanation, mainly for the fact that Quantity-derived sources for implicatures appear to be more objectively determined than Relevance-derived sources.
HYPOTHETICAL MODALITY AS A GRAMMATICAL CATEGORY
29
Horn (1984: 21) discusses the fact that R-based implicatures can be cancelled in certain circumstances, and he illustrates the cancellation with the use of a semi-modalised expression be able to, in (13): (13)
She was able to solve the problem
which R-implicates: (13′)
She solved the problem.5
The cancellation is obtainable without negation, using contrastive stress on the element in the sentence which derives the implicature, in this case, the expression of ability. Thus: (13″) She was able to solve the problem … could imply to the hearer that she only had the ability to solve it, but didn’t actually solve it. An appended clause introduced by an adversative conjunction such as but and followed by a negative can produce the same cancellation; e.g. (14)
She was able to solve the problem, but she didn’t have enough time.
The contrastive stress discussed by Horn (1984), as well as the expression of an explicit cancellation in an adversative clause, contributes to the conversion of a Q2 implicature of the prediction of the complement proposition (‘she solved the problem’) to become a Q1 implicature of negative prediction (‘she didn’t solve it’). The expression of past ability is then restricted quantificationally either by prosodic variation or by the introduction of a clause expressing contrast. The example of a semi-modalised expression serves as an illustration of the way in which Quantity implicatures may operate with modal expressions: the modal element acts as a quantifier over the potential realisation of its complement, and provides the grounds for predictive extensions of meaning. Used in the past, however, the predictions can be cancelled by the provision of further information about a situation (which is more likely to be available when referring to past), and a Q1 implicature of counterfactual prediction is the result, but in such cases it is an inference which is only contextually-derived from the presence of cancelling elements. The speech-act function of the apodosis clause, in a similar way, expresses a prediction, which Q2–implicates the realisation of the modal complement. Such predictions may stand alone in certain contexts, e.g. the ancient Egyptians would
5. The implicatures described here are alternatively referred to by Panther and Thornburg (1999) as a POTENTIALITY FOR ACTUALITY metonym.
30
HYPOTHETICAL MODALITY
have made this inscription (although the implicature in such predictions is weaker than in (13)) but the function of the protasis is generally to act as a modifier on the prediction expressed in the apodosis clause. According to Horn (1972: 56), suspension allows for something stronger to hold; e.g. This place is pleasant, if not idyllic in which the suspension is supplied in the if-clause. In a predictive conditional, the Q2 implicature of the realisation of the complement proposition in q is suspended by the provision of a condition in p: the pragmatic act of imposing a condition on the realisation of q might then be said to strengthen the likelihood of q being realised in the case of p being realised. However, the ‘givenness’ of p is not guaranteed, and the realisation of q when dependent on p is likely to be only weakly predictable. This is because p strengthens to imply a single and unique condition on q, and, given the relatively low probability for a unique condition to be realised and the causal relation implied between the two clauses, this unlikely probability in counterfactual protases is strengthened to imply a non-event. p then suggests a false proposition, and its original function of suspending the Q2 implicature is extended to act as a cancelling or contradicting device. This is seen most clearly when the protasis is postposed; e.g.: (15)
She would have become president, [and she did] if she had got enough votes [so maybe she didn’t].
The Q2 implicature of the realisation of the apodosis proposition (‘she became president’) is modified by the condition expressed in the protasis (if she had got enough votes), which suspends the Q2 implicature when it is not known whether she got enough votes, and cancels the Q2 implicature when it is already known that she didn’t get enough votes. However, in some cases it would not be known to the hearer whether or not she got enough votes, but the hearer assumes the speaker’s uncertainty in the act of imposing a condition on the outcome of q, and extends the uncertainty to infer the speaker’s beliefs to the contrary: the suspension of the prediction is therefore strengthened to suggest a total cancellation. The strengthening of the suspension to a cancellation is most likely when the condition for the realisation of q is a necessary one; e.g. in (15) it is necessary to get enough votes in order to become president. This is a matter of world-knowledge, though, and unrelated to the structure of a conditional, since the if-clause usually indicates only a sufficient condition. The strengthening which transforms sufficient conditions into unique and necessary ones is often referred to as conditional perfection (Geis and Zwicky 1971), but it is usually considered to be an inference of non-hypothetical or open conditionals, as in the time-honoured example:
HYPOTHETICAL MODALITY AS A GRAMMATICAL CATEGORY
31
(16)
If you mow the lawn, I’ll pay you $15 If you don’t mow the lawn, I won’t pay you $15 (therefore) Iff (If and only if) you mow the lawn, I’ll pay you $15. Used with counterfactual conditionals, though, the means by which conditional perfection is achieved, as pointed out by Horn (2000), often cannot apply without a change of meaning. But (17) still allows for the inference from if p to if ~p to hold: (17)
If she had got enough votes, she would have become president. If she hadn’t got enough votes, she wouldn’t have become president. But I don’t know whether she got enough votes — I wasn’t around at the time!
Usually there is an implicature (or presupposition, according to Horn, 2000) of falsity backing the utterance of counterfactual conditionals, which would make the two premises in (17) appear contradictory. (The additional information in (17) referring to the speaker’s knowledge of the true situation reveals most clearly that the speaker’s commitment to a factual premise for uttering a counterfactual conditional is not a necessary part of the conditional utterance itself.) However, if (15) is a Hypothetical conditional, it may or may not be known whether or not the subject referred to received the most votes, and as such, the conditional can be perfected in the same way as a non-hypothetical.6 In such cases it may be expressed as a unique condition: (18)
Only if she had got enough votes, would she have become president.
The relation between if p, q and if ~p, ~q > iff p remains the same, regardless of the temporal reference. Thus, the modifying protasis may be said to cancel the prediction in the apodosis when it is a counterfactual, but only suspend the prediction when it is a Hypothetical conditional, and in hypothetical conditionals there is a clear correlation between unique conditions and counterfactuals, and sufficient conditions and Hypotheticals. This is because unique conditions provide stronger justification as a cause for the possible realisation of q. Given that a unique condition and a sufficient condition are marked the same way in an if-clause, one means of disambiguating them might be to paraphrase the condition as a
6. One contributory factor which allows for such an interpretation is the presence of a third person subject, which is more distant from the speaker’s deictic centre, and thus evaluated by the hearer as more remote from the factual situation known to the speaker. Horn’s example used a second-person subject.
32
HYPOTHETICAL MODALITY
negative assertion. This may be done by encoding the negative assertion using a contradicting but (not) clause, since but (not) clauses as main clauses may contain assertions (the use of an adversative clause to cancel a prediction is discussed further in Ch. 3, Section 3.5).7 If the condition is sufficiently strong to withstand paraphrasing as a negative assertion without losing the implied causality between the clauses, it may be regarded as a unique, and therefore necessary condition.8 The but-clause can be appended to the predictive apodosis; for example: (19)
She would have become president, but she didn’t get enough votes.
The possibility for the if-clause to be paraphrased in such a way without losing the causal link between the two clauses suggests the potential for the clause to be considered a single necessary cause for the realisation of q, and therefore it can act as a cancelling device on the prediction contained in the apodosis. In order for the falsity of the protasis to be derived in such a manner, the type of the Quantity implicature involved must be taken as a Q2, implicating the positive realisation of the prediction contained in the modal complement. However, if the type of Quantity implicature is considered to be a Q1 implicature, the falsity of the protasis has to be accounted for in different terms. The strengthening of a suspending if-condition to become a cancelling iff-condition is held by Levinson (1995) and Van der Auwera (1997) to be illustrative of a Q2 implicature, via the intermediacy of an if-not condition. As noted above in (16) such inferences are usually only associated with present and future conditions; however, the strengthening of the sufficient condition to a unique and necessary one holds whatever the time reference, especially if (15) is considered to be a
7. Traugott (1997) provides a historical account of the relation between unless and but. But is translated as unless in some of her examples, and unless is defined as specifying a unique condition, while if … not specifies only one of several conditions available. Given such definitions, the relation between but and necessary conditions is easy to construe in a paraphrase, in which the but-clause, appended to a past prediction, specifies a unique condition (and therefore a necessary one). Furthermore, as Traugott and others have noted, unless rarely occurs in past counterfactual conditionals. The reason is that the function of unless is to restrict the number of possible alternative conditions to a unique condition; i.e. it can convert an if-clause to an iff-clause (as noted by Traugott). The criterion for its use, then, must be that the condition must be uttered without knowledge of any supporting facts on which to base it. A past counterfactual presupposes a factual basis, and therefore is unlikely to require restriction using unless. 8. To suggest that necessary conditions alone may provide the implicature of causality is inadequate, as pointed out to me by Barbara Abbott (p.c) — there may be any number of necessary conditions on the realisation of q, but the inferences contained in an iff p protasis suggest the construal of the condition as unique as well as necessary, and therefore also suggest the reduced probability of its realisation.
HYPOTHETICAL MODALITY AS A GRAMMATICAL CATEGORY
33
Hypothetical conditional. In such cases the implicature of if ~p from if p is an extension of the information available in if p, but it is doubtful whether it constitutes an example of a Q2 implicature (see also Horn, 2000, and Matsumoto 1995). The implicature of restricting the quantification of possibilities conditioning the realisation of q is more likely to be a Q1 implicature, suggesting that if p, q may be idealised as a point on an implicature scale, with (non-modal) q at the positive pole, and the modalised quantification in the apodosis (would have q) somewhere in between. The quantification becomes increasingly restricted as the scale descends (see (20)). (20)
(S) |- q | |- would (have) (q) | | (W) |- if p, would (have) (q)
In (20) the modal (+have) is considered to be a quantifier over the factuality of the bare proposition q, while the protasis condition may act as a quantifier over the prediction of q marked in the apodosis by a modal. According to scales of this nature (Horn 1984: 21), W (the weaker element, situated lower on the scale) (if p, q) Q1 implicates the negative of the stronger element (S), here, the proposition of q (whatever the condition). Thus, would (have) q Q1 implicates ~q, and if p, would (have) q Q1 implicates ~(would have) q. Given ~q, then, and the causal link between the two clauses (which is only a conversational implicature as suggested by Comrie 1986), there is a chain reaction from the negation of q to the negation of p. This chain reaction could contribute to the implicatures of falsity in the antecedent, so long as it is possible to idealise conditional clauses as quantifying restrictions over the realisation of the apodosis prediction, and so long as the type of implicature derivable from such quantification is a Q1 implicature. We are therefore presented with two possible explanations for the inferences of falsity in the antecedent clause of a hypothetical conditional: the first presupposes the positive extension of would (have) q to predict q, an inference which is brought about via the operation of Q2 implicatures (‘say no more than you must’) and then is subsequently cancelled by the sufficient condition in the antecedent, which increases in strength to become a necessary and unique condition (iff p). The second presupposes the negative extension of would (have) q to Q1 implicate ~q, and then via modus tollens, p itself is implicated as false. Although it cannot be stated with any certainty which of the two Quantity
34
HYPOTHETICAL MODALITY
maxims, Q1 or Q2, may be said to be basic to the prediction expressed in the apodosis, in the case of the second analysis, it is necessary that the causal link between the two clauses be already established in order to cancel p via the cancellation of q. If this is the case, then the first analysis may be primary, since it is via the strengthening of the protasis condition to a unique and necessary condition that the relationship of causality between the two clauses is established. Ch. 3 presents a historical viewpoint on the interaction of Quantity implicatures with the predictive senses conveyed by would (have), suggesting that Q2 implicatures may be basic in the development of hypothetical meaning as well as to future-projecting senses, and Q1 implicatures may therefore be derived. In either case, though, the relationship of implied causal dependency between the two clauses of a hypothetical conditional is crucial to the development of counterfactual inferences, and the implied causal dependency is created not only by the sequentiality of the clauses, but by the strengthening of the sufficient condition to become a unique one. It might therefore be suggested that the cancellation in (12) is possible because the sufficient condition in the protasis cannot be construed so readily as a unique one. The construal of a sufficient condition as a unique condition on a subsequent prediction contained in the apodosis is more difficult to cancel, and in examples like (12), appears to be not sufficiently closely related to the apodosis proposition to be causally linked with it. The strength of the causality (12) can therefore be tested by reversing the clause order and substituting the ifclause for a but (not) clause containing a negative assertion: (21) ?We would have found the clues, but the butler didn’t do it. (21) is questionable due to the weakness of the causal connection between p and q and hence (12) is not infelicitous. On hearing (21), one might be tempted to respond: ‘Well, did the butler have to do it for you to find the clues?’ The contradictory clause leaves the hearer in doubt as to the connection between the butler doing it and the finding of the clues.9 The processing of the inference from the butler’s actions to the finding of the clues requires more effort and a greater demand on the hearer’s imaginative powers. Thus, in order for the ifclause to be paraphrased as a but-not clause, there must be a sufficiently direct causal link between the protasis and the apodosis to constitute a necessary
9. According to Allan (forthc.) (amongst others), the use of but to suggest contrast or adversity is a conventional implicature, which cannot be cancelled without incoherence; e.g. We would have found the clues, but, not unexpectedly, the butler didn’t do it.
HYPOTHETICAL MODALITY AS A GRAMMATICAL CATEGORY
35
condition. Iff-clauses, then, as unique conditions, must provide an obvious cause for the realisation of the event described in q. If there is a closer, more direct link between the protasis and the apodosis propositions, then the cancellation is adequately conveyed in a but-paraphrase, because the causal link is stronger: (22)
We would have driven to the border, but the car didn’t have enough fuel (If the car had had enough fuel, we would have driven to the border.)
It is more difficult to treat the protasis clause in (23) as a suspension of the prediction, and this can be seen in the use of a cancellation of the falsity of the apodosis: (23) ?If the car had had enough fuel, we would have driven as far as the border, which we did. Since the speaker knows whether the car had made it to the border at the time of speaking, it seems almost redundant to suggest that even if the car had had sufficient fuel the border would have been reached, as cars need fuel to travel. The necessity of the car’s containing fuel for reaching the border is one possible cause which is too directly related to the effect to be felicitously cancelled in such contexts. The attempted cancellation does not permit the possibility of the car not containing enough fuel to be considered along with the possibility of it reaching the border in the same way as (12) permits the possibility of the butler not having done it to be considered with the possibility of our finding of the clues anyway, regardless of who committed the crime. The more direct the link between cause and consequent, the less justifiable a cancellation of the apodosis prediction becomes. In such cases the inferred counterfactuality cannot be read off the grammatical structure of the clauses, and the content of the propositions contained in the clauses has a positive role to play in creating the necessary causal links. It would appear that the modifying properties of an iff-condition imposed on an apodosis prediction, inferring causality between the protasis and the apodosis, may contribute a feature to a cluster of features which together contribute to inferences of counterfactuality in a conditional construction. Since past and perfective meanings can be realised grammatically as part of the cluster, one might assume, then, that there would be languages in which iff-conditions can be realised by distinguishing grammatical markers so that there exists one marker for Hypothetical if, and another deriving counterfactuality by means of expressing uniqueness (iff). It appears that such occurrences may be rare; Traugott
36
HYPOTHETICAL MODALITY
(1985) does not illustrate any distinctions, nor does Wierzbicka (1997) in her crosslinguistic summary of counterfactual conditionals. One possible instance may be in Korean, in which there exists a particle ya which if attached to the protasis of a conditional clause, denotes biconditionality (Alan H-O Kim, p.c.) but it does not appear to be used with past counterfactuals. Van der Auwera (1997) notes also the existence of zhiyao as a protasis marker in Mandarin, but Eifring (1995: 265–9) denies that this has any special function of marking counterfactual protases; the basic meaning ‘only need’ (need only) implies the ease of obtaining the required condition (a sufficient condition), rather than the difficulty of obtaining it (a unique condition), as in the English only if and its associations of impossibility. The likelihood that this member of the cluster, then, is grammaticalised, remains uncertain.10 2.1.4
Other features of the cluster
It has been shown in 2.1.1–3 that the presence of past tense or perfect morphology, as well as the extension of the condition in the protasis clause to become a bi-condition, and therefore a necessary one, are two factors which have been found to contribute to the evaluation of a conditional construction as more highly hypothetical. The latter is related more to the pragmatic interpretation of a conditional than to the presence or absence of specific grammatical clues, and thus may be a universal characteristic of conditionals across languages. However, there are other grammatical features which may contribute to the cluster of features that assist in providing a more highly hypothetical interpretation of a conditional construction, and such features are also associated, like past or perfect morphology, with the presence or absence of hypothetical meanings in constructions other than conditionals, e.g. the nominal clause complements of verbs of wishing, imagining, or desire, the predicates of predictive adverbials such as proximatives (almost; nearly), the past tenses of all stative verbs and past imperfective expressions, bare protasis expressions (e.g. if only …), and modal perfect constructions without subordination (e.g. You should have gone). The two factors are negation and the influence of subject person (Ziegeler 1994: 31).
10. Without a thorough crosslinguistic survey, it would be risky to suggest that iff is not grammaticalised in counterfactuals. However, the implicatures by which grammaticalisation usually proceed are Q2 or R-based implicatures, and the argument proposed in this chapter is that the extension from if to iff is a Q1, or Q-based implicature. Thus it would run counter to explanations of the strengthening of pragmatic inferencing in grammaticalisation to suggest that a grammatical item could gammaticalise further by the influence of Q1 implicatures.
HYPOTHETICAL MODALITY AS A GRAMMATICAL CATEGORY
37
In Ziegeler (1993 and 1994, discussed also in Ch. 4) a survey was conducted amongst a group of 31 native speakers of English and 60 L2 speakers of Singaporean English, in order to determine the possible level of hypotheticality interpretable from a short narrative in which all but one of the sentences implied negative counterfactuality in both clauses (the final one expressing negation in the protasis only). An example of one of the sentences held for evaluation was the following: (24)
If the water level had not dropped, the Japanese officer wouldn’t have spotted the wire.
The results of the exercise were then compared with those from another task, in which the counterfactuality of positive protasis clauses was assessed in the same way. For example: (25)
If the murderer had used a gun, he would have hidden it somewhere.
The results showed that the index of counterfactual interpretations, evaluated from the percentage of target answers, was very much higher for both groups for the negative counterfactual exercise than for the positive counterfactual one (the latter exercise also contained conditional clauses not expressing necessary conditions). The results are shown in Table 2.1: Table 2.1
Results in averages of two tasks eliciting counterfactual interpretations from native speaker and L2 speaker groups (Ziegeler 1993)
group native speakers L2 (Singaporean) speakers
negative counterfactual task
positive counterfactual task
62% 83%
03% 25%
There is clear evidence from this that counterfactual interpretations are favoured when negatives are present in at least the protasis clause. Wierzbicka (1997) also finds this in a number of crosslinguistic examples, and claims that the presence of negation (in conditional clauses) is a grammatical indicator of counterfactuality. The reason for negative counterfactuals to be more highly hypothetical is not discussed by Wierzbicka but can be explained by factors surrounding the use of a negative statement in general: a negative statement always presupposes the existence of an underlying positive proposition (Horn 1989: 173). Horn discusses the accessability of a corresponding positive proposition as essential in the interpretation of a negative; this is because the main function of a negative
38
HYPOTHETICAL MODALITY
utterance is to make a denial, and there must be some proposition to deny in the first place. The utterance of a negative statement, therefore, is held to be supported by information of a higher specificity than a positive statement, which requires no corresponding negative proposition with which to justify its utterance. The same conditions apply to the utterance of negative hypothetical conditionals: Wierzbicka (1997: 40) finds that in Russian and Polish, sentences with both a negative protasis and a negative apodosis are interpreted as unambiguously counterfactual: (26)
Polish (Wierzbicka 1997: 40) Gdyby nie pozsli te strone, to by sie nie zgubili ‘If they hadn’t gone in that direction, they wouldn’t have got lost.’
Wierzbicka also notes that first and second person subjects can occur with double negation in counterfactual conditionals to imply counterfactuality in Russian; e.g. (1997: 41): (27)
Esli by ne bylo dozdja ja by ne posla v kino ‘If it hadn’t rained I wouldn’t have gone to the cinema.’
She accounts for this as an unexpected phenomenon, since first and second person subjects usually express uncertain intentions rather than facts. However, as noted in note 6 and Ch. 6, the appearance of first and second person subjects (especially first person) is not unusual, and is likely to create a higher level of counterfactuality than a hypothetical sentence with a third person subject. This is because first and second persons are deictically closer to the speaker’s immediate domain of reference, and in the case of first person subjects, the subject and the speaker are the same. If the speaker and the subject are the same, then the speaker has the most intimate knowledge of the subject’s intentions, and therefore is in the best possible position to make a factually-based prediction about the past. This may be seen in the following example: (28)
a. b.
If I had been there at the time, I would have seen the thief. If he had been there at the time, he would have seen the thief.
The counterfactual implicatures in (28b) can be easily cancelled by adding ‘so let’s go and ask him if he was there’. (28a) uses a first person subject in both clauses, but even then, cancellation is possible, e.g. ‘but I didn’t know where I was at the time’. This indicates that only world knowledge and extra-textual information can truly determine the factuality of the premise upon which a counterfactual hypothesis is based; grammatical devices are available to indicate that the utterance is irrealis, but are not reliable indicators for disambiguating
HYPOTHETICAL MODALITY AS A GRAMMATICAL CATEGORY
39
levels of hypotheticality. It is also found that in the survey described in Ch. 6, sentences with first person subjects in hypothetical complement clauses (e.g. I wish I would come to your party) yielded the lowest number of acceptances overall (7.6% average), although the low acceptability could be equally due to the co-referentiality of the main clause and subordinate clause subjects. However, it is likely that the same implications of the presence of a first person subject (and a second person to a lesser degree) can apply in hypothetical contexts other than conditional constructions. The derivation of hypotheticality as a grammatical phenomenon, then, seems to be explained by the presence or absence of a number of features, which if considered as a cluster of features, together may contribute to an overall optimum situation for the expression of hypothetical notions. The more of the features that are present, the higher the level of hypotheticality implied, with counterfactuality obtainable as an inference or implicature deriving from the highest possible level of hypotheticality, as suggested by Comrie (1986). Thus, the continuum nature of Comrie’s description may correspond to variations in the number of features present in the hypothetical context. Some of the main features discussed so far are the following: (29)
a. b. c. d. e.
Past and perfect morphology, combined with an irrealis or future marker The causal link between the two clauses of a conditional Negation First person subjects Extra-linguistic world knowledge
(29b) is, of course, restricted specifically to conditional constructions; the other criteria may be found in association with any hypothetical context. Moreover, (29a, c, d, and e) may all contribute to the derivation of Q1, rather than Q2 implicatures in the apodosis clause of a counterfactual conditional. There may be features which can be added to the cluster; the list is by no means finite. But what is important is that the interpretation of hypotheticality, at least for the languages mentioned, as a grammatical or discourse notion requires reference to a range of features, not simply a single feature or a gram-type (of the kind first discussed by Bybee and Dahl 1989). In proposing a cluster definition of hypotheticality, it is possible to summarise the definition in the form of a principle, as discussed below.
40
HYPOTHETICAL MODALITY
2.2 A principle for determining the grammatical basis of hypothetical meaning In the above discussion, two main points are clarified with respect to the determination of hypothetical modality as a grammatical phenomenon in English: (i) it cannot be associated solely with past temporal reference, as counterfactual conditionals may appear with future time reference, given the appropriate circumstances, e.g. (3) (Type A conditionals). There is a suggestion in such examples that the original meanings of past time reference are semantically weakened in favour of the derived implicatures of (atemporal) counterfactuality; (ii) some conditionals with the same morphology as counterfactual conditionals may be found to be only Hypothetical (Type B — (9)); (iii) past temporal reference is insufficient to provide a metaphor of remoteness from fact, and, as suggested by Dahl (1997), it is a combination of both past time reference and prediction that eventually contributes towards the derivation of hypothetical inferences. Givón (1994: 310) raises the puzzling question why subjunctive clauses (usually associated with irrealis) are frequently found in combination with past or perfect morphology, which he describes as “quintessential realis tense-aspects”. It is noted above that past contexts are essential amongst a cluster of environmental conditions which provide the optimum situation for the interpretation of hypotheticality. Past and perfect morphology marks events and situations which are generally accepted as completed, and therefore have a higher epistemic value in the estimations of the speaker-hearer interaction. The termination of the event, in turn, indicates a greater potential for factual recall by the speaker, since it has already passed at the moment of speaking. The other environmental conditions noted in (29) also contribute to the likelihood of a factual supporting premise on which to base a hypothetical expression: negation, because it presupposes a corresponding positive proposition (positive hypotheticals do not necessarily presuppose a corresponding negative proposition, as seen in (12)); first person subjects, who have first hand information about the factual premise for the speakers’ utterance; the uniqueness of the condition in the protasis of hypothetical conditionals, resulting in a greater possibility of cancelling the prediction in the apodosis, and extra-world knowledge about the facts supporting a hypothesis, as would be possible in the case of (3). The enforcement of the inferences of counterfactuality/hypotheticality, then, depends only on the potential for the speaker to utter a statement about which true facts are known. As this potential increases with the addition of more members of the cluster, so the likelihood of defeasibility of those inferences decreases, and a counterfactual inference
HYPOTHETICAL MODALITY AS A GRAMMATICAL CATEGORY
41
becomes increasingly conventionalised. Given such features, a hypothetical implicature gains in strength, and this can be attributed to the strength and quantity of available factual information related to particular contexts, derived from the linguistic environment. What all the features in (29) have in common, then, is a specificity condition for the licensing of a counterfactual utterance — that is, the introduction of such features into the illocution increases the conceptual specificity of the supporting factual premise upon which the utterance is based. Thus the likelihood of the occurrence of an implicature in past contexts, for example, is minimal, since we normally have maximally-reliable facts about the past, and we do not need to derive pragmatic inferences for further information. Additional features added to the context such as negation or first person subjects increase the evaluations of the hypotheticality of the context, to the point at which implicatures of counterfactuality may be predicted. It may be questioned, then, why counterfactual implicatures are so often associated with environments with high information-density. The situation can be summarised in the form of a single principle, which will be labelled in the present study the CFI-(Counterfactual Implicature) Principle: (30)
The CFI-Principle: ‘The strength of an implicature is directly proportional to the specificity conditions in which it is located.’
(30) could also be paraphrased as: ‘The strength of an implicature is directly proportional to the information density of its context.’ In linguistic terms, it is hypothesised that by increasing the information specificity of an utterance, any hypothetical implicature or inference obtained from such an utterance will strengthen to counterfactuality; therefore, in a situation in which the specificity and/or factuality of an utterance is expected to be greater, as, for example, in past, negative contexts with first person subjects, or contexts about which the speaker is better informed of the facts (e.g. (3)), or in specific, only if protasis conditions, an implicature of highest hypotheticality or irrealis will be probable, having the potential for factual support. This also means that there is reason to believe that past contexts are more highly hypothetical not because they are remote or distant from reality, but because they are close to reality. Counterfactuality, then, runs parallel to factuality, and is not its polar opposite but merely an abstraction from the real world. A counterfactual utterance is thus an inverse construal abstracted from known facts about a situation. It does not serve to deny reality, it simply holds it in a different perspective. The knowledge of the factual premise is strictly within the purview of the speaker; the hearer is unable to deduce this from a
42
HYPOTHETICAL MODALITY
non-asserted utterance. As far as the hearer is concerned, there is no evidence in a counterfactual utterance that a factual premise actually exists, but only the potential for creating a factual premise. While counterfactuality cannot be determined logically (as noted by Comrie 1986), the grounds for the abstraction from reality associated with making a predictive hypothesis are the reality itself, and the stronger the evidential basis for assuming real facts, the stronger is the prediction to the contrary of what is known. A Hypothetical prediction formed on the basis of uncertain evidence will therefore be less hypothetical and weaker than one which is grounded with known facts. This is why it is possible to measurably predict the licensing of counterfactual inferences from the presence of certain grammatical and contextual pre-conditions.
2.3 Summary The above account has been provided mainly for the purpose of dealing with the grammatical phenomena surrounding the use of counterfactuals and hypotheticality in general, and examines conditional constructions as a representative case illustrating hypotheticality as a grammatical category. The interaction of formal versus pragmatic characteristics of the grammatical expression of counterfactuality is such that formal characteristics of particular grammatical morphology such as past and perfect markers, which locate the hypothetical prediction or utterance in a situation occurring before the moment of speech, or negatives and first person subjects, which also increase the grounds for information specificity, are able to provide clues to the level of factuality on which the hypothetical utterance was made. The hearer then bases his or her evaluations of the counterfactuality of the utterance on the presence of such formal characteristics, and by the operation of the CFI-Principle, determines the strength of its hypotheticality. Semantically, the content of a conditional construction will enable pragmatic strengthening to bi-conditionality, but this is not relevant in cases such as in (12), in which the same grammatical morphology used to indicate counterfactuality appears in a conditional which contains only a sufficient condition, as illustrated by its cancellation without redundancy. The causal link between the two clauses is then most likely the result of the frequent pragmatic strengthening of a sufficient condition to a unique and necessary one. Thus, the functional role of a conditional is to express in the main clause a prediction made about the past (in the case of counterfactuals), and the if-clause only modifies the prediction by a) suspending it; or b) cancelling it. Furthermore, the prediction of the complement of the modal, q, can be idealised as appearing
HYPOTHETICAL MODALITY AS A GRAMMATICAL CATEGORY
43
on a quantity scale, and the falsity of the protasis is deduced from the negation of q by Q1 implicatures. There is no specific grammatical morphology available to distinguish past Hypothetical from counterfactual conditionals, as seen in Type B (7). Thus, in some ways certain subtle nuances relating to the strength of the hypothetical implicature are not expressed in the grammatical form, and must be determined contextually. However, the probability of a counterfactual implicature is increased by the presence of a range of grammatical and/or contextual indicators which contribute to the hearer’s overall evaluations of the factuality or information density of the premise upon which the hypothetical utterance was made, and such indicators may be found as a crosslinguistic phenomenon. In later chapters the broader notions of counterfactuality and hypotheticality will be discussed in parallel with the historical development of certain modal verb forms in English, and the development of hypotheticality as an instance of grammaticalisation will be further examined.
C 3 A diachronic corpus study of would
In the previous chapter, the grammaticalised expression of counterfactuality and hypothetical meanings was examined and a number of contributory features was isolated as frequently occurring in association with such meanings. In English, the modal verb would grammaticalises one of the cluster of features (past tense and future marking) in that it expresses both past time reference and prediction in one grammatical form. It is well known that would grammaticalised from a lexical verb source expressing volition or desire (see for example, Bybee 1995), and in many other languages, similar (non-past) verbs expressing volition, intention or goal-direction are likely to grammaticalise into future meanings (Bybee, Pagliuca and Perkins 1991). Given that the same lexical sources of volition are also found in their preterite forms to be involved in the grammaticalisation of counterfactual and hypothetical meanings across many languages, it is seen therefore that such grammaticalisation paths are related by a common lexical source, but differ in the type of modality expressed as an outcome. The present chapter will examine the nature of the grammaticalisation of would by considering the historical factors which resulted in its present-day uses, and by investigating the means by which the pragmatic implicatures discussed in Ch. 2 evolved diachronically. The modal is highly productive in Present-Day English (PDE): the range of uses for both present and past tense forms extends from volitional and epistemic uses, to power, and habit, as well as futurity (Palmer 1990), and it may be questioned whether there exists a single diachronic explanation to account for so many different uses as well as the distinction between hypothetical and counterfactual meanings. In standardised varieties of English, the fluctuations between past tense forms and present tense modal forms have become conventionalised so that, in general, past tense forms express greater uncertainty than present tense forms, and distinctions between future and hypothetical meanings are formally defined. Similarly, as discussed in Ch. 2, counterfactual uses are often distinguished from merely hypothetical uses in English and many other
46
HYPOTHETICAL MODALITY
languages by additional morphology marking anterior or perfect aspect (Givón 1994: 315–7); e.g. would + have +V-en vs. simply would + V, in English, though, as also seen in Ch. 2, such distinctions are not always so easily determinable. The use of morphological means of marking different modal functions requires explanation in terms of a grammaticalisation hypothesis. As will be discussed later in the Chapter 3.2.1, Traugott (1989: 51, note) engages the explanation of Grice’s (1975) Quantity Maxim 2 and Horn’s (1984) R-Principle in accounting for the diachronic development of predictive, futureprojecting, and epistemic meanings from formerly deontic or root modals. Under such an analysis, the development of epistemic meanings from deontic modal meanings was attributed to the conventionalisation of conversational implicatures, in other words, pragmatic strengthening (Traugott 1989: 50–1), and past predictive meanings (relative futures) are seen to arise in the same way as present predictive meanings. The development of hypothetical meanings from preterite modal forms (including volitional modals), however, has been attributed to other causes (Bybee 1995), and will be discussed in 3.2.2. Counterfactual meanings are also derived in many languages from volition verb sources, developing alongside meanings of proximity, as discussed by Kuteva (1998), and it is questioned whether counterfactual meanings in English emerge in the same way. The aims of the present chapter are then to examine the grammaticalisation of modality in the English past modal verb WOULD and to investigate whether the same principles of the Quantity Maxim used to explain the development of futureprojecting senses from deontic modals are involved in the grammaticalisation of hypothetical and counterfactual meanings in English.1 The study will also determine whether factors such as variation in the presence of lexical retention or contextual features have any influence on grammaticalisation patterns. Comparison will be made with patterns of the grammaticalisation of volition verb sources in other languages, and a small sample of diachronic data will be examined in order to quantify the distribution of counterfactual, hypothetical, and predictive meanings of WOULD across an extended historical time span.
3.1 Grammaticalisation and grammaticality In order to place the pragmatic explanation within a context of grammatical-
1. A single cover term WOULD is used throughout to refer to both the present-day form and the various orthographic realisations of this form found in the diachronic data. Similarly, the term HAVE is used generally to cover the various manifestations expressed in the data.
A DIACHRONIC CORPUS STUDY OF WOULD
47
isation, it is necessary first to outline some of the assumptions basic to grammaticalisation which will underly the present analysis throughout this book. It must be emphasised, first, that grammaticalisation is understood in the present study to be a gradual process, and that this is reflected in the presence of coexisting old and new forms at one synchronic stage of development. Evidence of the co-existence of the older stages of development with new, grammaticalising meanings is discussed widely, especially in Hopper and Traugott (1993: 90) and in Hopper (1991) where it is termed ‘persistence’. This phenomenon was formerly observed by Bybee and Pagliuca (1987) who termed it ‘retention’, and it is shown to be valuable evidence for the gradualness of grammaticalisation changes. Retention has been thus defined as the phenomenon relating to the continued adherence of former senses associated with earlier historical uses of a grammaticalising expression which may be seen to affect the distribution of the expression in certain grammatical environments at later stages of its development. Most often, such senses are related to the earlier lexical meanings of the grammaticalising expression, and will therefore reveal at subsequent stages of grammatical development co-occurrence constraints associated with the lexical sources from which the expression was formerly derived. In the present study, retention will be seen as a means of indicating the direction of the grammaticalisation of WOULD within a paradigm of functions. The phenomenon of retention is usually associated with single tokens of a form observable at different stages of development and in different linguistic contexts. Because of this, retention can be statistically measured in the intuitions of native speakers across a community at one synchronic time-point (as in Ch. 6), and is often seen as a constraint on grammatical distribution. The fact that it can be measured in this way within a community of speakers is evidence also for the presence of linguistic variation as a manifestation of the grammaticalisation of certain features, which reflects their distribution within that community as well as within the language system itself. In this respect, variational data is a useful tool for examining levels of grammaticalisation at any given synchronic crosssection in the history of a grammaticalising feature. In the present chapter, retention will be taken as an indicator of the path of the grammaticalising modal, and its distribution across a selected sample of texts will be measured in terms of grammatical vs. lexical occurrences. The use of the terms lexical and grammatical does not presuppose rigid distinctions, but merely observable tendencies in which the context of use provides the basis for the categories. For this reason there may exist no ambiguity between lexical and grammatical functions on the level of the individual token, but there may be widespread variation in the representation of lexical vs. grammatical instances determined by contextual
48
HYPOTHETICAL MODALITY
features across an entire text sample. Ambiguity is thus represented most often in the distributional spread of the form at a particular time period, and distributional quantification is thus a valid measure of the development of a grammaticalising feature. A further point to be made, then, concerns the definition of grammaticalisation. To believe that grammaticalisation is observable only as a process in which a lexical form becomes used in a grammatical function and is ‘bleached’ of its lexical content is to accept too narrow a definition for the purposes of the present study. Meillet (1958 [1912]) had associated the term only with such definitions, referring to any additional generalisations of function as analogy. Analogy involves not only paradigmatic substitution of constitutents with the same functional specifications (Heine and Reh 1984: 28), but also rule spread within a community (Hopper and Traugott 1993: 32), such as that described above. Lehmann (1995: 11) cites Kuryłowicz (1965) in referring to the increase in grammatical status of a form undergoing grammaticalisation as inclusive of the definition of grammaticalisation; that is, once a form has started appearing in grammatical functions in one grammatical environment, the grammaticalisation will spread to other environments until there is a more generalised use of the form in a greater number of grammatical functions than lexical. For example, the level of grammaticalisation of modal verbs to have epistemic meanings may begin in some environments but may not be perceived until the verbs start to appear with inanimate subjects which are incompatible with their former functions of expressing desire, ability and obligation. Also consistent with the present definition may be an increase in functional scope of forms already possessing certain syntactic functions, such as that described in Traugott (1995b: 1). Thus it is the broader definition which is to be adopted in the present study: the present-day functions of WOULD are too many and varied to allow for a definition of grammaticalisation that will only refer to the processes of functional change, and the definition necessitates including the processes of generalisation across a range of functions.
3.2 The diachronic grammaticalisation of WOULD 3.2.1
The conventionalisation of implicature in the grammaticalisation of modal verbs
The evidence of lexical retention at various developmental stages as an indicator of the gradual nature of grammaticalisation is not conducive to explanations
49
A DIACHRONIC CORPUS STUDY OF WOULD
suggesting that grammaticalisation proceeds by metaphoric leaps across unrelated conceptual domains. Traugott (1989) proposed that the development of epistemic modal meanings in English was mainly the result of the gradual conventionalisation and strengthening of pragmatic implicatures, a process more closely associated with the effects of metonymic inferencing than with the abrupt shifts of meaning typical of bleaching or metaphor which is the mechanism espoused by Sweetser (1990) to explain the historical development of modal verbs. In a diachronic study of the modals shall (*sculan) will (*willan), and must (*motan), Traugott (1989) traces their development from main verb uses in Old English (600–1125 AD in her study), when they often appeared with transitive objects, while at the same time appearing as deontic modals; e.g: (1)
… þæt he geornor wolde sibbe wið hiene þonne gewinn that he rather wanted peace with him than conflict ‘… that he wanted peace with him rather than conflict.’ (Or. 96.17)
(2)
se de wunde lacnigean wille giote win on. he who wounds heal wants pour wine on ‘Whoever wants to heal wounds should pour wine on them.’ (CP 17 125.11)2
(Traugott 1989: 37–8) Some of the modals already had a use expressing general or typical occurrence, e.g. (3)
elpendes hyd wile drincan wætan elephant’s hide will absorb water (Or. 230.26)
(Traugott 1989: 39) Example (3) indicates that temporal inferences were beginning to appear as early as in the Old English period.3 However, such examples were described by Traugott (1989: 40) as relative tenses, that is, meaning simply ‘later time’ (than a given
2. Traugott notes that the historical references are found in Healey and Venezky’s (1980) microfiche concordance. 3. It could be argued that the temporal meanings were well established in spoken language by the Old English period due to the fact that they are appearing in written manuscripts. However, according to Burnley (1992: 2): “The scribes who wrote the manuscripts which are our only source of information about Old English lived in various parts of England and wrote a version of the language which they spoke every day.” In order to support such an argument, then, it would need to be proven that the language spoken by the scribes was different from the vernacular used at the time by other speakers.
50
HYPOTHETICAL MODALITY
reference point) but not necessarily dependent for their interpretation on speaker time as the point of reference; Traugott questions, therefore, whether there was actually a purely deictic future in Old English. Similarly, must was believed not to have developed epistemic meanings expressing the speaker’s conclusions and certainty until well after the Old English period. Using an example cited from Visser (1963–73: 1811), Traugott notes that the first uses of strong epistemic must occurring alone without a reinforcing adverb (such as nedes) do not appear until the 17th century: (4)
the fruit muste be delicious, the tree being so beautiful (1623 Middleton, Spanish Gipsie 1, i.16)
(Traugott 1989: 42) In showing the development of strong epistemic meaning in this way, she illustrates the way in which modal meanings became gradually more subjective over the period of time from Old English to the present day. Traugott also proposes that the processes by which this occurred did not involve metaphor, as suggested by Sweetser (1990), since it is difficult to understand how a proposition can be obliged to be true on the basis of the obligation contained in the proposition. The meanings of conclusion, belief, and knowledge do not exhibit the characteristics of analogical mapping found in metaphor; rather they are indicative of the conventionalising of conversational implicatures (Traugott 1989: 49–50). Old English *motan, according to Traugott, already was used as a deontic modal of permission (1989: 38); thus, if a speaker says ‘You must go’, meaning, according to the Old English interpretation of must, ‘you are allowed to go’, and is, according to Traugott, observing Grice’s conversational principle of Relevance, it can be inferred from the maxim ‘say no more than you must’ that the addressee is expected to go (1989: 50–1) and co-occurrence with an epistemic adverb such as nedes is presumed to have contributed to the developing epistemic inferences in the modal, Although Panther and Thornburg (1999) describe a similar type of inference as relating to a metonym labelled POTENTIALITY FOR ACTUALITY, rather than a conversational implicature of a Gricean nature (see also Ch. 2, note 4), the metonymic description is not necessarily inconsistent with the Gricean one, and as mentioned above, it is metonymic forces rather than metaphoric influences which motivate the progress of grammaticalisation as a product of the strengthening of pragmatic inferences.4 The
4. The principle of Relevance is also referred to in Traugott and König (1991) as a basis for the development of conversational implicatures. Traugott adds in a footnote (1989: 51) that the maxim ‘say no more than you must’, is also related to the Gricean Principle of Quantity 2, or Horn’s (1984)
A DIACHRONIC CORPUS STUDY OF WOULD
51
distinctions require greater clarification outside of conversational texts, but this is beyond the scope of the present study. Similarly, from volition (will) is implied intention, and hence the conversational implicature that the future action will be undertaken (Traugott 1989: 50–1). Bybee et al. (1994: 256) consider that predictive, future meanings developed via the stage of intention, in the following pathway: DESIRE > WILLINGNESS > INTENTION > PREDICTION. For example, an expression of intention, such as: She intends to go, conversationally implicates ‘She is/will be going’. This is not an entailment, as shown by the fact that such meanings may be cancelled or suspended without contradiction: She intends to go, but she will not be going, because they are not going to let her. Such cancellations must be relatively rare, though, and in most contexts the utterance of an intention produces a prediction without the possibility of cancellation; for example, I intend to go, but I’m not going is contradictory, indicating that the former conversational implicatures of prediction derived from the intention meanings of the modal have become conventionalised in such contexts. (However, it is possible to suspend the implicature: I intend to go, but I don’t know yet if it is possible, as I am waiting for some funding to come through.) The strengthening of the implicatures in this way suggests that the grammaticalisation of epistemic meanings from root modal meanings can be seen as a gradual process, in which epistemic meanings are already present as latent pragmatic inferences in the earlier root meanings. Another reason for rejecting the metaphor explanation, as noted earlier, is the verification of the simultaneous presence of both old and new meanings in the same synchronic data, even in the same utterance, indicating that both meanings were being used for different functions at the same period of historical development. Such observations are indicative of change in progress, which forms the basis of much research in variation studies, and which provides additional support as well as an extended application for the explanation itself. An example showing the co-occurrence of old and new meanings in a single utterance is shown in (5):
R-principle. In the previous chapter it was proposed that Quantity 2 describes the inferences in a less subjective way than Relevance, and therefore Q2 will be used in preference to Relevance throughout. Hopper and Traugott (1993: 72) follow Sperber and Wilson (1986) in maintaining that the Principle of Relevance alone is sufficient to account for all the other Gricean principles.
52
HYPOTHETICAL MODALITY
(5)
þa Darius geseah þæt he overwunnen beon wolde, þa wolde he when Darius saw that he overcome be would then wanted he hiene selfne on ðæm gefeohte forspillan himself in that battle kill- ‘When Darius saw that he would be overcome, he wanted to commit suicide in that battle.’ (c. 880, Orosius 3 9.128.5)
Hopper and Traugott (1993: 37) The use of wolde clearly shows that both predictive senses and volitional senses were being used simultaneously in the Old English period, and that the form was polysemous (although Hopper and Traugott 1993 agree with Bybee and Pagliuca 1987 in suggesting that the predictive meanings did not become well-established until the Middle English period). Epistemic meanings have been attested for the Old English period, as shown by Warner (1993) in the fact that modals such as wile (will), mæg (may), mot (must), and sceal (shall) occur with impersonal subjects, but pre-dating the developmental stages does not affect the fact that the route of development remains the same, with stronger subjective meanings not occurring until after weakly subjective ones, and epistemic senses not developing until after deontic or agent-oriented senses.5 All the evidence leads to a unidirectional path of development for modal verbs from main verb uses to premodals, and then from agent-oriented, weakly epistemic modals (including habitual and relative future uses), finally to strong epistemic meanings (Traugott 1989: 43). Moreover, the mechanism by which this path of development proceeds is seen to be the same as the mechanism illustrated in Ch. 2 (2.1.3) by which semimodalised expressions are shown to produce implicatures of the realisation of their predicates: the conventionalisation of a Quantity 2 implicature. It will be seen later in the chapter that the role of Q2 implicatures is important to developing meanings of hypotheticality as well. However, alternative arguments are discussed in 3.2.2. 3.2.2
Hypotheticality
Bybee (1995) discusses the lexical origins of the English modals would and
5. The term ‘agent-oriented’ is used first by Bybee (1985) to refer to root modals in general. According to Bybee, Pagliuca and Perkins (1994: 177), it is defined as ‘the existence of internal and external conditions on an agent with respect to the completion of an action expressed in the main predicate’. A more conservative view of the term ‘agent’ with regard to the subjects of modals is taken in the present study, since it is felt that actions which are not completed at the time of speaking may only have potential agents as subjects. My adoption of the term is therefore only notional.
A DIACHRONIC CORPUS STUDY OF WOULD
53
should appearing as past forms with hypothetical meaning in conditional apodoses as not the result of future meanings used in the past, but as having evolved independently from the use of the more concrete meanings of desire in combination with past temporality (e.g., wold- in Old English texts). Evidence that this is the route of development comes from early uses of the modal in Old English as a lexical verb meaning ‘want’ (Bybee 1995: 505):6 (6)
ðonne sweorda gelac sunu Healfdenes efnan wolde; ‘when the son of Healfdene wanted to practice sword-play’
By the Middle English period, the use of the past modal form could refer to present situations (ibid.): (7)
I wolde yowre wylnyng worche at my myŠt ‘ “I am willing to do your desire as far as I can…” ’
The present time use of the past form of a stative verb, such as a modal, suggests that the state of volition expressed was in existence before the moment of speech, as illustrated in (7), but the completion of the predicated event is not known (Bybee 1995: 506), and is not stated in the utterance. According to Bybee, this indicates that the modality may or may not be still in effect (1995: 507), since the hearer is likely to infer from the use of a past stative form whether or not the ‘desire’ of the past is complete. The expression of only the past volition of the subject to do whatever is described in the complement of the modal may imply that it was not achieved, especially if the past prediction of its achievement was cancelled by the use of another clause in the context, as discussed in Ch. 2 (2.1.3). As illustrated for modal expressions of past ability in (13), Ch. 2, and for would have + V-en in (19), Ch. 2, cancellation by a but (not) clause derives a scalar, Q1 implicature of negative prediction. Under Bybee’s (1995) analysis, the development of the grammaticallyderived conversational implicatures is the result of the use of past tense with the original modal forms which were in most cases stative lexical verbs (e.g. would carried the same meanings as wanted does today). Because of their stativity, there is no aspectual boundedness in their inherent lexical meaning, and the use of a past tense form leaves open the possibility that either the state has terminated or that it is not known whether it continues beyond past into the present (Bybee 1995: 506). That this is a conversational implicature can be ascertained
6. Bybee’s sources are Beowulf (for Old English) and Sir Gawain and the Green Knight (for Middle English).
54
HYPOTHETICAL MODALITY
from the fact that such meanings can be cancelled (as noted in Ch. 2); e.g. a sentence such as she was a teacher, and in fact she still is, with the use of a stative verb in the past, does not convey the definable sense of temporal boundedness inherent in non-stative examples such as he kicked a goal, *and in fact he is still kicking it. The implicature which developed out of the combination of past tense with stative verbs is most likely the result of the constraints of retention of former meanings: pasts were believed to have first grammaticalised out of perfectives, and then at later stages gradually generalised to cover imperfective, including stative environments (Bybee et al. 1994: 92). The presence of irrealis implicatures with past stative verbs is highly likely to be a by-product of the process of the generalisation of pasts to stative environments, with statives still retaining in some uses the former senses of perfectivity associated with the earlier functions of past meaning. Since states cannot be associated with an inherent lexical meaning of boundedness, the sense of perfectivity retained in the grammatical meaning of a past stative must be pragmatic rather than semantic, thus producing a implicature. Bybee et al. (1994: 77–8) also discuss the case of the Germanic preteritepresents, which included modal verb sources such as sculan, and which they believe were lexicalised to refer to present time before the preterite had spread to stative predicates (this is explained by a general grammaticalisation route whereby perfects or anteriors come to mark present states with stative verbs). However, the early appearance of preterite-presents does not account for the occurrence at the same time of examples such as (1) or (6) above in which the modals are used with past tense form and past lexical meaning. If the modals were already lexicalised with present time reference meaning at that time, such examples would have to represent a different grammaticalisation route. Also, the types of counterfactual implicatures obtained in this way are not Quantity implicatures with scope extending only over the complement of the modal (as discussed in 3.2.1), but are related to the grammatical context and the grammaticalisation of past tense in stative verbs, and as such, they have scope over the entire sentence. This contrast will be examined further below 3.2.3–4, and the importance of such implicatures in examining the Singaporean (L2) data will be explained in Ch. 5. 3.2.3
WOULD HAVE
Dahl (1997) discusses the grammaticalisation of counterfactual conditionals in the Germanic languages, and shows that the counterfactual pluperfect form was an innovation which was ‘grafted onto’ an older system that apparently was neutral
A DIACHRONIC CORPUS STUDY OF WOULD
55
to temporal reference: the situation is also exemplified in Bulgarian, in which the perfect auxiliary was introduced later as an optional element in the construction. Furthermore, Bybee (1995: 511) notes that the introduction of perfect have was due to the ‘fading’ of the meanings of pastness (in the modal should) by Middle English, as it became used in more and more present time contexts. The introduction of the perfect auxiliary in this way may then be identifiable with the notion of ‘reinforcement’ discussed by Lehmann (1995: 22 [1982]) as a process which occurs in grammaticalisation due to the decay or weakening of former meanings and which assists in conserving the older meanings. The ‘decaying’ meanings were in this case related to the loss of the retention of perfectivity associated with the first uses of the past tense (as discussed above). Perfect have was already grammaticalised by the Middle English period (see Carey 1994) and could serve the purpose of restoring the weakening senses of completedness or perfectivity which were essential to contribute to the counterfactual implicatures in the meaning of would. It was also likely to have increased the factual premise for a counterfactual utterance by situating the prediction within the speaker’s immediate domain of reference, perfect forms being associated with inferences of ‘current relevance’ (see Fleischman 1983). The counterfactual inferences are increased then not only by perfectivity, but also by relevance to a current reference point. Thus, as the use of WOULD as a counterfactual modal grammaticalised, the retention of lexical meanings of perfectivity, as well as those of volition, were becoming more opaque, and the conditional WOULD could be used to refer to any time, as Dahl suggests. The introduction of the have auxiliary also had the function of increasing the morphological complexity of the construction, making for more informational coding, and hence quantificationally marking the expression by comparison to the simple preterite form of the modal. Thus, the increase in expressive quantity, as well as the restoration of the bleaching meanings of pastness and the inferences of current relevance associated with perfect constructions generally, would have contributed to the redevelopment of counterfactual, Q1 implicatures in past volitional constructions, while the loss of perfective inferences, as described in 3.2.2, may have influenced the rise of hypothetical meanings. The usual readings of counterfactuality derived from would have+V-en expressions today are partly due to such reinforcement. There still remains, though, evidence that grammaticalised predictive meanings were possible in the early use of would; these are commonly known as future-in-the-past, and, like non-past futures, are not the result of Q1, but Q2, implicatures. Such meanings are discussed in 3.2.4.
56
HYPOTHETICAL MODALITY
3.2.4
Future-in-the-past
As noted in 3.2.1, Traugott (1989) gives a comprehensive account of the development of English epistemic modal verbs from their deontic or root sources in Old English, and attributes the development to the conventionalisation of conversational implicatures, invoking the explanation of the Relevance Maxim, which is related to the lower-bounding Maxim of Quantity 2. The expression of volition or intention to act may be taken as the prediction of that action being realised; in other words, a minimal expression is used to communicate a more extensive interpretation. Non-past modal forms such as those expressing meanings of ability, permission and volition were all regarded as the sources for future-projecting or epistemic meanings, derived in this way by the conventionalisation of Q2 implicatures. In Ch. 2 (2.1.3) it was suggested that past temporal reference, prosodic stress, and the presence of adversative clauses in the context could induce a drift from Q2 implicatures to Q1 implicatures in the use of past modal and semimodalised expressions such as in example (13): Ch. 2. Q2 implicatures are thus just as likely to be present in the use of the preterite form of the modal as they are in non-past forms of the modal. As mentioned earlier, Traugott refers to the development of a relative tense use of the modals in Old English which is likely to have preceded both the deictic tense use and the hypothetical/conditional uses, in which the time reference point is not the time of speaking (1989: 40), and this usage often carried implications of the realisation of a predicted event, rather than its non-realisation. In the following Old English example, wolde is listed by Warner (1993: 168) as expressing proximity (‘was about to’), meanings similar to those described by Heine (1994c) as the ‘proximative aspect’: (8)
Se untruma eac wacode oðþæt hit wolde dagian ‘the sick-man also stayed-awake until it was-about-to dawn’. (ÆLS i.21.123)
The use of wolde (‘would’) in (8) is clearly substitutable in present-day English by a proximative adverb such as almost, but the implicatures are of the realisation of the predicted event, rather than its non-realisation as is often the case with items expressing proximity to a predicted event. This demonstrates the fact that Q2 implicatures are basic in such contexts, and Q1 implicatures may be derived from expressions or features which modify the basic implicatures, such
A DIACHRONIC CORPUS STUDY OF WOULD
57
as conditional clauses.7 The subject in (8), furthermore, is impersonal, which places doubt on the possibility of volitional meanings. It is quite likely that the implicature of the realisation of the predicted event is due to certainties known which are extraneous to the construction, such as knowledge that the dawn always breaks. However, in addition, the predictive meanings of would in (8) may be due partly to its occurrence in a subordinate clause which contains a presupposition of factuality. This contrast is especially apparent in (5) above, which has wolde occurring in both a dependent clause and a main clause. The implied realisation of the proposition predicted in the subordinate nounclause in (5) is attributable to the fact that such clauses contain presupposed or given information, not assertions, and that the dependent clause is factive, since it reports perceived evidence. On the other hand, the use of wolde in the main clause function could derive hypothetical interpretations as it is new information and the outcome of the subject’s intentions is not known. It could well be the case that the early use of WOULD carried Q2 implicatures of the imminent realisation of the predicated event or Q1 implicatures of its non-realisation, according to whether the modal occurred in a (factive) subordinate clause or a main clause environment, or at least according to the presence of a time reference in the discourse, relative to which the prediction could be made. The level of factuality associated with the meaning of WOULD could therefore be determined by the context in which the modal appeared. It was suggested by Fleischman (1982) (cited in Bybee 1995: 515), that futures-in-the-past served as the source for the development of conditional meanings in Romance, and that the conditional use developed out of an earlier use of future-in-the-past; for example: Peter said he would come (if he got the money) illustrates the use of both future-in-the-past () and conditional use at the same time. This hybrid example, though, is not taken from any historical sources, and the argument is disputed by Bybee (1995), who finds examples in Classical Nahuatl in which a conditional has evolved from a past volitional verb without a corresponding future form developing from its present form. However, Traugott demonstrates that subjective futures, in which the time of reference is the time of speaking, did not develop until after a relative tense use had appeared in Old English, and provides a ME example in which a hypothetical meaning is derived from the use of a relative future form in a non-factive complement clause (Traugott 1989: 40):
7. Horn (1996: 21) maintains that almost is upward-implicating on a Quantity scale (this suggests that senses of prediction may be contained in its meaning).
58
HYPOTHETICAL MODALITY
(9)
Seynt Austyn … prayede to god þat alle childerne þat shulde bene borne afterwarde in þat citee of Rouchestre moste haue tailes. ‘St. Augustine … prayed to God that all the children that were to be born after that in the city of Rochester would have tails.’ c. 1400 Brut 1333 (Rwl B.171) 97/12
Although the example uses the past modal forms shulde and moste (implying obligation), not WOULD, it indicates, nevertheless, that the potential for a modal to express either hypotheticality or prediction is related primarily to the environment in which it occurs: the complement of a verb such as pray is likely to express less certainty than that of saw in (5), and as such, affects the modal reading. Furthermore, it is generally known that children do not have tails, and thus real-world knowledge will also play a role in determining the meaning of the modal. The source of future-in-the-past meanings as derived directly from past volition sources and not from grammaticalised present future forms is also well attested. Kuteva and Heine (1995) also note that in Old Bulgarian, a future-inthe-past meaning developed directly from a past volitional verb source and emerged before the deictic future had developed independently, and, in their data, counterfactuality is presented as a stage subsequent to future-in-the-past. The possibility of the relative future tense use for WOULD in Old English serving as the source for a conditional or hypothetical use is made more likely by the fact that predictive, or Q2 implicatures, were beginning to surface in the Old English modal WOULD, in particular grammatical environments such as factive subordinate clauses, as shown in (5). In contrast, Bybee (1995) demonstrates that main clause environments are often conducive to the development of hypothetical implicatures, but are derived in a different way. Evidence of this nature suggests that the type of implicature involved, and therefore the function of the modal, were determinable by the context in which the modal appeared. We are therefore presented with two possible factors contributing to the development of hypothetical meanings in WOULD: (i) the counterfactuality of the entire past stative clause, whether it is a modal or a non-modal form, leading to meanings of hypotheticality, as illustrated by Bybee (1995) and discussed in 3.2.2, and (ii) the counterfactuality of only the complement of the modal, which may have been derived from a relative future used in the past to suggest a prediction which was subsequently cancelled. Bybee (1995) uses the counterfactuality of the complement clause (Y did not happen) to account for the suspended counterfactuality of the entire sentence (X wanted to do Y, (but did not) and therefore possibly still wants to), but the counterfactuality of the entire
A DIACHRONIC CORPUS STUDY OF WOULD
59
sentence may be obtained without such justification, as discussed earlier and in 5.2. (Ch. 5). It will be seen that both situations are relevant to the grammaticalisation of the modal would in English, but that Quantity implicatures are relevant only to the counterfactuality of the complement clause. It could be held that some hypothetical functions of WOULD develop via (i), with counterfactual scope over the entire clause which weakens to hypothetical with the loss of retention of perfectivity, but that counterfactual meanings develop via type (ii), with the modal later reinforced and strengthened with the HAVE auxiliary to maintain the retention of perfectivity. This suggestion will be tested against the data below, illustrating the distribution of WOULD over a diachronic time span.
3.3 The data The diachronic data selected for the investigation were taken from the Helsinki Corpus of English Texts, and are representative of 3 time periods extending from 850 to 1570: 850–950 (Old English — OE); 1050–1420 (Middle English — ME); 1509–1570 (Early Modern English — EME).8 Two texts provided the sources for the Old English period, Alfred’s Boethius and Two of the Anglo-Saxon Chronicles; four texts were samples for the ME period, History of the Holy Rood Tree, Vices and Virtues, Chaucer’s A Treatise on the Astrolabe, and Chaucer’s Boethius. (The latter two were included because they are illustrative of a later period of the ME data.) The EME selection included Elyot’s The Boke Named the Gouernour, Ascham’s The Scholemaster, Colville’s Boethius, Fisher’s Sermons, and Latimer’s Sermon on the Ploughers. Also included were The Trial of Sir Nicholas Throckmorton; The Statutes of the Realm; and Fitzherbert’s The Book of Husbandry. The texts were examined for occurrences of WOULD, and examples were randomly selected from the available data, with sampling sizes enabling a minimum of 70 examples to be found per period. The extracted data was then classified, with the aid of published translations, into either a lexical function category or a grammatical function category, and ambiguous examples were listed separately (the sub-categories are explained below).9 The number of tokens was listed for each period according to the functional category, and in the
8. I am most grateful to Matti Rissanen and Merja Kyto for permitting me access to the diachronic portion of the Helsinki Corpus of English Texts. 9. The published translations were Giles (ed.) (1969 [1858]) for Alfred’s Boethius; Whitelock, Douglas and Tucker (eds.) (1961) for The Anglo-Saxon Chronicle, and the Early English Text Society for both The History of the Holy Rood Tree and Vices and Virtues (see the references for more detail).
60
HYPOTHETICAL MODALITY
sub-categories according to the grammatical environment, the hypothesis being that within the broader divisions of lexical or grammatical functions, the meanings implied in the use of the modal will vary according to the grammatical environment, a factive or a non-factive clause. It was predicted, for example, that a factive clause, carrying presuppositions, would provide the optimum environment for the development of predictive meanings in the use of WOULD appearing in it, while a non-factive (irrealis) environment would influence the development of hypothetical inferences. From this information, it was possible to determine the direction of the Quantity implicature and the likelihood of a predictive, a hypothetical or a counterfactual meaning emerging. It is also possible to demonstrate quantitatively a given trend in the grammaticalisation of WOULD across an extended time period. 3.3.1
Notes on the categories
As far as possible, the divisions between lexical and grammatical functions were determined on the basis of contextual clues, but it must be remembered, as noted earlier, that the divisions were in no way intended to be absolute, merely focal, and are presented in the data to provide a guide as to the progress of the grammaticalising modal at different time stages. Clearly ambiguous cases were isolated as a separate category; others were decided on the basis of contextual and co-occurrence features. Lexical or volitional uses were determined unambiguously by the presence of a nominal or directional object, but also often included a finite or a non-finite noun clause complement. These examples were evaluated on the strength of the available translations for the OE and ME periods, in which the modal would be interpreted usually with either would or another verb of volition, such as desire, or wish, or a construction such as be willing. Other factors taken into consideration were animacy of the subject, subject person (1st person subjects being most amenable to volitional interpretations), and negation, which also strongly suggests volitional interpretations in modals (as discussed in Ch. 6:1). Grammatical or predictive uses included all instances of WOULD in which there are clearly no ambiguities or lexical retentions, the latter such cases being listed under Ambiguous. Grammatical uses included the relative futures discussed above which are held to have a future, predictive meaning relative to a given past time reference contained elsewhere in the context. These uses were subclassified (as explained below) according to whether they occurred in a
A DIACHRONIC CORPUS STUDY OF WOULD
61
factive environment (inducing a Q2 implicature of the realisation of their complements) or a non-factive environment (inducing a hypothetical reading).10 Counterfactual environments were also included, as they express prediction and are indicated grammatically in English with the additional auxiliary HAVE, and as such represent a grammaticalised stage of development (no lexical retention is possible in most cases). Other less frequently represented grammatical environments are shown in the results. 3.3.1.1 Environments Lexical examples with nominal complements included those with nominal that-clauses as complements, as illustrated in (10), one or two noun phrase complements, and those with verbal complements which showed the function of an emerging auxiliary in the use of the modal in a function equivalent to a catenative verb of volition. Directional complements are also found (rarely) in the ME data, e.g. (19), and are possibly derived as a branch-off of the INTENTION stage illustrated by Bybee et al. (1994) in the grammaticalisation of volitional meanings to become futures. In such examples there is neither a nominal nor a verbal complement, but a locative one. Factive clauses in which the function is lexical are found as subordinate clauses throughout the data and include nominal clauses introduced by items equivalent to what, adverbial clauses of reason, relative clauses with referring head nouns, temporal adverbial clauses, and nominal clauses (objects of verbs of reporting or communicating, introduced by conjunctions equivalent to that). Non-factive subordinate clauses in which the function of WOULD is lexical include conditional clause complements introduced by conjunctions with the meaning ‘whether’, as well as those functioning as protases, introduced by if and equivalents, and adverbial clauses of manner occurring as imperatives and interrogatives (see for example, (29)). Grammatical environments were more diverse. Main clause evironments included conditional apodoses, and (rarely) future-in-the-past (only 2 examples appear in the entire sample), hypothetical interrogatives, and, especially in the EME data, non-conditional hypotheticals and counterfactuals, indicating the
10. The term ‘factive’ is not intended to be defined in exactly the same way as Karttunen (1971b) defines such predicates, though in some cases there is overlap with Karttunen’s definitions; e.g. a relative clause with a referring head noun will be classed as factive as it usually contains a presupposition. However, the factivity associated with complements of verbs of reporting or communication (that-clauses) depends on the nature of the matrix verb: e.g. pray/would that introduces a non-factive environment, while said that; promise that is taken in the present study to be factive.
62
HYPOTHETICAL MODALITY
extent of the grammaticalisation of such meanings in WOULD. The extent of grammaticalisation is also represented in the appearance of examples in which the modal co-occurs with another verb of volition, or with an adverb expressing volition (e.g. Šyrnende ‘fain’ in (17)). Such co-occurrences are indicative of the loss of lexical significance in the modal, and the need for reinforcement of such meanings with other items expressing volition. Grammatical functions in subordinate clauses include non-factive complements of verbs of considering, thinking, praying and wishing, in which the meaning of WOULD is hypothetical, non-factive clauses introduced by ‘unless’, comparative clauses including those introduced by items meaning ‘rather than’, relative clauses with non-referring head nouns, one rare example of a habitual sense in an if-clause (20), conditional protases with inanimate subjects (and some animate ones), relative clauses in which a habitual meaning is expressed, clauses introduced by items meaning ‘whether’, interrogative clauses, and occurrences in which the matrix clause is also WOULD (23).11 Factive environments included complements of verbs of communicating or reporting in which the meaning is predictive, nominal that-clauses and clauses of purpose, referring relative clauses, and occurrences in which there is a co-occurrence with another volition verb (e.g. wilniæn). Proximative uses (18) were listed as counterfactual, and directional uses with locative complements, including the one ablative directional sense mentioned earlier (19), have been considered separately although they appear in that-complements of verbs of reporting, as they are of special interest. Other grammatical subordinate clause categories include counterfactuals in which WOULD HAVE appears in the protasis (these occur in the EME data, e.g. (26)). Habitual uses of WOULD are very rare; in the entire sample, only one example is found in the EME data, and one example in the ME data (20), and both are in subordinate clauses.12
11. The habitual uses could equally be classed as factive, as, according to Givón (1994), habituals represent a ‘swing’ category between realis and irrealis meanings. 12. Visser (1969: 1710) claims that the use of WOULD in past habituals is very frequent in his data.
63
A DIACHRONIC CORPUS STUDY OF WOULD
3.3.2
Results
Results of the diachronic search13 Table 3.1
Distribution by function and environment of the modal forms wold-/ uuold/nold- in the OE sample lexical
grammatical
ambiguous
total N
main clauses sub-clauses
24 (32%).0 20 (26.5%)
10 (13%) 16 (21%)
0 (8%) 6 (8%)
34 42
totals
44 (58.5%)
26 (34%)
6 (8%)
76
Environments (main clauses)
N
%
lexical:
+nominal complement +verbal complement
09 15
12% 20%
grammatical:
conditional apodoses other irrealis
06 04
08% 05%
Environments (subordinate clauses) lexical:
factive clauses non-factive (irrealis)
15 05
20% 06.5%
grammatical:
factive clauses Non-factive (irrealis)
14 02
18% 03%
ambiguous:
factive clauses
06
08%
OE sample: 32,137 words
13. Note that all percentages in the tables, unless specifically indicated, are rounded to the nearest 1% or 0.5%.
64 Table 3.2
HYPOTHETICAL MODALITY
Distribution by function and environment of the modal forms wold-/uuold/ would/wald-/nold- in the ME sample lexical
grammatical
ambiguous
total N
main clauses sub-clauses
22 (25%) 20 (23%)
10 (11%) 24 (27%)
02 (02%) 10 (11%)
34 54
totals
42 (48%)
34 (38%)
12 (13%)
88
Environments (main clauses)
N
%
lexical:
+nominal complements +verbal complements +directional complements
03 17 02
03.5% 19.5% 02%
grammatical:
conditional apodoses other irrealis
02 06 02
02% 07% 02%
ambiguous:
conditional apodoses other irrealis
01 01
01% 01%
Environments (subordinate clauses) lexical:
factive clauses non-factive clauses
14 06
16% 07%
grammatical:
factive clauses non-factive clauses counterfactual clauses directionals
11 09 02 02
12.5% 10% 02% 02%
ambiguous:
factive clauses non-factive clauses
08 02
09% 02%
ME sample: 54,980 words
65
A DIACHRONIC CORPUS STUDY OF WOULD
Table 3.3
Distribution by function and environment of the modal forms wold-/would in the EME sample lexical
grammatical
ambiguous
total N
main clauses sub-clauses
05 (4.5%) 06 (5.5%)
41 (37%) 48 (43%)
07 (06%) 04 (04%)
053 058
totals
11 (10%)
89 (80%)
11 (10%)
111
Environments (main clauses)
N
%
lexical:
nominal clause complement verbal complement
03 02
02.7% 01.8%
grammatical:
conditional apodoses non-conditional hypotheticals non-conditional counterfactuals habituals
20 15 05 01
18% 13.5% 4.5% 01%
ambiguous:
verbal complements
07
06%
Environments (subordinate clauses) lexical:
non-factive clauses factive clause (proximative)
05 01
4.5% 01%
grammatical:
non-factive clauses factive clauses counterfactual clauses
24 16 08
22% 14% 07%
ambiguous:
factive clauses
04
04%
EME sample: (51,009 words)
66
HYPOTHETICAL MODALITY
3.4 Discussion For the most part, the differences between the categories lay in whether or not the item could be categorised into a subordinate clause category or a main clause one, and the function of the item could be determined on that basis. The data shown in Tables 3.1–3.3 clearly indicate a changing pathway of development in the generalisation of functions of WOULD from a predominantly lexical verb in OE with meanings of pure volition to a marker of mainly hypothetical functions, later reinforced with the perfect auxiliary HAVE (see discussion below). The direction of the Quantity implicature, though, implying either the predicted realisation or non-realisation of the complement of WOULD, appears to be determined by the grammatical environment. A review of the data may explain this. 3.4.1
The OE sample
In the OE sample, the primary function is the lexical function, as indicated by the proportionately high number of occurrences (58.5%). The following example illustrates a typical lexical use of the modal verb, appearing with a nominal clause as object: (10)
… ac ic wolde nu þæt ðu wende þin ingeþonc from þæm leasum gesaldum; ‘… but I am desirous that you turn your attention from the false goods;’ 850–950. Alfred’s Boethius (ed. Sedgefield) 33.78.3
It is interesting to note though, that even in this example in which the lexical meanings are quite clearly indicated, the temporal adverb nu (‘now’) was not included in the translation (in Giles, ed. 1969[1858]). The adverb is a crucial indicator that both past and present temporal meanings were possible in the use of the modal at this stage, and required additional contextual support to disambiguate them. It is such environments, in which past desire further implies present desire, that are claimed by Bybee (1995) to be instrumental in the development of hypothetical meanings generally, including those in conditional apodoses. However, their occurrence appears to be restricted, at least initially, to first person subjects expressing wishes, and a link would need to be established between such uses and the more general hypothetical uses (with a wider range of subjects) in conditional apodoses and other irrealis environments (e.g. (13)). The language used in the published translation is somewhat archaic, and the use of
A DIACHRONIC CORPUS STUDY OF WOULD
67
wolde in (10) could be translated in today’s English as ‘I would like you to’, in which an additional verb, like, reinforces the older meanings of desire lost in the grammaticalisation of would, or ‘I wish you would’ which has a slightly more manipulative meaning. The translation indicates the persistence of volitional meanings (but not past time reference) in PDE, demonstrating that volitional senses are retained longer with first person subjects; this may be for two possible reasons: (i) first person subjects, being usually of the same identity as the speaker, allow for a greater possibility of the speaker’s awareness of the subject’s volitional intentions, and therefore maximise the chances of a volitional reading (as shown in Ch. 6; see also 2.1.4); and (ii) if the speaker is most likely to be aware of the subject’s intentions, the need to hypothesise about the subject’s future actions is much less than it would be for a second or third person. Such reasons are important in considering the development of hypothetical meanings from volitional ones. In addition, the number of grammaticalising occurrences in the data is already rising at the OE stage, as indicated by the figures (34% + 8% classed as clearly ambiguous). A large proportion of the grammatical functions are provided by factive environments in subordinate clauses (18% overall), and relatively few examples occur in conditional apodoses (only 8%), or other irrealis environments (5%). The category of factive subordinate clauses includes noun clause complements (often introduced by verbs of speaking, quoting, or reporting) relative clauses with definite head nouns, and temporal adverbial clauses. Such environments may provide a high probability of predictive, (relative future) meanings, as being factive, they carry presuppositions of the truth of their complement sentences, and thus are more likely to predict the realisation of the complement of WOULD or any other modal contained in them; such predictive, future-projecting meanings are the result of inferences associated with the second maxim of Quantity, Q2 (as discussed in 3.2.1). It is for this reason that such environments may be seen as promoting the grammaticalisation of the modal to past predictive meanings under the operation of Q2 implicatures, and this stage of grammaticalisation has been hypothesised to have occurred at an early historical period, earlier than the grammaticalisation of subjective, non-past will to meanings of future prediction (Traugott 1989: 40). One example of a grammaticalising occurrence in a subordinate clause is the following: (11)
… & him eac geheton þæt hiera kyning fulwihte onfon wolde, & hie þæt gelæston swa; ‘and promised that their king would receive baptism, and they kept their promise’
68
HYPOTHETICAL MODALITY 850–950. Two of the Saxon Chronicles Parallel (ed. Plummer) 878.20 (p. 76)
The past use of wolde in (11) is predictive relative to the temporal context specified in the matrix verb, and the implicatures are that the promise was realised (confirmed by the clause which follows). A more ambiguous example, in which the lexical meanings may be seen to be still present alongside the predictive meanings, is illustrated in (12): (12)
… & sæde þæt he wolde oðer oððe þær libban oððe þær licgan. ‘… and he said that he wanted to/would either live there or die there.’ 850–950. Two of the Saxon Chronicles Parallel (ed. Plummer) 901.8 (p. 92)
The ambiguity is due to the content of the reported clause, which conveys a sense of the subject’s resolution and hence volitional intentions, but the environment also supports a predictive interpretation, as it expresses a future situation perceived as relative to a past time point indicated by the tense of the main clause verb. Such ambiguities reveal a possible environment in which grammaticalising predictive senses are seen to overlap with the older lexical senses, and it is from such environments that predictive senses may generalise to other environments in which the older volitional meanings are not possible, as will be discussed in 3.5. Non-factive, or irrealis environments also appear in the OE sample, as illustrated in the following subordinate clauses, when WOULD occurs with a grammatical function (in (13)), and with a lexical function (in (14)): (13)
… geþenc nu hu weorðlic & hu foremærlic þe wolde se man þincan. ‘… consider now, how honourable and how eminent the man would seem to you.’ 850–950. Alfred’s Boethius (ed. Sedgefield) 33.75.21
(14)
… swa þæt he mehte ægþerne geræcan gif hie ænigne feld secan wolden; ‘… so that he could reach either army if they wanted to come into the open country.’ 850–950. Two of the Saxon Chronicles Parallel (ed. Plummer) 894.6 (p. 84)
The non-factive (irrealis) environment in (13) is provided by the complement of an imperative and in (14) by the conditional protasis, and the translation in (13) indicates the use of wolde co-occurring with a stative verb, an environment which would be unlikely if the lexical meanings were still present. Environments such as (13), though not as frequent as factive environments, seem to suggest
A DIACHRONIC CORPUS STUDY OF WOULD
69
that WOULD was already grammaticalising to suggest unambiguously hypothetical meanings in subordinate clauses at the time when hypotheticality was beginning to appear as an inference in environments such as are illustrated in (10), and, furthermore, they carried no indication of any lexical retention of volitional senses. Thus, both source and target functions were co-occurring at the same time in the OE texts sampled. There are six examples of WOULD appearing in conditional apodoses (only 8%), including one example in which the main clause is the apodosis of a concession clause rather than a conditional clause. In only the latter instance is the subject a first person subject. One of the six examples appears as an interrogative with a conditional premise, thereby increasing the likelihood of a hypothetical reading: (15)
Gif þu nu deman moste, hwæðerne woldes þu deman wites wyrðran, ‘If you were now to judge, which would you judge more deserving of punishment,’ 850–950. Alfred’s Boethius (ed. Sedgefield) 38.122.28
Interestingly, there is one example of HAVE co-occurring with WOULD in the OE sample, and this instance certainly suggests a grammaticalised counterfactual meaning; there are no traces of volitional meaning: (16)
ða cwæð he: Mid hu micle feo woldest þu nu habban geboht þæt þu meahte ongitan hwæt þæt soðe god wære, … ‘Then said he: With how much money would you now have bought, that you might know what the true good was’ 850–950. Alfred’s Boethius (ed. Sedgefield) 34.89.26
The rarity of this example in the OE data, though, is possibly because, as explained in 3.2.3, HAVE was not required until the senses of past temporality had begun to fade in the lexical meaning of the modal, and, according to Bybee (1995: 511), this did not occur until the ME period. In general, then, the majority of OE examples are lexical, appearing in main clauses with a nominal or verbal complement. There are indications of hypothetical meanings emerging in non-factive environments, but many of the grammatical environments are factive (18% in subordinate clauses, compared with 3% non-factive), illustrating a relative future meaning in WOULD in which the past temporality is provided by the time reference of the main clause, e.g. (8) and (12). The distribution of such meanings will be seen to change slowly over the ME period.
70
HYPOTHETICAL MODALITY
3.4.2
The ME sample
In the ME sample, the number of lexical functions declines slightly (to 48%) relative to the grammatical functions which show a 4% rise. There is also an increase in the number of tokens which could be considered ambiguous (13%). While such changes are not significant, they do indicate the state of gradual generalisation of the predictive modal meanings at this time. There is also an increase in the range of grammatical environments in which the modal may be found, a further indication of its advancing grammaticalisation. The number of verbal complements of main clause uses increases significantly relative to nominal complements, the proportion of which drops by 8.5%. The use of the modal in conditional apodoses is still not proportionally high, remaining at around 8%. The proportion of factive subordinate clauses drops from 46% to 37.5%, and the number of non-factive environments increases by 10%. Clearly, the grammatical functions of WOULD appear to be emerging in subordinate clauses ahead of main clauses in the data, though the proportions of factive environments (rendering predictive meanings) versus non-factive environments (rendering hypothetical meanings) do not show significant differences during this period. One example with both a main clause and a subordinate clause grammatical use is the following: (17)
Ac ic wolde beon Šyrnende Šif hit godes willæ wære. & ic hit wurðe wære þt ic mid mine eaŠen iseon moste þt þt ic to ðe wilniæn wolde ‘But I would fain, if it were God’s will and I were worthy of it, be permitted to see with my own eyes that which I would ask of thee.’ 1150–1250. History of the Holy Rood-Tree (ed. Napier) p. 6
(17) is an important illustration of the extent of the grammaticalisation of WOULD at this time, as it is seen to co-occur with an adjective expressing desire (Šrynende), translated by ‘fain’ in the main conditional clause, and also co-occurs with a verb expressing desire (but not translated as such) in the subordinate factive clause (wilniæn). In order for such co-occurrences to be acceptable without senses of redundancy, the modal must be semantically bleached of many of its older volitional senses. Hagège (1993: 200–2) discusses a phenomenon known as the Proof by Anachrony (PA) Principle, in which a grammaticalised item can be used in contiguity with its historical source without redundancy, thus offering proof of the extent of its grammaticalisation. Although (17) does not illustrate an example of such contiguity, it is seen as contiguous with items with the same, or a similar semantic content, which suggests sufficient evidence of the loss of some of its lexical senses to propose classifying the
A DIACHRONIC CORPUS STUDY OF WOULD
71
example as having a grammatical function. The example is also typical of the environments indicated by Bybee (1995) as ambiguous between the uses illustrated in (10) and other conditional hypothetical uses. Other grammatical environments in subordinate clauses include two examples which have been labelled counterfactual clauses, as they contain uses of WOULD which illustrate a situation of proximity to an event, similar to that described in Heine (1994c) or Kuteva (1998). However, they are rare in the data, and also in Visser’s (1969) data, as mentioned earlier. One example is the following: (18)
Sonæ swa he heom on hond nimæn wolde þa wurdon heo feringæ all furenne swa ðæt þam cnihte forburnon ba twa þa handæn all buton ða earmæs. ‘As he was about to take them into his hands, they suddenly became all fiery so that the servant’s hands were entirely burnt except his arms.’ 1150–1250. History of the Holy Rood-Tree (ed. Napier) (p. 10)
In (18) there is a prediction of the realisation of the complement of wolde, derived by a Q2 implicature, which is subsequently cancelled by the event described in the main clause. The result is therefore one of the non-realisation of the complement of the modal and an implicature of counterfactuality instead. This is illustrative of the way in which Q1 inferences, associated with counterfactuality, are obtained by the cancellation of Q2 inferences of prediction, as discussed in Ch. 2 (2.1.3). Another unusual use of WOULD is found in two directional uses, in which the modal is complemented by a locative expression, with no complement verb, suggesting that the modal itself acts as a main verb of direction. The published translation, given in the following example, indicates the directional meanings intended in the use of nolden and offers a parenthesised verbal complement to provide PDE idiomaticity: (19)
ða ðe dauid þt onŠeat þt heo nateshwon of ðam stede noldon; … ‘When David perceived that they would by no means [be moved] from the spot’ 1150–1250. History of the Holy Rood-Tree (ed. Napier) (p. 22)
Interestingly, the direction is ablative, not allative as would be expected. Although there could be possible ambiguity between a lexical and a grammatical interpretation of nolden, the example has been classed as grammatical due to the constraints of the translation — the subject refers to inanimate objects (the rods), and volitional senses are therefore unlikely, even in the presence of a negative.
72
HYPOTHETICAL MODALITY
Grammatical functions in non-factive subordinate clauses were almost as frequent as those in factive clauses. In the example given below, the use of wolde is that of a past habitual marker in a conditional protasis, the conjunction Šif supplying the meaning of ‘whenever’; the presence of a past indicative verb in the apodosis clearly indicates that the function is not hypothetical. (20)
ac Šif min lauerd godd me wolde swingen mid ani swinge, al swa fader doð his sune, ic was ðar aŠean unþolemod, ‘but if my Lord God would chastise me with any rod, as a father does his son, I was impatient at it’ 1150–1250. Vices and Virtues, Part 1. (ed. Holthausen) (p. 13)
As noted above, habitual uses of WOULD are infrequent throughout the data, and thus it is unlikely that they might have had any influence on the development of the conditional use, in spite of the surface similarity of both types of constructions in many cases. The past indicative in the apodosis of (20) serves to distinguish such constructions from hypothetical conditionals which begin to appear in the OE data, as shown in (15) and (16) above, and which often appeared (in the data) with an adverb such as nu (‘now’), indicating that the finite verb, the modal, was non-past in time reference. Although many grammatical functions are emerging in the ME data, the use of WOULD as a main clause lexical verb continues, as indicated in the relatively high proportion of tokens (25%). Even in the later texts of Chaucer, such uses appear relatively frequently. One example is the following: (21)
Ensample as thus: The yeer of oure Lord 1391, the 12 day of March at midday, I wolde knowe the degre of the sonne. 1350–1420. Chaucer, A Treatise on the Astrolabe (ed. Benson) p. 669.C1
The use of wolde in (21) could be interpreted out of context as expressing a predictive, grammatical function, but the text clearly indicates that the meaning is one of past volition, as it continues: I soughte in the bakhalf of myn Astrelabie and fond the cercle of the daies, the past time reference conveyed in the past tense of non-modal verbs in the context. Lexical uses are also found in the ME sample in subordinate clauses, including the following, in a non-factive environment: (22)
and he hadde auŠene kere te donne hwaðer swo he wolde. ‘and he had [his] own choice to do whichsoever he would’ 1150–1250. Vices and Virtues, Part 1. (ed. Holthausen) p. 113
A DIACHRONIC CORPUS STUDY OF WOULD
73
The environment is non-factive because of the presence of a non-referring whitem introducing the complement clause, and is classed as lexical since the modal could be readily substituted with ‘wanted to’ or ‘desired’ in PDE. However, the presence of the non-referring wh-item also makes it possible to foresee the potential for hypothetical inferences to develop in such environments. The ME data, then, indicate some gradual changes taking place in the distribution and function of WOULD, but do not show any marked increases in the degree of grammaticality reached at this stage, which rises by only 4% from its OE level. Even in the later period in the Chaucer sample, the number of unambiguously grammatical examples is less than half the total (9/20 = 45%). The reason might be that the time periods of the OE and the ME samples are closer than those of the ME and the EME sample, the latter showing much more dramatic increases in the grammatical functions of WOULD, as discussed below. 3.4.3
The EME sample
As indicated in Table 3.3, the frequency of occurrence of WOULD as a grammatical item rises sharply (to 80%) in the sample taken from EME texts. There is also a rise in the frequency of occurrence overall compared with that of the ME period: from a sample of 54, 980 words, the number of ME tokens found was only 88 (probability = 0.0016); while for the EME sample, the total number of tokens was 111 in 51, 009 words (probability = 0.0022). The number of texts was far greater in the EME sample, so the increase in frequency in the EME data cannot be attributed to the idiomatic style of a particular author; rather, it is an indication of the advanced stage of grammaticalisation of the modal in that it becomes more versatile in distribution. The highest number of grammatical uses appear to be in conditional apodoses (18%), and non-factive subordinate clauses (22%), but there is also a new distribution which was not shown in data from earlier periods: non-conditional hypothetical and counterfactual main clauses. Most of these examples imply a condition which is absent from the context, suggesting that the hypothetical function of the modal is well established by this period. Lexical uses for the modal have decreased to only 10% of the total by this period, but counterfactual uses with HAVE are increasing (11.5% — there being no examples with HAVE in the ME data). The number of clearly ambiguous uses remains relatively unchanged, at (10%). Thus, it would appear that a rather sudden change has taken place by this period, in the increase in grammatical uses. Examples are listed as follows.
74
HYPOTHETICAL MODALITY
One example of a main clause lexical use, with a nominal clause complement, is shown in (23): (23)
Well, I woulde al men woulde loke to their dutie, as God hath called them, 1549. Latimer, Sermon on the Ploughers (ed. Arber). p. 29
The first use of woulde is the same as in (10) and is expressed in PDE as ‘wish’, or ‘would like’, the presence of a nominal clause complement indicating a persisting lexical use which appears to be declining elsewhere. The second example of woulde in (23), however, is already grammaticalised as a marker of hypotheticality, and has been listed as occurring in a non-factive environment for this reason. An example of a lexical use appearing with a verbal complement is given in the following: (24)
Whan a furious and wylfull yonge man in a sedicion had striken out one of the eies of kyng Licurgus, wherfore the people wolde haue slaine the transgressour, he wolde nat suffre them, but hauyng him home to his house, he by suche wise meanes corrected the yonge man, that he at the laste broughte hym to good maners and wisedome. 1531. Elyot, The Boke Named the Gouernour (ed. Rhys) p. 152
The lexical use is shown in the second example, and is suggested by the presence of a negative. The first example is listed as a counterfactual use appearing in a subordinate clause, and illustrates one of the early examples of WOULD HAVE (+V) in the data. The use of the auxiliary haue marks the meaning as counterfactual, and suggests a grammatical function, though it does not appear in a conditional sentence. An example of WOULD appearing in a conditional apodosis is given in the first occurrence in (25): (25)
What woldest thou say yf that a man had vtterly lost his sight and also hadde forgotten that euer he sawe, and yet dyd thynke that he lacked nothing of the perfection of a man would not we that saw the same iuge that he were blynde. 1556. Colville, Boethius (ed. Bax) p. 102
In the first instance above, WOULD appears in a main clause apodosis which is also interrogative; the second occurrence is in a non-factive subordinate clause, an embedded conditional protasis marked by inversion rather than a conjunction.
A DIACHRONIC CORPUS STUDY OF WOULD
75
The use of the modal in the protasis of conditionals does occur in the EME data, in some instances also with HAVE: (26)
And yet, if either ambition or voluptuouse idelnes wolde haue suffered that reder to haue sene histories, he shuld haue founden excellent princis, as well in payntyng as in keruynge 1531. Elyot, The Boke Named the Gouernour (ed. Rhys) p. 28
(26) is listed as a counterfactual subordinate clause because of the presence of HAVE. Apart from (24) above, other non-conditional counterfactuals occurring in the EME data include the following: (27)
For I know by good experience, that a childe shall take more profit of two fautes, ientlie warned of, then of foure thinges, rightly hitt. For than, the master shall haue good occasion to saie vnto him. (N. Tullie) would haue vsed such a worde, not this: would haue placed this word here, not there: would haue vsed this case, this number, this person, this degree, this gender: 1571. Ascham, The Scholemaster (ed. Arber) p. 184
The conditional sense is implied in the context, but expressed only as for than, and the use of would haue is clearly a non-conditional counterfactual, further indicating the extent to which the modal has grammaticalised (such uses do not occur in the OE data, except for (16) which is an isolated case). An example of a non-conditional hypothetical is given in the following: (28)
Touchynge this instruccyon thre thynges I wold do. First I wold shewe that the instruccyons of this holy gospell perteyneth to the vniuersal chirche of chryst. 1500–70. Fisher, Sermons. (ed. Mayor). 1, 314
Again, there is a conditional sense implied in the context, but not given, indicating that the hypothetical senses in WOULD had a fairly wide distribution by that time. Lexical senses still persist in subordinate clause environments as well as in main clauses. The following example in a non-factive environment illustrates the continuing use of WOULD to mean ‘wish’ or ‘would like’ in PDE, indicating that its function as a lexical item was still prevalent in some environments: (29)
How wilt thou be tried? (Throckmorton.) Shall I be tried as I would, or as I shuld? 1554. The Trial of Sir Nicholas Throckmorton (ed. Hargrave) p. 1.64.C2
76
HYPOTHETICAL MODALITY
As with (22), though, there is a possibility of hypothetical inferences being derived from such environments. Grammatical uses, however, are frequent in subordinate clauses as well as in main clauses, as seen in some of the examples above. One factive environment not illustrated so far is the following in which WOULD appears in the complement of a verb of reporting: (30)
he said to Moyses that he wold destroye them utterly, and make hym ruler of a moche greatter and better people. 1531. Elyot, The Boke Named the Gouernour (ed. Rhys) p. 151
The data shown in the EME sample illustrate that the use of WOULD is rapidly grammaticalising, and especially in main clause environments, where the rise in grammatical functions is 26% up from the ME figures, compared with a 16% rise for the same functions in subordinate clauses. (The difference between grammatical uses in OE subordinate clauses and ME subordinate clauses is only 6%.) This indicates that although grammatical uses were not significantly more frequent than lexical uses in the OE and ME sample, the main change which seems to take place in EME is a significant rise in grammatical uses of WOULD in main clauses. Given that very few of such uses are shown as , or predictive, the rise may be considered as attributable to an increase in hypothetical and counterfactual uses at that time, and, in fact, 36% of the 37% of grammatical functions in main clauses are found in conditional apodoses, and in non-conditional hypothetical and counterfactual uses. The spread in the statistical distribution of lexical and hypothetical uses from the OE period to the EME period is a clear indicator of the extent of the grammaticalisation of the modal, and is typical of the model outlined in Hopper and Traugott (1993: 36): (31)
A > A/B > B
in which the development of a grammatical function B, from a lexical function A, undergoes an intermediate or overlap stage when both functions share representation. In the data presented, the A/B stage is represented more in the statistical distribution of A versus B than in the ambiguity of individual tokens in which both A and B functions are possible. The interpretation of the overlap as statistical may be especially relevant when the modal appears in a conditional apodosis, since in such environments there is less chance of an overlap arising in individual tokens. This is because the presence of a prior condition in the context, the if-clause, lends itself more readily to a prediction in the main clause
77
A DIACHRONIC CORPUS STUDY OF WOULD
than an expression of volition.14 Thus it is unlikely that there would be many instances of retention creating ambiguity of the lexical and grammatical functions of WOULD in a conditional apodosis; prediction is the basic function and the hypothetical meanings are created by the modification of the main clause prediction by the conditional clause. Table 3.4 illustrates the distributional overlap represented in (31). Table 3.4
lexical grammatical
Distribution of lexical and grammatical functions of WOULD in the Old English, Middle English, and Early Modern English data, as shown in main clauses only OE (850–950)
ME (1050–1420)
EME (1509–1570)
32% 13%
25% 11%
4.5% 37%
The EME figures for the grammatical environments (37%) represent a roughly inverse correspondence to the OE figures for lexical environments (32%), and reflect patterns similar to those shown in Heine’s (1992: 351) nuclear grammaticalisation chain, in which primarily lexical functions are found in the beginning of the chain, lexical and grammatical both appearing in the middle stage, and primarily grammatical functions at the end. (However, in Heine’s grammaticalisation chain, overlap may be represented in the meanings of the individual token, as well as the statisitical distribution.) In Table 3.4, the stages are
14. Consider, for example, the following three conditional sentences: (i) If John came, Max left. (ii) If it rained, we wanted to cancel the game. (iii) If it rained, we were likely to cancel the game. There is no way that (ii) can support a felicitous reading if the conditional expresses a past indicative sense similar to that expressed in (i); knowledge of one’s own volition is not readily construed as dependent on a hypothetical condition (note, however, that when it rained, we wanted to cancel the game is quite felicitous). Thus it is unlikely that an overlap stage occurred when both (non-modal) volition and hypotheticality could be found occurring ambiguously in conditional apodoses. The only example in the (ME) data when such ambiguity is possible is in an example with a conditional clause acting as a mitigating effect of politeness on the possible imposition expressed in the main clause: (iv) Ane bene leof ic Šrynen wolde Šif hit min mæŠð wære … ‘One boon I would fain ask if I be worthy of it … ‘ 1150–1250. History of the Holy Rood Tree (ed. Napier), p. 28
but even then, wolde is likely to be bleached of much volitional content, as it co-occurs without redundancy with an adverb expressing desire.
78
HYPOTHETICAL MODALITY
represented in the statistical dominence of one function at a particular time period, and the sum of the figures at the intermediate ME stage is almost equal to the proportion of grammatical environments found in the final stage, the EME data. The increase in grammatical tokens for main clauses, therefore, coupled with the loss of lexical tokens, indicates that with the disappearance of volitional senses, the grammaticalisation of the modal must have extended to irrealis main clause environments which were formerly the domain of subjunctive morphology. One example from EME (though not from the sample) indicates this possibility: (32)
I am in blood stepp’d in so far that, should I wade no more, returning were as tedious as go o’er … c.1605–6. Shakespeare, Macbeth III.IV. 158–60
In the apodosis of (32), the older subjunctive inflection were survives, as the subject is only syntactic and the verb is stative; it is likely that WOULD at that time still retained sufficient of its former volitional senses to prohibit its generalisation to such environments. The subjunctive environment in the protasis, though, is already occupied by a modal, should, since it has an animate subject and an active verb. It seems clear, then, that the past form of the modal already carried meanings of past prediction, alongside its lexical uses, as early as OE times. Evidence for this appears in factive subordinate clauses, in which the modal derives Q2 implicatures of the realisation of its complement, since such clauses contain factive presuppositions in any case. A non-factive environment, such as a hypothetical dependent clause (as in (23)), or a conditional apodosis, could just as easily express the same predictive senses but modified to be hypothetical by the environment. There is no evidence in the data suggesting that hypothetical senses in conditional apodoses arose directly from volitional senses elsewhere; if this had been the case, one would anticipate a greater number of cases of lexical retention in the apodosis modal, and only one possible ambiguity is found, discussed in fn. 14 (even this example is doubtful). Instead, there is a marked correlation between a decline in the distribution of lexical functions and a rise in the distribution of hypothetical functions, likely to be due to a general loss of lexical meanings in main clause uses freeing the already grammaticalising predictive modal for a wider field of environments, including those in which the old subjunctive inflections might have been losing significance. (According to Traugott 1972, the modals had replaced the majority of subjunctive inflections by the Early New English period — the EME period in the present study.) The loss of lexical meanings also correlated with the rise
A DIACHRONIC CORPUS STUDY OF WOULD
79
in the use of the HAVE auxiliary, indicating the necessity for a marker of completive aspect to compensate for the bleaching senses of past temporality in the modal as it came to be used in more and more predictive functions, and also to contribute to the senses of counterfactuality by its inferences of current relevance associated with the perfect. Thus, as predictive WOULD generalised to main clause environments, it now could increase its range of grammatical functions from expressing past prediction in factive sub-clauses to expressing past prediction in conditional apodoses, modified by the presence of a conditional if-clause. This is not an instance of grammaticalisation from a lexical function to a grammatical one, but from a grammatical function to one which is more grammatical, indicating a generalisation of an already grammaticalised form, the mechanism of Q2 implicatures being now at work in both subordinate clauses and in hypothetical conditional apodoses. But the question remains how meanings of counterfactuality came to be associated with such extensions of function.
3.5 The context-dependent grammaticalisation of WOULD As noted earlier, the grammaticalisation of counterfactual meanings is often associated with the development of proximative meanings crosslinguistically (see Kuteva 1998). However, as is seen from the data, proximative uses of WOULD are not frequent in the data and are also negligible in Visser’s (1969–73) corpus. Examples do occur though, in subordinate clauses, e.g. (18), in which the Q2 implicature of the prediction of the event described in the complement is cancelled by the main clause proposition (‘he was about to take them into his hands [but didn’t] as they became all fiery’). In this example the cancellation is supplied in the context. In another example, the Q2 implicature appears in a main clause, and, as discussed in Ch. 2 (2.1.3), is cancelled by an appended but-clause indicating adversity or contrast: (33)
ða ðe he þa ea ofer faren hæfde ða wolde he forð on his weŠ to hierusalem. ac þt hors ðe he on rad natoþæshwon on þone rihte wæŠ faren wolde. ‘When he [David] had passed over the river, he wished to continue on his journey to Jerusalem, but the horse on which he was riding would by no means go along the right road.’ 1150–1250. History of the Holy Rood-Tree (ed. Napier) (p. 20)
The use of wolde in (33) is again directional, with no verbal complement expressed in the original text. But the Q2 implicature of prediction or destiny is
80
HYPOTHETICAL MODALITY
still present and its cancellation is also explicit in (33), in the but-clause which follows. With time, wolde may ‘absorb’ the cancellation and acquire senses of counterfactuality as an implicit metonymic extension without the expression of the cancellation being necessary in the context. The following example, which is taken from another text in the Helsinki Corpus (not in the present sample), illustrates the reinforcing HAVE-auxiliary in a counterfactual clause co-occurring with a cancelling clause: (34)
Vnto þe kirk he wald haue gane, Bot þederward way wist he nane. ‘He would have gone to the church, but he didn’t know the way’ 1350–1420. The Northern Homily Cycle (ed. Nevanlinna) p. II, 204
The co-occurrence of the cancelling clause with WOULD HAVE + V-en is still in use in PDE, as is obvious in the gloss, but the same inferences of counterfactuality can often be obtained, depending on the context, with a cancelling clause only implied; e.g. (35)
A. B.
Did he go to the church? He would have gone …
in which B implies a negative answer. However, B’s answer, taken out of context, can equally imply a past prediction of the realisation of the event described in the complement, indicating that the inferences of counterfactuality in English are not as grammaticalised as might be anticipated in the use of such forms. What is important in the examples given in the present study is the need for a modifying clause of some kind to either cancel, in the case of but, or suspend, in the case of if-clauses, the Q2 implicature of the prediction of the complement of WOULD (or WOULD HAVE). Even in (27), in which no explicit conditional or adversative clause is supplied, there is an implied condition in the context. It seems then, that WOULD HAVE (+V-en) has not yet reached the stage of expressing conventionalised meanings of counterfactuality in its morphology alone without appropriate contextual support. In the case of counterfactual conditionals, the contextual support is supplied by the suspension of the implicature by the protasis condition. The modal verb thus relies on its environment, and the discourse context, for its interpretation, and both tenets of the maxim of Quantity interact to determine the pragmatic inferences which supply either a hypothetical, counterfactual or a predictive interpretation. As demonstrated in the data, it seems that WOULD had already grammaticalised to refer to past prediction in factive subordinate clauses in OE, but still retained its lexical meanings in main clause environments and some irrealis subordinate clauses. However, evidence that the
A DIACHRONIC CORPUS STUDY OF WOULD
81
shift in meaning from volitional to predictive senses was a gradual shift is provided in example (12), in which both senses are possible. The loss of main clause lexical meanings is seen to coincide with the increase in past predictive meanings as they began to generalise to include conditional apodoses (and much less frequently, main clause uses). The data in the present study, therefore, suggest that it is most likely that hypothetical uses were generalised from grammaticalising relative futures, but that contextual features, such as a conditional clause, would suspend the implicature and thus modify the strength of the prediction. The spread of a relative tense form of WOULD to conditional apodoses was probably due to the fact that the subjunctive morphology was losing its significance, and the function it had expressed required renewal by an incoming, more periphrastic means of expression, as noted in Hopper and Traugott (1993: 122). The use described by Bybee (1995) as likely to be the source of hypothetical meanings is also a main clause environment, but it relies on the retention of volitional meanings (which are still present in similar environments in PDE); i.e. that the past volition may persist to become present volition (interpretable in PDE as ‘would like’, or ‘wish’ — see (23)). However, the implicatures in conditional main clauses are inherently predictive, Q2 inferences, and the modal does not show any lexical retention of volition in such environments in any of the data presented, suggesting that the senses of prediction were well established by the time uses in conditional apodoses began to appear. It is also likely that it is not Quantity implicatures that are motivating the development of hypothetical senses in Bybee’s (1995) data, but the retention of perfectivity in a past stative predicate, which creates modality via the conflict of lexical and grammatical aspect (also a feature of Dowty’s 1979 Imperfective Paradox). In this respect it is unrelated to the developments presented here. It could be the case, then, that this hypothesis illustrates a separate path of development.
3.6 Summary The present study has extended the predictions made in the previous accounts regarding the development of hypothetical and counterfactual meanings out of a volitional verb source. It is seen from the distributional spread of WOULD across the diachronic data that the counterfactual meanings associated with the function of volition verbs are found only in particular contexts related to the use of the modal in discourse. It is hypothesised that the default implicature produced by a past volition verb source is a Q2 implicature, predicting the realisation of
82
HYPOTHETICAL MODALITY
the proposition in its complement, and that this type of implicature is primary to the development of modal meanings from the preterite modal WOULD in English and most likely from past volition verb sources in other languages. However, the type of modality may vary with the type of clause in which the verb appears and the presence of other clauses in the environment. There are four conclusions which can be reached from this premise. First, counterfactual meanings can also be explained with reference to Q2 implicatures, but with added contextual features, such as the presence of a cancelling clause of adversity, motivating the drift from Q2 implicatures to the Q1 implicatures usually associated with counterfactuality. Second, hypothetical meanings are considered to be derived from relative futures, also the result of the mechanism of Q2 implicatures which had conventionalised over time, and such meanings had already appeared in OE before the deictic future began to emerge at later periods. In the context of a hypothetical clause, e.g. a conditional protasis, the past predictive inferences are suspended to imply hypotheticality. The spread of relative futures to conditional apodoses is seen in the present data to correlate most closely with the disappearance of lexical meanings of volition in main clauses (Table 3.4). Third, the emergence of the auxiliary HAVE is most evident in the EME data, also at the time when many of the lexical uses were disappearing from main clauses, suggesting that the loss of the lexical senses carried with it a loss of the time reference inherent in the volitional meanings. Because of this, a marker of aspectual completion was introduced to restore the senses of counterfactuality which usually are associated with reference to situations occurring before the moment of speaking. Finally, the account of hypotheticality proposed by Bybee (1995), which is derived from counterfactual implicatures of ‘perfective state’, may have a more restricted application than that which is explained by the conventionalisation of Quantity implicatures. In the data presented, the general interplay of discourse-contextual information with morphological markers for modality in the determination of pragmatic inferences could be suggested as associated with a process of emergent grammaticalisation (Hopper 1987), in which the grammatical function at any given stage of development is seen as more closely related to its immediate context in use than to an extended period of historical development. The data reveal some important trends relating to traditional paths of grammaticalisation, and show especially that the various stages of development can be perceived in statistical distribution. Many of the data at such stages reflect both the old, lexical meaning and the new, grammaticalising senses ambiguously in a single occurrence, and such ambiguities may be quantified and shown to illustrate a change in progress. This is not the case, though, with hypothetical
A DIACHRONIC CORPUS STUDY OF WOULD
83
apodoses, which are likely to be extensions from past prediction (relative tense) uses, illustrating grammaticalisation processes in the generalisation of predictive functions in the modal to new environments. Other functions such as past habituals, though, are not significantly represented in the data, and, furthermore, have not, to present knowledge, been a subject of very extensive analysis in the literature. The evidence generally suggests that the grammaticalisation of past volition verbs is a highly complex field of enquiry, and that it is necessary to broaden the field much further in future research endeavours, especially those of a crosslinguistic nature.
C 4 Singaporean English and substratum influences in the grammaticalisation of hypothetical modality
The objectives of the present study are not only to examine the means by which hypothetical modal meanings became associated with a particular range of grammatical encoding in L1 varieties of English. As noted in Chapter 1, it is further aimed to extend such investigation to L2 varieties, and to determine what kind of influences, whether related to substratum factors or to universal features of L2 varieties, may affect the representation of hypothetical meaning as it appears in conditional clauses and hypothetical complement clauses. The kinds of factors relating to substratum features are discussed in a study in Ch. 5; and those which may pertain to more general aspects of language acquisition in L2 varieties will be discussed in Ch. 6 and in remaining chapters. It will be seen that although in hypothetical conditional constructions the presence of features such as bi-conditionality, negation, and first person subjects ((29b–d) in Ch. 2), are not affected by language-specific constraints, perhaps the most important indicator of hypotheticality is the presence of past or perfect morphology combined with an irrealis or future marker ((29a) in Ch. 2). Even without the presence of the other features, this feature alone provides the necessary meanings of past prediction which are exploited by Gricean implicatures in the expression of hypotheticality, as shown in previous chapters. Furthermore, without the use of past tense marking, the potential for ambiguity to result between hypothetical meanings and other kinds of irrealis meanings is always a possibility. As noted in Chapter 1, the use of Singaporean English as part of the data base to this study was initially motivated by a previous descriptive study on the expression of counterfactuality and hypotheticality in this L2 variety of English (Ziegeler 1994), which left many problems unsolved for further investigation. A brief review of the results of that study will be given in 4.2.2. For the moment, it is necessary to clarify that the study showed considerable variation in the expression of hypothetical modality and in the selection of grammatical forms used, but that there was found no possible correlation between the ability to
86
HYPOTHETICAL MODALITY
produce counterfactual expressions and the ability of the informants to create a possible hypothetical or counterfactual belief-world when presented with such expressions in a narrative or in example sentences. This indicates that there is no underlying cognitive reason related to the absence of a particular grammatical marking in a speaker’s language or dialect which may affect the construal of counterfactual or hypothetical meanings in discourse. Variation in the selection of grammatical morphology for the expression of hypothetical meanings is therefore hypothesised to be related to factors of the diachronic development of the Singaporean variety of English, which, as an institutionalised L2 variety, must have followed a historical path significantly diverse from that of most L1 varieties. The present chapter will therefore first locate the variety known as Singaporean English in an appropriate linguistic perspective by outlining some of the sociolinguistic and L1 background features likely to have had an influence on its development, then will discuss the likely influence that L1 features may have on the grammaticalisation of hypothetical modality in the L2.
4.1 Singaporean English: a descriptive account 4.1.1
Substratum (L1) and contact languages
The present ethnic population ratios in Singapore are reported to stand fairly stable at 76.9% Chinese, 14.6% Malays, and 6.4% Indians and others. These figures, quoted by Ho and Platt (1993: 8), have remained relatively constant for at least a decade.1 There are four official languages used: English, Mandarin, Malay and Tamil, with Malay serving as the national language and English as the main inter-ethnic language of communication — equal in status to a lingua franca between groups of different racial and L1 background. English has a further important role as the language of administration, law and international affairs, and serves as the main medium of instruction in educational domains.2
1. Informal estimates communicated to me by local Singaporeans place the present ratio of Chinese at only 65%; however, I have no official confirmation of this figure. The remaining 2% in Ho and Platt’s figures is not accounted for. 2. Although English still remains the main medium of education at all levels, a government policy was announced in November 1991 reintroducing Chinese-medium education to be available to some children from Primary Four (Gupta 1991a: 13). It is also observed that Chinese children now begin to acquire literacy in both Mandarin and English from pre-school-age (4–5 years), and that the official language of the child’s ethnic group is taught as compulsory curriculum subject from primary level (see Ch. 5).
SINGAPOREAN ENGLISH AND SUBSTRATUM INFLUENCES
87
The Chinese dialects most likely to have had an influence on the early development of Singaporean English are Hokkien (the most widely spoken), Teochew and Cantonese, the first Chinese immigrants having arrived from the southern provinces of mainland China. There were virtually no native speakers of Mandarin — by 1957, only 0.1% of the population spoke Mandarin as a mother tongue (Newman 1988: 440). However, as early as 1930, Mandarin was being introduced into the educational system (ibid.). Today, the number of native speakers of Mandarin is likely to be very much higher, due to the increasing use of the language encouraged by the systematic imposition of a public language policy, the Speak Mandarin Campaign, which was launched in early 1978, as well as to the motivating effects of the economic rise in the region of Mandarinspeaking countries such as mainland China and Taiwan. Thus, Mandarin in Singapore has never held a place as a substratum language, but has received increasing importance as a language of communication between Chinese speakers from the various dialectal sub-groups, where it now appears to have gained prominence. As such, it is emerging as a contact language, rather than a substratum. In the eighties, Southern Chinese dialects, which include Hokkien, Teochew and Cantonese as the main ones, were noticeably being replaced in their former, typical domains, (described by Ho (1986: 15) as neighbourhood shops and eating-places), by Mandarin, indicating that the government policy was already taking effect. It was earlier reported that the use of Mandarin in such domains rose from 28% in 1991 to 36% in June 1993 (Straits Times: 4/9/93). While government statistics are often reputed to be notoriously optimistic, figures from other sources might also appear to indicate an increase in the use of the language. Asiaweek (27/10/88) reported that 69.1% of elementary schoolchildren named Mandarin as their most frequently used language, while only 23.3% named English, and 7.2% named a dialect. (In 1980, 64.4% named a dialect as their first choice.) Mandarin now appears to be in general use as a lingua franca amongst Singaporeans of Chinese ethnicity, as observed by Gupta (1994a: 151), who notes that in the home, Mandarin is now the first choice for many families, and that the motivation of the government policy is reaching the point where it is considered disloyal to one’s ethnicity not to know Mandarin. The ability of schoolchildren to maintain a number of dialects or contact languages has also been dramatically reduced, this being due to the separation of Chinese children from those of a different ethnicity for reasons of scheduling of language and religious instruction classes; the need to maintain a contact variety of colloquial English is therefore diminished as a consequence, to the advantage of Mandarin (Gupta 1994a: 152).
88 4.1.2
HYPOTHETICAL MODALITY
The status of Singaporean English
Although many Singaporeans would most likely still have a dialect of Chinese as a mother-tongue, in the sense that it would have been their primary language used since infancy, a considerable number of the younger generation are now true native speakers of both English and Mandarin. The status of English as a native language is illustrated by the fact that approximately 20% of the current generation of incoming schoolchildren were estimated by Gupta (1992: 323) to have acquired English in the home as a first language. Gupta (1994b: 15) placed this figure at a much higher level of “… up to half of all children starting nursery school … ”, indicating clearly the rise of a new generation of native speakers of Singaporean English. It is now common to find a situation of bilingualism in English and another of the official languages (Newbrook 1993: 1). Thus, whereas native speaker once referred only to speakers of L1 varieties of English, the native speaker in Singapore is now a speaker of an endonormative or local variety, Singaporean English, and one would hesitate, therefore, to accurately use the term non-native variety when discussing Singaporean English. However, the variety that is spoken as a native language in Singapore nevertheless retains some formal distinctions from other more established varieties of native speaker English, reflecting its origins as a second language variety. Thus it is a second language variety which is acquiring native speakers (Gupta 1992: 323), and it will be referred to, therefore, in the present study as an L2 variety, whether or not it has native speakers. As an L2 variety, its status is similar to that of a creolised variety, but not identical, since its origins were not those of a nativised pidgin, having been introduced into the 19th century Straits Settlements as a learner variety taught in the classroom. Gupta (1991a: 10) describes the language as a ‘mixed language’, a language which cannot be seen as having a genetic relationship with the superstratum language, and which exhibits a break in transmission. What relates it to a typical creole, she maintains, is the presence of grammatical and morphosyntactic features typical of languages which can be classified as creoles, such as deletion of the copula, reduced marking for tense and number, the preference for topic-comment discourse structure over subject-object types, and unmarked interrogative constructions (lack of inversion or do-auxiliaries). Gupta (1991a: 11) also notes there is no significant similarity of the colloquial sub-variety to China Pidgin English, a dialect formerly used in Hong Kong. She believes that the variety which provided the source for present-day colloquial Singaporean English may instead have arisen in the multilingual playgrounds of the English-medium
SINGAPOREAN ENGLISH AND SUBSTRATUM INFLUENCES
89
schools, and, as such, bears a formal affinity with a creole, though not a functional similarity. The dialect of Singaporean English describes a range of different subvarieties. Ho and Platt (1993: 1) use the term Singaporean English as a general term covering the English used by Singaporeans — they do not specify educational level — and Singapore English to describe only the variety used by speakers at the lower end of the educational and socio-economic scale. Platt (1977: 84) referred to the existence of a sociolectal continuum of speech varieties, ranging from the basilects at the lower end of the scale, through mesolects to acrolects used by the most educated speakers in the most formal environments. However, Gupta (1991a: 9) distinguishes only two varieties, the High (or H) which she considers equivalent to Standard English (she does not specify which standard), and the Low (L) variety which she defines as Singapore Colloquial English. In classifying the sub-varieties in this manner, she is adopting a reference to Singaporean English as that of a diglossic situation, using a description introduced by Ferguson (1959), which emphasises the functional specification of two separate varieties of a language in a community where they are spoken. There is little doubt that a form of English with only barely perceptible differences in form from established standards is used for the specified functions of public administration and the media, and in literature and formal styles. However, proponents of the diglossic approach seem to overlook the very real potential for style-shifting to occur between sub-varieties, depending on the situation, participants, topic and purpose of an interaction; the ability of the speaker to modify speech style according to these variables along a lectal continuum is an indication of a desire to assert in-group solidarity with the participants or to affirm the varying degrees of formality or informality of the situation of discourse (Ho and Platt 1993: 4–5). A description of a diglossic Singaporean variety would, therefore, necessarily include a Low variety which allows little provision to accommodate a range of sociolectal subvarieties. It is questionable, then, whether a generalisation of a Low diglossic form as a single variety is a necessarily accurate description of the range of lects available to a speaker of colloquial Singaporean English. The ability of the Singaporean to shift along the sociolectal continuum is not matched by speakers of more established varieties of English, who are less able to switch sociolects than to simply style-shift within the same sociolectal boundaries (Platt 1977: 90). However, in both instances, the shifting is the result of the speaker’s (conscious or unconscious) assessment of the formality of the situation or the sociolectal range of the participants in an interaction. Perhaps this
90
HYPOTHETICAL MODALITY
is indicative of the possibility that such shifting is often motivated not so much by the speaker’s assessment of the hearer’s proficiency with the language as by the need to establish social rapport and ethnic solidarity with the participants in an interaction, by using a sub-variety which could be considered less identifiable with varieties of English historically marked by their relation to a colonial past. In the present study, however, the data is obtained from speakers of a more educated variety in a relatively formal situation — university and higher secondary-school students were surveyed in a classroom situation, and using written data. Such conditions are those most likely to produce a style which displays fewer colloquial characteristics than would normally be found otherwise. 4.1.3
Characteristic features of Singaporean English: Chinese L1 influence
The acrolectal, or most educated variety has been widely acknowledged by speakers of established varieties (e.g. Gupta 1991a: 9, and Ritchie 1986: 17) and certainly by its own speakers to vary little from the general concept of a standard variety. However, one would feel that the basis for these assumptions is partially the result of an over-emphasis on analysis of the colloquial variety, which has been at the expense of a relative paucity of research on the more educated variety of Singaporean English. Foley (1988: xix) notes that variation may be found in the syntax of the formal educated variety as well as in the basilect, and that this reflects the structure of the background languages in Singapore. Given that the dominant substratum languages are Chinese dialects, it may be assumed that some of the grammatical variation found in the educated varieties will be likely to reflect substratum features and hence will relate to the grammar of Chinese. Platt and Ho (1993: 18) list a number of features, mainly found in the colloquial variety, which bear a significant resemblance to Chinese, including zero pronouns e.g.: (1)
We don’t have -- ;
the use of possession verbs to denote existence, e.g.: (2)
In China where got people go to English school?;
the use of topic-comment discourse structure and Chinese-type serial verb constructions, e.g.: (3)
That book on the TV, take come here, can or not? ;
they also note Chinese-type question tagging (or not? — cf. Mandarin shi bu shi?), and repetition for emphasis. Variation in the uses of a number of lexical
SINGAPOREAN ENGLISH AND SUBSTRATUM INFLUENCES
91
items are also observed to have their origins in equivalent Chinese forms; e.g. send for ‘bring’ or ‘give a lift to’. This feature would be likely to be found in more educated Singaporean English also, as would a similar lexical feature: follow meaning ‘accompany’, as in: (4)
I’ll send you to the airport — you can follow me,
with the intention that the speaker was offering a lift. Platt, Weber and Ho (1984: 78) also list, amongst other features, the omission of the copula in adjectival predicates, and a passive construction isomorphic to that of an agentless Chinese passive e.g. (1984: 80): (5)
The seeds (of a certain fruit) can eat (i.e. can be eaten).
Such examples illustrate clearly that general grammatical features of Chinese dialects have contributed significantly to the surface structure of many of the features of the colloquial variety. Structural features found in the more educated varieties may also be considered as related to Chinese substratum influences. Some of these might include use of the simple present for simple past (see Ch. 5), and the omission of third person -s on verb stems, most likely attributable to the general absence of verbal and nominal inflection in all the substratum languages. (Malay, also having an uninflected verb system, cannot be precluded from consideration as a possible substratum influence.) The use of lexical marking for aspect (already) is another feature usually reputed to be directly reflective of the substrate. It was formerly observed as especially frequent following such verbs as die or pass away, and forget, and combined with past tense marking in cases when the present perfect would normally be prescribed (Ho and Platt 1993: 145). However, in certain situations it cannot be assumed to indicate present perfect or perfectivity, but a change of state. Examples from personal (diary) observations collected between 24/10/94 and 1/9/95, illustrating the change-of state use, include the following: (6) (7)
(8)
I’m going over already, going to my Mum’s place (Discussing the film Schindler’s List): First it’s colour, after that it’s black and white already. … But at the end part it’s all coloured already. (In a baker’s shop) (A) What happened to the chicken pies? (B) No more already.
The change-of-state or inchoative use in Singaporean English (possibly translateable in English by ‘now’ or ‘then’) has also been observed by Bao (1995) as
92
HYPOTHETICAL MODALITY
parallel to the Hokkien particle liau or the Mandarin le which marks either perfective or inchoative uses depending on its position in the sentence (perfective for post-verbal le and inchoative for sentence-final le). However, there are as yet no comparative statistics to indicate whether or not this use is a recent innovation, or just a recent observation. Perhaps more speculatively, the frequent use of compared to or relative to as a means of expressing comparison (see Newbrook, Delikan and Dias 1987: 306) could have had its origins in an equivalent construction in Chinese dialects: Mandarin Chinese has no inflections or adverbs for expressing comparatives, one available device for making comparisons being by use of a connective bi, which is roughly translated as “compared to” or “relative to”, e.g. Ta bi wo gao, meaning (lit.) “He/she compared to me (is) tall.” Another example of substratum influence could be the use of the expression always not rather than never (ibid.: 314). There is no adverb lexicalised as never in Mandarin Chinese; one way of rendering this meaning is by the use of zongshi “always”, followed by bu, meaning “not” or “no”. This construction is commonly found in the writing of Chinese L1 students of English. Although the resemblance between these features described above and similar constructions in Chinese is relatively transparent, the argument for attributing these features to Chinese substratum languages does not go uncontested. Ho and Platt (1993: 18) mention that topic-comment structures, serial verbs, and repetition are also features of creoles and therefore it cannot be proven that they are the result of Chinese influence, but that they may be likely to occur universally in any contact language situation. Other features known to occur both in creoles and in Singaporean (colloquial) English include got as a locative verb of existence (Platt 1977: 87), used instead of there/is are.3 This verb is also used in Hawaiian Creole to mark existence (Bickerton 1981). However, semantic transfers may be attributable only to Chinese substratum influence and include the lexical items mentioned above: send for ‘bring’ or “give a lift” ; this is another instance of a form in Chinese having more than one meaning in English: song in Chinese has as its basic meaning ‘send’ but also is used to mean ‘to
3. In Chinese one verb you, is used to denote either existence or possession; its translation in English is have or have got conveying only the meaning of possession, or be for existentials and locatives. The existence of two verb forms in English, have and be, for one in Chinese, seems to have affected the semantic transfer of this substratum feature in basilectal Singaporean English. The availability of both have and have got having similar meanings may have been exploited, so that have in the basilect has come to denote possession and got seems to be used mainly for existence, as shown in Ho and Platt’s examples above (1993: 18).
SINGAPOREAN ENGLISH AND SUBSTRATUM INFLUENCES
93
escort’ or ‘accompany’, and follow used for ‘accompany’ is related to a Mandarin co-verb gen which means ‘follow’ or ‘accompany’ (which also functions as a comitative preposition meaning ‘with’). Arguments for substratum influence could also be supported by observations of Chinese Pidgin English, which has been shown to display features clearly reflective of influence from its substratum, Cantonese; such features are not explained by any theory of creole universals (Thomason and Kaufmann 1988: 187). Recent theories on the genesis of creoles tend to dismiss the possibility of a set of formal features to distinguish them from other languages: Mandarin Chinese has been found to exhibit all the features of a perfect creole, and yet it is not one (Mufwene 1990: 10). Researchers in creole linguistics often regard a combination of variable influences as complementary in the development of many creoles; these influences include not only the substrate, but also the superstrate, with universals acting only to constrain the form of the creole to be in accord with that of any natural language. The variety of English used in Singapore today is therefore recognised as an L2 variety with a history of development through the medium of school instruction rather than as a pidgin or creole, although it shares a number of similarities, both structurally and functionally, with creoles around the world (Gupta 1994b: 56). Certainly for speakers who would not have had the advantage of education, especially in the early history of Singapore, English must have had creole-like beginnings, and this cannot be discounted; it is likely that creole-like varieties were being used alongside instruction-based varieties, and not just in the school playgrounds, as Gupta maintains. However, present-day Singaporean English is not a learner variety, and any features characteristic of an interlanguage variety are more likely to be the fossilised evidence of earlier nonnative speaker variation which has now become conventionalised within the dialect (ibid.: 7) — such features are then acquired and passed on by successive generations of native speakers. The significance of the two facets of the historical development of Singaporean English, on the one hand as an L2 acquired through the medium of education, and on the other as a contact language outside of the classroom, will be made more apparent in later chapters. Some of the more characteristic features of Singaporean English are listed above; these are just a few of a wide range of identifiable items which mark the variety as distinct from other varieties. Amongst other characteristics is a systematic means of marking for past tense which has been found to interact very closely with lexical aspect, but which becomes less marked as the educational level of the speaker increases. In the study which follows (Ch. 5), it will be shown that this interaction contributes a major effect in the development of
94
HYPOTHETICAL MODALITY
hypothetical meaning in Singaporean English, providing an explanation for much of the variation associated with this form of expression. But it is necessary first to consider the ways in which hypothetical meaning may be marked in the L1 and contact languages.
4.2 Hypothetical modality 4.2.1
The expression of hypotheticality in Standard Chinese and substratum dialects
As noted above, the influence of Chinese dialects on the morphosyntax of Singaporean English is likely to be considerable, given that the majority of the population are bilingual in at least one Chinese dialect as well as English. The increasing emphasis placed by government campaigns to promote the use of Mandarin is also likely to be important as an influence on Singaporean English, and for ethnic Chinese speakers, English and Mandarin are now used side by side, depending on the situation and participants involved in the interaction. Although Mandarin is not a substratum language in Singapore, the concurrent use of the language by Chinese-English bilinguals is certain to have some effect on the variety of English spoken there, and the influence of substrata such as Hokkien and Cantonese is already shown to be significant. Because of the likelihood of L1 influences on the grammar of the English used in Singapore, it is justifiable, then, to include some discussion on the means of marking hypotheticality in Chinese. According to Chao (1968: 13), it is possible to consider Chinese almost as a single grammatical system, and that the largest amount of uniformity exists between the dialects in the grammatical structure.4 He notes the only major differences as order of direct and indirect objects in the Wu dialects, and slight differences in the order of negatives in potential complements in some southern dialects. The means of expressing hypotheticality, then, is anticipated not to vary significantly across the dialects. Li and Thompson (1989: 647) note that the conditional sentence in Mandarin Chinese is not marked for factivity or otherwise, nor is it marked for temporal
4. This is a very general claim, even an over-assumption. R. Cheng (1997) finds a number of grammatical differences in features between Beijing Mandarin, Taiwanese Mandarin, and the Hokkien spoken in Taiwan. However, none of these involves the function of expressing hypotheticality.
SINGAPOREAN ENGLISH AND SUBSTRATUM INFLUENCES
95
reference without the addition of specific lexical items. Thus, the use of the Chinese conditional, which is crucially dependent on context for its interpretation, assumes a knowledge of facts relating to the discourse situation and shared between the speaker and the hearer. This is also the case with English conditionals, but the interpretation of the level of factuality of English conditionals is aided by the propensity for the modal forms to co-occur with past tense marking and express past prediction. Li and Thompson provide an example of a conditional sentence in Mandarin which is ambiguous as to the level of factuality expressed. Decontextualised in such a way, it is impossible to ascertain whether or not the conditional expresses a possible, a hypothetical, or a counterfactual situation: (9)
Ruguo ni kan dao wo meimei, ni yiding if you see arrive I younger-sister you certainly zhidao ta huaiyun le. know 3 pregnant
This can be translated as: ‘If you see my younger sister, you will know she is pregnant’, ‘If you saw my younger sister, you would know she is pregnant’, or, ‘If you had seen my younger sister, you would have known she was pregnant’. Li and Thompson (1989: 647)5 What such ambiguity demonstrates is that counterfactual concepts can be constructed in Chinese without the presence of specific grammatical marking, and that any efforts to isolate a particular morphological means of marking counterfactuality only presuppose that such a means is required to express the concepts. This reflects the bias of a contrastive viewpoint adopted by speakers of languages in which de-contextualised examples can be often unambiguously interpreted as counterfactual. Actual language use, though, would hardly be likely to reveal situations in which counterfactual conditionals are decontextualised. Alleton (1994: 10) also notes the ambiguity created in Mandarin between nonfactual and counterfactual interpretations of sentences using the modal verb yinggai:
5. As was noted by a member of the audience at the 1994 Australian Linguistics Society conference (Latrobe University, Melbourne), the meaning of dao provided by Li and Thompson is the lexical meaning of this form; used as a grammatical marker it has a meaning similar to ‘manage to (see)’, that is, one of successful achievement.
96
HYPOTHETICAL MODALITY
(10)
Nei ge difang, jingguo nuli jinwan YINGGAI daoda. that place, through effort this.night YINGGAI reach a. ‘With some effort, we may certainly reach that place tonight.’ b. ‘With some effort, we certainly could have reached that place tonight.’
Comrie (1986: 91) makes the observation that in Mandarin, clauses can be unmarked for counterfactuality or hypotheticality. Chao (1976: 257) states that there is no specific means of expressing counterfactuality in conditional sentences in Chinese. Although he suggests that the least often used of all the if-words are those most frequently recruited to express an unlikely or impossible supposition (he lists eight forms), he adds that this is nevertheless not necessarily the case. It would seem that there are no categorical means of expressing counterfactual or impossible conditionals in Mandarin Chinese, though the use of certain markers such as adverbs and modals in contexts which are interpretable as remote from fact is likely to invoke a situation of higher hypotheticality than another. This is clear especially in examples in which the perfective marker le appears in the protasis, or in which the combination of sentence-final le in the apodosis with a temporal adverb referring to the past creates the temporal and aspectual conditions equivalent to those of the pluperfect in English. The following examples are provided by Chappell (ms.): (11)
Ruguo zhao-bu-dao shui, women zao jiu si le. if look--arrive water 1 early:on then die ‘If we hadn’t found water, we would’ve died.’
The likelihood of a counterfactual interpretation is increased by the use of the first person, and this is one of a cluster of features contributing to higher evaluations of the counterfactuality of a conditional sentence by the hearer (see (29), Ch 2). It would be, furthermore, pragmatically impossible to utter a sentence such as (11) if it hadn’t been counterfactual, and such information is not provided by any of the grammatical or lexical morphology used, but by extralinguistic information (also a member of the cluster of features, as noted in example (29): Ch. 2). However, what is significant is the combination of temporal marking with the sentence-final particle le, marked by Chappell as expressing inchoativity, or a change of state. This particle has been discussed by Li and Thompson (1989) as expressing ‘’ — ‘current relevant state’ — and is used to describe a situation which has some relevance to the particular situation which is current. In the apodosis of (11), the current situation is refers temporally to the past, and expresses the same aspectual inferences as are expressed by the present
SINGAPOREAN ENGLISH AND SUBSTRATUM INFLUENCES
97
perfect aspect (have + V-ed) in English, in combination with past time reference, that is, a meaning which situates the prediction in the apodosis relative to a current past time point. (The prediction, however, is not grammatically marked with a modal verb as it is in (10), it is only implied by the fact that the speaker is still alive and in the causal link between finding water and staying alive.) Thus, although the conditional morphology associated with highly inflected languages is not present in the Chinese construction, the presence of particles and adverbs and the appropriate contextual information can still contribute to a counterfactual interpetation. An even closer comparison can be made with the addition of the modal auxiliary hui (‘can’) in the apodosis. Chappell (ms.) illustrates the following as an example of a marked counterfactual conditional in Mandarin: (12)
Yaoshi ni lai le, wo de xiongdi jiu bu if 2 come 1 brother then hui si le. can die ‘If you had come, my brother wouldn’t have died’
In (12) the temporal reference is set as prior to the moment of speaking due to the presence of the perfective le in the protasis, and the combination of the modal expressing possibility relevant to the current situation (one of past-ness) supplies the same inferences as are obtainable by the modal perfect form in an English counterfactual apodosis (past modal + perfect aspectual form). The use of hui was also discussed by Cheng (1985) as an essential ingredient in a counterfactual conditional in Mandarin, but it must also be noted that such modals in Mandarin cannot be marked for perfective aspect in the same way as the English modals are marked by their preterite forms in more factually remote contexts, since they are stative verbs, and Chinese, which has not yet developed a tense system, cannot mark stative verbs with perfective aspect. This factor is at the core of most of the ambiguities regarding the factuality of hypothetical constructions in Chinese. Thus, it is possible to mark counterfactuality in conditionals in Mandarin by the use of the appropriate morphological equipment; however, this does not suggest that the unmarked types are not unambiguous, as seen above. The use of particular marking for counterfactual conditionals appears therefore not to be obligatory in Mandarin, and, given the proposals in Chs. 2 and 3 that counterfactuality is only an inference derived from a past prediction, the same conditions for substantiating the strength of that inference may be seen to apply in Mandarin as well as in English. Wierzbicka (1997: 34) also notes that the presence of certain adverbs in
98
HYPOTHETICAL MODALITY
Chinese marking the time reference of the conditional can create the inferences of counterfactuality: nashihou (‘at that time’) for past, and jianglai (‘in future’) for future conditionals. The use of negatives in her examples also provides evidence to distinguish past conditionals from future (bu referring to present or future conditions; mei you referring to past conditions). What is not always clear is how the distinction between future possible conditionals and future unlikely but not impossible conditionals is marked. When there is no reference to past, this distinction becomes blurred. The absence of such a distinction is not apparent in Eifring’s (1988: 198) examples, in which bu shi marks the conditional (see also Eifring 1995: Ch. 7): (13)
a. b. c.
Ta ruguo bu shi dao Meiguo qu nian shu (le), wo jiu hui qu zhao ta tan-tan. Ruguo ta bu shi dao Meiguo qu nian shu (le), wo jiu hui qu zhao ta tan-tan. Ruguo bu shi ta dao Meiguo qu nian shu (le), wo jiu hui qu zhao ta tan-tan.
All of (13) can mean: ‘If it were not for the fact that he has gone to the United States to study, I would (have) contact(ed) him to have a chat.’ Clearly, the use of the sentence-final particle, le, is sufficient to indicate the anteriority of the event described in the protasis, and the presence of a predictive modal, hui, in the apodosis, relates the prediction to the anterior event, but Eifring claims that the use of bu shi (lit. ‘is not’) before the verb of the protasis serves to mark counterfactual conditionals in Mandarin. However, he also notes that its use is restricted to informal or colloquial styles, and it does not have any abstract functions in scientific or technical discourse. It is also clear from (13) that the form bu shi is restricted to use with negatives and has no affirmative counterpart (Eifring 1988: 207). This supports the discussion in Ch. 2 that because negation is normally a marked category in most languages, a negative counterfactual is more easily perceived as contrary-to-fact than a affirmative one (see also Ziegeler 1994). One might conjecture then, that negative counterfactuals are seen to be more salient in some languages than affirmative counterfactuals, and it seems significant that a construction was selectively created to express negative counterfactuality but not affirmative counterfactuality. The likelihood of negative past conditionals being interpreted more frequently than positive ones as ‘true’ counterfactuals has also been noted by Wierzbicka (1997: 30), on the basis of crosslinguistic examples, as a possible universal (see 2.1.4). In consideration
SINGAPOREAN ENGLISH AND SUBSTRATUM INFLUENCES
99
of the naturalness of negation, then, it is not surprising that negative predictions made about the past (X would not have happened) lend themselves to more easily processed, stronger counterfactual implicatures. In Cantonese, as in Mandarin, there is no grammatical means of distinguishing non-hypothetical from hypothetical and counterfactual conditionals, the context usually providing the required inferences (Matthews and Yip 1994: 302). In the examples of explicit counterfactual conditionals given by Matthews and Yip, the context also supplies a negative proposition. For example: (14)
Yuhguo ngoh mh jidou ge wa jauh mh gam if I not-yet know say then not dare daaih-seng gong la. big-voice say ‘If I hadn’t known, I wouldn’t have said it so loud.’ Matthews and Yip (1994: 302)
The conjunctions yuhguo and ge wa emphasised in bold in the original, are suggested as indicative of the hypotheticality of the sentence, the expression ge wa interpreted as ‘suppose’ (this may be suggestive of a counterpart de hua in Mandarin, noted in Chappell (ms.) as introducing higher hypotheticality into the conditional). The idiom jou ji (lit. ‘early know’ — ‘if I had known earlier’) also conveys counterfactual meaning; this also has an equivalent in Mandarin: wo zao zhi dao. A Cantonese example is the following: (15)
Jou ji daai maaih leih heui la. early know take along you go ‘If I’d known I’d have taken you along’ Matthews and Yip (1994: 304)
(15) cannot be logically translated as ‘?If I knew I took you along’, mainly because of the presence of a first person subject — the speaker, as identifiable with the subject, has the most implicit knowledge of the subject’s extent of knowledge, or whatever action is referred to in the predicate; therefore, we give the highest credence to the possibility that the speaker, in complying with cooperative conventions, is implying that s/he did not know. As noted above and in Ch. 2, then, first person subjects as well as negatives in past conditionals are the most likely indicators of counterfactuality, and the idiom exemplified in (15) is apparently restricted to first person (singular) subjects. The counterfactuality of (14) is also strengthened by the presence of a first person subject. Examples of the means of expressing counterfactuality in Hokkien appear to be relatively under-researched so far; however, Wierzbicka (1997: 34) provides
100
HYPOTHETICAL MODALITY
examples from Taiwanese Min, a variety of Hokkien, which are transcribed below in (16) and (17) using Church Romanisation:6 (16)
hit le sichun goa na u chin, that time I if have money goa e chhoa i I will/would marry she/her
This can be compared with: (17)
i-ao/chionglai goa na u chin, goa e chhoa i later/future I if have money I will/would marry she/her
According to Wierzbicka, (16) can only be given a counterfactual interpretation; however, on consulting Taiwanese native speakers, it was revealed that (16) could be either counterfactual, Hypothetical, or an open conditional, depending on the context, and this would be known from the context. Thus, the translation of (16) would read ‘If at that time, I have/had/had had money, I will/would/ would have marry/ied her’, since the temporal adverbial provides no clue as to the time reference and there is no sentence-final aspectual marker. (Wierzbicka does not provide translations for these examples, but (17) might be glossed as ‘If later I have/had money, I will/would marry her’, the time reference being clearly marked, though the factuality is not.) Wierzbicka’s claims may be based on the fact that there is a first person subject, which contributes to meanings of counterfactuality (as suggested in Ch. 2) if the time reference is also taken to be past (not necessarily the case in (16)), as it would not be conversationally cooperative to hypothesise on the possibility of whether one had money in the past, when one usually knows such facts about oneself. The factual basis for expressing a counterfactual implicature is therefore easily inferred in such instances, but most likely, it is a combination of both factors. Wu (1994) suggests additional grammatical devices would make Li and Thompson’s (1989) example (9) more likely to express counterfactuality: (18)
Ruguo ni dangshi kan-dao wo meimei de hua, if you that:time see-arrive my younger:sister DE HUA ni jiu yiding hui zhidao ta huaiyun le. you then certainly would know she pregnant ‘If you had seen my younger sister at that time, you would have known that she was pregnant.’
6. The assistance of Cheng Hsiao-feng is gratefully appreciated for the transcription and gloss of these examples.
SINGAPOREAN ENGLISH AND SUBSTRATUM INFLUENCES
101
In (18), the counterfactual inferences are derived from a combination of the use of dangshi (‘at that time’), de hua (‘supposing’), and the presence of a modal auxiliary (hui) with a particle (‘current relevant state’), which, as noted above, provides a similar interpretation to that obtained by a perfect auxiliary in English (similar inchoative particles, such as a, cannot be used in the same construction in Hokkien — Lien Chin-fa, p.c.). Even though it would seem that the less hypothetical reading: ‘If you saw my younger sister at that time, you will have certainly known then that she was pregnant’ could not entirely be ruled out, there being no means of marking the pastness of the prediction, the likelihood of a counterfactual reading is held to be the most typical interpretation in such cases.7 More importantly for the purposes of the present study are the inferences of counterfactuality associated with the predicates of verbs of wishing. Mandarin has a single verb xiwang, which can be used to translate either ‘hope’ or ‘wish’ in English, there being no indication apart from context to show the likelihood of the wish or hope being realized. Eifring (1988: 194) recounts an instance when he observed a Taiwanese speaker utter the following sentence, while gazing at the top of a tall building: (19)
Wo zhen xiwang wo neng pashang qu.
This could be translated as either: ‘I really hope to be able to climb up there’ or ‘I really wish I’d be able to climb up there’ (sic) (Eifring 1988: 194). Eifring makes it clear that in such cases there is no question of ambiguity with regard to the meaning of the sentence, since it is obvious from the context in which it was uttered whether the desire expressed is non-factual or counterfactual. However, there is no marking on the verb to indicate this distinction, nor is there any lexical means of distinguishing the level of factuality associated with such expressions. An alternative means of expressing a counterfactual wish in Mandarin, which expresses desire contrary to the known situation, is found in the form of the construction yaoshi X, na jiu hao/gai duo hao (Lien Chin-fa, p.c.). This is a conditional construction, and may be translated as ‘if x, that will/would be good/should be better’. A similar type of construction has also been noted by Chao (1976: 160) (na duo hao being the form of the apodosis) and the same
7. Native speakers of Taiwanese Mandarin believe that the use of additional morphology creating the stronger sense of counterfactuality suggested by Wu is redundant, and makes the sentence appear like a literal translation from English.
102
HYPOTHETICAL MODALITY
means of expressing wishes and hopes is also found in Japanese and Korean (Akatsuka and Clancy 1993). However, there is still no unambiguous means of distinguishing the level of certainty attributed to the possible fulfillment of the wish. Chao also considers wo yuanyi followed by a counterfactual clause is translatable as ‘I wish’, as in the following (1976: 160): (20)
Wo yuanyi ni bie name long ‘I wish you weren’t so deaf’
The usual translation of yuanyi is ‘be willing to’ (Chao 1968: 738) or some expression of strong desire. Thus, the expression of strong desire in (20) has something of the function of a performative, as also in English, when it expresses a meaning similar to ‘I don’t want you to be so deaf’. The use of the verb wish as an expression of volition, as discussed in Palmer (1986: 152–3) is then clearly parallelled in the use of the verb yuanyi in Chinese. This expression has counterparts in Cantonese also, in a verb yuhnyi (‘be willing’); for example: (21)
Leih yuhn-mh-yuhnyi jipsauh ni go tiujin a? you will-not-willing accept this challenge ‘Are you willing to take on this challenge?’ (Matthews and Yip 1994: 237)
However, as can be seen by such examples, the volition verb is not used exclusively to express hypotheticality; only the context can provide the necessary inferences, there being no explicit grammatical markers of any description that may serve this purpose. In other contexts, a single form may be used to express either hypothetical or non-hypothetical desire for the realisation of a complement proposition, as in the Mandarin example (19), and in the Cantonese example (22): (22)
Ngoh heimohng leih wuih yuhnleuhng ngoh. I hope you will forgive me ‘I hope you’ll forgive me’ (Matthews and Yip 1994: 236)
Matthews and Yip’s translation uses the meaning of non-hypothetical desire, ‘hope’ to express the meaning; however, they note that the verb heimohng has both the meaning of ‘wish’ as well as ‘hope’, as in Mandarin. Both (19) and (22) contain a modal auxiliary (neng in (19) and wuih in (22)), but because there is no grammatical tense system in Chinese, the usual inferences of hypotheticality obtained in the English translation when the modal is marked for past tense are not available, leaving the factuality of the utterance ambiguous. In the conditional sentences shown above, the hypotheticality or counterfactuality is derivable from
SINGAPOREAN ENGLISH AND SUBSTRATUM INFLUENCES
103
the presence of temporal adverbs combined with modals of prediction, e.g. hui in (18). This combination serves to match the functions of past tense modals in English, providing a comparable translation (however, less hypothetical levels of factuality, expressed by forms such as will have + V-ed in English, are not completely ruled out in decontextualised examples). The addition of past tense adverbs, though, is not available in (19) and (22), and, apparently, is not required to define the level of hypotheticality of the utterance which could just as well be inferred from the context and the hearer’s evaluation of the speaker’s knowledge of the factual premise backing the expression. The importance of the use of past tense morphology as a grammatical indicator of counterfactual and hypothetical situations is outlined further in Ch. 5; for the moment, it is necessary to suggest that this single feature is responsible for disambiguating the hypothetical from the non-hypothetical meanings in languages which do express grammatical tense. In Hokkien, a similar form, hi-bang expresses the desire of the subject for the realisation of the complement proposition; the translation of this verb is given as ‘hope’ by R. Cheng (1978: 108) but it can also be used to translate ‘wish’, or express a hypothetical notion (Cheng Hsiao-feng, p.c.), e.g.:8 (23)
Goa hi-bang i khi ‘I hope he’ll go’
It is also noted by R. Cheng (1997: 53) that tense is not marked in Taiwanese Hokkien, and thus one would not expect to find any grammatical indicators of hypotheticality in the past form of a modal auxiliary in such constructions. In fact, unlike the Mandarin examples, there is no modal auxiliary appearing in (23) at all, although it is translateable using an English modal. From the above examples illustrating contexts which are most likely to receive a counterfactual interpretation in English, it is clear that grammatical morphology does play a role in disambiguating possible from hypothetical meanings, and hypothetical from counterfactual meanings, and that such morphology which is usually found in the preterite form of a predictive modal is absent from Chinese, in which predictive modals cannot be marked for past tense. However, it is also apparent that, given the appropriate past temporal reference in the context, a counterfactual or hypothetical interpretation of certain expressions can be construed with little processing effort. The major interpretation differences lie in constructions which would not usually appear with past time
8. While the data for Cheng’s (1997) study and (16) and (17) come from Taiwanese Hokkien, it is generally accepted that there are no major syntactic differences between the Hokkien spoken in Taiwan, and that used in Singapore (F-F Tsao, p.c.).
104
HYPOTHETICAL MODALITY
reference in Chinese, such as a present time reference wish for something that is unlikely to be associated with literal past time reference (e.g. (19)); that is, a construction in which the use of a past predictive modal in English has no relation to past time (‘I really wish I could climb up there’). Thus, it is only in past temporal conditionals that a counterfactual construal may be reliably obtained in Chinese, and such construals are most likely to be strengthened by the presence of contextual features such as the use of first person subjects, negation, and the impossibility of alternative readings due to factors known in the discourse, just as in English. The hearer does not have to use any grammatical devices to perceive a counterfactual situation, nor does the speaker to create one. At the same time, it must be still emphasised that the grammatical indicators usually applying to counterfactual and hypothetical expressions in English are by no means absolute either, as discussed in Ch. 2. The fact that counterfactual meaning is derived by implicature, and is not a grammatical category in the strict sense of the word, is sufficient to suggest that there is no one-to-one formfunction mapping in either English or in Chinese and that only contextual clues will provide the hearer with enough information to determine the factual premise underlying the speaker’s utterance of hypotheticality. However, in isolated, decontextualised examples the significance of grammatical devices may assist in disambiguating levels of factuality, and, certainly from the comparative data described above, it should be suggested that the grammatical morphology in the preterite form of a predictive modal present in English but not in Chinese is a likely index, if not a marker, of hypothetical or counterfactual meaning. 4.2.2
The expression of hypotheticality in Singaporean English
Past research on the expression of conditionals and counterfactuality in Singaporean English is relatively limited, although a few studies have provided examples which attest to the absence of the existence of a regular system. At the same time, they could be taken as random variation similar to that found in more established varieties of English. Some of the examples indicate a lack of crossclausal symmetry in the verb tenses; e.g.: (24) (25)
If I can induce anger in at least one reader, this article would have more than served its purpose. I don’t suppose you would be surprised if I tell you my ambitions. (Crewe 1984: 50)
Crewe’s examples are taken from an assortment of written texts, including newspapers, books, and students’ essays, and as such they could be representative
SINGAPOREAN ENGLISH AND SUBSTRATUM INFLUENCES
105
of formal usage. The following example avoids the use of modal verbs altogether (Crewe 1984: 51): (26)
They are now saying that plans had to be changed if a complete breakdown was to be avoided.
The patterns shown in (24–26) reflect the possibility that the use of a past tense form of the verb in the protasis is not obligatory and in (26) it appears that the use of a modal form in the apodosis is not obligatory either, though this is only a single example. Such patterns might be related to the substratum Chinese dialects, in which modal marking is not obligatory in hypothetical conditionals (see (9) and (11)). Non-hypothetical conditionals have been found to be expressed by simple juxtaposition of subordinate and main clauses, with the conjunction omitted; e.g.: (27)
You go by meter, you have to pay. (Platt, Weber and Ho 1984: 125)
This pattern appears to reflect a similar structure in Mandarin Chinese, as noted by Comrie (1986: 82): (28)
Zhangsan he jiu, wo ma ta (lit. Zhangsan drink wine, I scold him)
Brown (1992: 31) observes that the use of the past tense in the protasis and the auxiliary would in the apodosis are rarely found in Singaporean English conditionals of present or future time reference, and that these forms are replaced in most instances by the use of the simple present tense in both clauses, combined with the absence of the if conjunction, as shown in (29) (1992: 31): (29)
The Tans are late, we have to postpone the meeting.
Sometimes the form suppose or supposing is used as a protasis marker to replace if: (30)
Supposing that a student was to fail in the first year … (Platt, Weber and Ho 1984: 125)
Traugott (1985: 291) also notes the use of suppose as often the only marker of the protasis in creoles; e.g. Tok Pisin, though there does not appear to be an equivalent form in Mandarin Chinese according to Chao’s (1968: 116) detailed list of conditional conjunctions. However, in (14) and (18), there is a use of an expression which could be glossed as ‘supposing’, but, which, as noted, does not appear in clause initial position. It is considered in (18), though, to be only one of a combination of features contributing to meanings of counterfactuality.
106
HYPOTHETICAL MODALITY
Few examples of non-conditional hypothetical expressions appear in the literature; the following illustrate the ways in which the hypothetical complements of wish clauses are constructed, with non-obligatory marking for past tense in the verb form: (31) (32)
I wish I’ve read more, Kwang Meng said. I wish we have better bookshops here.
(Crewe 1984: 110)
Such examples, as well as (24) and (25) seem to suggest that past tense is not so frequently used to indicate factual remoteness as it is in more established varieties. There is also a possibility that it is avoided in (31) and (32) because the complement is stative, which would be less likely to be marked for past in any case (see discussion in Ch. 5, Section 5.1). However, the following example which contains a non-stative verb, observed informally of an educated speaker of Singaporean English (9/3/93), seems to suggest otherwise: (33)
I wish Princess Diana operates a ham radio.
There is obviously a reluctance to use past tense marking on verbs which do not refer to past time, and the hypothetical meaning is therefore derived directly from contextual inferences. The examples presented above certainly seem to indicate the absence of a regular system for marking hypotheticality in Singaporean English. 4.2.2.1 An earlier study The range of variation illustrative of the expression of hypotheticality in Singaporean English motivated the need for quantitative description in an earlier study (Ziegeler 1993, 1994), as noted in Ch. 1, and this will be reviewed briefly below. 60 tertiary-educated Chinese L1 speakers in Singapore were investigated using elicited written Singaporean English data, in order to assess whether the influence of L1 differences in the marking of counterfactuality had any cognitive effect on the comprehension and production of counterfactual expressions in an L2. The results failed to find any significant direct correspondence between the variable marking of distinctive morphology expressing counterfactuality and the perception of counterfactual notions expressed in English narrative discourse. What the study did uncover, however, was that there was a vast range of variation possible in Singaporean English for the expression of counterfactual complement clauses and past (pluperfect) counterfactual conditionals, 77% of which diverged from standard ways of marking counterfactual conditionals in standard English (i.e. using the pluperfect in the protasis and would + have + Ved in the apodosis). The data was elicited using a narrative which presented a problem to be
107
SINGAPOREAN ENGLISH AND SUBSTRATUM INFLUENCES
solved and offered a possible solution to that problem. The informants were requested to read the text and then hypothesise on possible alternative outcomes if the problem had not been solved in the way that it had. The exercise is illustrated in the following (Ziegeler 1994: 37): Please read the following story and answer the question below, using complete sentences: A farmer had to cross a river one day with a dog, a sheep, and some hay. The only boat he had was so small that he could only get one animal or the hay into it besides himself. The farmer knew that the dog was liable to attack sheep and that the sheep could not resist eating hay. Therefore he could not leave the sheep alone with the hay or the dog alone with the sheep. How did the farmer manage to get himself, the animals and the hay across the river safely? He took the sheep first, then returned and took the dog, returning with the sheep. He left the sheep then and took the hay. Then he returned to get the sheep again. Why couldn’t he have taken the dog or the hay first?
Much of the variation in the expression of counterfactual conditionals appeared at first to be random. In the protasis verb, the variation was reflected in the presence or absence of ‘back-shifting’, but in the apodosis, it was reflected in the alternation between past and non-past forms of the conditional modal would (as well as would have). A total of 11 different construction types were isolated, some with as few as one token. The combinations are illustrated in Table 4.1 below, together with the percentage rates of their occurrences. Table 4.1
Range of constructions used in Singaporean English to express a counterfactual conditional (Ziegeler 1994: 42)
type
A B C D E F G
description
tokens
protasis
apodosis
pluperfect pluperfect simple past simple past simple past pluperfect miscellaneous
would + would + would + will + V would + will + V
have + V-ed V V have + V-ed
23% 18.5% 15% 12% 09% 07% 15%
108
HYPOTHETICAL MODALITY
Examples of the different construction types are illustrated as follows: (34)
a. b. c. d. e. f. g.
If he had taken the dog first, the sheep would have eaten the hay. If he had taken the dog first, the sheep would eat the hay. If he took the dog first, the sheep would eat the hay. If he took the dog first, the sheep will eat the hay. If he took the dog or the hay first, the sheep would have eaten the hay or the dog would have eaten the sheep. If he had taken the hay first, the sheep will be bitten by the dog. If he have taken the dog first, the sheep will eat the hay.9
The data from a control group of 21 tertiary-educated native speakers of Australian English provided only 3 different types, as shown in the Table 4.2: Table 4.2
Range of constructions used by the Australian English control group to express a counterfactual conditional
type
A C G
description
tokens
protasis
apodosis
pluperfect past miscellaneous
would + have + V-ed would + V
59% 30% 11%
(The miscellaneous category (G) was represented by only 2 informants.) It is possible that the reduced number of available categories may be due to the smaller number of informants in the control group; however, since nearly 60% supplied a construction using the type of morphology typically associated with counterfactual conditionals, it would be unlikely that additional numbers of informants would make a great deal of difference to the proportions as shown. The variation in the ‘back-shifting’ between a past tense form, a pluperfect form (and even occasionally a present tense form) in the protasis verb was not considered surprising in the Singaporean data; it appeared to reflect only a temporal interpretation of such clauses, not a modal one. Ho and Platt (1993: 158) suggest that conditional pasts, being irrealis, were not treated as temporal (realis)
9. Type G, in which there were an additional 5 type variants, was classed as miscellaneous, since in this group there were no more than 2 respondents producing one construction type.
SINGAPOREAN ENGLISH AND SUBSTRATUM INFLUENCES
109
pasts in their data, and so were less frequently marked for past. However, the frequent variation between past and non-past modal forms in the apodosis is less easily explained, since both will and would could be categorised as irrealis modals. The use of modal forms in the hypothetical complement clauses of verbs of ‘wishing’ also appeared to be in a state of flux in the Singaporean English data in the study, as the following elicited examples demonstrate: (35) (36) (37)
I wish there is peace in the world. I wish that there will be peace in the world. I wish that there would be peace in the world.
These examples were produced by three different subjects in the same study, and in response to the same question. The present tense verb form in (35) can be explained with reference to the punctual/nonpunctual distinction: that statives and non-punctuals are less frequently marked for past in any case. The explanation for the alternation between will and would was previously attributed to a hypercorrection rule: would is commonly applied to non-hypothetical contexts where will would standardly be used (e.g. this office would be closed next Saturday (Tongue 1979: 42)) and is reported either to be considered a polite form and therefore more ‘correct’ (Newbrook and Chinniah 1987: 255), or an overgeneralisation of uses in the complements of verbs of reporting or communicating (Leslie 1981: 84). However, will and would are used in free variation in spoken Singaporean English according to Teow (1988: 53–4), and the same patterns applied to the modals can and could. There was no significant correlation of such variation to ethnic background found in Teow’s study, indicating that the features may be conventionalised as part of the dialect, rather than attributable to any particular type of L1 interference. Examples such as (36) and (37) above can also be considered overgeneralisations, the use of a modal auxiliary in such environments being redundant since the appropriate meaning could be derived from the use of a subjunctive, or back-shifted non-modal form: I wish there were (was) peace in the world. Because there are dialectal differences relating to the use of modals in such environments, a number of similar examples were put to the test of the intuitions of a group of native speakers of Australian English, along with examples similar to others which appeared overgeneralised in the 1994 study, such as: (38)
I wish that I would earn lots of money … .
The details of this later study are to be found in Ch. 6. In the meantime, it is clear from the data so far that the means of expression of hypothetical modality
110
HYPOTHETICAL MODALITY
in Singaporean English shows enormous variation and very little systematicity. While this variation may be unrelated to the presence of hypothetical or counterfactual concepts for such speakers, an explanation for such variation must nevertheless be provided.
4.3 Summary The present chapter discusses some of the more significant aspects of the background of Singaporean English, its status as a rapidly developing L2 variety with native speakers, and the importance of southern Chinese dialects and Mandarin as L1 and contact influences on the development of this variety. In many ways Singaporean English is unique in that it provides an excellent example of a L2 variety which has developed in ways similar to those of a creole, but is at the same time not a creole, in that it was introduced via the medium of education. Many of its distinguishing grammatical features, especially those in colloquial use, reflect the influence of Chinese dialects, and amongst these is included variable marking for hypothetical expressions, often involving ambiguity between different levels of irrealis. Hypothetical constructions are less grammaticalised in Chinese than in English, and often receive a counterfactual interpretation from only contextual inferences. The same inferences are used in the interpretation of Singaporean English hypothetical constructions, as the usual ways of distinguishing degrees of hypotheticality in most standard varieties, such as with preterite modal verb forms, are found to be less frequently used in Singaporean English. In the next chapter, one of the reasons for the variation found in the Singaporean data will be examined.
C 5 The interaction of tense and aspect in the grammaticalisation of counterfactuality
In the previous chapters, a number of contributory features were seen to be salient in the crosslinguistic expression of hypothetical modality, and these features included the presence of first person subjects, negation, the establishment of a causal relation between the protasis and the apodosis clause of a hypothetical conditional, and the presence of past tense or perfect morphology in combination with markers of irrealis or prediction in a hypothetical clause (this was also found to be salient in Givón’s 1994 study). The first two features and the last are relevant to any type of hypothetical clause, the third applies only to conditional constructions. In Ch. 4 it was seen that certain combinations of the features could produce a counterfactual interpretation even in the absence of others in some languages, that is, it was not necessary for all four features to be present to ensure a counterfactual interpretation. For example, in Taiwanese Southern Min (Ch. 4 (16)), there is no past tense morphology on the modal verb to distinguish degrees of factuality, but counterfactual meanings can be satisfactorily obtained by the combination of temporal adverbials such as hit le sichun (‘that time’) and first person subjects. The presence of such examples indicates that in many languages, counterfactuality is an inference: it is the hearer’s evaluations of the potential for the speaker to make a hypothetical abstraction backed by known facts to the contrary, and, as noted in Ch. 2, this potential is the sum of grammatical indicators, contextual clues, and extra world knowledge. It could be the case that the presence of a single gram (see Bybee and Dahl 1989) which can be isolated crosslinguistically to represent a one-to-one correspondence with a counterfactual meaning might be relatively rare in languages. Also observed in Ch. 4 was the fact that in Singaporean English there is a large amount of variation in the expression of a counterfactual conditional construction, that modal verbs are used in their preterite and non-preterite forms in free variation, and that in some cases, they are not even used at all. The substratum languages in Singapore, as noted previously, are languages such as
112
HYPOTHETICAL MODALITY
Hokkien, Cantonese, and Malay, all of which lack a past tense marking system, and it is hypothesised that such a feature has an influence on the development of grammaticalised forms of marking hypothetical modality in the L2. The present chapter discusses this point in the light of diachronic observations of the grammaticalisation of past tense modals in English, and offers an explanation in terms of the grammaticalised implicatures which created hypothetical meanings from past stative verbs in longer established varieties of English. It is believed that the only one feature of the four cluster features discussed in Ch. 2 as contributing to meanings of hypotheticality, the presence of past or perfect morphology combined with irrealis or predictive markers, provides the key to distinguishing counterfactual meanings from less hypothetical meanings in cases of ambiguity, but that even this feature is derived by pragmatic implicature and the distinctions are not part of the semantics of the grammatical form. Thus it is the interaction of tense and aspect which combine to produce hypothetical senses in certain constructions, and this combination will now be tested in speakers of Singaporean English, a variety which is influenced by substratum features of tense and aspect marking. By means of a brief questionnaire, it is further hoped to demonstrate that the acquisition by Singaporean speakers of English of the grammaticalised implicatures expressing hypotheticality may be seen to correlate with length and possibly frequency of exposure to such phenomena over a prolonged time period.
5.1 Tense marking in Singaporean English In Ch. 4 a number of distinctive features of more colloquial varieties of Singaporean English were described such as morphological aspects including lack of plural marking on nouns and absence of subject-verb concord (Gupta 1991b: 120–121), and such features have been found to be correlated with education and proficiency factors. Perhaps one of the most frequently-cited areas of such variation is found in past tense marking. This has been the basis of a study by Ho (1986), which was expanded and developed in Ho and Platt (1993). In the study, Ho examines the frequency of past-tense marking in Singaporean English, correlated with such factors as educational level of the subjects, phonological interference, and lexical aspect of the verb forms used (whether or not the verb refers to a punctual or non-punctual event, i.e. used duratively or iteratively, or stative). The most significant finding of this study was that punctual verbs are marked for past more frequently than stative, and stative verbs
THE INTERACTION OF TENSE AND ASPECT
113
more frequently than non-stative, non-punctual verb forms, e.g. those marked by durative or habitual aspect. Frequency of past tense marking was also found to increase with educational level (the sample consisted of speakers ranging from those with only primary education through to tertiary graduates). Ho and Platt include much discussion on the merits of Bickerton’s ‘Bioprogram Hypothesis’ as one possible explanation for the selective marking of past tense in the dialects of Singaporean English surveyed (see Bickerton 1981). According to this hypothesis, certain basic structures of language are encoded in the brain genetically and are activated in times of paucity of input (such as in creole genesis) (Bickerton 1981: 133). Bickerton’s studies of (two) creole languages revealed a clear punctual/non-punctual distinction in the marking for past tense, leading him to believe that it was verbal aspect rather than tense which was being marked (Ho and Platt 1993: 16), and that the punctual-nonpunctual distinction was primary in the grammaticalisation of the tense-moodaspect system (Bickerton 1981: 284). Punctuality is regarded broadly in the present context as entailing meanings of perfectivity, lacking reference to the internal temporal constituency of a situation (Comrie 1976: 21) and non-punctuality and stativity are seen as as entailing imperfectivity, or meanings which do have reference to the internal temporal constituency of a situation. Bickerton is cited as suggesting that semantically, the notion of punctuality must entail pastness (although the reverse situation does not necessarily hold) (Ho and Platt 1993: 154). For this reason, according to Ho and Platt’s data, punctual and ‘+ completive’ (i.e. telic) verbs are most frequently marked for past in Singaporean English, and non-punctual and stative verbs are treated as if referring to a situation which is still continuing, by exhibiting less frequent marking for past. Although perfective-imperfective distinctions are not marked as categories in the grammar of English, the relevance of their underlying semantic influence will become clear later in the discussion (see 5.2). In classifying the verbs used into punctual, non-punctual or stative types, Ho and Platt do not apply a rigid definition, but a broader one similar to that used by Bickerton, so that punctuality can be determined by contextual features as well as those which are inherent in the lexical verb semantics. For example, they find that verbs were also likely to be marked for past when they implied the attainment of a goal (telicity); e.g. He painted a picture (1993: 81), in which the context, not the verb, implies the completion of an act. A comparison of their Singaporean English data with that of Bickerton’s Guyanese Creole and Hawaiian Creole English shows that 56% of Singaporean English punctual verbs are marked for past, while 23% of non-punctual verbs have past marking; in
114
HYPOTHETICAL MODALITY
Hawaiian Creole English, the proportions are 53% of punctuals and 7% of nonpunctuals, whereas in Guyanese Creole the figures are 38% and 12% respectively (Ho and Platt 1993: 17). Ho and Platt do not discount the influence of substratum and contact tendencies, though, and they note that in Mandarin Chinese, the perfective aspect marker le is used in most cases where punctual aspect is indicated (1993: 150–1). However, it cannot occur in combination with habituals, duratives or statives. Li and Thompson (1989: 203) illustrate this in the following example from Mandarin: (1)
ta tian-tian hui-qu (*-le) 3 day-day return-go (*- S/He goes back every day.
The absence of perfective marking on past habituals in Chinese is reflected in Singaporean English, as shown in an example from personal diary observations collected from Singaporean speakers of English between 24/10/94 and 1/9/95: (2)
“When I was a child my mother goes to work and my father looks after us.”
Mandarin Chinese is not a substratum language of the majority of Singaporeans, but as noted in Ch. 4, it now plays an active role as a vital contact language in daily interaction at all levels. Notwithstanding, grammatical differences between the main (substratum) dialects of Chinese (e.g. Hokkien, Cantonese and Teochew) should also be taken into account, and a form cognate to le occurs in Hokkien and Cantonese, two of the major dialect groups. For example, in Hokkien, the form is represented as liau: (3)
Goa ka lin thiah ho-liau he-chhia phio. I for all-of-you bought train tickets [1 give you buy - train tickets — DZ] ‘I bought train tickets for all of you.’ (Chiang 1963: 36)
Substratum features in the dialects of Chinese are likely to have had some influence on the variation in past tense marking in Singaporean English, given that the majority of the population is of Chinese ethnicity. But even in minority substratum languages, such as Malay, tense is not indicated in the verb morphology, and aspectual verbs are considered the main means of supplying temporal meaning (Omar and Subbiah 1968: 4). Furthermore, the similarities between the ways in which past tense is marked in Bickerton’s creoles and in the Chinese substratum languages is not without problems in such an analysis, as noted in
THE INTERACTION OF TENSE AND ASPECT
115
Ch. 4, since Mandarin possesses characteristics similar to those of a creole, and there is bound to be overlap with certain features. Due consideration must then be given to patterns found in other local L1 and contact languages, such as Malay or Tamil. Tense in Malay is not indicated in the verb morphology, and the verb itself may refer to the present, past or future, depending on what lexical means of denoting temporality appear in combination with it (Omar and Subbiah 1968: 4). Such lexical forms are actually more comparable to temporal adverbs, but there are, in addition, a series of aspect verbs from which temporal meaning is inferrable (e.g. sudah, sedang, belum, telah, pernah, akan) (ibid.: 57). For example: (4)
‘completive’: saya sudah/telah/pernah menulis surat itu I .verb write letter that ‘I have written the letter.’ (Omar and Subbiah 1968: 57)
(The interlinear gloss is my own, DZ.) Although the Malay auxiliaries have been defined as referring to either tense or aspect, many accounts would now describe them as expressing a perfective/imperfective distinction (Alieva 1992: 179); such a distinction would not be unlike the semantic distinction which is at the basis of the variation in past tense marking in Singaporean English. Tamil, the official language of the Indian community in Singapore, does have explicit verb morphology for past tense marking as well as for present and future, although it is noted that the marking does not necessarily correlate one-to-one with past, present and future time (Asher 1985: 156). Aspect appears to be more salient, and there are various grammatical means for expressing an event as a single entity, the perfect being used to refer to a single point in the past (1985: 157–159). It is interesting to note that Platt (1989: 405) finds the punctual/non-punctual distinction evident even in the English of older Singaporean Tamils whose dialects are less likely to be influenced by the English of the Chinese majority. This is a clear indication that substratum influences could be playing a role in the semantic determination of past tense marking in Singaporean English. However, the question of universals should not be ignored, but should be approached from a different viewpoint. Previous studies on the acquisition of Spanish by English speakers (Andersen 1991) and the acquisition of English by multilingual groups of learners (Bardovi-Harlig and Reynolds 1995) indicate that lexical aspect of the verb plays a significant role in determining which verb classes will be the first to be marked for past tense for the learners acquiring the languages, and that verbs with inherent punctual or telic aspect are the first to
116
HYPOTHETICAL MODALITY
acquire marking for past tense in the learners’ interlanguages. Andersen attributes these findings to the operation of a Relevance Principle (following Bybee 1985), by which grammatical inflections denoting past meaning are more readily attracted to verbs with lexical meanings relevant to past time, such as punctual and telic verb types. Shirai and Andersen (1995) also discuss the fact that lexical aspect influences the acquisition of past tense marking for children acquiring English, and that the determining semantic factors are [+punctual], [+telic], and [+result] in the verbs first marked for past (see also Ch. 7). They explain the acquisitional route as an extension from a prototype, the most prototypical exemplars of past tense being verbs which are inherently punctual or resultative. These studies suggest that past tense is acquired first as perfective aspect, by both L1 and L2 learners alike. Typological evidence from Bybee, Pagliuca and Perkins (1994: 92) shows that in some languages perfectives do not apply at all to stative verbs. In their data, pasts are regarded diachronically as further developments from perfectives, and perfectivity, which is defined as referring to temporal boundedness, is entailed in the lexical meaning of punctual and telic verb types. As noted earlier (3.2.2), Bybee et al. predict that although pasts initially develop from non-stative perfectives, they gradually increase their range of environments to include statives as well. If the diachronic processes proposed by Bybee et al. can be verified, then it may well be the case that the L2 learner’s acquisition of past forms in a single language synchronically reflects the historical patterns of the grammaticalisation of past tense cross-linguistically, with the earliest-acquired uses of past tense occurring with non-stative verbs. Similar instances of parallels between diachronic and ontogenetic processes have been predicted in previous studies; for example, the acquisition by children of the English present perfect (Slobin 1994) and of Chinese classifiers (Erbaugh 1986) (see Ch. 7). It would be reasonable, therefore, to suggest that even if the substratum languages did not affect the grammaticalisation of past tense in some varieties of Singaporean English, naturalistic developmental routes of acquisition may also be found to have some influence.
5.2 Past modals: differences in counterfactual scope We have seen earlier that the process of the strengthening and conventionalisation of Quantity 2 implicatures is of considerable importance as an explanation for many of the meaning changes which take place during grammaticalisation. Such processes underly the historical development of epistemic modal
THE INTERACTION OF TENSE AND ASPECT
117
meanings out of former deontic ones (Traugott 1989), and the same processes can be held to explain the development of future meanings out of verbs of desire (Bybee, Pagliuca and Perkins 1994), and hypothetical meanings from past tense modal forms in English (see Ch. 3). However, as also noted in Ch. 3, Bybee (1995) suggests that hypothetical past modal forms (e.g. would) were derived directly from lexical origins as past forms of stative verbs, and did not evolve from the past forms of grammaticalised future modals, which suggests an alternative path of the development of hypothetical meanings in past modals. Bybee (1995) proposes that the present temporality often associated with past modal forms is accountable to the indeterminacy of the realisation of the modal complement. However, the present study maintains that the present (and/or future) time reference attributed to preterite modal forms is due to the retention of perfective meanings out of which preterite forms were believed to have evolved (Bybee, Pagliuca and Perkins 1994: 54), as discussed in (3.2.2). Modal verbs are inherently stative in lexical aspect, and the extension of past tense to stative predicates is not in accord with earlier stages of the development of pasts, which if derived from perfective aspectual categories, would not have been used with lexically-imperfective predicates. The situation of retention results in a conflict between (perfective) grammatical aspect and imperfective lexical aspect in the modal verb, resulting in senses of ‘perfective state’ modality, and suggesting that the termination of the past state of modality is only pragmatically implied and may therefore still hold. The speaker then exploits the use of the past form of the modal to express deference and politeness, and to indicate to the hearer that although the projected action may not have been achieved, it was within the speaker’s abilities or volitional intentions to undertake it. These inferences are believed by Bybee (1995) to have given rise to the senses of hypotheticality associated with the conditional use of would today. Any inferences which may evolve from such developments would have to be perceived, then, as having past volition as an earlier stage, rather than past prediction, and as such may be considered as grammaticalised from an earlier stage in the development of the volition modal, as Bybee suggests. Evidence from the corpus data in Ch. 3, though, suggests that such inferences may be associated with a different function from that found in conditional apodoses. If pragmatic inferences of hypotheticality arise with modal stative verbs, then the same kind of inferences should result from the use of the past tense with any stative predicate. Givón (1994: 316) notes the use of past tense and the progressive auxiliary ‘be’ in Israeli Hebrew, and of the perfect verb form in Early Biblical Hebrew, appearing in counterfactual conditional clauses alone with no modal or irrealis marker to contribute to the counterfactual meanings. This
118
HYPOTHETICAL MODALITY
suggests that the non-modal use of stative forms in the past may be sufficient to provide counterfactual inferences in some languages, even without a predictive marker; indeed, in the protasis clauses of English counterfactual conditionals, and in the complements of verbs of wishing, this is exactly the case: past counterfactuality is indicated simply by the use of the pluperfect, which contains a stative verb marked for past, the auxiliary had. In fact, in the history of counterfactual conditionals, it seems that the pluperfect was used in both clauses of a conditional, and that this use preceded the advent of modal forms in the main clause (Molencki 1998). The use of the pluperfect in the protasis and in past hypothetical wishes remains as a vestige of the period prior to the introduction of modals, and is rapidly being replaced in some dialects, e.g. North American English, with a modal construction identical to the form in the apodosis (Trudgill and Hannah: 1982: 47); the process by which this occurs is termed parallelism, or symmetry, and is discussed in Haiman (1985) (and Ziegeler 1994, with regard to developments in Singaporean English conditional protases). In the present chapter, the inferences produced by the conflict of grammatical and lexical aspect are tested in a study in which non-modal past stative and past progressive verb forms are used as stimulus items to elicit possible inferences of counterfactuality. The study described will demonstrate that there is a measurable statistical difference between the assessment of such inferences in the dialects of L2 speakers of Singaporean English and those of L1 speakers of English. The difference between this type of implicature and the Quantity implicatures described in Ch. 3 in contributing to hypothetical inferences is, as noted earlier, due to differences in the scope of the modality. Because the attribution of past to stative verbs only pragmatically suggests a terminative state, in a past predictive statement such as She wanted to become a teacher, there is hypotheticality, and potentially counterfactuality, in the possibility that the subject referred to still does want to become a teacher. This places the entire sentence within the scope of the modality, and offers a possible explanation for the development of preterite presents in Germanic and the subsequent use of the past forms of English modals to imply present time reference, discussed by Bybee et al. (1994: 77–8) and Bybee (1995). Quantity implicatures, however, have only the modal complement in their scope, and if one makes the statement She wanted to become a teacher, the expression of only the volition to undertake the action described suggests that the speaker is uncommitted to any further information, and could well be proposing a counterfactual meaning. It is felt that both types of implicature may combine together to produce pragmatic senses of hypotheticality, depending on the environment in which the modal occurs. However, in the examples from Chinese in [4.2.1], much of the hypotheticality
THE INTERACTION OF TENSE AND ASPECT
119
is derived from the context and from other features of the cluster, such as first person subjects. This is because perfective aspect has not yet generalised to stative environments in Chinese, and modals are stative verbs. The result is, as Comrie (1986: 91) suggests, languages such as Chinese and Indonesian (which is virtually indistinguishable from Malay in such respects) do not typically express degrees of hypotheticality in conditional constructions. Neither of these two languages has a grammaticalised past tense category, and both Malay and Chinese are major substratum languages in Singapore. This is not to suggest that an absence of tense marking can be correlated generally with an absence of grammaticalised ways of marking levels of hypotheticality; however, more research in this area could reveal that it could be a contributing factor.
5.3 The Study 5.3.1
Methodology
The study to be reported here is part of a longer and more comprehensive survey conducted in Singapore, using primarily L2 speakers of Singaporean English. The data were collected mostly in the classroom from four groups of secondary and tertiary students representing a total of 148 educated Singaporean speakers. They had all completed at least 10 years of English-medium education and were currently engaged in full-time education when the survey was taken.1 The use of a student sample was advantageous in that it provided an expedient sampling procedure which could also be more representative than many others. It was considered to represent an accurate profile of an educated speaker population in that it tested different proficiency levels of subjects who were fully engaged in full-time English-medium education at the time the survey was undertaken, and who were therefore closely involved in the everyday use of the language. The four groups are listed as follows:
1. The secondary 4 groups included 11 subjects whose primary education was in Indonesia. This factor was monitored closely, and the results of these subjects were checked carefully for any likelihood of skewing due to the subjects’ having less exposure to English and English-medium education. Since there were not found to be any significant differences between their results and those of the remainder of the Secondary 4 group, they were not eliminated from the study.
120
HYPOTHETICAL MODALITY
i)
(Sec.4 Express) Secondary 4 Express students from Bukit Merah Secondary School, Singapore. This group was considered by the teaching staff to be of a higher standard than the Cambridge O-Level class in the same school. The total number of students was 47 (comprising 2 classes) and the average age was 16. They were in their tenth year of English-medium education. ii) (Sec. 4) Secondary 4 and O-Level students from Bukit Merah Secondary School. The total number was 28. The average age for these students was 16.5 (some were as old as 18). Most of these students were in their fifth year of secondary education, and in their eleventh year of English-medium education. iii) (NUS 1) These were tertiary students from the National University of Singapore, and the total number was 39. They were mostly first-year undergraduates, although there were some from higher levels. All had a minimum of Cambridge A-Level education. The average age was 21.6. iv) (NUS 2) 34 second-year undergraduate students from the National University of Singapore. The average age was 21.5. There was also a Control Group of native L1 speakers, the total number of subjects being 38, although only 37 completed the entire questionnaire from which the questions below were extracted, one failing to complete Question 4, probably due to time constraints. The subjects were first and second year undergraduates from Monash University, Australia, and the average age was 22.6. The informants were asked to state their age and educational level; they were also asked to provide information regarding their mother tongue. This was difficult to obtain from the Secondary school students, since ‘mother tongue’ is actually a syllabus subject in Singaporean schools, where it refers to the official language of one’s ethnic group (i.e. Mandarin, Malay, or Tamil), and this need not be one’s true mother tongue. When the intended meaning of the term was pointed out to the students, most of them were willing to volunteer the correct information, though this cannot be guaranteed since unofficial languages are frequently regarded unfavourably in Singapore. 5.3.2
The Questionnaire
As noted above, the following four questions were extracted for specific analysis from a longer questionnaire containing 12 questions. The additional questions were intended to disguise the objectives of the present separate study, so assisting in eliciting more spontaneous responses from the subjects than would be possible otherwise. The subjects were given no time limit, and most were able
THE INTERACTION OF TENSE AND ASPECT
121
to complete the entire questionnaire in approximately five minutes. Any questions raised were answered after it was completed. The aim of the questions was to test for the presence of implicatures of counterfactuality present in the past tense forms of the stative verbs used, or as in Question 4, the past progressive aspect. The questions contained no past modal forms. The four questions to be analysed were presented as follows: What is most strongly suggested in the following italicised sentences? Please choose an answer from a), b), or c) below each one. 1) That lemon tree was in the ground for about 2 months. a) The tree has lemons on it. b) The tree is still in the ground. c) The tree is no longer in the ground. 2) He loved her all his life. a) He is no longer alive, but she is. b) They are both alive. c) She is no longer alive, but he is. 3) They used to live in Changi. a) They don’t live there now. b) They live in Changi now. c) They’re living in Changi now, but only for a short while. 4) This machine was running for a long time. a) The machine is now under repair. b) The machine is continuing to run now. c) The machine is not running at present. 5.3.3
Results
The Counterfactual Index The counterfactual index is measured by the sequence of responses: Q.1(c); Q.2(a); Q.3(a); Q.4(c), since it is this pattern which represents a measure of maximal counterfactual evaluation of the selected sentences. The totals for each group producing the indexical sequence are listed below in Table 5.2.
122 Table 5.1
HYPOTHETICAL MODALITY
Frequency of counterfactual implicatures for Singaporean and Australian groups, graded by educational level
Answer:
(a)
Question 1
Question 2
Question 3
Question 4
Table 5.3
(b)
Sec. 4 Express Sec. 4 NUS 1 NUS 2
04/47 = 8.5% 01/28 = 3.5% 03/39 = 7.6% 00
21/47 08/28 12/39 10/34
Control
02/38 = 5.2%
01/38 = 2.6%
35/38 = 92.1%
Sec. 4 Express Sec. 4 NUS 1 NUS 2
21/47 12/28 25/39 21/34
13/47 12/28 09/39 06/34
13/47 04/28 05/39 07/34
Control
32/38 = 84.2%
02/38 = 5.2%
04/38= 10.5%
Sec. 4 Express Sec. 4 NUS 1 NUS 2
46/47 26/28 38/39 34/34
01/47 = 2.1% 01/28 = 3.5% 01/39 = 2.5% 00
00 01/28 = 3.5% 00 00
Control
38/38 = 100%
00
00
Sec. 4 Express Sec. 4 NUS 1 NUS 2
04/47 01/28 06/39 02/34
20/47 13/28 09/39 08/34
Control
04/37 = 10.8%
= = = =
= = = =
= = = =
47% 42.8% 64.1% 61.7%
97.8% 92.8% 97.4% 100%
8.5% 3.5% 15.3% 5.8%
= = = =
(c)
= = = =
= = = =
44.6% 28.5% 30.7% 29.4%
27.6% 42.8% 23% 17.6%
42.5% 46.4% 23% 23.5%
02/37 = 5.4%
22/47 19/28 24/39 24/34
23/47 14/28 24/39 24/34
= = = =
= = = =
= = = =
46.8% 67.8% 61.5% 70.5%
27.6% 14.2% 12.8% 20.5%
48.9% 50% 61.5% 70.5%
31/37 = 83.7%
Proportion of deviation of the Singaporean counterfactual indices from the Control index
Sec. 4 Express Sec. 4 NUS 1 NUS 2
44.1% 46.1% 31.7% 23.4%
Average percentage of deviation from the Control index: 36.3%
123
THE INTERACTION OF TENSE AND ASPECT
100
% CF Responses
80
60
40 Q1 Q2 Q3 Q4
20
0 Sec 4 Exp
Sec 4
NUS 1
NUS 2
Control
Subject group
Figure 5.1
Graph showing percentages of counterfactual (CF) responses per group for each question
Table 5.2
Frequency of occurrence of the counterfactual index pattern, for Singaporean and Australian (control) groups
Sec. 4 Express Sec. 4 NUS 1 NUS 2
11/47 06/28 14/39 15/34
23.4% 21.4% 35.8% 44.1%
Control
25/37
67.5%
Average index of counterfactuality over all four Singaporean groups: 31.2% Average index of counterfactuality for the Control: 67.5%
124
HYPOTHETICAL MODALITY
%deviation from Control index
50% 45% 40% 35% 30% 25% 20% Sec. 4 Express
Sec. 4
NUS 1
NUS 2
Subject Group
Figure 5.2
Graph showing the percentage of deviation of the counterfactual indices of the Singaporean groups from that of the Control
5.4 Discussion In the following discussion, each of the questions will be analysed in turn, and a justification given for the use of each one in the questionnaire. A summary of the results will follow. 1) That lemon tree was in the ground for about 2 months. a) The tree has lemons on it. b) The tree is still in the ground. c) The tree is no longer in the ground. This sentence was very similar to one previously informally observed of an educated Singaporean speaker known to me. At the time at which it was uttered, the speaker was remarking on a tree which he had planted two months prior to making the statement, and which was still alive and growing in the ground. It was presumed that the most appropriate way of expressing such a notion for a native speaker would be by means of the present perfect, rather than the simple past, although naturally there would be expected considerable dialectal differences
THE INTERACTION OF TENSE AND ASPECT
125
even for native speakers.2 The use of the present perfect rather than the simple past in this instance would be more likely to convey to the speaker the fact that the tree was still in the ground, in other words, meanings of ‘current relevance’ usually associated with such forms (That tree has been in the ground for about 2 months). However, the consequences of the use of past tense with a stative verb were predicted to produce in this case an implicature of counterfactuality indicating that the state referred to in the past is no longer continuing. The additional use of an adverbial indicating a specific temporal duration, for about 2 months, when co-occurring with a past stative verb, increases the strength of the implicature, suggesting even more the likelihood of the past state having terminated prior to the moment of speaking. Durational adverbials such as this are found to be equally applicable to non-terminative senses, e.g. That tree has been in the ground for about 2 months, but this indicates that it is not the durational adverbial alone to which the terminative sense is attributed, but the past stative in combination with the durational adverbial. A (c) answer would suggest that the counterfactual implicature has been apprehended by the informant; a (b) answer would indicate that the informant either does not perceive the counterfactual implicature to be salient or that she has not apprehended it in the meaning of the construction. The (a) answer was the distractor. In the Secondary 4 Express group, which was the youngest of all the groups, the proportion of (c) answers was less than half (46.8%), suggesting that more than half of the informants did not perceive the counterfactuality of the construction. Of the remainder, 44.6% gave the construction a ‘factual’ interpretation (i.e. interpreted a meaning of factuality with regard to the present as well as the past) by choosing the answer (b) stating that the tree is still in the ground. Only 8.5% selected the distractor. With the increasing age of the informants, their willingness to select the counterfactual interpretation also increased. Although there is little difference between the figures for the three older Singaporean groups (67.8%, 61.5% and 70.5% respectively), there is quite a margin of difference between the frequency of the counterfactual responses for the youngest group (Sec. 4 Express — 46.8%) and the second-year university undergraduates (NUS 2 — 70.5%). Furthermore, the data from the Control group shows the proportion of counterfactual responses was much higher, with 92.1% of the subjects selecting (c) and only 2.6% selecting a (b) answer. This indicates that for this question there was a minimum
2. As noted by Palmer (1974: 53) and others, speakers of US English often use the simple past for what are present perfect contexts in other dialects. It would be interesting, though, to investigate whether stative verbs were affected as often as non-stative ones for US speakers.
126
HYPOTHETICAL MODALITY
of 21.6% difference between the Singaporean groups’ highest counterfactual response and that of the Control group. There is a strong possibility in all of these figures that the apprehension of counterfactual meaning in the form of a grammatical implicature is something which is acquired through prolonged exposure to the language. Such meanings may not be part of the classroom input; it is considered that this is not likely since they most probably would have been acquired by the tenth year of English-medium education, if this had been the case. With prolonged exposure to English, such implicative meanings may become gradually instilled in the learner’s ‘meta-knowledge’ of the meaning of certain constructions; however, the present data might suggest that for many Singaporeans, grammatical implicatures of counterfactuality are not acquired in the early stages of learning English. Question 2. He loved her all his life. a) He is no longer alive, but she is. b) They are both still alive. c) She is no longer alive, but he is. Hatav (1989: 497) discusses the distinction between states and events with reference to their points on an imaginery time line, suggesting that in order for a state to appear as a point on this time line, it must be marked off by a durational adverbial, such as for three days. As such, the situation is pragmatically inferred to be bounded in time, although logically it may persist for a longer period. The example given in Question 2 tests the presence of this pragmatic inference derived from the use of the durational adverbial all his life proposed as the premise for the inference or implicature. According to Hatav (1989) it is the durational adverbial which provides the pragmatic implicature of discontinuity into the present. It is proposed in the present study that the already-present implicature derived from a past stative is only reinforced or strengthened by the durational adverbial, which is not the only source for the development of this implicature. Adverbials of this nature can co-occur with the present perfect (as discussed above for Question 1), thus altering the focus of the temporal boundary to the initial point of the duration rather than end-point (see Depraetere 1995 for more discussion). The fact that the stative situation may be viewed from the initial point of the temporal boundary with durational adverbials used in present perfect constructions but from the end point of the temporal boundary with those used in past stative constructions suggests most strongly that it is the tense of the verb itself rather than the durational adverbial which is providing the implicature of present hypotheticality. The following examples (5)–(7) illustrate the likelihood that this is the case:
THE INTERACTION OF TENSE AND ASPECT
(5)
127
She has been a teacher for three years.
In (5) the focus of the temporal boundary is on the starting boundary of the three year time period, since its endpoint in the present does not imply that the period in which she was a teacher has necessarily terminated at the moment of speaking. However, in (6) the time period is interpreted as ending before the moment of speaking: (6)
She was a teacher for three years.
This is because the use of a past stative verb carries implicatures of irrealis, resulting in the focus of the temporal boundary shifting to the endpoint or ‘righthand boundary’ (Depraetere 1995: 3) of the duration expressed by the adverbial. Logically, however, this need not be so, as Hatav maintains (1989: 497), and evidence that it is the tense of the verb which is determining the shift in boundary focus is provided in examples such as the following: (7)
She had been a teacher for three years (and in fact she was still a teacher when I met her.)
In (7), the use of the perfect construction should, under the analysis for (5), enable an interpretation of continuity into the current frame of reference, whether at the present moment of speaking (as in (5)) or at the current temporal reference of a past situation (as in (7)), since it is the starting point of the temporal boundary which is in focus. However, when the use of the perfect is combined with past tense, the implicature of discontinuity beyond a ‘right-hand’ or terminal temporal boundary becomes evident, as is shown when a conjoined clause is used to cancel it. The evidence of the implicature is therefore an indication that it is the past tense marking in the perfect construction (had been) which alters the boundary focus in the durational adverbial from its starting point, as in a present perfect, to its endpoint, in a past perfect (cf. She has been a teacher for three years ?and in fact she still is, in which the conjoined clause appears redundant). However, the counterfactuality relating to the durational adverbial, all his life, is tested in Question 2 rather than the counterfactuality of the predicate loved. What is questioned, in fact, is whether the state of ‘loving’ demarcated by the temporal boundary all his life still holds, and from this it follows to question whether the endpoint of that temporal boundary has been reached. An (a) answer would then be the most appropriate answer supplied by an informant who had perceived the implicature of temporal termination associated with the use of the past stative verb and reinforced with the durational adverbial. A (b) answer probably would suggest that the counterfactual implicature had not been apprehended, or else that the situation of him loving her had been held by the
128
HYPOTHETICAL MODALITY
informant to be still continuing, since the adverbial states that the time period was ‘all his life’. (c) was the distractor, but its capacity in that function seemed less than obvious to some of the informants, as the results show. A (c) answer is possibly inferrable from He loved her alone. Another possible interpretation of Question 2 is: They are both no longer alive; but we have no information about whether ‘she’ is still alive or not, and the most readily accessible implicature is that suggested in (a). The results obtained were not unlike those of Question 1. The counterfactual implicature was interpreted by almost exactly the same number of informants at the Sec. 4 Express level (the youngest group with the shortest period of exposure to English) — 46.8% for Question 1 and 47% for Question 2. On the other hand, the proportions of (b) and (c) answers to Question 2 were the same (27.6%), perhaps because (c) appeared to be less of a distractor to some of the informants than it was intended to be. However, the number of (c) answers in the older Sec. 4 group, some of whom were in their eleventh year of English-medium education, was significantly lower at 14.2%. The proportions of counterfactual and non-counterfactual responses ((a) and (b) respectively) were exactly the same for this group (42.8%). There was little difference between the two Sec. 4 groups in the frequency with which the counterfactuality was perceived for this question (only 4.2%). For the two Singaporean university undergraduate groups, the percentages of counterfactual responses were higher than those of the Sec. 4 groups (64.1% for NUS 1 and 61.7% for NUS 2). 23% of the NUS 1 group provided a ‘factual’ response to Question 2, as did 17.6% of the NUS 2 group. The Control group provided the largest counterfactual response to this question (84.2%), indicating a margin of 20.1% between the highest counterfactual score for the Singaporean groups and that of the native speakers. Question 3. They used to live in Changi. a) They don’t live there now. b) They live in Changi now. c) They’re living in Changi now, but only for a short while. The form used to has been described as a marginal modal, though having the semantics of an auxiliary of tense and aspect (Quirk et al. 1985: 140). In function it has been described as frequently occurring to mark past habituality, but not an obligatory marker since it can alternate with other forms such as the simple past, as in the example given by Bybee (1994: 239): She sang to him every day. Would + V is another form used for this function, and this has also been observed in Singaporean usage (Ho and Platt 1993: 78).
THE INTERACTION OF TENSE AND ASPECT
129
The reasons for the inclusion of the past habitual in Question 3 are, firstly, to test the extent to which the counterfactual implicatures apparently present in the use of a stative verb with past tense marking are also present with another type of non-punctual aspect. The fact that the implicatures associated with past statives have been variably perceived by the Singaporean group is hypothesised to be related to the opposition between perfective and imperfective aspect in some varieties of Singaporean English and its compatability with past tense marking. It has been claimed (Comrie 1976: 28) that the meaning of discontinuity into the present with the use of used to is also merely an implicature, and is not part of the actual meaning of such constructions employing this form, as is suggested by the following example (ibid.: 29): (8)
[Bill] used to be a member of a subversive organisation, and he still is / but I don’t know whether or not he is now.
The proof that the meaning of discontinuity of a past habitual situation carries only an implicature of present irrealis is found in the conjoined clauses which act to cancel the implicature. The second reason for testing the implicature of the past habitual is that the form use to, without past tense inflection, has been observed in some Singaporean dialects to be a marker of present habituality (Ho and Platt 1993: 78, and Platt, Weber and Ho 1984: 71). The reasons for this are not clear. There is a remote possibility that use to can be traced back to an earlier period of colonisation when the present tense forms were more likely to be still in use: an example of the archaic present tense use of use to has been recorded to have appeared as late as 1928 (Visser 1969: 1411). Considering that a similar form, usa, marking the habitual present in Pitcairnese creole, has been noted by Labov (1990: 22–23), and that Pitcairnese has been known for a number of retained (but mainly lexical) archaisms (Kallgard 1993: 71), this possibility would not be unlikely. Labov ascribed the form to a back-formation from the habitual past used to, since the past is unmarked in Pitcairnese (ibid.: 23); the form has also been noted in West African pidgins (Keith Allan, p.c.) and attributed to a possible back-formation in such cases. If this is the case, then the languages using the form to mark present habituality may be seen to be overgeneralising the marking of habituality to environments in which it would appear there was no necessity for it, such as the present. Bybee (1994: 244) maintains that habituality is the default function of the present tense, and this might be one of the the reasons that the present tense forms of used to gradually became obsolete. It would seem unusual to find present habitual aspect marked in any language with a tense distinction for present and past, if this were the case. Therefore, the
130
HYPOTHETICAL MODALITY
appearance of an overt means of marking habituality in the present might suggest that aspectual distinctions were overriding tense distinctions, or that an absence of tense distinctions between present and past necessitated the presence of a general marker for habitual aspect in the aspectual system. This would mean that as tense distinctions gradually became more grammaticalised in varieties which previously did not mark them, and the meanings of an overt present habitual aspect were gradually taken over by those of ‘default’ present tense categories (as suggested by Bybee), then the use of the generalised form for habitual aspect in the present would disappear, remaining restricted to its standard past functions. In Question 3, an (a) answer would be an indication that the counterfactual implicature of discontinuity into the present has been perceived; a (b) answer would suggest that either the implicature has not been perceived by the informant, or that it has been perceived and has been suppressed (this cannot be discounted for any of the questions, and there is no way of determining whether this is so, short of questioning every informant on the motivations for his choice of answer. However, the instructions to select the ‘most strongly suggested’ answer were intended to direct the informant to the most readily interpretable answer). Answer (c) was intended as the distractor. The results of Question 3 suggest that for almost all of the informants, the counterfactual implicature was apprehended with hardly any variation in the interpretations. Nearly 100% of the informants in all the groups interpreted the sentence counterfactually and absolute scores of 100% were obtained for an (a) answer for the NUS 2 group and the Control. The results seem to indicate that the implicature of counterfactuality associated with the use of the past habitual marker used to may be well on the way to becoming conventionalised in the dialects represented in the present study. Certainly the results contradict the evidence from previous literature suggesting that used to, or a phonologically reduced form of this marker (use to), is generalised to present temporal uses in Singaporean dialects. While it is possible that the past tense inflection may be influencing selection by some Singaporean informants, it is equally likely that morphological markers of tense do not have as much significance as they do for native speakers for whom temporal distinctions have more grammatical salience. It could be predicted from these results that the use of use to in Singaporean English as a marker of present habituality is either disappearing or is restricted to very basilectal varieties which were not surveyed as part of the present study. Question 4. This machine was running for a long time. This was again an example based on a similar example observed informally of a Singaporean speaker: This machine was running for three years — at the time
THE INTERACTION OF TENSE AND ASPECT
131
at which it was uttered the machine was still running. It was considered that in this instance, the use of a present perfect form of the verb would be more appropriate to express the sense of continuity into the present. As with the previous examples in Questions 1–3, there is a likelihood that the past tense used with a non-punctual verb will produce an implicature of irrealis in the present, that is, suggesting that the machine is either not still running, or at least not known to be still running. Hatav (1989) finds sentences like Question 4 unacceptable, due to the fact that the progressive is normally used to deprive events of their endpoints, and examples such as (9)
*John was sleeping for ten hours (Hatav 1989: 501)
are considered anomalous because of the inability of the progressive to co-occur with duration adverbials. It is this factor, she claims, which distinguishes states from progressives, in that states can be delimited by duration adverbials, as shown in Question 1 above, but progressives cannot. Although this appears a valid argument for the incongruity of the progressive with the duration adverbial in (9), it does not account for the possibility that other types of duration adverbials may appear quite acceptable, e.g. (10)
John was sleeping all night.
This is similar to an example given by Palmer (1974: 55) (I was reading all morning), who further maintains that duration is a feature associated with the progressive, as shown by its collocation with such adverbials as all morning, for a long time, and continually. Thomson and Martinet (1988: 164) also note the possibility of the co-occurrence of the progressive with adverbials expressing a duration between two time points, as the following example indicates: (11)
From four to six Tom was washing the car.
Therefore, the argument that it is the specific endpoints indicated in the use of durative adverbs which are incompatible with the use of the progressive may not hold for examples such as (11). The question remains, then why (9) is apparently unacceptable, and examples such as (10) and (11) may not be. Perhaps the reason lies in the potential for durative adverbials to create ambiguities, as suggested by Binnick (1991: 309), and to the fact that the adverbial can refer either to the period of the activity or to its resultant state; thus, he subscribed to the magazine for four years could suggest that either the subscription ran for four years or that the activity of subscribing took place over a time period of four years. In the
132
HYPOTHETICAL MODALITY
same way (9) may be seen as unacceptable if the reference of the durative adverbial is on the accomplished state, the end of the ten-hour period, which would be incompatible with the use of the progressive. If the reference is on the period of the activity of sleeping, however, the sentence might have a more acceptable interpretation. The ambiguity thus results from whether the adverbial is sentential or phrasal. As a sentential adverbial, it seems acceptable, and can be disambiguated by moving the adverbial to a different position, for example: (12)
For four years, he subscribed to the magazine;
(13)
For ten hours, John was sleeping.
or
However, the most appropriate form might be the non-progressive: John slept for ten hours. What this seems to suggest is the ability of durational adverbs to place the focus on the conclusion of the activity rather than on the process of the activity itself, as it would be, for example, in (10) or (13). For this reason duratives with less definitive endpoints (for a long time; all night) might seem more compatible with the progressive. The use of past progressive then, must be stative-like in that the duration of the activity is more focussed than the temporal boundaries between which it occurred, and any senses of ‘perfectivity’ or temporal boundedness associated with the use of a duration adverbial must be no more than an implicature. There should be no reason, therefore, not to presume that the use of past tense with the progressive in Question 4 does not produce the implicature of present irrealis prevalent in Questions 1–3. The results indicate that this is indeed the case, showing patterns similar to those indicated for the other three questions. A (c) answer was intended to elicit a counterfactual response; a (b) answer would indicate that the implicature of counterfactuality either had not been apprehended or that it was considered not to be salient in the meaning of the sentence; and the (a) answer was intended to be the distractor. Implicit in the meaning of the distractor was the counterfactuality expressed in the (c) interpretation — that the machine was under repair would imply that it was not running. However, it would seem that if a respondent wished to express this interpretation of Question 4, s/he would be more likely to select the more explicit and direct means of doing so, by using (c), the implicature relating to a (c) interpretation being more directly derived from Question 4 than that of (a). In the youngest group, the Sec. 4 Express, nearly half of the group (48.9%) selected the counterfactual response in (c), while 42.5% selected the factual one (b). A small number (8.5%) chose the distractor (a) as their response, suggesting
THE INTERACTION OF TENSE AND ASPECT
133
that it was appropriately recognised as such by the informants. The figures were similar for the Sec. 4 group, of whom 50% gave a counterfactual response, and a slightly higher number than in the Sec. 4 Express group responded with a factual answer (46.4%). This figure for answer (b) fell to half as many for the two NUS groups, both of them supplying only around 23% of (b) answers. It was matched in the NUS 1 group by an increase to 61.5% of counterfactual responses — (c) answers, and 70.5% of the NUS 2 group responded in this manner. The NUS 1 group supplied the highest proportion overall of (a) answers (15.3%) though this was still not a high percentage. It is interesting to note that the proportions of counterfactual responses for both Question 1 and Question 4 were exactly the same for the two NUS groups: 61.5% for NUS 1 for both questions, and 70.5% for NUS 2 for both questions. Similarly, the counterfactual response for the Control group for Question 4 was almost the same as that for Question 2 (83.7% and 84.2% respectively). These statistical correlations are extremely significant in indicating the consistency with which the counterfactual implicatures can be interpreted across four different types of sentences, suggesting that there is little likelihood of chance probabilities affecting the answers to the questions. The frequency of counterfactual interpretations for the Control group (83.7%) showed that for Question 4, there was only a 13.2% margin of difference between that and the highest counterfactual response for the Singaporean group. The margin was narrower than those of Questions 1 and 2, and the Control counterfactual score was the lowest for this question. Nevertheless, the frequency of counterfactual responses for this question are still high, indicating that the implicatures relating to past stative verbs are common to past progressives as well. This stands as a justifiable argument that progressives may be interpreted pragmatically in the same way as stative verbs.
5.5 Concluding remarks Bybee, Pagliuca and Perkins (1994: 253) suggest that the examination of patterns of use to identify cases of retention of older senses in a grammaticalising feature can prove a highly efficient means of diagnosing earlier historical developments of that feature. The presence of such constraining factors or retained meanings reveals that a grammatical item can sometimes retain, amongst its various senses, a portion of meaning reflective of older uses, which has not been bleached before the more advanced stages of grammaticalisation are reached. Bybee et al. also note that evidence of the difference between a language that has a grammaticalised
134
HYPOTHETICAL MODALITY
past tense category and one that has a perfective is in the presence or absence of a past imperfective category. Given Ho and Platt’s (1993) data, it would appear that past tense marking has not yet been fully generalised to imperfective environments at all levels of Singaporean English, and is mainly restricted to marking perfectivity on punctual and telic verb types. In the present study, it is shown that the perception of counterfactuality as an implicature derived from the use of past tense with stative verbs will vary cross-dialectally, as the results from the Singaporean group surveyed indicated a lower average index of counterfactual perception overall (31.2%) than that of the Control group (67.5%). The explanation for such findings is firstly related to the tense/aspect system of Singaporean English, which does not categorically mark non-punctual verbs for past tense. This being the case, the co-occurrence of past tense with stative verbs will often not be present to provide the implicature of hypotheticality provided by such means in standard varieties, and if it is present, it is predicted to have less salience as a grammatical marker than in standard varieties. Thus, the general infrequency of past tense marking co-occurring with stative verbs is likely to have consequences for the expression and perception of counterfactual implicatures in many grammatical environments. Bloom’s (1981) assumptions, then, may have been well-founded, but it has been shown in the present study and in previous studies mentioned that it is not the absence of cognitive schemas related to counterfactual notions that may have affected the perception of counterfactuality by the Chinese L1 informants. Rather it is the fact that informants of L1 backgrounds in which stative verbs are not marked for past are less likely to recognise a particular grammatical form, the past tense form of a stative verb, as signalling inferences of counterfactual or hypothetical meaning in the L2. Furthermore, it appears that the grammaticalised implicatures derived from such forms are acquired over a prolonged period of time and exposure to English, as indicated in the present data by the steadily rising counterfactual responses which increase with age and educational level. The rise in the perception of counterfactual implicatures with age and educational level may be compared with Ho and Platt’s (1993) results, in which they also found an increase in the use of past tense marking with non-punctual verbs according to educational level (for example, the verb know was marked for past for 0% of the tokens of Primary-level speakers, but for 51.6% of tokens of Tertiary-level speakers — Ho and Platt 1993: 130). There is an indication in this and in the present data that the grammaticalisation of past tense amongst such speakers as it is generalised to non-punctual environments carries with it retained senses of perfectivity, but that such senses are not perceived as part of the meaning so long as the use of past marking in such environments may be in free
THE INTERACTION OF TENSE AND ASPECT
135
variation with its non-use. Semantic distinctions which are prominant in L1 usage may indeed influence the comprehension, as well as the production, of conflicting distinctions in L2 usage. For such speakers, it is not an issue whether the state referred to is marked as being in the past or present, so long as the lexical aspect defines it as a state; thus the use of past marking with a stative verb as pragmatically implying completion of that state, for some speakers, is still in competition with the inherent lexical aspectual senses of the verb signalling a semantic state of non-completedness. The state referred to in the verb may then be regarded by such speakers as past but not terminated prior to the moment of speaking. Clearly there is vacillation between the former lexical aspectual senses and the later pragmatic senses of ‘completed state’ (see Table 5.1). Such figures may be indicative of the mid-point of a developing change. The random alternation in Singaporean English between the modal forms will and would considered in previous studies to be related to hypercorrection rules or simply to the nature of free variation (Teow 1988) can now be more accurately explained. If past stative verb forms are not categorically perceived as conveying implicatures of present counterfactuality amongst Singaporean speakers of English, then it is hardly likely that the differences of irrealis meaning between will and would, as stative verbs, will have a great deal of significance either (see Ch. 4, Table 4.1). Under this analysis, the substitution of would for will, referred to as a hypercorrection in educated Singaporean English, may in fact be an overgeneralisation — this is evident in that it alternates with will in the same environments. The alternation may be an indication that some elements of the total meaning of the grammaticalising modal have not been fully acquired, or that earlier senses reflective of the diachronic development of the modal, the senses from which implicatures of irrealis were derived, past tense statives, are less salient to the L2 speaker than to the L1 speaker. The senses which may provide the implicatures of irrealis in L1 dialects are retentions of the earlier stages of the grammaticalisation of past tenses, which were likely to have initially affected only perfective environments (as suggested by Bybee et al. 1994). The overgeneralisation of will and would (and also can and could) in some Singaporean dialects is therefore predicted to be due to the absence in the meaning of would of former, retained senses of perfectivity associated with the first uses of the past tense. The overgeneralisation is a two-way process, since either form can occur in the same environment, showing that there is no difference in the factuality of the meanings conveyed by either form for the speakers who uttered them. There is thus less likely to be differentiation for the L2 speaker in the perception of degrees of hypotheticality which can be derived from the form of the modal,
136
HYPOTHETICAL MODALITY
whether past or non-past. While the lack of salience of past tense with stative verbs in the substratum language should not be disregarded, it must also be considered that for L2 speakers the modals have been acquired at more advanced stages of grammaticalisation, and hypothetical meanings associated with earlier diachronic processes are less likely to be salient in the composition of their meaning than for L1 speakers. The present findings clearly suggest, then, that the variation in the formal expression of hypotheticality by Singaporean speakers shown in previous studies is related directly to variation in the association of hypothetical meaning with past modals, which are stative verbs. Such variation is also applicable to the other past stative forms (as shown in Q. 1–2 in the study), and past progressives as well (Q. 4), indicating the stative senses associated with progressive aspectual meaning. Past habituals marked with used to (Q. 3) almost invariably receive a counterfactual irrealis interpretation according to the data, probably due to the fact that the implicatures associated with the use of this form have become more conventionalised than have the implicatures of other examples used in the study. The gradual increase in the degree of counterfactuality apprehended by the informants is concomitant with age and length of exposure to English-medium education. Such results indicate that grammatical implicatures are meanings which are progressively acquired as part of the learner’s increasing pragmatic competence — this notion has been proposed by Slobin (1994: 129–30) to explain the gradual acquisition of implicatures relating to the development of the use of the present perfect by children. The perception of hypothetical implicatures signalled in the past stative verb forms therefore develops in accord with individual maturity and exposure to the language in a formal environment. Thus, as discussed in the Introduction, it is seen that the grammatical combination of past or perfect morphology with irrealis or predictive markers in hypothetical clauses does have some reality as an indicator of counterfactuality, so long as the irrealis markers are in the form of stative verbs or express imperfective aspect. The implicatures of counterfactuality created in this way are the result of a conflict between perfective grammatical aspect retained in the meaning of the past tense, and imperfect lexical aspect inherent in stative verbs (imperfect grammatical aspect may also conflict with perfective grammatical aspect in the examples of the progressives — Q. 4). The pragmatic implicatures may not be available to L1 speakers of languages in which the relations between grammatical and lexical aspect are more semantically defined and in which tense is not an available category; e.g. Chinese. The lack of saliency attributed to preterite forms of modal verbs by such speakers, then, is merely a reflection of L1 influences.
C 6 Hypothetical WILL: A study in retention
6.1 Australian English In Ch. 3, the generalisation of the modal would to conditional apodoses was seen as a development subsequent to the grammaticalisation of the modal to express past prediction via the conventionalising of Quantity 2 implicatures. The generalisation was relatively late, as shown by the data in Ch. 3 and also Molencki’s (1998) study; i.e. for the most part, it did not take place until Early Modern English times, suggesting that when the modal generalised to use in conditional apodoses, it would have been bleached of most of its lexical source meanings of volition in other main clause environments. This was shown by the increase in main clause grammatical functions which corresponded with a general decrease in lexical functions, and the absence of examples in the data of counterfactual conditional apodoses in which retained volitional senses must have appeared as ambiguities of meaning. It was also noted in Ch. 3 that the development of hypothetical senses in would varied according to the environment in which the modal occurs. It could be suggested that most of Bybee’s (1995) OE and ME examples of wolde, in which the past form of the modal refers to present time, appear with first person subjects, and in quoted speech rather than in narrative. Such environments are likely to produce a higher incidence of retention, since the subject is also the speaker, who will have the most awareness of the subject’s intentions and desires, and as such will not need to make predictions. This is probably the reason that volitional lexical senses persisted even after the temporal lexical meanings of past had long faded. In a narrative context, though, in which wolde might often refer to the desires of third person subjects in main clauses, there is a greater likelihood that the desires can be subjectified as predictions, and if a past time reference point is supplied, as is often the case in narratives, the prediction is made relative to that time point, creating a meaning of future-in-the-
138
HYPOTHETICAL MODALITY
past. The evidence of retention of volitional senses is therefore a reliable indicator of the level of grammaticalisation achieved by the modal in a particular environment. In the present chapter, the use of the modal would will be investigated as a study in retention in the complement clauses of the verb wish. The evidence of the retention of volitional senses seems to indicate that the use of would in such contexts may be at a relatively early stage of grammaticalisation. (Evidence is also supplied by the continued use of older, pluperfect forms for past wishes; e.g.: I wish he had not entered the race, rather than I wish he would not have entered the race, which is disallowed in many dialects.) The study is undertaken first with Australian L1 speakers of English (part 1), then with Singaporean speakers of L2 English (part 2), and finally using 3 groups of both L1 and L2 speakers of British English (part 3). It is suggested that the levels of retention are slightly higher with L1 speakers, on average, than with L2 speakers. Various explanations are proposed to account for this in Ch.7. The investigation was motivated by an anomalous feature which emerged from the study of hypothetical expressions in educated Singaporean English discussed in Chapter 4 (Ziegeler 1994), in which were examined the use of forms marking hypotheticality in two environments: conditional clauses and hypothetical complement clauses of the verb wish. The previous results included elicited responses in the form of counterfactual conditional sentences and expressions of the informants’ personal wishes. The data revealed that in all the I wish constructions followed by a finite subordinate clause, will and would were observed to occur frequently, but in 5 of the 61 responses to the two questions which used a finite clause, the use of would appeared to be either overgeneralised or hypercorrect. Two of the examples in the previous study which seemed overgeneralised were referred to in Ch. 4: (37) and (38); they are repeated below in (1) and (5) with other similar examples: (1) (2) (3) (4) (5)
I wish that I would earn lots of money and help save (as far as possible) those that are in need. I wish that the ozone layer would not be depleted so quickly. I wish that there would be fairness and equalness in all … I wish that there wouldn’t be too many unfortunate or handicapped people. I wish that there would be peace in the world.
However, the inappropriateness of the use of would in these examples seemed to be a point of dispute when discussed later with native speakers. Under
HYPOTHETICAL WILL
139
the circumstances, it was considered that although the use of will (which also occurred in the data) in such contexts was inconsistent with patterns of standard usage elsewhere, the use of would in clauses of this type could often be acceptable. The modal would does appear to have a place in Hypothetical predicates, as noted by Leech (1987: 121, fn.) and Palmer (1986: 152), who describe the expressions containing them as wishes of the future; thus, the proposition expressed in the subordinate clause of I wish John would come is the remote counterpart, according to Palmer, of John will come. This is a wish of future temporality which is remote, but only in terms of the speaker’s opinion of the likelihood of the proposition being realised, rather than counterfactual in terms of the speaker’s knowledge of the realisation of the proposition, something which is usually possible to determine only in unreal clauses of present or past temporality.1 Thus, the association of the term ‘Hypothetical’ with specific temporal distinctions is made here on account of Palmer’s assumptions that the modal appearing in wish complements is a remote future modal. This association is consistent with earlier definitions of the distinction between Hypotheticality and counterfactuality (see Ch. 2) stating that counterfactuality applies only when the speaker is aware of known facts to the contrary of the counterfactual expression, and facts relating to the future are not usually verifiable at the time of speaking. Palmer’s treatment of would in counterfactual predicates is thus in accordance with the view that will and would carry the same modal meaning, but that would is simply a back-shifted form of will, used to express remoteness of mood, or factual remoteness. Others (e.g. Fleischman 1989 and 1995) have regarded the past forms of modals as expressive of less epistemic certainty than their present form counterparts, conveying in the use of past forms a metaphor of factual remoteness (see discussion in Ch. 2). The present study will then hold as a basic assumption that would in such clauses is a grammaticalising predictive marker expressing (remote) future meanings. Therefore will and would will be treated as basically two forms of a single lexeme WILL and not as separate modals, and the behaviour of would in such clauses will be examined from the perspective of will. The term covering both forms will be used throughout the present chapter, unless citation from specific sources requires the adoption of the form used in the work referred to. Thus, according to Palmer’s description, the likelihood of examples (1) to (5) being consistent with acceptable usage can only be considered if the temporal reference is perceived to be future. The question remains, though, why these particular examples appear intuitively marginal with respect to the use of WILL
1. Dudman’s (1984) example discussed in Ch. 2 (3) is a rare exception.
140
HYPOTHETICAL MODALITY
to express unlikely future wishes, and it is this question that will constitute the hypothesis to be tested in the present investigation. Although the use of WILL in hypothetical protases has been frequently discussed (see, e.g., Comrie 1982; Nieuwint 1986; and Haegeman and Wekker 1984), less attention appears to have been given to analysing its occurrence in other hypothetical contexts. One would feel somewhat hesitant, therefore, in subjectively evaluating the five examples listed above as evidence of the inappropriate use of the modal would, without at least some support from the unbiased assessment of a group of native speakers. It is for this reason that the present survey was undertaken. In the following sections (6.1.1–6.1.1.5) the distributional constraints on the use of WILL in hypothetical predicates as well as in non-hypothetical contexts will be discussed in the light of previous research into the grammaticalising nature of the modal and its functions as a future auxiliary. Therefore, throughout this discussion, it will be assumed, in accord with Palmer’s (1986) treatment, that WILL occurring in hypothetical predicates is simply a hypothetical variant of will, and therefore should behave in the same manner as its non-hypothetical counterpart. 6.1.1
Earlier discussions
Although discussion on the use of WILL in hypothetical predicates has received only minimal attention by previous researchers, some reference grammars summarise the limitations on the use of wish + subject + would as restricted to constructions in which there is an expressed interest on the part of the main clause subject in the subordinate clause subject’s willingness or unwillingness to perform a present action, or in which the subject of wish may express dissatisfaction with the present state of affairs and a desire for future change (Thomson and Martinet 1988: 261–2). In such cases the subject of WILL is usually animate or personal, and cannot be identical to the subject of wish in the main clause. Given such broad definitions, the decisions required by the speaker for the use or non-use of would in these contexts might appear quite arbitrary. If the modal did serve to mark only temporality in the hypothetical clause, as Palmer suggests, then it would seem unusual to find so many restrictions on its grammatical distribution, restrictions which would not normally be associated with its use in non-hypothetical contexts. That there is a necessity for subject-oriented control or potential agentivity in the hypothetical object clauses of the predicator wish seems highly likely, and given the constraints referred to by Thomson and Martinet (1988), it is hardly surprising that the examples (1) to (5) might appear to show either hypercorrect or inappropriate uses of WILL in such contexts. The
HYPOTHETICAL WILL
141
obvious importance of the role of the subject as a determining factor in constraining the use of WILL in predicates of remote future meaning seems to indicate that the modal occurring in such contexts cannot be interpreted categorically as expressing pure future temporality and that the underlying meanings reflecting its origins as a root modal of volition are still imposing constraints on its distribution in some grammatical environments. If these assumptions can be verified, then it might be possible to further suggest that this represents a stage in the grammaticalisation of the modal WILL at which both the old sense of volition and the new sense of prediction co-exist. This is clearly a case of lexical retention as defined in Ch. 3 (3.1). 6.1.1.1 Grammaticalisation and the epistemic status of WILL There is wide agreement that the historical processes of grammaticalisation are represented in the diachronic development of the modal WILL from a construction originally expressing only desire or intention into a form which now includes prediction or futurity amongst its meanings (see discussion in Ch. 3). Bybee (1985: 159) suggests that the development of future tenses is parallel to the grammaticalisation of epistemic modality from deontic and agent-oriented modality, and that this development was accompanied by a reduction in the application of the modality to the agent of the clause, as the modal meaning gradually extended its scope to cover the entire clause.2 As WILL has grammaticalised to include as one of its functions that of a future auxiliary, it might be considered to have followed the same path as the grammaticalisation of epistemic modality, especially if it can be ascertained that the apparent emphasis (referred to in 6.1.1) on the role of the subject (which is also the actor of the main verb in most cases) in hypothetical clauses could be linked to deontic or agentfocussed origins. Although previous accounts have not been entirely consistent over the use of the terms ‘deontic’ and ‘epistemic’ to refer to the volitional and the future meanings of WILL respectively, there is sufficient evidence (cf. Bybee, Pagliuca and Perkins 1991; Coates 1983; Fleischman 1982; Givón 1994; Goossens 1982; Heine, Claudi and Hunnemeyer 1991; Palmer 1987 and 1990; and Traugott 1989) to justify the use of these terms in order to locate WILL within a historical framework similar to that which traces the development of other English modals. From a comparative viewpoint, then, the terms ‘epistemic’ and ‘deontic’ may be used to refer respectively to the predictive and volitional meanings commonly associated with WILL.
2. See Ch. 3, note 5 regarding the use of the term ‘agent-orientation’.
142
HYPOTHETICAL MODALITY
The following sections (6.1.1.2–6.1.1.5) will discuss the use of hypothetical WILL in environments where the function of this modal as a future auxiliary is hypothesised to be affected by certain distributional constraints. The environments listed involve the use of specific features (stativity; progressive aspect; first person and inanimate subjects; subjects of equivalent cross-clausal identity; and negation), these having been selected for further investigation in the questionnaire (see 6.1.2.1). 6.1.1.2 Stative verbs Leech (1987: 121 note.) discusses the distinction between states and events when referring to the ‘hypothetical future’, claiming that in some dependent clauses, stative verbs cannot be combined with the modal would to form a hypothetical future, but instead use the hypothetical past tense, e.g. I wish that book belonged to me. Leech contrasts *I wish that book would belong to me, having a stative verb, as ungrammatical, compared with I wish the bus would arrive soon, in which there is an event verb. However, the use of a past stative verb is not totally restricted to providing future meaning, either, since I wish that book belonged to me can be quite easily understood as a wish about the present. Other accounts suggest that the hypothetical past tense form marks present hypothetical states which can be projected into the future, e.g. I wish I owned a car (Frawley 1992: 353); this is perhaps conceivable because of the cognitive notions of unboundedness often associated with states (ibid.: 149).3 It would seem from this, then, that hypothetical past stative verbs in counterfactual clauses can be thought of as having present temporality but with futurity implied in the continuation of that desired state, and because of this, there might be little cause to overtly mark them for future meaning with WILL in a hypothetical clause. However, future states appear quite felicitously in other (non-hypothetical) contexts in combination with the modal WILL, e.g. Next week that book will belong to me. The evidence leads again to a distributional analysis as a means of explaining the restrictions on the use of stative verbs, indicating that the cooccurrence of WILL with stativity might still be undergoing processes of grammaticalising in some environments. A stative verb could not combine easily in some contexts with a modal affected by the continued adherence of its original lexical meanings of volition or intention, since it would be difficult to
3. As has been pointed out to me by Keith Allan (p.c.), the use of a completive adverb such as already creates a notion of boundedness when combined with a stative verb, restricting its temporality to present (I wish that book already belonged to me).
HYPOTHETICAL WILL
143
find instances in which the subject could be said to actually impose any volitional control over a state, especially over a physical state. In order to test this, examples using stative verbs were included in the questionnaire (see 6.1.2.1 and Appendix). 6.1.1.3 The future progressive There are many accounts in the literature (e.g. Coates 1983: 180, Palmer 1990: 57, Leech 1987: 68, and Wekker 1976: 75) which attest to the future progressive (WILL + be + V-ing) as a means of disambiguating the deontic from the epistemic senses of this modal.4 (The perfect aspect is also observed to convey unambiguous epistemic senses — see Wekker 1976: 75; however, this will not be discussed since no examples using the perfect aspect were included for investigation in the present study.) That there might be a predominantly epistemic use of WILL when found in combination with the progressive seems highly likely, given that the progressive aspect is often considered to have a stativising effect on an action or event verb (cf. Vlach 1981: 274) and Frawley 1992: 314), and deontic modality has been regarded as incompatible with stativity (see above). It should also be noted that during the historical development of epistemic meaning from deontic meaning in modals, the progressive seems to have possibly served a disambiguating function when deontic and epistemic modality were being used contemporaneously, as the following example shows (c. 1740): Something must be hatching, I doubt! (Denison 1993: 430). The appearance of unambiguously future meaning with the progressive may then be predictive of a clear direction of the grammaticalisation of WILL, indicating the areas in which the processes of grammaticalisation are at a more advanced stage, beyond that at which old and new meanings still co-exist and produce ambiguities. The use of WILL as a future auxiliary would affect a limited number of grammatical environments in the beginning, then increase in scope over a period of time in which both old and new meanings may be said to co-exist, before the new meanings spread to include all environments in which the item may be used. This pattern of gradual diachronic semantic change has
4. For a counter-argument see Huddleston (1995), who gives the example: Will you please be all sitting quietly at your desks when the headmaster comes in at ten o’clock? However, without the use of a marker of politeness, please, and appropriate intonation (if spoken), Huddleston’s example might appear ambiguous between a somewhat more impolite request for action and a request for information. Furthermore, the use of please in a 2nd person interrogative seems almost nonsensical with other types of verbs: Will you please be coming to my party? Huddleston’s example appears as a request for a future state to be undertaken by the subject, and this restricts the type of main verb occurring in the construction to one which expresses an on-going ‘process’.
144
HYPOTHETICAL MODALITY
been described in Hopper and Traugott (1993: 36–7). Therefore if WILL is more grammaticalised as a future auxiliary with the progressive than in other environments, its use with this function should be categorical and acceptable in all contexts, including hypothetical ones. It is for this reason that the progressive was used in the questionnaire below (Question 2). 6.1.1.4 Subject sensitivity 6.1.1.4.1 First person subjects. It is widely agreed in descriptions of British English (see, e.g., Palmer 1990; Coates 1983; and Wekker 1976) that will cooccurring with a first person subject is most often found to have volitional rather than predictive meaning, or at least to be ambiguous between the two meanings. Coates observes this co-occurrence in 76% of all cases having a first person subject (1983: 176), and Wekker found this pattern in 78% of analysed cases (1976: 26). If the reasons for such patterns can be given as a ‘default’ explanation, then perhaps it could be suggested that WILL can be interpreted as a pure futurate modal only when there are no constraints on its use as such. That is, the maximum restriction would be likely to occur when the subject is in the first person, because the speaker and the subject then share the same identity and the speaker as the experiencer is likely to have more intimate knowledge of his or her own intentions than of those of another person. Furthermore, as discussed earlier, one is hardly likely to want to make predictions about one’s own actions, so the need for extension to such functions does not exist. When the subject is in the second or third person, WILL may sometimes be interpreted ambiguously, e.g. He won’t help us (Huddleston 1995: 440), since the speaker is less likely to have explicit knowledge of the subject’s volition or otherwise, and therefore the restrictions on the use of WILL with a maximally futurate interpretation are lifted. It might be assumed that shall has become available in some dialects to express pure futurity and that this is one of the reasons for the frequent interpretations of I/we WILL as having volitional meaning (see, e.g., Heine et al. 1991: 171, note.). However, many dialects do not regularly use shall for first persons, e.g. it is rare in North American English (Trudgill and Hannah 1982: 49) and not frequent in Australian or New Zealand English (ibid.: 19); Scottish English does not have any occurrence of shall at all (Perkins 1983: 47). Moreover, it has also been observed that WILL regularly occurs with a first person subject for pure future uses, even in dialects which do have a first person shall (Palmer 1990: 137). If shall and should do happen to serve as markers of unambiguously temporal meaning associated with first person subjects, then the use of these modals is certainly confined to a limited dialectal range, and is not
HYPOTHETICAL WILL
145
categorical within this range. The ambiguity between volitional and predictive senses associated with interpretations of I/we WILL still remains unresolved in many cases, and it is for this reason that the use of first person subjects in such constructions was not expected to force a volitional interpretation on the examples used in the questionnaire. 6.1.1.4.2 Inanimate subjects. Thomson and Martinet (1988) note the requirement that would in the complement clause of wish usually demands a personal or animate subject able to effect some kind of control over the actions expressed in the predicate. Bybee (1985: 168) associates animate subjects mainly with deontic modality, and with clauses involving agentive notions such as volition, ability, obligation or permission. Traugott (cited in Denison 1993: 299) considers that the presence of an inanimate subject with will is the only case in which it is inferred that the sense of the modal is primarily predictive, and that there appear to have been no occurrences of inanimate subjects in Old English, when the use of will was most likely to have been deontic. Therefore, the appearance of inanimate subjects in association with the modal WILL must have corresponded historically with the emergence of epistemicity in relation to the meaning of WILL. What this means for examples such as I wish it would stop raining (Thomson and Martinet 1988: 262) and other instances in which the subject of WILL is simply a formal grammatical device is that the only meaning possible for WILL with inanimate subjects in hypothetical clauses is epistemic or predictive, not a deontic meaning which would imply some kind of agentivity on the part of the subject, agentivity which cannot be said to apply to the dummy subject it. The same reasoning should apply then to the examples given in the Singaporean data above, in which appears the existential subject there, e.g. I wish there would be peace. Such examples ought to be interpretable by most speakers in terms of a purely futurative or epistemic meaning, since the subject there cannot be said to have control over the ‘action’ specified in the predicate. However, the examples still appear to be disputable; so perhaps the only basis for their being unacceptable must be the use of a stative verb be, over which no amount of agentive control can be applied (see 1.2). For this reason, the use of an existential subject was tested in the questionnaire (Q. 5) for comparison with other examples using stative verbs (Q. 1, 6, and 7.). If dummy subjects are possible, then, where do inanimate objects as subjects stand with regard to the modal WILL? Leech (1987: 121 note.), in discussing this question, claims that the difference between I wish this clock worked and I wish this clock would work is one of time, not modality, the former being a present wish and the latter a wish about the future. The former example may be expressive
146
HYPOTHETICAL MODALITY
of a continuous state of disrepair, and the use of the verb worked is therefore in a habitual sense. However, it must first be noted that this example contains an event verb, and there is no reason to suggest that a concrete inanimate object as subject cannot metaphorically have a will of its own (assuming that the verb it governs is not stative) as has been discussed by Coates (1983: 173). The negative presupposition to Leech’s example (This clock won’t work) may therefore be equally interpretable as an expression of the clock’s ‘refusal’ to work, thus questioning the necessity for the subject of the complement clause of wish to be always animate, and permitting a volitional reading of the modal. If Leech’s examples above show a distinction of time, rather than of modality, the use of a stative main verb should not preclude a futurate interpretation in similar examples. This was tested in Q. 6. 6.1.1.4.3 Co-referential subjects in both clauses. Thomson and Martinet (1988: 262) also find that the subject of the matrix verb wish cannot be the same as the subject of the verb in the dependent clause. The reasons for this restriction are obvious and reveal perhaps most clearly the fundamental argument underlying the present study. The utterance of a wish often implies the speaker’s present knowledge of facts to the contrary of those expressed in the predicate clause, and if the proposition in the predicate is futurate in meaning, no real facts can strictly be determined relating to it, although strong assumptions can be. The sense of counterfactuality usually implied by the utterance of a verb of wishing might appear to conflict with the sense of non-factuality often related to futurate expressions (see, e.g. Lyons 1977: 817). The reason for this incongruity to be so apparent in wishes in which the subject of the verb wish is the same as the subject of the dependent clause is because the ‘wisher’ is most likely to have the most knowledge regarding his or her ability to control the action expressed in the dependent clause, leaving no doubt as to the factuality of the meaning implied in that clause. Similar reasons seem to underly the restrictions on the future meaning of WILL when co-occurring with first person subjects in the hypothetical clause (see 6.1.1.4.1). Thus, the example: I wish I would earn lots of money … appearing in the Singaporean data (see (1) above), seems to invite the question: ‘Well, why don’t you?’ and a question such as this asked by the addressee would indicate his or her assumption that the speaker has some kind of control over whether or not she can earn lots of money. If, however, the modal WILL in the dependent clause has only a futurate interpretation with no residual volitional meaning, as is will in the non-factual utterance: I hope I will earn lots of money, then there seems no reason for disallowing constructions in which the subject of the main clause is identical to
HYPOTHETICAL WILL
147
that of the dependent clause. Evidence from the present investigation indicating rejection of the use of wish + would constructions in which the subject of the matrix clause and the subject of the dependent clause are identical would further support the hypothesis that the modal in the dependent clause is interpreted as deontic in nature rather than predictive or epistemic. This will be tested in Q. 3. 6.1.1.5 Negation The negation of will/would often renders the proposition in which they are contained open to a potentially volitional interpretation, resulting in a partial loss of future meaning. This has been previously observed by Wekker (1976), Ultan (1978), Palmer (1990), and Huddleston (1995), amongst others, but few suggestions have been offered as to the reasons for this. Perhaps one explanation could be provided by Horn (1989: 311), who maintains that the cognitive processing of negative concepts tends to be more complex than is required for affirmative structures; hence the diachronic development of such effects as negative raising, e.g. I think it’s not true > I don’t think it’s true, which evolved for the purpose of creating clarity and emphasis by fronting the negative element. The operation which created negative raising might then explain the leftward-extending realisation of negative scope with certain modals, especially if they are contracted, resulting in restrictively deontic or agent-focussed interpretations. Furthermore, if, according to Bybee (1985: 159), the historical shift from deontic to epistemic meanings was accompanied by a parallel shift in the scope of the modality to include the entire clause, then inevitably the modality would have to include its negation as part of the scope. The acceptability of mustn’t as epistemic found in educated northern English dialects and some Australian dialects (Newbrook 1992: 4) could then imply that the grammaticalisation of this modal might be at a more advanced stage in those dialects than it is in the southern English dialects, where the form is acceptable only with deontic meaning. This is a point worthy of futher investigation. The resistance of some epistemic modals to negation in some dialects might alternatively be explained from a logical viewpoint. It has been suggested that negative epistemicity is not a contrary of affirmative epistemicity, but an implicature of it (Huddleston 1995: 436). That is, to consider that a proposition might be true automatically implies its negative: that it might not be true; the contrary of possibility, though, is impossibility, which entails total falsity. This matter is also noted by Palmer (1990: 9) who finds logical equivalences between possibility and necessity concerning negation — that is, ‘not possible’ (cannot) is logically equivalent to ‘necessary not’ (mustn’t) — these equations thus provide the basis for the alternation between epistemic mustn’t and epistemic
148
HYPOTHETICAL MODALITY
can’t across some dialects. By all appearances, the development of epistemic modality had its logical repercussions, certainly in negative environments, and it is not surprising, therefore, that the expression of negative future WILL, especially in its contracted form, is frequently given a volitional interpretation more reflective of its lexical origins. This will be tested in Q.4. As discussed above, the occurrence of WILL in present-day hypothetical predicates is apparently incompatible with stative verbs and with some inanimate subjects. Negative constructions are domains in which the occurrence of WILL is likely to neutralise the interpretation of the modal as being purely futurate in nature, whereas the interpretation of WILL co-occurring with progressive aspect is generally considered to be unambiguously future. These factors were taken into consideration in the development of this investigation, which will endeavour to give evidence to support the hypothesis proposed earlier (6.1.1) that the constraints on the present-day use of future WILL need to be explained by taking into account the lexical origins of the modal and the degree to which its older meanings continue to impose a semantic constraint on its progress as a grammaticalising tense auxiliary. In order to test the hypothesis, the intuitions of native speakers were assessed in a questionnaire using WILL in hypothetical predicates; this can be found in (6.1.2.1). 6.1.2
Methodology
A questionnaire consisting of 7 sentences, all containing a main clause predicator, wish, and a hypothetical complement clause containing the construction would + V, was offered to a sample of 33 participants, who were to evaluate the sentences on an intuitive basis for acceptability.5 The anonymous subjects were asked to provide information on gender and date and place of birth. Four gradings were to be used in the responses: acceptable (A), questionable (Q), very questionable (VQ) or unacceptable (U). The informants were then asked to suggest possible changes to the sentences they had marked U or VQ. They were given very little information as to the aims of the exercise prior to attempting it, in order that a spontaneous response might be produced which might be more reflective of their immediate reactions than of carefully contemplated decisions. The research was conducted in the area of the south-eastern suburbs of Melbourne,
5. This questionnaire is extracted from a slightly longer questionnaire, containing an additional 3 sentences; it was felt, though, that the additional sentences were not strictly pertinent to the main direction of the argument presented in the current investigation, and so they were not analysed in this study.
HYPOTHETICAL WILL
149
Australia, and the participants were haphazardly selected from the campus of Monash University, Melbourne, or from amongst colleagues and acquaintances known to the researcher. All were native speakers of English, and almost all were known to be educated to post-secondary level. Twenty-two of the participants were Australian-born, 5 were of Irish origin, 2 were English, one was Scottish-born, and there was one South African and one participant born in Papua New Guinea. The group also included one participant of Slovenian ethnicity, who had been educated since childhood in Australia. The representation of nationalities amongst the participants was not intentional, but it was not considered prior to the investigation that the particular feature being investigated would be likely to be significantly affected by marked dialectal tendencies for the speakers being questioned, since the informants, unlike those of the previous study, were speakers of a first-language variety of (predominately Australian) English, and the results did not appear to convey any distinct bias with regard to their dialect origins either. Thus, by means of a questionnaire, the intuitions of mostly Australian speakers of English have been used to assess the acceptability of forms elicited from the previous survey of Singaporean speakers of a second-language variety of English. The questionnaire is presented below (6.1.2.1 — see also Appendix 1 for the full form). The results of the survey are provided in Tables 6.1–6.3. 6.1.2.1 The Questionnaire 1) 2) 3) 4) 5) 6) 7)
I wish you would be tall. I wish you would be coming to my party. I wish I would come to your party. I wish Pat wouldn’t come to the party. I wish there would be no fighting at the party. I wish the Porsche would belong to me. She wishes her parents would believe her about the party.
6.1.3
Results
Number of subjects: 20 F + 13 M = 33 Age range: 18–59
150
HYPOTHETICAL MODALITY
Table 6.1
Graded assessments of the main questionnaire by male and female Australian respondents, showing the acceptability of the distribution of WILL in hypothetical predicates by environment
Question
U
1. 2. 3. 4. 5. 6. 7.
13 17 16 00 04 10 00
Average probabilities
= = = = = = =
39.4% 51.5% 48.5% 0% 12.1% 30.3% 0%
VQ 8 7 9 1 5 3 1
0.26
= = = = = = =
24.2% 21.2% 27.2% 3% 15.1% 9% 3%
Q 09 07 08 09 16 13 04
0.14
= = = = = = =
27.2% 21.2% 24.2% 27% 48.5% 39.4% 12%
A 03 02 00 23 08 07 28
0.29
= = = = = = =
9% 6% 0% 70% 24.2% 21.2% 85%
0.31
(N = 33)
Table 6.2
Graded assessments of the main questionnaire by male respondents only
Question 1. 2. 3. 4. 5. 6. 7. Average probabilities (N = 13)
U 3 3 5 0 0 1 0
= = = = = = =
23% 23% 38% 0% 0% 7.7% 0%
0.13
VQ 3 4 4 1 1 2 1
= = = = = = =
23% 31% 31% 7.7% 7.7% 15.3% 7.7%
0.18
Q 5 4 4 3 9 7 3
= = = = = = =
38.4% 31% 31% 23% 69.2% 54% 23%
0.39
A 2 2 0 9 3 3 9
= = = = = = =
15.4% 15% 0% 69.2% 23% 23% 69.2%
0.31
151
HYPOTHETICAL WILL
Table 6.3
Graded assessments of the main questionnaire by female subjects only
Question
U
1. 2. 3. 4. 5. 6. 7.
10 14 11 00 04 09 00
Average probabilities
= 50% = 70% = 55% = 0% = 20% = 45% = 0%
VQ 5 3 5 0 4 1 0
0.34
= = = = = = =
25% 15% 25% 0% 20% 5% 0%
Q 4 3 4 6 7 6 1
0.13
= = = = = = =
20% 15% 20% 30% 35% 30% 5%
A 01 00 00 14 05 04 19
0.22
= = = = = = =
5% 0% 0% 70% 25% 20% 95%
0.31
(N = 20)
6.1.3.1 Results of Control Questionnaire Number of subjects: 10 M + 14 F = 24 Age range: 18–60 Country of origin: Australia (19); Ireland (2); South Africa (1); Papua and New Guinea (1); Slovenia (1 — educated from childhood in Australia) Table 6.4
Graded assessment of the example sentences in the control questionnaire (see Appendix) by both male and female respondents
Question 1 2 3 4 5 6 Average probabilities (N = 24)
U 5 3 1 3 1 3
= = = = = =
20.8% 12.5% 4.2% 12.5% 4.2% 12.5%
0.11
VQ 4 2 1 0 2 2
= = = = = =
16.7% 8.3% 4.2% 0% 8.3% 8.3%
0.08
Q 9 6 1 4 2 9
= = = = = =
37.5% 25% 4.2% 16.6% 8.3% 37.5%
0.21
A 06 13 21 17 19 10
= = = = = =
25% 54.2% 87.5% 71% 79.2% 41.7%
0.51
152
HYPOTHETICAL MODALITY
Table 6.5
Graded assessment of the control questionnaire by male respondents only
Question
U
1 2 3 4 5 6
3 1 1 3 1 1
Average probablities
= = = = = =
30% 10% 10% 30% 10% 10%
VQ 1 1 1 0 1 1
0.17
= = = = = =
10% 10% 10% 0% 10% 10%
Q 4 3 0 1 0 4
0.08
= = = = = =
40% 30% 0% 10% 0% 40%
A 2 5 8 6 8 4
= = = = = =
0.20
20% 50% 80% 60% 80% 40%
0.55
(N = 10)
Table 6.6
Graded assessment of the control questionnaire by female respondents only
Question
U
1 2 3 4 5 6
2 2 0 0 0 2
Average probabilities
= 14.3% = 14.3% = 0% = 0% = 0% =14.3%
VQ 3 1 0 0 1 1
0.07
= = = = = =
21.4% 7.1% 0% 0% 7.1% 7.1%
0.07
Q 5 3 1 3 2 5
= = = = = =
35.7% 21.4% 7.1% 21.4% 14.3% 35.7%
0.23
A 04 08 13 11 11 06
= = = = = =
28.6% 57.1% 93% 78.6% 78.6% 42.8%
0.63
(N = 14)
6.1.4
Discussion
In the following report, each question will be presented individually and the justification for its inclusion in the questionnaire will be given, before a summary of responses is analysed. Detailed quantification of specific questions in the results can be referred to in the tables (6.1.3). Question 1: I wish you would be tall. As discussed above (6.1.1.2), the co-occurrence of a stative verb with the modal WILL in clauses of future hypotheticality was considered unacceptable, the reasons postulated for this incongruity being that in such clauses, it was usual
HYPOTHETICAL WILL
153
that the use of WILL would necessitate some kind of agentive control on the part of the subject over the proposition expressed in the predicate (see 6.1.1.1). The inability of the subject in this construction to apply agentive control over the state of ‘being tall’ makes this example an unlikely candidate for acceptability, on those grounds. However, if, on the other hand, the use of WILL in this clause suggests that the wish is a wish for the future, and the modal has been grammaticalised in some dialects to include temporal grammatical marking in this environment as one of its functions, then there is no reason why the sentence should be rejected. Acceptance of this sentence would presumably indicate an interpretation of it as a future wish which is contrary to the speaker’s assumptions; that is, the addressee is unlikely to be very tall, but the speaker desires this nonetheless. The results show that only 9% of the entire group found Question 1 acceptable and almost 64% found it either unacceptable or very questionable. Of the males questioned, only 46% found the sentence unacceptable or very questionable but it was rejected more strongly by the females, of whom 75% found it unacceptable or very questionable, with only 5% accepting the sentence. Of the 19 respondents who offered alternative suggestions, 15 changed the construction to I wish you were tall, 1 changed it to I wish you were taller, 2 changed would to could, and one informant commented: “I can’t imagine anyone saying this … mainly because people can’t make themselves tall so it’s unlikely anyone would say it.” This comment reveals the sense of volition intuitively associated with the use of WILL in this context, a sense which is not always transparent when the main verb is an event or action verb, but which becomes evident as an underlying semantic constraint when WILL co-occurs with stative verbs.6 The possibility of a hypothetical future reading for the use of WILL in this context, in which the scope of the modality would extend over the entire clause, is limited by such intuitions of subject-orientation. Question 2: I wish you would be coming to my party. As discussed above, the use of the progressive aspect in the future has been frequently agreed upon as a disambiguator in the direction of a pure future
6. It could be argued that, for second person subjects, the volition sense is often transparent with an event verb, e.g. I wish you would close the window, when the wish functions rather as an indirect request than as a desire which is unlikely to be realised, as noted by Thomson and Martinet (1988: 262). However, this does not affect the fact that the unacceptabilty of Question 1 is most likely due to the use of a stative verb in the predicate, whether or not statives can be used in making requests of this kind.
154
HYPOTHETICAL MODALITY
meaning when there is any possibility of a volitional sense pervading the meaning of the modal. Assuming this to be a determining factor of futurity, a hypothetical future progressive has been included in the questionnaire to ascertain whether, as suggested by Palmer (1986) and Leech (1987), the proposition contained in the hypothetical clause is genuinely futurate in meaning, in which case the construction using the hypothetical future progressive (would be + V-ing) would be acceptable to most speakers. The use of other factors likely to inhibit a purely futurate interpretation (e.g. stative verbs, first person subjects and negatives) and features associated with imposing a volitive sense on the modal has been avoided. A rejection of this form in the hypothetical clause might indicate that, for some speakers in the present study, the meaning of WILL might be not so readily understood as primarily futurate (in a preterite form) as more reflective of its origins as a verb having volitional meaning. Such assumptions were indeed well-supported. The data show that a total of 72.5% of participants rejected Question 2 as either unacceptable or very questionable, 21% found it questionable and only a meagre 6%, representing 2 informants, found it acceptable. Within the female group, 85% either rejected it as U or VQ, while 54% of males responded in this way. No female respondent accepted the sentence. Amongst those who suggested changes to its form, 13 respondents changed it to I wish you were coming to my party, 6 respondents replaced would be coming with would come and 3 respondents changed it to could come. The selection of alternatives for this particular question is interesting in itself — there seems to be a clear preference for hypothetical past progressive forms over the use of a modal, even when a modal is acceptable in non-progressive form, e.g. would come or could come. The same preferences are seen in Question 1, with 15 out of 18 respondents changing the form to a hypothetical past tense, rather than using another possible modal. One might suggest that because counterfactuality is more readily associated with the past or the present (truth values being more easily determined in the past or present than in the future), past tense forms (marking remoteness of mood) lend themselves more naturally to expressing hypothetical propositions in the future, especially given the availability of present tense forms to express futurity in realis contexts. Question 3: I wish I would come to your party. This question was designed to test the participants’ reactions to the use of equivalent subjects in the dependent and matrix clauses, one of the constraints on the use of would referred to by Thomson and Martinet (1988). The use of an identical subject in a complement clause of wish is quite legitimate when the temporal reference is past (I wish I had come to your party) or when a hypo-
HYPOTHETICAL WILL
155
thetical past tense form is used to express a future wish (I wish I were coming to your party). The constraints against using identical subjects in each clause seem therefore not related to the notion of hypotheticality per se, nor to reasons which could be found to be basically syntactic. Furthermore, the use of a first person subject in Question 3 cannot be said to categorically exclude a purely predictive interpretation — the possibility of such a constraint is discussed in 6.1.1.4.1. Of all the sentences, this was the most strongly rejected by the participants. None of the group accepted it, and almost 76% found it either unacceptable or very questionable. 80% of females and 69% of males responded with U or VQ, showing this question to be the least preferred amongst males. However, of the 18 respondents who offered an alternative form, 13 corrected would to could in this sentence, preferring another form of modality by supplying a form which could in this context be interpreted as agent-oriented, showing (remote) ability (though its meaning could be ambiguous between ability and possibility). If could is regarded as dynamic, or denoting ability, then in replacing would with could these informants have demonstrated that the salience of showing agentive senses on the part of the subject of the clause overrides the need to retain the lexical meaning of the modal replaced. If the use of could, on the other hand, is regarded by those informants as implying possibility, then perhaps this indicates that the epistemic meaning of this modal has become more generalised than has that of WILL. There were also five corrections made to Question 3 which used a form other than a modal (one respondent supplied could as well as were coming, offering two alternatives); 3 altered the form to I wish I was coming and 2 to I wish I were coming, these latter using the more formal, subjunctive form of the remote be. The substitution by 5 respondents of a form having the same subject as that of the matrix clause indicates that it is not the use of an identical subject which is seen as inappropriate in the hypothetical clause, but the co-occurrence with would in this type of construction. The fact that this question was rejected the most strongly of all the questions might indicate that the influence of volitional meanings prevails over futurate meanings in such constructions, making them appear almost contradictory in nature (wishing against one’s own volition). Question 4: I wish Pat wouldn’t come to the party. This question was the only example of a negative form used in the questionnaire. As discussed above (see 6.1.1.5), negative constructions using WILL + V are more likely to have a volitional sense than a predictive meaning, although the latter meaning is possible. If this is the case when the modal is used in realis contexts, then the back-shifted form of the modal used in hypothetical contexts
156
HYPOTHETICAL MODALITY
may be open to similar interpretations of a volitional nature. However, this particular structure was accepted by as many as 70% of the participants, and none of the group rejected it. At the same time, only 3% (all males) found it very questionable and 27% found it questionable, represented by 30% of females and 23% of males. The differences between the statistics of male and female responses were not significant for this question. Only one respondent corrected this sentence, altering it to I wish Pat wasn’t coming to the party. The results for this question, therefore, seem to contradict assumptions discussed above in the literature that volitional meaning is very frequently found in association with negative WILL. The reasons for so many acceptances of this question appear somewhat difficult to determine; if would is acceptable in the volitional sense, then there is a suggestion that coming to the party is something done by Pat habitually — there is thus a counter-premise that Pat ‘insists’ on coming to a party which is regularly held (Pat will come to the party — but I wish she wouldn’t!). The use of WILL in this instance has the sense of typical behaviour (Palmer 1990: 136) and is therefore subject-oriented in nature. However, this would be unlikely, since the sentence has no adverbial qualification suggesting that ‘the party’ is a regular event. Thus, without any given contextual information, the sentence must be acceptable only on the basis that it is a wish for the future, and that wouldn’t is used in a predictive sense in this construction. The one participant who supplied another verb form by changing the modal to wasn’t coming, further suggested, with a hypothetical past progressive form, that the sentence was understood to be futurate in meaning. The acceptance of Question 4 as having future meaning by most of the present informants suggests that, for these speakers, the scope of the negation in such contexts must be seen to extend over the entire clause, in accordance with an epistemic interpretation of the modal, and not to be restricted to the scope of the modality alone, as would be the case if wouldn’t in Q. 4 were subject-oriented or volitional. An epistemic treatment of wouldn’t such as this indicates allows for a significant comparison with data from studies of the form mustn’t which is found to be commonly accepted as epistemic by some Australian speakers (see Newbrook 1992: 4), as discussed above (6.1.1.5). It also suggests that wouldn’t is understood as lexically analogous to a modal verb expressing necessity. Comparisons such as these could provide the basis for further, more penetrating investigations using other forms of modals and a range of different environments.
HYPOTHETICAL WILL
157
Question 5: I wish there would be no fighting at the party. This sentence is modelled on some of the apparently questionable examples taken from the Singaporean data (see (3)–(5) in the Introduction), in which an existential construction appears. The use of a dummy subject (there) in the hypothetical complement clause is discussed above (6.1.1.4.2), and it is concluded that if such constructions are acceptable, there should be a strict futurate meaning in the clause containing the dummy subject, since other types of dummy subjects are acceptable in this context (e.g. it in I wish it would stop raining) and the modal of which they are subject is interpreted as unambiguously epistemic. Any resistance to the tolerance of this construction would have to be as a result of the presence of the stative verb (be). Only 12% of all participants rejected this example, although at the same time only 24% of the group accepted it. The strongest tendency was found in the number of respondents who judged this sentence as questionable (48.5%); perhaps the reasons for rejecting it outright were not entirely perspicuous to many intuitions, since the question of the incompatability of a stative verb with an agentive subject may not arise, not only because the subject is inanimate and incapable of being an agent, but also because it is nothing more than a grammatical device. Amongst the female group, responses were quite evenly distributed: 20% rejected it, 20% found it very questionable, 35% questioned it and 25% found the sentence acceptable. Amongst the male informants, 69.2% found the sentence questionable, none rejected it, and 23% accepted it. The male informants were considerably less decisive in judging this sentence, as is shown by the fact that almost twice as many males as females judged it questionable. Some of the alternative suggestions offered for this construction included changing the ‘wish’ to a ‘hope’ — i.e. I hope there will be no fighting at the party, supplied by 3 out of the 8 informants who offered alternatives. The use of a non-factual alternative in the substitution of the verb hope was not offered for any of the other questions, and seems to suggest that some speakers might wish to avoid the implications of counterfactuality commonly associated with past or present wishes. Included in other suggestions for alternative forms were one correction adding that to the hypothetical clause (optional in any case) and another who changed would be no fighting to wouldn’t be any fighting, this preference giving further evidence to support the assumptions made above that hypothetical wouldn’t might be acceptable to some speakers as an epistemic negative modal. Another respondent supplied I wish there to be no fighting at the party and only one changed the time reference: I wish there were no fighting at the party. One informant suggested I wish there is no fighting at the party
158
HYPOTHETICAL MODALITY
(supplied by an Australian-born speaker), substituting a verb form unmarked by back-shifting for factual remoteness. The problems of using would in the hypothetical clause shown in Question 5 are reflected in the attempts shown by the respondents to provide a suitable alternative form. The fact that one respondent resorted to using a non-finite clause complement (I wish there to be no fighting at the party) is an indication of the difficulty of using a hypothetical future existential construction: this is the only example of a non-finite substitute appearing in any of the responses. Epistemic could is also avoided here. The use of a form with present time reference appears somehow paradoxical, since the adverbial phrase, at the party, denotes a point of reference either spatially or temporally (or both) removed from the moment of utterance, allowing for a future meaning to be easily extrapolated. Nearly half the responses to this question judged it Q, showing clearly the lack of confidence the informants might feel about using such a construction, and also suggesting that, for the present group, the acceptability of future hypothetical WILL in these contexts is far from categorical. Question 6: I wish the Porsche would belong to me. This question provides another example of a stative verb, but of a different type from that in Question 1 and in Question 5. As in Question 5, there is an inanimate subject, but this time the inanimate subject is a concrete object. Inanimate subjects have been considered as possible agents in clauses containing WILL + V (see the discussion in 6.1.1.4.2), therefore they can be considered as having a metaphorical volitional sense when co-occurring with event verbs in hypothetical predicates. The sentence in Question 6, however, was similar to an example used by Leech (1987: 121, fn.) to show that stative verbs cannot appear with WILL in such environments, and it is this factor, combined with the presence of an inanimate subject, that is being tested for acceptability in this question. Surprisingly, it was accepted by as many as 21% of the respondents, and 23% of males and 20% of females accepted it. 30% of participants rejected it outright, and this figure was represented by 7.7% of the male group and 45% of the females. The highest proportion of responses amongst the males were from those who found it questionable (54%). There was little difference overall between Question 6 and Question 5 in the number of responses accepting it (24% for Question 5 and 21% for Question 6); however, there were more than twice as many rejections for both groups for Question 6 (30%) as for Question 5 (12%). These figures, when compared with the figures for the rejections of Question 1 (39.4%), seem to suggest that when would co-occurs with a stative verb in hypothetical predicates, inanimate subjects are preferred over animate
HYPOTHETICAL WILL
159
ones, and, furthermore, dummy subjects are preferred over inanimate objects as subjects. This might provide convincing evidence (for the present study) for the increase of epistemic meaning in association with such contexts: an inanimate object as subject or a dummy subject could be regarded as less likely to be volitional (as suggested above) than an animate one, and there would therefore be no associations with lexical source meanings to preclude a futurate meaning on a clause containing one. 12 participants offered corrections to the sentence, 10 altering would belong to belonged, while one informant changed the sentence to I wish the Porsche were mine and only one substituted another modal, could. The fact that only one alternative modal was suggested seems to support our assumptions made earlier that non-modal verb forms are often preferred to modals in hypothetical clauses, where it is possible to use them. The substitution of a hypothetical past form, belonged, indicates the respondent’s preference for a form which allows for present time reference as well as for future (see discussion in 6.1.1.2). There is, nevertheless, a fairly low acceptance of Question 6 (21%), in spite of the fact that older volitional meanings connected with WILL are less likely to compete with newer predictive meanings than in Question 1. Question 7: She wishes her parents would believe her about the party. This sentence uses another stative verb, believe, of the type which has been classified by Leech (1987: 25) as verbs of ‘inert cognition’ or ‘unrestricted present’. Given that these verbs do refer to states, not actions, one would expect that their co-occurrence with would in the hypothetical predicate would be disallowed. However, with some of these verbs it might be possible to regard the subject as having the ability to effect control over (or change if desired) the situation described by the proposition in the predicate clause, and it seems that believe is one of these cases. In this way believe can be seen to behave much like an event verb; it could be said that it is by one’s volition that one believes, believe usually takes a human subject, and therefore the combination of believe with would in Q. 7 might be acceptable. This was indeed found to be true; the large majority of acceptances of the construction in Question 7 (85%) showed that the restrictions on the use of stative verbs in future hypothetical contexts is not categorical, and might only apply to certain ‘core’ statives. It is not possible, therefore, to define stativity alone as an index of the rejection of WILL in hypothetical clauses, since clearly, the use of the modal in Question 7 could rule out that possibility. In order to establish a basis by which the increasing range of acceptable functions is measured in the present data, other factors must come into play. Nevertheless,
160
HYPOTHETICAL MODALITY
the apparent unpredictability of the statistical patterning is evidence that the area under investigation is still semantically and grammatically in a state of development, at least according to the results of the present study. There were no rejections of this question, and only 15% found any degree of questionability in it. However, the male subjects were less likely to accept it than the females (69% of male subjects accepted it as against 95% of females). This shows a considerable divergence in the statistics, given that none of the males or females totally rejected it. It is possible that there might be psychological or cognitive as well as linguistic factors influencing the participants’ judgements, and these might be related more to the content of the questions than to the nature of the questionnaire, but it is well beyond the scope of the present study to investigate such possibilities. Only one informant altered this example, replacing would believe with believed. 6.1.4.1 The control questionnaire The average probability of rejections for females for the main questionnaire was more than double that for males (0.34 as against 0.13 respectively). The average probability of the questions being accepted, on the other hand, was identical in each group (0.31). Given that females are generally considered to be more conscious of linguistic norms (of the standard) than are males (see, e.g. Wardaugh 1992: 322–3 for more detail), the disparity in the figures showing rejections is not surprising. However, since it could not be determined for certain whether the tendency was due to a general fact about gender differences or to the specific nature of the questionnaire, a new questionnaire was devised, consisting of 6 sentences for evaluation in the same way as the first questionnaire. Each of the 6 sentences contained some feature of grammatical variation of the kind often used in colloquial speech, but not always condoned in formal, prescriptive writing. The questionnaire is to be found in Appendix 2. The participants used for this exercise were the same (apart from one male) as those used in the original exercise, but were fewer in number (24), due to the unavailability of some of the former participants. The results of the Control Questionnaire are to be found in Section 6.1.3.1 (Tables 6.4–6.6). The results of this additional questionnaire produced an anti-correlation, with the males scoring a higher average of the probability of rejections this time — 0.17 as against the females’ 0.07. The average number of acceptances was again the same for both males and females. It cannot, therefore, be concluded that the divergences in the figures for rejections were due to a general tendency for either males or females to reject features of doubtful acceptability; perhaps instead they might have been due to the content of the questions, to the nature of
HYPOTHETICAL WILL
161
the questionnaire, or even to the ordering of the options for selection on the Questionnaire (A; Q; VQ; U). Psychological or affective factors such as these could not, of course, be ruled out, but possible explanations such as a profound or marked cognitive proclivity amongst females which is manifest linguistically in the inhibited use of certain modals in particular contexts could not be entertained without considerably more penetrating research of a pyschological nature using different language samples from those used in the present study. Such a notion must surely be difficult to verify, though, and is not within the objectives of the present exercise. 6.1.5
A grammaticalisation chain
The present data indicate that the use of WILL in the object clause of the verb wish is most constrained when combined with the progressive aspect, as indicated by the highest percentage of rejections (51.5%). This may indicate that either the progressive form of the modal cannot be a categorical determiner of epistemic meaning, or more likely, that the meaning of WILL in these contexts is not considered futurate or epistemic by members of the present group. If so, then the failure by speakers in the present study to accept this meaning of the modal could be linked with the possibility that older, volitional senses still constrain its use as a future auxiliary in certain grammatical environments. Further evidence to support this proposition is provided by low acceptability of WILL co-occurring with a subject that is the same as the subject in the matrix clause (Q. 3 — rejected by 48.5% and not accepted by any informant) and by the failure of inanimate subjects (30.3% rejections, in Q. 6), incapable of volition, to be accepted by most speakers in this context. Even less acceptable, according to the present study, is the use of a stative verb co-occurring with a personal or animate subject (in Q. 1 — 39.4% rejections); however, there is variation of acceptability according to the nature of the stative verb, and verbs of certain cognitive states over which agentive control can be exercised are seen to be acceptable in most cases (85%) when co-occurring with WILL in hypothetical clauses. In all of these cases, the future modal form co-occurs quite felicitously with such elements in non-remote contexts. Also frequently acceptable in the questionnaire was the use of the contracted negative wouldn’t (70% accepted it and no informant rejected it outright). Much further research is required in the area of negative epistemic modals, and the results of the present survey indicate that there are possibly significant dialectal trends with regard to the grammaticalisation of negative epistemicity that might provide the basis of substantial investigation. In fact, any future studies involving cross-
162
HYPOTHETICAL MODALITY
dialectal comparisons might prove extremely fruitful, as would additional research into the evidence of the gender-specific tendencies found in the present study. In summary, the average probability for the acceptability of the examples provided is not outstandingly high (0.31), but the combined average probabilities for Q and A answers are significantly higher than the combined average probabilities for U and VQ answers (0.60 as against 0.40). This would suggest that, for the present group of participants, there is a gradual tendency for the use of hypothetical WILL in the complements of wish expressions to be increasing its distributional range to include areas normally considered as containing grammatical and/or semantic constraints on its use, i.e. stativity, aspect, agency of the subject, and negation. While these patterns certainly cannot be said to be representative of patterns of actual use among the participants, the tendencies illustrated here may be useful in providing the basis for future predictions of the grammaticalisation of WILL in the given environments. By considering the increasing percentages of acceptability, from 0% for identical cross-clausal subjects (Q. 3) to 85% for stative verbs expressing subject-oriented control (Q. 7), rather than the proportions of rejections, it is possible to perceive evidence of a process known as a grammaticalisation ‘chain’, such as that described in Heine et al. (1991) and Heine (1992), or a grammaticalisation ‘cline’ (Hopper and Traugott 1993).7 Chains may be said to have a synchronic as well as a diachronic dimension (Heine 1992: 345), the former represented by the presence of a number of stages simultaneously, which are evident in the appearance of variation of a contextdependent nature. The distribution of environments in the present study may therefore predict from a synchronic perspective the future direction of the grammaticalisation of hypothetical WILL according to this data, and the grammaticalisation chain would thus be represented diachronically in an expanding range of grammatical functions for the modal, which undergoes subtle semantic and pragmatic adjustments as its distribution area is increased. The process is demonstrated rather by the increase in the number of grammatical environments for a grammaticalising item (matched with a corresponding decrease in lexical functions) than by transformations in its morphological structure (see also Heine 1994a); in the present study this type of chaining effect is illustrated in Figure 6.1.
7. I am most grateful to an anonymous reviewer of Studies in Language for providing me with this enlightening suggestion.
163
HYPOTHETICAL WILL 100
80
% Acceptability
Acceptability
60
40
20
0 A
B
C
D
E
F
G
Environment Environment A: Would + believe — stative verbs expressing subject-oriented control (Question 7) B: Negatives (Question 4) C: Existential constructions (Question 5) D: Stative verbs — inanimate object subjects (Question 6) E: Stative verbs — animate subjects (Question 1) F: Progressives (Question 2) G: Identical subject in matrix clause (Question 3)
Figure 6.1
Degree of acceptability according to context found in the use of WILL in hypothetical predicates, indicating with a chain-like effect the predicted direction of grammaticalisation on a broader, diachronic scale
The reasons for the intrinsic ordering of environments as shown in Figure 6.1 are far from clear. In some instances, the rate of acceptability may be due to the likelihood that older deontic meanings, as in environment (A), permit
164
HYPOTHETICAL MODALITY
an acceptable reading, showing also the possible presence of competing deontic and epistemic meanings; thus, we are reminded of Palmer’s (acceptable) I wish John would come (see under 6.1) which is also open to a deontic interpretation. In other cases it could reflect the possibility that an environment which normally excludes a primarily deontic interpretation may be grammaticalising at a relatively faster rate, there being no possibility of the retention of volitional meanings to impede its progress; e.g. environment (C), which has a non-agentive, formal subject in the dependent clause is more acceptable than (D) and (E) in environments in which WILL co-occurs with a stative verb. The limitations of the current study preclude reasonable conjecture regarding the index for the acceptability judgements of the informants. A much more extensive survey, involving a larger number of participants and additional example sentences for evaluation would most likely be necessary in order to rank the elements in a natural order consistent with a single rule which might predict their diachronic progress from available evidence of a synchronic nature. What is clear from this investigation, though, is that the considerable variation in the acceptability of WILL in such domains can be related to the likelihood that its grammaticalisation into a modal which includes hypothetical futurity amongst its meanings has not yet extended into all possible environments. 6.1.6
Postscript: Variablity of retention in hypothetical clauses
The study presented in this chapter does not presuppose that retention is possible in all environments in which the modal would is used. The fact that there is retention of volitional source meanings in hypothetical complement clauses is most likely a factor of a) the function of the combining clause; b) the nature of a subordinate clause which does not express assertions; and c) (in)frequency of use. With regard to (b), it could be suggested that grammaticalisation will be slower in subordinate clauses than in main clauses; causing lexical retention to be more visible (though the data in Ch. 3 does not necessarily bear this out). James (1986: 103), however, notes that in subordinate clauses the marking of modality is not as important as it is in main clauses; subordinate clauses are usually inherently modal, and this is why they often require less overt means of marking modality (e.g. the counterfactual complement of a past wish-sentence is still expressed using only the pluperfect in many dialects). On the other hand, would in conditional apodoses is more advanced in its grammaticalisation path, and in only one example in the data examined in Ch. 3 did it ever show traces of retention of volition meanings in such environments (see Ch. 3, note 14). The reason, as stated earlier, is that it is difficult to perceive one’s desires as
HYPOTHETICAL WILL
165
dependent upon a conditional requirement; it is unlikely therefore that there will be an intermediate polysemous stage in the grammaticalisation of conditional would which accommodates both volitional and predictive meanings, as shown for Environment A, Figure 6.1 in the present study, the environment most accepted by the informants in the study. In the complements of verbs of wishing, though, there is no conditional dependence or contingency expressed, so the same environmental constraints on volitional meanings in conditional constructions do not apply, and there is nothing to prevent such senses persisting in the meaning for a longer period of time, unlike the situation in conditional apodoses. Thus, the nature of the clause with which the modal clause combines is important in determining the rate at which would grammaticalises. A third possible reason for the evidence of retention in wish-complements is (c): the possibility of their less frequent occurrence by comparison with conditional apodoses. Bybee (forthc.) maintains that frequency of textual occurrence, as well as being the result of loss of lexical meaning in a grammaticalising item, is also a motivation for further loss of meaning. In the EME texts investigated in Ch. 3, there was a total of 20 appearances of would in conditional apodoses, as against only 5 appearances of the modal in a hypothetical complement of wish or would. This may indicate that, at the time when use in conditional apodoses was becoming more frequent, it was not matched by the same frequency in hypothetical complements, which therefore represented a more slowly grammaticalising domain for the modal. A final possibility is the nature of cross-clause harmony: Bybee et al. (1994: 214) hypothesise that the modal appearing in the complement or dependent clause often displays a certain harmony of meaning with the main verb meaning. Given that the main verb wish expresses meanings of desire and volition, it is not suprising that would appears frequently in the complement of verbs of wishing, and that in former uses (e.g. (10) in Ch. 3) the main verb expressing desire and the complement modal were one and the same form, would. This chapter demonstrates, then, that retention of volitional senses in would may vary according to the grammatical function in which the modal appears, and that even within a functional category, the retention levels may vary first according to co-occurrence conditions. In addition, retention may vary according to the intuitions of the speakers in a community. In the remainder of this chapter, the differences between the retention levels of speakers from various dialectal backgrounds will be investigated (6.2 and 6.3).
166
HYPOTHETICAL MODALITY
6.2 Singaporean English It is recalled that in Ch. 5 the intuitions of a number of Singaporean informants were tested in a questionnaire which examined interpretations of counterfactuality arising from the retention of perfective inferences in past stative and nonpunctual verb types. The retention of perfectivity in such environments was found to be lower for L2 speakers than for L1 speakers, as evidenced by the lower incidence of counterfactual implicatures in the given environments. Ch. 6.1, investigated (primarily Australian) native speaker intuitions on the acceptability of a number of uses of would in wish-complements which were considered marginal in data from a previous study (Ziegeler 1994), and it was revealed that such uses are only variably accepted by such speakers, due to the retention of older, volitional senses in the modal. However, the examples of L2 usage which prompted the investigation in 6.1 were only few in number, and could be passed off as simply random variation. Therefore, in 6.2, the presence of retention in the intuitions of L2 speakers is tested in the grammaticalisation of the modal would in hypothetical complements, and compared with the results from 6.1 which used native speaker informants. The subjects surveyed in 6.2 of the study are speakers of Singaporean English, since the initial hypothesis was based on data collected from such speakers. A similar questionnaire to the one used in 6.1 is used for the study. The present section will summarise the findings. 6.2.1
Methodology
A questionnaire consisting of 7 sentences, similar to that used in 6.1, was offered to a haphazard selection of 33 educated Singaporean respondents, who were asked to evaluate the sentences on an intuitive basis for acceptability. The Questionnaire is to be found in 6.1.2.1, and in full form in Appendix 1. As with the Questionnaire in 6.1, there were 4 gradings of acceptability from which the subjects could choose: Acceptable (A); Questionable (Q); Very Questionable (VQ); or Unacceptable (U). The respondents were then asked to suggest possible alternative forms for the sentences they had marked U or VQ. As with the respondents in 6.1, they were given little information as to the objectives of the Questionnaire, in order to ensure a maximally spontaneous result, and were reminded that it was not a test of any kind, but merely a survey done for the purposes of collecting research data. Any questions about the questionnaire were answered after it was completed, but for the most part, the survey was undertaken in as informal an atmosphere as possible.
HYPOTHETICAL WILL
167
The group consisted of 22 first-year undergraduates, 5 other-year undergraduates, 5 school-teachers from a neighbourhood secondary school and one subject educated beyond tertiary level. The research was conducted mainly on the campus of the National University of Singapore, and at Gan Eng Seng Secondary School, Singapore. All the participants were aged between 19 and 50, and all were born in Singapore or Malaysia except for one informant who was born in mainland China.8 The informants were not asked to name their mother tongue, as those of the study in Ch. 5 had done, as invariably the decision over the correct use of this term is coloured by the effects of current language policy in Singapore (see 5.3.1). This factor is of only limited significance, though, as it must be remembered that the variety of English under investigation is not a learner variety, but an established L2 contact language which bears witness to the institutionalisation of changes and variation from previous generations. The variety, with all its institutionalised features, is now being passed on to succeeding generations by its own speakers, either at home or at elementary school. However, since it has been noted earlier that according to Gupta’s (1992) estimates, only 20–30% of the current generation of incoming schoolchildren have a knowledge of English acquired in the home, it was believed that not many of those surveyed at the time would have had mother tongue English. Thus, many of the speakers surveyed would have been L2 speakers of an L2 variety, with perhaps a small number being mother tongue speakers of an L2 variety. However, since it could be safely assumed that all of the subjects surveyed were educated speakers and would have acquired English from childhood at least through the education system, then the variation exhibited would be expected to be more indicative of dialectal patterns than of learners’ errors. The results of the questionnaire are presented below in the Tables 6.7 and 6.8, and in Figure 6.2. 6.2.2
Results
The following graph illustrates the differences in acceptance ratings between the Australian and the Singaporean informants.
8. Malaysian English is usually considered to be virtually indistinguishable from Singaporean English. Careful examination and comparison of the Chinese subject’s results showed no marked variation from the results of the other informants.
168
HYPOTHETICAL MODALITY
Table 6.7
Graded assessments of the questionnaire by male and female Singaporean respondents, showing the acceptability of the distribution of WILL in hypothetical complements by environment
Question
U
1. 2. 3. 4. 5. 6. 7.
11 08 16 02 06 10 04
VQ
= 33.3% = 24.2% = 48.5% = 6% = 18.1% = 30.3% = 12.1%
4 4 3 0 0 1 0
Average probabilities 0.25
= = = = = = =
12.1% 12.1% 9% 0% 0% 3% 0%
0.05
Q 14 05 09 10 09 10 08
= = = = = = =
A
42.4% 15.1% 27.2% 30.3% 27.2% 30.3% 24.2%
0.28
04 16 05 21 18 12 21
= = = = = = =
12.1% 48.5% 15.1% 63.6% 54.5% 36.3% 63.6%
0.42
N = 33 (19 males and 14 females)
Table 6.8
Comparative listing of the results of the Australian group (AG) with those of the Singaporean group (SG). U
Question 1. 2. 3. 4. 5. 6. 7.
VQ
Q
A
SG
AG
SG
AG
SG
AG
SG
AG
0033.3 0024.2 0048.5 0006 0018.1 0030.3 0012.1
0039.4 0051.5 0048.5 0000 0012.1 0030.3 0000
0012.1 0012.1 0009 0000 0000 0003 0000
0024.2 0021.2 0027.2 0003 0015.1 0009 0003
0042.4 0015.1 0027.2 0030.3 0027.2 0030.3 0024.2
0027.2 0021.2 0024.2 0027 0048.5 0039.4 0012
0012.1 0048.5 0015.1 0063.6 0054.5 0036.3 0063.6
0009 0006 0000 0070 0024.2 0021.2 0085
The scores are listed as percentages.
6.2.3
Discussion
The graph in Figure 6.2, in illustrating the acceptability ratings for both groups, provides a clear indication of the extent to which the environments are tolerated, thus showing the degree of constraint imposed by the influence of retention for each environment. The greatest proportion of differences between the two groups is shown in Environments C and D, and F and G. Clearly, the use of would in hypothetical complements appears to be grammaticalising at a faster rate for
HYPOTHETICAL WILL
169
Acceptability ratings - Singaporean and Australian subjects
% acceptable responses
100 80 60 Australian Singaporean
40 20 0 A
B
C
D E Environment
F
G
Environment A: Would + believe — stative verbs expressing subject-oriented control (Question 7) B: Negatives (Question 4) C: Existential constructions (Question 5) D: Stative verbs — inanimate object subjects (Question 6) E: Stative verbs — animate subjects (Question 1) F: Progressives (Question 2) G: Identical subject in matrix clause (Question 3)
Figure 6.2
Percentage of acceptable responses for both Singaporean and Australian groups, indicating the possible differences in rates of retention of former volitional meanings associated with the modal WOULD in hypothetical complements
Singaporean speakers than for Australian speakers, if indicated by the general tendencies shown of the present study. The greatest margin of difference between the two groups in acceptances (42.5%) came from Question 2: I wish you would be coming to my party, represented on the graph as Environment F. The reasons for such a difference in acceptability judgements for Q. 2 may provide the key to explaining the differences for the other questions; this will be discussed below. The environments that were most accepted by the Australian group, A and B, representing Qs. 7 and 4 respectively (i.e., the would + believe sentence and the negative), were found to be less acceptable by the Singaporean group, indicating, as shown in the graph, a reverse trend for these two environments.
170
HYPOTHETICAL MODALITY
This may reveal that, for some of the Singaporean group, factors other than the presence of retention might have been influencing the choices, or that the influence of retention might have been weaker and in competition with other influences. A closer analysis of the results of each question in turn should determine if this is the case. Q. 1: I wish you would be tall. Q. 1 showed little difference between the two groups in terms of A answers (SG — 12.1% and AG — 9%) or U answers (SG — 33.3% and AG — 39.4%), the two groups apparently agreeing on the more absolute measures of acceptability. The main differences occurred between the VQ and the Q answers, so that a higher number of Australians (24.2%) ranked the sentence as VQ than did Singaporean respondents (only 12.1%), but on the other hand, 42.4% of the Singaporeans ranked it as merely Q whereas only 27.2% of Australians provided this response. The distribution of the middle-range responses in such a way indicates a possible greater tolerance for the use of the modal in such environments by Singaporeans, and hence the likelihood that it is grammaticalising more rapidly amongst Singaporean speakers, whose preferences show a tendency towards the acceptable end of the scale (54.5% Q and A answers, while the Australians showed a total of 63.6% VQ and U answers). Thus, in spite of the similarity between the groups shown at the extreme ends of the scale, the differences in the middle of the scale were considerable and showed an interesting trend. Some of the alterations offered for this question included the following: I wish you were taller, suggested by 4 informants, I wish you were tall suggested by 2 subjects, I wish you are tall (3 informants), and one informant changed the modal to could: I wish you could be taller. The use of a present tense, non-modal form in I wish you are tall is common amongst some speakers of Singaporean English, and has been discussed at length in an earlier study (Ziegeler 1994) — non-modal verbs as well as modal verbs seem to be equally affected by the absence of marking for past tense. One female respondent commented: “— (U) — ‘tall’-ness isn’t within a person control” (sic), indicating, as did a similarly frustrated respondent in the Australian group (see 6.1) that for some members of the Singaporean group semantic constraints related to earlier agent-oriented senses of the modal as a main verb of volition were being activated by the agentive environment (human subject) and were affecting the distribution of the modal in environments where agentivity could not be applied to the main verb (be). However, the greatest proportion of responses came from those who replied that the use was merely questionable (42.4%), indicating that the constraints were
HYPOTHETICAL WILL
171
affecting the Singaporean group’s evaluations only to a limited degree. Q. 2. I wish you would be coming to my party. As discussed in 6.1, the use of the progressive with the modal will in its nonsubjunctive form has been frequently associated with exclusively predictive (hence, epistemic) meanings (see, for example, Coates 1983: 180; Palmer 1990: 57). Part of the reason for this is that the modal combines with a stative auxiliary, be, and, as noted for Q. 1 above, the agentivity derived from a human subject which activates the volitional senses in the modal cannot be extended to embrace stative verbs over which the subject has no volitional control. For this reason the modality is interpretable only as predictive and not as agent-oriented, and thus would include the entire proposition in its scope. If would occurring in a hypothetical complement is still being affected by volitional senses belonging to its former functions, then the progressive construction would be + V-ing should be disallowed, since only epistemic senses can be obtained from modals in progressive constructions. For this reason, the use of the progressive in such environments provides an excellent benchmark indicating the extent of the grammaticalisation of epistemic modality in hypothetical subordinate clauses. The results showed a striking degree of variation between the two groups. In the Singaporean group, the U responses amounted to less than half the proportion of Australian U-response (24.2% as against 51.5%), and the percentages of acceptances by the Singaporean group (48.5%) came to 41.5% more than the Australian group produced (6%). The percentage of VQ and Q responses varied little from each other in either group, showing only a 3% difference in the Singaporean group. This was a firm indication that the evaluations by the informants in both groups were quite decisive. In the Australian group’s results for this question outlined in 6.1, it was concluded that the strong tendency towards rejection of this sentence was due to the likelihood that an unambiguously epistemic modal construction could not be used in an environment which had not yet been affected by the grammaticalisation of the epistemic meanings of the modal; in other words, the functions of would in wish- complements were hypothesised to be fundamentally volitional, and could not permit an epistemic or predictive reading when one could not be implied alongside a volitional one. In I wish John would come, the example given by Palmer (1986) to illustrate his point that would in such contexts had a predictive meaning, a predictive sense is inferred from the volitional senses already present. The fact that the progressive construction was rejected by so many of the Australian group indicated that predictive readings could not be obtained in environments in which volition senses were not present first. The
172
HYPOTHETICAL MODALITY
results of the Australian group, then, indicated that the modal was gradually being generalised to imply purely prediction in hypothetical complements, but that the modal was not yet at the stage where bleaching would allow for such generalisation. The acceptance of the progressive in such environments, then, would be a certain indication that the bleaching of the modal to mark purely predictive meanings would have already taken place in environments in which volitional meanings were formerly possible, and that progressive environments were a later stage of grammaticalisation. This appeared to be the case for a high percentage of Singaporean speakers (48.5%), suggesting that the grammaticalisation of the modal in progressive environments was further advanced than for the Australian group. It is also interesting to examine the alternative suggestions for the rejections. Three informants changed the modal to could, but without retaining the progressive: I wish you could come to my party; and another 3 used the same modal in its non-preterite form: I wish you can come to my party. The absence of marking for the subjunctive of the modal, as discussed above and in Ch. 7, is relatively common in some speakers, the past inflection conveying probably less semantic salience for speakers who generally less frequently mark stative verbs for past (see discussion in Ch. 5). One of the latter three subjects offered both can and could as two alternatives, leaving it up to the reader to choose the option. This is an indication of the semantic equivalence for some Singaporean speakers of the preterite and non-preterite opposites of modal pairs. One subject suggested two alternatives: I wish you could come and I wish you will be coming, the latter being the only alternative suggested which maintained the use of the progressive. Another informant produced I wish you will be coming, and underlined the offensive would to indicate where her disagreements were centred; one informant simply used the non-progressive form: I wish you would come to my party and one changed the wish to a hope: I hope that you would come to my party. Another replied with a U answer but failed to produce an alternative. What was surprising about the alternatives was that only two informants maintained the use of the progressive aspect, while others dispensed with that feature altogether and also changed the lexical form of the modal. Not one Singaporean informant supplied the non-modal construction suggested by 13 out of 22 of the Australian group who supplied alternatives: I wish you were coming to my party. This demonstrated that for the Singaporean group surveyed, it was more important to use a modal in the clause than to maintain the progressive aspect of the main verb. For the Australian group, the hypothetical future meanings were equally possible in the use of a non-modal construction. It may also be noted that in the
HYPOTHETICAL WILL
173
data from the earlier study (Ziegeler 1994), the use of a modal in hypothetical complements was almost categorical for the Singaporean speakers surveyed i.e. there were very few cases of non-modal subjunctives (e.g. was and were) standing alone, in either past or present form. Perhaps a modal auxiliary stands as a general, unanalysed marker of irrealis for some speakers of Singaporean English. If this is the case, the use of a modal means to mark irrealis where formerly the notion had been coded with a subjunctive inflection is further advanced for speakers of an L2 variety than for the native speakers surveyed. Q.3 I wish I would come to your party. Question 3 was the only question which received 0% A answers in the Australian group, although 15% of the Singaporean group did select this option. On the other hand, the percentages of U answers for both groups were the same (48.5%) and the percentage of Q answers were approximately the same at 27.% for the Singaporean group and 24.2% for the Australian group. The greatest difference between the two groups was in the VQ responses — a difference of approximately 18%. This was due to the relatively low numbers of VQ answers from the Singaporean group overall. The option VQ was used quite infrequently by the Singaporean group by comparison with the Australian group, whose proportions of VQ answers in two questions exceeded their U answers. This was never the case with the Singaporean group, who supplied no VQ answers to three of the questions. However, in only two of the sentences, Q. 4 and Q. 7, did the combined values of the Singaporeans’ U and VQ answers exceed those of the Australian group, showing that the Australian group’s rejections were generally higher than those of the Singaporean group, (though varying in the degree to which the respondents were committed to their decisions). The presence of a subject in the would clause identical to that in the wish clause had been claimed by Thomson and Martinet (1988: 262) to be not possible, because of the somewhat contradictory meanings the expression implied — that the subject was wishing for something over which s/he had personal control. This is most lucidly illustrated with the use of a first-person subject, the subject and the speaker being in most cases of the same identity. However, the use of a third person subject could produce similar counter-inferences: ?Cynthiai wishes shei would come to your party when the subjects in either clause are still of identical reference. The reason for the contradictory inferences as stated earlier is that the subject of wish is most likely to have the most intimate knowledge of his or her own intentions or volition, and so if would appears in the subordinate clause with a subject that is co-referential to the subject in the main clause, the construction allows for maximum interpretation as a counter-
174
HYPOTHETICAL MODALITY
factual clause of volition, since notions of counterfactuality presuppose knowledge of facts to the contrary. However, hypothesising about one’s own known intentions seems illogical, and this is why constructions of this nature have been rejected in reference grammars (e.g. Thomson and Martinet 1988). As noted in 6.1, the rejection is not on the grounds that the two subjects are co-referential, as is shown by the alternatives offered by 5 members of the Australian group, for example, I wish I were/was coming to your party. Rather, the rejection of this sentence is due to the use of would in an environment in which the former, agent-oriented senses of volition are maximally transparent. Alternatives offered by the Singaporean group included 14 suggestions of I wish I could come to you party — this was the choice of 13 of the Australian group also; one informant used I wish I can come to your party with no use of the subjunctive modal form; another changed the entire sentence to I hope to come to your party and another, I hope to be able to come to your party, showing in the use of hope a rejection of any sense of counterfactuality implied by the complement of wish (both of the latter alternatives also dispensed with the finite clause). None of the Singaporean group used a non-modal past form expressing factual remoteness as had the 5 Australians mentioned above (I wish I were coming to your party), although subjunctive were does appear to be used by some speakers (see under Q. 6). This seems to indicate a general tendency in both groups for the use of a subjunctive modal form to be gradually ousting the use of a subjunctiveinflected lexical main verb in such environments, especially when the lexical verb is a verb expressing an action which cannot co-occur with the moment of speaking and must necessarily suggest future (or past) temporality.9 Such a situation is indicative of the grammaticalisation of would as a predictive, future modal in hypothetical environments. Question 4. I wish Pat wouldn’t come to the party. The grammaticalisation of epistemic sense from negative modals seems to have dialectal significance, as discussed in 6.1 for the grammaticalisation of epistemic mustn’t. Southern British dialects do not allow an epistemic interpretation of this form, although it is quite frequent in some northern English dialects, and mustn’t has been found to be interpreted epistemically by some Australian speakers (Newbrook 1992: 4), though this has not been shown to be categorical in Australian English. The constraints on the use of such modals as epistemic forms
9. We are reminded of the vestigial uses of subjunctive were in conditional apodoses, as illustrated in Ch. 3 (32).
HYPOTHETICAL WILL
175
may be attributed to the early stages of their development as deontic modals, and agent-oriented senses relating to their lexical origins as main verbs. The reasons that agent-oriented meanings seem to adhere longer in negative forms are not clear, but as discussed in 6.1 there may be a logical explanation for the redundancy of negative epistemic modality, and this may explain why forms such as mayn’t and mightn’t are extremely rare in some dialects (see Horn 1989: 259 for more discussion). Certainly, forms such as won’t are frequently cited as ambiguous between agent-oriented and epistemic meanings. However, for the speakers surveyed in the present exercise, the use of negation produced few problems, either for the Singaporean group or for the Australian group. The percentages for acceptances were quite close (70% for the Australian subjects and 63.6% for the Singaporeans). There was a small number of rejections by 2 members of the Singaporean group (6% U), one of whom offered the alternative: I wish Pat would come to the party, dispensing with the negative altogether. The other participant did not produce an alternative for that question nor for any of the others she had marked U. It would seem possible that wouldn’t is grammaticalising as a predictive modal in such environments for the speakers in the present study, since any context in which it could be assessed as a volitional modal would require the construction of a frame of reference in which the party was a regular event and Pat attended it habitually, i.e. Pat’s coming to the party was an action characteristic of her (as noted in 6.1). There is no contextual clue in the sentence to lead to such an interpretation though, and the fact that the one alternative offered was I wish that Pat wasn’t coming to the party seems to suggest that the party might have been most frequently understood to be an event taking place in the future. Such information could also be inferred from the use of a deictic verb such as come, as noted earlier, in which case, the reinterpretation of would as predictive in Question 4 may be induced by contiguous elements in the clause (see Heine et al. 1991 for more details on context-induced reinterpretation).10
10. The explanation of context-induced reinterpretation, at least for the present study, is considered to be only hearer- or addressee-oriented. Whether or not the evaluations of Question 4, as with all the other questions, are representative of actual production is a problem for further investigation. It seems unlikely that context-induced reinterpretation could be a speaker-oriented change in meaning since this would require explanation as to why the speaker uses the grammaticalising feature in an anomalous context, in which it is then reinterpreted with a new meaning. The presence of high acceptance scores such as for Question 7 would suggest otherwise: that the meaning changes begin from polysemous environments in which both the old and the new meanings are possible.
176
HYPOTHETICAL MODALITY
Question 5. I wish there would be no fighting at the party. This sentence was included as a means of testing the extent to which the predictive uses of would had generalised, since, as in Question 2, the environment precludes any reference to agentivity necessary for the activation of volitional senses in the modal. The subject is merely a formal grammatical device, there, and the main verb used is a stative verb indicating existence. If the construction were acceptable, it would be only on the grounds that it is predictive or epistemic and cannot contain any ambiguity with regard to agent-oriented senses in the modal. The scores for this question showed a high number of acceptances by the Singaporean group (54.5%), more than twice as many as the Australian group showed (24.2%). On the other hand, the highest ranking in the Australian group was for Q answers (48.5%), indicating with a percentage almost as high as the Singaporean group’s A responses that the tendencies for accepting the sentence were moving towards the generalisation of the modal as predictive in such constructions. The Australian group also showed a higher number of combined U and VQ answers than did the Singaporean group (27.2% as against 18.1% respectively). The hesitancy of many of the Australian group to totally accept the sentence was indicated, as shown in 6.1, in the variety of alternatives offered for this question — in fact, this was the only question in which it was felt by some Australian subjects that the hypotheticality implied in the complement clause was not appropriate: an alternative suggested by 3 informants was I hope there will be no fighting at the party. The reason for the rejection of hypotheticality by some of the informants might have been due to the improbability of predicting with assuredness the likelihood of fighting occurring at the party; the use of would as a hypothetical predictive modal would have to imply that the proposition there will be fighting at the party can be uttered with full knowledge of the events taking place. For some speakers, such a prediction may imply too much certainty of the future event about which the speaker is hypothesising, and invite the question: ‘Well, how do you know there will be fighting?’ It may seem that existential constructions occurring with future temporality are incompatible with notions of counterfactuality for many of the Australian group, possibly because the use of would in such environments co-occurs with a subject having no meanings at all associated with any residual lexical traces of agentivity necessary for the activation of volitional senses in the modal; that is, the meanings of volition necessary to provide inferences of prediction are not present and the modal has not yet bleached.
HYPOTHETICAL WILL
177
As with the Australian group, a wide range of options was offered by the Singaporean group as alternatives for unacceptable or very questionable ratings, and each of the options was suggested by only one informant each time, so that there was little agreement in any of the suggestions. One informant suggested I wish there were no fighting at the party, which seemed to indicate that the party was a regular event at which there was always fighting, since the phrase at the party might be interpreted as deictically removed from the place or time of speaking; another informant used I hope there will be no fighting at the party, as did 3 Australian informants; another suggested I wish there would not be any fighting — this was also suggested by one Australian; one more suggestion was I wish there will be no fighting at the party; and another suggested, somewhat amusingly, I wish everyone would be civilised at the party. The variety of alternatives offered by the participants suggests again that there was uncertainty over the use of an appropriate form for such a context. It is interesting to note that no Singaporean informant offered epistemic could as an alternative — this was totally avoided also by the Australian group for this question. Perhaps the difference in meaning between hypothetical prediction and hypothetical possibility was too great to allow for such substitutions; it seems from the results of the present study that could is only suggested as an alternative when there is agentivity implied in the context. Question 6. I wish the Porsche would belong to me. This sentence was used for evaluation on the basis of the fact that it had an inanimate object as subject, that is, a subject incapable of volitional intentions and so incompatible with the use of a agent-oriented modal expressive of volition. A similar expression had been rejected by Leech (1987: 121, fn.) to show that stative verbs cannot combine with would in wish-complements: I wish that book would belong to me; but, as discussed in 6.1, it was certainly not categorical that inanimate objects serving as subjects could not evoke senses of volition when an action verb was the predicate. What might at first appear to be a metaphorical extension of the notions of volition, associated with prototypical agentive subjects, could by all accounts be described as the retention of meanings of agentivity which were diachronically associated with the first uses of the category subject, as discussed in Ch. 7. Thus, the ambiguity between volitional and predictive meaning implied in the following: This key won’t open the door could be related to the retention of lexical properties originally ascribed to the grammatical subject (see also Schlesinger 1989). If the sentence in Q.6 were rejected, then it would be on the grounds that the retention of agentive senses in the subject would be inactive in an environment in which there is a stative verb, belong.
178
HYPOTHETICAL MODALITY
The proportions of rejections for this sentence were equal for both groups at 30.3%, and there was little difference in the percentages of Q and VQ responses either. The greatest amount of difference was in the A answers, the Singaporean group scoring 36.3% and the Australian group 21.2%. The difference was made up in the Australian group by slightly higher percentages in the middle range. Thus, although each group agreed on the extent of rejection of the sentence, there was some difference in the acceptability of such a construction. In the Australian group 10 out of the 12 who corrected the form changed it to I wish the Porsche belonged to me, and the same correction was used by 5 of the Singaporean group, showing that some of the informants found the use of would in such environments inappropriate. Three Singaporean informants changed the sentence to I wish the Porsche was/were mine — this was also used by one member of the Australian group. Another Singaporean informant offered the following alteration: I wish the Porsche belongs to me, and another I wish that the Porsche belongs to me, both using a non-modal form with no past tense to indicate the subjunctive mood. Only one suggested I wish the Porsche could be mine. This question appeared to elicit the highest number of non-modal forms in the alternatives out of all the questions, and one subject, who did not supply an alternative to either Question 1 or 6 despite marking them both U, commented “The Porsche has no free will, is an object” (her emphasis), suggesting most clearly the influence of the retention of volitional meanings apparent to her in interpreting such a construction. (This was the same subject who made a similar comment in response to Q. 1: “ ‘tall-ness isn’t within a person control’”). It would appear that for some of the Singaporean subjects surveyed in the present study, the influence of the retention of volitional meanings is stronger than for others, and in some environments more than others. The possible determining factors will be discussed below. Question 7. She wishes her parents would believe her about the party. It was anticipated that this question would elicit the highest number of acceptances and the lowest number of rejections, as the environment in which would occurs is most conducive to notions of agentivity, containing a human subject and a main verb which can be associated with senses of volition or control. As a future hypothetical modal, the interpretation of would in such contexts might appear marginal, allowing only for a future situation about which the subject already has known expectations (but not facts — she anticipates her parents intentions not to believe her). The most readily accessible context, though, seems to be that her wish reflects her present knowledge of the volitions of the complement clause subject (her parents), and her desire for a situation contrary
HYPOTHETICAL WILL
179
to her present knowledge. The use of a verb expressing a cognitive state illustrates more precisely the basic volitional senses of would used in hypothetical complements. If a verb of direction were used, for example, in Palmer’s (1986) example, I wish John would come, there is additional ambiguity created between the predictive and the volitional uses of the modal since the lexical verb of movement (come) contributes to the sense of futurity, and the predictive meanings may be context-induced. In such cases, it is uncertain, therefore, whether the inferences of prediction are derived from the modal auxiliary or from the senses in the main verb (or both). What is significant, then, is that cognitive state verbs do not express the same lexical inferences of futurity as do verbs of direction, but that volitional senses are present, suggesting that such environments may provide the optimum situations for the development of predictive senses. The responses indicated that the assumptions were well justified — and the highest scores for acceptability came from this question (63.6% for the Singaporean group and 85% for the Australian group). None of the Australian group rejected the sentence as unacceptable, although 12.1% of the Singaporean group made this evaluation. This was the same situation as in Q. 4, in which there were 0% U answers for the Australian group, but 6% for the Singaporean group. This reversal of the main tendencies is illustrated in Figure 6.2, environments A and B, in which the example sentences receiving the highest scores of acceptability by the Australian group are considered less acceptable by the Singaporean group. (The remainder of the pattern shows the Singaporean group expressing higher levels of acceptance for all the other environments.) It would be difficult to explain why this is the case, beyond suggesting that the Singaporeans were using different reasons for their evaluations from those of the Australian group; this would entail that the basis for their acceptances was not the presence of polysemous senses of volition and prediction. Only one alternative was offered for the U answers in Q. 7: She wish her parents believed her about the party (sic). The other 3 informants who marked Q. 7 as U did not offer any alternatives, one simply adding three question marks to his answer. 6.2.4
Summary
The results of the study described above, when compared with those in 6.1, show significant differences with regard to the degree of acceptability attributed to the example sentences. The widest margin of difference between the two groups appears in Q. 2 (the progressive — shown as Environment F in Figure 6.2) in which 48.5% of the Singaporean group but only 6% the Australian group rated
180
HYPOTHETICAL MODALITY
the sentence acceptable, and 24.2% of the Singaporean group and 51.5% of the Australian group rated it as unacceptable. If the use of the progressive acts as an unambiguous marker of epistemic modality, as discussed above, it indicates that for this question epistemic meanings in the environments surveyed are far more generalised in Singaporean English than in Australian English. This trend is further illustrated in the differences shown for the scores for Q. 5, in which a formal grammatical device, or ‘expletive’ subject containing no lexical meaning or traces of agentivity receives an acceptability rating of 54.5% by the Singaporean group and only 24.2% by the Australian group. The acceptability of environments such as these indicates the likelihood that the grammaticalisation of this modal to epistemic uses in such clauses is more advanced in the Singaporean group than in the Australian English sample, since Q. 2 (progressives) and Q. 5 (existential constructions) are two environments in which only bleached meanings are possible (and old and new meanings cannot co-exist). Thus, the use of would in some environments is ambiguous between the older, volitional meaning, and the newer, predictive meanings; e.g. I wish John would come to the party, and the initial development of new meanings will only be possible as inferences from the old meanings; in I wish John would come there is an inference that the speaker is aware of John’s intentions not to come, or may be unaware of what his intentions are, but in I wish you would be tall no such inference can be created, since the addressee cannot be said to have the intention to be tall. In the former sentence the hearer can infer from the use of a predicator such as wish that the content of the predicate is hypothetical, and the addition of a predictive modal such as would places the hypothetical meaning in the future; the use of a lexical verb of movement such as come also contributes to the development of such senses. Both volitional and predictive senses are simultaneously available at a stage of development which is intermediate between the old and the new meanings. For the Australian group, the lower acceptance of purely epistemic or predictive environments, as Q. 2 and Q. 5 show, is an indication that the grammaticalisation of the modal is still at the ‘A/B’ stage of an A > A/B > B path (Hopper and Traugott 1993: 36), where its acceptability is restricted mainly to environments in which both old and new meanings are still possible. On the other hand, the higher acceptablity of Q. 2 and Q. 5 in the Singaporean group shows that the modal has been further generalised to environments in which the former ‘A’ meanings are no longer possible. Also indicative of a more advanced generalisation to an environment where an older interpretation is not possible are the Singaporean acceptances for Q. 6., I wish the Porsche would belong to me. Although the rejections of this example are equal for both groups, there is a 15% difference between the acceptances,
HYPOTHETICAL WILL
181
the Singaporean group being higher. The possibility of acceptance as a predictive reading is increased by the inanimate subject and the stative verb, which would not permit a volitional interpretation; thus, such environments are most suggestive of the more rapid grammaticalisation of the form for Singaporean speakers. Perhaps the only anomalies are the responses to Q. 3 and Q. 7. In Q. 3, I wish I would come to your party, the rejection figures are equal for both groups, and it is believed that such judgements were made on the basis that the most likely interpretation of the modal is volitional, and this is incompatible with the environment. The zero acceptance rate by the Australian group confirms this; however, there is an acceptance rate of 15% for the Singaporean group. The possibility of its acceptance as a volitional modal is ruled out on the basis that we do not often have cause to hypothesise about our own desires (?I wish I intended to come to your party) — the interpretation is almost contradictory, as noted earlier. On the other hand, the predictive reading is also risky, since the volition readings still constrain a first person subject, unless the prevailing circumstances are beyond that person’s control (consider the non-hypothetical ?I hope I will come to your party, which seems to provoke the reply: “Well, can’t you do anything about it?” vs. I hope I will win in which there is less question of the subject’s control over the future situation). Would and will can rarely be purely predictive with a first person subject and an action predicate (but see 6.3), unless the accomplishment of the action is outside of the volitional control of the subject. Thus, the grounds for accepting the modal in Q. 3 as having either volitional senses or predictive senses are not strong, given the extent to which the former senses are retained in such environments, and a 15% acceptance of the form by the Singaporean group can only indicate that for such speakers, the retained volitional senses are not so apparent. Q. 7 is a similar anomaly: She wishes her parents would believe her about the party. The higher number of rejections and lower acceptance rate by the Singaporean group are not consistent with the group’s results of the other examples, in which there is an indication of more advanced grammaticalisation to predictive or epistemic uses of the modal. Moreover, this example is the most frequently accepted by the Australian group (85%), since the situation expressed in the predicate is one over which volitional control may be exercised (belief about a situation); there are also zero rejections for this example by the Australian group. If the basis for rejecting the sentence were that volitional meanings were the only ones possible, but were anomalous, as in Q. 3, such rejections would be conceivable. Furthermore, if the form is grammaticalising faster with the Singaporean group then there should be no reason to reject the form in Q. 7 either. But volitional meanings are not anomalous in this environment, as shown
182
HYPOTHETICAL MODALITY
by the Australian subjects, whose judgements seem to be controlled by such factors. There is clear evidence in this inconsistency that for some of the Singaporean speakers, their rejections might have been based on factors other than the presence or absence of a volitional reading. Although the average differences in the rates of acceptance and rejection for the questionnaire do not, in many cases, show a great deal of difference between the two groups, there are some differences in individual questions which are quite striking and indicate that for some speakers of Singaporean English, the average rate of retention of former meanings associated with earlier stages in the diachronic development of the grammaticalising modal are slightly lower than for a similar group of Australian speakers. The study represents a synchronic crosssection of a diachronic process seen across two linguistic communities, shown at a stage in which the older and the newer meanings are seen to co-exist in some environments, producing higher acceptance rates by the informants, while in other environments, and especially for the Singaporean speakers, the first indications of new meanings are occurring without the corresponding possibility of older meanings persisting in the same environments (e.g. Q. 2 and Q. 5). Eventually, it is presumed that would would be used indiscriminately in such environments, as the former volitional meanings are gradually bleached, and the generalisation of functions becomes unrestricted. However, the study predicts that the rate at which the modal grammaticalises to all possible environments will be considerably faster for speakers of the L2 dialect than for those who are native speakers of the L1 dialect, Australian English. This may be due to two contributory factors: (1) the need for greater generalisation of function by the L2 speakers, perhaps not a factor for native speakers; or (2) the controlling forces inhibiting the generalisation of functions for native speakers, which are not present to the same extent in L2 dialects. It is believed that these two causes interact and are not mutually exclusive; however, the reasons for the presence of such factors are not yet clear. Ch 7 will propose an explanatory hypothesis which may account for the these reasons.
6.3 British English — a control study In 6.2, it was seen that a number of differences existed between the acceptability of would by Singaporean speakers and by speakers of Australian English. Although it was proposed that the basis for such differences might be attributed to functional differences in the use of the item, the possibility cannot be ruled out that the differences may be associated simply with features of random
HYPOTHETICAL WILL
183
variation specific to the speech community in which the data was collected, and reflecting local practices. In order to assess whether or not this is the case, and to provide a more thorough investigation of usage variation amongst L1 speakers, further questionnaires were distributed amongst groups of British-born native speakers of English, including students in a central London secondary school and employees of the British Natural History Museum, and some non-British-born L2 speakers of English, who were also students of the same secondary school in central London. The students were all at Year 12 and 13 (final year secondary students), and their ages ranged from 16–18; that is, they were all at preuniversity level. Generally, it is found that speakers of this age (older teenagers) are more likely to be innovators of linguistic change than either older adults or younger teenagers, especially regarding lexical and phonetic changes (see, e.g., Finegan, Besnier, Blair and Collins 1992: 379–80). It would be assumed then, that young speakers of the 16–18 year age group in a cosmopolitan environment such as London would be likely to provide a stringent control test to determine whether the variation seen so far reflects a true contrast between L1 and L2-speaking communities. The homogeneity of the London group (LG) in terms of age and background by comparison to the other two groups was also felt to be an added advantage both in the sampling procedure and in its capacity to provide the optimum conditions for a control. 6.3.1
Methodology
The L1 student subjects (London L1 Group — LL1G) were born in Britain, many in London, and all named their first language used since infancy as English. They were all attending the North Westminster Community School, Upper School, and were surveyed under informal classroom conditions.11 The 17 L2 speakers of English (LL2G), surveyed under the same conditions and at the same time as the the L1 speakers, named various L1 backgrounds, such as Sudanese, Bengali and Arabic. As well as London, they listed their birth-places as the Sudan, Bangladesh, Iraq, and Tanzania. The Museum workers (London Museum Group — LMG) were all native speakers born in Britain (18 in number) and their average age was 41 — in this respect they represented an older, more educated, and possibly more conservative age-group. The latter group
11. Acknowledgement is especially due to Vivienne Heaton, Philip Blackwell, and Michael Marland MBE, without whose kind permission and cooperation the secondary school survey could not have been possible. Thanks are also extended to Jennifer Bryant for her assistance in collecting the data from the employees of the British Museum of Natural History, London.
184
HYPOTHETICAL MODALITY
were surveyed at their work-place, the questionnaire being distributed to them electronically. No distinctions were made in the sampling procedure between male and female respondents, since, as found with the earlier Australian study, differentiation by gender was not considered to be an influencing factor. Not all of the questions were answered; Questions 3, 5, and 7 were answered by only 36 out of the 37 respondents in the L1 student group. The questionnaire used was identical to that used for the Australian group (AG) in that it contained a number of distractors not used with the Singaporean group (SG).12 The respondents were given no information on the objectives of the questionnaire, but were free to include their comments after it was completed. The results and comments are listed below (6.3.2–3). 6.3.2
Results
The results of the LL1G survey are listed in Table 6.9 and Figure 6.3. Table 6.9 shows the acceptability of the distribution of WILL in hypothetical complements by environment. Figure 6.3 maps the figures for acceptability, showing likely trends for the grammaticalisation of would across the three main groups: the Singaporean Group (SG), the Australian Group (AG), and the L1 student group from the London secondary school (LL1G).
Table 6.9 Question 1. 2. 3. 4. 5. 6. 7. Average
London L1 Group (LL1G): Graded assessments of the questionnaire by both male and female respondents from the London secondary school U
VQ
Q
A
32.4 43.2 47.2 13.5 13.8 18.9 08.3
27 18.9 22.2 05.4 13.8 24.3 02.7
40.5 24.3 19.4 24.3 30.5 32.4 11.1
00 13.5 11.1 56.7 41.6 24.3 77.7
25.3
16.3
26
32.1
N = 37 (36 for Questions 3, 5, and 7). Figures are in percentages, rounded to the nearest 0.5%.
12. Distractors were not used with the Singaporean Group because the purposes of using them were not considered to be relevant or necessary for this group.
185
HYPOTHETICAL WILL
100
LG SG AG
% acceptability
80
60
40
20
0 A
B
C
D
E
F
G
Environment Environment A: Would + believe — stative verbs expressing subject-oriented control (Question 7) B: Negatives (Question 4) C: Existential constructions (Question 5) D: Stative verbs — inanimate object subjects (Question 6) E: Stative verbs — animate subjects (Question 1) F: Progressives (Question 2) G: Identical subject in matrix clause (Question 3)
Figure 6.3
Percentage of acceptable responses for LL1G, SG and AG, and indicating the possible differences in rates of retention and grammaticalisation rates associated with WOULD in hypothetical complements
186
HYPOTHETICAL MODALITY
Table 6.10 illustrates the distribution of responses from the older group of native speakers in London (the museum workers — LMG) whose average age was 41. The number of respondents was 18. Table 6.10 Question 1. 2. 3. 4. 5. 6. 7. Average
Distribution of responses for the London group of native speakers (older age museum workers group) U
VQ
Q
A
61.1 61.1 77.7 05.5 38.8 55.5 05.5
27.7 22.2 11.1 16.6 16.6 27.7 00
05.5 16.6 11.1 16.6 22.2 11.1 16.6
05.5 00 00 61.1 22.2 05.5 77.7
43.6
17.4
14.2
24.5
The figures are given in percentages and are rounded to the nearest 0.5%.
Table 6.10 is compared with Table 6.11, illustrating the distributional ratings for the secondary school respondents whose L1 was not English (LL2G). The total number of informants was 17, and the figures are given in percentages. Table 6.11 Question 1. 2. 3. 4. 5. 6. 7. Average
Distribution of responses for non-English L1 speakers (London secondary school) U
VQ
Q
A
35.2 29.4 41.2 05.9 05.9 05.9 00
11.8 29.4 17.6 00 23.5 17.6 00
35.2 29.4 29.4 23.5 29.4 41.2 23.5
17.6 11.8 11.8 70.6 41.2 35.3 76.5
17.6
14.3
30.2
37.8
187
HYPOTHETICAL WILL
Table 6.12 illustrates the distribution of acceptances of all groups, arranged in order from highest-scoring sentences to lowest-scoring ones, and indicating the possible general diffusion of a change in progress as it spreads across a paradigm of environments. The score for Question 2 is outstanding with regard to the Singaporean data, and will be discussed in detail below. For the remainder of the grid, it is seen that the changes illustrate a gradual, wave-like movement starting from the top left-hand corner, where the complete change begins with the L2 groups in all environments, then weakens to include only a restricted number of environments amongst the L1 groups. This type of wave-like spread is reminiscent of that discussed in early variation studies (see, e.g. Bailey 1973). The nonL1 speakers clearly indicate faster generalisation by the greater range of environments present, and the younger London native speaker group is not far behind. The pattern, however, should not be taken as representing any sort of implicational hierarchy, since the environments are not ordered with any particular purpose, but the patterns reveal which environments are most likely to be the first to receive a predictive reading of the modal in each group. The sociolinguistic reasons for the more advanced generalisation in the younger London group will be explained below. Table 6.12 Question 7 4 5 6 1 2 3 Average
Distribution of acceptances by all groups, arranged in order from highest to lowest scores SG
LL2G
LL1G
AG
LMG
63.6 63.6 54.5 36.3 12.1 48.5 15.1
76.5 70.6 41.2 35.3 17.6 11.8 11.8
77.7 56.7 41.6 24.3 00 13.5 11.1
85 70 24.2 21.2 09 06 00
77.7 61.1 22.2 05.5 05.5 00 00
41.9
37.8
31.9
30.7
24.5
Figures are given in percentages.
In Table 6.13, the variance of the scores for each question are calculated, in order to estimate the difference between each score for each question and the average for each question; i.e. to check the dispersion. The figures are all in percentages. As can be seen in Table 6.13c, there is an enormous range in the variance
188
HYPOTHETICAL MODALITY
Table 6.13 a.
b.
c.
Variance of the scores for each question
Differences (di) between actual acceptance scores for each group per question and the mean (M). Question
M
7 4 5 6 1 2 3
76.1 64.4 36.7 24.5 08.8 16 07.6
SG di
LL2 di
LL1 di
AG di
LMG di
−12.5 −0.8 17.8 11.8 3.3 32.5 7.5
0.4 6.2 4.5 10.8 8.8 −4.2 4.2
1.6 −7.7 4.9 −0.2 −8.8 −2.5 3.5
8.9 5.6 −12.5 −3.3 −0.2 −10 −7.6
1.6 −3.3 −14.5 −19 −3.3 −16 −7.6
Squares of the differences calculated in Table 6.13a Question
M
SG di2
LL2 di2
LL1 di2
AG di2
LMG di2
7 4 5 6 1 2 3
76.1 64.4 36.7 24.5 8.8 16 7.6
156.2 0.64 316.8 139.2 10.9 1056.2 56.2
0.16 38.4 20.2 116.6 77.4 17.6 17.64
2.56 59.2 24 0.4 77.4 6.25 12.2
79.2 31.4 156.2 10.9 0.04 100 57.76
2.5 10.9 210.2 361 10.9 256 57.76
Calculations of variances (V) per question (n = 5 x groups) Question
∑di
∑di2
V (∑di2 ÷ n−1)
7 4 5 6 1 2 3
0.0 0.0 0.2 0.1 0.2 0.2 0.0
0240.6 0140.5 0727.4 0628.1 0176.6 1436.0 0201.5
060.1 035.1 181.8 157.0 044.1 359.0 050.3
for some of the sentences, when compared with the mean for the same sentence. This indicates that the general distribution of responses may indeed be meaningful. The data will be discussed in detail below.
HYPOTHETICAL WILL
6.3.3
189
Discussion
The results will be examined from the point of view of the trends exhibited in Tables 6.12 and 6.13, as these contain the most revealing patterns in the presentation of the acceptability of the modal. The data is not to be understood as a direct indication of the grammaticalisation of the modal, as the environments represented are only seven of all the possible uses to which the modal may be extending; however, as acceptability judgements, they are interesting samples of a token representation of a possible grammaticalisation pattern across five different dialect types. Clearly, the averages shown in Table 6.12 indicate a gradually descending order of acceptance, with the two L2 groups at the higher end of the scale, the Australian group and the museum workers at the other end, and the younger L1 speakers in the middle. The differences are not great, but the ordering of the averages is important, and there is almost a 10% difference between the Singaporean Group and the highest scoring of the two L1 London groups (LL1) in total average acceptance rates. It could be suggested, at a brief glance, that the rates of grammaticalisation of the modal to new environments are affected in the first instance by whether the speaker is of an L1 or L2 background, and second, by the age group to which the speaker belongs, since the LL1 group was a younger group than the LMG. There is a clear indication that sociolinguistic variables may lie within the more general constraints of a speaker’s language background as indicators of the rate of grammaticalisation of a feature in a language. The pattern represented in Table 6.12, as with many variational studies, appears to illustrate an intermediate stage of change in the function of would in which only a limited number of dialects are represented in all of the environments, while some environments are not represented at all for some of the dialects. This may represent a dynamic model of variation, demonstrating a change in progress, cutting across both dialect types and grammatical environments. The change modelled in Table 6.12 appears to have begun in environments 7 and 4, rapidly spreading to 5 and all the other environments in its path, but at different rates for different groups of speakers, as shown by the smaller numbers in the bottom right hand corner of the grid which represent groups and environments not yet affected by the change. Clearly the speed at which the item grammaticalises is evident in the distribution of scores in this way, with age and language background as two interacting factors, and it may be predicted that the bottom right-hand corner of the matrix will eventually, with time, show higher scores as the modal grammaticalises to include an increasing number of environments amongst the L1 groups.
190
HYPOTHETICAL MODALITY
Question 2, in which the modal appears in a progressive construction, still stands out as the most marked indicator of the grammaticalisation rate for the Singaporean group. The reason for this will be investigated below. In the L1 groups there are scores of zero acceptability for Questions 1, 2 or 3. In particular, the score of zero for Question 2 by the museum workers’ group (LMG) stands in stark contrast to the 48.5% acceptance by the Singaporean group. The overall difference in averages between the highest scoring group, the Singaporean Group (rounded to 42%) as against the lowest scoring group, the London Museum workers (24.5%) is roughly 17%, and is significant. Although the numbers for this group were comparatively fewer than for the Singaporean group, they are indicative of a trend which would be likely to continue with the addition of more subjects of approximately the same background. The results will now be discussed in order of average acceptance from highest to lowest. Environment A (Question 7) There was a considerable margin of difference between the SG and the AG for Question 7 — nearly 22%, but this narrowed to a 14% difference between the SG and the LL1G and the LMG who seemed to find the sentence less acceptable than the Australian group. Such data indicate that, for the Australian group, the use of would in hypothetical clauses with a verb of cognition or at least a controllable state is well accepted, indicating the possible ambiguity for the Australian group between meanings of volition and meanings of prediction in such environments. For the other two groups, such ambiguity was not so pronounced: 8.3% of the LL1G found the sentence unacceptable, giving it a U response, while 2.7% marked it very questionable and 11.1% marked it questionable. Only one respondent supplied a correction to this sentence, and this was: “I wishes her parents belived her about the party” (sic). The fact that only one respondent volunteered an altered version is an indication of the extent of the grammaticalisation of the modal in such environments, as noted earlier. Both the modal form and the non-modal form seem to be generally acceptable, then, across all three groups. None of the LL2 respondents corrected the modal, although one changed wishes to ‘wished’. For that particular group, there were scores of zero for U and VQ answers, and 23.5% marked it as Q. For the LMG, there were 5.5% of U answers, zero VQ, and 16.6% Q. The one participant who marked this sentence as U did not provide a substitute. The results conform very much to those in 6.1 in which the use of would in such clauses is understood to be primarily volitional, since beliefs are usually felt to be stative situations over which agentive control can be imposed. The environment of a consciously
HYPOTHETICAL WILL
191
controllable state, then, provides the optimum conditions for the grammaticalisation of the modal, since both volition and prediction can be perceived to coexist at the same time. This represents a polysemous situation, and polysemous stages usually precede the development of more grammaticalised functions (Hopper and Traugott 1993: 92), as discussed in 6.1. Environment B (Question 4) Question 4 contained a negative: I wish Pat wouldn’t come to the party, and, as mentioned earlier (6.1), it was felt that a sentence of this type could only be possible if the reading were assumed to be volitional, and the volition had present time reference. Thus, since come is a non-stative verb, it would represent a wish for a particular habitual action (on the part of the subject) not to take place. As a predictive reading, it could suggest that the modal had grammaticalised to express future meaning in negative environments; however, grammaticalisation of the modal in negative environments is usually considered to be slower than in positive ones due to the a number of reasons, including negative raising, and the redundancy of negative epistemic possibility as an implicature of positive epistemic possibility (see 6.1). Such factors are likely to delay the rate at which negative root modals grammaticalise to have predictive and epistemic meanings. However, as is evident in Table 6.12, the use of hypothetical would in a negative appears to be unconstrained by such factors, and grammaticalisation is fairly well advanced amongst all groups. The average score for Question 4 was 64.4, well over 50% of the total number of respondents classing it as acceptable, and the Singaporean group’s score was very close to the mean (d = −0.8). This sentence also showed the lowest variance score amongst all groups (V = 35.1), indicating the degree to which all participants agreed on the level of acceptability of the sentence. The low level of variation suggests the rapid rate of acceptance and conventionalisation of the use of negative hypothetical would, and this may be because the alternative volitional reading is unsupported by any kind of adverbial clause indicating present time reference. Amongst the altered versions supplied by the LL1 group, 3 suggested: I wish Pat wasn’t coming to the party, indicating clearly, at least for the 3 who corrected it, as with the Australian group, that the sentence was understood to be an expression of predictive hypotheticality and did not refer to the regular habits of the subject referent. Another suggested using the modal could instead of would, which was a frequent substitution, suggesting its more advanced level of grammaticalisation as a hypothetical predictive modal. Of the LL2 group, none supplied an alternative: only 5.9% marking Question 4 as unacceptable and none
192
HYPOTHETICAL MODALITY
marking it as very questionable. There were 23.5% who marked it as questionable. In the LMG, the distribution was more evenly spread: 61.1% A; 16.6% Q; 16.6% VQ; and 5.5% U. This indicates a little more uncertainty about the conventionalisation of the modal to express negative hypothetical prediction. Altered versions for this sentence for the LMG included the following: I wish Pat wasn’t coming to the party (2 respondents); I wish Pat weren’t come to the party (1 respondent — it is assumed that the respondent omitted to change ‘come’ to ‘coming’); I hope Pat is unable to come to the party (offered as “perhaps in a more friendly context” by one participant who nevertheless marked it as A) the suggestion of improving the ‘friendliness’ of the context indicates that the participant had understood the sentence to have volitional meaning, or at least to be ambiguous between being volitional and predictive. (Perhaps in a future exercise, the kind of ambiguity apparent here could be resolved by discussing the results briefly with the participants following the exercise.) One participant altered the modal form to could; another corrected the sentence to I wish Pat wouldn’t be coming to the party. This particular respondent did not accept Question 2, in which there is a positive progressive, so the response to Question 4 is quite interesting, and suggests that he or she accepted the negative hypothetical predictive modal prior to the positive one. As discussed in 6.1, progressives are held to be unambiguously epistemic environments for the use of modals (alternatively, it could suggest that 3rd person subjects were preferable to 2nd person subjects in epistemic contexts). Although only one participant responded in this way, it would not be too speculative to suggest that this particular respondent had borrowed the progressive to express non-volitional future, and hence prediction, because the non-progressive context suggested too much ambiguity. Clearly, the results of Question 4 indicate a rapid trend towards the conventionalisation and generalisation of hypothetical predictive meanings in negative environments, which may be prior to generalisation in positive environments. Environment C (Question 5) Question 5, I wish there would be no fighting at the party, presents an environment in which there is a formal syntactic subject (existential there) and a stative verb, be. In 6.1, it was discussed that this environment represented the most indecisive for the respondents of all of those surveyed, since there was no likelihood of any ambiguity with volitional senses to provide an alternative means of interpretation. Since the changes from volitional to predictive have been found to be gradual and arising out of polysemous uses of a form (as volition or intention implies
HYPOTHETICAL WILL
193
prediction in most cases), the absence of the possibility for a polysemous interpretation of would in certain environments is a key indicator of its level of grammaticalisation. As for Question 7, when there is a likelihood of either volitional or predictive meanings, the acceptance rates rise accordingly. However, as shown in 6.1, the modal has not generalised completely as a predictive modal in such contexts, and this results in the indecision over environments such as Question 5 presents, in which there is no chance of a volitional reading. The results clearly indicate that the distribution of the modal represents a change in progress. 54.5% of the Singaporean group accepted the use, as did 41.2% of the LL2 group. The LL1 group showed a similar score (41.6%), and the scores for the Australian group and the London Museum group were quite low, at 24.2% and 22.2% respectively. In the LL1G, there were 30.5% who marked the sentence Q, 13.8% VQ, and 13.8% U. Amongst the LL2G, the results were quite similar: 29.4% Q; 23.5% VQ; and 5.9% U. In the LMG the scores were: 22.2% Q; 16.6% VQ; and 38.8% U. Some of the corrected versions which were offered in the LL1G were similar to those offered by the Australian group: in many cases, wish + would was substituted with hope + will: I hope there will be no fighting at the party (2 respondents); I hope there won’t be any fighting at the party (1 respondent); I hope there isn’t any fighting at the party (1 respondent); and I wish there to be no fighting at the party (1 respondent); I wish there would not be any fighting at the party (1 respondent). The rejection of the modal is clear in the substitution with a different expression, and as with the Australian subjects, this suggests a rejection of the possibility of hypothetical future contexts as an environment for the use of would. Hope + will suggests desire for realisation in the future, but not the hypothetical senses accompanying the use of wish + would. Extension of would in this way is clearly constrained for some respondents by the likelihood that the modal is interpreted by them as basically expressing hypothetical volition. When such meanings cannot co-exist with predictive or future hypotheticality, it appears to be less frequently accepted. Rejection of the modal was also evident in the LMG, in which a number of alternatives were offered besides hope + will. These included: I wish there will be no fighting at the party (1 respondent); I wish there could be no fighing at the party (1); I wish there was no fighting at the party (2); and another who discussed the sentence as follows: “ ‘wish’ should read ‘hope’ and ‘would’ should read ‘will’, unless the sentance (sic) as a whole is referring to a previous occasion when ‘wish’ should read ‘had wished’ or ‘had hoped’”. This respondent’s explanation summarises the confusion felt over the time reference of the event ‘the party’, presumed to be either a future occasion, as demonstrated by many of the hope-substitutions, or alternatively, possibly a habitual or regular event, as
194
HYPOTHETICAL MODALITY
indicated by the few who chose not to use a modal verb at all. Only 2 respondents supplied an alternative in the LL2 group, one suggesting I wish there to be no fighting at the party, and another I wish there will be no fighting at the party. The results thus indicate that, although the environment is acceptable to some speakers, it is by far not categorically acceptable, and perhaps represents a midpoint in the movement of the change across a selected range of environments as shown in the surveys. Environment D (Question 6) Question 6 uses an inanimate subject and a stative verb, both features being incompatible with the possibility of volitional readings of the modal. However, as discussed in 6.1, inanimate subjects may occasionally be found to co-occur with would if the verb is an action or non-stative verb. The presence of a stative verb in this sentence brought the average rate of acceptances down to 24.5% across all groups, although there was a clear margin of difference between the two L2 groups and the L1 groups: the SG and the LL2G were very close at 36.3 and 35.3 acceptances respectively, while the LL1G had an average acceptance rate of 24.3%, the AG 21.2, and the LMG had an average acceptability score as low as 5.5%. The wide-ranging dispersion of the scores in this way raised the variance monitor to 157 (see Table 6.13c). The LMG’s low score was represented by a difference of −19 from the mean of all the scores, the second largest difference in all the scores. This suggests that the modal is relatively restricted in certain environments for this group. The distribution of the scores for Question 6 is also interesting: for the LL1G the distribution is fairly close: 24.3% A; 32.4% Q; 24.3% VQ; and 18.9% U. For the LL2G the distribution is slightly more biased: 35.3% A; 41.2% Q; 17.6% VQ; and 5.9% U. The LMG demonstrates a converse picture: 5.5% A; 11.1% Q; 27.7% VQ; and 55.5% U. Of those in the LL1G who rejected Question 6, 9 changed the sentence to I wish the Porsche belonged to me, 3 changed it to I wish the Porsche could belong to me, and one respondent offered I wish I owned the Porsche. In the LL2G, 4 respondents suggested I wish the Porsche belonged to me, one simply crossed out the word would, one offered could for would, and another expressed the sentence as I wish the Porsche was mine. In the LMG, 8 respondents substituted belonged for would belong, one suggested could for would, another suggested either could belong or did belong, another I wish I owned a Porsche, using no modal either, and one discussed the sentence as follows: “Incorrect English. Also the implication that the Porsche has a will of it own. ‘would belong’ should read ‘belonged’ or ‘could belong’, or simply ‘…
HYPOTHETICAL WILL
195
[actual wording unclear here] mine’ replacing the whole of the sentence, and possibly ‘the’ was replaced by ‘that’.” The comments made by the last respondent clearly indicate the degree of retention of volitional senses evident in the use of the modal, expressed as actual intuitions in the speaker’s assessment of the modal meaning. The use of this sentence is a key determiner, therefore, in assessing the presence of retention in real time in the dialects of individuals. The importance of such indicators will be discussed later in Ch. 7. For the moment, it can be seen that inanimate subjects and stative verbs represent one environment in which the constraints of lexical retention in the meaning of the modal would in hypothetical predicates are fairly strong, and grammaticalisation to predictive senses is slow amongst all groups surveyed. Environment E (Question 1) Question 1, I wish you would be tall, illustrated the lowest acceptance rate for the SG (12.1%), and the second lowest variance overall (V = 44.1). The acceptance scores were generally very low for this example, and there was one score of zero for the LL1G, which is quite significant since this was the only zero score for that group, and also since this group was the largest, comprising 37 participants. The other scores for Question 1 for the LL1G were: 40.5% Q, 27% VQ, and 32.4% U. For the LL2G the scores were: 17.6% A, 35.2% Q, 11.8% VQ, and 35.2% U. The LMG had a range of scores as follows: 5.5% A, 5.5% Q, 27.7% VQ, and 61.1% U, showing a steady rise towards total rejection of the construction. The sentence, if accepted, could only have a predictive reading, as the cooccurrence of volitional senses with a stative predicate over which there is no agentive control is most unlikely. For this reason it was rejected by most of the participants, indicating a strong tendency for the interference of lexical retention in the distribution of would to such environments. Of the ‘corrections’ made to Question 1, one suggested I wish you would be taller, 8 offered I wish you could be tall(er), and 9 suggested I wish you were tall — these suggestions were offered by the LL1G. There appeared to be a still relatively strong motivation to retain the older subjunctive equivalent, without the modal, in the alternatives offered for this example, although there were almost as many respondents who wished to introduce a modal into the construction, in spite of the fact that it would involve a change of meaning. Amongst the LL2G, 6 respondents suggested I wish you were/was tall(er), and only one supplied could for would. In the LMG, the tendency again was to substitute the modal + be with were, and this alternative was chosen by 13 of the 18 respondents. Two respondents changed the sentence to I wish you could be tall, one suggesting as an additional
196
HYPOTHETICAL MODALITY
possibility, I wish you would be taller. The tendency to use a comparative, taller, may be attributed to the possibility that for some speakers, would was on the verge of becoming grammaticalised in their dialects, and the intended predictive meanings were already starting to emerge as ‘glimmers’ in certain constructions. The use of taller suggests a state which is likely to be realised at a future point in time, while the use of tall must refer to the present state of affairs, and hence in Question 1 must imply a counterfactual, rather than a hypothetical situation. However, counterfactual volition is precluded because the state of being tall cannot be volitional, and counterfactual prediction is also precluded because of the need for a factual premise underlying the utterance of a counterfactual statement (see 6.1), and factual premises, strictly speaking, cannot apply to the future. However, a hypothetical situation of being taller in the future is conceivable if the modal is treated as predictive, not volitional. The speaker alters the sentence then, to reduce the anomaly and to provide additional discourse clues to increase the likelihood of a predictive reading. Environment F (Question 2) Question 2, I wish you would be coming to my party, was the sentence which showed the most variance amongst all the sentences (V = 359). The high variance score is explained by the extremely wide margin of difference between the highest score for acceptability (48.5% in the SG) and the lowest score (5.5% in the LMG). The evidence of such a spread, however, is interesting, as it gives a good indication of the divergence of the grammaticalisation of the modal according to dialect. As noted in 6.1, the progressive aspect is clearly a disambiguator when there is potential ambiguity between deontic and epistemic senses of a modal, and the ambiguity is resolved in the direction of an epistemic, or predictive, interpretation. Thus, there is no reason at all to believe that Question 2 contains any residual senses of volition, as generalisation to progressive environments would have occurred only after the modal had grammaticalised from ambiguous environments elsewhere. The use of the progressive is therefore a key indicator of the extent of the grammaticalisation of would in hypothetical predicates. The distribution discussed suggests a far more rapid rate of grammaticalisation for the SG than for any of the other 4 groups. This is significant and supports the hypothesis that the SG is displaying characterstics of grammaticalisation which are more advanced than those of the other groups, either the London or the Australian groups, and the evidence is overwhelming. Given such evidence it could not be said that the variation shown is simply an indicator of
HYPOTHETICAL WILL
197
minor trends in local or regional dialects: the differences are too great in particular environments to pass them off as simply ‘variational tendencies’. Considering the role of the progressive aspect as an environment in which epistemic senses are most likely to be disambiguated from root or deontic modal senses, the evidence displayed in Question 2 is decidely in favour of a predictive meaning for would for the Singaporean speakers. The breakdown of the results for Question 2 clearly indicate the bias. In the LL1G, the proportion of U responses was little less than the proportion of A responses for the Singaporean group: 43.2%; the VQ responses amounted to 18.9%; and there were 24.3% Q, and only 13.5% A. Amongst the LL2G, the distribution was more evenly spread: 29.4% U; 29.4% VQ; 29.4% Q, and only 11.8% A. The low rate of acceptance amongst the LL2G was matched with an equal distribution of responses for the other categories, indicating a certain amount of indecisiveness over the use of would in such constructions for this group. For the LMG, the scores were much more decisive: 61.1% U, 22.2% VQ, 16.6% Q, and 0% A, indicating a steadily declining willingness to accept the sentence expressed over the four categories. The alternatives supplied for this sentence by the LL1G involved mainly either a substitute modal, a loss of the progressive aspect, or a loss of the modal altogether, as had been the case for the Australian group described in 6.1. One respondent changed the sentence to I hope you come to my party, and 4 retained the progressive, but changed the modal: I wish you could be coming to my party. 6 respondents dispensed with the modal altogether, producing I wish you were coming to my party, and 4 offered I wish you would come to my party. 4 changed the sentence to I wish you could come to my party, losing both the original modal and the progressive, and one respondent suggested I wish you would be able to come to my party. In the other groups, the suggestions offered in the LL2G were 3 for were coming, 3 for would come, one changed it to I wish you are coming to my party, 2 to I wish you come to my party, and 2 to I wish you could come to my party. In the LMG, the alternatives offered were: one for I wish you would come to my party, 3 for I wish (that) you could come to my party, 12 changed it to I wish you were coming to my party, and one suggesting: “The English is incorrect here. The sentance (sic) should read ‘I wish you would come to my party’, with the implication that the second party does not wish to come to the first party’s party, which may be correct, otherwise this could also read: ‘I wish you could come to my party’, with the implication that the second party cannot attend for some perfectly acceptable reason. I think that’s enough parties for today.” The last (prescriptive) judgement (offered by the same respondent who had offered the comments on Question 6) reflects clearly the senses that are conveyed by the
198
HYPOTHETICAL MODALITY
use of would in hypothetical complements; i.e. that the hypotheticality derives in the first instance from the speaker’s assessment of the subject’s willingness to do or not to do what is expressed in the complement of would. The main clause subject or speaker in these cases hypothesises about the subject’s volition, which then implies the likelihood or possibility of the subject to do whatever is described in the would-complement. The change of the sentence to a non-modal progressive by so many respondents is commensurate with the fact that hypothetical volition is retained in such clauses, and since the progressive with a modal verb in such contexts can only express hypothetical epistemicity, and the only possible way to express hypothetical predictive meanings is by using a modal which has become more grammaticalised on the pathway to predictive meanings (could). The older subjunctive equivalent (were coming) remains for a longer period with stative verbs as the hypothetical meanings are grammaticalising slowly and co-occurrence of would with statives is still prohibited to some degree by retention of lexical source senses of volition. The persistence of the former subjunctive-related form is stronger, it seems, amongst the LMG, who are of an older age-group and likely to be also of a higher educational level than many of the other groups. The underlying, unmarked senses of hypothetical volition conveyed by the modal in wish-complements is no more clearly manifested than in the responses to Question 3, which had the lowest overall acceptance rating across all groups. Environment G (Question 3) Question 3, I wish I would come to your party, was the least preferred of all the sentences in the survey. The average acceptance rate was as low as 7.6, just a little below the average rate for Question 1 (8.8%); however, it received no acceptances from two of the groups, the AG and the LMG. The sentence is a key indicator of the retention of volitional sentences in the use of would in hypothetical complements as the zero acceptance levels from these two groups reveal. The fact that there were acceptances in the SG, the LL2G, and the LL1G confirms the hypothesis that retention rates vary across individuals and across different dialect groups, with the L2 groups exhibiting generally lower rates of retention than the L1 groups. The variance for the data for Question 3, though, was not high, at only 50.3, showing general agreement amongst the groups that did accept the sentence. The figures for the LL1G illustrated a general decline in acceptability across the four categories: 11.11% A, 19.4% VQ, 22.2% Q, and 47.2% U, indicating an almost 50% rejection rate. The LL2G’s results were as follows: 11.8% A, 29.4%
HYPOTHETICAL WILL
199
Q, 17.6% VQ, and 41.2% U, and the LMG showed an even sharper decline towards rejection of this example: 0% A, 11.1% Q, 11.1% VQ, and an overwhelming 77.7% U, the same proportion as that of the acceptance scores for Question 7. The figures all very decisively indicate a tendency to reject Question 3, which is more strongly indicated than the tendency to accept Question 7. This may suggest that the constraints of retention are stronger than the motivation to extend the grammaticalised uses of the modal to new environments, but more data and a wider range of environments would need to be surveyed before such an assumption could be verified. The differences between the 5 groups are not all that great for Question 3, but the difference between the SG and the AG or the LMG, 15.1%, is significant enough to illustrate the differences in retention rates. Some of the comments and suggestions for altering Question 3 were as follows: (LL1G): 18 respondents supplied could for would, one offered I wish I were coming to your party and another, I wish I was coming to your party. The large number of respondents who substituted another modal reflects, as with the Australian group (part 1), the likelihood that the meaning of would is analysed as inappropriate or anomalous and should be replaced by another form of a different modal meaning; i.e. hypothetical ability implying possibility is substituted for what would otherwise be hypothetical volition implying prediction, since hypothetical possibility suggests that factors unrelated to the subject referent’s control may affect the predictive value of the complement. Those who supplied the progressive form obviously felt the need to express prediction, as the progressive aspect (used to express hypothetical past in such clauses) has a frequent function as a future marker. In the LL2G, 6 supplied could for would, one changed the subject of the clause from I to you and your party to my party, and another changed the complement verb itself: I wish I would go to the party. It is uncertain what the motivations were for this change; the respondent obviously felt uncomfortable with the deictics expressed in the use of come to refer to the addressee’s party (perhaps the meaning of go suggests a more predictive sense than come, since the perceived destination is distant from the deictic centre). There appears to be little evidence of lexical retention in the remainder of this respondent’s answers. In the LMG, there were 10 substitutions with could, 2 with was/were coming, and 2 suggesting either was/were coming or could. One additional comment was: “U. ‘would’ should read ‘could’, or ‘were able’, unless there is a severe mental problem with the first party and the other half of the first party is denying the willingness of the second half of the first party to come to the party.” (!!) These comments illustrate the degree to which the volitional senses are persisting in the intuitions of this individual, and are manifested as grammaticality judgements, or as constraints on the distribution of
200
HYPOTHETICAL MODALITY
the modal. Not all grammaticality judgements may be assumed to be representative of underlying intuitive senses of lexical retention, but the responses to the survey are useful in isolating some which are. It remains now to link the individual situation with the historical and diachronic patterns of development; this will be discussed later in Ch. 7. Although the sentences were not selected with any particular ordering in mind, the patterns represented in the data do appear to reflect a given order of preference for acceptance of the environments, and such patterns seem to be fairly consistent across all the groups. In general, then, the most acceptable environments seem to be those in which either a volitional reading with possible inferences of prediction is acceptable, followed by those in which only a predictive reading is possible, and the least preferred environments being those in which a volitional reading is anticipated because of the presence of a human subject, but is anomalous due to other linguistic variables such as stative predicates or co-referential main and subordinate clause subjects. Such environments cannot yield to the development of predictive senses, since in the first place, they do not allow for volitional interpretations to be possible. The patterns can be described in Table 6.14. Table 6.14
Correlation of average scores with functional description, illustrating the preferred acceptance patterns
Functional description
Question
Average scores (%)
Volitional/predictive
7 4
76.1 64.4
Predictive only
5 6
36.7 24.5
(Anomalous) volitional
2 1 3
16.0 08.8 07.6
It should be noted with reference to Table 6.14, that the arrangement of the environments in this way is the result of the outcome of the data and was in no way predicted prior to the commencement of the surveys. Thus, the organisation of the data enables the observation to be made that Questions 7 and 4 are the more preferred environments for predictive uses of would by the groups surveyed and that this is most likely due to the acceptability of volitional meanings with
HYPOTHETICAL WILL
201
predictive senses emerging as possible inferences as an additional polysemy. Because of the additional inferences, there is a higher probability for acceptance in these environments, and the average acceptance scores are more than double those for which only a predictive reading is possible. Environments which allow only a predictive reading are less preferred, as are those in which a volitional reading would be anomalous, indicating that the grammaticalisation of would is not very advanced in hypothetical complements of verbs of wishing, at least for the groups surveyed. Note also the difference in ordering between Table 6.14 and Table 6.13a–b; Table 6.13a–b was ordered with respect to the general trend across the groups; the ordering in Table 6.14 reflects only the total averages. 6.3.4
Summary
A close examination of the data across all 5 groups surveyed indicates distinct trends with regard to the rate at which would is grammaticalising in hypothetical complement clauses of wishing. The presentation of the data in the way illustrated, showing comparisons of acceptance rates, is in order to demonstrate most concisely the direction of extension of the functions of would, rather than the resistance to such extension. Although there is clear evidence that the modal is constrained by interference from lexical retention in its distribution, it is important to present the data as a means of making a prediction about future generalisations of the functions. In 6.1 it was explained that the current restrictions are due to the retention of lexical source meanings of volition; in the present series of surveys it was hoped to demonstrate the distribution over a wider range of dialects than had previously been surveyed. This task has been accomplished in the additional data from the London groups of speakers of British English. Thus, the trends illustrated can be explained by either or both of two competing factors: the restriction of the generalisation of the modal due to the persistence of lexical source meanings, or the dynamic tendency to generalise to new environments, which is demonstrated with greater variation across all the groups. Such variation may be equally accountable to non-linguistic factors as to linguistic ones. There is a clear tendency for the SG, and to some extent, the LL2G, to show more rapid rates of grammaticalisation of the modal, with scores reaching 10% or higher than the L1 groups in many cases. It is highly likely, then, that the differences cannot be simply due to regional or local tendencies, but must be related to non-linguistic factors such as communicative needs (see Ch. 7) or age, as the relatively advanced levels of grammaticalisation in the LL1G seem to suggest (innovations being often introduced by younger, adolescent speakers — as noted earlier).
202
HYPOTHETICAL MODALITY
The most likely indicators of the tendency for extension rather than constraint in distribution are found in Question 2, in which there is a progressive form, a clear indicator of an epistemic function in the modal. Another likely indicator is Question 5, which has a stative verb but a formal or syntactic subject. Such environments preclude the possibility of polysemous meanings and, if accepted, imply that the form may be well advanced in its grammaticalisation stages for the speakers who have accepted them: there can be no likelihood of volitional meanings co-existing alongside the predictive ones to provide alternative interpretations. The outstanding score by the SG for Question 2 in contrast to the other 4 groups indicates that for that group the competition between the constraints of lexical retention and the motivation to extend the uses of a form by generalisation are won out in favour of the higher levels of generalisation across a paradigm of uses: the form is already understood as predictive by many informants (as a hypothetical counterpart of predictive will). For the remaining groups, the appearance of zero scores for some of the L1 groups is evidence of the strong resistance provided by lexical retention in the environments concerned (the lowest score overall for the SG was 12.1% and for the LL2G 11.8%). Although the size of the groups is not the same in all cases, certainly across a range of speakers there is a statistical difference between them; therefore, it could be said that the SG, an L2 group in an L2 community, generally displays a tendency for more rapid grammaticalisation than either a group of L2 speakers in an L1 community (LL2G), or 3 groups of L1 speakers in L1 communities geographically distant from one another. The individual differences for some of the sentences may be the same for many speakers across all the groups, but the purpose of the data is to demonstrate a tendency only, not the presence of idiosyncratic variation. It is also possible to suggest that an older age group, or a more educated group of speakers, will show greater resistance to the changes represented as the generalisation of the modal spreads across an increasing number of environments. Changes described intuitively as ‘incorrect’ or ‘ungrammatical’ by many non-linguists are often those that lead the way to the generalisation of a form in grammaticalisation, but they are initially resisted by the more conservative speakers in a community at first, and regarded negatively as unconventional or innovative. L2 speakers and often younger L1 speakers may not even be aware of their proscription by other members of the speech community. In this way a grammaticalisation change can often be seen to be reflected in sociolinguistic attitudes to language use. As shown in Table 6.14, the change is seen to begin in environments in which the older meaning and the new meaning are both competing and either is possible; i.e. it develops out of a polysemous form. Eventually the older meaning
HYPOTHETICAL WILL
203
will be lost and the new one will be found in a larger number of environments, but this stage is not apparent to a large extent in the present data. Question 7 demonstrates a potential ambiguity in one enviroment, and the likelihood of acceptance of this sentence is doubled because of such a possibility; i.e. speakers may accept it for either of two reasons, depending on the extent to which the modal is grammaticalising in their individual dialects. Later, non-ambiguous environments will follow in the order of change, and Question 5 has an average acceptance score of approximately half that of Question 7, due to the absence of ambiguity (the reasons for accepting it are reduced by the exclusion of speakers for whom the modal is still basically volitional in meaning). Questions such as 1, 2 and 3 are more likely to be interpreted as volitional since they have human subjects, but are precluded from acceptability because of co-occurrence features. However, in spite of the relative uniformity of the ordering of the environments, the rate of change amongst the groups is still variable. The reasons for such variation will be discussed in Ch. 7.
C 7 The Lexical Memory Traces Hypothesis
In Ch. 3 it was noted that the study of grammaticalisation describes a gradual process in which the lexical semantic content of a grammaticalising feature is slowly eroded as the feature increases its range of grammatical functions and available environments of use. As noted in 3.1, evidence that this process, referred to in the literature as bleaching, is a gradual process, rather than a sudden change in the meaning of a grammatical morpheme, is provided in the incidence of the persistence of the earlier meanings associated with the original function of the grammaticalising form as a lexical item (Hopper and Traugott 1993: 90). The phenomenon was first defined by Bybee and Pagliuca (1987: 112) as retention. The phenomenon of retention was investigated in Ch. 6, in which a number of anomalous sentences containing the modal would were presented for evaluation by a group of native speakers of primarily Australian English (6.1), and compared with the evaluations of speakers of Singaporean English (6.2), and British English (6.3). As shown in the results of the investigation, it was found that not all of the example sentences offered for evaluation were accepted by the informants and that semantic intuitions varied over the acceptance of the anomalies in accordance with the presence or absence of meanings which could be related to the earlier historical functions of would as a verb expressing volition or intention on the part of the subject. For this reason, it was concluded that the informants’ intuitions were indicative of the presence of retention as a synchronic phenomenon in their dialects. The study showed that the environments which provided the optimum conditions for a volitional interpretation of the modal were those most readily accepted by the informants. By such evidence, the semantic intuitions of the informants could be correlated precisely with information referring to the diachronic evolution of the modal, indicating that retention was a phenomenon which could be observed and quantified via the evaluations of the speakers of a language at any given time. However, the question is how a speaker would have
206
HYPOTHETICAL MODALITY
the ability to retrieve information reflecting the historical origins of a grammaticalising expression, and how such information may be manifested in the expression of semantic intuitions in present-day uses of a grammaticalising form, when the speaker has not observed its evolutionary developments. It could not be the case that a linguistically-naïve native speaker may be consciously informed of the lexical origins of the grammaticalising expression in use, and, armed with this knowledge, proceed to cautiously apply constraints to its use in environments deemed inappropriate according to knowledge of earlier functions of the expression. The study discussed above clearly shows a collapse of the distinctions between diachronic and synchronic spheres of investigation, in that what has been provided as historical evidence for the evolution of a grammatical morpheme in the development of a given language can be reconciled feasibly with synchronic evidence from present-day informants, indicating the curious persistence of historical meanings in intuitive judgements, and reflecting earlier developments in the grammaticalisation of the expression. It is not difficult to survey languages and arrive at conclusions such as these, but it remains an enigma how psychological phenomena such as intuitive constraints can be seen to coincide so precisely with diachronic evidence of historical developments in the language itself. The present chapter contains a hypothesis to account for the nature of retention as evident in the semantic intuitions of present-day language users, drawing mainly on available literature sources in order to illustrate the ideas presented. The first part of the chapter defines more formally the notion of retention, and illustrates with two examples the occurrence of retention in the results of diachronic grammaticalisation processes (7.1.1–2). The studies in Ch. 6, illustrating the differences in the perception of retention by L1 and L2 speakers, will be reviewed briefly in 7.1.3, and possible explanations will be offered for such differences in 7.2. Previous studies of children’s generalisations show that parallel stages in some diachronic and acquisitional grammaticalisation paths can be attributed to similarities in the basic conceptual structures of diachronic and ontogenetic source material (7.3.2.1–5). The synchronic representation of retention in intuitive assessments will then be hypothesised to be not only the result of diachronic grammaticalisation routes of development, but also a product of acquisitional paths. Thus, the differences shown between the L1 and L2 speakers in the studies in Ch. 6 in accessing retention as a constraining device on grammatical distribution will be explained with reference to possible differences in the paths of acquisitional grammaticalisation for such speakers.
THE LEXICAL MEMORY TRACES HYPOTHESIS
207
7.1 The phenomenon of retention 7.1.1
Retention as illustrated in the results of grammaticalisation processes
According to Bybee and Pagliuca (1987: 112), retention is reflected in a grammaticalising form exhibiting a polysemous structure such as the English modal will which can be expressive of volitional meanings in some contexts and purely future meanings in others. The polysemy of a grammaticalising item demonstrates the co-existence of old and new meanings in one form, reflecting a stage in the grammaticalisation of the form in which both the earlier and the later stages can be used according to contextually-determined factors. In some cases this co-existence amounts to ambiguity, as in the use of a negative: He won’t join us, in which the modal is capable of expressing either volitive or predictive meanings. In other cases indications of retention are evident in the fact that old meanings from which a form has grammaticalised will often not be appropriate to new environments, and so will resist grammaticalisation at least initially. Hopper (1991) and Hopper and Traugott (1993), building on Bybee and Pagliuca’s definitions, were able to isolate a number of instances of retention or persistence (Hopper 1991), which were manifested as constraints on the distribution of a grammaticalising feature in certain synchronic environments. This is the type of retention described in Ch. 6 as evidenced in hypothetical would. Distributional constraints are the most frequently-cited and lucid indications of retention as the environments may be contrasted with already generalised instances of the same grammaticalising form, and also for that fact that they readily reveal the former lexical meanings of the grammaticalising form. Further examples of this type of retention are listed below (7.1.1.1–3). The three cases outlined below show the presence of retention as identifiable with a lexemic source. Retention has also been found to be present in grammaticalising expressions as an implicit lexical sense or property which does not affect the distributional range of the expression in present-day environments. Such instances are less perspicuous than those which do impose co-occurrence constraints, and are often imperceptible without the application of specific elicitation tasks. An example of such cases is in the retention of perfectivity in past stative predicates apparent in the results of the study in Ch. 5, which was detectable only by the presence of counterfactual implicatures. A similar example of this type will be given in (7.4.1); for the moment, however, the following sections (7.1.1.1–3) will focus on the type of retention exhibited in distributional constraints on the grammaticalising item, as this type will form the main basis for the development of the hypothesis in (7.4).
208
HYPOTHETICAL MODALITY
7.1.1.1 French The evidence of the retention of volitional source meanings of the English modal would, apparent in the data in Ch. 6, demonstrates clearly the way in which present-day distributional constraints in certain grammatical environments reflect former diachronic patterns of use. Evidence of similar phenomena is also found in other languages. Distributional constraints are most clearly illustrated in French in the examples given by Hagège (1993: 228), in which the form se voir (lit. ‘to see oneself’) is grammaticalised to become a static passive in which the subject is not responsible for, or is a victim of the event; for example: (1)
Il s’est vu offrir un livre ‘He was offered a book’ (Hagège 1993: 228)
Such uses are considered acceptable. However, using a sample of 52 native speakers of French, Hagège found that constructions such as (2) were accepted only with reluctance by the informants: (2)
L’aveugle, hésitant à traverser la rue, s’est vu soudain prendre par la main ‘The blindman, hesitating to cross the street, suddenly had his hands taken by someone.’ (ibid.)
This was due to the retention of the former lexical senses from which the expression grammaticalised, which were anomalous in a semantic environment relating to the later grammaticalised uses: a blind man does not see. The distribution of se voir is thus still affected by the co-occurrence constraints of its former lexical function. 7.1.1.2 Mandarin A similar instance of constraints from diachronically earlier meanings is given by Slobin (1997), who discusses the use of the co-verb ba, now interpretable as an object marker in Mandarin Chinese. Ba was derived originally in the fifth century B. C. from a full lexical verb with the meaning ‘take hold of’, later appearing during the Tang dynasty in serial-verb constructions. The grammaticalisation of ba in modern Mandarin has considerably restricted its use as a full verb, but it is not yet completely generalised in the grammatical function of an accusative marker, being still affected in certain environments by the retention of meanings associated with its original uses as a lexical verb meaning ‘take’ or ‘grasp’. This
THE LEXICAL MEMORY TRACES HYPOTHESIS
209
is especially apparent in the fact that it cannot co-occur with verbs of emotion with Experiencer subjects, such as love, miss, or verbs of cognition such as understand or see, as the following example illustrates: (3)
*ta ba Zhangsan kan-dao LE 3sg ba Zhangsan see-arrive / ‘S/He was able to see Zhangsan.’ (Li and Thompson 1989: 468)
Slobin’s explanation for such anomalies is that verbs of emotion or cognition are abstract and do not involve a physical manipulation or handling of the object; thus, the original meaning of ba continues to persist in the present-day functions of the grammaticalising form, and, as with se voir illustrated in 7.1.1.1, the grammaticalisation of the item is still inhibited in certain contexts. 7.1.1.3 Malay A third example of the presence of retention in a grammaticalising form is found in the grammaticalisation of the quantifier satu in Malay, from a reconstruction of a numeral plus a classifier (Hopper and Traugott 1993: 119–20). The development from watu (‘stone’) to sa watu (‘a + (classifier for smallish objects)’) and reconstructed in the form of suatu (‘a’; ‘one’) today (usually pronounced and written as satu), has its repercussions in present-day functions. As a numeral quantifier, it cannot co-occur with a classifier, since the sense of a classifier is already retained as part of it meaning. Thus, retention is evident in the redundancy in (4): (4)
*suatu buah rumah one house (Hopper and Traugott 1993: 120)
(4) would be impossible, although suatu rumah (‘one house’) is quite acceptable. The evidence of co-occurrence restrictions in (4) and in the other examples above is explained in terms of the continuing influence of the historical origins of the grammaticalising form, indicating that grammaticalisation progresses gradually across a range of environments, and that at any one stage, there may be some environments which have not yet been affected by the changes taking place, and some that have. The distributional knowledge about a grammaticalising item is present both in the speech community as a whole, and in the intuitions of the individual speaker in that community.
210 7.1.2
HYPOTHETICAL MODALITY
Defining the phenomenon
In the examples illustrated above (7.1.1.1–3), as well as in earlier chapters, the distributional constraints on grammaticalising items enable lexical retention to be clearly manifest in present-day anomalies. The present chapter, however, carries the definition further to encompass a wider range of situations, and to account for the presence of retention in the individual intuitions of a language user. The evidence of retention in the examples described above also suggests that grammaticalising meanings have an aggregate nature comprising not only meanings related to the later stages of their development, but also senses reflective of earlier diachronic uses. The term ‘compositional value’, first appearing in Langacker (1988), could be borrowed to apply to the composite meaning of a grammaticalising expression, and to refer to the accumulated storage of semantic and/or pragmatic information reflecting earlier uses of the grammaticalising expression which has not been ‘bleached’ as the expression progresses in its development. Retention of older lexical meaning is thus considered to be reflected in the compositional value of present-day grammaticalising entities, and may affect the range of uses available for them. It could be argued that the evidence of retention illustrated in 7.1.1.1–3 and in Ch. 6 represents nothing more than a summary of the selection restrictions or co-occurrence constraints pertaining to diachronically earlier, lexical uses of a grammaticalising form. However, retention can be related to ambiguity, as noted in 7.1.1, and again it is not a matter of constraints on distributional range; e.g. not only the modal will, but also may is often ambiguous: You may be late (if you wish (permission)/because there’s a traffic jam up ahead (prediction)). Furthermore, it is shown in 7.4.1 that retention can be perceived as part of the compositional value of grammaticalising categories without invoking co-occurrence constraints on distribution, and this can be demonstrated in exercises of elicitation. Thus, although retention is most perspicuous in co-occurrence constraints on grammatical distribution, it cannot be equated with such restrictions; it is merely a representation of historically earlier meanings in a grammaticalising expression. The phenomenon of retention illustrated in diachronic grammaticalisation may then be regarded as manifest in the effects on the synchronic functions of grammaticalising forms, resulting in either the presence of co-occurrence or distributional constraints or in ambiguity. For this reason, the description is not intended to include ‘relics’ of older forms which have become fossilised in certain uses, but are obsolete elsewhere; for example, the suffix -dom in seldom derives from an Old English dative case ending (Hopper and Traugott 1993: 164) but has no function in any other context. Such archaisms have no effect on
THE LEXICAL MEMORY TRACES HYPOTHESIS
211
synchronic grammaticalisation processes and are not to be classed in the present study as instances of retention. It is also intended to distinguish retention from similar processes such as ‘split’ (Heine and Reh 1984), or ‘divergence’ (Hopper 1991). In such cases the original lexical source of a grammaticalising item continues its existence as a lexical item, but does not undergo grammaticalisation as does its cognate; for example, French pas (‘step’) became grammaticalised to a negative marker, but the lexical functions of pas remain active in an independent function. Clearly, though, divergence cannot be confused with retention in such instances, as the lexical meanings in the grammaticalised form are often bleached to the extent that there is little likelihood of them ever imposing constraints on the distribution of the negative marker. Hagège (1993: 202) illustrates the autonomy of the original lexeme alongside its grammaticalising cognate in similar examples such as the movement futures of French and English: (5)
a. b.
Il va y aller He is going to go there.
Such examples show that the grammatical form is bleached to the extent that it can co-occur in the immediate environment of the lexical form without redundancy or contradiction. However, when the form is not yet fully bleached of its lexical origins, the situation is much less perspicuous, and the only possible means of discerning whether it is a genuine case of retention might be in the quantification of tokens of retained uses for particular environments relative to those of an autonomous lexical counterpart; for example, the volitional use of will is a case of retention and not split, as it appears with relative infrequency in more marginal environments such as subordinate clauses, but it is no longer productive as a full lexical verb with nominal objects. Aside from this, it seems that split is virtually indistinguishable from retention until the later stages of grammaticalisation, when the form is bleaching, but further investigation would be required to clarify such matters. The evidence of the examples described in 7.1.1.1–3 suggests that the following factors are therefore relevant with regard to the definitions of retention in grammaticalising expressions: (i)
Retention of original lexically-oriented meanings is most often visible in situations in which former meanings may inhibit the grammaticalisation of a feature to some grammatical environments until the feature is bleached of any residual lexical senses (if ever). This situation of distributional constraints has been illustrated in (2–4) above, and in the grammaticalisation of would in Ch. 6.
212
HYPOTHETICAL MODALITY
(ii) Retention may also occur as one of the polysemies of a grammaticalising expression, as discussed with regard to the modals will and may, which often exhibit ambiguity between the old and the new meanings in some contexts, but are not distributionally constrained. Both (i) and (ii) may be termed ‘direct retention’, since there is a directly perceptible formal association between the encoded lexemic source and the later grammaticalised function word. (iii) Retention may also occur as an implicit lexical property of a grammaticalising category, as is illustrated in Ch. 5 in the retention of perfectivity in past stative predicates. In such cases the presence of an encoded source lexeme may not apply since the lexical source may be present only as a prototype function, deriving its lexical character from association with other relevant or contiguous lexical elements in the sentence. This type of retention could then be termed ‘associative retention’. Another instance of this type is found in 7.4.1.1 7.1.3
L2 speakers’ overgeneralisations
As mentioned in the Introduction, the results of the studies in Ch. 6 revealed the presence of retention in the synchronic assessments of groups of native speakers and L2 (Singaporean English) speakers of the use of would in wish-complement clauses. The extent to which retention was present was found to depend on contextual features such as stativity of the complement main predicate, animacy and agentivity of the complement subject, negation, and aspect. Some of the stimulus sentences presented as part of the study were the following, repeated in (6)–(8) for convenience: (6) (7) (8)
I wish I would come to your party I wish the Porsche would belong to me I wish there would be no fighting at the party
A summary of the averages of acceptance rates of (6–8) by the (L1) speakers in the studies in Ch. 6 and the Singaporean (L2) speakers is shown below in Table 7.1.
1. It seems possible for associative retention to occur also in word-forming processes unrelated to grammaticalisation. Lehrer (1996) discusses the lexical associations of former uses of affixes which adhere to the affix when it generalises to new words; e.g. -o/aholic from alcoholic retains the sense of ‘addiction’ associated with its original stem when applied to neologisms such as workaholic or chocaholic. This area of research is worthy of further investigation.
THE LEXICAL MEMORY TRACES HYPOTHESIS
Table 7.1
213
Average acceptances of would in wish-complements in 3 environments
Question
Native (L1) speakers
Singaporean (L2) speakers
(6) (7) (8)
03.7% 17%.0 29.3%
15%.0 36.3% 54.5%
(6): containing first person, cross-clause subject co-reference; (7): containing an inanimate subject with a stative predicate; (8): containing an existential construction with a formal subject.
It is recalled from Ch. 6 that although retention appears in the results of the studies to be present in the dialects of both L1 speakers and L2 speakers, for some of the L2 speakers, who allowed a greater acceptance of the modal would co-occurring with non-agentive subjects and stative verbs, the level of retention of volitional meanings associated with earlier diachronic uses of the modal was considerably lower in certain environments, and the extent of the generalisation of this feature was much greater. This is shown by the markedly higher percentages of acceptances by the L2 speakers for (6)–(8) in Table 7.1. The question posed in the Introduction to this chapter, regarding the interaction of synchronic linguistic intuitions with information of diachronic developments, is now re-examined in the light of L2 intuitions, and it is further questioned why the L2 speakers surveyed do not appear to have accessed this information quite so readily as the native speakers. The hypothesis presented in the present chapter hopes to provide an answer to this question. The most obvious answer might be that the speaker’s intuitions are guided by current usage in the community, which will naturally follow historical patterns, and these may vary in L2 communities. This is possible since the L2 statistics discussed above may indicate a dialectal trend, the L2 variety of English spoken in Singapore being most often acquired in childhood, via the medium of education. Children acquiring English in Singapore are now exposed to a nativised variety, not an L1 variety. As such it will bear witness to changes which have taken place at earlier stages of development and which will be passed on to succeeding generations of speakers, even if those developments reflect a more rapid rate of grammaticalisation of some features than that reflected in standardised dialects. Whether or not the L2 speakers’ intuitions are indicative of institutionalised developments, though, the question is why they vary so markedly from those of L1 speakers. One would expect that the historical patterns reflected in the
214
HYPOTHETICAL MODALITY
synchronic state of the language originally acquired from L1 speakers should have been passed on to the first generation of L2 speakers, and that the first generations of L2 speakers would have encountered the same patterns of current usage familiar to L1 speakers. However, this does not seem to have been the case, as the variation indicates, and the accumulation of historical meanings in the compositional value of a grammaticalising item does not appear to have always been transmitted to L2 speakers as part of the package of acquiring the language. It may be necessary, therefore, to seek an explanation for the differences in the characteristics of acquisition processes themselves.
7.2 Towards an explanation 7.2.1
General considerations
The representation of retention in the synchronic behaviour of speaker’s intuitions carries with it the paradoxical problems of reconciling both the diachronic nature of retention with its synchronic manifestations, as well as, or probably including, the problem of reconciling the nature of retention as statistical variation across a community with that of individual speakers’ use. Certainly, intuitive assessments are no true guide to patterns of use, and therefore cannot be an accurate measure of the rate of grammaticalisation of a feature, and, furthermore, L2 speakers may be able to articulate an acquired distributional rule while not exhibiting evidence of application of the rule in production. Intuitive data, therefore, may not always accurately reflect current trends in the language community. At the same time, the results of the studies in Ch. 6, showing a stage of statistical variation in the synchronic representation of retention, suggest that the contribution of the linguistic conventions in a community of speakers is unlikely to be totally reliable as an influence constraining intuitive assessments either. If it were, then the L2 dialect, originally exposed to L1 conventions of input in the transmitting stages, would be expected to develop along similar lines as an L1 dialect — Table 7.1 shows otherwise. There are nevertheless factors relating to grammaticalisation within the language community which must be considered in seeking explanations for the differences between L1 and L2 speakers in rates of retention in certain contexts. These include frequency of use and analogical matching. 7.2.1.1 Frequency of use It could be argued that the lower degree to which retention is seen to influence
THE LEXICAL MEMORY TRACES HYPOTHESIS
215
grammatical distribution in L2 speakers’ dialects is related to the frequency with which the item occurs in the dialect, and that this may be expected to be lower than for an L1 dialect. The consequences of lower frequency of use may be incomplete learning due to the possibility that the term does not occur often enough for accurate assessments of its distributional range to be made by the L2 speaker in acquisitional stages. There are at least two reasons why this cannot be the only explanation for the differences. First, diachronic processes of grammaticalisation show that the more frequently a grammaticalising item occurs, the less likely it will be affected by retention of older meanings. Generalisation of meaning has long been associated with frequency of use (Bybee forthc.; Meillet 1958: 135), as discussed in 6.1, and forms become less analysable with each recurrent use (Langacker 1987: 100). Retention is discussed above as often associated with more marginal uses of a form; for example, there are no apparent constraints on the modal would in main clause conditional apodoses, although, as illustrated in Ch. 6, this form does exhibit signs of retention in the subordinate clause complements of wish. The diachronic process appears to be the reverse for speakers of an L2 dialect, though, or else very much accelerated, as revealed in the statistical differences summarised in 7.1.3 in which a broader generalisation of meaning seems to be appearing in environments which are likely to occur with less frequency, i.e. wish-complement clauses. The reasons for the generalisation of meaning must be different in either situation, and thus there is no correlation for L2 speakers between generalisation of meaning and frequency of use in the diachronic processes of grammaticalisation. Second, the very postulation of a hypothesis that requires frequency of use to explain incomplete learning of the distributional constraints on a grammaticalising item presupposes the need to otherwise ‘teach’ such constraints, which may not necessarily be present as part of the processes of naturalistic acquisition of the language; that is, they are not acquired along with the grammatical functions of the form in question. While the co-occurrence constraints on lexical items may be part of the learning process for L1 and L2 learners alike, and overgeneralisations may occur in either case, the distributional constraints on a grammaticalising item may be seen not always to require conscious heed to such constraints for L1 learners. This is demonstrated in at least one study of the L1–acquisition of the ba-construction in Mandarin Chinese (H-T Cheung 1992), which is acquired without overgeneralisation to stative verb environments by Taiwanese children. However, adults acquiring Mandarin as an L2 need to be informed of the distributional constraints of such forms (see also 7.1.1.2). Thus, for the reasons outlined above — the correlation of higher frequency with broader
216
HYPOTHETICAL MODALITY
generalisation, and the apparent need of instructional input of distributional constraints — the possibility of lower frequency of use of a grammaticalising item in the L2 dialect cannot be held as the only explanation for the occurrence of lower degrees of retention. 7.2.1.2 Analogical ‘matching’ As noted above, it could be argued that a speaker’s intuitions were guided by patterns and conventions of current usage in the linguistic community, and such conventions will naturally follow historical developments. If the individual speaker’s intuitions reflecting retention were guided by patterns of current usage in the community, then the intuitive judgements of individuals would be based on a system of analogical ‘matching’ with information deposited in memory storage; that is, input of instances of single uses of a structure would leave traces in the memory, which would then provide the basic pattern for subsequent uses of the same form (Langacker 1987: 100). In this case, intuitions regarding later uses would be constantly matched by analogy to the former traces of usage events, and would be rejected if found not to be appropriate to the environmental conditions laid down in the first instance. However, the hypothesis of analogous matching, in terms of language acquisition, presupposes that certain unacceptable uses, being rejected, would leave no traces in memory storage. The initial memory installation would consist then only of instantiations of acceptable uses. It does not explain how only the acceptable uses arrive in memory storage in the first place, and how the first instances are judged to be acceptable. It may be suggested that the speaker’s initial assumptions are regulated by conventions of standardness in the speech community, but as is evident in the results of variational studies, forms which vary from standard norms of acceptability are as frequently-occurring as acceptable forms, and thus, there is no reason why the former should not leave traces in memory storage as well. Because of the assumption that inappropriate forms are rejected as input, the hypothesis also does little to account for diachronic language change across a community. In order for language change to take place, a formerly unacceptable usage, matched by analogy to established instantiations in memory storage and previously rejected, would have to change to become acceptable and installed in memory storage as well. It could not be determined at what stage and for what reasons this reversal would occur. It certainly could not happen as a sudden change, as Sweetser (1990: 9) notes, without an intervening period when a formerly unacceptable usage and an acceptable one coexist. The hypothesis is also inconsistent with what is known about the processes of grammaticalisation
THE LEXICAL MEMORY TRACES HYPOTHESIS
217
itself, and the semantic changes necessary for a grammaticalising entity to gradually extend its uses in an increasingly wider distributional range of environments. The process of memory storage of usage instantiations could only serve, then, as an ancillary mechanism with a synchronic application, not as a central repository for generating semantic judgements relevant to diachronic facts. Thus, it can be seen that possible factors involving (in)frequency of use or analogical matching to entrenched usage patterns cannot provide the most accurate explanation for the differences in rates of retention between L1 and L2 speakers demonstrated in their intuitive assessments of distributional constraints on a grammaticalising item. 7.2.1.3 Parallels in language development In order to resolve the way in which individual semantic judgements coincide with diachronic facts, and to thus explain the differences between L1 and L2 intuitions, it is necessary to consider the synchronic evidence as running parallel to evidence of a diachronic nature. Sweetser (1990) has used a similar analysis in showing that the synchronic semantic links between lexical domains are parallel to those in the diachrony of a language, and may help to elucidate stages of semantic change historically; similarly, evidence of diachronic routes of change is useful for making predictions regarding synchronic connections between domains (1990: 45–6). Macro-scale evidence of the historical development of a language across a community might be matched with micro-scale evidence of language evolution in the individual, and the two processes appear to correlate. Under this analysis, the differences in retention rates between L1 and L2 speakers must be related to differences in the acquisition process. Underlying such a proposal is the possibility of the recapitulation of historical evolutionary processes in the individual organism — that diachronic and ontogenetic processes may be seen to follow a parallel course of development. This viewpoint could lead to an assumption that evolutionary processes in the species provide the blue-print for the development of some aspects regarding the grammatical system in the individual. However, ontogenetic/diachronic parallels in grammaticalisation studies have been recently proposed by Slobin (1994) and Erbaugh (1986) (discussed below), who have given clear evidence that there is no implication that such parallels can directly represent the transferral of any kind of genetic encoding in the organism, and it is far beyond the scope of the present study to attempt to substantiate such a possibility. For the moment, it has been suggested that the similarities found in the sources and outcomes of ontogenetic/diachronic processes of grammaticalisation are purely coincidental, and should not be considered as mutually relevant in any way
218
HYPOTHETICAL MODALITY
(Slobin 1997). The remainder of the chapter will review some of the previous studies in this area, providing examples of the independent but parallel grammaticalisation processes in diachronic and individual language development, and of similarities between the source concepts of diachronic grammaticalisation and those of acquisitional grammaticalisation. The coincidence of intuitive accounts of retention with diachronic evidence will then be explained by means of a model, proposing the parallel representation of retention as a factor in ontogenetic development.
7.3 Developing the hypothesis: parallels in historical and individual development 7.3.1
Correlations in source concepts
It is only recently, and with some reservations, that grammaticalisation studies have been extended to look towards co-evolutionary parallels between the ontogenetic and the diachronic development of grammatical morphemes. One notable study is that of Slobin (1994) who observes that the grammaticalisation of the English present perfect in child acquisition is seen to correlate quite closely with the grammaticalisation of the present perfect historically; that is, that more ‘concrete’ and readily accessible notions underly the first uses of the present perfect in early childhood. The same parallels between ontogenetic and diachronic development are seen in the grammaticalisation of the classifier system in Mandarin Chinese. According to Erbaugh (1986: 399) the acquisition of the noun classifier system by Chinese children and the historical development of the noun classifier system from 1400 B. C. to the present share a common order of development, with specific items, concrete objects, and valued items being marked earlier than generalised sets, abstractions, and ordinary items. In both the ontogenetic and diachronic development of such forms, there is a progression of meaning shifts from the concrete to the abstract, with the increase in the range of grammatical applications of the grammaticalising form being accompanied by a decrease in lexical content. More examples of the parallel nature of ontogenetic and diachronic grammaticalisation will be given below (7.4.2.1–5), and some of these will relate to the examples of retention in 7.2.1.1–2, showing the corresponding paths of development of such features in historical and in acquisitional stages. Meanwhile, it is necessary to consider the nature of the source notions from which the developmental paths proceed. The earliest concepts to grammaticalise both in individual and historical
THE LEXICAL MEMORY TRACES HYPOTHESIS
219
language development are those which are most concrete and readily accessible. Such concepts can be referred to as grammaticisable notions (Slobin 1997) (referring to either diachronic or ontogenetic sources) or lexical source concepts (Heine, Claudi and Hünnemeyer 1991) (referring to diachronic grammaticalisation sources). Lexical source concepts in diachronic grammaticalisation have been discussed by Heine et al. (1991: 32–3) as usually derived from basic entities such as concrete objects, processes, or locations. They are also found to be associated typically with immediate and elementary human experiences, behaviour and thought, and are most often represented in lexemic form. Some of the most typical lexemes which codify source concepts include verbs meaning do/make, come, arrive, or leave, and those expressing states such as stay or exist, want or like (1991: 35). The source concepts which may provide the lexical source material for grammaticalisation processes may be shared across a variety of languages from diverse families, and in some cases, for a number of different grammaticalisation processes in the same language, while at the same time, different languages can use the same source for grammaticalisation in diverging directions. The elementary nature of lexical source concepts suggests that the range will not be extensive cross-linguistically; there is usually considered to be a limited pool of predictable sources for many of the most frequently-occurring grammatical morphemes (see, for example, Bybee et al. 1994). In acquisitional grammaticalisation as well, there is a tendency for basic categories to appear as the source material for the development of more complex structures. Slobin (1985) acknowledges the pre-existence of basic semantic values as a point from which a child’s mastery of grammatical functions proceeds. The fundamental determination of a semantic source for the generation of grammatical sentences had been previously introduced by Bickerton (1981) (as noted in Bowerman 1985: 1283), although earlier accounts such as these still bore traces of the generative grammarians’ mechanistic models and formulae, and made no attempt to relate the processes of language acquisition to other processes of human cognition (such attempts have been made more recently by Givón 1995, and will be discussed in [7.4.3]). Slobin (1985) discusses the presence of a Language Making Capacity and a Basic Child Grammar, considered to be a universal set of semantic values, which would at subsequent stages of acquisition develop into categories in accordance with the input language. In later accounts (Slobin 1997), the role of the input language was seen to figure more prominently in shaping the grammaticizable notions from initial stages of acquistion, and grammaticizable notions were regarded as contributing to the structuring of ways of thinking in such a way necessary for the generation of similar future codifications of the notions themselves. For example, a child
220
HYPOTHETICAL MODALITY
acquiring English learns to mark the concept of Path using independent particles such as in and out, and uses them freely for both caused and spontaneous events, but a child acquiring Korean must pay attention to distinctions of transitivity as the notion of Path is marked on the verb, and there are different Path expressions for caused and spontaneous events (Bowerman 1994: 42). The systematic features of the language itself are reinforced during the learning process, and the regularity of this form-function mapping then sets the template for subsequent use. This indicates that grammaticizable notions may be governed by the salience of form-meaning correspondences specific to the input language, which may vary considerably between languages. The influence of source structures in ontogenetic processes will be revealed in [7.3.2.1–5], and a number of utterances made by small children will reveal the retention of the semantically-based grammaticisable notions in the developing language which can be compared directly with evidence of similar source concepts in the diachronic situation. This data will also show that the evidence provided of parallels in diachronic and acquisitional grammaticalisation processes may be attributed quite feasibly to similarities between the grammaticisable source notions of diachronic and acquisitional processes. The situations to be discussed are modal verbs in English and other languages, the ba particle in Mandarin, the development of past tense marking, grammatical subjects, and space-to-time mapping. Although the last three situations are illustrated with errors from children learning English, the phenomena being referred to as grammaticalisation sources in each case could well serve a more general, crosslinguistic domain of reference. 7.3.2
Actual instances of co-evolutionary parallels, and evidence from children’s errors
7.3.2.1 Modal verbs The diachronic grammaticalisation of modal verbs has been discussed in Ch. 3 in association with the development of more abstract meanings from more concrete meanings, the concrete meanings being related to the initial functions of the modals as lexical verbs. As noted earlier, historically later, more abstract epistemic meanings are described as centred in the speaker’s rather than the subject’s domain of perception (Traugott 1995a), while earlier uses of modals have been specifically related to deontic or agent-oriented meanings such as obligation and permission (Bybee, Pagliuca and Perkins 1994). Correspondingly, it is deontic modality which has been most often associated with the early acquisition of modal forms by children. Shepherd (1982) compares the acquisition of
THE LEXICAL MEMORY TRACES HYPOTHESIS
221
modals by children acquiring American English with both child and adult uses of modals in Antiguan creole and notes that children acquiring L1 English and also those acquiring creole, as well as adult creole speakers, use a greater number of deontic modals than epistemic ones. Shepherd does not discount the possibility that children may be subject to a relatively high frequency of input of deontic notions (1982: 321 fn.); however, she also notes that a form such as mosa in Antiguan creole, glossed as ‘have to’, is used exclusively as an epistemic modal in adult language, in spite of the fact that it goes through a deontic stage in children’s acquisition of the form. This raises interesting prospects to consider with regard to the volitional meanings in would discussed above: would has been found to be rarely used as a root modal of past volition. Goossens (1995) reports 2% of uses in one corpus and no more than 4% in another, and in all the cases surveyed, it occurred only as a negative modal. Yet, clearly in the study discussed in Ch. 6 its constraints reflected a basic root sense of volition. Moreover, Goossen’s data seems to suggest that the semantic constraints are not likely to result from the input to memory storage of instances of previous adult usage events, since the form appears to be rarely used by adults in a volitional sense. Such evidence offers reason to propose that the semantic constraints reflect an independently ordered pattern of development, similar in both the acquisitional and the diachronic evolution of a form, and that is what is being called upon in exercises of intuitive judgement. The grammaticalisation of modal verbs in child language acquisition has also been reported by Stephany (1988), from studies of children acquiring British and American English, and is compared with data from studies of the acquistion of L1 Greek by children. It was shown in these studies that the earliest uses of modalised expressions in the acquisition of both languages were to indicate deontic meanings. Can, could and may were found to be associated with deontic, action-oriented possibility, and will, want to, going to, would, shall, have (got) to, must, should, had better, and ought to were used to express deontic necessity (1988: 390). In a report of the acquisition of British English modals, based on data by Wells (1979), Stephany finds that will, want to, going to, and would are volition-centred in the child’s first uses. It is interesting to note the child’s earliest uses of would are volition-centred, even though, as indicated above in the data surveyed by Goossens (1995) noted above, it could hardly be claimed that such meanings continue to predominate in adult usage. Will is also noted as initially expressing intention in child language (Stephany 1988: 392), although volitional meanings in adult usage most often seem to appear only ambiguously alongside predictive meanings (as discussed above 7.1.1). It could be inferred from these findings that children’s early uses of modal verbs indicate a proclivity
222
HYPOTHETICAL MODALITY
towards meanings which are most likely to be closely associated with the original lexical meanings of the verbs before grammaticalisation to modal auxiliaries begins, and that this could be a pre-determined stage through which a child passes en route towards the acquisition of more abstract and speaker-subjective meanings of modals. The correlations between historical and ontogenetic processes of development are thus clearly illustrated in the example of modal verbs, in English, Greek and Antiguan Creole. 7.3.2.2 Mandarin ba The effects of retention in the present-day meanings of the Mandarin particle ba have been discussed in 7.1.1.2, where it is noted that its development began in the fifth century B. C. from a source which functioned as a full lexical verb meaning ‘grasp’ or ‘take hold of’. By the time of the eighth century A. D. the verb had already started to show signs of grammaticalising to become an accusative case marker, as illustrated in the example of a Tang Dynasty poem, given by Li and Thompson (1976: 485), and cited in Slobin (1997): (9)
Zui ba zhu-gen-zi xi kan drunk ba dogwood careful look2
The progress of the grammaticalisation of ba is evident in (9), as the meaning of the ba is ambiguous between a full-verb interpretation, meaning ‘take’: ‘While drunk, (I) took the dogwood and looked at it’; and a grammaticalising accusative marker, so that the sentence can also translate as: ‘While drunk (I) carefully looked at the dogwood.’ Slobin notes that in Modern Mandarin, the form can no longer function as a full verb with the lexical meaning of ‘take’, but at the same time, the restrictions on its use to co-occurrence with definite objects and objects which undergo some physical effect (usually termed ‘disposal’ senses) indicate that it has not yet grammaticalised to become an object marker in all environments. Evidence from acquisitional studies of children learning L1 Mandarin indicates that the route of acquisition of ba may indeed follow a similar or parallel path of development as the diachronic route, with full-verb uses appearing at earlier stages as learning ‘errors’. In a study by Erbaugh (1992) three of the four Taibei children in the study used the particle as a full verb with the meaning assumed by Erbaugh to be ‘use’ or ‘do’. In another study by Jepson (1989), ba was used by Beijing children aged between 2;2 and 4;2 with the actual meaning of ‘take’; for example (1989: 204–5): 2. This example appears exactly as cited in the original given references; however, I have been advised by Alain Peyraube (p.c.) that the word for ‘dogwood’ should be zhu-yu-zi, not zhu-gen-zi.
THE LEXICAL MEMORY TRACES HYPOTHESIS
(10)
a.
b.
223
Wo ba yizi ‘I take chair’ (as child picks up chair and begins to carry it to another room) Ba jeige! ‘Take this’ (handing book to another child)
What is even more compelling is that, like the Antiguan creole modal verb mosa, discussed in 7.3.2.1, and, to some extent, the volitional meanings of would, the child’s uses of ba show evidence of the presence of archaic sources which are no longer used in the adult present-day language, and therefore cannot be part of the child’s input. Such uses cannot be explained without reference to the correspondence of similar source concepts in acquisitional and diachronic grammaticalisation. Furthermore, it could not be the case that a similar route of acquisition would be followed by L2 learners of Mandarin, who would acquire the co-verb as a grammaticalised element and require instruction as to its distributional limitations. The possible differences between the ontogenetic grammaticalisation of L1 and L2 learners are most clearly visible in the example of ba. 7.3.2.3 Past tense The example of retention revealed in Ch. 5 is illustrative of the presence of retained meanings of perfectivity which produce pragmatic inferences of irrealis when past tense is combined with a stative predicate, and may recall an earlier stage in the development of past tenses when only verbs with inherent lexical aspect of telicity or punctuality attracted the first meanings of deictic pasts. Bybee, Pagliuca and Perkins (1994: 54–5) distinguished perfectives from pasts in that perfectives view the situation as bounded temporally, while pasts simply indicate that a situation occurred before the moment of speech. Furthermore, as noted earlier, they show that perfectives interact with the lexical semantics of the predicate more than pasts do, perfectivity being entailed in the meaning of punctual and telic classes of verbs. This evidence led to the hypothesis that pasts are more grammaticalised than perfectives and are considered diachronically as further developments from perfectives. Thus, as perfectives developed into pasts, they gradually increased their range of applications to include stative verbs as well. Historically, there are records in Indo-European languages testifying to the hypothesis which trace the development of past tenses from the Proto-IndoEuropean aspects of perfect and aorist, both perfective aspects, but distinguished only by whether they were [+momentary, +punctual] (aorist), or [−momentary, −punctual] (perfect) (Lehmann 1974: 144). It is also believed that the aorist and the perfect later amalgated, as shown by a form in Rigveda, to produce a
224
HYPOTHETICAL MODALITY
meaning little different from the perfect in Latin and the preterite in Germanic (1974: 145). Tense was not a category of Proto-Indo-European, but the forms of the perfect and the aorist came to be used for past time when the shift to a tensebased system later took place. Thus, lexical aspect in historical sources appears to have played a prominent role in determining the developmental route of past tenses as a diachronic process. Lexical routes of development appear to be followed also by learners acquiring the past tense. It was discussed in Ch. 5 that the significance of lexical aspect in the acquisition of English past tenses by children was borne out in a study by Shirai and Andersen (1995), in which not only punctual lexical aspect, but also telicity and resultant state (marking perfectivity) were the characteristics of the first verbs to receive past tense marking in acquisitional stages. Shirai and Andersen attribute their findings to the operation of an analysis by the children in which the prototype of the category past is marked on verbs with [+telic], [+punctual], and [+result] characteristics. Thus, the prototypes of past tense and perfective aspect, which are very similar, provide the basis for future extensions of the category of past tense to less prototypical uses until the adult norms are eventually acquired (1995: 759). It will be recalled that similar accounts have been described of the acquisition of past tense by English speakers learning Spanish (Andersen 1991), and multilingual groups learning English (BardoviHarlig and Reynolds 1995), as discussed in Ch. 5. If past tenses developed out of perfectives, as suggested by Bybee et al. (1994), then there are close parallels in the developmental routes shown by both children and adults acquiring the category. Parallels in the historical and individual developmental routes of the past tense are also supported by evidence from children’s learning errors. An example of the correlation of similar lexical source concepts in acquisitional grammaticalisation and diachronic grammaticalisation can be found in the observation made Herring (1981), cited in Slobin (1985: 1183–4), of the acquisition of past tense by English-speaking children. A child of 2;7 years was reported to have produced (11)
Cowboy did fighting me
which was assumed by Slobin to be an attempt to mark both the process (with -ing) and the endpoint (with did) of a past action. The adult English form of past progressive marks only whether or not the past event was in progress; there is no specific marking for completion (although it is an implied meaning — see Ch. 5 for supporting evidence). The attempt by the child to combine marking for completion with progressivity may be evidence that past tense is grammaticalising in the child’s acquisition from its most prototypical uses — punctual,
THE LEXICAL MEMORY TRACES HYPOTHESIS
225
completed events — to the less prototypical uses — non-punctual actions in progress. (11) may thus be seen as an intermediate stage, showing the continuing adherence of lexically-oriented meanings most closely associated with prototypical notions of perfectivity in the use of a completive operator did rather than the stative was. It could also be conjectured that the child who uttered (11) was not yet aware of the possible pragmatic implicatures which may derive senses of ‘perfectivity’ when past tense is used with progressives, and was finding his own way of marking the combination of both perfectivity and progression. Whatever the motivations for the use of such a form, it is likely that it reflects the retention of the lexical content of the child’s early semantic concepts, in which perfectivity might have provided a lexical conceptual route for the development of past marking (as is predicted for the diachronic grammaticalisation of pasts by Bybee et al. 1994). If this is the case, it indicates that for both the ontogenetic development and diachronic routes of the grammaticalisation of pasts, the source concepts may be the same. 7.3.2.4 Grammatical subjects The merging of topic and agentive nominal to become subject in English is considered by Shibatani (1991) to be a clear instance of grammaticalisation, exactly as lexemic structures may grammaticalise to become abstract grammatical structures. The evidence of earlier discourse notions in the grammaticalisation of subject is found in the fact that English discourages the use of indefinites as subjects, which would be inappropriate to earlier topic functions, for example, There is a boy reading a book in the garden is preferred to A boy is reading a book in the garden (1991: 129). This example in itself constitutes a case of retention of the continuing presence of earlier topic functions from which subjects grammaticalised in English. It is also noted by Shibatani that the grammatical category of subject has been generalised in English much further than in other languages in which the occurrence of non-agentive nominals may be limited in subject position. Languages such as Spanish and Italian, for example, indicate the absence of agency associated with stative verbs of experiencing, such as like, for example, Me gusta la cerveza — (Sp.) ‘I like beer’, which expresses the experiencer in the dative rather than as a nominative subject as in English (1991: 101–2). Shibatani concludes that the category of subject is the grammaticalised result of the generalisation of agent over other semantic roles, and that this generalisation varies in extent from one language to another (1991: 103). Evidence that English subjects are more grammaticalised than those in
226
HYPOTHETICAL MODALITY
Spanish and Italian is apparent in the existence of ‘subjectless’ sentences in Old English which do possess subjects in present-day English. Traugott (1992: 208–9) notes that predicates referring to natural phenomena such as the weather, sensory or cognitive experience and mental experience such as hyngrian ‘be hungry’, sceamian ‘be ashamed’, and lician ‘please, like’, which semantically involved an animate experiencer and a stimulus, appeared without a lexical subject in Old English, indicating that the constraints of agentivity on the expression of a grammatical subject were much stronger than they are today. Because of the restriction of subject category in this way, experiencers were usually marked in the dative, as in the present-day Spanish and Italian examples discussed above. (Warner 1993: 170) provides the following example from Old English: (12)
him sceal sceamian ætforan gode ælmihtigum … ‘he [lit.: him (dative)] shall be-ashamed before God Almighty’ (ÆLS i.12.169)
The lexical meanings of agentivity, then, appear to be fundamental to the grammaticalisation of subjects in English and in other languages, and as illustrated in (12), agentivity must have been closely associated with the development of grammatical category of subject, as non-agentive nominals in pre-verbal position were distinguished by the dative or accusative marking in the historical development of the language. Agent was also amongst a number of semantic concepts first suggested by Schlesinger (1971) as being instrumental in forming relations basic to sentence production in children; such concepts included agent, action, object and location. It was also noted that in children’s first sentences, the expression of agent normally preceded the action performed, such that the two concepts were regarded as closely related conceptually (Bowerman 1994: 37). These ‘primitive’ semantic relations were found to influence children’s first sentence production in a number of varied languages.3 Cross-linguistic evidence of the primacy of agent/action relations comes from languages in which ergative marking appears first with past tense verbs in Kaluli (Schieffelin’s 1985 study cited in Slobin 1985: 1176–7). The first uses of agentive marking by children are shown in Kaluli to be prototypically associated with verbs which refer to a clear achievement or result: this is most likely due to the fact that evidence of an achieved
3. A counterargument is offered in Lieven et al. (1997), in which non-agentive verbs such as want, and see are produced by the child at the same age (1;8), in the first ten utterances of verb-object structures. Bowerman (1973: 208) notes that verbs such as want, see, and fall are the only exceptions to the agentive prototypes.
THE LEXICAL MEMORY TRACES HYPOTHESIS
227
action implies evidence of an agent’s intervention in its accomplishment. Thus, pragmatic reasons may account for the undergeneralisation of agentive marking by children. The association of subject with agent is also clear in the fact that young children cannot comprehend without difficulty English passive sentences and treat the syntactic subjects as though they were the agent rather than the patient or location (Johnston 1985: 988). This seems to indicate that the children are relying on word order cues to determine the agent of an action; however, word order cannot be the only means of determining agentivity, as is illustrated in Kaluli above. Evidence that agent selection is marked morphologically is found in the errors of young American children learning English, in a study which reveals that subject pronouns for self-reference used by children vary according to the type of verb with which they occur (Budwig 1989: 272–4): (13)
a. b.
My cracked the eggs I like my breakfast
Budwig noted that I tended to be used by the children in her study when they wanted to express internal states and intentions, or utterances expressing little agentivity. On the other hand, my was used for highly-agentive actions which expressed the child’s control over a situation, or desire to bring about a change. Thus, agentivity is seen as a distinguishing characteristic in the marking of nominals in subject position, just as in the diachronic development of grammatical subjects shown above. Again there are correlations between the historical situation and the individual acquisitional situation, though there is still no reason to postulate an interdependent relation between the two processes. 7.3.2.5 Space-to-time mapping A frequently-cited source concept for the development of grammatical meanings is the domain of spatial concepts. As well as being a source for diachronic grammaticalisation, spatial concepts appear evident in children’s early source concepts. Space has been noted as a source schema for the expression of temporality across languages, and in fact is shown to be contiguous with time in a continuum of metaphorical categories, illustrated in Heine et al. (1991: 55). The likelihood of spatial concepts serving as the source schemas for the development of temporal concepts in children is illustrated in the example of children’s utterances given by Bowerman (1985: 1292) in which the following overgeneralised uses were observed: (14)
Can I have any reading behind the dinner? (= ‘after’)
228
HYPOTHETICAL MODALITY
(15)
Do we have room before we go to bed for another reading? (= ‘time’)
According to Bowerman, spatial and temporal categories had already long been distinguished in the speech of the children who produced the utterances in (14) and (15). This would appear to weaken the assumption that what is happening is that the less abstract, lexical source of spatial relations serves as a template for the generation of temporal metaphors, suggesting rather a kind of haphazard vacillation between the concepts of space and time in the child’s language. However, although the overgeneralisations produced by the children in (14) and (15) do show evidence of a conflation of the notions of space and time, in both cases it is space which has been generalised to functions of temporal meaning, not the reverse. If there were also evidence of temporal concepts being overgeneralised to spatial environments, then one could quite feasibly suggest that the generalisations were indicative of a child’s cognitive dimension in which the notions of space and the notions of time were initially indistinguishable from one another. But without such evidence, it is possible only to suggest that the space to time meaning shifts are in accord with the unidirectionality of grammaticalisation processes in general, and that a child’s semantic basis for the development of temporal concepts is that of spatial concepts, as is maintained for the same development in diachronic studies (see above). The lexical sources for temporal meanings derive from spatial concepts, and the children’s errors in (14) and (15) clearly indicate retention from such lexical sources in early acquisitional stages, and the possibility that spatial notions provide the lexical source material for temporal notions, in acquisitional and in diachronic grammaticalisation as well. 7.3.3
Provisional conclusions
On the basis of the evidence given above, then, it is evident that there is at least some correlation between what Bowerman (1985) labels the ‘starting semantic space’ for the development of children’s grammars, and the source concepts and propositions that are described by Heine et al. (1991) as the cognitive bases from which the processes of grammaticalisation are initiated. The examples reveal retention in the developing grammar, manifested in the child’s speech as errors rather than constraints as reflected the in adult intuitions discussed in 7.1.1.1–2. The correlations may be coincidental, as implied by Slobin (1997) in that they use similar structures (basic words and concepts or basic-level categories Lakoff 1987: 13 and immediate and highly accessible notions) for the grammaticalisation of more complex categories, but they cannot be interdependent. The continued development of fundamental semantic concepts requires the abstraction and all
THE LEXICAL MEMORY TRACES HYPOTHESIS
229
the mechanistics of cognitive and highly-developed pragmatic inferencing procedures usually only associated with more advanced stages of grammaticalisation, and children cannot be considered as contributing to such development (Slobin 1994). The maturity of the language in the developing individual, as well as in the history of the society in which it is spoken, results in the production of efficient pragmatic devices for the extension of basic ideas and concrete meanings; this is a factor of economy in the language itself. However, it could not be denied that the initial sources for the development of grammaticalising features happen to show a striking conformity in acquisitional and diachronic accounts. It must be emphasised, though, that the comparisons between ontogenetic evidence and the diachronic evidence of early historical records are not drawn on the basis of similar text material.4 In comparing, say, the child’s first uses of the present perfect as a perfect of result with Old English telic verbs as the most frequent verb type used to express the present perfect, the common features observed in the two situations are a similarity of function, regardless of the differences in context of use, speaker and style, which naturally cannot be compared over such a time scale, and under diverse conditions. Slobin concedes that the development of research into the historical treatment of grammaticalisation is invaluable in helping to explain the nature and origins of grammaticizable notions, but stresses that the processes of historical grammaticalisation are quite different from those of acquisitional grammaticalisation, in spite of having a superficial similarity (1997: 301–2). The superficial parallels between the two processes centre around the functional similarity of the source concepts from which grammaticalisation proceeds in either case: the sources are ‘basic words’ in both cases of grammaticalisation — but for children these are acquired first because they are frequently used and easy to learn, whereas the reason such concepts appear at the beginning of diachronic routes of grammaticalisation is because they happen to be also frequent and general enough to provide the basis for the development of more specific inferences (1997: 304). The parallels drawn in the preceding sections have also been made on the evidence of correlated stages of development, which by all accounts do bear a remarkable similarity. Whatever the means by which these stages are reached in either parallel situation, it is clearly the case that children in acquisition reflect and recapitulate, within a highly accelerated time scale, the patterns of historical change in an entire speech community. However, the similarity in the stages must eventually be attributed to the similarity in the starting notions and source
4. I am grateful to Keith Allan for drawing my attention to this point.
230
HYPOTHETICAL MODALITY
concepts from which the processes develop in the first instance: given a small range of simple source structures, there is only a limited number of directions in which further development may inevitably proceed in either individual or diachronic processes. Thus, the reasons for the correlation of similar stages of development are not complex.
7.4 A theoretical account of retention 7.4.1
Retention as a product of acquisition
In 7.3.2.4 there was shown evidence of the retention of basic notions of agentivity as a grammaticisable source in individual grammatical development, as children’s errors distinguished agentive from non-agentive subjects in early acquisitional stages. Schlesinger (1988), building on the notions of semantic relations proposed by Bowerman, proposes that a process of semantic assimilation in acquisitional stages allows the agent-action relation to gradually develop into the subject-predicate relation. Schlesinger’s account is supported by a later study (Schlesinger 1992), in which he provides empirical evidence from groups of adult native speakers of English to support his earlier claims. On the basis of his analysis of agent as a cluster concept, he investigates the degree to which notions of Control and Intention, members of the cluster, are accorded to the subjects of verbs of mental experience and cognition, firstly in sentences in which the subject normally acts as an experiencer; for example (Schlesinger 1992: 316): (16)
John admires Paul,
and secondly in sentences in which the experiencer is the object of the verb and the subject acts as the stimulus; e.g.: (17)
John impresses Paul
Although the subject in (17) can be described as possessing one of Schlesinger’s (1992) features for agentivity, Cause, and possibly Control, but not necessarily the feature Intention, in all the investigations, it was found that the subject of the verb was accorded a higher degree of Control or Intention than the object, even when the subject is only the stimulus of the mental experience. In one experiment, this situation was found to hold even when the subject was an inanimate object, as in (18) (ibid.: 326): (18)
The food delighted Mary
THE LEXICAL MEMORY TRACES HYPOTHESIS
231
On the basis of these results, Schlesinger concluded that the grammatical category of subject is found to contain the semantic residue of prototypical functions of agents, and that such properties originate in the first uses of the subject category by children: “the formal categories of the adult linguistic system retain some of the ‘flavor’ of the semantic notions out of which they developed.” (1992: 329). This retention was represented in the ratings of the sentence subjects by the informants as having more Control and Intention than the objects. As such, it represents an instance of associative retention (iii). What is interesting about Schlesinger’s study is that it revealed the presence of retention without using evidence of constraints on grammatical functions to isolate the retained senses; this indicates that although retention is most visible in semantic constraints, it is an often imperceptible element in the compositional value of any grammaticalising feature, at least until ‘bleaching’ has taken place. The study also provides compelling evidence of a psycholinguistic nature to support the hypothesis that the semantic intuitions of adult individuals, reflecting the retention of diachronic senses in the compositional value of a grammaticalising expression, can be attributed to the development of grammatical meaning from lexically-related sources in childhood acquisitional processes. It is proposed therefore that the retention of lexically-oriented diachronic meanings in grammaticalising forms, represented in the psychology of the individual as constraints on semantic intuitions, is the result and the product of acquisitional processes and is related to the grammaticalisation of pre-grammatical cognitive schemas in the individual, so that the continuing influence of earlier stages of language acquisition may well result in such a phenomenon. The synchronic representation of retention, therefore, can be identified and correlated with earlier stages in the development of the grammaticalising feature diachronically. 7.4.2
The Lexical Memory Traces Model
Rather than memory traces of the isolated usage instantiations discussed above in 7.2.1.2 as ‘analogical matching’, memory traces of pre-grammatical stages of acquisition providing the semantic bases for further linguistic development can now be referred to; these may be described as Lexical Memory Traces (LMTs), and are regarded as a grammaticalised constituent of the compositional value of a grammatical morpheme. As noted in 7.1.2 it is in some cases useful to discuss the continuing influence of the selectional restrictions formerly associated with the lexical origins of the grammaticalising expression, when the constraints imposed by such restrictions on the LMTs at later stages of development will reveal evidence of their continuing presence. However, as Schlesinger’s (1992)
232
HYPOTHETICAL MODALITY
study reveals in 7.4.1, retention may be observed and quantified in the perceptions of present-day speakers, and correlated with acquisitional stages without the need to refer to the constraints laid down by selectional restrictions. For this reason, the use of a description such as the Lexical Memory Traces model provides may be a more comprehensive means of accounting for the phenomenon of retention in its various manifestations. A schematic diagram of the diachronic process by which retention appears in the compositional value is represented in Figure 7.1.
Lexical sources
Figure 7.1
LMT
→
Retention Retention
Grammaticalising form
Representational model of the development of retention as an ontogenetic/diachronic phenomenon in the compositional value of the grammaticalising expression
Figure 7.1 shows that the realisation of retention appears as Lexical Memory Traces in the grammaticalising form, the shaded area indicating the areas of overlapping influence in the transferral of lexical meaning from the Lexical Sources to the Grammaticalising Form. In Figure 7.2, the model is applied to illustrate the retention of volitional senses relating to the grammaticalisation of the hypothetical modal would, as discussed in Ch. 6.
THE LEXICAL MEMORY TRACES HYPOTHESIS
Lexical sources of volition in the modal would
Figure 7.2
LMTs of volition
→
Retention Retention
233
Grammaticalising senses of predictive would in affected environments
Representational model of the retention of volitional senses in the grammaticalising predictive meanings of the modal WOULD in affected environments
The shaded area in Figure 7.2 represents an environmental distribution bias of LMTs as part of the larger paradigm in which would is used, represented by the entire right-hand area. Examples of the environments affected by the retention of volitional senses in hypothetical would include those discussed such as stative predicates, negation, and inanimate and co-referential subjects. The environments most clearly affected occur in examples such as (6) — I wish I would come to your party; and others such as I wish the Porsche would belong to me (7); it is in such environments that the developing predictive senses are constrained by the retention of volitional senses as part of the compositional value of the meaning of the modal. The model proposed suggests that the psychological force of the acquisitional patterning laid down in childhood and forming in semantic memory as Lexical Memory Traces may run directly parallel to the phenomenon of the retention of earlier meanings in the diachronic development of a language and is often revealed as a constraint on the grammatical distribution of a form synchronically. It is important to realise that, despite their similarities, the two parallel processes are not interdependent but are independent of each other, and bear similarities simply because the cause and the motivation for grammaticalisation processes are the same in either case. The motivations for both courses of grammaticalisation come as no mystery, as discussed by Slobin (1997); simple cognitive structures will always form the basis for more complex ones, by nature of human reasoning and inference processing. It must be remembered though, that grammaticalisation in actual time, however, works in a complementary manner with the pragmatic force of the inferences and mechanisms of grammaticalisation in historical time, and is tempered at any given time by current sociocultural needs and language conventions. Conventions, in turn, interact with the frequency of acceptable instantiations in memory storage, but, as discussed
234
HYPOTHETICAL MODALITY
in 7.2.1.2, such an account on its own is insufficient to explain the nature of retention. Perhaps it may be suggested at this stage, then, that it is the constraints of LMTs and static memory storage instantiations reflecting conventions which together combine to influence the progress of a grammaticalising expression. The examples illustrated in 7.3.2.1–5 demonstrate the way in which a given lexical source concept can appear as part of the initial semantic space from which children’s acquisitional grammaticalisation proceeds, or as the source of grammaticalised structures across languages and the origin of the development of grammatical morphology across time. The types of source concepts may be initially realised as complete lexemes or identifiable only as a prototypical lexical property of grammatical categories (e.g. agentivity), but both types comprise an intrinsic element in the compositional value of a grammaticalising expression. In some cases, semantic constraints on the distribution of a feature will be invoked by the presence of retention in the compositional value in certain environments. Retention, however, appears as a synchronic phenomenon, while at the same time, it reflects processes of change which have taken place over sometimes hundreds of years, and the correlation of synchronic evidence of retention with historical evidence of the development of a language still raises questions of the interdependency of the two parallel courses. The preceding sections have illuminated a number of instances giving evidence to show the correlation of similar lexical origins in co-evolutionary parallel. The next section relates this evidence to a recent account of the neurological nature of language development. 7.4.3
Integrating the model into a theory: neurological order of development
Givón (1995: 391–5) discusses the evolution of language as concomitant with the development of learning, craft-making, and cultural transmission of knowledge. He maintains that the first cycle of symbolisation to emerge was a codified lexicon, followed by a second cycle of grammatical development, and he goes on to discuss the presence of a module in the human communicative system known as the conceptual lexicon, which is the accumulation of conventionalised experiences known as concepts. This has been labelled by psychologists as the ‘permanent semantic memory’, and it is the conceptual lexicon, he claims, that is first acquired by children. In addition, pre-grammatical pidgin communication also makes use of the lexicon earlier than the grammar — this pattern is also followed by learners of a second language in a naturalistic setting (1995: 400). Evidence for the possibility that the lexicon emerges first in the acquisition of both first and second language is found in the iconic, non-abstract and less symbolic nature of the rules of pidgin grammar, and its heavy reliance on the
THE LEXICAL MEMORY TRACES HYPOTHESIS
235
vocabulary as a means of pre-grammatical communication (1995: 402). The likelihood of this order of acquisition — both ontogenetically and phylogenetically — is strengthened by the evidence of studies of a neurological nature: Givón cites Greenfield in concluding that the area of the brain known to be used for grammatical processing, Broca’s area, is simply the evolutionary extension of the primary motor cortex which was used for co-ordinated motor skills of a sequential-hierarchic kind (1995: 420). The semantic areas discovered in the prefrontal area of the brain, as well as Wernike’s area, are all adjacent and wellconnected to the grammar-processing area. Givón emphasises the significance of these discoveries, since grammar has been found to be highly affected by lexical semantics; for example, features of generic classification of grammatical categories include concreteness, animacy, humanity, and agency (as have been discussed above). These findings are further related to the fact that lexical sources provide the diachronic derivation of grammatical morphemes in grammaticalisation (ibid.). In the development of the individual, the maturity of the area known as Broca’s area is correlated with the age at which grammar begins to develop in the child (approximately 2 years) (Givón 1995: 420). Evidence to indicate that the development of grammar in the child runs parallel to an evolutionary process is suggested by the fact that non-human primates, though capable of acquiring a limited range of lexical code-labels for nouns, verbs, and adjectives, cannot be taught to use any kind of morphological or syntactic system (1995: 400). The inability of chimpanzees to acquire a grammatical system is explained by Bates and Goodman (1997) as due to the fact that there are limitations on the animals’ lexical development — they cannot acquire more than 200–300 words. Previous studies have shown that children with vocabularies under 300 words have extremely limited grammatical development; this confirms that the mechanism for acquiring semantic language is developed at an earlier stage on the evolutionary scale than that for acquiring grammar, and that non-human primates are still at a pre-grammatical stage of development. More iconic, concrete representations of experiences are codified before the more arbitrary, abstract and symbolic representations at later stages, both in the evolution of the language capacity as a phylogenetic phenomenon as well as in ontogenetic history (Givón 1995: 441). However, such parallel processes, as noted above in 7.3.3, are linked only by a common motivation for development, and are not causally related in any way. The evidence is in place to further develop the explanation for the phenomenon of retention in future studies, both as a diachronic and a synchronic, psychological feature of language use. Clearly there are parallels between first language acquisitional grammaticalisation and the diachronic processes which
236
HYPOTHETICAL MODALITY
refer to the traditional use of the term — this evidence cannot be denied. We may also come closer now to explaining with neurological accounts why semantic intuitions bear witness to earlier, diachronic developments in the history of the language itself, even though there are no apparent relations of causality. The interaction between the semantic processing area and the grammar processing area in early stages of acquisition might appear to be significant to the retention of knowledge of the lexical origins of grammatical morphemes, especially given that the acquisition of semantic knowledge of language is both ontogenetically and phylogenetically prior to the development of grammatical knowledge. It remains open to future investigation to provide a neurological justification showing that the interaction of the lexical processing area with the grammatical processing area extends beyond early acquisitional stages in the individual. Later investigation might involve the application of psychological experimental techniques such as priming.5 In order to complete such tasks, though, a study much broader in scope than the present one would be required.
7.5 Applying the hypothesis to the results of the studies 7.5.1
Lexical Memory Traces in past tense stative verbs
The Lexical Memory Traces model hypothesises nothing more than the potential for lexical meanings representing earlier stages of the diachronic development of grammaticalising entities to be correlated with lexical meanings representing earlier stages of acquisition in childhood. The perception of retention by the adult speaker is related to stages of childhood acquisition which recapitulate historical stages of development. The LMT Hypothesis can apply equally to each of the situations outlined above (i-iii) as examples of retention in grammaticalising features. In the case of (iii) then, retention had the effect of producing implicatures of ‘perfective states’. In the extension of past tenses to stative verbs, the retention of perfectivity from earlier diachronic uses of past tenses continues to adhere to such environments, and this produces a pragmatic meaning of counterfactuality with regard to the continuation of the state into the present. It is important to note, though, that the sense of perfectivity represents only a semantic continuity from older meanings, not a morphological continuity as in (i) and (ii).
5. Acknowledgement is due to Talmy Givón (personal communication) for this helpful suggestion.
THE LEXICAL MEMORY TRACES HYPOTHESIS
237
As far as the L1 speakers were concerned, the results showing the interpretation of a counterfactual implicature in such environments were quite clear — an average of 67.5% of counterfactual interpretations was the overall result for all the environments surveyed (see Ch. 5). However, the L2 speakers showed an average of only 31.2% across all the environments, less than half that of the L1 speakers. Given that the presence of the counterfactual implicatures is due to a retention factor associated with past tense in stative environments, one would believe that the same reasons that were causing the lower rates of retention in the study in Ch. 6 applied to the study on counterfactual implicatures. That is, that the absence of retention was due to the by-passing of lexical stages in early acquisition, so that the L2 speaker is likely to experience an accelerated rate of grammaticalisation and bleaching occurs at a much faster rate than for L1 speakers. However, this cannot be the case for the L2 speakers in the study in Ch. 5, for two reasons. First, the data reveal that, as the education and age of the L2 speaker increases, the rate of retention increases concomitantly, so that younger speakers will have lower rates of retention of perfectivity than older, more educated speakers; e.g. for Question 1, the implicature of counterfactuality for the youngest (Secondary Year 10) group was shown by 46.8% of the group, while for the same environment, the second-year university students rated around 70.5% of counterfactual interpretations. If retention is a product of acquisition, then one would not expect to find more evidence of its presence in older speakers. Rather, one would assume that, with maturity and more exposure to the grammaticalised conventions of the language, retentive influences would diminish. However, this could not possibly be so, and it suggests that the situation of retention shown in the L2 speakers’ data in Ch. 5 is quite different from that expressed in Ch. 6 in the grammaticalisation of would. The second reason relates to the nature of the retention itself — perfective marking, and the dialectal patterns exhibited in Singaporean English of past tense marking. As noted in Ch. 5, apart from the phonological features of consonant clusters associated with weak verbs, lexical aspect plays a key role in determining which verbs are selected for past tense marking in colloquial varieties of Singaporean English, so that punctual verbs are marked more frequently for past than stative verbs and non-punctual verbs such as those expressing habitual or durative aspect (Ho and Platt 1993). Thus, within the present stage of development of the colloquial subvariety, stative environments are not selected for past tense marking in any case. According to Mufwene (1996: 9 fn.), grammatical strategies in contact languages are often modified to suit the new systems into which they are integrated, rather than being transferred intact. If this is the case with past tense
238
HYPOTHETICAL MODALITY
marking in Singaporean English, the use of past tense has been modified to accommodate the semantic restrictions of the substratum languages (e.g. Chinese and Malay). These restrictions are still present in the dialects of successive generations of speakers, and continue to interfere with the morphological marking of the current L2 variety of English. Furthermore, Hopper and Traugott (1993: 3) note that the co-presence of a grammaticalising feature and its lexical counterpart still functioning in a language may inhibit the grammaticalisation of that feature, resulting in slower development. If the contact languages are not as grammaticalised as the L2 with regard to certain features, they are also likely to play an inhibiting role on the development of those features. In such cases, then, the presence of retention is not yet relevant, as the grammaticalisation of past tense in such varieties has not yet extended to all stative and non-punctual environments. Thus, although the use of forms such as was/were may indicate to the L2 speakers that a state was in existence before the moment of utterance, semantic interference from local contact and substratum languages may inhibit the pragmatic interpretation that the past state has terminated before the moment of utterance. This is a very different pattern from that which describes the presence of retention in the modal would in the study in Ch. 6. It is also the reason that the absence of retention can be said in Ch. 5 to be due to undergeneralisation, or what could be termed ‘hypo-grammaticalisation’, rather than overgeneralisation, which could be described as ‘hyper-grammaticalisation’. 7.5.2
By-passing the lexical stages of acquisition
The situation of the lower rates of retention in past statives therefore indicates that the category of past tense amongst the L2 speakers surveyed is still affected by aspectual constraints, while in the study of the grammaticalisation of hypothetical would, the modal has grammaticalised further for such speakers in the environments investigated than in the L1 dialect. The LMT Hypothesis shows that the generalisation is accounted for by the possibility that the first generations of L2 speakers have either by-passed or passed more rapidly through the earlier stages of acquisition in which lexical sources of the grammaticalising features were relevant, and therefore subsequent generations will exhibit an accelerated rate of grammaticalisation of such features. This is not to suggest that the children of L2 speakers, acquiring the form from their parents, do not pass briefly through an early stage of acquisition when the lexical senses of the form are most relevant. If the form is already bleached in the parent dialect, though, then the children will naturally proceed to the bleached meanings with more
THE LEXICAL MEMORY TRACES HYPOTHESIS
239
rapidity than if it is not. The same situation exists in L1 dialects for forms which are already bleached of traces of older lexical meanings. However, this also raises the question that children are creating anew, out of their own competence, lexical meanings which are not necessarily present in adult input. There is evidence in a few rare examples in the children’s errors discussed in 7.3.2.1–5 that this is actually happening: the case of mosa in Antiguan creole has been discussed, as well as Mandarin ba, and the predominance of root modals amongst children acquiring English may be compared with corpus data showing a much lower frequency of such senses in adult use (7.3.2.1). In a recent study of the acquisition of ba in Taiwanese Mandarin (Ziegeler, ms.), it was discovered that some children were relexifying the grammatical nature of the adult input, possibly due to use in particularly frequent distributional contexts such as imperatives directed at the child, when ba is sentence initial and can be categorised in analogy with other verbs in the same sentence position. This evidence does not account for the cases of mosa or the English modals; however, it does explain one way in which the child assimilates more grammaticalised adult input in conformity with a conceptually-concrete world-view. A parallel situation to the study of hypothetical would appears to be found in the grammaticalisation of going to futures in Nova Scotia Black English (Poplack and Tagliamonte 1995). Mufwene (1996: 8) comments that the grammaticalisation of gon(na) in AAVE (Afro-American Vernacular English) indicates that it is at a stage which is more advanced than the lexifier, since it allows copula deletion. He also discusses Poplack and Tagliamonte’s (1995) study, noting that there are semantic constraints on the use of going to in Nova Scotia English, such as animacy of the subject and immediate future meanings, which are not adhered to by black speakers living in the same speech community, who use the future form indiscriminately (Mufwene 1996: 10). It is interesting to note that there are differences between black and white speakers in the same community — this lends support to the claims made earlier that exposure to the constraints of standardised uses bears little relation to the presence of retentive constraints in an L2 community. Mufwene suggests that these examples are indications of a greater extent of grammaticalisation of the future form in contact varieties, and that they are extensions of developments already in progress in the lexifier languages (ibid.). The situation in Nova Scotia Black English may be compared with the evidence of more rapid grammaticalisation of would for Singaporean L2 speakers, in which accelerated grammaticalisation patterns are attributed to either a bypassing or a reduction of the time-span of earlier lexical stages in the transmission of the language from its L1 sources. That is, the accelerated rates of
240
HYPOTHETICAL MODALITY
grammaticalisation are concomitant with lower rates of retention. Thus, just as the former lexical meanings of volition in would appear to have been circumvented to some extent by L2 learners, the first generations of black speakers of English in Nova Scotia must have acquired the future form going to by passing through the lexical stages of acquisition with more rapidity. The lexical stages in acquisition would parallel earlier diachronic stages in the development of the form; that is, when the use of the form applied only to animate subjects capable of movement. The situation presents a case which is directly parallel to the case of would in hypothetical complement clauses: it does not represent so much the retarded progress of L1 dialects, but rather a situation of accelerated progress in L2 dialects. Subsequent generations of speakers would then acquire the broader distributional range of a grammaticalising item of the previous generation, even though they are likely, for whatever reason, to pass through a brief stage at which the item was used mainly in lexically-oriented functions. In this way, the accelerated patterns of grammaticalisation associated with the first generation L2 speakers are passed on intact to the next generation. 7.5.3
Reasons for the differences in rates of grammaticalisation
Heine and Reh (1984: 89–90) discuss the evolution of grammaticalising features in pidgins, suggesting that developments that may take up to a millenium in natural languages can be accomplished within three generations in pidgin languages. Although the L2 variety discussed in the present study is not a pidgin (as noted in Ch. 4), the processes of grammaticalisation in such a variety may be compared, given the preceding data. Most significant to the study of grammaticalisation in such varieties is the fact that (as also noted above by Mufwene 1996), pidginisation appears to anticipate developments which are destined to occur in the lexifier language at later stages (1996: 87), and such developments appear as “a time-acceleration device in language evolution” (Heine and Reh 1984: 87). It may be questioned why this is so. In the case of the modal would investigated in the study in Ch. 6, the retention is manifest as a paradigmatic bias, that is, would is grammaticalised in some environments but not in others, in which it is still constrained by its lexical meanings. This means that the range of environments overall in which the modal is grammaticalised is more reduced for L1 speakers than for L2 speakers. Lexical Memory Traces of earlier stages of acquisition are intuitively accessed by speakers in encountering the modal in such environments, but to a lesser degree by L2 speakers, since the modal is more grammaticalised in their dialects. The accelerated grammaticalisation of the features concerned would appear to be the
THE LEXICAL MEMORY TRACES HYPOTHESIS
241
result of overgeneralisations occurring in the transmission of the language to the first generation of speakers. The overgeneralisations, in turn, could be brought about by a common motivation for greater communicative needs (termed Communicative Pressure (CP) by Hagège 1993: 130) amongst speakers who do not have the L2 variety as a mother tongue. In the grammaticalisation of L1 varieties, a number of forms will be present at the same time to express a single function (and also a number of functions may be expressed by a single form). This is discussed as layering (Hopper 1991), and it indicates in many ways a formal redundancy as there may be multiple forms for some functions. L2 speakers first acquiring the language may possess a restricted inventory of forms available for expressing the functions required relative to L1 speakers (so that layering is less likely for such speakers) and will overextend the functions of existing resources under Communicative Pressure. This would account for overgeneralisation, hypergrammaticalisation, and more rapid rates of generalisation to new grammatical environments amongst L2 speakers. 7.5.4
LMTs as prototype extensions
There are clear differences shown between the types of retention listed in 7.1.2 and their manifestations in present-day uses of grammaticalising forms. The types of retention described in the distributional constraints on the use of would in hypothetical wish-complements, and in the use of going to futures in the Nova Scotia example are instances of (i) listed in 7.1.2, in which the overgeneralisations of L2 speakers reveal environments which have not yet been affected by grammaticalisation in L1 dialects, and so create semantic anomaly. Ambiguity, or polysemy, is described in (ii) as a stage at which the form is grammaticalising in an environment as an inference from older meanings. The retention of perfectivity, however, discussed in Ch. 5 as producing counterfactual implicatures in stative predicates, is from a prototype extension, behaving in the same way as the retention of agentivity in grammatical subjects discussed in 7.4.1. Both types can be described as associative retention, and are revealed as present in an implicit meaning in the grammaticalising structure, even though there may be no immediate effects in the grammatical environment. For this reason it may be questioned whether such cases as the retention of perfectivity in stative verbs described in (iii) or the retention of agentive senses in grammatical subjects (7.4.1) are genuine cases of retention at all. The present study claims that they are, while they are expressed in a different form from the other cases described in (i) and (ii). What makes type (iii) different from (i) and (ii) is that in (i) and (ii) the retention is perceptible as a form which is clearly
242
HYPOTHETICAL MODALITY
identifiable with an earlier morphological representation as an individual lexeme, while in (iii) the diachronic origins of the lexical source morpheme are more opaque, and may, in fact, have less relevance to the current state of retention than the presence of other lexical information in the environment with which the grammaticalising item most protoypically co-occurs, and which ‘rubs off’ onto it through frequency of use. However, this is not to suggest that the historical sources for such categories are not important to this study; data from other languages is useful in providing a comparative framework from which to hypothesise the possible developments taking place in the early history (and prehistory) of English. What is interesting is that the ontogenetic development of the language in the individual still suggests parallels with historical sources, whether the grammaticalisation is of a lexemic form or not. Lakoff (1987: 65) has discussed the centrality of the members of prototypical agents and prototypical topics to the characterisation of the category of subject, and evidence for their prototypicality is given in 7.4.1. Similarly, Shirai and Andersen (1995) derive their analyses from the prototype theory of Rosch (1973 and 1978), which states that the prototype of a cognitive category describes the best exemplars of that category, and other category members are related gradiantly by the features they share with the prototype. Shirai and Andersen apply the theory to language acquisition — suggesting that the prototypical exemplars of past tense are acquired first before the more peripheral ones. Lakoff (1987: 65) suggests that the category of grammatical subject is a radial category, and that peripheral members of that category are motivated by family resemblances to the central core. He uses the term first used by Wittgenstein (1953) and later adopted by Rosch and Mervis (1975) to describe the links in the relationship between prototypes and peripheral members of a category. Heine (1992) has also discussed family resemblances with regard to grammaticalisation chains. The relevance of prototype theory to examples of grammaticalisation is readily perceived in grammatical subjects and the category of past tense, and both of these examples can be related to diachronically earlier meanings (which are discussed in 7.3.2.3–4). Thus, there is justification for describing the prototype senses which prevail in some grammatical categories as retention from earlier stages of development. The formal differences shown in respect to the different types of retention may thus be open to question, and the type of retention defined in (iii) and illustrated in 7.4.1 as the result of extension from a prototype meaning may instead be related to the simple processes of analogy. But this would still imply retention as long as the analogy was brought about as a grammaticalisation
THE LEXICAL MEMORY TRACES HYPOTHESIS
243
process. According to Heine et al. (1991: 33), lexical sources for grammaticalisation processes ultimately are encoded as lexemes, although there are also a number of basic propositions from which grammaticalisation processes may proceed (1991: 36). If the source concepts for grammaticalisation are ultimately lexemic in form, it may be questioned why the retention shown in (iii) is not identifiable with a lexemic source. It could be conjectured that there was at a very much earlier stage of the ancestor language, a lexemic source from which the development of grammatical categories such as perfective aspect and subject marking proceeded. However, in the present-day uses of English past tense and grammatical subjects, lexemic sources are not transparent morphologically, and the glimmers of lexical properties associated with their uses in particular environments are only implicit. Grammaticalisation has been defined in the present study (Ch. 3) as not only the development of grammatical status from lexical forms, but also the development of more grammatical status from an erstwhile grammaticalised form (Lehmann 1995 [1982]: 11). It would seem that if the meanings of past tenses and grammatical subjects were grammaticalised in the same way as other categories from ultimately lexical forms, then they are now grammaticalised but developing to achieve still more grammatical status. As such, the earlier, lexical meanings will not appear as constraints in any environment across a paradigm, since there are no identifiable lexemic sources present. The LMTs, therefore, appear to be present in the grammaticalising forms only as inferential extensions from a semantic prototype: perfective meanings extend by inference to nonprototypical, stative environments (as suggested by C. Smith 1991: 221), and the category of grammatical subject extends to include non-prototypical, nonagentive nominals. As inferences, the LMTs are pragmatically, not semantically derived, and in this way allow for new environments to be grammaticalised.
7.6 Concluding remarks Retention, as shown in the present model, appears as much a product of the ontogenetic development of grammatical morphemes as it is of their diachronic development. It is often represented in the individual as semantic constraints on grammatical functions and, from the evidence provided in the present study, appears to be reflective of earlier stages of language development in the child, when concrete, lexically-oriented concepts formed the basis for communication and grammatical structures were more iconically encoded on the lexical apparatus. For this reason, differences in the acquisitional processes may well result in
244
HYPOTHETICAL MODALITY
differences in the rates at which the Lexical Memory Traces are retained in the semantic composition of a grammaticalising item. Hence, the overgeneralisations of L2 speakers discussed in 7.1.2 and 7.5 may be indicative of acquisitional processes which have either by-passed or significantly reduced the earlier lexical stages of naturalistic learning; that is, the route of acquisition might be found to more directly lead to the bleached, grammaticalised meanings, suggesting that the L2 speaker is less likely to be constrained by the LMTs of earlier stages of acquisition that affect native speakers’ judgements. Future investigation in the direction of L2 studies might reveal interesting patterns amongst separate generations of speakers in L2 communities, also indicating that grammaticalisation rates are more accelerated for such speakers. In earlier chapters, it was seen that retention occurs as Lexical Memory Traces in the grammaticalising modal would in hypothetical complement clauses, and also in the pragmatic meanings of perfectivity in stative verbs. Such examples show that retention may be identified in more than one type, and the types are listed in 7.1.2 as (i)–(iii). The LMT Hypothesis also proposed that the differences between L2 speakers and L1 native speakers in the presence of retention in the features under investigation was attributed to differences in acquisition. The fact that the L2 was not a mother tongue for many of the L2 speakers surveyed was shown to lead to a need for more rapid grammaticalisation of some functions which do not have alternative means of expression in the dialect; hence there is overgeneralisation or ‘hyper-grammaticalisation’ of features which would normally occur at later stages in the input language in any case. In the grammaticalisation of past tenses to stative environments investigated in Ch. 5, the evidence of retention as irrealis implicatures is less likely to be present amongst the Singaporean L2 groups due to the interference of contact features and substratum languages in which past tense is not marked in stative and non-punctual enviroments (this is borne out by the variability of marking for past in Ho and Platt’s 1993 data). The case of retention cannot be compared to that in the study of hypothetical would for the reason that the grammaticalisation of past has not extended beyond punctual environments for many of the speakers surveyed. The LMTs of former lexical senses are not operating for the L2 speakers in such environments as the data suggests that the inferences necessary for the extension of the prototypical past tense meanings have not yet been formed by many of the subjects. The influence of retention in the two studies, therefore, is different in either situation. The degree to which the LMTs are held to maintain influence over the developing grammaticalised forms, though potentially for a long period of time, cannot be construed as fixed indefinitely. The prevailing forces of language
THE LEXICAL MEMORY TRACES HYPOTHESIS
245
change in the society, variation in function from discourse pragmatic factors, and social factors such as pressure from language contact in the community or dialect convergence and other external influences are likely to provide countering effects in a trade-off between the generalisation of a grammaticalising expression in an individual’s environment and the likelihood of its acceptance and adoption in the individual’s dialect. Such external factors are likely, in the long run, to weaken considerably the constraints imposed by LMTs on grammaticalising features. For this reason, so often, intuitive judgements and the retentive constraints which influence them should be regarded merely as a guide indicating usage patterns, not a de jure representation of the grammaticalisation of a feature in any language community.
C 8 Conclusions
The broader aims of the present study were initially to provide a theoretical explanation for a large amount of variation in the expression of counterfactual and hypothetical modality in an L2 variety of English, but in addition, it was intended to demonstrate more concisely the ways in which the grammatical meaning of hypothetical modality may be described within the context of recent theories of grammaticalisation. In working towards these two broad aims, it was also seen that a more specific hypothesis could be developed to describe and explain the ways in which L1 and L2 phenomena differ in terms of grammaticalisation processes. The following summary explains how this was achieved. Earlier studies in Chinese counterfactuals, which assumed a deterministic role for grammatical morphology, maintained that the presence or absence of explicit counterfactual marking in the grammar of a language may be directly linked to the presence or absence of cognitive schemas for counterfactual reasoning; such studies cannot be regarded with any circumspection. In the first place, it is shown in the present study that explict formal devices indicating hypothetical or counterfactual modality are difficult to isolate crosslinguistically, and even languages which encode the past tense of a stative predictive verb for hypothetical modality only pragmatically implicate such meanings by a combination of Quantity 2 implicatures and discourse features. Furthermore, grammatical morphology is only one member of a cluster of features which together may contribute to an overall sense of counterfactuality or hypotheticality, as shown in Ch. 2. The perception of such implicatures by L2 speakers may vary due to factors such as interference from semantic constraints in substratum and contact languages. However, the perception of counterfactuality through implicatures is unrelated to the ability to construct cognitive schemas for counterfactual reasoning, which is independent of language, and in the context of a counterfactual utterance, may be the result of extra-linguistic factors brought to bear upon the speech situation. The linguistic means of expressing counterfactual and hypothetical ideas might also vary from one language to another, and there is no cause to
248
HYPOTHETICAL MODALITY
assume that the conventions used to encode such ideas in one language may be the same as those in another language. The present study has shown that L1/L2 differences in the perception of English hypothetical meanings may vary according to the relative salience of the linguistic means of expressing them. The use of grammaticalisation theory to explain such variation has been invaluable as it reveals that present-day meanings of modality are derived diachronically by the increasing conventionalisation of pragmatic implicatures relating to source meanings. Hypothetical modality is a meaning which is implied in the use of a past tense modal form, and English has exploited such means to create hypothetical meanings. The development of hypothetical modal meanings in this way has been concomitant with the grammaticalisation of epistemic meanings in modal verbs, and the meanings have developed gradually over the history of the language. It is quite possible that variation in the expression or interpretation of hypothetical modality in present-day varieties of English may be attributable to synchronic, dialectal tendencies. However, given that the dialects under investigation are contrasted as L2 and L1 varieties, the tendencies shown are likely to reflect either anticipated developments in the future grammaticalisation of such forms (in the case of the L2 data) or evidence of diachronically earlier meanings (more obvious in the L1 data). By comparing the data from L1 and L2 sources, both these phenomena were observable at one synchronic crosssection in time, as shown in Ch. 6. The grammaticalisation of meanings in an L2 variety highlights the importance of continuity in the historical processes of grammaticalisation, and two opposing factors emerge from the comparative studies. First, it is demonstrated that in languages with a discontinuous historical development and a break in the transmission of the language from one generation to the next, there will also occur a break in the grammaticalisation processes of certain features of that language. Thus: Ch. 6 shows the relatively rapid grammaticalisation of the modal would as evidenced by its general acceptance amongst the L2 speakers surveyed, progressing to become grammaticalised in environments which in L1 varieties are still affected to a greater extent by the meanings of older uses of the modal. Second, the continuity of grammaticalisation processes historically is delayed in L2 varieties by interference from semantic constraints relating to substratum and contact languages. This is illustrated in the lower incidence of counterfactual implicatures derived from the past tense of stative predicates (see Ch. 5). The result of such delays is the interchangeable use of both past and present tense forms of the modal which appear to be in free variation, regardless of the modality of the environment. In these two respects, the study demonstrates that the continuity of historical processes in grammaticalisation may be interrupted in
CONCLUSIONS
249
the transmission to L2 dialects, and significant questions are raised regarding the interaction of the individual’s acquisitional processes within the diachronic grammaticalisation processes of the speech community. In Chapter 7 a hypothesis was proposed to answer such questions. The hypothesis claimed that the continuity in the historical processes of grammaticalisation is maintained in the acquisition of grammaticalising features in a language as they are passed on from one generation to the next, so that each subsequent generation recapitulates the historical processes in acquisition, but that the parallel processes are not interdependent. Because of such parallels, the distributional constraints affecting grammaticalising features in new environments are available to L1 speakers as a result of the route of acquisition undertaken, which reflects, independently, the historical stages. Speakers of L2 varieties for whom the language is not a mother tongue, however, are likely to advance more rapidly to the bleached stage in acquisitional grammaticalisation due to the increased pressure to ensure maximum automation of communication with a minimum inventory of available terms. This results in overgeneralisation (‘hypergrammaticalisation’) and a break in the continuity of the grammaticalisation of certain features as they are passed on from an L1 variety to an L2 variety. The hypothesis is discussed further below. It could be argued that the historical development of grammaticalising morphemes is of little relevance to the synchronic situations as they are presented in previous chapters, given that the acquisitional routes of children are not interdependent with, but only parallel to, the historical routes. If the present study had used only L1 data, then this argument would need to be addressed. However, the fact that differences in the rates of retention are shown to occur between L1 and L2 groups is sufficient to raise the question of consistency in acquisition routes in either case, and to prompt a need for further investigation in such areas. The possibility of finding cases (such as Mandarin Chinese ba: Ch. 7), in which parallels may be made with historical lexical sources without the lexical functions being represented in adult input, provide significant and exciting prospects for any theory of acquisition or language development as well. The notion of an association of pragmatic implicatures with counterfactual and hypothetical meaning was introduced in Ch. 2. It was concluded, following Comrie (1986), that the grammatical expression of counterfactuality in conditional clauses was derived partly from the causal link between the protasis and the apodosis, which was, in fact, found to be conversationally implied and hence defeasible. The use of past tense as an indication of temporal remoteness to metaphorically express factual remoteness in hypothetical clauses was also discussed, and it was concluded, following Dahl (1997), that it was not simply
250
HYPOTHETICAL MODALITY
past tense, but past tense in combination with imperfective aspect which provided the historical justification for the development of hypothetical meanings in Romance languages. The use of past tense to express factual remoteness is found to be cross-linguistically significant, as shown in Fleischman (1989 and 1995) and James (1982), but this may only be due to the fact that the past is usually known; language therefore exploits such presuppositions for the pragmatic extension of basic ideas and to refer to concepts which otherwise would not be possible to encode linguistically. Past is therefore associated more with realis concepts than irrealis, and the example of past as realis provides the grounds to invoke counter-hypotheses from factually-specific premises; this was basis for the establishment of a principle to describe counterfactual implicatures (the CFI Principle). Under the assumption that the hypotheticality of an irrealis expression is increased by the use of past or perfect morphology, it is concluded that any linguistic features present which increase the potential for forming realis concepts, will, in turn, increase the potential to derive a corresponding counterfactual interpretation, given the appropriate grammatical context. This means that counterfactual concepts are only those which can be abstracted from conceptually-specific premises, and features such as negation and first person subjects were added to the cluster of specificity pre-conditions which also includes past and perfect morphology. The result is that counterfactuality can be predicted to a certain degree, based on the hearer’s evaluation of the speaker’s knowledge of facts to the contrary of what is uttered. The predictability can be increased by the strengthening of the information specificity of the utterance via such linguistic tools as are found in Ch. 2 (29). The diachronic evolution of counterfactual and hypothetical meanings associated with the modal verb would was examined in Ch. 3, in a corpus study, and it was found that the grammaticalisation of the modal could be described as as much a case of emergent grammaticalisation as one of diachronic grammaticalisation. There do not appear to exist clear, well-defined stages in which the modal may be classed as representing either lexical, grammatical, or ambiguous functions, but instead, it appears to demonstrate both lexical (volitional) and grammatical (predictive) functions from as early as Old English times. The interpretation of the modal is dependent upon the discourse context for its grammaticalisation: in factive clauses which contain a presupposition of the truth of the proposition expressed, would clearly demonstrates a predictive (past future) function. In non-factive clauses, the function is either hypothetical or ambiguous with volitional meanings, according to the presence of other factors such as subject person and animacy, negation, and stativity of the complement verb. If there is a diachronic form of grammaticalisation taking place, it is
CONCLUSIONS
251
illustrated in the extension of the grammatical uses to main verbs, the frequency of which increases suddenly in the Early Modern English data, most probably due to the increase in distribution to areas formerly occupied by recessive subjunctive inflections. This kind of grammaticalisation offers an example of an increase in grammaticality, rather than one of a steady series of changes in the evolution of a former lexical item. The association of hypothetical meanings with past imperfectives was explored further in Ch. 5 in a study of the counterfactual implicatures resulting from the combination of past tense with stative predicates. Two types of stative verbs (be and love), a past habitual (used to), and a past progressive were used to elicit responses of counterfactuality from selected groups of speakers of Singaporean English, whose responses were compared with those of a control group of native speakers undertaking the same task. In colloquial use, past tense morphology in Singaporean English encodes mainly perfective aspect, use with imperfective predicates being less grammaticalised than in L1 varieties, and therefore the meanings associated with the combination of past tense and nonpunctual lexical aspect were often not available in the first place to create implicatures of counterfactuality in past tense stative verbs. Furthermore, it was considered that the implicatures of counterfactuality were the result of semantic retention of perfectivity from earlier stages in the development of past tenses, and it was found, accordingly, that the speakers of Singaporean English did not exhibit a high frequency of counterfactual implicatures. The inferences suggesting termination of the past state prior to the moment of speaking, revealed in the meanings of past stative and non-punctual predicates for L1 speakers, were not so frequently available to L2 speakers. The reasons were considered to be interference factors from contact and substratum languages in which perfectivity is not marked on stative predicates. Other factors contributing to the creation of hypothetical inferences include the use of a future marker in a past temporal context. As noted by Bybee, Pagliuca and Perkins (1991), futures are often derived from volition verbs crosslinguistically. Ch. 6 described a study in which the retention of older senses of volition in would was found in the hypothetical complements of the verb wish. For many of the subjects surveyed, the degree of retention was evident in the acceptablity judgements of a number of sentences. Acceptability ratings varied for each sentence, but it was concluded that the highest levels of acceptability were attributed to sentences in which both the older meanings of volition associated with would and new meanings of prediction could co-exist in the same environment; for example, She wishes her parents would believe her about the party (85% acceptable). This was considered to be due to the fact that newer,
252
HYPOTHETICAL MODALITY
grammaticalising meanings tend to develop out of older meanings in which they can be implied, and once grammaticalised in such environments, later spread by a process of analogy or generalisation to environments in which they cannot initially be implied (hence there were lower scores for those environments in which volitional meanings were impossible, suggesting that the grammaticalisation of would in such clauses is still at an early stage). Thus, the observation of the grammaticalisation of would in hypothetical complement clauses, using the intuitive data of a selected sample of respondents, illustrates with a synchronic cross-section one stage of paradigmatic variation in the broader processes of diachronic development, and suggests that for some environments in which would is used, the grammaticalisation to predictive meanings is still in progress. A number of reasons were proposed to account for this. The investigation was also undertaken using Singaporean speakers of English (Ch. 6: 2), as the data which motivated the survey originally came from Singaporean subjects (see Ziegeler 1994). The initial data revealed an overgeneralisation of would in such environments, and the subsequent survey undertaken with a group of Singaporean L2 speakers tested sentences similar to those tested with the native speakers. Acceptability judgements varied according to the nature of the sentence, but in general, the rates of acceptance were lower than for the native speakers surveyed (30.7% for the Singaporean group, and 42% for the Australian group, on average). For some of the sentences, however, there was as much as a 42% difference in the acceptability ratings between the two groups (e.g. the progressive: I wish you would be coming to my party). Given that the progressive is an environment highly indicative of epistemic meanings in modal verbs and of predictive meanings in the grammaticalisation of future will, it was considered that this was an indicator that the grammaticalisation of would in such environments was at a stage more advanced than for the native speaker group surveyed. Such differences in the rates of grammaticalisation provided the basis for the hypothesis in Ch. 7. However, the original data came from relatively small numbers of informants, and the trends shown may only have been representative of regional variation, and not related in particular to more general trends found in L1 and L2 communities and relating to grammmaticalisation processes. For this reason, the data obtained from the comparative study in Chapter 6 was further tested against the data from a control group of speakers of British English. The results from the British speakers showed that L1 acceptability evaluations, on average, were still somewhat higher than those of L2 groups, even the groups of L2 speakers living in an L1 speech community in Britain. There was also a noticeable diffference between the evaluations of older and more educated L1 speakers and younger L1
CONCLUSIONS
253
speakers, the older speakers being more conservative in judgement. This was felt to be related to sociolinguistic reasons: that younger speakers, especially those surveyed who were of adolescent age, were often the innovators of change across a speech community. It is important to remember that extra-linguistic factors relating to the current synchronic context of the grammaticalising item will often have an affect on the rate of grammaticalisation in a community. The higher levels of acceptance of the modal would in hypothetical complement clauses by L2 speakers was then considered to be the result of differences in the rates of grammaticalisation, and demonstrated that features in L2 dialects tend to grammaticalise faster than in native speakers’ dialects. The reasons for the different rates of grammaticalisation were explained by the Lexical Memory Traces Hypothesis, whereby diachronic information regarding earlier stages of development of a grammaticalising feature was correlated with information retained in memory from earlier stages in the acquisition of the form by the individual. In this way retention was seen as an ontogenetic factor as well as a diachronic factor. The means by which such correlations could explain the differences in the L2 data were considered to be due to differences in the acquisition of the grammaticalising feature by L2 speakers — factors such as Communicative Pressure (Hagège 1993) motivates the need to express a wider range of functions with a smaller inventory of forms in the dialects of speakers who are acquiring a language which is not their mother tongue, and in a situation in which it is used as a lingua franca. The Hypothesis predicts that such speakers might progress in acquisition more rapidly to grammaticalised forms with fewer distributional restrictions, or even by-pass earlier stages of acquisition which reflect the lexical origins of the grammaticalising form. The range of distribution of a grammatical item is much broader than it is for a lexical item; this reason also contributes to the accelerated rates of grammaticalisation in pidgins and creoles. The Hypothesis in its application to L2 data, does, however, leave one or two questions unanswered. It may be presumed, for instance, that the Communicative Pressure which is responsible for the accelerated rates of grammaticalisation in L2 varieties may indeed also apply to L1 learning situations. That is, a child with a limited inventory of forms to express grammatical functions should just as easily overextend the uses of available forms under Communicative Pressure, resulting in faster grammaticalisation of such features in acquisition. While the situations may be different in the two cases, and certainly the learning processes are different, it remains for future investigation to examine L1–L2 studies of acquisition, comparing the degree to which grammaticalising
254
HYPOTHETICAL MODALITY
features have been overgeneralised. In the case of child acquisitional grammaticalisation, the constraints imposed by the use of a grammaticalising form in its former lexical context are set against the child’s communicative needs which may predict a similar increase in grammatical distribution. The present study suggests that a child will not overgeneralise forms which are acquired initially with more concrete, lexical functions and are still constrained in early stages by the co-occurrence conditions associated with them. However, in the case of L1 acquisition, if there were overgeneralisation, an explanation would be needed to account for the speaker reverting to conventional restricted usage at later (adult) stages. This could provide a fruitful area for further research. Further studies should also be directed to cases similar to that of the acquisition of the Mandarin particle ba, which for L1 speakers of Mandarin, appears to go through an acquisitional stage in which it is used as a full lexical verb, reflecting its historical origins (see 7.3.2.2). This is not a use derived from adult input, in which it is grammaticalising as an object marker. Though L1 speakers of Mandarin may have an intuitive knowledge of its distributional restrictions based on its earlier use as a verb meaning ‘take’, an L2 learner of Mandarin must acquire such knowledge via instruction and rules, usually referring to terms such as ‘disposal’ sense to elucidate its functional limitations (see, e.g., Li and Thompson 1989). As noted in Ch. 7, recent research (Ziegeler ms.) indicates that ba may be acquired initially as a verb because of the child’s ability to analogise its function in frequently used input contexts, such as imperatives. However, other examples such as the modal mosa in Antiguan Creole would require independent explanation. Such differences in acquisition reflect the need for more intensive research into grammaticalisation as phenomenon of individual language development. Finally, the study leaves open the question of the rate of grammaticalisation of would, and the evidence of retention of lexical senses in some environments (hypothetical complements) but not in others (conditional apodoses). Clearly the functions of would are many and varied, and each function needs to be considered in isolation. If there were ever a stage at which lexical and hypothetical meanings could be found as ambiguously co-existing throughout an entire paradigm of uses, it is not clearly demonstrated in every grammatical environment. The matter is complex, and requiring of much further attention in diachronic research. In conclusion then, the study raises a number of points of enquiry for future investigation. It has presented a panchronic perspective on grammaticalisation studies, using diachronic explanations to explain variation in synchronic linguistic behaviour, demonstrating the role of retention in accounting for semantic
CONCLUSIONS
255
anomaly, and relating retention with ontogenetic factors of individual acquisition. However, it remains for future enquiry to test the acquisitional routes of grammaticalisation in L2 situations and to determine whether there are differences between naturalistic and instructional routes of acquisition of certain features. It is also hoped that further efforts might attempt to verify the persistence of retention from lexical sources as a neuro-cognitive phenomenon prevailing beyond the early pre-grammatical stages of acquisition in children, and to apply findings to other areas such as word-forming processes. The present study is, furthermore, limited to the examination of synchronic anomaly in grammaticalising hypothetical modal functions; more research using different features is needed in order to justify the range of applications the LMT Hypothesis may have. For these reasons and others outlined above, it is hoped that the study will offer an attractive invitation for further exciting research in the field.
Appendices
Appendix 1: Chapter 6 Questionnaire Sex:
Date and place of birth:
Please evaluate the following sentences, determining whether they seem:
or
Acceptable — (A) Questionable — (Q) Very Questionable (VQ) Unacceptable — (U)
1) I wish you would be tall. 2) I wish you would be coming to my party. 3) I wish I would come to your party. 4) I wish Pat wouldn’t come to the party. 5) I wish there would be no fighting at the party. 6) I wish the Porsche would belong to me. 7) She wishes her parents would believe her about the party.
For the sentences you have marked (U) or (VQ), can you suggest any changes that might make them more acceptable?
258
APPENDICES
Appendix 2: Chapter 6 Control Questionnaire Sex:
Date and place of birth:
Please evaluate the following sentences, determining whether they seem:
or
Acceptable — (A) Questionable — (Q) Very Questionable (VQ) Unacceptable — (U)
1) This is done better than what the other side could do. 2) I insist that she changes her plans for the tour. 3) There were a number of errors on the official form. 4) None of you have any idea what I mean. 5) If you have 8 items or less, you can join this queue. 6) This book is a good read.
For the sentences you have marked (U) or (VQ), can you suggest any changes that might make them more acceptable?
Bibliography
Selected bibliography Abraham, Werner. 1992. The aspectual source of the epistemic-root distinction of modal verbs. Paper presented at the Symposium on Mood and Modality, University of New Mexico, Albuquerque, May 8–10, 1992. Ahlqvist, Anders (ed.). 1982. Papers from the Fifth International Conference on Historical Linguistics. Amsterdam and Philadelphia: John Benjamins. Akatsuka, Norika, and Clancy, Patricia M. 1993. “Conditionality and deontic modality in Japanese and Korean: Evidence from the emergence of conditionals”. In Japanese/Korean Linguistics, P. M. Clancy (ed.), 177–192. Stanford: Centre for Language and Information. Alieva, Natalia F. 1992. “Problems of tense in general linguistics and the temporal system in Malay/Indonesian”. In Language and the Perception of Time and Space: Proceedings of the International Conference on Language and the Perception of Time and Space, 174–182. Organised by The Language Centre, University of Malaya, Kuala Lumpar. Kuala Lumpar: Dicetak Oleh City Reprographic. Allan, Keith. 1986. Linguistic Meaning, Vols I and II. London: Routledge and Kegan Paul. Allan, Keith. (forthc.). Natural Language Semantics. Oxford: Blackwell. Alleton, Viviane. 1994. “Some remarks about the epistemic values of auxiliary verbs yinggai and yao in Mandarin Chinese”. In In Honor of William S-Y Wang. Interdisciplinary Studies on Language and Language Change, Matthew Y. Chen and Ovid J. L. Tzeng (eds.), 1–16. Taipei: Pyramid Press. Andersen, Roger W. 1991. “Developmental sequences: the emergence of aspect marking in second language acquisition”. In Crosscurrents in Second Language Acquisition and Linguistic Theories, T. Huebner and C. A. Ferguson (eds.), 305–324. Amsterdam and Philadelphia: John Benjamins.
260
BIBLIOGRAPHY
Antinucci, Francesco, and Parisi, Domenico. 1971. “On English modal verbs”. In Papers from the Seventh Regional Meeting of the Chicago Linguistic Society 28–39. Asher, R. E. 1985. Tamil [Croon Helm Descriptive Grammars]. London: Croon Helm. Athanasiadou, Angeliki and Dirven, René (eds.). 1997. On Conditionals Again. Amsterdam and Philadelphia: John Benjamins. Atlas, Jay D. and Levinson, Stephen C. 1981. “It-clefts, informativeness, and logical form: Radical pragmatics. In Radical Pragmatics, P. Cole (ed.)., 1–61. New York: Academic Press. Au, Terry K-F. 1983. “Chinese and English counterfactuals: The Sapir-Whorf hypothesis revisited”. Cognition 15: 155–187. Baker, Philip. and Syea, Anand (eds.). 1996. Changing Meanings, Changing Functions. London: University of Westminster Press. Bao, Zhiming. 1995. “Already in Singapore English”. World Englishes 14 (2): 181–188. Bardovi-Harlig, Kathleen, and Reynolds, Dudley W. 1995. “The role of lexical aspect in the acquisition of tense and aspect”. Tesol Quarterly 29 (1): 107–129. Bates, Elizabeth, and Goodman, Judith C. 1997. “On the inseparability of grammar and the lexicon: Evidence from acquisition, aphasia, and real-time processing”. Language and Cognitive Processes 12: 507–584. Bickerton, Derek. 1981. Roots of Language. Ann Arbor: Karoma. Binnick, Robert. I. 1991. Time and the Verb. A Guide to Tense and Aspect. New York: Oxford University Press. Bloom, Alfred H. 1981. The Linguistic Shaping of Thought: A Study in the Impact of Language on Thinking in China and the West. Hillsdale, N. J.: Lawrence Erlbaum. Bowerman, Melissa. 1973. Structural relationships in children’s utterances: syntactic or semantic? In Cognitive Development and the Acquisition of Language, T. Moore (ed.), 197–213. New York: Academic Press. Bowerman, Melissa. 1985. “What shapes children’s grammars?”. In Dan. I. Slobin (ed.), Vol. 2, 1257–1319. Bowerman, Melissa. 1994. “From universal to language-specific in early grammatical development”. Philosophical Transactions of the Royal Society of London B 346: 37–45. Brinton, Laurel J. 1988. The Development of English Aspectual Systems. Cambridge: Cambridge University Press.
BIBLIOGRAPHY
261
Brown, Adam. 1992. Making Sense of Singapore English. Singapore: Federal Publications. Budwig, Nancy. 1989. “The linguistic marking of agentivity and control in child language”. Journal of Child Language 16: 263–284. Burnley, David. J. 1992. The History of the English Language: A Source Book. London: Longman. Bybee, Joan. 1985. Morphology. A Study of the Relation between Meaning and Form. Amsterdam and Philadelphia: John Benjamins. Bybee, Joan. 1994. “The grammaticalization of zero. Asymmetries in tense and aspect systems”. In William Pagliuca (ed.), 235–254. Bybee, Joan. 1995. “The semantic development of past tense modals in English”. In Bybee and Fleischman (eds.), 503–517. Bybee, Joan. 1998. “ ‘Irrealis’ as a grammatical category”. Anthropological Linguistics 40 (2): 257–271. Bybee, Joan. forthc. “The role of repetition in grammaticization: can in English”. To appear in Handbook of Historical Linguistics, Richard Janda and Brian Joseph (eds.), Oxford: Blackwell. Bybee, Joan and Dahl, Östen. 1989. “The creation of tense and aspect systems in the languages of the world”. Studies in Language 13 (1): 51–103. Bybee, Joan, and Fleischman, Suzanne (eds). 1995. Modality in Grammar and Discourse. Amsterdam and Philadelphia: John Benjamins. Bybee, Joan, and Pagliuca, William. 1985. “Cross-linguistic comparison and the development of grammatical meaning”. In Historical Semantics and Historical Word-Formation, J. Fisiak (ed.), 59–83. Berlin: Mouton de Gruyter. Bybee, Joan, and Pagliuca, William. 1987. “The evolution of future meaning”. In Papers from the Seventh International Conference on Historical Linguistics, A. G. Ramat, O. Carruba, and G. Bernini (eds.), 109–122. Amsterdam and Philadelphia: John Benjamins. Bybee, Joan, Pagliuca, William, and Perkins, Revere D. 1991. “Back to the future”. In Traugott and Heine (eds.), Vol. 2, 17–58. Bybee, Joan, Pagliuca William, and Revere D. Perkins. 1994. The Evolution of Grammar: Tense, Aspect and Modality in the Languages of the World. Chicago: University of Chicago Press. Carey, Kathleen. 1994. “The grammaticalisation of the perfect in Old English”. In Pagliuca (ed.), 103–117. Chafe, Wallace. 1988. “Extending the frontiers of grammaticalization”. Paper presented at the Symposium on Grammaticalization, University of Oregon, Eugene, 12–15 May.
262
BIBLIOGRAPHY
Chafe, Wallace. 1995. “The realis-irrealis distinction in Caddo, the Northern Iroquoian languages, and English”. In Bybee and Fleischman (eds.), 349–365. Chappell, Hilary. (ms.) “Conditionals and counterfactuals in Mandarin and other Sinitic languages”. Chao, Yuen Ren. 1968. A Grammar of Spoken Chinese. Berkeley and Los Angeles: University of California Press. Chao, Yuen Ren. 1976. Aspects of Chinese Sociolinguistics. Stanford: Stanford University Press. Cheng, P. W. 1985. “Pictures of ghosts: A critique of Alfred Bloom’s The Linguistic Shaping of Thought. American Anthropologist 87: 917–922. Cheng, Robert. 1978. “Modality in Taiwanese”. Paper presented at the Annual Meeting of the Modern Languages Association, New York, Dec. 29, 1978. Compiled in Tai Yu Min Nan Yu. Yufa Lunwen Xuanji 2. (Collected Papers on the Grammar of Taiwanese Southern Min 2.). Institute of Linguistics, , Hsinchu, Taiwan. Cheng, Robert. 1997. “A comparison of Taiwanese, Taiwanese Mandarin and Peking Mandarin”. In Tai, Huayu de Jiechu Xing Tongyi Yu de Hudong (Contacts between Taiwanese and Mandarin and Restructuring of their Synonyms), R. Cheng (ed.), 27–62. Taipei: Yuan-Liou Publishing Co. Cheung, Hin-tat. 1992. The Acquisition of ba in Mandarin. Unpublished Ph.d dissertation. University of Kansas. Chiang Ker Chiu. 1963. Progressive Hokkien (Amoy) Readers. (Book One). Singapore: The Chin Fen Book Store. Chung, Sandra, and Alan Timberlake. 1985. “Tense, aspect and mood”. In Language Typology and Syntactic Description. Vol. 3., T. Shopen (ed.), 202–258. Cambridge: Cambridge University Press. Coates, Jennifer. 1983. The Semantics of Modal Auxiliaries. London: Croon Helm. Coates, Jennifer. 1995. “The expression of root and epistemic possibility in English”. In Bybee and Fleischman (eds.), 55–66. Coates, Jennifer and Geoffrey Leech. 1980. “The meanings of modals in modern British and American English”. York Papers in Linguistics 8: 23–34. Comrie, Bernard. 1976. Aspect. Cambridge: Cambridge University Press. Comrie, Bernard. 1982. “Future time reference in conditional protases”. Australian Journal of Linguistics 2: 143–52. Comrie, Bernard. 1986. “Conditionals: A typology”. In E. C. Traugott, A. ter Meulen, J. Snitzer Reilly, and C. A. Ferguson (eds.), On Conditionals, 77–99. Cambridge: Cambridge University Press.
BIBLIOGRAPHY
263
Crewe, William J. 1984. Singapore English and Standard English: Exercises in Awareness. Singapore: Eastern Universities Press. Cruse, D. Alan. 1973. “Some thoughts on agentivity”. Journal of Linguistics 9: 11–23. Dahl, Östen. 1997. “The relation between past time reference and counterfactuality: a new look”. In Athanasiadou and Dirven (eds.), 97–114. Dancygier, Barbara. 1993. “Interpreting conditionals: time, knowledge, and causation”. Journal of Pragmatics 19: 403–434. Dancygier, Barbara. 1998. Conditionals and Prediction. Time, Knowledge and Causation in Conditional Constructions. Cambridge: Cambridge University Press. Dancygier, Barbara and Eve Sweetser. 1997. “Then in conditional constructions”. Cognitive Linguistics 8 (2): 109–136. Davies, Eiran C. 1979. On the Semantics of Syntax. Mood and Condition in English. London: Croon Helm. Delancey, Scott. 1984. “Notes on agentivity and causation”. Studies in Language 8 (2): 181–213. Denison, David. 1993. English Historical Syntax: Verbal Constructions. London: Longman. Depraetere, Ilse. 1995. “On the necessity of distinguishing between (un)boundedness and (a)telicity”. Linguistics and Philosophy 18: 1–19. Dowty, David R. 1979. Word Meaning and Montague Grammar. Dordrecht: Reidel. Dowty, David. 1989. “On the semantic content of the notion of ‘thematic role’”. In Properties, Types and Meaning. Vol. 2. Semantic Issues, G. Cherchia, B. H. Partee, and R. Turner (eds.), 69–129. Dordrecht: Kluwer. Dowty, David. 1991. “Thematic proto-roles and argument selection”. Language 67 (3): 547–619. Du Bois, John. 1988. “Discourse as pattern model for grammar: the possessor = ergator affiliation”. Paper presented at the Symposium on Grammaticalization, University of Oregon, Eugene, 12–15 May. Dudman, Victor H. 1984. “Conditional interpretations of if-sentences”. Australian Journal of Linguistics 4 (2): 143–204. Duffley, Patrick J. 1994. “Need and dare: The black sheep of the modal family”. Lingua 94: 213–243. Eifring, Halvor. 1988. “The Chinese counterfactual”. Journal of Chinese Linguistics 16 (2): 193–217. Eifring, Halvor. 1995. Clause Combination in Chinese. Leiden: Brill.
264
BIBLIOGRAPHY
Erbaugh, Mary S. 1986. “Taking stock: the development of Chinese noun classifiers historically and in young children”. In Noun Classes and Categorization, C. Craig (ed.), 399–436. Amsterdam and Philadelphia: John Benjamins. Erbaugh, Mary S. 1992. “The acquisition of Mandarin”. In Dan. I. Slobin (ed.), Vol. 3, 373–455. Fauconnier, Giles. 1985. Mental Spaces: Aspects of Meaning Construction in Natural Language. Cambridge, Mass.: MIT Press. Ferguson, Charles A. “Diglossia”. Word 15 (2): 325–340. Fillmore, Charles J. 1990. “Epistemic stance and grammatical form in English conditional sentences”. Papers from the 26th Regional Meeting of the Chicago Linguistics Society, 137–162. Finegan, Edward, Besnier, Niko, Blair, David, and Collins, Peter. 1992. Language. Its Structure and Use. Sydney Harcourt Brace Jovanovich. Fleischman, Suzanne. 1982. The Future in Thought and Language. Cambridge: Cambridge University Press. Fleischman, Suzanne. 1983. “From pragmatics to grammar”. Lingua 60: 183–214. Fleischman, Suzanne. 1989. “Temporal distance: A basic linguistic metaphor”. Studies in Language 13 (1): 1–50. Fleischman, Suzanne. 1995. “Imperfective and irrealis”. In Bybee and Fleischman (eds.), 519–551. Foley, Joe (ed.). 1988. New Englishes: The Case of Singapore. Singapore: Singapore University Press. Foley, William A., and Van Valin Jr., Robert D. 1984. Functional Syntax and Universal Grammar. Cambridge: Cambridge University Press. Frawley, William. 1992. Linguistic Semantics. Hillsdale, NJ: Lawrence Erlbaum. Gazdar, Gerald. 1979. Pragmatics, Implicature, Presupposition, and Logical Form. New York: Academic Press. Geis, Michael and Zwicky, Arnold. 1971. “On invited inferences”. Linguistic Inquiry 2: 561–66. Giles, J. A. (ed.). 1969 [1858]. The Whole Works of King Alfred the Great. Vol. 2. New York: AMS Press. Givón, Talmy. 1976. “Topic, pronoun, and grammatical agreement”. In Charles Li (ed.), Subject and Topic, 149–88. New York: Academic Press. Givón, Talmy. 1979. On Understanding Grammar. New York: Academic Press. Givón, Talmy. 1989. Mind, Code and Context: Essays in Pragmatics. Hillsdale, NJ.: Lawrence Erlbaum.
BIBLIOGRAPHY
265
Givón, Talmy. 1991. “Serial verbs and the mental reality of ‘event’: Grammatical vs. cognitive packaging”. In Traugott and Heine (eds.), Vol. 1, 81–127. Givón, Talmy. 1994. “Irrealis and the subjunctive”. Studies in Language 18, 2: 265–337. Givón, Talmy. 1995. Functionalism and Grammar. Amsterdam and Philadelphia: John Benjamins. Goossens, Louis. 1982. “On the development of modals and of epistemic function in English”. In A. Ahlqvist (ed.), 74–84. Goossens, Louis. 1987a. “The auxiliarization of the English modals: A functional grammar view”. In Historical Development of Auxiliaries, M. Harris and P. Ramat (eds.), 111–43. Berlin: Mouton de Gruyter. Goossens, Louis. 1987b. “Modal shifts and predication types”. In Ins and Outs of Predication, J. van der Auwera and L. Goossens (eds.), 21–37. Dordrecht: Foris. Goossens, Louis. 1995. “The English modals as grounding predications: A synchronic-dynamic view”. Paper presented at the International Cognitive Linguistics Conference, University of New Mexico at Albuquerque, 16–21 July 1995. Grice, H. Paul. 1975. “Logic and conversation”. In Speech Acts [Syntax and Semantics 3], P. Cole and J. Morgan (eds.), 41–58. New York: Academic Press. Grice, H. Paul. 1981. “Presupposition and conversational implicature”. In Radical Pragmatics, P. Cole (ed.), 183–198. New York: Academic Press. Gupta, Anthea, F. 1991a. “Almost a creoloid: Singapore English”. California Linguistic Notes 23 (1): 9–21. Gupta, Anthea F. 1991b. “Acquisition of diglossia in Singapore English”. In Child Language Development in Singapore and Malaysia, A. Kwan-Terry (ed.), 119–160. Singapore: Singapore University Press. Gupta, Anthea F. 1992. (In) “Panel discussion: A dictionary of English for Singapore and its practical implications”. In Words in a Cultural Context, A. Pakir (ed.), 216–237. Singapore: Unipress. Gupta, Anthea F. 1994a. The Step-Tongue. Children’s English in Singapore. Clevedon: Multilingual Matters. Gupta, Anthea F. 1994b. “The truth about Singapore English”. English Today 38, Vol.10 (2): 15–17. Hacking, Jane. 1998. Coding the Hypothetical. A Comparative Typology of Russian and Macedonian Conditionals [Studies in Language Companion Series 38]. Amsterdam and Philadelphia: John Benjamins.
266
BIBLIOGRAPHY
Haegeman, Liliane and Wekker, Herman. 1984. “The syntax and interpretation of futurate conditionals in English”. Journal of Linguistics 20: 45–55. Hagège, Claude. 1993. The Language Builder: An Essay on the Human Signature in Linguistic Morphogenesis. Amsterdam and Philadelphia: John Benjamins. Haiman, John. 1985. Natural Syntax. Cambridge: Cambridge University Press. Haspelmath, Martin. 1989. “From purposive to infinitive — a universal path of grammaticization”. Folia Linguistica Historica 10 (1–2): 287–310. Hatav, Galia. 1989. “Aspects, aktionsarten and the time line”. Linguistics 27: 487–516. Heine, Bernd. 1992. “Grammaticalization chains”. Studies in Language 16 (2): 335–368. Heine, Bernd. 1993. Auxiliaries. Cognitive Forces and Grammaticalization. New York: Oxford University Press. Heine, Bernd. 1994a. “Grammaticalization as an explanatory parameter”. In Pagliuca (ed.), 255–287. Heine, Bernd. 1994b. “Areal influence on grammaticalization”. In Language Contact and Language Conflict, M. Puetz (ed.), 55–68. Amsterdam and Philadelphia: John Benjamins. Heine, Bernd. 1994c. “On the genesis of aspect in African languages”. Proceedings from the 20th Annual Meeting of the Berkeley Lingusitics Society, 35–46. Heine, Bernd. 1995. “Agent-oriented vs. epistemic modality. Some observations on German modals”. In Bybee and Fleischman (eds.), 17–53. Heine, Bernd, and Ulrike Claudi. 1986. On the Rise of Grammatical Categories. Berlin: Dietrich Reimer. Heine, Bernd, and Mechthild Reh. 1984. Grammaticalization and Reanalysis in African Languages. Hamburg: Helmut Buske. Heine, Bernd, Ulrike Claudi, and Hünnemeyer, Friederike. 1991. Grammaticalization: A Conceptual Framework. Chicago: University of Chicago Press. Herring, S. 1981. “Tense vs. aspect and focus of attention in the development of temporal reference”. Unpublished paper, University of California. Herring, Susan C. 1991. “The grammaticalization of rhetorical questions in Tamil”. In Traugott and Heine (eds.), Vol. 1, 253–84. Ho, Mian Lian. 1986. The Verb Phrase in Singaporean English. Ph. D. thesis, Monash University. Ho, Mian Lian, and Platt, John. T. 1993. Dynamics of a Contact Continuum. Singaporean English. Oxford: Clarendon Press. Holisky, Dee Ann. 1987. “The case of the intransitive subject in Tsova-Tush (Batsbi)”. Lingua 71: 103–132.
BIBLIOGRAPHY
267
Hopper, Paul J. 1987. Emergent grammar. Proceedings of the 13th Annual Meeting of the Berkeley Linguistics Society, 139–57. Hopper, Paul J. 1991. “On some principles of grammaticization”. In Traugott and Heine (eds.), Vol.1, 17–35. Hopper, Paul and Elizabeth C. Traugott. 1993. Grammaticalization. Cambridge: Cambridge University Press. Horn, Laurence R. 1972. On the Semantic Properties of the Logical Operators in English. Los Angeles: University of California dissertation. Horn, Laurence R. 1984. “Towards a new taxonomy for pragmatic inference: Qbased and R-based implicature. In Meaning, Form and Use in Context: Linguistic Applications [GURT ‘84], D. Schiffrin (ed.), 11–42. Washington, DC: Georgetown University Press. Horn, Laurence R. 1989. A Natural History of Negation. Chicago: University of Chicago Press. Horn, Laurence J. 1996. “Exclusive company: Only and the dynamics of vertical inference”. Journal of Semantics 13: 1–40. Horn, Laurence R. 2000. “From if to iff: Conditional perfection as pragmatic strengthening”. Journal of Pragmatics 32(3): 289–326. Huddleston, Rodney. 1989. Introduction to the Grammar of English. Cambridge: Cambridge University Press. Huddleston, Rodney. 1995. “The case against a future tense in English”. Studies in Language 19 (2): 399–446. Jackendoff, Ray. 1969. Some Rules of Semantic Interpretation for English. Unpublished Ph. D. dissertation. Massachusetts Institute of Technology. Jackendoff, Ray. 1990. Semantic Structures. Cambridge MA. MIT Press. Jacobsson, Bengt. 1974. “The auxiliary need”. English Studies 55: 56–63. James, Deborah. 1982. “Past tense and the hypothetical. A cross-linguistic study”. Studies in Language 6 (3): 375–403. James, Francis. 1986. Semantics of the English Subjunctive. Vancouver: University of British Columbia Press. Jepson, Jill. 1989. “The acquisition of categories in Chinese”. Lingua 78: 193–216. Johnston, Judith R. 1985. “Cognitive prerequisites: the evidence from children learning English”. In Dan I. Slobin (ed.), Vol. 2, 961–1004. Kallgard, Anders. 1993. “Present-day Pitcairnese”. English World-Wide 14 (1): 71–114. Karttunen, Lauri. 1971a. “Counterfactual conditionals”. Linguistic Inquiry 2: 566–569.
268
BIBLIOGRAPHY
Karttunen, Lauri. 1971b. “Some observations on factivity”. Papers in Linguistics 4 (1): 55–69. Karttunen, Lauri, and Peters, Stanley. 1977. “Requiem for presupposition”. Proceedings of the Third Annual Meeting of the Berkeley Linguistics Society, 360–71. Katz, Jerrold J. and Postal, Paul M. 1964. An Integrated Theory of Linguistic Descriptions. Cambridge MA: MIT Press. Klinge, Alex. 1993. “The English modal auxiliaries: from lexical semantics to utterance interpretation”. Journal of Linguistics 29: 315–357. Kuteva, Tania. 1998. “On identifying an evasive gram: Action Narrowly Averted”. Studies in Language 22 (1): 113–160. Kuteva, Tania, and Heine, Bernd. 1995. “The proximative”. Paper presented at the 4th International Cognitive Linguistics Conference, University of New Mexico at Albuquerque, 16–21 July 1995. Kuryłowicz, Jerzy. 1965. “The evolution of grammatical categories”. Reprinted in Esquisses Linguistiques. Vol. 2, J. Kuryłowicz, 1976, 38–54. München: Fink. Kytö, Merja (comp.). 1996. Manual to the Diachronic Part of the Helsinki Corpus of English Texts: Coding Conventions and Lists of Source Texts. (3rd edition). Helsinki: Department of English, University of Helsinki. Labov, William. 1990. “On the adequacy of natural languages”. In Pidgin and Creole Tense-Mood-Aspect Systems, J. V. Singler (ed.), 1–58. Amsterdam and Philadelphia: John Benjamins. Lakoff, George. 1987. Women, Fire, and Dangerous Things. Chicago: University of Chicago Press. Langacker, Ronald W. 1977. “Syntactic reanalysis”. In Mechanisms of Linguistic Change, C. Li (ed.), 57–139. Austin and London: University of Texas Press. Langacker, Ronald W. 1987. Foundations of Cognitive Grammar. Vol. 1: Theoretical Prerequisites. Stanford: Stanford University Press. Langacker, Ronald W. 1988. “An overview of cognitive grammar”. In Topics in Cognitive Linguistics, B. Rudzka-Ostyn (ed.), 3–48. Amsterdam and Philadelphia: John Benjamins. Leech, Geoffrey. 1985. Principles of Pragmatics. London: Longman. Leech, Geoffrey. 1987. Meaning and the English Verb. Harlow: Longman. Lehmann, Christian. 1995. Thoughts on Grammaticalization. München: Lincom Europa. Originally published as: Christian Lehmann, 1982. Thoughts on Grammaticalization. A Programmatic Sketch. Vol. 1 [Arbeiten des Kölner Universalienprojekts 48]. Köln: Institut für Sprachwissenschaft der Universität.
BIBLIOGRAPHY
269
Lehmann, Winifred P. 1974. Proto-Indo-European Syntax. Austin: University of Texas Press. Lehrer, Adrienne. 1996. “Identifying and interpreting blends: An experimental approach”. Cognitive Linguistics 7 (4): 359–390. Leslie, Andrew. 1981. Oddities in Language. Singapore: Federal Publishers. Levinson, Stephen. 1995. “Three levels of meaning”. In Grammar and Meaning. Essays in Honor of Sir John Lyons, F. R. Palmer (ed.), 90–115. Cambridge: Cambridge University Press. Lewis, David. 1973. Counterfactuals. Cambridge, Mass.: Harvard University Press. Lewis, David. 1981. “Counterfactuals and comparative possibility”. In Ifs, Conditionals, Belief, Decision, Chance, and Time, W. Harper, R. Stalnaker, and G. Pearce (eds.), 57–85. Dordrecht: Reidel. Lewis, David. 1986. On the Plurality of Worlds. Oxford: Blackwell. Li, Charles, N. and Thompson, Sandra. 1976. “Development of the causative in Mandarin Chinese: Interaction of diachronic processes in syntax”. In The Grammar of Causative Constructions [Syntax and Semantics 6], M. Shibatani (ed.), 477–92. New York: Academic Press. Li, Charles, N. and Sandra A. Thompson. 1989. Mandarin Chinese. A Functional Reference Grammar. Berkeley: University of California Press. Lieven, Elena V. M., Pine, Julian M., and Baldwin, Gillian. 1997. Lexicallybased learning and early grammatical development. Journal of Child Language 24: 187–219. Lightfoot, David. 1979. Principles of Diachronic Syntax. Cambridge: Cambridge University Press. Lightfoot, David. 1991. How to Set Parameters: Arguments from Language Change. Cambridge, Mass.: MIT Press. Liu, Lisa G. 1985. “Reasoning counterfactually in Chinese: Are there any obstacles?”. Cognition 21: 239–270. Lucy, John A. 1992. Language Diversity and Thought. Cambridge: Cambridge University Press. Lucy, John A. 1996. “The scope of linguistic relativity: An analysis and review of empirical research”. In Rethinking Linguistic Relativity [Studies in the Social and Cultural Foundations of Language 17], J. J. Gumperz and S. C. Levinson (eds.), 37–69. Cambridge: Cambridge University Press. Lyons, John. 1968. Introduction to Theoretical Linguistics. London: Cambridge University Press. Lyons, John. 1977. Semantics, Vols I and II. Cambridge: Cambridge University Press.
270
BIBLIOGRAPHY
Lyons, John. 1995. Linguistic Semantics. An Introduction. Cambridge: Cambridge University Press. Matisoff, James A. “Areal and universal dimensions of grammaticization in Lahu”. In Traugott and Heine (eds.), Vol. 2, 385–53. Matsumoto, Yo. 1995. “The conversational condition on Horn Scales”. Linguistics and Philosophy 18: 21–60. Matthews, Stephen and Virginia Yip. 1994. Cantonese: A Comprehensive Grammar. London: Routledge. McCawley, James D. 1981. Everything that Linguists have Always Wanted to Know about Logic* … *but were ashamed to ask. Chicago: University of Chicago Press. Meillet, Antoine. 1958. “L’évolution des formes grammaticales”. In Linguistique Historique et Linguistique Générale, A. Meillet (ed.), 130–48. Paris: Champion. Originally appearing in 1912, Scientia 12, No. 26 (6). Mey, Jacob L. 1993. Pragmatics. An Introduction. Oxford: Blackwell. Mithun, Marianne. 1995. “On the relativity of irreality”. In Bybee and Fleischman (eds.), 367–388. Molencki, Rafal. 1998. “Modals in past counterfactual protases”. In Advances in English Historical Linguistics 1996, Jacek Fisiak and Marcin Krygier (eds.), 241–251. Berlin: Mouton de Gruyter. Mufwene, Salikoko S. 1990. “Transfer and the substrate hypothesis in creolistics”. Studies in Second Language Acquisition 12: 1–23. Mufwene, Salikoko S. 1996. “Creolization and grammaticization: what creolistics could contribute to research on grammaticization”. In Baker and Syea (eds.), 5–28. Newbrook, Mark (ed.). 1987. Aspects of the Syntax of Educated Singaporean English. Frankfurt: Peter Lang. Newbrook, Mark. 1992. “Unrecognised grammatical and semantic features typical of Australian English: A checklist with commentary”. English World-Wide 13, 1: 1–32. Newbrook, Mark. 1993. “Which English? Institutionalised second-language varieties of English in Asia and the implications for educators: The cases of Singapore and Hong Kong”. Journal of Intercultural Studies 14 (1): 1–17. Newbrook, Mark and Chinniah, Y. A. 1987. “Aspects of the Singaporean English verb phrase: Norms, claims and usage”. In Newbrook (ed.), 339–79. Newbrook, Mark, Delikan, Shyleja, and Dias, Shamini S. E. 1987. “Syntacticsemantic isomorphism in Singaporean English — some neglected phenomena and a possible explanation”. In Newbrook (ed.), 298–338.
BIBLIOGRAPHY
271
Newman, John. 1988. “Singapore’s Speak Mandarin Campaign”. Journal of Multiligual and Multicultural Development 9: 437–448. Newmeyer, Frederick J. 1998. Language Form and Language Function. Cambridge, MASS.: MIT Press. Nicolle, Steven. 1996. Conceptual and Procedural Encoding in Relevance Theory: A Study with Reference to English and Kiswahili. D.phil. thesis: University of York. Nieuwint, Peter. 1986. “Present and future in conditional protases”. Linguistics 24: 371–392. Omar, Asmah Hagi, and Subbiah, Rama. 1968. An Introduction to Malay Grammar. Kuala Lumpar: Dewan Bahasa dan Pustaka. Oxford English Dictionary. 1970. Oxford: Clarendon Press. Oxford English Dictionary. 1971. Oxford: Oxford University Press. Second Edn., Oxford: Clarendon Press. 1989. Oxford Illustrated Dictionary. 1984. Oxford: Oxford University Press. Pagliuca, William (ed.). 1994. Perspectives on Grammaticalization. Amsterdam and Philadelphia: John Benjamins. Palmer, Frank R. 1974. The English Verb. London: Longman. Palmer, Frank R. 1986. Mood and Modality. Cambridge: Cambridge University Press. Palmer, Frank R. 1987. The English Verb. Harlow: Longman. Palmer, Frank R. 1990. Modality and the English Modals. London: Longman. Palmer, Frank R. 1994. Grammatical Roles and Relations. Cambridge: Cambridge University Press. Panther, Klaus U. and Thornburg, Linda. 1999. “The Potentiality for Actuality Metonymy in English and Hungarian”. In Metonymy in Cognition and Language, Klaus U. Panther and Günter Radden (eds.), 333–357. Amsterdam and Philadelphia: John Benjamins. Perkins, Michael R. 1983. Modal Expressions in English. London: Francis Pinter. Picallo, M. Carme. 1990. “Modal verbs in Catalan”. Natural Language and Linguistic Theory 8: 285–312. Plank, Frans. 1984. “The modals story retold”. Studies in Language 8, 3: 305–64. Platt, John T. 1977. “The subvarieties of Singapore English and their sociolectal and functional status”. In The English Language in Singapore, W. J. Crewe (ed.), 83–95. Singapore: Eastern Universities Press. Platt, John T. 1989. “The nature of indigenized Englishes: Interference, creativity, universals”. Language Sciences 11 (4): 395–407. Platt, John T., Weber, Heidi and Ho, Mian Lian. 1984. The New Englishes. London: Routledge and Kegan Paul.
272
BIBLIOGRAPHY
Poplack, Shana and Tagliamonte, Sali. 1995. “It’s black and white: the FUTURE of English in rural Nova Scotia”. Paper presented at the NWAVE 24 Conference, University of Philadelphia, 12–15 October. Quirk, Randolph, Greenbaum, Sidney, Leech, Geoffrey and Svartvik, Jan. 1985. A Comprehensive Grammar of the English Language. London: Longman. Quirk, Randolph, Leech, Geoffrey and Svartvik, Jan. 1987. A Grammar of Contemporary English. Harlow: Longman. Ramat, Anna Giacalone. 1992. “Grammaticalization processes in the area of temporal and modal relations”. Studies in Second Language Acquisition 14: 297–322. Ritchie, William C. 1986. “Second language acquisition research and the study of non-native varieties of English: Some issues in common”. World Englishes 5 (1): 15–30. Roberts, John R. 1990. “Modality in Amele and other Papuan languages”. Journal of Linguistics 26: 363–401. Romaine, Suzanne. 1995. “The grammaticalization of irrealis in Tok Pisin”. In Bybee and Fleischman (eds.), 389–427. Rosch, Eleanor H. 1973. “On the internal structure of perceptual and semantic categories”. In Cognitive Development and the Acquisition of Language, T. F. Moore (ed.), 111–44. New York: Academic Press. Rosch, Eleanor H. 1978. “Principles of categorization”. In Cognition and Categorization, E. Rosch and B. B. Lloyd, 27–48. Hillsdale, NJ: Lawrence Erlbaum. Rosch, Eleanor H., and Mervis, Carolyn B. 1975. “Family resemblances: studies in the internal structure of categories”. Cognitive Psychology 7: 573–605. Saussure, Ferdinand de. 1968 [1907–1912]. Cours de Linguistique Générale. Wiesbaden: Otto Harrossowitz. Scatton, Earnest A. 1984. A Reference Grammar of Modern Bulgarian. Columbus, Ohio: Slavica. Schlesinger, Izchak M. 1971. “Production of utterances and language acquisition”. In The Ontogeny of Grammar, Dan I. Slobin (ed.), 63–101. New York: Academic Press. Schlesinger, Izchak M. 1988. “The origin of relational categories”. In Categories and Processes in Language Acquisition, Y. Levy, I. M. Schlesinger, and M. D. S. Braine (eds.), 121–178. Hillsdale, NJ: Lawrence Erlbaum. Schlesinger, Izchak M. 1989. “Instruments as agents: on the nature of semantic relations”. Journal of Linguistics 25: 189–210. Schlesinger, Izchak M. 1992. “The experiencer as an agent”. Journal of Memory and Language 31: 315–332.
BIBLIOGRAPHY
273
Schieffelin, Bambi B. 1985. “The acquisition of Kaluli”. In Slobin (ed.), Vol. 1, 525–593. Searle, John R. 1979. Expression and Meaning: Studies in the Theory of Speech Acts. Cambridge: Cambridge University Press. Shepherd, Susan C. 1982. “From deontic to epistemic: an analysis of modals in the history of English, creoles and language acquisition”. In Ahlqvist (ed.), 316–323. Shibatani, Masayoshi. 1991. “Grammaticization of topic into subject”. In Traugott and Heine (eds.), Vol. 2, 93–133. Shirai, Yasuhiro, and Andersen, Roger W. 1995. “The acquisition of tense-aspect morphology: A prototype account”. Language 71, 4: 743–762. Singh, Rajendra. 1995. “’New/non-native’ Englishes revisited: a reply to my colleagues”. Journal of Pragmatics 24: 323–333. Slobin, Dan I. (ed.) 1985; 1992; 1997. The Crosslinguistic Study of Language Acquisition, Vols I-II; III; V. Hillsdale, NJ: Lawrence Erlbaum. Slobin, Dan I. 1985. “Crosslinguistic evidence for the language-making capacity”. In Dan I. Slobin (ed.), Vol. 2, 1157–1249. Slobin, Dan. I. 1994. “Talking perfectly. Discourse origins of the present perfect”. In W. Pagliuca (ed.), 119–133. Slobin, Dan I. 1997. “The origins of grammaticizable notions: Beyond the individual mind”. In Slobin (ed.), Volume 5, 245–323. Smith, Carlota. 1991. The Parameter of Aspect. Dordrecht: Kluwer. Sperber, Dan, and Wilson, Deidre. 1986. Relevance. Oxford: Blackwell. Steele, Susan. 1975. “Past and irrealis: just what does it all mean?”. International Journal of American Linguistics 41 (3): 200–217. Stephany, Ursula. 1988. “Modality”. In Language Acquisition: Studies in First Language Development, P. Fletcher and M. Garman (eds.), 375–400. Cambridge: Cambridge University Press. Sweetser, Eve E. 1988. “Grammaticalization and semantic bleaching”. Proceedings of the 14th Annual Meeting of the Berkeley Linguistics Society, 389–405. Sweetser, Eve E. 1990. From Etymology to Pragmatics. Cambridge: Cambridge University Press. Sweetser, Eve E. 1996a. “Reasoning, mappings, and meta-metaphorical conditionals”. In Grammatical Constructions: Their Form and Meaning, M. Shibatani and S. Thompson (eds.), 221–233. Oxford: Oxford University Press. Sweetser, Eve. E. 1996b. “Mental spaces and the grammar of conditional constructions”. In Spaces, Worlds, and Grammar, G. Fauconnier and Eve Sweetser (eds.), 318–333. Chicago: Chicago University Press.
274
BIBLIOGRAPHY
Talmy, Leonard. 1985. “Force dynamics in language and thought”. Papers from the 21st Regional Meeting of the Chicago Linguistic Society, 293–337. Tang Boyland, Joyce. Ms. Evidence and explanations for change in ‘would have done’ constructions. Tedeschi, Philip J. 1981. “Some evidence for a branching-futures model”. In Tense and Aspect [Syntax and Semantics 14], P. J. Tedeschi and A. Zaenan (eds.), 239–269. New York: Academic Press. Teow, Ai Lan. 1988. Modal auxiliaries in Singaporean English. Academic exercise, National University of Singapore. Thomason, Sarah G. and Kaufman, Thomas. 1988. Language Contact, Creolization, and Genetic Linguistics. Berkeley and Los Angeles: University of California Press. Thomson, A. J., and Martinet, A. V. 1988. A Practical English Grammar. Oxford: Oxford University Press. Tongue, Ray K. 1979. The English of Singapore and Malaysia. Singapore: Eastern Universities Press. Tottie, Gunnel, and Övergaard, Gerd. 1984. “The author’s would — a feature of American English”. Studia Linguistica 38 (2): 148–164. Traugott, Elizabeth C. 1972. A History of English Syntax. New York: Holt, Rinehart and Winston. Traugott, Elizabeth C. 1982. “From propositional to textual and expressive meanings: Some semantic-pragmatic aspects of grammaticalization”. In Perspectives on Historical Linguistics, W. P. Lehmann and Y. Malkiel (eds.), 245–71. Amsterdam and Philadelphia: John Benjamins. Traugott, Elizabeth C. 1985. “Conditional markers”. In Iconicity in Syntax, John Haimen (ed.), 289–307. Amsterdam and Philadelphia: John Benjamins. Traugott, Elizabeth C. 1986. “From polysemy to internal reconstruction”. Proceedings of the 12th Annual Meeting of the Berkeley Linguistics Society, 539–50. Traugott, Elizabeth C. 1988. “Pragmatic strengthening and grammaticalization”. Proceedings of the 14th Annual Meeting of the Berkeley Linguistics Society, 406–416. Traugott, Elizabeth C. 1989. “On the rise of epistemic meanings in English”. Language 65: 31–55. Traugott, Elizabeth C. 1992. “Syntax”. In The Cambridge History of the English Language. Vol. 1, R. Hogg (ed.), 168–289. Cambridge: Cambridge University Press.
BIBLIOGRAPHY
275
Traugott, Elizabeth C. 1995a. “Subjectification in grammaticalization”. In Subjectivity and Subjectivisation in Language, D. Stein and S. Wright (eds.), 31–54. Cambridge: Cambridge University Press. Traugott, Elizabeth C. 1995b. “The role of the development of discourse markers in a theory of grammaticalization”. Paper presented at the 12th International Conference on Historical Linguistics, University of Manchester at Manchester, 13–18 August. Traugott, Elizabeth C. 1997. “Unless and but conditionals: A historical perspective”. In Athanasiadou and Dirven (eds.), 145–167. Traugott, Elizabeth C. and Heine Bernd (eds). 1991. Approaches to Grammaticalization, Vols I and II. Amsterdam and Philadelphia: John Benjamins. Traugott, Elizabeth C., and König, Ekkehard. 1991. “The semantic-pragmatics of grammaticalization revisited”. In Traugott and Heine (eds.), 189–218. Traugott, Elizabeth C. and Smith, Henry. 1993. “Arguments from language change”. Journal of Linguistics 29: 431–447. Trudgill, Peter, and Hannah, John. 1982. International English: A Guide to Varieties of Standard English. London: Edward Arnold. Ultan, Russel. 1978. “The nature of future tenses”. In Universals of Human Language. Vol. 3, J. Greenberg, C. A. Ferguson, and E. A. Moravcsik (eds.), 83–123. Stanford: Stanford University Press. Van der Auwera, Johan. 1983. “Antecedent possibilities”. Journal of Pragmatics 7: 297–309. Van der Auwera, Johan. 1997. “Conditional perfection”. In Athanasiadou and Dirven (eds.), 169–190. Van Valin, Robert D. Jr. 1993. “A synopsis of role and reference grammar”. In Advances in Role and Reference Grammar, R. D. Van Valin (ed.), 1–164. Amsterdam and Philadelphia: John Benjamins. Vendler, Zeno. 1967. Linguistics in Philosophy. Ithaca: Cornell University Press. Venezky, Richard L. and Healey, Antoinette DiPaolo. 1980. A microfiche concordance of Old English. Dictionary of Old English Project, Center for Medieval Studies: University of Toronto. Visser, F. Th. 1969–73. An Historical Syntax of the English Language. Parts I-III. Leiden: EJ Brill. Vlach, Frank. 1981. “The semantics of the progressive”. In Tense and aspect [Syntax and Semantics 14], P. J. Tedeschi and A. Zaenan (eds.), 271–292. New York: Academic Press. Wardaugh, Ronald. 1992. An Introduction to Sociolinguistics. Oxford: Blackwell. Warner, Anthony R. 1993. English Auxiliaries. Structure and History. Cambridge: Cambridge University Press.
276
BIBLIOGRAPHY
Waterlow, Sarah. 1982. Passage and Possibility. A Study of Aristotle’s Modal Concepts. Oxford: Clarendon Press. Wekker, Herman C. 1976. The Expression of Future Time in Contemporary British English. Amsterdam: N. Holland. Wells, C. G. 1979. “Learning and using the auxiliary verb in English”. In Language Development: A Reader, V. Lee (ed.), 250–70. London: Croon Helm. Werner, Heinz, and Kaplan, Bernard. 1963. Symbol-Formation: An OrganismicDevelopmental Approach to Language and the Expression of Thought. New York: Wiley. Werth, Paul. 1997. “Remote worlds: The conceptal representation of linguistic would”. In Language and Conceptualization [Language, Culture and Cognition 1], Jan Nuyts and Eric Pederson (eds.), 84–115. Cambridge: Cambridge University Press. Whitelock, D., Douglas, D., and Tucker, S. I. 1961. The Anglo-Saxon Chronicle. London: Eyre and Spottiswoode. Wierzbicka, Anna. 1997. “Conditionals and counterfactuals: Conceptual primitives and linguistic universals”. In A. Athanasiadou and R. Dirven (eds.), 15–59. Wittgenstein, Ludwig. 1953. Philosophical Investigations. New York: Macmillan. Wu, Cynthia H.-F. 1994. “If Triangles were Circles …”. A Study of Counterfactuals in Chinese and in English. Taipei: Crane Publishing Co. Ziegeler, Debra P. 1993. Conditionals and Counterfactuality in Singaporean English. Unpublished Honours dissertation. Monash University. Ziegeler, Debra P. 1994. “Conditionals and counterfactuality in Singaporean English”. Journal of Intercultural Studies 15 (1): 29–49. Ziegeler, Debra P. 1995. “Diachronic factors in the grammaticalization of counterfactual implicatures in Singaporean English”. Language Sciences 17 (4): 305–328. Ziegeler, Debra P. 1996. “A synchronic perspective on the grammaticalisation of WILL in hypothetical predicates”. Studies in Language 20 (3): 411–442. Ziegeler, Debra P. 1997. “Retention in ontogenetic and diachronic grammaticalization”. Cognitive Linguistics 8 (3): 207–241. Ziegeler, Debra P. 2000. “The role of quantity implicatures in the grammaticalisation of would”. Language Sciences 22: 27–61. (Presented under the same title at the New Reflections on Grammaticalization Symposium, University of Potsdam at Potsdam, 17–19 June 1999.). Ziegeler, Debra P. (ms.). “The ontogenetic grammaticalisation of Mandarin ba”. Zipf, G. K. 1949. Human Behaviour and the Principle of Least Effort. Cambridge: Addison-Wesley.
BIBLIOGRAPHY
277
Primary sources Old English 850–950. Alfred’s Boethius. = W. J. Sedgefield (ed.). 1899. King Alfred’s Old English Version of Boethius de Consolatione Philasophiae. Oxford: Clarendon. 850–950. Two of the Saxon Chronicles Parallel = C. Plummer (ed.). 1965 [1892]. Two of the Saxon Chronicles Parallel, Vol. 1. Oxford: Clarendon.
Middle English Benson, L. D. (ed.). 1987. The Riverside Chaucer, Boston: Houghton Mifflin. Chaucer, Geoffrey. 1350–1420. A Treatise on the Astrolabe = In L. D. Benson (ed.). 1987, 662.C1.1–673.C2.7. Chaucer, Geoffrey. Boethius. = In L. D. Benson (ed.). 1987, 429.C1.1–454.C2.376. 1150–1250. History of the Holy Rood Tree = Arthur S. Napier (ed.). 1974 [1894]. A History of the Holy Rood Tree [The Early English Text Society, O. S. 103]. London: Kegan Paul Trench. 1150–1250. Vices and Virtues (Part 1) = F. Holthausen (ed.). 1967 [1888] Vices and Virtues [The Early English Text Society O. S. 89]. London: Oxford University Press. Early Modern English: 1571. Ascham, The Scholemaster = E. Arber (ed.). 1870. Roger Ascham, The Scholemaster. London: English Reprints. 1556. Colville, Boethius = E. B. Bax (ed.). 1897. George Colville, Boethius. Boethius’ Consolation of Philosophy [The Tudor Library V]. London: David Nutt. 1531. Elyot, The Boke Named the Gouenour = E. Rhys (ed.). 1907. Thomas Elyot, The Boke Named the Gouenour. London and New York: J. M. Dent and Co. 1500–70. Fisher, Sermons = J. E. B. Mayor (ed.). 1935 [1876]. John Fisher, Sermons by John Fisher. The English Works of John Fisher, Bishop of Rochester. Part I [The Early English Text Society]. London: Oxford University Press.
278
BIBLIOGRAPHY
1549. Latimer, Sermon on the Ploughers = E. Arber (ed.) 1868; 1869. Hugh Latimer, Sermon on the Ploughers, 18 January 1549; Seven Sermons before Edward VI, On each Friday in Lent. London: Alex Murray and Son. 1350–1420. The Northern Homily Cycle = S. Nevanlinna (ed.) 1973, 1984. The Northern Homily Cycle, Parts II, III. Helsinki: Societé Neophilologique de Helsinki. Skeat, W. W. (ed.) 1965 [1882] = Fitzherbert, 1534. The Book of Husbandry [The English Dialect Society 37]. Vaduz: Kraus Reprint Ltd. The Statutes of the Realm. Printed by Command of his Majesty King George the Third in Pursuance of an Address of the House of Commons of Great Britain, Vol. III. Dawsons of Pall Mall (1963 [1817]). 1554. The Trial of Sir Nicholas Throckmorton = F. Hargrave (ed.). 1730. The Trial of Sir Nicholas Throckmorton. A Complete Collection of State-Trials and Proceedings for High Treason, and Other Crimes and Misdemeanours; from the Reign of King Richard II to the End of the Reign of King George I, Second Ed., Vols. I and IV. London: J. Walthoe Sen. etc. The Works of William Shakespeare. London: The Waverley Book Co. (1949).
Name Index
A Akatsuka, Norika 102 Alieva, Natalia F. 115 Allan, Keith 34, 129, 142 fn. Alleton, Viviane 95 Andersen, Roger W. 115–6, 224, 242 Asher, R.E. 115 Atlas, Jay D. 28 Au, Terry K-F. 1 B Baker, Philip 4 Baldwin, Gillian 226 fn. Bardovi-Harlig, Kathleen 115, 224 Bates, Elizabeth 235 Besnier, Niko 183 Bickerton, Derek 92, 113, 219 Binnick, Robert I. 131 Blair, David 183 Bloom, Alfred 1, 134 Bowerman, Melissa 219–20, 226, 226 fn., 227–8, 230 Budwig, Nancy 227 Burnley, David L. 49 fn. Bybee, Joan 3, 5, 6, 9, 9 fn., 16, 22, 39, 45–7, 51–2, 52 fn., 53–4, 55, 57, 58, 66, 71, 81, 82, 116–8, 128–30, 133, 141, 145, 147, 165, 205, 207, 215, 219–20, 223–4, 251 C Carey, Kathleen 55
Chappell, Hilary 96–7, 99 Chao, Yuen Ren 94, 96, 101–2, 105 Cheng, Robert 94 fn., 97, 103, 103 fn. Cheng, Hsiao-feng 103 Cheung, Hin-tat 215 Chiang, Ker Chiu 114 Chinniah, Y.A. 109 Clancy, Patricia M. 102 Claudi, Ulrike 141, 144, 162, 219, 227–8, 243 Coates, Jennifer 5, 141, 143–4, 146 Collins, Peter 183 Comrie, Bernard 8, 15, 26–7, 33, 39, 42, 96, 105, 113, 119, 129, 140, 249 Crewe, William J. 105, 106 D Dahl, Östen 19, 21–2, 39–40, 54, 55 Dancygier, Barbara 18, 27 Davies, Eiran C. 23–4 Delikan, Shyleja 92 Denison, David 143, 145 Depraetere, Ilse 126–7 Dias, Shamini 92 Douglas, D. 59 fn. Dowty, David R. 81 Dudman, Victor 18, 139 fn. E Eifrin, Halvor 36, 98, 101 Erbaugh, Mary 116, 217–8, 222
280
NAME INDEX
F Fauconnier, Giles 22 Ferguson, Charles 89 Fillmore, Charles 22 Finegan, Edward 183 Fleischman, Suzanne 7, 20–1, 55, 57, 139, 141, 250 Foley, Joe 90 Frawley, William 142 G Gazdar, Gerald 23 Geis, Michael 8, 30 Giles, J.A. 59 fn., 66 Givón, Talmy 22, 40, 46, 62 fn., 111, 117, 141, 219, 234–5 Goodman, Judith C. 235 Goossens, Louis 141, 221 Grice, H. Paul 8, 16, 23, 28, 46, 50 Gupta, Anthea 86 fn., 87–90, 93, 112 H Hacking, Jane 7, 20 Haegeman, Liliane 140 Hagège, Claude 70, 208, 211, 241, 253 Haiman, John 118 Hannah, John 118, 144 Hatav, Galia 126, 131 Healey, Antoinette DiPaolo 49 fn. Heine, Bernd 48, 56, 58, 71, 77, 141, 144, 162, 211, 219, 227–8, 240, 242–3 Herring, Susan 224 Ho, Mian Lian 9, 86, 86 fn., 87, 89–92, 92 fn., 105, 108, 112–4, 128–9, 134, 237, 244 Hopper, Paul J. 47–8, 51–2, 76, 81, 82, 144, 162, 180, 205, 207, 209–11, 241 Horn, Laurence R. 8, 28–9, 30, 33, 37, 57, 147 Huddleston, Rodney 19, 143–4
Hünnemeyer, Friederike 141, 144, 162, 219, 227–8, 243 J James, Deborah 7, 20, 250 Johnston, Judith R. 227 K Kallgard, Anders 129 Karttunen, Lauri 3, 26–7, 61 fn. Kaufmann, Thomas 93 König, Ekkehard 50 fn. Kuryłowicz, Jerzy 48 Kuteva, Tania 46, 58, 71, 79 Kytö, Merja 59 L Labov, William 129 Lakoff, George 228, 242 Langacker, Ronald 211, 215, 216 Leech, Geoffrey 17, 18, 26, 139, 142–3, 145–6, 154, 159, 177 Lehmann, Christian 48, 55, 243 Lehmann, Winifred 223 Lehrer, Adrienne 212 Leslie, Andrew 109 Levinson, Stephen 28, 32 Lewis, David 3, 7, 22, 26 Li, Charles N. 28, 94–5, 95 fn., 96, 100, 114, 209, 222, 254 Lien, Chin-fa 101 Lieven, Elena V.M. 226 fn. Liu, Lisa G. 1 Lucy, John 1 Lyons, John 19, 146 M Martinet, A.V. 131, 140, 145, 154, 173–4 Matsumoto, Yo 33 Matthews, Stephen 99, 102 McCawley, James D. 3 Meillet, Antoine 48, 215
NAME INDEX Mervis, Carolyn B. 242 Molencki, Rafal 118, 137 Mufwene, Salikoko S. 93, 237, 239–40 N Newbrook, Mark 88, 92, 109, 147, 174 Newman, John 87 Nieuwint, Peter 140 O Omar, Asmah Hagi 114–5 P Pagliuca, William 3, 5, 6, 9, 45, 47, 52, 52 fn., 54, 116–8, 133, 141, 205, 207, 219–20, 223–4, 251 Palmer, Frank R. 5, 5 fn., 19, 24–5, 102, 139–41, 143–4, 147, 154 Panther, Klause-Uwe 8, 29 fn., 50 Perkins, Revere D. 3, 5, 9, 45, 52 fn., 54, 116–8, 133, 141, 219–20, 223–4, 251 Peters, Stanley 26–7 Pine, Julian M. 226 fn. Platt, John T. 9, 86, 86 fn., 89–92, 92 fn., 105, 108, 112–5, 128–9, 134, 237, 244 Poplack, Shana 239 Q Quirk, Randoph 17, 18, 26 R Reh, Mechthild 48, 211, 240 Reynolds, Dudley W. 115, 224 Rissanen, Matti 59 Ritchie, William C. 90 Rosch, Eleanor H. 242 S Schlesinger, Izchak M. 177, 226, 230–1
281
Schieffelin, Bambi B. 226 Shepherd, Susan C. 221 Shibatani, Masayoshi 225 Shirai, Yasuhiro 116, 224, 242 Slobin, Dan I. 11, 116, 208, 217, 218, 219, 222, 224, 226, 228–9, 233 Smith, Carlota 243 Sperber, Dan 8, 51 Stephany, Ursula 221 Subbiah, Rama 114–5 Svartvik, J. 17, 18, 26 Sweetser, Eve 22, 25, 49–50, 216–7 Syea, Anand 4 T Tagliamonte, Sali 239 Tedeschi, Philip J. 22 Teow, Ai Lan 109, 135 Thomason, Sarah G. 93 Thompson, Sandra A. 28, 94–5, 95 fn., 96, 100, 114, 209, 222, 254 Thomson, A.J. 131, 140, 145, 154, 173–4 Thornburg, Linda 8, 29 fn., 50 Tongue, Ray K. 109 Traugott, Elizabeth C. 3, 8, 25, 28 fn., 32 fn., 35, 46–7, 48–9, 49 fn., 50–2, 56–7, 67, 76, 78, 81, 105, 117, 141, 144–5, 162, 180, 205, 207, 209–10, 220, 226 Trudgill, Peter 118, 144 Tucker, S. I. 59 fn. U Ultan, Russel 147 V Van der Auwera, Johan 3, 32, 36 Venezky, Richard L. 49 fn. Visser, F. Th. 50, 62 fn., 79, 129 W Wardaugh, Ronald 160
282 Warner, Anthony 52, 56, 226 Weber, Heidi 91, 105, 129 Wekker, Herman 140, 143–4, 147 Wells, C.G. 221 Werth, Paul 17 Whitelock, D. 59 fn. Wierzbicka, Anna 36–8, 97–100 Wilson, Deidre 8, 51 Wittgenstein, Ludwig 242 Wu, Cynthia H-F. 1, 100, 101 fn.
NAME INDEX Y Yip, Virginia 99, 102 Z Ziegeler, Debra P. 2, 36–7, 85, 98, 106–7, 118, 138, 166, 173, 252, 254 Zipf, George K. 28 Zwickey, Arnold 8, 30
Subject Index
A acquisition 11, 12, 115–6, 206, 215–24, 229–31, 234, 236–8, 240, 242, 244, 249, 253–5 agentivity 145, 153, 157, 170–1, 176–8, 195, 225–7, 226 fn., 230–1, 241–2 analogical matching 216, 231 animacy 145–6, 148, 158, 161, 239 antecedent (see also protasis) 7, 17, 18, 22, 23, 28, 33 Antiguan Creole 221, 223, 239, 254 apodosis 25–27, 29, 30, 32–35, 40, 74, 76–78, 96–97, 106–107, 109, 164, 249, 254 aspect (see also lexical) 9, 10, 12, 21, 22, 40, 97, 113, 115, 223, 238 conflict 9, 12, 81, 136, 117–8 habitual 17, 62, 62 fn., 72, 113–4, 129–30, 136, 146, 156, 237 imperfective 12, 13, 21, 117, 129, 134, 250–1 perfective 8, 12, 54, 92, 97, 114, 116–7, 119, 129, 166, 223–4, 236–7, 243, 251 progressive 21, 121, 131–3, 136, 143–4, 153, 161, 171–2, 179–80, 190–2, 196–9, 202, 224–5, 252 + stative verbs 9, 10, 12, 13, 53–54, 97, 112–3, 116–8, 125, 127, 134–6, 166, 223, 236–7, 244, 251
telic 113, 115–6, 134, 223–4 assimilation (semantic) 230 Australian English 2, 10, 108, 137, 144, 174, 180, 182 B ‘back-shifting’ 18, 19, 108 bi-condition 8, 16, 36 Bioprogram Hypothesis 113 bleaching 12, 48–49, 70, 137, 172, 205, 211, 231, 237–9, 249 boundedness (aspectual) 53–54, 116, 132, 142, 142 fn., ‘branching futures’ model 22 British English 10, 144, 182, 252 Bulgarian 55, 58 C Cantonese 87, 93–4, 99, 102, 114 causal link 27, 32–35, 42, 249 Chinese (see also Mandarin) 1, 2, 3, 10, 11, 12, 24, 87, 90–92, 94–95, 97, 104, 110, 114–6, 119, 247 Chinese Pidgin English 93 classifiers 116, 209, 218 cluster, of features 16, 35, 36, 39, 40, 96, 112, 119, 230, 247, 250 Communicative Pressure 241, 253 community (speech) 11, 47–48, 183, 202, 209, 213–4, 216–7, 239, 249, 252–3 compositional value 210, 214, 231–4 conceptual lexicon 234
284
SUBJECT INDEX
conditional (see also counterfactual; hypothetical) 3, 5, 7, 15, 17, 18, 22–27, 30–31, 33–36, 37, 38, 54–55, 58–59, 69, 71, 73–75, 77, 81–82, 94–100, 104–105, 107–108, 111, 117–9, 138, 164–5, 249 conditional perfection (see bi-condition) consequent (see also apodosis) 23 context-induced reinterpretation 175, 175 fn. co-reference 146, 173, 200 counterfactuality 1, 2, 8, 15–16, 17, 18, 26, 37–38, 40–43, 54–55, 58–59, 69, 71, 73–75, 80–82, 95–104, 101 fn., 106, 111–2, 116, 118, 121, 125–7, 134, 136, 139, 146, 154, 173–4, 196, 247, 249–1 Counterfactual Implicature (CFI) Principle 41, 250 creole 3, 4, 12, 13, 88, 92–93, 110, 113, 115, 129, 253 current relevance 55, 125 D deixis 19, 31 fn., 38, 175, 177 dependency relation 20 dialect 4, 5, 6, 9, 10, 12, 15, 87, 89, 91, 94, 110, 114, 118, 135, 167, 174–5, 182, 189, 197, 201, 213–5, 238–41, 245, 248, 253 digossia 89 directional (function) 61, 71, 79 distance (see remoteness) divergence (see split) durational adverb 125–7, 131–2 E epistemic conditional 25 evolution/co-evolution 7, 19, 206, 217–8, 220, 234–5, 240 existential 145, 158, 176, 192
F factive (clause) 60–61, 72, 61 fn., 250 family resemblances 242 French 208, 211 frequency of use 164–5, 214–7, 221, 233, 239, 242 future 19, 22, 46, 51, 56–58, 67, 98, 117, 139–44, 143 fn., 147–8, 153–4, 158, 176, 178, 180, 193, 196, 211, 239–41, 251 -in-the-past (FITP) 56–58, 61, 67, 76, 250–1 G Germanic 54–5, 118, 224 grammaticalisation 3, 4, 5, 7, 10, 11, 12, 13, 15, 19, 36 fn., 46–49, 51, 54–55, 62, 70, 76, 79, 113, 116–7, 133–6, 138, 141, 143, 147, 161–2, 164, 171–2, 180, 189, 190–1, 196, 202, 206–7, 209, 211, 214–9, 231, 234, 243, 247–54 accelerated 7, 136, 180–1, 196, 201–2, 213, 237–40, 244, 253 chain 161–2, 242 emergent 82, 250 hyper-/hypo- 12, 238, 244, 249 L1/L2 5, 6, 11, 135–6, 182, 186, 189, 202, 237–9, 240–1, 244 parallels (ontogenetic/diachronic) 6–7, 116, 218, 220, 223–5, 228–9, 233, 249 grammaticizable notions 219–20, 229 Guyanese creole 113 H harmony 165 Hawaiian creole 113–4 Hebrew 117 Hokkien 87, 92, 94, 99–101, 103, 103 fn., 114 hypotheticality 2, 4, 9, 10, 12, 13, 15–17, 26, 39–43, 45–46, 52–54,
SUBJECT INDEX 57–59, 67, 69, 77–78, 77 fn., 81–82, 85–86, 94, 96, 99, 102–104, 106, 112, 117–9, 126, 134–6, 138–40, 142, 176, 180, 198, 247–51 hypothetical modality 3, 4, 5, 7, 8, 9, 10, 12, 15, 16, 40, 85, 94, 111, 247–8 I implicatures (see also Quantity) 5, 8, 18, 26, 27, 29–31, 32–34, 39–41, 51, 127, 129, 134, 136, 225, 244, 247, 249 cancellation of 22–25, 27, 29, 30–35, 42, 51, 53, 58, 71, 79, 80, 82, 127, 129 conversational 3, 8, 16, 22, 26, 27, 33, 50 fn., 53–54, 56–57, 249 conventionalisation of 46, 49–52, 56, 248 counterfactual 7, 18, 24, 25, 27, 38, 39, 41, 43, 54, 55, 82, 99, 126, 133–6, 241, 248, 250–1 hypothetical 12, 18, 41, 43, 58, 112, 251 suspension of 30–31, 35, 42, 51, 58, 80–82 inchoative 91 indicative 25, 72, 72 fn. intuition 10, 11, 153, 157, 165–6, 195, 199, 205–6, 209–10, 213, 216–7, 231, 236 irrealis 38–39, 40, 41, 68, 109–110, 127, 129, 131–2, 135–6, 173, 223, 250 Italian 225–6 K Kaluli 226–7
285
L L1 2, 3, 4, 6, 9, 10, 11, 12, 13, 86, 88–90, 106, 110, 115, 134–6, 138, 183–4, 187, 190, 202, 213–5, 217, 223, 238–40, 244, 247–9, 251–2, 254 L2 2, 3, 4, 6, 7, 9, 10, 11, 12, 13, 86, 88, 93, 134–6, 138, 166–7, 182–3, 187, 189, 202, 212–5, 217, 223, 238–41, 244, 247–9, 252–3, 255 lexemic source 207, 212, 243 lexical 5, 10, 11, 12, 47–48, 60–61, 66, 68, 72–74, 76–77, 112, 115, 118, 135–6, 176–7, 180, 207–8, 211–2, 218, 224–6, 235, 237–40, 244, 249–51, 254 aspect/Aktionsart 9, 12, 21, 53–54, 81, 93, 112, 116–8, 135–6, 223–4, 251 Lexical Memory Traces 6, 11, 12, 13, 205, 231–3, 236, 240–1, 244, 253, 255 retention 5, 6, 11, 12, 13, 47–48, 55, 77–78, 81, 117, 133, 135, 137–8, 141, 164–5, 166, 168, 170, 177–8, 182, 195, 198–9, 200–2, 205–10, 211–5, 212 fn., 218, 229–4, 236–8, 241–4, 249, 251, 253–5 associative 212, 212 fn., 231, 241 direct 212 sources/origins 3, 4, 6, 82, 117, 142, 148, 159, 175, 198, 201, 211–2, 218–9, 222, 224, 228, 230, 234–7, 242–3, 249, 253–4, 255 lingua franca 86–7, 253 LMT (see lexical) M Malay 10, 86, 91, 114–5, 119, 209 Malaysian English 167 fn.
286
SUBJECT INDEX
Mandarin (see also Chinese) 86–8, 86 fn., 92, 94–5, 94 fn., 97, 99, 101, 114–5, 208, 215, 218, 222–3, 239, 254 metaphoric extension 16, 19, 20–21, 40, 49–51, 158, 177 metonymic extension 19, 21, 29 fn., 49–50, 80, 228 modal/modal verbs 3, 5, 5 fn. 8, 9, 10, 12, 13, 15, 17, 20, 21, 22, 27, 29, 33, 36, 45, 48–50, 52–54, 52 fn., 58, 71, 73, 75, 78, 80, 95–97, 102–105, 109, 111, 117–8, 135, 137, 139, 141, 147, 154–6, 159, 161, 165, 171–2, 177, 180–2, 189–3, 195, 197–9, 205, 207, 210, 212–3, 220–3, 233, 238–40, 248, 250, 252, 254–5 complement/predicate 8, 29, 32, 42, 67, 71, 117–8 deontic/epistemic 5, 7, 8, 28, 46, 49, 56, 116, 141, 143, 145, 147, 171, 174–5, 177, 180, 191–2, 196–7, 199, 220–1, 248 mood 5, 178 N native speaker 88, 93, 110, 130, 138, 140, 182, 206, 213, 244, 253 negation 36–39, 60, 98–99, 147, 156, 175 non-factive (clause) 57, 60–62, 61 fn., 68, 70, 72, 73–74, 78, 250 Nova Scotia 239–41 O ontogenetic (see grammaticalisation) overgeneralisation 6, 11, 12, 109, 135, 212, 215, 228, 238, 241, 244, 249, 252, 254 P panchronic 3, 255
past tense 9, 10, 12, 13, 19–21, 45, 53–55, 93, 95, 102–103, 105–106, 111–7, 119, 121, 125, 127, 129–31, 134–6, 142, 178, 223–5, 236–8, 242–4, 247–51 person (subject) 31 fn., 36, 38, 39–42, 60, 67, 96, 99–100, 104, 119, 144–5, 153 fn., 155, 173, 181, 192 Pitcairnese creole 129 Polish 38 polysemy 165, 175 fn., 179, 191–3, 201–2, 207, 212, 241 pragmatic (see also implicature) 3, 7, 8, 12, 13, 15, 22–23, 42, 49, 50, 54, 117–8, 126, 135–6, 223, 225, 227, 229, 243, 248–50 prediction 24 fn., 25, 29–32, 33–35, 38, 40, 42, 51, 53, 57, 57 fn., 58, 61, 71, 76–77, 79–81, 95, 97–99, 137, 141, 144, 172, 176, 179, 190–1, 196, 199–200, 251 presupposition 7, 22, 23 fn., 27, 31, 57, 80, 82, 250 preterite-present 54 priming 236 progressive (see aspect) Proof by Anachrony Principle 70 protasis 15–18, 23–28, 30–32, 32 fn., 33–36, 96–98, 105, 107–108, 118 Proto-Indo-European 223–4 prototype 12, 116, 212, 224, 241–3 proximative 56, 62, 79 Q Quantity (see also implicature) 5, 8, 28, 28 fn., 29, 32, 33, 34, 46, 52, 59, 60, 66, 80, 82, 118, 247 Quantity 1/Q1 maxim/implicatures 8, 28, 29, 32, 33, 34, 36 fn., 39, 43 Quantity 2/Q2 maxim/implicatures 7, 8, 28, 28 fn., 29–30, 32–34, 36 fn., 39, 46, 50–51 fn., 56, 67
287
SUBJECT INDEX R R-Principle/R-based 8, 28, 29 36 fn., 46, 51 fn. radial categories 242 realis 16, 22, 24 fn., 40, 250 reinforcement 55, 62 relative tense (see also future-in-thepast) 49, 56–57, 81 Relevance Principle 116 relics 210 remoteness, factual or modal 16, 19, 20, 27, 40, 106, 139, 249 retention (see lexical) Russian 38 S Singaporean English 2, 9, 12, 37, 86–91, 93–94, 104–106, 109–15, 118, 130, 134–5, 138, 166, 167 fn., 170, 173, 180, 182, 212, 237–8, 251 sociolinguistic variables 189 Spanish 224–6 specificity 8, 38, 41 split 211 subjunctive 5, 25, 40, 78, 81, 155, 172–4, 174 fn., 178, 195, 198, 250 substratum/a 4, 9, 12, 85–87, 90–94, 114–6, 238, 244, 247–51
T Taiwanese 94, 100, 103, 103 fn., 215, 239 Tamil 10, 86, 115 temporality (see also past tense) 17, 18, 23, 79, 94, 96–97, 108, 115, 117, 126–7, 130, 132, 139, 141–2, 174, 227–8, temporal adverb 115 Teochew 87 transmission discontinuity 13, 88, 239, 241, 248 U undergeneralisation 12, 238 unique condition 30–31, 32 fn., 34–36 V volition 5, 8, 10, 12, 13, 45–46, 51, 53, 58, 60–62, 67, 69–70, 77, 77 fn., 81, 117–8, 137–8, 141, 144–7, 153–4, 153 fn., 156, 159, 164–5, 170–4, 176–9, 181, 190–3, 195–6, 198, 200–1, 203, 205, 207, 213, 221, 232–3, 250–2 W wave (theory) 187 West African pidgin
129
In the STUDIES IN LANGUAGE COMPANION SERIES (SLCS) the following volumes have been published thus far or are scheduled for publication: 1. ABRAHAM, Werner (ed.): Valence, Semantic Case, and Grammatical Relations. Workshop studies prepared for the 12th Conference of Linguistics, Vienna, August 29th to September 3rd, 1977. Amsterdam, 1978. 2. ANWAR, Mohamed Sami: BE and Equational Sentences in Egyptian Colloquial Arabic. Amsterdam, 1979. 3. MALKIEL, Yakov: From Particular to General Linguistics. Selected Essays 1965-1978. With an introd. by the author + indices. Amsterdam, 1983. 4. LLOYD, Albert L.: Anatomy of the Verb: The Gothic Verb as a Model for a Unified Theory of Aspect, Actional Types, and Verbal Velocity. Amsterdam, 1979. 5. HAIMAN, John: Hua: A Papuan Language of the Eastern Highlands of New Guinea. Amsterdam, 1980. 6. VAGO, Robert (ed.): Issues in Vowel Harmony. Proceedings of the CUNY Linguistics Conference on Vowel Harmony (May 14, 1977). Amsterdam, 1980. 7. PARRET, H., J. VERSCHUEREN, M. SBISÀ (eds): Possibilities and Limitations of Pragmatics. Proceedings of the Conference on Pragmatics, Urbino, July 8-14, 1979. Amsterdam, 1981. 8. BARTH, E.M. & J.L. MARTENS (eds): Argumentation: Approaches to Theory Formation. Containing the Contributions to the Groningen Conference on the Theory of Argumentation, Groningen, October 1978. Amsterdam, 1982. 9. LANG, Ewald: The Semantics of Coordination. Amsterdam, 1984.(English transl. by John Pheby from the German orig. edition “Semantik der koordinativen Verknüpfung”, Berlin, 1977.) 10. DRESSLER, Wolfgang U., Willi MAYERTHALER, Oswald PANAGL & Wolfgang U. WURZEL: Leitmotifs in Natural Morphology. Amsterdam, 1987. 11. PANHUIS, Dirk G.J.: The Communicative Perspective in the Sentence: A Study of Latin Word Order. Amsterdam, 1982. 12. PINKSTER, Harm (ed.): Latin Linguistics and Linguistic Theory. Proceedings of the 1st Intern. Coll. on Latin Linguistics, Amsterdam, April 1981. Amsterdam, 1983. 13. REESINK, G.: Structures and their Functions in Usan. Amsterdam, 1987. 14. BENSON, Morton, Evelyn BENSON & Robert ILSON: Lexicographic Description of English. Amsterdam, 1986. 15. JUSTICE, David: The Semantics of Form in Arabic, in the mirror of European languages. Amsterdam, 1987. 16. CONTE, M.E., J.S. PETÖFI, and E. SÖZER (eds): Text and Discourse Connectedness. Amsterdam/Philadelphia, 1989. 17. CALBOLI, Gualtiero (ed.): Subordination and other Topics in Latin. Proceedings of the Third Colloquium on Latin Linguistics, Bologna, 1-5 April 1985. Amsterdam/Philadelphia, 1989. 18. WIERZBICKA, Anna: The Semantics of Grammar. Amsterdam/Philadelphia, 1988. 19. BLUST, Robert A.: Austronesian Root Theory. An Essay on the Limits of Morphology. Amsterdam/Philadelphia, 1988. 20. VERHAAR, John W.M. (ed.): Melanesian Pidgin and Tok Pisin. Proceedings of the First International Conference on Pidgins and Creoles on Melanesia. Amsterdam/Philadelphia, 1990.
21. COLEMAN, Robert (ed.): New Studies in Latin Linguistics. Proceedings of the 4th International Colloquium on Latin Linguistics, Cambridge, April 1987. Amsterdam/ Philadelphia, 1991. 22. McGREGOR, William: A Functional Grammar of Gooniyandi. Amsterdam/Philadelphia, 1990. 23. COMRIE, Bernard and Maria POLINSKY (eds): Causatives and Transitivity. Amsterdam/Philadelphia, 1993. 24. BHAT, D.N.S. The Adjectival Category. Criteria for differentiation and identification. Amsterdam/Philadelphia, 1994. 25. GODDARD, Cliff and Anna WIERZBICKA (eds): Semantics and Lexical Universals. Theory and empirical findings. Amsterdam/Philadelphia, 1994. 26. LIMA, Susan D., Roberta L. CORRIGAN and Gregory K. IVERSON (eds): The Reality of Linguistic Rules. Amsterdam/Philadelphia, 1994. 27. ABRAHAM, Werner, T. GIVÓN and Sandra A. THOMPSON (eds): Discourse Grammar and Typology. Amsterdam/Philadelphia, 1995. 28. HERMAN, József: Linguistic Studies on Latin: Selected papers from the 6th international colloquium on Latin linguistics, Budapest, 2-27 March, 1991. Amsterdam/Philadelphia, 1994. 29. ENGBERG-PEDERSEN, Elisabeth et al. (eds): Content, Expression and Structure. Studies in Danish functional grammar. Amsterdam/Philadelphia, 1996. 30. HUFFMAN, Alan: The Categories of Grammar. French lui and le. Amsterdam/Philadelphia, 1997. 31. WANNER, Leo (ed.): Lexical Functions in Lexicography and Natural Language Processing. Amsterdam/Philadelphia, 1996. 32. FRAJZYNGIER, Zygmunt: Grammaticalization of the Complex Sentence. A case study in Chadic. Amsterdam/Philadelphia, 1996. 33. VELAZQUEZ-CASTILLO, Maura: The Grammar of Possession. Inalienability, incorporation and possessor ascension in Guaraní. Amsterdam/Philadelphia, 1996. 34. HATAV, Galia: The Semantics of Aspect and Modality. Evidence from English and Biblical Hebrew. Amsterdam/Philadelphia, 1997. 35. MATSUMOTO, Yoshiko: Noun-Modifying Constructions in Japanese. A frame semantic approach. Amsterdam/Philadelphia, 1997. 36. KAMIO, Akio (ed.): Directions in Functional Linguistics. Amsterdam/Philadelphia, 1997. 37. HARVEY, Mark and Nicholas REID (eds): Nominal Classification in Aboriginal Australia. Amsterdam/Philadelphia, 1997. 38. HACKING, Jane F.: Coding the Hypothetical. A Comparative Typology of Conditionals in Russian and Macedonian. Amsterdam/Philadelphia, 1998. 39. WANNER, Leo (ed.): Recent Trends in Meaning-Text Theory. Amsterdam/Philadelphia, 1997. 40. BIRNER, Betty and Gregory WARD: Information Status and Noncanonical Word Order in English. Amsterdam/Philadelphia, 1998. 41. DARNELL, Michael, Edith MORAVSCIK, Michael NOONAN, Frederick NEWMEYER and Kathleen WHEATLY (eds): Functionalism and Formalism in Linguistics. Volume I: General papers. Amsterdam/Philadelphia, 1999.
42. DARNELL, Michael, Edith MORAVSCIK, Michael NOONAN, Frederick NEWMEYER and Kathleen WHEATLY (eds): Functionalism and Formalism in Linguistics. Volume II: Case studies. Amsterdam/Philadelphia, 1999. 43. OLBERTZ, Hella, Kees HENGEVELD and Jesús Sánchez GARCÍA (eds): The Structure of the Lexicon in Functional Grammar. Amsterdam/Philadelphia, 1998. 44. HANNAY, Mike and A. Machtelt BOLKESTEIN (eds): Functional Grammar and Verbal Interaction. 1998. 45. COLLINS, Peter and David LEE (eds): The Clause in English. In honour of Rodney Huddleston. 1999. 46. YAMAMOTO, Mutsumi: Animacy and Reference. A cognitive approach to corpus linguistics. 1999. 47. BRINTON, Laurel J. and Minoji AKIMOTO (eds): ollocational and Idiomatic Aspects of Composite Predicates in the History of English. 1999. 48. MANNEY, Linda Joyce: Middle Voice in Modern Greek. Meaning and function of an inflectional category. 2000. 49. BHAT, D.N.S.: The Prominence of Tense, Aspect and Mood. 1999. 50. ABRAHAM, Werner and Leonid KULIKOV (eds): Transitivity, Causativity, and TAM. In honour of Vladimir Nedjalkov. 1999. 51. ZIEGELER, Debra: Hypothetical Modality. Grammaticalisation in an L2 dialect. 2000. 52. TORRES CACOULLOS, Rena: Grammaticization, Synchronic Variation, and Language Contact.A study of Spanish progressive -ndo constructions. 2000. 53. FISCHER, Olga, Anette ROSENBACH and Dieter STEIN (eds.): Pathways of Change. Grammaticalization in English. 2000. 54. DAHL, Östen and Maria KOPTJEVSKAJA TAMM (eds.): Circum-Baltic Languages. Volume 1: Past and Present. n.y.p. 55. DAHL, Östen and Maria KOPTJEVSKAJA TAMM (eds.): Circum-Baltic Languages. Volume 2: Grammar and Typology. n.y.p. 56. FAARLUND, Jan Terje (ed.): Grammatical Relations in Change. 2001. 57. MEL’C UK, Igor: Communicative Organization in Natural Language. The semanticcommunicative structure of sentences. n.y.p. 58. MAYLOR, Brian Roger: Lexical Template Morphology. Change of state and the verbal prefixes in German. n.y.p.