The Present Perfect in Non-Native Englishes: A Corpus-Based Study of Variation 9783110255027, 9783110255010

Citation preview

The Present Perfect in Non-Native Englishes

Topics in English Linguistics 77

Editors

Bernd Kortmann Elizabeth Closs Traugott

De Gruyter Mouton

The Present Perfect in Non-Native Englishes A Corpus-Based Study of Variation

by

Julia Davydova

De Gruyter Mouton

ISBN 978-3-11-025501-0 e-ISBN 978-3-11-025502-7 ISSN 1434-3452 Library of Congress Cataloging-in-Publication Data Davydova, Julia, 1977 — The present perfect in non-native Englishes : a corpus-based study of variation / by Julia Davydova. p. cm. — (Topics in English linguistics ; 77) Includes bibliographical references and index. ISBN 978-3-11-025501-0 (alk. paper) 1. Language and languages — Variation. 2. English language — Globalization. I. Title. P120.V37D38 2011 427—dc23 2011024518

Bibliographic information published by the Deutsche Nationalbibliothek The Deutsche Nationalbibliothek lists this publication in the Deutsche Nationalbibliografie; detailed bibliographic data are available in the Internet at http://dnb.d-nb.de. © 2011 Walter de Gruyter GmbH & Co. KG, 10785 Berlin/Boston Cover image: Brian Stablyk/Photographer’s Choice RF/Getty Images Printing: Hubert & Co. GmbH & Co. KG, Göttingen © Printed on acid-free paper Printed in Germany www.degruyter.com

Contents

Acknowledgements

xn

1. Introduction 1.1. General background 1.2. Research questions and goals 1.3. Previous studies on the present perfect 1.4. Principal definitions of the study 1.4.1. Working terminology 1.4.2. Native speaker vs. non-native speaker 1.4.3. Second language vs. foreign language 1.4.4. Language acquisition vs. language learning 1.4.5. Simplification 1.4.6. Avoidance strategies 1.4.7. Transfer 1.4.8. Varieties vs. interlanguage 1.4.9. Acrolect, mesolect, basilect 1.5. Outline of the book

1 1 4 6 7 7 8 9 9 11 12 12 14 15 16

2. Non-native varieties of English 2.1. Foreign-speaker varieties of English 2.1.1. Case study: English in Russia 2.1.1.1. History 2.1.1.2. English in Russia: some social aspects 2.1.1.3. Essential characteristics of the English spoken in Russia: from acrolect to basilect 2.2. Second-language varieties of English 2.2.1. Case study: Indian English 2.2.1.1. History 2.2.1.2. Indian English: some social aspects 2.2.1.3. Essential characteristics of Indian English 2.2.1.3.1. Acrolectal varieties of Indian English 2.2.1.3.2. Basilectal varieties of Indian English 2.3. Summary

18 18 23 23 24

3. Some theoretical preliminaries 3.1. The category of tense 3.2. The category of aspect 3.2.1. Aktionsart

42 42 45 47

26 29 29 29 32 34 35 39 40

vi

Contents

3.3. The present perfect: tense or aspect? 3.4. Summary

49 50

4. Towards a theoretical explanation for variation between the present perfect and preterite 52 4.1. Thepretente 52 4.2. The present perfect 54 4.2.1. The meaning of the present perfect 56 4.2.1.1. The resultative perfect 57 4.2.1.2. The extended-now perfect 58 4.2.1.3. The experiential perfect 60 4.2.1.4. The perfect of recent past 61 4.3. Semantics of the perfect and preterite: Contrastive analysis.... 62 4.4. Limitations of the semantic analysis of the perfect and preterite 66 4.4.1. Analysis of the collocation of the perfect and preterite with other elements in the sentence 67 4.4.2. Pragmatic analysis of the perfect and preterite 70 4.4.2.1. The preterite and pragmatic discourse 71 4.4.2.2. The perfect and pragmatic discourse 72 4.5. Summary 73 5. The notion of complexity and the English present perfect 5.1. Complexity in the history of linguistics 5.1.1. Absolute approach vs. relative approach to measuring complexity 5.2. Complexity in language 5.2.1. Phonology 5.2.2. Grammar 5.2.2.1. Syntax 5.2.2.2. Inflectional Morphology 5.3. Definitions of complexity adopted in this study for evaluating the linguistic status of the English present perfect 5.3.1. Complexity in morphology 5.3.2. Complexity as a function of strategies employed in structure formation 5.3.3. Complexity of temporal relations 5.3.4. Complexity in semantics 5.3.5. Learners'complexity 5.3.6. Complexity and frequency 5.4. The English present perfect as a complex category 5.4.1. Structural make-up ofthe English present perfect

75 76 78 80 80 80 81 81 82 82 83 84 84 84 85 85 .87

Contents vn 5.4.2. 5.4.3. 5.4.4. 5.4.5. 5.4.6.

Expression of temporal relations by the English present perfect Semantic composition and polysemy of the English present perfect The order of acquisition of the English present perfect The English present perfect and frequency The English present perfect as a complex category: Some additional arguments

6. Metrics of complexity 6.1. Complexity as a degree of language-internal variation 6.2. Complexity as L2 acquisition difficulty 6.3. Summary

88 89 90 97 99 101 102 107 109

7. Empirical design of the study and methodology 110 7.1. Data: Some general comments 110 7.1.1. Large-scale corpora Ill 7.1.2. Small-scale corpora 112 7.1.2.1. Small-scale corpora: sample design and sociolinguistic hypotheses 112 7.1.3. Large-scale and small-scale corpora: hypotheses concerning language-internal variation 115 7.2. Collecting data 118 7.3. Identifying the variable context and coding the data 119 7.3.1. Classifying tokens according to semantic context 124 7.3.1.1. Identifying extended-now contexts 125 7.3.1.2. Identifying resultative contexts 126 7.3.1.3. Identifying experiential contexts 128 7.3.1.4. Identifying contexts of recentpast 129 7.3.2. Classifying tokens according to Aktionsart 131 7.3.3. Classifying tokens according to transitivity 138 7.4. Methods 142 7.5. Summary 143 8. The reference variety of Standard English English (LLC) 8.1. Data 8.2. Quantitative analysis 8.2.1. Distributional analysis 8.2.1.1. Variation in present perfect contexts 8.2.1.2. The present perfect and preterite in present perfect contexts

144 144 145 145 145 147

viii

Contents

8.2.2. Multivanate analysis 8.2.3. Discussion of findings 8.3. Qualitative analysis: The present perfect in definite past time reference contexts 8.4. Summary

152 157 156 159

9. Acrolectal variety of Indian English (HCNVE) 9.1. Sociolinguistic history of Indian English 9.2. Data 9.3. Quantitative analysis 9.3.1. Distributional analysis 9.3.1.1. Extra-linguistic factors 9.3.1.2. Language-internal factors 9.3.1.2.1. Variation in present perfect contexts 9.3.1.2.2. The present perfect and preterite in present perfect contexts 9.3.2. Multivariate analysis 9.3.3. Discussion of findings 9.4. Qualitative analysis: The present perfect in definite past time reference contexts 9.5. Summary

160 160 160 161 161 161 162 162

10. Upper-mesolectal variety of Indian English (ICE) 10.1. Data 10.2. Quantitative analysis 10.2.1. Distributional analysis 10.2.1.1. Variation in present perfect contexts 10.2.1.2. The present perfect and preterite in present perfect contexts 10.2.2. Multivariate analysis 10.2.3. Discussion of findings 10.3. Qualitative analysis: The present perfect in definite past time reference contexts 10.4. Summary

174 174 175 175 175

11. Mesolectal variety of Indian English (HCNVE) 11.1. Data 11.2. Quantitative analysis 11.2.1. Distributional analysis 11.2.1.1. Extra-linguistic factors 11.2.1.2. Language-internal factors 11.2.1.2.1. Variation in present perfect contexts

192 192 192 192 192 193 193

163 167 168 170 173

182 185 188 188 190

Contents

11.2.1.2.2. The present perfect and preterite in present perfect contexts 11.2.2. Multivariate analysis 11.2.3. Discussion of findings 11.3. Qualitative analysis: The present perfect in definite past time reference contexts 11.4. Summary

ix

197 200 201 202 203

12. Basilectal variety of Indian English (HCNVE) 12.1. Data 12.2. Quantitative analysis 12.2.1. Distributional analysis 12.2.1.1. Variation in present perfect contexts 12.2.1.2. The simple past tense in present perfect contexts 12.2.2. Multivariate analysis 12.2.2.1. Stages in the formal and functional development of the English tense-aspect morphology 12.2.3. Discussion of findings 12.3. Qualitative analysis: The present perfect in definite past time reference contexts 12.4. Summary

205 205 205 205 205 211 214

13. Upper-mesolectal variety of East African English (ICE) 13.1. Sociolinguistic history of East African English 13.2. Data 13.3. Quantitative analysis 13.3.1. Distributional analysis 13.3.1.1. Variation in present perfect contexts 13.3.1.2. The present perfect and preterite in present perfect contexts 13.3.2. Multivariate analysis 13.3.3. Discussion of findings 13.4. Qualitative analysis: The present perfect in definite past time reference contexts 13.5. Summary

220 220 222 223 223 223

14. Upper-mesolectal variety of Singapore English (ICE) 14.1. Sociolinguistic history of Singapore English 14.2. Data 14.3. Quantitative analysis 14.3.1. Distributional analysis 14.3.1.1. Variation in present perfect contexts

215 217 218 218

225 229 230 231 234 235 235 236 237 237 237

x

Contents 14.3.1.2. The present perfect and preterite in present perfect contexts 14.3.2. Multivariate analysis 14.3.3. Discussion of findings 14.4. Qualitative analysis: The present perfect in definite past time reference contexts 14.5. Summary

15. Mesolectal variety of English spoken in Russia (HCNVE) 15.1. Sociolinguistic history of English spoken in Russia 15.2. Data 15.3. Quantitative analysis 15.3.1. Distributional analysis 15.3.1.1. Extra-linguistic factors 15.3.1.2. Language-internal factors 15.3.1.2.1. Variation in present perfect contexts 15.3.1.2.2. The present perfect and preterite in present perfect contexts 15.3.2. Multivariate analysis 15.3.3. Discussion of findings 15.4. Qualitative analysis: The present perfect in definite past time reference contexts 15.5. Summary 16. Mesolectal variety of English spoken in Germany (HCNVE)... 16.1. Sociolinguistic history of English spoken in Germany 16.2. Data 16.3. Quantitative analysis 16.3.1. Distributional analysis 16.3.1.1. Extra-linguistic factors 16.3.1.2. Language-internal factors 16.3.1.2.1. Variation in present perfect contexts 16.3.1.2.2. The present perfect and preterite in present perfect contexts 16.3.2. Multivariate analysis 16.3.3. Discussion of findings 16.4. Qualitative analysis: The present perfect in definite past time reference contexts 16.5. Contrastive summary

246 250 252 253 255 257 257 257 258 258 258 259 259 265 268 270 271 272 274 274 275 275 275 275 276 276 281 284 287 288 288

Contents

xi

17. Developing a bird's-eye perspective on the variability of the present perfect across non-native Englishes 17.1. Metrics of complexity 17.1.1. Complexity as adegree of language-internal variation 17.1.2. Complexity as L2 acquisition difficulty 17.2. Complexity of the English present perfect and its repercussions for a non-native grammar 17.3. The concept of current relevance 17.4. Summary

295 296

18. Concluding remarks

306

Abbreviations References Corpora Appendixl Appendix2 Index

310 311 331 332 334 338

291 291

297 301 305

Acknowledgements

The present work is the result of a two-year project (from 2007 till 2009) earned out at the collaborative research centre (SFB 538) on multilinguahsm and at the University of Hamburg. First and foremost, I would like to thank Prof. Peter Siemund, my Doktorvater, for giving me a chance to embark upon a journey that has taken me as far as Bnansk (Russia) and New Delhi (India); for providing input to my work in the form of discussions; for challenging and nurturing and, more importantly, for giving me enough time and space for my ideas to grow. Funding by the Deutsche Forschungsgememschaft is gratefully acknowledged. I am grateful to many institutions and people for granting their permission to reproduce material from previously published sources and hereby acknowledge that two text passages from Housen (2002: 165, 166), two tables from Housen (2002: 158, 162, Table 1 and Table 2) and the graphs from Radden and Dirven (2007: 204, 205) were reproduced with kind permission by John Benjamins Publishing Company, Amsterdam/Philadelphia, www.benjamins.com. Five charts from Petersen (2004: 57, 105) were reproduced with permission by Peter Lang. One table from Bnnton (1988: 29, Table 1.6) was reproduced with permission by Cambridge University Press. The text passage from Melchers and Shaw (2003: 180) was reproduced by permission of Hodder Education (p. 20). The text passage from Winford (2009: 208) was reproduced with permission by Taylor and Frances. The text passages from Leather and James (1996: 272) and from McCoard (1978: 135) were reproduced with permission by Elsevier. One text passage from Gobelsmann (1995: 116) was reproduced with kind permission by Riidiger Kopper. I thank Marianne Hundt for allowing me to reproduce one table from Hundt and Smith (2009: 63). While getting this book ready for publication, I made every effort to obtain necessary permissions for the publication of the previously published material. If any omissions are brought to my attention, I will be happy to provide appropriate acknowledgements in subsequent editions of this work. I also feel much indebted to Prof. Bernd Kortmann, Birgit Sievert, Ulnke Swientek and Frank Benno Junghanns for their invaluable help at various stages of the publication process. I would also like to thank Prof. Anvita Abbi, Prof. Osten Dahl, Prof. Peter Trudgill, Prof. Donald Winford, Dr. Martin Elsig, Dr. Shantanu Gosh,

Acknowledgements xin Dr. Lukas Pietsch, Dr. Devyam Sharma, Dr. Bidisha Som and Tomke Briiggemann for shanng their expertise and providing constructive criticisms in the form of discussions and (written) comments. I thank Prof. Dan I. Slobin for his generous encouragement in the initial stages of my project and Prof. Thomas Berg, from whom I learnt that it takes a critical mind to tackle and conquer the most complex issues in linguistics and in life. I thank my numerous informants for the time they took to explain to me the intricacies of the tense-aspect distinctions inherent in their native-language grammars. I am just as grateful to those informants who consented to give me an interview during the data collection period. My special thanks are due to L.M. Khokhlova, Flonan Zieger and Mayank Jain for their invaluable help in organising the interviews and introducing me into their communities. This project started to become a reality as a direct result of their kind support. Finally, I would like to thank Prof. Sail A. Taghamonte for her encouragement and support at various stages of my Ph.D. project and her genuine interest in my work, and, more importantly, for being who she is: the perfect role model that shaped my linguistic thinking in an essential way. It goes without saying that all remaining shortcomings are my own responsibility. Last but not least, I thank and dedicate this work to my family: Jean Remy, Robert, Dasha, Zhenja and my Mom. This work would not have been possible if it were not for their unconditional love and belief that everything will turn out just fine.

1. Introduction

1.1. General background The overarching aim of this study is to provide a comprehensive account of the category of the English present perfect (also referred to as the perfect or HAVE-perfect) in the light of data obtained from non-native varieties of English, both second-language (or L2) varieties such as Indian English (IndEng), East African English (EAfEng) and Singapore English (SingEng) as well as foreign-speaker varieties of English exemplified by the English spoken in Russia and Germany (RusEng and GerEng respectively). We focus on the present perfect because "the sheer complexity and abundance of grammatical apparatus concentrated in this area of the grammar make it an excellent site for examining the differences and similarities amongst related [forms of English]" (Taghamonte 1996: 351). In more concrete terms, the study aims at investigating the perfect and other surface variants in what has become known as present perfect contexts in the relevant theoretical literature. Thus, some scholars distinguish as many as three dominant contexts for the present perfect (cf Jespersen 1924; Zandvoort 1932; Bauer 1970; Fenn 1987; Winford 1993; Taghamonte 2000). These contexts are (i) resultative contexts, (h) extended-now or continuative contexts and (m) experiential contexts. They are illustrated in (1) through (3). (1)

resultative context He has broken his arm.

(2)

extended-now context / have lived in Hamburg since 2001.

(3)

experiential context I have never been to Russia.

In addition, some researchers single out a context of recent past and its subtype, a hot-news context, as a distinct semantic environment requiring the present perfect in Standard English (Leech 1971a; McCawley 1971; Comne

2

1. Introduction

1976; Bnnton 1988; Klein 1994; Huddleston and Pullum 2002; Siemund 2004; Radden and Dirven 2007), as exemplified in (4).1 (4)

context of recent past The Prime Minister has resigned recently.

In a nutshell, the resultative context suggests that a past action results in a change of state at the moment of utterance, whereas the extended-now context implies that a situation that started in the past still obtains at the moment of utterance. In contrast, the experiential context typically refers to a situation or an event that occurred once or several times prior to the moment of utterance. Finally, the context of recent past is taken to describe a recent event. (A meticulous description of these semantic environments is given in Chapter 4.) The project was initiated by the basic empirical observation that forms other than the present perfect surface in present perfect contexts across nonnative varieties of English. These forms are the present tense, the simple past tense, the past perfect, lone past participle, etc. Moreover, the previous research has revealed that the only other form which alternates with the present perfect in all present perfect contexts across all above-mentioned varieties of English is the simple past tense, otherwise known as the preterite (cf Davydova 2008). Even in standard varieties of English, namely British English and American English, there is always a considerable amount of variation between both forms in present perfect contexts despite clear preferences in favour of either one form or the other in a specific context in a given standard variety. Such a robust variation between the present perfect and other forms does not seem to be a matter of a mere cross-varietal coincidence. Rather, what we deal with here is a subtle phenomenon that needs to be elaborated on. Along these lines, the study proposes to consider the matter by adopting a second- (and foreign-) language learner perspective and to examine the variation between the perfect and other verb forms in varieties that have been labelled together as non-native Englishes.

1 We use the term 'Standard English' to refer to the variety of British English promoted by normative pressures and codified in various (grammatical and typological) descriptions of English (e.g., Leech 1971a, 1971b; Comrie 1976, 1985; Quirk et al. 1985; Huddleston and Pullum 2002; Radden and Dirven 2007, etc.).

1.1. General background

3

But why should one study variation in non-native varieties of English? To start with, variation in non-native Englishes has remained a much neglected issue because for a long time non-native English was thought of as simply wrong English. The present study is thus supposed to bridge this gap by providing a comprehensive description of the category of the English present perfect across different forms of non-native English. Moreover, the existing studies dealing with non-native forms of English (mostly indigemsed varieties) are largely descriptive, thus frequently failing to provide a differentiated account of various morpho-syntactic phenomena. The present study is therefore concerned with working out a methodologically sound and theoretically insightful framework within which non-native varieties of English (both second-language varieties and foreign-speaker varieties) can be examined and compared across the board. Since many morpho-syntactic peculiarities of non-native varieties of English have been described as having emerged as a result of (imperfect) second-language acquisition, studying non-native variation of English may help us to understand what co-occurrences are results of the processes frequently claimed to underlie L2 acquisition (e.g., transfer or substrate influence) and are thus of a specific and localized nature, and what patterns of variation can be claimed to possess a more general character. Thus, studying non-native Englishes should ideally lead us to a better understanding of cross-varietal patterns, their pervasiveness as well as their limits. By the same token, studying variation across non-native varieties of English seems to be a promising approach since it may provide us with additional insights and clues leading to a better understanding of mechanisms governing language variation because variation attested in second-language output is arguably just as rule-governed as the native-speaker variation. Finally, the spread of English as a global lingua franca has repercussions for its non-native users, who - to put it in Berns' (1995: 10, cited in Jenkins 2003: 43) words - find themselves "in the midst of an exciting, challenging, and creative social and linguistic phase of their history". Studying this "sociolinguistic history-in-the-making" becomes particularly important as empirical evidence becomes increasingly available through, for instance, the Internet and other advanced communication technologies (cf Jenkins 2003). Since most non-native Englishes are to a large extent products of educational systems, language-internal variation observed in non-native varieties is compared to the variety of Standard English English (StEngEng), a form of English spoken by the educated native speakers in England. This perspective on studying non-native variation is very attractive for two major reasons. First, it provides a researcher with a straightforward design for

4

1. Introduction

empirical work. Thus, it seems to be natural to "take some norm as a base of reference and to investigate how and why the learner misses it" (Klein and Perdue 1997: 307). Second, it allows a researcher to establish the degree of affiliation between a reference variety (which serves as a yardstick against which the data is measured and compared) and a respective nonnative variety. This procedure might in its turn enable us to reveal patterns of pervasive regularities across different varieties of English which might lead to a better understanding of the systematicity of processes underlying the occurrence of the present perfect across non-native Englishes.

1.2. Research questions and goals The perfect is one of the most intricate phenomena of the English morphosyntax primarily due to its formal and functional complexity. Its full use is one of the last features of English acquired by native learners (cf Van Herk 2008). It therefore should not come as a surprise that a non-native speaker frequently substitutes this structure with other variants in the process of language learning what often gives rise to a robust variation of forms in contexts that require the perfect according to the traditional descriptive accounts of English grammar. Hence, from a wide perspective, the major question addressed in the study can be formulated as follows: How can we explain variation between the present perfect and other verb forms in present perfect contexts in theoretically insightful ways? In order to account for the patterns of variation observed in the data, we rely and elaborate on the notions of complexity employed in linguistics. We first of all show that the English present perfect is a linguistically complex category whose full system of uses is mastered only by the most advanced learners of English. Second, we demonstrate how variation of morphological variants populating present perfect contexts can be studied in terms of varying complexity levels across non-native Englishes. The other issue dealt with in this study concerns the mechanism that underlies variation between the perfect and its major rival: the preterite. Relying on the results of previous research (cf. Davydova 2008), we seek to ascertain in how far the co-occurrence patterns of these verb forms can be explained in terms of the notion of current relevance. To that end, both distributional and multivariate evidence is put into a comparative perspective. While describing and explaining patterns of variability, we rely on the results of the distributional and multivariate analyses, which help to uncover

1.2. Research questions and goals

5

the area of the present perfect marking in a (non-)native English grammar. The goal of the distributional and multivariate analyses is to find out what the studied varieties have in common and along what dimensions they differ. Together, both analyses allow a researcher to establish shared and idiosyncratic patterns, elucidating common linguistic variables (or factors) conditioning the occurrence of the present perfect form. Such internally differentiated comparisons across varieties should shed some light on the mechanisms shaping variation in present perfect contexts (cf Davydova et al. 2011). Yet another question addressed in this study concerns the provenance of nonstandard verb forms used in present perfect contexts in place of the perfect. Where do they come from? Are they a result of a substrate /mothertongue influence or have they emerged due to some general mechanisms involved in L2 acquisition? Alternatively, can these forms be explained in terms of contact with native varieties of English? Carried out within the sociolinguistic (vanatiomst) paradigm, this study aims at assessing the role of the sociolinguistic variable of sex in the production of the HAVE-perfect across non-native Englishes. More specifically, females are assumed to use the more "correct" HAVE-perfect more frequently and consistently than males (see also Chapter 7, section 7.1.2.1. for the elaboration of this hypothesis). Furthermore, non-native speakers use the perfect in contexts from which it is banned in Standard English, at least according to the descriptions available in modern reference grammars (cf. Quirk et al. 1985; Huddleston and Pullum 2002). To provide an example, Mesthne and Bhatt (2008) report the occurrence of the HAVE-perfect with definite past time adverbials such as last month, many years ago, etc. across various "New Englishes", a finding which is in line with our results. Given the keen academic interest that this feature of non-native Englishes has evoked in the past few decades, these tokens were included in the study but analysed qualitatively rather than quantitatively due to low token counts, which often jeopardize a quantitative (i.e. multivariate) analysis. The central issue surrounding these forms concerns their origin. In other words, we seek to discover whether the occurrence of the present perfect in definite past time reference contexts is better analysed in terms of processes underlying second-language acquisition (i.e. transfer, overgenerahsation) or whether they are triggered by other factors such as pragmatic constraints imposed by the speaker on an utterance in the discourse. Last but not least, one of the goals of this study is to show how studying variation attested in non-native Englishes can be informed by the findings

6

1. Introduction

from the fields of Second-Language Acquisition (SLA) and contact linguistics. In other words, while analysing variation between the perfect and other verb forms in the same contexts, we take recourse to various concepts traditionally employed within both disciplines (e.g., transfer, or substrate influence, substrate-independent learner strategies such as overgeneralisation, simplification, etc.). We furthermore rely on the major findings concerning the development of the tense and aspect system in a secondlanguage learner of English (e.g., Inherent Lexical Aspect Hypothesis). In doing so, an attempt is made to bridge the "paradigm gap" (Sndhar and Sndhar 1986) that has existed between the SLA studies and the studies of English for quite some time now (cf Mesthne and Bhatt 2008: 156). Though not longitudinal in its design, the study nevertheless comprises "synchronic snapshots" of data, which enable us to trace the trajectory of the development of the English present perfect in a non-native grammar.

1.3. Previous studies on the present perfect This section provides an overview of the literature dealing with category of the English present perfect. The English perfect has been a subject of enormous interest to linguists of various affiliations. Early structuralist accounts of the perfect can be found in Jespersen (1924) and Zandvoort (1932), whereas more recent descriptive reports are represented by Bauer (1970), Quirk et al. (1985), Huddleston and Pullum (2002) as well as Declerck, Reed, and Cappelle (2006). A typological perspective on the English perfect is provided in Comne (1976), Dahl (1985) and (1999), Dahl and Hedin (2000) and Haspelmath et al. (2005). A meticulous semantic-pragmatic account of the category can be found in Fenn (1987), whereas its description from the cognitive perspective is given in Radden and Dirven (2007). The historical development of the HAVE + participle construction has been extensively studied by various authors (cf. Visser 196373; Bnnton 1988; Demson 1993). The present perfect has received much attention in the sociolinguistic literature on non-standard varieties such as African-American Vernacular, Tnnidadian English and Samana English (cf. Winford 1993; Tagliamonte 1996, 1997, 2000 and Van Herk 2008). Studies on the so-called shift variety of Irish English have on the other hand focused on what might be termed as the functional equivalents of the English perfect, i.e. verb forms used in place of the perfect (cf. Hams 1984a, 1984b; Filppula 1997a, 1997b, 1999; Siemund 2004; Pietsch 2005b, 2007, 2009; Hickey 2004b,

1.4. Principal definitions of the study

7

2007; Kirk and Kallen 2006). Elsness (1997) is a large corpus-based study focusing on variation between the present perfect and "its chief rival, the preterite" in British and American English (Elsness 1997: 1). Last but not least, emergence and development of the category of the English present perfect has been closely examined in accounts on both firstand second-language acquisition. Hence, Chomsky (1969), Nussbaum and Naremore (1975), Gathercole (1986) and Slobin (1994) deal with the acquisition of this structure in a native English grammar, whereas Felix (1978) and Housen (2002) report the development of the English perfect in a nonnative grammar. Furthermore, Odlin and Alonso-Vazquez (2006) discuss the role that the native language exerts on the acquisition of the present perfect by a second-language learner. Finally, Agmhotn, Khanna, and Mukherjee (1998) is a sociolinguistic study designed to crystallize socio-psychological variables underlying the acquisition of the tense and aspect system including the English perfect by second-language learners in New Delhi. As is clear from this quick overview, researchers working within the sociolinguistic paradigm have focused mainly on those varieties of English which are spoken natively. By contrast, very little vanatiomst research has been conducted on non-native varieties of English. This study attempts to fill this gap by giving an exhaustive description of the English present perfect across non-native Englishes of various types. 1.4.

Principal definitions of the study

This section elaborates on some key concepts and important terminological distinctions drawn in the present study. 1.4.1. Working terminology There is no unanimous agreement in the relevant literature on how to refer to the category of the English present perfect, its surface realisations as well as its semantic readings or use types. This study employs the terms "the perfect", "the present perfect" and "the HAVE-perfect" interchangeably to refer to the morphological variant consisting of the auxiliary HAVE and a past participle of the main verb (e.g., have talked, have spoken, etc.). It also uses the terms "resultative perfect/context", "extended-now perfect/context", "experiential perfect/context", "perfect/context of recent past" to refer to the major functions or semantic environments of the English present perfect.

8

1. Introduction

The alternating variants of the present perfect are labelled as follows. The terms the "preterite" or the "simple past tense" are used to refer to verb stems inflected for the past tense (e.g., asked, sard, etc.). The label "past perfect" is employed to designate constructions of the type 'had + past participle'. The label "lone past participles" is related to forms such as gone, been, etc., and "lone present participle" to forms such as gomg, smoking, etc. Constructions comprising the auxiliary BE and a past participle are referred to as the "BE-perfect", whereas the label "present (tense)" is used to describe both the inflected simple present forms (e.g., he goes, she does etc.) and the present progressive forms (e.g., he is going, I am swimming, etc.). Moreover, "bare verb stems" are invariant verb forms such as come, travel, like, etc., whereas "three verb clusters" are constructions of the type am done spent, etc. Finally, the terms "morphology" (e.g., perfect and preterite morphology) and "morpho-syntactic form" are used to refer to surface realisations, or variants, of the linguistic variable studied here.

1.4.2. Native speaker vs. non-native speaker Relying on traditional accounts, Mesthne and Bhatt (2008: 36) draw the following distinction between a native and a non-native speaker of a language: Traditionally a native speaker is assumed to be one who has learnt a language from birth without formal instruction. By contrast a non-native speaker of a language has learnt it as a second (or later) language some time after being initiated into his/her native language. The most important characteristics of non-native speakers of English is that they (i) do not use English as a (primary) means of communication in the family and (n) live in an environment where English is not a prevailing means of communication. Given the obvious differences in the acquisitional contexts obtaining between second-language varieties and foreign-speaker varieties on the one hand and (monolingual) native-speaker varieties on the other, a distinction between native and non-native English in the sense of Mesthne and Bhatt (2008: 36) is maintained in this study.

1.4. Principal definitions of the study

9

1.4.3. Second language vs. foreign language Since this study deals with variation observed in second-language and foreign-speaker varieties of English, we should also look at the differences between a second language and a foreign language. Traditionally, a second language plays an important role as a means of instruction at school and in the academic environment, and as a means of interethmc communication. By contrast, a foreign language is used for international communication, teaching and research and is not used in everyday life (cf. AH 1999: 4). Thus, Englishes spoken in India, Malaysia, Singapore, Bangladesh, Pakistan, etc. are good examples of a second language in contrast to Englishes spoken in China, Japan, Russia and most European countries, where they perform a function of a foreign language. In this study, we will use the terms 'non-native varieties of English' or 'non-native Englishes' to referto second-language and foreign-speaker varieties of English. 1.4.4. Language acquisition vs. language learning In studies on language acquisition a distinction is usually drawn between the process of language acquisition and that of language learning. The former is postulated to be a "natural process of internalizing linguistic rules without formal instruction or conscious efforts" (cf. AH 1999: 5). It is believed that children learn their native language this way. The latter on the other hand is a process requiring a conscious effort on the part of the learner. Hence, a first (native) language is acquired, whereas a non-native language (be it a second language or a foreign language) is learned. Having said that, we must add that such a rigorous terminological distinction is not always maintained in the relevant literature and both terms may be used interchangeably to refer to the processes and outcomes of second-language acquisition (cf. Edmondson 1999: 7). From the perspective of second-language acquisition theory the processes underlying the acquisition of a second language as well as a foreign language are considered to be essentially the same, with no distinction being drawn between the two (cf. Winford 2003). It follows that the major cognitive mechanisms underlying the emergence of second-language and foreignlanguage varieties can be argued to be by and large identical. Second language (acquisition) is thus understood as both second and foreign language (acquisition). It has been noticed in the relevant literature that language learners make various 'errors' (omissive, additive, substitutive errors or errors related to

10

1. Introduction

word order) in their L2 production. Here are some examples illustrating these 'errors': (5)

Learners'errors a. Iask0 him yesterday, (omissive error, simplification ) b. James and Henry like quarrelling with themselves. 'each other' (substitutive error, overgeneralisation/reanalysis)

The term 'error' focuses on the outcome of L2 acquisition. The processes that produced these errors are simplification and overgeneralisation (or reanalysis) amongst others. These processes are also referred to as a set of principles that inform second-language acquisition (Winford 2009: 205). In the ensuing chapters we will use the generic terms "cognitive (or learner) strategies", "universal strategies of second-language acquisition", "universal (language) learning strategies" or "umversals of language creation" to refer to the processes and the outcomes of second-language acquisition of mistype. In fact, all these cognitive strategies have been argued to be the most influential factors in adult language learning (King 1969), accounting for the generally simplified outcome of most L2 systems. The follow-up studies will make it clear that second-language learners' systems are not always 'simpler' than those of native speakers of English. In fact, the learners' attempts to come to terms with a linguistically and cogmtively unusual phenomenon of the English present perfect may give rise to quite complex systems of variation between the perfect and other verb forms in present perfect contexts across non-native varieties of English. We shall elaborate on the notion of complexity in the following chapters. To sum up the preceding discussion, no strict distinction between a foreign language and a second language is drawn in this study as it is assumed that the cognitive mechanisms underlying the acquisition of a second language and a foreign language are fairly similar (cf Winford 2003). Following Edmondson (1999), we will use the terms learn and acquire interchangeably to refer to the processes and outcomes of second-language acquisition.

1.4. Principal definitions of the study

11

1.4.5. Simplification Perhaps the most important process (i.e. strategy) "employed by learners as compensation for partial or incomplete acquisition is simplification" (Winford 2003: 217). It therefore requires a special word of explanation here. Simplification generally refers to the non-realisation (i.e. reduction, omission or deletion) of morphological markings and grammatical words, a strategy designed to ease the learner's perception and production. Consider, for instance, the case where a non-native speaker uses forms such as gone, come, done, etc. instead of have gone, have come, have done, etc. in the target language. Furthermore, bare verb stems (e.g., But other state never visit) are frequently outcomes of inflection reduction. More importantly, simplification is a process whereby an L2 learner replaces a complex form of the target language with a simpler one. For instance, a German speaker using the simple past tense in a context where Standard English and spoken German vernacular require the present perfect can be conceived of as simplification. (6)

Mesolectal German English (HCNVE: GE08) ci I visited French lots oftifnes ' 'I have visited France many times.' German b. Ich habe Frankreich mehrmals besucht. I have France many times visited. 'I have been to France many times'

Additionally, non-native speakers tend to use the structurally and semantically simpler present tense in contexts where a native speaker of English is very likely to employ the HAVE-perfect. Such cases can also be accounted for in terms of simplification strategies. Consider, for instance, (7). (7)

Indian English (ICE: Sla-030) No, initially it [the climate] didn 't affect me but now my resistance is much less. '...my resistance has subsided (decreased).'

Thus, simplification is understood as a very general (and largely substrateindependent) strategy, whereby learners use the semantically and morphosyntactically simpler, i.e. "least marked" (Housen 2002: 160) or more "natural" (Davydova et al. 2011), variant in their L2 production where the

12

1. Introduction

target language requires a semantical^ and a morpho-syntactically more complex form2

1.4.6. Avoidance strategies Avoidance strategies (Edmondson 1999: 96) is another key concept in SLA studies used to refer to a process, whereby learners tend to avoid producing those forms in their target language that they perceive as difficult. To provide an example from Edmondson (1999), learners of German tend to produce diminutive forms of nouns with the -chen suffix, which always signals a neuter form, if they are not sure what gender they should assign to the corresponding noun. So they might say das Ttschchen instead of der Ttsch, das Stuhlchen instead of der Stuhl. The ensuing chapters will clarify how the rare occurrence of the HAVE-perfect in some varieties of English can be viewed as a result of speakers using avoidance strategies in order to come to terms with the complex phenomenon of the English present perfect.

1.4.7. Transfer For the purposes of the present study it is also crucial to draw a distinction between learner strategies and language transfer. The former were treated in some detail in the foregoing sections. What still needs to be elaborated on is the notion of language transfer. First and foremost, language transfer can be described as reinforcing influence of the mother tongue on the second language, resulting in a language change (cf Winford 2005: 373; Kirk and Kallen 2006: 88; Odlin 2009: 265). Other labels (i.e. 'interference', 'substratum influence' 'imposition') have been used to refer to this type of linguistic phenomenon. There seems to be consensus that learners employ features of their mother tongue to compensate for their limited proficiency in a second language (cf. Winford 2005: 379). Such direct mappings include vocabulary and semantics but can also extend to phonology, morphology and syntax. In more concrete terms, transfer means that certain structures of one's LI can be used to express an L2 meaning. To provide an example, a German speaker of English may say Peter probably sings tomorrow meaning 2 It must be noted here that simplification may also refer to other processes such as rule regularisation (see also Winford (2003: 217-219) for a detailed discussion).

1.4. Principal definitions of the study

13

Peter will probably sing tomorrow. The "mcorrect" sentence is the result of a direct influence from German, in which one can use the present tense to express futurity (e.g., Peter singt wahrscheinlich morgen). What is apparently happening here is a speaker employing an LI structure (present tense) to express an L2 meaning (futurity in English) in analogy to German.

German (LI)

English (L2)

Semantic domain: Future reference

Semantic domain: Future reference

Structure: Present tense

Structure: Present tense

Figure 1.1. The process of language transfer The speaker thus mentally projects the structure available in the semantic domain of her mother tongue (i.e. present tense for future reference) onto the corresponding semantic domain of her second language (i.e. future reference). The result of this direct mapping is the use of the present tense to express future meaning in English. For this process to become possible the speaker also needs a structure in her L2 that can be analysed as a corresponding LI structure (for instance, present tense in English and German). To put it differently, transfer requires cross-linguistic identification of two linguistic forms in LI and L2. When the connection between an LI and an L2 form has been established, a whole set of LI structure-meaning correspondences is activated within the speaker's mind. These structuremeaning mappings available in LI are then projected onto L2. In this study, the terms "language transfer", "mother-tongue interference" and "substrate influence" are used interchangeably to refer to the process as well as its outcomes outlined in the preceding paragraphs. The approach used in this study to detect substrate influence on nonnative English is to contrast learners' interlanguage performances with respective native languages such as Hindi, German, Russian, etc. While describing contrasts and similarities obtaining amongst distinct language forms, we rely on actual data produced in the interlanguages labelled col-

14

1. Introduction

lectively as non-native Englishes (cf. Odlin 2009: 266). By contrast, patterns of mother-tongue use draw on the structural descriptions of the native language provided in grammar books. We will also use the introspective method, relying on the native speakers' translations of the English sentences, in order to elicit what verb structures are used in the speaker's mother tongue in contexts where Standard English requires the HAVEperfect (see also Appendix 1).

1.4.8. Varieties vs. interlanguage Variety is one of the key concepts in this study and is in need of an explanation. To start with, a variety is a distinct form of language spoken by a particular group of people or in a particular region (sociolect vs. dialect). A time dimension could be added to introduce historical varieties of a particular language (for instance, the 19th-century Irish English). In addition, all recognised varieties are rule-governed and relatively homogenous, i.e. they exhibit a number of fixed idiosyncratic features of their own in the area of phonology or morpho-syntax what accounts for their unique flair when contrasted with other varieties of the same language. Interlanguage on the other hand has traditionally been described as a linguistic system developed by a non-native language learner who has not yet become fully proficient in the target language, her non-native output constantly approximating to native speakers' proficiency. Moreover, interlanguage developed by each speaker has generally been characterised as a variety in its own right exhibiting a number of idiosyncratic characteristics coming about as a result of the learner's unique experiences with the L2. More importantly, non-native language learning is constrained by processes and principles that are universal in nature. In other words, learners make use of similar strategies or take recourse to some general mechanisms while learning another language. Additionally, non-native language learners sharing the same mother tongue and an equal exposure to the target language tend to form a relatively homogeneous community of speakers whose interlanguage exhibits a number of relatively fixed peculiarities in the area of phonology and morpho-syntax. From the perspective developed in this study, interlanguage is a language variety shaped by mother-tongue influence, substrate-independent learner strategies as well as input from native-speaker varieties (see also Chapter 2).

1.4. Principal definitions of the study

15

1.4.9. Acrolect, mesolect, basilect Since varieties discussed in this study reflect various stages of the longitudinal development of the interlanguage grammar, the notions of acrolect, mesolect and basilect are also in need of an explanation. To start with, the three terms originate from the studies on decreolisation (cf DeCamp 1971; Bickerton 1975 cited in Schumann and Stauble 1983). Being a situation in which a superstate language (i.e. the socially and politically dominant language) and a Creole exist side by side, decreolisation gives rise to a variety of lects that form a post-creole continuum. Hence, the lect closest to the Creole is called a basilect, the one closest to the superstate language is referred to as acrolect. Finally, the intermediate varieties of the Creole are known as mesolects in the relevant literature (cf. Mesthne and Bhatt 2008: 226). Bickerton (1975, cited in Schumann and Stauble 1983: 261) was the first to suggest that the developmental stages exhibited in the process of second-language acquisition may parallel those exhibited in the process of decreohsation since in both cases the speakers' position along the continuum is motivated by the degree of contact they have had with the target language. Drawing on the analogy of decreolisation continuum, non-native Englishes can therefore be systematically described if we conceive of these varieties as exhibiting various degrees of conformance to the standard pattern. We can thus frequently encounter different sub-vaneties within a nonnative variety of English (for instance, IndEng). Depending on the type of the learner, some of these varieties can be described as closely resembling Standard English, whereas others deviate from the standard pattern to such a considerable extent that they are virtually incomprehensible to a native speaker of English. Yet other varieties fall somewhere in between these two extremes. Non-native varieties of English can be said to form a continuum on which basilectal forms are most distantly located from the standard language and acrolectal forms are most closely approaching it. Mesolectal varieties occupy an intermediate position on this continuum. Acrolect, mesolect and basilect can in principle be construed as interlanguage systems (or varieties) within which the observed variation is constrained by speakers' mother tongue, universal processes of second-language acquisition and the amount of exposure to the target language in both classroom and natural environment among a few other factors. Basilectal varieties are represented by speakers with very little exposure to the target language. By contrast, speakers with intensive exposure to the target Ian-

16

1. Introduction

guage constitute acrolectal varieties of a language. Mesolectal speakers demonstrate a moderate amount of contact to the reference variety. The boundaries between one form and another on this continuum are not always clear-cut and in fact speakers may gradually move from one point on this continuum to another as their knowledge of English expands. Moreover, the speakers' position on that continuum can be said to mirror their level of competence (i.e. knowledge of grammar) in Standard English. But how do we know what level of competence a particular group of speakers has in English? The most straightforward answer to this question appears to be that linguistic competence is presumably reflected in the patterns of language use (cf Cedergren and Sankoff 1974: 333-334). In order to elicit the level of competence of a non-native speaker we need to establish the patterns of use of the present perfect in the standard variety of English which will then serve as a point of comparison against which to compare variation of the present perfect in non-native varieties of English. The level of L2 competence is then a function of the degree of conformance to the standard pattern.

1.5. Outline of the book The issues mentioned in this introduction will be elaborated on in the ensuing chapters of the monograph. Having developed a general perspective on studying non-native Englishes (Chapter 2), we shall turn to a more theoretical discussion of the English present perfect. Chapters 3 and 4 present a comprehensive analysis of the English present perfect in the light of the current theoretical accounts of tense and aspect. Chapter 5 elaborates on the contention that the English present perfect is a linguistically and cogmtively complex phenomenon, whereas Chapter 6 shows how variation observed in present perfect contexts across non-native Englishes can be studied and explained in terms of varying complexity levels. Chapter 7 introduces the empirical design of the study and its methods; the major criteria (or independent variables) according to which the corpus data will be analysed in the subsequent analyses as well as the hypotheses underlying the choice of these variables. In what follows, data analyses are presented. Thus, Chapter 8 discusses the patterns of occurrence of the present perfect in the standard variety of English. Chapters 9 through 14 present analyses of variation attested in second-language varieties of English exemplified by different varieties of IndEng as well as upper-mesolectal varieties of EAfEng and SingEng. Chapters 15 and 16 deal with foreign-speaker varieties of English

7.5. Outline of the book

17

spoken in Russia and in Germany. Chapter 17 develops a bird's-eye perspective of the variability of the present perfect across non-native Englishes. It provides a synthesis of the results of this study by taking recourse to the concepts elaborated on in the theoretical part of the book. Chapter 18 provides some concluding remarks about the most important findings of this study, pinpointing their relevance for the ongoing and future linguistic research.

2. Non-native varieties of English

This chapter focuses on developing a perspective that would help to bring the existing differences between second-language (or indigemsed) and foreign-speaker varieties of English to a common denominator. Taking the variety of IndEng and the English spoken in Russia as examples, we will demonstrate what these different varietal types have in common and along what dimensions they may differ. The procedure will enable us to develop a common ground allowing for a comparison of these quite distinct forms ofEnglish. 2.1. Foreign-speaker varieties ofEnglish As a global language, English boasts a multitude of speakers that use the language for different purposes and in different forms (cf Crystal 2004). Despite this heterogeneity, three principal groups of speakers of English speakers have been distinguished (cf. Quirk et al. 1985; Kachru 1985, 1986; Crystal 2004): those who speak English as a native language and thus belong to countries of the inner circle; those who speak English as a second language and hence inhabit countries of the outer circle, in which English enjoys the status of an official or co-official language; and, finally, those who use English mainly as a foreign language and belong to the countries of the expanding circle, where English is not recognised on the official level and is mainly used as a means of communication with foreigners. The immediate questions arising from the preceding considerations are twofold. First, one might wonder if it is justified to accord a variety status to those non-native forms ofEnglish that are spoken in countries where the language does not have an official status. Second, one might also ask why we need to study foreign-speaker varieties ofEnglish. Since the answer to the second question seems to be more straightforward, we shall consider it first. To start with, as the process of globalisation proceeds, English is becoming a common "linguistic denominator" worldwide. It is one of the very few features that a Korean executive in Russia, a German Eurocrat in Brussels and a Chinese geneticist at an international symposium in Sweden have in common (cf. Power 2005: 65). Native speakers of English tend to become more and more disadvantaged by this phenomenon, failing to communicate their message to foreigners and thus

2.1. Foreign-speaker varieties of English

19

losing out on deals when they do not folly understand how English is being used by non-natives (cf. Power 2005). Furthermore, non-native speakers of English (including foreign speakers) outnumber native speakers 3:1 (cf. Crystal 2004). This trend is likely to continue, as English has become a language of technology and commerce and is a target language for an ever growing number of population groups. Consider the following figures from Graddol (1997: 10): First-language speakers: 375 million Second-language speakers: 375 million Foreign-language speakers: 750 million The following citation illustrates the impact the English language has had on various spheres of life in different countries all over the world: One out of five of the world's population speak English to some level of competence. Demand from the other four fifths is increasing ... By the year 2000 it is estimated that over one billion people will be learning English. English is the main language of books, newspapers, airports and air-traffic control, international business and academic conferences, science technology, diplomacy, sport, international competitions, pop music and advertising. Graddol (1997: 2) Thus, some thirty years ago English was used for work only by elites such as diplomats and CEOs. Nowadays English has spread into other social strata, gaining in importance among staff of workers, guides, taxi drivers and ordinary citizens in countries such as China, Russia, the Ukraine, Poland, Hungary and many others. The situation is very similar in Western European countries such as Germany, France, the Netherlands, etc., i.e. countries that boast long traditions of teaching and learning the English language. It is probably for this reason that English is taught and learned at an increasingly earlier age in these countries. To provide an example, English is now taught starting at the age of 6 in many schools in Russia. In 2004, primary schools in major Chinese cities began offering English in the third grade, rather than in middle school (cf. Power 2005). English is taught as a first foreign language in schools in all of above-mentioned countries, which accounts for its special status in these countries (cf. Crystal 2004). A similar idea is expressed in Melchers and Shaw (2003):

20

2. Non-native varieties of English Once a language becomes widely known, schools tend to make its dominance self-perpetuating. If it is observed that English is a useful language on the world stage, then schools start to teach it. Once more people in more countries have learnt it at schools, it becomes more useful because there are more foreigners with whom it can be used. So, if following the wishes of parents and pupils, schools teach the language even more, so even more people learn it, it becomes even more useful. [...] At present the world's schools are forces to strengthen the position of English. English is the main foreign language taught in Japan, China and other Asian countries. The EU issues education figures for 26 expanding circle European countries which are members of the EU or EEA or candidates for that status (Pilos 2001), and in all but two English is the most studied foreign language. ... The age of beginning English study is gradually being lowered and it is not unusual to start before the age of 10 (Pilos 2001). Melchers and Shaw (2003: 180)

Finally, English is spreading not only in the top-down direction, i.e. through educational institutions, but also bottom-up, i.e. through individuals representing subcultures in countries all over the world. In such social contexts, English is used as a lingua franca by the members of subcultures associated with computers and hip-hop, heavy-metal and rock music (cf Melchers and Shaw 2003). These subcultures enjoy covert prestige and often represent a network of knowledgeable individuals that make use of English for communicative purposes. Interestingly enough, the English spoken within such subcultures is characterised by a standard conformant (technical) vocabulary but varied grammar. As can be seen, a whole new world of Englishes has been emerging, a world that needs to be investigated in 'real-time'. Since no in-depth descriptions have been provided for the varieties of English of the expanding circle so far, future research should focus on studying these forms of English. While approaching a more complicated issue as to whether or not foreign-speaker Englishes can be classified as self-contained varieties of English, we might want to consider a few general factors that shape variation of both foreign-speaker varieties and second-language varieties of English. First, similar to speakers of second-language varieties (e.g., IndEng), speakers of the expanding circle are exposed to the input from the native-speaker varieties (mostly British English and American English) - by and large through the media - and can thus be said to be native norm sensitive and in many cases - native norm dependent. Second, just like speakers of the outer circle, speakers of the expanding circle learn English from non-native

2.1. Foreign-speaker varieties of English

21

speakers of English. For instance, in Russia secondary school students learn English from native speakers of Russian that have had a professional training in the English language; the same applies, by and large, to secondary school students in Germany. We can conclude, therefore, that the English spoken in an ESL classroom in Russia and in Germany but also in France, the Netherlands, etc. is indeed characterised by a number of idiosyncratic features of its own in the areas of phonology, lexicon and grammar, this specificness being due to the influence from the teachers' and students' respective mother tongues. We should, however, bear in mind that foreignlanguage learning is always standard conformant in that the foreignlanguage learner is always, or at least most of the time, native-speaker competence oriented. However, despite their striving for native speakers' competence, most foreign-language learners are bound to make errors as a result of native language transfer processes and universal mechanisms of second-language acquisition, which become activated in a language learning situation. Interestingly enough, a considerable number of peculiarities exhibited by the second-language varieties of English are, to a great extent, the result of the substrate influence and learners' strategies employed during secondlanguage acquisition. Here the question is to what extent foreign-speaker and second-language varieties of English are shaped by substrate influence and what linguistic features can be ascribed to (the) 'universal', i.e. cogmtively salient, mechanisms of second-language acquisition. The case studies in the ensuing chapters address this question. Finally, similar to second-language varieties, foreign-speaker varieties of English can be described as extremely heterogeneous forms of English, as speakers master their English to various degrees. This difficulty notwithstanding, many (if not all) second-language varieties of English, also referred to as New Englishes (cf Schneider 2007; Mesthne and Bhatt 2008), have been frequently described in the relevant literature as possessing relatively consistent forms in the area of phonology, lexicon and even grammar. By contrast, no attempt has been made so far to provide a systematic description of the linguistic properties of foreign-speaker forms of English. It is, however, important to realise that varieties of both types share one essential characteristic, i.e. they come into existence as a result of 'imperfect' second-language acquisition and are thus subject to similar constraints that shape the linguistic systems of non-native speakers of English. Factors constraining the variation found in non-native varieties of English are summed up in Figure 2.1.

22

2 Non-native varieties of English Input from the standard varieties ofEnglisMBritishEnglishand American English)/ Superstate influence

4 Universal (learner) strategies of L2 acquisition

Non-native grammar of English

Degree of proficiency in (or amount of contact with) the target language

tf " Language transfer/ Substrate influence

Figure 2.1. Factors shaping non-native forms of English Other factors constraining non-native variation include the motivation of learners and the social function of the second language (cf Winford 2009: 223). Being by and large compatible with the analyses of constraints on nonnative-speaker variation provided in Klein and Perdue (1997) and Winford (2006, 2009), the account presented above establishes a general perspective on studying non-native varieties of English: variation attested in non-native varieties of a language is just as rule governed as variation attested in fullyfledged languages (for instance, native English). More importantly, this analysis allows us to grant foreign-speaker Englishes a variety status. To be more exact, all non-native Englishes (be it second-language or foreignspeaker varieties) are systematically shaped by similar factors such as speakers' mother tongue, speakers' proficiency, degree of input from the standard varieties of English and universal mechanisms of L2 acquisition. It follows that language forms constrained by identical parameters should in principle be comparable across the board. It is thus a highly interesting task (as well as a challenge) for the analyst to assess the exact role of each factor in the formation of non-native Englishes.

2.1. Foreign-speaker varieties of English

23

2.1.1. Case study: English in Russia Most expanding-circle varieties of English have been described only scantily in the relevant literature. The English spoken in Russia is no exception to the rule. These difficulties notwithstanding, the ensuing sections will attempt to show that phonology, lexis as well as the grammar of English spoken in Russia can be systematically portrayed in analogy to the existing descriptive accounts of the indigemzed varieties of English. In doing so, we assess the role that historical, sociolinguistic as well as cultural context may have played in shaping the specific character of English spoken in Russia.

2.1.1.1. History Russia is probably best described as a country in which studying European foreign languages has formed an essential part of good education. Starting from the 18th century up until the revolution of 1917 various European languages were used by the Russian elite. German, for example, was very popular in the 18th century since many spouses of the Russian Tsars came from Germany including Catherine the Great. To provide a more vivid example, most professors of Moscow University founded in 1755 were Germans, the only exception being Mikhail Lomonosov, the founder of the university, who, however, received his education in Germany (cf TerMinasova 2005: 445). In the 19th century French became the language of education and a means of communication for Russian nobility. To give a notorious example, Ekatenna Dashkova (1743-1810), Head of the St. Petersburg Academy of Sciences and the Russian Academy, knew four foreign languages but could barely speak Russian (cf. Ter-Minasova 2005: 445). In the 20-century, especially after the Second World War, the focus of public attention shifted towards English. During the Soviet era, English was an indispensable part of the Russian education system, although the attitude towards foreign languages in general and to English in particular was somewhat negative. In fact, during the first decades of the 20th century foreign languages were declared to be a sign of bourgeoisie and aristocracy and therefore banned from the school curricula in Russia and did not return until the late 1920s. After heated debates, English was re-introduced into educational institutions including universities and since that time English was an invariable part of the curriculum in the Soviet system of education. However, people studying foreign languages including English were looked

24

2. Non-native varieties of English

down upon as potential spies, emigrants or cosmopolitans lacking loyalty and patriotism (cf. Ter-Minasova 2005). For years and years studying English was a "love-for-love's-sake" enterprise (Ter-Minasova 2005). This kind of attitude towards students of English and the language in general remained dominant up until the end of the Soviet period and had repercussions for the conditions under which English was acquired during those times. Because the Russian community was cut off from the rest of the world during the Soviet era, generations of teachers were raised who never had a direct contact to a native-speaker variety of English during their acquisition period. The English acquired behind the Iron Curtain was the language that was learnt under conditions where learners did not have an access to the English radio, television, newspapers as it was the language of the "enemy". The situation was in many cases aggravated by inadequate teaching materials. As is clear, this type of the acquisition setting is very similar to that of secondlanguage varieties of English (for instance, IndEng) where most learners do not get to interact with native speakers of English and, in fact, have comprehension difficulties, both phonological and contextual, or for that matter cultural (cf. Sndhar and Sndhar 1986: 6).

2.1.1.2. English in Russia: some social aspects "A most striking change this country is now experiencing is, among other thrilling novelties, an unprecedented and ever-increasing urge for foreignlanguage learning - mostly and overwhelmingly English" (cf. Ter-Minasova 2005: 451). Since the times of perestroika English has enjoyed a prestigious status amongst professionals and the more general population alike, although it is a younger generation, i.e. the generation born in the 1980s, that is more exposed to English and, as a result, more proficient in the language. Russian learners of English can be subdivided into four major groups: (i) schoolchildren counted in millions, (h) students of higher education counted in millions, (hi) the new class of Russian capitalists and (iv) middle-class people of all ages with various goals in mind such as business, career, travelling, and emigration (cf. Ter-Minasova 2005). As in many countries where English has no official status, English is used in Russia in four principal domains - advertising, the media, business, interaction with outsiders and education. First, English shapes Russian advertising culture: it is one of the salient features of Russian TV commercials, exerting influence on the advertising discourse structural patterns of commercials produced in Russia (cf. Ustinova

2.1. Foreign-speaker varieties of English

25

and Bhatja 2005). Thus, the majority of Russian commercials employ a 'bilingual mix' (cf. Ustinova and Bhatja 2005) which inevitably increases English awareness among the general population of Russia. Similarly, a vast majority of Russian advertisements are at least partially in English, and these appear in supermarkets, on the sides of buses, in newspapers, etc. English has therefore established itself as a language of advertising creativity in Russia. Second, English is the primary language for addressing foreigners and tourists in Russia. Since the Internet has become a popular medium of communication, English has been increasingly used as a lingua franca in this social domain so that it could be argued that in Russia, English is fairly closely associated with the Internet. Third, English has been making inroads into the Russian media as well. Thus, some Russian newspapers (ARI, Russian Information Agency, The Chechen Times, Delovoj Peterburg, Extrabalt, Ichkena, MosNews, Neva News, Tatarskaya Gazeta, ZR Press, etc.) have an English edition. All these newspapers host a site on the Internet and supply their English versions in electronic format as well. Naturally, the newspapers are written and produced by Russian native speakers for whom English is a non-native language. Furthermore, popular culture in Russia is intrinsically connected to English. For instance, a substantial proportion of pop-song lyrics heard on Russian radio and TV channels are in English. In addition, Englishlanguage programmes have recently become available on satellite TV and their audience is increasing constantly. Fourth, many young and educated native speakers of Russian come into contact with the English language through the software installed on their personal computers. Although major international products are available in Russian, more knowledgeable users prefer authentic versions of software as they are associated with a better quality. Fifth, multinational companies (e.g., Airbus, Ford) operating in Russia adopt English as a company language for newsletters, correspondence, documentation, etc. In addition, negotiations within international companies are typically held in English. Finally, English is sometimes used as a medium of instruction in institutions of tertiary education, while the national language is widely employed as the medium of education in primary and secondary schools. But even this situation is slowly changing. More and more Russian parents are hiring private tutors of English so that their children start acquiring English simultaneously with other subjects. Thus, Russia has more than 350 Englishlanguage schools, where English is used as a major medium of instruction.

26

2. Non-native varieties ofEnglish

In addition, more and more students attend various courses taught in English such as business studies, law, advertising so that they learn the language along with the subject. These socio-economic pressures stimulate many Russian universities (for instance, Moscow University) to offer courses taught in English and to attract the best researchers they can get from all over the world. Last but not least, Russian Ph.D. theses are often written in English which once again confirms its status as a lingua franca of the academic world and shows that Russia is no exception to that rule. We should, however, bear in mind that most domains of Russian social life are unaffected by English. English has not as yet penetrated such spheres of communication as home and family life, religion and courts, national administration and national politics, and some others. Moreover, English is not used amongst friends in Russia as is the case with, for instance, IndEng, nor is it used as a 'link language' (cf Sndhar 1991) to unite people constituting different communities of the country. Summing up, a close examination of the sociolmguistic history of RusEng revealed that even though English fulfils an extensive range of functions in Russia, its functional domains are not as wide-spread as they are in the countries where English is granted an official status. However, the context in which English is acquired in this country bears one crucial similarity to the acquisition^ context of second-language Englishes: the language is acquired mostly through non-native speakers in the countries of the outer and expanding circle.

2.1.1.3. Essential characteristics of the English spoken in Russia: from acrolecttobasilect As elaborated on in the introductory part of the study, a variety can be understood as a distinct form of language that is spoken by a particular group of people in a particular area. The expression 'distinct form' implies that a variety must possess a number of morpho-syntactic, lexical and phonological features that would distinguish this form of speech from any other forms of speech. The major purpose of this section is, therefore, to describe some distinct features of RusEng in the area of phonology, lexicon and grammar.1 1 The ensuing analysis draws on data obtained from the Hamburg Corpus of NonNative Varieties of English (see the chapter on the empirical design of the study for further details).

2.1. Foreign-speaker varieties of English

27

In the field of phonology, the basic distinctive features of RusEng can be summed up as follows. To start with, mesolectal and basilectal RusEng lacks aspirated consonants /p/, hi and /k/ typical of native-speaker varieties of English. Thus, the phonological feature of aspiration is by and large absent in these varieties of RusEng. Second, the dental fricative consonants /0/ and 16/ are frequently replaced by the alveolar stops hi and /d/ in basilectal and mesolectal RusEng. Third, many Russian speakers of English tend to make no distinction between long and short English vowels as vowel length is not a distinctive phonological feature in Russian. For example, there may be no distinction between the pronunciations of bin and bean in mesolectal and basilectal varieties of RusEng. It should be noted that all phonological peculiarities of RusEng presented here correlate with the proficiency level of a given Russian speaker of English. Thus, advanced speakers of English are likely to have mastered aspiration so that their English will be standard conformant in this respect. By contrast, mesolectal and basilectal speakers' pronunciation will exhibit deviations from the standard pattern. It is this particular circumstance that makes this foreign-speaker variety very similar to second-language vaneties (for instance, IndEng), in which variation observed within a particular linguistic domain is also dependent on the speaker's proficiency level. Similar to phonology, the morphological and syntactic differences between RusEng and Standard English are variable rather than categorical. Thus, the acrolectal variety of RusEng may be argued to exhibit only minimal deviations from the native-speaker pattern. In fact, this variety can be expected to exhibit the highest degree of conformance to the reference variety. It is the mesolectal variety that deviates from the norms of the target language. The ensuing description of the RusEng morpho-syntax is therefore derived from the latter. To start with, the mesolectal variety of RusEng is characterised by frequent omission of articles both definite and indefinite as this category is not attested in the Russian language. (8)

Russian English (HCNVE: RE06; RE01) / have 0 very good family. YMeHHeoHeHBxopomaHceMBH.

Second, the present tense of the copula BE tends to be omitted, whereas the past tense of the copula BE is rarely omitted. This peculiarity may also be explained as a result of substrate influence from Russian, which lacks the copula BE in the present but not in the past tense.

28

2. Non-native varieties of English

(9)

Russian English (HCNVE: RE13) Maybe a few years ago we were weak, now we 0 strong. HecKOJiBKO Jiex ra MM 6BUIH CJia6He, cennac MM 0 CHJIBMBIC

Moreover, some mesolectal speakers of RusEng tend to overgeneralise the progressive aspect to habitual events and situations, as in (10) Russian English (HCNVE: RE13) And every day I am sitting in the Internet for thirty minutes. Other interesting features in the domain of tense marking concern the use of the present tense in extended-now contexts and the simple past tense in resultative contexts, as in (11) Russian English (HCNVE: RE05; RE01) a. / m studying French for five years. b. - For how long has she been running it [the beauty salon]? -May be three weeks, but she progressed a lot. Finally, the present perfect is sometimes used in RusEng to describe events with definite past time reference, as in (12) Russian English (HCNVE: RE10) / don't think that Europe has done the good thing when they united in one. Note that many morpho-syntactic peculiarities of RusEng are, in fact, frequently encountered in second-language varieties of English (cf Mesthne 2004a; Kortmann and Schneider 2004). The mesolectal domain of the RusEng lexis reflects the sociolinguistic and cultural background of the local ethos like no other linguistic domain. As in other non-native varieties of English, Russian words are introduced into English. (13) Russian torn*-beatroot soup pelmeni - meat balls sci - cabbage soup khlebandsoV- bread and salt

2.2. Second-language varieties of English

29

The expression khleb and sol' can also be idiomatically used to mean 'Russian hospitality', as in (14) Russian English (HCNVE: RE04) It \Russian culture] is not only khleb and sol' as all the people think. Summing up, RusEng exhibits some distinct linguistic features of its own. More importantly, many of these features (i.e. overgeneralisations, tense and aspect marking, etc.) bear striking similarities to those attested in second-language varieties of English.

2.2.

Second-language varieties of English

In his discussion of vernacular data, Siemund (2009) observes that specific properties attested in different varieties of English must be considered against their socio-cultural and historical background (cf Siemund 2009: 335). The following sections will show how extra-linguistic context may affect language-internal structure of a given second-language variety. Our focus is on IndEng, one of the most important varieties of English throughout the globe in terms of number of speakers (cf. Schneider 2007).

2.2.1.

Case study: Indian English

2.2.1.1. History First and foremost, IndEng is marked by enormous internal differentiation: it is partly a native variety, partly a shift variety2 and partly a second-language variety (cf. Siemund 2009). To be able to understand what gave rise to such enormous linguistic heterogeneity, we need to consider the development of this variety against its historical and socio-economic background. To start with, Engfish was introduced into India in the early if h century (cf. Schneider 2007) but it was not until the beginning of the 18th century 2 Following Thomason (2001) and Siemund (2009) the term 'shift variety' is used for those varieties that emerge when a community of non-native speakers of English gives up its mother tongue as a means of communication and shifts to English.

30

2. Non-native varieties of English

that it was in regular use. The gradual spread of English in India is intrinsically linked to the establishment of the British influence on the subcontinent. British sailors and traders were the first to bring English to India's shores followed by missionaries, who established English-medium schools in various parts ofthe country throughout the 17th, i s l a n d 19th centuries. The early English input was far from elitist since those that came to India belonged to the uneducated strata ofthe British society (cf Schneider 2007). In fact, Mehrotra (1998, cited in Schneider 2007) asserts that Northern British dialects and Cockney represented the biggest English language input at the initial stages ofthe development of English on the subcontinent. At the same time it was noted in, for instance, Mesthne and Bhatt (2008) that most ofthe missionaries in India were not ofthe working class. Thus, out of 550 missionaries based on the subcontinent in the period 17891859, only 114 were working class (cf. Mesthne and Bhatt 2008: 195). In fact, "teachers who became missionaries in India then, were drawn from the better educated, professionally conscious, ranks of teachers" (Piggin 1984: 34, cited in Mesthne and Bhatt 2008: 195). In the second half of the 18th century the British East India Company rose to a prominent position in Bengal, thus taking more and more power from the Mogul emperor, and eventually became responsible for civil government and education. Since then, the Indian population has been divided into two large groups as regards the question of whether or not English should remain the language of education. Thus, "onentalists" see English as a threat to indigenous Indian culture and languages, whereas "westernize d believe that mastenng English was the best way to gain access to Western knowledge (cf. Melchers and Shaw 2003: 137). Thus, some indigenous authors have voiced their concern that English would take over other languages of India. For instance, Y.C. Bhatnagar (1988) emphasises in his monograph "Foreign-language teaching (FTL) in India" that it is necessary to develop "operational efficiency" in one ofthe indigenous languages at the national level and other national at the state level, thus encouraging 'literate multilingualism' (cf. Bhatnagar 1988: 38). The author regards English as a threat to Indian indigenous cultures and languages. In this view, "unless the honzontal unity of the ruling class based on English is broken, India is headed for a slow and sure disintegration due to linguistic and cultural intolerance" (Bhatnagar 1988: 26). An essentially similar stand has been taken more recently in Dua (1994), who claims that "the question of switch-over from English to Indian languages in higher education is most crucial to cultural and linguistic renaissance in India". Last but not least, Bhatnagar claims that English has influenced the

2.2. Second-language varieties of English

31

study of other foreign languages (for instance, Russian, German, French, etc.) in the country negatively (cf Bhatnagar 1988: 37). Other authors, however, regard English as the principal means to solve linguistic problems that one inevitably encounters in a multilinguistic society like India. This stand is taken in Meena (1991: 104), Bailey (1996: 48), Ferguson (1996: 36) and Agmhotn and Khanna (1997: 14). The general impression is that the long-standing controversy between "orientalists" and "westermzers" led to the victory of the latter (cf. Schneider 2007: 161). It is, however, interesting to note that criticisms with respect to English as a pan-Indian medium can still be encountered in the relevant academic literature. In the 19th and the beginning of the 20th century English established itself as the language of the elite. English has become even more widespread since India gained its independence in 1947. Even the introduction of Hindi as one of the official languages of India did not stop English from becoming the most prestigious means of communication among well-to-do people and those aspiring for more prosperous positions in life alike. English is thus the principal language of higher education and government, justice and legislation, diplomacy and trade, journalism and tourism. There are basically two reasons for this phenomenon. First and foremost, English is perceived as a neutral language (as opposed to Hindi) by a vast majority of the population. As a result, it is the language in which people coming from different parts of India (for instance, Bihar and Tamil Nadu) are likely to converse. By contrast, Hindi is a language associated with a specific ethnic group. Introducing Hindi as a lingua franca implies that Hindus have a certain degree of privilege over other ethnic groups. It is for this reason that the introduction of Hindi as a national language met with great resistance on the part of many people in India. The second reason for English being so wide spread in India is that it is a language normally associated with literacy, privileged social status and economic prosperity. To provide an example, mastering English is often the only way to get a well-paid job and have a career. There are, however, many social domains from which English is barred even today. For instance, 95% of the time on state radio and TV is devoted to broadcasting in Hindi; the milestones, routes of state transport buses, number plates of private and public offices including shops and departmental stores are all in the regional languages (cf. Bhatnagar 1988: 15). In addition, results of the survey on language use in Himachal Pradesh reported by Mishra and Raj (1980) reveal that more than 40% of the listeners report listening to news and programmes in Hindi, about 15% to 19% in dialects

32

2. Non-native varieties ofEnglish

and only 13% in English (cf. Mishra and Raj 1980, cited in Bhatnagar 1988). A similar idea is expressed in Schneider (2007), who points out that "the majority of realities and lives of people on the subcontinent are untouched by the presence of English" (Schneider 2007: 161).

2.2.1.2. Indian English: some social aspects This section elaborates on some aspects of the sociolinguistic reality of IndEng. In particular, it focuses on the social characteristics of speakers of IndEng as well as the differences in conditions under which the language is acquired by different social groups. To start with, although English is spoken by a considerable proportion of the Indian population, only a very small group of South Asians (around 100,000) has English as their mother tongue. These are so-called "AngloIndians", descendants of mixed marriages between Indians and Britons. Furthermore, it has been estimated by some researchers that only 6% of the entire population of India speak English with "native-like" competence (cf. Ghosh Shantanu November 2007: personal communication).3 However if we consider the entire population that has acquired English as a second language and is proficient in it to an extent ensuring successful communication, the figure is likely to increase up to 40%. Another observation is that despite its official status, English is not always acquired as a second language in schools, in the sense that it is not always used as a means of instruction in secondary education. More often than not it has to be taught as a foreign language in many Indian schools due to the lack of financial resources. In particular, many government schools in India do not offer English as a medium of instruction, and even if they do, the teachers' level of proficiency in English is very often below the marginal mark so that subjects are often taught either in Hindi or in other indigenous languages. This has in its turn repercussions for the acquisition of English in Indian government schools: English is acquired as a foreign language rather than a second language. As a result, the level of proficiency in English varies dramatically from school to school as well as from region to region. To give an example, New Delhi is often described as the heartland of English language proficiency, i.e. most fluent speakers of IndEng come from New Delhi since the best educational institutions of secondary and tertiary levels are located here. Thus, Indians educated in private (or 3 Prof. Shantanu Ghosh, Indian Institute of Technology (IIT) in New Delhi.

2.2. Second-language varieties of English

33

public) schools of New Delhi are better at English than Indians that got their education in private (or public) schools in, for instance, Bihar. The importance of attending a good public school for receiving good education in English has been pointed out in Agmhotn, Khanna, and Mukherjee (1998): In India, being able to attend a good public school generally indicates greater exposure to English, relatively high socio-economic status, a more positive attitude towards English and a greater use of English in different domains of activity. .. .There is a very sharp contrast between 'good public schools' and 'vernacular government schools' which represent the two extreme ends of the continuum reflecting schooling. It is only in good public schools that facilities are provided to create a pleasant and comprehensible input in anxietyfree situations (Krashen 1982). These schools generally have a highly qualified teaching staff, very good libraries and several recreational facilities where children can learn while playing. It is obligatory to use English all the time in these schools. On the other hand, English is hardly used outside the classroom in government schools and the teaching generally focuses on discreet grammatical issues. The ordinary government schools generally cater to children from poor socio-economic backgrounds. There are minimal opportunities to use English, the libraries are not well-stocked and recreational facilities available are very limited. The medium of instruction in the good public schools is invariably English, whereas in the government schools it is the mothertongue. Agnihotri, Khanna, andMukherjee (1998: 114-115) Summing up the preceding discussion, it is necessary to point out that there is a considerable social stratification as well as regional variation in terms of conditions under which the English language is taught and acquired in India. The better-off upper-middle classes acquire English as a second language, whereas a considerable proportion of Indian population acquires English under conditions which are very similar to those of foreign-language acquisition. It follows that varieties of IndEng form a linguistic continuum, relating to social as well as regional factors. Furthermore, lack of sufficient knowledge of English arguably hinders the process of education at the tertiary level. This is particularly true of secondary school students from government high schools. Summing up, English has established itself as a major medium of communication (or as an inter-state lingua franca for that matter) in India, being acquired as a second language by the upper-middle class of the Indian society. English is a language that can be characterised by the upward mobility, being the most prestigious means of communication, the language of administration and the sign of power, both social and economic.

34

2. Non-native varieties ofEnglish

2.2.1.3. Essential characteristics of Indian English There has been an ongoing controversy concerning the term 'Indian English' (McArthur 1992; Hickey 2004a). The obvious problem here is that the term 'Indian English' "implies a certain unity which many scholars would maintain does not exist" (Hickey 2004a: 542). Thus, some linguists are likely to argue that referring to the Englishes spoken on the Indian subcontinent as a single variety of IndEng carries the danger of simplification, as these are extremely heterogeneous forms of English, exhibiting no linguistic characteristics common to all (cf Singh 1998; Mufwene 2000). An essentially similar view is defended by Anvita Abbi ( November 2007: personal communication),4 who contends that there are basically as many Indian Englishes as there are indigenous languages in India, the English spoken by a particular group of native speakers (Hindi, Urdu, Panjabi, Bengali, etc.) being by and large determined by the respective substrate language. Other scholars on the other hand argue that the label 'Indian English' should be preserved as a cover term comprising a host of varieties differing from each other along various dimensions such as speaker's proficiency, ethnicity, and speaker's social and regional background, etc. (cf. Schneider 2007: 172). Indeed, variation that can be observed in IndEng is frequently a matter of regional diversity, speakers' respective mother tongues exerting the major influence on their second-language output. Furthermore, variation in IndEng can be explained in many cases in terms of social stratification. We have seen in the previous section that acquisition of English in India is determined by the type of school that a learner attends, which is in its turn related to the learner's socio-economic background. Thus, in Hindimedium and also in many government schools the manner of acquisition of English often resembles that of foreign-language acquisition, most instruction being provided in Hindi or in other regional languages. By contrast, English-medium (i.e. private and public) schools including convents provide excellent facilities and a well-trained teaching staff, thus creating all necessary conditions for acquiring English as a second language rather than a foreign language. To sum up, it seems that varieties of IndEng should be stratified according to the respective substrate language and the amount of exposure to the target language including formal learning at school in order to provide a detailed account of variation of different language forms (cf.

4 Prof. Anvita Abbi, Jawaharlal Nehru University in New Delhi.

2.2. Second-language varieties ofEnglish

35

Melchers and Shaw 2003). An important observation at this point is that foreign-speaker varieties can be described along these lines as well. Speakers of IndEng (and other Englishes as well) can thus be described as forming a continuum demonstrating various degrees of proficiency in the target language. It is, however, the IndEng of speakers with high socioeconomic status that approximates the standard variety of British English most closely, while still exhibiting some idiosyncratic features of its own. Summing up the preceding discussion, varieties of IndEng produced by speakers educated in (private) English-medium schools can be defined as acrolectal varieties of IndEng. By contrast, English produced by speakers who got their instruction in (government) Hindi-medium or regional language-medium schools could be described as mesolectal and basilectal vaneties of IndEng. As a matter of fact, there is a sharp contrast between acrolectal varieties of IndEng on the one hand and basilectal varieties on the other in terms of phonology, lexis and morpho-syntax. Mesolectal forms of IndEng fall somewhere in between incorporating the features of both. We now turn to describing some salient features typical of the acrolectal and basilectal varieties of Ind-Eng, hence demonstrating to what extent both forms of IndEng differ from each other.5

2.2.1.3.1. Acrolectal varieties of Indian English Phonology First and foremost, many (if not all) varieties of IndEng are rhotic, thus demonstrating an obvious similarity to Standard American English. The IndEng M is, however, often trilled or tapped (as in Scottish English). Second, StEng alveolar hi and /d/ may be pronounced as retroflex. Similarly, the voiceless stops /p/, hi, Dd may be unaspirated. Here, a certain amount of variation is expected as a result of different levels of mastery of English. Furthermore, standard conformant varieties of IndEng lack diphthongs /ei/ and W, replacing these phonemes with long vowels /e:/ and /o:/ respectively so that words such as name and coat are pronounced as /ne:m/ L /ko:t/ (cf Mukherjee 2007). Finally, British English / 3 / is replaced with the voiceless post-alveolar fricative /J/ by some IndEng speakers. However, this phonological feature is particularly characteristic of Bengali speakers of IndEng.

5 The ensuing analysis employs the data from the Hamburg Corpus ofNon-Native Varieties ofEnglish. All remaining sources are indicated accordingly.

36

2. Non-native varieties of English

Moreover, IndEng has a number of characteristic prosodic features. A general pattern is that most IndEng sentences end with a rising tone, which serves as a marker of politeness. Furthermore, word stress has been described as non-distinctive (cf Melchers and Shaw 2003) and subject to individual variation. In addition, word stress in IndEng is presumably influenced by the speakers' respective native languages. Moreover, sentence stress and intonation system does not appear to be as important to conveying the message as they are in British English and American English (cf. Melchers and Shaw 2003). Furthermore, IndEng can be described as a syllable-timed variety of English as opposed to stress-timed native varieties of the language. The pitch range is, however, much wider than that of Standard English and the function of volume change may have different functions when compared to the native standard. This short survey of the major phonological features of acrolectal vaneties of IndEng reveals that these varieties indeed have systematic phonological characteristics which makes them quite distinguishable from other varieties of English, both native and foreign-speaker varieties. A noteworthy observation is that the degree to which all above-mentioned features are maintained within a particular speaker strongly depends on her level of proficiency in English. Thus, in certain cases, features just described may not be attested in speech of many educated speakers of English in India as they may have trained away some phonological features traditionally associated with IndEng. As is clear, this observation provides a parallel to the one we have made about Russian speakers of English, the preceding discussion having demonstrated that variation in non-native varieties of English is largely proficiency-dependent. Morpho-syntax In the area of morpho-syntax the contrasts between acrolectal IndEng and Standard English are less obvious. Furthermore, the formal written register of IndEng generally shows signs of being a British English conformant variety. In this respect, formal written IndEng can be compared to the English written in Russia, which has always been under the great influence from Standard British English. At the same time, some researchers observed that the grammar of IndEng vernacular (i.e. spontaneous speech) is as diverse as the indigenous languages spoken on the subcontinent (cf. Abbi, Anvita November 2007: personal communication). However, it is so far unclear which features systematically represent which variety of IndEng since there are to the author's best knowledge - no in-depth studies of Indian Englishes

2.2. Second-language varieties ofEnglish

37

representing different substrate languages (for example, Tamil English, Bengali English, and so on). Yet some grammatical features can be described as characteristic of all acrolectal (or educated) varieties of IndEng irrespective of the speakers' mother tongue. Hence, one grammatical feature that has been traditionally analysed as being typical of different vernacular (or informal) varieties of IndEng is invariant tag questions. In particular, the invariant tags isn't it and no have been described as being characteristic of the informal register of educated IndEng (cf Bhatt 2004: 1021). (15) Acrolectal Indian English (spoken vernacular; from Bhatt 2004: 1022) a. We can sit here, no? b. You said you will do the fob, isn 't it? However, it must be noted that some speakers have a preference for one form rather than the other. For instance, Bengali speakers of English use the particle no as the invariant tag form. Furthermore, one important empirical observation is that the invariant tag isn 7 it is not very frequent in the speech of Hindi speakers, being often replaced by the invariant tag M, which is a clear substrate influence from Hindi. To be more exact, ha is a literal translation of 'yes' in Hindi and performs a number of pragmaticdiscourse functions in an utterance, signalling politeness and interest on the part of the speaker. This analysis of the functions of the invariant tag ha is compatible with the analysis of the invariant tag isn 7 it proposed in Bhatt (2004: 1022). Another salient feature typically associated with acrolectal IndEng is the expansion of function of the progressive forms. In educated IndEng vernacular progressive forms frequently surface in contexts from which they would be barred in Standard English. The progressive aspect expanded its functions to habitual contexts and to stative verbs in IndEng with the result that progressives are perceived as normal in these contexts by most educated speakers of IndEng. (16) Acrolectal Indian English (spoken vernacular) a. I am thinking that it is a good idea. b. Sometimes I am easily falling ill. Furthermore, IndEng educated vernacular is characterised by the absence of the auxiliary-subject inversion, as in (17).

38

2. Non-native varieties ofEnglish

(17) Acrolectal Indian English (spoken vernacular) Why I would tell this story? Finally, IndEng has undergone some morphological innovations. First, some English nouns have been converted to verbs in IndEng. Take, for instance, words such as horn, airline, slogan and public that can surface only as nouns in native English. In IndEng, however, these nouns can be used as verbs as well. Lexis It is impossible to give a comprehensive account of the varieties of IndEng, both acrolectal and basilectal varieties, without mentioning some lexical features constituting the idiosyncratic character of these forms of English. First and foremost, some indigenous words have become a part of the vocabulary stock of some northern varieties of IndEng, where Hindi predominates as a regional language. As a rule, these words describe objects and phenomena essential to Indian culture and Indian way of life. Here are some examples demonstrating which English nouns have become permanently replaced with equivalents from Hindi: (18) Indian English roti,paratha- bread

cW-nce masala- spice daal- lentils prarihna-wxf* puja- the act of worshiping Second, some Hindi verbs have found their way into the vocabulary of IndEng even though they are considered to belong to the informal style of IndEng (cf Som, Bidisha November 2007: personal communication)6 (19) Acrolectal Indian English (spoken vernacular) a. Could you please sit still, don't hilo. '...don't fidget' from Hindi hilo meaning 'to make fidgeting movements while sitting' 6 Dr. Bidisha Som, Jawaharlal Nehru University in New Delhi.

2.2. Second-language varieties ofEnglish

39

b. Stop chatofying me! 'Stop boring me by talking too much!' from Hindi ca^a meaning'to talk' Third, some Hindi nouns are used in the function of verbs in IndEng, as in (20) Acrolectal Indian English (spoken vernacular) He wets line fitciYofvxfiQ with her 'He was engaged in mild flirtations with her.' from Hindi line mama meaning 'mild flirtation' Furthermore, some English words have undergone a change of meaning in IndEng. To give an example, the intransitive verb bunk, which has a meaning 'to sleep in someone else's house' has become transitivised and has undergone a complete change of meaning in IndEng. (21) British English (Longman Dictionary of Contemporary English) You can bunk down on the sofa for tonight. 'You can sleep on the sofa tonight' (22) Acrolectal Indian English (spoken vernacular) / bunked the lecture today. 'I missed the lecture today.' Finally, the vernacular vocabulary of IndEng can be characterised by the abundant presence of abbreviations specific to the country's cultural realities. To provide some examples, CNG (Compressed Natural Gas), ST (scheduled tribes) and SC (schedules castes) reflect local and social reality of India.

2.2.1.3.2. Basilectal varieties of Indian English As mentioned above, basilectal varieties are produced by speakers who have typically not been exposed to English as a medium of instruction, having attended Hindi-medium government schools. Such speakers are characterised by a very limited amount of contact with the target language. Basilectal varieties of IndEng exhibit a number of idiosyncratic features in the area of phonology and grammar which make them quite distinct from acrolectal varieties.

40

2. Non-native varieties of English

The first empirical observation is that only basilectal (and sometimes mesolectal) speakers of IndEng insert an initial vowel before the consonant clusters /sk/, /si/ and /s P / in onset positions, as in Askul/ school, /istejan/ station, /is P i:ty speech, A s l a n t / student, etc. Second, omission of copula BE and (in)defimte articles is typical of basilectal varieties of IndEng. (23) Basilectal Indian English (spoken vernacular; from Mukherjee 2007) a. This BaranasQ very old city. b. Varanasi 0 our very oldest city in India. Third, basilectal speakers of IndEng do not usually mark the verb for the 3 rd person singular, as in (24) Basilectal Indian English (spoken vernacular) This radio soundQ good. Finally, neither the progressive nor the perfect surfaces regularly in basilectal IndEng, which forms a rather sharp contrast to the acrolectal varieties of IndEng. Hence, we could assume that morphological and semantic complexity of these morpho-syntactic structures leads to their underrepresentation in the speech of those speakers who have not attained sufficient levels of proficiency in their target language.

2.3. Summary The major purpose of this chapter was to work out a common ground which would allow us to compare varieties of the outer and expanding circle despite the marked difference obtaining between them. Several generalisations can be made at this point of our discussion. First and foremost, although foreign-speaker Englishes are not used in as many domains as second-language Englishes, linguistic variation observed across varieties of the outer and expanding circles yields consistent descriptive accounts (see also Kachru, Kachru, and Nelson 2006). Examples from RusEng showed that the foreign-speaker varieties exhibit structural properties fairly similar to those attested for second-language varieties of English. It follows that providing systematic accounts of languageinternal variation across non-native Englishes might shed some light on the

2.3. Summary

41

universal mechanisms underlying the occurrence of language forms in these varieties (cf Davydova et al. 2011). Second, varieties of both types exhibit extreme language-internal heterogeneity, resulting in a continuum of basilectal to acrolectal forms. This heterogeneity is best understood in terms of socio-economic conditions under which English is acquired in the countries of the outer and expanding circle and in terms of the sociohistoncal contexts in which these non-native Englishes emerged (cf. Schneider 2007: 4). More importantly, we have seen that the description of phonological and morpho-syntactic features found in these varieties largely depends on parameters such as speakers' proficiency (or the amount of contact with the target language), speakers' native tongues, stylistic register, etc. It follows that studying variation of different linguistic forms in varieties of both types must ideally include the aforementioned (extra)-linguistic variables. Only a detailed investigation of these variables will allow us to assess their exact role in constraining variation of the linguistic phenomenon under analysis. Last but not least, we can generally expect that differences in (extra)linguistic settings will give rise to the marked differences between individual varieties (cf. Schneider 2007: 4). By contrast, similarities in the sociolinguistic background will arguably help to account for similar outcomes in variation of linguistic forms observed across different Englishes.

3. Some theoretical preliminaries Before embarking upon the task of analysing the occurrence of the perfect across different varieties of English, it is necessary to elaborate on some theoretical issues concerning the status of the perfect as a verb category. Furthermore, it is important to consider the relations that obtain between the perfect and another form frequently surfacing in all present perfect contexts in all varieties of English studied here: the preterite. The major purpose of the ensuing chapters is, hence, to unify the existing contrasts between these verb forms. It will be without any doubt very useful to consider the present perfect and preterite in the light of the categories "most commonly referred to in discussions of how the English verb expresses time: tense and aspect" (Elsness 1997: 14). 3.1. The category of tense Despite the fact that there is no unanimous agreement in the academic world with respect to the question of how or in what terms the category of tense should be defined, many researchers agree that tense is a deictic category, i.e. it relates actions, events or situation to a certain time point also known as the deictic centre. In most languages, the time of utterance (the present moment to be more exact) forms the deictic centre, and tense performs the function of locating situations and events before, after or simultaneously with the time of utterance. This view is explicit in Reichenbach (1947), Lyons (1969), Comne (1985), Komg (1995) and Le Poidevin (2002), who believe that (i) tenses express relations such as 'before' (), 'is included in' (C) and 'is simultaneous with'; (n) the arguments of these relations are speech time (S), event time (E) and some other time point known as reference time ( R ) / A different stance on the issue of tense is taken in Klein (1994), who argues that the category of tense "expresses a relation between the time of utterance and some time for which the speaker wants to make an assertion - the 'topic time'" (Klein: 1994: 24).2 Probably the most recent approach to tense is found in Radden and Dirven (2007: 201-232). Since it will be adopted for the analysis of the 1 Comrie (1985) uses a slightly different but entirely compatible terminology: the speech time (S), the event time (E) and the reference point (R). 2 See Klein (1994) for further details.

3.1. The category offense

43

present perfect in the ensuing chapters, we examine their position on the category of tense in some detail here. First, Radden and Dirven (2007) state that tense "relates to the way a situation is located in time from the speaker's viewpoint" (Radden and Dirven 2007: 201). Tense, however, also relates to the presentation of a situation as a mental space in the speaker's and hearer's mind (cf Radden and Dirven 2007: 202). Speech time (S), event time (E) and reference time (R) are treated as parts of mental spaces. The moment of speaking serves as a deictic centre, which allows the speaker to refer to three different time spheres representing deictic times: the present time sphere, the past time sphere lying behind the speaker and the future time sphere lying ahead of the speaker. Deictic times, which relate to speech time, are typically expressed by what Radden and Dirven call simple tenses. Simple tenses are the present tense, which locates the situation at, around, or included in speech time; the past tense, which locates a situation at a time earlier than speech time, and the future tense, which locates a situation at a time later than speech time. By contrast, complex times always involve two temporal relations: relation between speech time and a deictic time and a relation between the deictic time as a reference time and the time of the anterior or posterior event (cf. Radden and Dirven 2007: 204-205). In other words, complex times are characterised by a reference time, which links an event to speech time (cf. Radden and Dirven 2007: 207). Complex times are expressed by complex tenses. Complex tenses in English include past perfect (had left) and past prospective {was going to leave), present perfect {has left) and present prospective (is going to leave), future perfect (will have left) and future prospective (? will be going to leave). As is clear, the major difference between simple tenses and complex tenses is that the latter involve a reference time, and thus express more elaborate temporal relations, whereas the former do not. Moreover, some researchers are likely to support the idea that a verb form marked for tense can be treated as a polysemous category in natural languages. For instance, in his monograph on tense and aspect, Petersen (2004: 156) contends that English verb forms are "non-monadic", that is to say that one verb form may have more than one meaning. The basic function of tense is to locate situations and events on the time axis, thus grounding a situation in time, and its meaning can in principle be defined in terms of temporal relations.3 Furthermore, in some studies it has 3 Yet proponents of the tradition of Functional Semantics believe that the meaning of the category of tense cannot be described in time referential terms; in fact, Functional Semantics does not define linguistic meaning in terms of reference

44

3. Some theoretical preliminaries

become a tradition to distinguish between absolute tenses, relative tenses and absolute-relative tenses (see Comne 1985 for a detailed discussion). Another important facet of the linguistic category of tense is its formal representation in a given language. It is for this reason that tense is defined as a "grammaticahsed expression of location in time" (Comne 1985: 9). Tense is, therefore, a grammatical category, i.e. it is integrated into the grammatical system of the language. Tense is indicated on the verb, either by verb morphology (for example, affixes, vowel alternations, etc.) or by grammatical words such as auxiliaries. In some languages, tense is expressed by particles (cf Klein 1994). Many traditional grammars argue that tense is a verbal category on the basis of its morphological attachment to the verb. Comne's claim is that this category must be ascribed to the whole sentence or to the whole proposition since "it is a truth value of the proposition as a whole... that must be matched against the state of the world at the appropriate time point" rather than just a property of the verb. This comment is quite stimulating since it implies that elements other than those pertaining to the field of semantics should be considered when giving an exhaustive account of the grammatical category of tense. It will be suggested here that the description of the meaning of tense cannot and - more importantly - should not be restricted to semantics only, the discipline traditionally associated with the investigation and explanation of the invariant meaning, and must involve the investigation of pragmatic elements as well. This claim can be substantiated by the following observation. In German, the so-called present tense can be employed to refer to the future: (25) German Peter kommtjeden Tag um 6 Uhr abends nach Hause. 'Peter comes home every day at 6 p.m.' (26) German Peter fahrt morgen nach Berlin. 'Peter is going to Berlin tomorrow.' The overall tense meaning of (25) and (26) can be deciphered with the help of lexical modifiers jeden Tag, morgen used in the sentence. In some cases, and referential categories at all (Petersen 2004: 199). It does not fall within the scope of this book to discuss all the advantages and disadvantages of that approach. Let it suffice to refer here to the exhaustive account of this approach provided in Petersen (2004).

3.2. The category of aspect

45

however, the meaning of a particular tense form cannot be inferred from the overall meaning of the sentence. Linguistic information is not enough in such cases; it is at this point that extra-linguistic factors start playing a decisive role in analysing the meaning of the grammatical category under study. Let us consider the following example: (27) German — Wo ist Peter? 'Where is Peter?' — Er koifijfit UYH 6 Uhr yictch Hctusc

'He comes home at six.' The linguistic context alone does not allow for the correct interpretation of this bit of dialogue. The hearer needs additional information about the participants of the conversation (such as the conversation between Peter's daughter and wife), some particular knowledge about Peter's habits (for example, the fact that he regularly comes back from work around 6 o'clock) as well as some knowledge about the time at which the conversation takes place (let us say 5.30 p.m.) to be able to interpret the proposition Er kommt nach House um 6 Uhr as Er wird um 6 Uhr zu House sem (He will be coming home oround 6 o 'clock) and, therefore, as an utterance with the future reference. The extra linguistic context, or rather an implicature which arises out of it, thus allows the hearer to attribute the correct time reference to the verb form. An important inference can be made from this observation: the description of tense, a traditionally grammatical category, thus involves not only morphological and semantic criteria but a pragmatic parameter as well. Later discussion will demonstrate the role played by the pragmatic context in defining the meaning of the present perfect and another closely related form-the preterite. At this point, however, we raise an issue concerning the grammatical status of the perfect: Is it a tense or is it something else? Since there is no hard-and-fast answer to this question, we consider this verb form in the light of another linguistic category typically mentioned in discussions on time encoding of the English verb - aspect.

3.2. The category of aspect Aspect is yet another fairly controversial term in linguistics. To start with, 'aspect' is a translation of Russian vtd (which is in its turn a translation

46

3. Some theoretical preliminaries

from Latin) traditionally used to refer to perfective/imperfective opposition. Nowadays this term is used to cover a variety of other "oppositions based upon the notions of duration, instantaneity, frequency, initiation, completion" (Lyons 1977: 705) that receive a grammatical encoding in a given language. The most prominent aspectual opposition in English is that between progressive and non-progressive verb forms, "which at least in its most central uses can be explained in terms of the fundamental aspectual concepts of perfectivity" (Elsness 1997: 17). Klein (1994) views aspect as the relationship that obtains between "the time of some possible situation", or time of situation in Klein's terminology, and "the time span to which the speaker's claim is confined", i.e. topic time. Klein's (1994) contention is that aspect expresses the way in which some situation is hooked up to some topic time (cf Klein 1994: 6). Lyons (1977: 689) notes that there is an obvious connection between temporal notions and other notions typically classified as aspectual, pointing out that "anteriority is not always distinguishable from completion and termination" (Lyons 1977: 689). For this reason some linguists find it difficult to decide whether the present perfect and past perfect in English differ from the respective non-perfect forms in terms of tense or in terms of aspect (cf. Lyons 1977: 689-690). Anderson (1973a, cited in Elsness 1997: 18) suggests that aspect is "concerned with the relation of an event or state to a particular reference point: it is located before (retrospective), after (prospective) or simply at (AORIST) a particular point in time" (Anderson 1973a: 39-40, cited in Elsness 1997: 18). The major problem with this approach is that the boundaries between the categories of tense and aspect become blurred under this definition of aspect as a reference point is frequently included into the description of the category of tense (cf. Comne 1985: 65). The stance taken here is that aspect is "the grammatical form used by a speaker in taking a particular view of a situation" (Radden and Dirven 2007: 175). In most general terms, aspect evaluates a situation or an event as it progresses in time irrespective of the moment of utterance (cf. Maslov 1988: 63). For that reason it is characterised as a non-deictic category primarily concerned with describing an "internal temporal constituency of a situation" (Comne 1976: 3), thus viewing it [the situation] as if from within. Aspect is in this sense very different from tense, which is a deictic category. Summing up, aspect has been defined differently by different authors. The answer to the question about whether or not the perfect can be considered to be an aspectual category is by and large determined by the definition that one chooses to operate with (cf. Elsness 1997: 18).

3.2. The category of aspect

47

3.2.1. Aktionsart Another category closely connected (and sometimes confused) with the category of aspect is that of Aktionsart, i.e. a lexical category which basically expresses the same type of notions (duration, frequency etc.) as aspect does. This notion plays an important role in this study since one of its objectives is to find out whether or not the lexical meaning of the verb is implicated in the choice of the present perfect in the standard variety of English and in non-native varieties. We will therefore discuss this term in some detail here. The term Aktionsart was introduced into the study of language by the linguists and philosophers interested in demonstrating (i) how verbs can incorporate the features of duration, dynamic processes and change; (n) how the categorical interplay between tense, aspect, and Aktionsart occurs. The best-known typology establishing the ontological division state/process/ event with regard to verb semantics is the one provided by Vendler (1967). His classification is based on the assumption that all verbs possess inherent lexical properties such as states, processes, occurrences, achievements, and so on, and can be categorised according to these features. Vendler thus distinguished between four classes of verbs expressing states, activities, achievements and accomplishments which can be best illustrated as demonstrated below.

state

6

3

activity

g

A-

achievement accomplishment Figure 3.1. Vendler's verb classes typology (adapted from Petersen 2004: 57) The clearest way to describe Vendler's four categories is by means of a number of semantic features, namely [± stative, ± durative, italic, ±voluntary]. The four classes can thus be characterised as follows:

48

3. Some theoretical preliminaries

state

+stative + durative - telic - voluntary

activity

+ stative + durative -telic ± voluntary

accomplishment

- stative + durative + telic ± voluntary

achievement

-stative -durative ± voluntary

Figure 3.2. Semantic features of Vendler's verb classes (adaptedfromBrintonl988:29) These lexical verb classes will not be discussed in any detail here as a comprehensive account of all verbs is provided in the chapter on methods, including criteria for classifying specific verbs found in the data as states, activities, accomplishments or achievements. The notion of Aktionsart is also important here because it is directly related to what has become known as the Inherent Lexical Aspect Hypothesis (IAH) in second-language acquisition studies, which predicts the path of acquisition of the tense and aspect morphology (cf Andersen 1991; Housen 2002). In his account of the development of tense and aspect in English as a second language Alex Housen (2002) notes: In its most general reading, the Inherent Aspect Hypothesis states that the emergence, early use and development of TA [tense and aspect] morphology in language acquisition is influenced by the inherent semantic properties of the verb predicate which the learner uses to refer to a particular situation (Andersen 1991; Andersen and Shirai 1996; Bardovi-Harlig 1999). These semantic properties are aspectual in nature and are most commonly defined in terms of Vendler's (1967) model of inherent verb semantics. Housen (2002: 165) Housen (2002) furthermore provides specific claims with respect to the development of the past tense morphology in English: a. During at least some early stage of morphological development, learners associate and use past and perfect tense morphology (hereafter PAST) predominantly with prototypical punctual-telic predicates, or achievements (e.g., fell, dropped); atelic and durative verbs tend to remain uninflected (e.g., want, play, grow up). At this stage, learner's use (or non-use) of PAST morphology is largely independent of the grammatical and discourse-

3.3. The present perfect: tense or aspect?

49

pragmatic values of the TL [target language] (e.g., past and perfect time reference, grounding status in narrativediscourse). b. At later stages of development, the initial strong bias of PAST is progressively extended, first to less prototypical achievements and then to other predicate types, following a systematic pattern of lexical diffusion... The final stage of development is all verbs in past- and perfect-time contexts are properly marked for tense, irrespective of their inherent aspectual values. Housen (2002: 166) In most general terms, the Inherent Lexical Aspect Hypothesis predicts that Aktionsart must be implicated in the choice of the present perfect in mesolectal varieties of English (i.e. in varieties whose speakers have had moderate exposure to the target language). More specifically, this hypothesis predicts that less advanced speakers of English (such as mesolectal speakers) should favour the present perfect (i.e. constructions comprising an auxiliary HAVE and a past participle) and the preterite (i.e. verb stems inflected for past time reference) with verbs of achievement and accomplishment. By contrast, stative verbs and verbs of activity (otherwise known as dynamic verbs) can be expected to be disfavoured with both verb forms.

3.3. The present perfect: tense or aspect? Having sketched the major features characteristic of tense and aspect, we can now pose a legitimate question with respect to the relevance of these categories for the perfect. Should the perfect be regarded as a tense, an aspect or a separate category in English? This issue has remained unsolved so far, researchers frequently expressing divergent opinions on the subject. Thus, there are some who believe that the perfect should be subsumed under the category of aspect (Comne 1976; Quirk et al. 1985; Klein 1994; Biber et al. 1999), although some of them admit that "it is an aspect in a rather different sense from other aspects" (Comne 1976: 52). Others (Mustanoja 1960; McCoard 1978; Inoue 1979; Fenn 1987; Elsness 1997; Hantson 2005; Declerck, Reed, and Cappelle 2006; Radden and Dirven 2007) consider the perfect to belong to the category of tense on the basis of the time reference expressed by this verb form. In addition to that, some authors tend to assign the perfect to a separate grammatical domain (Dahl 1999; Musan 2002) since "it is sufficiently uniform in its semantics and morphosyntactic behaviour to be considered a cross-linguistically valid category" (cf Dahl 1999: 290). More recently, Michaelis (2006: 224) suggested that the perfect

50

3. Some theoretical preliminaries

be viewed as an aspectual construction that "may function as a tense without losing its aspectual properties"4 In the section dealing with aspect it was mentioned that whether the perfect can be recognised as an aspect depends on the definition of aspect one operates with. Taking the definition of aspect adopted in this study into account, the category of the perfect can hardly be regarded as an aspect in a strict sense, although one could argue that the perfect possesses some sort of aspectual meaning with respect to "the inherent structure of the overall situation" (Radden and Dirven 2007: 206). The position advocated here is to regard the English perfect as a complex tense in the sense of Radden and Dirven (2007). Such an approach seems to have its own advantages for the study of linguistic variation since it allows us to regard the perfect and other forms such as the present tense or the simple past tense as representatives of the same linguistic category (i.e. the tense). It is in this way that we can ensure a homogenous perspective on comparing the perfect and other tense forms and investigate these grammatical phenomena systematically. To give an example, the present perfect and preterite are semantical^ quite distinct; yet these forms share the same cognitive meaning (i.e. reference to the past) and can for this reason be treated as variants of the same linguistic variable or functional realisations of a "common underlying formin present perfect contexts (cf Tagliamonte 2006: 72). Similarly, the present tense and the perfect are far apart from each other in terms of their semantics. However, when used in extended-now contexts, both verb forms refer to an event or situation that started in the past and continues to exist at the moment of utterance. In fact, "the linguistic variable need not to be confined to cases in which the variants necessarily mean precisely the same thing" (Tagliamonte 2006: 76). It is "the sameness of cognitive meaning" or general functional equivalence that is believed to be the relevant criterion for identifying grammatical forms under analysis as variants of the same linguistic variable (cf. Winford 1993: 142).

3.4. Summary We have so far established that tense is a deictic category, whereas aspect is not. Furthermore, a verb form marked for tense can be regarded as a non4 For further discussion see Michaelis (2006: 223-224).

3.4. Summary

51

monadic (polysemous) category, which means that "it is not the case that there is one and only one meaning of a single [verb] form" (Petersen 2004: 156). Additionally, aspect is not to be confused with Aktionsart, which is a lexical, not grammatical category. An excursion into recent accounts on tense, aspect and Aktionsart has shown that three categories are closely intertwined; it will, therefore, be interesting to try to find out if the inherent lexical meaning of the verb is implicated in the choice of the perfect in a non-native grammar. Last but not least, following the definitions of tense and aspect proposed in this study, we argued that the English perfect is better analysed as a tense category rather than an aspect category.

4. Towards a theoretical explanation for variation between the present perfect and preterite

As mentioned earlier, the only other form which alternates with the present perfect in all present perfect contexts is that of the preterite. Indeed, the preterite is the major functional competitor of the HAVE-perfect across different varieties of English and needs to be considered in some detail here. In order to determine what mechanism may underlie variation between these verb forms, we need to elaborate on a general theoretical perspective which will help us to describe the most essential characteristics of the present perfect in StEngEng and oppose them to other verb forms, in particular to the preterite. This chapter argues that the most salient features of the present perfect are best understood in the light of the notion of current relevance.

4.1. The preterite Since there is a general agreement with regard to the grammatical status of the preterite, we will start the discussion with this verb form. The first observation is that the preterite is - to use Comnan terminology - an absolute tense, i.e. it "includes as part of its meaning the present moment as its deictic centre" (Comne 1985: 36). The preterite is thus deictic, i.e. the meaning of this verb form depends on a certain contextual feature (the moment of utterance to be more exact) which forms a deictic centre in relation to which the event expressed by the preterite is located on the time axis. Furthermore, the English preterite has a morphological marker -ed inflected on the verb, which is one of the strongest arguments in favour of the inclusion of the preterite into the tense system of English. The preterite places a situation or an event prior to the time of utterance on the time axis (cf Petersen 2004: 105): El

i r

TU Figured.

Thepasttense

4.1. Thepreterite

53

The basic function of the preterite is hence the location of the situation prior to the present moment. Moreover, the preterite performs another important role in Standard English, i.e. the one of the narrative tense. It is therefore useful to make a distinction between these two functions of the preterite: (i) the reference to past actions and events occurring at a specific moment in the past and (11) the narrative function, used for describing a string of events which occurred one after the other at some time in the past; these events are viewed by the speaker as not being connected to the time of utterance. Special attention needs to be given to contexts featuring time adverbials such as never, ever, always etc.. Such environments are associated with the perfect in some varieties of English and with the preterite in others. (28) American English (from Fenn 1987: 127) Did you have lunch (yet, already)? (29) American English (from Vanneck, 1958: 237, cited in Fenn 1987: 95) IUve in New York. But I never saw the St. Patrick's Day Parade. (30) Standard British English I have never been to New York. (31) Irish English (HCIE: Dunne 12) I never had so much to do as~at present. Another important observation here is that the narrative context is generally the domain of the preterite across varieties of English.1 It follows that the occurrence of ever, never, yet, etc. in narrative contexts is not relevant for our discussion of the perfect. (32) Standard British English (beginning of a narration) My grandmother was a queer creature. She never wore pearls like most women ofher generation did. The meaning of the preterite can be defined as the one of definite past time, the element of defimteness being suggested either by the speaker or by adverbial specification (Elsness 1997). The hearer frequently decodes this element of definite meaning relying on the context in which the proposition 1 Australian English is perhaps the only exception to this - otherwise pervasive pattemofuse.

54

4. Explaining variation between the present perfect and preterite

is being uttered. As a rule, the speaker and the hearer are assumed to have some shared knowledge of the world upon which their interaction is based. The information about the temporal location of the verbal situation (i) can be inferred from a wider linguistic context or (11) can be provided by some extra-linguistic situation. If no such information is available in the context, the communication will probably fail. Defined as a notion of definite past time, the preterite often appears with overt indicators of definite past time such as past time adverbials yesterday, in 1999, some day ago, etc. or in contexts suggesting a definite past time reference (i.e. narrative contexts). By contrast, the HAVE-perfect is ungrammatical in such environments - at least in Standard English: (33) Standard English a. / bought it years ago. b. */ have bought it years ago. However, it is sometimes the case that verbs marked for past tense as well as the HAVE + past participle morphology occurs without any explicit time signal in a sentence: (34) Standard English a. I wrote a letter. b. / have written a letter. In such cases, the preterite relates an event or a situation to some definite time in the past and it often becomes clear from wider contexts (either linguistic or extra-linguistic) to what time in particular the speaker refers. Later in this chapter we will see how situational (extra-linguistic) context can convey the idea of definite past time reference and thus make the hearer decode the message as the one referring to a particular moment in the past.

4.2. The present perfect The analysis of the perfect is generally highly controversial and, as shown earlier in the discussion, there is little agreement about the grammatical status of the English perfect. Moreover, there is just as little consensus on the issue of the semantics of the English perfect. Some scholars argue that semantical^, the perfect can be analysed in terms of several independent meanings or readings (see Leech 1971a, 1971b; McCawley 1971; Comne

4.2. The present perfect

55

1976; Dahl 1999; Dahl and Hedin 2000; Lindstedt 2000) and, therefore, understood as an ambiguous verbal category. Others (McCoard 1978; Inoue 1979; Meyer 1992, 1995; Klein 1994; Musan 2002; Radden and Dirven 2007) postulate one central meaning for this verb form, claiming that all other readings are mere pragmatic realisations of the prototypical meaning. However, opinions are divided as to what meaning should be considered the fundamental property of the perfect. Consonant with the approach defended in Leech (1971a, 1971b), Comne (1976), Dahl (1999), Dahl and Hedin (2000) and Lindstedt (2000), the stance taken in this study is that the perfect is a polysemous or, to use Petersen's (2004) terminology, non-monadic verbal category. It differentiates between as many as four established meanings (or senses) which can be described in terms of its basic use types.2 The major argument in favour of this approach is that not all languages that have forms with perfect meaning have the full range of the meanings rendered by the English perfect, while in some languages there are distinct forms for some of these meanings" (Comne 1976: 56). This typological observation captures one important insight: Various manifestations of the perfect can be regarded as cross-linguistically valid semantic categories that may (or may not) receive a grammatical encoding in a language (cf Dahl 1985: 133, 142; Haspelmath et al. 2005: 271). In other words, specific languages may grammaticalise either some or all of these meanings. It is on these grounds that the English perfect can be assumed to be a polysemous linguistic category, which has one grammatical marking for the expression of several cross-linguistically valid semantic properties of the perfect. Additionally, all the types of the English perfect are established readings in the sense that they are described at some length in traditional grammars of English (cf. Quirk et al. 1985; Huddleston and Pullum 2002). Following Cruse (2000: 109), established readings of a linguistic item are meanings (or senses) of that item.

2 This approach is also roughly compatible with that of Declerck, Reed, and Cappelle (2006), who distinguish between two meanings of this verb form - a 'before now' meaning and a 'co-extensive' meaning (cf Declerck, Reed, and Cappelle 2006: 234). The similarity between the two approaches is revealed by the fact that the authors acknowledge that the present perfect can be regarded as a category with more than one meaning or function.

56

4. Explammgvanation between the present perfect and preterite

4.2.1. The meaning of the present perfect In most general terms, the English present perfect can be characterised in terms of three properties: focus on the present time, current relevance and indefimteness (Radden and Dirven 2007: 212). More specifically, the focus on the present time makes the present perfect compatible with present time adjuncts (e.g., Our tram has just left and now we 're stranded here. Cited from Radden and Dirven 2007). The component of current relevance implies that a past situation is still (perceived as) relevant at speech time. The overall indefimteness expressed by the perfect manifests itself in the fact that this verb form is incompatible with adjuncts that specify a definite setting (cf Radden and Dirven 2007: 213). This definition of the English perfect can be conceived of as a default reading, i.e. "a meaning that would come to the mind in the absence of contextual information" (Cruse 2000: 108). It is this reading that every competent native speaker of English should in principle be able to reconstruct when presented with the HAVE-forms (e.g., have written, have done, have been, etc.) without any context. The major uses of the English perfect listed below may be viewed as more specific and contextual^ restricted senses, representing subvaneties of this very general sense.3 What all the senses of the category have in common is a semantic component of current relevance. This makes possible an account of the present perfect in terms of polysemy (cf. Cruse 2000: 109) 4 The ensuing analysis, however, will show that some use types are more central to the category of the perfect than others. In what follows, we will elaborate on the semantics of these use types, while relying on the state-of-the-art typological and cross-varietal accounts of the perfect. This procedure will allow us (i) to establish the ground for a contrastive account of the perfect and preterite in the subsequent discussion and (ii) to build up hypotheses for the ensuing analyses of the present perfect in the standard variety of English and related non-native varieties.

3 This type of polysemy was referred to as autohyponymy in Cruse (2000: 110). 4 In his account of polysemy Cruse (2000) states that a linguistic item can be can be viewed as polysemous if its two or more established senses are related to each other.

4.2. The present perfect

57

4.2.1.1. The resultative perfect The resultative perfect (referred to as the strive perfect by McCawley 1971) refers to a situation that inherently involves a change of state (Huddleston and Pullum 2002: 145), thus expressing a resultant state. (35) Standard English (from Huddleston and Pullum 2002: 145) a. She has broken her leg. He has closed the door. She has gone. b. She's been to the bank. She has run ten kilometres. As the preceding examples demonstrate, resultative perfects are often (but not always!) associated with verbs of accomplishment and achievement. The meaning of the resultative perfect is considered to be a "central manifestation of a perfect" (cf Comne 1976; Siemund 2004) and seems to be the most explicit with respect to the expression of the notion of current relevance, a concept typically associated with the category of the present perfect. (36) Standard English (from Siemund 2004: 414) She has eaten the pizza all on her own. (There is no pizza left on the table) In order to understand how the notion of current relevance is manifested in the meaning of the resultative perfect, let us consider the sentences John broke Ms leg two months ago vs. John has broken Ms leg [and can't walk right now]. Thus, in the first sentence, a simple reference to an action which took place at some time in the past is being made and it is not at all clear form the proposition made in the first sentence whether John's leg has since then been healed or is still broken, whereas in the second sentence the result of a prior situation still holds for the present, i.e. John's leg is still broken at the moment of utterance. This type of result is known as a continuing result: the resultant state begins at the time of occurrence of the past situation and continues through into the present (cf. Huddleston and Pullum 2002: 145). Continuing results generally correlate with verbs denoting a change of state and current relevance can thus be said to derive directly from the meaning of these verbs. Yet in other cases current relevance is heavily dependent upon pragmatics of resultative perfects. For instance, in She's been to the bankthe meaning of current relevance cannot be derived directly from verb semantics; it hinges upon the circumstances in which the proposition is pronounced: for instance, current relevance is contained in resulting

58

4. Explammg variation between the present perfect and preterite

implicates like "She has some money" or "The cheques are deposited" (cf. Huddleston and Pullum 2002: 145). Finally, current relevance might also be implied in so-called 'ml results'. Hence, in I've tried to phone her but she is not answering current relevance manifests itself as a failure to obtain the expected or intended result (cf. Huddleston and Pullum 2002: 145). Furthermore, the fact that resultative meaning is a central manifestation of the perfect is corroborated by data on non-standard native varieties of English. For instance, Miller (2004) observes that resultative contexts are commonly associated with the present perfect morphology even in nonstandard varieties of English, whereas the simple past tense morphology is widely used for all other meanings of the category (cf. Miller 2004: 305). To sum up, the hotly debated semantic component of current relevance, unmistakably recognisable in resultative contexts, should be taken into the ensuing theoretical account of the opposition 'present perfect : preterite'. More specifically, the perfect can be expected to be preferred over the preterite in resultative contexts in the corpus data.

4.2.1.2. The extended-nowperfect The type of meaning conveyed by the extended-now perfect has been labelled in literature in many different ways: 'inclusive past-and-present', 'universal perfect', 'continuative perfect' or the 'perfect of persistent situation' (cf. Filppula 1999: 123). This perfect describes a situation that started in the past and continues or persists into the moment of utterance. As a consequence, current relevance is quite tangible in such contexts. (37) Standard English It has been raining since morning. It is therefore no surprise that the perfects of this type often combine with verbs of activity or duration in Standard English. It has furthermore been noted in the relevant literature that the continuative perfect is frequently associated with certain time adverbial specifiers such as since 1950, for a longtime,allhislife,uptillnow,ctc. The extended-now perfect subsumes another type known as habitual perfect, which indicates that an event started to occur at a certain point in the past and has occurred regularly up to the moment of utterance as in / have watched CNN on Saturdays for more than three years now. Both uses

4.2. The present perfect

59

are expected to exhibit a propensity to occur with the perfect rather than the preterite in the corpus data, the notion of current relevance being overtly manifest in both types of contexts (cf Siemund 2004: 414). When considered cross-linguistically, the perfect in extended-now contexts seems to be an idiosyncratic property of Standard English as a vast majority of other languages employs the present tense to convey a continuative meaning (Dahl 1985: 137; Dahl, Osten 2007: personal communication). 5 Here are only a few examples: German Ich lebe sett 2001 in Hamburg; French J'habite a Hambourg depuis 2001; Russian Ja zhivu v Gamburge s 2001 goda6 Furthermore, cross-varietal evidence is highly consonant with these typological findings. For instance, in Irish English the present is fairly frequently used in the contexts under analysis: And they are fighting ten years in the North for an all-Ireland republic (Filppula 1999: 122). Moreover, the present tense in extended-now contexts has been attested in all varieties of non-native English studied here. Yet another observation is that the present tense was used in the contexts under discussion at earlier stages ofEnglish. (38) Early Modern English (from Henry VI, I, 1.2.40, cited in Siemund 2004: 409) Let's leave this town; for they are hare-brained slaves, And hunger will enforce them to be more eager: Of old I know them; rather with their teeth The walls they 11 tear down than forsake the siege. '...Ihave known them foralong time...' In fact, the present tense was used in the continuative contexts throughout the Old English and Middle English periods. The perfect appears in these environments only in later Middle English (cf. Traugott 1972: 145). (39) Late Middle English (from Canterbury Tales, Knight A.929, cited in Traugott 1972: 146) We han ben waitynge al this fourtenycht.

5 Prof. Osten Dahl, Stockholm University. 6 Languages that use the perfect in continuative environments similar to Standard English include Amharic, Estonian, Finnish, Kammu, Kikuyu, Limouzi (present tense is also given as an alternative), Spanish, Swedish and Wolof (cf. Dahl 1985: 137).

60

4. Explaining variation between the present perfect and preterite

It continues to compete with the present tense throughout the Early Modern English period before ousting the latter from the standard dialect a few centuries later (cf Traugott 1972).

4.2.1.3. The experiential perfect The experiential perfect, also known as 'existential' perfect or 'indefinite anterior' perfect, suggests that a situation or an event has taken place once or several times during a period of time leading up to the present. Thus, a sentence / have been to America implies that the speaker has visited the country at least once up to the moment of utterance. Experiential perfects frequently appear with the stative verbs such as be or have but also with the verbs indicating duration or activity: (40) Standard English (from Siemund 2004: 414) / have played tennis, but not very often. Klein (1994) notices that the distance between the topic time and the time of the situation is much longer in case of the experiential perfect in contrast to resultative perfects, where the temporal distance between the two is very short indeed (cf. Klein 1994: 112). The notion of current relevance can be described in terms of iterativity or 'repeatability', i.e. the situation takes place in a time span which is not over yet (such as one's life time) and it can thus be repeated (cf. Dahl and Hedin 2000: 388). It is this unspecified connection to the present moment which basically accounts for this use type in Standard English. We could thus claim that current relevance is not as explicitly expressed in experiential contexts as in resultative contexts, simultaneously hypothesising that experiential contexts are less favoured by the perfect when compared to resultative contexts or extended-now contexts. The latter assumption is in line with an empirical observation provided in Miller (2004) that "in spontaneous speech and in non-standard English the experiential interpretation is regularly (and frequently) assigned to clauses with Simple Past" (Miller 2004:315).

4.2. The present perfect

61

4.2.1.4. The perfect of recent past The English perfect is also used for situations where the present relevance of a past situation is simply the one of temporal closeness, that is, the past situation is very recent (cf Comne 1976: 60). (41) Standard English (from Huddlston and Pullim 2002: 145) a. It has been a bad start to the year, with two fatal road accidents overnight. b I ve discovered how to YHQYICI the fuse c. She has recently/just been to Pans. ' It is for this reason that the perfect of recent past collocates with the adverbs recently, lately, this year, etc. Moreover, Klein (1994) contends that the perfect of recent past seems to be compatible with the past tense, i.e. it expresses the relation between the time of utterance and the time of the situation, not the relationship between the time of the situation and topic time, characteristic of all other uses of the HAVE-perfect. Therefore, this usage looks like the beginning of a new development in the course of which the perfect develops into the past tense (cf. Klein 1994: 113). Due to its very general semantic nature, the perfect of this type can surface with lexical verbs of virtually all groups in Vendler's typology: state, activity, accomplishment and achievement. The use of the perfect in this function seems to be a unique property of Standard English as well since all other European languages use the preterite or 'historical perfect' in this case (cf. Dahl and Hedin 2000)7 In addition, studies on non-standard varieties of English report that the simple past tense is used to express recent past time. Thus, in his account of Scottish English, Miller (2004: 313) mentions that recent past time can be signalled by the use of the simple or progressive past with there in this dialect, as in (42) Scottish English (from Miller 2004: 314) / was speaking to Susan there. It follows from this discussion that the recent past meaning of the perfect is not its prototypical use, which should not come as a surprise. What is, however, important for the present study is the assumption that the recent past 7 For the definition of a 'historical perfect' see, for example, McCoard (1978: 235).

62

4. Explammgvanation between the present perfect and preterite

meaning does not play a crucial part in triggering the HAVE-perfect in our data since cross-linguistically "recency" does not seem to be a decisive semantic feature in determining the choice in favour of the perfect (cf Dahl and Hedin 2000). The overall conclusion is that contexts of recent past are less explicit with regard to the expression of the semantic component of current relevance. It seems that the only way to link the past event to the moment of utterance in contexts of recent past is to say that the event is very recent. Hence, it can be generally expected that contexts of recent past are less favoured by the perfect in our data.

4.3. Semantics of the perfect and preterite: Contrastive analysis The ensuing discussion is devoted to establishing a common theoretical ground for unifying the existing contrasts between the English present perfect and its major competitor in the domain of past tense marking: the preterite. The procedure, by no means simple, should contribute to the longstanding scientific debate about the relationship obtaining between the present perfect and preterite in English. Moreover, working out a consistent approach is no doubt helpful in explicating the patterns of use of these verb forms in the reference variety of English and its non-native varieties. We saw that the perfect is generally characterised by its four basic use types: the resultative perfect, the experiential perfect, the extended-now perfect and its subtype, habitual perfect, and the perfect of recent past. As mentioned above, what these all use types have in common is the notion of current relevance, i.e. "the linking of a situation which began or happened in the past to the time or moment of utterance" (Siemund 2004: 414). A similar idea is expressed in Huddleston and Pullum (2002), who claim that the major use types of the HAVE-perfect can be thought of as classification of different ways in which the past situation may have 'current relevance' (cf. Huddleston and Pullum 2002: 143). Since the semantic feature of current relevance, inherent in all perfect use types, makes the perfect quite distinct from the simple past tense, it is suggested that the semantic meaning of the perfect and preterite can be described in terms of the following compositional analysis:* 8 This analysis could be extended to include the simple present tense, which would appear to be the most unmarked member of the paradigm lacking the features

4.3. Semantics of the perfect and preterite: Contrasts analysts 63 Table 4.1. Semantic composition of the perfect and preterite reference to current thepast relevance preterite perfect

+ +

+

focus on the present

focus on the past

+

+

indefinite- definite+

+

The first observation is that both perfect and preterite are characterised by the 'reference to the past'. It is on this ground that both verb forms can be regarded as functionally equivalent, although semantical^ not identical.9 To be more precise, the proposed semantic analysis reveals that the perfect possesses one semantic component more when compared to the preterite: current relevance. It appears that current relevance is a feature that makes the perfect stand out as a verb form. It is this notion of current relevance coupled with the 'focus on the present' which allows the perfect to collocate with the time adverbials whose semantics includes both the time of the past situation and speech time (today, these two years, for the past twenty years, etc.) in Standard English (cf. Radden and Dirven 2007: 215). By contrast, the absence of this feature in the semantic make-up of the preterite coupled with the 'semantic focus on the past' nicely explains why the simple past tense is normally used with the definite past time adverbials (yesterday, two hours ago, etc.) in English: a situation that occurred at a specific moment in the past cannot be in any way connected to the present moment. What still needs to be elaborated on is the notion of current relevance. Traditionally, current relevance is believed to convey "resultativeness" (cf. Elsness 1997: 68), i.e. it shows that the results of some prior action are still tangible at the moment of utterance and are in this way relevant for the speech time. Dahl and Hedin (2000) give an essentially similar definition of current relevance, endorsing the idea of "the continuance of a result" as being indispensable to the interpretation of current relevance (Dahl and Hedin 2000: 392). Moreover, Siemund (2004) argues in much the same vein that current relevance is "the linking of a situation that began in the past to the moment of utterance" (Siemund 2004: 414), whereas Declerck, Reed, and 'reference to the past', 'current relevance', 'focus on the past' and 'definiteness'. 9 This observation is particularly important for the subsequent analyses of our data, in which the preterite and the perfect are viewed as verb forms sharing a similar function, i.e. reference to the past.

64

4. Explaining variation between the present perfect and preterite

Cappelle (2006: 213) contend that current relevance is a "speaker's concern with NOW". All preceding definitions of current relevance capture the major insight that current relevance is a semantic component that links a past situation to the moment of utterance. They can hence be adopted for this study. Moreover, this notion has been described as a "condition of repeatability" (Inoue 1979: 561). According to the latter analysis, current relevance describes a discourse topic which refers to a situation as "either repeated or repeatable at the time of speech act" (Inoue 1979: 574). Current relevance implies that a situation occurred or started to occur at some time in the past and is either being repeated at the moment of speaking or can be repeated later on. Thus, Inoue (1979) understands current relevance as the component of meaning that focuses the speaker's and hearer's attention on the present or even the future rather than the past. In this respect, current relevance can be said to be opposed to singular non-repeatable actions in the past, typically expressed by the preterite. Inoue's analysis of current relevance has nevertheless some limitations despite its undeniably useful insights into the semantic nature of the perfect. Thus, one of the consequences of the definition of current relevance provided in her analysis is the assumption that the present perfect cannot be used if the discourse topic refers to a non-repeatable activity. This claim is, however, untenable since there are instances of the usage of the perfect in non-repeatable contexts. Compare: (43) Standard English (from Elsness 1997: 35) John has completed his thesis. The discourse topic introduced by this sentence is non-repeatable (since a particular thesis can be completed only once) and yet the use of the perfect is totally acceptable in this case (cf Elsness 1997: 35). Being strongly associated with resultative contexts, the verbs of accomplishment and achievement introduce non-repeatable activities in the discourse, thus accounting for the inconsistency of the analysis of current relevance in terms of repeatability. Our conclusion is that Inoue's analysis of current relevance is best applied to experiential contexts such as Tom has seen this film twice, which typically imply that the event occurred once or several times before the moment of utterance during a time span which is not over yet and can for this reason be repeated in the future. Finally, current relevance has been defined as an implicature (Kortmann 1995: 194) which over time was grammaticalised and became a part of the

4.3. Semantics ofthe perfect and preterite: Contrasts

analysts

65

inherent meaning ofthe perfect, a verb form that gradually developed into a semantical^ more complex member of the opposition 'perfect: preterite': On the one hand, it has been suggested in literature on conversational implic a t e s that whenever there is a choice between two forms differing in structural complexity the speaker should choose the structurally simpler form. If the speaker selects the complex form instead, the interpreter is entitled to believe that the relevant utterance may conversationally implicate something "by virtue of its 'taking more effort' or 'taking the speaker further out of his way' than some alternative utterance." Kortmann (1995: 194) It follows from this observation that current relevance can also be understood as exerting a complicating effect on the semantic opposition 'perfect: preterite', in which the perfect is a semantically more "loaded" member and thus a more complex one. Such a definition of current relevance is completely in line with the analysis ofthe perfect and preterite elaborated on so far. Current relevance is, hence, a semantic component that implies the linking of a past situation to the moment of utterance, thus rendering the semantic composition ofthe English present perfect fairly complex. However, the degree to which current relevance is manifested in a sentence varies from use type to use type. Thus, it is obvious in resultative contexts, which have been claimed to be a central manifestation ofthe perfect (cf. Comne 1976; Siemund 2004), and needs little motivation in ext e n d e d ™ contexts, which typically describe a situation that started in the past as persisting into (and, hence, being valid at) the moment of utterance. In contrast, in contexts of recent past and in experiential contexts the notion of current relevance has been claimed to be less vivid. To give an example, the proposition The Prime Minister has resigned this year is not necessarily associated with any change of state at speech time. In other words, a recent (or a fairly remote) event has no immediate repercussions for the speakers at the moment of utterance. In contrast, an event immediately preceding speech time frequently involves a change of state in the present, which in its turn might have direct consequences for the participants ofthe situation (cf. Dahl and Hedin 2000). Similarly, the only way to justify the semantic component of current relevance in experiential contexts is to describe it in terms of repeatability, i.e. by saying that the past situation or event logically applies to a time span which is not over yet (e.g., / have been to America three times in my whole life) and can thus be repeated. It is this unspecified connection to the present moment which basically accounts for this use type in Standard English.

66

4. Explaining variation between the present perfect and preterite

Following this assumption, we can construct a hierarchy of uses of the present perfect as regards the expression of current relevance, with the resultative and extended-now perfects occupying the top positions in the ranking followed by the recent past and experiential use types. resultative extended-now recent past

experiential

Figure 4.2. The hierarchy of uses of the present perfect Taking this idea one step further, we can say that resultative and extendednow perfects are the most salient representatives of the category of the English perfect as opposed to experiential contexts and contexts of recent past because the notion of current relevance is manifested more vividly in the former contexts. Note that this claim is made with respect to the English language alone without any attempt at a cross-linguistic generalisation. Indeed, this hypothesis will hardly hold cross-linguistically since the vast majority of languages employ the present tense in extended-now contexts, as mentioned earlier. Moreover, it is resultative and experiential perfects that appear to be the most robust semantic categories cross-linguistically (cf Haspelmath et al. 2005: 271). Finally, the category of the present perfect is generally very unstable cross-linguistically, easily moving into cogmtively related domains: its shifting of meaning can be described as developing from the meaning of the result of a past action, further to anteriority relative to speech time and finally to the past proper (cf. Khlebmkova 1973: 127).

4.4. Limitations of the semantic analysis of the perfect and preterite The proposed semantic analysis of the perfect and preterite suggests that the co-occurrence patterns of these verb forms are context sensitive in Standard English. However, this analysis cannot account for all the instances of use of the perfect and preterite. Neither can it always explain their specific distribution mechanism across varieties of English. Evidently,

4.4. Limitations of the semantic analysts ofthe perfect and preterite

67

this approach must be supplemented with an investigation of other factors that might influence the choice between the perfect and preterite in StEngEng and its varieties. As mentioned above, the analysis of tense meaning - or the meaning of the perfect and preterite for that matter - cannot be restricted to the analysis of its semantics. McCoard (1978: 4), for instance, observes that it is a fruitful strategy to study "how available form-choices are used, rather than to restrict ourselves to describing the explicit, invariant semantic content of lexical and semantic constituents." Hence, additional linguistic phenomena such as components constituting a sentence and pragmatic discourse will be considered with the purpose of enlarging the scope ofthe analysis.

4.4.1. Analysis of the collocation of the perfect and preterite with other elements in the sentence It has been recognised that factors other than semantic contexts featuring the perfect and preterite condition their variation (cf Winford 1993; Taghamonte 2000; Van Herk 2008). For instance, occurrence of these morphological variants may be triggered by some other sentence constituents. In particular, time adverbials are felt to affect the choice between the perfect and preterite (cf. McCoard 1978; Inoue 1979; Van Herk 2008). Current literature reports that the choice of a particular verb form (either the perfect or the preterite) often depends on the time adverbial with which it appears in a sentence. To that effect, several groups of time adverbials have been identified: adverbials that collocate with the preterite only, adverbials that require the perfect exclusively, and adverbials that can generally collocate with both verb forms, although in certain varieties strong preferences in favour of one ofthe verb forms are attested.10 Time adverbials of all these groups are believed to differ from each other with respect to the expression of a certain semantic feature which, as will be claimed, can be defined in terms ofthe notion of current relevance. In his useful account, McCoard describes this feature as ± THEN and proposes the following classification of time adverbials (McCoard 1978: 135): 10 For instance, some speakers of British English are likely to claim that adverbials like just, already, and never can go with the perfect only. In contrast, speakers of American English feel that the usage of the preterite with these adverbials is perfectly natural. It is for this reason that McCoard assigns them to the ±THEN group.

68

4. Explaining variation between the present perfect and preterite

Table 4.2. McCoard's classification of time adverbials + THEN

±THEN

-THEN

long ago five years ago

long since today in my life today recently just now often always

at present up till now so far as yet not yet during these five years since the war before now

yesterday the other day those days last night in 1900 at 3:00 after the war no longer

already before

This classification provides a good example of how the co-occurrence patterns of various adverbials with the perfect and preterite can be described in a systematic way. However, such an analysis presents one with a problem since the above presented classification suggests that all the -THEN adverbials should be compatible with the present tense. In terms of logic, the semantic feature -THEN should be equated to some semantic feature +NOW, and therefore to the moment of utterance. It is nevertheless not always the case: */ am trying to get in touch with you during these five hours is unacceptable in Standard English. McCoard admits that the -THEN feature cannot equated with the present tense. However, he does not provide any specific definition of the -THEN feature nor does he say what temporal frame this feature should be assigned to. McCoard's classification clearly presents us with a difficulty of defining a specific semantic feature allowing for a consistent arrangement of time adverbials. It is therefore suggested that the time adverbials under analysis can be organized into distinct cohorts according to their ability to express the notion of current relevance. The difference between an event which occurred at a particular time in the past and an event which occurred during the time span stretching up to the moment of utterance can be summarised as follows: an event which took place at a particular moment in the past (three weeks ago, yesterday) cannot be perceived as being currently relevant any longer. Thus, in / broke my arm yesterday the adverbial yesterday refers to a specific past moment during which a singular event (one's breaking his arm) occurred rather than to the results of that event in the present (one's broken arm).

4.4. Limitations of the semantic analysis of the perfect and preterite

69

By contrast, it seems that time adverbials of the -THEN group in McCoard's classification contribute to conveying the meaning of current relevance in a sentence since their inherent meaning is to link some past situation to the moment of utterance. The results of some prior action are perceivable in the present because time adverbials of the type these five years, so far, up to now, etc. make the connection between some prior event and its result in the present palpable, as in / have run seven miles up till now. Here is another example to substantiate this view: (44)

Standard English a. I learned a poem yesterday. b. I have learned five poems so far.

In the second sentence, it is the adverbial so far that links a situation that happened in the past (the act of learning the poems) to the time of utterance and, hence, makes the whole event relevant to the present. The results of the event are still tangible at the moment of speaking (one has learned five poems and can tell them by heart, for example). It is therefore suggested that time adverbials should be classified according to the parameter of '± current relevance': Table 4.3. Semantic classification of time adverbials (adapted from McCoard 1978)" -current relevance

± current relevance

+ current relevance

long ago five years ago once yesterday the other day those days last night in 1900 at 3:00 after the war no longer in the past

never ever always just (now) today in my life recently lately often before before now

at present (now) up till now so far as yet not yet already during these five years past herewith since (the war) for (three years) long since

11 Similar to adverbs yet and so far, the adverb already is semantically compatible with the resultative meaning (cf. Fenn 1987: 127) and therefore with the notion of current relevance. It is thus classified as an adverb of the '+current relevance' group. The adverbial long since is compatible with the extended-now meaning and is assigned to the '+ current relevance' group. Finally, time adverbials before and before now are amenable to the interpretation in terms of the experiential meaning and are classified as belonging to the '± current relevance' group.

70

4. Explaining variation between the present perfect and preterite

What is important for the present discussion is the observation that Standard English does not allow the present perfect to collocate with adverbials of definite past time (or of the ' - current relevance' group in our interpretation). One cannot say: (45) Standard English */ have arrived in Hamburg five years ago/ in 2001. Current relevance interpretation, typically associated with the perfect, seems to lose its validity with indicators of definite past time reference, at least in Standard English. Results of the previous research suggest that time adverbials of the ' - current relevance' group do not collocate with the perfect, whereas time adverbials of the '+ current relevance' group do not go with the preterite (McCoard 1978; Elsness 1997). There are, however, a number of time adverbials for which it has turned out difficult to decide to which of the above-mentioned groups they should be assigned since such adverbials generally collocate with both forms, although speakers of a certain variety may have strong intuitions favouring the choice of just one verb form. Such time indicators have been labelled as the '± current relevance' group here. At the moment, it is not quite clear how these adverbials contribute to the overall meaning of a sentence. What is, however, obvious is that the time adverbials of the '± current relevance' group are less explicit with respect to the expression of the component of current relevance when compared to the time adverbials of the '+ current relevance' group. This issue will be elaborated on in the subsequent analyses of the distribution of the perfect and preterite across varieties of English. Summing up, we have seen that the perfect and preterite are distinguishable as far as their semantic composition is concerned. Furthermore, we have hypothesised that the occurrence of both verb forms may depend on the elements constituting the sentence. In particular, time adverbials were assumed to play a significant role in determining the choice in favour of a specific verb form in a given sentence.

4.4.2. Pragmatic analysis of the perfect and preterite The preceding analysis has been trying to pinpoint the differences between the perfect and preterite relying on their semantic properties as well as their co-occurrence capacities with other elements in a sentence. However, it is

4.4. Limitations of the semantic analysts ofthe perfect and preterite

71

by no means sufficient to define the nature of interaction between the perfect and preterite in those terms only. Many researchers (Inoue 1979; McCoard 1978; Elsness 1997; Dahl and Hedin 2000) have acknowledged the importance of pragmatic discourse in reconstructing the meaning of both verb forms. In sentences with an overt time specification, the hearer relies on the time adverbial for the interpretation of an event as either being currently relevant or not. The fact is, however, that both perfect and preterite often appear without any time anchors in a sentence. (46) Standard English (from Dahl and Hedin 2000: 387, 390) The Prime Minister has been killed. Did John wink? It is at this point that the hearer has to make inferences about the propositions relying on a context wider than a sentence. The following sections demonstrate how one can interpret a proposition as either being referred to a particular moment in the past or as being currently relevant relying on pragmatic knowledge only.

4.4.2.1. The preterite and pragmatic discourse If you live in a house where the door is locked at bedtime, then the sentence Did you lock the door? pronounced after everyone has gone to bed would be an equivalent to Did you lock the door at bedtime? (cf Quirk et al. 1985: 184). It is obvious that the knowledge about the semantics ofthe preterite is not enough to reconstruct the meaning ofthe sentence. One needs pragmatic knowledge about the house, its inhabitants and their habits in order to grasp the meaning ofthe sentence. A conclusion at this point is that "a full account ofthe meaning ofthe preterite, and ofthe tense as such is not to be confined to semantics but has to bring elements from pragmatics, as the line of division between the two disciplines is often drawn" (Elsness 1997: 7). We may take this idea one step further, arguing that pragmatic factors play a crucial role as well in defining the meaning ofthe perfect.

72

4. Explaining variation between the present perfect and preterite

4.4.2.2. The perfect and pragmatic discourse To be able to understand how pragmatic discourse contributes to our understanding of the meaning of the perfect, let us consider the following example from Dahl and Hedin (2000: 391-392): (47) Standard English A bank has been robbed. Dahl and Hedin (2000) claim that this sentence sounds perfectly natural if uttered by the sheriff of Tomstone, Arizona, who is trying to make his subordinates leave the room where they are playing cards. Evidently, the sheriff is not simply informing his deputies about some event that is still valid with regard to the moment of utterance, but he is also urging them to do something about the situation, i.e. he tries to make them draw a conclusion that they have to get to the scene of the crime. Therefore, current relevance is understood here in terms of repercussions that the event has for the participants of the discourse: In contrast to the prototypical cases of.. .perfects, these repercussions are not directly derivable from the meaning of the verb. In many cases, one has to rely on specific knowledge about the situation or about some convention. Dahl and Hedin (2000: 392) Current relevance in this sense is thus regarded not only as "a condition on the world", i.e. the continuance of a result that is still tangible at the moment of utterance, but also as "a condition on the discourse" in that the speaker portrays the consequences of an event as somehow essential to the point of what she is saying (Dahl and Hedin 2000: 392). In the study, the notion of current relevance will be operationalised in terms of the inherent meaning pertaining to the verb form (quantitative, i.e. distributional and multivariate, analyses) as well as in terms of conditions (or constraints) which current relevance can impose on the pragmatic discourse and, hence, determine the choice of the perfect (qualitative analyses). The findings of preceding two sections can be summed up as follows: the interpretation of the perfect and preterite often depends on the speaker's view of the event (cf McCoard 1978: 47). The choice in favour of either the perfect or preterite often reflects the speaker's perspective on the identification of the situation as either being relevant to the moment of utterance or not.

4.5. Summary 73 In view of the evidence presented above, we can assume that variation between the perfect and preterite may be motivated by the constraints that speakers impose on the pragmatic discourse. We therefore postulate a connection between the perfect and pragmatic discourse.

4.5. Summary This chapter has demonstrated that the perfect is semantically more complex than the preterite as it can be used in four major readings, whereas the major use of the preterite is to express a reference to an event, or a string of events, that took place at some definite time in the past not connected to the moment of utterance. Furthermore, the perfect and preterite show considerable differences with respect to their distribution with time adverbial specification as well as pragmatic inferences that they invoke. Thus, the perfect and preterite collocate with different types of time adverbials, the latter being primarily used with adverbials of definite past time. The former, on the other hand, combines with adverbials whose inherent semantic meaning can be defined in terms of the '+ current relevance' feature. Furthermore, extra-linguistic discourse, too, exerts a persistent influence on the choice of the present perfect. The preceding analysis has revealed that the principal contrast between the perfect and its major competitor seems to be the one of current relevance, a feature whose range of application transcends semantic frontiers and finds its pragmatic manifestation in an English utterance as well. Furthermore, various form elements (mainly time adverbials) enhance our understanding of the notion of current relevance. Current relevance can thus be defined as a three-dimensional theoretical concept which plays a decisive role for the choice between the perfect and preterite in Standard English. Table 4.4. The concept of current relevance Semantics

Semantic property inherent in all use types of the perfect, rendering the perfect a semantically more complex member of the opposition 'perfect:preterite'.

Form elements

Linking of a past event to the moment of utterance by means of time adverbials of the '+ current relevance' group.

Pragmatics The condition on the discourse which reflects the speaker's perspective on the utterance as being relevant with regard to speech time.

74

4. Explainingvanation between the present perfect and preterite

Even though the boundaries between these linguistic domains are not always clear-cut and it is sometimes difficult to determine one single factor (semantic, formal or pragmatic) accounting for the notion of current relevance in a sentence, such a concept of current relevance seems nevertheless to be a rewarding approach since it comprises various aspects of language and aims at a comprehensive analysis of the phenomenon under investigation. Given the role current relevance plays for the choice between the perfect and preterite in Standard English, it seems reasonable to suggest that in other (standard conformant) varieties of the language, the patterns of use of these verb forms could be described in terms of this notion as well. This premise goes back to the universale idea of all languages exploiting common principles of language architecture and language acquisition.12 Having taken that idea one step further, we assume that a language and its varieties must share some common mechanism underlying the occurrence of functionally equivalent linguistic forms.

12 In dialectology, a similar idea became known under the label of panlectal identity hypothesis. According to this hypothesis, dialects of a single language "share a common grammatical 'core' and differ only in matters of low-level realisations" (cf Harris 1984b: 303).

5. The notion of complexity and the English present perfect

Up until now we have been mostly concerned with the semantic/pragmatic properties of the English present perfect, which make it stand out in the cohort of temporal markers. It is these semantic/pragmatic intricacies that have been argued to present learners of English with enormous difficulties (cf. Radden and Dirven 2007: 212). However, to be able to give a comprehensive account of this category, we need to consider it from different angles, thus enlarging the scope of our investigation. In his functional-typological introduction to syntax Givon (1984) observes that of all tense-aspects in human language, the so-called perfect is by far the most complex (Givon 1984: 269). This observation may be used as a starting point of the discussion of the English present perfect in the light of the notion of complexity. First and foremost, definitions of complexity differ according to the framework or the perspective adopted in the study (cf. McWhorter 2001: 133; Miestamo, Sinnemaki, and Karlsson 2008). It is, however, possible to conceive of complexity in two major ways: (i) by providing a list of criteria for classifying language categories into the classes of linguistically complex or linguistically simple phenomena and (h) by developing a metric according to which not only one specific language form but also clusters of forms in a particular language domain can be assessed and measured. In other words, complexity is a notion that can be applied in relation to (i) language categories and (n) language domains (or systems). In this study, the notion of complexity is operationalised both as a reference to a language category (Chapter 5), and as a reference to a language subsystem (Chapter 6). In doing so, we draw a distinction between an absolute and a relative approach to measuring complexity. Such a distinction is particularly important in view of the fact that complexity has frequently been criticized as "an ambiguous and malleable concept" (McWhorter 2001: 133). The purpose of the following two chapters is, hence, twofold. Chapter 5 gives an account of the English perfect as a linguistically complex category, presenting a list of criteria suitable for classifying a category into the class of linguistically complex phenomena. Chapter 6 presents metrics of complexity which provides a useful theoretical framework for studying language-internal variation across non-native varieties of a language. In doing

76

5. The notion of complexity and the English present perfect

so, we rely and elaborate on the existing models for measuring complexity frequently discussed in the current theoretical literature. But before that we need to be clear about in what terms complexity has been defined in the previous research. Let us, therefore, start this chapter by providing an historical overview of the notions of complexity.

5.1. Complexity in the history of linguistics The notion of complexity can be traced back to the works of the German thinkers of the first half of the 19th century. Linguistic complexity of many Indo-European languages was assumed to be related to highly sophisticated European culture. Language was therefore believed to be a direct manifestation of the character of a people or a nation. Particular emphasis was placed on inflectional morphology to the effect that typologies were established to explain the differences between isolating, agglutinative, inflexional and incorporating languages. The unanimous linguistic consensus of the time was that languages possessing elaborate inflectional systems were superior to other languages. In fact, some authors considered inflectional languages to be more "organic" (cf Schlegel 1808, cited in Kusters 2003) in contrast to agglutinative counterparts, whereas others (cf. von Humboldt 1836, cited in Kusters 2003) assumed that inflectional rather that agglutinative or incorporating word formation was better suited for the expression of complex thoughts and ideas. As is clear, these authors implicitly assumed that all languages of the world form an ascending continuum, on which they could be contrasted to one another in terms of increasing morphological complexity. The position advocated by Schlegel and von Humboldt is that Indo-European languages are clearly more complex in terms of morphological structure. However, structurally complex languages were not always considered to be the most optimal vehicles of communication. In fact, French Enlightenment thinkers argued that languages abounding in various morpho-syntactic embellishments were less suited for efficient communication what could impede social and cultural progress (cf. Kusters 2003: 4). The idea of language complexity was, therefore, frequently employed in discussions about the relationship between language and culture. With the advance of structuralism, language complexity was rarely discussed since its proponents (for example, Sapir, Boas and de Saussure) believed that each language should be defined in its own terms. Since comparison of languages was no longer the focus of linguistic inquiry, the notion of complexity was by and large avoided in linguistic debates for a long

5.1. Complexity in the history of linguistics

11

period of time. It is, however, noteworthy to mention one exception: Otto Jespersen (1922) was probably the first professionally trained linguist to postulate a connection between inflectional complexity and communicative efficiency (cf Kusters 2003). In his view, the language in which the greatest amount of meaning can be expressed with the simplest linguistic mechanisms should be ranked higher. Thus complexity was defined in terms of relationships that hold between form and function in a particular language. Indeed, many so-called communicatively efficient and therefore structurally simple languages in which one form has several meanings were seen as more complex from the perspective of language processing (but not from language production) because of the lack of transparency between form and meaning. On the other hand, we have languages in which the relationship between form and meaning is more straightforward. Such languages, however, frequently possess a considerable amount of morpho-syntactic inventory (that is forms), thus making their speakers put more effort into language production. In other words, languages in which the relationship between form and meaning is transparent but which contain more morpho-syntactic forms are more complex for the speaker, whereas structurally simple languages with one form mapping onto a number of different meanings are expected to be more complex (i.e. difficult) for the hearer but probably less so for the speaker.1 Complexity was thus also defined in terms of speaker/ hearer difficulty in the history of linguistics. The second half of the 20th century is probably best described in terms of the Chomskyan revolution and a gradual shift of focus towards cognitive mechanisms that individuals employ while reconstructing the relationship between form and meaning in language. Furthermore, the advent of cont r a s t s linguistics, applied linguistics and Creole and typological studies meant a revival of interest in the issues of language complexity. Complexity has proven to be a useful explicatory tool in various domains of linguistics. It is thus an established notion in dialectology (especially in works by Peter Trudgill, perhaps, one of the most ardent advocates of the theoretical concept), contact linguistics (cf. Kusters 2003, 2008), typological studies (cf. Bisang 2009; Dahl 2004, 2009; Miestamo 2009; Nichols 2009; Sinnemaki 2009) and - to some extent - in studies on varieties of English (cf. Szmrecsanyi and Kortmann 2009). Furthermore, complexity is a viable notion in cognitive linguistics. The principle developed by Rohdenburg (1996) predicts that "in the case of 1 AsimilarstanceistakeninMcWhorter(2001: 135).

78

5. The notion of complexity and the English present perfect

more or less explicit grammatical options the more explicit ones will tend to be favoured in cogmtively more complex environments" (Rohdenburg 1996: 151). Rohdenburg (1996) points out that there is a morpho-syntactic contrast between the more and less explicit variant: the more explicit form contains an additional bound or free morpheme (Rohdenburg 1996: 173). This approach to complexity was successfully implemented in studies dealing with language processing.

5.1.1. Absolute approach vs. relative approach to measuring complexity The most recent research on complexity has it that complexity can generally be approached from two different - albeit not at all incompatible - perspectives: the absolute (objective) approach, which views complexity as "an objective property of the system", defining it in terms of the number of the parts of the system, and the so-called relative (user-oriented, subjective) approach, which defines complexity in terms of cost and difficulty to language user (Miestamo 2008: 24). Both approaches are of great relevance to the present study and will be discussed in some detail here. An absolute approach to complexity is defended in McWhorter (2001, 2007, 2008), Dahl (2004, 2009), Miestamo (2008, 2009) and Nichols (2009). The basic idea behind this approach is that the more parts a system has, the more complex it is (cf Miestamo 2008: 24). By this definition, a language that has 34 phonemes has a more complex phoneme inventory than the one that has only 18 (cf. Miestamo 2008: 24). Relying on the complexity measure known as Kolmogorov complexity (see Miestamo 2008: 24), Dahl (2004: 21-24) argues that complexity of a linguistic phenomenon may be measured in terms of the length of the description of that phenomenon; the longer a description a phenomenon requires, the more complex it is. The past time reference system of Kokongo, in which four past tense distinctions are attested, requires a longer description than the past time reference system of Japanese, which has only one overt past tense marker (example cited in McWhorter 2001). McWhorter (2001) understands complexity as a "degree of overt signalling of various phonetic, morphological, syntactic and semantic distinctions beyond communicative necessity" (McWhorter 2001: 125). From this perspective, languages that have more alternatives to express a semantic distinction are considered to be more complex. Older languages are believed to be more complex because of the natural course of history, which inevitably leads to languages accumulating random (and often unnecessary or redun-

5.1. Complexity in the history of linguistics

79

dant) overt markers of semantic distinctions as they grow older. By contrast, new languages, i.e. Creoles, are conceived of as structurally more simple. This does not mean to say that Creoles exhibit no complexity whatsoever. In fact, some Creoles make distinctions that analytic older languages do not. For instance, defimteness is distinguished in Saramaccan Creole but not in Laku, an older analytic language of Burmu-Tibetian group (cf McWhorter 2001). The crucial point is, however, that Creoles do not exhibit as much overall complexity as other grammars do. In other words, they do not make more overall overt distinctions than older languages. Miestamo's (2008) approach to complexity combines those proposed in McWhorter (2001, 2007, 2008) and Dahl (2004). This researcher suggests two principles that can be used in an absolute approach to complexity: the principle of Fewer Distinctions and One-Meaning-One-Form principle. The former states that language A which requires a longer description of a functional domain because more distinctions are grammaticalised within this domain is more complex than language B that makes fewer distinctions and consequently requires a shorter description. The latter principle implies that a linguistic domain in which one form is preserved for one meaning is less complex than a domain in which the form-function correspondences are not one-to-one since the description of the latter domain requires additional specification concerning these form-function relationships (cf. Miestamo 2008: 34). A relative approach to measuring complexity is provided in Kusters (2003) and more recently in Kusters (2008), who following Trudgill (2001) argues that complexity should allow examining "how the forms in questions are processed by different kind of users like foreigners or children" (Kusters 2003: 6). Kuster thus defines complexity as an amount of effort a non-native speaker (an 'outsider' in his terminology) has to make in order to learn the language in question. From this perspective, linguistic phenomena that are more difficult for a non-native speaker to learn are consequently more complex. The principal advantage of the approaches to complexity discussed here is that they are not embedded in a particular linguistic theory. Their modus operandi is thus very similar in spirit to the approach advocated in this study, whose major aim is to contribute to the building of linguistic theory, while relying on empirical facts. More importantly, both approaches can and actually will - be applied to assessing complexity of a language category and a language domain in the ensuing discussion.

80

5.2.

5. The notion of complexity and the English present perfect

Complexity in language

Like any linguistic term, the notion of complexity needs to be defined in a precise manner while being supported by empirical evidence (cf Kusters 2003: 2). Before providing definitions of complexity in this study, let us take a closer look at the existing definitions of linguistic complexity and discuss the suitability of this term to different components of language: phonology, syntax and morphology. Since it has been argued that some language phenomena are more complex than others, we explain which language features can be regarded as more complex in which language domains and why.

5.2.1. Phonology In no other area of linguistics has complexity been discussed so widely and intensively as in phonology. More often than not, complex phonological systems are described as possessing cross-linguistically unusual, i.e. marked, sounds such as click sounds (Kusters 2003: 13). Such a classification is based on empirical facts in the Greenbergian sense: languages with typologically rare sounds (for instance, click sounds) always have typologically usual sounds since a language cannot consist of marked sounds only. The reverse relationship, however, does not apply. Phonological complexity has thus been defined in linguistics in terms of typological frequency. The question as to whether a phonologically complex sound requires maximum articulatory effort on the part of the speaker on the one hand and hearer's mastery of processing a larger set of contrasts on the other has so far been regarded as difficult to tackle and, therefore, very much neglected. However, the notion of phonological complexity should be considered in the light of the evidence from studies on language acquisition and language processing. For instance, the order of acquisition of phonemes by LI and L2 learners should provide clues to their complexity status (cf. Kusters 2003: 13).

5.2.2. Grammar The general stance is that a grammar is more complex if it has a greater number of grammaticahsed expressions to convey fine-grained semantic and pragmatic distinctions (cf. McWhorter 2001; Kusters 2003; Szmrecsanyi

5.2. Complexity in language

81

and Kortmann 2009). The following sections explain what this somewhat general statement entails for specific areas of grammar: syntax and inflectional morphology.

5.2.2.1. Syntax It has been argued that in syntax some word orders are more natural (or unmarked) and hence better suited for communicative purposes than others. Thus, the Subject-Verb-Object (SVO) order has been claimed to be more natural (cf Croft 1990). This assumption has been corroborated by the fact that most Creoles have an SVO word order. In addition, when languages come into contact and influence each other, the direction of change seems to be towards the SVO order (cf. Kusters 2003: 13). Furthermore, languages allowing for more word order options can be regarded as more complex for L2 processing and production since a learner has to be aware of a number of semantic-pragmatic distinctions conveyed by various word order strategies to be able to interpret (or produce) the meaning of the sentence correctly. Although syntax does not fall within the scope of this investigation it is rather worth mentioning as this area of grammar captures an important insight that certain linguistic rules (and linguistic forms for that matter) are more difficult to produce and process in a specific context such as L2 acquisition. Moreover, the number of rules a language employs in a particular area seems to be have repercussions for the development of the L2 grammar. By this logic, languages with fewer rules (for instance, word order options) are easier to master.

5.2.2.2. Inflectional Morphology Languages differ from one another with respect to inflectional morphology. Thus, there are languages with reduced inflectional systems, English being a case in point, whereas others tend to exhibit overwhelmingly extensive inflectional systems. This circumstance in its turn offers an excellent opportunity for measuring complexity across languages and its varieties: languages with more extensive inflectional systems should be considered to be more complex when compared to languages with reduced inflectional systems. Inflection is generally considered to be a complexifying factor because it (i) exerts a load upon processing (cf. McWhorter 2001: 137) by (n) introducing irregularity into language (cf. Kusters 2003: 18).

82

5.3.

5. The notion of complexity and the English present perfect

Definitions of complexity adopted in this study for evaluating the linguistic status of the English present perfect

To be able to account for the present perfect as a linguistically and cogmtively complex category, we define complexity with respect to various aspects of language: its surface structures, their semantics, their acquisition, their frequency as well as temporal relations these structures are used to express.

5.3.1. Complexity in morphology Some authors have argued that derivational rules regularise the lexicon, reducing the amount of information needed to learn a language and, therefore, minimising complexity, whereas inflectional morphology may have a complicating effect because inflections extend the number of words (cf Kusters 2003). Let us notice here that not all strategies employed in inflectional morphology necessarily increase complexity. Let us consider a hypothetical example in a language X which has a suffix -y as a past tense marker which is equally attached to all verb stems. The language X thus has an inflectional rule regularising the whole past tense paradigm. Now consider a language Z which has a regular suffix -x as a past tense marker. In addition, this language Z has a number of other strategies to express the simple past tense such as umlaut or ablaut. In this language, the past tense paradigm emerges as a result of two irreconcilable forces - regular suffix adding, which aims at introducing more transparency into the language, thus simplifying it, on the one hand, and various ablaut strategies, resulting in opacity of the past tense paradigm, therefore contributing to ultimate language complexity since irregular verb forms (i.e. verbs formed via ablaut) are presumably stored in the lexicon separately. The morale of this story is that inflectional morphology cannot be defined in absolute terms (as, for instance, complexity increasing). Instead, we should take a closer look at the inflectional rules, trying to distinguish regularising strategies from those introducing irregularities. Regular inflectional suffixes (for instance, suffix -ed in English) are therefore considered morphologically less complex word formation strategies when compared to irregular (ablauting) strategies of word formation. If word formation strategies employed in inflectional morphology were to be understood as a continuum of the type "simplex strategies vs. complex strategies", then zero morpheme strategy would occupy the leftmost place on that continuum, thus representing the

5.3. Definitions oficomplexity adopted in Ms study

83

least complex morphological process, followed by regular inflection strategies and finally, by irregular (or ablaut) strategies. This observation is important for the ensuing discussion of the present perfect as a linguistically complex category and the way it contrasts to other verb forms (for instance, the preterite). Let it suffice to mention here that both regular and irregular strategies are involved in the formation of both present perfect and preterite. To sum up the preceding discussion, the processes involved in inflectional morphology vary from each other with respect to the amount of complexity they can add to the structure of a particular language. Regularising strategies (such as suffix -ed) can thus be viewed as complexity reducing since they require that the learner masters only one rule (i.e. adding suffix -ed to the stem of the verb). By contrast inflectional irregularities (ablaut) can be argued to be complexity enhancing since they seem to present a non-native learner with enormous difficulties, demanding that each separate (ablauted) verb form should be stored separately in the speaker's mind.

5.3.2. Complexity as a function of strategies employed in structure formation If a language requires an explicit expression of a particular category, this can be achieved in two ways: We can either build an analytic structure or express the meaning through various synthetic means (such as inflections). It has been argued that analytic strategies are simpler because they are, for instance, easier to process (cf Kusters 2003). From this perspective, an analytic structure is less complex than a synthetic structure. However, what we also need to take into consideration while evaluating the complexity status of a particular linguistic category is the number of strategies involved in the formation of a language structure. The general rule of thumb is that the more strategies a category involves in its formation the more complex it is. By this definition, a category involving both analytic and synthetic strategies is more complex than a category built up through synthetic means only. Summing up this short discussion, with respect to strategies involved in structure formation, complexity is a function of strategies employed in a particular language to build a linguistic structure.

84

5. The notion of complexity and the English present perfect

5.3.3. Complexity of temporal relations As elaborated on in the preceding discussion, we can generally distinguish between simple tenses and complex tenses (cf. Radden and Dirven 2007: 204). Simple tenses express deictic times which relate an event to the moment of utterance. By contrast, complex tenses express complex times, which involve two kinds of temporal relations: a relation between speech time and a deictic time, and a relation between the deictic time as a reference time and the time of an anterior or posterior event. 5.3.4. Complexity in semantics Semantic complexity is best understood in the light of the principle of compositionality which claims that "the meaning of a grammatically complex form is a compositional function of the meanings of its grammatical constituents" (Cruse 2000: 68).2 The definition implies that meanings can be combined together to yield more complex meanings and that the meaning of a complex expression is determined by the meanings of its constituents, every (grammatical) constituent possessing a meaning which contributes to the meaning of the whole expression. The principle of compositionality thus predicts that the more semantic constituents a particular category or expression contains, the more complex it is. This semantic complexity must be, however, consistently reflected by the structural complexity of the expression, according to the principle.3 5.3.5. Learners' complexity Complexity of a linguistic form is a relation between this form and the one who evaluates it (cf. Kusters 2003). With respect to non-native speakers, complexity can be defined as the amount of effort a learner has to invest in order to master a particular category of the target language. In other words, learners' complexity is the amount of difficulty involved in the acquisition 2 See also Cruse (2000: 239-242) for a detailed argumentation in favour of the componential analysis. 3 The principle of compositionality does not give an account of the linguistic elements that are structurally simple but semantically complex (for instance, inflections in Latin). Since such elements do not fall within the scope of this study, the topic is not pursued here any further.

5.4. The English present perfect as a complex category 85 of the target category in question. Under this definition, structures whose acquisition is delayed are more complex, i.e. difficult to acquire, than structures whose acquisition is not postponed.

5.3.6. Complexity and frequency It has been noted in, for instance, Mayerthaler (1981), Croft (1990) and more recently in McWhorter (2001) that unusual, i.e. marked, linguistic phenomena (for instance, click sounds) are crosslinguistically less frequent than more natural categories (such as vowels), which tend to be pervasive. This general observation has one important implication: unusual (i.e. marked) categories tend to add to the overall complexity of a particular language domain. For instance, languages containing unusual sounds (such as ejectives, clicks, rounded back vowels, etc.) will also contain natural (i.e. unmarked) sounds. McWhorter writes in this connection: ...an inventory with a great many marked sounds (e.g., a click language) is more complex than one with all or almost all unmarked sounds (e.g., a Polynesian language) because the former type of inventory has marked members in addition to unmarked ones. The marked sounds IMPLY the concurrent existence of unmarked ones (there exist no phonemic inventories with only crosslinguistically marked phonemes). McWhorter (2001: 135-136) It is in this sense that unusual (or marked) categories can be considered to be complexity enhancing when contrasted with unmarked categories. Since it has been noticed that unusual categories are less frequent than natural categories, we can use frequency as a criterion for diagnosing complexity of a particular language category provided we agree to view marked features as complexity-enhancing features. To put it differently, frequency can be used for diagnosing both markedness and complexity of a linguistic feature: less frequent features will tend to be both marked (i.e. in some way linguistically unusual) and overall language complexity increasing (i.e. linguistically weird features will exist side by side with linguistically normal features, thus expanding the paradigm of a particular language domain).

5.4. The English present perfect as a complex category In order to evaluate the overall complexity of the English present perfect, we provide a list of criteria suitable for classifying language categories into

86

5. The notion of complexity and the English present perfect

the classes of complex and simple phenomena. The criteria proposed here are consonant with the definitions of complexity elaborated on in the foregoing sections. 1. Structural (morpho-syntactic) make-up of the linguistic form. First, structures represented by multiple morphemes and compound expressions are more complex than grammatical structures consisting of single morphemes and single words (Croft 1990: 173). Furthermore, structures involving more strategies for their formation are considered to be more complex. 2. Temporal relations. Tenses expressing two kinds of temporal relations are more complex than tenses expressing one kind of temporal relation (cf.Radden and Dirven 2007: 204). 3. Semantic composition. The more semantic components a particular language category possesses, the more complex it is. 4. Polysemy. Polysemous categories are more complex than monosemic categories as they possess more meanings. 5. Order of acquisition. First-language acquisition. Second-language acquisition. More complex categories are acquired after less complex categories have been mastered. 6. Frequency. More natural categories have been postulated to be more pervasive (therefore more frequent) in a language (cf Mayerthaler 1981; Croft 1990). A less frequent category is linguistically more unusual, adding to the overall language complexity. It is clear that the criteria listed here draw a strict line between the meaning and form of a linguistic structure. The first four criteria are compatible with the absolute approach to complexity, according to which a more complex system or entity consists of more parts or elements. The fifth criterion is in accordance with the relative approach to complexity, which states that complex language forms are more difficult from the perspective of language user than simple language forms. The last criterion does not seem to fall under either definition of complexity. However, it appears to be a suitable parameter for diagnosing the status of a language category and will be applied here. The final observation is that all the claims made in the ensuing sections are made with respect to the English present perfect only. In other words, it is not our aim to argue here that the cross-linguistic category of PERFECT (cf. Dahl 1985: 129) is a complex category.

5.4. The English present perfect as a complex category

87

5.4.1. Structural make-up of the English present perfect Table 5.1 provides a schematic overview of the structural composition of the present perfect and other related categories: the simple past tense and the preterite. Table 5.1. Present perfect as a structurally complex category Present perfect

Preterite

Present

HAVE + V(2) ablaut HAVE + V(ed)

V(2) ablaut V(ed)

V©

The first observation is that both regular and irregular strategies are involved in the formation of both present perfect and preterite. The present perfect thus cannot be argued to be more complex than, for instance, the preterite on the ground that it involves irregular inflectional strategies (such as ablaut) in its formation, which have previously been argued to be more complex (because more difficult) from the perspective of an L2 learner. Both verb forms are formed with the help of regular suffixes and irregular stem vowel alternations in English. The present perfect is nevertheless a compound category consisting of an auxiliary (or a periphrastic marker) and a past participle of the main (lexical) verb (cf Quirk et al. 1985: 151). Clearly, this form possesses much more weight' in terms of syllabic structure when compared to the simple present tense and preterite, i.e. forms that frequently surface in present perfect contexts in non-standard varieties of English. The present perfect is thus a complex structure provided that complex structures are compound expressions as opposed to single words (cf. Croft 1990: 173). This type of evidence is, however, not entirely sufficient to make a claim concerning the morpho-syntactically complex status of the present perfect. To be able to understand why the present perfect is structurally more complex when compared to, for instance, the simple present tense, we also need to consider the strategies involved in its construction and compare these strategies with those employed in the construction of other verb forms. To start with, the formation of the present tense in English does not involve any change of verb structure (with the exception of the 3d.pers.sg.): nothing is added to the verb stem; the stem vowel remains intact. This type of coding is iconic (see, for instance, Mayerthaler 1981: 14, 25) with the semantics of this verb structure since the present time reference seems to be more central to human organisation of the concept of time (as we all live in

88

5. The notion of complexity and the English present perfect

the present). The semantic basicness of the category of the present tense is vividly manifested on the structural level in English. By contrast, the category of the preterite can be said to be semantical^ more marked as it involves reference to the past and is thus cogmtively more complex. The increase of semantic (cognitive) complexity is mirrored in the structural composition of this verb form: in English, the simple past tense is synthetically marked. Two basic strategies are involved in this process. The first one can be described as a segment additive change represented in this case by the regular suffix -ed (e.g., ask-ed). The second strategy became known under the label of modulatory processes (cf Mayerthaler 1981, "modulatonsche Prozesse") and is manifested by various irregular ablaut-mechanisms (e.g., find-found, write-wrote, etc.). Being semantical^ more loaded than the preterite, the English present perfect boasts an even more complex surface structure. This interrelation between meaning and form is nicely captured by the principle of constructional icomcity (cf. Mayerthaler 1981: 25, konstruktioneller Ikonismus): "Was semantisch mehr ist, sollte auch konstruktionell mehr sein" [Semantic complexity is reflected in structural complexity]. Similar to the preterite, the present perfect employs synthetic strategies for its construction - the regular suffix -ed and ablaut (e.g., ring- past participle rung). Additionally, it employs an analytic marker have/has for its formation. The foregoing analysis thus makes it clear that the English present perfect is structurally more complex than the preterite not only because of the greater number of morphosyllabic strings it provides for online processing but also because of the greater number of linguistic strategies it employs for its formation. It is quite true to say that an analytic strategy is to a large extent simpler than a synthetic one. However, when combined with synthetic ones it seems to contribute to the structural weight of a linguistic form what the example of the English present perfect demonstrates quite clearly. Summing up, employing a greater number of strategies (analytic + synthetic) for its surface formation, the English present perfect constitutes a case of a structurally more complex category when compared to either the present or the simple past.

5.4.2. Expression of temporal relations by the English present perfect The ensuing analysis is consistent with that proposed in Radden and Dirven (2007: 202-208). The English present perfect is a complex tense in terms of temporal relations that it is used to express. Whereas the simple past and

5.4. The English present perfect as a complex category 89 present locate the situation before or at (or around) speech time respectively, the perfect expresses more complex temporal relations. Not only does the perfect locate a situation before speech time; it also expresses a relation between a reference time, which is the present tense in this case, and the time of an anterior event. In other words, the present perfect involves "a background looking stance from a viewpoint at a reference time" (Radden and Dirven 2007: 204). It follows that the simple past and present express a relation between two types of times: speech time (S) and deictic time, which is also known as event time (E) in the relevant literature. By contrast, the English present perfect expresses more complex relations that hold among speech time (S), event time (E) and reference time (R). These fundamental differences between these categories are represented in Figure 5.1 below. Speech time (S)

, speech time

Event time (E)

past time

preslnttime

Simple Tenses:

past tense

present tense

Speech time (S)

speech time

Reference time (R)

present time

Event time (E) Complex Tense:

anterior present perfect

Figure 5.1. Simple English tenses vs. complex English tenses (adapted from Radden andDirven2007:204,205)

5.4.3. Semantic composition and polysemy of the English present perfect In the preceding chapter, a detailed account of the semantics of the present perfect was presented. It was mentioned among other things that in terms of its semantic composition, the English present perfect has elaborate semantic structure, the component of current relevance representing a complicating effect in the semantics of this verb form in contrast to the simple past tense, which lacks this semantic component. Another observation is that the formal

90

5. The notion of complexity and the English present perfect

structure of the perfect consistently reflects its rich semantic nature: the past participle denotes its relation to the past, whereas the auxiliary HAVE signals reference to the moment of utterance (cf Elsness 1997: 67). We can thus conclude that the complex semantics of the English present perfect is predicted by the principle of compositionahty spelled out in, for instance Cruse (2000: 68), according to which a meaning of a morpho-syntactically complex form is a sum of the meanings of its constituents. In other words, the meanings of the auxiliary HAVE and the past participle can be said to combine together to give rise to a complex meaning expressed by the present perfect in English. Furthermore, as has been explained in the preceding discussion, the English present perfect is a polysemous category, exhibiting as many as four principal meanings (cf. Comne 1976). By contrast, the preterite distinguishes at best between two functions (see section 4.1.), whereas the major function of the present tense is to locate a situation at (or around) the present moment (Comne 1976: 36; Quirk etal. 1985: 175).

5.4.4. The order of acquisition of the English present perfect It has been pointed out in the literature that ontogenetic order of acquisition can be regarded as an index of cognitive complexity of a linguistic category (cf. Slobin 1977: 206). The English present perfect can thus be argued to be a complex category from the perspective of language acquisition, both firstand second-language acquisition. In this sense, complex would mean "difficult to acquire". To start with, some authors noticed that the full use of the present perfect develops at later stages of LI acquisition (cf. Van Herk 2008). In addition, problems that the English present perfect causes secondlanguage learners have been described in the literature on second-language acquisition (cf. Felix 1978; Housen 2002). The following section presents data on acquisition of the English present perfect. The major purpose is show that the acquisition of the present perfect is delayed in both first- and second-language acquisition. In addition, it demonstrates what patterns of the ontogenetic development of the form are consonant with synchronic variation of the present perfect across non-native varieties of English and along what dimensions the developmental patterns differ from the kind of variation attested in synchrony. The first study to be mentioned here is that of Nussbaum and Naremore (1975), who demonstrated that the present perfect is indeed one of the last features to be acquired by native children of English. The major finding of

5.4. The English present perfect as a complex category

91

the study is that the use of have is not stabilised by the age of six even though the form begins to appear around age four (cf Nussbaum and Naremore 1975: 219). As Nussbaum and Naremore point out, their result nicely dovetails with that of Carol Chomsky (1969), who showed that certain syntactic structures including the perfect tend to be acquired by some children later on in life, in some cases past the age of nine. Furthermore, the results of this study are very much in line with reports by Cromer (1968, 1971), Brown (1973), de Vilher and de Villier (1973), which indicated that children learning English acquire the perfect relatively late. Thus, Nussbaum and Naremore (1975) report that the perfect begins to surface in an overwhelming majority (99%) of all possible cases (i.e. cases in which an adult speaker of English would have used the perfect) only at the age of six, i.e. at the age when the present tense and the preterite have long been acquired. Richard Cromer (1969) provides similar evidence demonstrating that the present perfect appears only after the simple present and the preterite have been acquired. Interestingly enough, our synchronic data obtained from pools of informants with various linguistic backgrounds (i.e. Russian, German, Hindi) shows that only and only acrolectal second-language speakers, i.e. speakers that have been exposed to the medium of English for a considerable period of time (i.e. more than 15 years), demonstrate the use of the present perfect consistent with descriptions found in traditional grammars of Standard English. Second, Nussbaum and Naremore report a variety of substitutions for the present perfect at the stages where the form is not fully mastered by the informants, i.e. around the age of four and five. The following substitutions for the present perfect have been attested in the individual LI acquisition of the present perfect: He already was finished, He been in the house, She is already washed the dishes. In addition, Gatherole (1986) reports the use of the present tense in present perfect contexts, as in Now she gots a new baby (cf. Gathercole 1986: 551). Synchronic variation between the perfect and other forms in non-native varieties of English demonstrates striking similarities: lone past participle, BE-perfect, the present tense and preterite are all attested in the upper-mesolect varieties of English (IndEng, SingEng and EAfEng) as well as in the mesolectal grammars of Russian, German and Hindi speakers of English. Finally, basilectal Hindi speakers of English demonstrate a cohort of variants in present perfect contexts used in place of the present perfect. Third, Nussbaum and Naremore (1975) report that the only form most frequently substituted for the perfect is the preterite. This finding is again

92

5. The notion of complexity and the English present perfect

consistent with the analysis of present perfect contexts across non-native varieties of English. Data from both second-language (e.g., IndEng, SingEng, EAfEng) and foreign-speaker (e.g., RusEng, GerEng) varieties of English confirms that the preterite is the only form frequently distributed across all perfect contexts in all varieties of English under investigation, namely IndEng, SingEng, EAfEng as well as RusEng and GerEng. This finding suggests that the preterite seems to be one of the most readily available alternatives to the perfect in naturalistic discourse. This assumption can be corroborated by historical data: The perfect and preterite were mutually interchangeable for a long period of time in the history of the English language and it was only in Early Modern English period (cf Elsness 1997; Taghamonte 2000) that the perfect gradually started to oust the preterite from its hitherto traditional environments. While looking for a functional explanation of this phenomenon, one might add that the perfect and the preterite differ from each other with respect to one semantic feature only (current relevance). Thus, this apparent similarity of function (i.e. reference to the past) may account for the fluidity between the two forms in both LI and L2 discourse. The next two studies (Felix 1978 and Housen 2002) deal with the L2 development of morpho-syntactic structures. To start with, Linguistische Untersuchungen zum naturhchen Zweitsprachenenverb by Felix (1978) is a qualitative study investigating acquisition of linguistic forms by children between the age of five and seven. Felix demonstrates that 'Aux + V structure is the last one to be acquired by children learning a second language (either German or English). More specifically, Felix (1978) refers to the work by Huang (1971) demonstrating that the 'Aux + V pattern presents a five-year-old Chinese learner of English with extreme difficulties. In fact, the structure was not fully acquired by the end of the investigation period (19 months). Furthermore, Housen (2002) shows that the emergence of morphology and meaning of the English present perfect are deferred in classroom second-language acquisition. Housen (2002) draws an important distinction between the formal and functional development of tense and aspect morphology, showing that the English present perfect develops late on both levels in a non-native grammar. Table 5.2 demonstrates the order of emergence of the tense and aspect markers in an L2 English grammar.

5.4. The English present perfect as a complex category 93 Table 5.2. Order of emergence of morphological categories (from Housen 2002: 158, Table I) 4 Stage Category

Comment

Example

0

Invariant V

esp. base form V

«^/qy

1

Present Participle Ving

intitially without Aux.fe

Irregular Past of be

was

2

Irregular Past (other verbs)

had, got

3

Regular Past Ved Future be going + Vinf

4

Perfects+F Presents Future will + V

allomotphs: without Aux. be, to,-ing; gonna allomotphs: Aux. be and have; initially V=V0

played, worked is going married; are go dancing; am going to take, is gonna happen have see,is fall, is fallen, has fall, have fallen goes, comes, does will make, will see

Housen's analysis of the data from 46 Dutch-speaking and French-speaking L2 learners of English (ages 9,11,13,15,17) suggests that the present perfect morphology appears at the final stage of the L2 development of the English tense and aspect system. This finding is consonant with that of Dietrich, Klein, and Noyau (1995) and Giacalone-Ramat (1995) who observed naturalistic adult L2 learners from a variety of first-language backgrounds (cf Housen 2002: 157). Relying on previous empirical research, Housen (2002) also contends that the meaning of the present perfect is one of the last tense-aspect meanings to be acquired by a non-native learner of English (cf. Housen 2002: 162). Table 5.3 below shows the order in which TA meanings emerge and become grammatically marked in his data.

4 Housen (2002) uses traditional terminology here for the interlanguage verb forms, without implying that the forms in question are fully targetlike or that they are used with their standard meanings.

94

5. The notion of complexity and the English present perfect

Table 5.3. Order of functional marking of tense-aspect meanings and their respective markers (from Housen 2002: 162, Table 2) Meaning

Form

0

Pre-functional stage

random and complementary distribution

1

anteriority (past and perfect) 1. Past of Be (was) 2. Irreg. Past (other verbs) 3. PerfectTWife + K 4. Reg. Past Kerf

Stage

imperfectivity/progressivity 1. Aux. Be + Ving 2

3

futurity

l.Aux.Be + Going+Vinf 2. Aux. Be + Ving 3. Aux. Will+V 4. P r e s e n t ^

habituality

I. Aux. Be + Ving

present

1. P r e s e n t ^

simple past

l.Pastof2fe( w a s , w e «0 2. Irreg. Past 3. Reg. Past Kerf

present perfect past perfect

l.iWffos+K^/reg 2.^+^/F/rreg

Housen furthermore emphasises that the distinction between present perfect and simple past meanings is mastered only by the most advanced learners of English (cf Housen 2002: 163). The results presented in Housen (2002) are in line with those provided in Agmhotn, Khanna, and Mukherjee (1998), who show that the English present perfect together with the past perfect and the future perfect appears to be an area of difficulty to first year undergraduates at the University of Delhi. Summing up, the evidence presented so far points to the correctness of the general thrust of our argumentation that the English present perfect is a complex category from the perspective of language development resulting in acquisition difficulties and subsequently in acquisition delay. But how can we account for this relatively late acquisition of the present perfect? It has sometimes been argued in studies on first-language acquisition that the reason for the prolonged development of the perfect can be attributed to the cognitive complexity of the form under study. Thus, Cromer (1968,1971) and Kuczaj and Daly (1979) pointed out that the present perfect

5.4. The English present perfect as a complex category

95

requires an "ability to consider the relevance of one timed sequence to another" (Cromer 1976: 301, cited in Gathercole 1986: 538). In other words, children seem to fail to appreciate the relation between past events and present consequences (cf Gathercole 1986: 538). It is exactly this inability "to decentre in time" (Cromer 1976) that leads to acquisition difficulties. A similar explanation is provided in Slobin (1973: 186) who states that the acquisition of the present perfect in English is initiated by cognitive development which gives rise to semantic intentions for which new means of expression must be forged (Slobin 1973: 186). As is obvious, children seem to need a long time to come to terms with the semantic feature labelled as "current relevance" in this study in that they learn first of all to refer to simple past events and only after that do they learn to refer to past events obtaining relevance at the moment of utterance. We should, however, bear in mind that all these suggestions have only a limited explanatory power as they cannot explain why adolescent and adult learners of English experience a delay in the acquisition of the English present perfect. Their cognitive structures are fully developed; yet many L2 speakers have difficulty in using the HAVE-perfect in accordance with the native-speaker patterns5 Whereas cognitive complexity was suggested as one possible explanation for the delayed acquisition of the present perfect in a child's language, morphosyntactic and semantic complexity of this periphrastic structure appears to be responsible for its relatively late emergence in an L2 grammar (cf. Housen 2002: 164; Winford, Donald December 2008: personal communication).6 The study earned out by Gathercole (1986) adds a new dimension to the issue of the cognitive complexity of the present perfect. While examining the acquisition of the perfect in 3- to 5-year-old monolingual children of Scottish English, Gathercole (1986) establishes the ranking order of the major uses of the perfect in present perfect contexts. Her finding is that children employ the perfect in the contexts of recent past and experiential contexts more frequently than in resultative and extended-now contexts. As is clear, the order of acquisition of the present perfect by children is the reverse image of the hierarchy of present perfect uses postulated in the previous chapter, the hierarchy that reflects the system of uses of the verb form by adult speakers of English whose grammar has been fully developed. 5 See also Fletcher (1981), Weist (1986) and Gathercole (1986) for other arguments against a cognitive developmental factor in language acquisition. 6 Prof. Donald Winford, Ohio State University.

96

5. The notion of complexity and the English present perfect

To be more specific, the child appears to start using the perfect in contexts which are semantical^ closer to the simple past time reference (i.e. recent past time and indefinite past time). Following Fletcher (1981), Gathercole (1986) observes in this connection that "the child may use the present perfect without recognising that the present perfect entails present relevance, in that the use of the present perfect requires that the state be repeatable" (Gathercole 1986: 555). In other words, while using the perfect in such contexts a child may not recognise that the HAVE-perfect involves relevance to the time of utterance. Why do children use the perfect in contexts of recent past and experiential contexts more frequently than in resultative and extended-now contexts? It seems that children start learning a new form (i.e. the present perfect) in the contexts that resemble the context they are already familiar with - past time reference contexts. Indeed, indefinite and recent past time reference is the closest option of past time expression on the continuum "past time reference - past time reference with current relevance". Having thus acquired the perfect as an alternative to the preterite in the contexts of recent and indefinite past time (i.e. contexts of recent past and experiential contexts), a child might proceed to more complex environments (i.e. contexts where current relevance has been postulated to be more vividly manifested). She thus elaborates on the ever more subtle differences between the present perfect and preterite, while acquiring a better understanding of the major function of the present perfect: reference to the past with current relevance. It follows that children may start acquiring the perfect as an alternative to the preterite without really recognising the meaning of the former (that is, the one of current relevance) and may do so in contexts which are functionally fairly close to the simple past time reference. Only after they have matured for the concept of current relevance, do they start using the perfect massively in contexts closely associated with this notion: resultative contexts. In fact, that the acquisition of the present perfect meaning begins with resultative contexts has been convincingly argued for in Slobin (1994). Although Gathercole (1986) does not address the question of "when the child begins to appreciate that the present perfect, in contrast to the simple past tense, entails relevance to the time of utterance", the scenario outlined in the foregoing paragraphs may provide a sound hypothesis for future research. To sum up, studies discussed above demonstrate that the acquisition of the English present perfect is delayed in both first- and second-language acquisition. This is viewed as an additional piece of evidence that the English present perfect is a (cogmtively) complex phenomenon.

5.4. The English present perfect as a complex category 97 5.4.5. The English present perfect and frequency Since the perfect is crosslingmstically less frequent than the past tense (cf Dahl 1985 115-190),7 the original hypothesis was that the present perfect with the functions similar to those found in English will always exist side by side with another category expressing simple past time reference (for instance, the simple past tense). To put it slightly differently, the initial assumption was that languages whose entire past tense paradigm would consist of the present perfect with functions similar to those of the English perfect should in principle not be attested. Notice that this assumption provides a nice parallel to McWhorter's (2001) observation that there exist no phonemic inventories with only linguistically marked phonemes (cf. McWhorter 2001: 136). We could view a language containing the perfect of the type attested in English and a simple past tense marker as more complex than a language containing only one simple past tense. However, it turns out that it is possible to find languages which have grammaticalised resultative and/or experiential meanings without having developed a general past marker (cf. Dahl 1985). In fact, general pasts often arise from perfect-like constructions; so if a language could not have a perfect without having a past, it would be rather tricky for pasts to arise in the first place (Dahl, Osten 9 January 2009: personal communication). To sum up, it seems that the present perfect cannot be regarded as a complexityenhancing feature from a crosslinguistic perspective. Although our hypothesis could not be substantiated by typological observations, some empirical facts obtained from different varieties of English suggest that the English perfect can be viewed as a complexity increasing feature at least from a crossvanetal perspective. Relevant evidence shows that there exist no forms of English in which the verb form with the functions of the Standard English present perfect would constitute the entire past tense paradigm. The present perfect typically alternates with the preterite in all vaneties of English known to the author8 By contrast, it is possible to

7 In Dahl's (1985) analysis, the past tense is twice as frequent as the perfect (cf. Dahl 1985: 183). Furthermore, Haspelmath et al. (2005) observe that the past tense marking is found in 134 cases in a sample comprising 222 languages, whereas the perfect is found in only 108 cases (Haspelmath et al. 2005: 269, 271). 8 This includes English-based Creoles, which seem to be in accordance with this statement. The analysis of 44 Creoles provided in Holm (1989: 405-551) suggests

98

5. The notion of complexity and the English present perfect

find varieties of English in which the present perfect is not attested but the simple past tense is. The most prominent examples are basilectal varieties of English. It follows that the perfect is less frequent across varieties of English than the preterite: the appearance of the perfect suggests the presence of the other less marked member of the paradigm in a given variety. This observation implies that the present perfect can and should be viewed as a complexity-enhancing verb form, at least in English. Furthermore, vanatiomst studies have provided ample empirical evidence that when encountered, the English present perfect is generally less robust in the data than, for instance, the preterite. To start with, Elsness (1997), and more recently Elsness (2009) reports the fairly infrequent percentage of use of the present perfect across genres and discourse types in contemporary varieties of British and American English (cf Elsness 1997: 107). Text categories studied by Elsness include science, drama, business correspondence, magazines, newspapers, radio news, telephone dialogues and face-to-face conversations. The author shows that the preterite outnumbers the present perfect by a wide margin in most kinds of texts in both varieties (Elsness 2009: 229). Second, results of the study on past time reference system in Samana English reported in Tagliamonte (1997) provide an even more clear-cut picture: out of the total of 7878 past time reference tokens attested in Samana English, only 86 tokens were found to have the 'HAVE + past participle' surface realisation.9 This forms a striking contrast to the number of preterite forms (the total of 4997) manifested by the -eJ/suppletion surface morphology. Results of this study are in accordance with those reported in Tagliamonte (1996) for Early Black English: 1162 tokens were found to be the preterite and only 18 tokens assumed the form 'HAVE + past participle'. Yet another piece of evidence is provided in Lawrence (2000) who similarly demonstrates a scarce percentage (only up to 4%) of the present perfect forms in her sociolinguistic interviews of York English. Further-

that whereas an anterior tense (which is analysed as a past tense marker) is found in all English-based Creoles, a completive marker (which has been argued to semantically overlap with the English perfects, cf. Holm 1989: 161-162) has been attested to be absent in, for instance, Barbados creole in the Caribbean (Holm 1989: 407). 9 This included have/has/'s as well as a following verb form which could include unmarked weak verbs and strong verbs with preterite morphology in addition to Standard English past participle (cf. Tagliamonte 1996: 369).

5.4. The English present perfect as a complex category

99

more, Van Herk (2008) reports 1337 forms of the present perfect and 2131 surface realisations of the pretente for African American Vernacular. Finally, Davydova (2007) analysed the distribution of the present perfect and pretente in a corpus consisting of letters of Irish emigrants and their families back in Ireland, the analysis resulting in a finding confirming the infrequent status of the present perfect. Thus, in her corpus of an overall of 7466 tokens, only 1467 had present perfect morphology. Summing up, the cross-varietal evidence presented in this section suggests that the fairly infrequent present perfect can generally be viewed as complexity enhancing in English.

5.4.6. The English present perfect as a complex category: Some additional arguments Following are some additional arguments substantiating the assumption that the English present perfect is a complex category. To start with, the discussion in the preceding chapter made clear that certain forms are cross-linguistically more salient in some functions rendered by the English present perfect. To provide an example, the present perfect required in extendednow contexts in Standard English is, in fact, cross-linguistically very unusual since most languages feature the present tense in this context (Dahl, Osten 20 October 2007: personal communication). We could therefore argue that the use of the present perfect in this context requires more attention, mental effort or processing time on the part of the non-native learner of English whose native language employs the present tense in extendednow contexts. Third, natural categories are believed to be resistant to the effects of language contact (cf Mayerthaler 1981: 4), whereas unnatural categories are assumed not to survive in a language-contact situation. Remarkably, empirical evidence suggests that the English present perfect gets easily lost in a contact situation, thus reducing the complexity of the language emerging in a contact situation. In fact, the present perfect has been claimed to be either nearly non-existent in English-based Creoles or so alien that the constraints on its use are not fully controlled, even by near-acrolectal speakers (cf. Bickerton 1975: 122-130, cited in Van Herk 2008). Finally, studies in psycholinguists (cf. Slobin 1973: 200, Slobin 1977: 195) have demonstrated that from the point of view of the perceptual strategies it is continuous structures that are easier to process. Being discontinuous in nature, the English present perfect could be argued to be more

100

5. The notion of complexity and the English present perfect

complex in terms of language perception and processing in contrast to the simple present and past tense which are continuous.10 Having presented the major arguments that the English present perfect is a linguistically and cogmtively complex category, we are now in a position to ask the following question: what repercussions does the alleged complexity of the present perfect have for the occurrence of this verb form across non-native varieties of English? We will address this issue in Chapter 17. Let us, however, now turn to elaborating on the perspective from which language-internal variation attested in non-native Englishes could be explained in terms of varying complexity levels (cf Szmrecsanyi and Kortmann2009).

10 The term 'continuous structure' refers to a structure that resists intervention of any linguistic material between its parts, whereas the term 'discontinuous structure' signifies a linguistic item that allows for the insertion of language elements between its parts, as in / have already read this paper (cf. Slobin 1973, 1977).

6. Metrics of complexity

Not only can the notion of complexity be employed to describe the status of a particular language category; more importantly, it has been implemented in linguistics in order to provide generalisations across languages and its varieties. To be able to account for the distribution of competing constructions surfacing in present perfect contexts across varieties of English in terms of differences in complexity levels, we need to consider the ways in which the notion has been operationalised and implemented in previous studies. There is to date no "conventionally agreed-upon" metric for measuring complexity in grammar (cf McWhorter 2001; Sampson, Gill, and Trudgill 2009). We saw earlier that that the two basic approaches to measuring complexity comprise an absolute approach and a relative approach. These general ways of viewing complexity have given rise to two principle methods of defining complexity levels in language. They are summed up for convenience here. 1. Ornamental rule or feature complexity traditionally refers to a number of "ornamentally complex" (cf. McWhorter 2001) features or accretions attested in a language domain. Representing an absolute approach to measuring complexity, this metric has been defined and operationalised in different research domains such as typology (e.g., McWhorter 2001, 2007, 2008, 2009) and vanatiomst studies (e.g., Szmrecsanyi and Kortmann 2009). McWhorter (2007, 2008) proposes three criteria for measuring ornamental complexity: (i) overspecification, which designates that a language A is more complex than a language B to the extent that it makes more (unnecessary) overt distinctions in its grammar, (n) structural elaboration which refers to a "number of rules mediating between underlying forms and surface forms", the complexity of a language increasing with the number of rules in its grammar (cf. Miestamo 2008: 30) and (in) irregularity, according to which the more irregularities a grammar contains the more complex it is. 2. L2 acquisition difficulty, also known as 'outsider complexity' (cf. Kusters 2003, 2008), is equated with difficulties of learning for adults, thus representing the relative approach to assessing complexity (cf. Trudgill 2001: 371). From this perspective, some language features are L2 learner friendly, whereas others are not. We can thus distinguish between the L2

102

6.

Melricsofcomplexity

easy and L2 difficult features that tend to recur in interlanguage varieties. Complexity of linguistic features or feature clusters is understood in relation to the one who evaluates them: the language user. This metric of complexity has been successfully implemented in Kusters (2003, 2008), and Szmrecsanyi and Kortmann (2009). When combined, both methods of studying complexity capture a very useful insight that the metric of complexity should be defined both in relation to language form and in relation to language use. Such a strict dichotomy is particularly important for studying non-native variation because linguistic variants surfacing in a particular variable context in a learner's grammar may not necessarily represent the respective L2 category (cf Winford, Donald December 2008: personal communication). For instance, a German learner of English using the present tense in a sentence He studies here for five years signals that the category of the present perfect has not been completely acquired because the speaker does not use the form of the present tense to express the category of the English present perfect. She most probably uses the present tense to express the meaning of the present tense (i.e. present tense with extended-now meaning) attested in her native language: German. Therefore, a strict distinction between a linguistic form and a meaning that non-native speakers attach to this form in their language use is crucial for understanding variation in a non-native grammar. Being in line with the major concepts of vanatiomst linguistics and with SLA accounts of tense and aspect development, a method has been developed in this study that comprises the level of the variable context in which language forms co-occur and the level of language use. The ensuing sections elaborate on this method of measuring complexity, while drawing on the existing tools known as ornamental complexity and outsider complexity.

6.1. Complexity as a degree of language-internal variation The definition of ornamental complexity implies that some languages have more formal means to refer to the same linguistic meaning (e.g., plurality, gender, etc.) than others. For instance, Kikongo has developed four formal markers to express the idea of pastness, whereas Japanese has only one overt marker to express the meaning of the past tense (cf. McWhorter 2001: 127). Taking this idea one step further, one could consider complexity to be a function of the number of alternating linguistic variants (or surface forms) used to express a linguistic meaning in a given variety. A variety in which

6.1. Complexity as a degree ojlanguage-internal variation

103

two or three formal means are used to render a particular meaning is more complex than a variety exploiting only one form for exactly the same purpose. To give an example, Irish English, in which progressive meaning can be rendered through the means of a special prepositional construction (e.g., He is at the milking of the cow) in addition to the more usual 'BE + V/«g' construction, is more complex than Standard English, where only the latter is attested. Despite its indisputable merits, this metric of complexity cannot be applied to studying non-native variation because, as mentioned above, not all forms within a circumscribed context are used by a non-native speaker to convey the meaning of the category in the target language. It is for this reason that we need to modify the existing definition of ornamental complexity in order to be able to operate with this term in relation to surface forms, while projecting it onto studying a non-native grammar of a language. It is therefore proposed that complexity should be defined as a degree of language-internal variation attested within a circumscribed variable context. This type of complexity can also be referred to as complexity of the variable context. It is manifested through (i) the number of variants surfacing in a variable context and (11) the number of language-internal factors (or rules) constraining the occurrence of a linguistic variant in a variable context. Notice that this definition of complexity is consistent with the notion of absolute complexity (cf Miestamo 2006, 2008), which broadly defines complexity "in terms of the number of parts in a system" (Miestamo 2008: 24), thus viewing the language as an autonomous entity, not related to the experiences of a particular kind of language user (cf. Kusters 2008). In other words, the more parts a system has, the more complex it is. Furthermore, this definition is compatible with McWhorter's approach to measuring complexity of language systems, who views complexity not only in terms of the number of linguistic elements (accretions) a particular language domain has; he also considers complexity to be a function of the number of rules involved in the occurrence of surface forms in a given language domain. Let us first of all consider complexity as a varying number of linguistic items surfacing within a defined variable context in an interlanguage system. A variable context (VC) of the variety A (VarA) can be considered to be more complex than the same variable context of the variety B (VarB) if and only if it contains more surface variants that are frequent enough to be included into a description of this variety. Consider Figure 6.1.

104

6. Metrics of complexity

Variety A: variable context X

variant A - frequent variant variant B - frequent variant variant C-infrequent variant variant D-infrequent variant

°£

Variety B: variable context X

°A

variant A - frequent variant variant B - frequent variant

=>VC(VarA) = VC(VarB) Figure 6.1. Infrequent variants in a variable context

This example illustrates the case where a variable context X of a variety A contains two more elements C and D when contrasted with the same context in a variety B. These elements, however, account for less than 1% of the entire data and need not be included in the description of the variable context of the variety A because (i) they arguably lie outside the core variable context and do not participate in the variation of linguistic items that constitute it, (ii) they do not provide any robust generalisations about variation of linguistic items on a global, i e. cross-varietal, scale and, consequently, (Hi) they do not necessarily enhance our understanding of variation at large in any insightful way. This means that the description of the variable context X in the variety A is in principle as long as the description of the variable context X in the variety B: both descriptions will include elements A and B only. A conclusion is then that the variable context of the variety A is not more complex than the variable context of the variety B in terms of surface variants alternating within a variable context. Consider now a different case illustrated in Figure 6.2.

6.1. Complexity as a degree of language-internal variation Variety A: variable context X

variant A-frequent variant B - frequent variant C-frequent variant D-frequent

105

Variety B: variable context X

variant variant variant variant

variant A-frequent variant variant B-frequent variant

=)• VC (VarA) > VC (VarB), where > stands for 'more complex' Figure 6.2. Frequent variants in a variable context

Now the variable context of variety A can be considered to be more complex than that of variety B because elements C and D are essential parts of that variable context. The description of the variable context X in a variety A comprises four elements, whereas its description in a variety B contains only two elements. The former description is longer than the latter and is in this sense more complex. Let us now consider complexity as a number of (language-internal) factors constraining variation of a linguistic variant in a variable context. Generally there are two types of variation: (a) non-systematic, i.e. random or free, and (b) systematic variation. Non-systematic variation is always more complex than systematic variation because it is irregular, and irregularity always involves more complexity than regularity (cf. Mainzer 2008: 48; Miestamo 2009: 81).1 In a linguistic system exhibiting free variation, elements alternate with each other in unpredictable patterns not constrained by any independent factors. Hence, free, non-systematic variation is more complex than systematic variation. Systematic variation on the other hand is always constrained by a number of independent factors. Since constraints should be regarded as increasing complexity (cf. Nichols 2009: 112), a variable context X of a variety A can be considered to be more complex than the same variable context X of 1 In Kolmogorov complexity, the highest complexity is attributed to random order (see also Fenk-Oczlon and Fenk 2008: 44).

106

6. Metrics of complexity

a variety B if and only if the linguistic variants surfacing in the variable context are constrained by more independent factors. Consider Figure 6.3. Variety A

Variety B

Variable context X

Variable context X

Variant Y

Variant Y

-

T "

1

Factor 1

F^toTT] Factor 2 Factors

Figure 6.3. Factors constraining variation in a variable context Variant Y in the variable context X of the variety B is constrained by more independent factors which build up hierarchies, some factors exerting more influence on the occurrence of the linguistic variant than others. In our example, independent factors build up constraint ranking in the variety B, with Factor 1 producing the strongest effect on the occurrence of the linguistic variant followed by Factor 2 and, finally, by Factor 3, whose influence is the least. We are now in a position to spell out the hierarchy of linguistic subsystems (i.e. variable contexts) in terms of varying complexity levels, illustrated in Figure 6.4.

Free variation

>

Regularvariation constrained by 3 factors

Regularvariation constrained by 2 factors

>

Regularvariation constrained by 1 factor

>

where > stands for 'more complex than' Figure 6.4. Hierarchy of linguistic subsystems in terms of varying complexity levels

6.2. Complexity as L2 acquisition difficulty 107 Linguistic subsystems with free variation are more complex than linguistic subsystems exhibiting systematic variation because the description of the former requires the listing of potentially infinite possibilities of language patterns. Regular linguistic subsystems in which the occurrence of a linguistic variant is constrained by two, three and more independent factors (or rules) are more complex than regular linguistic subsystems in which the occurrence of the same linguistic variant is constrained by only one factor as the former require a longer description than the latter. Let us, for instance, consider a hypothetical variety in which the occurrence of the linguistic variant going to is constrained by the point of reference and compare it to the variety in which going to is constrained not only by the point of reference but also by the type of clause and the ammacy of subject. It is the latter variety that is more complex when compared to the former since in the latter, more rules operate on the linguistic variant, thus enhancing the complexity of the variable context in question. In other words, the description of rules operating on the linguistic item is longer in the latter variety than in the former. It is in this sense that the latter variety can be considered to be more complex than the former. Under this metric, complexity is thus operationalised as (i) a number of verb forms surfacing in present perfect contexts in a given variety and (11) a number of independent language-internal factors triggering the occurrence of the HAVE-perfect in a given variety. The greater the number, the more complex variation in a given variety is. Having provided a definition of complexity on the level of variable context, we will explain how complexity can be defined with regard to language use.

6.2. Complexity as L2 acquisition difficulty As has been noted in the relevant literature (cf Kusters 2003, 2008), complexity can be defined in relation to the one who evaluates it (i.e. language user). Thus, L2 acquisition research has shown that some language features are more difficult for an L2 learner to master than others. It is for this reason that a distinction between L2 easy and L2 difficult features has been drawn in, for instance, Szmrecsanyi and Kortmann (2009). L2 difficult features are considered to be more complex than L2 easy features under this definition, which represents a relative approach to measuring complexity. Furthermore, not always can a feature be described as either L2 difficult or L2 easy; more often than not the target language's features form a continuum from very/rather easy to rather/very difficult. Let us now arrange the major

108

6. Metrics of complexity

English verb forms on the continuum of L2 easy/difficult features. We could come up with the following diagram.

I fj

• complexity as L2 acquisition difficulty

Figure 6.5. Continuum of L2 easy/difficult features (adapted from Housen 2002)2 The continuum mirrors the major stages of development of the English tense and aspect system in a non-native English grammar irrespective of LI, as elaborated on in, for instance, Housen (2002).3 Unmflected (or bare) verb stems used to express present, past and future time reference are the easiest forms from the learner perspective, followed by the present participles (e.g., going, coming, etc.) initially used to express both progressivity and present time reference, and the simple past tense used to express first anteriority and then simple past time reference. According to this metric, the English present perfect is a very difficult (and - therefore - complex) category requiring a lot of time and effort on the part of the L2 learner. As elaborated on in the previous chapter, the reason for that is not only the morphologically complex make-up of the HAVE-perfect but also - more importantly - the elaborate system of its uses and its rich semantic meaning. Under this metric, complexity is operationahsed as the extent to which a given variety of English attests one of the L2 most difficult features in the area of tense marking - the present perfect. The greater the percentage of the correct uses of the HAVE-perfect is, the more complex a given variety is. It can therefore be expected that only the most advanced learners of English will use it consistently, i.e. in accordance with native-speaker patterns. This hypothesis is in keeping with an observation provided in Kusters (2003), 2 Note that the present participle does not include the progressive form, which according to Housen (2002) is acquired after the regular past. 3 Note that this continuum does not account for some individual differences inherent in L2 acquisition.

6.3. Summary

109

who states that "the upper regions of complexity are populated by language users that have invested extra amounts of energy in learning and adapting their mind to highly complex structures" (Kusters 2003: 12). Less advanced speakers on the other hand can be expected to use L2 easy forms in place of the HAVE-perfect. The metrics of complexity as a degree of language-internal variation and complexity as an L2 acquisition difficulty have been formulated as logically independent and must be kept apart; their relationship is, however, worth examining. The former type of complexity does not have to entail the latter; but to which extent they correlate in language is an interesting question (cf Miestamo 2008: 29). After having examined complexity of the variable context, we might be able to explain our findings in terms of secondlanguage acquisition difficulty. Examining possible trade-offs between two types of complexity may provide evidence for (or against) the general cognitive mechanisms responsible for such trade-offs.

6.3. Summary Summing up, the discussion of the preceding two chapters has made clear that despite controversies surrounding this notion in linguistics, complexity can be conceived of in two major ways: (i) by providing a list of criteria suitable for classifying language categories into the classes of linguistically complex or linguistically simple phenomena and (n) by developing metrics according to which not only one specific language form but also clusters of forms in a particular language domain can be assessed and measured. Under the former definition, complexity presents an important explicatory tool for elucidating the origin of patterns of use of the perfect in a non-native grammar. Under the latter definition, complexity aims at providing generalisations across languages or varieties thereof, which in its turn may help to understand the nature of variation of the present perfect and other forms in present perfect contexts across non-native Englishes.

7. Empirical design of the study and methodology

The present study is earned out within the methodological framework of variations studies (as elaborated on in Tagliamonte 2006), thus aiming at determining language-internal as well as extra-linguistic factors underlying the occurrence of the present perfect. Moreover, the study was inspired by the ongoing research on second-language acquisition in the area of tense and aspect. Being not longitudinal in nature, it considers synchronic pools of data allowing a researcher to discover how the category of the present perfect develops in a non-native English grammar as we move from one speaker group to another speaker group.

7.1. Data: Some general comments The data for analyses consist of large-scale and small-scale corpora that are assessed both quantitatively and qualitatively (see the section on methods in this chapter). The major focus is on (non-)native vernaculars, i.e. "the style in which the minimum attention is given to the monitoring of speech" (Labov 1972c: 208, cited in Tagliamonte 2006). In other words, we study "every day speech", "real language in use", "spontaneous speech reserved for intimate or casual situations" or "informal speech" (cf Tagliamonte 2006: 8). Studying vernacular data has its own indisputable merits. First, vernacular is assumed to be a variety most free from hypercorrections and styleshifting, which are considered to be overlays of the original linguistic system (cf. Tagliamonte 2006: 8). Second, vernacular is the style from which all other styles must be calibrated (Labov 1984: 29, cited in Tagliamonte 2006). Third, the vernacular is positioned maximally distant from the idealized norm (Poplack 1993: 252, cited in Tagliamonte 2006), which means that "once the vernacular baseline has been established, the multidimensional nature of speech behaviour can be revealed" (Tagliamonte 2006: 8). In sum, observing the most "relaxed" type of L2 speech will enable us to tap into those forms of non-native English in which the specific L2 morpho-syntactic properties are presumably the most abundant.

7.1. Data: Some general comments 111 7.1.1. Large-scale corpora Large-scale non-native corpora compnse informal vernaculars as represented in the International Corpus of English (ICE) including IndEng, EAfEng (Kenya and Tanzania), SingEng (see Greenbaum 1996 for further details on the sample design of ICE). These corpora were supplemented with the data obtained from the London-Lund Corpus (LLC) of Spoken Enghsh (cf. Svartvik 1990), which represents the standard variety of native English spoken by educated speakers in England. In addition, we take recourse to the Hamburg Corpus oflrtsh English (HCIE), which is used for illustrative purposes in this chapter and for contrastive purposes in the discussion of SingEng. Analysed both quantitatively and qualitatively, the large-scale corpora were primarily used to elicit linguistic factors underlying the occurrence of the present perfect. In other words, extra-linguistic independent variables were not controlled for in either ICE or LLC because it was unreasonable to expect the compilers of the corpora to match speakers exactly in biographical features such as sex, educational level, occupation, etc. (cf. Greenbaum 1996: 5). However, the corpora yield global comparisons in terms of language-internal variation. Furthermore, checking for the respective substrate language was a methodological challenge in large-scale corpora due to the fact that conversations frequently took place among speakers with various mother tongues. Furthermore, the information on substrate languages was frequently not available in various corpora (see, for instance, Hudson-Ettle and Schmied 1999 for their discussion of EAfEng). We thus need to be aware of the limitations of the analyses of the possible origin of verb forms in present perfect contexts in these corpora. Because the sample design of the large-scale corpora did not allow a researcher to assess the role of the social factors or to study the exact effects of a particular native language, a number of supplementary (small-scale) corpora were built into the study in order to test the significance of some (extra)-linguistic variables on the use of the present perfect by non-native peak rs of English.1

1 In the ICE corpora speakers are generally not coded for either sex or their respective substratum language. In the LLC corpus, the information on the speaker's sex is available. However, female speakers are slightly overrepresented (207 male vs. 241 female speakers) in this corpus. The role of sex was therefore not investigated in the distribution of the present perfect in large-scale corpora data.

112

7.1.2.

7. Empirical design of the study and methodology

Small-scale corpora

Small-scale corpora form a part of the Hamburg Corpus of Non-Native Varieties of English (HCNVE), which is a collection of interviews earned out within the framework of a transfer project at the collaborative research centre (SFB 538) on multilingualism in Hamburg. The interviews were obtained through the 'fnend-of-a-fnend' technique (Taghamonte 2006: 21-33). An informed consent comprising (i) consent for the interview, (n) guaranteed anonymity and (m) access to the researcher was obtained for each informant of the study (cf Taghamonte 2006: 33). One of the biggest advantages offered by these databases is that they allow a researcher to investigate the role of sex and the length of exposure to the target language. Moreover, they help to assess the influence exerted by respective native languages on the occurrence of the present perfect. The following section elaborates on the sampling strategies used in collecting the corpora and the major (socio)-linguistic hypotheses underlying these sampling strategies.

7.1.2.1. Small-scale corpora: sample design andsociolinguistic hypotheses To start with, it has become a widely recognised fact in studies on LI vaneties of English that speech of female speakers differs from that of male speakers. Numerous sociolinguistic studies carried out in Western societies have come to the conclusion that women tend to employ prestigious language more often than men (cf. Romaine 2004; Cheshire 2007). To give an example, one of the findings of the study on Reading English was that female teenagers used standard variants more frequently and consistently than their male companions (cf. Cheshire 1978). Similar results were obtained in Trudgill (1974), and Nordberg and Sundgren (1999), to mention just a few. If extended to second-language acquisition, this hypothesis should claim that women tend to acquire standard forms in other languages much more quickly than the opposite sex because of their general orientation towards the linguistic norm. Indeed, some researchers acknowledge that females are "better learners" of foreign languages than males (Trudgill, Peter December 2007: personal communication).2 For instance, studies on foreign-language acquisition confirm that women generally do better in achieving a native2 Prof. (Emeritus) Peter Trudgill, Universite de Fribourg.

7.1. Data: Some general comments 113 like accent in the target language (cf. AH 1999: 14). An explanation to this sociolingmstic pattern was proposed by AH (1999), who claims that women have to try harder to be accepted by society which leads to a higher degree of proficiency in their target language (cf. Ah 1999: 15). Moreover, that gender (or sex) might be implicated in second-language acquisition is explicit in Leather and James (1996):3 The gender (and presumably also sociosexual disposition) of the learner may indirectly constitute a constraint on the variety of L2 speech learned. A survey among Dutch students of English (Broders, 1982) revealed that female learners were significantly more favourably disposed towards a "prestige accent" of L2 - British "Received Pronunciation" - than male: the stronger orientation of women towards prestige speech in LI (see, e.g., Hudson, 1980) would seem to carry over to the learning of other languages. Leather and James (1996: 272) Therefore, the principal sociolingmstic hypothesis underlying the study reads as follows: L2 female speakers tend to use the prestige variant, i.e. the present perfect, in present perfect contexts much more consistently than L2 male speakers as the present perfect is a verb category strongly associated with StEngEng. In other words, we should expect female speakers to produce more target-like uses of the perfect from the point of view of the prescriptive grammar. By contrast, male speakers are expected to use more different variants (preterites, bare verb stems, present tense, three verb clusters, etc.) in their speech because they are less oriented towards linguistic norms as learners, placing more emphasis on communicating their message regardless of the correctness of its form. The following sample design should help to explore this sociolingmstic issue.

3 In contrast to gender studies, the terms sex and gender are used synonymously here to refer to males and females considered as a group (cf. LDCE 2003: 669). No strict distinction is drawn between the terms because gender differences are based on sex differences (cf. Chambers 1995: 103) so that the former is frequently the corollary of the latter. In fact, sex and gender have not been distinguished consistently in most variationist studies because information about the individual's sex is easily accessible (cf. Eckert 1989: 246-247) and male-female differences evidently persist even in the absence of well-defined gender roles (cf. Chambers 1995: 104).

114

7. Empirical design of the study and methodology

Table 7.1. Sample design of the small-scale corpora Community (Native language)

Male (between 15 and 25 years of age)

Female (between 15 and 25 years of age)

Total

Russian (Russian) German (German) Indian (Hindi) Grand Total

6 6 6 18

6 6 6 18

12 12 12 36

The second issue addressed in the sample is a speaker's native language background. For instance, German native speakers are generally expected to use the perfect frequently in contexts where it should not occur (for instance, in definite past time reference contexts) as a result of substratal influence from German. Hindi speakers of English on the other hand are expected to use the perfect in their speech consistently as Hindi has a verb form which has traditionally been described as the present perfect tense (cf Kachru 2006: 151; Jain 1995: 185), whose functions roughly correspond to those of the standard English perfect. We could thus expect a greater amount of standard-conformant patterns from Hindi speakers than from German speakers of English. Finally, Russian speakers represent an interesting study case since the Russian language lacks the category of the present perfect. These speakers of English are therefore expected to exhibit the greatest difficulties with the HAVE-perfect when compared to other groups since neither the perfect morphology nor its meanings form a part of the linguistic knowledge of a Russian speaker. Summing up, the study starts with the assumption that the native language must exert a persistent influence on the occurrence of the category of the present perfect. This is expected to be especially true of those speakers that have not considerably advanced along the acrolect/basilect continuum (e.g., basilectal and mesolectal speakers). This hypothesis can be traced back to Slobin (1996: 89-91), who argued that tense and aspect categories are particularly susceptible to transfer during second-language acquisition. Finally, some informants were stratified according to the length of exposure to the medium of English. Thus, in IndEng corpus all informants were stratified according to the following sample.

7.1. Data: Some general comments 115 Table 7.2. Sample design of the small-scale corpora of IndEng Less than 5 years More than 10 years but less than 15 years More than 15 years

4 informants (two males and two females) 4 informants (two males and two females)

Total:

12 informants (six males and six females)

4 informants (two males and two females)

By contrast, informants on both RusEng and GerEng are represented by relatively homogeneous groups of speakers that have been exposed to English for more than 5 but less than 10 years. Generally, it was hypothesised that the longer non-native speakers are exposed to English, the less learning difficulties they are expected to experience with the category of the English present perfect and the more standard conformant patterns of use of the present perfect they are likely to exhibit. This assumption thus implies that the length and (the quality) of exposure to the medium of English is important for establishing a native-speaker command of the verb form under study because the English present perfect is a cogmtively complex phenomenon taking a considerable amount of time and learning effort to master. It follows that examining synchronic patterns reflecting different stages of L2 acquisition of this verb form should allow the analyst to reveal various insights about the development of the present perfect in a non-native grammar.

7.1.3. Large-scale and small-scale corpora: hypotheses concerning language-internal variation Quantitative analysis of data implies investigation of internal factors underlying variation in present perfect contexts. Previous research has illuminated a number of linguistic variables conditioning the occurrence of the present perfect morphology. These variables are semantic contexts (or uses) of the present perfect (cf Elsness 1997; Winford 1993; Tagliamonte 2000), time adverbials (cf. Elsness 1997; Fenn 1987; Tagliamonte 2000; Van Herk 2008), polarity (cf. Elsness 1997; Mazzon 2004; Van Herk 2008), Aktionsart (cf. Bybee, Perkins, and Paghuca 1994; Shirai and Andersen 1995; Housen 2002; Odlin and Alonso-Vazquez 2006; Van Herk 2008) and transitivity (cf. Davydova 2008). Our task is therefore to put all these variables to the test, elucidating the factors that play a significant role in the appearance of the morpho-syntactic form of the present perfect in non-native

116

7. Empirical design of the study and methodology

varieties of English. In what follows, we will elaborate on the set of hypotheses, thus highlighting the expected patterns of variation. To start with, resultative and extended-now contexts are generally expected to be more favoured by the perfect since the notion of current relevance is more vividly manifested in these contexts, as has been suggested in the theoretical part of the study. By contrast, experiential and recent past time contexts are expected to be either less favoured or disfavoured by the present perfect morphology in StEngEng and in non-native varieties of English oriented towards the British Standard. Formulated as a hierarchy of uses of the present perfect, this trend is expected to become particularly obvious with respect to the distributional patterning of the surface realisations of the present perfect and preterite, as these verb forms are (i) the two most robust variants across non-native Englishes studied here and (n) alternate with each other in all present perfect contexts, in contrast to, for instance, present tense, past perfect and medialobject perfect, which tend to be circumscribed to just one of these contexts. The preterite can thus be expected to favour experiential and recent past environments. It has been noticed in earlier studies on, for instance, American English (cf Vanneck 1958 and Duskova 1976, cited in Fenn 1987) that there is a certain amount of free variation between the perfect and preterite as regards their distribution with time adverbials ever, never, always, before, etc. Furthermore, Davydova (2008) showed that that the preterite rather than the present perfect surfaces with the time adverbials of the '± current relevance' group in Irish English. Therefore, adverbials of the type ever, never, always, etc. can be expected to pattern with the preterite rather than the perfect because (as could be argued) the component of current relevance is not vividly manifested in the semantics of the time adverbials of this type. By contrast, time adverbials of the '+ current relevance' group are expected to show a strong favouring effect for the perfect as they are very explicit with respect to the expression of current relevance. Furthermore, functioning as an indefinite past time reference marker (cf. Leech 1971a; Quirk et al. 1985; Radden and Dirven 2007), the English present perfect is assumed to be preferred in contexts without any time adverbial specification. By contrast, with the simple past tense greater importance is attached to temporal adjuncts in interpreting its meaning (cf. Huddleston and Pullum 2002: 137). We can therefore hypothesise that with the exception of several time adverbials strongly associated with the present perfect (those of the '+ current relevance' group), the preterite attracts temporal specification in contrast to the present perfect.

7.1. Data: Some general comments

111

The factor group 'polarity' (or 'negation') tests the claim that there is a close association between negation (both ^-negation and ^-negation, or constituent negation) and the English present perfect (Elsness 1997; Van Herk 2008). However, constituent negators such no, none, nothing may also be expected to disfavour the perfect since they were demonstrated to favour the preterite in some varieties such as Irish English (cf Davydova 2008). Furthermore, some studies demonstrated a favouring effect of the affirmative contexts on the occurrence of the preterite (Van Herk 2008). The next factor group to be discussed here is that of transitivity. As is well known, the origin of the perfect can be traced back to constructions of the type HABBAN/WESAN (BEON) + past participle. At the incipient stages of grammaticahsation of the perfect, the auxiliary HABBAN was used with transitive verbs, whereas WESAN/BEON occurred with intransitive and mutative verbs. As the category of the present perfect advanced along its grammaticalisation path in the standard variety of English, the auxiliary HAVE gradually replaced the auxiliary BE, which was constantly losing ground as a 'perfect' auxiliary. Thus, "throughout the ME period a steady increase is noticeable in the use of HAVE as an auxiliary for the perfect" (Mustanoja 1960: 501). However, it was not until 1900 that the change was complete (cf. Demson 1998: 136). As a result of this historical development, the present perfect is hypothesised to favour both transitive and intransitive (including mutative) verbs in StEngEng. Acrolectal/upper-mesolectal non-native varieties of English oriented towards the educated native-speaker Standard English are expected to reflect this tendency showing a preponderance of the HAVE-perfect in both transitive and intransitive (including mutative) environments. This in its turn could be interpreted as a sign that this verb form is a well-established (or even fully-fledged) grammatical category in these varieties. In less standard-conformant (i.e. mesolectal and basilectal) varieties, on the other hand, the HAVE-perfect is not expected to be particularly robust in either transitive or intransitive contexts as the perfect is hypothesized to be used in a reduced number of environments in these varieties of English. Finally, it was demonstrated in, for instance, Davydova (2008: 7) that there is a fairly close association between transitive verbs and the preterite. We can therefore assume that the preterite favours transitive contexts and put the hypothesis to the test in the basilectal variety of IndEng, where the preterite is the most frequent form in present perfect contexts. The final language-internal variable expected to be implicated in the occurrence of the present perfect is Aktionsart. To be able to understand what factors within this factor group might govern variability of the present

118

7. Empirical design of the study and methodology

perfect in our data, we need to take a closer look at the history of this verb form in Standard English. To start with, the English perfect can be traced back to the so-called "statal resultative" constructions (Hams 1984b) which denote "a state that was brought about by some action in the past" (Bybee, Perkins, and Pagliuca 1994: 63) and are originally associated with and applied to telic verbs, i.e. verbs which describe events with inherent end points (Bybee, Perkins, and Pagliuca 1994: 54). As the process of grammaticalisation proceeds and resultatives develop into the perfect proper, perfect constructions gradually expand their use from telic to dynamic verbs and finally to strive verbs (Bybee, Perkins, and Pagliuca 1994: 68). Semantically, the change from resultative to perfect means the generalisation of meaning from "current result" to "current relevance" (cf Lindstedt 2000: 368). The present perfect can be expected to favour the verbs of all four lexical types in StEngEng as well as the acrolectal and upper-mesolectal corpora as this would suggest a complete development of the HAVE-perfect meaning (i.e. the one of current relevance) in those varieties of English. By the same token, if the perfect extends its meaning to all lexical verbs, it can be said to be a fully-fledged (grammatical) category in a given variety. Second, in the basilectal and mesolectal varieties, both the perfect and preterite morphology is expected to favour verbs of accomplishment and achievement while at the same time shunning stative verbs and dynamic verbs, or verbs of activity. This assumption can be traced back to the Inherent Lexical Aspect Hypothesis (cf. Andersen 1991), which claims that the morphological development of tense and aspect system in a learner's grammar starts with learners assigning the past and perfective morphology to punctual-tehc verbs, gradually extending these uses to other lexical predicates (such as stative and dynamic) (compare the section on Aktionsart in Chapter 3). Indeed, research on second-language acquisition has provided evidence demonstrating that non-native speakers of English tend to associate past forms with telic and punctual verbs at the initial stages of language acquisition (cf. Shirai and Andersen 1995; Housen 2002; Odlin and Alonso-Vazquez 2006).

7.2. Collecting data The data for the small-scale corpora were collected by means of a sociolmguistic interview, one of the most widely employed methods in sociolinguistics (cf. Tagliamonte 2006). This method of data collection was particularly

7.3. Identifying the variable context and coding the data

119

well suited for the present study because it represents a well-developed strategy aimed at eliciting the vernacular, i.e. a relaxed type of speech. In order to conduct a sociolinguistic interview, a questionnaire was developed (see Appendix 2). It consisted of a series of hierarchically structured sets of questions all of which were specifically designed to elicit the use of the present perfect in a non-native speaker's speech production. The interview was constructed in such a way that half of the questions featured the present perfect form (e.g., Have you ever been to an English speaking country?, What have you been doing recently? etc.). Yet in other questions the present perfect did not surface; rather a present perfect context was suggested. For instance, informants were asked to describe the results of environmental pollution or globalisation. Other questions not featuring the perfect were 'Is this week different from other weeks?' or 'Is the education system in India (Russia, etc.) different now from what it used to be?' etc. The major goal of this technique was to find out whether or not semantic contexts of the English present perfect presented non-native speakers of English with difficulties. It was assumed that only speakers with native-like competence in English would freely use the present perfect in both types of questions, whereas other (less advanced) speakers would have difficulties in identifying the uses of the present perfect in contexts not featuring this form. Finally, speakers with very limited access to the target language were hypothesised to have problems with both morphology and semantics of the present perfect.

7.3. Identifying the variable context and coding the data Since the present perfect is a morpho-syntactic form demonstrating astonishing semantic complexity in Standard English, it should not come out as a surprise that its variable context could be defined in semantic-pragmatic terms only (cf Winford 1993). This procedure, in turn, imposes an enormous methodological challenge on a researcher working on this linguistic variable (cf. Hams 1984b: 316). A particular problem, as it has been described in the literature, arises when one tries to establish an equivalence of meaning at the morphological level. Thus, Lavandera (1978), Romaine (1980) and Milroy (1987) argue that the notion of the sociolinguistic variable cannot be extended beyond the levels of phonology because of the problem of establishing semantic equivalence at the syntactic level of linguistic structure. This theoretical controversy notwithstanding, quite a few researchers (Wolfram 1969; Fasold

120

7. Empirical design of the study and methodology

1972; Trudgill 1974) convincingly demonstrated how the scope and relevant contexts of morpho-syntactic variables can be defined. Following the Labovian definition of the linguistic variable, i.e. a set of variants sharing the "sameness of cognitive meaning" (cf Winford 1993: 142), studies carried out by Cheshire (1982), Hams (1984b), Taghamonte and Poplack (1988), Winford (1993), Taghamonte (1996, 1997, 2000), and Van Herk (2008) refined the notion of the linguistic variable, thus providing further insights into the nature of morpho-syntactic variation. In particular, it has been observed, for instance, in Hams (1984b: 316) that in order to determine the semantic equivalence of different morphosyntactic forms, a wider perspective including semantic, discourse and pragmatic considerations is required. In the following, we describe the specific techniques and methods employed to identify the variable context and extract the data from the corpora, thus providing a comprehensive account of the criteria used to determine the linguistic variable under study. Yet another purpose of this procedure is to ensure that the present study can be replicated (cf Taghamonte, Sail 18 July 2007: personal communication)4 First and foremost, transcribed texts were read one by one in search of contexts that have been traditionally described as present perfect contexts (cf. Jespersen 1924; Zandvoort 1932; Bauer 1970; Leech 1971a; McCawley 1971; Comne 1976; Fenn 1987; Bnnton 1988; Tagliamonte 1996, 1997, 2000; Siemund 2004), automatic extraction of data being virtually impossible in this case. Essentially, the focus was on verb forms that surfaced in contexts in which Standard English allows the present perfect. But how are these contexts to be identified in the corpus? In order to answer that question, we once again need to consider very briefly the general meaning of the contexts in which the verb form under study occurs. To start with, what all present perfect contexts have in common is that they refer to some state at speech time which arises from some past situation (cf. Fenn 1987; Winford 1993). In other words, a present perfect context always implies that a past situation (or an event) is oriented towards the moment of utterance. In order to identify such a context in the discourse, we need to single out (i) events that are absolutely past in relation to speech time, (ii) events that occur at the moment of utterance (i.e. neither prior nor posterior to speech time), and (in) past events that are in some way linked to the moment of utterance (cf. Winford 1993: 155). Our concern here is with events of the third type since the Standard English present perfect is associated with the meaning described in (m). The meaning of the type (i) 4 Prof. Sail Tagliamonte, University of Toronto.

7.3. Identifying the variable context and coding the data

121

is typically expressed by the preterite in this variety (cf. Elsness 1997), whereas the meaning of the type (n) is rendered by the present tense. It is therefore crucial for our analysis to identify those temporal instantiations in the discourse that occurred (or began to occur) in the past but are intrinsically oriented towards the moment of utterance. This can be best achieved by working out strategies allowing the analyst to differentiate between definite and indefinite past time reference. Whereas indefinite past time events (i.e. events not modified by adjuncts specifying a definite setting, cf. Radden and Dirven 2007: 213) are to a considerable extent linked to speech time, definite past time events always draw a distinct line between a moment in the past and the moment of utterance. First, in order to distinguish between past events that are oriented towards speech time and those that have definite past time reference, a method of elimination was employed (cf. Winford 1993). As the name suggests, this method aims at excluding all contexts that are clearly marked as definite past time reference contexts. In fact, it has been observed in the literature that time adverbials play a decisive role in determining the meaning of the entire proposition as either having relevance with respect to the moment of utterance or referring to a specific moment in the past (cf. Crystal 1966; Quirk et al. 1985; Winford 1993). As Crystal (1966) suggests, "time relations in English are handled more by the careful use of adverbials... than by any other means" (Crystal 1966: 7). It is therefore time adverbials that make temporal distinctions of the types discussed above clear. Following this logic, there are certain time adverbials in English that are inherently associated with definite past time reference contexts as opposed to present perfect contexts and require the preterite in Standard English. In his useful account, McCoard (1978: 135) singles out time adverbials that belong to former contexts. These time adverbials are long ago, five years ago, once, yesterday, the other day, those days, last night, in 1900, at three o'clock, after the war. As is clear, these time adverbials are incompatible with the moment of utterance. Hence, all utterances featuring these explicit definite past time indicators were excluded from the analysis. (48) Indian English (ICE: Sla-052) See no, the scene exactly happened you know the other day. In addition, sentences containing subordinate clauses specifying the exact time when the event took place were excluded from the analysis. To illustrate this point, sentences of the type / never ate sweets when I was a child were not extracted since the subordinate clause provides a clear definite

122

7. Empirical design of the study and methodology

past time reference. The following example provides an unambiguous definite past time reference (49) Indian English (ICE: Sla-031) Because that was in the tenth standard when my parents stop fed J me. Similarly, some adverbials of place (e.g., in Bangalore, in the restaurant, etc.) refer an event or a situation to a particular moment in the past and therefore require the preterite in Standard English. Hence, it was necessary to exclude from the analysis sentences featuring explicit reference to a place where the event occurred. (50) Indian English (ICE: Sla-049) A: Where you have fallen down? B: / fell down at my native place. In the foregoing example, the answer contains a reference to the moment in the past which is not oriented towards speech time. More specifically, it features the adverbial of place at my native place referring the event of falling down to a particular moment in the past. The answer can thus be interpreted as Ifell down when I was at my native place. Furthermore, the preterite is required in English to describe a string of events that occurred one after another in the past and are not connected to the moment of utterance. In such narrative contexts the speaker relates a series of real or Active events in the order they took place (cf Dahl 1984: 116). This temporal domain of the preterite has no overlap with present perfect contexts in Standard English. In fact, Dahl (1985) argues that "one salient property of PFCT [perfect] in general is the fact that it is not used in narrative contexts" (Dahl 1985: 113). As a result, all past narrative contexts were excluded from the analysis, as in (51) Indian English (ICE: Sla-004) Panduranga once visitedPundah, Pundalika's house, you know, and Pundalika was so engrossed and busy serving Ms parents, you know, and Panduranga. As he went there, you know, Panduranga wanted Pundalika to pay attention to htm but Pundalika, being a very obedient son and very again we can say devoted to his, uh parents you know, he had no time to pay attention to Vithal. So he threw a piece of brick you know and made him stand or made him wait there until he finishes his service to his parents.

7.3. Identifying the variable context and coding the data

123

Whereas adverbials indicating definite past time reference can be clearly differentiated as definite past time markers, adverbials now, at present, at the moment are on the other hand compatible with both the present tense and the present perfect. Sometimes these adverbials are used by the speaker to reinforce the orientation of the past bounded event towards the moment of utterance (cf Winford 1993: 156). In such cases, lexical characteristics of the verb play an important role in determining whether the meaning of these time adverbials is compatible with the meaning conveyed by the present perfect. A second strategy consisted in identifying contexts through time adverbials whose semantics are compatible with the meaning of the present perfect (cf. Fenn 1987: 209). As explained in the theoretical part of this book, a considerable number of adverbials are unambiguously associated with the meaning rendered by the present perfect in Standard English. These adverbials include since last year, for the last two days, for two weeks, up till now, so far, as yet, not yet. Some authors (for instance, Quirk et al. 1985; McCoard 1978) are very likely to add the adverb lately to the paradigm. Thus, propositions featuring these time adverbials presented no difficulty for token extraction. However, some adverbials are neutral with respect to the orientation towards the moment of utterance and are for this reason associated with both definite past time reference and present perfect contexts, adverbials before, often, always, ever, never representing a case in point. In order to identify the cases where propositions featuring the abovementioned time adverbials are linked to speech time, larger discoursepragmatic context needed to be taken into consideration (cf. Winford 1993: 157). Yet another (additional) technique used to identify the variable context and extract the data can be described as analysis of the sequences of syntactic structures preceding and following present perfect contexts. The basic empirical observation here is that present perfect contexts are generally either preceded or followed by the speech chunks featuring the present tense, asm (52) Indian English (ICE: Sla-022, Sla-029, Sla-033, Sla-039) a. / think yeah we have passed a very good way. b.It's something that we have not experienced before. c. Because all the while I was in the company of very nice people educated, again refined, those who have very nice ideas to vrovazate. d. And now the Kukis they refused to pay any more and so they are moving azainst them.

124

7. Empirical design of the study and methodology

The general idea behind this technique is that since the major function of the present perfect is to relate a past event to the present, a present perfect context is often surrounded by speech chunks featuring the present tense which helps the speaker to convey the idea that a past event is intrinsically linked to the moment of utterance. Having outlined the major strategies employed for identifying the vanable context in the corpora, we turn to a description of the criteria used to classify the data after respective tokens have been extracted.

7.3.1. Classifying tokens according to semantic context The extracted tokens were first of all coded for semantic context. To start with, each token was analysed step by step according to the event schema presented below. Past event (or state) persisting into the moment of utterance?

JO extended-now context

*, change of state at speech time due to some previous action?

resultative context

repeatability reading?

experiental context

unique event in recent past?

context of recent past

context discarded from the analysis

Figure 7.1. Event schema for the categorisation of present perfect contexts First, it was ascertained if the extracted token represented a past event or situation persisting into the moment of utterance. If this was the case, the token was coded as an extended-now context. If it was not the case, it was necessary to pose a question as to whether or not the event implied a change of state at speech time. If the question was answered positively, the sentence was coded as a resultative context. If not, the analyst proceeded

7.3. Identifying the variable context and coding the data

125

with the question of whether an event could be described in terms of repeatability. If the context suggested that an event has occurred several times during the time span leading up to the present and could be repeated, the token was coded as an experiential context. Finally, if the event did not obey the repeatability constraint but rather referred to a unique event that occurred in recent past, the token was classified as the context of recent past. Having described the most general procedure of distinguishing between various types of semantic contexts, we will now turn to describing more specific techniques involved in the process, providing concrete examples from the data. 7.3.1.1. Identifying extended-now contexts As noted in the theoretical part of the study, extended-now contexts frequently collocate with adverbials such as for X time or since X time since the extended-now perfect always refers to a situation that stared in the past and continues or persists into the moment of utterance. As a result, these contexts are relatively easy to identify. (53) Indian English (ICE: Sla-016; Sla-009; Sla-004) a. / have been staying for thirteen years only in hostel. b. What have you been doim since then? c. Yes, in my house she stayed with us for five months [and she is still staying with us]. As can be seen, the predicates involved in extended-now contexts are fairly frequently dynamic in nature. Such predicates help to convey the sense of a past action persisting into speech time. Moreover, extended-now contexts are more often than not associated with the "progressive form" have been +ing (cf. Bauer 1970: 193). This form was, therefore, a good indicator of an extended-now context. However, not all cases were clear cut. First, time adverbials featuring since were sometimes used with verbs expressing punctual events such as get (a letter), receive (a letter), etc. In cases like that the entire proposition can be said to acquire an experiential reading, as in (54) Irish English (HCIE: Sproul04) / received your letter on the second of December and one from D Herr and I send one to him with this one I want you to write several letters than I do for I have only got 3 [letters] from you since I left home.

126

7. Empirical design of the study and methodology

Second, some verbs describing momentary actions (for instance, forget) acquired a strive meaning when used in negative clauses. Thus, a sentence like / hope you have not forgot the last words I ever spoke to you to meet me in heaven. (IrE: Fife 09) can be said to convey the following meaning 'I hope you have been keeping in your memory the last words that I spoke to you all the time up till now'. Such cases were therefore treated as instantiations of extended-now contexts. By contrast, when used in a declarative sentence, forget tends to assume a resultative meaning. Hence, / suppose you almost forgot me. (IrE: Wyly 04) can be interpreted as 'You don't think of me anymore', a proposition that implies a change of state at the moment of utterance. These tokens were coded as resultative contexts. Finally, although the use of the present tense copula BE is often intended to refer to the activity continuing into the present (e.g., Since morning there is no hot water, IndEng ICE: Sla-054), there were some examples featuring copula BE that were difficult to classify as extended-now contexts. (55) Indian English (ICE: Sla-030; Sla-018) a. No initially it didn't affect me but I think now my resistance is much less and I have got sore throat, cold. '...my resistance has decreased...' b. But now this is connected with STD facility so that people ofKhadaklat need not depend upon Nipani telephone exchange. '...now this has been connected...' Both contexts strongly suggest the use of the present perfect as the speakers clearly refer to changes obtaining at the moment of utterance. Such cases were therefore coded as resultative contexts.

7.3.1.2. Identifying resultative contexts It was stated elsewhere that resultative contexts always suggest a change of state incurred by some past action at the moment of utterance. In other words, resultative contexts typically describe a past event as bringing about some current state (cf Winford 1993: 167). Since a change of state is inherent in the semantics of some verbs, tokens featuring such verbs were relatively easy to categorise as resultatives. In fact, sentences containing die, get married, improve can be rightly described as prototypical resultative contexts.

7.3. Identifying the variable context and coding the data

127

(56) Irish English (HCIE: DunneJO; ForreEOl; Dunne_09) a. But he is much improvefdj now thank God he is able to go out now. b. To the Estate Commissioners, I regret to inform you that my Mother (M=rs = Hannah Forrest) has died. c. Pers. Brady that is James Bradys youngest son got married to his first cousin a Miss O Farrellfrom Melbourne with thirty Thousand pounds fortune. A discourse marker frequently featuring in resultative contexts and giving some 'resultative' force to the entire proposition is the adverb already. (57) Indian English (ICE: Sla-035) Now we 've already, actually, we've turned it to rubber. Similarly, the use of adverbial now seems to orientate a past bounded event to the present, as in (58) Indian English (ICE: Sla-057; Sla-086) a. Now we have reduced the reservation from sixty-nine percent to fifty percent. b. My daughter's name is Aloka, she is in Holy Cross. She got now ninety-five marks in the unit test. Cases involving the adverb yet posed a special methodological problem (cf Winford 1993). The reason for this is tint yet has two meanings. According to Fenn (1987), yet has basically two interpretations, i.e. it can be construed as either "a point in time (in a sense a 'modal' variant otnow), or as a timespan in the sense up to now^ (Fenn 1987: 126). The former has a more resultative sense, while the latter is more experiential (cf. Winford 1993: 171). Hence, the following sentences seem to be resultative. (59) Indian English (ICE: Sla-035; Sla-095) a. Yeah they call me every night and they say the rains have not yet started. b.Uhl don % we haven >t decided anything yet_actually. It's a little too early for us. In both cases the speakers seem to be concerned with the present state, focusing their attention on the absence of the result at the moment of utterance.

128

7. Empirical design of the study and methodology

Accordingly, such cases were interpreted as resultatives. In some cases, however, yet seems to have a time span reading, as in (60) Indian English (ICE: Sla-057) Those people have notyetbeen given a chance to [get a job]. In this case, the speaker seems to be concerned with the non-occurrence of a series of events up to the moment of utterance. Following Winford (1993), such contexts were interpreted as experiential. It is important to mention here that yet with a resultative reading is generally much more frequent in the data than yet with an experiential reading. This observation suggests that s e m a n t i c a l ^ is still strongly associated with speech time. Finally, most resultatives do not require any adverbial specification because this present perfect context involves no time span as a rule (cf Winford 1993), as in (61) Indian English (ICE: Sla-051) A: Uhwhere have you kept your account? B: Which account? A: Money account... Uh you have opened it? B: YeahinChandargilput... A: Which bank? B.Uhl have put in bank Tokyo. Here only there is a new bank man and that he has uh kept it a nick name.

7.3.1.3. Identifying experiential contexts Since the experiential perfect indicates a situation or an event that has taken place once or several times during a period of time leading up to the present, adverbials referring to a general time span such as always, in my whole life or adverbials indicating activity within a time span such as often, sometimes, many times are good signals of an experiential context (Fenn 1987: 78; Winford 1993: 147). Furthermore, never and ever are frequently used to render an experiential reading, as in (62) Indian English (ICE: Sla-034; Sla-014; Sla-014) a. Mysore is a beautiful city. I never imagined that it's a beautiful city. b. Have you ever been to Ooty? c. Did you ever had paper on Telugu?

7.3. Identifying the variable context and coding the data

129

In other cases, however, no explicit reference to the time span was made, but the discourse context made it clear that the speaker was referring to her experience up to the present (cf Winford 1993: 140) (63) Indian English (ICE: Sla-014; Sla-095; Sla-014) a. / have been to a few places and the remaining few I plan to see later. b. A: Hello, I hear that you had gone to Baroda. '...you have been to...' B: Haan/Yes. A: For some symposium or conference? B: Seminar. CBSC seminar. c. A: He has not seen any ofthe important places. Have you not seen? B: Yeah I visited this Palace. In the latter example, even though no explicit time reference is given in response, the temporal frame of the question suggests an experiential context and the ensuing answer implies that a past action can be described in terms of repeatability. Thus, the speaker in (63c) suggests that a past event (i.e. visiting the Palace) applies to a time span (i.e. her entire life) which is not over yet and can therefore be repeated. Such cases were therefore included into the analysis and coded as experiential contexts. Notice also that most of the examples in (63) feature stative and dynamic verbs. As is clear, atelic verbs without adverbial specification often give rise to an experiential reading, as in / have lived in London (cf. Bauer 1970; Crystal 1966). Such cases were also coded as experiential contexts.

7.3.1.4. Identifying contexts of recent past First and foremost, just, lately, recently are the time adverbials associated with recent past and with the HAVE-perfect in, at least, Standard English (Quirk etal. 1985; Klein 1994). (64) Indian English (ICE: Sla-001; Sla-005) a. I'm justcome here on a holiday. b. Israel has got a government recently and that government and that government is formed by three parties.

130

7. Empirical design of the study and methodology

Other discourse markers used to identify the context of recent past are adverbials this year {month, week) and today since the major function of these adverbials is to signal recency. (65) Indian English (ICE: Sla -097; Sla-009; Sla-073) a. Today I heard that you plan going to Mysore. b. What has happened this year is really a political disaster. c. So this vear also I saw after my diploma also, many of the guys like, my classmates they got job. However, in some cases semantic contexts featuring the time adverbial today were interpreted as resultative, as in (66) Indian English (ICE: Sla-100) A: Ramakrishna has come today? B: No, hers on leave today. A: Ahha, Ramakrishna is on leave. B: / think so because he has not applied; he may apply tomorrow. In this case, the speaker's focus appears to be on the repercussions that a past event has for the moment of utterance. In other words, it is not a recent event as such that is at issue here; what seems to be important to the speaker is Ramaknshna's presence in the office because his assistance appears to be needed at speech time. This is how a past event (i.e. Ramaknshna's not coming to the office) seems to result in some current state (i.e. Ramaknshna's not being at the office when his presence seems to be required). Therefore, it could be argued that the whole context in (66) has some resultative force. In some contexts of recent past no time adverbial specification is found, asm (67) Indian English (ICE: Sla-100) A' Sir about that Chcivci I received a letter fvoYH the director of ' school education that Chaya Katti. Could you tell me something about it sir? B: Even I don't know anything about it.

7.3. Identifying the variable context and coding the data

131

(68) Irish English (HCIE: CarroJll) Dear Brother William, I have received your letter of the i t November with great pleasure to hear you and your family are all well as these few lines leaves me andfamily at present, thank Godfor his goodness. Contexts of this type were nevertheless included into the analysis since the discourse context unambiguously implies that the event occurred fairly recently and the event as such is of a certain relevance for the speakers at the moment of utterance. Because contexts in (67) and in (68) do not necessarily imply any change of state at the moment of utterance, they were treated as contexts of recent past in the data. Finally, since the main point of a discourse opening is frequently to report some recent event (cf Dahl and Hedin 2000: 385), the verb forms used to introduce a new topic can be described as pertaining to recent past contexts. (69) Indian English (ICE: Sla-050) A: So what do you think about the disaster happened, Dakshin Maharashtra about earthquake? B: Actually Umesh, as we have seen, there are a lot of people died. So according to scientists they are expecting more but this is a very serious disaster happened last month. The foregoing example contains the perfect referring to a recent event (i.e. the act of seeing the consequences of a disaster). The act of seeing as such does not necessarily incur a change of state at the moment of utterance. As a result, the present relevance of the past situation referred to is simply the one of temporal closeness (Comne 1976: 60) and the context under study is best described as a context of recent past.

7.3.2. Classifying tokens according to Aktionsart As mentioned earlier, Aktionsart is assumed to be highly implicated in the distribution of the verb forms under analysis in present perfect contexts across non-native varieties of English. This section elaborates on the major techniques used to classify the verbs in the data according to their semantics. Because there is a substantial interaction between the semantics of present perfect contexts and the meaning that a lexical verb acquires in an

132

7. Empirical design of the study and methodology

utterance (as explained in, for instance Tagliamonte 2000: 350),5 that is to say that their semantics are often conflated, the verbs were categorised according to the kinds of situations they initially describe to avoid interactions between the factor groups 'Aktionsart' and 'semantic context' in the upcoming multivariate analyses. To give a preliminary example, the inherent meaning of the verb discuss is a process; however, in a sentence We have discussed that several times, the verb discuss can be treated as an accomplishment because of the semantics of the context it occurs in, which refers to an action completed in the past. In order to avoid that kind of interdependences that exist between the Aktionsart of the verb and the semantics of the context it occurs in, the verb discuss was classified as a dynamic verb or a verb of activity irrespective of the utterance it occurred in. Similarly, to write a letter was coded consistently as an accomplishment irrespective of whether it occurred in a sentence like / have been writing this letter for two hours or in a sentence like I have written letters but not very often. Hence, all relevant lexical verbs were classified according to the typology of verbs proposed by Vendler (1967), who distinguished between states, activities, accomplishments and achievements. Because Vendler (1967) does not provide an exhaustive account of all verbs and verb groups, his methodology was expanded by the accounts of verb semantics available in the relevant literature. To start with, stative verbs generally denote emotional states {love, like, hate, feel, etc.), attitudes {believe, want, wonder, be bothered etc.), mental states {know, think (that), understand, consider, remember etc.) and sensory perception {see, hear, smell, etc.). In addition, stativity is inherent in the meaning of the verbs such as remain and contain, which basically express a "potential for permanence" (cf Mufwene 1984). As is evident, stative verbs describe situations that occur for an indefinite period of time and have no inherent end point (cf. Bnnton 1988: 28). Furthermore, states can be regarded as qualities. Thus, hearing and seeing are abilities inherent in any human being and are, therefore, states (cf. Vendler 1967: 108). Here are some examples demonstrating stative verbs in the corpora. (70) Indian English (ICE: Sla-009; Sla-007) a. Now, I've understood much from you... b. I've never seen anybody taking that in my family. 5 In her account of Samana English, Sail A. Tagliamonte (2000) points out that continuative perfects tend to be stative, resultative perfects tend to punctual.

7.3. Identifying the variable context and coding the data

133

(71) Irish English (HCIE: Johnst03) Dear Brother, as I have not heard from you for Some Years, I concluded you ware Either dead or left the Country whar you was farmarly settled. Stative verbs are, however, to be distinguished from the verbs denoting activities (cf. Vendler 1967), sometimes referred to as dynamic verbs (cf Comne 1976: 25, 49). In the following, we will use the terms "verbs of activities" and "dynamic verbs" synonymously. Dynamic verbs describe processes and activities that can last for an indefinite period of time with no inherent goal (cf. Bnnton 1988: 28). Stative verbs and verbs of activities demonstrate, however, two principal contrasts. First, activities involve some change, whereas states do not (Bnnton 1988: 28). Second, verbs of activities can be used in the progressive, whereas stative verbs cannot be used with the progressive, as a rule (cf. Vendler 1967: 99). Relying on these distinctions, we can thus single out the following verbs of activities: walk, go, flow, travel, stay, visit {stay at someone 's place), wait, listen, keep, fight, look, watch, read, learn, discuss, study, etc. (72) Indian English (ICE: Sla-016) I have been staying for thirteen years only in hostel. (73) Irish English (HCIE: GibsonOl) He thinks it very strange that he gets no letters from his brother James. He has wrote to him often and received no answers. Similarly, verbs of reporting (answer, ask, tell, say, etc.) were classified as activities because reporting is a process that involve some change and can be both voluntary (e.g., [a person] talk) and involuntary (e.g., [an article] report) (cf. Bnnton 1988: 29). In other words, the one who reports is engaged in a verbal activity just as the one who runs is engaged in a physical one. Moreover, activities are relatively homogeneous processes, i.e. "they consist of successive phases following one another in time" (cf. Vendler 1967: 99). Yet another observation is that the verbs of activities possess no inherent end points in contrast to other verbs whose semantics sets a terminal point. However, as Bnnton (1988: 26) points out, we must distinguish between verbs denoting an arbitrary terminal point and verbs descnbing a necessary goal. The verbs of the latter type are refened to as verbs of accomplishment. As the name suggests, verbs of accomplishment presuppose that the

134

7. Empirical design of the study and methodology

situation they describe has a necessary goal and are thus tehc. In contrast to verbs describing punctual events, verbs of accomplishment refer to durative situations. Vendler (1967) provides examples of accomplishments consisting of a verb and an object (e.g., paint a picture, write a letter, play a game of chess, deliver a sermon, etc.). Likewise, most of our examples for this category are represented by complex verb phrases, i.e. V + Object: choose a pair of shoes, form a party, fill petrol (into the car), complete (write) a thesis, acquire significance. Furthermore, some intransitive verbs were treated as accomplishments, for instance, come (to a place), update, or bring back. For instance, come just like go describes a process of walking; however, in contrast to go, the semantics of the verb come always presupposes a final destination (since we usually come to a place) and is telic in this respect. Finally, expressions of the type 'get + past participle', i.e. get used to, get accustomed to, get adjusted were regarded as accomplishments. Verbs of accomplishment were identified according to the test proposed in Vendler (1967: 104). The test can be described as follows: if I do A in a period of time T, then I can say that I am doing A at any moment during that period of time T. For instance, the proposition It took me three hours to come to Peter's house implies that I was in the process of approaching Peter's house at any moment of the three hours. The following examples feature the verbs of accomplishment attested in the data. (74) Indian English (ICE: Sla-002) This is first time flood came in Raelsima district. (75) Irish English (HCIE: Fife_02) We have everything about the house as neat as usual. I have run a wall across the pig trow and broke out a door opposite to the road and built a wall from the stabling hedge to the corner of the pig trow. Finally, punctual situations that take place at a specific (i.e. definite) point in time were singled out (cf Bnnton 1988: 28). Verbs describing momentary actions (cf. Leech 1971a: 18) with no inherent goal {drop, fall, snatch, die, be born, etc.) were coded as achievements. The following test was applied to differentiate between achievements and accomplishments: if I am doing A and get interrupted, can I still declare that I have been doing A? For example, if I am writing a letter and get interrupted, I can still argue a certain accomplishment of an action. Even though I may not have completed writing

7.3. Identifying the variable context and coding the data the whole letter, I can still claim that I have written at least a part of it. By contrast, if I am getting ready to get married but get prevented due to some circumstances I cannot claim to have got married because one cannot get married in a partial way. These examples illustrate verbs of achievement: break, get divorced, give, lose, pick up, submit, open, shut, cancel, leave, go abroad, etc. In addition, the verbs notice and mention were classified as achievements (cf Bnnton 1988: 33). This decision is due to the following reason: a momentary change of state implied in both notice and mention cannot express either state or activity. Of the remaining two possibilities, i.e. achievement and accomplishment, achievement seems to be a better choice because spontaneous and abrupt changes cannot in principle be durative. The following examples feature some verbs of achievement attested in the data. (76) Indian English (ICE: Sla-020) That's one of the reasons why these uhyou know schemes have failed. (77) Irish English (HCIE: Carsen02) James has finished his apprenticeship and is engaged at Ten Dollars per week without Board. The last general technique used for classifying verbs according to their Aktionsart is trying to use them in the progressive. Thus, strive verbs as well as verbs of achievement are usually incompatible with the progressive, whereas verbs of activities and accomplishments can be used as a continuous form (cf. Bnnton 1988: 39). This technique is visualised in Table 7.3. Table 7.3. Lexical verbs in the progressive accomplishment and/or activity = progressive 'states m d / o r a c h i e ; e m e r t s

-

progressh;e

It is true that stative verbs do occur occasionally in the progressive (e.g., She is smelling the roses) and in this case states can be said to denote a 'contingent activity' (cf. Hirtle 1967). However, following Bnnton (1988: 40), it is argued that states cannot be normally viewed progressively since their semantics does not presuppose any development or change. Moreover, Hirtle (1975) notices that a state "exists as a whole during each and every instance of its duration" (Hirtle 1975: 27-28, cited in Bnnton 1988: 40). A state is then to be regarded as a monolithic, unfractionable entity and its dynamic reading can be ascnbed to the context.

135

136

7. Empirical design of the study and methodology

Similarly, achievements sometimes occur in the progressive. For instance, we can say She is recognising a lot effaces here. In this case, the punctual verb recognise acquires an iterative meaning. It could, however, be quite plausibly argued that the meaning of repeatability is inherent in the context (for example, due to the interaction of the progressive form and the verb) rather than in the meaning of the lexical verb itself. We can thus conclude that verbs describing momentary actions cannot generally be viewed as durative despite the fact that there are certain contexts that allow their use in the progressive form. Finally, it must be mentioned that not all verbs presented a clear-cut case for the classification: some verbs tend to change the aspectual schema inherent in their meaning according to whether they are used transitively or intransitively, the verb to write representing a case in point. Thus, when used intransitively (e.g., to write from France) this verb can be viewed as an activity, however the transitive reading (e.g., to write a letter) gives rise to the accomplishment meaning. Thus, transitive and intransitive uses of the verb write were kept distinct during the coding. A similar (although not entirely identical) problem was caused by the verb to make: the semantics of the verb to make is not entirely immune to the semantics of the direct object following it. It is probably for this reason that the verb to make has various meanings such as create, cause to happen, reach, etc. registered in the dictionary (cf OALD 2000: 775), all of which depend on respective nouns that the verb is used with. To illustrate this point, make noise, make use (of) or make friends are activities, whereas make a cake, make a selection, make preparations are accomplishments. Finally, make a mistake and make a choice can be (and actually were) treated as achievements. Thus, while coding the verb to make for Aktionsart, special attention was paid to the semantics of the argument following it. The verb to have demonstrates a similar behaviour. Table 7.4 summarises major collocations with have. Table 7.4. Collocations with have Verb to have

used in the progressive

state (possessing) activity (experiencing) achievement (receiving)

not (usually) used in the progressive have a car, have a business, etc.

have a cold, have a warm summer, have troubles, etc. have a letter (i.e. get or receive a letter)

7.3. Identifying the variable context and coding the data

137

To bring the whole discussion to a common denominator, the influence of the present perfect context on the lexical meaning of the verb was largely ignored in the process of coding verbs for Aktionsart. Close attention was, however, paid to semantics of the arguments accompanying verbs. The major lexical verbs analysed in the study are listed for convenience below. Aktionsart Stative verbs (states): be; be bothered; be exposed; be married; believe; benefit (to be useful to someone); consider; desire; dislike; dominate; feel; hate; have (in the sense of possess); hear; imagine (that); know; like; live (i.e. be alive); love; possess; preserve (retain in a particular way); remain; remember; represent; rule; see; think that; understand; want; wonder. Dynamic verbs (processes and activities): accompany; affront; answer; apply; ask; assault; associate; attack; attempt; attract; beat; blame (also charge or accuse) (a person); change; connect; comment; deal (with a problem); decline; describe; develop; discuss; do; dream; educate; expect; experience; explain; fight; flourish; follow; go; increase; inform; improve; keep; last; learn; listen; live (i.e. spend your life in a particular way); look (at); maintain; make (noise); miss (be longing for someone); paint; participate; pass (about time); phone; plan (i.e. make arrangements); practise; postpone (i.e. keep something from happening); pull; push; read; reply; ring; run; say; seek; show (demonstrate); sow; stay; study; support (someone); swim; talk; teach; tell; think (about, of); travel; try; visit (stay at someone's house); use; wait; watch; work; write. Verbs of accomplishment (processes with an inherent goal that end up in a change of state): acquire; alter (make something become different); arrive (in a place); be brought up (in the atmosphere of freedom); become (a writer); be (get) extended; bring (a chair); choose (a pair of shoes); come (to a place); complete a thesis; construct (homes); convey (a message); convince (a person); diagnose (a disease); divide (work); disintegrate into different segments; destroy (a place); establish a proximity; evolve (a formula); fill petrol (into the car); fill up (a form); form (a party); get accustomed with (a place); get adjusted (to a new life); get excelled (in education); get settled; get standardised; get used to something; grow up; impose a concept (force people to accept it); make (a pie); move (into another place); prepare (lunch, a document); put on (weight); remove (something from one place into another); set everything (into place); transfer to (one place); transform (something into something); turn to rubber; return (to a place); write a book; undergo some changes; update.

138

7. Empirical design of the study and methodology

Verbs of achievement (momentary actions): accept; achieve; allow; arrest; be born; begin; be (or get) selected; break; buy; cancel; close; decide; declare; die; disappear; donate (money); encounter; expire; fail; find; finish; fix; forget; get; get admission; get (take up) a job; get divorced; get drowned; get married; give; go abroad; happen; have (children); hire; identify; include; introduce; invite; invest; join or unite (a group); knock (on the door); lock (a house); leave; lose; make (an observation); make up one's mind; mention; meet; miss (the bus); notice; offer (a job, a party); open; pass (an examination); pick up; receive (a letter); recognise; refuse; register; retire; select (a member); sell; send (a letter, a message); shut; spare; start doing something; stop doing something; submit; take (the subject); turn up; vanish. Having examined the major techniques used to classify all the verbs attested in the data into four major categories (states, activities, accomplishments and achievements), let us now consider methods employed for coding the data according to the parameter of transitivity.

7.3.3. Classifying tokens according to transitivity Since transitivity is believed to underlie the choice of the present perfect in the data, all verbs were also coded with respect to their ability to take arguments in a sentence. Thus, verbs that take a direct object were coded as transitive, as in (78) Irish English (HCIE: McCanc02) / have seen nobees as yet either tame or wild although I beheve that there is some in this colony. Verbs taking both direct and indirect objects were coded as transitive. (79) Irish English (HCIE: CrockeOl) Ourpubhc prints has given us several accounts of the French making a descent upon England and Ireland but that they were in every attempt disappointed. Verbs that do not take an object were coded as intransitive, as in (80) Irish English (HCIE: McCanc02) Our great railway is now all the talk but is not begun yet.

7.3. Identifying the variable context and coding the data

139

Intransitive verbs denoting motion and a change of state were coded as mutative. The following verbs attested in the corpora were coded as mutative: change, leave, go, come, vanish, die, recover, become, marry, move out (into), turn out, grow up, fail, settle (down), improve, return, breake up (down), decline, arrive, fall, increase, get thin (bolder). Special attention was paid to the verbs that can be used both transitively and intransitively. Let us consider, for instance, the verb to write. (81) Irish English (HCIE: Brenna02; Madill02) a. / rejoice to let you know that on the day I got your letter I got one from Thomas. He is in good health and he informs me that he has had letters from you frequently though you say you have wrote often toMm and got no answer. b. My Dear Margretl lift my pen to write you these few lines hopping it will find you in good health as it leaves us all at present. ...I wrote twice or three times and I got no answer. In the foregoing examples, the verb to write is used intransitively and was therefore coded as such. In other cases, the verb was used with a direct object and hence was classified as a transitive verb. See an example below. (82) Irish English (HCIE: Wyly_07) I have written three letters for the one you wrote. Similar to the verb to write, the verbs to change, to improve and to marry can be used both transitively and intransitively. Therefore, a distinction needed to be drawn between transitive and intransitive uses of these verbs. (83) Irish English (HCIE: Wyly_03; QuinnW06) a. Dear Sister I must tell you a little about work. lam still in the same place only in a different name, I told you {of}. The times have changed. Now it is the firm ofGault & Scott. b. Address Kalgoorhe Hannans Western Australia. They have changed the name you need not vut Perth on it. Whenever to change, to improve and to marry were used intransitively, they were coded as mutatives. In other cases, they were assigned a transitive label.

140

7. Empirical design of the study and methodology

7.4. Methods As mentioned above, the data is analysed both qualitatively and quantitatively in this study. Those variants used in place of the HAVE-perfect in non-native Englishes are analysed both quantitatively and qualitatively. These forms include the preterite, the present tense, the BE-perfect, bare verb stems, lone past participles, etc. Careful analysis of the use of these verb forms in the interlanguage, in the respective native language as well as in contact dialects of English (such as various settlers' dialects) can help to detect the source of the non-standard verb form in a given non-native variety. This procedure allows the analyst to determine what verb forms are amenable to an explanation in terms of substratum/superstratum influence and what variants are better analysed as a result of substrate/superstrate-independent acquisition manifested through various learner strategies. Moreover, qualitative analysis is provided for the present perfect in definite past time reference contexts because the HAVE-perfect is scarce in these linguistic environments and thus cannot be analysed with the help of the multivariate techniques.6 The major purpose of the quantitative analysis is to evaluate the role of internal and external factors underlying the occurrence of the present perfect. Both distributional and multivariate evidence is employed to that effect. The distributional analysis provides a general understanding of patterns attested in the data, whereas the multivariate analysis helps to elucidate factors that exert a persistent and statistically significant influence on the occurrence of the present perfect in a given set of data when all the factors are considered simultaneously. Hence, the multivariate analysis, or variable rule analysis (cf Tagliamonte 2006), is "the probabilistic modelling and the statistical treatment of discrete choices and their conditioning" in language (Sankoff 1988: 984). Developed by Cedergren and Sankoff (1974) and elaborated on by Rand and Sankoff (1990), Robinson, Lawrence, and Tagliamonte (2001), Sankoff, Tagliamonte, and Smith (2005), the variable rule analysis allows the analyst to model a grammar that has heterogeneity with contextual^ conditioned 'order' (cf. Tagliamonte 2006: 130) in contrast to other statistical procedures such as analysis of variance, or ANOVA, which turned out to be unsuitable for language data (cf. Rousseau and Sankoff 1978; Tagliamonte 2006). 6 In fact, definite past time reference contexts tend to be near-categorical in different Englishes (e.g., StEngEng, EAfEng, etc.). This makes the variable rule analysis impossible in such data sets.

7.4. Methods

141

The multivariate procedure is in fact the most appropriate tool for the statistical analysis of linguistic data. Because the distribution of the data among the cells cannot be controlled when a natural speech sample is used and because many different factors may influence the probability of the linguistic variant, the final configuration of data represents "high-dimensional array" where many or even most of the possible cells are empty (Sankoff 1985: 76). Estimation methods based on sum-of-squares approximations are therefore inappropriate for handling such data and the parameters of the model must be assessed with the exact maximum likelihood methods (cf Sankoff 1985). Although many statistical computing packages can carry out this type of analysis, most of the linguistic research has made use of one or the other version of the variable rule programme. Thus, Winford (1993), Tagliamonte (2000), Pietsch (2005a), Levey (2006), Szmrecsanyi (2006), Tagliamonte and D'Arcy (2007), Walker (2007), Van Herk (2008), Elsig (2009), to mention a few, are the studies employing the Goldvarb method in assessing language data. In this study, the multivariate analyses were carried out by the Goldvarb 2001, which provides users with the Windows version of the original Macintosh application (cf. Tagliamonte 2006: 158). The programme performs a multiple regression procedure (binominal step-up/step-down), determining the influence of each independent variable (or factor group) such as Aktionsart, transitivity, negation, etc. on the dependent variant (e.g., a verb form). Since the multivariate methodology requires that variants must surface frequently enough in all the contexts under study in order to be amenable to the statistical analysis, rare verb forms were excluded from the analysis (cf. Elsig, Martin 2008: personal communication).7 Thus, various sporadic items such as BE-perfects, lone past participles, lone present participles, three verb clusters, bare verb stems, etc. were banned from the variable rule analysis. Variants restricted in their distribution were likewise excluded from the analysis (cf. Winford 1993: 178). For instance, the present tense occurs predominantly in extended-now contexts across different varieties, whereas the past perfect tends to favour experiential contexts in the uppermesolectal IndEng. Because these variants showed very bad distribution (empty cells and poorly distributed cells) in the first step of the Goldvarb 2001 analysis when cross-tabulated with other factors, they were removed from the analyses (Winford 1993: 178; Tagliamonte 2006: 183). The procedures described above ultimately lead to the configurations of data featuring 7 Dr. MartinElsig, University of Hamburg.

142

7. Empirical design of the study and methodology

two dependent variants: the present perfect and pretente. These were tested in the ensuing multivariate analyses across varieties.8 Another issue that often needs to be handled during the binominal stepup/step-down analysis concerns interactions in the data. An interaction can be regarded as "an interplay of three variables: two independent variables and one dependent variable" (Elsig 2009: 37). For instance, it turned out that semantic context (i.e. the first independent variable) often interacts with time adverbial specification (i.e. the second independent variable) with respect to the choice of the HAVE-perfect. Such interaction could be resolved by merging factors within the factor group of time adverbial specification. Yet other ways of resolving interaction include removing one of the factor groups from the variable rule analysis and creating a cross-product of the two factor groups (cf Taghamonte 2006: 151-153, 185-187; Elsig 2009: 37). Interpreting results of the multivariate analysis generally involves three "lines of evidence": 1) statistical significance, i.e. factors that are selected as statistically significant by the step-up/step-down method of multiple regression 2) relative strength, i.e. factor groups that exhibit the biggest/smallest range and are thus the most/least significant and 3) constraint hierarchy, i.e. the arrangement of factors within the same group (cf. Tagliamonte 2006: 235). Decimal numbers are the so-called factor weights which "provide a numerical measure of the strength or influence of each factor relative to other factors in the same group, on the linguistic variable under investigation" (Bayley 2002: 126). The general rule of thumb for assessing the factor weights can be summarised as follows: the closer the decimal number is to 1, the more favourable the effect is. By contrast, the closer the decimal number is to 0, the more disfavouring the effect is (cf. Levey 2006: 62). The measure range indicates a relative strength of each factor group, i.e. the contribution of each factor group to the "probability of occurrence of the linguistic variable in question" (Levey 2006: 62). The total number of contexts that are considered in each step of the analysis is recorded under N, which stands for a 'total number'. Results for the factor groups that were not selected as statistically significant are presented in square brackets in the multivariate analyses of data.

8 The basilectal variety of IndEng, where the HAVE-perfect is very rare, is the only exception to that rule (see the chapter on basilectal IndEng for further details).

7.5. Summary

143

7.5. Summary The purpose of this chapter was to elaborate on the general sample design of the study and to provide an overview of the major methodological issues involved in identifying the linguistic variable and extracting the data. First, we elaborated on the principal techniques used for circumscribing the variable context. Moreover, we have given a comprehensive account of the strategies used to classify the data according to semantic context, Aktionsart and transitivity. Finally, we have accounted for the major methods employed in the study, while explaining their appropriateness for this investigation.

8. The reference variety of Standard English English (LLC)

8.1. Data In order to understand the mechanism that triggers the occurrence of the HAVE-perfect in various non-native Englishes, we need to single out constraints that underlie the use of this verb form in the variety of StEngEng. This variety will serve as a yardstick against which we can study nonnative varieties in order to detect and explain the similarities and differences in the patterns of use of the present perfect. The decision to explore StEngEng rather than American English is due to three major reasons. First, the present perfect is much more frequently attested in the former rather than in the latter (cf Elsness 1997). Second, most of the non-native varieties under study are oriented towards the standard spoken in England rather than in the United States. For instance, it is the standard variety of English English which enjoys a prestigious status in India and in many parts of Africa (cf. Schneider 2007: 171, 194, 198, 202, 207, 216). Additionally, the foreign-speaker varieties of English studied here are also oriented towards StEngEng. More importantly, theoretical descriptions of the present perfect provided in previous chapters draw on evidence obtained from descriptive grammars of StEngEng such as Quirk et al. (1985), Huddleston and Pullum (2002), etc. The authors of such grammars often use the norms of the educated native speakers in England as a reference point for their discussions of the Standard English dialect. It is such grammars that shape the acquisitional context of English (at least in a classroom setting) in many countries where the language is spoken non-natively. Analysis of StEngEng draws on data obtained from the London-Lund Corpus (LLC) of Spoken English (cf. Svartvik 1990), collected and transcribed in London. Informants selected for this corpus therefore represent educated native speakers of English. The analysed corpus material encompasses spontaneous surreptitiously recorded (telephone) conversations between intimates and "distants" and non-surreptitious private and public conversations. Excluded from the analysis were the texts containing monitored speech (sports and radio commentaries, demonstrations of experiments, speeches in parliamentary debates, sermons, political speeches, etc.) because this type of data exhibits less robust variation of verb forms in the

8.2. Quantitative analysts

145

contexts under analysis. The analysed corpus contains a total of 76 texts (some 400,000) words. This chapter focuses on the distributional and multivariate analyses of factors constraining variation of the present perfect (i.e. auxiliary HAVE + past participle) in present perfect contexts and then proceeds to presenting the results of the qualitative analysis dealing with the use of HAVE-perfects in definite past time reference contexts in StEngEng. 8.2.

Quantitative analysis

8.2.1.

Distributional analysis

8.2.1.1. Variation m present perfect contexts Following is the distributional analysis of the linguistic items in present perfect contexts in StEngEng. Table 8.1. The overall distribution of variants in StEngEng HAVE-perfect preterite present BE-perfect reduced perfect

1812 197 4 2 1

Total N

2016

(90%) (9%) (0%) (0%) (0%)

As is clear, the HAVE-perfect is the predominant verb form in the data, which is not surprising given the descriptions of this verb form in traditional grammars and empirical studies. The second most frequent variant is the preterite. This is again not surprising; the simple past tense has been frequently described as the major rival of the perfect across varieties of English (cf. Elsness 1997, 2009; Van Herk 2008). In more general terms, the overall distributional analysis corroborates some important findings of the previous research. First, it nicely tallies with the claim that the present perfect is a fairly frequent item in British English (cf. Elsness 1997: 83-84; Gathercole 1986: 540).1 Indeed, the perfect is preferred over the preterite in 1 By contrast, in American English the preterite is in many cases preferred over the present perfect and is therefore more robust in this variety.

146

8. The reference variety of Standard English English (LLC)

many cases in this variety of English, a tendency that is reflected in the results of the overall distributional analysis. Second, the preterite is somewhat infrequent in present perfect contexts what empirically confirms an assumption often expressed in traditional grammars of English that the chief function of this verb form is that of the narrative tense (cf Quirk et al. 1985:185; Elsness 1997: 81; Biber, Conrad, and Leech 2002: 151). To provide an example, Elsness (1997: 84) reports a higher proportion of the preterites in the fictional text categories, typically associated with narration. He furthermore observes that the higher (relative) frequency of the preterite in fictional texts corresponds to a higher frequency of the present perfect in the informational prose (e.g., articles in newspapers), which is typically more oriented towards the present (cf. Elsness 1997: 86). Rare forms are the present tense, the BE-perfect and the so-called reduced perfect represented by construction of the type been + present participle (e.g., been going). (84) Standard English English (LLC_08) Tell me about yourself. What you been doing? The use of the present tense in StEngEng is restricted to the copula BE in extended-now contexts. In other words, in StEngEng the present tense is only attested in sentences of the type It's ages since I have seen you (LLC 01) and It's a long time since you read any Chaucer (LLC 03). The attested two tokens of the BE-perfect surface with the lexical verb go. Thus, once robust competitors of the present perfect (Mustanoja 1960; Siemund 2004), the present tense and the BE-perfect have become rudimentary elements in the temporal domain of the present perfect marking in contemporary StEngEng, surfacing only very infrequently in highly (lexically) restricted contexts. In what follows, we will therefore focus on the examination of the distributional mechanism of the HAVE-perfect and preterite. As elaborated on in the preceding chapters, several factors have been identified as affecting the choice of the present perfect: semantic contexts, Aktionsart, polarity (i.e. negation), transitivity and time adverbial specification. Preliminary (distributional) analysis reveals some important trends emerging in the data, showing how the verb forms under study pattern with the independent features of the linguistic environment.

8.2. Quantitative analysis

147

8.2.1.2. The present perfect and preterite in present perfect contexts The first important observation arises out of the examination of the distribution of the present perfect and preterite across semantic contexts, as Figure 8.1 shows. 100% • perfect • preterite 80%

60%

I

40%

I

I

20%

0% resultative

-

extended-now

•

experiential

•

recent past

Figure 8.1. Distribution of the perfect and preterite across semantic contexts in StEngEng As one can see, the present perfect is robust in all present perfect contexts. Yet the most preferred contexts are resultative and extended-now contexts. In experiential and recent past time contexts the perfect is less robust, constituting only 81% and 76% of the data respectively. These results are by and large compatible with those provided in Gathercole (1986), who studied the use of the present perfect in speech of four native-born Scottish women. Gathercole's finding is that the perfect is used in resultative and extended-now contexts almost exclusively, whereas it is used interchangeably with the preterite in contexts of recent past, although the perfect is preferred in these environments (cf Gathercole 1986: 549). From a wider (historical) perspective, our results tally quite nicely with the finding provided in Elsness (2009), who showed that the present perfect slowly loses its ground against the preterite in British English as exemplified by the Freiburg LOB corpus of British English (FLOB) * Yet another factor which is presumably implicated in the choice of the perfect is time adverbials. As suggested in the theoretical part of the study, these form elements differ from each other with respect to the expression of 2 See Elsness (2009) for further details.

148

8. The reference variety of Standard English English (LLC)

current relevance: thus current relevance was postulated to be inherent in the meaning of the time adverbials of the '+ current relevance' group, whereas it was claimed to be less vividly manifested with other time adverbials which have been labelled as '± current relevance' group. The following figure demonstrates the distribution of the perfect and preterite with various time adverbials. perfect

1

11

1

preterite

11 1 1

y*/

Figure 8.2. Distribution of the perfect and preterite by time adverbial specification inStEngEng Notice that the preterite occurs fairly frequently with time adverbials never (30%), ever (20%), always (26%), before (31%), recently (17%) and just (17%). Interestingly, all these time adverbials belong to the '± current relevance' group for which current relevance have been postulated to be less vividly manifested. The meaning of the adverbs ever, never, always and before can be described as the one of iterativity which lends the idea of recurrence or "repeatability" (Fenn 1987) to the predication. In other words, semantically these time adverbials refer to a time span which is not yet over, thus suggesting that the event can be repeated. We can conclude that semantically ever, never, always and before are compatible with the meaning of an experiential context, which also tends to pattern with the preterite. Consider now the meaning of the adverbial so far, which gives a resultative force to the entire proposition, suggesting that a past event has led to a current result or state (Fenn 1987). The adverbs already, yet and now seem to convey a similar meaning orienting the past event towards the present. The distribution of these time adverbials (which have been classified as '+ current relevance' group in the theoretical discussion) suggests that the

8.2. Quantitative analysts

149

preterite is almost non-existent in these linguistic environments because its semantics (i.e. definite past time reference) clashes with the explicit orientation towards the moment of utterance inherent in the semantics of the time adverbials so far, already, yet and now. Another important observation is that the HAVE-perfect is more frequent in contexts without any time adverbial specification (e.g., I have broken my leg). The hypothesis that temporally unspecified contexts should be preferred by the perfect has been borne out by this data set. Furthermore, this finding is very much in line with that provided Hundt and Smith (2009), who report in their survey of the present perfect and preterite in British and American English that the HAVE-perfect is used without time adverbial specification much more frequently in both varieties of English. Table 8.2. Co-occurrences of the present perfect and temporal specification, adapted from Hundt and Smith (2009: 63) LOB

FLOB

Brown

Frown

with temporal spec. without temporal spec.

275(33.4%) 548(66.6%)

248(31.7%) 535(68.3%)

240(34%) 466(66%)

211(31.5%) 459(68.5%)

^tam

8^

7^

m

6^

As is obvious, semantic contexts and semantics of time adverbials seem to be highly implicated in the distribution of the present perfect in StEngEng. Yet the present perfect patterning with Aktionsart and transitivity offers further insights into the nature of the variability of the linguistic variable under study. The distributional analysis of the present perfect and preterite by Aktionsart shows that the present perfect is very frequent with both stative and dynamic verbs on the one hand and verbs of achievement and accomplishment on the other. Table 8.3.

Distribution of the perfect and preterite by Aktionsart in StEngEng HAVE-perfect

stative dynamic accomplishment achievement Total N (2009)

578(87%) 645(91%) 130(92%) 459(90%) 1812

preterite 81(12%) 57(8%) 10(7%) 49(9%) 197

150

8. The reference variety ofStandardEnglish English (LLC)

Note also individual distributions of the HAVE-perfect and the preterite across lexical verbs of various types, which reveal that both verb forms exhibit a similar distribution - albeit at different rates - with various semantic predicates.3 • HAVE-perfect • preterite 2000 1800 1600 1400 1200 1000 800 600 400 200

L stative

dynamic

accomplishment

achievement

Total N

Figure 8.3. Distribution of the perfect and preterite across lexical verbs in StEngEng4 These results can be nicely tailored with the findings of the LSWE corpus (cf Biber, Conrad, and Leech 2002) which also demonstrate that the English present perfect generally collocates very well with the verbs of all types in Vendler's classification. What this kind of distribution seems to suggest is that the StEngEng present perfect, which is historically associated with telic verbs, has become extended to atelic verbs (i.e. stative and dynamic verbs), having thus generalised its meaning from "current result" to "current relevance" in StEngEng (Lindstedt 2000: 368). Yet, on taking a closer look at the distributional patterns, we might notice that the present perfect is a little bit less robust in stative contexts, constituting only 87% of the data in contrast to achievements and accomplishments (90% and 92% respectively). This difference is not dramatic but one might still wonder why it is so. A more careful observation of patterns in the data reveals that 3 The lower rates of the verbs of accomplishment in Figure 8.3 can be related to verb frequency: whereas the classes of stative and dynamic verbs as well as verbs of achievement contain some highly frequent lexical items (e.g., go, see, hear, etc.), the class of verbs of accomplishment is composed of predicates that are much less frequent (e.g., write a letter, chose a pair of shoes, etc.). 4 This figure contains raw token counts.

8.2. Quantitative analysts

151

there are two lexical verbs that show a propensity to surface with the preterite rather than the perfect in the data: (85) Standard English English (LLC_02) ci I YIQVQY heard from her since b. The best thing I saw ever like that was Rosemary Baby. The tendency exhibited by some stative verbs to resist the present perfect is therefore still existent in the standard dialect of English but seems to be restricted to certain lexical verbs such as see and hear. Notice also that both see and hear are high frequency verbs in English what might account for their reluctance to adjust to new patterns since high frequency items tend to be more "entrenched" in the human mind and thus could plausibly be argued to be less prone to processes triggering changes in language (cf Croft and Cruse 2004). Finally, co-occurrence patterns of the present perfect with the verbs of the factor group 'transitivity' reveal the highly grammaticalised status of the HAVE-perfect in StEngEng which has expanded its functions from transitive to intransitive (including mutative) verbs and is highly frequent in both environments. Table 8.4. Distribution of the perfect and preterite by transitivity in StEngEng HAVE-perfect

preterite

transitive intransitive Copula BE (HAVE BEEN vs. WAS)

1153(90%) 397(89%) 262(90%)

124 (9%) 47(10%) 26 (9%)

Total N (2009)

1812

197

Yet another observation is that the copula HAVE BEEN is much more frequent in the data than the copula WAS. Not only has the perfect expanded its functions to virtually all lexical verbs; it is also very robust with function words such as copulas (what can be viewed as an additional indication of its advanced position along the grammaticalisation path in StEngEng). Our last observation in this section concerns the factor group of negation. The claim that negation is associated with the present perfect (cf. Elsness 1997) is not borne out here, as the present perfect is as frequent in affirmative contexts as in negation environments including constituent negation

152

8. The reference variety of Standard English English (LLC)

(no, none, nothing, etc.). More importantly, this factor group was not selected as statistically significant in the upcoming multivariate analysis. Table 8.5.

Distribution of the perfect and preterite by negation in StEngEng HAVE-perfeet

preterite

affirmative negation

1520(90%) 292(89%)

161 (9%) 36(10%)

Total N (2009)

1812

197

The preliminary analysis thus reveals that the distributional patterns of the present perfect across semantic contexts and time adverbials are consistent with the notion of current relevance, which has been postulated to underlie the choice of the perfect in StEngEng in the theoretical discussion. In particular, semantic contexts and time adverbials for which current relevance has been argued to be more vividly manifested appear almost exclusively with the perfect. By contrast, semantic contexts and time adverbials that have been claimed to be less explicit with respect to the expression of current relevance demonstrate a noticeable occurrence with the preterite. Furthermore, distribution of the present perfect by Aktionsart and transitivity has corroborated the initial assumption that the present perfect is a fullyfledged (grammatical) category in StEngEng as it is used in an overwhelming majority of present perfect contexts.

8.2.2. Multivariate analysis We have seen so far in which linguistic environments the StEngEng present perfect tends to surface in spontaneous speech. What still needs to be uncovered is which language-internal categories actually trigger the occurrence of the linguistic form under study, in which order and to which extent. To be able to provide a definite answer to this question, we need to carry out an analysis that would consider all factors simultaneously, thus revealing the underlying system that produces the variation in the body of material. Let us therefore consider the results of the multivariate analysis, a research method specifically designed to deal with the issue.

8.2. Quantitative analysis

153

Table 8.6. Multivariate analysis of the contribution of factors selected as significant to the probability of the perfect in StEngEng (LLC) Perfect Input value

Factor weight

% 90%

N 1812/2009

1. Semantic context resultative extended-now experiential recent past

.69 .63 .27 .19

96% 95% 81% 76%

788/813 394/411 430/525 200/260

range

50

2. Time adverbial specification unspecified time adverbial

.54 .39

92% 83%

1340/1446 472/563

range

15 [-51] [-47]

90% 90%

1153/1277 659/732

[-54] [-49]

89% 90%

292/328 1520/1681

[-53] [-48] [-48]

91% 90% 87%

645/702 589/648 578/659

3. Transitivity transitive intransitive range 4. Negation negation affirmative range 5.Aktionsart dynamic change-of-state verbs5 stative range

Two factor groups were selected as statistically significant - time adverbial specification and semantic context. However, the most important factor group underlying the use of the present perfect in StEngEng is that of 5 Henceforth, we use the term "change-of-state verbs" to refer collectively to verbs of achievement and verbs of accomplishment. A linguistic justification for merging verbs of both types into one group runs as follows: verbs of accomplishment and verbs of achievement always express a change of state. It is on this ground that we can combine both to create a new linguistic factor.

154

8. The reference variety of Standard English English (LLC)

semantic context exhibiting the range of 50. This finding is consistent with the initial theoretical assumption that the use of the present perfect in StEngEng impinges to a large extent on its semantic/pragmatic realisations. More important, however, is the order in which these semantic contexts are arranged within the factor group. A closer observation of the constraint hierarchy of this independent variable reveals that in fact, there are two contexts favouring the use of the present perfect in StEngEng - resultative and extended-now contexts. As expected, resultative and extended-now contexts occupy the first two positions in the ranking followed by experiential and recent past use types. The order of constraint ranking (i.e. from resultative through extended-now contexts to experiential and recent past time contexts) has confirmed the initial hypothesis formulated as a hierarchy of uses of the present perfect in Chapter 4. Yet another factor group implicated in the distribution of the perfect and preterite in StEngEng is that of time adverbial specification. Although the effect demonstrated by this independent variable is fairly slight (range 15), results of the analysis vividly indicate that the StEngEng perfect tends to surface in sentences not featuring any time adverbial specification, favouring temporally unspecified contexts at .54. Furthermore, cross-tabulating factor groups of semantic context and adverbial specification revealed that a relatively large proportion of all unspecified contexts (slightly less than 50%) are represented by resultative uses. This finding is consistent with the description of resultatives provided in Winford (1993), who observes that we often find no time adverbial specification in resultative contexts because "the resultative perfect does not crucially require adverbial specification for its meaning to obtain" (Winford 1993: 148). It is interesting to observe how semantic context patterns with adverbial specification triggering the occurrence of the present perfect in our data. We could now formulate the linguistic environment in which the appearance of the present perfect in StEngEng is the most probable - a resultative context that is temporally unspecified. This linguistic environment is a logical configuration substantiated by theoretical and empirical research on the present perfect. First, it corroborates the theoretical insight that resultativeness is the most vivid manifestation of the present perfect in StEngEng (cf Comne 1976; Siemund 2004). Second, the pattern of unspecified contexts favouring the perfect suggests that indefimteness is one of the major properties of the English perfect as opposed to the defimteness of the simple past tense (cf. Radden and Dirven 2007: 213, 219).

8.2. Quantitative analysts

155

8.2.3. Discussion of findings How can we account for the fact that resultative and extended-now contexts are favoured by the present perfect more than experiential and recent past time contexts in StEngEng? Relying on the line of argumentation defended in the theoretical part of the study, we suggest that the key to an explanation of this particular pattern lies in the semantics of present perfect contexts which were claimed to differ from one another with respect to their orientation towards speech time. In Chapter 4, it was argued that semantically both resultative and extended-now contexts are intrinsically linked with the moment of utterance what explains the frequent occurrence of the perfect rather than the preterite in these contexts. By contrast, the orientation towards speech time is less explicit in the semantics of experiential and recent past time contexts which is why the use of the past tense is more expected in these semantic environments. It seems that the notion of current relevance (i.e. "the linking of a past situation towards the moment of utterance" (Siemund 2004: 414), which implies a direct orientation towards the moment of utterance) underlies the occurrence of the HAVE-perfect in StEngEng. The second issue that needs to be addressed here is why neither Aktionsart nor transitivity was selected as statistically significant for the use of the present perfect in StEngEng. The non-significance of both factor groups can be explained by the fully developed grammatical status of the present perfect in native English. As the perfect starts its long way along the grammaticalisation path its occurrence can be triggered by certain verbs (for instance, transitive verbs denoting a completed action) and less so by others (for example, intransitive verbs denoting mental activity, visual perception, etc.). At this stage both Aktionsart and transitivity can be said to be highly implicated in the occurrence of the linguistic form under study (cf Carey 1990, cited in Slobin 1994; Mustanoja 1960). However, as soon as the perfect has overgenerahsed its meaning from current result to current relevance, the distinction between lexical verbs (both transitive and intransitive) does not seem to be relevant any longer: semantic context becomes more decisive for the use of the present perfect. At more advanced stages of the grammaticalisation of the perfect, current relevance is no longer an imp l i c a t e that a hearer can infer relying on the meaning of the entire proposition; it is an essential part of the semantic meaning of the present perfect underlying its appearance. Summing up, one might say that linguistic factors underlying the use of the present perfect in StEngEng is a logical synchronic reflex of its diachromc development.6

156

8. The reference variety of Standard English English (LLC)

Finally, why should the present perfect be favoured in temporally unspecified contexts? The answer can be once again related to the semantics of the present perfect. To start with, one of the functions of the present perfect has been described as that of indefinite past time reference (cf Leech 1971a) or indefimteness (cf. Quirk et al. 1985; Radden and Dirven 2007) in the relevant literature. While using the perfect, one often makes a statement concerning an event in the past without making explicit when this event occurred. According to Leech (1971a), "it is natural to start a conversation indefinitely, and then to progress to definite reference (past tense, definite article, personal pronouns) once a frame of reference has been established" (Leech 1971a: 37). Typically, indefinite past time reference does not call for time adverbials, which explains why the perfect is frequently used without temporal specification. To sum up, the indefinite past time theory, which assumes that the major function of the perfect is that of indefinite past time reference, nicely explains the frequent occurrence of the HAVEperfect without any time adverbial specification in our data.

8.3. Qualitative analysis: The present perfect in definite past time reference contexts Up until now we have been concerned with cases in which the present perfect occurs in contexts described as regular present perfect contexts in traditional grammars of English. The LLC corpus, however, contains 17 cases where the perfect surfaces with definite past time adverbials, as in (86) StEngEng (LLC_02) a. Charlie Wilson's just been here - he's been here two or three weeks ago. I left on July the twenty-eighth. b. It is a community where people seem to be seriously influenced by all these things which have happened so long ago but those things are so much more intense over there. Notice that the use of the perfect in this type of contexts is not in accordance with the descriptions provided in grammars on Standard English. Although rather infrequent, these instantiations of the present perfect are nevertheless 6 It must be mentioned in brackets that supportive diachronic evidence in other research is hard to obtain as most historical studies on the present perfect do not employ variationist methodology.

8.3. Qualitative analysis

157

attested in spontaneous speech of adult educated native speakers of English and require an explanation concerning their occurrence. To start with, constructions of the type They have done a course two years ago are attested across regional varieties fairly frequently and have traditionally been analysed as overgenerahsations of the present perfect to simple past contexts. (87) New Zealand English (from Bauer 1989: 71) Sanctions have been imposed by the UN thirteen years ago. (88) South British English (from Trudgill 1984: 42) I've seen him last year. However, considered within a context of a native variety of English such an explanation can hardly be regarded as a satisfactory one since overgeneralisation is a term traditionally employed to refer to context involving nonnative acquisition of language. Yet another possible scenario is that the perfect with definite past time reference is a residue of a historical pattern, once well and alive in English (cf. Gorlach 1991: 111; Elsness 1997: 248). Dating back as early as 1920, the following example demonstrates the occurrence of the perfect with definite past time adverbials, a trend which is still extant in StEngEng vernacular: (89) Modern English (from Galsworthy, In Chancery, 1920, cited in Elsness 1997: 250) I have been to Richmond last Sunday. At first glance, such constructions attested in our data might indeed look like a reflex of an earlier development in the distribution between the perfect and preterite. However, on closer observation of the LLC corpus it becomes clear that the HAVE-perfects forms are often triggered by certain environments. To start with, some of the perfects featuring definite past time specification appear immediately after another perfect or a present tense form in the discourse. Second, many HAVE-perfects are punctualtelic predicates (for instance, send, meet, occur, happen, etc.). Consider, for instance,

158

8. The reference variety of Standard English English (LLC)

(90) StEngEng (LLC_09)

a. / have had no reply. I mean, I have sent the first letter two or three months ago and the last one about a month ago. b. A: Well, I don't know that I'd go as far as to say that I was fairly well equipped. I mean I've read generally. B: What is your favourite reading novel? A: Any purpose in mind, you know. B:No. A: But this has only occurred to me during the last year. It seems that / have had no reply takes the scope over the entire proposition so that the event of sending a letter anchored in the definite past is still perceived as currently relevant. Furthermore, semantics of the verb of achievement send contributes to the speaker's viewing results of a past action as completed at the moment of utterance and still valid to the point she is making. Similarly, the sentence I've read generally establishes the temporal framework for the following discourse. The effect is that the speaker regards the event as being connected to speech time despite definite past time adverbial specification. Following the logic provided in Dahl and Hedin (2000), one could argue the speaker imposes constraints on the pragmatic discourse by viewing an event that indeed occurred at some definite time in the past as being somehow relevant to her claims made at the moment of utterance; the notion of current relevance is arguably contained in the speaker's attitude towards contents of utterance. For instance, in (90a) above, the speaker's focus is on the entire proposition rather than single events (i.e. sending letters) which she wants to portray as being valid with regard to speech time, thus using the present perfect in contexts that would be unacceptable under normal conditions. The speaker's attitude towards the proposition^ content of utterance, i.e. her regarding the situation as related to the moment of utterance, can be analysed as a constraint conditioning the use of the present perfect in definite past time reference contexts in StEngEng. Contextual clues such as verb semantics and the HAVE-perfect introducing the discourse topic play a vital role in reconstructing the speaker's viewpoint on the entire proposition.

8.4. Summary

159

8.4. Summary The major purpose of this chapter was to take a closer look at the variation of the present perfect and other verb forms in the educated native vanety of English spoken in England. First and foremost, the overall distributional analysis revealed an extremely high overall rate of the HAVE-perfect. Furthermore, we saw that the only other variant competing with the present perfect in present perfect contexts is the preterite. Second, the factor-by-factor analysis demonstrated that the present perfect virtually dominates all inherent lexical contexts (state, activity, accomplishment and achievement) on the one hand and transitive and intransitive environments on the other. We interpreted this type of distribution as a sign of a fully grammaticalised status of the present perfect in StEngEng. Third, multivariate procedure revealed that there are two independent factors underlying the occurrence of the HAVE-perfect in StEngEng: semantic context and time adverbial specification. As far as semantic contexts are concerned, the order of constraint ranking (i.e. from resultative through extended-now contexts to experiential and recent past time contexts) corroborates the initial theoretical assumption that the resultative and extended-now contexts are most closely associated with the category of the English present perfect. Being manifested with various degrees of intensity in different use types, the semantic component of current relevance thus underlies the occurrence of the present perfect in StEngEng. Moreover, that the StEngEng present perfect is preferred in temporally unspecified contexts nicely tallies with the theoretical observation concerning the major function of the present perfect as indefinite past time marker. Finally, the qualitative analysis revealed that the occurrence of the present perfect in definite past time reference contexts may be pragmatically constrained by the speaker's attitude towards the proposition^ content of the utterance, which she might view as relevant to (and still obtaining at) speech time.

9. Acrolectal variety of Indian English (HCNVE)

9.1. Sociolinguistic history of Indian English The sociolinguistic history of IndEng has been properly dealt with elsewhere (see the chapter on non-native speaker varieties of English). Hence, we will not consider this topic in any close detail here. One remark is, however, in order. The previous discussion has revealed that IndEng is a highly heterogeneous variety that can be subdivided into various lects the speakers of which exhibit varying degrees of proficiency in the target language. Competence or proficiency in IndEng can thus be ranked on a "cline of bilingualism" (Kachru 1965: 393, cited in Gargesh 2006: 92), "an ascending scale that begins with the most pidgimsed variety and ends with an educated variety, with intermediate points with more or less language mixing" (Gargesh 2006: 92). In the following chapters, we will describe, analyse and discuss variation in present perfect contexts in different types of IndEng, representing various stages on the scale of proficiency in English.

9.2. Data The data for this variety of English was elicited through interviews with 4 informants (two males and two females) from Jawaharlal Nehru University in New Delhi. All interviewees claimed to have been in contact with the medium of English for more than 15 years. All informants referred to Hindi as their native language.1 English, on the other hand, was referred to as a second language which had never been used as a (predominant) means of communication within the family. However, English existed in the linguistic repertoire of all informants at least since primary school. Since all the informants representing this group went to prestigious public schools, they received instruction in English in all subjects from very early on. In general, public schools ensured close and continuing access to the Standard 1 It must be mentioned here that two informants described Punjabi as one of their substrate languages and one informant referred to Sindi as one of his native languages besides Hindi. In view of the fact that all of them showed extremely high proficiency in English so that substrate influence effects were minimal, these informants were included in the sample.

9.3. Quantitative analysts

161

(British) English model. More specifically, all informants admitted that they aspired to Standard English as their target. Last but not least, English was used as a means of communication with friends and outsiders. Given such an intense contact with the language, this group of informants was classified as acrolectal speakers of IndEng 2 Each informant was interviewed in a separate session, each session lasting for about an hour. The material collected for this case study is represented by approximately 4 hours of recorded and transcribed spontaneous speech data yielding 26,350 words. In the following sections we will put the hypothesis that female speakers use the HAVE-perfect much more frequently to the test. Furthermore, we will elaborate on the major verb forms used in present perfect contexts in the acrolectal variety of IndEng and the mechanism underlying their occurrence. In doing so, we will present patterns accounting for the close affinity between acrolectal IndEng and StEngEng. The final section deals with the qualitative analysis of the present perfect in definite past time contexts.

9.3.

Quantitative analysis

9.3.1.

Distributional analysis

9.3.1.1. Extra-linguistic factors Figure 9.1 demonstrates that males use the present perfect more frequently than females. Female speakers, on the other hand, use the simple past tense more often.

2 Although acrolectal speakers of IndEng came into contact with English fairly early in life, they never really used the language in the most private spheres of their lives (such as family and close relatives). Furthermore, their exposure to English was by and large limited to the input from advanced second-language learners of English. It is in this sense that the English spoken by the acrolectal IndEng speakers can be characterised as non-native (see also discussion in Mesthrie and Bhatt 2008: 36-37).

162

9. Acrolectal variety ofIndian English (HCNVE)

100% I male

• female

80%

60%

40%

20%

0% HAVE-perfect

J

preterite

present tense

BE-perfect

three verb clusters

Figure 9.1. Percentage distribution of variants by sex in acrolectal IndEng The hypothesis that female learners should use the HAVE-perfect more frequently because of its association with prestige and Standard English is not corroborated by this data.

9.3.1.2. Language-internal factors 9.3.1.2.1. Variation in presentperfect contexts Table 9.1 illustrates the major verb forms used in present perfect contexts in the acrolectal variety of IndEng. Table 9.1.

Overall distribution of variants in acrolectal IndEng N

%

HAVE-perfect preterite present BE-perfect three verb cluster

182 54 3 2 2

74% 22% 1% 0% 0%

Total N

243

Notice first of all the robust occurrence of the HAVE-perfect (74%) in the data. Indeed, all informants on the acrolectal variety of IndEng used the perfect consistently in their answers to questions that both featured and lacked the perfect.

9.3. Quantitative analysts

163

The only other verb form regularly surfacing in present perfect contexts is that of the simple past tense. Notice also a proportional correlation between these two forms that is close to the standard. Thus, in the IndEng acrolect we attest the present perfect (74%) vs. the preterite (22%), whereas in StEngEng this correlation is as follows: the present perfect (90%) vs. the preterite (9%). Similar to StEngEng, the use of the present tense is limited to contexts featuring the copula BE. (91) Acrolectal Indian English (HCNVE: IE01; IE03) a. It is a culture which is alive for three to five thousand years with such a rich history of art and dance and music and language. b. Since then, anybody who comes from that political section, he's called Naxal. The BE-perfect and three verb clusters are also sporadic and will not be treated here in close detail. In what follows, we will focus on the two most frequent items in our data - the present perfect and the preterite.

9.3.1.2.2. The present perfect and preterite m present perfect contexts As noted above, the HAVE-perfect is the most frequent form in the corpus what provides a nice parallel to StEngEng, where this surface variant is also very robust. The mere existence of a form is, however, not sufficient to identify the underlying grammatical mechanism that produced it. We will therefore present a detailed analysis of factors conditioning variation in present perfect contexts in acrolectal IndEng below. Consider, first of all, distribution of the present perfect and preterite across semantic contexts in the acrolectal variety of IndEng (Figure 9.2). Clearly, there are two contexts in this set of data in which the present perfect is the most frequent - resultative and extended-now contexts. By contrast, the preterite is more robust in experiential and recent past time contexts. Another factor group which might be implicated in the occurrence of the HAVE-perfect in this variety of English is that of time adverbial specification. Figure 9.3 demonstrates the overall distribution of the present perfect and the preterite in temporally (un)specified contexts.3 3 More differentiated analysis featuring single time adverbials (ever, never, etc.) was not possible due to low token counts.

164

9. Acrolectal variety ofIndian English (HCNVE)

100%

•

perfect preterite

80%

60%

40%

20%

0% resultative

extended-now

experiential

recent past

Figure 9.2. Distribution of the perfect and preterite across semantic contexts in acrolectal IndEng 100% • perfect 80%

60%

1

^m

40%

20%

0% ,

^H time adverbial

^ | unspecified

Figure 9.3. Distribution of the perfect and preterite by time adverbial specification in acrolectal IndEng Similar to StEngEng, the present perfect is more frequent than the preterite in temporally unspecified contexts. This finding substantiates the claim that being closely associated with the meaning of indefimteness, the present perfect is preferred in contexts without any time specification (cf Leech 1971a; Radden and Dirven 2007). The perfect is, however, also preferred in contexts featuring time adverbials. The factor group 'Aktionsart' shows further insights into the nature of variability of the present perfect in the variety of English under study.

9.3. Quantitative analysts

165

Table 9.2. Distribution of the perfect and preterite by Aktionsart in acrolectal IndEng HAVE-perfect

preterite

^twe

Ju^%)

1502%)

dynamic accomplishment achievement

66(75%) 32(84%) 33 (73%)

21(24%) 6(15%) 12(26%)

Total N (236)

M

54

Table 9.2 shows that the HAVE-perfect is fairly well established with the lexical verbs of all types in Vendler's typology. Similar to StEngEng, the perfect shows a preponderance of stative and dynamic verbs on the one hand and verbs of accomplishment and achievement on the other. A nearly complete generalisation of meaning of the HAVE-perfect is thus confirmed for this standard-conformant non-native variety of English. It means that the meaning of current relevance is associated with (and applied to) not only punctual-telic but also atelic verbs in acrolectal IndEng, thus encompassing a wide range of contexts. A similar picture emerges from the observations of the distributional patterns by transitivity. Table 9.3. Distribution of the perfect and preterite by transitivity in acrolectal IndEng HAVE-perfect

preterite

transitive

95(75%)

31 (24%)

intransitive

87(79%)

23 (20%)

Total N (236)

M

54

The HAVE-perfect is highly frequent with both transitive and intransitive (including mutative) contexts. This distributional pattern suggests a highly developed grammatical status of the HAVE-perfect in the IndEng acrolect. Note that this pattern also exhibits a striking parallel to StEngEng, where the HAVE-perfect was shown to be highly frequent with both transitive and intransitive environments at the overall rate of 90%.

166

9. Acrolectal variety ofIndian English (HCNVE)

Finally, the factor-by-factor analysis of the factor group 'negation' reveals a positive correlation between the HAVE-perfect and contexts featuring negation (both ^-negation and ^-negation). Table 9.4. Distribution of the perfect and preterite by negation in acrolectal IndEng HAVE-perfect

preterite

affirmative negation

164(76%) 18(85%)

51 (23%) 3 (14%)

Total N (236)

M

54

As can be seen, the perfect is used with negative statements (85%) more frequently than with affirmative ones (76%). The claim that the present perfect should attract negation environments (Elsness 1997) is borne out by this set of data. Recall that in StEngEng the perfect was used in both negative and affirmative environments equally frequently. Summing up, co-occurrence patterns have revealed close parallels between the acrolectal variety of IndEng and StEngEng. We saw that with the exception of the factor group of negation, the HAVE-perfect patterns with independent linguistic variables of semantic context, time adverbial specification, Aktionsart and transitivity in a fashion very similar to that attested in the native standard variety of English. Our task now is to identify those factors (both extra-linguistic and language-internal) that actually trigger the occurrence of the HAVE-perfect in the IndEng acrolect, demonstrating the relative strength of each group and constraint ranking within each independent variable. We will perform a multivariate analysis of factors in order to assess the contribution of each variable to the probability of the present perfect in this variety of English. This procedure will enable us to pronounce final judgements as to in how far the grammar of the acrolectal variety of IndEng and that of StEngEng are related to each other.

9.3. Quantitative analysis

167

9.3.2. Multivariate analysis Table 9.5. Multivariate analysis of the contribution of factors selected as significant to the probability of the perfect in acrolectal IndEng (HCNVE) Perfect Input value l.Sematic context extended-now resultative experiential recent past range 2. Sex male female range S.Aktionsart4 change-of-state verbs stative and dynamic range 4. Negation negation affirmative range 5. Transitivity intransitive transitive range 6. Time adverbial specification time adverbial unspecified range

Factor weight .77

o/o

77%

N 182/236

.70 .69 .43 .06 64

88% 89% 75% 34%

32/36 74/83 66/88 10/29

.74 .35 39

85% 72%

72/84 110/152

[-55] [-47]

78% 76%

65/83 117/153

[-62] [-48]

85% 76%

18/21 164/215

[-52] [-47]

79% 75%

87/110 95/126

[.63] [.47]

87% 75%

28/32 154/204

4 For this analysis to work out it was necessary to merge stative and dynamic verbs on the one hand and verbs of achievement and accomplishment on the other. Here is a linguistic explanation for this slight rearrangement of factors. Stative and dynamic verbs generally describe events as lacking change. By contrast, verbs of achievement and accomplishment always imply a change of state. Similar to this distinction is the distinction made between telic and non-telic verbs because telicity, i.e. goal-orientedness, always involves a change of state (cf.Brintonl988:26).

168

9. Acrolectal variety ofIndian English (HCNVE)

Table 9.5 shows that only one linguistic factor actually conditions the appearance of the HAVE-perfect in acrolectal IndEng: semantic context. With the relative strength of 64, this factor group represents the strongest constraint conditioning the variation of the HAVE-perfect in this variety of English. Recall that in StEngEng semantic context was also selected as the most important factor group underlying the use of the present perfect. This finding provides a clear indication that both varieties are closely related to each other. Notice also that semantic context exhibits a constraint hierarchy in acrolectal IndEng very similar to that of StEngEng. Similar to StEngEng, extended-now and resultative contexts are favoured by the perfect at .70 and .69, whereas experiential and recent past time contexts are disfavoured at .43 and .06 respectively. The hypothesis formulated as a hierarchy of uses of the present perfect has thus been borne out by this data set, revealing that the underlying grammatical mechanism that produces the HAVE-perfect in the acrolectal variety of IndEng is very similar to that of StEngEng. Furthermore, results of the multivariate analysis show that the extralinguistic variable of sex is implicated in the distribution of the present perfect. Contrary to the initial hypothesis that the present perfect should be favoured by female speakers, our results demonstrate unambiguously a strong favouring effect of the HAVE-perfect for male speakers. 9.3.3. Discussion of findings Results of the foregoing multivariate analysis show that similar to StEngEng, neither Aktionsart nor transitivity was selected as statistically significant in the acrolectal variety of IndEng. In the chapter on StEngEng, we interpreted the non-significance of both factor groups as an indication of the fully developed grammatical status of the present perfect in this native variety of English. Along these lines, we interpret the statistical non-significance of both Aktionsart and transitivity as a sign of the complete development of the HAVE-perfect in acrolectal IndEng because - as will become clear in the ensuing chapters - it is precisely these factors that play a significant role in the distribution of the present perfect in less advanced viz. mesolectal varieties of English. The final issue that needs to be addressed here concerns the extra-linguistic factor group of sex, which has been shown to underlie the appearance of the HAVE-perfect in the variety of English under study. We have a clear pattern indicating that the linguistic variable under study is dependent on the sex of the speaker who uses it. Contrary to the intuitive assumption

9.3. Quantitative analysts

169

that females rather than males should be using the prescnptively correct HAVE-perfect more frequently, results of the multivariate analysis show a close association between male speakers and the standard conformant verb form. This finding is in sharp contrast with those on native languages spoken in Western societies (cf Trudgill 1974; Nordberg 1997). Various sociolinguistic studies revealed a close association between women and standard speech or more prestigious language in countries such as England, Netherlands, Sweden, Austria (cf. Romaine 2004: 100-102). At the same time it has been noted in, for instance, Chambers (1995: 140-142) and Romaine (2004: 109) that in many contemporary nonWestern cultures women are further away from the prestige norms of the society. For instance, in many parts of the Middle East and Africa women normally do not have as much education as men, thus lacking access to the norms of written language (cf. Romaine 2004: 109). This contention can hardly be regarded as a satisfactory explanation in our case since all informants on the acrolectal IndEng received a similar amount of formal education in English (i.e. private school and two or three years at the university). Why is it then that acrolectal IndEng male speakers use the perfect more consistently than their female counterparts? It is suggested by way of an explanation that men in India are by and large perceived as a socially dominant and upwardly mobile group. This group is in turn expected to speak English conforming to the norms of Standard (British) English as closely as possible because this variety of English is very prestigious in India. The high degree of conformance to its patterns is, therefore, a good indicator of a high social status and good breeding. By this token, male speakers use the HAVE-perfect to mark their orientation towards prestigious norms, which are directly linked to success and prosperity. In other words, identifying themselves with the Standard (British) English, Indian men may want to claim their socially and economically superior status. It follows then that the social and cultural structure of the Indian society is reflected in the patterns of linguistic variation attested in this data. This explanation is in principle compatible with that of Bakir (1986: 6, cited in Chambers 1995: 140), who studied gender variation in Basrah, Iraq. Bakir (1986) shows that men use standard forms of Arabic much more often than women. The author believes that an explanation to this pattern is to be found in the social structure of the Arab communities, which are segregative in essence. Bakir (1986: 6) says Although there are many types of institutions where men and women meet and work together, the men's society and the women's society are still separate,

170

9. Acrolectal variety ofIndian English (HCNVE)

and women are expected not to trespass on men's grounds by doing men's work or assuming roles and participating in functions that men are expected to perform. More generally, Labov (1982: 79, cited in Chambers 1995: 141) asserts that "where women have not traditionally played a major role in public life, cultural expectations will lead them to react less strongly to the linguistic norms of the dominant culture". Summing up, the IndEng male speakers favouring the standard variant seem to represent a pattern characteristic of "a Nearand South-Asian Sprachbund, a linguistic area that has developed a distinct pattern of sexual differentiation" (Labov 1983: 7, cited in Chambers 1995: 141-142).

9.4. Qualitative analysis: The present perfect in definite past time reference contexts Similar to StEngEng, acrolectal IndEng contains cases featuring the present perfect in definite past time reference contexts. Their occurrence is, however, not very frequent: in fact, the entire corpus attests only 11 tokens. Here are some examples. (92) Acrolectal IndEng (HCNVE: IE03; IE04; IE02) a. Once you have fingers raised onyourcharacter, this girl is like that, I have seen her at two in the mornim and sort of all places, it's a problem. Problem for you to get married. b. Even as a child, I've never paid too much attention to what my grandmother would tell me. In India it's more than... if I think how it was when I have been learning languages in another country, when I was learning Spanish or French or even German. c. Well, because I wanted to go on with something and JNU itself is such a prestigious university, the name speaks for itself, and then when I knew that I have been taken for the French course in JNU, that's something simply awesome. Something unbelievable as such. But what is the possible source of these variants? Is it the result of learner strategies, substrate influence from Hindi or other phenomena similar to those attested in the native variety of English? Let us consider the latter possibility first.

9.4. Qualitative analysis 111 In the chapter on StEngEng, the HAVE-perfect in definite past time reference contexts was analysed as being arguably conditioned by the speaker's attitude towards the entire proposition. More specifically, we saw that the speaker's regarding the situation as related to the moment of utterance could be analysed as a constraint underlying the use of the present perfect in definite past time reference contexts. Two contextual hints turned out to be particularly useful for this pragmatic-discourse analysis: the present perfect introducing a discourse and verbs denoting a change of state, i.e. telic and punctual verbs. If the occurrence of the present perfect in acrolectal IndEng is constrained by the speaker's attitude towards her utterance in analogy to StEngEng, then we should in principle be able to find similar patterns in this set of data. A closer examination of the acrolectal IndEng corpus reveals that this is not the case. First, the HAVE-perfect appears with both telic and atelic verbs. For instance, as the examples above demonstrate, the HAVE-perfect appears with both telic (pay, take) and atelic (learn, see) verbs in our data. These co-occurrence patterns contrast with those attested in StEngEng, where the HAVE-perfect tends to surface with verbs denoting a change of state in definite past time reference contexts. Second, the patterns featuring the perfect opening a discourse and taking a scope over the following discourse, in which another perfect appears (this time modified by a definite past time adverbial) were not attested in the data on acrolectal IndEng. Another possible scenario is that the present perfect in definite past time contexts is a result of processes involving mother-tongue influence, as suggested in Sanyal (2007). The [...] example [I have bought the book yesterday] (use of the present perfect where the simple past is idiomatic) is [...] intriguing. All Indian languages have the simple past. Why then this confusion? The simple Hindi version of the correct form would be main nay woh kitaab kal kharidi. But confusion begins when one mentally adds the unnecessary thalthi at the end (to say main nay who kitaab kal kharidi thi) in the belief that it serves more emphatically to convey completion of the action. One then mentally translates that unnecessary thalthi into the present perfect. That's how the unidiomatic have intrudes where the simple past suffices. Sanyal (2007: 182-183) There is, however, one major problem with the analysis proposed in Sanyal (2007). What is of crucial importance here is that the auxiliary thi is a past tense (not a present tense) marker in Hindi. The sentence main nay who kitaab kal kharidi thi is therefore better translated as / had bought the book

172

9. Acrolectal variety ofIndian English (HCNVE)

yesterday. If we assume that Hindi speakers "mentally translate" the Hindi sentence of the type presented above into English, the outcome of this translation should be a sentence featuring the past perfect rather than the present perfect. (93) Hindi (from Agmhorti 2007: 22) Ham kal baazaar gaye the. we yesterday market gone BE-PAST * 'We had gone to the market yesterday.' This is, however, not the case. We can therefore conclude that mental projection of the mother-tongue forms onto English sentences, i.e. substrate influence, is not the most convincing scenario is this case. It seems that a different process underlies this structural realignment of the past time reference system in Hindi IndEng. Apparently, the Hindi speaker of English reanalyses the English auxiliary have as a definite past time reference marker relying on the Hindi auxiliary thalthi as a model for the re-analysis. The present perfect in definite past time contexts is thus better explained in terms of learner strategies manifested in this case by the process of reanalysis. More importantly, similar to Standard English, Hindi does not allow the present perfect, which is formed with the help of the auxiliary BE (Hindi honaa) in the present tense and a perfect participle (cf Montaut 2004: 102; Agmhotn 2007: 78), in definite past time contexts (cf. Sharma, Devyam 24 November 2008: personal communication).5 Consider, for instance, a sentence (94) Hindi *Vah kal cinema gayaa hai. he yesterday cinema gone is *'He has gone to the cinema yesterday.' Since the present perfect cannot be used in definite past time contexts in Hindi, the analysis of the present perfect structures with simple past time reference in IndEng in terms of language transfer from Hindi seems to be highly implausible. Taking all the above-presented facts into account, we can conclude that acrolectal Hindi speakers of English seem to have re-analysed the meaning 5 Dr. Devyani Sharma, QueenMary University of London.

9.5. Summary

173

of the HAVE-perfect occasionally extending its uses to definite past time contexts in analogy to the model (perfective participle + auxiliary marker) available in their mother tongue.

9.5. Summary The major highlights of this chapter can be summed up as follows. The distributional analysis revealed that the present perfect is used in a wide range of contexts in the acrolectal variety of IndEng. Not only is it robust with both telic and atehc predicates; it is also highly preferred in transitive and intransitive contexts alike. These patterns are an indication of a highly developed grammatical status of the HAVE-perfect in the IndEng acrolect. Moreover, the close association between the present perfect and temporally unspecified contexts have been confirmed for this variety of English in analogy to StEngEng. The multivariate analysis further demonstrated that similar to StEngEng, semantic contexts are highly implicated in the occurrence of the present perfect in acrolectal IndEng. More importantly, constraints within this factor group are arranged in the order very similar to that attested in StEngEng. Hence, there is a major dividing line between resultative and extended-now contexts on the one hand and experiential and recent past time contexts on the other in both StEngEng and acrolectal IndEng. The underlying grammatical mechanism producing the HAVE-perfect has thus been argued to be identical in both varieties. Qualitative analysis showed that similar to StEngEng, acrolectal IndEng exhibits the occurrence of the present perfect in definite past time reference contexts. In contrast to the native variety of English, in which the appearance of the HAVE-perfect in contexts normally rendered by the simple past tense was argued to be pragmatically constrained by the speaker's attitude towards the entire proposition, such structures were shown to be better analysed in terms of re-analysis strategies in the acrolectal variety of IndEng. Finally, the statistically significant pattern of male speakers using the perfect more consistently than female speakers has been argued to mirror the social and cultural structure of the Indian society.

10. Upper-mesolectal variety of Indian English (ICE)

10.1. Data This analysis is based on the data drawn from 100 texts of the Indian component of the International Corpus of English (ICE), which is regarded as the upper-mesolectal variety ofIndEng in this study.1 The subcorpus contains a total of 200,000 words. The focus is on the informal spoken register represented by direct conversations (S1A-001 to S1A-090) and telephone calls (S1A-091 to S1A-100). Most of the texts are drawn from the trained ELT teachers, who received varying degrees of instruction in English. The corpus ICE-India studied here comprises both private conversations and discussions, which are frequently mixed. Although a strict distinction between the two categories was not drawn by the authors of the corpus, it was noted in, for instance, Shastn (2002: 2) that many genuinely private conversations constitute a bulk of the private-spoken data of ICE-India. The ICE-corpus studied here is therefore by and large compatible with the standard vernacular represented by the LLC. Of particular importance for the description of this data set is the fact that IndEng has generally been described as a variety marked for its desire for greater approximation to Standard [British] English (Snvastava and Sharma 1991: 192). At the same time it was noticed in, for instance, Sharma (2001) that many Indians ceased to aspire to a British acrolectal standard in their English. As a result, the range in bilingual competence increased (Sharma 2001: 345). Both tendencies are crucial for understanding the type of variation attested in this set of data, which presumably draws on both standard-conformant and standard deviant speakers of English. Moreover, this corpus contains speakers of the major Indian languages included in the Indian constitution. Additionally, some minor languages such as Konkam and Tulu were included in the sample (cf Shastn 2002: 2). 1 Several researchers (for instance, Peter Siemund and Michaela Hilbert June 2008: personal communication) have acknowledged that IndEng (ICE) is a variety well advanced on the acrolectaVbasilectal continuum. However, this variety differs in many important ways from the truly acrolectal variety of IndEng represented in this study by a small-scale corpus (HCNVE). It is for this reason that the Indian component of ICE is referred to as the upper-mesolectal variety in this study.

10.2. Quantitative analysts

175

The main objectives of this chapter are as follows. First, it aims at portraying the overall distribution of data, while trying to assess the possible origin of non-standard variants. Second, this chapter provides a comprehensive analysis of the HAVE-perfect and the mechanism underlying its occurrence in the upper-mesolectal variety of IndEng. StEngEng provides a descriptive point of comparison which allows us to reveal standardconformant patterns in the data. In addition, patterns attested in this body of material are contrasted with those obtained for acrolectal IndEng and mesolectal varieties of English including learner Englishes, i.e. RusEng and GerEng. The aim of this procedure is to show what patterns of variability account for the close affinity obtaining between upper-mesolectal IndEng and the standard grammar of English, and what patterns make this variety similar to other non-native Englishes. The final section depicts the results of the qualitative analysis of data, which deals with those uses of the present perfect that are not in accordance with the descriptions of the present perfect provided in traditional grammars.

10.2.

Quantitative analysis

10.2.1.

Distributional analysis

10.2.1.1. Variation in present perfect contexts Let me first of all consider all the verb forms that have been found in present perfect contexts in upper-mesolectal IndEng. Table 10.1 represents the overall distribution of variants in present perfect contexts in the variety under study. Table 10.1. Overall distribution ofvariants in upper-mesolectal IndEng %

HAVE-perfect preterite present BE-perfect past perfect bare verb stem lone past participle lone present participle three verb cluster ToM^

715 471 60 16 40 9 7 6 1 1325

53% 35% 4% 1% 3% 0% 0% 0% 0%

176

10. Upper-mesolectal variety ofIndian English (ICE)

First, variability between the present perfect and preterite is robust. Second, two other forms surfacing in the data rather frequently are the present tense and the past perfect. Furthermore, forms such as BE-perfects, bare verb stems, lone past participles, lone present participles, three verb clusters are also attested in the data. As is clear, this variety of IndEng exhibits a system of present perfect contexts in which there is an alternation among many competing forms. The overall rate of forms attested in this variety is in contrast with that observed in StEngEng and acrolectal IndEng. This finding is somewhat surprising given the fact that we deal with the upper-mesolectal variety of English for which a considerable number of similarities with the Standard have been attested, as we shall see in a moment. The fairly robust variation of surface forms can partially be explained by the sample design of the ICE-India, whose creators aimed at compiling a corpus comprising speakers with various regional, ethnical and - to some extent - educational backgrounds. However, this type of variation may also be indicative of the gradual restructuring processes within the present perfect system that are under way in the IndEng upper-mesolect, a variety representing a step by step transition from the mesolect to the acrolect. The second issue that needs to be addressed here is the origin of nonstandard forms. Are they a result of a substrate/superstrate influence or have they emerged due to some general mechanisms (or strategies) involved in L2 acquisition? Obviously, these non-standard verb forms cannot be equated with the English present perfect and therefore cannot be attributed to an English-like grammar. But do they have essentially the same functions as those attested in other non-standard varieties of English or can they perform other functions as well? Let us start the analysis by considering the present tense first.2 Present tense Similar to other varieties of non-native English (for instance, mesolectal varieties of RusEng and GerEng represented by HCNVE), the present tense is attested in extended-now contexts in the upper-mesolectal IndEng, as in (95) Indian English (ICE: S la-052; S la-040) a. They 've advantage because they are practising since May. b. He's explaining the difference between charming graceful and beautiful for half an hour. 2 Neither preterites nor BE-perfects or bare verb stems will be considered in this section as they are analysed elsewhere (see the chapter on basilectal IndEng).

10.2. Quantitative analysis

177

Furthermore, the use of the copula BE in the present tense is occasionally attested in the corpus which provides a nice parallel to the type of variation attested in Indian South African English (InSAfEng) and South African English (SAfEng) more generally, where the use of the present BE for ext e n d e d ™ contexts have been reported (cf Mesthne 2004b: 1134). (96) Indian English (ICE: Sla-054) Since morning there is no hot water. In contrast to InSAfEng and SAfEng, however, the present tense has also been attested in resultative and recent past time contexts in IndEng, as in (97) Indian English (ICE: Sla-018; Sla-26; Sla-008; Sla-030) a. Now this Khadaklat exchange is recently connected. '...has been recently connected.' b. Smce the autormzatton before a year the staff is reduced by ten people. '...has been reduced...' c. Recently she is appointed as a tutor. '...has been appointed...' d. Recently we do have one snake project. '...have had...' e. No initially U didn't affect me butlthmknow my resistance is much less. '...has decreased' Figure 10.1 illustrates the distribution of the present tense across semantic contexts in ICE-IndEng. I present tense 40%

20%

0% extended-now

1 resultative

recent past

experiential

Figure 10.1. Distribution of the present tense across present perfect contexts in uppermesolectal IndEng

178

10. Upper-mesolectal variety ofIndian English (ICE)

As is clear, the present tense has expanded its functions to contexts other than continuative contexts in IndEng, although it is the most frequent variant in extended-now contexts. The present tense in extended-now contexts has been analysed as a substrate influence in some accounts on, for instance, Irish English (cf Filppula 1999). Abbi (November 2007: personal communication) proposes analysis of these forms in much the same vein for IndEng. Her contention is that since most wide-spread languages of India (for instance, Hindi, Bengali, Tamil and Malayalam) employ the present tense in extended-now contexts Indian speakers simply map the mother-tongue form, i.e. the present tense, on extended-now contexts while speaking English. If this hypothesis were right, we should have expected the occurrence of the present tense in extended-now contexts only. The analysis presented above reveals, however, that an explanation in terms of substrate influence alone is somewhat problematic with this set of data because the present tense is fairly robust in other present perfect contexts and thus does not replicate the pattern attested in respective substrate languages (i.e. present tense in extended-now contexts). Hence, distribution of the present tense across present perfect contexts cannot be entirely explained in terms of direct form-meaning mappings from mother tongue onto the target language. It seems that substrate influence effects closely interact with other more general (or universal) phenomena (such as learners replacing the HAVEperfect with the less complex present tense form) in this variety of English. The analysis of origin of the present tense thus yields both interpretations depending on the context in which the verb form appears. The appearance of the present tense in extended-now contexts is amenable to an explanation in terms of substrate influence, whereas its occurrence in resultative and recent past contexts is better understood in terms of substrate-independent learner strategies. Past perfect Another form surfacing rather frequently in present perfect contexts in IndEng is the past perfect, as in (98) Indian English (ICE: Sla-041; Sla-052) a. Recently Raj had called me up. (context of recent past) b. Even I had heard about it but even I've not seen this girl but many of my friends you know heard, (experiential context)

10.2. Quantitative analysts

179

This finding is consonant with that of Sharma (2001) who discovered that IndEng differs from native usage by associating had + V-ed with present perfect meaning (Sharma 2001: 343). What is again at issue here is the origin of the past perfect in present perfect contexts in IndEng. To start with, the past perfect has been attested in present perfect contexts in some native varieties of English (for instance, Samana English) (cf Tagliamonte 1997: 47). Moreover, the past perfect apparently used to surface in present perfect contexts at earlier stages of English (cf. Mustanoja 1960: 497). In the light of this evidence one would assume that the superstate influence scenario cannot be entirely excluded in this case. However, in native varieties the past perfect in present perfect contexts is generally extremely rare. Thus, in her corpus on Samana English consisting of more than 7,000 tokens Tagliamonte (1997) reports only 4 past perfect tokens (