Structural Nativization in Indian English Lexicogrammar
Studies in Corpus Linguistics (SCL) SCL focuses on the use of...
70 downloads
1265 Views
1MB Size
Report
This content was uploaded by our users and we assume good faith they have the permission to share this book. If you own the copyright to this book and it is wrongfully on our website, we offer a simple DMCA procedure to remove your content from our site. Start by pressing the button below!
Report copyright / DMCA form
Structural Nativization in Indian English Lexicogrammar
Studies in Corpus Linguistics (SCL) SCL focuses on the use of corpora throughout language study, the development of a quantitative approach to linguistics, the design and use of new tools for processing language texts, and the theoretical implications of a data-rich discipline. For an overview of all books published in this series, please see http://benjamins.com/catalog/scl
General Editor
Consulting Editor
Elena Tognini-Bonelli
Wolfgang Teubert
The Tuscan Word Centre/ The University of Siena
University of Birmingham
Advisory Board Michael Barlow
Graeme Kennedy
Douglas Biber
Geoffrey N. Leech
Marina Bondi
Michaela Mahlberg
Christopher S. Butler
Anna Mauranen
Sylviane Granger
Ute Römer
M.A.K. Halliday
Jan Svartvik
Yang Huizhong
John M. Swales
Susan Hunston
Martin Warren
University of Auckland Northern Arizona University University of Modena and Reggio Emilia University of Wales, Swansea University of Louvain University of Sydney Jiao Tong University, Shanghai University of Birmingham
Victoria University of Wellington University of Lancaster University of Nottingham University of Helsinki University of Michigan University of Lund University of Michigan The Hong Kong Polytechnic University
Volume 46 Structural Nativization in Indian English Lexicogrammar by Marco Schilk
Structural Nativization in Indian English Lexicogrammar Marco Schilk Justus Liebig University, Giessen
John Benjamins Publishing Company Amsterdamâ•›/â•›Philadelphia
8
TM
The paper used in this publication meets the minimum requirements of American National Standard for Information Sciences – Permanence of Paper for Printed Library Materials, ansi z39.48-1984.
Cover design: Françoise Berserik Cover illustration from original painting Random Order by Lorenzo Pezzatini, Florence, 1996.
Library of Congress Cataloging-in-Publication Data Schilk, Marco. Structural nativization in Indian English lexicogrammar / Marco Schilk. p. cm. (Studies in Corpus Linguistics, issn 1388-0373 ; v. 46) Includes bibliographical references and index. 1. English language--India--Grammar. 2. English language--India--Usage. 3. Lexicology 4. Grammaticality (Linguistics) 5. Linguistic analysis (Linguistics) I. Title. PE3502.I61.S35â•…â•… 2011 427’.954--dc23 2011018552 isbn 978 90 272 0351 9 (Hb ; alk. paper) isbn 978 90 272 8508 9 (Eb)
© 2011 – John Benjamins B.V. No part of this book may be reproduced in any form, by print, photoprint, microfilm, or any other means, without written permission from the publisher. John Benjamins Publishing Co. · P.O. Box 36224 · 1020 me Amsterdam · The Netherlands John Benjamins North America · P.O. Box 27519 · Philadelphia pa 19118-0519 · usa
To Mara Moësha
Table of contents
Tables and figures Acknowledgements chapter 1 Introduction and overview 1.1 Introductionâ•… 1 1.2 Structural nativization in Indian English lexicogrammar: Previewâ•… 2
x xiii 1
chapter 2 5 Aspects of structural nativization 2.1 From English in India to Indian Englishâ•… 5 2.2 Models of World Englishes: The situation of Indian English todayâ•… 7 2.2.1 Static and dynamic models of World Englishesâ•… 7 2.2.2 Indian English in the dynamic evolutionary model: Aspects of ongoing nativization and endonormative stabilizationâ•… 11 2.3 Different influences on structural nativizationâ•… 15 chapter 3 17 Aspects of lexicogrammar: Collocation and verb-complementation 3.1 Introduction – The interdependence of lexis and grammarâ•… 17 3.2 Collocationâ•… 19 3.2.1 Introductionâ•… 19 3.2.2 Quantitative approaches to collocationsâ•… 19 3.2.3 Phraseological approaches to collocationsâ•… 24 3.2.4 Collocations in Indian Englishâ•… 26 3.2.5 Collocation in the lexis-grammar continuumâ•… 27 3.3 Verb-complementationâ•… 27 3.3.1 Introductionâ•… 27 3.3.2 Verb-complementation in descriptive grammar (Quirk et al. 1985) and corpus-based descriptive grammar (Biber et al. 1999)â•… 28 3.3.3 Verb-complementation in cognitive grammar and construction grammarâ•… 32 3.3.4 Verb-complementation in Indian Englishâ•… 36
viii Structural Nativization in Indian English Lexicogrammar
chapter 4 Methodology 4.1 Introduction: A corpus-based description of structural nativization in Indian English lexicogrammarâ•… 41 4.2 Corpora and toolsâ•… 42 4.2.1 The International Corpus of English – The British and Indian components (ICE-GB and ICE-India)â•… 42 4.2.2 The Times of India corpusâ•… 44 4.2.3 Corpus analysis software: WordSmith Tools V4.xâ•… 48 4.3 Verb-complementational profilesâ•… 49 4.4 Collocational profilesâ•… 54 4.5 Verbs under scrutiny: GIVE, SEND and OFFER – A quantitative overviewâ•… 57 4.6 Summaryâ•… 62
41
chapter 5 63 GIVE 5.1 Introduction: GIVE as a prototypical ditransitive verbâ•… 63 5.2 Distribution of complementation patterns in different text categoriesâ•… 66 5.3 Verb-complementational patterns of GIVE in spoken Indian and British Englishâ•… 69 5.4 Verb-complementational patterns of GIVE in written Indian and British Englishâ•… 76 5.5 Collocational profiles of GIVE in Indian and British Englishâ•… 84 5.6 GIVE: A brief summaryâ•… 92 chapter 6 95 SEND 6.1 Introduction: SEND as a less-prototypical ditransitive verbâ•… 95 6.2 An overview of the verb-complementation patterns of SEND in ICE-GB and ICE-Indiaâ•… 97 6.3 Distribution of complementation patterns in different text categoriesâ•… 100 6.4 Verb-complementational patterns of SEND in spoken Indian and British Englishâ•… 102 6.5 Verb-complementational patterns of SEND in written Indian and British Englishâ•… 104 6.6 SEND in the International Corpus of English: A synopsisâ•… 109 6.7 Verb-complementational patterns of SEND in Indian and British newspaper Englishâ•… 110
Table of contents
6.8 Collocational profiles of SEND in Indian and British newspaper Englishâ•… 117 6.9 SEND in Indian and British newspaper English: A brief summaryâ•… 128 chapter 7 OFFER 7.1 Introduction: OFFER as a low-frequency ditransitive verbâ•… 129 7.2 Verb-complementational patterns of OFFER in spoken and written Indian and British Englishâ•… 131 7.3 OFFER in British and Indian newspaper Englishâ•… 132 7.4 Collocational profiles of OFFER in Indian and British newspaper Englishâ•… 133 7.5 OFFER: A brief summaryâ•… 144 7.6 Analysis of GIVE, SEND and OFFER – Concluding remarksâ•… 146 chapter 8 Evaluation and discussion 8.1 Introductionâ•… 147 8.2 Evaluation of the analysisâ•… 148 8.2.1 Corpora and softwareâ•… 148 8.2.2 Verb-complementational patternsâ•… 152 8.2.3 Collocational profilesâ•… 155 8.3 Discussionâ•… 158 8.3.1 Introductionâ•… 158 8.3.2 Od-collocate analysis and collostructional analysisâ•… 161 8.3.3 Lexicogrammatical variation in varieties of Englishâ•… 161 8.4 Towards a model of lexicogrammatical nativizationâ•… 163 chapter 9 Conclusion and prospects for future research 9.1 Conclusionâ•… 171 9.2 Prospects for future researchâ•… 173 References Index
129
147
171
175 181
ix
Tables and figures
Table 2-1. The evolution of New Englishes as a cyclic process (Schneider€2003:â•›255) Table 2-2. Classifying Indian English today: Characteristics of stage 3: Nativization and stage 4: Endonormative stabilization Table 3-1. The lexis-grammar interface Figure 3-1. Expected and observed co-occurrences (Evert 2005:â•›76) Figure 3-2. Comparison of asymptotic hypothesis tests with exact testing (Evert 2005:â•›111) Table 3-2. Sub-categories of word-like combinations (cf. Cowie 1998:â•›7) Table 3-3. Degrees of schematicity (cf. Langacker 1999b:â•›122) Figure 3-3. Relations among constructions (Goldberg 1995:â•›91) Table 4-1. The International Corpus of English (cf. Nelson et al. 2002:â•›307–308) Table 4-2. Approaches to the use of the Internet as a linguistic corpus (cf. Thelwall 2005:â•›529) Table 4-3. Collostructional attraction (+) and repulsion (–) in ICE-GB (cf. Mukherjee 2009:â•›133) Table 4-4. Complementation patterns of ditransitive verbs (cf. Mukherjee & Hoffmann 2006:â•›172) Table 4-5. Frequency of the verbs under scrutiny as counted by WST 4.x / BNC-online Table 4-6. Distribution of GIVE, SEND and OFFER in the spoken and written sections of ICEâ•‚India and ICEâ•‚GB Table 4-7. Comparison of GIVE, SEND and OFFER across modes and varieties Table 5-1. Complementation of GIVE in ICE-India and ICE-GB (cf. Mukherjee & Hoffmann 2006:â•›172) Table 5-2. Complementation of GIVE in ICE-India and ICE-GB – focus on central patterns Table 5-3. Distributional differences of GIVE: A chi-square test Table 5-4. Distributional differences of GIVE according to text types Table 5-5. Complementation of GIVE in ICE-India and ICE-GB – spoken data
10 14 18 20 23 25 33 36 43 45 51 53 59 60 61 64 65 66 68 69
Tables and figures
Table 5-6. Pronominal Oi in spoken GIVE type-I patterns Table 5-7. Complementation of GIVE in ICE-India and ICE-GB – written data Table 5-8. Pronominal Oi in written GIVE type-I patterns Figure 5-1. Type-II pattern in ICE-India: W1A (student writing) Figure 5-2. Type-II pattern in ICE-GB: W1A (student writing) Table 5-9. Od-collocates of GIVE in ICE-India Table 5-10. Od-collocates of GIVE in ICE-GB Table 5-11. Od-collocates of GIVE sorted by variety Table 5-12. Complementation patterns of GIVE (only significant collocations) Table 5-13. Frequent complementation of GIVE in correlation to Od-collocates in ICE-India Table 5-14. GIVE in correlation to Od-collocates in ICE-India (no specific trend) Table 5-15. GIVE in correlation to Od-collocates in ICE-GB Table 5-16. Complementation patterns of GIVE (only shared Od-collocates) Table 6-1. Complementation of SEND in ICE-India and ICE-GB (cf. Mukherjee & Hoffmann 2006:â•›173) Table 6-2. Complementation of SEND – focus on central patterns Table 6-3. Distributional differences of SEND: A chi-square test Table 6-4. Distributional differences of SEND according to text types Table 6-5. Complementation of SEND in ToI1.75 and BNCNews.1.77 Table 6-6. Distribution of SEND in ToI and BNC – focus on central patterns Table 6-7. SEND as a particle verb in ToI and BNC Table 6-8. Od-collocates of SEND in the Times of India (9 million word subcorpus) Table 6-9. Od-collocates of SEND in the BNC newspaper section Table 6-10. Od-collocates of SEND sorted by variety Table 6-11. Verb-complementational distribution of SEND + Od-collocates Table 6-12. SEND in correlation to Od-collocates in Times of India9m (type-II dominant) Table 6-13. SEND in correlation to Od-collocates in Times of India9m (type-III dominant) Table 6-14. SEND in correlation to Od-collocates in Times of India9m (no pattern bias) Table 6-15. SEND in correlation to Od-collocates (BNCNews)
72 77 79 82 82 85 86 87 87 88 89 90 92 98 99 99 101 112 113 116 118 118 120 121 122 123 124 125
xi
xii Structural Nativization in Indian English Lexicogrammar
Table 6-16. Verb-complementational profile of SEND + Od-collocates (shared) Table 7-1. Verb-complementational distribution of OFFER in ICE-India and ICE-GB Table 7-2. Distribution of OFFER in ToI1.75 and BNCNews.1.77 – focus on central patterns Table 7-3. Od-collocates of OFFER in the Times of India (9 million word subcorpus) Table 7-4. Od-collocates of OFFER in the BNCNews Table 7-5. Od-collocates of OFFER sorted by variety Table 7-6. Verb-complementational profile of OFFER + Od-collocates Table 7-7. OFFER in correlation to Od-collocates (type-III dominant) Table 7-8. OFFER in correlation to Od-collocates (type-II dominant) Table 7-9. OFFER in correlation to Od-collocates (type-I dominant) Table 7-10. OFFER in correlation to Od-collocates (no pattern bias) Table 7-11. OFFER in correlation to Od-collocates (type-III dominant) Table 7-12. OFFER in correlation to Od-collocates (type-I dominant) Table 7-13. OFFER in correlation to Od-collocates (no pattern bias) Table 7-14. Verb-complementational profile of OFFER + Od-collocates (shared) Table 8-1. Degrees of schematicity (cf. Langacker 1999b:â•›122) Figure 8-1. Lexicogrammatical nativization in cause-receive processes
127 131 133 134 135 136 136 137 139 139 140 141 143 143 144 164 166
Acknowledgements
I wish to express my gratitude to many people who supported me with their help and comments while I was writing this book. An enormous debt is owed to my academic teacher and advisor, Joybrato Mukherjee, for his advice throughout the project, his enthusiasm, wisdom and patience. Without his constant feedback and support, this book would not have been possible. I am grateful to the Justus Liebig University, Giessen, the German Research Foundation (DFG) and the International Graduate Center for the Study of Culture (GCSC) for providing me with project funding and travel grants. Further, I would like to thank the conference delegates at the ICAME conferences in Helsinki (2006), Stratford-upon-Avon (2007), Ascona (2008), Lancaster (2009) and Giessen (2010) as well as the ISLE conference in Freiburg (2008) for their stimulating questions and helpful feedback to the presentation of various stages of the project. Special thanks go to Sebastian Hoffmann for introducing me to the LWPÂ�module in Perl and providing me with the newspaper section of the BNC, to Stefan Th. Gries for his statistical insights and the time he devoted to the identification of statistical bugs in some of the software I used and to Rosemary Bock who proofread the original manuscript. Thanks are also due to my collegues at the linguistics department in Giessen. Sandra Götz, who assisted me with coding when she was still a student and became a great office-mate and friend later; Christiane Brand for all the time she devoted to linguistic discussion and personal conversation over many cups of coffee and Tobias Bernaisch for many fruitful talks about verb-complementation. Finally I would like to thank my family: my parents Ulrike and Bernd-Rüdiger, for their unfailing love and support, my wife Manuela for her love and patience as a “PhD-widow” and for adding a non-linguists perspective. Last but not least I wish to thank my beloved daughter Mara Moësha for reminding me that there is more to life than work. To her this book is dedicated.
chapter 1
Introduction and overview
1.1
Introduction
English has global currency, be it as a native tongue, a foreign language or as a lingua franca. Its spread was originally in part due to a long history of British political and economic dominance in every continent of the world. Apart from communities in which English is the dominant or single language, English functions as an additional or link-language in many countries that had colonial ties to Great Britain in the past. In those countries English is not the native language of the majority of its speakers, but plays an important role in many international and intranational functions (cf. Görlach 1991:â•›13). One of these non-native varieties, Indian English, stands out owing to its sheer number of speakers. Although it is very difficult to assess the exact number, some estimates are as high as 200 million (cf. Crystal 2004:â•›101), thus, from a purely numerical standpoint, ranging amongst or even outnumbering native-Â�varieties of English. There have been many studies on nativization in Indian English since Braj B. Kachru published his fundamental collection of papers The Indianization of English (cf. Kachru 1983). However, early studies of Indian English often described the variety based on a very limited set of linguistic data, and were often of an anecdotal nature. Over the last two decades since the compilation of the Kolhapur Corpus of Indian English and later the Indian component of the International Corpus of English, this has gradually changed, so that an ever increasing body of corpuslinguistic descriptions of Indian English has now been developing. However, these descriptions often concentrate on the lexical description of the variety, since structural innovations in new varieties of English were traditionally believed to be relatively “unobtrusive” (Quirk et al. 1985:â•›17). Although there have been recent studies that aim at a more detailed description of lexicogrammatical features, such as Schneider (2004) and Olavarría de Ersson & Shaw (2003), these studies were mainly of a pilot character. Other recent research, such as Sedlatschek (2009), provides a good description of many features of Indian English, including lexicogrammatical features, but does not focus on them in detail. The aim of the present study is to provide insights into structural nativization
˘
Structural Nativization in Indian English Lexicogrammar
on the level of collocation and verb-complementation in Indian English with a specific focus on ditransitive verbs.
1.2
Structural nativization in Indian English lexicogrammar: Preview
The present work is organized into nine chapters. The second chapter lays the foundations by providing a general description of Indian English. Firstly, the historical development of the variety from its early colonial stages until the present day is summarized in order to show how the present linguistic situation on the subcontinent developed. In the second part of Chapter 2, different models that are currently used for the description of World Englishes will be discussed. This discussion will treat classical static models, such as Kachru’s (1985) highly influential three-circle model, as well as more dynamic models such as Schneider’s (2003) cyclic evolutional model; a discussion of influential factors of structural nativization concludes the chapter. Chapter 3 deals with theoretical preliminaries concerning the main linguistic areas in the focus of the present analysis, i.e. collocation and verb-complementation. These two lexicogrammatical areas represent different points within the lexis-grammar continuum. Collocation, situated more towards the lexical end of this continuum, is described from a quantitative and a phraseological viewpoint. Additionally, earlier studies on collocation in Indian English will be briefly discussed. The section on verb-complementation describes different influential models for the description of ditransitive verbs. The introduction in Chapter 3 thus lays the foundation for the analysis of the ditranstive verbs GIVE, SEND and OFFER that is carried out in Chapters 5, 6 and 7, respectively. Chapter 4 introduces the descriptive apparatus and the corpus data of the study. Since the study is based on two standard corpora of British and Indian English, as well as on a web-derived newspaper corpus, the design of the standard corpora and the retrieval process and structure of the web-derived corpus will be described in detail. The second part of Chapter 4 introduces the concept of collocational profiles and verb-complementational profiles that form the basis of the descriptive analysis of the current study. The third part of Chapter 4 provides a quantitative overview of the usage of the verbs that are analysed in detail within the next three chapters. In Chapters 5, 6 and 7 the findings of the analysis are in focus. For each of the three chosen ditransitive verbs, GIVE, SEND and OFFER, collocational and verb-complementational profiles are derived from the different corpora. In turn, similarities and differences between the varieties are discussed in order to show
Chapter 1.╇ Introduction and overview
the extent to which structural nativization is taking place in Indian English with respect to the lexis-grammar interface. Chapter 8 evaluates the methodological framework introduced in Chapter 4 in the light of the findings discussed in the previous chapters. After this methodological evaluation a model for the description of lexicogrammatical nativization with a focus on the linguistic encoding of cause-receive processes is proposed. A brief conclusion of the present analysis and an outlook for further research into structural nativization in Indian English is given in Chapter 9.
˘
chapter 2
Aspects of structural nativization
2.1
From English in India to Indian English
India has been called an “expanding network of languages” (Annamalai 2004:â•›151), a network of a large variety of languages from four different language families (with Indo-Aryan and Dravidian languages accounting for more than 95% of the languages spoken). English entered this network in the 17th century and although it was not the first European language to be introduced into the subcontinent as a contact language, it became the most influential of these in the colonial history of India. Mehrotra (1998) divides the colonial history of the English language in India into two different phases; an early phase from 1579 to 1834 and the phase of the British Raj from 1835 to 1947. The initial period starts with the arrival of the missionary Thomas Stephens, who is said to have been the first Englishman in India (Mehrotra 1998:â•›3). During this phase, no clear-cut language policy was carried out by the colonial power. The final upshot of this unsettled situation was the Orientalist-Anglicist Controversy that ranged from the 1780s to the 1840s, where the Orientalist position aimed at governing India by use of the “languages that had been in use before, such as Persian” (Schiffmann 2005:â•›2112), while the Anglicist position aimed at systematically introducing the English language into the colony. Schiffmann (2005) summarises this situation as follows: Ironically during the first 250 years of British colonialism in India, the East India Company ran things and their policy was to use whatever languages had been in use before […], although English gained some ground. This did not work everywhere and some local elites resisted it. […] [After 1850] the rule of the Company came to an end and a colonial office was established in London that undertook to train people professionally to govern India. But even before this, English-educated elites challenged the “Orientalist” position, and wished to use more English, (Schiffmann 2005:â•›2112) and after 1830, this became the policy.
This policy culminated in the Minute on Indian Education (cf. Macauley 1835), which institutionalized English language teaching in India on a large scale and . For a thorough discussion of the Orientalist-Anglicist Controversy see Rahman (1996).
˘
Structural Nativization in Indian English Lexicogrammar
which marks the end of the first of Mehrotra’s (1998) phases. This Minute can be seen as the linguistic beginning of the phase of the British Raj, during which the colonial power actively promoted the English language by “making English the medium of all education in the schools and universities in India” (Mehrotra 1998:â•›4). This systematic introduction of the English language into the multilingual setting of the subcontinent, which subsequently allowed for the use of English as a medium of administration and law throughout India, “changed the position of English in the multilingual network in India, endowing it with a dominant position together in the three domains of commerce, administration and politics” (Annamalai 2004:â•›153). This change of position had several effects on Indian society as well as on the English language spoken on the subcontinent itself. Within Indian society, a good command of English became an asset for Indians since it opened the door to positions within the colonial administration and the field of trade. This aspired status of the English language was actively initiated by the British administration in Harding’s proclamation (1844) that “made recruitment for public offices in India available only through English schools – this was of course focused on elites only” (Schiffmann 2005:â•›2106). Since it was not the native language of any group of Indians, it “was not perceived to be under the control of any native elite, based on caste and religion and blocking access to it (as Sanskrit was under the control of the Brahmins and Persian under the Muslim elite)” (Annamalai 2004:â•›153), an important point in a highly heterogeneous India. Access to the English language may now no longer have been blocked by traditional Indian barriers such as caste or religion, but it was controlled by the non-native group of British citizens, who, by founding English-based schools and universities, gradually estaÂ�blished English “as the official and academic language of India” (Kachru 1983:â•›23). However, at least during the early stages, only a very limited number of native Indians achieved bilingualism in more than a very limited set of registers of English and a native Indian language, so that Kachru has suggested viewing bilingualism in India as a cline: […] This cline has three ‘measuring points’, i.e. the zero point, the central point, and the ambilingual point. An English speaking bilingual who ranks just about the zero point is considered a minimal bilingual. […] a bilingual who has adequate competence in one or more registers of IE […], may rank around the central point. A standard (or educated) IE bilingual may be defined as one who is intelligible not only to other Indians in the different parts of the subcontinent, but ideally speaking, to the educated native speakers of English. (Kachru 1983:â•›129, his italics)
Chapter 2.╇ Aspects of structural nativization
As implied in this quotation, Kachru does not view complete ambilingualism as a necessary goal of a bilingual, but sees ambilingualism as a “rare if not impossible phenomenon” (Kachru 1983:â•›129). At this point, we need to take a look at the second turning point within the history of English in India, the situation after India (and Pakistan) became independent in 1947. Whereas the introduction of English in 1835 and the enforced strengthening of the English language in the multilingual setting of India were initiated by the colonial power, from 1947 onwards the decisions about language planning lay with the Indian government and the Indian population. After the independence of India, the initial plan to substitute English with Hindi after a transitional period (1950–1965) met with enormous resistance, especially from the non-Hindi speaking parts of the population, leading to violent unrest and language riots (cf. Schiffmann 2005). The multilingual situation in India thus demanded a “neutral” link language that did not favour any one ethnic group in India. As a result, the National Policy Resolution of 1968 settled on a threeÂ�language formula that was reiterated in the National Policy on Education in 1986 (cf. Schiffmann 2005). This formula gives the states of India limited autonomy on the choice of the language of education and basically promotes Hindi, English and a local Indian language. However, the actual realization differs in the various states of India, so that some states pursue a two-language formula (e.g. Tamil Nadu), while others have integrated four official languages. Although these formulae regard English only as an additional language, knowledge of English has never lost its importance as an asset for socio-economic advancement in India, a situation that is the topic of an ongoing debate about the status of English in India (cf. e.g. Tully 1997 and Matthew 1997). As the multilinguistic status of the subcontinent is likely to prevail, there is a need for a language of wider communication. Promoting the dominance of any one specific Indian language is undesirable for social reasons and introducing a completely new non-Indian language would be extremely uneconomic. Thus the English language in India may well retain and consolidate its status.
2.2
Models of World Englishes: The situation of Indian English today
2.2.1
Static and dynamic models of World Englishes
As the function and the status of the English language (or variety of English) in a speech community is central to the description of World Englishes, there is a tradition in research into varieties of classifying different Englishes according to different models, resulting in a rather wide range of descriptive terminology.
˘
˘
Structural Nativization in Indian English Lexicogrammar
A central issue within all of these models is the question of target norms for the different varieties of English and their relation to other varieties, especially the so-called input-varieties, i.e. native-language varieties that had at least historical influence on the emergent or new variety. Within the range of models, there are static models that describe the function of the English used in the different speech communities. Examples of this traditional modelling are the classification of English as a Native Language (ENL), English as a Second Language (ESL) and English as a Foreign Language (EFL) (cf. Görlach 1991). This is partially mirrored in Kachru’s community-based “ThreeCircle-Model”. Kachru points out that the majority of English speakers in the world are not native speakers of English (i.e. not members of the “inner circle”), but use the language either as a second language (in the “outer circle”) or as a foreign language (in the “expanding circle”) (cf. Kachru 1985). Apart from these widespread static models for the categorization of specific varieties of English, there is a growing tendency to describe the status of varieties with respect to their development. The underlying assumption of such approaches is that different varieties are subject to a comparable developmental process. An early example of such a model is Moag (1982), which describes the development of post-colonial varieties of English as a dynamic diachronic process, during which English starts out as a foreign language in the new region, then becomes increasingly nativized and institutionalized, developing into a second language and, at the end of the cyclic process, is (in some cases) replaced by another national language, shifting back to foreign language status. This life cycle is influenced by different factors or processes: Four processes are posited as significant constituents of the life cycle: transportation, indigenization, expansion in use and function and institutionalization. A fifth, restriction of use and function, does not apply in all cases. It is not possible to regard these as stages in the strict sense since they are not fully consecutive. Each process begins in the order stated, but once under way it overlaps with suc(Moag 1982:â•›270–271) ceeding processes.
A newer and slightly different dynamic model is proposed by (Schneider 2003); this model is comparable to Moag’s (1982) model in that it describes the development of the respective varieties during different stages and allows for positioning present-day varieties within its framework. It is, however, more finegrained since . At the time of writing of the present work, Schneider republished the proposed (2003) model in Schneider (2007). Although this new model is slightly expanded and the repective stages are discussed in more detail, this republished model generally follows the same line of argumentation than Schneider (2003).
Chapter 2.╇ Aspects of structural nativization
it includes different linguistic and metalinguistic fields (such as historical and political developments) that are influenced during the developmental process. The framework of this model is summarized in Table 2-1. On the (vertical) diachronic axis this model distinguishes between five different stages of development: – Stage 1, entitled foundation, describes the earliest contact stage, in which the settlers arrive and early communication systems, such as pidgins, begin to evolve. – Stage 2, exonormative stabilization, describes the colonial phase, during which an increasing number of the indigenous population acquire language skills in the language of the settlers, while at the same time the languages of the indigenous population exert influence on the language of the settlers (such as the borrowing of lexical items to describe exotic flora, fauna or topology). – Stage 3, nativization, describes the later colonial and early postcolonial phase. In this phase, the new variety of English becomes increasingly nativized. Language creativity and conservativism are in constant equilibrium so that new usages only evolve slowly, but become increasingly accepted. – Stage 4, endonormative stabilization, describes the postcolonial phase, during which the new variety starts to be treated as a variety in its own right. Norms come from within the language community rather than from the former colonial power, so that codification and standardization of the variety are of central importance during this stage. – Stage 5, differentiation, is not of interest for the present study, since IndE has not arrived at this stage. During this stage the accepted variety undergoes a process of diversification and dialectal development, as can be shown in varieties such as American English or Australian English. While there is not yet an indigenous-strand dominated variety, Singapore English is a close candidate to such a variety (cf. Lim 2004:â•›6–7).
. It seems doubtful whether IndE will ever arrive at this stage, since it is an indigenous-strand dominated variety. In these varieties of English the indigenous population of the contact area keep up the language of the settlers after independence, even if there are hardly any settlers left. Settler-strand dominated varieties, on the other hand, are varieties where the original settlers (or their descendents) stayed in the region of contact, as is the case, for example, in Australia. Furthermore, the Indian subcontinent is highly panethnic and the number of different Indian languages on the subcontinent is overwhelming. In this regard India differs from the situation in Singapore, where it has been argued that Singapore English is developing into a native speaker-variety (cf. Lim 2004).
˘
stable colonial status
weakening ties; often political independence but remaining cultural association
2: Exonormative stabilization
3: Nativization
Linguistic development/ structural effects
widespread and regular contact, accommodation IDG: common bilingualism, toward language shift STL: sociolinguistic cleavage between innovative speakers (approximating IDG) and conÂ�servative speakers (upholding external norm ‘complaint tradition’
STL: acceptance of original norm, expanding contact IDG: spreading (elite) bilingualism
heavy lexical borrowing IDG: phonological innovations (‘accent’ possibly due to transfer; structural nativization (in wordformation, phrases, prepositional usage, verb complementation) spreading from IDG to STL
lexical borrowing (esp. fauna and flora, cultural terms) ‘-isms’
STL: cross dialectal contact, limited STL: koinézation exposure to local languages toponymic borrowing IDG: minority bilingualism (acquisition of English)
Sociolinguistics of contact/use/ attitudes
5: Differentiation
stable young nation, internal sociopolitical differentiation
group specific (as part of over�arching new national identity)
network construction (increasingly dense group-internal interactions)
dialect birth: group specific (ethnic, regional, social) varieties emerge (as L1 or L2)
acceptance of local norm, positive stabilization of new variety, 4: Endonormative post-independence, self- (member of) new nation, homoÂ�geneity, codification stabilization dependence (possibly after territory based, increasingly attitude to it (residual conservativpanethnic ism); literary creativity in new variety (dictionary writing) “Event€X”)
STL: permanent resident of English origin IDG: permanent resident of indigenous origin
STL: outpost of original nation, ‘English-plus-local’ IDG: individually ‘localplus-English’
STL: colonial expansion: STL: part of original nation trade, military outposts, IDG: indigenous missionary activities, emigration/settlement IDG: occupation, loss/ sharing of territory, trade
1: Foundation
Identity construction
History & politics
Stage
Table 2-1.╇ The evolution of New Englishes as a cyclic process (Schneider€2003:╛255)
10 Structural Nativization in Indian English Lexicogrammar
Chapter 2.╇ Aspects of structural nativization
On the horizontal axis Table 2-1 shows the sociolinguistic processes that take place during these stages, structured into four self-explanatory categories: – – – –
History & politics Identity construction Sociolinguistics of contact/use/attitudes Linguistic development/structural effects
If we apply the framework of Schneider’s dynamic model to the Indian variety of English, a clear-cut specification of the present stage of the variety is hardly possible, since the variety displays features that point towards a variety in stage€3, while at the same time other features point towards a stage 4 variety. For this reason, I will sketch out the features of IndE that are typical of stage 3, as well as those typical of stage 4. Although nativization is still ongoing, I argue that there is a sufficient degree of endonormative stabilization within the variety to place IndE at the beginning of stage 4 of the model. 2.2.2
Indian English in the dynamic evolutionary model: Aspects of ongoing nativization and endonormative stabilization
The process of nativization that constitutes stage 3 of the model is discussed at length in Kachru (1983); therefore I will only briefly sketch the situation in India during this stage. Concerning history and politics, India is an independent nation, although not without ties to Great Britain (in terms of population it is the largest member of the Commonwealth of Nations). This notwithstanding, there are hardly any British citizens living on the subÂ�conÂ�tinent today, so that in terms of identity construction Indians no longer feel that they are under a strong British influence. To use Schneider’s terms, the number of “permanent residents of English origin” (cf. Table 2-1) is negligible, while at the same time the number of “permanent residents of indigenous origin” (cf. Table 2-1) accounts for the vast majority. Concerning the sociolinguistics of contact/use/attitudes, the situation in India differs from other contact situations. Firstly, in contrast to Australian English or American English, for example, Indian English is spoken as a non-native variety by almost all of its speakers, so that the classical label ESL applies here (cf. Görlach 1991:â•›12–13). Furthermore, owing to the large number of indigenous . This has, of course always been the case, but with no British residents remaining in positions of political power, the few British that do still live in India do not exert the same influence as the few powerful British residents in the earlier days.
11
12
Structural Nativization in Indian English Lexicogrammar
Indian languages, phenomena of linguistic transfer are not limited to one L1, but are due to a vast number of L1s that all have an influence on the resulting ESL variety. Bilingualism in India is not as common as the model in question suggests, and complete language shift does not seem likely; nevertheless, the speakers of IndE constitute the powerful elite of Indian society, so that speakers of IndE have an enormous influence within this society. The “sociolinguistic cleavage between innovative and conservative speakers of the settlers’ strand” (cf. Table 2-2) is no longer applicable to the Indian situation, since hardly any settlers are left in India; hence this part of stage 3 may be seen as already lying in the past. However, this situation does to some extent still prevail in the indigenous strand where the “complaint tradition” (cf. Milroy & Milroy 1985) “in which conservative language observers typically claim that linguistic usage keeps deteriorating, that in the new country ‘corrupt’ usage can be heard which, however, should be avoided” (Schneider 2003:â•›248), can still be observed in “letters to the editor” and other discussion channels. Many transparent linguistic developments of the nativization of Indian English (such as the retroflexed realization of the pronunciation of the alveolar series, or new word-formation processes like hybrid compounding) have already been discussed by Kachru (1983) and others (cf. e.g. Mehrotra 1998 and Nihalani et al. 2004), while other more opaque features are yet to be described in detail (cf. Shastri 1992). Although the process of nativization does not seem to be at an end, the emancipatory stance of scholars of IndE, such as Kachru (1985), and the compilation of Indian English corpora and dictionaries (such as Nihalani et al. 1979 and 2004) implies that endonormative stabilization (the title for stage 4 of Schneider’s model) is well under way. With regard to the four dimensions of Schneider’s (2003) model, this stage can be described in the following way: In the history and politics dimension India fulfils the criteria for stage 4 of the model quite well. India has been independent since 1947; it still has ties to the United Kingdom but is no longer under any British political control. In India the violent consequences of independence with the resulting separation of India and Pakistan may be seen as an “Event X”, as Schneider calls it. His definition is as follows:
. It may, for example, be possible for native speakers of Punjabi to use features of another Indian language, such as Telugu, that have been transferred to IndE and standardized within the variety, without having knowledge of Telugu or even being aware that these features are the product of transfer (and standardi�zation) from another Indian language.
Chapter 2.╇ Aspects of structural nativization
[…] an incident which makes it perfectly clear to the settlers that there is an inverse misrelationship between the (high) importance which they used to place on the former mother country and the (considerably lower) importance which the (Schneider 2003:â•›250) (former) colony is given by the homeland.
This definition shows that the terminology Schneider uses is only to a certain extent applicable to the Indian situation, since for an indigenous-strand dominated variety the mental state of the settlers after independence is no longer of much importance. At this point it should be noted that Schneider’s model is at times better suited to describe the development of settler-strand dominated varieties, such as Australian English, since there is a certain amount of emphasis on the sociolinguistic factors in the settler-strand. Thus, interpreting the turmoil of independence as “Event X” in the Indian context is not strictly in the sense that Schneider implies. Still, the loss of influence and interest of Great Britain plays an important role in the development and acceptance of nativized features of IndE. The independence of India and the separation of the subcontinent is a crucial turning point for the variety; I have therefore expanded the use of Schneider’s term to describe a development in the indigenous-strand that is comparable to the development described as “Event X” within the settler-strand in Schneider’s model. In the dimension of identity construction, Indians no longer consider themselves dependent on British influence. In view of the conflicts with Pakistan, territory plays an imÂ�portant role in the Indian situation. India has always been highly panethnic and multilingual, a fact that tends to impede the homogenization of the variety. In the dimension of sociolinguistics of contact/use/attitudes, the aboveÂ�mentioned tendencies of emancipation play an important role. Although the acceptance of Indian English by native speakers is still not extremely high according to empirical surveys (cf. Shaw 1981:â•›119–120; and Hohenthal 2003), the variety is used confidently by many writers and speakers of IndE and, after some discussion within India, creative writing in Indian English is now largely accepted (cf. Kachru 1992:â•›67). The criteria for stage 4 that Schneider subsumes under structural effects are fulfilled for IndE with some exceptions. The variety is becoming increasingly stabilized, and codification is under way, as the dictionary of Nihalani et al. (2004) shows. Homogeneity of the variety, however, is a problematic topic with respect to IndE, since there are so many different L1s involved in the development of the L2.
. For a detailed description of phase 4 elements of Indian English cf. Mukherjee (2007).
13
History & politics
Late phase of the Raj, Independence in 1947, separation of India and Pakistan, English increasingly becomes a link language
post-independence, self-dependence (possibly after “Event€X”)
Stage
Stage 3: Nativization
Stage 4: Endonormative stabilization
widespread and regular contact, accommodation IDG: bilingualism of social elites; total language shift not likely STL: sociolinguistic cleavage between innovative speakers (approximating IDG) and conÂ�servative speakers (upholding external norm), ‘complaint tradition’, phenomenon more important in the IDG strand after Independence emerging acceptance of local norm, positive attitude to it (residual conservativism); literary creativity in new variety
new nation, territory based, highly panethnic and multilingual
Sociolinguistics of contact/use/ attitudes
STL: only a few, but in powerful positions, semi-permanent residents of English origin IDG: majority of the population, permanent resident of indigenous origin
Identity construction
stabilization of new variety, codi�fica�tion (dictionary writing, compilation of corpora, emerging grammars)
heavy lexical borrowing IDG: phonological innovations (‘accent’ possibly due to transfer; structural nativization in wordformation; phrases, prepositional usage; verb complementation) spreading from IDG to STL
Linguistic development/ structural effects
Table 2-2.╇ Classifying Indian English today: Characteristics of stage 3: Nativization and stage 4: Endonormative stabilization
14 Structural Nativization in Indian English Lexicogrammar
Chapter 2.╇ Aspects of structural nativization
Table 2-2 summarizes the features of Indian English at present, outlining the features that are treated as stage 3 features, as well as those that are treated as stage€4 features. Although it is still only a minority of Indians that use the English language in their daily lives, the fact that English is used for all communicative functions, including creative writing and very prominently in the print media (cf. Mehrotra 1998:╛10), shows that IndE is much more than a foreign language in India.
2.3
Different influences on structural nativization
With regard to the similarities between World Englishes, Schneider (2004) claims that there is: a uniform underlying process of mutual identity adjustments and linguistic accommodation between the parties involved in a colonization process. At the heart of this process there is a phase of nativization when varieties of English develop and adopt distinctive linguistic features of their own on all levels of lan(Schneider 2004:â•›227) guage organization.
While this process and the phases of nativization and endonormative stabilization were described at some length in the last section, it is helpful at this stage to put this process into a wider functional framework that accounts for the different factors shaping nativizational processes. The development of a new variety of English is a complex process that involves a wide variety of different factors. Some of these are dependent on the input variety, while others depend on the language users of the new variety. On the one hand, there are different input variety-based factors, such as superstrate retention and exonormative stabilization. With regard to these factors a two-fold process may lead to the development of distinctive features of the new variety that are not primarily induced by its speakers, but by the speakers of the exonormative target model. As this target model, however, is subject to diachronic changes from within its own community, these changes may be institutionalized in the new variety only after a time lag, or possibly not at all. Thus, superstrate retention of the original superstrate and later innovations in the development of the input variety are not always directly transported to the new
. The original style of Schneider’s model has been kept for this table. However it should again be pointed out that IndE displays stage 3 and stage 4 features at the same time, so that the vertical axis in Table 2-2 is not, strictly speaking, a diachronic axis with a clear transition point between the stages.
15
16
Structural Nativization in Indian English Lexicogrammar
variety by means of exonormative stabilization, the time-lag thereby becoming part of the nativizational process. There are also several indigenous as well as settler-based influences on the nativization process. Traditionally, it is often claimed that substrate influences are the most important of these indigenous factors. However, recent publications such as Schneider (2004) or Sand (2004) show that there are far too many similarities between world varieties of English with very different substrate languages for substrate influences to be the sole or even strongest source of influence for nativization processes. A possible explanation for the claim of strong substrate influence may be found in the history of the analysis of varieties of English rather than in the importance of the factor itself. Early studies, such as Kachru (1983) or Nihalani et al. (1979), focused mainly on the levels of phonology and lexicology; levels, where substrate influences are probable and easily detectable. The levels at which the angloversal features span across varieties, for example, similar article use or tense use (cf. Sand 2004) or the use of particle verbs (cf. Schneider 2004), have only recently been placed in the focus of research. Thus, if substrate influence is not necessarily the strongest factor for nativization, there are other important factors on the indigenous side. As on the settlers’ side, the new variety of English is also subject to inherent language change, driven by the relation of innovative and conservative trends. All of these factors in combination will then give the new variety its characteristic features, which in turn may be endonormatively stabilized and institutionalized.
. Sand (2004) calls these universal features of varieties of English Angloversals.
chapter 3
Aspects of lexicogrammar Collocation and verb-complementation
3.1
Introduction – The interdependence of lexis and grammar
Traditional grammatical models did not incorporate lexis and worked as slotfiller models where certain syntagmatic relations open paradigmatic choices that are relatively independent. While accepting the syntagmatically dependent paradigms, Firth challenged the traditional models by contending that “the main concern of descriptive linguistics is to make statements of meaning” (Firth 1957:â•›190). By means of the concepts of collocation and collocability, Firth (1957) introduces a level of meaning that is on the borderline between lexical semantics and grammatical meaning: It must be pointed out that meaning by collocation is not at all the same thing as contextual meaning, which is the functional relation of a context of situation in the context of culture. […] Meaning by collocation is an abstraction on the syntagmatic level and is not directly concerned with the conceptual or idea approach (Firth 1957:â•›195–196) to the meaning of words.
Picking up on this, Halliday (1966) uses the examples of strong argument and powerful car to illustrate his point. He explains that items such as strong are members of a specific set that also includes powerful. While some items within this set can be used in collocation with items of different sets, such as argument, others cannot and are instead used with items of other sets again, such as car. In Halliday’s (1966) framework, item, set and collocation “are definable without reference to restrictions stated elsewhere in the grammar”. This is to say that distinct members of specific sets tend to collocate with specific members of other sets. There are restrictions on paradigmatic choices allowed for at the syntagmatic level that cannot be accounted for by structural restrictions, so that in “lexicogrammatical statements collocational restrictions intersect with structural ones” (Halliday 1966:â•›163) without being dependent on these. The notion of sets of lexical items that have probabilistic collocational tendencies towards members of different sets is an abstract concept that was brought into connection with actual language data only at a later stage. Especially in the
18
Structural Nativization in Indian English Lexicogrammar
1980s, when the English Language Research team at the University of Birmingham worked together with Collins publishers on corpus-based lexicographical research, the views on how lexicography (and lexis in general) relates to grammar shifted. As the evidence started to accumulate it became clear that the accepted categories of lexicography were not suitable; […] Three major areas of language patterning, besides grammar could not be comprehensively presented in a dictionary, […]. These are collocation, semantics, and pragmatics. (Sinclair 1991:â•›2–3, his italics)
To describe the properties of collocation within this research, Sinclair proposes two different “models of interpretation” to explain the “way in which meaning arises in text”, the open-choice principle and the idiom principle (Sinclair 1991:â•›109), which have received much attention in the past two decades. While the idiom principle is useful to account for collocation, it is not limited to the explanation of this linguistic phenomenon but has much wider implications on the interdependence between lexis and grammar: Many uses of words and phrases show a tendency to co-occur with certain grammatical choices. For it was pointed out […] that the phrasal verb set about […] is closely associated with a following verb in the -ing form. […] What is more the second verb is usually transitive, for example set about testing it. Very often set will be found in co-occurrence patterns. (Sinclair 1991:â•›112)
Francis et al. (1996) link their work to the concept of co-occurrence of patterns by describing verbal patterns in relation to semantically defined verb groups that are typically associated with the pattern at hand. These verb groups are semantically defined and relate to specific patterns. Hunston (2002) shows that there is often a certain amount of overlap of typical patterns, with the result that the idiom principle at work leads to a phenomenon she calls pattern flow, i.e. a specific pattern or frame will be part of a new pattern or frames. These patterns constructed mainly on the level of the idiom principle can then additionally be combined with paradigmatic choices at the level of the open-choice principle. Table 3-1 shows the lexis-grammar-interface as a cline between the abstract concepts of the lexicon and the grammar of a language. Of the depicted concepts, collocation and verb-complementation in particular will be the focus in Table 3-1.╇ The lexis-grammar interface Lexis-grammar interface Levels of description
Lexicon word-formation collocation
Grammar phraseology
verb-complementation […]
Chapter 3.╇ Aspects of lexicogrammar
the remainder of this chapter, as the present work aims to describe structural nativization in Indian English at these lexicogrammatical levels. Thus, Section€3.2 will deal with different descriptions of collocation, a level on the lexis grammar continuum that is situated close to the lexis end of the cline. Here the focus will primarily be on two different approaches to collocation, the frequency-based approach and the phraseological approach. A first discussion of earlier findings concerning collocations in Indian English will end this section. Section 3.3 will deal with verb-complementation that is situated closer to the grammatical end of the lexis-grammar continuum. In this part, different traditions of grammatical description of the verb phrase will be described and related to earlier findings concerning verb-complementation in Indian English.
3.2
Collocation
3.2.1
Introduction
Although, as shown above, the notion of collocation has been discussed for at least 50 years, there is so far no consistent definition of the term. It therefore encompasses a variety of different phenomena of lexical co-occurrence. Nesselhauf (2004) gives a comprehensive overview of the different approaches to collocations and phenomena of lexical co-occurrence, which could be viewed as complementary descriptions of the ongoing discussion of collocations as a phenomenon on the interface between grammar and the lexicon. She distinguishes between “two main views” (Nesselhauf 2004:â•›11), a statistical, frequency-based approach and a phraseological approach, which both concentrate on the lexical side of collocations. In the following, I will briefly outline these two views. As the present work is, however, mainly based on the quantitative approach to collocation, this level will be described in more detail than the phraseological approaches. These observations will then be connected to the specific case of variety-dependent collocations. 3.2.2
Quantitative approaches to collocations
The statistical approach to collocations is based on the use of large-scale computer readable corpora that are accessed by means of specific software (such as WordSmith Tools (cf. Scott 1998) or Collocate), which can be used to count cooccurrences of words in predefined spans of co-text. Rooted in the neo-Firthian tradition (e.g. Sinclair and Halliday), this is a standard approach in corpus
19
20 Structural Nativization in Indian English Lexicogrammar
linguistics, although there is ongoing discussion as to whether other factors should also be taken into consideration. At a very basic level collocates are identified within a short span of co-text: The qualification for an instance being scrutinized is co-occurrence within four words of back [the node], on either side, this being the cut-off point […]. No account is taken of syntax, punctuation, change of speaker, or anything other than (Sinclair 1991:â•›117) word-forms themselves.
At this point either raw frequencies of the co-occurring lexical items are used as a basis for the definition of collocations, or more sophisticated statistical procedures are used to identify co-occurrences that are not random. These statistical procedures are based on hypothesis testing by refuting the null-hypothesis of random co-occurrence. Although there is no unanimous agreement as to whether this approach is suitable for the underlying nonâ•‚randomness of language, the resulting values often give a better insight into relations of co-occurrence than raw frequencies do.10 As some of these procedures will play a role in the analysis of Indian and British collocations, I will introduce some of the algorithms that are most commonly used for collocational hypothesis testing. Underlying all the algorithms is the relation of observed occurrences to occurrences that could be expected by chance (cf. Figure 3-1). Figure 3-1 illustrates how randomly distributed co-occurrences can be calculated from observed co-occurrences of words. The right-hand matrix shows the observed frequencies of joint occurrence of two lexical items (O11), the V=v U=u
E11 =
U =/ u
E21 =
R1C1 N
R2C1 N
V =/ v E12 = E22 =
R1C2 N
R2C2 N
V=v
V =/ v
U=u
O11
O12
= R1
U =/ u
O21
O22
= R2
= C1
= C2
=N
Figure 3-1.╇ Expected and observed co-occurrences (Evert 2005:╛76) . Criteria for defining collocations based on raw frequencies vary from counting all co�occurring lexical items as collocations (e.g. Halliday 1966) to only counting recurrent or frequently co-occurring lexical items (e.g. Stubbs 2001). 10. For a discussion of the problems of the underlying null-hypothesis see Kilgarriff (2005).
Chapter 3.╇ Aspects of lexicogrammar
occurrences of the items in the cases where they do not co-occur with another (O12 and O21) and the number of other possible bigrams of lexical items (O22). By using the column totals (C1 and C2) as well as the row totals (R1 and R 2) and the total corpus size (N), the expected frequencies for all fields in the matrix can be calculated using the formulae in the left-hand side of the figure. Based on these values, different statistical relations can be calculated to analyze whether the items co-occur significantly more frequently than expected by chance. The most frequently applied tests in corpus linguistics are the Mutual Information (MI) test and the different asymptotic hypothesis tests t-score, z-score and chi-square (χ²). However, it has to be pointed out that all of these tests have specific shortcomings, hence their popularity in corpus linguistic research seems to be mainly based on the relative ease of application. The MI-score is calculated by dividing O11 by E11 converted to a base-2 logarithm. While the results give an indication of what has been called “strength of collocation” (Hunston 2002:â•›71), the algorithm favours infrequent lexical items that have a tendency to co-occur. An alternative to the use of the MI ratio is the use of asymptotic hypothesis tests, namely t-score, z-score and the chi-square (χ²) test. Although they do not suffer from the shortcomings inherent in the MI test, they still do not yield exact statistical p-values. As Evert (2005) points out: Asymptotic hypothesis tests address two problems of the exact tests: (i) their numerical complexity and (ii) the difficulty of defining more “extreme” outcomes. Instead of the exact p-value, they use a much simpler equation called a test statistic. This statistic is chosen in such a way that its distribution under the null hypothesis converges to a known limiting distribution for large samples (N →€∞). The limiting distribution can then be used to derive an approximate p-value corresponding to the observed value of the test statistic. (Evert 2005, his emphasis)11
Z-score values are calculated by applying the formula (f1) while the related t-score values are calculated by dividing by the square root of O11 as shown in formula (f2). (f1) z =
O11 – E11 E11
(f2) t =
O11 – E11 O11
As z-score values can be inflated with low expected ratios, the use of the t-score is advisable, although, with expected values < 100, the t-score may lead to values 11. Available at http://www.collocations.de.
21
22
Structural Nativization in Indian English Lexicogrammar
too conservative compared with exact testing (cf. Evert 2005:â•›114). Overestimation of significance is also the central problem of the χ²-test that is calculated by applying formula (f3).12 (Oij – Eij)2 Eij i, j
(f3) χ2 = ∑
The limitations of these tests as well as their underlying assumptions have therefore led to some discussion about their usefulness. While it is largely agreed that the results gained by applying them will yield more solid results than counting raw frequencies, it has been pointed out that only exact testing would lead to statistically sound results. Gries (2006), for example, in a programmatic paper on the difficulties of the use of association statistics in linguistic works, strongly advises exact testing: To my mind, it […] does not make much sense to ask whether, on one particular occasion, the t-test yields better results than the binomial test or whether the z-test yields better results than the Fisher-Yates exact test. Even if the z-test proved superior on one occasion, the mathematical assumption underlying it would still be violated, so the question arises whether one should really use measures whose results look promising, but which are based on mathematical assumptions that are violated in one’s own data. I believe the answer to this question is ‘No!’, which is why one should always use exact statistical tests (cf. Stefanowitsch & Gries 2003) as well as simulation and/or resampling methods (cf. Gries 2005[…]) wherever possible. Given the state-of-the-art in modern desktop computing, these techniques are all at our fingertips and there is no reason not to use them if (Gries 2006:â•›199) we are interested in meaningful results.
The last comment on the increased power of desktop computing notwithstanding, exact testing can still be an enormous drain on computational resources, especially if these values are desired for many different collocation candidates. The reason for this is in the hypergeometric distribution underlying Fisher’s exact test shown in formula (f4): (f4) Fisher =
C2 min{R1,C1} ( C1 ) × (
∑
k = O11
k
N R
R1 – k
)
; P =
C1! C2! R1! R2! N! O11! O12! O21! O22!
13
C1! (as well as the other expressions containing !) in this term is a factorial, the product of C1 and all the whole numbers less than it, down to one (cf. Rayson 12. The conservative nature of the t-score and the overestimation of significance of the χ²-test are visualized in Figure 3-1. 13. For a detailed discussion cf. Fisher (1922).
Chapter 3.╇ Aspects of lexicogrammar
2003:â•›48).14 Especially for calculating the association measures for all possible collocations in a corpus (a feature, for example, offered by WordSmith Tools), it would not be feasible to apply the Fisher exact test, thus for applications like this an alternative seems necessary. By relating different association measures to the benchmark of the Fisher exact test, Evert (2005) showed that the log-likelihood function (formula 5) “gives an excellent approximation to the Fisher p-values across the entire range of frequency signatures” (Evert 2005:â•›112) as shown in Figure 3-2. (f5) log-likelihooddunning = –2log
L (O11, C1, r) × (O12, C2, r) L (O11, C1, r) × (O12, C2, r)
L = (k, n, r) = rk (1 – r)n-k ; r =
R1 N
; r1 =
O11 C1
C2
t 200 400 600 800 0
0
200 400 600 800
200 400 600 800
0
0
Poisson
Fisher
200 400 600 800
200 400 600 800
Fisher
G2
O12
0
200 400 600 800 0
X2corr
(cf.€Dunning€1993)
; r2 =
0
200 400 600 800 Fisher
0
200 400 600 800 Fisher
Figure 3-2.╇ Comparison of asymptotic hypothesis tests with exact testing (Evert 2005:╛111)
14. The factorial for 0 equals 1.
23
24
Structural Nativization in Indian English Lexicogrammar
Figure 3-2 summarizes Evert’s (2005) comparison of asymptotic hypothesis tests with the exact p-values according to the Fisher exact test. As can be seen in the top row, “chi-squared overestimates significance dramatically […] [whereas] [t]he t-score measure […] turns out to be highly conservative” (Evert 2005:â•›112). The results of Dunning’s (1993) log-likelihood (G2), however, give a very close approximation of exact p-values, as shown in the bottom left panel.15 In view of the vast computational resources that are needed to apply exact testing, a more feasible method in the majority of cases is the log-likelihood test. This procedure makes it possible to calculate statistics for a large number of collocation candidates, while it is only necessary to use the Fisher exact test for cases where the log-likelihood value is very close to the critical values for significance. 3.2.3
Phraseological approaches to collocations
Phraseological approaches to collocations differ from quantitative approaches, as they put only minor, if any, emphasis on frequency. Instead, syntactic and semantic relations of word co-occurrences are central to the definition and description of collocations. In those approaches in particular the term collocation is by no means used consistently, so that besides the term collocation a number of different terms for related concepts of word combinations are used. While some of them define different semantic or syntactic aspects of the combinability of words, others are largely used synonymously and only differ to a marginal extent. As frequency plays only a minor role in phraseological definitions of word-combinations or phraseological units (cf. Cowie 1998), these combinations are categorized mainly by semantic properties. The main criteria here are: (a) semantic transparency vs. opacity and (b) motivation (cf. Table 3-2). Table 3-2 summarizes the different terminology used for related concepts by different authors over the last five decades. The middle column shows the different terms for semantically opaque, unmotivated word combinations. Examples of these would be spill the beans or kick the bucket, whose meanings can be construed neither compositionally nor through metaphoric extension. In the most recent of these works, the terms idiom or pure idiom are used.16 Partially motivated units are word combinations with meanings that are still not construable compositionally, but can be assigned by means of metaphorical extension: examples 15. Poisson testing is rarely used for statistic testing in linguistic research. 16. Note, however, that the term idiom used in this sense is not congruent with the way it is used in Sinclair’s (1991) idiom principle. While in Sinclair’s model the focus is on holistic storage and reproduction, in phraseological models opacity is the defining criterion for idioms.
Chapter 3.╇ Aspects of lexicogrammar
Table 3-2.╇ Sub-categories of word-like combinations (cf. Cowie 1998:╛7) Author
General category
Opaque, invariable unit
Partially motivated unit
Phraseologically bound unit
Vinogradov (1947) Amosova (1963) Cowie (1998) Mel’čuk (1988b) Gläser (1988a) Howarth (1996)
Phraseological unit Phraseological unit Composite
Phraseological fusion Idiom
Phraseological unity Idiom (not differentiated) Figurative idiom
Semantic Phraseme Nomination
Idiom
Phraseological combination Phraseme, or Phraseoloid Restricted collocation Collocation
Composite unit
Pure idiom
Pure idiom
Idiom
Idiom (not differentiated) Idiom (not differentiated) Figurative idiom
Restricted collocation Restricted collocation
here would be blow off steam (in the non-technical sense) or hit rock bottom (also in the metaphorical sense) (cf. Cowie 1998:â•›5). The right-hand column shows the different terminology used for collocations in the most frequently used sense. A defining feature of these combinations is a restricted commutability of the elements. Hausmann (1984) differentiates between the elements of a collocation: one element is called the base (German Basis) the second the collocator (Kollokator). While the base is freely chosen, only a restricted set of words is commutable with this base and can be used as collocators. A restricted set of commuting items is, however, not a sufficient defining condition for restricted collocations, as many combinations are restricted semantically without forming restricted collocations. Cowie (1998) therefore distinguishes free combinations (which can be restricted on semantic grounds) from restricted collocations (with arbitrary limitations). Besides defining collocations on grounds of motivation and commutability and selectional restrictions, in phraseological approaches collocations are also frequently subclassified on syntactic grounds, so that, for example, adjective + noun (heavy smoker), adverb + adjective (deeply disappointed) or verb + noun (make a decision) can be viewed as subgroups of collocations (cf. Nesselhauf 2004:â•›22). With regard to the last group, it should be pointed out that this group includes stretched- or light-verb constructions (cf. Allerton 2002), which can be defined as follows: [T]he noun is eventive and carries the bulk of the meaning, while the verb contributes comparatively little to the lexical meaning of the combination and can (Nesselhauf 2004:â•›20) therefore be called a ‘light verb’.
25
26 Structural Nativization in Indian English Lexicogrammar
These constructions have special importance in the description of second language varieties of English, as it has been pointed out that there is a tendency to use light-verb constructions differently and more frequently in second language varieties of English (cf. Mukherjee & Hoffmann 2006:â•›154–155). They show, for example, that the verb give is used frequently as a light verb in light-verb constructions where speakers of British English would prefer a construction based on the simplex root verb. This extended use of light-verb constructions in collocations that would be unusual for speakers of British English may therefore be seen as an important feature of Indian English at the intersection between collocation and verb-complementation. 3.2.4
Collocations in Indian English
Although the notion of collocational differences between varieties of English in general, and British English and Indian English in particular, has been discussed for some time, so far there are only a few studies that deal with collocational phenomena in Indian English. As most of them were carried out before the corpus revolution (cf. Rundell & Stock 1992), they are either based on intuitive observations or a very small database. Furthermore, studies such as Kachru (1983) use the term collocation in a very general sense; therefore, neither statistical measures are used to provide information on significance of collocation, nor are the phraseological distinctions introduced in Section 3.2.3 applied. A large variety of Indian English collocations can be found in the dictionary section of Nihalani et al.’s (1979/2004) Indian English: Handbook of usage and pronunciation. However, Nihalani et al. (1979/2004) do not provide a stringent definition of the term, either. As I show elsewhere (Schilk 2006), the lack of a more sophisticated definition makes it relatively difficult to make statements on the typicality and the actual currency of many of the collocations described by Nihalani et al. (2004). Furthermore, concerning verb + noun collocations, Nihalani et al. (2004) include light-verb constructions in their definition of collocation without explicitly pointing out their specific status (cf. e.g. the entry for the verb do in Nihalani et al. 2004:â•›67). However, these constructions have been shown to be of particular importance in Indian English, as Mukherjee & Hoffmann point out: “[o]ne particularly striking phenomenon in IndE is the use of give in a light-verb construction, i.e. a construction with give as a semantically fairly empty (“light”) verb and a lexically more specific (“heavy”) noun (e.g. give explanation)” (Mukherjee & Hoffmann 2006:â•›154). If this is the case, then the examples included for light verbs such as do in Nihalani et al. (2004) cannot be seen as a comprehensive description
Chapter 3.╇ Aspects of lexicogrammar
of the specific IndE use of these verbs. This is especially the case, since light-verb constructions not only often form collocations unusual in British English, but are also influential at other levels of the lexis-grammar interface, such as, for example, verb-complementational patterns of specific verbs. 3.2.5
Collocation in the lexis-grammar continuum
In order to analyze collocations on a corpus basis, both quantitative approaches that make use of the statistic methodology introduced in Section 3.2.2 and phraseological approaches as described in Section 3.2.3 can be of high descriptive value. A preference for one of these approaches, therefore, may be motivated by two considerations. The first is often a pragmatic consideration: especially in the case of second language varieties of English, balanced large-scale corpora do not yet exist; hence a purely statistic analysis may not yield such strong results as the statistic analysis of large-scale corpora of native varieties of English, such as the British National Corpus (BNC) or the Bank of English. The second consideration is of a linguistic nature. Although statistic significance can be claimed as a conditio sine qua non, such as, for example, by Gries (2006), stating significance for specific collocations in a variety by itself often does not shed light on the reasons for the specific preferences. As collocations of specific words influence other parts of the lexicogrammar, neglecting phraseological considerations only allows for insight into one part of a larger and more complex system. Therefore, in the following, I will describe a different level of the lexis-grammar interface that may be related to variety-specific collocational preferences, namely the different description of verb-complementation and complementational profiles of verbs and interdependencies between lexical items and complementational structures.
3.3
Verb-complementation
3.3.1
Introduction
In Section 3.1 I introduced the interrelation between lexical choice and syntactic pattern. In the following section I will set the ground for the systematic description of the possible patterns at the level of verb-complementation. As a comprehensive description of verb-complementation would be beyond the scope of the present work, I will highlight some influential grammatical models that focus on the verbal phrase. In the first two parts of this description, models of verb-complementation
27
28
Structural Nativization in Indian English Lexicogrammar
in two influential reference grammars of the English language, namely the Comprehensive Grammar of the English Language (Quirk et al. 1985) and the Longman Grammar of Spoken and Written English (Biber et al. 1999) will be the focus. The second subsection will deal with verb-complementation from a cognitive perspective, the main focus being on construction grammar. After these descriptions of theoretical models for the description of verb-complementational patterns with a special emphasis on (di)transitive event types, a brief overview of earlier work concerning verb-complementation in Indian English will be provided. 3.3.2
Verb-complementation in descriptive grammar (Quirk et al. 1985) and corpus-based descriptive grammar (Biber et al. 1999)
In this section The Comprehensive Grammar of the English Language (Quirk et al. 1985) as one of the most influential reference grammars of English will be used as a starting point for the discussion of different models of verb-complementation. A basic notion of Quirk et al. (1985) is the distinction between formal and functional grammatical categories. Formal refers to the internal structure of a unit. At this level, units such as the verb phrase or the noun phrase can be classified. These formal elements are in a paradigmatic relationship with corresponding formal elements and may, but need not be, in paradigmatic relationship to non-corresponding elements, so that, for example, a noun phrase can always be substituted by a different noun phrase, but only on some occasions by other kinds of phrases, such as, for example, an adverb phrase (cf. Quirk et al. 1985:â•›48). Functional grammatical categories, on the other hand are not defined by their internal structure, but according to their function: “by function is meant a unit’s ‘privilege of occurrence’ in terms of its position, mobility, optionality, etc, in the unit of which it is a constituent” (Quirk et al. 1985:â•›48–49). Thus, these higher units, such as clauses, or sentences, may be defined formally by referring to the formal smaller units that they are composed of, or functionally by referring to the functions that are realized by the formal elements. Therefore, clauses and sentences consist of formal units that realize different functions. In the case of simple sentences, that is, sentences that consist of a single independent clause, a different number and different combinations of the five functional categories (subject, verb, object, complement and adverbial) of clause constituents can be realized: one two-constituent pattern, consisting of a subject and a verb, three three-constituent patterns consisting of a subject, a verb and either an object, a subject complement or an adverbial, and three four-constituent patterns consisting of a subject, a verb, a direct object and either an indirect object, an object complement or an adverbial (cf. Quirk et al. 1985:â•›58).
Chapter 3.╇ Aspects of lexicogrammar
This underlying schema allows for the categorization of functions of verbs according to the number of possible arguments they may take. Thus, verbs functioning in the two-constituent pattern are called intransitive verbs, verbs in the three-constituent pattern are called copular verbs (in the case of SVC and SVA patterns) or monotransitive verbs (in the case of the SVO pattern) and verbs functioning in the four-constituent patterns are called complex transitive verbs (in the case of the SVOC and SVOA pattern) or ditransitive verbs (in the case of the SVOO pattern) (cf. Quirk et al. 1985:â•›1171).17 The different types of complementation can be realized by a variety of different formal patterns. Ditransitive complementation, for example, can be achieved in six different ways: both objects can be realized as noun phrases (D1), the indirect object may be realized as a to-prepositional phrase (D2) and the direct object can be realized in a number of different ways, i.e. as a that-clause, a wh-clause, a wh-infinitive clause or a to-infinitive (D2–D6) (cf. Quirk et al. 1985:â•›1171). From these patterns, a number of structurally related patterns may also be derived, such as, for example, fronted object constructions or passive constructions (where the subject can, but need not be realized as a by-agent). Quirk et al. (1985) go on to show that the potential for taking arguments is, at least partially, a property of the respective verb. There is, for example, only a relatively small number of verbs, apart from the principle copula be, that allow copular complementation, which can be distinguished into current and resulting copulas. Examples of the first group would be verbs of appearance, such as appear, feel, look or seem, while examples of the second group would be resultative verbs such as become, grow or turn (cf. Quirk et al. 1985:â•›1172). It should be pointed out, however, that these verbs are not limited to copular complementation, as this usage is defined by a specific meaning of the verb. The verb grow, for instance, can take other complementation patterns if it is used with different aspects of meaning, e.g.: The girl has grown over the last year (intransitive) or He has grown a beard (monotransitive) or, possibly, even I grew myself a nice crop of tomatoes (ditransitive). The same distinction holds for the other complementation patterns as well. While there are verbs that typically take a specific complementation pattern and thus may be called monotransitive, complex transitive or ditransitive verbs, they often also allow a variety of different complementation patterns when associated 17. The use of the terms monotransitive verb or ditransitive verb etc. may be misleading, as many verbs can function in more than one complementation pattern, so, strictly speaking, these verbs could be called verbs in monotransitive (ditransitive etc.) function (cf. Quirk et al. 1985:â•›1168). However, it should be pointed out that some verbs are typical of verbal functions; e.g. the verb give is typical of a ditransitive complementation pattern and the verb become of copular complementation patterns.
29
30
Structural Nativization in Indian English Lexicogrammar
with different meanings. A case in point here would be the verb give, which could be viewed as the central ditransitive verb. However, it is often also used with monotransitive complementation patterns. It has also been pointed out that typically intransitive verbs such as cry may be used ditransitively in specific situations, e.g. in Cry me a river (cf. Goldberg 1995:â•›36). It should, however, not go unmentioned that the organization of different formal realizations of functional categories in the transitivity framework proposed by Quirk et al. (1985) is not entirely uncontroversial, as the neatness of this framework comes at the price of being, at times, not fully stringent (cf. Standop 2000:â•›231). For example, the category D2 (ditransitive complementation with a prepositional object) may be interpreted as an example of complex transitive complementation rather than ditransitive complementation, as the use of a to-prepositional phrase may be interpreted as the realization of an adverbial rather than an (indirect) object. Consider Example (1): (1) He gave a ring to Mary.
Prepositional phrases, however, cannot realize objects in the framework of Quirk et al. (cf. 1985:â•›60). Thus, the prepositional object is, strictly speaking, the noun phrase following the preposition. If this were the case, however, the preposition to would be part of a prepositional verb give to. If we consider the properties of prepositional verbs, the preposition must still occur in the passivized pattern, as shown in Example (2): (2) a. She looked after her children. b. Her children were looked after (by her).
If give to were a prepositional verb, the passivization of (1) should then be (1b), which is not grammatical: (1) b. *Mary was given a ring to.
This structural inconsistency is strongly criticized by Standop (2000), who considers examples like (1) to be instances of complex transitive complementation. On the other hand, this syntactic criticism, although plausible, does not take into account the semantic equivalence of ditransitive complementation with two objects realized by noun phrases with examples like (1). At least semantically there is a difference between examples like I left the key at home and Example (1). This difference as well as the semantic synonymy of a prepositional realization of the indirect object with a ditransitive noun phrase realization is nicely demonstrated in Goldberg (1995), who argues for introducing related constructions that, although structurally distinct, are semantically synonymous. This will be discussed in detail in Section 3.3.3.
Chapter 3.╇ Aspects of lexicogrammar
The Longman Grammar of Spoken and Written English (Biber et al. 1999), the second reference grammar to the English language described here, differs from the approach of Quirk et al. (1985) and other reference grammars in the underlying data set and its mode of description.18 Unlike other works, Biber et al. (1999) base their grammar entirely on a 40-million-word corpus of spoken and written English, the Longman Corpus of Spoken and Written English (henceforth LSWE).19 While Quirk et al. (1985) use a combined approach of intuitive findings and consult three relatively small corpora (cf. Quirk et al. 1985:â•›33), in Biber et al. (1999) the emphasis is on the actual use of English in the spoken and written domain, as well as across different registers represented by the LSWE. The use of a large computer-readable corpus not only offers the opportunity to make statements about actual language performance, but also shows frequencies and distribution of linguistic features. With regard to the valency of verbs, they show that verbs are related to transitivity patterns, i.e. specific verbs occur in one or a number of different transitivity patterns. Thus valency is not strictly a property of the verb, but different valency patterns can be used for different verbs, as the majority of the verbs under scrutiny (all verbs that occur in the LWSE corpus at least 300 times per million words) occur in different transitivity patterns: The large majority for notably common verbs in the LSWE corpus occur with transitivity patterns. C. 47% of these verbs occur with both transitive and intransitive patterns. C. 36% of these verbs occur only with transitive patterns. The number of common verbs taking transitive intransitive and copular patterns is (Biber et al. 1999:â•›382) not distributed evenly across semantic domains.
The last point can be exemplified by taking a closer look at the verbs that Biber et al. (1999) included in this description. The verbs are chosen on the basis of their frequency and are, in a second step, sorted according to the semantic domains to which they belong. Although Biber et al.’s (1999) findings regarding different transitivity patterns in relation to specific verbs cannot be discussed comprehensively here, some major points should nevertheless be highlighted. They describe a strong relation between the meaning of a verb and the transitivity patterns it€takes:
18. It should be pointed out, however, that in its general approach Biber et al. (1999) follows the notions of Quirk et al. (1985). This becomes especially clear when contrasting Biber et al. (1999) with Huddleston & Pullum (2002). 19. For a detailed description of the 40-million-word LSWE corpus see Biber et al. (1999:â•›24–35).
31
32
Structural Nativization in Indian English Lexicogrammar
Verbs occurring with transitivity patterns are common across all semantic domains. In contrast there are very few exclusively intransitive verbs. Only three semantic domains have common verbs that are exclusively intransitive: activity (Biber et al. 1999:â•›383) verbs, occurrence verbs and existence verbs.
Although Biber et al. (1999) do not, at this point, differentiate between different transitive patterns (monotransitive, ditransitive and complex transitive), the findings already serve to underline the point made in the last section, namely that the meaning of the verb has a strong influence on the pattern that it takes. A strong point of the LGSWE lies in the different registers that are included in the LWSE corpus. The text included covers four different registers: conversation, fiction, news and academic writing. As specific lexicogrammatical patterns may be preferred in a given register, this distinction allows for a more fine-grained description of verbs in relation to the patterns in which they occur. This point should not be neglected in relation to variety-specific lexicogrammatical differences. Note that it has been shown that register differences are often stronger than variety differences, so the comparison of two varieties in many different registers is often problem-ridden (cf. Sand 2004:â•›294–295), a point that I will return to in the following chapters. Biber et al. (1999) show that, especially for verbs with a high versatility, i.e. verbs that allow a great variety of patterns and uses, the use of these verbs in specific patterns is register-dependent. For example, the verbs show and get, which take almost all possible patterns, show a large amount of register dependency. 3.3.3
Verb-complementation in cognitive grammar and construction grammar
The notion of a lexis-grammar continuum is also found in cognitive grammar. Based on the view that language is used to symbolize basic cognitive conceptualizations, linguistic structure is seen as being shaped by several non-linguistic factors that influence the language users’ physical experience and possibilities of linguistic representation of this experience (cf. Langacker 1999a:â•›14–15). In this view, language is not an autonomous cognitive faculty since in principle the same cognitive processes are those applied to other cognitive tasks. Sounds and utterances here form the input and output of the cognitive process that governs speaking and understanding. Langacker (1999b) posits three essential components for the symbolization of cognitive conceptualization: semantic structures, phonological structures and symbolic links between the two, pointing out that “the central claim of CG [cognitive grammar] is that nothing else is needed” (Langacker 1999b:â•›1). From this point of view “[l]exicon, morphology, and syntax
Chapter 3.╇ Aspects of lexicogrammar
Table 3-3.╇ Degrees of schematicity (cf. Langacker 1999b:╛122) Degree of schematicity
Example
mainly specific schematic in several positions constructional schema with a specific element all elements schematic
crane X’s neck X take Y over X’s knee and spank Y [[send] [NP] [NP]] [[V] [NP] [NP]]
form a gradation claimed to be fully describable as assemblies of symbolic structure” (Langacker 1999b:â•›122). The elements of this gradation may vary in a degree of schematicity. Here, what traditionally forms part of the lexicon is situated at the most specific end of the gradient, whereas with increasing schematic symbolization, items would be traditionally seen as syntactic structures. Table 3-3 gives specific examples of symbolic structures that vary in their degree of schematicity and shows that it is hard to categorize these examples according to whether they are lexical or syntactic items. Using these examples, Langacker (1999b) argues that there is no evident reason why a constructional schema that incorporates a specific element, e.g. [[send] [NP] [NP]], should not also be a lexical item. That in turn is only one step away from according lexical status to assemblies like [[V]€[NP] [NP]], all of whose components are schematic. My point, of course, is that lexicon and grammar grade into another so that any specific line of demarca(Langacker 1999b:â•›122) tion would be arbitrary.
Construction grammar expands the notion of varying degrees of schematicity by considering the units within the lexis-grammar continuum as different constructions that are mostly atomic and substantive at the lexical point of the continuum and become increasingly schematic. Within this continuum the notion of construction differs from traditional grammatical theories by not upholding the atomistic view present in these, but offering a more integrated model that does not, for example, have to treat idiomatic expressions as awkward exceptions on the performance side of language, but rather integrates them into a regular system. To highlight the problems that an atomistic view may cause, Fillmore et al. (1988) point out: Under the idealization just discussed, any sentence in a language can be resolved into configurations containing only constituents of the designated types, arranged according to the standard rules, and yielding interpretations which follow from regular principles of compositional semantics. […] As useful and powerful as the atomistic schema is for the description of linguistic competence, it does not allow the grammarian to account for absolutely everything in its terms. (Fillmore et al. 1988:â•›503–504)
33
34
Structural Nativization in Indian English Lexicogrammar
The notion of construction as a double continuum between the atomic and the complex, as well as the substantive and schematic, however, allows for descriptions of form as well as meaning in a more thorough way. Idiomatic expressions – e.g. the let alone construction described by Fillmore et al. (1988), or the way construction described by Goldberg (1995) – are integrated into a model that “conforms to Langacker’s content requirement for a grammar: the only grammatical entities that are posited in the theory are grammatical units and schematizations of these units (Langacker 1987:â•›53–54)” (Croft & Cruse 2004:â•›256). Thus constructions pair the cognitive side of the event type with the formal side of linguistic representation, or form and meaning. Goldberg (1995) describes how lexis and syntax as well as form and meaning are integrated in construction grammar: Lexical constructions and syntactic constructions differ in internal complexity, and also in the extent to which phonological form is specified, but both lexical and syntactical constructions are essentially the same type of declarative repre(Goldberg 1995:â•›7) sented data structure: both pair form with meaning.
This pairing of form and meaning of syntactic constructions can be exemplified by means of the ditransitive construction. On the meaning side there is the underlying cognitive event type of TRANSFER that may be schematized by the formula ‘X causes Y to receive Z” – a volitional agent causes a willing recipient to receive a patient –, whereas on the formal side this cognitive event is represented by a subject (representing X), two objects (representing Y and Z) and a specific syntactic pattern SVO1O2 (representing the transfer event). In Goldberg’s (1995) description we find a number of other syntactic constructions that are related to the ditransitive construction in terms of argument realization. In the following I will sketch out two of these, as they will be picked up in the present study, namely the transfer-caused-motion construction and the caused-motion construction. The cognitive representation of the caused-motion construction is ‘X causes Y to move Z’, which is structurally defined as [SUBJ [V OBJ OBL]], where the oblique is a directional (prepositional) phrase (cf. Goldberg 1995:â•›152). In the terminology of Quirk et al. (1985), this construction would be regarded as complex transitive, realized by the pattern SVOA (C3). However, Goldberg (1995) argues for the necessity of the level of construction, because “the semantic interpretation cannot plausibly be attributed to the main verb and other means of deriving the semantics compositionally also fail” (Goldberg 1995:â•›152). Thus, it is the construction that adds further meaning to the compositional meaning of the arguments and their realizations.
Chapter 3.╇ Aspects of lexicogrammar
The transfer-caused-motion construction is closely related to the ditransitive construction and the caused-motion construction. The transfer-caused-motion construction is considered to be a metaphorical extension of the caused-motion construction, as the transfer of ownership is represented as a physical transfer.20 The structural representation of this construction is [[S] [V] [OBJ] [OBL]], where the oblique element is a to-prepositional phrase. The relation to the causedÂ�motion construction is structurally evident, while the relation to the ditransitive construction is a semantic relation, as ownership is being transferred. This can be exemplified by drawing on the terminology of Quirk et al. (1985), who subsume the structural representation of this construction under the representations of ditransitive verb complementation with the oblique described as a prepositional object (D2). However, Goldberg (1995) emphasizes that there is no syntactic relation between the ditransitive construction and the transfer-caused-motion construction, while there is such a relation to the caused-motion construction, as “their semantic synonymy […] does not constitute a motivation link” (Goldberg 1995:â•›91). While there is semantic synonymy between the caused-motion construction and the ditransitive construction, there is no pragmatic synonymy, as some events of transfer can only be represented by one construction, but not the other; consider Examples (3) and (4): (3) He sent the book to storage. (4) ?He sent storage a book.
Figure 3-3 summarizes the relations among the caused-motion construction, the ditransitive construction and the transfer-caused-motion construction. As shown in Figure 3-3, the caused-motion construction is related to the transfer-caused-motion construction on the level of its syntactic realization. The arrow labelled IM shows the semantic relation between the caused-motion construction and the transfer-caused-motion construction as a metaphorical extension. The underlying semantic pattern of the caused-motion construction, a CAUSE to MOVE relation with the potentially profiled participants cause (realized by the subject), goal (realized by the oblique) and theme (realized by the object) are translated to the CAUSE to RECEIVE relation of the transfer-caused-�motion construction with the participants agent (realized by the subject), recipient (realized by the oblique) and patient (realized by the object) due to this extension. The ditransitive construction and the transfer-caused-motion construction, on
20. Note that, due to its nature of a metaphorical extension of the caused-caused motion construction, the transfer-caused-motion construction is no longer described as a subcategory in its own right in Goldberg (2006).
35
36
Structural Nativization in Indian English Lexicogrammar
Sem cause-move pred Syn
< cause
goal theme >
<
v
> subj
obl
obj
Im: Transfer of ownership as physical transfer Ditransitive construction Sem cause-receive pred
Sem cause-receive < agt pred Syn
v
rec
pat >
<
>
rec
pat >
<
> subj obj obj2
v
Syn
Transfer-caused-motion construction
< agt
S-synonymous (→ ~ P-synonymous)
subj obl obj
Figure 3-3.╇ Relations among constructions (Goldberg 1995:╛91)
the other hand, are related semantically as their underlying semantic patterns and potentially profiled participants are synonymous, while the syntactic realizations differ. Thus, the use of this model of a cognitively based construction family allows for the position of a semantic relation of the different constructions semantically, while taking into account their different syntactic realizations. 3.3.4
Verb-complementation in Indian English
Whereas in the last sections theoretical approaches to verb-complementation were discussed, in this section some relatively recent work on structural nativization in Indian English lexicogrammar will be introduced. In research into varieties of English, it has often been assumed that distinct grammatical features of these varieties are relatively sparse (cf. Quirk et al. 1985:â•›19), hence studies, especially
Chapter 3.╇ Aspects of lexicogrammar
those undertaken before the compilation of large-scale corpora, such as Kachru (1983), concentrated on the description of lexical or phonological innovations as well as on innovations at the level of text types. In more recent studies, partly due to the increased availability of larger corpora of text, it has been found that there are also many innovations at the interface between grammar and the lexicon: [I]t has been found in many earlier listings of characteristic features of WEs (e.g. Bauer 2002) that distinctive phenomena tend to concentrate at the interface between grammar and the lexicon, concerning structural preferences of certain words (like the complementation patterns that verbs allow), co-occurrence and collocational tendencies of words in phrases, and also patterns of word forma(Schneider 2004:â•›229) tion.
Accordingly, there has been an increasing interest in studies of World Englishes that concentrate on the lexis-grammar interface. There are two basic trends in corpus linguistic research into different varieties of English. One approach makes use of different corpora of a relatively large number of varieties to investigate fields in which varieties of World Englishes display differences and similarities, while the other restricts itself to a comparison of a specific second language variety (in this case Indian English) and a native-language variety, in most cases – for historical reasons – British English (BrE). Examples of the former are the works of Schneider (2004), a study of particle verbs in different World Englishes, and Sand (2004), who compares article use in different World Englishes, while examples of the latter are the works of Shastri (1996), Olivarría de Ersson & Shaw (2003) and Mukherjee & Hoffmann (2006). Olivarría de Ersson & Shaw (2003) show, for example, that significant differences can be observed for the frequencies of the complementation patterns of pelt-verbs in Indian and British English (cf. Olivarría de Ersson & Shaw 2003:â•›154). Specifically, they observed clear variety-specific tendencies in the verb-complementational patterns of the pelt-verbs PELT and SHOWER: while in British English one of the most frequent patterns is the complementation with a with-prepositional phrase, in Indian English there are strong tendencies to complement PELT with an at-prepositional phrase and SHOWER with an on-prepositional phrase (cf. Olivarría de Ersson & Shaw 2003:â•›154). Note in this context that in the with-phrase pattern, which is preferred in British English, the syntagmatic positions of the goal/beneficiary and of the transferred/theme are different from the positions of these semantic roles in the patterns preferred in Indian English. While in the with-pattern the theme is in end-focus position, in the on/at-patterns the goal is in this position. Mukherjee & Hoffmann (2006) pick up on Olavarría de Ersson & Shaw’s (2003) findings and look at the complementation of ditransitive verbs in Indian English. In a first step they employ the British and the Indian component of the
37
38
Structural Nativization in Indian English Lexicogrammar
International Corpus of English (ICE) to categorize all the instances of verbs such as GIVE and SEND into five basic complementation types, which are described and exemplified in (5)–(9). In the type-I pattern both objects are realized as noun phrases; this pattern is viewed as being the most basic type of ditransitive verbcomplementation (cf. Mukherjee & Hoffmann 2006:â•›151). In the type-II pattern, the indirect object is realized as a to-prepositional-phrase and placed after the direct object. In type-III, the indirect object is not made explicit, while in typeIV both objects are deleted, and in type-V it is the direct object that is not made explicit. As these patterns are prototypical patterns, it is also possible to derive structurally related patterns, e.g. with specific elements in fronted positions, relative clause structures, participle constructions and passive constructions. (5) a.
I (S) GIVE [Oi:NP] [Od:NP]
b. On Tuesday members of Parliament gave the government their over� whelming support (6) a.
II (S) GIVE [Od:NP] [Oi:PPto]
b. I meant to give it to you earlier (7) a.
III (S) GIVE [Od:NP] Oi
b. he wanted physical love and I couldn’t give that (8) a.
IV (S) GIVE Oi Od
b. If you give and take when there’s that close bodily contact it’s great (9) a.
V (S) GIVE [Oi:NP] Od
b. I didn’t give Lakshmi I had just given Sumi you know
Mukherjee & Hoffmann (2006) point out that there are significant quantitative differences in the preference for complementation patterns of the verbs GIVE and SEND between the two varieties. For example, there is a clear tendency in British English to use GIVE in the type-I pattern, while in Indian English the most frequent pattern of GIVE is the type-III pattern (cf. Mukherjee & Hoffmann 2006:â•›152). As earlier descriptions such as Nihalani et al.’s (1979) dictionary assumed that the range of verbs that are used in the basic ditransitive pattern, i.e. type-I, differs between the varieties, Mukherjee & Hoffmann (2006) compiled a 31-millionword newspaper database for Indian English derived from the online archive of
Chapter 3.╇ Aspects of lexicogrammar
the Calcutta-based national newspaper The Statesman.21 They show that there are many verbs that Indian English users use in this pattern which are not admissible in the type-I pattern in British English, for example the verb GIFT: (10) He was forced to bring down Nabi in the danger zone after gifting him the ball. (11) Delay means serious risk of gifting Islamabad a talking point. (12) She said she wanted to gift him a dream.
Mukherjee & Hoffmann (2006) explain the extension of the ditransitive construction to new verbs like GIFT in terms of nativized semantico-structural analogy, which is defined as: [a] process by means of which non-native speakers of English as a second language are licensed to introduce new forms and structures into the English language because corresponding semantic and formal templates already exist in the (Mukherjee & Hoffmann 2006:â•›166f.) English language system.
Thus, Mukherjee (2009:â•›127) suggests that it is the “analogy between the semantics and syntax of give […] on the one hand and semantically closely related verbs such as gift on the other” (Mukherjee 2009:â•›127) that leads to the innovative ditransitive complementation pattern of the verb. These examples show that while there are relatively few studies focusing on variety-specific differences in verb-complementation, all of these studies can show that these differences exist. However, some of these differences are mainly found on a quantitative level. While it is only rarely the case that users of Indian English use verb-complementational patterns that are not applicable in British English, they seem to favour different complementational patterns. While, for example, instances of ditransitive use of verbs that are not used in the ditransitive pattern in British English are a low frequency phenomenon, as shown above, the monotransitive use of verbs that are preferably used in the ditransitive pattern in British English seems to be a salient feature of Indian English.
21. A second, although relatively similar, edition to Nihalani et al. (1979) now exists, published in 2004. As Shastri (1996), however, is based on the first edition, I have decided to use this edition for the sake of consistency.
39
chapter 4
Methodology
4.1 Introduction: A corpus-based description of structural nativization in Indian English lexicogrammar In order to carry out a corpus-based description of collocation and verb-complementation, several methodological questions have to be addressed that will be discussed in this chapter. Although Indian English displays characteristics of an endonormatively stabilizing variety (cf. Section 2.2), at a lexicogrammatical level a description of the variety without reference to a native-speaker variety may not show the distinct features of Indian English in full detail. Therefore, the Indian English data will be related to reference corpora of British English. British English, being the historical input variety, seems the logical choice for reference corpora, although this choice is not entirely unproblematic, as each variety is constantly changing and the reference variety may thus be something of a moving target. Furthermore, there are very few entirely comparable corpora of Indian and British English, namely the corpora included in the International Corpus of English (ICE) and the Indian and British members of the Lancaster-Oslo/Bergen (LOB)/ Brown family, the Kolhapur corpus of Indian English and the LOB corpus. The ICE corpora in particular can be seen as a representative sample of the varieties, as they include both spoken and written data, covering a large variety of genres and text types, and were therefore chosen as the standard corpora for this study. In Section 4.2.1, these corpora will be described in some detail. However, with a size of 1 million words each, the ICE corpora often do not contain enough data to analyse low- and medium-frequency features of the respective varieties. As no balanced larger databases of Indian English existed at the time of writing, I compiled a very large database of Indian newspaper English by collecting data published in a large national Indian newspaper that can be accessed freely on the internet.22 Although this database only covers one specific genre, the sheer mass of data as well as the influence of this genre on the variety seem to make this procedure a worthwhile approach for tapping data available on the world-wide web.
22. As a suitable reference corpus the newspaper a part of the British National Corpus (BNC) that contains nine million words of British newspaper English will be used.
42
Structural Nativization in Indian English Lexicogrammar
The description of the compilation of this database as well as a description of the database itself will be covered in Section 4.2.2. As most of the corpora do not include corpus analysis software and are thus just computerized text collections, it is necessary to use software tools for the analysis of these corpora. For the present study WordSmith Tools (WST) V4.x will be used to process the different corpora. A detailed description of the software is provided in Section 4.2.3. Sections 4.3 and 4.4 deal with the concepts of verb-complementation and collocation and the methodology used for the description and analysis of these features in this study. In Section 4.3 the approach to different verb-complementational profiles of verbs that are potentially used in the ditransitive and the transfer-caused-motion construction will be discussed. In Section 4.4 these complementational profiles will be related to a model of collocational profiles of these verbs with special attention to collocates in the position of the direct object of the verb. Finally, Section 4.5 describes the procedure of profiling different trivalent verbs according to the models discussed in 4.3 and 4.4 and comparing them across the different varieties of English.
4.2
Corpora and tools
4.2.1
The International Corpus of English – The British and Indian components (ICE-GB and ICE-India)
The ICE corpus project works on the compilation of comparable subcorpora of different varieties of English. During the launch phase of this project 18 different subcorpora were planned for the different regions across the world where English is spoken either as a native or a second language (cf. Greenbaum 1996:â•›3–4). This number has now risen to well beyond 20, with some corpora still under construction. Each of these subcorpora contains about 1 million words of spoken and written English, organized into 500 texts with about 2,000 words per text. 300 of these texts are spoken, while the remaining 200 are written texts, so each subcorpus contains about 600,000 words of speech and 400,000 words of writing (cf. Greenbaum 1996:â•›30). Table 4-1 illustrates the corpus design of the subcorpora included in the ICE corpus. Spoken texts are subcategorized into dialogues and monologues, while written texts are subcategorized into printed and non-printed material. The author and speaker selection of the ICE corpora focuses on the standard variant of the variety by including only authors and speakers that fit a specific profile, as described in Greenbaum (1996):
Chapter 4.╇ Methodology
Table 4-1.╇ The International Corpus of English (cf. Nelson et al. 2002:â•›307–308) Spoken (300) Dialogues (180)
Private (S1A) (100)
€
€
Public (S1B) (80)
€
Monologues (120) Unscripted (S2A) (70)
Scripted (S2B) (50)
Written (200) Non-printed (50) €
€
€
Printed (150)
€
€
€ €
€ €
€
€ €
Conversations (90) Phonecalls (10) Class Lessons (20) Broadcast Discussions (20) Broadcast Interviews (10) Parliamentary Debates (10) Cross-examinations (10) Business Transactions (10) Commentaries (20) Unscripted Speeches (30) Demonstrations (10) Legal Presentations (10) Broadcast News (20) Broadcast Talks (20) Non-broadcast Talks (10)
Student Writing (W1A) (20) Student Essays (10) Exam Scripts (10) Letters (W1B) (30) Social Letters (15) Business Letters (15) Academic (W2A) (40) Humanities (10) Social Sciences (10) Natural Sciences (10) Technology (10) Popular (W2B) (40) Humanities (10) Social Sciences (10) Natural Sciences (10) Technology (10) Reportage (W2C) (20) Press reports (20) Instructional (W2D) (20) Administrative Writing (10) Skills/hobbies (10) Persuasive (W2E) (10) Editorials (10) Creative (W2F) (20) Novels (20)
In general the population represented in the corpus will be adults of 18+ who have received formal education through the medium of English to the completion of high (secondary) school, but second language countries might require a (Greenbaum 1996:â•›3) university degree.
This speaker selection bears specific implications for Indian English, as well as other second language varieties, as only a relatively small proportion of the speakers who frequently use English are included in the potential corpus population. Owing to the correlations of social standing and English medium education in
43
44 Structural Nativization in Indian English Lexicogrammar
second language communities, only a highly educated élite is therefore included (cf. Lange 2007:â•›98–99). Basing the population of the corpus on educational criteria thus has a certain impact on the comparability of the corpora, especially as native corpora include a much wider educational spectrum of the speech community and do not only focus on an educational élite. However, the education-based speaker selection also has some advantages for the description of second language varieties. By only focusing on the acrolectal use of English, the identification of variety-specific features as opposed to learner mistakes becomes much more reliable than it would be if also mesolectal or basilectal speakers were included. A great advantage of the ICE corpora is the inclusion of the large variety of different text types and genres and the fact that they contain a large amount of spoken data. In particular for Indian English no other balanced corpora of spoken language exist to date, so ICE-India offers unique opportunities to gain insight into communicative processes. However, this balanced design is not entirely without drawbacks. As the corpora are relatively small, a fact that is connected with the immense work involved in the compilation of spoken corpora, the number of words contained in each category is at times also very small. For instance, only 40,000 words of creative writing are included in the corpora, which makes it difficult to find sufficient examples of any lexicogrammatical phenomenon within some genres. Nevertheless, with most of the originally envisioned corpora compiled, and much work carried out on the basis of these corpora, the foundations for the description of many varieties of English have been laid. However, as of today, more than a decade after the first ICE corpora became available, the only corpus that has been completely parsed grammatically is ICE-GB. Therefore, for most of the corpora, including ICE-India, only plain-text editions exist at this stage. These can only be analyzed with ICE-independent software, whereas ICE-GB is linked to its own software package, although a plain-text version of ICE-GB can also be processed by other software. These points will be taken up in 4.2.3. 4.2.2
The Times of India corpus
As earlier studies on Indian English lexicogrammar have indicated, the 1-millionword standard corpora of Indian English are often not large enough to make concrete statements on salient features of the variety (cf. e.g. Shastri 1996; Olavarría de Ersson & Shaw 2003). The possibilities of gaining additional data from internet sources have played a role in variety research for some time. The discussion about the use of the web as a linguistic corpus, however, is not limited to variational corpus linguistics but has also been the focus of many more general linguistic studies.
Chapter 4.╇ Methodology
Table 4-2.╇ Approaches to the use of the Internet as a linguistic corpus (cf. Thelwall 2005:╛529) Approach
Data Ease quality of use
Sophistication Example
Direct search engine queries Search engine intermediate interfaces Webcrawler for stylized web genres Webcrawler for small sites Webcrawler for large multi genre sites
low low
high low medium medium
Google or site-based queries WebCorp
high
low
Newspaper archives
high
medium medium high low medium high
Forum / Newsgroups National university webs
Therefore, in this section I will give a brief overview of different possibilities for using the world-wide web as a linguistic database in general and the implications of these possibilities for variational research in particular. Table 4-2 shows different approaches to linguistic research using internet data. While direct search-engine queries are very easy to use, these queries are beset with many problems, such as unknown data sources and authorship, preference of search engines for commercial websites or, especially in the case of sitebased search engines, user interfaces that do not fulfil linguistic requirements. On the other hand, webcrawlers, i.e. computer programmes that automatically access websites and save the data contained, although more difficult to use, offer various options for customization that is often needed in order to select relevant data. For the present study, I developed a specific webcrawler that accesses the freely available online archives of The Times of India (cf. http://timesofindia.indiatimes. com/), one of the largest English newspapers published in India. This crawler is specifically tailored to access the data of news archives and would thus fall under the category of “Webcrawler for stylized web genres”. Picking up on Hoffmann (2007), who describes a process of obtaining large amounts of text from internet sources, I compiled a 110-million-word database of Standard Indian English, based on the online archive of The Times of India. In order to achieve this I developed a webcrawler based on the LWP-Module in Perl that saved the data from all the issues of the years 2002–2005.23 The Times of India website contains a very large online archive, where articles from 1999 until 2005 can be accessed easily and freely. The articles included are 23. This is the same time period that is covered by the Statesman Database, a comparable database that underlies Mukherjee & Hoffmann (2006), so these databases can also be combined to gain even more data from two different news sources.
45
46 Structural Nativization in Indian English Lexicogrammar
digitalized print articles. On any given day of the above period, there are approximately 200–400 news articles stored in this archive, each containing an average of 300 to 500 words; hence any single day contains between 60,000 and 200,000 tokens of running text. As this vast amount of data makes manual extraction almost impossible, the “Web-page to megacorpus” method proposed by Hoffmann (2007) is ideally suited to the task of retrieving all the relevant data. The recursive webcrawler developed for this purpose starts at a source page, here the entry point to the different years of news archives, and in turn opens the page for every day in a year, then opens all news articles of the specific day and saves the text of the article on the hard drive. After this the program repeats this procedure for the following years, recursively. At the core of the underlying Perl script are different regular expressions (cf. Friedl 1997) that identify the relevant links that are to be followed and the actual body of text from the articles. As this body is separated from other content on the website by specific mark-up, the automated search for these mark-up tags and the text covered between them, using regular expressions, is an economical and easy way to find and download all of the relevant text. However, not only the relevant text is downloaded, but also irrelevant mark-up and other undesirable material, such as e.g. advertising. Therefore, in an additional process the data gathered is then rid of HTML mark-up and any other undesirable content that may be embedded in the downloaded texts. In this way, the complete content of a newspaper archive or a similar website can be saved on a computer and in turn processed by corpus-linguistic tools, such as WordSmith Tools. The news included in The Times of India is organized in the following categories: Politics (c.€22%), Sports (without Cricket; c.€20%), India (c.€12%), Cities (c.€12%), Infotec (Information and Technology; c.€10%), Cricket (c.€4%), Entertainment (c.€4%), Health and Science (c.€4%), India Business (c.€4%), International Business (c.€4%) and World (c.€4%). However, category-based access is possible only for the actual day of publication, while the archives contain only unsorted headlines.24 It is therefore not possible to automatically sort the text ex-ante according to category. Ex-post sorting may be possible, but is not feasible under the conditions mentioned above. The choice of a newspaper corpus is, of course, to some degree motivated by the availability of such huge data sources. There is virtually no other genre that can be accessed and processed in the way that online archives of printed newspapers can. However, this practical reason is not the only motivation for the preference of this genre over others. Several other reasons make the genre a good choice for 24. In a newer version of the archives many of the article categories, especially city news can be derived from the respective URL, at the time of access (2005) this was not possible.
Chapter 4.╇ Methodology
a corpus of standard Indian English. Firstly, the authors of these newspapers can be considered very proficient users of the English language, so nativized features are relatively unlikely to be what are often considered learner mistakes. Secondly, this genre plays a very important role in the formation and nativization of Indian English. It can reasonably be argued that besides English-medium schools, Indian English newspapers have one of the biggest influences on the development of the variety.25 While in inner-circle countries the role of spoken language is of the highest importance for the development of these varieties, in outer-circle countries the use of English is more strongly writing-oriented. English newspapers here fulfil a dual function of identity construction and language standardization. Since no concise grammars or dictionaries exist for the variety, publication in newspapers often serves as an instrument of codification of variety-specific features. The majority of the national newspapers in India are published in English and educated Indians often do not consider reading regional newspapers, which are published in native Indian languages, as socially prestigious behaviour (cf. Mehrotra 1998:â•›10–11). Regarding circulation, English-medium newspapers are the largest group of newspapers in India: It may […] be pointed out that in terms of circulation the English newspapers stand first among all newspapers in India. In 1989 the English dailies had the highest circulation of 2.6 million copies, Hindi being second with 2.2 million (Mehrotra 1998:â•›10) copies.
These figures are somewhat dated, but the situation today shows that there is an even greater market for English newspapers. The Times of India alone claims a “daily circulation level of 22 lakh [2,200,000] copies”, being the “largest selling English daily between Tokyo and Paris” and thereby forming the “Masthead of India, an opinion-forming instrument for India and Indians” (cf.: http://syndication. indiatimes.com/articlelist/12972317.cms). A third and very significant reason for the choice of online newspapers is the fact that a variety of Indian English newspapers is available on the internet, connecting Indians living in different parts of the world to the subcontinent, which can be seen as a further influence of this genre on both the linguistic currency of the variety and an instrument of identity construction in Schneider’s (2003) sense. However, some of the drawbacks of the use of newspaper material should not go entirely unmentioned. Firstly, it is often not possible to estimate the extent of native-speaker post-editing, since most newspapers do not offer concise information on their post-editing practice. Furthermore, even if material that is explicitly marked as containing news-agency material is excluded, some of this 25. For a discussion of English medium teaching by Indian teachers cf. D’Souza (1997:â•›94–95).
47
48 Structural Nativization in Indian English Lexicogrammar
material may still be used in slightly modified fashion without explicit mention of the source agency. 4.2.3
Corpus analysis software: WordSmith Tools V4.x
The empirical part of the corpus analysis, based on the corpora described above, will be carried out with the analysis tool WordSmith Tools V4.x (cf. Scott 1998). WordSmith Tools (WST) is a type of multi-purpose corpus-processing software. It includes three basic tools, a concordancer (Concord), a wordlist and lexical statistics package (Wordlist), and a keyword analysis tool (Keywords). In the present work Keywords is not used and will therefore not be described in detail. Concord is a concordancer that allows even larger amounts of text to be processed relatively fast. The basic output of Concord consists of KWIC (keyword in context) concordances that can be sorted according to the user’s needs in a three-layer sorting algorithm. In addition to the basic concordancing functions, a number of other functions are included in Concord, such as displaying collocates, providing the user with frequently occurring multiword clusters of the keyword, and plotting the occurrence of the keyword graphically for the whole corpus. However, the display of collocates is more sophisticated in the Wordlist tool, which is used for this purpose in the present study. Wordlist, apart from its basic function of creating corpus-based wordlists with frequency counts, also allows the user to index a corpus for further work. This is necessary for a comprehensive computation of the existing collocations in the corpus. This function counts the occurrence of every word in a corpus in possible co-occurrence with every other word in the corpus and offers the user a selection of different statistic measurements for the resultant collocations.26 This procedure makes it possible to obtain a comprehensive overview of the collocations of any word in a corpus. However, this is a purely statistic approach to collocations. Any word that co-occurs at least five times with another word is considered a collocation by WST, while the statistics give further information about possible significance. Therefore, the results obtained here will only be used as a starting point, after which further steps will be taken to classify the collocation candidates obtained by Wordlist; these steps will be explained in Section 4.4. I will end this brief explanation of the basic features of WST with a reference and a caveat. WST is constantly updated, so for this reason alone it is not
26. In earlier versions of WST 4.x, some of the underlying algorithms were flawed. Most of these have been updated. Nevertheless, owing to these original flaws, all of the material used in the present study has been manually checked for the correct statistics.
Chapter 4.╇ Methodology
possible to give an all-encompassing overview of all of its features. Furthermore, as WST has been chosen for this study due to its versatility, not all of the features contained in the software package were used and are therefore not explained in detail at this point.27 The caveat concerns the use of WST in combination with ICE corpora. As WST is not limited to use with a specific corpus, but can process any plain-text corpus, there are some issues of compatibility with the ICE corpora, since they contain specific mark-up that is not compatible with WST. Two points in particular have always been problematic when using WST to process ICE corpora (cf. e.g. Lange 2007:â•›99). The first is that not all the texts in the ICE corpora are exactly 2,000 words long. In most cases the texts are slightly longer and the part that surpasses the 2,000 word limit is given a mark-up symbol for so-called extracorpus material. However, this mark-up is not processable by WST. Furthermore, ICE-India in particular includes metacomments on the corpus that cannot be identified as such by WST. For these reasons I have decided to delete extra-corpus material and metacomments wherever possible.28
4.3
Verb-complementational profiles
Two basic considerations underlie the choice of verbs that will be analyzed empirically according to the complementation patterns that they frequently take in correlation with specific collocates. In the following, these considerations will be briefly discussed. The first of these considerations is based on earlier work on verb-complementation in Indian English, while the second is rooted in the construction grammar approach to trivalent verbs proposed by Goldberg (1995). It has been shown in the past that trivalent and ditransitive verbs differ in the distribution of possible complementation patterns between different varieties of English (cf. Olavarría de Ersson & Shaw 2003 and Mukherjee & Hoffmann 2006), and thus potentially contain nativized features of the varieties. While Olavarría de Ersson & Shaw (2003) concentrate on a class of trivalent verbs that could be classified as verbs used in the caused-motion construction in the Goldberg (1995) framework, Mukherjee & Hoffmann (2006) concentrate on a class of verbs that are associated with the ditransitive construction. Goldberg (1995) describes a relation 27. A comprehensive description of the software can be found at http://www.lexically.net, M.€Scott’s website. 28. In order to achieve this I used a Perl-script that automatically accesses all 500 files of the respective ICE corpus and deletes all extra-corpus material (marked <X> ) and all editorial comments (marked <&> &>).
49
50
Structural Nativization in Indian English Lexicogrammar
between these constructions and a third construction, the transfer-caused-motion construction, as explained in Section 3.3.3. While the caused-motion construction and the transfer-caused-motion construction are related syntactically, as they both profile the indirect object with a prepositional phrase, the relation between the ditransitive construction and the transfer-caused-motion construction is of a semantic nature (cf. Figure 3-3). For many verbs it has been shown that they function in the ditransitive construction as well as in the transfer-caused-motion construction, a phenomenon often referred to as the dative-alternation. Stefanowitsch & Gries (2004:╛106) show that different verbs are more commonly used with either the ditransitive or the transfer-caused-motion construction. However, while they show that specific verbs that are potentially used in both constructions have preferences for one of these constructions (they speak of verbs being attracted or repelled by a specific construction), the reason for these preferences is not entirely clear. This is particularly the case for verbs that appear frequently in both constructions. In this context, Goldberg (1995) argues for a pragmatic distinction between these constructions. From the point of construction grammar this makes sense, since there is only a need for two distinct constructions if these constructions are not completely synonymous.29 Goldberg (1995) demonstrates this pragmatic non-synonymy by using different examples of non-focused themes or recipients: If the theme is not focused (e.g. by encoding it in a personal pronoun), the ditransitive construction is infelicitous. If the recipient, on the other hand, is not focused, the transfercaused-motion construction is infelicitous, since the focused theme should be in end-position (cf. Goldberg 1995:╛92). Consider Examples (13) and€(14): (13) a. #She gave an old man it. b. (She gave it to an old man.) (14) a. #She sold a slave trader him. b. (She sold him to a slave trader.) (cf. Goldberg 1995:╛92)
In the following I aim to take the concept of non-P-synonymy a step further by empirically analyzing the different collocates of trivalent verbs that are potential themes (and thus function as direct objects), since the choice of the theme and its focus are at least to some extent part of the pragmatics of the specific verb. The concept of collocational profiles of the verbs in the present analysis will be picked
29. Cf. Goldberg (1995:â•›67): “The Principle of No Synonymy: […] Corollary A: If two constructions are syntactically distinct and S(emantically)-synonymous, then they must not be P(ragmatically)-synonymous.”
Chapter 4.╇ Methodology
Table 4-3.╇ Collostructional attraction (+) and repulsion (–) in ICE-GB (–log Fisher exact, 10, based on Coll. Analysis 3) (cf. Mukherjee 2009:â•›133)30 Complementation type in ICE-GB
ask
give
offer
send
tell
ditransitive (active) ditransitive (passive) complex-transitive monotransitive dimonotransitive intransitive
58.0 + 13.0 + â•⁄ 5.8 – 15.0 – 73.2 + â•⁄ 3.1 –
â•⁄╛╛Inf. + 117.5 + â•⁄ 11.8 – â•⁄â•⁄ 0.6 – â•⁄â•⁄ 0.2 – 104.2 –
20.6 + 21.2 + â•⁄ 0.9 – 10.3 + â•⁄ 0.5 + 20.1 –
68.9 + â•⁄ 3.8 + â•⁄ 5.8 + 10.6 + â•⁄ 0.3 + 32.7 –
â•⁄╛╛Inf. + â•⁄ 61.4 + â•⁄â•⁄ 9.5 – â•⁄ 97.5 – 260.4 + â•⁄ 65.0 –
up in Section 4.4, â•›while here the focus is on the possible complementational patterns of verbs. Most trivalent verbs can be used in more than one complementation pattern. This is exemplified in Table 4-3, which subsumes five of the most frequent trivalent verbs that occur in ICE-GB, and their complementation patterns in terms of the classification of the ICE-GB parsing system.31 Based on a collostruction analysis (cf. Stefanowitsch & Gries 2004), attractions and repulsions of these verbs with regard to the specific complementation patterns or constructions are given in p-values (–log, 10-based) in Table 4-3. The verbs that attract a ditransitive complementation pattern are give and tell, with give as the prototypical ditransitive verb showing a stronger attraction to ditransitive complementation in the passive than tell. Send, ask and offer also attract ditransitive complementation, albeit less strongly than give and tell with a relatively weak attraction for passive ditransitive complementation, especially in the case of send. The five verbs in Table 4-5 do not attract complex transitive complementation, with the exception of send, as shown in Example (15): 30
(15) She would have said oh God I wouldn’t send my boy there. <s1a–23€329>
As complementation with a to-prepositional phrase used for the profiling of the recipient is subsumed under monotransitive complementation, the instances that 30. Coll. Analysis 3 is a program-script developed by Stefan Th. Gries based on the statistic package R, available under: http://www.linguistics.ucsb.edu/faculty/stgries/research/overviewresearch.html. 31. Note that only trivalent verbs that occur both in the ditransitive and in the transfer-causedmotion construction have been chosen. Furthermore, it should be pointed out that the classification of transitivity in ICE-GB differs slightly from the terminology used through most of this work, as the semantic synonymy of prepositional realization of the indirect object is not reflected in this classification. These instances are subsumed under monotransitive complementation.
51
52
Structural Nativization in Indian English Lexicogrammar
would be classified as belonging to the transfer-caused-motion construction in the present work are found here. This complementation is most strongly attracted by the verbs offer and send. In ICE-GB, verbs that take an indirect object but do not have a profiled direct object are classified as dimonotransitive. Here tell and ask are the verbs that most strongly attract this pattern, as in (16): (16) And she came back and told her husband, you see. <s1a-052 108>
Finally and unsurprisingly, all of the verbs are only rarely used intransitively, so that this pattern is repelled by all five verbs, most strongly by give. While the grammatical mark-up of ICE-GB makes this kind of analysis possible, there are two points regarding this mark-up that should be briefly discussed here. The first point is a general problem with all grammatical mark-up. As syntactic parsing is always dependent on specific underlying assumptions and classifications, an analysis of a syntactically parsed corpus always comes at the price, at least initially, of accepting these choices. Furthermore, because manually parsing a corpus is very time and resource consuming, these projects will always be carried out over a longer period of time with many different people involved, a fact that may lead to inconsistencies in the mark-up. In the present case, the classification of semantically ditransitive sentences in which the recipient is profiled with a to-prepositional phrase as monotransitive complementation is insofar problematic as no distinction is made between sentences that are semantically as well as structurally monotransitive from those that are only structurally so. For the present analysis, therefore, a classification system has been chosen that is not identical to the one used in the parsing of ICE-GB. Although theoretically Goldberg’s (1995) framework of constructions (especially the ones discussed above) underlies the analysis, its design is suited to illustrating the mental organization of these constructions rather than to describing quantitative distributions of specific constructions or complementation patterns in which verbs are used. For this reason, I have decided to utilize an adaptation of the descriptive framework that Mukherjee & Hoffmann (2006) used for the description of ditransitive verbs. Formally based on the system for the description of ditransitive verb-complementation in the Comprehensive Grammar of the English Language (Quirk et al. 1985), this model differs in that it only accepts verbs as being ditransitive if they occur in the ditransitive construction, or the “basic type of ditransitive complementation [D1]” (Mukherjee 2005:â•›14). In the present work, this system will be modified to the extent that, although the different complementation patterns will be labelled according to the same types (Type-I to Type-V, as described in 3.3.4), they will be related to Goldberg’s (1995) construction framework. This approach has two advantages: firstly a comparison of the results from the present study with
Chapter 4.╇ Methodology
Table 4-4.╇ Complementation patterns of ditransitive verbs (cf. Mukherjee & Hoffmann 2006:╛172) Type
Pattern
I
(S) GIVE [Oi:NP] [Od:NP]
Ia Ib Ic Id
(S) GIVE [Od:NP] [Oi:NP]
[Od: NPantecedent] (rel. pron.) [S] GIVE [Oi:NP]
[Oi:NPantecedent] (rel. pron.) [S] GIVE [Od:NP] [Od:NPfronted] [S] GIVE [Oi:NP]
miscellaneous IP IPb
[S < Oi active] BE given [Od:NP] (by-agent)
[Od:NPantecedent] (rel. pron.) [S < Oi] BE given (by-agent)
miscellaneous II IIa IIb IIc
(S) GIVE [Od:NP] [Oi:PPto]
(S) GIVE [Od:NP] [Oi:PPfor]
[Od:NPantecedent] (rel. pron.) [S] GIVE [Oi:PPto] (S) GIVE [Oi:PPto] [Od:NP] miscellaneous
IIP IIPb
[S < Od active] BE given [Oi:PPto] (by-agent)
[antecedent]co (S < Od)co (BE) given [Oi:NPto] (by-agent)
miscellaneous III IIIb
(S) GIVE [Od:NP] Oi
[Od:NPantecedent] (rel. pron.) [S] GIVE Oi miscellaneous
IIIP IIIPb
[S < Od active] BE given Oi (by-agent)
[antecedent]co (S < Od)co (BE) given Oi (by-agent)
miscellaneous IV
(S) GIVE Oi Od
miscellaneous IVP V
given Oi Od (by-agent) (S) GIVE [Oi:NP] Od
the earlier results by Mukherjee & Hoffmann (2006) will be facilitated. However, the second reason may be seen as the more important: Mukherjee & Hoffmann’s (2006) model is used for ditransitive verbs, owing to the procedure of classifying all verbs with all their possible complementation patterns as ditransitive if they can be found in the type-I pattern. Although there is no explicit reference to Goldberg’s (1995) transfer-caused-motion construction, by classifying prepositional complementation (i.e. the type-II pattern) as a semantically motivated
53
54
Structural Nativization in Indian English Lexicogrammar
alternative to the type-I pattern, the semantic synonymy described by Goldberg (1995) is reflected in this model.32 However, there are some differences between the classification system used in the present work and that used by Mukherjee & Hoffmann (2006). Mukherjee & Hoffmann differentiate between patterns that are structurally related to the five basic complementation patterns that they describe (cf. Table 4-4). Besides passive patterns, this differentiation includes, for example, patterns with fronted objects or objects in antecedent position. These related patterns are labelled type-Ia, typeIb etc. Of these related patterns, only the passive patterns have been retained in the quantitative analysis of the present study, while all other patterns related to the basic patterns are subsumed as derivational patterns, labelled type-Ider, typeIIder etc. Moreover, the total number of patterns described in the present analysis is smaller, as the type-IV and type-V patterns are not included and are subsumed under others. The reason for this lies in the integrated approach to verb-complementation and collocation in which only collocates in the direct object slot are taken into consideration. As the direct object is not profiled in the type-IV and type-V patterns, a correlation between Od-collocate and the respective verb cannot be analyzed. This approach to collocational correlations with verb-complementational patterns will be the topic of the following section (4.4).
4.4
Collocational profiles
Trivalent verbs used in the ditransitive or the transfer-caused-motion construction usually collocate with formal realizations of up to three functional and semantic categories. Semantically these can be described as an agent, a recipient and a patient involved in a transfer process, which are realized by the subject, the direct object and the indirect object or a prepositional oblique, respectively. As discussed in Section 4.3, the ditransitive and the transfer-caused-motion construction, albeit semantically synonymous, differ in their pragmatics, so that, depending on the focused participants, there are many instances where one or the other construction is not applicable. While this is relatively straightforward in the examples given by Goldberg (1995:â•›92), in which the non-focused elements are realized by prepositions, non-P-synonymy can also be extended to the realizations of the semantic categories involved in the transfer process. The working hypothesis for the present analysis therefore involves the lexical realization of the 32. Note, however, that Mukherjee & Hoffmann’s (2006) model does not include verbs that are only used in the transfer-caused-motion construction (type-II pattern), but not in the ditransitive construction (type-I pattern), owing to their focus on ditransitive verbs.
Chapter 4.╇ Methodology
participants of the process: it is assumed that specific collocates of different trivalent verbs will correlate with the choice of construction that the verb is used in. The base for the collocation in this respect is therefore always the trivalent verb under scrutiny, while the collocation candidates of the quantitative analysis are restricted both quantitatively (by only accepting collocation candidates that are statistically significant) and functionally (by only including candidates that potentially function as the theme in the construction, realized as a direct object).33 This definition of collocation is partly inspired by Bartsch (2004), who also defines collocations on a quantitative and a qualitative basis. However, for statistic and systematic reasons the model used by Bartsch (2004) is only to a certain degree applicable to the present study. Bartsch (2004:â•›76–77) defines collocation candidates on statistic grounds, while certain qualitative criteria must be met for the collocation to qualify as a genuine collocation: (1) Within a span of 3:3 (or 5:5) words to the left and right of the node word (the search word), (2) two (or more) words which co-occur recurrently with the node word, (3) and whose frequency of co-occurrence can be said to be statistically significant according to at least one of the three statistic algorithms employed (the default values are MI ≥ 10 and t-score ≥ 2,576 for 95% certainty and a significantly high chi-square rating), are said to constitute collocation candidates. These collocation candidates are subsequently subjected to qualitative criteria. In order to qualify as a genuine collocation, the constituents (4) must be in a direct syntactic relation with each other, and (5) display either lexically and/or pragmatically constrained lexical selection, or (6) have an element of semantic opacity such that the meaning of the collocation cannot be said to be deducible as a function of the meanings of the constitu(Bartsch 2004:â•›76–77) ents.
In the present study, point 1 is restricted to a 5:5 span, as direct object collocates are not limited to a 3:3 span, and moreover a much larger span would yield too many results.34 Point 2 is modified by defining recurrent co-occurrence as a minimal 33. In the basic type-I, type-II and type-III patterns, the direct object is always realized as a noun phrase. However, only the head of this noun phrase will be considered the collocate to the verb. 34. It could be argued that positions to the right of the verb are more important for direct object collocates, and therefore the span should be extended. However, in quantitative research into collocation, a certain degree of arbitrary span settings cannot be avoided. Positions to the left of the verb are important in order to identify passive patterns and patterns in which the Od is realized as an antecedent NP. Furthermore the setting of this span is only relevant for the
55
56
Structural Nativization in Indian English Lexicogrammar
co-occurrence of five instances. This is done for practical as well as systematic reasons. A lesser minimal co-occurrence would yield many marginal results of infrequent co-occurrences that would be of little significance to the description of possible correlations between lexical and grammatical co-occurrence. The main modification of the quantitative side of Bartsch’s (2004) model concerns point 3. As I do not believe that an optional significance for any of the tests that Bartsch applied can be a sufficient criterion for statistic significance, I opted for a log-likelihood approach as an approximation to (cpu-intensive) exact testing. Particularly the optional chi-square significance implemented in Bartsch’s (2004) model will lead to a qualification of collocation candidates that are not significant according to exact testing (cf. Section 3.2.2) so that chi-square testing is not deemed a useful criterion for collocational significance. Due to the conservative nature of the t-score, virtually any collocation candidate that is significant according to the t-score will also be significant according to chi-square testing.35 As I discussed in Section 3.2.2, the log-likelihood score, however, gives a good approximation of exact p-values. Therefore, a significance according to the G2-value is used as the only criterion for statistic significance. As this categorization of statistic significance based on G2-values alone still may not be as reliable as exact testing, a very high level of significance (p ≤ 0.0001) is used in order to avoid allowing for candidates that might not be significant according to exact testing. Additionally, certain qualitative criteria, besides collocating with the predefined verb, must be met. These differ from the qualitative criteria underlying Bartsch’s (2004) definition, because only a very specific set of collocations is under scrutiny in the present study. As described above, only collocates of the verb that fulfil the semantic function of a patient or theme are considered, although, of course, the verb will have many more different collocates. Semantic opacity, a criterion in Bartsch’s (2004) model, does not play a role in this definition, as collocation is not limited to restricted collocations in Nesselhauf ’s (2004:â•›14) terms, but also includes what Nesselhauf calls free combinations (cf. Section 3.2.3). Collocation candidates and Od-collocates are thus defined according to five conditions: Condition 1: Collocation candidates must occur in a span of 5:5 of a node word that is a predefined trivalent verb. selection of collocation candidates. Once the threshold of five occurrences within this span is met, other ocurrences in the same clause are also counted for the creation of a verb-complementational profile. 35. Note that in all the cases in Bartsch (2004) where the MI-criterion or the t-score criterion are violated, the chi-square criterion allows for collocation candidate status (cf. Bartsch 2004:â•›216–221).
Chapter 4.╇ Methodology
Condition 2: A minimum number of five co-occurrences must be met. Condition 3: Collocation candidates must be statistically significant according a log-likelihood value of ≥ 15.13 (p ≤ 0.0001) as an approximation to exact testing. Condition 4: Collocation candidates qualify as Od-collocates if the collocate fulfils the semantic function of a patient in a cause-receive process. This definition differs from the phraseological definitions discussed in Section€3.2.3 by accepting collocations that are only restricted on the grounds of their semantic function. This represents a quantitative approach to what has been called semantic preference, a “relation, not between individual words, but between a lemma or word forms and a set of semantically related words” (Stubbs 2001:â•›65). To apply this methodology to the different corpora, in a first step the corpora are indexed using WST. From this index it is possible to calculate all the collocations in the corpus that fulfil conditions 1 and 2. The collocations for the relevant lemmata are extracted from this list and are tested for condition 3. Subsequently, the remaining collocation candidates are manually tested for condition 4. This procedure results in a set of collocation candidates for the respective trivalent verbs that are statistically significant collocations that fulfil the function of the direct object in ditransitive or transfer-caused-motion patterns, so that it is possible to create a comparable collocational profile for each verb in each variety.
4.5
Verbs under scrutiny: GIVE, SEND and OFFER – a quantitative overview
Based on the methodological framework discussed in the previous sections, a short overview of the three verbs under scrutiny is useful for the understanding of the detailed analysis provided in the following chapters. As a comprehensive description of all ditransitive verbs is beyond the scope of this analysis, a set of three verbs that are typically used in the ditransitive and the transfer-caused-Â�motion construction was chosen. The choice of these verbs is based on their overall frequency and, thus, their currency in the English language and the preference of these verbs in the different constructions.36 Initially, the study will deal with the verb GIVE, which is the most frequent and most prototypical ditransitive verb in English. A less frequent verb with a tendency to be used in the transfer-caused36. Preference of the verbs in the specific constructions can be inferred, for example, from Stefanowitsch & Gries’s (2004) work on collostructions.
57
58
Structural Nativization in Indian English Lexicogrammar
motion construction as well as in the ditransitive construction is SEND, which will be the second verb under scrutiny. The third verb analyzed here is OFFER, also a verb much less frequent than GIVE, which has been shown to be associated with both constructions.37 Besides differing in terms of frequency, these three separate verbs also have different semantics, which, it is supposed, also influence the preferred argument structures. Whereas GIVE shows typical ditransitive semantics in the sense of an agent transferring a patient/theme to a recipient, with SEND and OFFER things are somewhat different. SEND, although often profiling the same thematic roles as GIVE, also displays some semantic differences. Firstly, the actual transfer does not have to be completed, as is the case when the sent entity gets lost on the way; so in this case the situation schema also includes incomplete transfer. Furthermore, there are certain restrictions on the choice of argument realization depending on different factors. These will be discussed in detail in Chapter 6, which deals with SEND. One example here is the “[a]nimacy restriction on [the] doubleÂ�object construction” (Levin 1993:â•›46); the animacy of the recipient or goal of the transfer process influences the grammaticality of the double-object construction. Consider Examples (17) and (18): (17) a. Please send us the rest of the copies […] b. Please send the rest of the copies to us. (18) a. Hairy’s parents had enough resources to send their only son to a university b. *Hairy’s parents had enough resources to send a university their son.
While (17a) can be paraphrased as (17b), albeit with slight semantic differences (cf. Pinker 1989), (18a), in which the recipient of the transfer is an inanimate entity, cannot be paraphrased as (18b) without jeopardizing the grammatical wellformedness of the sentence. In the case of OFFER, the transfer process that is usually a complete process with GIVE and that can be incomplete with SEND, is only a potentially intended process of transfer. This has been described as not being “changes of possessions but proactive commitments of some sort guaranteeing them” (Pinker 1989:â•›111). Thus, the potentially intended transferred entity of the transfer scheme in OFFER never actually has to be transferred; indeed the proactive commitment of the agent might in reality not be given, that is to say, transfer might not even be intended by the agent, despite the fact that he or she is offering.
37. Furthermore, all three verbs allow other complementation patterns, the most frequent case being the type-III pattern, in which the indirect object is not made explicit.
Chapter 4.╇ Methodology
Table 4-5.╇ Frequency of the verbs under scrutiny as counted by WST 4.x / BNC-online Verb (lemma) GIVE SEND OFFER
ICE-India
ICE-GB
Times of India
BNC (periodicals)
total
pmw
total
pmw
total
pmw
total
pmw
1,989 380 136
1,768 337 122
1,249 360 198
1,156 333 187
114,935 33,301 28,799
1,087 313 272
32,856 7,133 10,577
1,159 252 373
Table 4-5 gives an overview of the total numbers of instances of the three verbs in the corpora used in the study.38 The overview in Table 4-5 alone bears some interesting implications for the distribution of the three verbs across the two varieties and the differences between the corpora, which are mainly differences of mode and genre.39 If we consider the left-hand side of the table, which displays the distribution of GIVE, SEND and OFFER in ICE-India and ICEâ•‚GB, there are clear quantitative differences in the use of these verbs in the two varieties. GIVE is used much more frequently in the Indian data, while the normalized distribution of SEND is roughly equal and OFFER is more frequently used in ICEâ•‚GB. A more finegrained picture is displayed in Table 4-6, where the different distribution of the verbs under scrutiny is shown according to the number of occurrences in the written and spoken parts of the ICE corpora. With regard to GIVE, Table 4-6 shows that it is used significantly more frequently in the spoken part of ICE-India than in the written part, while in ICEGB, although GIVE is also used slightly more frequently in the spoken part, this difference is not statistically significant. Comparing the spoken and written use of GIVE across the varieties, it can additionally be seen that GIVE is used much more frequently in the spoken part of ICE-India than in the spoken part of ICEGB and also more frequently in the written part of ICE-India compared with the written part of ICE-GB. However, here the discrepancies are much larger in the spoken mode. With regard to the data from the purely written newspaper corpora (Times of India and BNC periodicals), it should be pointed out, however that here GIVE is used significantly less frequently in the Indian corpus (G2 = 102.87). 38. Note that at this point particle verbs and non-verbal uses have not been discarded. 39. Note here that the figures given for OFFER in the Times of India corpus are estimates. Due to homography of the nominal singular with the verb form offer and the nominal offering with the continuous form of the verb, these had to be manually extracted from the concordances. Since a sum total of more than 30,000 concordance lines is unwieldy for manual deletion of these nominal forms, a sample of 200 lines for each of the potential homograph sets were analyzed and the percentages of the nominal forms were extrapolated from the total number of instances.
59
60 Structural Nativization in Indian English Lexicogrammar
Table 4-6.╇ Distribution of GIVE, SEND and OFFER in the spoken and written sections of ICE╂India and ICE╂GB40 Verb (lemma) ICE-India spoken ICE-India written ICE-GB spoken GIVE SEND OFFER
ICE-GB written
total
pmw
total
pmw
total
pmw
total
pmw
1,382 147 54
2045.0 217.6 79.9
609 233 82
1343.0 513.8 180.8
748 208 81
1173.0 326.2 127.0
501 152 117
1131.0 343.1 264.0
At this point, the reasons for this higher frequency of GIVE in ICE-India compared with ICE-GB are largely unclear, but the fact that GIVE is used significantly more frequently in the spoken mode within ICE-India may lead to the assumption that speakers of Indian English use GIVE in a variety of cases where speakers of British English would use a near-synonym, such as HAND or PASS. Consider Example (19), where a speaker of British English might prefer hand in: (19) And I completely forgot that we had to give in the literature paper tomorrow.
Hence, under the pressure of online speech production in particular, there may be a tendency for Indian English speakers to use GIVE as a sort of ‘lexical teddy bear’ (cf. Hasselgren 1994). With regard to SEND, the picture is quite different. In ICE-India there is a significant tendency to use SEND more frequently in written language than in spoken language, whereas in ICE-GB there is no significant difference concerning the use of SEND between the modes. Furthermore, in the comparison across the varieties, SEND is used significantly less frequently in spoken Indian English compared with spoken British English. At the same time, in the written part of ICE-India, SEND is used significantly more frequently than in the written part of ICE-GB. Thus, SEND differs from GIVE in being used more frequently in written Indian English, compared to spoken Indian English as well as written British English. As with GIVE, the tendencies observed in the case of SEND in the ICE corpora are also reflected in the written newspaper corpora, where SEND is used significantly more frequently in the Indian than in the British data (G2 = 155.86). OFFER differs from GIVE and SEND by being used less frequently in ICEIndia than in ICE-GB in both the spoken and the written component. Comparing the use of OFFER in the spoken and written parts of ICE-India and ICE-GB, it can be shown that OFFER is used significantly less frequently in the spoken part 40. Per million word (pmw) freuquencies are an extrapolation, since the ICE-corpora only contain about 600,000 words of spoken and 400,000 words of written language.
Chapter 4.╇ Methodology
of ICE-India compared to the written part of the same corpus. This also holds true for the comparison of OFFER across the modes in ICE-GB. Thus, OFFER is used significantly more frequently in the written mode in both varieties of English. Comparing the use of OFFER across the varieties, OFFER is used significantly less frequently in Indian-English in the spoken parts and in the written parts of the corpora, without marked differences between the modes. As with GIVE and SEND, the tendencies of use in spoken and written Indian and British English derived from the ICE corpora are in line with the observations about the frequencies in the purely written newspaper corpora. Again, the less frequent use of OFFER in Indian English can also be claimed for the Times of India corpus compared to the BNC periodicals with a high level of significance (G2 = 726.69). Table 4-7 summarizes the above results by displaying the different corpora that have been compared. In the tendency column, the direction of higher or lower frequency is displayed by the use of > and <.41 Table 4-7.╇ Comparison of GIVE, SEND and OFFER across modes and varieties42 Item
Corpus 1
Tendency
Corpus 2
G2
sig. level
GIVE GIVE GIVE GIVE SEND SEND SEND SEND OFFER OFFER OFFER OFFER
ICE-India spoken ICE-GB spoken ICE-India spoken ICE-India written ICE-India spoken ICE-GB spoken ICE-India spoken ICE-India written ICE-India spoken ICE-GB spoken ICE-India spoken ICE-India written
> NA > > < NA < > < < < <
ICE-India written ICE-GB written ICE-GB spoken ICE-GB written ICE-India written ICE-GB written ICE-GB spoken ICE-GB written ICE-India written ICE-GB written ICE-GB spoken ICE-GB written
â•⁄ 78.00 â•⁄â•⁄ 0.11 155.86 â•⁄â•⁄ 8.17 â•⁄ 68.91 â•⁄â•⁄ 0.22 â•⁄ 14.35 â•⁄ 12.82 â•⁄ 22.33 â•⁄ 26.21 â•⁄â•⁄ 7.11 â•⁄â•⁄ 7.02
p ≤ .0001 p ≥ .05 p ≤ .0001 p ≤ .01 p ≤ .0001 p ≥ .05 p ≤ .001 p ≤ .001 p ≤ .0001 p ≤ .0001 p ≤ .01 p ≤ .01
41. The column sig. levels gives the level of statistic significance according to log-likelihood values (G2); p > .05 implies that no significance can be stated (probability of error is larger than€5%). 42. For a detailed discussion of log-likelihood ratios cf. Dunning (1993). The following thresholds for the different levels of statistic significance apply: p ≤ .05: G2 ≥ 3.84; p ≤ .01: G2 ≥ 6.63; p ≤ .001: G2 ≥ 10.83; p ≤ .0001: G2 ≥ 15.13.
61
62
Structural Nativization in Indian English Lexicogrammar
4.6
Summary
In this chapter I laid the foundations for a quantitative and qualitative analysis of trivalent verbs that are chosen on the grounds of their frequency. In order to carry out this analysis, the British and Indian component of the International Corpus of English (cf. Section 4.2.1), as well as a very large internet-derived database of Indian English newspapers, the Times of India corpus (cf. Section 4.2.2), will be used. Since the different corpora differ in terms of corpus annotation, they have been converted to a plain-text format and cleaned from additional mark-up so that they can be comparably processed by the Wordlist and Concord tool of the corpus software bundle Wordsmith Tools V4.x (cf. Section 4.2.3). Initially a verb-complementational profile for each of the verbs that are potentially used in both the ditransitive construction and the transfer-caused-motion construction is created according to a slightly revised version of the framework proposed by Mukherjee & Hoffmann (2006) (cf. Section 4.3). These varietyÂ�specific verb-complementational profiles are compared in order to find out if significant differences for the complementation of the verbs exist between the varieties. As many factors seem to underlie the selection of possible patterns of verbcomplementation, one of these factors, namely collocation, has been selected as a possible influential factor for pattern selection. The definition of collocation in this work is to some extent restricted, as only collocates of the verbs that are found in the direct object slot of the verbal phrase are taken into consideration, and a number of statistic conditions must be met. These conditions allow a comprehensive profile to be created regarding Od-collocates for each of the verbs in both varieties (cf. Section 4.4). Since it is assumed that specific collocates of different trivalent verbs will correlate with the choice of construction that the verb is used in, additional verb-complementational profiles for the specific verb-Od-collocate collocations are created in order to attest for possible correlation. Thus, potential differences between the varieties can be identified according to the overall picture of verb-complementational profiles and differences in collocational profiles, as well as on the basis of a combination of both. The final part of Chapter 4 covered an overview of the four ditransitive verbs under scrutiny – GIVE, SEND and OFFER –, their respective semantics and their distribution in the underlying corpora. Here also a differentiation between the different uses according to the written and spoken modes was presented. The following chapters build on these basic observations by giving a detailed “thick description” for each of these verbs.
chapter 5
GIVE
5.1
Introduction: GIVE as a prototypical ditransitive verb
GIVE is the most frequent verb that is used in what Mukherjee (2005) describes as “[…] ‘explicit ditransitive syntax’: both objects are made explicit, with the indirect object as a noun phrase and the direct object as a noun phrase or a clause” (Mukherjee 2005:â•›79). Owing to this frequency GIVE can be called a prototypically ditransitive verb, although from a purely cognitive perspective, other verbs that profile a more direct transfer process, such as HAND, may also be argued to be among the most central ditransitive verbs. As mentioned above, the transfer scheme profiled by GIVE can be realized in a number of different syntactic ways, most notably the options offered in the so-called dative-alternation, or – in Goldberg’s (1995) terms – the ditransitive and transfer-caused-motion constructions. Furthermore, with GIVE the recipient in the transfer process is frequently not explicitly profiled, for example, when the information is accessible from the context. It is only rarely the case that the transferred entity is not profiled, so that only the indirect object or no object at all is profiled. Mukherjee & Hoffmann (2006) summarize these possibilities of different patterns by using a classification system that relates to the systemic-functional as well as the construction grammar approach. Patterns are formulated that are defined by the formal realization of the clause in which the verb is used. Table 5-1 summarizes these different patterns for the use of GIVE in ICE-GB and ICE-India. The figures in Table 5-1 differ somewhat from the figures in Table 4-5, because Mukherjee & Hoffmann (2006) discarded phrasal and prepositional verbs such as GIVE up or GIVE away from the original dataset represented by Table 4-5. Since these verbs are associated with different semantics from GIVE, this practice will also be adopted for the present study. Furthermore, it can be inferred from Table 5-1 that, besides the basic patterns (type-I to type-V), GIVE also functions in a number of structurally related patterns, for example in relative clauses or with fronted elements. Since these patterns are relatively infrequent, and a distinctive analysis with regard to questions of pattern selection and its correlation to Od-collocates is not likely to yield salient results, in the following these patterns will be collapsed as derivative patterns type-Ider, type-IIder etc.
64 Structural Nativization in Indian English Lexicogrammar
Table 5-1.╇ Complementation of GIVE in ICE-India and ICE-GB (cf. Mukherjee & Hoffmann 2006:╛172) Type
Pattern
ICE-India sum
I Ia Ib Ic Id
(S) GIVE [Oi:NP] [Od:NP] (S) GIVE [Od:NP] [Oi:NP]
[Od:NPantecedent] (rel. pron.) [S] GIVE [Oi:NP] [Oi:NPantecedent] (rel. pron.) [S] GIVE [Od:NP] [Od: NPfronted] [S] GIVE [Oi:NP] miscellaneous
IP
[S < Oi active] BE given [Od:NP] (by-agent)
IPb
[Od:NPantecedent] (rel. pron.) [S < Oi] BE given (by-agent)
II
(S) GIVE [Od:NP] [Oi:PPto]
miscellaneous IIa IIb IIc
(S) GIVE [Od:NP] [Oi:PPfor]
[Od:NPantecedent] (rel. pron.) [S] GIVE [Oi:PPto] (S) GIVE [Oi:PPto] [Od:NP] miscellaneous
IIP
[S < Od active] BE given [Oi:PPto] (by-agent)
IIPb
[antecedent]co (S < Od)co (BE) given [Oi:NPto] (by-agent)
III
(S) GIVE [Od:NP] Oi
miscellaneous
IIIb
[Od:NPantecedent] (rel. pron.) [S] GIVE Oi
%
407 22.6
ICE-GB sum
%
404 38.0
3
0.2
1
0.1
14
0.8
23
2.2
1
0.1
2
0.2
3
0.2
1
0.1
11
0.6
10
0.9
130
7.2
84
7.9
3
0.2
12
1.1
4
0.2
0
0.0
310 17.3
123 11.6
5
0.3
4
0.4
12
0.7
7
0.7
4
0.2
2
0.2
6
0.3
6
0.6
70
3.9
23
2.2
19
1.1
17
1.6
2
0.1
2
0.2
528 29.4
247 23.2
17
0.9
16
25
1.4
3
0.3
123
6.8
38
3.6
IIIPb [antecedent]co (S < Od)co (BE) given Oi (by-agent)
49
2.7
28
2.6
miscellaneous
10
0.6
0
0.0
IV
(S) GIVE Oi Od
miscellaneous IIIP
IVP V
[S < Od active] BE given Oi (by-agent)
1.5
24
1.3
10
0.9
miscellaneous
0
0.0
1
0.1
given Oi Od (by-agent)
5
0.3
0
0.0
6
0.3
0
0.0
6 0.3 1,797 100 < 0.001
0 1,064
0.0 100
(S) GIVE [Oi:NP] Od
miscellaneous sum total p
As can be seen in Table 5-1, there are significant differences in the choice of the complementation patterns between the varieties, especially when it comes to the type-III pattern that is used much more frequently in ICE-India than in ICEGB. This and the other differences in pattern selection for GIVE with regard to
Chapter 5.╇ GIVE
Table 5-2.╇ Complementation of GIVE in ICE-India and ICE-GB – focus on central patterns Type I Ider IP IPder II IIder IIP
Pattern
ICE-India
ICE-GB
sum
%
sum
%
(S) GIVE [Oi:NP] [Od:NP]
407
22.6
404
38.0
32
1.9
37
3.5
[S < Oi active] BE given [Od:NP] (by-agent)
130
7.2
84
7.9
7
0.4
11
1.0
(S) GIVE [Od:NP] [Oi:PPto]
310
17.3
123
11.6
all related to type-II
27
1.5
19
1.9
[S < Od active] BE given [Oi:PPto] (by-agent)
all related to type-I
all related to type-IP
70
3.9
23
2.2
IIPder
all related to type-IIP
21
1.2
19
1.8
III
(S) GIVE [Od:NP] Oi
528
29.4
247
23.2
42
2.3
19
1.8
[S < Od active] BE given Oi (by-agent)
123
6.8
38
3.6
IIIder IIIP
all related to type-III
IIIP der all related to type-IIIP
59
3.3
28
2.6
IV
24
1.3
11
1.0
5
0.3
0
0.3
6
0.3
0
0.0
6 0.3 1797 100 < 0.001
0 1063
0.0 100
IVP V
(S) GIVE Oi Od
given Oi Od (by-agent) (S) GIVE [Oi:NP] Od
Vder all related to type-V sum total p
the different varieties will be discussed in detail in the following. Table 5-2 shows the distribution of GIVE according to pattern and variety with a focus on the central patterns, collapsing all derivative patterns into single categories. In terms of a chi-square test of Table 5-2, the distribution is highly significant (χ² = 125.34; critical value for p ≤ 0.001 with df = 15 is 37.7). However, as Table€5-2 contains a number of cells where the expected values are below 5, this result has to be taken with a certain measure of caution. The relevant cells in this context are the marginal patterns type-IVP, type-V and type-Vder. Although they were kept in Table 5-2, I collapsed all the marginal patterns (including the type-IV pattern) into a single category for an additional statistic analysis. Here the contingency table only contains 13 instead of 16 patterns (thereby lowering the number of rows to 13 and thus df to 12). Without these problematic cells, the distribution is still highly significant (χ² = 120.50; critical value for p ≤ 0.001 with df€= 12 is 32.9), so it seems safe to assume that collapsing Mukherjee & Hoffmann’s (2006) original categories does not have an influence on the significance of the distribution.
65
66 Structural Nativization in Indian English Lexicogrammar
Table 5-3.╇ Distributional differences of GIVE: A chi-square test Type I Ider IP IPder II IIder IIP IIPder III IIIder IIIP IIIPder others sum total
ICE-India observed 407 32 130 7 310 27 70 21 528 42 123 59 41 1797
ICE-India expected 509.57 43.35 134.46 11.31 272.06 28.90 58.43 25.13 486.95 38.33 101.16 54.66 32.67 1796.98
χ²
ICE-GB observed
20.65 2.97 0.15 1.64 5.29 0.13 2.29 0.68 3.46 0.35 4.72 0.34 2.12 44.79
404 37 84 11 123 19 23 19 247 19 38 28 11 1063
ICE-GB expected 301.43 25.65 79.54 6.69 160.94 17.10 34.57 14.87 288.05 22.67 59.84 32.34 19.33 1063.02
χ² 34.90 5.03 0.25 2.78 8.94 0.21 3.87 1.15 5.85 0.59 7.97 0.58 3.59 75.72
Table 5-3 shows how the distributional differences for some complementation patterns of GIVE are more influential in terms of the total distributional difference than others. The chi-square test in Table 5-6 shows how some patterns in the different varieties are used with a high degree of variation, resulting in high partial χ²-values. Others display less variation and, thus, have a much smaller impact on the total χ²-value. For example, the more frequent proportional use of the type-I pattern in ICE-GB, which results in a partial χ²-value of 34.9, alone surpasses the critical χ²-value that needs to be reached for the distribution to be significant. Although these figures by themselves do not provide us with any information about the actual differences of the use of GIVE in the different complementation patterns, let alone the underlying reasons for them, they may nevertheless serve as a valuable starting point, because they show where the largest differences can be found and, thus, where lexicogrammatical variation is most likely to be identified. The focus will, therefore, be more on these patterns in the following.
5.2
Distribution of complementation patterns in different text categories
As discussed above, there are marked differences between the use of GIVE in the spoken and written components of ICE-India compared with ICE-GB. In
Chapter 5.╇ GIVE
ICE-India, GIVE is used significantly more frequently in spoken text types, both compared with ICE-GB and with the written part of ICE-India. This observation along with earlier findings about variation according to variety and text type (cf. Sand 2002) led to the decision to sketch a more detailed picture of these distributional variations. While the data Mukherjee & Hoffmann (2006) present give a good overview on variation between the varieties, text type-dependent variation is not included in their figures. Therefore, in the following, the occurrences of GIVE in the different complementation patterns will be related to their occurrence in different text types. This is carried out based on the text category system provided by the ICE corpora (cf. Table 4-1). Table 5-4 provides an overview of the differences in the complementation of GIVE according to variety and text type. After this introductory overview, a more detailed examination of the distributional differences in the spoken and written modes and the different text types will be provided. As can be inferred from Table 5-4, there are prominent differences in the usage of GIVE in the different text categories, both between the varieties and within variety-specific corpora.43 However, since the data in this table only shows raw frequencies and does not provide a more detailed picture, it is useful to analyse the usage of GIVE in the different categories in more detail. As in the ICE corpora, not all categories have the same size; the differences shown above also have a different influence on the total results, so strong distributional differences in a small category may have less influence on the cumulative figures than small differences in a large category.44 The strong differences in the use of GIVE between the spoken and written modes of Indian and British English will be used as a starting point for the ensuing description of the complementation of GIVE. Therefore, I will set out to analyze the different complementation tendencies with regard to their use in the different modes and provide a more detailed examination of some differences in the specific registers and text types.
43. The numbers given in Table 5-4 are slightly different from those provided by Mukherjee & Hoffmann (2006). There may be several reasons for this: different usages may have been sifted out as self-corrections, some particle verbs may have been treated differently and the classification of more than 3,000 concordance lines may have led to erroneous results in some cases. 44. The differences in category size are intended to be representative of the use of English. As with all matters of corpus design, this assumption should, of course, be taken with a grain of salt, especially when native varieties of English are compared with second language varieties of English. In the latter, some categories might be of different importance compared with British English, which served as a model for the corpus design.
67
68 Structural Nativization in Indian English Lexicogrammar
Table 5-4.╇ Distributional differences of GIVE according to text types ICE-India Cat.
S1A S1B S2A S2B W1A W1B W2A W2B W2C W2D W2E W2F sum
Type I
I der
I P
IP der
II
II der
II P
87 76 77 35 18 31 3 21 9 6 3 39 405
4 8 4 3 0 2 0 0 1 0 0 2 24
21 30 13 15 9 5 9 10 7 0 2 1 122
1 2 0 0 1 0 0 0 0 1 0 0 5
44 50 55 36 24 18 16 19 14 4 4 11 295
4 8 5 3 0 1 1 0 0 0 0 0 23
15 23 16 16 4 5 2 1 2 2 1 0 86
IIP der
III
III der
III P
IIIP der
IV
V
0 127 6 128 0 98 0 30 0 29 1 29 0 27 3 30 0 18 0 18 1 4 0 7 11 540
11 9 5 2 0 2 0 0 0 0 0 0 29
28 35 14 21 4 5 28 7 4 4 1 0 147
3 4 4 2 1 7 8 3 1 1 0 2 35
9 2 4 0 0 0 1 1 0 0 0 0 17
6 3 0 0 0 0 0 0 0 0 0 0 9
ICE-GB Cat.
S1A S1B S2A S2B W1A W1B W2A W2B W2C W2D W2E W2F sum
Type I
I der
I P
IP der
II
II der
II P
IIP der
III
III der
III P
IIIP der
IV
V
103 100 60 23 7 34 12 28 18 16 4 31 436
8 6 2 1 0 3 1 0 0 2 0 1 24
8 14 10 12 9 6 4 6 7 9 1 5 85
1 0 1 2 0 0 0 1 0 2 0 0 7
21 25 15 15 10 8 5 10 4 4 5 5 127
4 4 5 2 0 3 1 0 0 1 0 0 20
0 5 6 4 0 1 2 3 1 5 1 0 28
0 1 5 2 0 8 0 2 1 2 0 1 15
18 36 64 25 13 8 22 34 4 15 3 9 251
3 9 2 2 0 0 0 1 0 0 1 1 19
1 2 7 9 2 9 6 1 1 11 1 0 50
0 5 1 2 1 4 1 2 1 7 0 0 24
4 3 0 0 3 0 0 0 0 0 0 1 11
0 0 0 0 0 0 0 0 0 0 0 0 0
5.3
Chapter 5.╇ GIVE
Verb-complementational patterns of GIVE in spoken Indian and British English
The ICE corpora contain about 600,000 words of spoken English, organized into a dialogue section and a monologue section. The dialogue section contains 100 texts of private dialogue and 80 texts of public dialogue (each text containing 2,000 words). The monologue section is subcategorized into two sections containing unscripted monologues (70 texts) such as commentaries and scripted monologues (50 texts), e.g. news broadcasts. Table 5-5 shows the distribution of the central complementation patterns of GIVE in the spoken components of ICE-India and ICE-GB. In Section 4.5 I already pointed out that the verb GIVE is used much more frequently in Indian English. The figures provided at that point, however, did include a number of instances that have been excluded from the present analysis. Especially particle verbs, the use of given as a preposition and repetitions and selfcorrections have been discarded for Table 5-5. Nevertheless, GIVE is used almost twice as frequently in ICE-India compared to ICE-GB. Table 5-5.╇ Complementation of GIVE in ICE-India and ICE-GB – spoken data Type
Pattern
ICE-India sum
I Ider IP IPder II IIder IIP IIPder III
(S) GIVE [Oi:NP] [Od:NP]
%
ICE-GB sum
% 43.5
275
22.90
286
all related to type-I
19
1.60
17
2.6
[S < Oi active] BE given [Od:NP] (by-agent)
79
6.60
44
6.7
all related to type-IP
(S) GIVE [Od:NP] [Oi:PPto]
3
0.30
4
0.6
185
15.40
76
11.6
all related to type-II
20
1.70
15
2.3
[S < Od active] BE given [Oi:PPto] (by-agent)
70
5.80
15
2.3
all related to type-IIP
6
0.50
8
1.2
(S) GIVE [Od:NP] Oi
383
31.90
143
21.7
IIIder
all related to type-III
27
2.30
16
2.4
IIIP
[S < Od active] BE given Oi (by-agent)
98
8.15
19
2.9
all related to type-IIIP
13
1.10
8
1.2
(S) GIVE Oi Od
15
1.30
7
1.0
8
0.70
0
0.0
1 0.08 1,202 100.00 < 0.001
0 658
0.0 100.0
IIIPder IV V
(S) GIVE [Oi:NP] Od
Vder all related to type-V sum total p
69
70 Structural Nativization in Indian English Lexicogrammar
In Section 4.5 I suggested a hypothesis for the reasons for the more frequent overall usage of GIVE in Indian English, namely that there may be a tendency to use GIVE in circumstances where speakers of British English might prefer a different verb. In fact, there are a number of examples that are in accordance with that assumption, as shown in (20) to (23): (20) Then you just give a topping you give uh cheese (21) They will give horrible trouble and it would be (22) […] one year British Government refused to give oil (23) First of all the board will give an advertisement in the paper […]
In Example (20), speakers of British English would not use the phrase give a topping but possibly prefer add a topping. Similarly in Example (21), speakers of British English would probably prefer the expression make trouble to give trouble. In Example (22), instead of the more general verb give, a more specific verb like deliver or provide may be the preferred British variant, whereas in Example (23) put an advertisement in the paper is likely to be the British variant. In some instances in ICE-India, GIVE is also used as a general verb and then replaced by a more specific verb by means of self-correction, which also seems to be in line with the assumption that GIVE is often used as a versatile alternative to – and stands proxy for – other more specific verbs. Consider Example (24): (24) And those that party which never gave money uhm gave shelled out money like that
In Example (24) we can see how GIVE is initially used in the Type-III pattern gave money, then, after a brief filled pause, the verb is repeated, to be finally replaced with shelled out.45 Besides the more frequent use of GIVE in Indian English for the above reasons, different tendencies for the choice of specific patterns can also be shown.
45. This example also shows why different researchers might get different total results, as discussed above. The first instance of give money could be classified as complementation in the Type-III pattern and the other corrected instance could be neglected. However, it might also be possible to neglect both instances because of self-correction or because the final form is a particle-verb.
Chapter 5.╇ GIVE
In ICE-India, the most frequent complementation pattern for GIVE is the type-III pattern, where the indirect object is not made explicit. This pattern occurs 383 times, or in 31.9% of all instances, compared to 143 occurrences (or in 21.7%) in ICE-GB. The examples below may also shed some light on the preference for this pattern in ICE-India. Especially in cases like (27), there is no genuinely admissible possibility to use a pattern with an explicit indirect object. Besides these general tendencies there are also, however, some very interesting text-type specific examples of a nativized use of GIVE in Indian English in the type-III pattern. Some very prominent examples of this can be found in the legal texts of section S1B (public dialogues / cross examinations) and S2A (unscripted monologues / legal presentations): Consider Examples (25) to (28). (25) He has given the history of taking alcohol when <,> he was examined by the doctor (26) Mr Angale while giving the evidence give small sentences because it is to be typed (27) As Sir <,> Your Honour will will see Sir <,> here he has given <,> that accused number one <,> assaulted <,,> Iyengar <,> on chest <,,> stomach chest <,> hands and legs <,> (28) P W number eight is Doctor Balkrishna krishna Deshpande who performed the <,> post-mortem <,> He has given all the injuries on the person of the deceased 46
In all of these instances, GIVE is used in the meaning of giving evidence before a court. Thus, the indirect object is implicitly provided by the discourse situation. Furthermore, GIVE is often not used in combination with the collocate evidence explicitly mentioned, as this combination is also understood implicitly. These sentences suggest that if somebody gives something in court they make an official statement. Thus, in a highly formalized register, GIVE is used in a routinized way untypical of British English. The type-III complementation pattern has been shown to be used much more frequently in Indian English than in British English, to some extent for the reasons described above. Firstly, it is used as a sort of “teddy bear expression” (cf. Hasselgren 1994) when speakers of other varieties of English may prefer different,
46. These examples are not an exhaustive list. In the 20 legal texts (40,000 words) there are 62 occurrences of GIVE in the type-III pattern with similar meaning (16.2 % of GIVE in the typeIII pattern in the spoken part of ICE-India). Sometimes GIVE is used in light-verb constructions, such as GIVE explanation or GIVE statement, but in many cases the eventive noun of the construction is not made explicit, as in Example (31) below.
71
72
Structural Nativization in Indian English Lexicogrammar
Table 5-6.╇ Pronominal Oi in spoken GIVE type-I patterns Pronoun
her
ICE-India ICE-GB
7 7
him him-self 26 19
0 1
it
me
12 22
46 49
my-self them them-selves 0 1
33 16
0 1
us
you total
39 31
86 92
249 239
more specific verbs. Secondly, it is frequently employed in nativized text typeÂ�specific uses, such as, for example, in legal language. In summary, in the spoken part of ICE-India, 275 instances of the type-I pattern are attested (22.9% of all instances), while in ICE-GB 286 instances can be found (43.5% of all instances).47 Particularly in the dialogue sections (S1A and S1B) this pattern seems to be the preferred way of using GIVE in British English with 103 and 100 instances respectively, compared to, for example, only 18 and 36 instances of type-III complementation. In both corpora the type-I pattern is used with the indirect object realized as a pronoun in the vast majority of the cases. In the spoken part of ICE-India about 90% of the sentences complemented in the type-I pattern have a pronominal indirect object; in ICE-GB the figure is slightly lower (approx. 83%). The choice of the pronominal indirect object is also relatively stable between the varieties, as shown in Table 5-6.48 Thus, it can be pointed out that the type-I pattern for GIVE is used with almost the same frequency in the 600,000 words of spoken English in both varieties. Further, the Oi is typically profiled by a pronoun in both varieties with only slight differences in the choice of the respective pronoun. While the type-I pattern is evenly distributed across the varieties, it is the more frequent use of the type-III and type-II patterns in Indian English that make the type-I pattern less prototypical of GIVE in Indian English than in British English. As in the case of the type-III pattern, the type-II pattern is also used much more frequently in Indian English than in British English. As shown in Table 5-7, with 185 instances it is used more than twice as often in the spoken part of ICEIndia compared to ICE-GB with 76 instances. In relative terms, this discrepancy is not so marked with 15.4% of the total occurrences of GIVE compared to 11.6% of the total occurrences. In ICE-GB the type-II pattern is often used in relatively fixed expressions, such as give way to (12 times) and give rise to (6 times), whereas in ICE-India the expression give way to is not attested, though give rise to (11€times) 47. This discrepancy between raw and relative figures also influences the high chi-square values shown in Table 5-6. 48. Note that there are 13 occurrences of giving with the Oi â•›you in ICE-India due to the phrase “This is All India Radio giving you the news” that precedes many of the news broadcasts in the corpus.
Chapter 5.╇ GIVE
is relatively frequent here, too. However, by looking at the relative frequencies for both of these expressions, we see that they account for about 24% of all type-II pattern instances in ICE-GB and only for approximately 6% of all type-II pattern instances in ICE-India. This is due to the non-existence of the expression give way to in the Indian corpus, but it may also lead to the assumption that this pattern has a different productivity in Indian English than in British English. Similar to the use of the type-III pattern in legal texts, the type-II pattern is very productively used in Indian English in the section concerning sports commentary (S2A texts 1–20). Out of a total of 55 occurrences of the type-II pattern in the S2A category, 32 are found in the 20-text subcategory that contains spontaneous commentary. In 20 out of these 32 examples, the referent of GIVE is a ball being passed on. Compared with ICE-GB, the use of GIVE in these circumstances is much rarer here, with 8 occurrences in the sports commentary section, only two of which refer to playing or passing a ball. In ICE-GB, the verb that is most frequently used in this context is PLAY, which occurs 41 times in the respective section, referring to a ball being passed on. At the same time, in the commentary section of ICE-India, PLAY is only used 6 times in this specific setting. So in this case GIVE seems to be a genuine alternative to PLAY in Indian English, while in British English, albeit not incorrect, GIVE is used much less frequently in the type-II pattern to encode the transfer of a ball in sports commentary. Thus, for the type-II pattern in spoken English in the ICE corpora, GIVE is used to a larger extent in formulaic sequences like give way to in British English but used less frequently in direct transfer actions, especially in sports commentaries. Apart from the differences in the three central patterns for the complementation of GIVE in the two varieties, there are also some differences in the less central patterns, especially in the passive patterns. While the type-I passive pattern, typeIP, is used in relative terms equally frequently in the two varieties (79 occurrences in ICE-India to 44 occurrences in ICE-GB, 6.5% and 6.7% of all the occurrences in the respective variety), there is a certain amount of variation in the use of the type-IIIP pattern and the type-IIP pattern. Besides being used more frequently overall in ICE-India compared with ICEGB, the type-IIIP pattern is also found to a much larger extent in the spoken part of ICE-India. There are in total 98 occurrences of GIVE in the type-IIIP pattern in the spoken part of ICE-India compared to 19 in the spoken part of ICE-GB. This pattern is also not used as frequently in the written part of ICE-India (49 occurrences), whereas in ICE-GB the pattern is used more frequently in written English (31 occurrences). The reason for the frequent use of GIVE in the type-IIIP pattern in Indian English is comparable to the frequent use of the type-III pattern described above. GIVE is used in a wider range of settings and with direct object collocates that do not collocate with the verb in British English, a point that I will
73
74
Structural Nativization in Indian English Lexicogrammar
return to later. It is also used in light-verb constructions in cases where speakers of other varieties might prefer simplex verbs. Consider Example (29): (29) Thereafter she was discharged she was given discharge
This example shows a self-correction where first the semantically heavy verb discharge is used to be replaced later with the light-verb-construction give discharge, where give is used as the light – in the sense of semantically empty – verb and discharge as an eventive noun. Interestingly, the self-correction is not being performed the other way around: the speaker seems to be more comfortable with the light-verb construction than with the simplex verb, a fact that hints at a certain degree of nativization of such light-verb constructions for GIVE. Examples (30) to (33) also show uses of GIVE in the type-IIIP pattern which are typical of the use of GIVE in Indian English, but also contain many other typical features of this variety that seem worth pointing out: (30) So whereas if free atmosphere is given then you will come with new ideas no? (31) […] more than thirty uh three hundred crore rupees have been uh given as loan and not a single money has been recovered from the uh people who got it as loan (32) And I think uhm perhaps you could throw some light on the fact that you see Jammu and Kashmir was treated on a seperate footing and the certain assurances that were given which were legal enshrined in a special status for Jammu and Kashmir (33) But uh it is also important to remember here that an assurance was given by the uh government in New Delhi <,>
Example (30) is part of a discussion about the innovative power of societies with regard to their political and social system. Atmosphere therefore relates to the social climate or the status of democratization of a society. One typical feature of Indian English in Example (30) is the use of atmosphere without a determiner. Secondly, although the indirect object, namely the people in the society, or the general you in the second clause, is implicit, its omission as well as the passivization of the sentence in general would presumably not be the choice of a speaker of British English. If such a speaker talked about atmosphere in this case at all, an active sentence with a different verb, such as, for example If you live in a free atmosphere then you will come up with new ideas, won’t you would be the more likely choice. Example (30) also displays two other typical Indian English features,
Chapter 5.╇ GIVE
namely the different use of particle verbs, here realized by the omission of the particle up in come up and the invariant question tag no. Example (31) is not as marked as Example (30) in the choice of the type-IIIP pattern. However, the filled pause before the verb indicates that the speaker briefly thinks about the appropriate verb. This may be further evidence that speakers of Indian English prefer GIVE over more specific verbs when under time pressure, even when the sentence does not comply to the basic ditransitive sentence pattern that is the preferred pattern for this verb in British English.49 Both Examples (32) and (33) have implicit direct objects and are to a certain extent in accordance with the observations concerning the use of GIVE in legal proceedings described for the type-III pattern above and the preferred use of light-verb constructions exemplified in (29). In both cases the direct object is assurance or assurances. The collocation of GIVE with assurance is also found in ICE-GB, albeit less frequently (4 times in ICE-GB, 9 times in ICE-India). In ICEGB, however, sentences where GIVE co-occurs with assurance are usually complemented in the type-I pattern. Furthermore, the only use of the type-IIIPâ•›der pattern occurs in the written part of the ICE-India, whereas the type-IIIP pattern in this collocation is found in both parts of the corpus. This pattern is even used more frequently in the spoken part (5 times spoken, 2 times written), showing that Indian speakers prefer that phrase in speech, while British speakers use it exclusively in writing. The last basic pattern that will be covered in this section is the passive pattern where the indirect object is realized as a to-prepositional phrase, the type-IIP pattern. Again, this pattern is used significantly more frequently in the spoken part of ICE-India than in the spoken part of ICE-GB. In ICE-India the pattern is used 70 times, while in ICE-GB it only occurs 15 times. The distribution across the text categories is relatively stable in ICE-India, with the exception of spoken public dialogue (S1B) (cf. Table 5-5). While in ICE-GB this pattern is rare in spoken and written modes, in ICE-India it is much more frequent in the spoken than in the written mode (17 occurrences). However, regarding the type-IIP pattern, no clear trends for the choice of this pattern apart from its frequency of use can be derived from the data. The direct objects have a certain tendency to be concrete rather than abstract, and in some cases it is possible to assume that British speakers 49. There are two further lexical features of Indian English, which, although not exerting influence on verb-complementation, would be worth pointing out here. Firstly (as has been demonstrated in many earlier studies of Indian English), when citing numbers, especially in financial sums, speakers of Indian English very often use the Indian system of crores (10,000,000) and lakhs (100,000); in this example two hundred crore rupees. Secondly, the non-count noun money is used as a count noun in this sentence: not a single money.
75
76
Structural Nativization in Indian English Lexicogrammar
would choose a more specific verb. This, as already mentioned above, also leads to the higher overall frequency of GIVE. However, these cases are relatively rare, so, in effect, apart from the greater tendency to use GIVE in this pattern, no specific Indian uses were found. In summary, this section has shown that GIVE is used more frequently in spoken Indian English overall and that it is complemented in a wider range of patterns. The type-I pattern, the favourite complementation pattern of British speakers, is used to roughly the same extent and in this pattern the indirect object is usually realized by a pronoun in both varieties. The type-II pattern in British English is used frequently in fixed expressions such as give way to, whereas speakers of Indian English use this pattern more frequently in situations of actual transfer, especially in sports reportage when players pass a ball. The type-III pattern is the most frequent form of verb-complementation of GIVE in the spoken part of ICEIndia; British speakers use this pattern to a much lesser extent. This frequency is partly due to the use of GIVE where speakers of British English may opt for a more specific verb, as well as the use in light-verb-constructions where a simplex verb would be an alternative. Furthermore, in the case of legal Indian English it was shown that GIVE in the type-III pattern seems to be a routinized expression for making official statements in court. The passive patterns type-IIP and type-IIIP were also used significantly more frequently in spoken Indian English than in the British variety. In the case of the type-IIIP pattern, the use of GIVE in light-verb constructions was also salient, whereas no specific explanations for the frequent use of the type-IIP patterns could be provided.
5.4
Verb-complementational patterns of GIVE in written Indian and British English
The written component of the International Corpus of English is smaller than the spoken component and contains about 400,000 words of written English. It is divided into 8 different text-type categories that strongly differ in size. Some categories, such as W2A (learned professional writing) and W2B (popular professional writing), contain up to 40 texts of 2,000 words, while others, such as W2E (news editorials) and W2F (creative writing) contain only 10 or 20 texts. Although the similar design of all the ICE subcorpora proves valuable for the description of different varieties of World Englishes, it should be borne in mind that these similarities might not represent the actual importance of a specific text type in a given variety. Furthermore, the bias towards academic texts has to be taken into account when it comes to the question as to what the corpus is representative of, since it could be argued that the average speaker of a given variety might be
Chapter 5.╇ GIVE
Table 5-7.╇ Complementation of GIVE in ICE-India and ICE-GB – written data Type I Ider IP IPder II
Pattern (S) GIVE [Oi:NP] [Od:NP] all related to type-I
[S < Oi active] BE given [Od:NP] (by-agent) all related to type-IP
(S) GIVE [Od:NP] [Oi:PPto]
IIder
all related to type-II
IIP
[S < Od active] BE given [Oi:PPto] (by-agent)
IIPder III
%
sum
%
130
23.8
150
34.2
5
0.9
7
1.6
43
7.9
41
9.3
2
0.4
3
0.7
110
20.1
51
11.6
3
0.5
5
1.1 3.0
2.9
13
5
0.9
7
1.6
(S) GIVE [Od:NP] Oi
157
28.8
108
24.6
all related to type-III
[S < Od active] BE given Oi (by-agent)
V
sum
16
IIIP IV
ICE-GB
all related to type-IIP
IIIder IIIPder
ICE-India
all related to type-IIIP (S) GIVE Oi Od
(S) GIVE [Oi:NP] Od
Vder all related to type-V sum total p
2
0.4
3
0.7
49
9.0
31
7.1
22
4.0
16
3.6
2
0.4
4
0.9
0
0.0
0
0.0
0 0.0 546 100.0 < 0.001
0 439
0.0 100.0
less prone to reading academic texts than, say, newspapers. Additionally, this may be variety-dependent, as speakers of non-native varieties have often undergone tertiary education, while for speakers of native varieties secondary education is the minimal qualification for inclusion in the corpus, so the choice of reading material of the average speaker may differ according to their education. Table 5-7 shows the distribution of the different complementation patterns of GIVE in the written parts of ICE-India and ICE-GB with regard to the actual number of occurrences as well as to the percentages of use of a specific pattern in the given variety. As mentioned in Section 4-5, GIVE is used less frequently in the written parts of both ICE corpora, though only significantly so in ICE-India. It is also used significantly more frequently in the written part of ICE-India than the written part of ICE-GB. As in the spoken part, the main differences are the more frequent use of GIVE in the type-I pattern in ICE-GB and at the same time the higher frequency in the other basic patterns in ICE-India. While in raw figures the difference in the type-I pattern is not so pronounced (130 occurrences compared to 150), in relative figures GIVE is complemented by the type-I pattern in 23.8% of all occurrences of GIVE in ICE-India and in 34.2% of occurrences in ICE-GB. In
77
78
Structural Nativization in Indian English Lexicogrammar
ICE-India this relation to the overall use of GIVE is comparable to the use in the spoken part of this corpus (22.9% type-I complementation). In British English, however, it is used much more frequently in the spoken part (43.5%) than in the written part, so there is a clear preference for this pattern in spoken British English (G2€= 5.83, p ≤ .05), while this preference is not significant in Indian English (G2€= 0.14, p ≥ .05). However, if we take the text type-specific complementation of GIVE in the type-I pattern into account (cf. Table 5-4), it can be shown that in ICE-India the text types in which GIVE is most prominently used are W1B (letters) and W2F (creative writing). Together they contain 70 out of the 130 examples. Both text types are more closely related to the spoken mode than others. Letters are often written in an informal spoken style and novels and short stories contain dialogue and narration, which are intended to resemble natural speech. This distinction has been pointed out, for example, by Biber & Finegan (1989), who show that both fictional texts and letters are closer to the spoken end of the three dimensions that they analyzed. Biber & Finagan use a model based on three factors that contain a large number of weighted linguistic features, such as number of nouns, word length or number of attributive adjectives, in order to account for “three dimensions that are associated with ‘oral’ and ‘literate’ differences in English” (Biber & Finegan 1989:â•›489). These three dimensions, each to be considered a cline, include distinctions of style between informational and involved, elaborated and situation-dependent, as well as abstract and non-abstract (cf. Biber & Finegan 1989:â•›494–497).50 Although there is also an overrepresentation of type-I complementation of GIVE in British English in these categories, the use of the type-I pattern is more equally distributed across text types, so that in this case these categories only contain 65 out of 150 examples. While there is no significant difference in the choice of the type-I pattern between the modes in Indian English, the results show that it is used more frequently in the speech-related categories W1B and W2F. However, the results also show a significant tendency for this pattern to be used in spoken British English, while its use in writing is more text-type independent. As was the case in the spoken parts of the corpora, in the written part the indirect object is realized as a pronoun in the majority of cases. However, here the figures are lower. In ICE-India 83% of all type-I sentences have a pronominal Oi (compared with 90% in the written part), while in ICE-GB pronominal realization is used much less in the written part of the corpus (59% of all type-I sentences) in relation to the spoken part (83%). Although for Indian speakers there 50. With regard to the ICE corpora, it should be kept in mind that W1B contains business as well as social letters. In the model of Biber & Finnegan (1989), business letters range closer to the written mode in all dimensions.
Chapter 5.╇ GIVE
Table 5-8.╇ Pronominal Oi in written GIVE type-I patterns Pronoun
her
him
it
me
ICE-India ICE-GB
5 8
19 16
â•⁄ 2 11
21 11
them themselves 16 â•⁄ 6
2 1
us
you
total
21 â•⁄ 9
22 26
108 â•⁄ 88
is no bias towards the use of the type-I pattern when it comes to the difference between the two modes, there is a certain bias towards the use of this pattern within the written mode. In the written part of the corpus this pattern is mainly used in speech-related text types. British speakers who use this pattern more frequently in the spoken mode, however, show a greater dispersion of the pattern within the written mode. They use it more frequently in non-speech-related text types and use a larger variety of indirect objects. Table 5-8 shows the use of the different pronominal indirect objects in the two corpora. When compared to Table 5-6 – the distribution of pronominal indirect objects in the spoken mode – we can see that there are some noteworthy differences. While the use of me as an indirect object did not show any large variation in the spoken parts of the corpora (46 ICE-India, 49 ICE-GB), in the written part it is used much more frequently in the Indian corpus. Looking at the use of the type-I pattern with the indirect object me in the written part of ICE-India, it can be seen that this combination is almost exclusively used in the categories discussed above with 6 occurrences in category W2F (creative writing) and 14 occurrences in category W1B (letters). Regarding the use of me in ICE-GB, a similar picture can be observed, here all of the 12 instances of GIVE + me are found either in W2F (5 occurrences) or W1B (7 occurrences). Thus, while the use of the pronoun me is text type-dependent in both varieties, it is used more frequently in the Indian corpus, which is also connected to the higher preference for this pattern in these text types in Indian English in general. If we look at the difference in the use of the type-I pattern, in which the indirect object is us, the situation at first sight appears similar to the use of me. While us was used only slightly more often in Indian English in the spoken part of the corpus (39 compared to 31 times), it is used much more frequently in the written part of ICE-India than in this part of ICE-GB. However, if text types are taken into account, the use of us clearly differs from the use of me in the type-I pattern. Whereas me was most frequently used in categories W1B and W2F in both varieties, there are variety-specific differences in the use of the type-I pattern with us. In ICE-India this pattern is used 10 times in category W1A (student writing), while it is not found in this category in ICE-GB. Interestingly, there is no clear preference between the varieties for the use of this pattern in the other text categories. It is, for example, used 5 times in both varieties in category W1B and
79
80 Structural Nativization in Indian English Lexicogrammar
3 times in category W2B, so that the difference in the use of GIVE + us between the corpora mainly seems to be due to a preference for this pattern in the writing of Indian students. Therefore, the differences in the use of the type-I pattern are more prominent in the written parts of the ICE corpora than in the spoken parts. It is used less frequently in Indian English than in British English, whereas this distinction was not as prominent in the spoken part in absolute terms. This difference is due to the higher frequency of use of GIVE complemented in the other patterns, especially the type-III pattern and the type-II pattern. As in the spoken mode, the type-III pattern is also the most frequent complementation pattern of GIVE in the written mode in ICE-India. It is, however, used less frequently than in the spoken part of the corpus. In the written part of the corpus this pattern is used in 28.8% of all instances where GIVE is used in any of the patterns discussed, while this figure amounts to 31.9% in the spoken part. Furthermore, by looking at the figures in relation to corpus size we can see that this pattern is used 0.64 times ptw in the spoken part and only 0.39 times ptw in the written part of the corpus. In relation to the other instances of complementation of GIVE, the discrepancy seems minimal (3.1%). However, in relation to the size of the corpus parts it is in fact quite prominent. In British English, on the other hand, this pattern is only the second-most frequent complementation pattern after the type-I pattern. Here this pattern is also used less frequently in the written part in relation to the other patterns (21.7% in the spoken part, 24.6% in the written part). However, while in British English the lower relative figures can be explained by the higher tendency to use the type-I pattern in the spoken mode, in Indian English the higher tendency for type-II complementation in the written mode leads to the lower percentage of type-III complementation in that mode. If the total use of GIVE in the type-III pattern in relation to corpus size is taken into account, another large difference can be shown between the varieties. In the spoken part of ICE-GB, this pattern is used 0.24 times ptw in the spoken part and 0.27 times ptw in the written part. Thus, while this pattern is used significantly less frequently in relation to corpus size in the written part of ICE-India (G2€=€27.97, p ≤ 0.0001), it is used slightly more frequently in the written part of ICE-GB, though this difference is not significant (G2€= 0.95, p ≥ 0.05). The use of this pattern in the different text types can be shown to be most frequent in the text-type categories W1A (student writing) and W1B (letters) in Indian English. In category W1A it is used 29 times in 20 texts, in category W1B 29 times in 30 texts. In comparison, this pattern is used much less frequently in the same categories in ICE-GB (13 times in W1A, 8 times in W1B). Thus, in Indian English the frequency of this pattern is markedly higher both in spoken texts and in unedited written texts than in ICE-GB. Furthermore, this pattern is used
Chapter 5.╇ GIVE
to a greater extent in the category of news reportage (W2C) in Indian English, occurring 18 times in 20 texts compared with only 4 times in ICE-GB. In the following, I will discuss a few examples of typical use of this pattern in the different text-type categories described above. Examples (34) to (36) show typical usages of this pattern in personal as well as business letters in Indian English: (34) Therefore, he didn’t see any need to give any letter of introduction. (35) Please consider for giving training in your organization for Two Months. (36) […] Please find out donors, commercial companies willing to give ads etc.
Example (34) is in accordance with many of the observations for the use of GIVE in the spoken part of the corpus. Here GIVE is used as an equivalent for possible different verbs in other varieties, such as SEND. Furthermore, comparable to the use in legal English, it is used in the meaning of ‘(officially) stating something’. Example (35) shows the use of GIVE in a light-verb construction. In this construction the recipients of the training do not have to be profiled, while with the simplex verb TRAIN the recipient of the training would have to be profiled, as otherwise the letter would be ambiguous, because the addressee of the letter might also be meant. Example (36) is comparable to Example (23), as GIVE is used with the direct object ads (36) and advertisement (23). Here GIVE is again used as a generalmeaning verb where a more specific verb that collocates more strongly with ad, such as PLACE, could also be used. Regarding the type-II pattern, the comparison of Table 5-8 and Table 5-10 shows that while the use of this pattern reveals no significant difference in the spoken and written parts of the British corpus, it is used more frequently in the written part of the Indian corpus. According to text type, the type-II pattern is used more frequently in Indian English in almost all categories, with the exception of instructional and persuasive writing (W2D and W2E), where it is used either with equal frequency (W2D: 4 times), or slightly more frequently in ICEGB (W2E: 5 times ICE-GB; 4 times ICE-India). Relative to category size, this pattern is used most frequently in Indian English writing in the categories W1A (student writing) with 0.6 times ptw, W2C (press reports) with 0.35 times ptw and W2B (letters) with 0.3 times ptw. In the British counterpart of the corpus it is mainly used in the categories W1A with 0.25 times ptw and W2E (persuasive writing) with 0.2 times ptw. Thus, the type-II pattern is used most frequently in category W1A in both corpora, although its presence is much stronger in the
81
82
Structural Nativization in Indian English Lexicogrammar
N Concordance 1 ystem Act, Sati Abolition Act & many more. They 2 ouse compulsory compulsorily in the dark room. If she did not 3 up of a new and refreshing outlook. The progress of science 4 through a valuable phase. Nature and environment, which 5 erci. These are hemimetabolous insects and gives 6 ffected. Forests provide protection to animals & humans And 7 plan roots of plants suck absorb water from the earth & 8 demerits which may be listed as - - Chemical control does not 9 is recongnized democracy can servive survive. Political parties 10 to look after children. Without working only would have to 11 the scientist arrives at explanatory hypothesis which should 12 waste it in day-dreaming. 3) Do not prejudice be prejudiced, 13 organisations. Banks or many financial institutions are ready to 14 for navigation. Government or semi-Government does not 15 respective co-operative society Act and made provisions to 16 surprising, thus, to note that the new government decided to 17 unbalanced his equation of life. He never wanted to fro 18 compulsory compulsorily in the dark room. If she did not gave 19 fish’. Body is covered with scales wich are is silver and 20 0.2 to 0.3 mm long and it gives rise to pupa. Pupa is again 21 which is full grown measures about 0.2 to 0.3 mm long and it 22 in November & It it may go under hibernation. The pupa 23 malformations The incidence of ‘Pulmonary pathology’ alone 24 about consumption of fish 5) In tropics Government is not
gave free education the to much many of the women. They p gave give the birth to son she was severly critized criticized from gave rise to a scientific outlook to life. Blind faith has been gave birth to man and nursed him, is in peril. Environmental give rise to indivisuals with the characters i.e. which are similar give valuable food to these. But if we do deforestation we will give it to the air by leaves. And this fall into rains. And if we cut give a permanent solution to a problem even though it shows give proper direction to the efforts of the people & as such they give more and more importance to the children. Mother would give explanation to the problem. 4 A good hypothesis should give a fair chance to the speaker to express his views and give credit to big organisations. Thus, raising the finance is not a give amount or any financial help to those rivers. 5) In rivers, give 30% representation to women members on the Board of give highest priority to agriculture and rural development give pain to both the Ladies. Thus only solution he could find is give the birth to son she was severly critized criticized from each gives silver appearance to the insect. LIFE CYCLE: gives rise to adult. The pupal period is 5 to 7 days and after gives rise to pupa. Pupa is again gives rise to adult. The p gives rise to adult which is a stout moth. It is buff coloured. The giving rise to still births and neonatal deaths is very high and giving finance to fisherman fishermen like to build buiding
Figure 5-1.╇ Type-II pattern in ICE-India: W1A (student writing) N Concordance 1 Love of the transcendental image of the Virgin Mary was what 2 where the snake’s eternal shedding of skin (ego) and rebirth 3 minerals and are then known as amygdales. Pillow lavas also and in basic it is known as scourie. The size of the vesicles 4 5 first with the finer sediment on top. eg Other structures as deltas, formed by the varying effects on the estuary 6 7 a pattern as in FIG I. Fluvial, and marine environments also 8 with unavoidable unpredictable environmental change has 9 onto on to the next lines. At the end of the poem the speaker 10 produced. It is wealth, the fact that Pierce had it and is
gave birth to the glorious cathedrals of the Gothic world and the gave the initiate access to Her wisdom. However, in a culture give rise to way up as the they generate a cusp like feature due give rise to way up as the large vesicles form near the top of the give rise to way-up being established. They are generally less give rise to top, fore and bottom sets which show way-up. On give rise to characteristic bedding such as cross bedding and given rise to selection pressures responsible for the evolution of gives an ultimatum to God. Which, which could at first seem giving it to Oedipa, which sets her on her isolated quest.
Figure 5-2.╇ Type-II pattern in ICE-GB: W1A (student writing)
Indian part. By looking at the concordances for the use of GIVE in this pattern in category W1A in ICE-India (Figure 5-1) and ICE-GB (Figure 5-2), we can see€– as already noted in Section 5.2.2 – that the use of GIVE in the fixed expression GIVE rise to accounts for a large number of instances in ICE-GB (60%). Although this expression is also frequently used in ICE-India, it accounts for only 25% of the uses of GIVE in the type-II pattern. If we consider the whole written part of the ICE corpora, we will find that the underlying trend of the higher frequency of use of the expression GIVE rise to in the British corpus is also attested, though
Chapter 5.╇ GIVE
not as prominently as in this category (15.5% of all type-II patterns in ICE-India compared to 24% of all type-II patterns in ICE-GB). Thus, it seems that in Indian English the type-II pattern is more versatile and is more often seen as a genuine alternative to the type-I pattern. Furthermore, the concordance in Figure 5-1 shows the above-mentioned use of GIVE in light-verb constructions in line 8 and line 24; where speakers of British English may prefer the simplex verbs explain and finance respectively. Of further interest here is the use of the type-II pattern in the reportage category in ICE-India, since with 0.35 times ptw it is used about as frequently as in the spoken category S2B (scripted monologue: 0.34 ptw), that is, news broadcasts, broadcast talks and non-broadcast talks. So this pattern is used with roughly the same frequency in the news-related categories of the written and spoken parts of ICE-India. At the same time, the fixed expression GIVE rise to, which accounts for 25% of all uses of this pattern in category W1A, is only found rarely in these categories. In the written news category W2C it does not occur at all and in the spoken scripted monologues category S2B it only accounts for three out of 36 instances (8.3%), so in the written news category this pattern is clearly underrepresented, while in the spoken categories it is slightly overrepresented. However, it should be noted that it is only used once in the S2B texts that contain broadcast news reports, the category closest to press reportage, and twice in texts that contain broadcast talks. Finally, regarding the passive patterns type-IP, type-IIP and Type-IIIP, Tables 5-5 and 5-7 show that, while there are large discrepancies in the use of these patterns in the spoken parts of the corpora, in the written parts they are used much more homogeneously. Here only the more frequent use of the type-IIIP pattern in ICE-India (49 occurrences compared to 31 ocurrences in ICE-GB) is statistically significant (G2 = 4.08; p ≤ 0.05). The higher tendency for the use of this pattern in ICE-India was also attested in the spoken parts of the corpora. However, in the spoken part the use of this pattern was, by comparison, distributed relatively equally across the different text types, while in the written part the use of this pattern is clustered in two different text categories in the different varieties. In ICE-India 28 out of 49 (ca. 57%) of the occurrences are found in category W2A (academic writing), whereas in ICE-GB this pattern is mainly used in category W2D (instructional writing) with 11 out of 31 occurrences (ca. 35%), and only 6 times in category W2A. The frequent use of the type-IIIP pattern in category W2A stems mainly from the use of sentences such as that in Example (37): (37) The WQI of river water thus calculated is also given in table 2.
83
84
Structural Nativization in Indian English Lexicogrammar
In eight instances the type-IIIP pattern is used as in (37) with given preceeding in table and in six further instances the pattern is used for the same situation, but with the phrase in figure instead of in table. Hence, at this point it is difficult to say whether the frequency of use of the type-IIIP pattern in ICE-India has variety-specific reasons or is more related to the choice of corpus material. Sentences like Example (37) are also encountered in ICE-GB and are relatively common in academic writing. Thus, although there may be some intervarietal differentiation in the verbs used in this construction that might further strengthen the theory that users of Indian English use GIVE in a number of instances where users of British English would prefer other verbs, the evidence does not hold in this case. In ICE-GB the phrase in table (occurring 4 times) is preceded by three different verbs, namely GIVE, LIST and DESCRIBE. In ICE-India the phrase itself is more frequent (17 times) but there is also a much wider range of verbs preceding it. Besides GIVE, a number of other verbs, such as SHOW, SUMMARIZE, LIST, PRESENT and MENTION are used in this construction. So the frequent use of the type-IIIP pattern in this category will at least partly be due to the inclusion of many texts that contain tables in the corpus.
5.5
Collocational profiles of GIVE in Indian and British English
In the following section, I set out to provide collocational profiles for GIVE in Indian and British English according to the methodology described in Section 4.4. The first step is to index the ICE corpora with WST V4.x to calculate possible collocation candidates. These candidates will then be submitted to a set of testing conditions to find genuine collocations to the verb under scrutiny. These conditions – as explained in Section 4.4 – consist of a set of quantitative restrictions and a further qualitative restriction. These restrictions (repeated from Chapter 4) are as follows:51 Condition 1: Collocation candidates must occur in a span of 5:5 of a node word that is a predefined trivalent verb. Condition 2: A minimum number of five co-occurrences must be met. 51. At this point I would like to include a brief note on lemmatization. In the present collocational analysis the co-occurrence of the lemma of the verb with the lemma of the collocate has been analyzed. However, due to the restrictions of WST 4.x, occurrences of a word form of a given lemma are only taken into account if they themselves occur with a minimum frequency of 5. For these reasons collocates have not been lemmatized in the sections on SEND and OFFER where the larger underlying datasets made it possible to analyze individual word forms, a procedure suggested by Michael Stubbs (2008, pc).
Chapter 5.╇ GIVE
Condition 3: Collocation candidates must be statistically significant according to a log-likelihood value of ≥ 15.13 (p ≤ 0.0001) as an approximation to exact testing. Condition 4: Collocation candidates qualify as Od-collocates if the collocate fulfils the semantic function of a patient in a cause-receive process. The candidates that fulfil the first four conditions can be found by calculating a collocational profile for GIVE with the use of WST, condition five is checked manually. After these different tests, collocations for GIVE in both Indian and British English are identified and can thus be compared across the varieties to account for collocations shared in both varieties and collocations only found in one variety. In a next step the sentences in which GIVE is used in a genuine collocation with specific Od-collocates will be analyzed according to their verbcomplementational patterns to test for correlations between collocational and complementational preference in the varieties. After these steps, a total of 22 Od-collocates of GIVE can be identified in ICEIndia, as shown in Table 5-9. Table 5-9.╇ Od-collocates of GIVE in ICE-India ICE-India Od-collocates
log-likelihood
address advice amount answer attention award chance description details example explanation history idea importance information meaning money news party priority report rise
214,66 â•⁄ 80,92 â•⁄ 85,10 167,52 â•⁄ 31,34 â•⁄ 52,26 195,56 â•⁄ 63,08 240,30 260,06 â•⁄ 80,92 â•⁄ 40,92 165,26 269,77 209,16 â•⁄ 58,30 192,12 211,27 â•⁄ 98,00 110,02 â•⁄ 99,48 382,45
85
86 Structural Nativization in Indian English Lexicogrammar
Table 5-10.╇ Od-collocates of GIVE in ICE-GB ICE-GB Od-collocates
log-likelihood
chance details evidence example idea impression indication information opportunity prescription ring rise sense support time way
139,60 314,42 303,92 104,58 106,38 171,46 152,33 164,78 201,69 149,71 â•⁄ 93,06 187,39 â•⁄ 78,95 â•⁄ 81,36 365,39 277,42
If the same methodology is applied to GIVE in ICE-GB, a similar table for a collocational profile of GIVE in British English can be constructed. In the British corpus, however, there are fewer significant Od-collocates to GIVE, so in this case the table only includes 16 different collocations with their respective G2-values (Table 5-10). By comparing these two profiles it can be shown that a number of Od-collocates of GIVE are salient in both varieties of English, while others are preferred in one of the varieties. Table 5-11 gives an overview of variety-specific collocates (left and right column) and collocates shared by the two varieties (middle column). Against the background of the collocational profiles for GIVE in both varieties, we can now look at the complementational patterns of the respective verb in those instances where it collocates with one of the items above. Although these occurrences only account for a relatively small proportion of the overall occurrences of GIVE (299 out of 1063 concordance lines in ICE-GB, 375 out of 1797 concordance lines in ICE-India), Table 5-12 shows that there are still significant distributional differences between the varieties even if only these instances are taken into account.52 52. Note that Table 5-12 shows only the basic patterns, subsuming non-basic patterns under others and collapsing the passive patterns with the respective derivative patterns. This step seems reasonable as, especially in the following, the relatively low frequency of the single collocation would not profit from an overly in-depth analysis.
Chapter 5.╇ GIVE
Table 5-11.╇ Od-collocates of GIVE sorted by variety ICE-GB
ICE-GB and ICE-India
ICE-India
evidence impression indication opportunity prescription ring sense support time way
chance details example idea information rise
address advice amount answer attention award description explanation history importance meaning money news party priority report
Table 5-12.╇ Complementation patterns of GIVE (only significant collocations) Corpus/Pattern
I
ICE-India 97 ICE-India % 25.87 ICE-GB 87 ICE-GB % 29.10 p ≤ 0.01 (# of occurrences)
Ider
II
IIder
III
IIIder
other
sum
24 â•⁄ 6.40 34 11.37
72 19.20 37 12.37
14 â•⁄ 3.73 â•⁄ 5 â•⁄ 1.67
111 â•⁄ 29.60 â•⁄ 74 â•⁄ 24.75
33 â•⁄ 8.80 27 â•⁄ 9.03
24 â•⁄ 6.40 35 11.71
375 100 399 100.00
Though only the part of the total population of instances of GIVE that occurs with significant Od-collocates is taken into account, the distribution in Table 5-12 is largely in line with Table 5-5 in showing significant distributional differences. Preference for type-I pattern complementation in British English and the preference for type-III and type-II pattern complementation in Indian English are also still represented in this smaller dataset. Many specific Od-collocates favour different complementation patterns. Table 5-13 shows the different Od-collocates of GIVE in Indian English. As Table€5-13 shows, specific collocates to GIVE are preferred in specific complementation patterns. When the items news, advice, chance, idea and importance are used with GIVE, the resultant sentences all show a tendency towards type-I pattern or type-Ider pattern complementation. The clearest cases of this are news, advice and chance. However, as mentioned earlier, the use of the type-I pattern in combination
87
88
Structural Nativization in Indian English Lexicogrammar
Table 5-13.╇ Frequent complementation of GIVE in correlation to Od-collocates in ICE-India Od-collocate Type-I Type-Ider Type-II Type-IIder Type-III Type-IIIder other
N
news advice chance idea importance priority rise address money answer details information explanation
20 â•⁄ 8 19 20 25 10 29 21 21 19 19 25 â•⁄ 8
85.0% 75.0% 52.6% 35.0% â•⁄ 0.0% â•⁄ 0.0% â•⁄ 0.0% 14.3% 28.6% 15.8% â•⁄ 5.3% 20.0% 12.5%
â•⁄ 0.0% â•⁄ 0.0% 31.6% â•⁄ 5.0% 44.0% â•⁄ 0.0% â•⁄ 0.0% â•⁄ 0.0% â•⁄ 0.0% â•⁄ 5.3% â•⁄ 0.0% â•⁄ 0.0% â•⁄ 0.0%
â•⁄ 0.0% 12.5% 15.8% â•⁄ 5.0% 58.0% 70.0% 96.6% 19.0% â•⁄ 4.8% â•⁄ 0.0% â•⁄ 0.0% â•⁄ 0.0% 12.5%
â•⁄ 0.0% â•⁄ 0.0% â•⁄ 0.0% 35.0% â•⁄ 0.0% 20.0% â•⁄ 3.4% â•⁄ 0.0% â•⁄ 0.0% â•⁄ 0.0% â•⁄ 0.0% â•⁄ 4.0% 12.5%
10.0% 12.5% â•⁄ 0.0% â•⁄ 5.0% â•⁄ 0.0% 10.0% â•⁄ 0.0% 47.6% 61.9% 68.4% 73.7% 64.0% 62.5%
â•⁄ 5.0% â•⁄ 0.0% â•⁄ 0.0% â•⁄ 0.0% â•⁄ 8.0% â•⁄ 0.0% â•⁄ 0.0% 19.0% â•⁄ 4.8% 10.5% 21.1% 12.0% â•⁄ 0.0%
â•⁄ 0.0% â•⁄ 0.0% â•⁄ 0.0% 15.0% â•⁄ 0.0% â•⁄ 0.0% â•⁄ 0.0% â•⁄ 0.0% â•⁄ 0.0% â•⁄ 0.0%â•⁄ â•⁄ 0.0% â•⁄ 0.0% â•⁄ 0.0%
with news has to be taken with a pinch of salt, as 13 of the 20 total occurrences stem from the sentence “This is All-India-Radio giving you the News”, which seems to be overrepresented in the corpus. In the cases of advice and chance there is no such distortion, although the relatively low number of total instances of advice should be taken into account. idea is used equally frequently in the type-I and the type-II patterns, whereas importance is used mainly in the type-I pattern when the sentences are in the passive voice and in the type-II pattern when the sentences are in the active voice. In the case of priority we find a strong tendency towards type-II pattern complementation. While the same holds true (to an even larger extent) for rise, the formulaic sequence give rise to has basically no alternative way of complementation (?give X rise). The largest group of Od-collocates in Table 5-13 consists of those used in sentences that are preferably complemented by the type-III pattern. This group includes address, money, answer, details, information and explanation. As the polysemous nature of address may be of some relevance to the complementation pattern, a look at the different meanings and their use may be of interest here. However, as the ‘speech’ meaning of address is only encountered twice (both in type-III pattern complementation), more data would be needed in order to account for complementational differences on grounds of semantic distinction. The ‘place to live’ meaning is attested in all three of the basic complementation patterns, with a clear preference, however, for the type-III pattern and its derivatives. Example (38) provides a typical example of this use:
Chapter 5.╇ GIVE
(38) Please acknowledge this letter giving your home address
The cases of money, answer and details are relatively straightforward in their tendency towards type-III pattern complementation, although in the case of money there is also some variation towards type-I complementation. Information and explanation are the two items in Table 5-13 that are potentially attested in light-verb constructions, a construction that seems to play a certain role in the nativization of Indian English, as pointed out, for instance, in Example (30) in Section 5.2.1. By looking at the concordance lines for GIVE + information, eight out of the total 25 instances can possibly be classified as lightverb constructions, i.e. the use of the simplex verb INFORM would be a genuine alternative to the construction, as exemplified in (39): (39) The article does give fairly comprehensive information about what the ICE corpus consists of […]
With regard to verb complementation, all of these eight instances are complemented by the type-III or type-IIIder patterns, so this use may also be responsible for the association of GIVE + information with these patterns. As GIVE + explanation is relatively rare in general, the number of possible light-verb constructions here – only two instances – is too small for substantial claims to be made about this construction correlating with a specific pattern, as one is encountered in the type-II pattern and one in the type-III pattern. There are a number of other Od-collocates of GIVE in ICE-India that do not cluster around specific complementation patterns or are too rare for claims to be made about correlations between verb complementation and collocation. These Od-collocates and their distribution according to verb-complementational patterns are given in Table 5-14. The collocates meaning and award with seven and eight occurrences, respectively are relatively infrequent and both are used in a variety of patterns, so that here there is no favoured pattern. Example and party are both attested quite Table 5-14.╇ GIVE in correlation to Od-collocates in ICE-India (no specific trend) Od-collocate Type-I Type-Ider Type-II Type-IIder Type-III Type-IIIder other
N
meaning party example award report amount
â•⁄ 7 15 32 â•⁄ 8 14 12
â•⁄ 0.0% 26.7% 28.1% â•⁄ 0.0% 14.3% 16.7%
â•⁄ 0.0% â•⁄ 0.0% â•⁄ 0.0% 12.5% â•⁄ 0.0% â•⁄ 0.0%
42.9% â•⁄ 0.0% â•⁄ 3.1% 25.0% â•⁄ 7.1% â•⁄ 8.3%
14.3% 33.3% â•⁄ 0.0% â•⁄ 0.0% â•⁄ 7.1% â•⁄ 0.0%
42.9% â•⁄ 0.0% 31.3% 50.0% 35.7% 16.7%
â•⁄ 0.0% â•⁄ 0.0% 15.6% 12.5% 14.3% 41.7%
â•⁄ 0.0% 40.0% 21.9% â•⁄ 0.0% 21.4% 16.7%
89
90 Structural Nativization in Indian English Lexicogrammar
Table 5-15.╇ GIVE in correlation to Od-collocates in ICE-GB Od-collocate Type-I Type-Ider Type-II Type-IIder Type-III Type-IIIder other
N
ring sense chance opportunity prescription idea rise way evidence details indication example impression
â•⁄ 9 11 15 18 11 14 18 40 32 18 12 15 13
100.0% â•⁄ 81.8% â•⁄ 73.3% â•⁄ 44.4% â•⁄ 81.8% â•⁄ 42.9% â•⁄â•⁄ 0.0% â•⁄â•⁄ 0.0% â•⁄â•⁄ 0.0% â•⁄â•⁄ 5.6% â•⁄ 41.7% â•⁄ 20.0% â•⁄â•⁄ 7.7%
â•⁄ 0.0% 18.2% 20.0% 55.6% â•⁄ 9.1% 40.0% â•⁄ 0.0% â•⁄ 0.0% â•⁄ 3.1% â•⁄ 0.0% â•⁄ 0.0% â•⁄ 0.0% â•⁄ 7.7%
â•⁄â•⁄ 0.0% â•⁄â•⁄ 0.0% â•⁄â•⁄ 0.0% â•⁄â•⁄ 0.0% â•⁄â•⁄ 0.0% â•⁄â•⁄ 0.0% 100.0% â•⁄ 42.5% â•⁄â•⁄ 3.1% â•⁄â•⁄ 5.6% â•⁄â•⁄ 0.0% â•⁄â•⁄ 0.0% â•⁄â•⁄ 0.0%
0.0% 0.0% 0.0% 0.0% 9.1% 0.0% 0.0% 0.0% 3.1% 0.0% 0.0% 0.0% 7.7%
â•⁄ 0.0% â•⁄ 0.0% â•⁄ 6.7% â•⁄ 0.0% â•⁄ 0.0% â•⁄ 7.1% â•⁄ 0.0% 25.0% 65.6% 55.6% 58.3% 53.3% 61.5%
â•⁄ 0.0% â•⁄ 0.0% â•⁄ 0.0% â•⁄ 0.0% â•⁄ 0.0% â•⁄ 7.1% â•⁄ 0.0% â•⁄ 0.0% 25.0% 33.3% â•⁄ 0.0% â•⁄ 0.0% 15.4%
â•⁄ 0.0% â•⁄ 0.0% â•⁄ 0.0% â•⁄ 0.0% â•⁄ 0.0% 14.2% â•⁄ 0.0% 30.0% â•⁄ 0.0% â•⁄ 0.0% â•⁄ 0.0% 26.7% â•⁄ 0.0%
frequently and show certain variability in pattern use. There are, however, many occurrences where the collocate is not in the position of the direct object, subsumed under others in Table 5-15. Finally, report and amount were also categorized as not clustering around specific patterns. Although there is a certain tendency towards type-III and type-IIIder complementation respectively, the relatively low frequency of these collocates and the variation between patterns does not rule out a random distribution. The verb-complementation patterns for the Od-collocates in ICE-GB are displayed in Table 5-15. As can be seen in Table 5-15, the overall trend for the use of the type-I pattern in British English, as discussed in the previous sections, is also attested for this smaller dataset. While in ICE-India the sentences that showed a tendency to be used in the type-I pattern were also used in the type-II and typeIII pattern, in ICE-GB these sentences are almost exclusively used in the type-I pattern or in one of its derivatives. The phrase GIVE X a ring, as shown in (40), in relation to a telephone call, for example, appears to be highly institutionalized, so that there seems to be virtually no genuine alternative to the type-I pattern (?give a ring to X). (40) I’ll give you a ring next time I’m down and we’ll try and meet OK
If we look at opportunity, another of the Od-collocates that is preferred in the type-I pattern, we can see that this pattern is very pervasive, including in cases where many more instances are attested. Although there is some variation, mainly based on the use of the passive pattern – which is here subsumed under
Chapter 5.╇ GIVE
type-Ider€– again there are no uses in patterns not related to the type-I pattern. Example (41) gives a typical passive example: (41) You would do if you were given the opportunity
The two Od-collocates that have a tendency to be used in the type-II pattern are rise and way. In the case of rise all of this collocate forms are part of the formulaic expression GIVE rise to X, where the realization of the indirect object in X is an inanimate or abstract entity in the vast majority of cases, as exemplified in (42): (42) For the size of the countries concerned these figures did not yet give rise to alarm
In the case of way there are some similarities to the case of rise. Here we also encounter give way to x as a formulaic sequence, where x is often inanimate and possibly abstract, as in (42), but there is also a physical sense to this expression, where an animate entity provides another animate entity with an opportunity or physical space for movement, or some entity moves out of the way of another. In the latter case, the type-II pattern and the type-III pattern are frequently used, see (43)–(44), whereas the former case seems to be limited to the type-II pattern. (43) Beneath it an ocean ten thousand feet deep that will kill you in minutes if the ice gives way (44) Now I give way to the honourable gentleman if he wishes to come in
Thus, while in the case of GIVE rise to x the pattern was very fixed, in the case of way there is a possibility for pattern variation that is influenced by the different semantics of way in a physical or figurative sense. The Od-collocates that are most frequently used in type-III pattern complementation in ICE-GB are EVIDENCE and DETAIL. The use of the type-III pattern makes perfect sense insofar as in both cases the recipients of the transfer process are understood implicitly, so no specification (as an indirect object) is necessary. Consider (45) and (46): (45) The witness, Mr Kenneth Kgase, admitted on his third day of giving evidence that he had misled the court […] (46) Details are given in the separate leaflet Applying for a reader’s pass.
As shown in Table 5-11, there are a number of significant Od-collocates used with GIVE in both Indian and British English. By looking at the verb-distributional
91
92
Structural Nativization in Indian English Lexicogrammar
Table 5-16.╇ Complementation patterns of GIVE (only shared Od-collocates) Corpus/Pattern
I
ICE-India 32 ICE-India % 22.2 ICE-GB 27 ICE-GB % 27 p ≤ .20 (# of occurrences)
Ider
II + IIder
III
IIIder
other
sum
â•⁄ 7 â•⁄ 4.9 14 14
42 29.2 19 19
41 28.5 25 25
12 â•⁄ 8.3 â•⁄ 9 â•⁄ 9
10 â•⁄ 6.9 â•⁄ 6 â•⁄ 6
144 100 100 100
patterns of these Od-collocates we can see that there is still a certain amount of variation between the varieties. However, this distribution is no longer significant, as can be seen in Table 5-16. This result corroborates the hypothesis that the choice of an Od-collocate exerts an influence on the complementational profile chosen. Thus, different collocational profiles in the individual varieties are one of the factors that underlie verb-complementational differences between the varieties.
5.6
GIVE: A brief summary
In summary, there are several factors underlying the distributional differences in the verb-complementational profiles of GIVE in Indian and British English. Firstly, there are clear differences between the use of GIVE in the written and the spoken mode. Indian English speakers use GIVE much more frequently than speakers of British English. To some extent, this may be explained by the use of GIVE as a substitution for more precise transfer verbs. The reason behind this is€– at least with regard to the spoken data – that speakers of ESL-varieties may use a very familiar verb in situations when they are under time constraints in online speech production, whereas speakers of ENL-varieties may not suffer these constraints in the same way. Furthermore, it was shown that speakers of Indian English have a certain tendency to use light-verb constructions where speakers of British English would opt for an alternative, such as e.g. a different simplex verb. Example (35) aptly illustrates a situation where a speaker chooses the light-verb construction over the related simplex verb in a self-correction. This can be seen an indication of the currency of this expression, especially as the repeated use of this construction is more pervasive in the Indian than in the British data. Another frequent use of GIVE is the text type-dependent use in legal proceedings. GIVE€+ evidence is quite frequent and the use of GIVE in the meaning of officially stating something seems to be a nativized feature in Indian English. Finally, I discussed the different collocational preferences of GIVE in the varieties. GIVE frequently
Chapter 5.╇ GIVE
collocates with more different items in the position of the direct object in ICEIndia than in ICE-GB. This is partly motivated by the lower frequency of GIVE in ICE-GB generally. However, there is only a limited number of shared Od-collocates between the varieties, so the description of variety-specific collocational profiles seems to be a worthwhile endeavour. This is particularly true if the collocational preferences of a verb are related to preferences at complementational level. As specific collocates correlate with specific complementation patterns, a difference in the collocational profile of a verb can be seen as one of the factors for variation in pattern selection.
93
chapter 6
SEND
6.1
Introduction: SEND as a less-prototypical ditransitive verb
Compared with GIVE, SEND is a far less prototypical ditransitive verb, on semantic grounds as well as on the grounds of frequency. Semantically, while GIVE implies a successful transfer of an affected entity towards a recipient, SEND only implies an intended transfer, since during the process of sending the transferred entity might be affected in a way that precludes the completion of the process. In Goldberg’s (1992) terms, SEND functions more at the periphery of a radial network of polysemous senses of the ditransitive construction, whereas GIVE (along with other verbs such as PASS and HAND) functions in the central sense of the construction. Goldberg’s (1995) framework also helps to deal with an additional property of SEND that clearly differentiates it from GIVE. Besides the semantic differences, there are also structural properties of SEND that differ from those of GIVE. Particularly in patterns that contain a to-prepositional phrase, it is at times difficult to decide whether the prepositional phrase encodes an object or an adverbial; consider Examples (47) to (49): (47) Britain is to send another fourteen thousand troops to the gulf (48) USA has sent a signal a clear signal to Pakistan on terrorism (49) a. People sent letters of complaint to the newspaper. b. People sent the newspaper letters of complaint.
In Example (47) the type-I pattern is not an admissible alternative, which may be grounds for arguing that the to-prepositional phrase encodes an adverbial of place rather than an indirect object. In Example (48), however, this is no longer as clear, as the type-I alternative sent Pakistan a signal does not sound ungrammatical. Example (49) illustrates this problem on a slightly finer scale. Whereas the animacy restriction on the double-object construction (cf. Levin 1993:â•›46) states that if the recipient is non-animate, the double-object construction is not admissible, institutions are often treated as animate recipients, so that the newspaper as an institution would not fall under this restriction. On the other hand, it is often
96 Structural Nativization in Indian English Lexicogrammar
difficult to decide what constitutes an institution that can function cognitively as an animate entity (consider again Example (18) in Section 4.5). If Pakistan and the USA in Example (48) are considered institutions that could cognitively function as animate entities, the type-I pattern is admissible, while this is clearly not the case for the gulf in (47). Thus, in these cases the use of the term type-II pattern is not strictly stringent, as the to-prepositional phrase does not in all cases encode indirect objects, but may also function as an adverbial of place. Mukherjee (2005) uses the label type-II pattern “for reasons of semantic and structural analogy” (Mukherjee 2005:â•›182) in cases such as Example (48). Goldberg (1995) treats this in a slightly different way, although with a comparable outcome. In her framework, the roles in the cause-receive pattern, namely agent recipient and patient (or goal) are the same in both the ditransitive and the transfer-caused-motion construction. In the ditransitive construction they are, however, realized by a subject and two objects, while in the transfer-caused-motion construction the recipient is realized by an oblique (cf. Chapter 3). Thus, for the sake of consistency in the present work, cases like Example (48) will be treated as instances of the type-II pattern, although they might be better thought of as cases of the transfer-caused-motion construction.53 In terms of frequency, SEND is also much less central a ditransitive verb than GIVE; firstly its overall frequency is much lower than that of GIVE and additionally it is used much less frequently in the type-I pattern, i.e. the basic ditransitive pattern. Because of the lower frequency of SEND in the ICE corpora overall, and the overrepresentation of the verb in one of the text categories included in the corpora (W1B, social letters and business letters), in this section (as well as in the next on OFFER) the ICE data will be complemented with data from the much larger Times of India Corpus (ToI). Since the complete ToI corpus of 110 million words is too large to yield manageable results (even though SEND is much less frequent than GIVE), a random subset of the data has been chosen, consisting of 20 full editions of The Times of India (henceforth ToI1.75). This randomized subset contains about 1.75 million words of texts in which SEND is used approximately 500 times, a number of concordance lines that is taken as being representative of the total corpus, as this number is representative of the distributions of the word forms of a given lemma. As a reference corpus, a comparable random subset of the newspapers contained in the periodical section of the British National Corpus (BNCNews.1.77) is used in the analysis of the larger dataset. 53. Note that Goldberg (2006) does not make a distinction between the transfer-caused-Â�motion construction and the caused-motion construction, but merges these into a single causedÂ�motion construction.
6.2
Chapter 6.╇ SEND
An overview of the verb-complementation patterns of SEND in ICE-GB and ICE-India
As in Chapter 5, this chapter will start with a brief overview of the complementation patterns of SEND in ICE-India and ICE-GB. As a starting point, I will again draw on the findings of Mukherjee & Hoffmann (2006), which provide a more detailed overview of the possible derivative patterns of SEND than the present study, albeit without differentiating between modes or text types. Table 6-1 summarizes their findings with regard to SEND. Mukherjee & Hoffmann (2006) again did not include phrasal and prepositional verbs, such as SEND back, SEND out and SEND off, because of the semantic differences from SEND proper. This procedure will be retained in the present study and additionally the idiomatic expression SEND Od Ving – represented as (S) SEND [Od:NP] [CompC:VPcont], a pattern that is not used for the other verbs under scrutiny – will be treated as a pattern in its own right. However, as this pattern is only used once in the ICE data (ICEGB), it will initially be subsumed under others. Thus, this pattern is of negligible impact in terms of the ICE corpora, but will be of some importance in the section on newspaper English (Section 6.7). As can be seen in Table 6-1, there are significant differences in the distribution of verb-complementational patterns of SEND between ICE-India and ICE-GB. While in ICE-GB the frequencies of the three main patterns, the type-I, type-II, and type-III pattern, are roughly equal, in ICEIndia there is a clear preference for the use of the type-III pattern, followed by the type-II pattern, while the type-I pattern is used to a much lesser extent. As Table 6-1 shows, the majority of the instances of SEND in the two corpora fall within the three basic pattern types and their respective passive patterns, while the peripheral patterns, type-IV and type-V, are only used rarely. The same holds true for the derivative patterns that will – analogous to the procedure in Chapter€5 – be collapsed under type-Ider to type-Vder, respectively. Table 6-2 shows the distribution of SEND as counted by Mukherjee & Hoffmann (2006) with a focus on the central patterns, including the type-IV and type-V patterns, respectively, but collapsing the derivative patterns. As discussed in Chapter 5, collapsing the derivative patterns into single categories makes it necessary to recalculate the statistic significance of Table 6-2, as the underlying degrees of freedom of the distribution change. However, since in some of the cells the expected values are below five, which makes a chi-square test problematic, these cells have been further collapsed into a single category termed others. This recalculation is shown in Table 6-3, which gives the observed and expected frequencies of the complementational patterns in both corpora and adds the chi-square values. When these values are added up, the distribution has a total chi-square value of 27.67, which is significant at a level of p ≤ 0.01. Thus, as
97
98 Structural Nativization in Indian English Lexicogrammar
Table 6-1.╇ Complementation of SEND in ICE-India and ICE-GB (cf. Mukherjee & Hoffmann 2006:╛173) Type I Ib
Pattern (S) SEND [Oi:NP] [Od:NP]
[Od: NPantecedent] (rel. pron.) [S] SEND [Oi:NP] miscellaneous
[S < Oi active] BE sent [Od:NP] (by-agent)
ICE-India
ICE-GB
sum
%
sum
%
54
15.4
62
21.8
4
1.1
6
2.1
1
0.3
1
0.4
0
0.0
7
2.5
0
0.0
1
0.4
(S) SEND [Od:NP] [Oi:PPto]
78
22.3
64
22.5
0
0.0
3
1.1
IIc
(S) SEND [Oi:PPto] [Od:NP]
0
0.0
2
0.7
IIP
[S < Od active] BE sent [Oi:PPto] (by-agent)
IP IPb II IIb
[Od:NPantecedent] (rel. pron.) [S < Oi] BE sent (by-agent) [Od:NPantecedent] [S] (rel. pron.) SEND [Oi:PPto] miscellaneous
IIPb
[antecedent]co (S < Od)co (BE) sent [Oi:NPto] (by-agent)
III
(S) SEND [Od:NP] Oi
miscellaneous
IIIb
[Od:NPantecedent] (rel. pron.) [S] SEND Oi
miscellaneous IIIP IIIPb
[S < Od active] BE sent Oi (by-agent)
[antecedent]co (S < Od)co (BE) sent Oi (by-agent)
Miscellaneous IV V
(S) SEND Oi Od
(S) SEND [Oi:NP] Od
miscellaneous sum total p
5
1.4
5
1.8
29
8.3
27
9.5
9
2.6
4
1.4
2
0.6
1
0.4
130
37.1
69
24.2
6
1.7
4
1.4
3
0.9
2
0.7
11
3.1
20
7.0
10
2.9
3
1.1
1
0.3
0
0.0
7
2.0
2
0.7
0
0.0
1
0.4
0 0.0 350 100.0 < 0.001
1 285
0.4 100.0
in the case of GIVE, collapsing the derivative categories into single categories has no negative influence on the significance of the distribution of verb-complementational patterns of SEND in ICE-GB and ICE-India. Apart from stating statistic significance for the distribution as a whole, Table€6-3 also shows in which of the patterns the differences between the corpora are particularly marked. Of special interest is the type-II pattern, where no variation between the two varieties can be attested. Thus, the main variety-specific differences in the complementation of SEND – at least when different modes or text types are not taken into account – are found in the basic ditransitive pattern, typeI, the monotransitive pattern, type-III, and in the passive of the type-III pattern.
Chapter 6.╇ SEND
Table 6-2.╇ Complementation of SEND – focus on central patterns Type
Pattern
ICE-India sum
I (S) SEND [Oi:NP] [Od:NP] Ider all related to type-I IP [S < Oi active] BE sent [Od:NP] (by-agent) IPder all related to type-IP II (S) SEND [Od:NP] [Oi:PPto] IIder all related to type-II IIP [S < Od active] BE sent [Oi:PPto] (by-agent) IIPder all related to type-IIP III (S) SEND [Od:NP] Oi IIIder all related to type-III IIIP [S < Od active] BE sent Oi (by-agent) IIIPder all related to type-IIIP IV (S) SEND Oi Od V (S) SEND [Oi:NP] Od Vder all related to type-V sum total p
%
54 15.5 5 1.4 0 0.0 0 0.0 78 22.3 5 1.4 29 8.3 11 3.2 130 37.1 9 2.8 11 3.1 11 2.2 7 2.0 0 0.0 0 0.0 350 100.0 < 0.01
ICE-GB sum
%
62 7 7 1 64 10 27 5 69 6 20 3 2 1 1 285
21.8 2.5 2.5 0.4 22.5 3.6 9.5 1.8 24.2 2.1 7.0 1.1 0.7 0.4 0.4 100.0
Table 6-3.╇ Distributional differences of SEND: A chi-square test Type I Ider II IIder IIP IIPder III IIIder IIIP IIIPder others sum total
ICE-India observed 54 5 78 5 29 11 130 9 11 11 7 350
ICE-India expected 63.94 6.61 78.27 8.27 30.87 8.82 109.69 8.27 17.09 7.72 10.47 305.02
χ²
ICE-GB observed
1.54 0.39 0.00 1.29 0.11 0.54 3.76 0.06 2.17 1.40 1.15 12,41
62 7 64 10 27 5 69 6 20 3 12 285
ICE-GB expected 52.06 5.39 63.73 6.73 25.13 7.18 89.31 6.73 13.91 6.28 8.53 284.98
χ² 1.90 0.48 0.00 1.59 0.14 0.66 4.62 0.08 2.66 1.72 1.41 15.26
99
100 Structural Nativization in Indian English Lexicogrammar
However, if we compare Table 6-3 with Table 5-3 in Section 5.1, which contained detailed chi-square test data for the verb GIVE, we find that the differences in the complementation of SEND are less prominent. While in the case of GIVE the type-I pattern alone would have sufficed to account for significant differences in the distribution, in the case of SEND the differences are relatively evenly distributed among many different patterns. After this overview of the total differences of the complementation of SEND in ICE-India and ICE-GB, in the following a more detailed study of the differences between the spoken and the written mode and the different text types will be provided.
6.3
Distribution of complementation patterns in different text categories
The overview of the verb-complementational patterns of SEND provided in the previous section serves as a good starting point to describe the differences between Indian and British English. It is remarkable, for example, that there does not seem to be a default pattern for SEND in British English, as the type-I, the type-II and the type-III patterns are represented in relatively equal proportions. In ICE-India the situation is different, insofar as there is a clear tendency towards the use of the type-III pattern. The type-II pattern is represented equally in both corpora, while the type-I pattern is clearly underrepresented in ICE-India. However, the lack of a favoured default pattern for SEND in British English may to some extent be misleading. In this regard, Mukherjee (2005) points out that: […] from larger-scale corpus analyses the conclusion may be drawn that the type-III pattern functions as the default pattern of SEND […]. The Longman Grammar, in which frequency information is obtained from a database that is much larger than ICE-GB […], points out that the pattern ‘SVOd(NP)’ (which is equivalent to the type-III pattern) clearly outnumbers the patterns ‘SVOiOd(NP)’ and ‘SVOdOi(PrepP)’ (i.e. the type-I pattern and the type-II pattern respectively) in all genres (cf. Biber et al. 1999:â•›390). (Mukherjee 2005:â•›181)
Although the differences in corpus size between ICE-GB and the Longman Grammar database will without doubt influence the verb-complementational profile of a verb, there may also be differences in terms of corpus design, which makes a closer look at the categories included and the dispersion of the different complementation patterns within these text types a worthwhile endeavour. Particularly in the case of the type-I pattern, the corpus design of the ICE corpora plays a major role in terms of its relative frequency (especially in British English). Table€6-4
Chapter 6.╇ SEND 101
Table 6-4.╇ Distributional differences of SEND according to text types ICE-India Cat.
S1A S1B S2A S2B W1A W1B W2A W2B W2C W2D W2E W2F sum
Type I
I der
IP
IP der
II
II der
IIP
IIP der
III
III IIIP IIIP der der
8 1 0 0 0 45 0 0 0 0 0 1 55
1 0 0 0 0 4 0 0 0 0 0 0 5
0 0 0 0 0 0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0 0 0 0 0 0
5 6 7 13 1 39 2 2 3 0 1 3 83
2 0 0 3 0 0 0 0 0 0 0 0 4
2 7 3 5 1 10 1 1 3 0 0 0 33
0 1 2 1 0 1 0 0 2 0 0 1 8
24 11 14 7 4 59 6 2 4 2 2 5 140
1 0 0 0 0 10 0 0 0 0 0 0 11
III IIIP IIIP der der
1 3 0 2 0 5 0 1 0 0 0 0 12
0 0 0 0 0 6 0 3 0 0 0 1 10
IV
V
4 1 2 0 0 0 0 0 0 0 0 0 7
0 0 1 0 0 1 0 0 0 0 0 0 2
IV
V
1 0 1 0 0 0 0 0 0 0 0 0 2
0 0 0 0 0 0 0 0 0 0 0 0 0
ICE-GB Cat.
S1A S1B S2A S2B W1A W1B W2A W2B W2C W2D W2E W2F sum
Type I
I der
IP
IP der
II
II der
IIP
IIP der
III
11 3 4 3 0 35 0 1 0 6 0 2 65
2 1 1 0 0 2 0 0 0 0 0 1 7
1 0 0 0 0 1 0 0 0 2 0 0 4
2 0 0 0 0 0 0 0 0 0 0 0 2
18 10 6 7 1 14 2 2 2 4 0 1 67
0 0 4 2 0 4 0 0 1 0 0 0 11
0 1 3 1 3 6 0 3 1 0 1 0 19
3 0 2 0 0 1 0 1 0 1 0 0 8
17 6 21 9 0 12 0 3 1 1 1 3 74
0 0 1 0 0 1 0 0 0 0 1 0 3
1 5 2 3 0 3 0 2 0 3 0 2 21
3 0 0 0 0 0 1 0 0 0 1 1 6
gives an overview of the dispersion of the different complementation patterns across the different modes and text types.54 54. Note again that the figures in this table differ slightly from the data given by Mukherjee & Hoffmann (2006).
102 Structural Nativization in Indian English Lexicogrammar
As shown in Table 6-4, the majority of the uses of the type-I pattern are clustered within a single text category in both corpora. This text category W1B contains both social and business letters and is thus a category that naturally contains many instances of the verb SEND (as sending and receiving is an integral part of correspondence). With the exception of this category, the type-I pattern is only used in the spoken part of ICE-India and more frequently in the spoken part of ICE-GB. This dominance of the W1B category for the use of SEND also plays a certain role with regard to the other basic patterns, especially concerning ICEIndia. However, these patterns are also used quite frequently in the other text categories, while this is much less the case for the type-I pattern. In the following, I will discuss these differences of pattern usage in the spoken and written modes according to different text types and varieties, in order to shed some light on possible explanations for the different preferences in the varieties.
6.4
Verb-complementational patterns of SEND in spoken Indian and British English
As shown in Table 4-7, SEND is used significantly less frequently in the spoken section of ICE-India than in the spoken section of ICE-GB. This trend reflects the much more frequent use of GIVE described in Chapter 5, as in many cases GIVE and SEND can be used nearly synonymously. As speakers of Indian English tend to use GIVE in a larger variety of contexts, it seems plausible that other ditransitive verbs that can in theory be substituted by GIVE are used less frequently in Indian English than in British English. This general underuse, on the other hand, makes it somewhat difficult to draw conclusions about the reasons for the complementational differences attested between the corpora, as the overall frequencies are relatively low and the complementational differences are not as marked as in the case of GIVE (especially regarding the type-II pattern where no significant differences are attested). If we look at the four different text categories of the spoken sections of ICEGB and ICE-India, we find several properties of SEND in the different varieties that are not attestable if we view the data as a whole. With regard to the central patterns, the main differences are found in the use of the type-I pattern, although some differences are also attested for the use of the type-III pattern. In the Indian corpus, the use of the type-I pattern is restricted to the dialogue part of the spoken section (9€instances, 6.52%), whereas in ICE-GB, although used more frequently in dialogue (14 instances, 12.65%), it is also found in the monologue section (7€instances). In terms of the dialogue section, it occurs more often in direct conversations than in distanced conversations, even when the
Chapter 6.╇ SEND 103
smaller size of the S1B category is taken into account. In almost all of the cases, the type-I pattern is used with pronominal indirect objects; the only exception here is found in the unscripted monologue section (S2A) in ICE-GB. The derivative patterns, type-Ider, are only attested once in the spoken section of ICE-India (S1A) and with four instances a little more frequently in ICE-GB, so the type-I pattern seems to be more versatile in the British data. However, with figures so low, no concrete claims can be made about this on the basis of the data. The passive type-IP is virtually not used in the spoken parts of the corpora (one instance in ICE-GB). This is not surprising, as the passive is used relatively infrequently in spoken language. The distributional differences in the type-II pattern – as pointed out in Table€6-3 – are relatively small, and no statistic significance can be claimed for any specific tendency. This pattern is used in 23.2% of all instances of SEND in the spoken section of ICE-GB and in 24.7% of all instances of SEND in ICE-GB, which shows that differences are indeed minimal. This notwithstanding, there are some differences between the varieties in the preference for the passivization of the type-II pattern. While passivization is relatively rare in ICE-GB (5 instances, 3.01%), this pattern is used relatively frequently in ICE-India (17 instances, 12.32%). Thus, the pattern is used almost twice as frequently in the spoken section of ICE-India as the type-I pattern. In ICEGB, this pattern is solely used with to-prepositional phrases denoting locations such as prison, England or market – the transfer-caused-motion pattern discussed above€– where the ditransitive construction is not a permissible alternative. While this is also true for ICE-India in the majority of cases, the type-II pattern is also used in a transfer-caused-motion construction that has an admissible ditransitive alternative, as shown in Example (50): (50) Reports were sent to parents on twenty-third November ninety-four
However, with only two instances of this use in ICE-India, it remains doubtful whether this difference in use may serve as an explanation for the stronger tendency to use the type-II passive pattern in the spoken section of ICE-India, although it is somewhat surprising that the stronger preference for passivization in the spoken data of ICE-India is attested only for this pattern and not for the type-I or type-III patterns. The type-III pattern is clearly the default complementation pattern of SEND in spoken British and spoken Indian English. In the spoken part of ICE-GB it is attested 53 times (31.93%), while in ICE-India it is used even more frequently, namely 56 times (40.58%). The tendency of Indian speakers to use SEND more frequently in monotransitive complementation than their British counterparts
104 Structural Nativization in Indian English Lexicogrammar
reflects British speakers’ tendency towards a more frequent use of the type-I pattern. However, this observation is not sufficient for the simplified assumption that Indian speakers use the type-III pattern in instances where British speakers use the type-I pattern. The type-III pattern is not used more frequently in Indian English than in British English in the categories where Indian speakers do not use the type-I pattern at all (S2A and S2B). On the other hand, speakers of British English use the type-III pattern less in the S1A-category, where most of their use of the type-I pattern is attested. Thus, the type-I pattern may often be an alternative for the type-III pattern for speakers of British English, while at the same time the type-III pattern is not used as an alternative for the type-I pattern in the Indian data. In summary, the analysis of the complementation patterns in the spoken parts of ICE-India and ICE-GB has shown that SEND is generally used less frequently in spoken Indian English than in spoken British English. While in spoken British English the type-I pattern is used in 12.7% of all instances, this pattern is much rarer in ICE-India (6.5%). No clear differences are attested for the type-II pattern, although Indian speakers use the passive subtype of this pattern more frequently than speakers of British English. The type-III pattern is the default pattern in both varieties, with speakers of Indian English (as in the case of GIVE) using this pattern proportionally more often than speakers of British English.
6.5
Verb-complementational patterns of SEND in written Indian and British English
As has been mentioned above, in the written components of ICE-India and ICEGB SEND is used very frequently in the text category W1B, which contains 15 texts of 2,000 words of business letters and 15 texts of social letters. In ICE-India SEND is used 180 times in W1B alone, which amounts to 78.3% of the 232 occurrences in the whole written component and 48.5% of the use of SEND in terms of written and spoken components. In ICE-GB these figures are a little lower, with 79 out of 135 occurrences (58.5%) in the written corpus; and 27.2% of all instances of the verb in ICE-GB. Although these figures seem to indicate that SEND is vastly overrepresented here, it is not surprising that figures are very high in this category. The question is not one of an overrepresentation of SEND in the letters-category, but more a question of corpus design. If a category containing business and social letters is included in a corpus of written and spoken English, how large should that category be? Are 60,000 words out of 1,000,000 representative of the average language user’s contact with business and social letters? There is an ongoing discussion
Chapter 6.╇ SEND 105
of these questions concerning the representativity and balancedness of corpora. Leech (2007), for example, points out that: […] balancedness is very difficult to demonstrate, even for such painstakingly constructed corpora [as the Brown corpus, the BNC and ICE-GB]. An obvious way forward is to say that a corpus is ‘balanced’ when the size of its subcorpora (representing particular genres or registers) is proportional to the relative frequency of occurrence of those genres in the language’s textual universe as a (Leech 2007:â•›136) whole.
However, this proportionality of the genres to the population of texts is often very hard to establish, especially when questions of varieties and diachronic variation are taken into account. The textual universe arguably differs in various speech communities and the fact that the compilation of ICE corpora is an ongoing process further complicates matters. Regarding comparable corpora, Leech (2007) goes on to describe this with the example of the LOB/Brown corpus family: This brings me to a more fundamental challenge to comparability: Genre Evolution […]. It is increasingly being recognized that the genres on which stratified sampling of many corpora is based are themselves subject to change. New genres emerge; old genres decay […]. As a case in point, we had problems filling the slots in the Lancaster 1931 Corpus for science fiction and sociology texts€– two sub-genres that were emergent at that time. (Leech 2007:â•›143)
The genre evolution is not as drastic in the case of the ICE corpora because the diachronic gaps between them are not as large as between the corpora of the LOB/Brown family. However, with regard to letters, there have been immense changes over the last few decades.55 Arguably, both social and business letters no longer play much of a role in many modern speech communities, as this form of communication has mainly been taken over by email.56 Thus, although there are grounds for accepting the high frequency of SEND in this category, I would still advise taking the results here with a certain degree of caution, as it is not certain to what extent the inclusion of this category reflects the actual importance of letters for language users in both India and Great Britain. 55. A further issue regarding the ICE corpora in comparison to the LOB/Brown family is that the diachronic gap in the latter is intentional, while for the former it is more of a result of the long periods of time that the compilation of each ICE corpus takes. This makes it difficult – if not impossible – to include comparable spoken texts. In order to avoid a diachronic gap within a single ICE corpus, it is necessary to include written texts from the period of corpus compilation, rather than from the early 1990s as originally envisioned. 56. This is a point reflected in the substitution of letters by email in some of the newer ICE components, such as, for example ICE-Singapore.
106 Structural Nativization in Indian English Lexicogrammar
For these reasons, the following analysis will be twofold. In a first step the distribution of the different verb-complementational patterns of SEND will be analyzed, neglecting the W1B category. Especially for ICE-India, however, this will lead to very low figures for some of the complementation patterns, so that it will not be possible to provide sound results for some of the less central patterns. On the other hand, this also puts the results of Table 4-7 into perspective, where it was shown that SEND is significantly less frequently used in spoken than in written Indian English and significantly more frequently used in written Indian English than in written British English. As almost 80% of the total occurrences of SEND are to be found in the W1B category, this category alone is responsible for these differences and should therefore be analyzed in its own right. This will be the second step of the present analysis – a comparison of the data found in W1B in ICE-India and ICE-GB. If we neglect the data from W1B, we find that SEND is used 52 times in total in ICE-India, compared to 56 times in ICE-GB. The main differences in usage regarding verb-complementation is again attested in the type-I and the type-III patterns. In terms of the type-I pattern, SEND is used nine times in this pattern (16.1%) in ICE-GB, whereas it is only attested once in ICE-India (1.9%). Six out of the nine occurrences in ICE-GB are found in the category W2D (instructional writing); in all of these cases the indirect object is realized by the pronoun you. Two further instances are attested in the category W2F (creative writing); here the indirect objects are her and each other. The last instance of the type-I pattern in ICE-GB is in category W2B (informational popular), where the pronoun you is used in the indirect object position. The only instance of this pattern attested in ICE-India is found in the creative writing category (W2F), where the indirect object is also realized by a pronoun; you in this case. Thus, this pattern is more frequently used in ICE-GB, where it is also used in a larger spectrum of text categories. In ICE-India this pattern seems largely restricted to spoken dialogue and creative writing (if W1B is ignored). As has been shown in the case of GIVE and in the discussion of the spoken data of SEND, this pattern is mainly used with pronominal indirect objects in both Indian and British English. In terms of the derivative and passive patterns of the type-I pattern, these are not attested in the written part of ICE-India and are relatively rare in ICE-GB too. Regarding the type-II pattern, no significant differences between the varieties are attested. This pattern is used 12 times in both the written datasets, representing 21.5% of the uses in ICE-GB and 23.1% of the uses in ICE-India. There is also no clustering of this pattern in specific text categories, so it is used in most of the categories in both corpora. The derivative patterns of the IIder type are virtually not used in the written parts of ICE-GB and ICE-India (one occurrence in ICE-GB). There is also very
Chapter 6.╇ SEND 107
little variation concerning the passive pattern type-IIP, which is used six times in ICE-India compared to eight times in ICE-GB. Thus, there are no variety-specific differences for this pattern attested in the data, either if we look at the cumulative picture, or analyze its use in different modes or text-types. In the analysis of the spoken data, the type-III pattern was attested as the default pattern for the complementation of SEND in both varieties, but the picture is not so clear when it comes to the written data. In ICE-India this pattern is very frequent (25 instances, 48.1%) and can be seen as the default pattern. However, in ICE-GB this pattern is only used nine times (16.1%), thus (without W1B) the type-II pattern is the most frequent pattern here, with the type-I pattern and the type-III pattern used with equal frequency. In ICE-India, the type-III pattern is used in all text categories, varying between six instances in W2A (informational learned) and two instances in W2D and W2E. In ICE-GB this pattern also occurs in the majority of text-types (albeit not in W1A and W2A), though only between one (W2C, W2D, W2E) and three (W2B and W2F) times. The type-IIIder pattern is only attested once in the dataset under scrutiny (ICE-GB, W2E), so a discussion of this pattern cannot be undertaken. The passive pattern, type-IIIP, however, is interesting in that the passive pattern is found more frequently in ICE-GB (seven instances, 12.5%) than in ICEIndia (one instance, 1.9%), whereas the basic pattern is used much less frequently. The categories where this pattern is used are instructional writing (three times) creative writing and informational popular (both twice). Thus, especially in the W2D category instructional writing, SEND is used more frequently in ICE-GB overall and with a greater variation of patterns. However, a qualitative analysis of these differences will not be undertaken at this point, as the possibility of corpusbased idiosyncrasies is too large considering the low frequencies overall. In the social and business letter category W1B, SEND is used 180 times in ICE-India and 79 times in ICE-GB. Thus, the more frequent use of SEND in written Indian English, compared to written British English and spoken Indian English described in Table 5-3, is largely based on this category. Accounting for 45 of the total 55 instances of the type-I pattern in ICE-India, this pattern is mainly used in this category. However, the type-I pattern is not the default pattern for SEND in the Indian data from W1B; with 25% of the 180 instances it is clearly used less frequently than the type-III pattern (32.8%). The derivative pattern, type-Ider, only plays a minor role (four instances, 2.2%) and passivization of this pattern is not attested in W1B in ICE-India. In ICE-GB there are some strong differences from ICE-India in the W1B category. SEND is used much less overall in this category, with the result that the category has less influence on the cumulative pattern distribution compared to ICE-India.
108 Structural Nativization in Indian English Lexicogrammar
However, owing to the very frequent use of the type-I pattern in this category, the cumulative picture still becomes distorted. While in most other text categories of ICE-GB the type-III pattern is the default pattern for the complementation of SEND, in W1B the type-I pattern is used 35 times (44.3%), compared to the use of this pattern in the spoken part of the corpus (12.7%) and the rest of the written part of the corpus (16.1%). As in ICE-India, the type-Ider pattern is rarely used in W1B of ICE-GB (three instances) and there are no instances of the passive pattern type-IP. Regarding the type-II pattern, the differences between ICE-India and ICEGB are not as prominent. In ICE-India, the type-II pattern is used 39 times in total (21.7%), whereas it is used slightly less frequently (14 instances, 17.7%) in ICE-GB. Although these differences are not as distinctive as the differences between the varieties in the type-I and the type-III patterns, they stand out when compared to the cumulative picture. Derivative pattern type-IIder is not attested in ICE-India, while there are some occurrences of this pattern in ICE-GB (four instances), all of which contain fronted direct objects, so this pattern is slightly more productive in British English. The type-IIP passive pattern is used slightly more frequently in W1B of ICEIndia in absolute terms (ten times in ICE-India, six times in ICE-GB) but slightly more frequently in ICE-GB in relative terms (7.6% compared to 5.6%). However, due to the scarcity of the data, a more detailed analysis of these differences cannot be undertaken. As mentioned above, there are also marked differences between the varieties regarding the use of the type-III pattern. In ICE-India, with 59 instances (32.5%), this pattern is used only slightly less frequently in this category than in the spoken part of the corpus, which is in accordance with the more frequent use of the type-I pattern in this category. In the W1B category of ICE-GB, the type-III pattern is used 12 times (16.1%) and is, thus, used to roughly the same extent in this category as in the rest of the written data of ICE-GB.57 Thus, while in ICE-India the type-III pattern is the most frequent in both the spoken and the written component (38.1% and 32.8% respectively), in ICE-GB the picture is quite different. Here the type-III pattern is the default pattern in the spoken component of the corpus (31.9%), whereas in the written component the type-I pattern is the most frequent pattern overall (32.6%), followed by the typeII pattern (19.3%) and only then the type-III pattern (15.6%). A large majority of the instances of the type-I pattern are found in the W1B category; discounting 57. Note here, however, that in this category the type-I pattern is used 35 times, so the percentages will take this frequent use into account, while in the rest of the data both the type-I and the type-III pattern are used nine times.
Chapter 6.╇ SEND 109
this category, we find that the type-III pattern and the type-I pattern are used with equal frequency (16.1%), while the type-II pattern is used slightly more frequently (21.4%). Thus, in ICE-GB the use of the type-III pattern seems far more mode-dependent than the equivalent pattern in ICE-India. The type-IIIder derivative pattern is mainly attested in ICE-India (ten instances, 5.6%); in ICE-GB this pattern occurs only once. Typically fronted objects or relative clauses are used here, as Example (51) demonstrates: (51) The balance of Rs. 267/– we are sending by D.D. No. 391460 dated 28 – 8 – 95 drawn on Bank of Baroda, service Branch N. Delhi.
The only occurrence of this pattern in ICE-GB is shown in (52): (52) Just a reminder that you need to let me know the names of people whom you want to send on first aid courses.
Although Examples (51) and (52) illustrate why this pattern may be more productive in ICE-India, as the fronting of the direct object in this case may not be the preferred choice of speakers of British English, the lack of additional reference data makes such claims tentative. The passive pattern type-IIIP is relatively infrequent in both datasets (five instances in ICE-India, three instances in ICE-GB). Typical of the use of this pattern is the omission of the by-agent, which is the default case in the ‘correspondence’ category, as the subject that would be realized by the by-agent is very often (in the present dataset always) the author of the respective letter, and therefore deducible from the context.
6.6
SEND in the International Corpus of English: A synopsis
The analysis of SEND in the Indian and British components of the International Corpus of English has shown that there are distributional differences in the complementation patterns with regard to text type and mode, both within the two varieties and across the varieties. As SEND is much less frequent than GIVE in both varieties, the main focus of the analysis has been on the central patterns, type-I, type-II and type-III. The type-I pattern is mainly used in the spoken parts and the W1B (letters) category of the corpora. In the spoken part of ICE-GB this pattern is used both in monologue and dialogue (10.1% of all spoken uses of SEND), while in ICE-India it is only attested in the dialogue sections of the spoken part of the corpus (6.1% of all spoken uses). In the written part of the corpora, SEND is used most frequently in category W1B in both corpora. In this category, the use of the type-I pattern is the preferred use in British English (44.3% of all uses
110 Structural Nativization in Indian English Lexicogrammar
in this category). In ICE-India, although the type-I pattern is mainly used in this category, at 25% it is only the second most frequent pattern for the complementation of SEND after the type-III pattern (32.8%). For the type-II pattern no significant differences between the varieties were attested in the data analysis of the whole corpora. Also with regard to different text types and modes, variety-specific differences are relatively small. An explanation for this may lie in the fact that in many cases there is no genuine alternative to this pattern. If we consider the type-II pattern as a realization of the transfer-causedmotion construction, it is often not possible to substitute it with the ditransitive construction, such as in cases where the recipient is inanimate. The type-III pattern is the default complementation pattern for SEND in the spoken parts of both corpora (38.1% of all spoken instances in ICE-India, 31.9% in ICE-GB). When the W1B category is not discounted, this pattern is also the default pattern in the written data in ICE-India; when this category is discounted, the type-II pattern is the most frequent pattern used in ICE-GB. Besides the fact that the type-III pattern is the most frequent one in both varieties overall, it is used significantly more frequently in Indian English, which is in line with the result for GIVE. Hence, a more frequent monotransitive use of potentially ditransitive verbs in Indian-English was attested for GIVE and SEND in the ICE-data. The passive patterns type-IP, type-IIP and type-IIIP are relatively infrequent in both corpora, so text type-specific or mode-specific results are hardly interpretable. Regarding the whole corpora, there is a higher tendency for the use of the passive patterns in British English. The type-IP pattern is used in 1.4% of the total number of instances of SEND in both corpora, the type-IIP pattern is used in 8.7% of the cases in ICE-India and in 9.2% of the uses of SEND in ICE-GB. The type-IIIP pattern is used in 3.15% of all instances in ICE-India compared to 6.9% in ICE-GB. Thus, while the type-III pattern is more productive in Indian English, its passive version is more frequent in British English.
6.7
Verb-complementational patterns of SEND in Indian and British newspaper English
As described above, SEND is much less frequent in both Indian and British English than GIVE. Owing to this lower frequency, many points about possible variety-specific differences could not be discussed in the previous ICE-based section. Firstly, peripheral patterns of verb-complementation were not discussed in detail, as they are only used marginally. Secondly, and maybe more importantly for the description of possible nativization in the case of SEND, the low overall frequencies of the verb do not allow for a meaningful collocational analysis
Chapter 6.╇ SEND
based on the criteria described in Chapter 4. To provide further insights into the preferred use of SEND in the two varieties, two additional analyses were undertaken. These two analyses, one concerning verb-complementational profiles, the other one concerning collocational profiles of SEND, are based on two different datasets. The dataset used for the verb-complementational analysis is a randomly derived subset of approximately 1.75 million words from the Times of India corpus (ToI1.75) and 1.77 words from the newspaper section within the BNC periodical section (BNCNews.1.77). These reduced corpora contain 528 (ToI1.75) and 503 (BNCNews.1.77) instances of SEND respectively. At this point it should be kept in mind that this additional dataset only contains data from newspapers and is clearly genre-restricted. For the collocational analysis that follows the verb-complementational analysis, an even larger dataset was used, as with approximately 500 total occurrences of the verb there are still too few collocation candidates. This additional dataset contains all newspaper texts from the BNC amounting to nine million words of text (BNCNews) and a randomly chosen part of the ToI of comparable size (ToI9m). As discussed in the previous section, phrasal and prepositional verbs were not taken into account in the first step of the verb-complementational analysis of SEND. The idiomatic expression SEND Od Ving – schematized as (S) SEND [Od: NP] [CompC:VPcont], as in Example (53) – has been treated as a complementation pattern of SEND in its own right. (53) Torrential rain put a damper on the event, sending bedraggled guests squelching across lawns to seek shelter. (BNC k41)
It could be argued that Example (53) is a case of type-III complementation, as it is semantically analogous to sentences that realize the gerund with a to-infinitive as in Example (54): (54) ?Torrential rain put a damper on the event, sending bedraggled guests to squelch across lawns to seek shelter.
However, as the to-infinitive sentence is an invented example and the expression SEND Od Ving is the more natural expression (at least in British English), this expression has been treated as a special case. Furthermore, differences in the use of phrasal and prepositional verbs, which are used much more frequently in the BNC data than in the ToI data, will be discussed at the end of this section, in order not to neglect findings on the level of lexicogrammatical nativization on purely methodological grounds. Table 6-5 gives an overview of the complementational profile of SEND in the ToI and BNC regarding absolute and relative frequencies.
111
112 Structural Nativization in Indian English Lexicogrammar
Table 6-5.╇ Complementation of SEND in ToI1.75 and BNCNews.1.77 Type
Pattern
ToI1.75 sum
I Ider IP IPder II IIder IIP
3.80
18
4.93
all related to type-I
0.21
0
0.00
[S < Oi active] BE sent [Od:NP] (by-agent)
2
0.42
6
1.64
0
0.00
1
0.27
140
29.54
91
24.93
all related to type-IP
(S) SEND [Od:NP] [Oi:PPto]
all related to type-II
[S < Od active] BE sent [Oi:PPto] (by-agent)
5
1.05
11
3.01
96
20.25
59
16.16
2.11
16
4.38
30.17 101
27.67
10
III
(S) SEND [Od:NP] Oi
143
all related to type-III
[S < Od active] BE sent Oi (by-agent)
IIIPder
all related to type-IIIP
IV
(S) SEND Oi Od
V
%
1
all related to type-IIP
IIIP
sum
18
(S) SEND [Oi:NP] [Od:NP]
IIPder IIIder
%
BNCNews.1.77
(S) SEND [Oi:NP] Od
SEND Od Ving (S) SEND [Od:NP] [CompC: VPcont] sum total
2
0.42
4
1.10
35
7.38
21
5.75
10
2.11
8
2.19
1
0.21
1
0.27
0
0.00
0
0.00
10
2.11
28
7.67
474
100.00
365
100.00
The figures in Table 6-5 show that there are far fewer distributional differences between the varieties than had been the case for both SEND and GIVE in the ICE data. To a certain extent this is not surprising, since news reportage is a strongly stylized text type, especially when it is compared to some of the spoken data contained in the ICE corpora. As Sand (2004) points out:58 […] the findings presented above confirm the importance of text type in linguistic analysis. In all varieties, differences across text types are observable and genre differences within one variety are practically always more pronounced than (Sand 2004:â•›294–295) overall variation across varieties.
These observations lead Sand (2004) to the conclusion that generalizations about the nature of any variety of English based on a single text type would be highly questionable (cf. Sand 2004:â•›295). Although this is certainly true, for many varieties of English, including Indian English, large-scale balanced corpora comparable to the BNC simply do not yet exist. In order to carry out a lexicogrammatical analysis, as undertaken in the present study, the data in the ICE corpora are simply insufficient, so there is a need for a larger database, even if it contains 58. Note that Sand’s (2004) study analyzed article use in different varieties of English.
Chapter 6.╇ SEND 113
Table 6-6.╇ Distribution of SEND in ToI and BNC – focus on central patterns Type I II IIder IIP IIPder III IIIP IIIPder SEND Od Ving others sum total
ToI1.75 observed
ToI1.75 expected
χ²
18 140 5 96 10 143 35 10 10 7 474
20.34 130.51 9.04 87.57 14.69 137.85 31.64 10.17 21.47 10.73 474.00
0.27 0.69 1.81 0.81 1.50 0.19 0.36 0.00 6.13 1.30 13.06
BNCNews.1.77 BNCNews.1.77 observed expected 18 91 11 59 16 101 21 8 28 12 365
15.66 100.49 6.96 67.43 11.31 106.15 24.36 7.83 16.53 8.27 365.00
χ² 0.35 0.90 2.34 1.05 1.94 0.25 0.46 0.00 7.96 1.69 16.94
only a single text type. Although, from the overview provided in Table 6-5, there does not seem to be much variation in the complementation of SEND across the varieties, it is still possible to look for variety-specific differences between the two datasets if a more detailed approach is adopted. The first step, therefore, is a closer examination of the distribution of the central patterns as shown in Table 6-6. With a total chi-square score of 30.11, this distribution is significant at the p€≤€.001 level for which the critical threshold is 27.88 (at df€=€9). Therefore, it is worth taking a closer look at the different complementation patterns and their influence on the total chi-square value, especially since some of these patterns have a large influence on the cumulative chi-square, while in others the difference between observed and expected values is almost negligible. Especially in the case of the type-III pattern and the type-IIIP patterns almost no differences between the varieties are observable. The type-III pattern, as the most frequent pattern in both varieties, is thus the preferred choice for the complementation of SEND in newspaper articles. This is in line with the findings from the ICE corpora, as there the main variation in the use of this pattern was attested in the spoken part and in the letter category. It is also in line with the observations of Biber et al. (1999:â•›390), who show that the pattern SVOd (NP)€– the corresponding pattern to the type-III pattern – is the most frequently used (between 50% and 75% of all instances) for SEND in all the genres they described, including newspaper writing. In the case of the type-II patterns and their derivatives, variety-specific variation has a stronger influence on the total distribution. By looking at the relation between the type-II pattern and its derivative patterns, and the type-IIP pattern with its derivatives, we see that the ToI corpus clearly favours the basic patterns,
114 Structural Nativization in Indian English Lexicogrammar
which are both used more frequently than expected, while in the BNC these patterns are used less frequently. The opposite is the case for the derivative patterns, which are used more frequently in the BNCNews.1.77 (compared to expected values) than in the ToI1.75, both actively and passively. The type-I pattern, finally, is infrequent in both corpora and the contribution to the cumulative χ² is thus relatively small. This reflects the observation from the ICE corpora where SEND, used in the ditransitive type-I pattern, is used more frequently in spoken text or text types closely related to the spoken mode. Besides the basic complementation patterns, type-I to type-III, SEND is frequently used in a pattern that is not applicable to the other verbs under scrutiny. As discussed above, the pattern SEND Od Ving was therefore treated as an additional possible pattern for SEND. As shown in Table 6-6, the differences between the varieties in the use of this pattern of SEND contribute strongly to the cumulative chi-square value. With 28 instances, this pattern is almost three times as frequent in the BNCNews.1.77, compared to the ToI1.75. This more frequent use is mainly due to a usage of this pattern in the BNCNews.1.77 where movements on the financial market are described. Consider Example (55): (55) But no-one complained about yesterday’s figures, which sent the shares soaring 25p to 334p in a plunging stockmarket.
Out of the 28 attested instances of this pattern, in 14 cases SEND is used in sentences similar to Example (55), i.e. describing changes in the price of shares or currencies in the BNC newspaper data, while no similar use of this pattern is attested in the ToI data. The 14 other uses in the BNCNews.1.77 are not as homogeneous in terms of the topic. Here, six instances are from sports commentaries, where verbs in the complement clause include packing, crashing, sprawling and hurtling, while the other eight instances stem from a variety of topics, such as politics or cultural reportage. In the ToI1.75 data, there is also a tendency to use this pattern in connection with sports reportage (five instances), while, like in the BNC, the rest of the uses are found in a larger variety of topics. Apart from the quantitative differences in the use of this pattern, it should be pointed out that the pattern is used in a relatively stable way: in almost all instances, in both corpora, this pattern describes an action that is conceived negatively, so that it can be argued that the whole construction has a negative semantic prosody (cf. Louw 1993:â•›158–159).59 The only possible exception for this negative prosody of the 59. Louw provides the following definition of semantic prosody: “… the phenomenon which Sinclair (personal communication 1988) has begun to refer to as a ‘semantic prosody’, applying the term ‘prosody’ in the same sense that Firth (in Palmer 1966:â•›40) used the word to refer to phonological colouring which was capable of transcending segment boundaries. The nasal
Chapter 6.╇ SEND 115
pattern is found in the ToI1.75, but even in this example it can be argued that by use of the negatively connotated pattern a sense of irony is created: (56) Pakistan-bashing on the silver screen, specially when it is a cat and mouse game between an Indian spy and the Pakistani authorities, offers tremendous scope for high-pitched dialogues and confrontations. But is that enough to send the cash register ringing? Well, cricket has shown that Pakistan bashing is a hit for sure! (ToI 37697)
By looking at the context of the sentence, which makes use of the SEND pattern in question, it can clearly be seen that the author employs the negative prosody of the pattern as a stylistic device. An action that might be conceived positively (ringing cash register as a metaphor for earning money) receives a negative connotation through the use of this pattern in order to emphasize the author’s regret that “Pakistan-bashing” appears to be popular in India. This example shows that, although the use of this pattern in business reportage is not attested in the ToI and, thus, seems to be restricted to British usage, the pattern is still used quite productively and creatively in Indian newspapers. Bearing in mind the low number of differences in pattern selection, especially in the basic patterns, a closer look at SEND used in particle verb constructions will conclude this section. There are considerable differences between the varieties when the use of SEND proper is compared to the use of particle verbs. Out of the total of 548 instances of SEND in the ToI1.75 and 503 instances in the BNCNews.1.77, 74 instances in the ToI1.75 and 138 in the BNCNews.1.77 were disregarded for the analysis. Of those, 65 instances in the ToI1.75 and 126 instances in the BNCNews.1.77 are uses of SEND as part of a particle verb (all phrasal verbs). The rest of the disregarded instances were either duplicates, or SEND was not used as a verb, and in a few cases sentences were not analyzable, as obligatory phrase elements are missing. In particular the difference in the use of phrasal verbs is remarkable, as in the British newspapers SEND + particle is used almost twice as frequently as in the Indian data, although the Indian data contains slightly more instances of SEND overall.60 Table 6-7 gives an overview of the distribution of SEND used in phrasal-verb constructions with regard to the different particles used in the ToI and the BNC datasets.
prosody in the word Amen would be an example: we find the vowels are imbued with a nasal quality because of their proximity to the nasals m and n. In the same way, the habitual [negative€– M. Sch.] collocates of the form set in are capable of colouring it, so it can no longer be seen in isolation from its semantic prosody […]” (Louw 1993:â•›158–159, his italics). 60. That is the grapheme SEND, including SEND proper and all neglected forms.
116 Structural Nativization in Indian English Lexicogrammar
Table 6-7.╇ SEND as a particle verb in ToI and BNC SEND
Particle
ToI1.75
%
BNCNews.1.77
%
back out in up on off over down across away through sum
24 16 8 1 3 1 1 5 4 1 1 65
36.92 24.62 12.31 1.54 4.62 1.54 1.54 7.69 6.15 1.54 1.54 100.00
30 26 9 1 4 52 2 2 0 0 0 126
23.81 20.63 7.14 0.79 3.17 41.27 1.59 1.59 0.00 0.00 0.00 100.00
The observation that SEND is used more frequently in particle-verb constructions in the BNCNews.1.77 data is in line with an earlier work by Schneider (2004), who compared the use of various particle verbs in five different ICE corpora. Although he only investigates intransitive particle verbs in his analysis of the total use of particle verbs (and therefore does not deal with SEND), he shows that there is an overall lower tendency to use particle verbs in India than in Great Britain: The first interesting result concerns the overall propensity to use PVs: It is very high in Singapore (at 68 occurrences per million words), where PVs are used even more frequently than in Great Britain (55), while, conversely, it is lower everywhere else and very low in Tanzania (29), India (26) and the Philippines (Schneider 2004:â•›235) (35[…]).
Besides the overall higher frequency of SEND as part of a phrasal verb, Table 6-7 also shows that there are strong differences in the use of this verb depending on the particle used. While in the BNCNews.1.77 SEND off is the most frequent particle verb (52 occurrences, 41.3%), in the Times of India SEND back and SEND out are used most frequently (24, 36.9% and 16, 24.62%). While these verbs are also used more frequently in the BNCNews.1.77 than in the ToI1.75 (30 and 26 occurrences), they are used less frequently in relation to the proportion of all particle verbs (23.8% and 20.63%). The very frequent use of SEND off in the BNCNews.1.77 is mainly due to its use in sports reportage in sentences like (57): (57) […] David Heron, has received the now customary eight-match ban after being sent off for a high tackle in the Regal Trophy quarter-final replay with Wigan.
Chapter 6.╇ SEND 117
From the 52 instances of SEND off in the BNCNews.1.77 data, 46 have a meaning comparable to Example (57), so that one reason for the frequent use of this pattern in the BNCNews.1.77 will be the inclusion of rugby reportage, where this type of occurrence is more frequent than in the cricket reportages that form the largest part of sports reportage in the ToI1.75. Thus, extralinguistic reasons exert a certain influence on the higher tendency towards the use of this particular phrasal verb. However, it should be pointed out that the only instance of SEND off in the ToI is also from a sports commentary, while there are six instances in the BNCNews.1.77 where SEND off is used in sentences where the particle does not contribute significantly to the meaning of the sentence, such as in Example (58): (58) I am always sending my producers off to the BBC music library, telling them to seek a breadth of listening: […]
This is in line with Schneider’s (2004) observation that there is a stronger tendency in British English towards the use of particle verbs that are nearly equivalent to the verbs proper than there is in Indian English. However, in the data analyzed by Schneider (2004) these differences were mainly found in the spoken data (cf. Schneider 2004:â•›241). The same holds for the verb SEND out, which is used 26 times in the BNCNews.1.77 – of which in 23 usages the particle does not add much information to the sentence compared to the use of SEND proper – in comparison with 16 instances of SEND out in the ToI data. As the figures in the Indian and British data for SEND back do not vary significantly, and the use of the particle is for the most part semantically obligatory, no further analysis of this form will be provided in the present study.
6.8
Collocational profiles of SEND in Indian and British newspaper English
In order to create collocational profiles for SEND that are comparable to the profiles provided for GIVE in Section 5.5, more data is required than can be provided by the ICE corpora, as SEND is far less frequent than GIVE. Therefore, the complete newspaper section of the BNC (BNCNews / approx. 9 million words) and a comparable random subset of the Times of India (ToI9m) have been used for the collocational profiles of SEND. The ToI9m subset contains the data that was used in Section 6.7 and additional random data. Owing to the vast number of collocation candidates that fulfil condition 1 and 2 of the restrictions for Od-collocates (cf. Section 5.5), these candidates were checked manually for the fulfilment of condition 4 prior to statistic testing. Thus, only nominal or pronominal candidates were
118 Structural Nativization in Indian English Lexicogrammar
Table 6-8.╇ Od-collocates of SEND in the Times of India (9 million word subcorpus) Times of India9m Od-collocates
log-likelihood
Times of India9m Od-collocates
log-likelihood
bells bill cassette children circular complaint copy data delegation fax feelers him information letter letters mail mails message messages money names note
â•⁄â•⁄ 98.09 â•⁄â•⁄ 44.72 â•⁄â•⁄ 83.39 â•⁄ 145.50 â•⁄â•⁄ 47.22 â•⁄â•⁄ 23.36 â•⁄â•⁄ 95.42 â•⁄â•⁄ 31.60 â•⁄â•⁄ 91.28 â•⁄ 101.63 â•⁄ 153.60 â•⁄â•⁄ 74.51 â•⁄â•⁄ 22.44 â•⁄ 714.64 â•⁄ 408.75 â•⁄ 190.44 â•⁄ 286.32 1015.82 â•⁄ 383.12 â•⁄â•⁄ 44.83 â•⁄â•⁄ 41.48 â•⁄â•⁄ 93.98
notice notices proposal recommendation recommendations reminder reminders reply report request samples signal signals sms smss sos students team teams them troops wards
255.60 163.43 368.35 118.91 â•⁄ 59.70 â•⁄ 59.62 164.54 â•⁄ 52.89 251.83 â•⁄ 65.42 215.07 215.79 467.09 302.56 â•⁄ 75.22 110.42 â•⁄ 67.67 484.35 225.79 170.81 230.70 â•⁄ 49.36
Table 6-9.╇ Od-collocates of SEND in the BNC newspaper section BNCNews Od-collocates
log-likelihood
BNCNews Od-collocates
log-likelihood
answers appeal card cash cheque children copy details donations envoy fax him letter letters
â•⁄ 42.66 â•⁄ 17.34 â•⁄ 82.35 â•⁄ 34.74 103.42 â•⁄ 75.20 237.59 â•⁄ 58.96 130.18 227.14 108.62 104.20 625.11 479.43
message messages money name picture prices representatives sae shares signal someone them troops video
633.92 143.20 â•⁄ 30.44 â•⁄ 12.47 â•⁄ 23.43 â•⁄ 38.22 â•⁄ 51.01 109.01 108.78 â•⁄ 90.91 â•⁄ 28.12 158.11 207.71 â•⁄ 24.23
Chapter 6.╇ SEND 119
retained and additionally checked for their use as a patient in a cause-receive process. Only instances where this condition was fulfilled were used to calculate the collocational statistics underlying conditions 2 and 3. The base, (i.e. SEND) has been lemmatized, although only collocation candidates that co-occur at least five times with a single word form are identified by WST 4.x. The collocates were not lemmatized, since in most cases this would have distorted the results, as most collocates fulfil the minimal occurrence criterion only in the singular or in the plural form. Collocates that fulfil the conditions described above in both singular and plural were treated separately. Table 6-8 shows all Od-collocates of SEND in the Times of India dataset. In a similar fashion, a comparable table can be generated for the BNC newspaper data. Table 6-9 thus shows all collocations in the position of the direct object in the BNC newspaper section. By comparing Table 6-8 and Table 6-9, it can be shown that there are a number of variety-specific Od-collocates as well as an area of overlap between the varieties, as shown in Table 6-10. In the following, verb-complementation patterns in relation to the Od-collocates of the two varieties will be analyzed and discussed. The first step is to compare the verb-complementational distributions of all the sentences that contain variety-specific Od-collocates in order to account for overall differences between the varieties with regard to SEND and its collocates. Subsequently, the verb-complementational profiles of sentences that contain specific Od-collocates will be provided, in order to show possible correlation of verb-complementational preferences with collocational preferences. In this analysis, Od-collocates that are attested in both varieties will be left out, as they are the focus of the third step of the analysis, in which possible differences in verb-complementation between the varieties in cases where the same Od-collocates are used will be addressed. Table 6-11 provides a chi-square test for the distribution of verb-complementational patterns for all the instances where significant Od-collocates co-occur with SEND. Table 6-11 allows several points to be interpreted. Firstly, in the Indian data there is a higher number of instances where SEND co-occurs with different Odcollocates, a point that is not very surprising, as the total number of Od-collocate-types is higher in the ToI9m data (44 types) than in the BNCNews data (28 types). However, the higher frequency of Od-collocate tokens in the ToI9m data is not only based on this higher number of Od-collocate types, but also on a higher average number of tokens per type (ToI9m: 22.03 tokens per type, BNCNews: 17.43 tokens per type). SEND is therefore used more frequently in co-occurring patterns in the ToI9m data than in the BNCNews data, both with regard to Od-collocate types and to Od-collocate tokens. Secondly, with a total χ²-value of 34.29,
120 Structural Nativization in Indian English Lexicogrammar
Table 6-10.╇ Od-collocates of SEND sorted by variety Od-collocates Times of India9m only
Od-collocates Times of India9m and BNCNews
Od-collocates BNCNews only
bells bill cassette circular complaint data delegation feelers information mail mails names note notice notices proposal recommendation recommendations reminder reminders reply report request samples signals sms smss sos students team teams wards
children copy fax him letter letters message messages money signal them troops
answers appeal card cash cheque details donations envoy name picture prices representatives sae shares someone video
the verb-complementational distribution of SEND co-occurring with significant Od-collocates shows significant variation between the varieties (p ≤ 0.0001, min. χ²-value at this level of significance = 33.72). Additionally, Table 6-11 shows which of the complementation patterns of SEND vary most strongly between the varieties, as in these patterns the discrepancies between observed and expected values and the resulting partial χ²-values are higher than in patterns that only vary minimally between the varieties. Thus, the more frequent use of the type-I pattern in British English is again attested in the collocation-based dataset. (This
Chapter 6.╇ SEND 121
Table 6-11.╇ Verb-complementational distribution of SEND + Od-collocates Type I II IIder IIP IIPder III IIIP IIIPder SEND Od Ving others sum total
ToI9m observed
ToI9m expected
χ²
BNCNews observed
BNCNews expected
χ²
41 419 7 94 45 271 32 32 17 11 969
51.21 425.64 10.64 83.80 39.90 259.38 29.26 27.27 25.27 16.63 969.00
2.04 0.10 1.25 1.24 0.65 0.52 0.26 0.81 2.71 1.90 11.48
36 221 9 32 15 119 12 9 21 14 488
25.79 214.36 5.36 42.20 20.10 130.62 14.74 13.73 12.73 8.37 488.00
4.04 0.21 2.47 2.47 1.29 1.03 0.51 1.63 5.38 3.78 22.81
pattern was also attested for GIVE and to a lesser extent for the non-collocationbased dataset in the last section on SEND.) Further, the pattern SEND Od Ving is more frequently used in the British data than in the Indian data, although the differences here are not as prominent as in the non-collocation-based sample of SEND (cf. Table 6-6). In the most frequent patterns, i.e. the type-II and the typeIII patterns, the variation between the varieties is relatively small, so it could be assumed (at least in the case of SEND) that nativization takes place more strongly at the periphery of verb-complementational patterns than at the centre. While Table 6-11 shows the verb-complementational distribution of all the sentences in the different corpora that contain significant Od-collocates, in the following a closer examination of the variety-specific Od-collocates will be provided. As was shown in Table 6-10, there are differences in the collocational profiles of SEND between the varieties. Therefore, the intention is to take a closer look at the correlations between the use of Od-collocates of SEND in the different varieties and the complementation patterns used. Analogous to the procedure in Section 5.5, the different significant Od-collocates in both varieties are sorted according to the complementation pattern preferred in sentences that contain these collocates. Table 6-12 shows the Od-collocates in the Times of India that correlate with type-II complementation. Out of the 32 Od-collocates of SEND that are only significant collocates in the ToI data, a majority of 18 correlate with the type-II pattern. As Table 6-12 shows, the strength of this correlation differs according to the collocates. In the case of the collocate wards, 85.7% of sentences are complemented by the type-II pattern, whereas this correlation is much weaker in the case of other collocates, such as bill or notice. The two Od-collocates that correlate most strongly with the type-II
122 Structural Nativization in Indian English Lexicogrammar
Table 6-12.╇ SEND in correlation to Od-collocates in Times of India9m (type-II dominant) Type II dominant (%) wards sos recommendation proposal reminders report feelers teams reminder students reply request information recommendations samples note bill notice
I
II
IIP
IIP der
III
IIIP
0.0 0.0 7.7 4.0 7.7 8.7 9.1 0.0 0.0 0.0 0.0 7.7 22.2 0.0 4.0 6.3 0.0 11.1
85.7 77.8 53.8 50.0 46.2 45.7 45.5 44.8 42.9 40.9 40.0 38.5 33.3 33.3 32.0 31.3 27.3 25.0
0.0 11.1 7.7 10.0 15.4 15.2 0.0 27.6 14.3 4.5 10.0 0.0 11.1 33.3 24.0 12.5 9.1 11.1
0.0 0.0 0.0 0.0 7.7 10.9 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 8.0 18.8 9.1 16.7
14.3 0.0 0.0 0.0 23.1 0.0 18.0 4.0 23.1 0.0 13.0 2.2 27.3 9.1 13.8 3.4 28.6 0.0 22.7 0.0 20.0 10.0 15.4 0.0 22.2 0.0 22.2 0.0 8.0 4.0 18.8 0.0 18.2 0.0 16.7 2.8
IIIP V-ing oth. der
N
0.0 0.0 0.0 4.0 0.0 0.0 0.0 0.0 0.0 13.6 0.0 7.7 0.0 0.0 8.0 6.3 0.0 8.3
7 9 13 50 13 46 11 29 7 22 10 13 9 9 25 16 11 36
0.0 0.0 0.0 10.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
0.0 11.1 7.7 0.0 0.0 4.4 9.1 10.3 14.3 18.2 20.0 30.8 11.1 11.1 12.0 6.3 36.4 8.3
pattern are low-frequency collocates. In contrast, Od-collocates that are used in the type-II pattern between 40% and 55% are mainly medium to high frequency collocates (seven to 46 occurrences). As has been discussed above, in many cases complementation with the type-I pattern is not a genuine alternative for the use of the type-II pattern, as the recipient in the cause-receive process is not an animate entity. This is reflected in the scarce use of the type-I pattern displayed in Table 6-12. However, while in some cases (such as teams), the type-I pattern is not used at all and would not be applicable, in other cases (such as reminders), both type-I and type-II complementation are used and type-II complementation is preferred, even if type-I complementation is applicable, as shown in Example (59): (59) I have sent several reminders to the commissioner […]
Thus, in the case of reminders, the placement of the recipient in end focus may be a reason for the frequent use of the type-II pattern. While a detailed discussion of all items shown in Table 6-12 would be beyond the scope of this work, one of the items, namely feelers, should be discussed here, as the collocation of SEND + feelers appears to be a specific Indian English
Chapter 6.╇ SEND 123
Table 6-13.╇ SEND in correlation to Od-collocates in Times of India9m (type-III dominant) Type-III dominant (%) smss data mails delegation names team mail sms casette
I
II
II der
IIP
IIP der
III
III der
IIIP
IIIP der
oth.
N
0.0 14.3 0.0 0.0 0.0 0.0 22.2 11.8 0.0
14.3 28.6 32.0 35.7 11.1 30.5 14.8 17.6 12.5
0.0 0.0 0.0 0.0 0.0 0.0 3.7 0.0 0.0
0.0 0.0 4.0 14.3 11.1 15.9 3.7 8.8 12.5
0.0 0.0 0.0 0.0 22.2 4.9 0.0 0.0 0.0
71.4 57.1 56.0 50.0 44.4 41.5 40.7 38.2 37.5
0.0 0.0 0.0 0.0 0.0 0.0 0.0 2.9 12.5
0.0 0.0 4.0 0.0 0.0 2.4 3.7 2.9 0.0
0.0 0.0 0.0 0.0 0.0 3.7 3.7 2.9 25.0
14.3 0.0 4.0 0.0 11.1 1.2 7.4 14.7 0.0
7 7 25 14 9 82 27 34 8
collocation that is rarely (if ever) found in British English. SEND + feelers has the meaning of discreetly and carefully conveying a message or evaluating somebody’s inclinations. Consider Example (60): (60) […] Musharraf said India had sent “some feelers” to Pakistan to resolve the Kashmir issue […]
Although the status of SEND + feelers as a unique Indian English collocation cannot be claimed solely on the basis of this data, the absence of this collocation in the BNCNews supports the hypothesis that it is a nativized Indian English collocation. While both in the Times of India9m data and in the BNCNews data the type-II pattern is the most frequent complementation pattern of SEND in relation to Odcollocates, there are also a number of collocates that are found predominantly in the type-III pattern. Table 6-13 thus complements Table 6-12 in showing all the variety-specific Od-collocates of SEND in the ToI9m data that are mainly complemented in the type-III pattern. Regarding the Od-collocates that correlate with type-III pattern complementation in the ToI data, a number of interesting occurrences should be remarked upon. Firstly, some of these collocates are only found in the Times of India9m for more corpus-specific than variety-specific reasons. This is the case for sms as well as for mail and mails, where e-mail and e-mails are included. These Od-collocates will be encountered much more frequently in the Times of India9m compared to the newspapers included in the BNCNews, as the Times of India9m data covers more recent editions than the BNC data and, therefore, in the BNCNews these new technologies are rarely to be found. The plural form smss, however, seems to be a preferred way of pluralizing sms in Indian English, while in British English
124 Structural Nativization in Indian English Lexicogrammar
Table 6-14.╇ SEND in correlation to Od-collocates in Times of India9m (no pattern bias) No specific tendency circular complaint notices signal
I
IP
II
IIP
IIP der
III
IIIP
IIIP der
oth.
N
0.00 0.00 5.26 0.00
0.00 0.00 5.26 0.00
42.86 33.33 21.05 27.91
0.00 0.00 26.32 0.00
0.00 33.33 0.00 0.00
42.86 33.33 26.32 27.91
0.00 0.00 10.53 2.33
0.00 0.00 5.26 4.65
14.29 0.00 0.00 37.21
7 6 19 43
the abbreviation sms in general is rarely used, as the terms text or message are preferred. A second interesting finding in Table 6-13 is the use of the collocate team. While the plural teams was used more frequently in the type-II pattern, the singular is clearly type-III dominant, as demonstrated in Examples (61) and (62): (61) On Wednesday the Trinamul would send two teams to Rishra and Bankura, on a fact-finding mission (62) Saudi Arabia sent a team of investigators to take part in the interrogation of the cell
In examples such as (61) the goal as well as the purpose of the team is profiled, while in Example (62) only the purpose is profiled. This difference in profiling is pervasive in the use of singular and plural of team and leads to the differences in pattern use. Regarding correlations of Od-collocates with verb-complementational patterns of SEND in the ToI data, there are finally a number of collocates that do not display a specific tendency towards any complementation pattern. These are shown in Table 6-14. Although the Od-collocates shown in Table 6-14 are not used predominantly in a specific complementation pattern, neverthless the overall tendency for SEND to be mainly complemented in the type-II or type-III pattern and only rarely in the type-I pattern is also given here. Of the four items, only notices is found as a direct object in a type-I pattern sentence, and in this case there is only a single occurrence. Circular and complaint are both infrequent, so it would be sensible to make no claims here, apart from the fact that they do not prefer a specific complementation pattern. In the case of signals, the high number of others should be noted. These are mainly particle verb constructions. For signals + SEND proper there is an equal distribution between the type-II and the type-III pattern and a rare use of the type-IIIP and type-IIIPder passive patterns. The last variety-specific Od-collocate of SEND in the Times of India9m dataset is the item bells. This item is mainly used in the phrase SEND alarm bells ringing,
Chapter 6.╇ SEND 125
Table 6-15.╇ SEND in correlation to Od-collocates (BNCNews) Od-collocate / pattern %
I
name sae answers envoy appeal representatives card video cheque someone donations shares details prices cash picture
0.0 0.0 0.0 0.0 0.0 0.0 23.1 16.7 7.7 0.0 0.0 0.0 16.7 0.0 10.0 16.7
IP
II
II der
IIP
III
IIIP
0.0 100.0 0.0 0.0 87.5 0.0 0.0 83.3 0.0 0.0 70.0 0.0 0.0 66.7 0.0 0.0 62.5 0.0 0.0 61.5 7.7 0.0 50.0 0.0 7.7 46.2 0.0 0.0 42.9 0.0 0.0 28.6 0.0 0.0 19.0 0.0 8.3 16.7 8.3 0.0 11.1 0.0 0.0 0.0 0.0 16.7 0.0 16.7
0.0 0.0 16.7 15.0 0.0 0.0 0.0 0.0 0.0 0.0 28.6 0.0 16.7 0.0 0.0 0.0
0.0 12.5 0.0 10.0 16.7 37.5 7.7 16.7 23.1 42.9 35.7 23.8 16.7 33.3 60.0 33.3
0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 7.1 0.0 0.0 0.0 0.0 0.0
IIIP V-ing oth. der 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 7.7 0.0 0.0 0.0 0.0 0.0 0.0 0.0
0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 42.9 0.0 55.6 0.0 0.0
0.0 0.0 0.0 5.0 16.7 0.0 0.0 16.7 7.7 14.3 0.0 14.3 16.7 0.0 30.0 16.7
N 5 8 6 20 6 8 13 6 13 7 14 21 12 9 10 6
an expression that may be typical for Indian English as it does not occur in any of the British data and it is found on many Indian websites on the world-wide web. Of the ten occurrences of this phrase, eight display semantics similar to Example€(63): (63) The bird flu (avian flu) is sending alarm bells ringing across Southeast Asian countries […]
In addition to this usage, there are two instances in which the complement clause is not made explicit, as demonstrated in Example (64): (64) […] that boys and girls are indulging in sexual activities frequently is sending alarm bells in the minds of the guardians of young siblings […]
While the pattern SEND Od Ving is less frequent in Indian English than in British English, the use of SEND alarm bells ringing shows that this pattern is productive in Indian English. Furthermore, the implicit complement clause in Examples such as (65) can be seen as a specific variability of this pattern in Indian English. As shown in Table 6-10, there are relatively few Od-collocates of SEND that are only attested in the BNC. Therefore, all 16 Od-collocates will be subsumed in Table 6-15.
126 Structural Nativization in Indian English Lexicogrammar
Table 6-15 shows that the majority of the instances of SEND + Od-collocate in the BNC newspaper data are complemented in the type-II pattern, while relatively few Od-collocates are used with preference in the type-III pattern. Unlike the Times of India data, however, there are more collocates that are used in sentences complemented in the type-I pattern. This reflects observations so far that British speakers use this pattern more frequently for transfer verbs than their Indian counterparts do. However, owing to the lower number of collocate tokens in the BNCNews dataset compared to the ToI dataset, only a few instances of this pattern and of other non-central patterns of SEND are attested. This scarcity of data also has an influence on the interpretation of the collocates that are predominantly found in the type-II and type-III pattern. Name and answer, for example, co-�occur only five and six times respectively with SEND (at least as a patient in the Od-position), so that the high percentages still only represent a relatively low number of instances. Regarding envoy and card, the picture is somewhat clearer, as these Od-collocates are much more frequent. With respect to the type-III pattern, only two Od-collocates are clearly used predominantly in this pattern, namely cash and picture. For someone and donations no clear tendency can be derived from Table 6-15. While someone is used with equal frequency in both the type-II and the type-III patterns, donations is used slightly more frequently in the type-III pattern than in the type-II pattern, but there are also a number of instances where it is used in the type-IIP pattern. As shown in Table 6-6 (Section 6.7), the pattern SEND Od Ving has a strong influence on the significance of the differences in the verb-complementational profile of SEND in Indian and British newspaper English. By looking at the verbcomplementational distribution of sentences that contain significant Od-collocates, we will see that this pattern is also used more frequently in the BNCNews dataset compared with the ToI9m dataset. Although there are a number of instances of this pattern observed in the collocational profile of the ToI9m corpus, it is still used more frequently in the BNCNews. As pointed out in Section 6.7, there is a tendency to use this pattern in British newspaper English in the context of business and financial news. Table 6-15 illustrates that the Od-collocates mainly occurring in this pattern are shares and prices; (65) exemplifies a typical use of this pattern with the collocate prices: (65) WHEN the Eighties boom sent property prices skyrocketing, London set the pace.
There is an area of overlap between the varieties in terms of the use of this pattern in expressions such as SEND someone packing. However, the divergence between the varieties in terms of overall frequency and the functional differentiation between expressions comparable to Example (65) in British English and expressions
Chapter 6.╇ SEND 127
Table 6-16.╇ Verb-complementational profile of SEND + Od-collocates (shared) Type I II IIder IIP IIPder III IIIP IIIPder others sum total
ToI9m observed
ToI9m expected
χ²
BNCNews observed
BNCNews expected
χ²
11 203 5 23 19 95 16 10 9 391
20.47 190.11 5.92 24.24 18.31 94.79 14.54 9.69 12.93 391.00
4.38 0.87 0.14 0.06 0.03 0.00 0.15 0.01 1.19 6.83
27 150 6 22 15 81 11 8 15 335
17.53 162.89 5.08 20.76 15.69 81.21 12.46 8.31 11.07 335.00
5.11 1.02 0.17 0.07 0.03 0.00 0.17 0.01 1.39 7.97
comparable to Example (63) in Indian English, corroborate the hypothesis that this pattern is used quite differently in Indian English and in British English. Table 6-10 showed that there are 12 significant Od-collocates that occur in the datasets of both varieties. By looking at the total verb-complementational distribution for occurrences of SEND in combination with these shared Od-collocates in Table 6-16, we can see that there are no significant differences in verb-complementation in the case of the shared Od-collocates. With a total chi-square value of 14.8, the distribution shown in Table 6-16 is not significant, the critical threshold for significance at the p ≤ 0.05 level being 15.5. This result strengthens the hypothesis that differences in verb-complementation between Indian and British English correlate with different collocational preferences, which is in accordance with the results for GIVE. However, it should be pointed out that even in the case of shared Od-collocates there is a tendency towards more frequent use of the type-I pattern in British English, which contributes greatly to the total chi-square value (9.49 / 14.8 of the total χ² are based on the type-I pattern). Therefore, even in the case of this non-significant distribution, an overall trend for a more frequent use of the ditransitive pattern in British English is observable. Especially in the case of three Od-collocates, namely letters, letter and copy, this tendency can be clearly identified: In the case of letter, 12 out of 74 co-occurrences with SEND are used in the type-I pattern in the BNC data, compared to four out of 84 in the Times of India9m data. In the case of letters the type-I pattern in the BNCNews is used four times out of 51 total co-occurrences with SEND, compared to two instances out of 42 in the Times of India. Finally, in the case of copy, the type-I pattern is used eight times out of 25 in the BNC compared to one occurrence out of 13 in the Times of India.
128 Structural Nativization in Indian English Lexicogrammar
6.9
SEND in Indian and British newspaper English: A brief summary
The analysis of the different datasets in Sections 6.7 and 6.8 has shown that the differences in verb-complementation between Indian and British English are less prominent when one looks at a single specific text type. The analysis of the verbcomplementational profiles of SEND without relation to Od-collocates showed that the main differences between the varieties were attested in the peripheral pattern SEND Od Ving, which is more productive in the BNCNews data, as well as in the more frequent use of particle verb constructions. Although these constructions were initially not included in the statistic test for significant complementational distribution (cf. Table 6-5), the preference of British speakers for these constructions, especially in cases where the particle does not add much to the meaning of the sentence compared with SEND proper, nicely complements the findings of the variation of SEND between the varieties which were attested more at the periphery than in the case of central complementation patterns. The analysis of verb-complementational patterns and their correlation to significant Od-collocates of SEND in Section 6.8, based on a larger dataset, shows that a more focused examination provides further insights into variety-specific variation. Firstly, speakers of Indian English use SEND with a higher number of collocates in terms of both Od-collocate types and tokens. While there is an area of overlap of significant Od-collocates, there are also a number of collocates that are only attested in one variety and not in the other. Again, more of these collocates could be attested in the ToI9m data than in the BNCNews data. By correlating the Od-collocates with patterns of verb-complementation, Section 6.8 demonstrated how variation between Indian and British English is more pervasive when lexical and grammatical elements are examined jointly. While in both the Indian and the British data the type-II pattern is the most frequent complementation pattern in the majority of cases, there is a slightly stronger tendency to use the type-III pattern in the ToI data. At the same time the basic ditransitive pattern, type-I, is used more frequently in the BNCNews data. Regarding the Od-collocates that are found in both the BNCNews and the ToI9m, no significant difference in the overall verb-complementational distribution was attested, which is in line with the results for GIVE in Section 5.2. Although in the case of these shared Odcollocates a slight tendency towards a more frequent use of the type-I pattern was also attested, differences in verb complementation between the varieties are clearly more prevalent in the cases of variety-specific Od-collocates, which further strengthens the hypothesis that there is a clear interrelation between lexical and grammatical nativization.
chapter 7
OFFER
7.1
Introduction: OFFER as a low-frequency ditransitive verb
The third verb analyzed in the present study, OFFER, is different from the prototypical ditransitive verb GIVE and the non-prototypical verb SEND for quantitative as well as qualitative reasons. Mukherjee (2005) groups this verb (together with SEND and some other verbs) into the category of habitual ditransitive verbs, which are defined “as those verbs that occur frequently in ICE-GB (i.e. 200 to 700 occurrences), but that are not strongly associated with an explicit ditransitive syntax […] in the clear majority of all cases” (Mukherjee 2005:â•›147). With 196 occurrences in ICE-GB and 131 occurrences in ICE-India, OFFER only barely meets the frequency criterion provided by Mukherjee (2005) in ICE-GB and does not meet it in ICE-India. Thus, this verb is defined in the present work as a lowfrequency habitual ditransitive verb, since it is on the border between habitual ditransitive verbs and peripheral ditransitive verbs.61 Besides this quantitative categorization of OFFER, it is also possible to draw some qualitative distinctions between the use of OFFER and other ditransitive verbs. In the framework of Goldberg (1992) – a discussion of the polysemous senses of the ditransitive construction – OFFER would be categorized within the group of verbs whose satisfaction conditions imply that a Subject causes Object1 to receive Object2. In this regard, no actual transfer is necessarily encoded in sentences in which OFFER is used. Firstly, offers can be declined and secondly some offers violate the satisfaction conditions in not actually being offers at all, when the offerer has no intention to actually fulfil his or her offer. Apart from these quantitative and qualitative differences of OFFER compared to GIVE and SEND, there are some differences in the admissible patterns for OFFER. Here two specific patterns will be taken into account, which are either not associated or occur extremely rarely with GIVE and SEND. The patterns in question were labelled type-IIId and type-IIIh/t pattern by Mukherjee (2005) and
61. Peripheral ditransitive verbs are defined by Mukherjee 2005 as verbs that occur less than 200 times or occur in less than 5% of their uses in explicit ditransitive syntax (cf. Mukherjee 2005:â•›84).
130 Structural Nativization in Indian English Lexicogrammar
will be subsumed under the type-III derivative patterns type-IIIder here. These two patterns are described in a formulaic way in (66) and (97): (66) type-IIId pattern: (S) OFFER [Od:to-infinitive] Oi (67) type-IIIh/t pattern: (S) HAVE [Od:NP] to offer Oi
The type-IIId pattern shown in Example (66) is a specific use of OFFER that is not, or is only rarely, encountered in combination with GIVE or SEND; in this pattern the direct object does not profile a physical or abstract entity but a future action as shown in Example (68): (68) ALI NORRISH, the Oxford University Boat Club president, offered to resign last night after failing her first-year engineering examinations for the second time.
The type-IIIh/t pattern is another specific use for OFFER, as the pattern in itself contributes to the meaning of a sentence by being frequently used in cases where the direct object or transferred entity is non-existent, or at least very small, so that the choice of this pattern has a certain semantic prosody with regards to the direct object. Consider Examples (69) and (70): (69) The Opposition has nothing to offer. (70) TO MANY she’s still the chirpy Cockney who made it on her looks and has little else to offer.
Due to the low frequency of OFFER in the ICE corpora, the procedure applied in the present section differs slightly from the procedures used in the sections on GIVE and SEND. In the ICE corpora, OFFER is not used frequently enough to conduct a useful analysis of its use according to text type. Therefore, the discussion of the ICE data will be limited to differences between the written and the spoken mode, without any further delving into the different text types included in the two mode-based categories. Similarly to the procedure applied for SEND, this data will be complemented with an analysis of a dataset of British and Indian newspaper English derived from the BNC newspaper section (BNCnews) and the randomized extraction of the Times of India dicussed earlier (ToI9m). In addition, a collocational analysis will be provided for OFFER in order to account for correlations of pattern selection and collocational preferences in the varieties. This collocational analysis draws upon the same dataset that was used for SEND, i.e. the total number of newspaper texts included in the BNC and a comparable subset of the Times of India.
7.2
Chapter 7.╇ OFFER 131
Verb-complementational patterns of OFFER in spoken and written Indian and British English
As discussed above (Section 7.1), OFFER is far less frequent than both GIVE and SEND in Indian and British English. For this reason, no attempt is made to specify different uses according to different text types, and only differentiation between written and spoken texts will be the focus at this point. Table 7-1 provides an overview of the verb OFFER used in the different complementation patterns in ICE-GB and ICE-India. In sum, there is a higher frequency of OFFER in the British component of ICE€– 196 occurrences compared to 131 in ICE-India. In both varieties, the typeIII pattern is the default pattern for OFFER in both the spoken and the written mode. In spoken Indian-English, the type-III pattern accounts for about 34% of all uses; in written Indian English this pattern is used in 46% of all instances. In spoken British English it accounts for about 37% of the uses, while in written British English it accounts for 47%. Thus, variation between the varieties regarding the use of the type-III pattern is limited in the ICE corpora. Patterns that are structurally related to the type-III pattern – that is type-IIIder, type-IIIP and type-IIIPder€– are found relatively infrequently. However, especially in the case of the type-IIIder and the type-IIIPder pattern, there is a tendency to use these patterns more frequently in ICE-India than in ICE-GB. The frequent use of the type-IIIPder pattern in written Indian English is particularly noteworthy, as this pattern mainly represents participle constructions that, therefore, seem to be more popular with writers of Indian English. Table 7-1.╇ Verb-complementational distribution of OFFER in ICE-India and ICE-GB ICE-India Cat.
spoken written
Type I
I der
I P
IP der
II
II der
II P
IIP der
III
III der
5 14
1 0
3 2
0 1
7 4
1 0
1 0
0 0
18 36
10 2
III IIIP IV P der
V
sum
2 0
2 0
53 78
III IIIP IV P der
V
sum
0 0
81 115
3 6
0 13
ICE-GB Cat.
spoken written
Type I
I der
I P
IP der
II
II der
II P
IIP der
III
III der
18 15
1 1
6 10
0 0
9 7
3 0
0 1
0 0
30 54
8 14
4 8
1 5
1 0
132 Structural Nativization in Indian English Lexicogrammar
A higher amount of variability is attested for the use of the type-I pattern. In terms of this pattern, there is a clear distinction between its use in spoken and written Indian English as well as between the varieties. While in the spoken part of ICE-India this pattern accounts for only 9.4% of all instances, it is used in 18% of all cases in the written part. In ICE-GB we find a different relation regarding the use of this pattern in the different modes. In the spoken part of ICE-GB it is used in approximately 22% of all occurrences, whereas in the written part it is used only in about 13%. Thus, in terms of the type-I pattern, variety-specific variation is attested in the more frequent relative use of this pattern in both spoken and written British English compared to Indian English, while at the same time there is a preference for this pattern in the spoken mode in ICE-GB and a preference for this pattern in the written mode of ICE-India. The structurally related patterns of the type-I pattern are also scarcely represented, although it should be pointed out that the higher frequency of use of the type-I pattern in British English overall is also reflected in a more frequent use of these patterns, especially the type-IP pattern in ICE-GB. The use of the type-II pattern is rare in both varieties, so in this case the database is not sound enough to draw meaningful conclusions about possible varietyspecific variation. The same also holds true for the patterns related to the type-II pattern, as well as the marginal patterns type-IV and type-V.
7.3
OFFER in British and Indian newspaper English
As the overall frequency of OFFER is relatively low in both ICE corpora and particularly so in ICE-India, the description of the verb-complementation of OFFER will be accompanied in the following by an examination of a larger dataset from the Times of India corpus and the newspaper section of the BNC. The underlying procedure for this analysis is identical to the one that is used in Chapter 6 on SEND. A query for OFFER in the datasets described in Section 6.7 yields a total number of 434 instances in the Times of India1.75 dataset and 619 instances in the comparable BNCNews.1.77 data. Thus, analogous to the findings in the previous section that were based on ICE-India and ICE-GB in the two newspaper corpora, OFFER is also used significantly more frequently in British English than in Indian English (G2 = 29.98; p ≤ 0.0001). Apart from this overall higher frequency of OFFER in the BNC dataset, however, there are only slight differences in the distribution of complementation patterns, so again genre conventions may compensate variety-specific variation. Consider Table 7-2. Table 7-2 shows that there are no significant differences in the distribution of verb-complementational patterns of OFFER in the Indian and British news
Chapter 7.╇ OFFER 133
Table 7-2.╇ Distribution of OFFER in ToI1.75 and BNCNews.1.77 – focus on central patterns Type I IP II IIP III IIIder IIIP IIIPder others sum total
ToI1.75 observed 49 18 53 7 200 52 19 21 15 434
ToI1.75 expected 47.40 24.73 42.45 7.01 212.67 48.22 15.66 22.67 13.19 434.00
χ²
BNCNews.1.77 observed
BNCNews.1.77 expected
χ²
0.05 1.83 2.62 0.00 0.76 0.30 0.71 0.12 0.25 6.64
66 42 50 10 316 65 19 34 17 619
67.60 35.27 60.55 9.99 303.33 68.78 22.34 32.33 18.81 619.00
0.04 1.28 1.84 0.00 0.53 0.21 0.50 0.09 0.17 4.66
corpora (the total χ² in table = 11.3 at 8df for p ≤ 0.05 a value of 15.51 has to be met). In both corpora the type-III pattern is the default pattern for OFFER, while there is some minor variation between the type-IP pattern that is used slightly more frequently in British English and the type-II pattern that is used slightly more frequently in Indian English. A typical example of the use of the type-IP pattern in British Newspaper English is shown in (71); a typical example of the type-II pattern in Indian newspaper English in (72): (71) Turner has been offered employment by the company […] of Newport’s new coach, Gareth Evans at a better salary than he currently enjoys (72) We would like to offer our sincere apologies to the gentleman
Although these differences in the overall distribution between the varieties are minimal, there is some variation on the level of collocation and verb-complementation in the case of the use of OFFER in combination with Od-collocates, which will be described in the following section.
7.4
Collocational profiles of OFFER in Indian and British newspaper English
In order to create collocational profiles for OFFER in both Indian and British English, the datasets described in 6.8 were used, with the application of the same methodology as for the creation of these profiles for SEND. In the case of OFFER, however, a further manual process was included in order to eliminate the homographic noun forms offer, offers, offering. The first step in this process was to take the initial list of collocation candidates that was compiled automatically
134 Structural Nativization in Indian English Lexicogrammar
Table 7-3.╇ Od-collocates of OFFER in the Times of India (9 million word subcorpus) Od-collocate Times of India9m
log-likelihood
Od-collocate Times of India9m
log-likelihood
assistance berths bribe concessions courses discounts equity explanation facilities facility flights help incentives job loan money namaz opportunities opportunity packages potential
â•⁄ 82.81 â•⁄ 44.65 â•⁄ 80.48 107.85 439.31 133.98 â•⁄ 42.59 â•⁄ 73.43 191.99 134.72 â•⁄ 26.61 103.93 132.26 â•⁄ 36.40 â•⁄ 24.88 â•⁄ 63.20 197.50 â•⁄ 76.93 236.48 â•⁄ 95.46 â•⁄ 22.53
prayers price quality rates resistance role RS schemes scholarships service services shares solutions something support terms ticket tickets tiles treatment value
620.45 â•⁄ 36.16 â•⁄ 29.05 â•⁄ 65.40 â•⁄ 94.45 217.73 â•⁄ 52.64 â•⁄ 57.28 â•⁄ 83.21 110.88 514.71 â•⁄ 42.85 â•⁄ 95.00 â•⁄ 32.12 â•⁄ 96.34 â•⁄ 18.70 â•⁄ 31.64 â•⁄ 44.74 â•⁄ 55.61 â•⁄ 37.35 115.89
by WST€4.x and lemmatize it for the node word; items that cannot function as a direct object were excluded. The next step was to exclude homographic noun forms so that finally only the remaining instances of OFFER (V) + Od-collocation candidate were used to calculate collocational statistics. By applying this procedure, a list of 42 collocations was generated for both the Times of India9m and the BNCNews data. Table 7-3 shows the list of all the Od-collocations in the Times of India9m data. The same procedure is applied to the BNCNews data, and the resulting Odcollocation candidates of OFFER in British Newspaper English are shown in Table€7-4. These collocational profiles of OFFER have an area of overlap, i.e. some Odcollocates are shared by Indian and British newspaper English, while at the same time there are also a number of collocates that are found in one variety but not in the other, as shown in Table 7-5.
Chapter 7.╇ OFFER 135
Table 7-4.╇ Od-collocates of OFFER in the BNCNews Od-collocate BNCNews
log-likelihood
Od-collocate BNCNews
log-likelihood
advice alternative assistance cash chance choice compensation contract counselling deals discounts drugs evidence excuses explanation facilities guarantee help hope information job
396.98 164.52 â•⁄ 58.86 â•⁄ 15.20 536.18 145.94 â•⁄ 56.30 228.98 â•⁄ 69.00 104.38 â•⁄ 93.08 â•⁄ 30.94 â•⁄ 75.80 â•⁄ 75.53 106.57 105.19 â•⁄ 81.75 235.81 138.03 â•⁄ 63.54 154.95
lift money opportunities opportunity post prize protection rate rates relief resistance reward service services shares something sum support terms use value
â•⁄ 24.74 â•⁄ 74.65 103.18 279.81 â•⁄ 19.42 â•⁄ 67.09 111.89 â•⁄ 21.60 â•⁄ 53.94 â•⁄ 51.76 â•⁄ 64.68 399.81 237.43 264.81 â•⁄ 67.70 â•⁄ 33.47 â•⁄ 30.04 151.04 â•⁄ 68.50 â•⁄ 26.32 â•⁄ 40.74
As has been shown in the previous sections of this chapter, the occurrences of the verb in combination with its Od-collocates are analyzed for the verbÂ�complementation patterns that are typical for the respective verb + Od-collocate combination. Table 7-6 displays the distributions of all OFFER + Od-collocate combinations for both the Times of India9m and the BNC datasets analyzed for variation between the varieties. The distribution in Table 7-6 shows several interesting points. Firstly, OFFER in combination with frequently recurring Od-collocates is more frequent in British newspaper English than in Indian newspaper English. Although the number of Od-collocate types is the same for both varieties, the number of collocate tokens is significantly higher in British English (G2 = 16.23) with a mean token-per-type ratio of 19.1 in the British data compared to 15.5 in the Indian data. Secondly, in contradistinction to the non-collocation based distributions of the complementation patterns of OFFER discussed in the previous section, significant differences in the distribution of these patterns can be attested for the complementation of OFFER in combination with Od-collocates. The partial χ²-values in Table 7-6
136 Structural Nativization in Indian English Lexicogrammar
Table 7-5.╇ Od-collocates of OFFER sorted by variety Od-collocates Times of India9m only
Od-collocates Times of India9m and BNCNews
Od-collocates BNCNews only
berths bribe concessions courses equity facility flights incentives loan namaz packages potential prayers price quality rates role RS schemes scholarships solutions ticket tickets tiles treatment
assistance discounts explanation facilities help job money opportunities opportunity resistance service services shares something support terms value
advice alternative cash chance choice compensation contract counselling deals drugs evidence excuses guarantee hope information lift post prize protection rate rates relief reward sum use
Table 7-6.╇ Verb-complementational profile of OFFER + Od-collocates Type I IP II IIder IIP III IIIder IIIP IIIPder others sum total
ToI9m observed
ToI9m expected
χ²
BNCNews observed
BNCNews expected
χ²
71 23 79 9 9 352 18 40 34 15 650
86.40 45.66 75.65 7.16 9.40 331.27 20.14 28.68 32.23 13.43 649.99
2.74 11.25 0.15 0.47 0.02 1.30 0.23 4.50 0.10 0.18 20.94
122 79 90 7 12 388 27 24 38 15 802
106.60 56.34 93.35 8.84 11.60 408.73 24.86 35.35 39.77 16.57 802.01
2.22 9.11 0.12 0.38 0.01 1.05 0.19 3.64 0.08 0.15 16.95
Chapter 7.╇ OFFER 137
Table 7-7.╇ OFFER in correlation to Od-collocates (type-III dominant) Type-III dominant (%) flights loan tiles prayers namaz rates treatment concessions facility courses berths potential schemes solutions RS quality packages
I
II
II der
IIP der
III
III der
IIIP
IIIP der
oth.
N
0.0 0.0 0.0 0.0 0.0 0.0 0.0 9.1 5.6 0.0 40.0 0.0 0.0 8.3 19.2 14.3 0.0
0.0 0.0 0.0 4.2 0.0 0.0 0.0 9.1 16.7 2.1 0.0 40.0 0.0 33.3 15.4 0.0 10.0
0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 14.3 0.0
0.0 0.0 0.0 0.0 0.0 8.3 0.0 9.1 5.6 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
100.0 100.0 100.0 93.8 85.7 75.0 75.0 72.7 66.7 66.0 60.0 60.0 55.6 50.0 46.2 42.7 40.0
0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 2.1 0.0 0.0 11.1 0.0 0.0 0.0 0.0
0.0 0.0 0.0 2.1 7.1 0.0 12.5 0.0 0.0 17.0 0.0 0.0 22.2 8.3 0.0 0.0 20.0
0.0 0.0 0.0 0.0 7.1 16.7 12.5 0.0 5.6 12.8 0.0 0.0 11.1 0.0 3.9 28.6 30.0
0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 15.4 0.0 0.0
5 5 6 48 14 12 8 11 18 47 5 5 9 12 26 7 10
show that variation is strongest in the type-IP pattern and the type-IIIP pattern, followed by the type-I pattern and the type-III pattern (all shown in boldface). These observations are in accordance with the results of the previous sections inasmuch as type-I pattern verb-complementation is used more frequently in British English and type-III complementation is used more frequently in Indian English, while there is much less variation in the use of the type-II pattern. In order to find which of the Od-collocates correlate with which pattern, Table 7-6 can be analyzed in its component parts, namely the verb-complementation patterns of all the sentences in which OFFER is used in combination with an Od-collocate. Similar to the procedure in Section 6.8, this analysis is carried out in three steps. Firstly, all the sentences containing OFFER in combination with Od-collocates that only occur in the Indian dataset are analyzed for verb-complementation, then the same is carried out for the group of Od-collocates that only occur in the British data. Finally, this will also be undertaken for the shared collocates. Table 7-7 shows the complementation patterns of all the sentences containing OFFER and a significant Od-collocate that occur in the Times of India9m, but not in the BNCNews data, which are predominantly complemented in the type-III pattern. The top three collocates flight, loan and tiles are all relatively infrequent; hence, a certain amount of idiosyncrasy may play a role here. Nevertheless, in the case of
138 Structural Nativization in Indian English Lexicogrammar
tiles in particular, it can be shown relatively easily why a potential recipient of the offering process does not need to be made explicit; consider Example (73): (73) Kajaria Ceramics offers wall tiles in sizes like 20 cm × 20 cm and 30 cm × 30€cm.
The next two collocates, prayers and namaz clearly show the cultural influence of the Indian background, as religion plays a more central role in the panethnic and multi-religious setting of the subcontinent.62 As both prayers in general and namaz in particular play a central social role, the fact that these items collocate more frequently with OFFER in Indian English than in British English (with its group of more secularized speakers) is not surprising; it is also not surprising that the recipient, namely the deity to whom prayers are offered, rarely needs to be explicitly mentioned. In the case of namaz there is only one possible deity, and in the case of other religious ceremonies the deity in question is often accessible from discourse. Rates and treatment are again infrequent collocates, but comparable to the first three, recipients are not profiled, as these are apparent from the context. This is somewhat different in the cases of concessions, facility and courses. TypeIII complementation is also the most frequent with these collocates. However, in these cases recipients are sometimes profiled as nominal and prepositional objects (type-I and type-II pattern) in the case of concessions and facility and as a prepositional object (type-II pattern) in the case of courses. The collocates berths, potential and schemes are again low-frequency collocates of OFFER, which, however, unlike the ones discussed earlier, show a certain amount of variation in terms of pattern selection. Berths and potential are used in the type-I and the type-II pattern respectively, while schemes (mainly in the financial sense) shows some variation within the different type-III derivative patterns. Of the remaining four, RS is the most interesting, insofar as it is frequent and displays a certain variability. Although the type-III pattern is still the most frequent pattern, in this case a larger number of sentences with explicit recipients are attested, which makes sense considering that in monetary transfer processes there is often the need to profile the recipient. Note here that the 15.38% of cases subsumed under others are instances of the type-IP pattern, which was not used for any of the other type-III dominant collocates. Consider Example (74) as a typical example of the type-IP pattern of OFFER in collocation with RS: (74) Teachers from India will be offered the equivalent of Rs 1 lakh per month […] 62. Namaz is the Persian term for the ritual prayers in the direction of Mecca in Islam.
Chapter 7.╇ OFFER 139
Table 7-8.╇ OFFER in correlation to Od-collocates (type-II dominant) Type-II dominant (%) bribe tickets scholarship
I
II
IIder
IIPder
III
IIIder
IIIP
N
11.1 0.0 12.5
66.7 57.1 50.0
0.0 0.0 0.0
0.0 0.0 0.0
22.2 28.6 25.0
0.0 0.0 0.0
0.0 14.3 12.5
9 7 8
Table 7-9.╇ OFFER in correlation to Od-collocates (type-I dominant) Type-I dominant (%) role
I
Ider
IP
IIP
IIIPder
N
62.9
2.9
17.1
14.3
2.9
35
As the type-III pattern is clearly the default pattern for OFFER in Indian newspaper English, there are only relatively few collocates that are dominantly used in one of the other basic patterns. Table 7-8 shows the Od-collocates that are most frequently used in the type-II pattern. The three Od-collocates that are used primarily in the type-II pattern in the Times of India are bribe, tickets and scholarship. Although these three collocates are all relatively infrequent, it can be seen that they all denote patients of the transfer process that are concrete rather than abstract or informational (cf. Bresnan & Hay 2008). This concreteness may be one of the reasons why the recipient is profiled more frequently in sentences containing these collocates than for many of the collocates shown in Table 7-7. Only one of the Od-collocates that occur frequently with OFFER in the Times of India9m but not in the BNC news section most frequently co-occurs with the type-I pattern. The complementational distribution of this collocate, namely role is given in Table 7-9. In Table 7-9 it can be seen that in the majority of cases OFFER + role is complemented in the type-I pattern or one of its derivatives. This collocate is relatively frequent in the Times of India owing to the importance of the Bollywood movie industry, which leads to a relatively large number of interviews with actors in the paper as shown in Example (75): (75) a. I was having coffee at Holiday Inn when Pallavi Chatterjee spotted me and offered me this role. b. Vidya was, reportedly, very close to him and had also been offered a role in an upcoming movie through him […].
Examples (75a) and (75b) show typical uses of OFFER + role in the type-I and the type-IP pattern, which together account for 80% of the 35 instances of the OFFER€+ role co-occurrences.
140 Structural Nativization in Indian English Lexicogrammar
Table 7-10.╇ OFFER in correlation to Od-collocates (no pattern bias) No specific tendency (%) equity incentives price ticket
I
IP
II
II der
IIP
III
III der
IIIP der
N
0.0 7.7 33.3 16.7
0.0 7.7 0.0 50.0
42.9 30.8 0.0 0.0
0.0 0.0 22.2 0.0
14.3 7.7 0.0 16.7
42.9 30.8 22.2 16.7
0.0 0.0 11.1 0.0
0.0 15.4 11.1 0.0
7 13 9 6
As was the case with GIVE and SEND in the previous sections, there are a number of Od-collocates that show no specific tendency for co-occurrence with a single pattern in the Times of India. These are shown in Table 7-10. Table 7-10 contains the four Od-collocates that do not show a tendency to occur in a specific complementation pattern, but display variability across many different patterns. Equity is mainly found in phrases complemented in the type-II and type-III pattern with an additional instance of the type-IIP pattern. Incentives is also mainly found in the type-II and the type-III pattern, with two additional instances attested for the type-I and IP pattern, respectively. Price is mainly found in the type-I pattern and the type-III pattern with its derivatives, while ticket is mainly found in the type-IP pattern. Applying the same procedure to the BNC news-specific collocates, i.e. all the Od-collocates that are significant in the BNC newspaper section but not in the equivalent Times of India data, we find that there are differences in the preferred complementation patterns containing the collocates that are preferred in a specific variety. Table 7-11 shows the 16 Od-collocates of OFFER that are attested most frequently in the type-III pattern. Compared with the ToI data, we find one collocate type less. However, the number of collocate tokens is slightly higher than in the corresponding Times of India Table (274 compared to 248). The top three Odcollocates in Table 7-11 are guarantee, evidence and reward. Guarantee is mainly used in a general sense, where the guarantee is offered by a firm or institution on a product or service, as shown in Example (76): (76) Now one firm offers a free guarantee that you won’t lose out.
Since such offers are valid to any customer of the product or service, there is no need to profile the recipient. The second collocate, evidence, as already discussed in relation to GIVE, also rarely needs a profiled recipient, since the settings in which evidence is offered are rather limited and the recipient is accessible from the context. Reward, in the same way as guarantee, is mainly used in general offers to anyone who performs a specific action as exemplified in (77):
Chapter 7.╇ OFFER 141
Table 7-11.╇ OFFER in correlation to Od-collocates (type-III dominant) Type-III dominant (%) guarantee evidence reward sum excuses information rate choice advice relief cash deals alternative protection hope prize
I
IP
II
III
III der
IIIP
IIIP der
oth.
N
0.0 0.0 2.6 20.0 0.0 14.3 0.0 18.2 2.0 0.0 0.0 0.0 13.6 18.8 11.5 0.0
9.1 0.0 0.0 0.0 0.0 0.0 0.0 9.1 0.0 0.0 0.0 14.3 13.6 0.0 0.0 0.0
0.0 0.0 0.0 0.0 0.0 7.1 0.0 0.0 20.4 33.3 16.7 7.1 4.6 18.8 34.6 8.3
90.9 86.7 81.6 80.0 75.0 71.4 71.4 68.2 67.4 66.7 66.7 64.3 63.6 56.3 53.9 50.0
0.0 0.0 0.0 0.0 0.0 0.0 14.3 0.0 4.1 0.0 16.7 0.0 0.0 0.0 0.0 0.0
0.0 6.7 10.5 0.0 25.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 16.7
0.0 6.7 2.6 0.0 0.0 0.0 14.3 0.0 4.1 0.0 0.0 7.1 0.0 6.3 0.0 25.0
0.0 0.0 2.6 0.0 0.0 7.1 0.0 4.6 2.0 0.0 0.0 7.1 4.6 0.0 0.0 0.0
11 15 38 5 8 14 7 22 49 9 6 14 22 16 26 12
(77) Now store owner Paul Harper has offered a £500 reward for information leading to a conviction.
The next two collocates, sum and excuses, are both relatively rare, so a detailed analysis is not possible. It is noteworthy, however, that excuses collocates with negative items such as no or idle, so the pattern itself has a negative semantic prosody. Information is used less frequently without an explicit recipient. If we examine four specific instances of the 14 cases in which information is profiled, we find the type-I pattern in two instances, type-IPder in the third and type-II in the fourth. Rate is used in the same way as guarantee and reward as a standing offer by a (mainly) financial institution to potential customers. Choice also functions in a general way, comparable to that just discussed; however, general recipients are relatively frequently profiled, mainly in the type-I pattern, as exemplified in (78): (78) a. Holiday company EuroSites offer a choice of tents or mobile homes […] b. The Questor Selection offers readers a choice of shares […]
As shown in (78b), the recipients in the type-I patterns for OFFER + choice are generally not individual people, but rather groups of people. As shown throughout this chapter, there is a tendency for British English to use the type-I pattern more frequently than Indian English. This phenomenon may mean that recipients
142 Structural Nativization in Indian English Lexicogrammar
are profiled in British English even when their inclusion is not necessary for semantic reasons. Advice, relief and cash – besides the default type-III complementation patterns – are all used in the type-II pattern in a number of instances. As relief and cash are both relatively infrequent, at this point only the more frequent advice will be discussed. The two most frequent patterns for advice are the type-III pattern, frequently realized as sb + OFFER + advice on sth, and the type-II pattern in which the theme of advice is usually not profiled.63 Consider Example (79): (79) a. Decide on Darlington is a new guide for tourists […] and offers advice on a range of places to stay. b. […] the Inland Revenue’s Belfast offices are offering free advice to those with tax problems.
Alternative, protection and hope, although still type-III dominant, have a tendency to be used in a large spectrum of complementation patterns. In the case of alternative, if the recipient is profiled, this is mainly done by the use of the type-I or its corresponding passive pattern. In the case of protection, both the type-I and the type-II pattern are used with equal frequency, and in the case of hope the type-II pattern is preferred over the type-I pattern for the profiling of recipients. Prize, finally, is mainly complemented in type-III and its derivative patterns, most notably type-IIIPder, which mainly encodes participle constructions. In contrast to the Times of India data, no Od-collocates that clearly favour type-II complementation are attested in the BNCNews data. Although the type-II pattern is quite frequent for some of the sentences that contain significant Od-collocates, for example in the case of hope, relief and drugs, all of these collocates are used with preference in different patterns. The same holds true, albeit to a lesser extent, for the type-I pattern. Again, no collocate for which the use of the type-I pattern would constitute a majority usage is attested. If, however, we include the type-I passive and derivative patterns, there are five Od-collocates that can be found in one of these patterns in the majority of all instances, as shown in Table 7-12. In Table 7-12 the cases of contract and chance in particular are important, since these two collocates are used very frequently with OFFER. Therefore, the tendency to complement these sentences in the type-I or type-IP pattern has a strong influence on the differences in the overall pattern distribution between the varieties. Contract in this table can be compared to role in Table 7-11; both are of approximately equal frequency and both denote a concrete offer for a certain
63. One instance was attested of the type-II pattern in which the theme of advice is profiled.
Chapter 7.╇ OFFER 143
Table 7-12.╇ OFFER in correlation to Od-collocates (type-I dominant) Type-I dominant (%) contract chance post lift drugs
I
IP
II
III
III der
IIIP
IIIP der
oth.
N
45.5 45.3 28.6 20.0 14.3
30.3 22.7 71.4 60.0 57.1
â•⁄ 9.1 â•⁄ 1.3 â•⁄ 0.0 â•⁄ 0.0 28.6
â•⁄ 6.1 21.3 â•⁄ 0.0 20.0 â•⁄ 0.0
3.0 1.3 0.0 0.0 0.0
0.0 0.0 0.0 0.0 0.0
0.0 2.7 0.0 0.0 0.0
6.1 5.3 0.0 0.0 0.0
33 75 â•⁄ 7 â•⁄ 5 â•⁄ 7
Table 7-13.╇ OFFER in correlation to Od-collocates (no pattern bias) No specific tendency (%)
I
IP
II
III
III der
IIIP
IIIP der
N
compensation counselling use
11.1 12.5 33.3
11.1 37.5 â•⁄ 0.0
11.1 12.5 â•⁄ 0.0
33.3 37.5 22.2
â•⁄ 0.0 â•⁄ 0.0 44.4
22.2 â•⁄ 0.0 â•⁄ 0.0
11.1 â•⁄ 0.0 â•⁄ 0.0
9 8 9
kind of job to an individual, which is typical for recipient profiling in the type-I pattern. An interesting point about chance is that chance is a significant Od-collocate with GIVE in both varieties (cf. Section 5.5), while with OFFER it is only a significant collocate in British English. This can be seen as another indication of a higher range of transfer verbs in British English where speakers of Indian English might opt for the more central GIVE. Post, lift and drugs are all infrequent, so statements about these are somewhat tentative; however, there is a tendency, at least in the case of drugs, to profile the possible recipient differently in active and in passive phrases. In the active voice the recipient is frequently realized with a to-prepositional phrase (type-II), while in the passive voice the nominal type-I pattern is the preferred choice. As was the case in the Times of India data, there are also a number of Od-collocates that show no preference for a specific pattern; these are shown in Table€7-13. As the three Od-collocates shown in Table 7-13 are all relatively infrequent and are not used with preference in a specific pattern, no closer analysis of these collocates is undertaken at this point, since the amount of speculation based on single instances would be too high owing to the variability in pattern selection. Finally, by looking at the verb-complementational profiles for the two varieties with a focus on the Od-collocates that are found in both datasets, we find that with these collocates the differences in the overall complementation patterns of OFFER + Od-collocate are much smaller than in Table 7-6. In contrast to GIVE and SEND, however, they are still significant. Consider Table 7-14.
144 Structural Nativization in Indian English Lexicogrammar
Table 7-14.╇ Verb-complementational profile of OFFER + Od-collocates (shared) Type I Ider IP II III IIIder IIIP IIIPder others sum total
ToI9m observed
ToI9m expected
χ²
BNCNews observed
BNCNews expected
χ²
â•⁄ 31 â•⁄ 10 â•⁄â•⁄ 9 â•⁄ 40 161 â•⁄ 15 â•⁄ 22 â•⁄ 12 â•⁄â•⁄ 8 308
â•⁄ 34.72 â•⁄â•⁄ 5.41 â•⁄ 16.69 â•⁄ 41.49 150.17 â•⁄ 14.43 â•⁄ 15.78 â•⁄ 16.69 â•⁄ 12.63 308.01
â•⁄ 0.40 â•⁄ 3.89 â•⁄ 3.54 â•⁄ 0.05 â•⁄ 0.78 â•⁄ 0.02 â•⁄ 2.45 â•⁄ 1.32 â•⁄ 1.70 14.15
â•⁄ 46 â•⁄â•⁄ 2 â•⁄ 28 â•⁄ 52 172 â•⁄ 17 â•⁄ 13 â•⁄ 25 â•⁄ 20 375
â•⁄ 42.28 â•⁄â•⁄ 6.59 â•⁄ 20.31 â•⁄ 50.51 182.83 â•⁄ 17.57 â•⁄ 19.22 â•⁄ 20.31 â•⁄ 15.37 374.99
â•⁄ 0.33 â•⁄ 3.20 â•⁄ 2.91 â•⁄ 0.04 â•⁄ 0.64 â•⁄ 0.02 â•⁄ 2.01 â•⁄ 1.08 â•⁄ 1.39 11.62
Table 7-14 shows how verb-complementational variation of OFFER + shared Od-collocate takes place mainly at the periphery. Not many differences can be attested for the central patterns of type-I, type-II and type-III, while there is a certain amount of variation in the passive patterns and the derivative patterns, such as in the case of the type-Ider pattern that is used ten times in the Times of India data and only twice in the BNC data. This pattern mainly encodes the type-I pattern with a fronted element, such as in (80): (80) The two services which airlines offer their passengers, […], is [sic] safety and comfort.
Apart from these differences in the use of the type-Ider pattern, there is also a certain amount of variation when it comes to the passive patterns of the type-IP and type-IIIP. Type-IP is used more frequently in the British corpus, while typeIIIP is used more frequently in the Indian corpus. However, these differences are quite minor and are based only on relatively few concordance lines. Although these findings are in line with the overall trend of using more monotransitive complementation in Indian English, this possible conclusion should be taken with a pinch of salt.
7.5
OFFER: A brief summary
As described in the introduction to this section, OFFER is not as frequent as GIVE or SEND, especially in the relatively small ICE subcorpora. This fact made a texttype specific analysis – as was undertaken for the other verbs – impossible, so Section 7.2 only differentiated between spoken and written modes in the two vari-
Chapter 7.╇ OFFER 145
eties. A central finding in this section was the tendency to use OFFER more frequently in British English than in Indian English, a fact that might to some extent complement the finding that GIVE is used more frequently in Indian English than in British English, since users of Indian English substitute OFFER with GIVE in specific circumstances.64 In both ICE subcorpora, OFFER is used more frequently in the written mode than in the spoken mode, and the type-III pattern is the default pattern. The basic ditransitive pattern, type-I, is used more frequently in British English, a point that was also observed for GIVE and SEND. Furthermore, this pattern is used more frequently in the spoken mode than in the written mode in British English, while in Indian English we find the opposite to be true. Owing to the low frequency of OFFER in the ICE data, two subcorpora, one of the BNC news section and one of the Times of India corpus (approx 1.76 million words), were analyzed for the use of OFFER in different verb-complementation patterns. The findings were in accordance with the findings in the ICE corpora in terms of the more frequent use of OFFER in British English compared to Indian English and in that the type-III pattern is the default complementation pattern in both varieties. This second analysis, however, revealed no significant differences between the varieties in the verb-complementational profile of OFFER, although there was slight variation in the use of the type-IP and the type-II pattern, where the type-IP pattern was more frequently used in British English and the type-II pattern in Indian English. The third part of the analysis of OFFER contains a combined collocational and verb-complementational analysis. Based on the complete BNC newspaper section and a comparable random set of the Times of India, collocational profiles of OFFER with respect to collocates that function in the position of the direct object were generated for both varieties. This analysis showed that British and Indian English both significantly collocate with 42 Od-collocate types, but that in British English significantly more tokens of these types are used, which is also in line with the observations of an overall higher frequency of OFFER in British English. An examination of the complementational profiles of OFFER + Odcollocate showed significant differences between the varieties, based mostly on the higher frequency of the type-I and the type-IP pattern in British English and the higher relative frequency of the type-III and the type-IIIP pattern in Indian English. The following analysis of the complementation of OFFER in relation to specific Od-collocates showed that OFFER is used most frequently in the type-III pattern with the majority of all Od-collocates in both varieties. In the Times of India data, a number of collocates was used most frequently in the type-II pattern 64. The frequent use of OFFER + chance in British English, discussed in Section 7.4, strengthens this assumption.
146 Structural Nativization in Indian English Lexicogrammar
while for the BNC-specific collocates no such trend was attested. One Od-collocate, namely role, was found most frequently in the type-I pattern in Indian English, while in British English no collocate was used dominantly in the type-I pattern, but a number of collocates were found to correlate with the type-I pattern and its structurally related pattern type-IP. Regarding the Od-collocates that were significant in both varieties, the verb-complementational variation was still significant between the varieties, but less so than in the case where variety-specific and shared collocates were analyzed together. In the case of the shared collocates, in particular the more frequent use of the peripheral type-Ider pattern (fronting of the direct object) and the passivized type-III pattern led to the differences in verb-complementation, which is in accordance with some of the observations on SEND described in Chapter 6.
7.6
Analysis of GIVE, SEND and OFFER – concluding remarks
In the last three chapters, I have shown how British and Indian English differ with respect to verb-complementation and collocation. Lexical preferences on the level of collocation influence the verb-complementational patterns chosen, and some of these lexicogrammatical combinations have distinct variety-specific functions. Although variation on the lexicogrammatical level is a pervasive element in varieties of English, it is at times difficult to interpret. For this reason, the present chapter made use of what could be called a micro-approach. A relatively small set of verbs was analyzed based on large sets of data. The use of these large datasets in general, and the combination of standard corpora with a web-derived megacorpus in particular, made it possible to analyze the differences between the varieties in minute detail. Although this approach is in contrast to many earlier descriptions of Indian English, the results of the present analysis show that this depth of detail greatly improves the tracking of individual features, especially with regard to different modes and text types. The downside of this approach, however, is partly due to a lack of larger scale annotated corpora. The manual analysis of a huge dataset clearly places restrictions on the scope of the analysis. The evaluation of this methodology as well as the implications of the results of this work for the analysis of lexicogrammatical nativization in varieties of English from a more abstract point of view will, therefore, be the main focus of Chapter 8.
chapter 8
Evaluation and discussion
8.1
Introduction
The analysis in Chapters 5, 6 and 7 showed how lexicogrammatical nativization in Indian English can be traced with respect to verb-complementation and collocation. Each of the verbs under scrutiny displays a variety-specific collocational profile. The collocates of the verbs, a feature situated towards the lexical end of the lexis-grammar continuum, correlate with the verb-complementational profiles of the verbs. Collocation, therefore, is one of the underlying factors for variety-specific differences in the field of verb-complementation, where verbcomplementation can be regarded as being situated closer to the grammatical end of the continuum. In the present chapter I will set out to evaluate the methodology applied in the analysis and relate the results to other approaches towards lexicogrammatical description. This evaluation and the consequent discussion will deal with a number of different issues: – Firstly, some of the methodological decisions underlying the analysis in the previous chapters are revisited in order to account for their strengths and weaknesses. Firstly, the choice of corpora and software will be discussed with respect to their suitability for the description of lexicogrammatical nativization. – Secondly, a closer examination at the categorization model for verb-complementational patterns will be provided. – In a third step, the choice of conditions to identify Od-collocates will be related to the possible influence of methodological decisions on analytical results. – In the subsequent section, the approach as a whole will be related to other possible approaches for the description of verb-complementation and collocation. These approaches all have elements in common with the present study, but also differ in various aspects. – Finally, I will show how these different approaches deal with different aspects of lexicogrammatical nativization on different levels of schematicity in relation to the present study, and how the results of this study can be translated into a model for the description of lexicogrammatical nativization in Indian English in the case of the encoding of cause-receive processes.
148 Structural Nativization in Indian English Lexicogrammar
8.2
Evaluation of the analysis
8.2.1
Corpora and software
The present study is based on a combined dataset that makes use of balanced standard corpora as well as a large web-derived newspaper corpus. The use of web-based data is an area of ongoing discussion within corpus-linguistics and its usefulness is still under debate. Mair (2007), for example, discusses the question as to whether the web is an “unwanted corpus” (Mair 2007:â•›235), pointing out that: The “accidental corpus” (from the title of Renouf et al. 2004) seems an appropriate phrase to capture a widespread ambiguous mood: the web will have to be used because it is there, but clearly it is not the corpus that linguists would have compiled. Outside the corpus-linguistic community the mood tends to be even more reserved as expressed in Brian Joseph’s “caveat googlator”, a sternish warning addressed to the community from the prestigious position of the Language editor’s column (Joseph 2004:â•›382). (Mair 2007:â•›236)
These cautious remarks notwithstanding, he also makes a case in favour of using internet data for corpus-linguistic research, since the data on the world-wide web offers opportunities that cannot easily be disregarded: […] despite its obvious drawbacks the English-language web is an inevitable source of data for studies on change in present day English. By its very nature, work on this topic always requires recent data, […], [and] it also requires more data than even the bigger linguistic corpora such as the British National Corpus (BNC) can provide. (Mair 2007:â•›236)
Mair’s points on the inevitability of recent data for studies on change in English also hold for the research of many varieties of English, firstly, because they are not static, but steadily evolving, and secondly, because there are no corpora that are comparable in size to the BNC. In the present study, as discussed in Chapter€4, I opted for an approach that does not make use of the internet in its boundless form, but merely uses the internet as a type of data mine in order to obtain a very large amount of Indian newspaper English. This was carried out by following Hoffmann’s (2007) reasoning that “Internet-derived corpora offer the researcher an opportunity to greatly expand the range of available data without having to unduly compromise the application of standard corpus linguistic methodology” (Hoffmann 2007:â•›70–71). After this general overview of the benefits and pitfalls of using web-data for linguistic research, I wish to present some observations that
Chapter 8.╇ Evaluation and discussion 149
are drawn from the present research, describing the effect of combining standard corpus data with web-derived data. The ICE corpora have proven to be a valuable source for my analysis. Without these, many of the finer points on the use of GIVE, SEND and OFFER could not have been made. One of the main advantages of the ICE corpora is the inclusion of a large amount of spoken data. Without this resource, results such as the markedly more frequent use of GIVE in spoken Indian English than in both spoken British English and written Indian English cannot be attested. Another benefit of the ICE corpora is the large variety of text types included. Especially for frequent verbs such as GIVE, this genre scope is very useful, as variation according to genre often compensates variety-specific variation. Therefore, an analysis within a given genre often proves more insightful than a general overview of a combination of many different genres. Although such an overview can yield interesting results, it often obscures the influence text types have on language use. A good example of this is the very frequent use of the verb SEND in the letter sections in the ICE corpora. As discussed in Chapter 6, the uses of this verb in this category account for the vast majority of the overall use of SEND in written Indian and British English. If text types were not considered, these uses would be associated with the written mode in its entirety, rather than with the strong influence of a single text type. However, this scope is at times also a drawback, since many text categories include only a very limited number of words. Particularly in the case of relatively infrequent items such as OFFER, a consistent text type-specific analysis is not feasible. The limitation in size is also a further downside of the ICE corpora for the purpose of collocational analyses. The creation of collocational profiles of specific words, as undertaken in this study, is rather limited. This is especially the case when these profiles only include content words. Therefore, such a collocational analysis based on the ICE corpora is only feasible for high-frequency items. One solution to these drawbacks is the additional use of data that is available in online archives. Although these resources lack many of the advantages of standard corpora, the combination of both datasets may level out the disadvantages of the individual studies. Mode and text type-dependent features can be identified using standard corpora, while some of the features, the analysis of which requires a larger size, can be attested within web-derived datasets. The dataset drawn from The Times of India that has been used in the present study thus proved to be a valuable additional resource. However, generating such a database as well as relating it to a British English reference corpus is never entirely unproblematic. The automated nature of such procedures can create a certain amount of unwanted text, most notably in the
150 Structural Nativization in Indian English Lexicogrammar
form of duplicates. In the present study, these unwanted texts amounted to approximately 10% of the total data. Regarding the verb-complementational study this was of minor importance, as double instances of the same sentence can easily be identified and eliminated. For the collocational analysis this proved a bigger problem, as some items were automatically identified as collocation candidates when in fact their frequent co-occurrence with a specific verb was due to multiple instances of the same text. Although this can be – and has been – solved by manual deletion of the instances, an automated process to clean the data is desirable for future projects.65 Relating the data from the Times of India corpus to a suitable British reference corpus also raises a number of issues. Although British English is often used as a point of reference for post-colonial varieties of English because British English is the historical input variety, this is not entirely uncontroversial. Firstly, today’s British English has undergone tremendous change since the time in which it served as the input variety to most of the post-colonial varieties of English, where the time lag may convert the reference model into something of a “moving target” (Mair 2002:â•›111). Although British English is still upheld as a prestigious model for many speakers of post-colonial varieties of English, its influence today may have waned in favour of other influences. Secondly, the process of corpus compilation in itself creates specific time lags, so additional diachronic discrepancies between the different corpora exist, an issue which may accelerate the velocity of the target, metaphorically speaking. Hunston (2002) addresses this point in relation to corpus annotation: The work involved in annotation acts as a constraint against updating or enlarging a corpus. For example, the 1961, 1 million word LOB corpus is still used as a source of data, small and outdated though it is, precisely because it is parsed. (Hunston 2002:â•›92)
Hunston’s (2002) point applies to other older corpora too. The London-Lund Corpus (LLC), which contains prosodic annotation, as well as the parsed version of ICE-GB, suffer to a certain extent from the situation described by Hunston (2002), as the value of their annotation is often regarded as being more important than the date of their collection. The choice of the newspaper corpora employed in the present study is, therefore, not without problems either, as the texts included in the Times of India corpus are more recent than the texts included in the BNC newspaper section. Particularly on the level of lexical change this has a 65. For the present analysis this drawback was of minor importance, as all collocation candidates had to be manually checked for their possible function as a patient in a cause-receive process. During this step redundant candidates were identified.
Chapter 8.╇ Evaluation and discussion 151
certain influence on the results. SEND in combination with email is a case in point for this. These drawbacks notwithstanding, the collocational analysis showed that there is a relatively large area of overlap between the corpora of the different varieties, which can be taken as an indication of their comparability. At the same time, many variety-specific features that are not subject to the time lag between the corpora have been pointed out. Good examples of this are the culturally motivated collocations GIVE + prayer and GIVE + namaz in the Times of India or the topicbased use of the pattern SEND Od V-ing in financial texts in the BNC data. The software used in this study consisted mainly of the corpus analysis package WST v4.x and a number of custom-made Perl scripts. WST has proven to be an invaluable resource for corpus-based analyses of language. It is very fast in comparison to other linguistic tools and it provides a number of functions that are essential for the analysis of unmarked corpora. Especially the possibility of indexing huge corpora in order to provide a comprehensive set of collocates for any given word in a corpus has been of tremendous value for my analysis. However, there are some disadvantages in working with this tool that should not go unmentioned. As WST is a toolset that is not tailor-made to fit a specific corpus, at some points there are incompatibilities between the functions of the toolset and the design of a given corpus. A good example of this is the SGML mark-up in ICE-India, which includes commentaries that cannot be easily filtered by WST. If this is not addressed by certain changes to the mark-up, it leads to WST counting mark-up items as regular words.66 A further drawback is related to the possibilities for lemmatization. While it is possible to lemmatize tokens, this function does not carry over to the indexing function that is needed for creating a collocation list. For this reason, only lemmata whose single word forms meet a quantitative criterion are listed as collocates in WST, which in combination with small corpora may well lead to very sparse results. Another issue is that although WST is not an open-source package, it is still regularly updated to fit different users’ needs and, additionally, some errors in the source code have been corrected in various updates, which leads to updated programmes sometimes working slightly differently from earlier versions. These small changes in the source code aptly illustrate the fact that blind trust in software tools is often not advisable. Danielsson (2004) addresses some of these problems and advises customizing linguistic work by using simple Perl scripts:
66. This can sometimes yield ludicrous results. For example, a naїve keyword analysis of ICEGB and ICE-India will show that the mark-up item letterhead is one of the key-words in Indian English vocabulary.
152 Structural Nativization in Indian English Lexicogrammar
Today many of the available corpora are marked up with SGML […] or XML […], however, most corpus tools work on plain text […]. Further […] many texts have additional breaks in each line […]. When using a concordance tool such as WordSmith […] this will alter the display of text and therefore may require the removal of these additional breaks. […]. All that is needed to rectify each of these (Danielsson 2004:â•›227) problems is a small program.
In the present work the main concern was the above-mentioned difficulty with the SGML mark-up of ICE-India, and rectifying this with a small programme was a useful step. However, Danielsson’s argument that linguists should make more use of program languages such as Perl has wider implications. Firstly, it would not have been possible to generate the Times of India corpus without automatic procedures. Additionally, even smaller tasks, such as comparing lists of collocates in order to account for what is shared and what is variety-specific, are greatly facilitated by such programmes. Thus, the use of corpus-linguistic software, such as WST, is indispensible for many corpus studies, but owing to the universal nature of such tools it is highly advisable to customize any approach to the needs at hand. Ultimately, however, some basic programming expertise must become a staple part of any corpus linguist’s methodological toolkit. 8.2.2
Verb-complementational patterns
The categorization system for verb-complementational patterns used in the present study is based on Mukherjee & Hoffmann (2006) and Mukherjee (2005). This system provides a very detailled approach, since Mukherjee’s (2005) original goal was to provide a comprehensive analysis of the complementation of a selection of ditransitive verbs: It should be noted that the distinction of patterns here is much more fine-grained than, say, the traditional ‘clause patterns’, because, for example, each clause pattern (being defined on the basis of configuration of clause elements) covers a wide range of different realisations of a particular clause element (cf. Biber et al. (Mukherjee 2005:â•›93) 1999:â•›141).
For the description of variation between Indian and British English, in particular with a specific focus on modes and text types, this high degree of specialization was renounced in favour of a simplified system. Whereas the system proposed by Mukherjee (2005) contains a number of structurally related patterns for each of the basic patterns, these were collapsed into one category for all the derivative patterns of any given basic pattern. This decision was helpful, as it was possible to reduce the total number of patterns without jeopardizing the statistic results
Chapter 8.╇ Evaluation and discussion 153
for pattern distribution. Therefore, the results based on the collapsed distribution were comparable to the ones supplied by Mukherjee & Hoffmann (2006) for GIVE and SEND. Furthermore, this simplification made it far easier to account for the variation of patterns within the different text-categories of the ICE corpora, where an overly detailed approach would have been problematic rather than helpful. A large number of derivative patterns, useful for the description of all possible realizations of clause elements, become very difficult to account for in the specific – relatively small – text type-based categories of the ICE corpora. With the BNC newspaper section and the Times of India, I used corpora that are much larger than the ICE corpora. In contrast, the works of Mukherjee (2005) and the description of SEND in Mukherjee & Hoffmann (2006) are based exclusively on ICE corpora. Therefore, a specific pattern of SEND that has a certain currency in British as well as Indian newspaper English was not specified in these earlier studies. The pattern in question – SEND Od V-ing, schematized as (S) SEND [Od:NP] [CompC: VPcont] – was thus added to the pattern-set of SEND. This pattern is of specific importance in the description of variation in newspaper Englishes. Although genre conventions tend to homogenize variation (cf. Sand 2004), there are marked intervarietal differences in the use of this pattern. This result underlines the importance of the use of databases larger than the ICE corpora, as many products of nativization will firstly occur at the peripheral use of linguistic items, and small corpora such as the ICE corpora can only serve to describe central uses. While the pattern SEND Od V-ing is quite salient in British English, there is only one attested use of it in ICE-GB, which is the reason why Mukherjee (2005) listed this occurrence under miscellaneous, rather than treating it as a pattern in its own right. In sum, the revised categorization model proved useful to describe the differences in terms of the complementation of potentially ditransitive verbs. At some points, however, the functional nature of this model makes it necessary to add additional reasoning to the classification of specific sentences. Especially in the case of the type-II pattern it is noteworthy that in many cases the type-I pattern is not a genuine alternative, such as, for example, in the case of inanimate recipients. For this reason I have argued in favour of adding a constructional approach to the model by considering these instances as realizations of the transfer-causedÂ�motion construction proposed by Goldberg (1995). Combining construction-grammar approaches with corpus-linguistic analyses, however, bears its own challenges. Like many construction-grammar approaches, Goldberg’s (1995) model is highly schematic and is better suited to explaining relatively simple invented examples than to account for the complexity of natural language. An advantage of using this schematic model is, for example, that the transfer-caused-motion construction does not suffer from possible
154 Structural Nativization in Indian English Lexicogrammar
confusion of objects and adverbials when it comes to the profiling of recipients, because the realization of the recipient is simply coded as an oblique element. However, if this this model is applied to the large amount of natural language that formed the basis of my analysis, many of the finer points that are captured with this more fine-grained functional model are lost. The difficulty in combining corpus-linguistic methodology with construction-grammar approaches is not always apparent at first glance. Mak (2005), for example, emphasizes the similarities in the underlying assumptions of both disciplines: […] Construction Grammar is a theory which models language as a conceptual structure grounded in the way it is used with an experiential correspondence with the world. This ties in neatly with the basic philosophy of corpus linguistics that language can be better understood through occasions of use, and that meaning often emerges from linguistic collocational patterns which in themselves are evident in their repeated occurrence in everyday language use. (Mak 2005:â•›14)
On the other hand, however, many works in construction grammar seem more interested in acceptable peripheral structures and their explanations than in frequent and typical expressions. Hanks (2008), for instance, discusses an example from Goldberg and Jackendoff (2004), namely Jack watered the plants flat, with a certain amount of reservation: [This] is an example concocted by Goldberg and Jackendoff and, according to them, it is idiomatic. But water, unlike, say, hammer, is not a verb that normally participates in this construction. In the unlikely event of such a sentence being found in a real text, we should say that it is an exploitation of a norm, not an idiomatic use. When, occasionally, they cite an authentic example from Google, they are apparently unaware of the argument by corpus linguists that authenticity alone is not enough: evidence of conventionality is also needed. (Hanks 2008:â•›228)
There are recent trends to overcome this discrepancy between the similarity of underlying assumptions between corpus-linguistics and construction grammar on the one hand, and the mismatch of the objects of investigation on the other. Bergs (2008), for example, suggests ways to merge these fields in a more systematic manner, referring to the example of collostruction analysis: The descriptions and analyses in CxG [construction grammar] are usage-based, i.e., they begin with actual language data often extracted from large corpora of spoken and written language […]. So-called collostructional analyses – hugely complex statistical operations on enormous amounts of corpus data – have shown that in some cases minute differences in use can be indicative of different constructional and cognitive status for superficially very similar items. (Bergs 2008:â•›271)
Chapter 8.╇ Evaluation and discussion 155
The present analysis is similar to these procedures, and some of the results show that the use of Goldberg’s (1995) framework combined with a classification model based on natural language data helps to solve some of the problems.67 For instance, the distinction between two versions of the type-II pattern that have slight semantic differences – where one has an admissible correspondence with the type-I pattern and the other does not – can be treated in a way suggested by Bergs (2008). While both are superficially similar and are, therefore, coded as instances of the type-II pattern, there are slight constructional and cognitive differences between the patterns. Although the ditransitive construction is largely semantically synonymous with the transfer-caused-motion construction, there are pragmatic differences. If the recipient is inanimate, for example, the transfer-caused-motion construction is the only admissible construction, while with animate recipients often both constructions are admissible. However, even if both constructions are admissible, the use of one of them adds to the meaning of the sentence. In this light, it is remarkable that speakers of British English have a strong bias towards ditransitive usage, mainly manifest in the use of the type-I pattern, when compared to speakers of Indian English. Additionally, in the case of SEND in spoken language, British speakers use the type-II pattern exclusively with inanimate recipients, while speakers of Indian English use this pattern with inanimate as well as animate recipients. Thus, the incorporation of Goldberg’s constructional framework into the revised pattern model has added some important information to the study, which the use of one of these models alone may have obscured. 8.2.3
Collocational profiles
The collocational analysis of the verbs in this work was based on a combination of quantitative and qualitative conditions that had to be met in order to identify collocates that function in the position of the direct object, so-called Od-collocates. In the following I will revisit these conditions in order to provide a post-hoc evaluation of the influence of each of them: Condition 1: Collocation candidates must occur in a span of 5:5 of a node word that is a predefined trivalent verb. Condition 1 is the basis for the identification of all possible collocation candidates. The span that is set in this condition is largely arbitrary and has been chosen for traditional reasons, rather than inherent linguistic reasoning. A 5:5 span 67. A discussion of similarities and differences to collostructional models is undertaken in Section 8.3.
156 Structural Nativization in Indian English Lexicogrammar
is often used as a standard span and there seems to be some agreement that symmetric spans between three and five are suitable for collocational analyses. To my knowledge, spans of these sizes underlie almost all work on collocation and have remained fairly unchallenged.68 However, there may be some benefits in using different spans. For example, it might be argued that in the verb phrase most of the elements follow the verb, which would be grounds for setting an asymmetric span that covers more elements to the right of the verb. For the present analysis, however, the drawbacks of an asymmetric span seemed larger than the possible benefits. Since the focus lies on the combination of collocation and verb-complementation, an asymmetric span may have favoured active and basic patterns over passive and derivative patterns. This is clearly undesirable, as the results would have been unduly influenced by an arbitrary choice. The arbitrary choice of a symmetric 5:5 span may also have influenced the results, as elements that function as direct objects outside this span are not identified. However, this is only the case in the initial automatic identification of collocation candidates. Once a collocation candidate had been identified, the manual steps that were applied allowed (rare) occurrences to be integrated where these candidates did not occur within the given span. Condition 2: A minimum number of five co-occurrences must be met. Condition 2 is also an arbitrary condition that has been found to be useful for many of the standard corpora and underlies the majority of the present work on collocation. In the work with the larger corpora in this analysis, the threshold of five often seemed to be too low, as this condition allows for some candidates that are later filtered out by statistic testing. It has been retained, however, in order not to jeopardize the comparability of the collocational profiles of GIVE, based on small corpora, and those of SEND and OFFER, which are based on larger corpora. Additionally, the other conditions eliminate many items that occur only five times in the larger corpora so this low figure could be regarded as a minor inconvenience for the researcher that does not, however, influence the analysis in any major way. Condition 3: Collocation candidates must be statistically significant according to a log-likelihood value of ≥ 15.13 (p ≤ 0.0001). Condition 3 is the statistic condition that underlies the identification of collocation candidates. The significance level chosen for log-likelihood is very high, 68. It should be pointed out, however, that there are some recent proposals for less arbitrary spans, as for example Daudaravičius & Marcinkevičienė (2004).
Chapter 8.╇ Evaluation and discussion 157
which is due to the fact that log-likelihood testing only resembles exact testing (albeit closely) (cf. Evert 2005:â•›111); possible discrepancies are avoided by this very high setting. In retrospect, the choice of this statistic condition has proved to be a good way of analyzing large numbers of possible collocation candidates without having to apply exact testing to each of them. There has, of course, been some debate as to the importance of statistic significance for the description of linguistic findings. Hundt (1998), for example argues: [T]he real problem is to find hypotheses for statistically significant differences in word frequencies that are both easy to verify or falsify and that gave a linguistic explanation. In other words, statistic procedures, in this study, were not found to be a useful tool for the discovery of grammatical differences between national varieties of English. Statistical significance in itself tells us nothing about the worth (Hundt 1998:â•›130) of our findings (see Butler 1985:â•›68).
While it is certainly true that statistic significance in itself does not reveal much about the value of linguistic findings, statistic significance is arguably a conditio sine qua non. Even if significance is not a sufficient condition to describe linguistic variation, it is a necessary condition that has to be taken into account before any data is interpreted. Gries (2006), for example, makes a case for the application of up-to-date descriptive methodology, pointing out that: […] in spite of the methodological advances and the overall very promising development, many corpus-based studies exhibit a variety of what may be considered methodological shortcomings. This is particularly astonishing since, (i)€presumably, for many scholars part of the reason to turn to corpus-linguistic methods may well have been a perceived dissatisfaction with methods from socalled ‘armchair linguistics’ (in the sense of Fillmore 1992) and (ii) there are other scientific disciplines which have already successfully coped with many of these (Gries 2006:â•›192) problems.
From this point of view, applying the statistic procedures described above has been very helpful in order to identify the lexical items that were of specific relevance for the collocational profiles of the verbs in question. However, not every item that co-occurred statistically significantly with any of the verbs was ultimately treated as a relevant collocate for this study, owing to the inclusion of a final qualitative condition. Condition 4: Collocation candidates qualify as Od-collocates if the collocate fulfils the semantic function of a patient in a cause-receive process. Condition 4 was applied in order to obtain the collocates that are used as potential patients, thus encoding direct objects. These have been labelled Od-collocates.
158 Structural Nativization in Indian English Lexicogrammar
This condition was of central relevance for the present study, because the purely numerical conditions include a high number of elements that do frequently cooccur with the verbs in question, but do not have much influence on the complementation patterns that ditransitive verbs can take.69 For example, auxiliaries will always collocate relatively strongly with any verb and will therefore show statistic significance. It is, however, not useful to include all these elements when the point of interest is the influence of – partially culturally motivated – Od-collocates on the complementation of a verb. By taking Hundt’s (1998) observation on the usefulness of statistic testing into account, this additional qualitative condition made it possible to isolate the elements that have been the focus of the analysis.
8.3
Discussion
8.3.1
Introduction
The present analysis ties in with earlier work on the complementation of ditransitive verbs in different varieties of English. In the following, I will briefly compare the methodology and possible results of this analysis to other existing models for the description of lexicogrammatical patterns both in native and in second language varieties. I will set out to relate the procedures applied in this study to Stefanowitsch & Gries’ (2004) work on collostructions as well as to Stefanowitsch & Gries’ (2005) work on the covariation of collexemes. Subsequently, I will compare the present work to Mukherjee & Gries’s (2009) work on collostructional nativization in New Englishes. Finally, I will discuss the framework of this work in relation to Bresnan & Hay’s (2008) work on different covariates of the dative alternation in New Zealand and American English. 8.3.2
Od-collocate analysis and collostructional analysis
Stefanowitsch & Gries’ (2004) collostructional model has some parallels with the work that underlies the description of collocation and verb-complementation in this analysis. They show how specific verbs display a tendency to be used in either alternative of the dative alternation: We propose a similar method for the analysis of alternating pairs, differing from Church et al. [1991] in that we look at near-synonymous (or functionally 69. Alternatively they are in their very nature determinants or parts of a specific pattern such as the preposition to in the case of the type-II pattern.
Chapter 8.╇ Evaluation and discussion 159
near-equivalent) constructions rather than words, and that we focus on words appearing in particular slots in these constructions rather than at all words within a given span (we refer to such words as collexemes of the construction(s) in question). (Stefanowitsch & Gries 2004:â•›101)
This quotation aptly illustrates some of the similarities as well as some of the differences between collostructional analysis and the procedure opted for in the present work. Firstly, although both approaches deal with potentially ditransitive verbs that have the constructional alternatives of the ditransitive and the to-Â�dative€– to use Stefanowitsch & Gries’s terminology – the present analysis also accounts for all other possible patterns of verb-complementation for these verbs. Secondly, Stefanowitsch & Gries (2004) describe the verbs in question as collexemes of the construction. They find that some verbs are more closely connected with one of the two options that they consider, while others are attracted by the other. In this work, the same could also be shown (though to a somewhat finer degree), by firstly describing the use of the verbs in different patterns in the different varieties and secondly relating the trends for the verbs to occur in a specific construction to a further element, namely the Od-collocate. Thus, while the present study is largely consistent with Stefanowitsch & Gries’s (2004) approach, it also makes a case for integrating further elements into the model and looking at the full spectrum of the possible complementation patterns of the verbs. Concerning the integration of more elements into a collostructional analysis, Stefanowitsch & Gries (2005) themselves extend their collostructional model: In this paper we extend a ‘single-slot’ methodology developed in Stefanowitsch and Gries (2003) to the investigation of potential interactions between two slots and apply it to the into-causative. We show that such interactions exist, i.e. that cause and result predicates ‘covary’ systematically. We then consider two factors influencing this covariation: a cognitive one, based in causative event types; and a cultural one, based in knowledge about frames and possible cause-effect relations (Stefanowitsch & Gries 2005:â•›225) between them.
This extended model bears some similarity to the analysis undertaken in this study, as not only one element is related to a specific construction, but also the interrelation with other elements is integrated. In the present work, however, the relations of the collocates to the verbs, and the tendency of these verb-collocate pairs to be frequently used in different complementation patterns were analyzed separately and to a large extent manually. This is based on the assumption that a stringent collostructional analysis of the sort that Stefanowitsch & Gries (2004) propose is only possible for fully parsed corpora, such as ICE-GB. This is due to the fact that in order to achieve the statistic values for attraction of a lexical element into or repulsion of a lexical element from a construction, all instances of
160 Structural Nativization in Indian English Lexicogrammar
this construction need to be known. With unparsed corpora, however, it is not possible to gain clearcut figures on how often a construction occurs, regardless of a specific verb. This limitation of the collostructional model is arguably its largest weakness.70 The problem of having to rely on a fully parsed corpus is less prominent in Stefanowitsch & Gries (2005), because the construction that they analyze can be more easily traced than the ditransitive construction and its possible alternatives, and can therefore also be traced by using a corpus that is not fully parsed, but only POS-tagged. In contrast, regarding the ditransitive construction, Mukherjee & Hoffmann (2006) propose a method of retrieving possible instances from a large POS-tagged corpus: A syntactically parsed corpus like ICE-GB is of course ideally suited to an analysis of grammatical patterns such as ditransitive verb-complementation. In the absence of this type of information, sequences of part-of-speech tags can be defined to retrieve potentially relevant constructions. In the case of ditransitive verbs with the type-I pattern, such a sequence would need to match all verbs that are immediately followed by two noun phrases. Since noun phrases can be extremely complex, it is not feasible to make a complete list of all potential realizations. (Mukherjee & Hoffmann 2006:â•›160–161)
Although this method proved to be very valuable for retrieving low-frequency phenomena of verbs that enter the ditransitive construction in Indian English€– one of the aims of Mukherjee & Hoffmann (2006) – the above quotation shows the problems for adjusting this method to collostructional analyses. With the ditransitive construction, or the type-I pattern alone, it is not possible to reach a recall of 100% of all possible instances of this construction. In order to analyze alternations of this construction, these alternations would also have to be retrieved semi-automatically from a POS-tagged corpus, further lowering recall. In sum, the methodology applied in this study has proven to be a profitable alternative to a collostructional analysis. In contradistinction to a collostructional approach, grammatical patterns were not treated as countable elements in their own right, because a comprehensive description of all instances of all verbs is clearly not feasible for any larger body of unparsed text. This is especially true when we consider the type-III pattern as one of the alternative patterns for verbs that have been shown to function in both the ditransitive and in the transfer-caused-motion construction. Alternatively, these patterns were treated as possible patterns in which specific verbs occur. Although this does not yield a statistic measurement for the attraction or repulsion of a specific verb to a specific construction or 70. This problem can be avoided by extrapolation, such as, for example, in Mukherjee & Gries (2009), which will be discussed in the following section.
Chapter 8.╇ Evaluation and discussion 161
pattern, clear trends for verbs can be shown. By adding a further element, collocation, to the analysis, it is also possible to show how verbs, patterns and collocates are in different relations in different varieties of English. 8.3.3
Lexicogrammatical variation in varieties of English
While the previous section related the present analysis to other frameworks of the description of co-occurrence between lexical and grammatical elements, this section describes two recent studies, both of which focus on the differences between varieties of English. Both of these studies are related to the present work, although both apply different methodologies and reach different, but not contradictory, conclusions. The first is the work of Mukherjee & Gries (2009), which provides a collostructional analysis of verbal constructions for the Englishes spoken in Hong Kong, India and Singapore. Their focus lies on the description of the possible attraction of a set of verbs to specific complementation patterns in the different varieties. The methodology used in this study is largely analogous to the methodology applied in Stefanowitsch & Gries’s (2004) original work on collostruction. The above-mentioned difficulties with unparsed corpora have been avoided by a combination of extrapolation and manual coding. Their focus on structural nativization at the level of grammar is somewhat different from the focus of the present analysis, as they include more verbs and categorize these into cases of ditransitive, monotransitive and intransitive complementation. As in Stefanowitsch & Gries (2004), no further covariates, such as for example collocations of the verbs, are taken into account. The results in terms of Indian English, however, are largely in line with the overall results of the present study. This is particularly the case regarding the observation of a preference for monotransitive complementation in Indian English with verbs that are more closely associated with ditransitive complementation in British English: In particular, there are various verbs in this group that do not prefer the ditransitive construction in one of the Asian Englishes (e.g. convince in Hong Kong English, cost in Indian English and lend in Singapore English); in all of these cases the verbs at hand prefer the monotransitive construction in the Asian English variety. That is to say, if a verb is repelled by the ditransitive construction in the course of the evolution of a New English variety, it tends to be attracted to the (Mukherjee & Gries 2009:â•›47) monotransitive construction instead.
Owing to the quantitative nature of Mukherjee & Gries’s (2009) study, however, which focuses on the description of statistically significant similarities and differences between the varieties, the reasons for this trend remain largely unclear. Several hypotheses for the trend towards monotransitivity in Indian English have
162 Structural Nativization in Indian English Lexicogrammar
been discussed in the previous chapter, collocational preference and the use of stretched-verb construction being two of them. In a further work on differences in verb-complementational variation of the verb GIVE, Bresnan & Hay (2008) show that there are several additional factors that may be of relevance for the difference within verb-complementation between varieties of English. Although they focus on two native-speaker varieties of English, they show that in different socio-cultural settings the influence on factors such as animacy of a given element may vary. In order to achieve this, Bresnan & Hay (2008) code their data for a variety of features, including numerical variables such as the length of specific elements, as well as categorical variables such as animacy or discourse accessibility. Their results show that these variables have a predictive value for the choice of either alternative of the dative alternation. Especially in the case of animacy, there is a significant degree of variation between speakers of New Zealand English and American English: Finally, there is an interaction of variety with animacy […]. Non-animate recipients are more likely to be used in the double object construction in the NZ than in US spoken data, and this is so, of course, independently of the other variables. (Bresnan & Hay 2008:â•›252)
Bresnan & Hay (2008) do not regard this result to be particularly surprising, as it is in line with several other studies on the effect of animacy in New Zealand English (cf. e.g. Hundt 1998), and animacy furthermore plays a role for word order choices “in ditransitive constructions with dative verbs” in “a number of unrelated languages” (Bresnan & Hay 2008:â•›18). For them, however, this co-variation between animacy and variety is grounds to argue that: [t]he variability we have found provides evidence in favor of models of grammar which are quantitative and learned from exposure to other speakers. Any such grammar is likely to display some variability, depending on the nature of the exemplars that successive generations are exposed to. (Bresnan & Hay 2008:â•›255)
Bresnan & Hay’s (2008) observations are drawn from the observation of nativespeaker varieties of English over time and geographical area. It is therefore not surprising that they do not use the terms nativization or endonormative stabilization for the description of their findings. However, these observations can to some extent be transferred to the variational model used in the present study, as exposure to other (non-native) speakers will also influence the development of second language varieties, a point discussed in Section 6.4. Besides the co-variation of animacy and variety, Bresnan & Hay (2008) also show that there is a large number of other underlying factors for the choice of the
Chapter 8.╇ Evaluation and discussion 163
dative construction or its prepositional alternative, such as log-length of elements or pronominality.71 Compared to Bresnan & Hay’s (2008) work, the present study is both more restricted and more complex in the description of the verb-complementation of ditransitive verbs. It is more restricted in only including one possible covariate for verb-complementation. Other factors, such as pronominality and discourse type have been discussed, but were not subject to statistic testing. The reasons for this lie in the more complex nature of the present study with regard to the inclusion of a much larger spectrum of possible patterns and the description of verbs that are less frequent than GIVE. In many of the patterns described in the present work, specific elements of the cause-receive process are not profiled, so a statistic comparison of the influence of the log-length of elements is problematic. A second reason concerns the feasibility of such an analysis within the scope of the present project. Since most of the underlying corpora are plain-text corpora, roughly 10,000 sentences had to be categorized manually for their complementation patterns. A further categorization of elements such as syntactic complexity or discourse accessibility may have shed more light on some of the influences of these factors, but was not feasible within the scope of the present work. However, some of these factors may well prove to be influential for the description of nativizational processes on the level of verb-complementation. A recent feasibility study by Cook (2008), for example, shows that “relative complexity is an extremely predictive covariate when trying to predict the dative alternation” and that “covariates relating to recipient attributes appear to be more predictive than covariates relating to patient attributes” (Cook 2008:â•›53).
8.4
Towards a model of lexicogrammatical nativization
In Chapters 5, 6 and 7, the interrelations of collocation and verb-complementation in Indian and British English were analyzed quantitatively and qualitatively, while in the first sections of the present chapter the underlying framework was compared to other studies that describe lexicogrammatical interrelations as well as nativization at the lexis-grammar interface. The quantitative results in presented in these chapters confirm that Indian English displays structural features that differ markedly from British English. These findings are in contrast to the traditional view that variety-specific innovation and nativization are mainly detectable within phonology and lexis and that “grammatical variation tends to be less extensive and certainly less obtrusive” (Quirk et al. 1985:â•›17). This view has been challenged by 71. Bresnan & Hay (2008) weight elements according to the number of graphemic words contained. To surpress outliers they convert these weights to a logarithmic scale.
164 Structural Nativization in Indian English Lexicogrammar
Table 8-1.╇ Degrees of schematicity (cf. Langacker 1999b:╛122) Degree of schematicity
Example
mainly specific schematic in several positions constructional schema with a specific element all elements schematic
crane X’s neck X take Y over X’s knee and spank Y [[send] [NP] [NP]] [[V] [NP] [NP]]
several recent studies, such as the ones discussed in the previous sections. However, much of this work has been carried out in the form of pilot studies that have only rarely been followed up. Furthermore, these studies differ to a certain extent from the analytical procedure used in the present study, as they either describe only verb-complementational nativization (cf. e.g. Olavarría de Ersson & Shaw 2003; Mukherjee & Hoffmann 2006), or relate a small set of alternatives, such as the dative alternation, to a larger set of highly schematic covariates (cf. e.g. Bresnan & Hay 2008). In contrast, the present analysis has opted to focus on several degrees of schematicity, following Langacker’s (1999b) reasoning “that lexicon and grammar grade into another so that any specific line of demarcation would be arbitrary” (Langacker 1999b:â•›122). To exemplify this procedure, Table 3-3 is repeated here as Table 8-1. The present analysis uses the highest degree of schematicity as the starting point, since it deals with verbs that are potentially used in the ditransitive construction (in Table 8-1 shown as a verb that is followed by two noun phrases). The choice of several specific verbs is thus one degree less schematic than the starting point. At this level of schematicity, the choice of the specific item forms the basis for the following steps. In the present work three specific verbs were chosen on the grounds of their frequency and differences in semantics. All three verbs are admissible in the type-I, the type-II and the type-III pattern in both varieties. Notably, it is also possible at this level of schematicity to investigate specific elements that are admissible in a given schema in one variety but not in the other, such as, for example, in Mukherjee & Hoffmann’s (2006) description of new ditransitives. The focus of Bresnan & Hay’s (2008) study lies somewhere between the one used by Mukherjee & Hoffmann (2006) and the one adopted in the present study in terms of the description of schematic and specific elements. Bresnan & Hay (2008) deal with two schematic variants, namely the dative alternation, for the specific verb GIVE. In contrast to Mukherjee & Hoffmann (2006), Bresnan & Hay (2008) further analyze specific realizations of the schematic elements that encode recipients and patients. However, these specific realizations are not dealt with in
Chapter 8.╇ Evaluation and discussion 165
their concrete lexical forms, but are analyzed on a certain level of abstraction by assigning them numerical and categorical values in order to construct a statistic model that allows for a prediction of the chosen alternative. As discussed above, this procedure is problematic when more than two constructional options are taken into consideration, since various elements may not be profiled in a specific construction. The present study therefore includes specific lexical elements that function in the position of direct objects and are realized as NPs. Although this procedure only takes a limited number of variables into account, its focus on specific lexical realizations allows for a description of variety-specific innovations that are based much more strongly on meaning and content. Especially for the description of variety-specific variation, this focus on meaning and content is necessary, since the speakers decide what they want to encode linguistically before using a pattern that is associated with the specific elements. The present analysis, for example, showed that speakers of Indian English use the verb GIVE with a markedly greater frequency than speakers of British English when they want to encode transfer processes. Furthermore, different specific Od-collocates are used with the verbs by speakers of different varieties of English. Therefore, the speaker makes the choice of a specific verb with its Od-collocate before encoding the extralinguistic situation in one of the patterns that is admissible for the verb and typical for the verb + Od-collocate combination in the variety. An example from the analysis in Chapter 7 is the use of the verb OFFER in combination with namaz or prayers when the speaker wants to encode a situation of prayer. The choice of the habitual ditransitive verb OFFER in combination with one of these collocates, rather than the choice of the simplex verb PRAY is firstly in line with the observation that speakers of Indian English often prefer light-verb constructions over simplex verbs. Secondly, this may also be culturally motivated if the process of praying is cognitively regarded as an offer to a deity, rather than a direct and possibly unilateral act towards a deity. The salient combination of OFFER with the Od-collocate namaz in turn leads to the choice of the type-III pattern in the vast majority of cases, as the respective recipient is an implicit part of the meaning of the collocate. As this socio-culturally motivated verb + Od-collocate combination has far less currency in British English, this in turn has a certain influence on the distribution of the different patterns that are frequently used in combination with OFFER. Furthermore, the frequent use of the respective verb in collocation with its current Od-collocate strengthens the collocational status of the collocate diachronically. This process is depicted in Figure 8-1. Figure 8-1 shows the process of encoding cause-receive processes with regard to lexicogrammatical variation. The extralinguistic situation of a cause�receive process evokes the ditransitive situation schema in which the acting
166 Structural Nativization in Indian English Lexicogrammar
Extralinguistic situation cause-receive process Evokes Acting entity
Direct object
Collocation
Ditransitive verb
Collocation
Indirect object
Encoding
Collocational nativization
Profiling Selection of lexical elements
Verb-complementational nativization
Ditransitive verb
Transferred entity
realization Type-I pattern
Type-II pattern
Type-III pattern
…
Figure 8-1.╇ Lexicogrammatical nativization in cause-receive processes
entity causes an affected entity to receive a transferred entity. The lines between the acting entity, the transferred entity and the affected entity show the transfer process in chronological order, while the dotted line between the acting entity and the affected entity indicates an indirect interrelationship (cf. Mukherjee 2005:â•›37). The possible transferred entities within the ditransitive situation schema range along a cline from concrete to increasingly abstract entities. On the concrete side of this cline, actual physical objects are transferred from one person to another. Transfer of information could be considered to range within the middle of the cline, while the transfer of abstract concepts such as for example give an advantage is situated towards the abstract end of the cline.
Chapter 8.╇ Evaluation and discussion 167
The cause-receive process is encoded by the use of a ditransitive verb, a trivalent verb that allows for the profiling of all three elements. However, although all ditransitive verbs allow for the profiling of three elements, it is not obligatory to actually profile all three. In particular the affected entity is often not profiled, which leads to a monotransitive use of the ditransitive verb. In order to profile these arguments, the speaker makes a selection of the actual lexical elements. An important factor for the selection of these elements is the linguistic experience of the speaker with the language or variety. Since any speaker of a variety is exposed to natural language of the respective variety, this experience works as a filter for the choice of actual elements. While the choice of the lexical item that represents the transferred entity is mainly based on the extralinguistic situation, the choice of the ditransitive verb is influenced by these conventions, as specific verbs collocate with specific lexical realizations of transferred entities. The collocational relation of the verb with the realization of the affected entity is weaker, as the element is often not profiled and the main choice in selecting the element is between nominal and pronominal realization. After the selection of lexical elements, the speaker encodes the cause-Â�receive process in a sentence that is built in one of the sentence patterns associated with the chosen verb. If all three elements are profiled, the sentence is encoded in one alternative of the dative alternation (type-I or type-II pattern). Here a number of factors influence the realization of one of the alternatives, such as, for example, variational collostructional preferences, as described by Mukherjee & Gries (2009), or pronominal realization of the indirect object. If the affected entity is not profiled, the sentence is encoded in the type-III pattern. In this case, variational conventions also influence the lexicogrammatical choices, as speakers of Indian English, for example, tend to profile fewer elements than speakers of British English (cf. Mukherjee & Schilk 2008). Additionally, the frequent collocation of ditransitive verbs with eventive nouns in light-verb constructions in Indian English that was pointed out in Chapter 5 often occurs in sentences that are complemented in the type-III pattern. Furthermore, sentences can be encoded in any of the peripheral patterns; for example when only the affected entity is realized as an indirect object, or when both the transferred and the affected entity are not profiled. These possibilities are indicated by the dotted lines that connect the direct object and the indirect object with the placeholder for peripheral patterns (…). The concrete lexical encoding of the cause-receive process in turn becomes part of the body of natural language that other language users experience. This can happen on a small scale in conversations, where only a single recipient is exposed to the concrete sentence, but it also happens on a much larger scale when the sentence is published. Particularly in the case of large newspapers, such as The Times of India, this multiplying effect is significant, owing to the large circulation
168 Structural Nativization in Indian English Lexicogrammar
figures. This process is represented by the two upwards arrows labelled collocational nativization and verb-complementational nativization in Figure 8-1. Although the structural differences between varieties such as Indian and British English may be more opaque, for example, than lexical differences, this steady process of language use and language exposure has a strong long-term influence on the features of a variety. Schneider’s (2004) comment on Algeo (1989) describes precisely how these opaque features serve to differentiate varieties on a finer scale than, for instance, differences in pronunciation or the lexicon: Algeo (1989) is a masterly demonstration of how subtle but nevertheless consistent and unavoidably conspicuous the differences are: in terms of lexical choice, mode of expression, choice between structurally equivalent alternatives, and the like, each variety has its own distinct preferences, and speakers of one variety typically rate the expressions used by the other as certainly not wrong but somehow odd and alien, simply not the way they would express things themselves. (Schneider 2004:â•›228)
Owing to the dynamic process of language use and language exposure, the preferences for verb + Od-collocate and verb-complementation patterns that have been shown to exist from a synchronic viewpoint drive the development of the characteristics of Indian English. Although there are no prescriptive norms for the use of these specific features, their constant use in large media, such as The Times of India or Bollywood films, serves as a yardstick for speakers of Indian English, so these media can be seen as norm-providing institutions for Indian English. Thus, as far as structural nativization in Indian English lexicogrammar on the level of collocation and verb-complementation is concerned, the corpus-based study in Chapters 5, 6 and 7 has shown that there are marked differences between Indian English and British English at both levels. Moreover, clear correlations between these levels have been proven. While conservative descriptions of varieties often claim that nativization at the structural level is minimal, these correlations in particular show that nativization at the lexical level influences other points within the lexis-grammar continuum. Therefore, although it is possible to describe variational innovation and nativization at a purely lexical level, such restrictions neglect the importance of interrelations between the lexicon and grammar. Many of these lexicogrammatical correlations are, however, only traceable if large amounts of natural language data are examined, which may be one of the reasons why such integrated approaches have so far been largely disregarded. This is especially the case because the available standard corpora for Indian English are relatively small and only available in plain-text formats. Although the use of the large amount of data that has been drawn from The Times of India has been of tremendous value for the present analysis, the
Chapter 8.╇ Evaluation and discussion 169
compilation of larger balanced corpora similar to, for instance, the BNC, would be highly desirable. This is particularly true in the light of the results for GIVE drawn from the ICE corpora, which showed that differentiation according to text type and mode is important, since differences between varieties are also detectable at this finer level, as each speech community has different conventions for different text types.
chapter 9
Conclusion and prospects for future research
9.1
Conclusion
The immediate aim of the present study was to account for structural nativization in Indian English lexicogrammar with a specific focus on collocation and verbcomplementation of ditransitive verbs. Contrary to conservative notions about variation in World Englishes, the recent study demonstrates that Indian English displays many variety-specific features at the structural level. Although many features of Indian English have been described thoroughly in the past, the majority of previous studies were only concerned with what Shastri (1996) calls “transparent features”, that is, features on the lexical or phonological level. Furthermore, early research into Indian English was often based on very limited datasets, which often leads to anecdotal results that describe peripheral “Indianisms”, rather than pervasive features of the variety. In order to show that Indian English has undergone a process of nativization that is not limited to these transparent features but is in fact all-pervading, I opted for an analysis that is based on large amounts of language data. Based on a combination of balanced standard corpora and a web-derived newspaper corpus, the present study shows how the interrelation between lexis and grammar influences nativization. In order to demonstrate these interrelations, I started out by defining the two main levels of analysis, namely collocational profiles and verbcomplementational profiles of ditransitive verbs. This step was especially necessary in the case of collocational profiles, since there is no unanimous definition of what exactly constitutes collocation. The underlying definition of collocation in the present work is a combination of quantitative and functional criteria, since all collocation candidates of the ditransitive verbs under scrutiny need to fulfil several quantitative conditions and have to be used in the position of the direct object. Thus, the collocational profiles are limited functionally to Od-collocates, i.e. collocates that function as patients or themes in cause-receive processes. Verb-complementational profiles of the verbs were based on a combination of a classification system proposed by Mukherjee & Hoffmann (2006) and the constructional framework proposed by Goldberg (1995). The combination of collocational profiles and verb-complementational profiles within an integrated
172 Structural Nativization in Indian English Lexicogrammar
functional and constructional framework has provided valuable insights for the description of the lexis-grammar interface in general and lexicogrammatical variation in Indian English in particular: – A usage-based functional model that integrates a construction grammar framework has a clear advantage over one of the models alone. – A clear correlation between the lexical and the syntactical side of the lexis-grammar continuum exists in the form of a correlation between verbÂ�complementational patterns and Od-collocates. – Structural nativization in Indian English lexicogrammar can be shown at both levels – collocation and verb-complementation – and in the interaction of those levels. – Collocational nativization and verb-complementational nativization are part of a dynamic nativizational process: the lexicogrammatical choices of speakers are influenced by the body of natural language they experience. In turn those choices become part of that body of language, which has a multiplying effect on the development of the variety (cf. Figure 8-1). In addition to these central findings, the present analysis also raised a number of questions, especially with regard to the availability of natural language data for Indian English. The limited nature of standard corpora makes it impossible to account for many linguistic features that are not highly frequent. However, language change is arguably slow, therefore innovations are relatively infrequent in the early stages of dynamic nativizational process of the specific feature. A case in point is the differences in the use of the pattern SEND Od V-ing. Based on standard corpora alone, these differences are not attestable. In the present study I opted for a combination of small balanced standard corpora and a much larger web-derived newspaper corpus. The use of the standard ICE corpora made it possible to describe central differences between the varieties and allowed for a distinction between different modes and text types. The larger web-derived corpus on the other hand provided insights into low-frequency and peripheral features. Although this corpus only contains a single text type, namely newspaper articles, this limitation is not without advantages. Since English-medium newspapers in general and The Times of India in particular are read by millions of speakers of Indian English, the above-mentioned multiplying effect on the nativization of Indian English is quite high. Additionally this paper contains only acrolectal standard English in a highly formalized genre, so differences between the varieties in this genre are less likely to be considered “learner mistakes”, but are the results of an ongoing endonormative process. This process is in turn accelerated further when speakers of Indian English look towards quality newspapers as a linguistic yardstick for standard Indian English.
9.2
Chapter 9.╇ Conclusion and prospects for future research 173
Prospects for future research
The present study entails a systematic corpus-based description of lexicogrammatical nativization in Indian English and provides a ‘thick description’ of nativization on the level of three frequent ditransitive verbs. Naturally the results of this study are by no means comprehensive, and only provide an analysis of a small area within lexicogrammatical nativization. However, the choice of the object of investigation – ditransitive verbs and their Od collocates – proved to be a good starting point for a more comprehensive description of lexicogrammatical features of Indian English. As shown above, the combination of standard and web-derived corpora allow for many observations that would be unobtainable had the research been limited to one of these datasets alone. Nevertheless, the lack of larger balanced corpora and especially the lack of syntactically parsed corpora for Indian English made it impossible to analyze a more extensive spectrum of factors that may influence structural nativization. In view of this, I would like to make a plea for the collection of larger corpora of Indian English. Since it is almost a truism that any corpus-based study will make a plea for more data, it seems even more important to emphasize that not only the amount of data but also the quality of annotation influences the feasibility of large-scale descriptions of language. In this light it will be fascinating to observe whether ongoing technological development will be a driving factor for a second “corpus revolution”. In the last two decades there has been a certain dominance of lexicological analyses within corpus linguistics, a dominance that is partly motivated by higher feasibility of those studies in the light of plain-text corpora. With probabilistic parsing becoming more and more accurate, corpus linguists will have to seek more interdisciplinary exchange with computational linguistics in order to carry corpus linguistics from its early stages towards a new level. Large-scale parsed corpora of different varieties of English will have a direct effect on the possibilities for the description of structural nativization. Several recent pilot studies, such as Cook (2008), Nam (2009) and Schilk (2009), suggest that there are a number of additional driving factors for the choice of verb-complementational patterns, which have not been the focus of the present study. The isolation of such factors is based on sets of highly annotated language data and the use of sophisticated statistic methodology. Since most of the current corpora of English are only sparsely annotated much of the work on influential factors of verb-complementation is still at its beginning. This is especially true for postcolonial varieties of English, since corpus collection and annotation for these varieties provides challenges in its own right. These challenges are often greater than they are for native varieties, since automatic annotation, such as syntactical parsing has to be based on a source-model. If this
174 Structural Nativization in Indian English Lexicogrammar
source-model is a native variety of English these procedures will not be adequate for non-native varieties. Changing this source model to a non-native model is also inherently difficult, as these models can only be formulated after describing nativized features and integrating them into a grammar. In summary the description of structural nativization of different varieties of English is highly dependent on the data basis and the different possibilities of data annotation. The development of densely annotated corpora, however, will require much interdisciplinary work. Some of the fields that may contribute to this may be variational (socio)linguistics, descriptive syntax and computational linguistics.
References
Algeo, J. 1989. Queuing and other idiosyncrasies. World Englishes 8: 157–163. Allerton, D. J. 2002. Stretched Verb Constructions in English. London: Routledge. Annamalai, E. 2004. Nativization of English and its effect on multilingualism. Journal of Language and Politics 3: 151–162. Bartsch, S. 2004. Structural and Functional Properties of Collocations in English: A Corpus Study of Lexical and Pragmatic Constraints on Lexical Co-occurrence. Tübingen: Narr. Bauer, L. 2002. An Introduction to International Varieties of English. Edinburgh: EUP. Bergs, A. 2008. Can we take construction grammar beyond sneezing napkins off tables? In Anglistentag 2007 Münster – Proceedings, K. Stierstorfer (ed.), 269–276. Trier: WissenschftÂ� licher Verlag Trier. Biber, D. & Finegan, E. 1989. Drift and evolution of English style: A history of three genres. Language 65: 478–517. Biber, D., Johansson, S., Leech, G., Conrad, S. & Finegan, E. 1999. Longman Grammar of Spoken and Written English. Harlow: Pearson Education. Bresnan, J. & Hay, J. 2008. Gradient grammar: An effect of animacy on the syntax of give in New Zeeland and American English. Lingua 118: 245–259. Butler, C. 1985. Statistics in Linguistics. Oxford: Blackwell. Church, K. W., Gale, W., Hanks, P. & Hindle, D. 1991. Using statistics in lexical analysis. In Lexical Acquisition: Exploiting On-line Resources to Build up a Lexicon, U. Zernik (ed.), 115–164. Hillsdale NJ: Lawrence Erlbaum Associates. Cook, M. 2008. Statistical Analyses of the Different Varieties of English. MA thesis, University of Warwick. Cowie A. P. 1998. Introduction. In Phraseology: Theory, Analysis and Applications, A. P. Cowie (ed.), 1–20. Oxford: OUP. Croft, W. & Cruse, D. A. 2004. Cognitive Linguistics. Cambridge: CUP. Crystal, D. 2004. The Cambridge Encyclopedia of the English Language, 2nd edn. Cambridge: CUP. Danielsson, P. 2004. Simple Perl programming for corpus work. In How to Use Corpora in Language Teaching, J. McH. Sinclair (ed.), 225–248. Amsterdam: John Benjamins. Daudaravičius, V. & Marcinkevičienė, R. 2004. Gravity counts for the boundaries of collocations. International Journal of Corpus Linguistics 9: 321–348. Dunning, T. 1993. Accurate methods for the statistics of surprise and coincidence. Computational Linguistics 19: 61–74. Evert, S. 2005. The Statistics of Word Cooccurrences: Word Pairs and Collocations. PhD dissertation, University of Stuttgart. Fillmore, C. J. 1992. Corpus linguistics vs. computer-aided armchair linguistics. In Directions in Corpus Linguistics, J. Svartvik (ed.), 35–60. Berlin: Mouton de Gruyter.
176 Structural Nativization in Indian English Lexicogrammar
Fillmore, C. J., Kay, P. & O’Connor, M. C. 1988. Regularity and idiomaticity in grammmatical constructions: The case of let alone. Language 64: 501–538. Firth, J. R. 1957. Papers in Linguistics 1934–1951. Oxford: OUP. Fisher, R. A. 1922. On the interpretation of χ² from contingency tables and the interpretation of p. Journal of the Royal Statistical Society 85: 87–94. Francis, G., Hunston, S. & Manning, E. 1996. Collins Cobuild Grammar Patterns, I: Verbs. London: Harper Collins. Friedl, J. E. F. 1997. Mastering Regular Expressions. Cambridge MA: O’Reilly and Associates. Görlach, M. 1991. Englishes: Studies in varieties of English 1984–1988 [Varieties of English around the World G9]. Amsterdam: John Benjamins. Goldberg, A. E. 1992. The inherent semantics of argument structure: The case of the English ditransitive construction. Cognitive Linguistics 3: 37–74. Goldberg, A. E. 1995. Constructions: A Construction Grammar Approach to Argument Structure. Chicago IL: The University Press of Chicago. Goldberg, A. E. 2006. Constructions at Work: The Nature of Generalizations in English. Oxford: OUP. Goldberg, A. E. & Jackendoff, R. S. 2004. The English resultative as a family of constructions. Language 80: 532–568. Greenbaum, S. 1996. Comparing English Worldwide: The International Corpus of English. Oxford: Clerendon Press. Gries, S. T. 2005. Resampling corpora. Paper presented at the workshop on Corpus Statistics: Objectives, Methods, Problems. University of Leipzig, Germany, September 29, 2005. Gries, S. T. 2006. Some proposals towards more rigorous corpus linguistics. Zeitschrift für Anglistik und Amerikanistik 54: 191–202. Halliday, M. A. K. 1966. Lexis as a linguistic level. In In Memory of J. R. Firth, C. E. Bazell, J.€C. Catford, M. A. K. Halliday & R. H. Robins (eds), 148–163. London: Longman. Hanks, P. 2008. The lexicographical legacy of John Sinclair. International Journal of Lexicography 21: 219–229. Hasselgren, A. 1994. Lexical teddy bears and advanced learners: A study into the ways Norwegian students cope with English vocabulary. International Journal of Applied Linguistics 4: 237–260. Hausmann, F. J. 1984. Wortschatzlernen ist Kollokationslernen. Praxis des neusprachlichen Unterrichts 31: 395–406. Hoffmann, S. 2007. From web-page to mega-corpus: The CNN transcripts. In Corpus Linguistics and the Web, M. Hundt, N. Nesselhauf & C. Biewer (eds), 69–85. Amsterdam: Rodopi. Hohenthal, A. 2003. English in India – Loyalty and attitudes. Language in India 3. Huddleston, R. D. & Pullum, G. K. 2002. The Cambridge Grammar of English. Cambridge: CUP. Hundt, M. 1998. New Zealand English Grammar, Fact or Fiction? A Corpus-based Study in Morphosyntactic Variation. Amsterdam: John Benjamins. Hunston, S. 2002. Corpora in Applied Linguistics. Cambridge: CUP. Joseph, B. D. 2004. On change in Language and change in language. Language 80: 381–383. Kachru, B. B. 1983. The Indianization of English. Dehli: OUP. Kachru, B. B. 1985. Standards, codification and sociolinguistic realism: The English language in the outer circle. English in the World: Teaching and Learning the Language and Literatures, R. Quirk & H. G. Widdowson (eds), 11–30. Cambridge: CUP.
References 177
Kachru, B. B. 1992. The Other Tongue: English across Cultures. Chicago IL: University of Illinois Press. Kilgarriff, A. 2005. Language is never ever ever random. Corpus Linguistics and Linguistic Theory 1: 263–276. Langacker, R. W. 1987. Foundations of Cognitive Grammar, vol. 1: Theoretical prerequisites. Stanford, CA: Stanford University Press. Langacker, R. W. 1999a. Assessing the cognitive linguistic enterprise. In Cognitive Linguistics: Foundations, Scope and Methodology, T. Jansen & G. Redeker (eds), 13–59. Berlin: Mouton de Gruyter. Langacker, R. W. 1999b. Grammar and Conceptualization. Berlin: Mouton de Gruyter. Lange, C. 2007. Focus marking in Indian English. English World-Wide 28: 89–118. Leech, G. 2007. New resources, or just better old ones? The holy grail of representativeness. In Corpus Linguistics and the Web, M. Hundt, N. Nesselhauf & C. Biewer (eds), 133–149. Amsterdam: Rodopi. Levin, B. 1993. English Verb-classes and Alternations: A Preliminary Investigation. Chicago IL: The University of Chicago Press. Lim, L. 2004. Singapore English: A Grammatical Description [Varieties of English around the World G33]. Amsterdam: John Benjamins. Louw, W. 1993. Irony in the text or insecurity in the writer? The diagnostic potential of semantic prosodies. In Text & Technology, M. Baker, G. Francis & E. Tognini-Bonelli (eds), 157–176. Amsterdam: John Benjamins. Mair, C. 2002. Three changing patterns of verb complementation in late Modern English: A realtime study based on matching text corpora. English Language and Linguistics 6: 105–131. Mair, C. 2007. Change and variation in present-day English. Integrating the analysis of closed corpora and web-based monitoring. In Corpus Linguistics and the Web, M. Hundt, N. Nesselhauf & C. Biewer (eds), 233–247. Amsterdam: Rodopi. Mak, K. T. 2005. The Dynamics of Collocation – A Corpus-based Study of the Phraseology and Pragmatics of the Introductory – it Construction. PhD Dissertation, University of Austin TX. Matthew, R. 1997. English in India: A response to Mark Tully. ELT Journal 51: 165–168. Macauley, T. B. [1835] 1965. Minute on Indian education. In Selections from Educational Records, Part I: 1781–1839, H. Sharp (ed.), 107–117. Delhi: National Archives of India. Mehrotra, R. R. 1998. Indian English – Texts and Interpretation [Varieties of English around the World T7]. Amsterdam: John Benjamins. Milroy, J. & Milroy, L. 1985. Authority in Language. London: Routledge. Moag, R. F. 1982. The life cycle of non-native Englishes: A case study. In The Other Tongue, B.€B. Kachru (ed.), 270–288. Oxford: Pergamon. Mukherjee, J. 2005. English Ditransitive Verbs: Aspects of Theory, Description and a Usage-based Model. Amsterdam: Rodopi. Mukherjee, J. 2007. Steady states in the evolution of New Englishes: The case of present day Indian English. Journal of English Linguistics 35: 157–187. Mukherjee J. 2009. The lexicogrammar of present-day Indian English. Exploring the LexisGrammar Interface [Studies in Corpus Linguistics 35], U. Römer & R. Schulze (eds), 117– 135. Amsterdam: John Benjamins. Mukherjee, J. & Hoffmann, S. 2006. Describing verb-complementational profiles of New Englishes: A pilot study of Indian English. English World Wide 27: 147–173.
178 Structural Nativization in Indian English Lexicogrammar
Mukherjee, J. & Schilk, M. 2008. Verb complementational profiles across varieties of English: Comparing verb-classes in Indian and British English. In The Dynamics of Linguistic Variation. Corpus Evidence on English Past and Present [Studies in Language Variation 2], T.€Nevalainen, I. Taavitsainen, P. Pahta & M. Korhonen (eds), 163–182. Amsterdam: John Benjamins. Mukherjee, J. & Gries, S. T. 2009. Collostructural nativization in new Englishes. Verb-construction associations in the International Corpus of English. English World-Wide 30: 27–51. Nam, C. 2009. A Statistical Analysis of Syntactic Choice in Varieties of English. MA thesis, University of Warwick. Nelson, G., Wallis, S. & Aarts, B. 2002. Exploring Natural Language: Working with the British Component of the International Corpus of English [Varieties of English around the World G29]. Amsterdam: John Benjamins. Nesselhauf, N. 2004. Collocations in a Learner Corpus [Studies in Corpus Linguistics 14]. Amsterdam: John Benjamins. Nihalani, P., Tongue, R. K. & Hosali, P. 1979. Indian and British English. A Handbook of Usage and Pronunciation. Dehli: OUP. Nihalani, P., Tongue, R. K., Hosali, P. & Crowther, J. 2004. Indian and British English. A Handbook of Usage and Pronunciation, 2nd edn. Dehli: OUP. Olavarría de Ersson, E. O. & Shaw, P. 2003. Verb complementation patterns in Indian Standard English. English World-Wide 24: 137–161. Palmer, F. R. 1966. Selected Papers of J. R. Firth 1952–1959. London: Longmans. Pinker, S. 1989. Learnability and Cognition: The Acquisition of Argument Structure. Cambridge MA: The MIT Press. Quirk, R., Greenbaum, S., Leech, G. & Svartvik, J. 1985. A Comprehensive Grammar of the English Language. London: Longman. Rahman, T. 1996. Language and Politics in Pakistan. Karachi: Oxford University Press. Rayson, P. 2003. Matrix: A Statistical Method and Software Tool for Linguistic Analysis through Corpus Comparison. PhD dissertation, Lancaster University. Renouf, A., Kehoe, A. & Mezquiriz, D. 2004. The accidental corpus: Some issues in extracting linguistic information from the web. In Advances in Corpus Linguistics. Papers from the 23rd International Conference on English Language Research on Computerized Corpora (ICAME 23) Göteborg 22–26 May 2002, K. Aijmer & B. Altenberg (eds). Amsterdam: Rodopi. Rundell, N. & Stock, P. 1992. The corpus revolution. English Today 8: 21–32. Sand, A. 2002. English in the Carribean. In Studies in English as a World Language. D. Allerton et al. (eds). 79–91. Basel: Schwabe. Sand, A. 2004. Shared morpho-syntactic features in contact varieties of English: Article use. World Englishes 23: 281–298. Schilk, M. 2006. Collocations in Indian English: A corpus-based sample-analysis. Anglia 126: 276–316. Schilk, M. 2009. Complementation of ditransitive verbs in South Asian Englishes. A multifactorial analysis. Paper presented at ICAME (30), Lancaster. Schiffmann, H. F. 2005. Bilingualism in South Asia: Friend or foe? In ISB4: Proceedings of the 4th International Symposium on Bilingualism, J. Cohen, K. T. McAlister, K. Rolstad & J.€MacSwan (eds), 2104–2114. Somerville MA: Cascadilla Press. Schneider, E. W. 2003. The dynamics of new Englishes: From identity construction to dialect birth. Language 79: 233–281.
References 179
Schneider, E. W. 2004. How to trace structural nativization: Particle verbs in world Englishes. World Englishes 23: 227–249. Schneider, E. W. 2007. Postcolonial English: Varieties around the World. Cambridge: CUP. Scott, M. 1998. WordSmith Tools, Version 4.x Oxford: OUP. Sedlatschek, A. 2009. Contemporary Indian English: Variation and Change [Varieties of English around the World G38]. Amsterdam: John Benjamins. Shastri, S. V. 1992. Opaque and transparent features of Indian English. In New Directions in English Language Corpora: Methodology, Results, Software Developments, G. Leitner (ed.), 263–275. Berlin: Mouton de Gruyter. Shastri, S. V. 1996. Using computer corpora in the description of language with special reference to complementation in Indian English. In South Asian English: Structure, Use and Users, R. J. Baumgardner (ed.), 70–81. Urbana IL: University of Illinois Press. Shaw, W. D. 1981. Asian student attitudes towards English. In English for Cross-cultural Communication, L. E. Smith (ed.), 108–122. New York: St. Martin’s Press. Sinclair, J. 1991. Corpus, Concordance, Collocation. Oxford: OUP. Standop, E. 2000. Englische Verbkomplementation. Anglia 118(2): 217–237. Stefanowitsch, A. & Gries, S. T. 2003. Collostructions: Investigating the interaction between words and constructions. International Journal of Corpus Linguistics 8: 209–243. Stefanowitsch, A. & Gries, S. T. 2004. Extending collostructional analysis: A corpus-based perspective on ‘alternations’. International Journal of Corpus-Linguistics 9: 97–129. Stefanowitsch, A. & Gries, S. T. 2005. Covarying collexemes. Corpus Linguistics & Linguistic Theory 1: 1–48. Stubbs, M. 2001. Words and Phrases: Corpus Studies of Lexical Semantics. Oxford: Blackwell. Thelwall, M. 2005. Creating and using web corpora. International Journal of Corpus Linguistics 10: 517–541. Tully, M. 1997. English: An advantage to India? ELT Journal 51: 157–168.
Index
A acting entityâ•… 166 adverbialâ•… 28, 30, 95–96 affected entityâ•… 95, 166–167 animacy restrictionâ•… 95 antecedent (NP)â•… 53, 64–65, 98–99 asymptotic hypothesis testsâ•… 21, 23–24 B basic patternâ•… 75, 97, 152 BNC news (subcorpus)â•… 139– 140, 145 British Englishâ•… 37–39, 41, 149–150, 155, 161 British National Corpus (BNC)â•… 27, 59, 61, 96, 105, 148, 150–151 C caused-motion constructionâ•… 34–35, 49–50, 52–54, central patternsâ•… 65, 97, 99, 113, 133 chi-square testâ•… 23, 66, 97, 99–100, codificationâ•… 9–10, 13–14, 47 collocationâ•… 17–19, 21–22, 24–27, 54–57, 84–85, 117, 119–123, 133–135, 155–158, 165, 167–168 collocational profileâ•… 57, 85–86, 147, 155–158 collostructionâ•… 51, 154, 161 colonial historyâ•… 5 complementâ•… 28, 114, complement clauseâ•… 114, 125 complex transitiveâ•… 29–30, 32, 34 concordâ•… 48 concordanceâ•… 83, 152
constructionâ•… 32–36, 49–50, 52–55, 95–96, 153–155, 159–165 corporaâ•… 27, 31, 37, 41–44, 48–49, 59–62, 96–98, 105– 106, 148–154, 171–174 corpus annotationâ•… 62, 150
G genreâ•… 41, 43–47, 105, 149, 172 genuine alternativeâ•… 73, 83, 110, 153 GIVEâ•… 38, 53, 57–93, 149, 151, 162–165
D databaseâ•… 41–42, 45, 112, 149 dativeâ•… 50, 158–159, 162–164, 167 dative-alternationâ•… 50, 63 default patternâ•… 100, 104, 131, 145 derivative patternâ•… 67, 107–109 dialogueâ•… 43, 69, 72, 75, 78, 102, 106, 109 differentiation (cf. stage 5)â•… 9–10 direct objectâ•… 28–29, 38, 62–63, 155, 167, ditransitiveâ•… 29–30, 32, 34–35, 37–39, 42, 49–54, 158–167 ditransitive constructionâ•… 34– 36, 39, 95–96, 160–162, 164 ditransitive situation schemaâ•… 165–166
H habitual ditransitive verbâ•… 129, 165
E encodingâ•… 50, 147, 165–167 endonormative stabilization (cf. stage 4)â•… 9–12, 14–15, 162 Englishesâ•… 2, 7, 10, 15, 161, 171 exonormative stabilization (cf. stage 2)â•… 9–10, 15–16 F fisher exact testâ•… 22–24 foundation (cf. stage 1)â•… 9–10 fronted elementâ•… 63
I Indian Englishâ•… 1–3, 5, 7, 11–15, 26, 36–39, 165–168, 171–173 indigenous (IND) strandâ•… 12 indirect objectâ•… 28–30, 38, 50, 52, 167 International Corpus of English (ICE)â•… 38, 41, 43 intransitiveâ•… 29–32, 51 K keywordâ•… 48 Kolhapur corpusâ•… 41 L Lancaster Oslo Bergen (LOB) corpusâ•… 41, 105, 150 lemmaâ•… 57, 59–60 lexicogrammarâ•… 2, 17, 27, 44, 168, 171–172 lexiconâ•… 18–19, 33, 37, 164 lexisâ•… 17–19, 32–34, 171–172 log-likelihoodâ•… 23–24, 56–57, 156–157 Longman grammarâ•… 28, 31, 100 M MI-scoreâ•… 21 monologueâ•… 69, 102–103
182 Structural Nativization in Indian English Lexicogrammar
monotransitiveâ•… 29–30, 32, 51–52, 98, 161, 167 multilingualâ•… 6–7, 13–14 N namazâ•… 134, 136–138, 151, 165 nativization (cf. stage 3)â•… 1–3, 9–12, 14–16, 146–147, 153, 161–166, 171–174 newspaper corpusâ•… 46, 148, 171–172 node wordâ•… 55–56, 155 normâ•… 10, 14, 168 noun phraseâ•… 28, 30 O oblique elementâ•… 35, 154 Od-collocateâ•… 54, 57, 62, 157–159, 165, 168 OFFERâ•… 57–62, 129–146, 165 P Pakistanâ•… 7, 12–14 parsed corporaâ•… 159, 173 parsingâ•… 51–52, 173 particle verbâ•… 30, 115–116 passive constructionâ•… 29, 38 Perlâ•… 45–46, 151–152 phrasal verbâ•… 18, 116–117 phraseological approach (to collocation)â•… 19, 24–26 pragmatic synonymyâ•… 35 prepositional objectâ•… 30, 35 prepositional phraseâ•… 29–30, 35, 37, 50–52, 95–96 probabilistic parsingâ•… 173 p-valueâ•… 21, 23–24, 51, 56 Q quantitative approach (to collocation)â•… 19–24, 57 R recipientâ•… 34–35, 50–52, 154–155, 163–165, 167 registerâ•… 32, 71
S semantic prosodyâ•… 114, 130, 141 semantic synonymyâ•… 35–36 SENDâ•… 57–62, 95–128, 149, 151, 153, 172 settler (STD) strandâ•… 9, 12–13, 16 sociolinguisticsâ•… 10–11, 13–14 spanâ•… 20, 55–56, 84, 155–156 spoken corporaâ•… 44 spoken modeâ•… 59–60, 78–80, 92, 114, 130, 132, 145 stabilization (endonormative)â•… 9–12, 14–15, 162 stabilization (exonormative)â•… 9–10, 15–16 stage 1 (foundation)â•… 9 stage 2 (exonormative stabilization)â•… 9 stage 3 (nativization)â•… 9, 11–12, 14–15 stage 4 (endonormative stabilization)â•… 9, 11–14 stage 5 (differentiation)â•… 9 Statesman (newspaper, web-archive)â•… 39 statistical significanceâ•… 20–24, 157 structurally related patternsâ•… 29, 38, 63–64, 152 structural nativizationâ•… 1–3, 5, 10, 14–15, 19, 166–168, 171–174 subcontinentâ•… 2, 5–7, 9, 11, 13, 47 subcorpusâ•… 42 T text-categories (ICE-corpora)â•… 43, 153 Times of India (ToI) corpusâ•… 44–48, 62, 150, 152 Times of India (ToI) newspaperâ•… 44–47, 167–168, 172 to-prepositional phraseâ•… 29–30, 35, 95–96 transfer-caused-motion constructionâ•… 34–35, 42, 50, 155, 160
transferred entityâ•… 166–167 transitivityâ•… 30–32 trivalent verbâ•… 55–56, 84, 155, 167 t-scoreâ•… 21–22, 24 type-Ider patternâ•… 64–69 type-IIder patternâ•… 54, 64–69 type-IIIder patternâ•… 64–69 type-III patternâ•… 38, 64–69, 160, 164–165, 167 type-IIIPder patternâ•… 64–69 type-IIIP patternâ•… 64–69 type-II patternâ•… 38, 53, 64–69, 153, 155, 167 type-IIPder patternâ•… 64–69 type-IIP patternâ•… 64–69 type-I patternâ•… 38–39, 53–54, 64–69, 153, 155, 160 type-IPder patternâ•… 64–69 type-IP patternâ•… 64–69 type-IV patternâ•… 38, 54, 64–69 type-V patternâ•… 38, 52, 54, 64–69 V valencyâ•… 31 variationâ•… 66–67, 98, 161–163, 165, 171–172 varietyâ•… 1–2, 5, 7–16, 19, 26–27, 161–165, 167–168, 171–172, 174 verb-complementationâ•… 2, 17–19, 26–28, 32, 36, 38–39, 52–54, 156, 158–160, 162–163, 168, 171–173 verb-complementational profileâ•… 49–54 verb phraseâ•… 19, 28, 156 W Webcrawlerâ•… 45 wordlistâ•… 48, 62 WordSmith Toolsâ•… 19, 23, 42, 46, 48 World Englishesâ•… 2, 7, 15, 37, 171 written modeâ•… 61, 75–80, 149