Misunderstandings in English as a Lingua Franca: An Analysis of ELF Interactions in South-East Asia 9783110288599, 9783110286519

Citation preview

Misunderstandings in English as a Lingua Franca

Developments in English as a Lingua Franca 1

Editors

Jennifer Jenkins Will Baker

De Gruyter Mouton

Misunderstandings in English as a Lingua Franca An Analysis of ELF Interactions in South-East Asia

By

David Deterding

De Gruyter Mouton

ISBN 978-3-11-028651-9 e-ISBN 978-3-11-028859-9 ISSN 2192-8177 Library of Congress Cataloging-in-Publication Data A CIP catalog record for this book has been applied for at the Library of Congress. Bibliographic information published by the Deutsche Nationalbibliothek The Deutsche Nationalbibliothek lists this publication in the Deutsche Nationalbibliografie; detailed bibliographic data are available in the Internet at http://dnb.dnb.de. ” 2013 Walter de Gruyter, Inc., Boston/Berlin Printing: Hubert & Co. GmbH & Co. KG, Göttingen 앝 Printed on acid-free paper 앪 Printed in Germany www.degruyter.com

Table of contents List of abbreviations .................................................................................... ix Keywords for Vowels .................................................................................. ix Transcription conventions............................................................................. x Chapter 1 Introduction ............................................................................. 1 1.1. Scope of the investigation ..................................................................... 2 1.2. Overview of the book ........................................................................... 3 Chapter 2 ELF, intelligibility and misunderstandings ........................... 4 2.1. ELF and World Englishes ..................................................................... 4 2.2. The Lingua Franca Core (LFC) ............................................................ 7 2.3. Intelligibility ......................................................................................... 9 2.4. Misunderstandings .............................................................................. 12 2.5. Repairs ................................................................................................ 14 2.6. Accommodation .................................................................................. 16 2.7. Conversation Analysis ........................................................................ 17 2.8. Conclusion .......................................................................................... 19 Chapter 3 Data and methodology............................................................ 20 3.1. CMACE .............................................................................................. 20 3.1.1. Speakers ................................................................................... 21 3.1.2. Recordings ............................................................................... 23 3.1.3. Transcription ............................................................................ 24 3.2. Analysis .............................................................................................. 27 3.2.1. Instances and tokens ................................................................ 27 3.2.2. Classification ........................................................................... 30

vi

Table of contents

Chapter 4 Pronunciation ......................................................................... 33 4.1. Consonants .......................................................................................... 34 4.1.1. TH ............................................................................................ 34 4.1.2. Initial [h] .................................................................................. 40 4.1.3. [n],[l], [r] and [w] ..................................................................... 41 4.1.4. Initial consonant cluster simplification .................................... 44 4.1.5. Final consonant cluster simplification ..................................... 47 4.1.6. Dark-L...................................................................................... 51 4.1.7. Aspiration and voicing ............................................................. 55 4.1.8. Fricatives and affricates ........................................................... 58 4.1.9. Other missing consonants ........................................................ 61 4.1.10. Consonants: summary ............................................................ 63 4.2. Vowels ................................................................................................ 63 4.2.1. NURSE....................................................................................... 64 4.2.2. TRAP ......................................................................................... 65 4.2.3. FACE ......................................................................................... 68 4.2.4. Vowel length ............................................................................ 68 4.2.5. Vowel reduction....................................................................... 70 4.2.6. Miscellaneous vowels .............................................................. 71 4.2.7. Vowels: summary .................................................................... 73 4.3. Spelling pronunciation ........................................................................ 73 4.4. Stress ................................................................................................... 74 4.4.1. Word stress .............................................................................. 74 4.4.2. Utterance stress ........................................................................ 77 4.5. Rhythm, intonation and speaking rate ................................................ 79 4.5.1. Rhythm .................................................................................... 79 4.5.2. Intonation ................................................................................. 80 4.5.3. Speaking rate ........................................................................... 81 4.5.4. Syllables ................................................................................... 83

Table of contents

vii

4.5.5. Laughter ................................................................................... 85 4.6. Phonetic accommodation .................................................................... 85 4.6.1. Listening accommodation ........................................................ 86 4.6.2. Speaking accommodation ........................................................ 87 4.7. Summary of pronunciation ................................................................. 87 Chapter 5 Lexis, grammar and code-switching ..................................... 92 5.1. Lexis ................................................................................................... 92 5.1.1. Words....................................................................................... 92 5.1.2. Phrases ..................................................................................... 94 5.1.3. Idioms ...................................................................................... 98 5.1.4. Shifted meaning ..................................................................... 100 5.1.5. Polysemes .............................................................................. 105 5.2. Grammar ........................................................................................... 107 5.2.1. Plural nouns ........................................................................... 108 5.2.2. Articles ................................................................................... 111 5.2.3. Verb suffixes .......................................................................... 114 5.2.4 Missing verbs ......................................................................... 117 5.2.5. Prepositions............................................................................ 119 5.2.6. Word order ............................................................................. 120 5.2.7. Other grammar issues ............................................................ 121 5.3. Code-switching ................................................................................. 123 5.4. Miscellaneous ................................................................................... 128 5.5. Summary of lexis, grammar and code-switching ............................. 130 Chapter 6 Repairs .................................................................................. 131 6.1. Self-initiated repairs .......................................................................... 131 6.1.1. Correcting oneself .................................................................. 131 6.1.2. Unprompted paraphrase ......................................................... 132 6.1.3. Asking for help ...................................................................... 136

viii

Table of contents

6.2. Responses to misunderstandings ...................................................... 138 6.2.1. Asking for clarification .......................................................... 139 6.2.2. Correcting .............................................................................. 144 6.2.3. Silence.................................................................................... 145 6.2.4. Backchannels ......................................................................... 147 6.2.5. Selecting part of the utterance ............................................... 150 6.2.6. Changing the topic ................................................................. 152 6.2.7. Laughter ................................................................................. 154 6.2.8. Non-awareness ....................................................................... 156 6.3. Avoiding misunderstandings ............................................................ 157 6.3.1. Topic fronting ........................................................................ 157 6.3.2. Lexical repetition ................................................................... 159 6.3.3. Echoing .................................................................................. 162 6.3.4. Collaborative completions ..................................................... 164 6.4. Summary of repairs........................................................................... 167 Chapter 7 Implications for pedagogy................................................... 168 7.1. Implications of the findings ............................................................ 1699 7.2. Pronunciation teaching ..................................................................... 172 7.3. Materials for ELF-based teaching..................................................... 174 7.4. Teaching accommodation ................................................................. 175 7.5. Testing ............................................................................................ 1777 7.6. Prognosis for the future..................................................................... 177 Appendix The tokens of misunderstanding in CMACE ..................... 179 References................................................................................................. 191 Author Index ............................................................................................. 202 Subject Index ............................................................................................ 204

List of abbreviations

ACE CA CAA CMACE COCA ELF FASS ICE LFC RP UBD VOT

Asian Corpus of English Conversation Analysis Civil Aviation Authority Corpus of Misunderstandings from ACE Corpus of Contemporary American English (available from http://corpus.byu.edu/coca/) English as a Lingua Franca Faculty of Arts and Social Sciences International Corpus of English Lingua Franca Core Received Pronunciation University of Brunei Darussalam Voice Onset Time

Keywords for Vowels The following keywords are used for referring to various vowels. Not all the keywords suggested by Wells (1982) are listed, as only some are referred to in this book. The RP British English pronunciation of each keyword is shown after it. FLEECE KIT DRESS TRAP

L‫ޝ‬ /ܼ/ /e/ /æ/

NURSE STRUT BATH LOT

/‫ޝܮ‬ /‫ݞ‬/ /‫ޝܤ‬ /‫ܥ‬/

FACE GOAT PRICE

/eܼ/ /ԥ‫ݜ‬/ /aܼ/

In addition: TH dark-L

represents the sound at the start of words such as thin and this represents the [l] in a syllable coda, such as in call and world

Transcription conventions The transcription conventions used in this book are adopted from the VOICE project (VOICE 2011). (.) (2) hh @@@ words

word: word:: (word) WORD . ? words word u k xxx ... [MIn:name] FTw, MLs

short pause (less than 0.5 seconds) pause of about 2 seconds audible intake of air ODXJKWHUHDFKµ#¶LQGLFDWHVRQHV\OODEOH words spoken accompanied by laughter non-linguistic sound word that is extended word that is extended considerably uncertain transcription word (or syllable) said with extra emphasis falling intonation rising intonation overlapping speech word that is misunderstood word that is spelt out XQFOHDUVSHHFKHDFKµ[¶ indicates one syllable indicates some omitted material the name of a participant; names are removed to protect anonymity refers to a female (F) or male (M) participant, with the country shown after the initial F or M, as follows: Br : Brunei In : Indonesia Ma : Malaysia

Ch+Br : 55

Ch : China Hk : Hong Kong Jp : Japan Ls : Laos Ng : Nigeria Tw : Taiwan

refers to a recording involving the two people indicated by the initials; the number shows the time in seconds from the start of the recording

Chapter 1 Introduction

This book is about misunderstandings in English as a Lingua Franca (ELF), so it is best if my use of these terms is immediately explained. First, misunderstandings. In this book I will use the term to include all cases where a listener does not understand something that is said to them. Note that this does not necessarily involve a breakdown in communication, as the interaction often proceeds quite smoothly even when a few words are not understood. Nevertheless, in considering what contributes to intelligibility, it is informative to analyse the words that are not understood even in cases where the listener can in fact grasp the overall gist of the utterance and so the conversation appears to continue without a problem. The methodology of identifying such instances of misunderstanding will be discussed in Chapter 3. Next, ELF. Seidlhofer (2011: 7) defines it DV³any use of English among speakers of different first languages for whom English is the communicative medium of choice, and often WKHRQO\RSWLRQ´. Note that this is a fairly broad definition, as it includes native speakers so long as their interactional partners are not native speakers. An alternative is that by House (1999: 74) who deILQHV(/)LQWHUDFWLRQVDV³between members of two or more different linguacultures in English, for none of whom English is the mother WRQJXH´ This alternative definition excludes native speakers. Actually, the concepts of mother tongue and native speaker are not easy to define (McKay 2002: 28; Davies 2003). For example, there are some people in Brunei who grow up speaking English at home and it may be their best language although they also regularly speak Malay. So should they be classified as native speakers or not? In describing my data, I will make reference to the Three Circles model of English proposed by Braj Kachru (1985). The conversations analysed in this book do not include speakers from Inner-Circle countries such as Britain, the USA or Australia; but I will not worry about whether the participants are native speakers or not. So the term ELF as it is used in this book can be understood as follows: it is the English used by speakers from postcolonial Outer-Circle countries such as Brunei, Malaysia and Nigeria as well as Expanding-Circle places such as Taiwan, Japan and Laos when they are conversing with speakers from other countries in the Outer or Expanding Circles.

2

1. Introduction

I should emphasize that this indicates the scope of the data analysed in this book but it does not attempt to provide a definition of ELF. Other researchers will adopt alternative, perfectly valid definitions of ELF that are appropriate for their work, and in many cases they will include speakers from the Inner Circle to a certain extent. Furthermore, the focus on speakers coming from different countries is not one shared by all researchers into ELF. It just characterizes the data on which my research is based. This book, then, is about the factors that cause misunderstandings to occur in spoken interactions in English between people from different countries in the Outer and Expanding Circles. In addition, it considers how those misunderstandings are dealt with and how misunderstandings in ELF conversations are avoided.

1.1. Scope of the investigation Most people would probably agree that intelligibility is of primary importance in ELF interactions: if you cannot be understood, then your language use is not successful. However, intelligibility is actually a complex, multifaceted concept (Nelson 2011), something I will consider in greater depth in Chapter 2. The research in this book builds on the seminal work of Jenkins (2000) in analysing what enhances and what hinders intelligibility in ELF interactions, though I will extend the investigation beyond her main focus on pronunciation to consider grammatical and lexical issues, and also to discuss how misunderstandings are dealt with and avoided. Much research on misunderstandings in ELF tends to investigate breakdowns in communication: it analyses data to find out what causes a breakdown in communication to occur and how it is repaired. This book is rather different. Most of the participants themselves were subsequently involved in the transcription or else they contributed to the analysis by providing invaluable feedback about what they had said and also what they had not understood. This has allowed me to find numerous instances where a participant does not understand something but where the conversation continues with no apparent problem, and I only know about the misunderstanding as a result of the subsequent feedback from the participants. This provides a rich source of data on intelligibility. Even though the majority of the instances that I will analyse involve no overt breakdown in communication, nevertheless it is important to consider which features of speech have an impact on intelligibility, and what it is about the pronunciation, lexis,

1.1. Scope of the investigation

3

grammar, and general patterns of usage that have caused certain words or phrases not to be understood by the listeners. At this point, it is important to emphasize that misunderstandings are very much the exception rather than the rule in my data, and the conversations that constitute my corpus generally proceed smoothly. I will therefore endeavour not just to consider features of speech that cause misunderstandings to occur but also those that serve to enhance intelligibility. Examples of non-standard features (in terms of Inner-Circle Englishes) that might be suggested as improving intelligibility in an ELF setting are: the occurrence of a full vowel rather than a reduced vowel like [ԥ] in the first syllable of a word such as computer or consider and in function words such as of and at; the use of a plural suffix on logically countable words such as furnitures and equipments; and prominent topic fronting often accompanied by a resumptive pronoun, as in my sister, she lives in Singapore.

1.2. Overview of the book In Chapter 2, I will consider the background to this study, including the concepts of intelligibility, misunderstanding and repair, and I will also provide a brief overview of Conversation Analysis. Chapter 3 describes the Corpus of Misunderstandings from the Asian Corpus of English (CMACE), including the speakers, the recordings, and the methods of identifying and analysing misunderstandings. In Chapter 4, the role of pronunciation is investigated, and then in Chapter 5, other features that sometimes cause misunderstandings are discussed, particularly grammar and lexis. In Chapter 5, I will additionally consider code-switching. Then in Chapter 6, I analyse how misunderstandings are dealt with, occasionally by the interactants asking for clarification though more usually by them adopting a µOHWLWSDVV¶ strategy (Firth 1996) in the hope that things will sort themselves out naturally. Chapter 7 deals with the pedagogical implications of my research and offers a few conclusions. Finally, a full listing of all the tokens of misunderstanding from the CMACE corpus is provided in the Appendix.

Chapter 2 ELF, intelligibility and misunderstandings

The patterns of usage found in English as a Lingua Franca (ELF) have been treated as a serious topic of investigation particularly since the publication of Seidlhofer (2001), and the concept of a limited set of pronunciation features for ensuring intelligibility in ELF interactions, the Lingua Franca Core (LFC), has been the focus of substantial discussion since the publication of Jenkins (2000). Nevertheless, the status of ELF and especially the proposed LFC continue to be controversial, and many writers have criticized them in derLVLYHWHUPVUHPLQLVFHQWRIWKH³KDOI-EDNHGTXDFNHU\´ used by Quirk (1990) to dismiss the emergence of varieties of World English as models for teaching English around the world. This chapter reviews the status of ELF, in particular how it differs from World Englishes (Kachru 2005: 15). Then there is an overview of the LFC and a discussion of some of the hostility that has been targeted at it. Next I will consider the nature of intelligibility, including the distinction between intelligibility, comprehensibility and interpretability proposed by Smith (1992), and I will offer a brief overview of some research on the relative intelligibility of different varieties of English around the world. Closely linked to intelligibility is the concept of misunderstandings, and their nature will be discussed before I consider repairs and the typical ways that misunderstandings are dealt with and avoided. One way of avoiding misunderVWDQGLQJVLVE\PHDQVRIDFFRPPRGDWLRQWKHDGDSWDWLRQRIRQH¶VOLVWHQLQJ or speaking WR WKH QHHGV RI RQH¶V FRQYHUVDWLRQDO SDUWQHUs, so some basic concepts in accommodation will be discussed. Finally, I will offer a brief overview of Conversation Analysis (CA), the discipline that often provides the basis for research on misunderstandings and repairs, and I will consider the ways in which research into ELF interactions may differ from analysis of the patterns of native-speaker conversations that are usually investigated in CA.

2.1. ELF and World Englishes The term English as a Lingua Franca (ELF) has been in use at least since the mid-1990s (Jenkins 1996). However, it has only been widely adopted as a formal term to describe English as it occurs in international settings since

2.1. ELF and World Englishes

5

2000, partly as a result of Barbara Seidlhofer¶V SDSHU2QHPD\QRWH that although one of the most prominent figures in this field of study, Jennifer Jenkins, discussed the term ELF in her 2000 book (Jenkins 2000: 11), she chose at the time to use the alternative term English as an International Language (EIL) and only adopted the term ELF more widely in later publications (e.g. Jenkins 2005). However, research on ELF is now firmly established, with its own journal (Journal of English as a Lingua Franca), a dedicated series of conferences (the fifth ELF conference was held in Istanbul in May 2012), and a burgeoning array of monographs (e.g. Jenkins 2007; Smit 2010; Kirkpatrick 2010; Seidlhofer 2011; Cogo and Dewey 2012) as well as edited volumes (e.g. Mauranen and Ranta 2009; Archibald, Cogo, and Jenkins 2011). One problem with the term ELF is that the concept of a lingua franca often has negative connotations (Seidlhofer 2011: 80), referring to an impoverished code that has limited domains of use and is merely adopted as a last-resort means of communication between speakers with no other means of talking to each other. In fact, recent work on ELF shows that it can be immensely sophisticated, characterized by highly resourceful patterns of interaction. It is therefore inappropriate to regard it as an impoverished code, and furthermore it is of considerable interest to investigate the innovative ways that ELF speakers ensure that they can interact successfully. In this respect, I will now briefly discuss one feature of language usage: the occurrence of idioms. While one might expect ELF speakers to avoid the most obscure English idioms as well as colourful but opaque sayings transferred from their respective first languages, one actually finds that they often incorporate quite a lot of idioms into their English and they even sometimes develop fresh ones during their interactions. In fact, it has been suggested that ELF speakers may be particularly adept at this innovative extension of language usage. For example, Seidlhofer and Widdowson (2009) show how participants at an international conference created and then adopted the idiom endangered fields (by deliberate analogy with endangered species) to refer to areas of study that were perceived to be under threat. So, far from being an impoverished code, we generally find that ELF is rather creative, partly because it adopts expressions and patterns of usage from a wide range of different backgrounds, and also because its interactants frequently do not feel constrained by native-speaker normative rules. Indeed, it has been suggested (e.g. Cogo and Dewey 2012) that ELF interactants tend to pay little attention to standard grammar as they focus much more on the most effective ways of getting their message across, and the patterns of English that they adopt can be highly sophisticated. One of

6

2. ELF, misunderstandings and intelligibility

the goals of the research in this book is to investigate the extent to which the use of non-standard grammar as well as idioms might result in misunderstandings among ELF interactants from different backgrounds. It is important to distinguish the study of ELF from that of World Englishes, as the two reflect quite different perspectives. The World Englishes paradigm investigates new varieties of English as independent, named, regional varieties, such as Singapore English, Indian English and Nigerian English, and it generally focuses on features of pronunciation, lexis, grammar and discourse that make each variety distinct from the others (Kirkpatrick 2007; Jenkins 2009). One of the key aspects of studies into World Englishes is how they emerge with their own independent identity, with endonormative standards of pronunciation, lexis and grammar that are not linked to the standards found in traditional Englishes from places such as the UK and the USA (Schneider 2007). In contrast, research into ELF typically considers how people from different countries interact. In other words, while studies of World Englishes generally focus on the distinct features of national varieties of English, research on ELF usually considers common patterns that are shared by speakers from disparate backgrounds. Nevertheless, it is fundamentally incorrect to suggest that research on ELF is proposing the emergence of a single variety of English. Indeed, even though it seeks to investigate some of the shared patterns by which people from different backgrounds communicate, it always acknowledges and indeed celebrates the fact that there continues to be wide variation in the ways that English is used around the world. A model that has been highly influential in the study of World Englishes is that of the Three Circles of English (Kachru 1985, 2005), in which the traditional varieties of English such as those of the UK, the USA and Australia are classified as being in the Inner Circle, postcolonial varieties such as those of Singapore, India and Nigeria are in the Outer Circle, and varieties in places with no colonial background and where English is taught as a foreign language such as Japan, Germany and Brazil are in the Expanding Circle. While there are some issues with this model because it is geographically and historically based (Jenkins 2009: 20±21) and it fails to reflect some of the dynamic ways that English is nowadays being used around the world (Cogo and Dewey 2012: 9), it still offers a constructive way of conceptualizing some of the different kinds of English that exist. In this book, I will make reference to the Three Circles, specifying, for example, that all the speakers in the recordings on which my research is based are from the Outer and Expanding Circles.

2.1. ELF and World Englishes

7

Despite its wide acceptance nowadays, the study of ELF still encounters substantial opposition, particularly from those who believe that we should continue to focus on native-speaker norms and patterns of English usage. Such opposition, even entrenched hostility, is especially targeted against the LFC, so I will discuss that in the next section.

2.2. The Lingua Franca Core (LFC) The Lingua Franca Core (LFC) represents a finite set of pronunciation features which, it is claimed, are necessary for achieving international intelligibility in spoken English. It is further suggested that non-core features which occur in native-speaker pronunciation are not needed for maintaining intelligibility, and indeed some features of pronunciation (such as the use of reduced vowels in the weak forms of function words) might actually interfere with international intelligibility. As a result, it is proposed that it is not necessary to teach non-core features, though some learners may decide that they want to approximate to native-speaker patterns of speech, and this choice should of course be respected. The features of the LFC were set out in Jenkins (2000), and were then summarized in various subsequent papers and books (e.g. Jenkins 2005: 201, 2006: 37, 2007: 23±24). The following are considered core features:      

all the consonants of native-speaker English except >ș@DQG>è@ aspiration on initial voiceless plosives initial and medial consonant clusters vowel length distinctions the quality of the NURSE vowel the placement of the intonational nucleus

In contrast, the following features are outside the LFC, so there should be flexibility in how they are realized:        

the TH sounds final consonant clusters dark-L vowel quality (except for the NURSE vowel) vowel reduction (especially in the weak forms of function words) rhythm word stress the shape of the intonational contour (rising, falling, etc.)

8

2. ELF, misunderstandings and intelligibility

While many people would accept that some features of pronunciation are more important than others, and furthermore they would agree that there is no need for learners always to mimic speakers from the UK or USA too closely, the details of what should be included and what might be excluded from the LFC remain contentious. For example, many teachers assert that word stress is vitally important; and furthermore excluding the quality of vowels from the LFC would be alarming to many teachers and learners. In contrast, not all teachers would agree that vowel length distinctions and aspiration on initial voiceless plosives are essential for a speaker to be intelligible in international settings. It is one of the main goals of the current study to provide further data that allows us to extend the evaluation about which features of pronunciation should be included in the LFC and which might be excluded. The LFC proposals were derived on the basis of just 40 tokens of misunderstanding between speakers from Japan and Switzerland who were engaged in comparing different versions of a map with each other (Jenkins 2000: 85). Only 27 of these tokens involved phonology, so there is clearly a need to extend the research, to analyse more instances of misunderstanding from a wide range of speakers in different conversational settings, and a few other studies have done that. Matsumoto (2011) basically concurs with the findings of Jenkins, though the suggestion that the final [t] in present is a core feature (p. 102) is a little surprising given that it is part of a final consonant cluster which is usually considered non-core. Osimk (2011) shows that the voiceless TH sound at the start of things is understood better ZKHQ LW LV SURQRXQFHG DV >W@ WKDQ ZKHQ LW LV >ș@ IRU OLVWHQHUV IURP 6SDLQ, Italy, France and Germany, though the performance when it is pronounced as [s] is not so good; and she also confirms that initial voiceless plosives are recognized best with substantial aspiration. However, McCrocklin (2012), challenges the LFC proposals when she asserts that word stress is important for intelligibility, based mainly on evidence from studies of listening by native speakers, though she provides no new data from ELF settings to support this claim. Unfortunately, there have not been many other attempts to replicate Jenkins¶RULJLQDOILQGLQJV-HQNLQV, Cogo, and Dewey 2011: 288). Varying attitudes have been reported among international students towards different features of the LFC. Fowler (2010) reports that, while the majority of international students EHOLHYHWKDWXVHRI>ș@DQG>è@IRUWKH7+ sounds is important, most of them do not feel that use of weak forms is helpful. There is some logic to this: pronunciation of voiceless TH as [t] LQVWHDGRI>ș@Fan be regarded as a loss of information, as it results in tin

2.2. The Lingua Franca Core (LFC)

9

and thin not being distinguished; but the avoidance of weak forms retains useful distinguishing information, so for example have and of will always be differentiated (while they may both be said as [ԥv] by speakers in the Inner Circle). In fact, the attitudes of ELF speakers towards the TH sounds may depend on how they are realized. Shaw, Caudery, and Petersen (2009: 195) found that, over a period of several months, exchange students in universities in Sweden and Denmark stopped using [s] and [z] for the TH sounds, but the use of [t] and [d] remained stable. This finding supports the suggestion of Osimk (2011) that the latter pronunciation is more acceptable in ELF settings. Although it seems that some international students may welcome the LFC proposals, many English language teachers have less positive attitudes. For example, Scheuer (2005) asserts that LFC-based teaching is harmful for students, and Sobkowiak (2005) insists that it is not suitable for adoption in Poland. Jenkins (2005) contends that this opposition arises because of misconceptions about what the LFC represents, and Kirkpatrick (2007) argues that ELF-based teaching should be empowering and liberating for nonnative English teachers. But perhaps we should not be too surprised if teachers are reluctant to adopt quite radical proposals that seem to challenge so much that they believe in. There is, therefore, a need to consider in depth the extent to which LFCbased teaching might interfere with intelligibility, or whether conversely it might actually enhance the ease with which speakers can make themselves understood in international settings.

2.3. Intelligibility Smith (1992) makes a helpful distinction between three different kinds of understanding:   

intelligibility: recognition of words and utterances comprehensibility: understanding the meaning of words and utterances interpretability: understanding the meaning behind words and utterances

In other words, intelligibility refers to our ability to identify the words in an utterance, comprehensibility is about whether we know what the utterance means, and interpretability is concerned with the pragmatic implication of an utterance.

10

2. ELF, misunderstandings and intelligibility

While these three concepts usefully remind us that intelligibility is a multifaceted concept, so it is not just about correctly identifying words, in fact applying the terms when dealing with real data can be problematic. For example, Nelson (2011) discusses the three-way distinction in some detail, and he suggests (p. 63) that the following quote from a novel by Chinua Achebe raises issues for interpretability: I want one of my sons to join these people and be my eyes PHDQLQJµ,DPVHQGLQJ\RXDVP\UHSUHVHQWDWLYH¶  But is it really true that this would be hard for people from other cultures to understand? Although some people around the world might find it an odd way of saying something, there does not seem to be much difficulty in comprehending what it means. In contrast, Nelson claims (2011: 108) that when his Australian-born sister-in-law noticed some yoghurt was spoiled and said that it was off, this was not an issue for intelligibility, but rather a ³liveO\ DQG LQWHUHVWLQJ´ way of expressing herself. But on what basis can we determine that saying some food is off is understandable to people from other backgrounds while asking someone to join these people and be my eyes causes problems for interpretability? This illustrates that identifying the interpretability of utterances is tough, as it is difficult to know the extent to which people really understand the implications of everything that others are saying. Pickering (2006) similarly notes that the concept of interpretability is hard to measure. For this reason, in my research on misunderstandings, I will be focusing mostly on intelligibility at the word and phrase level: I classify something as an instance of misunderstanding if there are some key words that the listener cannot identify or does not understand, even if at the wider level they may be able to follow the gist of the utterance quite successfully. And the two principle questions I investigate are: which phonological, lexical and grammatical factors have an influence on intelligibility? And how are misunderstandings dealt with and avoided? Munro, Derwing, and Morton (2006) also make a useful but different three-way distinction between intelligibility, comprehensibility, and accentedness. Intelligibility involves the recognition of words and sentences, so this is similar to the way the term is used by Smith (1992). However, the other two terms are different: comprehensibility is concerned with the ease with which listeners understand an utterance; and accentedness is the degree to which the pronunciation of an utterance deviates from a norm. The research of Munro et al. is based in Canada, and their classification only really makes sense in an Inner-Circle setting where a norm is reasonably well specified, as it is not clear what accentedness would mean in many Outer-Circle contexts, a point that Munro (2008: 193) acknowledges when

2.3. Intelligibility 11 he notes that the distinction between a foreign and a native accent is blurred in the context of nativized varieties of English. For example, if one asked listeners in Singapore to estimate the accentedness of an utterance, some of them would judge RP British pronunciation to be more accented than Singapore speech, while others would make the opposite judgment. Therefore, while Munro et al. make some insightful observations about the multifaceted reactions to accented speech, for example showing that familiarity with a variety of English does not always enhance the intelligibility of utterances in that variety, I will not adopt their classification here. It is of course important to recognize that intelligibility is not just about whether something is understood or not, and Munro and his colleagues conduct valuable research into the ease with which listeners understand various kinds of speech. However, assessing the comprehensibility of conversational data remains a problem. While it is not too difficult to determine whether listeners understand an utterance, for example by asking them to transcribe the words that they hear, it is rather harder to measure the ease with which interactants in a conversation understand the words, and any such evaluation is inevitably rather subjective. Some interesting work in this respect has been done by Björkman (2009), who uses questionnaires to determine how irritating certain non-standard features of speech are judged to be by ELF listeners in a Swedish university, and she finds (p. 242) that disrupted word order is the feature which is most often reported to be irritating, while tense usage and non-marking of plural nouns are among the features that are judged to be the least irritating. Quite apart from the classification of the different aspects of intelligibility, there seems to be widespread agreement that English spoken by people from the Inner Circle is not necessarily more intelligible than that produced by people in the Outer and Expanding Circles. For example, Smith and Rafiqzad (1979) report that the speech of someone from the USA was found to be less intelligible than that of someone from Malaysia, and Smith and Bisazza (1982) found the same when comparing an American speaker with someone from India. Furthermore, there is plenty of anecdotal evidence that many ELF speakers have problems understanding people from the Inner Circle. House (2003: 567) reports that few misunderstandings emerge in ELF discourse in contrast with the many misunderstandings she found in her native±non-native data. And Shaw, Caudery, and Petersen (2009: 192) quote exchange students at universities in Sweden and Denmark who observed that although Americans and Australians were friendly, they were hard to understand, partly because they did not pronounce all the sounds. In fact, there is widespread agreement that ELF speakers are able to

12

2. ELF, misunderstandings and intelligibility

make their speech more intelligible if they do not try to approximate nativespeaker norms, often because they are good at adopting suitable accommodation strategies such as repetition and paraphrasing (Cogo 2009: 257). One might expect that familiarity with a variety of English should improve the degree to which it is intelligible. However, in a review of research on intelligibility in ELF, Pickering (2006) concludes that this is not necessarily true. She also notes that there is a wide range of factors that influence intelligibility, including the attitude of the listener, familiarity with the speaker or the topic, and level of tiredness. Something that might be discussed in connection with intelligibility is the concept of fluency. Most speakers of English hope that their speech is judged to be fluent, and achieving fluency seems to be one of the central goals of language teaching. But what do we actually mean by fluency? Is it connected with rate of speaking? Or with linking words together? Or with avoiding pauses? In fact, Hüttner (2009) observes that the concept of fluency becomes even more problematic in an ELF setting. And one might observe that all three features just mentioned may under some circumstances interfere with intelligibility. In fact, speaking more slowly, avoiding too much linking between words, and using appropriate pausing would seem to be quite advantageous in making oneself easily understood. And, as will become apparent, a fast speaking rate is one of the factors that contributes most often to misunderstandings occurring. One other aspect of fluency suggested by Prodromou (2008) is the easy use of idiomatic fixed chunks. However, although use of these established phrases certainly facilitates the production of speech, it can also lead to misunderstandings if listeners do not know the idioms. I will discuss the occurrence of unfamiliar idioms in Chapter 5.

2.4. Misunderstandings Misunderstandings of course occur in all kinds of communication, nativespeaker as well as ELF. The question arises, therefore, whether misunderstandings are more frequent in ELF settings or not. Conventional wisdom suggests that they are. Yet research indicates that this common-sense assumption is not borne out in reality, as ELF speakers tend to be particularly adept at avoiding misunderstandings, and ELF discourse is actually usually rather successful (Mauranen 2006; Kaur 2010: 205). It is, nevertheless, of considerable interest to investigate what kinds of misunderstandings occur, what causes them, and how they are dealt with.

2.4. Misunderstandings

13

Kaur (2009, 2010) makes a useful distinction beWZHHQµPLVXQGHUVWDQdLQJV¶DQGµQRQ-understandings¶ZLWKPLVXQGHUVWDQGLQJVWKHOLVWHQHUWKLQNV they know what is said but gets it wrong; but with non-understandings, the listener does not know what is said. However, while this is an insightful distinction in theory, in practice it is not always easy to apply. For example, someone often has an idea about what is said but is not too sure, and it is difficult to categorize instances such as these. Similarly, there are tokens in which listeners may be able to make a correct guess about the identity of the words but they are still not certain and therefore ask for clarification. I will include such instances in my data and analyse how they are dealt with. But should they be classified as misunderstandings or non-understandings? Kaur (2010: 194) also notes that there are various degrees of understanding. Clearly, instances in which a listener is able to guess a word but seeks to confirm it are at a different position on the continuum from instances in which the listener is unable to make any guess. Yet objectively placing each instance somewhere along this continuum is not straightforward. In addition to considering a possible distinction between misunderstandings and non-understandings, and also the degrees of understanding, one might consider the level of analysis. Somewhat similar to the distinction made by Smith (1992) between intelligibility, comprehensibility and interpretability that I discussed in the previous section, misunderstandings can be analysed at the word level, the utterance level, or the overall pragmatic level. In fact, as will be seen in Chapter 3, in this book a systematic distinction will be made between word-level misunderstandings (which are here termed µWRNHQV¶ RI misunderstanding) and utterance-level misunderstandings (which I will UHIHUWRDVµLQVWDQFHV¶RIPLVXQGHUVWDQding). In the sections on pronunciation and lexis, I will mostly discuss tokens of misunderstanding; but in the sections on grammar and repairs, I will deal with instances of misunderstanding. While discussion of the classification of misunderstandings is important, we must also consider what constitutes an instance of misunderstanding. Most research on ELF interactions analyses only those instances in which there is some obvious breakdown in the interaction, for example when one participant asks for clarification or else when someone clearly gets the wrong meaning and so answers inappropriately, though Kaur (2009) also looks at cases where a pause may signal a breakdown in communication. However, in addition I will include instances in which there is no apparent breakdown so the conversation seems to progress smoothly, but one of the interactants subsequently reported that there were some things they had not

14

2. ELF, misunderstandings, and intelligibility

understood. The methodology for identifying such instances will be presented in Chapter 3. While the focus of this book is on misunderstandings, it is important to keep in mind the fact that ELF interactions are usually rather successful, so in addition to analysing what causes misunderstandings to happen, I will consider features of speech that reduce their occurrence, including for example the use of full vowels rather than reduced vowels such as [ԥ]. The wider issue of how misunderstandings are avoided can be considered under the heading of repairs, which I will now discuss.

2.5. Repairs Repairs to misunderstandings can firstly be analysed as self-initiated or other-initiated (Kitzinger 2013). In addition, they can be self-repairs or other-repairs (Have 1999: 116; Liddicoat 2011: 210). Typical instances of each of these categories are:   



self-initiated self-repairs: a speaker realizes something may not be clear so provides a paraphrase or explanation without being prompted. self-initiated other-repairs: the speaker is unable to think of a word and asks for help, and the other person then offers a suggestion. other-initiated self-repairs: a listener does not understand something so asks for clarification, and the original speaker explains what they meant. This might be regarded as the archetypical kind of repair, but it is actually quite rare in ELF interactions, though Smit (2010) suggests that it may be more common when the interactants have got to know each other well so they feel that they can be quite direct in asking for clarifications. other-initiated other-repairs: the listener does not understand something and then provides the answer. This category includes instances where the listener corrects the original speaker, sometimes as a checking strategy to ensure that they got it right. Instances of correction are rare in ELF interactions (though it is possible that they are more common when a native speaker is involved), but they do sometimes occur, and I will discuss a few such instances.

Liddicoat (2011: 248) suggests that there is a strong preference for selfrepair, which is hardly surprising, as asking for help (self-initiated otherrepair) and correcting (other-initiated other-repair) are not common strategies. But, in fact, one common kind of repair that has been suggested in

2.5. Repairs

15

ELF interactions is when the listener does not actually indicate that there is a problem and instead adopts WKH µOHW LW SDVV¶ strategy (Firth 1996) in the hope that things will sort themselves out. It is hard to classify these repairs under the four categories listed above ± maybe these repairs could be described as zero-initiated self-repairs. Or maybe they could be classified as natural repairs, as one can say that the problem naturally fixes itself. Firth VXJJHVWVWKDWWKHµOHWLWSDVV¶strategy is particularly common in ELF interactions, especially the telephone conversations in an international company that he studied, though it remains uncertain whether the strategy is really more common in ELF settings than among native speakers. Perhaps it is actually the normal way we all behave when faced with something we do not understand. Indeed, Mortensen  FRQFOXGHVWKDWWKHµOHWLWSDVV¶ VWUDWHJ\LV³DEDVLFLQWHUSUHWLYHSURFHGXUHWKDWKXPDQEHLQJVIROORZLQRrder to make sense of the woUOG´ Misunderstandings involving silence are sometimes overlooked in research on misunderstandings in ELF because they generally do not become apparent from the transcript. Furthermore, we might also say that a few misunderstood words often do not have too much impact on the successful continuation of ELF interactions, as it is normal for things to sort themselves out naturally, and also we do not need to understand every single word in order to follow the gist of what is being said. Ehrenreich (2009: 144) quotes someone working in a German multinational company who said that he knew what a Chinese colleague wanted, but he could not tell what the individual words were. And maybe this is a common pattern. Nevertheless, in the current study, I will be looking at individual words that are not understood, as this gives important information about what enhances and what interferes with intelligibility. Furthermore, as I will VKRZ VRPHWLPHV WKH µOHW LW SDVV¶ VWUDWHJ\ GRHV QRW ZRUN DQG it subsequently becomes apparent to the listener that they have to sort out the identity or meaning of misunderstood words. Indeed, this is illustrated in the study by Firth (1996: 244) where one of the interactants realized that he had to find out what blowing cheese meant even though he had let it pass the first time it was said. Similar to the strategy of keeping quiet, minimal backchannels such as mm or yeah are comPRQ ZD\V RI DGRSWLQJ WKH µOHW LW SDVV¶ strategy. The primary role of such backchannels is to confirm that the listener is following what the speaker is saying, but they can also be used to conceal the fact that the listener does not actually understand something in the hope that it will all get sorted out in time. As we will see, such use of backchannels is common in ELF discourse, though once again it is not certain if the patterns

16

2. ELF, misunderstandings, and intelligibility

of usage are special to ELF interactions or are broadly the same as in native-speaker conversations. Finally, Kitzinger (2013: 255) notes that reSDLUV³FDQEHXVHGDVDZD\ of managing interpersonal conflicts DQG GLIILFXOWLHV´ VR WKH\ DUH QRW MXVW about dealing with problems of understanding. Indeed, under the rubric of repairs, we can consider some strategies which are adopted to ensure that there is no breakdown in communication. For example, House (2003: 568) suggests that a kind of response which she calls µUH-present¶ LV FRPPRQ among Asian speakers. This involves echoing part of the previous speaNHU¶V words, partly to encapsulate the information, but also to support working memory and to create a linkage between turns. Whether this kind of echoing of words is more common in Asian discourse is hard to evaluate; but I will offer a few examples of it occurring in my data, and I will suggest that one of the reasons it is sometimes adopted is to conceal the fact that the listener does not understand everything. Many of these kinds of conversational strategies involve a high degree of cooperation between interactants, so they might be described as forms of accommodation. I will now briefly consider that.

2.6. Accommodation There is sometimes a perception that the proposals for ELF-based teaching are based on negative things: students do not need to aspire to nativespeaker norms in pronunciation or grammar, and according to the LFC suggestions, it is not necessary to learn to use some of the inventory of English sounds. However, this overlooks an extremely important component of the proposals: the need for speakers to accommodate to their listeners. The importance of accommodation in ELF discourse has long been emphasized (Jenkins 2000: 167±175) and has subsequently been investigated in further depth (Jenkins 2007: 25). $FFRPPRGDWLRQ XVXDOO\ UHIHUV WR DGDSWLQJ RQH¶V SDWWHUQV RI VSHHFK WR PDNHLWPRUHVLPLODUWRWKDWRIRQH¶VLQWHUORFXWRUWKRXJKRIFRXUVH divergent behaviour sometimes also occurs, such as when someone wants to emphasize that they are different from their conversational partners, for example in order to exert power over them. Much of accommodation is connected ZLWK PRGLI\LQJ RQH¶V SURQXQFLDWLRQ LQFOXGLQJ VXFK WKLQJV DV speaking rate and pausing frequency (Giles, Coupland, and Coupland 1991: 7). However, there are different kinds of accommodation, affecting more than just pronunciation. First, there is listening accommodation, which

2.6. Accommodation

17

PHDQVJHWWLQJXVHGWRWKHSDWWHUQVRIVSHHFKRIRQH¶VFRQYHUVDWLRQDOSDUtners; and furthermore, there is accommodation in terms of grammar and word usage, so for example, Shaw, Caudery, and Petersen (2009: 193) report that, over a period of several months, exchange students in Sweden and Denmark sharply reduced the frequency of using the phrase I think, while at the same time well and like became used substantially more often, and they argue that this was a sign of them accommodating their speech patterns to those of the other speakers around them, suggesting a shared trend towards the use of a wide range of discourse management devices. In Chapter 4, I will consider phonetic accommodation, both listening and speaking, in Chapter 5, I will discuss code-switching, a form of accommodation in which conversational interactants seek to use all the linguistic resources that they share, and in Chapter 6, I will analyse pragmatic accommodation that is found in ELF data as speakers adopt a range of discourse strategies to try to ensure that they are not misunderstood.

2.7. Conversation Analysis Conversation Analysis (CA) originated in the work of Harvey Sacks on the description of interactional behaviour in audio recordings from a suicide hotline in Los Angeles in the 1960s, and it has been extended to provide an extensive methodological framework for the analysis of all kinds of conversational interactions (Have 1999: 3). Key aspects of the methodology associated with CA which I will adhere to in this book are:    

the descriptive, non-judgmental reporting of conversational interactions, ZKLFKDYRLGVUHIHUHQFHWRSUHFRQFHLYHGQRWLRQVRIµFRUUHFW¶XVDJH the use of natural conversations rather than specially designed experimental materials or read data the attempt to find out not just what people are saying (the form) but why they are saying it (the function) the inclusion within the transcription of lots of detail, including information about overlaps, pauses, changes in the tone of voice, and anything else that may help to explain what is going on

In one other respect, the current work might be seen as following the principles of CA. In his original work on telephone conversations, Harvey Sacks found just one exception to the conventional behaviour where the receiver speaks first, but instead of dismissing this instance as deviant, he focused on it, to try to determine why this unexpected behaviour occurred.

18

2. ELF, misunderstandings, and intelligibility

We can say, therefore, that one crucial aspect of work on CA is that deviant cases should be taken seriously (Have 1999: 39). And similarly, in the current work, misunderstandings are very much the exception rather than the norm, as most ELF interactions progress remarkably smoothly without too many misunderstandings. So, in this sense, by focusing on misunderstandings, I am conducting a detailed analysis of the abnormal cases. However, the analysis in this book is distinct from much work in Conversation Analysis in two respects. First, CA tends not to refer to accounts by the participants themselves, on the basis that they often do not know what they are doing or why they are doing it (Have  ,QGHHGµSXUH &$¶WUHDWVWKHFRQYHUVDWLRQDOUHFRUGLQJVDQGWUDnscripts as pure data which should be interpreted in their own right, unencumbered by distracting information about the backgrounds or the settings (Have 1999: 55). However, in this work I make extensive use of feedback from participants in two crucial ways: I get them to tell me what they said and also what they did not understand. Although it is certainly true that speakers often do not know why they say something, they usually do know what words they are trying to say. And this feedback from the speakers is vital, as I will show in Section 3.1.3, as there are many instances when I was simply unable to decipher the words without the input of the speakers. In addition, and equally importantly, I depend substantially on feedback from the participants to determine what they do not understand. This feedback must be treated with caution, as sometimes participants may subsequently claim they did not understand something when in fact they did understand it, possibly because they had no problems in context but later on were no longer able to decipher what was said when they listened to the recording. Nevertheless, as we will see, this feedback provides insightful information about intelligibility, and I believe that the overwhelming majority of the misunderstandings that are reported are genuine. Second, most work in CA is qualitative, with detailed analysis of transcripts but little attempt at quantification, and while adverbs such as commonly or typically may occur, numerical quantification is often avoided (Liddicoat 2011: 11). In fact, the usual practice is to find one or two key exemplars of a conversational exchange, examine them in considerable detail, and then investigate whether they are characteristic of interactions in general (Have 1999: 144). In contrast, in my investigation, I will try to quantify some aspects of the data, in particular what features of pronunciation are most likely to give rise to misunderstandings and which have little impact on intelligibility, and also the frequency of use of various repair strategies. The work of Smit (2010) is similarly grounded in CA but at the

2.7. Conversation Analysis 19 same time attempts to quantify the relative frequency of occurrence of the patterns that are found, and my research takes that as a model to follow. In many ways, therefore, the current study can be described as a mixedmethods study, combining qualitative and quantitative research, something that Dörnyei (2007) suggests is rather common in linguistic investigations. Let me finally discuss one further way in which analysis of ELF interactions may not fit into CA as it was originally conceived. One fundamental conjecture of CA is that all conversations are highly structured and orderly, and it is the goal of the analyst to work out this structure. Liddicoat (2011: 7) obseUYHVWKDWLWLVDVVXPHG³participants share the same procedures for GHVLJQLQJDQGLQWHUSUHWLQJWDON´. However, Firth (1996: 252) suggests that this may not apply to ELF interactions, as the interactants do not necessarily share the same knowledge about how to participate in conversations, and as a result the structure of their interactions may be more flexible than that of native-speaker conversations. One of the goals of my work is to investigate some of the ways that the participants in ELF settings negotiate their interactions in order to achieve mutual understanding, avoid miscommunication, and deal with those misunderstandings that do occur.

2.8. Conclusion Research on ELF interactions is becoming an increasingly important and vibrant field of study, offering substantial insights into the ways that people around the world are communicating in English. ELF interactants are generally focused on maintaining intelligibility, but it seems that they are sometimes not too concerned about adhering to native-speaker norms, so it is vitally important to determine what features of speech enhance intelligibility and which are more likely to cause misunderstandings to occur. Analysis of this is the main focus of this book. It is also instructive to investigate how repairs are achieved. We need to be aware that many of the patterns involved may differ from the ways that repairs are enacted in native-speaker discourse, so some of the complex turn-taking routines reported in Conversational Analysis may not be applicable in ELF interactions. I will now discuss the data that I collected and how it is analysed in this book.

Chapter 3 Data and methodology

One way of conducting research on misunderstandings is by analysing a corpus of spoken data. In this chapter, the Corpus of Misunderstandings from the Asian Corpus of English (CMACE) is described, and then the methodology for the analysis of the data is discussed.

3.1. CMACE Corpora of naturally-occurring English from around the world are now widely used in the analysis and description of speech. One of the most influential of these is the International Corpus of English (ICE), which provides a carefully balanced sample of spoken and written material in a range of different Englishes from places such as Britain, Singapore, India, Hong Kong, and the Philippines, and some substantial analyses have emerged (e.g. Nelson, Wallis, and Aarts 2002). The ICE corpus is restricted to Englishes from the Inner and Outer Circles, so it excludes speakers of English in the Expanding Circle. However, Seidlhofer (2011) has argued that there is no reason to continue to exclude speakers in the Expanding Circle from corpora of English usage. They now outnumber all other speakers of English, so it is important to conduct research into the ways that they interact. Indeed, the VOICE corpus that Barbara Seidlhofer and her colleagues have assembled in Vienna quite deliberately includes lots of speakers from the Expanding Circle (VOICE 2011). The Asian Corpus of English (ACE) is inspired by the VOICE corpus, and it adopts much of the same methodology as well as most of the transcription conventions. The ACE corpus is headed by Andy Kirkpatrick and is led by a team at the Hong Kong Institute of Education, and it consists of conversational data collected and transcribed by a number of teams, including those in Brunei, Hong Kong, Malaysia, the Philippines and Singapore. The Brunei component of the ACE corpus has been collected at the University of Brunei Darussalam (UBD) by a team led by Salbrina Sharbawi. It consists of 14 recordings, each lasting between 20 minutes and one hour, totalling about ten and a half hours. Five of these recordings involve three or four people having a discussion in a classroom at UBD, while the

3.1. CMACE

21

other nine are conversations between two people, with one interviewing the RWKHULQDOHFWXUHU¶VRIILFHDW8%'$OOWKHUHFRUGLQJVZHUHPDGHXVLQJD Handy H4n recorder, to ensure a high quality of recording for the speech of all the participants, and they are saved in WAV format. Nine of these ACE recordings made in Brunei, totalling about six and a half hours, have been investigated for this book in order to find instances where one speaker fails to understand something that is said by another speaker. These nine were selected because the people involved were able to come back subsequently and help identify where misunderstandings occur and also to clarify what is actually said. One of the recordings involves a discussion between four people, while the other eight are conversations involving two participants, the interviewer and the interviewee. The speakers, further details about the recording sessions, and the methodology adopted in identifying the instances of misunderstanding are described below. In this book, I will refer to these nine recordings as the ACE data; and I will refer to the corpus of misunderstandings derived from the ACE data as CMACE. 3.1.1. Speakers Five female and four male participants are involved in the current study. They are identified by their gender (F or M) followed by a two-letter code indicating their country of origin. All of them speak English fluently and proficiently, but some of them have a distinctive regional accent that can sometimes cause problems in intelligibility. They are listed in Table 3.1. Table 3.1. Participants in the recordings Code FBr FCh FJp FMa FTw MHk MIn MLs MNg

Speaker Female from Brunei Female from China Female from Japan Female from Malaysia Female from Taiwan Male from Hong Kong Male from Indonesia Male from Laos Male from Nigeria

Age 23 22 22 24 56 56 58 34 28

Background research student exchange student exchange student research student housewife university academic university academic government officer undergraduate student

One consideration in making the recordings was whether native speakers might be included. Some writers have observed that the dynamics of

22

3. Data and methodology

interactions can change when native speakers are involved. For example, one of the participants in the study by Matsumoto (2011: 109) observed that he hesitates in the presence of native speakers and only feels at ease when he is talking to non-native speakers. We might note that Jenkins (2000) excluded native speakers from the data which provided the original basis for the LFC, as her research mainly focused on interactions between learners of English from Japan and Switzerland. However, native speakers are not completely excluded from other ELF corpora, such as the ELFA corpus being compiled in Helsinki (Mauranen 2006: 129). Indeed, in more recent work, Jenkins has acknowledged that native speakers may be included in ELF data (e.g. Jenkins, Cogo, and Dewey 2011: 283). One issue, as noted in the introduction to this book, is that the concept of µQDWLYHVSHDNHU¶LVUDWKHUKDUGWRGHILQHMcKay 2002: 28; Davies 2003). In the current study, two of the participants, FBr and FMa, state that English is their best language, even though both have at least one other language that she uses regularly at home. So, are they native speakers or not? They would certainly be classified as coming from the Outer Circle, as they are from post-colonial countries in South-East Asia. But if their first and best language is English, surely they should be classified as native speakers. They are included in the study, because the conversations they are involved in yield lots of highly informative instances of misunderstandings. The data that constitute my corpus are rather different from some other ELF data in one important respect: all of the speakers were complete strangers, as none of them had ever met any of their conversational partners before the recordings. Much other work on ELF interactions focuses on communities of practice, people who form a group because they work or study together or else have shared interests, and these groups can develop innovative idioms and other shared patterns of language usage over a period of time. For example: Ehrenreich (2009) reports on the English usage in multinational companies based in Germany where some special terminology and modes of interaction have become established among the participants; and Smit (2010) conducts a detailed investigation into the English used by students from a range of countries involved in a two-year hotel management course at a school in Vienna. However, in the current study, all the participants were together for less than one hour, so we would expect to find far fewer examples of shared innovative patterns of language usage emerging in their conversations. Furthermore, as we will see, there are rather more instances of misunderstanding arising because of pronunciation in the data investigated here than reported by Smit (2010). And finally, some of the patterns of dealing with repairs are likely to be substan-

3.1. CMACE

23

tially different. Smit (2010) reports that the participants she studied could be quite direct in asking for clarification, as one might expect for people who know each other well. In contrast, in the data I am analysing here, the participants are likely to be less direct, and it is possible that WKHµOHWLWSDVV¶ strategy is adopted rather more frequently in my data. Of course, both kinds of data provide important insights into various aspects of ELF interaction, but my recordings provide data that is conducive for the study of misunderstandings. 3.1.2. Recordings The nine recordings that are analysed in depth are listed in Table 3.2. The identifying code for each recording consists of a two-letter code for the country of origin of each of the two participants, the first country listed being the country of the interviewee and the second being that of the interviewer. For the most part, therefore, I am investigating how well the second participant is able to understand the first, though in a few cases the misunderstanding goes the other way. For example: in one case in the Jp+Br recording, FJp gets the wrong meaning for find said by FBr; in an instance in the Hk+Ma recording, MHk is unable to understand the phrase fell into place said by FMa; and at one point in the In+Tw recording, MIn misunderstands the word Hash said by FTw and hears it as harsh. Table 3.2. Recordings Code Ch+Br Hk+Ma Hk+Tw In+Ma In+Tw Jp+Br Ls+Br Ng+Br Tw+Ma

Participant 1 FCh MHk MHk MIn MIn FJp MLs MNg FTw

Participant 2 FBr FMa FTw FMa FTw FBr FBr FBr FMa Total :

Duration (min:sec) 51:03 23:13 47:14 36:29 42:03 52:28 45:07 58:17 29:01 6:24:55

The Ng+Br recording is a group discussion involving four participants. However, two of them, a male from Pakistan and a male from Brunei, say very little during the recording, so the analysis here focuses on the interaction between MNg and FBr, in particular what FBr is unable to understand in the speech of MNg. The Ng+Br recording is therefore treated as an inter-

24

3. Data and methodology

action between two people, so it is similar in this respect to the other eight conversations. I will refer to extracts from these recordings by means of the identifying code of the recording together with the time in seconds from the start of the recording, so for example Ch+Br:357 refers to an extract which begins 357 seconds from the start of the conversation between FCh and FBr. 3.1.3. Transcription In the early days of the development of large-scale corpora of conversational speech, Sinclair (1985: 251) observed that there was a need to overhaul the systems for the description of English to deal with real conversational data; and Seidlhofer (2009: 38) suggests that the creation of largescale ELF corpora has necessitated a similar paradigm shift in our descriptive systems. However, this has already largely been achieved, particularly arising out of the work of Barbara Seidlhofer and her colleagues in Vienna in transcribing and analysing their VOICE corpus. I will therefore adopt the VOICE transcription conventions. They have been used in transcribing the ACE data, and they will be adopted in presenting extracts from the CMACE data in this book. They are summarized at the front of the book. The conventions used for the VOICE data are fairly consistent with those adopted in other work in Conversation Analysis (Have 1999: 87). One salient difference is the use of anglHG EUDFNHWV!« ! ! « etc. to show overlaps instead of the traditional use of square brackets µ>«@¶$QDGYDQWDJHRIWKHDQJOHGEUDFNHWVconvention is that it facilitates showing multiple overlaps that can occur in interactions with lots of participants, as details of the progression of the overlaps can sometimes get confusing with the simpler square bracket notation. In fact, in the current research, nearly all of the interactions involve just two participants; but it was decided to stick with the convention adopted in most work on ELF nowadays, to make the current work consistent with other research on ELF. One other decision was whether to use upper case to show all stressed syllables or whether it should only be used for unexpected stress (Have 1999: 88). Here, it was decided only to show the unexpected occurrence of stress, particularly when this may lead to misunderstandings (though, as will be seen in Chapter 4, misunderstandings arising from stress placement are in fact infrequent). See the Transcription Conventions on page x. One of the unusual aspects of the CMACE data is the extensive involvement of the participants in the analysis. House (1999: 76) observes that it is difficult to derive a watertight definition of a misunderstanding, so

3.1. CMACE

25

she suggests that it should depend on whatever the participants perceive as a misunderstanding. But if we do not obtain their detailed feedback, how can we find out what they perceive to be a misunderstanding? Therefore, I involved the participants in the transcription and analysis to SURYLGH D YDOXDEOH LQVLGHU µHPLF¶  SHUVSHFWLYH RQWR WKH GDWD 6PLW 2010: 97). Seidlhofer (2011: 56) has noted that there is a tendency for people working with speech corpora to focus too much on form and pay too little attention to function, and there is a need to get behind the data, to find out what is really going on. I believe that the involvement of the participants in the analysis facilitates obtaining the kind of in-depth insights that she envisages. The methodology adopted here offers two crucial advantages: first, it has enabled me to correct the transcription in many places when what is said is not clear (though, inevitably, it has not been possible to resolve all such instances); and secondly, it has made it possible for me to identify instances of misunderstanding that are not actually signalled in the recordings, which actually represents the overwhelming majority of the tokens. In this way, I can find instances where one of the participants does not understand something even when there is in fact no apparent breakdown in communication. Let me illustrate this with an extract from the Ng+Br recording. Extract 3.1. Ng+Br : 2742 (Token 173) Context: MNg is talking about how it is easy for some people to get a job after graduating from university. 1 2 3 4 5

MNg FBr MNg

«they are not looking for anything. that time they will be calling them because they are hot cake (1.3) they need their service. yeah yeah. not like. oh i want to no no no no no no no when they see this one you do in brunei today. if you have anything «

In Extract 3.1, MNg says it is easy for some people to get a job because they are hot cake (line 2), meaning that they are so talented that employers automatically come looking for them ± so it refers to people rather than things as is more common with the idiom hot cakes in British English. FBr does not understand this, partly because there is no [h] at the start of hot and furthermore she is not familiar with this use of the idiom, but there is no indication of her lack of understanding from listening to the recording. In fact, in line 3 FBr provides the backchannel yeah in response. I only found out that she does not understand it because she subsequently tran-

26

3. Data and methodology

scribed it as out kick; and the only way I know that it is actually hot cake is because MNg came and explained it to me. The value of the substantial involvement of the participants in the analysis can be appreciated when comparing the CMACE corpus with other studies of misunderstandings. For example, Mauranen (2006: 132) only found six clear cases of misunderstanding in her five hours of data from Finnish universities; Kaur (2010) based her analysis of problems of understanding on a corpus of fifteen hours of speech, but her report includes a detailed consideration of just twelve extracts; Björkman (2009) analysed 70 hours of lectures and group work in English in a Swedish university and lists only four tokens in which paraphrasing or self-correction occurs because of overt disturbance in the discourse, and she observes (p. 237) that it is very hard, or maybe impossible, to assess comprehension of the lectures; and Matsumoto (2011) analysed three recordings totalling nearly one and a half hours, but only six instances of misunderstanding are discussed. While it is likely that rather more tokens of misunderstanding were found in most of these studies and just a few key instances were selected for detailed consideration in order to illustrate the issues that are under investigation, it also seems probable that in all cases there were rather a lot of instances of misunderstanding that were not identified. In contrast, in the six and a half hours of data analysed here, 147 instances of misunderstandings have been identified and analysed, and the overwhelming majority of these would not have become apparent without the invaluable feedback from the participants. Even with such extensive feedback from the participants, inevitably there remain some tokens which I cannot decipher. For example, consider Extract 3.2. FBr subsequently transcribed the final word in line 6 in brackets, to indicate that she could not determine what word it is. So what in fact is FCh trying to say? Extract 3.2. Ch+Br : 2083 Context: They are discussing movies that they like to watch. 1 2 3 4 5 6 7 8 9

FBr FCh FBr FCh FBr FCh FBr

DOOULJKWLGRQ¶t particularly watch chinese movies? (.) cos i GRQ¶WNQRZDORW? EXW WKHUH¶V21(WKDWHUPLNQRZLW¶s quite f- LW¶s really funny erm (.) kung fu hustle? have you heard of it? kung fu hustle? yeah. kung fu (.) the (chose) kung fu hustle erm: (1.6) maybe i know the (.) chinese name but not the english o:h okay

3.1. CMACE

27

It is loud and clear as [t‫ݕ‬o‫ݜ‬s], but I cannot determine what the word is. When I asked her about it, FCh herself suggested that she might have started to say Chinese, but despite careful listening, she was also unable to work it out. Instances like this are not included in the corpus of misunderstandings because, in the absence of knowing what the word is, I cannot draw any conclusions about what the problem might be. Fortunately, with the help of the participants, there are not too many instances like this. I am very grateful to the generous contribution of all these people in the project.

3.2. Analysis 147 instances of misunderstanding were identified and collated in the CMACE database, which is available on-line. This section describes how the analysis of the misunderstandings was conducted. 3.2.1. Instances and tokens Many instances of misunderstanding involve more than one token. Here, I will explain the difference I make between instances and tokens. Extract 3.3 shows an instance of misunderstanding from the Ng+Br recording. FBr understands most of what MNg is saying, but she does not get some of the details, including in line 2 the word portion (a Nigerian term for an instance of punishment) and in line 3 the words weed (a misunderstanding that may arise out of the length of the vowel) and cutlass (perhaps because in Brunei one would refer to a parang instead). Extract 3.3. Ng+Br : 693 (Tokens 151 to 153) Context: MNg is talking about the punishments given in schools in Nigeria. 1 2 3 4 5 6

FBr MNg FBr MNg

what what what kind of punishment. how far yeah the punishment will be (.) giving you portion (.) go there (.) you have to go and weed the something with cutlass (.) or you have to kneel down inside the sun (.) for some hours under the hot sun YEAH that is it

Therefore, although this extract represents one instance of misunderstanding, it includes three separate tokens. While the features that contribute to each of these tokens can be analysed separately, at the same time it is im-

28

3. Data and Methodology

portant to consider them as a single instance, because sometimes there is substantial linkage between the tokens. For example, consider Extract 3.4. Extract 3.4. Hk+Tw : 999 (Tokens 50 & 51) Context: MHk is talking about practising what you are good at, such as swimming. 1 2 3 4 5

MHk FTw MHk

«you can swim but you never swim you know you never enjoy swimming (1.0) you never enjoy you know the freedom you know floating mm on the water you know  DOOWKDWVWXII«

In line 3, MHk says floating, but FTw does not understand this word, largely because the [l] is missing, and she subsequently left a question mark in her transcript. Then in line 5, she hears under instead of on the, partly because the consonant at the start of the is pronounced as [d]. But if she had understood floating, she would probably also have understood on the water. So these two tokens of misunderstanding, floating and on the, are linked. Of course, a single token may sometimes consist of more than one word, particularly when a fixed phrase is involved. In Extract 3.5, FCh has been talking about films she likes to watch, and even in this context about watching recent films (in a recording made in February 2011), FBr is unable to understand Black Swan until FCh adds that it stars Natalie Portman. Extract 3.5. Ch+Br : 2013 (Token 10) 1 2 3 4 5

FBr FCh FBr FCh FBr

... what about movies. you like to watch movies? yes: i just saw the (.) black swan. i liked it. rex one? yeah the natalie portman OH: RIGHT:

Here, Black Swan consists of two words, but it is just one entity, so it is treated as a single token of misunderstanding. Inevitably, it is sometimes not clear whether a phrase should be analysed as consisting of one token or more than one token. For example, consider Extract 3.6. Subsequently, FMa reported that she initially heard want the wall (line 2) as won the war, principally because there is no [l] at the end of wall. Although want and war could be treated as separate tokens within this instance, here I regard them as a single token because they are part of the same short phrase.

3.2. Analysis

29

Extract 3.6. Hk+Ma : 710 (Token 19) Context: MHk is talking about painting murals in America. 1 2 3

MHk

you all you know have this kind of background okay (0.8) then (.) i want the wall (.) this is i the wall i (.) everyday i will (.) go WRZRUNLZLOOVHHLJRWRVFKRROLZLOOVHHWKLVZDOO«

There is one other issue with regard to Extract 3.6. FMa misunderstands wall both times it occurs in line 2 (though she finally figures it out by the time it is said again in line 3). However, I do not treat the second wall as a separate token of misunderstanding as it occurs so soon after the first. In fact, if a word is misunderstood more than once within the same instance, it is classified as a single token of misunderstanding. In contrast, if the same word is misunderstood again somewhat later and it therefore belongs in a different instance of misunderstanding, then it is classified as a separate token. In Extract 3.7, FTw misunderstands the word plastic said by MHk, hearing it as past (largely because the [l] is missing, but also because there is no final [k]). Extract 3.7. Hk+Tw : 1405 (Token 56) Context: MHk is talking about wastage. 1 2

MHk

«IRUH[DPSOHKRZPXFK\RXNQRZplastic you use. if you XVLQJVKRSSLQJEDJ«

Then about two and a half minutes later, FTw again misunderstands the word plastic spoken by MHk, as shown in Extract 3.8, hearing it as past once more. As these two tokens of plastic are separated by over two minutes, they are in different instances and so they are treated as separate tokens. Extract 3.8. Hk+Tw : 1564 (Token 63) Context: MHk is talking about how long it takes for garbage to decay. 1 2

MHk

... how long you know for the (.) nature you know to digest this plastic FRQWDLQHU«

In the 147 instances of misunderstanding, there are 183 tokens, all of which are listed in the Appendix. The largest number of tokens in a single instance is four, as shown in Extract 3.9, in which FTw is unable to understand the words Tech, Lubbock, northern and west spoken by MHk, and she

30

3. Data and Methodology

subsequently transcribed them as there, lot of, low term and where respectively. Extract 3.9. Hk+Tw : 274 (Tokens 31 to 34) Context: MHk is talking about when he got married. 1 2 3 4

MHk

«LQWH[DV\RXNQRZDWWKDWWLPH ZHDUH\RXNQRZVWXG\in texas tech you know in the graduate school. in lubbock texas. er northern you know er (.) west you know part of the texas. so (.) thaWIHEUXDU\¶V WKDWGD\LVVQRZLQJWRR VR«

The first two of these tokens involve names, though FTw does not realize that they are names, while northern is pronounced as [l‫ܥ‬ftԥn], and west has no final [t]. 3.2.2. Classification After all the tokens of misunderstanding were collated, I attempted to classify the factors that give rise to each one, principally as involving pronunciation, lexis, or grammar. Inevitably, it is not always possible to identify a single factor, as quite often there are multiple causes. Consider for example Extract 3.10. Extract 3.10. Hk+Ma : 1211 (Token 26) Context: MHk is discussing the problems of moving to a new place. 1 2 3 4

MHk FMa MHk

«EHFDXVHHYHU\time when i relocate you know. either really (.) cold freezing cold mm or flaming hot  LW¶VXVHGWREHLW¶VRND\«

FMa subsequently reported that she was unable to understand flaming hot, and she guessed that the first word was fuming. The immediate cause of this can be assumed to be phonetic, as MHk simplifies the initial [fl] consonant cluster in flaming by omitting the [l], something he does rather often (as we will see in Section 4.1.4). However, he similarly omits the [r] from freezing in line 2, and FMa has no problem understanding this word. How can this be explained? The answer is that the phrase freezing cold is rather common in English, occurring 197 times in the on-line Corpus of Contemporary American English (COCA 2013), while the phrase flaming hot is much less common, occurring just twice. So we should actually conclude

3.2. Analysis

31

WKDW)0D¶VLQDELOLW\WR understand flaming hot is a combination of pronunciation and lexis, something which is quite common. Another instance from the same recording is shown in Extract 3.11. Extract 3.11. Hk+Ma : 154 (Token 14) Context: MHk is talking about why he decided to become an artist. 1 2 3

MHk

«ZKDWVXEMHFW\RXFKRVHWR VWXG\ IRU\RXU m a ? your you know f- (.) terminal degree? (.) base on (1.0) your interest

In this case, FMa is unable to understand terminal, and she subsequently guessed that it could be common (though that does not make much sense in this context). This might be considered a problem with pronunciation: terminal is pronounced as [t‫ޝܮ‬PX@ZLWKMXVWWZRV\OODEOHV But that does not tell the whole story. The real problem here is that the phrase terminal degree is frequently used LQWKH8QLWHG6WDWHVWRUHIHUWRRQH¶VILQDOGHJUHHVR for example it occurs 40 times in the COCA data. However, it is not so widely used outside the United States. Indeed, I had to get MHk to come and explain it to me. The conclusion is this: if terminal degree had been a familiar phrase for FMa, such as freezing cold in Extract 3.10 above, then saying it fast with some phonetic reduction would not matter too much. But given that it is not such a common phrase outside the USA, phonetic reduction is more likely to cause a problem. So, once again, I conclude that this token involves a combination of pronunciation and lexis. One should note that the issues illustrated in Extracts 3.10 and 3.11 are subtly different. In Extract 3.10, MHk is using his own idiosyncratic phrase flaming hot which does not occur frequently elsewhere, while in Extract 3.11, he is using a phrase terminal degree which is common in the USA but not elsewhere. Nevertheless, both of these extracts illustrate that many tokens of misunderstanding involve a range of different factors, and it is not always possible to identify a single cause. Despite these difficulties, I have attempted to classify the tokens of misunderstanding as involving pronunciation, grammar or lexis. In cases where there appear to be multiple causes, I cross-classify the tokens, so Extracts 3.10 and 3.11 are classified as involving both pronunciation and lexis. The overall results of this classification of the 183 tokens of misunderstanding are shown in Table 3.3, in which the percentages add up to more than 100% because of the cross-classification of tokens.

32

3. Data and Methodology

Table 3.3. Classification of tokens of misunderstanding Classification Pronunciation Lexis Grammar Code-Switching Miscellaneous

Total (%) 158 (86.3%) 41 (22.4%) 25 (13.7%) 4 (2.2%) 2 (1.1%)

It can be seen from Table 3.3 that the overwhelming majority of tokens of misunderstanding arise because of pronunciation, though it should be noted that 45 out of these 158 tokens involve something else as well. The figures shown here are broadly in line with those of Jenkins (2000: 85±87), who reports that among her 40 tokens, 27 tokens were classified as involving pronunciation, eight as lexis, just one as grammar, and four others. We might further note that, although the percentage of tokens involving grammar is rather higher than that reported by Jenkins, in fact in most of the 25 tokens, grammar appears to be quite a minor factor, as we will see in Chapter 5. The proportion in the CMACE data arising from pronunciation is much higher than the percentage found by Smit (2010: 202) who reports that just 4.1% of the repairs in her data occurred because of pronunciation. However, the data is very different, as her data was collected from participants who were together for two years while they studied together on a Hotel Management Programme, so it is hardly surprising that they got used to the patterns of speech of their classmates and teachers. Moreover, Smit reports that the occurrence RIµPLVKHDrLQJV¶LQKHUGDWDsome of which are related to pronunciation, decreased substantially over time, so 54.5% of these repairs occurred in the first few weeks when the participants had just started to get to know each other, while only 13.6% occurred during the third and final time that she investigated the same participants, almost two years later. In contrast, she reports that there was no such reduction over time in repairs arising out of issues of fact or because of vocabulary. As pronunciation seems to give rise to the majority of the tokens of misunderstanding in my data involving speakers who have not met each other before, I will consider it first in Chapter 4, before I discuss other factors in Chapter 5.

Chapter 4 Pronunciation

As outlined in Chapter 3, the overwhelming majority of the tokens of misunderstanding that have been identified, over 86% of them, involve pronunciation, either in conjunction with something else or on its own. This chapter investigates the features of pronunciation that cause misunderstandings to occur. I will first analyse consonants, and then vowels, before discussing suprasegmental features such as stress, rhythm and intonation. Finally, I will consider phonetic accommodation. As mentioned in Chapter 2, the LFC proposals were based primarily on the analysis of only 40 tokens, 27 of them involving pronunciation, and with speakers from just two countries: Japan and Switzerland (Jenkins 2000: 84±87). As is acknowledged by Jenkins, it is important to extend the investigation by analysing a larger corpus of misunderstandings involving speakers from a wide range of different backgrounds. For example, word stress is excluded from the LFC on the basis that, in the original data, unexpected word stress did not cause misunderstandings. But can we be sure that this would be true for speakers and listeners from other countries? This chapter therefore aims to extend the proposals concerning the LFC in the light of data that has been recorded in South-East Asia. It focuses on whether the features that were originally included in the LFC are confirmed to be essential for intelligibility in the CMACE data, and also the extent to which features that are outside the LFC might in fact sometimes cause problems. At the same time, it must be acknowledged that my research also has its limitations: it is likely that many of the features that I suggest are not important in maintaining intelligibility in the CMACE data will in fact be found to be crucial for speakers and listeners from other countries. There is always a need to extend the investigation to include participants from as many different backgrounds as possible, so that a comprehensive picture can emerge about factors that affect the intelligibility of English in international settings. Although the focus here is on misunderstandings, we should remember that most of the ACE recordings that are analysed actually reflect a high degree of mutual understanding between the participants, something that has been noted in virtually all work on ELF interactions (e.g. Mauranen 2006; Kaur 2009; Smit 2010). In other words, misunderstanding is the ex-

34

4. Pronunciation

ception rather than the norm. This chapter therefore also considers some of the non-standard features of pronunciation (from a native-speaker perspective) that do not result in misunderstandings, including some such as avoidance of vowel reduction and use of syllable-based rhythm that are widely found in ELF speech and may actually enhance the intelligibility of English spoken in international contexts.

4.1. Consonants In the LFC proposals, it is suggested that all English consonant sounds except >ș@ and [ð] are essential for intelligibility (Jenkins 2000: 132). However, simplification of some final consonant clusters is considered to be acceptable, and there is leeway in the pronunciation of dark-L. Here, I first analyse the TH sounds; then I discuss other sounds such as [h], [n], [l], [r] and [w], initial consonant clusters, final consonant clusters, dark-L, plosives, fricatives and affricates, and finally other omitted consonants. 4.1.1. TH The dental fricatives >ș è@ DUH WKH RQO\ VRXQGV IURP WKH LQYHQWRU\ RI twenty-four consonant phonemes of most native Englishes that are excluded from the LFC. These sounds are rare in the languages of the world with the result that many people find them difficult to produce, and furthermore some speakers in all three circles use other sounds at the start of words such as think and in the middle of other: in the Inner Circle, speakers in London may use [f] and [v] (Wells 1982: 328), and many in Ireland and New York use [t] and [d] (or their dental equivalents) (Wells 1982: 429, 515); in the Outer Circle, [t] and [d] tend to occur in Singapore (Deterding 2007: 14), Malaysia (Baskaran 2004a), and Brunei (Deterding and Salbrina 2013: 25), but in Hong Kong [f] is more common at the start of words such as three and think (Deterding, Wong, and Kirkpatrick 2008); and in the Expanding Circle, [s] and [z] often occur in China (Deterding 2006) and Germany (Swan 1987). Here, following the practice of Wells (1982), to DYRLGSUHVFULELQJKRZWKHVRXQGVµVKRXOG¶EHSURQRXQFHG,UHIHUWRWKHP as voiceless TH and voiced TH. And the question is this: does variability in the pronunciation of the TH sounds cause misunderstandings to occur? There is some previous evidence that pronunciation of initial voiceless TH as [f] may occasionally cause utterances to be misunderstood, as Deterding (2005) reports that three nights spoken by a young speaker from

4.1. Consonants

35

England was understood as free nights by four out of six listeners in Singapore while a fifth listener was unable to come up with any suggestion for this phrase. However, this involves a speaker from the Inner Circle, so it might be regarded as irrelevant for the current focus on ELF interactions. I will now consider the results from the CMACE data. There are a total of seventeen tokens in which variable pronunciation of the TH sounds may have contributed to an instance of misunderstanding, eleven involving voiceless TH and seven involving voiced TH. (One of them, Token 167, has both voiceless TH and voiced TH, so it is included in both totals.) I will first consider those with voiceless TH, which are listed in Table 4.1. In the first three, Tokens 13, 35 and 61, voiceless TH is pronounced as [f]; in the final one, Token 180, it is pronounced as [s]; in all the others it is pronounced as [t]. Table 4.1. Misunderstandings in words with voiceless TH No 13 35 61 98 112 159 160 165 167 170 180

Spk. MHk MHk MHk MIn MIn MNg MNg MNg MNg MNg FTw

List. FMa FTw FTw FTw FTw FBr FBr FBr FBr FBr FMa

Word north three throw it three ethnic three three three another thing cathode thought

Heard as law of few for two/three? other (tere) (tere) theory under them cut off saw

Context on the north hemisphere three feet high of snow just throw it as trash and er (.) i have three children have many ethnic GROUPS is the way we we use 6334 primary school. three like two master three phd you LW¶Vanother thing entirely this wire is cathode this one my children thought in

In all these tokens except the fourth and the last (Tokens 98 and 180), something other than the voiceless TH is the major factor causing the problem, and the pronunciation of the TH sound can be regarded as a minor issue: in Token 13, the initial sound in north is [l] rather than [n] (see Section 4.1.3); in Tokens 35 and 61, there is no [r] following the initial TH in three and throw (see Section 4.1.4); in Token 112, ethnic is spoken fast and not very loudly, and it seems that [n] is omitted from the word (see Section 4.1.9); in Token 159, three is part of the phrase six three three four (characterizing the education system in Nigeria) which is said rather fast (see Sections 4.5.3 and 5.1.2); in Tokens 160 and 165, three is pronounced with two syllables (see Section 4.5.4); in Token 167, another thing is pronounced as [‫ݞ‬nd‫ݞ‬tܼ], so the second syllable of another is missing (see Section 4.5.4); and in Token 170, the listener, FBr, does not in fact know the word cathode,

36

4. Pronunciation

as becomes clear a few seconds later when she is similarly unable to understand anode, transcribing it as anod (see Section 5.1.1). In fact, in addition to Tokens 159, 160 and 165, MNg says three on four other occasions during the 58 minutes of the Ng+Br recording, and FBr can understand him in these other instances; so it is only when the word is pronounced as [tԥri] (with a vowel inserted into the initial consonant cluster) that there is a problem. In other words, voiceless TH pronounced as [t] seems only to be problematic when it is combined with some other unexpected feature of pronunciation. Finally, note that in Token 165, FBr correctly interprets initial [t] as voiceless TH, but she is still unable to identify the word as three, hearing it as theory instead. Tokens 98 and 180 are perhaps the only ones in which the problem can be directly attributed to the pronunciation of voiceless TH. So let us analyse them in more detail. The context for Token 98 is shown in Extract 4.1. Extract 4.1. In+Tw : 161 (Token 98) 1 2 3 4 5 6 7

FTw MIn FTw MIn FTw MIn

so all your family are here?. yeah eventually er (.) i have three children (.) how many you have we got two oh great @@@ and how about you. do you (have three) three i have three i have three childrens yeah

In this extract, FTw is unsure if MIn says two or three, which is why she asks him to repeat himself. Furthermore, though the speech at the end of line 6 is not too clear because it is overlapping with the start of the response by MIn in line 7, FTw seems to have guessed three correctly, so maybe there is no misunderstanding at all. (It is classified as an instance of misunderstanding primarily because FTw asks for clarification.) Token 180 is shown in Extract 4.2. In this extract, FMa hears both tokens of thought as saw. Extract 4.2. Tw+Ma : 620 (Token 180) 1 2 3 4 5 6 7

FTw FMa FTw

the reason we (1.0) we move to singapore is because i want them to learn chinese as well (0.9) because my children thought in england uh-uh er in england when we were in england. my children thought the whole world (1.7) mummy is the only o- (.) only one speaking (.) mandarin

4.1. Consonants

37

This misunderstanding occurs principally because of the [s] at the start of thought, though the occurrence of a glottal stop rather than [t] at the end of the word may also be a contributory factor. Before leaving voiceless TH, we might consider two tokens in which the misunderstood words do not have voiceless TH, but where initial [t] is mistakenly heard as voiceless TH, presumably influenced by the pronunciation of voiceless TH as [t] elsewhere. (These two tokens are not included in Table 4.1 because there is no voiceless TH in the target word.) In Extract 4.3, FBr subsequently transcribed take logic and thinking as think logical thinking. The voiceless TH at the start of thinking is pronounced as [t], so perhaps this has influenced FBr into believing that the [t] at the beginning of take also represents voiceless TH. Extract 4.3. Ng+Br : 2147 (Token 166) Context: MNg is talking about the courses that students take in the Faculty of Arts and Social Sciences (FASS) at UBD. 1 2

MNg

... u b d in year one now i can say if you are in fass you will take logic and WKLQNLQJ ULJKW"«

In Extract 4.4, FTw hears tow as throw, even though the pronunciation is actually perfectly standard [to‫]ݜ‬. It seems that FTw has become used to MHk omitting [r] from an initial cluster (so she has accommodated to his speech in this respect), and she is also aware that voiceless TH can often be pronounced as [t] (though she has failed to notice that MHk actually pronounces it as [f], so she has not accommodated to this feature of his speech), and these two factors might explain why she hears tow as throw. Extract 4.4. Hk+Tw : 2117 (Token 71) Context: They have been talking about wastage in the United States. 1 2 3

MHk

... one guy why you need to buy v h engine big truck you know. why you know you tow everything every day. there is nothing to tow. there is nothing you know

As Extracts 4.3 and 4.4 both seem to involve listening accommodation, they will be discussed again in Section 4.6.1. I will now analyse the seven tokens of misunderstanding that involve a target word with voiced TH being heard as a word with no voiced TH (or, in the case of Token 72, no guess about the identity of the word was made). They are listed in Table 4.2.

38

4. Pronunciation

Table 4.2. Misunderstandings involving voiced TH No. 2 33 51 72 77 135 167

Spk. FCh MHk MHk MHk MHk MLs MNg

List. FBr FTw FTw FTw FTw FBr FBr

Word in that northern on the other this weather another thing

Heard as internet low term under ?? days waiter under them

Context (.) she also in that company texas. er northern you know floating «on the water (.) the other you know big city i think this you know rain the food (.) for the weather LW¶V another thing entirely

In nearly all of these instances, the problem arises because of factors other than the variable pronunciation of the voiced TH. In Token 2, in that is pronounced as [ܼn‫]ݦܭ‬, so the voiced TH is actually omitted. (The quality of the vowel in that is discussed in Section 4.2.2.) In addition, there is no verb in the sentence (see Section 5.2.4). In Token 33, northern is pronounced as [l‫ܥ‬ftԥn], with [l] instead of [n] at the start in addition to [ft] for the medial TH sound (see Section 4.1.3), so it is not surprising that FTw is unable to understand it (and, for this token, I had to rely on input from MHk to determine what he had said). In Token 72 there is no consonant at all for the TH sound, so other is pronounced as [‫@ޝݞ‬,Q7RNHQ0/VLQGLFDWHVWKDW he is going to talk about food, but then he suddenly changes his mind and discusses the weather instead, so it is perhaps not too surprising that FBr hears the food-related word waiter instead of weather (see Section 6.3.1 for a discussion of topic fronting). And in Token 167, as was discussed above in connection with voiceless TH, another thing is pronounced as [‫ݞ‬nd‫ݞ‬tܼ], so the middle syllable is missing from another (see Section 4.5.4). Of the seven tokens listed in 4.2, three seem to arise primarily because of the realization of the voiced TH sound. In Token 72 there is no audible consonant for the medial TH in other, which explains why it is not understood. In Tokens 51 and 77, the misunderstanding seems to arise largely because of the use of [d] for voiced TH. I will therefore analyse these two tokens in more detail. Token 51 has already been discussed in connection with Extract 3.4 in Section 3.2.1, and it is now shown again in Extract 4.5. Extract 4.5. Hk+Tw : 999 (Tokens 50 & 51) Context: MHk is talking about practising what you are good at, such as swimming. 1 2 3 4

MHk FTw MHk

«you never enjoy swimming (1.0) you never enjoy you know the freedom you know floating mm on the water you know (.) all that stuff «

4.1. Consonants

39

In line 2 of this extract, floating is pronounced with no [l] in the initial cluster (see Section 4.1.4), so FTw cannot understand it. If she had understood floating, she would almost certainly have heard on the water rather than under water. So we might conclude that hearing on the as under in Token 51 is actually related to the earlier failure to understand floating in Token 50, and the use of [d] at the start of the is a minor contributory factor. The wider context for Token 77, in which this is misunderstood as days, is shown Extract 4.6. Extract 4.6. Hk+Tw : 2799 (Tokens 77 & 78) 1 2

MHk

«LWKLQNthis you know rain forest (.) is really important. EHFDXVH\RXNQRZEUXQHLDFWXDOO\LV«

In this extract, in addition to misunderstanding this, FTw hears rain forest as wind forced (see Section 4.1.3). It is hard to tell whether this contributes to the misunderstanding of this or not; maybe, even if FTw had understood rain forest correctly, she would still not have understood this. Finally, I will consider one instance in which [d] is mistakenly heard as voiced TH. In Token 118, dishes is heard as this is. It is illustrated in Extract 4.7. Extract 4.7. In+Tw : 1721 (Token 118) 1 2

MIn

«LQHYHUWDVWHDQ\ GLIIHUHQWNLQGRIHUdishes except H[FHSWHUFKLQHVHLQGLDDQGRIFRXUVH«

In this extract, the main problem is that the medial consonant in dishes is pronounced as [s] rather than [‫( ]ݕ‬see Section 4.1.8); so, although it is possible that 0,Q¶V pronunciation of voiced TH as [d] elsewhere might have contributed to this token of misunderstanding (suggesting that FTw has accommodated to this feature of his pronunciation), it seems that the pronunciation of the medial consonant is a more important factor. In summary, although there are a few tokens where variable pronunciation of the TH sounds may have contributed to the misunderstanding, there are not many in which it is the main cause of the problem. Indeed, there are hundreds of cases in the ACE data in which voiceless TH is pronounced as [t] (or as [f] by MHk or [s] by FTw), and in which voiced TH is pronounced as [d], and this rarely causes any difficulty. In conclusion, therefore, the results support the exclusion of the TH sounds from the LFC. However, a word of caution is appropriate here: all the recordings took place in South-East Asia, where it is common for voiceless TH to be pro-

40

4. Pronunciation

nounced as [t] and voiced TH as [d], so the listeners will have been familiar with these patterns of speech (Deterding and Kirkpatrick 2006); and it is possible that variable pronunciation of TH may be more problematic in other places where listeners are less familiar with this feature of pronunciation. I will finally note WKDWXVHRI>W@LQVWHDGRI>ș@IRUYRLFHOHVV7+ may in fact enhance intelligibility in some contexts. The Civil Aviation Authority (CAA) in the UK stipulates that the pronunciation of both three and thousand should have initial [t] for the crucially important domain of air traffic communication (CAA 2012: ch. 2 p. 3), so it seems that in an international setting, the CAA believes that TH pronounced as [t] is more easily intelligible than >ș@ And Deterding and Salbrina (2013: 26) report that the occurUHQFH RI >W@ UDWKHU WKDQ >ș@ IRU LQLWLDO YRLFHOHVV 7+ LV VLJQLILFDQWO\ PRUH frequent when speakers in Brunei are reading a text than when they are engaged in conversation, and this suggests that, for some speakers at least, [t] might be the more prestigious form.

4.1.2. Initial [h] Omission of [h] at the start of a word, otherwise known as H-dropping, is characteristic of London speech (Wells 1982: 321) and its absence has a ORQJKLVWRU\RIEHLQJUHJDUGHGDVµYXOJDU¶DQGDVPDNLQJWKHVSHDNHUVRXQG µLJQRUDQW¶0XJJOHVWRQH 2003: 95). In fact, most speakers often omit the [h] from the start of the weak forms of pronouns such as him and her (Roach 2009: 91); and furthermore [h] is generally absent from some words borrowed from French, including hour, heir, honest and honour. Nevertheless, appropriate use of [h] at the start of a word seems important for maintaining intelligibility. Here I will consider whether H-dropping (or H-insertion) causes problems in ELF conversations in South-East Asia. In the CMACE data, there are six tokens of misunderstanding arising from H-dropping and two (Tokens 150 and 155) occurring because of Hinsertion, all of them in the speech of MNg. These eight tokens are listed in Table 4.3. For token 161, it may be noted that has is a function word, so omission of the [h] might be expected if has is pronounced using its weak form. However, in this case, the word has a full vowel, and this is incompatible with its use as the weak form of this function word. So I conclude that H-dropping is the key factor in this misunderstanding, though the occurrence of has rather than have and also rise rather than risen may also have contributed to the problem (see Section 5.2.3).

4.1. Consonants

41

Table 4.3. Misunderstandings involving H-dropping and H-insertion No. 150 154 155 161 162 168 173 174

Spk. MNg MNg MNg MNg MNg MNg MNg MNg

List. FBr FBr FBr FBr FBr FBr FBr FBr

Word us in higher all means has rise hold handle hot cake help us

Heard as house (naya) (home is) are try old undo (out kick) a post

Context not prepare (.) us for the real is why in higher institution know something by all means some of them has rise to (.) see somebody hold (.) seven you have to handle me if i because they are hot cake (.) LW¶Vhelp us again to know

Although in some cases something else is involved, such as the unfamiliar use of the idiom hot cake in Token 173 to refer to someone who is in demand from employers (as discussed in Section 3.1.3 and to be analysed in Section 5.1.3), it seems that most of these misunderstandings arise largely as a result of H-dropping or H-insertion. I conclude, therefore, that initial [h] is important, which is consistent with the LFC proposals. One final comment about H-dropping and H-insertion: it is found in the speech of only one of the speakers, MNg; and that is why it causes a problem. As discussed above, if a feature of pronunciation, such as the use of [t] for voiceless TH, is found among a range of different speakers, then listeners may already be accustomed to it and it is not generally a problem (Deterding and Kirkpatrick 2006). But if there is a feature which occurs only among a small subset of speakers, then it is more likely to cause misunderstandings. Indeed, this is consistent with many of the features of pronunciation discussed in this chapter, as each problematic feature is often characteristic of the speech of people from just one or two different countries. 4.1.3. [n],[l], [r] and [w] Making a clear distinction between consonants such as [n], [l], [r] and [w] is regarded as important in the LFC proposals. Unfortunately, initial [n] and [l] are typically confused by speakers from Hong Kong (Deterding, Wong, and Kirkpatrick 2008), and [l] and [r] are notoriously difficult for speakers from Japan to keep apart (Riney, Takagi, and Inutsuka 2005). Here, I will consider confusion between [l] and [n], then between [l] and [r], and finally the occurrence of [w] in place of [r]. There are six tokens of misunderstanding in which [l] occurs instead of [n] in word-initial position, four in the speech of MHk and two in that of FCh (who comes from Guangdong, the province next to Hong Kong in

42

4. Pronunciation

South China). In addition, there is one token (Token 8) in which FCh uses >O@LQVWHDGRIPHGLDO>ƾ@7KHVHVHYHQWRNHQVDUHVKRZQLQ7DEOH Table 4.4. Misunderstandings involving [l] instead of a nasal No. 5 8 9 13 20 23 33

Spk. FCh FCh FCh MHk MHk MHk MHk

List. FBr FBr FBr FMa FMa FMa FTw

Word nearby hanging noisy north next next northern

Heard as ?? (haling) (loisy) law of (lext) OHW¶V low term

Context india (.) and bhutan (.) LW¶s nearby went to the hanging bridge too noisy based on the north hemisphere art project i GLG«ZRXOGEHnext the day next will be tomorrow texas. er northern you know er (.)

In two of these tokens, some other phonological issue contributes to the problem: final TH is pronounced as [f] in north in Token 13, and medial TH is [ft] in northern in Token 33 (see Section 4.1.1). But in the remaining five tokens, the use of [l] in place of a nasal consonant seems to be the sole phonological factor that causes the word to be misunderstood, though it might be noted that in Token 5, FBr is unable to understand the earlier word as Bhutan, largely because it is pronounced with a medial [d], and this probably interferes with her locating the place as near to India, and in Token 23, the day next has unusual word order (see Section 5.2.6). Furthermore, even though another phonological factor has played a part in Tokens 13 and 33, the use of [l] for initial [n] seems to be a major factor. It appears, therefore, that failure to distinguish between [n] and [l] is problematic. All the tokens shown in Table 4.4 involve [l] in place of [n] RU>ƾ@. The reverse occurs with MLs using [n] in place of [l] in the coda of a syllable, a pattern that has been reported in Deterding and Kirkpatrick (2006) where listeners from Malaysia and the Philippines were unable to understand a speaker from Laos who pronounced holes with [n] instead of [l]. (This will be discussed in Section 7.1.) In the CMACE data, there are three tokens of misunderstanding arising from this substitution, as shown in Table 4.5. Table 4.5. Misunderstandings involving [n] instead of [l] in the syllable coda No. 145 147 149

Spk. MLs MLs MLs

List. FBr FBr FBr

Word football old old

Heard as food burn own own

Context LGRQ¶WWDONDERXWWKHfootball. the old (.) people sometimes usually sarong for: the old man

In both tokens involving old heard as own, there is no [d] at the end of the word. However, consonant cluster simplification in a phrase such as old

4.1. Consonants

43

man is not too surprising (Cruttenden 2008: 304) and would probably not cause a misunderstanding (see Section 4.1.5). In all three tokens, the use of [n] in place of [l] is therefore assumed to be the main cause of the problem. Next, let us consider misunderstandings arising because of confusion between [r] and [l], which occurs with three speakers: FCh, FJp and MLs. Eight tokens involve [r] in place of [l] and one (Token 134) has the reverse pattern, [l] instead of initial [r]. In Token 132, it is the second [l] of fluently that is pronounced as [r]. These nine tokens are shown in Table 4.6. Table 4.6. Misunderstandings involving confusion between [r] and [l]. No. 6 10 128 131 132 133 134 138 141

Spk. FCh FCh FJp FJp FJp MLs MLs MLs MLs

List. FBr FBr FBr FBr FBr FBr FBr FBr FBr

Word club Black Swan literature Labuan fluently like the real sultan lunchtime

Heard as crowd (rex one) ?? ?? poetry rather (leal) (suratan) runtime

Context the international club they i just saw the black swan major is literature in japan to labuan with my friends can: speak english fluently very (.) er: like the civilisation the real situation when i come former president to your sultan the lunchtime is hh usually

For FCh, the use of [r] in place of [l] only occurs in consonant clusters, at the start of club and Black in Tokens 6 and 10 respectively. In contrast, for some of the remaining tokens involving FJp and MLs, the substitution may also affect [l] on its own at the start of the word. In all the tokens listed in Table 4.6, confusion between [l] and [r] seems to be a substantial factor in the misunderstanding, though in addition in Token 138, sultan is pronounced as three syllables (see Section 4.5.4), and in Token 141, [t‫ ]ݕ‬is omitted from lunchtime (see Section 4.1.8), so maybe these are more serious issues. It is also possible that the use of the definite article the before the name of the film Black Swan might have confused FBr in Token 10 (see Section 5.2.2). Finally, for Token 131, the conversation has been about films, so FBr may be puzzled about the relevance of Labuan (a small island near Brunei). This token is discussed again in Section 6.3.1, in connection with the importance of clearly establishing a topic. I will now analyse [w] in place of [r]. There are five tokens in which [r] at the start of a word is heard as [w], all in the speech of MHk. They are shown in Table 4.7. In all these tokens, the use of [w] in place of [r] seems to be a major factor, though in addition in Tokens 12 and 16, the medial consonant in reverse is pronounced as [w] rather than [v] (see Section 4.1.8) and in both Tokens 65 and 78, FACE is heard as KIT (see Section 4.2.3).

44

4. Pronunciation

Table 4.7. Misunderstandings involving [w] instead of [r] No. 12 16 60 65 78

Spk. MHk MHk MHk MHk MHk

List. FMa FMa FTw FTw FTw

Word reverse reverse wrapping related rain forest

Heard as we worse we worse weapon we live wind forced

Context know er reverse (.) you know \RXNQRZ LW¶V reverse paper wrapping stuff (.) a lot that art is related with truth know rain forest (.) is really

It seems that failure to distinguish the consonants [n], [l], [r] and [w] is a substantial cause of misunderstanding. Riney, Takagi, and Inutsuka (2005: 463) suggest that substitutions might be allowed for [l] and [r], as they are so widely confused by speakers in Japan. However, it appears that using [r] in place of [l] is likely to cause speakers from Japan to be misunderstood by listeners in the rest of the world, as predicted by the LFC proposals. In addition, confusion between [l] and [n] and also between [r] and [w], as occurs widely in South China and Hong Kong, seems to be problematic. 4.1.4. Initial consonant cluster simplification In the LFC, the treatment of consonant clusters depends on where in the word they occur. In initial and medial position, simplification is discouraged; but in final position, some degree of simplification is allowed, particularly as this is common among native speakers. In this section, I will consider consonant cluster simplification in the onset of a syllable, and then in the next section, I will discuss consonant clusters in the coda. There are twenty-one tokens in which omission of [r] or [l] from a consonant cluster in the onset of a syllable has contributed substantially to the occurrence of a misunderstanding, making it the most common phonological source of misunderstandings in the CMACE data. The eleven tokens in which [r] is absent are shown in Table 4.8, and then the ten tokens involving omission of [l] are listed in Table 4.9. In a few cases in Table 4.8, something else contributes to the problem. In Token 17, there is no verb in everybody afraid about the same thing (see Section 5.2.4), in Tokens 35 and 61 initial voiceless TH is pronounced as [f] (as discussed in Section 4.1.1), in Token 42 the final [t‫ ]ݕ‬in church is absent (see Section 4.1.8), in Token 76 [w] instead of [v] occurs in the middle of provide and there is no final [d] (see Sections 4.1.8 and 4.1.9), and in Token 137 there is no final [t] in present (see Section 4.1.5). But in Tokens 24, 45, 55, 140 and 142, omission of [r] seems to be the sole issue causing the misunderstanding, and in all cases the absent [r] is a major factor.

4.1. Consonants

45

Table 4.8. Misunderstandings involving omission of [r] from an initial cluster No. 17 24 35 42 45 55 61 76 137 140 142

Spk. MHk MHk MHk MHk MHk MHk MHk MHk MLs MLs MLs

List. FMa FMa FTw FTw FTw FTw FTw FTw FBr FBr FBr

Word afraid phrase three different church process process throw it provide present coffee-break treaty

Heard as are feel phase few (deventray) causes courses for to why (pisen) coffee bake ??

Context everybody afraid about you know phrase (.) you have three feet high of (.) er different church the process you know know process take more just throw it as trash and UBD provide you know saw: some present from serve in (.) coffee-break the treaty agreement or

The ten tokens involving omission of [l] from a consonant cluster at the start of a word all occur in the speech of MHk. They are listed in Table 4.9. Table 4.9. Misunderstandings involving omission of [l] from an initial cluster No. 26 30 36 37 50 56 57 59 63 67

Spk. MHk MHk MHk MHk MHk MHk MHk MHk MHk MHk

List. FMa FTw FTw FTw FTw FTw FTw FTw FTw FTw

Word flaming class plough close down floating plastic plant planting plastic meal plan

Heard as fuming car power goes down ?? past pound pounding past view pen

Context freezing cold ... or flaming hot not skip the class you know you know snow plough truck they close down the university freedom you know floating you know plastic you use (.) if then you can plant you know you planting something you digest this plastic container the meal plan (.) just killing

As before, something else can be identified as contributing to the misunderstanding in some of the tokens. In Token 26, the phrase flaming hot is an unusual collocation (see Sections 3.2.2 and 5.1.2); in Token 30, [s] is omitted from class (see Section 4.1.8); in Token 36, there is no verb in they no you know snow plough truck (see Section 5.2.4); in Token 37, there is little aspiration on the [k] in close (see Section 4.1.7); in Token 56, there is no final [k] in plastic (see Section 4.1.9); and in Token 67, the phrase meal plan for a schedule of payment for student food rarely occurs outside the USA (see Section 5.1.2). But in Tokens 50, 57, 59 and 63, there seems little apart from the omitted [l] to cause a misunderstanding. It appears that omission of [l] from a word-initial consonant cluster can cause a problem. Before leaving initial clusters, I will consider two instances from MHk involving [j]. In Extract 4.8, FMa hears companies rather than communist,

46

4. Pronunciation

partly because of the absence of [j] at the start of the second syllable, though we might note that (as expected) the second syllable of communist is unstressed, so in reality [j] would not generally be very prominent even if it were present. In addition, two other features of pronunciation contribute to this misunderstanding: the absence of [t] at the end of communist; and the use of [‫ ]ݞ‬rather than [‫ ]ܥ‬in the first syllable. These two issues will be discussed in Sections 4.1.5 and 4.2.6 respectively. Extract 4.8. Hk+Ma : 577 (Token 18) Context: MHk is talking about painting murals 1 2

MHk

«DQGWKHQWKHcommunist you know using this huge painting (.) as a propaganda (.) you know that  VR«

In Extract 4.9, FTw fails to hear an approximant [j] at the start of curious, and she subsequently transcribed it as kill it. We might note that, in addition, FTw does not know the word collage (see Section 5.1.1), but this probably does not affect the misunderstanding of curious. (The reason for the laughter by FTw in line 6 will be discussed in Section 6.2.7.) Extract 4.9. Hk+Tw : 875 (Tokens 47 & 48) Context: MHk is talking about different kinds of art 1 2 3 4 5 6

MHk FTw MHk FTw

«RKif you touch it and then you find oh this is a (.) paper you know mm collage on the top you know (.) and (.) then you¶re curious you know they ask questions @@

It seems, therefore, that retention of [r] and [l], and maybe sometimes [j], in clusters in the onset of a syllable is important for maintaining intelligibility. This is consistent with the LFC proposals. Finally, while discussing the pronunciation of initial consonant clusters, I will analyse Extract 4.10, in which FBr subsequently transcribed trekking as checking, because the word is said with [t‫ ]ݕ‬rather than [tr] at the start. Extract 4.10. Ch+Br : 1415 (Token 7) Context: 7KH\DUHWDONLQJDERXW)&K¶VWULSWR7HPEXURQJDGLVWULFWRI%UXQHL 1 2 3

FBr FCh FBr

so what did you do in temburong er rafting? and trekking oh WOW

4.1. Consonants

47

One might note that FCh uses a pronunciation of trekking that is rather similar to how the word might be said in the UK, where train and chain are often almost homophones. This is a clear example in which imitating InnerCircle pronunciation may not be the best way of speaking in an ELF setting. FCh might be advised to have a clear [tr] at the start of words such as trekking, even if that is not how some people in Britain would say the word. 4.1.5. Final consonant cluster simplification Final consonant clusters are very different from initial ones, and it is normal for native speakers to omit sounds from final consonant clusters under some circumstances. For example, Cruttenden (2008: 304) gives a long list of phrases in which word-final [t] or [d] is omitted from the end of a consonant cluster by speakers in the UK when the next word begins with a consonant, including soft centres, left wheel, drift by, mashed potatoes, finished late, pushed them, bend back, cold lunch, found five, dined well, tinned meat, loved flowers, and many more. If omission of [t] and [d] in these phrases is the norm among native speakers, it would seem strange to expect ELF speakers to retain them. But let us consider cases in which omission of a consonant from a word-final consonant cluster may be a factor in the occurrence of a misunderstanding in the CMACE data. I will first analyse cases in which the omission of a word-final [t] may have contributed to a misunderstanding. Table 4.10 lists eleven tokens in which a word with [t] at the end of a word-final cluster is understood as a word with no final [t], or for which the listener was subsequently unable to come up with any hypothesis (Tokens 89 and 158). Table 4.10. Misunderstandings involving omission of [t] from a word-final cluster No. 18 19 34 52 89 90 104 137 139 143 158

Spk. MHk MHk MHk MHk MIn MIn MIn MLs MLs MLs MNg

List. FMa FMa FTw FTw FMa FMa FTw FBr FBr FBr FBr

Word communist want the wall west environment i went to cairo joint fact present just their accent board test

Heard as companies won the war ZKHUH¶V that long ?? changing like (pisen) their the option ??

Context then the communist you then (.) i want the wall you know er (.) west you control the environment i went to cairo a little i have a joint supervision despite the fact that they i saw: some present from i think (.) just meat. is er: their accent (.) er like matriculation board test i

48

4. Pronunciation

The first four tokens involve MHk. In three of them, other phonetic factors seem to be key: for Token 18, the absence of [j] in communist was discussed in Extract 4.8 in Section 4.1.4, and [‫ ]ݞ‬rather than [‫ ]ܥ‬in the first syllable of the word is considered in Section 4.2.6; in Token 19, the main problem seems to be the absence of [l] at the end of wall (see Section 4.1.6), resulting in the word being heard as war, and then it is hardly surprising that the phrase is heard as won the war; and in Token 52, the main issue is probably the use of [w] rather than [v] in environment (see Section 4.1.8). Tokens 89, 90 and 104 all involve MIn. In Token 89, i went to cairo is pronounced as [awentԥkær‫]ݜ‬. One problem is that Cairo is said quite fast and not very loudly as [kær‫]ݜ‬, with the five syllables in the clause i went to cairo taking a total of 0.7 seconds. Although it is true that there is no separate [t] at the end of went, in fact pronouncing went to as [wentԥ] is perfectly standard, and producing an extra [t] for went might be regarded as hyper-articulation. Similarly, both Tokens 90 and 104 are spoken quite fast and not very loudly, but there does not seem to be anything non-standard involved. We might note that omission of the [t] at the end of fact in despite the fact that would be expected in British English (Cruttenden (2008: 304). The speaking rate in Tokens 89, 90 and 104 is analysed in Section 4.5.3. Tokens 137, 139 and 143 all involve MLs, and in each case there is another substantial issue in the pronunciation that has caused the problem: in Token 137, there is no [r] in present (see Section 4.1.4); in Token 139, there is no [s] in just (see Section 4.1.8); and in Token 143, there is no [k] in the middle of accent (see Table 4.12, later in this section). Token 158 is by MNg, and the wider context is shown in Extract 4.11. The phrase joint admission and matriculation board test is said quite fast as [d‫ܼܧݤ‬nt‫ݞ‬dmܼ‫ݕ‬ԥnԥm‫ݞ‬trܼkj‫ݜ‬le‫ݕ‬ԥnb‫ܧ‬dԥt‫ݞ‬s], the twelve syllables taking just 1.62 seconds (see Section 4.5.3). It seems to be an established phrase in Nigeria (Wikipedia 2013). Although the omission of [t] from the end of test is one factor in this token of misunderstanding, there seem to be multiple other issues involved, including an epenthetic vowel after board and a vowel that sounds like [‫ ]ݞ‬in test. In fact, the main problem here is that MNg seems to treat this as a common fixed phrase, so he says it rather quickly without noting that people in Brunei may not be familiar with it. I will discuss 01J¶VXVHRIfixed phrases such as this in Section 5.1.2. Extract 4.11. Ng+Br : 894 (Token 158) Context: MNg is talking about the education system in his country 1 2

MNg

so if i now pass that joint admission and matriculation board test i pass it. i will now enter the school «

4.1. Consonants

49

Finally, we should look at Token 34, the only one in which the absence of word-final [t] on west seems to be the main cause of the misunderstanding. The wider context for this is shown in Extract 4.12. Extract 4.12. Hk+Tw : 274 (Tokens 31 to 34) 1 2 3 4

MHk

«LQWH[DV\RXNQRZDWWKDWWLPH ZHDUH\RXNQRZVWXG\LQ texas tech you know in the graduate school. in lubbock texas. er northern you know er (.) west you know part of the texas. so (.) WKDWIHEUXDU\¶V WKDWGD\LVVQRZLQJWRR so «

In this extract, there are four tokens that are misunderstood (as discussed in connection with Extract 3.9 in Section 3.2.1), including northern just before west. We could say that this has had an influence on the failure of FTw to understand west, which she subsequently transcribed as ZKHUH¶V. However, the only phonetic feature of the word west itself that contributes to it being misunderstood seems to be the lack of a final [t]. Now I will analyse tokens in which the omission of [d] from a wordfinal cluster may have contributed to a misunderstanding. The six tokens containing words with a word-final consonant cluster ending in [d] that are understood as words without [d] (or, in the case of Token 156, the listener was subsequently unable to make a guess) are listed in Table 4.11. In two of these, Tokens 107 and 156, [d] in fact occurs in the pronunciation, but they are included in Table 4.11 because they involve a word with [d] at the end of a word-final consonant cluster. Table 4.11. Misunderstandings involving [d] at the end of a word-final cluster No. 86 107 108 147 149 156

Spk. MIn MIn MIn MLs MLs MNg

List. FMa FTw FTw FBr FBr FBr

Word by the end .. world cold war old old you attend a

Heard as biology war guru own own ??

Context yeah by the end of the day er in the world at the time so have the (.) cold war (.) so the old (.) people sometimes sarong for: the old man in my you attend a (.) brunei school

In Token 86, by the end of the day is pronounced as [baܼ‫ܥ‬lԥd‫ݤ‬eܼ], with multiple issues, including [d‫ ]ݤ‬instead of [d] at the start of day (see Section 4.1.8), so it is hardly surprising that it is not understood. In Token 107, the main issue seems to be the absence of [l] from world (see Section 4.1.6). In Token 108, cold war is pronounced as [ko‫ݜ‬lw‫@ޝܧ‬VRLQWKLVFDVHWKHRPLssion of [d] seems to be the main problem, though minimal aspiration on the initial [k] might also be a contributory factor (see Section 4.1.7). In Tokens

50

4. Pronunciation

147 and 149, as already discussed in Section 4.1.3, [n] occurs instead of [l], and this seems to be the main issue. And in Token 156, attend a is pronounced as [‫ݞ‬t‫ܭ‬nd‫]ݞ‬, with a full vowel in the first syllable of attend and also in a, and no clear stress on the second syllable of attend. These issues will be discussed in Sections 4.2.5 and 4.4.1. In summary, therefore, the omission of [d] from the end of a word-final consonant cluster appears to be the major factor causing a misunderstanding in only one case: Token 108, in which cold war is heard as guru. In dealing with consonant clusters, I will mention three further tokens that involve the omission of non-final consonants (other than dark-L, which is discussed in the next section, and fricatives, which are dealt with in Section 4.1.8). They are listed in Table 4.12. Table 4.12. Misunderstandings involving the omission of non-final consonants No. 96 143 148

Spk. MIn MLs MLs

List. FTw FBr FBr

Word correct accent point

Heard as great option poet

Context er the correct one [MIn:name] er: their accent (.) er like the first GRQ¶WPDNHWKHpoint for the

In Token 96, [k] is omitted from the final cluster in correct, and the word is heard as great. Aspiration on the initial [k] is also a factor here (see Section 4.1.7). In Token 143, [k] is omitted from the middle of accent. And in Token 148, both [n] and [t] are omitted from point, but the word is heard as poet (which is why the token is not included in Table 4.10), so it seems that the omission of the [n] is a bigger problem than the absence of the [t]. Finally, I will discuss Token 1, shown in Extract 4.13, in which garment is pronounced as [g‫ޝܤ‬WPܼQ@VRWKH>W@VHHPVWRKDYHPRYHGWRWKHHQGRIWKH first syllable. Although omission of [t] from the word-final consonant cluster may be a contributory factor, the bigger problem seems to be insertion of [t] earlier in the word. (In fact, in this instance FBr guesses the word correctly, but it is still treated as an instance of misunderstanding because she asks for confirmation that she has got it right.) Extract 4.13. Ch+Br : 357 (Token 1) 1 2 3 4 5 6 7

FCh FBr FCh FBr FCh FBr

er: my father and my mother (.) they own a company (.) yeah business company? yeah business company about the the: (.) garment garment yeah clothes clothing oh: okay

4.1. Consonants

51

In conclusion, there are only two tokens where the main issue seems to be the omission of [t] or [d] from the end of a word-final consonant cluster: Token 34, in which west is heard as ZKHUH¶V; and Token 108, in which cold war is heard as guru. It appears, therefore, that omission of [t] or [d] from the end of a word-final consonant cluster is not much of a problem, which is consistent with the LFC proposals. I started this section by stating that native speakers often omit [t] and [d] from word-final consonant clusters. But perhaps this is irrelevant for ELF speakers who, it can be argued, should not be bound by the same rules as native speakers. If retention of [t] and [d] at the end of a word enhances intelligibility, then retention of these sounds should be encouraged. However, I have found little evidence that omission of [t] or [d] causes much of a problem. Furthermore, there is evidence that, in their attempt to speak clearly and correctly, people from China and Korea are sometimes so concerned about omitting a final [t] or [d] that they insert an extra vowel, making fast sound like faster (Deterding 2006; Couper 2011). So it seems that these speakers might be advised to omit the final [t] or [d] under some circumstances. Omission of final [t] or [d] from the end of consonant clusters rarely causes misunderstandings, and retention of these sounds can be problematic if it is accompanied by an epenthetic vowel. 4.1.6. Dark-L In most varieties of English, dark-L (transcribed phonetically as [ܽ]) occurs in the coda of a syllable (Roach 2009: 49). However, it is common in a range of Englishes for dark-L to be pronounced as a close back vowel such as [‫]ݜ‬, a process known as L-vocalization (Wells 1982: 20). This happens especially in London English (Wells 1982: 313) and is one of the most salient features of the variety sometimes known as Estuary English (Cruttenden 2008: 79), but it is also found elsewhere in the world, including Inner-Circle varieties such as those of Australia and New Zealand (Horvath and Horvath 2001) as well as Outer-Circle ones such as that of Singapore (Tan 2005). Here I will discuss the extent to which variable pronunciation of [l] in a syllable coda gives rise to misunderstandings in the CMACE data. In total, there are 23 tokens in which a word with [l] in the coda of a syllable is understood as a word with no [l], or where the listener was unable to come up with any guess. Three of these involve [n] instead of [l] by MLs, so football is heard as food burn in Token 145, and old is heard as own in Tokens 147 and 149. These three have already been discussed in Section 4.1.3. In addition, Token 138 involves [r] in place of [l] in sultan, so it has

52

4. Pronunciation

also been dealt with in Section 4.1.3; and Token 108 involves omission of [d] in cold war (see Section 4.1.5). These five tokens are quite distinct from the use of a vowel for dark-L, so they are not analysed further here. Next, let us consider the three tokens shown in Table 4.13 where the listener does not know the word, so the main issue is classified as lexical rather than phonological. In Tokens 74 and 101, FTw simply does not know the words vinyl and inculcate respectively (see Section 5.1.1). In Token 181, FTw says knuckle to refer to pork knuckle (Schweinshaxe), a common dish in South Germany, but FMa is not familiar with this term, especially when the word pork is omitted (see Section 5.1.4). It would not be appropriate, therefore, to conclude that the pronunciation of the [l] is a major factor in these three misunderstandings. Table 4.13. Misunderstandings with [l] in the coda where the word was not known No. 74 101 181

Spk. MHk MIn FTw

List. FTw FTw FMa

Word vinyl inculcate knuckle

Heard as ?? got ??

Context still the vinyl plastic (.) is (.) you i (.) i inculcate er disseminate er drank beer (.) and knuckle eating

The two tokens shown in Table 4.14 might also be considered under lexical issues, as the phrase is not familiar to the listener (see Section 5.1.2). In Token 14, the phrase terminal degree to refer to the final degree that one studies is not familiar to FMa, and in Token 67 the phrase meal plan to refer to the payment for a schedule of meals for university students is a term that is not known by FTw. Furthermore, in both these tokens, there is something more important in the pronunciation: in Token 14, terminal is said with just two syllables (see Section 4.5.4); and in Token 67, plan has no [l] in the initial cluster (see Section 4.1.4). Table 4.14. Misunderstandings with [l] in the coda in an unfamiliar phrase No. 14 67

Spk. MHk MHk

List. FMa FTw

Word terminal meal plan

Heard as common view pen

Context f- (.) terminal degree (.) base on the meal plan (.) just killing

Let us next analyse the four tokens involving MNg, which are listed in Table 4.15. In three of these, Tokens 155, 168 and 174, the main issue seems to be H-insertion or H-omission, as discussed in Section 4.1.2, though in all cases L-vocalization might also be a contributory factor. And in Token 157, there is little aspiration on cut (see Section 4.1.7), and furthermore, the grammar is unexpected, as one might expect a main verb to

4.1. Consonants

53

occur between will and out, or maybe we can say that in this case out is being used as a verb (see Section 5.2.4). Therefore L-vocalization is not the main factor causing the misunderstanding in any of these four tokens. Table 4.15. Misunderstandings from MNg with [l] in the coda of a syllable No. 155 157 168 174

Spk. MNg MNg MNg MNg

List. FBr FBr FBr FBr

Word all means will out handle help us

Heard as (home is) way out undo a post

Context to know something by all means that exam (.) the cut off will out you have to handle me if i come LW¶Vhelp us again to know

That leaves nine tokens involving [l] in the coda of a syllable. They are listed in Table 4.16. A contributory factor in two of these, Tokens 2 and 39, is grammatical: in Token 2, the most important issue seems to be the absence of a main verb in the clause she also in that company (see Section 5.2.4), though it is true that L-vocalization may be a factor, as also is pronounced as [‫ޝܧ‬VR‫( ]ݜ‬and see Sections 4.1.1 and 4.2.2 for the pronunciation of in that); and in Token 39, there is no preposition of between couple and father (see Section 5.2.5), though we might note that couple is pronounced as [k‫ݞ‬p‫]ݜ‬, so L-vocalization may again be a contributory factor. Table 4.16. Miscellaneous misunderstandings involving [l] in the coda No. 2 19 22 27 39 75 103 107 115

Spk. FCh MHk MHk FMa MHk MHk MIn MIn MIn

List. FBr FMa FMa MHk FTw FTw FTw FTw FTw

Word also want the wall we call fell into place couple tile until now world still virgin

Heard as owns won the war record ?? ?? tire as you know war in the region

Context she also in that company then (.) i want the wall the same (.) we call today right all fell into place couple you know father like the tile okay ten year so until now meaning in the world at the time especially still virgin in

Semantic shift might be implicated in Token 103, as until now is used (instead of still) to refer to something that is currently true, a usage that is common in the English spoken in much of South-East Asia, including Singapore (Deterding 2007: 53), but a meaning for the phrase that may be unexpected for FTw. (For further discussion of this issue, see Section 5.1.4.) In fact, there is no L-vocalization in this token, as until has a clear-L at the end, probably influenced by the clear-L that tends to occur at the end of a syllable in spoken Malay (Clynes and Deterding 2011).

54

4. Pronunciation

In Tokens 19 and 22, the absence of [l] at the end of wall and call respectively seems to be the main issue. In Token 19, the absence of [t] in want may also be a contributory factor (as already discussed in Section 4.1.5), but it seems a minor factor. One issue in Token 22 might be that MHk elsewhere uses [w] in place of [r] (for example, at the start of reverse in Tokens 12 and 16, as discussed in Section 4.1.3), and it is possible that FBr has become accustomed to this and so she mistakenly hears the [w] at the start of we as [r] (see Section 4.6.1). However, the main issue concerning the pronunciation of we call is the absence of [l] at the end of call. Token 27 is the only instance in the Hk+Ma recording in which MHk fails to understand FMa rather than the other way round, but this token occurs because the phrase all fell into place is accompanied by laughter (see Section 4.5.5), and in fact L-vocalization does not occur in fell, as the [l] is used as a linking consonant in fell into. In Token 75, MHk is talking about maintaining his house in the USA, but even in this context, FTw hears tile as tire, partly as a result of Lvocalization. One factor here is that it sounds as if there might be two syllables in tile, and this bisyllabic pronunciation of the word is more typical of the triphthong in [taܼԥ] than the diphthong in [taܼl]. One could say, therefore, that syllabification might also be an issue (see Section 4.5.4). In Token 107, world is pronounced as [w‫ޝܮ‬G@ZLWKD>G@DWWKHHQGDQG the expected vowel quality for the NURSE vowel, but it is heard instead as war, so I conclude that the missing [l] is what causes this misunderstanding. Finally, in Token 115, there is no L-vocalization in still, as MIn tends to use a clear-L when it occurs at the end of words (just like with until in Token 103). The problem in Token 115 is with the pronunciation of virgin as [vܼԥd‫ܼݤ‬n]. This might be regarded as a spelling pronunciation, something that will be discussed in Section 4.3. In conclusion, there are four tokens in which L-vocalization is the main factor in the misunderstanding: Tokens 19, 22, 75 and 107. One might note that three of these tokens involve L-deletion rather than L-vocalization: in Token 19, wall is pronounced as [w‫ @ޝܧ‬in Token 22, call is pronounced as [k‫@ޝܧ‬DQGLQ7RNHQworld is pronounced as [w‫ޝܮ‬G@-HQNLQV (2000: 139) suggests that pronouncing dark-L as [‫ ]ݜ‬is not problematic; but perhaps omitting the sound entirely can sometimes give rise to misunderstandings. Before leaving dark-L, we might look at one token where dark-L is mistakenly heard when it does not exist. In Token 177, shown in Extract 4.14, FMa subsequently reported that she heard the first token of taught as told, though she managed to figure it out correctly by the time of the second token. (This token will be discussed again in Section 6.3.2.)

4.1. Consonants

55

Extract 4.14. Tw+Ma : 328 (Token 177) 1 2 3

FTw

so i love sitting by the fire drinking (.) whisky with my husband so (0.8) he taught me (1.2) well not taught me L¶GOHDUQLWP\VHOI

One possibility here is that FMa has accommodated to the widespread Lvocalization in the speech of FTw by this time (five and a half minutes from the start of their conversation); so the occurrence of L-vocalization elsewhere contributes to this token of misunderstanding. (Instances of listening accommodation such as this will be discussed in Section 4.6.1.) 4.1.7. Aspiration and voicing In English, the distinction between voiced and voiceless initial plosives is largely conveyed by means of aspiration. However, in other languages such as Malay, aspiration is less prominent, and initial voiceless plosives are more often unaspirated (Clynes and Deterding 2011). Does this cause problems when speakers of English are influenced by their home language so they produce voiceless plosives with little aspiration? The LFC proposals suggest that the distinction between voiced and voiceless consonants is important, and for word-initial plosives, it is best if it is conveyed by means of aspiration (Jenkins 2000: 140). Let us therefore analyse the extent to which lack of aspiration contributes to the occurrence of misunderstandings in the CMACE data. Table 4.17 lists nine tokens in which a voiceless plosive at the start of a syllable is heard as a voiced plosive. Table 4.17. Misunderstandings where a voiceless plosive is heard as a voiced one No. 4 37 84 96 101 105 108 109 179

Spk. FCh MHk MIn MIn MIn MIn MIn MIn FMa

List. FBr FTw FMa FTw FTw FTw FTw FTw FTw

Word Bhutan close academic correct inculcate Karl Marx cold war anti P6

Heard as ?? goes i got ?? great got Dalai Lama guru undie B6

Context india (.) and bhutan  LW¶V they close down the university meaning aCADemic or (.) go the correct one [MIn:name] i inculcate er disseminate er er (.) karl marx for example the (.) the cold war (.) so is basically anti religion so p p six p five

In Tokens 37, 84, 96, 101 and 108, [k] is heard as [ܳ@,n Token 105, initial [k] is heard as [d]. In Tokens 4 and 109, [t] sounds like [d], and as a result

56

4. Pronunciation

in Token 4 FBr is unable to make a guess about the word. And in Token 179, [p] is heard as [b]. One might note that all but the first and third of these tokens (Tokens 4 and 84) are mis-heard by FTw, who spent many years in the UK where initial voiceless plosives would be expected to be fully aspirated. It seems that lack of aspiration on voiceless plosives does not cause too many problems for other listeners, possibly because it is a widespread feature of pronunciation in South-East Asia. Acoustically, the degree of aspiration on a plosive can be estimated by measuring the Voice Onset Time (VOT), which represents the delay between the release of the plosive and the start of the voicing for the following sound (Ladefoged and Johnson 2011: 151), so a voiceless plosive is expected to have longer VOT than a voiced one. In British English, the VOT for [k] is usually about 70 msec, for [t] it is expected to be about 60 msec, and for [p] it would be approximately 50 msec (Docherty 1992: 25). In all the tokens listed in Table 4.17, the VOT for the plosive is substantially less than these values for British English, so the measurements confirm that the plosives have little aspiration: in Token 4 the plosive in the middle of Bhutan has a VOT of just 19 msec, in Token 37 the VOT for [k] is 29 msec, in Token 84 the VOT for the first [k] in academic is 20 msec, in Token 96 it is 23 msec, in Token 101 it is 32 msec, in Token 105 it is 29 msec, in Token 108 it is 18 msec, in Token 109 the VOT for [t] is 18 msec, and in Token 179 the VOT for [p] it is just 10 msec. However, in all cases apart from Tokens 4, 101 and 179, another feature of pronunciation has contributed substantially to the misunderstanding: in Token 37, there is no [l] in close (see Section 4.1.4); in Token 96, the final [kt] cluster in correct is pronounced as a glottal stop [‫( ]ݦ‬see Section 4.1.5); in Token 105, there is no [s] at the end of Marx (see Section 4.1.8); in Token 108, there is no [d] in cold (see Section 4.1.5); and in Token 109 the vowel at the end of anti is [aܼ], using the American pronunciation rather than the [i] that is expected in Britain (Wells 2008: 34 ± see Section 4.2.6). The issue with Token 4 is that FCh seems to be using the Chinese pronunciation of Bhutan, ᶵᷡ EGƗQ, and the medial consonant is [d] rather than [t]. It is therefore not surprising that FBr cannot understand what country FCh is referring to. (FBr subsequently confirmed that she knows about Bhutan, but she could not understand it when it was said by FCh.) Finally, lexical factors are implicated for Tokens 101 and 179: in Token 101, FTw does not know the word inculcate; and in Token 179, FMa assumes that FTw knows about the Singapore school system, where P6 refers to the sixth year of primary school, but in fact FTw is not familiar with this term. These two tokens will be discussed again in Sections 5.1.1 and 5.1.2.

4.1. Consonants

57

It seems, therefore, that aspiration on initial plosives does not cause much of a problem in the CMACE data from South-East Asia, except perhaps occasionally when FTw is the listener. But we need to be cautious in concluding that aspiration is not important in ELF interactions, as it is possible that it could result in more misunderstandings with speakers and listeners from elsewhere. As always, we need more data from a wider range of participants before drawing any firm conclusions about what is important for intelligibility in international settings and what is not. I will now analyse some voicing contrasts involving fricatives. Seven such tokens are listed in Table 4.18. In the last four of these (Tokens 54, 73, 83, and 99) a word with a voiced fricative is heard as a word with a voiceless one, and in the other three (Tokens 25, 41 and 43) the devoicing of a fricative may have contributed to the misunderstanding. Table 4.18. Misunderstandings involving the devoicing of a fricative No. 25 41 43 54 73 83 99

Spk. MHk MHk MHk MHk MHk MIn MIn

List. FMa FTw FTw FTw FTw FMa FTw

Word river measure zen zone over vagrant visit so

Heard as ?? ?? zen? so all full fragrant faces it

Context never step in the same river twice WKHUH¶VQRmeasure you know they call zen you know yeah the time zone you know is change LQWKHSDVWLW¶VoVER you know my grandfather (.) is vagrant you they visit so meanings like you

The first five of these tokens all involve MHk. In Token 25, he seems to pronounce river with a medial [f], so maybe this contributes to FMa being unable to guess the word. However, perhaps it is actually the context that is the main problem (see Section 5.4). In Token 41, measure is pronounced as [m‫ݕޝܮ‬ԥ], and it is probably the quality of the vowel in the first syllable rather than the use of [‫ ]ݕ‬instead of [‫ ]ݤ‬as the medial consonant that causes the problem (see Section 4.2.1). In Token 43, MHk pronounces zen as [s‫ܭ‬n], and FTw asks to confirm if he means zen. The open-mid quality of the vowel may also be an issue (see Section 4.2.6), but the voicing of the initial consonant seems to be the main problem. In Token 54, zone is pronounced as [s‫ܧ‬Ѻ‫@ޝ‬ZLWKDEDFNPRQRSKWKRQJUDWKHUWKDQ>o‫ ]ݜ‬and a nasalised vowel instead of a final [n]. It is hard to know which of these features of pronunciation contributes most to the misunderstanding: the voiceless initial fricative, the quality of the vowel, or the loss of the final nasal consonant. Token 73 also has multiple contributing factors: over is pronounced as [o‫ޖݜ‬f‫]ݜ‬, with stress apparently on the second syllable (to be discussed in Section 4.4.1). But one further issue is the grammar, so the wider context is

58

4. Pronunciation

shown in Extract 4.15. MHk is saying that people in Brunei once earned more than those in Singapore, with over PHDQLQJ µPRUH WKDQ¶ 8QIRUWunately, the over gets separated from Singapore by two tokens of you know, and this probably contributes to the misunderstanding. Token 73 will be discussed again in Section 5.2.7. Extract 4.15. Hk+Tw : 2240 (Token 73) Context: MHk is suggesting that people in Brunei used to be richer than elsewhere, but maybe they no longer are. 1 2 3

MHk

... brunei is you know (.) they said that you know in the past LW¶VoVER you know er (.) you know singDSRUH ZKHQZH¶UH WDONLQJDERXW\RXNQRZWKHSHRSOHLQFRPH\RXNQRZ«

Finally, in Table 4.18, we have Tokens 83 and 99, both said by MIn. In Token 83, vagrant is pronounced as [fr‫ܳݞ‬r‫ݞ‬n], with what sounds like a spurious [r] in the onset of the first syllable. However, the biggest problem is probably the semantic shift involved when vagrant is used to refer to someone who travels to different countries in pursuit of work, with no negative connotation intended for the word, unlike the way it is more commonly used. This token will be discussed again in Section 5.1.4. In Token 99, they visit so was transcribed by FTw as their faces it, and this arises because visit is pronounced as [fܼsܼ‫]ݦ‬, with both [v] and [z] being devoiced (as well as a final glottal stop in place of [t]). So in this case, the devoicing of the fricatives seems to have caused the problem. In conclusion, there are just a few instances in which devoicing a fricative, particularly in initial position, causes a problem. One might further note that none of the instances involve final fricatives, for example raise being heard as race. In reality, it is normal in all varieties of English for final [z] to be devoiced (Docherty 1992: 35), and there is no evidence from the CMACE data that this causes problems. A much bigger problem is if the fricative is omitted, and this will be discussed in the next section.

4.1.8. Fricatives and affricates In the previous section, I considered the devoicing of fricatives. In this section, I discuss further issues regarding fricatives and affricates, particularly their replacement with other sounds as well as their omission. There are four tokens involving use of [w] in place of [v], all by MHk. They are listed in Table 4.19.

4.1. Consonants

59

Table 4.19. Misunderstandings involving [w] in place of [v] No. 12 16 52 76

Spk. MHk MHk MHk MHk

List. FMa FMa FTw FTw

Word reverse reverse environment provide

Heard as we worse we worse that long to why

Context know er reverse (.) you know LW¶V reverse you know we control the environment the UBD provide you know

In both Tokens 12 and 16, in addition to the use of [w] for the medial [v], [w] occurs in place of [r] at the start of reverse (see Section 4.1.3). In Token 52, the words that are heard by FTw, that long, do not have [w], yet it still seems that the occurrence of [w] in place of [v] in environment is the main factor causing the misunderstanding. In Token 76, provide is pronounced as [po‫ݜ‬waܼ], so the absence of both [r] from the initial cluster and [d] at the end of the word are additional factors in giving rise to the misunderstanding (see Sections 4.1.4 and 4.1.9). Three tokens involve the occurrence of [s] in place of [‫]ݕ‬, all in the speech of MIn. They are listed in Table 4.20. Table 4.20. Misunderstandings involving [s] in place of [‫]ݕ‬ No. 85 118 125

Spk. MIn MIn MIn

List. FMa FTw FTw

Word cash dishes pressure

Heard as caste this is ??

Context consider as (.) non (.) cash subject different kind of er dishes except because of the (.) time er pressure

In Token 85, the unconventional use of non-cash to refer to a subject (such as literature) that does not earn much money is one issue (see Section 5.1.4), though the use of [s] in place of [‫ ]ݕ‬is also a substantial factor. (See also Section 4.2.2 for a discussion of the vowel in cash.) In Token 118, the use of [d] for voiced TH elsewhere by MIn may contribute to the problem (see Extract 4.7 in Section 4.1.1), but in dishes, the only unexpected feature of pronunciation is [s] for the medial fricative. In Token 125, the misunderstanding seems to arise solely because of the use of [s] in pressure. Five tokens involving the affricate [t‫ ]ݕ‬are listed in Table 4.21. Table 4.21. Misunderstandings involving [t‫]ݕ‬ No. 42 97 124 141 183

Spk. MHk MIn MIn MLs FTw

List. FTw FTw FTw FBr FMa

Word different church eitch niche lunchtime orchard

Heard as (deventray) it needs runtime orchids

Context (.) er different church so eitch is not pronounce they are giving us niche break or the lunchtime like a apple orchids LVQ¶W

60

4. Pronunciation

In Token 124, niche is pronounced as [nܼt‫]ݕ‬, using what might be classified as a spelling pronunciation (see Section 4.3); but FTw does not in fact know the word niche, so this token should actually be classified as lexical rather than phonological (see Section 5.1.1). In Token 183, FTw intends to say orchard but she says orchids instead, getting the two words confused. In fact, FMa guesses the meaning correctly, but then asks for confirmation, which is why it is regarded as an instance of misunderstanding. This is an issue with word selection, so it will be classified as a lexical issue and discussed in Extract 5.1 in Section 5.1.1. In the other three tokens in Table 4.21, the omission of [t‫ ]ݕ‬appears to be the main issue. In Token 42, there is no [t‫ ]ݕ‬at the end of church, and though the absent [r] in different also contributes to the problem (see Section 4.1.4), the missing [t‫ ]ݕ‬is probably more serious. In Token 97, eitch WKHOHWWHUµK¶ ends with [t] rather than [t‫]ݕ‬. In addition, the FACE vowel is pronounced as [ܼ] (see Section 4.2.3), but [t] instead of [t‫ ]ݕ‬at the end of the word is probably the main issue. Finally, use of [r] at the start of lunchtime in Token 141 is a contributory factor, but the absence of [t‫ ]ݕ‬is probably the bigger problem. Classification of Token 141 is discussed in Section 4.7. There are five tokens involving [d‫]ݤ‬. They are listed in Table 4.22. Table 4.22. Misunderstandings involving [d‫]ݤ‬ No. 47 86 87 127 136

Spk. MHk MIn MIn FJp MLs

List. FTw FMa FMa FBr FBr

Word Heard as collage come rough by the end of the day biology Berlitz ?? major ?? young junk

Context collage on the top by the end of the day was taken from Berlitz my major is literature before i was young (.)

In Token 47, collage ends with [d‫ ]ݤ‬rather than the expected [‫ ;]ݤ‬but FTw does not know this word anyway, so this token should be classified as lexical. In Token 87, Berlitz LVSURQRXQFHGDV>‫ޖ‬E Oܼd‫]ݤ‬, which may be a spelling pronunciation (see Section 4.3). In fact, FMa is not familiar with the use of English-learning materials from Berlitz, so again this should be classified as lexical. Both these tokens are discussed in Section 5.1.1. The other three tokens in Table 4.22 are all primarily phonological. In Token 86, MIn says by the end of the day as [baܼ‫ܥ‬lԥd‫ݤ‬eܼ], so the use of [d‫]ݤ‬ instead of [d] at the start of day is probably the biggest problem, though the pronunciation of end is a contributory factor (see Section 4.1.5). In Token 127, the use of [‫ ]ݤ‬instead of [d‫ ]ݤ‬as the medial consonant in major is what causes the misunderstanding, though presumably the occurrence of [r] at the start of literature compounds the problem (see Section 4.1.3). And in

4.1. Consonants

61

Token 136, the use of [d‫ ]ݤ‬at the start of young is the main issue, though in addition when i was young would be a more standard way of saying it. Finally, there are four tokens involving the omission of [s] or [z] from the coda of a syllable. They are listed in Table 4.23. Table 4.23. Misunderstandings involving omission of [s] or [z] in the syllable coda. No. 30 53 105 139

Spk. MHk MHk MIn MLs

List. FTw FTw FTw FBr

Word class because Karl Marx just

Heard as car at Dalai Lama their

Context we not skip the class you know and because the time zone er (.) karl marx for example i think (.) just meat. is nothing

In Token 30, class is pronounced as [k‫ @ޝܤ‬VR LW LV KDUGO\ VXrprising that FTw hears it as car. Classification of this token is discussed in Section 4.7. In Token 53, because is pronounced as [bܼkԥ], and FTw is probably using its context before the time in guessing the word as at, especially when she also hears zone as so (see Section 4.1.7). In Token 105, Karl Marx is pronounced as [k‫ޝܤ‬OP‫ޝܤ‬N@VRWKHRPLVVLRQRIWKHILQDO>V@LVWKHFUXFLDOIDFWRU And in Token 139, just is pronounced as [d‫]ݞݤ‬, so the missing [s] is key. 4.1.9. Other missing consonants There remain a few tokens involving missing final plosives and other sounds. Four tokens involving omitted final [k] are listed in Table 4.24. Table 4.24. Misunderstandings involving omitted word-final [k]. No. 31 56 63 112

Spk. MHk MHk MHk MIn

List. FTw FTw FTw FTw

Word Tech plastic plastic ethnic

Heard as there past past other

Context study in texas tech you know how much plastic you use to digest this plastic container have many ethnic GROUPS

In Token 31, there is a glottal stop at the end of Tech, but FTw fails to detect this and she hears the word as there. Another problem is with the name Texas Tech, which she is not familiar with (see Section 5.1.1). In both Tokens 56 and 63, the main issue is the absence of the [l] in the initial cluster in plastic (as discussed in Section 4.1.4). Actually, the absence of [k] in Token 63 is not surprising given that the next word begins with [k]. Similarly, in Token 112, the omitted [k] in ethnic is to be expected given the [ܳ] at the start of groups. In fact, the key issue with Token 112 is the missing [n], so it is listed again in Table 4.26 later in this section.

62

4. Pronunciation

Five tokens involve missing word-final [t] or [d] (other than those in a cluster ± see Section 4.1.5). They are shown in Table 4.25. Table 4.25. Misunderstandings involving omitted final [t] or [d]. No. 58 68 76 88 146

Spk. MHk MHk MHk MIn MLs

List. FTw FTw FTw FMa FBr

Word garden spade add UBD provide spirit hot

Heard as gardens where air you need to why speed ??

Context the garden spade you meal you know add (.) the UBD provide you the spirit of learning RK\RXGRQ¶Who:t? @

In Tokens 58 and 68, the missing [d] in spade and add respectively seems to be the main issue, though in addition in Token 68, the grammar is a bit odd (see Section 5.2.7). In Token 76, the omission of [r] in provide and the use of [w] instead of [v] both contribute to the problem (see Sections 4.1.4 and 4.1.8). In Token 88, perhaps the biggest problem is the fast speaking rate, with the five syllables in spirit of learning said in less than 0.6 seconds (see Section 4.5.3). Finally, in Token 146, [t] in fact occurs at the end of hot, but it is delayed and partly lost in the following laughter (see Section 4.5.5). However, the main issue with this token seems to be grammatical, with no main verb like feel in the sentence \RXGRQ¶WKRW (see Section 5.2.4). That leaves the four miscellaneous tokens involving omitted consonants that are listed in Table 4.26. In Token 110, there is no [b] in Begawan; in Token 112, there is no [n] in ethnic (though in fact, this word is rather quiet, so it is hard to tell) and also no final [k] (as discussed above); in Token 163, there is no [n] in seven; and in Token 176, there is no [p] in leap. Table 4.26. Miscellaneous tokens involving consonants. No. 110 112 163 176

Spk. MIn MIn MNg FTw

List. FTw FTw FBr FMa

Word Seri Begawan ethnic seven leap forward

Heard as Sudar other same with ??

Context bandar seri begawan dialect have many ethnic GROUPS hold (.) seven degrees a great (.) leap forward

In three of these tokens, there is another contributory factor. In Token 110, the phrase Bandar Seri Begawan (the capital city of Brunei) is spoken rather quickly, with the seven syllables spoken in just 0.88 seconds, presumably in the expectation that FTw will understand it as she currently lives there, so the main problem is the fast speaking rate (to be discussed in Section 4.5.3). In Token 163, the main problem probably lies with the previous word hold, which FBr hears as old (see Section 4.1.2). And in Token

4.1. Consonants

63

176, the use of the idiom great leap forward to describe a sudden big advance LQRQH¶VSHUVRQDONQRZOHGJHLVunexpected for FMa. This use of the idiom will be discussed in Section 5.1.3. Finally, there is one further token that involves the listener not hearing a final consonant even though it is actually there. In Extract 4.16, FCh hears jury instead of to read, but it is uncertain why she does not hear the final [d], because it seems to be reasonably clear in the recording. Extract 4.16. Ch+Br : 2853 (Token 11) 1 2 3 4

FBr FCh FBr FCh

so erm (.) speaking of books. do you like to read? jury? do you like to read er yes (.) i do @@

Perhaps instead it is the quality of the initial [t] in to that is problematic, as FCh hears it as the affricate [d‫]ݤ‬. When FBr repeats to read in much the same fashion, FCh understands it correctly. 4.1.10. Consonants: summary At the end of this chapter, I will try to estimate the main phonological factor which causes each token of misunderstanding discussed in the chapter; but first let me summarize the findings from the analysis of consonants. The biggest issues seem to be the replacement of consonants such as [l], [r], [n] and [v], the omission of [l] and [r] from initial clusters, and Hdropping or insertion. In contrast, the realization of the TH sounds and the omission of [t] and [d] from word-final clusters rarely cause problems. Lvocalization is seldom a problem, though complete omission of a dark-L may sometimes be. All these findings are in line with the LFC suggestions of Jenkins (2000). However, I have found few issues with lack of aspiration on initial voiceless plosives, and this is not consistent with the LFC proposals, though it should be acknowledged that unaspirated initial voiceless plosives might be more problematic for listeners from elsewhere. I will now consider the extent to which the quality and duration of vowels contribute to misunderstandings.

4.2. Vowels The LFC proposals suggest that, while vowel quantity (the distinction between long and short vowels) is important, vowel quality (apart from the

64

4. Pronunciation

vowel) is not important for maintaining intelligibility, and initially, this seems rather surprising. However, one should remember that variation in vowel quality is widespread in Englishes around the world, so for example people in the north of England tend to pronounce STRUT as [‫ ]ݜ‬rather than [‫]ݞ‬, most varieties of American English have [‫ @ޝܤ‬UDWKHU WKDQ>‫ @ܥ‬for LOT and they also have [æ] rather than [‫@ޝܤ‬IRU BATH, speakers in Australia usually have a wide diphthong for FACE and GOAT, people in New Zealand typically have a close quality for DRESS and a central vowel for KIT, and those in Scotland and Wales often have monophthongs for FACE and GOAT (Wells 1982). In other words, listeners are accustomed to hearing substantial variability in vowel quality, in contrast with the relative stability of pronunciation for English consonants around the world, and so they can deal with shifts in vowel quality more easily than variation in the pronunciation of consonants. But to what extent does variability in vowels give rise to misunderstandings in ELF interactions? Many of the misunderstandings in the CMACE data involve the word that is heard having a different vowel from the intended word; yet it is not clear that variation in vowel quality is what has caused the misunderstanding. Listing all the tokens would be overwhelming, so here I will just focus on the quality of three vowels: NURSE, TRAP, and FACE, before I consider vowel duration and vowel reduction. Finally, I will analyse a few miscellaneous tokens in which vowel quality seems to be a substantial factor. NURSE

4.2.1. NURSE There are seven tokens involving the NURSE vowel, and they are shown in Table 4.27. In Token 41, the vowel in the first syllable of measure is pronounced as [‫@ޝܮ‬so we can say that NURSE rather than the expected DRESS occurs in the first syllable of this word, while in the other six tokens, the NURSE vowel is heard as something else. Table 4.27. Misunderstandings involving NURSE. No. 14 41 42 107 115 121 129

Spk. MHk MHk MHk MIn MIn MIn FJp

List. FMa FTw FTw FTw FTw FTw FBr

Word terminal measure different church world still virgin return early morning

Heard as common ?? (deventray) war region leave alimony

Context f- (.) terminal degree (.) WKHUH¶VQRmeasure you (.) er different church in the world at the time especially still virgin in i REturn to jakarta i very early morning

4.2. Vowels

65

In three of these tokens, it is the vowel that seems to cause the problem: in Token 41, measure is pronounced as [m‫ݕޝܮ‬ԥ], and the vowel is probably the main issue rather than the voiceless fricative (see Section 4.1.7). In Token 115, the vowel in virgin is pronounced as [ܼԥ] in what might be called a spelling pronunciation (to be discussed in Section 4.3), and this is why the word is heard as region. And in Token 129, early is said with an open vowel in the first syllable, and FBr is unable to understand it. In the other four tokens, something else seems to be a bigger issue. In Token 14, the phrase terminal degree is unfamiliar to FMa, so this is a lexical issue (see Section 5.1.2). In Token 42, church has no [t‫ ]ݕ‬at the end (see Section 4.1.8). In Token 107, the vowel in world is in fact the expected [‫@ޝܮ‬ and the only non-standard feature is the omission of the [l] (see Section 4.1.6). In Token 121, return is stressed on the first syllable (to be discussed in Section 4.4.1), and this is why it is heard as leave. In conclusion, the quality of NURSE occasionally causes problems; but there seem to be only three tokens of misunderstanding arising from the quality of this vowel in the six and a half hours of conversational data. 4.2.2. TRAP In the ACE data, there are two ways in which the pronunciation of TRAP may deviate from the [æ] that would be expected in most varieties of British or American English: it may be pronounced as a less open vowel, [‫]ܭ‬, in which case it may be merged with DRESS, as is typical in places such as Singapore, Malaysia and India (Deterding 2007; Baskaran 2004a; Gargesh 2004); or it may be pronounced as a less front vowel such as [‫]ݞ‬, in which case it may be merged with STRUT, as is typically found in Nigerian English (Gut 2004). The LFC proposals suggest that this should not matter, as TRAP, DRESS and STRUT are all short vowels, so the pronunciation only involves vowel quality, not quantity. But let us see the extent to which misunderstandings involving TRAP occur in the CMACE data. Table 4.28 lists the seven tokens in which TRAP is understood as a less open vowel. It is mostly heard as DRESS, but in Token 15, it is heard as KIT. Tokens 2 and 10 are both by FCh. In Token 2, also in that is pronounced as [‫ޝܧ‬VR‫ܼݜ‬n‫]ݦܭ‬, so in addition to the open-mid vowel in that, there is L-vocalization in also and omission of voiced TH in that; but in fact the problem probably arises because of the absence of a verb, so the main issue is classified as grammatical rather than phonological (see Section 5.2.4). In Token 10, the [l] in black is pronounced as [r] (see Section 4.1.3), and this seems to be the main problem, so the quality of the vowel is a minor issue.

66

4. Pronunciation

Table 4.28. Misunderstandings involving TRAP heard as a less open vowel. No. 2 10 15 40 60 67 144

Spk. FCh FCh MHk MHk MHk MHk MLs

List. FBr FBr FMa FTw FTw FTw FBr

Word also in that Black Swan matter what Italian wrapping meal plan Man U

Heard as owns an internet (rex one) middle war i tell you weapon view pen menu

Context she also in that company saw the (.) black swan in no matter what they you know italian father paper wrapping stuff the meal plan (.) just VLULGRQ¶WOLNHman u

Tokens 15, 40, 60 and 67 are all spoken by MHk. In Token 15, the vowel in the first syllable of matter seems to be [ܼ], but in addition the flapped consonant in the middle of matter and the glottal stop at the end of what are contributory factors in the phrase being heard as middle war. In Token 40, Italian is pronounced as [ܼt‫ܭ‬lܼn], and it is probably the use of [ܼ] in place of [ܼԥ] in the final syllable that is the most important factor in the misunderstanding (see Section 4.2.6). In Token 60, [w] instead of [r] at the start of wrapping seems to be the biggest problem (see Section 4.1.3). And in Token 67, the absence of [l] in plan is the main phonological cause of the misunderstanding (see Section 4.1.4), though we should also note that the phrase meal plan is not familiar to FTw (see Section 5.1.2). Finally, there is Token 144, in which Man U (referring to the English football team, Manchester United) is understood as menu by FBr, and the quality of the vowel in the first syllable is the only feature of pronunciation that contributes to the misunderstanding. However, in this case, we need to consider the wider context, which is shown in Extract 4.17. Extract 4.17. Ls+Br : 1853 (Tokens 144 & 145) Context: MLs is talking about experiences eating in an Indian restaurant. 1 2 3

MLs FBr

oh everybody open your PHQX RKQRVRUU\VLULGRQ¶t like man uHUP VR\RXVHH QRLGRQ¶WWDONDERXWWKHfootball @@@@

In Extract 4.17, MLs is actually making a joke about someone confusing menu and Man U, and he seems to be putting on an Indian accent at this point, in imitation of the pronunciation of the waiter. So it seems that use of [‫ ]ܭ‬is part of his attempted Indian accent. Although FBr laughs at the end, in fact she does not understand the joke, partly because she does not understand football, which is pronounced with a final [n] (see Section 4.1.3). In summary, pronouncing TRAP as [‫ ]ܭ‬does not seem to cause too many difficulties for listeners, though pronouncing it as [ܼ] (in matter) might be

4.2. Vowels

67

more problematic. These findings therefore support the claims of Jenkins (2000) that small shifts in vowel quality do not matter. Let us now consider misunderstandings involving TRAP heard as a more back vowel. There are six such tokens, and they are shown in Table 4.29. Table 4.29. Misunderstandings involving TRAP heard as a more back vowel. No. 84 85 91 109 161 168

Spk. MIn MIn MIn MIn MNg MNg

List. FMa FMa FMa FTw FBr FBr

Word academic cash pattern anti has rise handle

Heard as i got ?? caste part of undie are try undo

Context profession meaning aCADemic or consider as (.) non (.) cash subject LW¶VQRWDVDVpattern this and so is basically anti religion some of them has rise to the way you have to handle me if

The first four of these are all spoken by MIn. In Token 84, the main issue seems to be that the stress is on the second syllable of academic rather than the penultimate syllable (see Section 4.4.1). In addition, there is little aspiration on the first [k] in academic (see Section 4.1.7). In Token 85, the vowel in cash is actually pronounced as [æ], so vowel quality is not an issue. In this case, there are two contributory factors: the final consonant is pronounced as [s] instead of [‫( ]ݕ‬see Section 4.1.8); and FMa is unfamiliar with the use of non-cash to describe an unfashionable subject (see Section 5.1.4). In Token 91, pattern is pronounced as [p‫ޝܤ‬WԥQ@VRWKHYRZHOseems to be the main pronunciation issue, though we should note that [‫ @ޝܤ‬LV D long vowel, so maybe vowel length rather than quality is key (see Section 4.2.4). In addition, FMa is unfamiliar with the use of pattern in this context to describe something that is repeated and predictable (see Section 5.1.4). And in Token 109, the main problem is that FTw is confused by the pronunciation of anti with [aܼ] instead of [i] at the end (see Section 4.2.6), and lack of aspiration on the [t] may also be an issue (see Section 4.1.7). Tokens 161 and 168 are both spoken by MNg, and the omission of the initial [h] appears to be the main issue (see Section 4.1.2), though the back quality of the vowel is probably a contributory factor. It seems, therefore, that TRAP being pronounced as [‫ ]ܭ‬does not cause a problem in South-East Asia, where this pronunciation is widespread. However, if it is pronounced with a more back quality, this may occasionally contribute to a loss of intelligibility, because this pronunciation occurs less widely in the region. This concurs with the observation of Deterding and Kirkpatrick (2006), as already mentioned regarding TH, that variations in pronunciation that occur widely in the region do not generally cause problems, but unexpected features of pronunciation may be more troublesome.

68

4. Pronunciation

4.2.3. FACE One pattern that seems to occur sometimes in the data is FACE being heard as a close monophthong. Six such tokens are listed in Table 4.30. In Token 17, the vowel in the second syllable of afraid is heard as FLEECE, while in the other five tokens, FACE is heard as KIT. Table 4.30. Misunderstandings involving FACE heard as a close monophthong. No. 17 65 78 97 166 173

Spk. MHk MHk MHk MIn MNg MNg

List. FMa FTw FTw FTw FBr FBr

Word afraid related rain forest eitch take logic and hot cake

Heard as are feel we live wind forced it think logical out kick

Context everybody afraid about art is related with truth you know rain forest is so eitch is not pronounced take logic and thinking because they are hot cake

In all these tokens, another factor has contributed to the misunderstanding. In Token 17, [r] is omitted from afraid and there is also no verb in the sentence (see Sections 4.1.4 and 5.2.4); in Tokens 65 and 78, [w] occurs at the start of related and rain respectively (see Section 4.1.3); in Token 97, [t] rather than [t‫ ]ݕ‬occurs at the end of eitch (see Section 4.1.8); in Token 166, take LVSURQRXQFHGDV>WH‫ ]ݦޝ‬with a glottal stop rather than [k] at the end (but note that FBr actually hears a word with a final [k]); and in Token 173, there is no [h] at the start of hot and furthermore the phrase hot cake as an idiom for people who are in demand from employers is not familiar to FBr (see Sections 4.1.2 and 5.1.3). Perhaps only in Token 166 is the close quality of the FACE vowel a major factor in causing the misunderstanding. 4.2.4. Vowel length Jenkins (2000: 144) suggests that vowel length is important, particularly the distinction between long and short vowels but also the difference in duration that signals the contrast between voiced and voiceless final consonants. For example, the vowel in bid is longer than that in bit because the [t] in bit shortens the preceding vowel (Roach 2009: 28). But what evidence is there in the CMACE data that the absence of such distinctions in vowel length causes misunderstandings? Aside from the items listed in the previous section (where confusion between FACE and KIT may be seen as one of length as well as quality), there are just three tokens in which FLEECE is heard as KIT, and a single case

4.2. Vowels

69

(Token 114) where KIT is heard as FLEECE, and there do not seem to be any other obvious cases of confusion between a long/short pair of vowels (apart from Token 91, in which pattern is heard as part of, as discussed in Section 4.2.2). These four tokens are shown in Table 4.31. Table 4.31. Misunderstandings involving confusion between FLEECE and KIT. No. 114 119 152 155

Spk. MIn MIn MNg MNg

List. FTw FTw FBr FBr

Word living senior weed all means

Heard as leading similar with (home is)

Context i am: er the living old- eldest and then er (.) more senior friend have to go and weed the something to know something by all means

In Token 114, living is pronounced as [lܼvܼƾ@ZLWKZKDWVRXQGVOLNH>ܼ] in the first syllable and [v] as the medial consonant, so it is not clear what causes the problem. It is actually probably word order that is the issue in this token, as i am the living eldest is an unexpected way of saying that he is the oldest of his siblings still alive (see Section 5.2.6). In both Tokens 119 and 152, the only problem seems to be that FLEECE sounds like [ܼ], which is why senior is heard as similar in Token 119 and weed is heard as with in Token 152. In Token 155, in addition to means being heard as is, there are two other issues: all starts with what sounds like [h], and there is no [l] at the end of the word (see Sections 4.1.2 and 4.1.6). These are probably the main issues with this token. The distinction between FLEECE and KIT involves long and short vowels. In addition, we should look at the distinction between word-final voiced and voiceless consonants which, as mentioned above, may be signalled by the duration of the vowel. There is just one token in this category. In Token 28, shown in Extract 4.18, FTw hears need as meet. In this Token, need is pronounced as [ni‫]ݦ‬, and perhaps the short duration of the vowel is a factor in FTw hearing the final consonant as [t] rather than [d]. Extract 4.18. Hk+Tw : 11 (Token 28) Context: MHk is talking about his name, particularly his lack of a middle name. 1 2

MHk

but \RXNQRZDVWKHUHLGRQ¶WUHDOO\\RXNQRZI- need you know the i mean the middle

In conclusion, there seem to be few tokens in which vowel length is a factor in causing a misunderstanding. But we need to be cautious here: there are no vowel length distinctions in most of the indigenous languages of South-East Asia such as Malay or Chinese, and furthermore it is com-

70

4. Pronunciation

mon in local varieties of English such as those of Brunei, Singapore and Malaysia for long and short vowels to be merged (Salbrina 2006; Deterding 2007; Baskaran 2004a). It seems, therefore, quite likely that vowel length distinctions have less impact on the intelligibility of English in South-East Asia than in some other parts of the world, and a wider range of data should be analysed before concluding that vowel length does not play a substantial role in intelligibility in ELF. Jenkins (2000) suggests that vowel length is important, but her data involved listeners from Japan, and vowel length in Japanese is crucial in indicating if there is one mora or two in a syllable. 4.2.5. Vowel reduction The use of reduced vowels in the weak form of function words and the unstressed syllables of polysyllabic words is a key feature of most InnerCircle Englishes. Field (2008) claims that the weak forms of function ZRUGV SURYLGH WKH µPRUWDU¶ WKDW KROGs WRJHWKHU WKH µEULFNV¶ RI WKH FRntent words, and this suggests that appropriate use of reduced vowels in function words may be important when conversing with speakers in the Inner Circle. But what evidence is there that this is also true for ELF interactions? The use of weak forms is excluded from the LFC (Jenkins 2000: 146), suggesting that ELF speakers do not need to learn this feature of pronunciation, except of course in order to understand native speakers. So, in the CMACE data, does the absence of reduced vowels ever cause misunderstandings? There are three tokens in which a full vowel instead of [ԥ] may be one factor. They are shown in Table 4.32. Table 4.32. Misunderstandings involving lack of vowel reduction. No. 17 156 182

Spk. MHk MNg FMa

List. FMa FBr FTw

Word afraid you attend a agenda

Heard as are feel ?? agent now

Context everybody afraid about the you attend a (.) brunei school the main agenda would be to

There are other issues in Token 17. There is no [r] in afraid (see Section 4.1.4), and there is no verb in everyone afraid about (see Section 5.2.4). The full [‫ ]ݞ‬vowel in the first syllable seems, therefore, to be a minor issue. In Token 156, attend a is pronounced as [‫ݞ‬t‫ܭ‬nd‫]ݞ‬, so it seems that the full vowel at the start of attend and also for the indefinite article is the source of the problem. Similarly, in Token 182, agenda is pronounced as [ed‫ܭݤ‬nd‫ @ޝܤ‬ZLWK D IXOO YRZHO LQ ERWK WKH ILUVW DQG WKLUG V\OODEOH VR WKLV seems to be the main problem. Alternatively, one could say that there is no

4.2. Vowels

71

clear stress on the second syllable in either of these tokens, so perhaps stress placement might be identified as a key issue. Word stress will be discussed in Section 4.4.1. There therefore seems to be little evidence from the CMACE data that absence of vowel reduction causes too many problems, and this supports the exclusion of weak forms from the LFC. In her interviews of 17 international students at a language school in the UK, Fowler (2010) reports that 15 of them felt that it was not necessary to learn vowel reduction, so it seems that this aspect of the LFC proposals may be easily accepted by ELF speakers. Indeed, one could make a strong case that avoidance of vowel reduction actually enhances intelligibility in an international context, and many speakers of English around the world find it difficult to parse utterances by Inner-Circle speakers with their extensive use of reduced vowels. One might further note that the UK Civil Aviation Authority specifies that seven, hundred and thousand all be said with no vowel reduction in the second syllable for the crucial domain of air traffic communication (CAA 2012: ch. 2, p. 3). Apparently they believe that avoiding vowel reduction enhances intelligibility in an international setting. 4.2.6. Miscellaneous vowels There remain eight miscellaneous tokens in which the quality of the vowel may be a substantial factor in the misunderstanding, but which do not fit into any of the categories already discussed. They are listed in Table 4.33. Table 4.33. Miscellaneous misunderstandings arising from vowel quality. No. 18 40 43 79 89 100 109 120

Spk. MHk MHk MHk MHk MIN MIn MIn MIn

List. FMa FTw FTw FTw FMa FTw FTw FTw

Word Heard as Context communist companies then the communist you know Italian i tell you you know italian father you zen zen? they call zen you know yeah heart hot in the heart of you know this Cairo ?? i went to cairo a little bit now no master then now work anti undie is basically anti religion and going native good to eat is er going native meaning of

In Token 18, communist is misunderstood as companies partly because of the occurrence of [‫ ]ݞ‬rather than [‫ ]ܥ‬in the first syllable. Although it is true that [‫ ]ܥ‬would not be expected in an American accent, in fact MHk uses [‫]ܥ‬ for the LOT vowel elsewhere, for example in the first syllable of propaganda just three seconds later; so it seems that this occurrence of [‫ ]ݞ‬in

72

4. Pronunciation

communist may have confused FMa. As already discussed in Sections 4.1.4 and 4.1.5, the absence of [j] and also the omission of the final [t] may also have contributed to this misunderstanding, but it seems that the vowel in the first syllable is probably the main issue. In Token 40, Italian is pronounced as [ܼt‫ܭ‬lܼn]. Though the vowel in the second syllable is heard as DRESS rather than TRAP (see Section 4.2.2), it is probably the [ܼ] rather than the expected [ܼԥ] in the final syllable that is the key issue in this misunderstanding. The lack of an -s on father may be another factor, but this seems to be minor (see Table 5.7 in Section 5.2.1). In Token 43, zen has an open-mid vowel [‫ ]ܭ‬that sounds as if it might be TRAP. In addition, the initial consonant is [s] (see Section 4.1.8), and this is probably the key factor in prompting FTw to ask for clarification. In Token 79, heart is said as [h‫ޝܤ‬W@ as expected. It may be that FTw hears it as hot because that is how hot is pronounced in the USA; but as just discussed, MHk tends to use [‫ ]ܥ‬for LOT elsewhere (though not in communist). One issue here is that FTw fails to understand rain forest said just afterwards, hearing wind forced instead, so the misunderstandings are compounded. In Brunei, there is a Heart of Borneo initiative to protect the rain forest, but if you do not hear rain forest, maybe heart does not make sense. In Token 89, Cairo is pronounced as [kær‫]ݜ‬, so maybe it is the vowel in the first syllable that causes the problem. However, another factor is the fast speaking rate for the words i went to cairo (see Section 4.5.3). In Token 100, it is hard to tell why FTw hears now as no. The wider context is shown in Extract 4.19. Maybe the combination of then plus now is confusing, or else the lack of -s on work, so the main issue for this token should probably be classified as grammatical (see Section 5.2.3). Extract 4.19. In+Tw : 609 (Token 100) Context: MIn is talking about what his children are doing. 1 2 3 4

MIn FTw MIn

the oldest one er in er (.) what do you call canada actually study? yeah er after completing master then now work (.) so meaning LW¶V#LW¶VYHU\GLIILFXOWWRR@ «

In Token 109, anti ends with [aܼ], using an American pronunciation instead of the [i] expected in Britain (Wells 2008: 34), and this confuses FTw, who spent nine years in the UK. In addition, aspiration on the [t] and the quality of the first vowel may also be factors (see Sections 4.1.7 and 4.2.2). Finally, in Token 120, native is pronounced as [nætܼv], and this is why FTw fails to understand it. Her transcription of going native as good to eat was probably just a wild guess. However, it turns out that FTw is not famil-

4.2. Vowels

73

iar with the phrase going native, used here by MIn to refer to eating local food while he was living in Egypt. So we might classify this as a lexical issue (to be discussed in Section 5.1.2). 4.2.7. Vowels: summary In many tokens, the words that are understood have different vowels from the words that are intended. However, the main issue seems only rarely to be related to the quality or length of the vowels. The conclusion that the quality of vowels seldom causes problems is consistent with the LFC proposals; but the finding that few misunderstandings are caused by a lack of distinction between long and short vowels does not concur with the LFC proposals, as Jenkins (2000: 144) claims that vowel length is important. It is possible that vowel length is more important outside South-East Asia.

4.3. Spelling pronunciation One common stimulus for language change is that the pronunciation of words may shift to reflect their spelling, especially in the modern age in which virtually everyone is literate (Algeo 2005: 46). For example, forehead used to be pronounced as [f‫ܥ‬rܼd] (so it rhymed with horrid), but now it is generally [f‫ޝܧ‬KHG@HVSHFLDOO\DPRQJ\RXQJSHRSOH and Wells (2008: 317) reports WKDWRIµ\RXQJHU¶VSHDNHUVLQ%ULWDLQSUHIHUWKHSURQXQFLation with an [h], indicating that it is becoming the norm. Similarly, often was once pronounced as [‫ܥ‬fԥn] (with no [t]), but now it is sometimes pronounced as [‫ܥ‬ftԥn], though this new pronunciation is not yet so widely established, with only 27% of speakers in Britain and 22% in America preferring the word with [t] (Wells 2008: 560). These are trends that are found in Inner-Circle Englishes. But the additional occurrence of spelling pronunciation is widespread in new varieties of English around the world. For example, in places like Brunei and Singapore, salmon tends to have an [l] (Deterding and Salbrina 2013: 42). In fact, from the perspective of making oneself understood, use of [l] in salmon almost certainly enhances intelligibility in South-East Asia, and anyone who goes to the shop and asks for [sæmԥn] may be met with a blank stare. Although well-established examples of spelling pronunciation such as [l] in salmon seldom result in misunderstandings in the region, and in this case the absence of the [l] is more likely to cause confusion, it seems that idiosyncratic spelling pronunciation may cause problems. And there are a few

74

4. Pronunciation

instances like this, all in the speech of MIn. The reason seems to be that he has a sophisticated knowledge of complex written English, but he may not have heard some of the words spoken. Four tokens that might be classified as spelling pronunciations are listed in Table 4.34. Table 4.34. Misunderstandings arising from spelling pronunciation. No. 82 87 115 124

Spk. MIn MIn MIn MIn

List. FMa FMa FTw FTw

Word tubers Berlitz still virgin niche

Heard as (tabrus) ?? in the region needs

Context local roots namely tubers taken from berlitz actually especially still virgin in the so they are giving us niche

In Token 82, tubers (referring to root vegetables) is pronounced [t‫ݞ‬b‫ݜ‬s], in Token 87 Berlitz is pronounced as [bælܼd‫]ݤ‬, and in Token 115 virgin is pronounced as [vܼԥd‫ܼݤ‬n], so in these three cases, the pronunciation seems to have been influenced by the spelling (though for Berlitz, it is hard to see where [bælܼd‫ ]ݤ‬comes from). Token 124, in which niche is pronounced as [nܼt‫]ݕ‬, is a little different. According to Wells (2008: 541), 5% of people in Britain prefer [nܼt‫ ]ݕ‬inVWHDG RI >QL‫ ]ݕޝ‬for the pronunciation of this word, but in fact [nܼt‫ ]ݕ‬is the standard pronunciation in the USA. So we might say that in this case MIn is adopting the American pronunciation of the word. In any case, in Tokens 87 and 124, FMa and FTw respectively do not know the target words, so the main problem with these two is assumed to be lexical (see Section 5.1.1). The other two tokens, tubers and virgin, are classified as involving idiosyncratic spelling pronunciation.

4.4. Stress Stress can be divided into two kinds: word stress, or the syllable in a polysyllabic word which receives the most prominence; and utterance stress, otherwise known as intonational nucleus placement, referring to the focus of information within an utterance. These two kinds of stress will be discussed separately. 4.4.1. Word stress Jenkins (2000: 150) describes word stress DVD³JUH\DUHD´,Wseems to be crucial for many native speakers, but it might be excluded from the LFC as it appears not to be so important for intelligibility in ELF contexts.

4.4. Stress

75

This may seem surprising, as many would expect word stress to be of central importance, and everyone appears to be able to think of anecdotes when unexpected stress caused a problem. For example, I remember listening to a speech by a government minister in Singapore in which I heard CIGar with initial stress as seagull, and also IMportant with initial stress as impotent. Similarly, Pointon (2012) notes that he was confused by DEBacle said with initial stress by the UK politician David Miliband. However, this is anecdotal evidence involving listeners from the Inner Circle, and it is entirely possible that word stress plays a less important role in ELF contexts. In Section 4.2.5, I discussed lack of vowel reduction, and it is likely that word stress is not very salient if most of the syllables in a word have a full vowel rather than a reduced vowel. But let us consider the extent to which word stress may be implicated in tokens of misunderstanding in the CMACE data. Six such tokens are listed in Table 4.35. Table 4.35. Misunderstandings involving shifts in word stress. No. 73 84 117 121 156 182

Spk. MHk MIn MIn MIn MNg FMa

List. FTw FMa FTw FTw FBr FTw

Word over academic pre-islamic return you attend a agenda

Heard as all full i got ?? ?? leave ?? agent now

Context the past LW¶V oVER you know meaning aCADemic or (.) go the: PRE-islamic architecture if i re- i REturn to Jakarta you attend a (.) brunei school the main agenda would be to

In the first four of these tokens, word stress placement seems to be the main factor, with stress occurring on the second syllable of oVER, the second syllable of aCADemic, the first syllable of PRE-islamic and the first syllable of REturn, though other factors may also have played a part, such as [f] as the consonant in over in Token 73 (see Section 4.1.7), and minimal aspiration on the first [k] in academic in Token 84 (see Section 4.1.7). In Tokens 156 and 182, there seems to be no clearly stressed syllable in the misunderstood word, so perhaps we might conclude that the absence of stress is an issue. Alternatively, the absence of vowel reduction can be regarded as the main issue for these two tokens, as discussed in Section 4.2.5. Half of the tokens listed in Table 4.35 involve MIn, but we might note that, although his word stress is quite often unexpected, it rarely seems to cause problems. Table 4.36 lists some instances where unusual word stress by MIn does not appear to be an issue. So perhaps we can conclude that use of unexpected word stress only occasionally causes misunderstandings in ELF contexts.

76

4. Pronunciation

Table 4.36. Instances of unexpected stress from MIn that do not cause a problem. Recording In+Ma:531 In+Ma:674 In+Ma:1183 In+Ma:1355 In+Ma:1444 In+Ma:1375 In+Ma:1594 In+Ma:1624 In+Ma:1686 In+Ma:1695 In+Ma:1782 In+Ma:1990 In+Tw:13 In+Tw:589 In+Tw:769 In+Tw:1015 In+Tw:1326 In+Tw:1371 In+Tw:1719 In+Tw:1995 In+Tw:2254 In+Tw:2320

Words as (.) non (.) cash subject (.) so meaning these PAriah subject of course my children for example PREfer to live in brunei of course i reaLIZE that mcgill is ranked (.) number seventeen LW¶VQRWHDV\WRVWXG\LVODPIURPWKH3(56SHFWLYHRI KLVWRU\ er not abandoning fully over the study of er (.) er arCHIVE er muslim developmenTALists and anthropologists despite the fact that there are many diffiCULties meaning particularly context (.) so thereFORE one would er part of your british emPIRE you know of course friend with many example that er (.) despite the (.) stereoTYPE of brunei from scriptural religion (.) er religious PERspective (.) now meaning give you (.) SYMbolic kind of expression so the COMmittee put the right way of pronouncing it so this make them comFORtable here very er (.) very very conserVAtive in in that sense and thereFORE of course our general IDea which is not true when er we were in kinderGARten especially still virgin «QRWPXFK 32/OXWHGE\WKHWRXULVWV i was still (.) very (.) er what you call er inNOCent guy then these are partly DIScourage me indeed there was a push eventually there¶s bad er bad EFfect on my stomach because of the (.) time er pressure so er i i PREfer to to do it

One might note three cases in Table 4.36 in which unexpected word stress is not a problem, but some other segmental issue earlier in the utterance causes a misunderstanding: in In+Ma:531, FMa is unable to understand cash, hearing it as caste (Token 85 ± see Section 4.1.8), but she has no problems with PARiah, even though it is stressed on the initial syllable; in In+Tw:1371, FTw is unable to understand still virgin, hearing it as in the region (Token 115 ± see Section 4.3), but she has no difficulty with POLluted with stress on the first syllable; and in In+Tw:2320, FTw cannot understand pressure, because of the medial [s] (Token 125 ± see Section 4.1.8), but she has no problem with PREfer even though it has initial stress. One other extract from Table 4.36, In+Ma:1594, is interesting. As someone from the Inner Circle, I initially heard diffiCULties as different companies, but when I played it to FMa, she had no trouble immediately understanding the intended word. So perhaps word stress placement is, by and large, only really important for Inner-Circle listeners. The conclusion that variable or unclear word stress rarely causes misunderstandings in ELF settings is consistent with the LFC proposals.

4.4. Stress

77

4.4.2. Utterance stress In contrast with the suggestion that word stress may not be important for ELF intelligibility, Jenkins (2000: 153) proposes that the placement of the intonational nucleus within an utterance should be included in the LFC, as it serves to highlight the key part of the message. Let us now analyse tokens of misunderstanding that seem to involve prominence on an unexpected word. There are just three such tokens, all involving MIn, and they are shown in Table 4.37. Table 4.37. Misunderstandings involving unexpected phrasal and utterance stress. No. 93 113 123

Spk. MIn MIn MIn

List. FMa FTw FTw

Word religious elite vocabularies put yourself

Heard as religiosity ?? producer

Context the (.) brunei reLIGious elite many maLAY vocabularies you correctly put YOURself

In Token 93, one might expect the main stress in the phrase religious elite to fall on elite, but in fact there is little stress on elite. However, maybe another problem here is the speed at which the phrase is spoken. In fact, the same phrase is said again about one minute later (in Token 95) and it is similarly misunderstood by FMa, who again transcribed it as religiosity, and in this second case it seems purely to be caused by speaking rate. Token 95 will be analysed in Section 4.5.3. In Token 113, the main stress is on Malay, which is unexpected for a phrase consisting of an adjective followed by a noun. Another problem is that vocabularies is said rather softly and quite fast. Of course, the plural vocabularies for a noun that in Inner-Circle Englishes would be non-count might also be a contributory factor in this case. Token 113 will be discussed again in Section 5.2.1 when I consider plural nouns. In Token 123 the stress is on the first syllable of yourself, and this seems to be the main cause of the misunderstanding, though one should also note that this token could alternatively be regarded as involving word stress, as yourself might have been understood if the second syllable were more prominent. One pattern commonly found in the English spoken in South-East Asia is pronouns being given prominence when no contrast is implied (Levis 2005; Deterding and Kirkpatrick 2006). However, there is little evidence of this giving rise to misunderstandings. For example, in line 2 of Extract 4.20, MIn puts substantial emphasis on both ME and I, but this does not seem to cause a problem for FMa.

78

4. Pronunciation

Extract 4.20. In+Ma : 1375 (Token 90) Context: MIn is talking about his research when doing his PhD 1 2 3 4

MIn

and so eventually then i had er anthropologists who want to (.) take that issue with ME and (.) so then I have a joint supervision between (.) historians (.) er muslim developmenTALists and anthropologists

While it is true that FMa cannot understand joint supervision immediately afterwards, this seems to be related to the way these words are spoken (to be discussed in Section 4.5.3) rather than the emphasis on the pronouns. In conclusion, there seems to be little evidence that stress on an unexpected word causes too many problems in the data analysed here. This is not in line with the suggestions of the LFC. One possibility is that extra prominence on a word sometimes has a special discourse role in the Englishes spoken by many people in South-East Asia. For example, consider Extract 4.21, in which MIn puts substantial emphasis on school in line 2, girl in line 4, and son in line 6. It seems that he is using emphasis in this way to mark the end of a sentence. This is a pattern reported for the Englishes spoken throughout South-East Asia (Deterding and Kirkpatrick 2006). Extract 4.21. In+Tw : 817 Context: MIn is talking about the multicultural education of his children. 1 2 3 4 5 6

MIn

«but for example what is important is that i sent them to (.) chinese SCHOOL you know secondary school so meaning that saint andrews (.) mainly er what you call school for: the chinese boys and GIRL (.) er so m- (.) my my s- also in my daughters er (.) er my my two daughters my: 621HU EXWHUPHDQLQJWKH\PLQJOHZLWKPDQ\GLIIHUHQW«

,QDVLPLODUIDVKLRQHPSKDVLVPD\EHXVHGWRPDUNWKHHQGRIRQH¶VWXUQ An apparent instance of this use of emphasis by MIn is in Extract 4.22, in which of at the end of line 3 seems to be stressed. Extract 4.22. In+Tw : 1300 Context: MIn is talking about the background of his family in Indonesia. 1 2 3 4 5

MIn FTw MIn

«we abandon that er (.) that profession. my brother actually still (.) run some business then. but not actually the business that my father (.) was proud OF mm (.) we went to: yogyakarta once wonderful

4.4. Stress

79

In this extract, FTw successfully picks up this end-of-turn signal, and she comes in appropriately with a shift of topic. One might say, therefore, that utterance stress has an important discourse function in these ELF conversations, but it is a different role from that expected in most varieties of Inner-Circle English. Grabe, Kochanski, and Coleman (2005) report that the location of the intonational nucleus (utterance stress) is consistent in the English spoken throughout the UK (in contrast with the selection of intonational tone, which is highly variable). However, it seems that the same rules for the placement of utterance stress may not extend to Englishes in some other parts of the world.

4.5. Rhythm, intonation and speaking rate In this section, I will consider the impact on intelligibility of other suprasegmental features of speech, primarily rhythm, intonation and speaking rate. Also in this section I will consider the syllables in a word. Although this might seem to belong more with the discussion of consonants and vowels in previous sections, it is included here because the omission of syllables is generally connected with the rate of speaking: when people talk fast, they tend to omit syllables. Finally, I will discuss laughter.

4.5.1. Rhythm Stress-based rhythm is often assumed to be a key feature of the pronunciation of English. For example, Ladefoged and Johnson (2011: 118) state that stressed syllables in English WHQG³to come at regular interYDOV´DQGLaver    REVHUYHV WKDW ³English is often said to use a stress-based rhythm´. Indeed, many textbooks on English pronunciation regard the successful adoption of stress-based rhythm as essential for fluency. For instance, Teschner and Whitley (2004) introduce the metrical foot in their first chapter, considerably before they deal with the vowels and consonants of English, in the belief that stress-based rhythm and the associated alternation of strong and weak syllables is the fundamental framework on which the whole sound system of English is based. Similarly, Celce-Murcia, Brinton, and Goodwin (1996: 26) claim that the adoption of stress-based rhythm LV ³the most widely experienced pronunciation challenge for speakers of other lanJXDJHV´.

80

4. Pronunciation

However, many varieties of English around the world are reported to have more syllable-based rhythm, including those of Brunei (Deterding 2012), Singapore (Deterding 2007), Malaysia (Baskaran 2004a: 1044), Nigeria (Gut 2004: 927), and India (Sailaja 2009: 34). And there seems to be little evidence that speakers from these countries are less intelligible than those from the UK, the USA or Australia. Crystal (2003: 172) warns against imposing norms of rhythm where they are not appropriate, and Jenkins (2000: 150) excludes rhythm from the LFC. It is hard to separate out the effects of rhythm from the presence or absence of vowel reduction, and Cruttenden (2008: 265) comments that ³the occurrence of full vowels generally predicts the rhythm of English rather more useIXOO\ WKDQ DQ\ QRWLRQ RI VWUHVV´. Indeed, this is why one of the most widely used metrics for measuring rhythm depends on comparing the vowels of neighbouring syllables (Low, Grabe, and Nolan 2000). In Section 4.2.5, I discussed the possibility that lack of vowel reduction could contribute to misunderstandings, and although I suggested that there might be two or three such tokens, I concluded that it is rare. In fact, the tokens involving lack of vowel reduction might also be said to involve syllablebased rhythm, so here I will draw the same conclusion: use of syllablebased rhythm does not seem to cause too much of a problem. Furthermore, in an international setting (though maybe not in an Inner-Circle context), the use of syllable-based rhythm, with its relatively clear enunciation of distinct syllables, probably enhances intelligibility. This concurs with the suggestions of Jenkins (2000) that rhythm should be outside the LFC. Cruttenden (2008: 323) suggests that the problem of learners of English failing to use stress-EDVHGUK\WKP³is especially great and will require proORQJHGDWWHQWLRQ´; but he is assuming that speakers want to approximate to native-speaker norms of pronunciation, so this observation may not be relevant from the ELF perspective. 4.5.2. Intonation The placement of the intonational nucleus indicates what speakers choose to highlight as important in their message (Wells 2006: 7), and this is therefore included in the LFC (Jenkins 2000: 153). I have dealt with this in Section 4.4.2, where I suggested that unexpected placement of utterance stress only occasionally causes misunderstandings in the CMACE data. Here I will discuss the tone that is used. The selection of an intonational tone has a range of functions: it conveys the attitude of the speaker, it shows whether a statement or question is in-

4.5. Rhythm, intonation and speaking rate

81

tended, and it helps to indicate whether something is foreground or background information (Wells 2006: 11). However, while nucleus placement is quite stable in a range of Inner-Circle Englishes, the selection and realization of tone is highly variable (Grabe, Kochanski, and Coleman 2005). If Inner-Circle speakers vary so much in their use of tones, why should ELF speakers be expected to imitate one particular realization? The question then arises: does the selection of tone ever give rise to misunderstandings in the ACE data? It all depends on what you mean by misunderstanding. In terms of words or utterances being misunderstood, it is hard to find any tokens where tone selection plays a part. But there are almost certainly cases of overlapping speech that are caused by misinterpretation of the intonation patterns used, and there may also be cases where a question is interpreted as a statement and vice versa. These considerations are beyond the scope of the current investigation, but they certainly warrant further research. Hughes (2005: 124) suggests that many of the overlaps in conversations between Singaporean female undergraduates and a speaker from Britain DULVH EHFDXVH RI ³FRQIXsing turn-signals´. But note that this involves a speaker from the Inner Circle. Further research might investigate overlaps in the ACE data and the extent to which some of them might be classified as misunderstandings. I will consider overlapping speech in Section 6.3.4, but only in connection with collaborative completions. 4.5.3. Speaking rate One of the most common causes of misunderstanding in the CMACE data is simply that some of the participants, particularly MIn, tend to speak quite fast, and sometimes not very loudly. Let us now consider these issues. In discussing speaking rate, it would be good to have a benchmark: how fast is English usually spoken? Of course, this depends not just on the speaker but also the context, so for example sports commentary can be very rapid, while poetry reading is usually slower. Fletcher (2010: 571) gives figures of 5.2 and 5.9 syllables per second, while Roach (1998: 153) suggests a range between 3.3 and 5.9 syllables per second. However, this is for English in the Inner Circle, and the norm for ELF interactions may be different. Hüttner (2009: 288) gives a rate of 3.3 syllables per second for one conversation, though it is not clear if this is normal or not. Perhaps we should take the midpoint of the range from Roach: 4.6 syllables per second. There are fourteen tokens in the CMACE data in which the speed of speaking seems to be a significant factor that impacts on intelligibility.

82

4. Pronunciation

These fourteen tokens are all listed in Table 4.38. Ten of them occur in the speech of MIn. Table 4.38. Misunderstandings involving fast speech Token 29 32 86 88 89 90 95 103 104 110 111 117 158 159

Speaker MHk MHk MIn MIn MIn MIn MIn MIn MIn MIn MIn MIn MNg MNg

Listener FTw FTw FMa FMa FMa FMa FMa FTw FTw FTw FTw FTw FBr FBr

Words you know Lubbock by the end of the day spirit of learning i went to Cairo joint supervision religious elite until now despite the fact that Bandar Seri Begawan Jakarta pre-Islamic board test six three three four

Heard as no lot of biology speed of learning ?? changing ?? religiosity as you know like that Bandar Sudar which country ?? ?? his (tere) for

To investigate speaking rate further, we need to consider a little more context for these tokens. This is done in Table 4.39, in which the duration of the underlined words as well as the calculation of speaking rate in terms of syllables per second are shown in the last two columns. Table 4.39. Duration (seconds) and speaking rate (syllables per second) in tokens involving fast speech Tok. 29 32 86 88 89 90 95 103 104 110 111 117 158 159

Spk. MHk MHk MIn MIn MIn MIn MIn MIn MIn MIn MIn MIn MNg MNg

Words you know they have [you] know first name in the graduate school (.) in lubbock texas then er yeah by the end of the day er (.) we meaning that the spirit of learning in mcgill i went to cairo a little bit frustrated so then I have a joint supervision between this (.) er religious elite in brunei is that they so on you know (.) so until now meaning despite the fact that they insist on religious of er: bandar seri begawan dialect of jakarta dialect of jakarta (.) but even within that meaning the: PRE-islamic architecture joint admission and matriculation board test that is the way we we use six three three four

Dur. 1.18 0.80 0.78 0.81 0.70 0.78 1.31 0.58 1.29 0.88 0.75 1.37 1.62 1.09

Rate 5.08 5.00 7.69 7.41 7.14 6.41 7.63 8.62 6.20 7.95 8.00 5.84 7.41 5.56

4.5. Rhythm, intonation and speaking rate

83

Although in all of these tokens the speaking rate is above the 4.6 syllables per second that was suggested as average for English, some are only a little faster than this, and two are slower than the rates given by Fletcher (2010). Some of these cases should be analysed further. In Token 29, the overall speaking rate for the six syllables in the phrase they have [you] know first name is only just over five syllables per second. In fact, the problem in this token is that you is omitted entirely, so we might say that just the phrase you know is spoken so fast that half of it is missing, and FTw hears no instead. In addition, the absence of the indefinite article a before first name could be considered a contributory factor. Omission of articles will be discussed further in Section 5.2.2. Many of the other tokens involve fixed phrases which the listener may not be familiar with. For example, in Token 32, MHk says Lubbock Texas, but FTw has never heard of Lubbock (see Section 5.1.1). And in Tokens 158 and 159, MNg says joint admission and matriculation board test and six three three four respectively, both of which leave FBr somewhat bemused. The fact that MNg utters phrases like these without slowing down suggests that he sometimes fails to accommodate to the needs of his listeners. Tokens 158 and 159 will be discussed again in Section 5.1.2. Finally, one might note that most of the ten tokens from MIn involve not just a fast speaking rate but also a soft tone of voice. In fact, sometimes fading away in his utterances seems to be characteristic of 0,Q¶V speech. For example, Cairo in Token 89, joint supervision in Token 90, despite the fact in Token 104, Seri Begawan in Token 110, Jakarta in Token 111, and Islamic in Token 117 are all so soft that they are almost inaudible. Perhaps this is the real reason behind the loss of intelligibility in these tokens.

4.5.4. Syllables One result of the fast speaking rate that sometimes occurs in the CMACE data is that syllables may be omitted. For example, as mentioned in the previous section, in Token 29 you is omitted from the phrase you know. Furthermore, in Token 110, the first syllable of Begawan is omitted from the phrase Bandar Seri Begawan, which partly explains why FTw fails to understand the name of this place even though she actually lives in Bandar Seri Begawan, the capital of Brunei. In addition, there are five other tokens in which a missing syllable seems to contribute substantially to the misunderstanding. They are listed in Table 4.40.

84

4. Pronunciation

Table 4.40. Misunderstandings involving a missing syllable. No. 14 38 46 49 167

Spk. MHk MHk MHk MHk MNg

List. FMa FTw FTw FTw FBr

Word terminal Italy related with instead of another thing

Heard as common ?? believer of as an under them

Context know f- (.) terminal degree father you know from italy know related with arts too question instead of answer LW¶Vanother thing entirely

In Token 14, terminal is pronounced as [t‫ޝܮ‬PX@, with no second syllable and L-vocalization in the final syllable. But the main issue with this token is lexical, as FMa is not familiar with the phrase terminal degree (see Section 5.1.2). In Token 38, Italy is pronounced as [ܼti], with just a hint of a second syllable, and FTw was subsequently not able to make a guess about it despite the context of discussing &DWKROLF IDWKHUV LQ 0+N¶V VFKRRO ,Q Token 46, related with is pronounced as [riliwiv], and maybe it should really be transcribed as relate with, with no -ed suffix at the end of the first word, so this might partly be a morphological issue (see Section 5.2.3). In Token 49, instead of is pronounced as [ste‫ܥݦ‬v], with no first syllable in instead. And in Token 167, another thing is pronounced as [‫ݞ‬nd‫ݞ‬tܼ], so the omission of the second syllable of another is the main issue, though the pronunciation of the TH sounds is also a factor (see Section 4.1.1). In summary, apart from Token 14, terminal degree, which might be classified as lexical, the missing syllable in the other four tokens is probably the key factor in the occurrence of the misunderstanding. There are also some misunderstandings that seem to occur because of an additional syllable. Four such tokens are listed in Table 4.41. Table 4.41. Misunderstandings involving an added syllable. No. 75 138 160 165

Spk. MHk MLs MNg MNg

List. FTw FBr FBr FBr

Word tile sultan three three

Heard as tire (suratan) (tere) theory

Context like the tile okay ten year you know my former president to your sultan in the primary school three like degrees two master three phd

In Token 75, MHk tries to articulate tile really carefully, but as a result of L-vocalization, the pronunciation as [taܼԥ] ends up sounding like tire, (see Section 4.1.6). The absence of a plural -s on tile is discussed in Section 5.2.1. In Token 138, in addition to the added syllable in sultan, the use of [r] rather than [l] is another factor (see Section 4.1.3). In Tokens 160 and 165, FBr cannot understand three, and it seems that the inserted vowel is the problem. As discussed in Section 4.1.1, )%UFDQXQGHUVWDQG>WUL‫@ޝ‬DV three with no problem; but [tԥUL‫@ޝ‬LVnot so easily understood.

4.5. Rhythm, intonation and speaking rate

85

4.5.5. Laughter Laughter can occasionally be the cause of an instance of misunderstanding. There are three examples in which words that are accompanied by laughter are misunderstood. They are shown in Table 4.42. Table 4.42. Misunderstandings involving laughter. No. 27 122 146

Spk. FMa MIn MLs

List. MHk FTw FBr

Word fell into place administrators hot

Heard as Context ?? .. all fell into place ambitious be administrators ??  RK\RXGRQ¶Who:t @@@

In Token 146, laughter may partly mask the delayed final [t] in hot. However, the main problem seems to be the absence of a main verb (to be discussed in Section 5.2.4). In the other two tokens, the simultaneous laughter is categorized as the main cause of the misunderstanding. I will discuss the role of laughter in ELF discourse in Section 6.2.7.

4.6. Phonetic accommodation There are two ways that phonetic accommodation can occur: listeners can adapt to the pronunciation of speakers, by getting accustomed to their patterns of pronunciation; and speakers can adjust their pronunciation to the needs of their listeners. I will consider these two separately. But first I will consider the extent to which phonetic accommodation is successful in reducing the incidence of misunderstandings. A crude estimate of this is to consider where in a conversation the misunderstandings occur: if there is phonetic accommodation, either listening or speaking, then we would expect fewer misunderstandings towards the end of the recordings than at the beginning. In fact, on average, the misunderstandings occur at 45.8% of the duration of the recordings, so it seems that there are a few more misunderstandings in the earlier part of each conversation than in the later part. However, the average is only a little less than 50%, so there is little evidence either that the listeners are successful in getting used to the pronunciation habits of the speakers or that the speakers accommodate very much towards the needs of their listeners. If we recall that D OLVWHQHU¶V tiredness is one factor that has a negative impact on intelligibility (Pickering 2006 ± see Section 2.3), it is possible that this counteracts the benefits of the listener becoming familiar with the pronunciation patterns of the speaker. But it would be hard to evaluate the competing effects of tiredness and accommodation in the CMACE data.

86

4. Pronunciation

4.6.1. Listening accommodation There is occasional anecdotal evidence that some listeners manage to get used to the patterns of speech of the other person. In Token 128, which occurs in the first minute of the Jp+Br recording, FBr is unable to understand literature spoken by FJp as it is pronounced with an initial [r] (see Section 4.1.3). However, FBr said that she subsequently got used to the lack of a distinction between [l] and [r] by FJp and it became less of an issue as the conversation progressed. However, as suggested above, there is in fact little evidence overall that the listeners improved their ability to understand their conversational partners by very much. Conversely, there are four tokens in which accommodation may actually have played a role in causing the misunderstanding. These four tokens are shown in Table 4.43, and it may be significant that all but the last one occur late in the conversation, by which time the listener may have become accustomed to the pronunciation patterns of the speaker. Table 4.43. Misunderstandings in which accommodation may have played a role No. 22 71 166 177

Spk. MHk MHk MNg FTw

List. FMa FTw FBr FMa

Word we call tow take logic and taught

Heard as record throw think logical told

Context the same (.) we call today you tow everything take logic and thinking so (.) he taught me

In Token 22, which occurs nearly 19 minutes into the 23-minute Hk+Ma recording, FMa hears initial [w] as [r], as use of [w] in place of [r] is characteristic of MHk¶V VSHHFK elsewhere (e.g. Tokens 12 and 16 involving reverse heard as we worse ± see Section 4.1.3). However, I assume that Lvocalization is the main problem in Token 22 (see Section 4.1.6). Token 71 (shown in Extract 4.4 in Section 4.1.1) occurs 35 minutes from the start of the Hk+Tw conversation, and FTw hears [to‫ ]ݜ‬as throw, perhaps because she has already got used to 0+N¶V omission of [r] from word-initial consonant clusters and so she assumes that [r] may be missing from this word. However, we might also observe that she has failed to notice that MHk actually usually pronounces voiceless TH as [f], and he has in fact already said throw nine times in their conversation, pronouncing it with an initial [f] every time and never with initial [t]. So, in this respect, we can conclude that FTw has not accommodated to the speech patterns of MHk. Token 166 (shown in Extract 4.3 in Section 4.1.1) occurs nearly 36 minutes into the Ng+Br recording. FBr hears the [t] at the start of take as

4.6. Phonetic accommodation

87

voiceless TH, partly because MNg tends to pronounce voiceless TH as [t]. In fact, he uses [t] at the start of thinking just half a second later. Token 177 (shown in Extract 4.14 in Section 4.1.6) is the only one of these four tokens that is near the beginning of the recording, occurring just five and a half minutes from the start. But FMa may already have got used to L-vocalization in the speech of FTw, particularly as L-vocalization is a common pattern among many different speakers, and this is why she hears taught as told even though there is no [l] in the word as it is said. 4.6.2. Speaking accommodation To what extent do the speakers in the ACE data successfully adapt their pronunciation to the needs of their listeners over the course of a conversation? There is little evidence of them achieving this. One basic reason is that some speakers simply do not know what the problems are or how they can fix them. I will return to this in Chapter 7 when I discuss the pedagogical implications of my research, and I will further discuss the case of MHk, who is aware that people often do not understand him but apparently does not know what to do about it. For example, he lives in an apartment with valley in the name, but people often have trouble understanding him when he says it with an initial [w]. In addition, as already discussed in Sections 4.1.6 and 4.5.4, he realizes that dark-L may be problematic, so he tries to say tile really carefully, but as a result he pronounces the word with two syllables and FTw hears it as tire (Token 75). The conclusion seems to be this: some of the speakers do not know how to improve the clarity of their speech even when they realize that they are not being understood. Raising awareness about how students speak, how they are heard, and how they might fix problems when they are misunderstood seems to be one of the most important goals that pronunciation teachers should focus on.

4.7. Summary of pronunciation In this section, I offer a summary of the findings on pronunciation. First, I will attempt to identify the main cause of each token of misunderstanding. Inevitably, this is rather subjective; yet at the same time, some features of pronunciation seem to be more important than others in affecting intelligibility, and it is useful to try to identify which are the most important.

88

4. Pronunciation

Many of the tokens of misunderstanding involving pronunciation have been discussed more than once in this chapter, as it is common for a token to involve a combination of various features. For example, Token 30 is shown in Extract 4.23. In line 3, FTw hears car instead of class. Extract 4.23 Hk+Tw : 262 (Token 30) Context: MHk is talking about his simple wedding in the USA. 1 2 3 4 5

MHk FTw MHk FTw MHk

... and then you know er and then we you know er was a big ceremony there we not skip the class you know no @@ we we just go to the court house you know. in february.

This misunderstanding occurs because class is pronounced as [k‫@ޝܤ‬ZLWKQR [l] or [s]. So, which of these two features of pronunciation is more important? It is assumed that the absent [s] is more important, because [k‫ޝܤ‬V@ might have been understood as class while [kl‫@ޝܤ‬ZRXOGSUREDEO\QRWKDYH been. But this is just a guess. An alternative hypothesis is that either [k‫ޝܤ‬V@ or [kl‫@ޝܤ‬ZRXOGKDYHEHHQXQGHUVWRRGVRLWLVRQO\ZKHQWKHWZRIDFWRUVDUH combined that a misunderstanding occurs. Nevertheless, Token 30 is classified as being caused primarily by the absence of the final fricative. Similarly, consider Extract 4.24, illustrating Token 140 in which FBr hears coffee-break as coffee bake as well as Token 141 in which she hears lunchtime as runtime. Extract 4.24. Ls+Br : 961 (Tokens 140 & 141) Context: MLs is talking about the food he likes to eat. 1 2

MLs

so usually: i: see: er: the food er they serve in (.) coffee-break or: the lunchtime is hh usually er have the chicken.

In Token 140, coffee-break is pronounced as [k‫ܥ‬fibeܼ], but FBr has no problems with the omission of the final [k] because she hears the second word as bake. It is the missing [r] that causes the problem, so it is straightforward to classify this as arising from the simplification of the initial consonant cluster, [b] instead of [br]. However, classification of Token 141, in which lunchtime is heard as runtime, is more problematic. It is pronounced as [r‫ݞ‬nt‫ޝܤ‬P@with [r] in place of initial [l] and the omission of [t‫( ]ݕ‬as well as the PRICE vowel pronounced as [‫@ޝܤ‬LQVWHDGRI>Dܼ@LQWKHVHFRQGV\OODEOH  But which factor is more important? Here, it is assumed that the loss of [t‫]ݕ‬ has a greater effect on intelligibility, as [r‫ݞ‬nt‫ݕ‬t‫ޝܤ‬P@might have been under-

4.7. Summary of pronunciation

89

stood as lunchtime but [l‫ݞ‬nt‫ޝܤ‬P@ZRXOGQRWKDYHEHHQ%XWLWPXVWEHDdmitted that this assumption is rather speculative, and my attempts to identify the main cause of the problem need to be treated with caution. In total, 158 tokens have been discussed in this chapter, most of them more than once, and each of them is now classified under one main issue. For 20 tokens, even though they have been discussed in this chapter, the main factor appears actually not to be phonological. But for 138 tokens, it seems that pronunciation is probably the main issue. Those involving consonants are listed in Tables 4.44 to 4.46. Table 4.44. Tokens involving replacement of an approximant or nasal Factor [l] Æ [n] [n] RU>ƾ@ Æ [l] [l]Æ[r] [r] Æ [l] [r] Æ [w] [v] Æ [w] dark-L

Tokens 145 (football), 147 (old), 149 (old) 5 (nearby), 8 (hanging), 9 (noisy), 13 (north), 20 (next), 23 (next), 33 (northern) 6 (club), 10 (Black Swan), 128 (literature), 131 (Labuan), 132 (fluently), 133 (like the) 134 (real) 60 (wrapping), 65 (related), 78 (rain forest) 12 (reverse), 16 (reverse), 52 (environment), 76 (provide) 19 (wall), 22 (we call) , 75 (tile), 107 (world) Total

Total 3 7 6 1 3 4 4 28

Table 4.45. Tokens where the main factor is another individual consonant Factor voiceless TH voiced TH H-dropping H-insertion inserted [t] fricative voicing aspiration [d‫]ݤ‬ [t‫]ݕ‬ [tr] Æ [t‫]ݕ‬ no final fricative [‫ ]ݕ‬Æ [s] final plosive no [n]

Tokens 98 (three), 180 (thought) 51 (on the), 72 (other), 77 (this) 154 (in higher), 161 (has rise), 162 (hold), 168 (handle), 173 (hot cake), 174 (help us) 150 (us), 155 (all means) 1 (garment) 43 (zen), 54 (zone), 83 (vagrant), 99 (visit) 4 (Bhutan), 179 (P6) 86 (by the end of the day), 127 (major), 136 (young) 97 (eitch), 141 (lunchtime) 7 (trekking) 30 (class), 53 (because) 85 (cash), 118 (dishes), 125 (pressure) 11 (to read), 31 (Tech), 58 (spade), 68 (add) 112 (ethnic), 163 (seven) Total

Total 2 3 6 2 1 4 2 3 2 1 2 3 4 2 37

90

4. Pronunciation

Table 4.46. Tokens where the main factor is an omitted consonant from a cluster Factor initial clusters

final [t] or [d] no [s] no [k] no [n]

Tokens 17 (afraid), 24 (phrase), 26 (flaming), 35 (three), 36 (plough), 37 (close), 42 (different), 45 (process), 48 (curious), 50 (floating), 55 (process), 56 (plastic), 57 (plant), 59 (planting), 61 (throw), 63 (plastic), 67 (meal plan), 137 (present), 140 (break), 142 (treaty) 34 (west), 108 (cold war) 105 (Karl Marx), 139 (just) 96 (correct), 143 (accent) 148 (point) Total

Total 20

2 2 2 1 27

The tokens where the main issue seems to be a vowel are listed in Table 4.47, and those for other features of pronunciation are in Table 4.48. Table 4.47. Tokens where the main phonological factor involves a vowel Factor NURSE TRAP

vowel length vowel reduction miscellaneous

Tokens 41 (measure), 129 (early morning) 15 (matter), 91 (pattern), 144 (Man U) 28 (need), 119 (senior), 152 (weed) 156 (attend a), 182 (agenda) 18 (communist), 40 (Italian), 79 (heart), 89 (Cairo), 109 (anti) Total

Total 2 3 3 2 5 15

Table 4.48. Tokens involving suprasegmental and miscellaneous features Factor spelling word stress utterance stress speaking rate

missing syllable extra syllable laughter accommodation

Tokens Total 82 (tubers), 115 (virgin) 2 73 (oVER), 84 (aCADemic), 117 (PRE-islamic), 4 121 (REturn) 93 (reLIGious elite), 113 (maLAY vocabularies), 3 123 (put YOURself) 29 (you know), 32 (Lubbock), 88 (spirit of learning), 10 90 (joint supervision), 95 (religious elite), 104 (despite the fact that), 110 (Seri Begawan), 111 (Jakarta), 158 (board test), 159 (6334) 38 (Italy), 46 (related), 49 (instead), 167 (another) 4 138 (sultan), 160 (three), 165 (three) 3 27 (fell into place), 122 (administrators) 2 71 (tow), 166 (take logic), 177 (taught) 3 Total 31

4.7. Summary of pronunciation

91

Tables 4.44 to 4.46 suggest that the greatest impact on intelligibility comes from consonants, which is consistent with the LFC proposals. Furthermore, the biggest problems with consonants are the loss of the second consonant from initial clusters such as [pl] and [fr], H-dropping, omission of single final consonants such as [s] and [d], and confusion between sounds such as [l], [n], [r], [w] and [v]. In contrast, omission of [t] or [d] from final consonant clusters does not cause too many problems, and L-vocalization also does not give rise to many misunderstandings (though omission of dark-L may occasionally be an issue), which again is in line with the LFC. However, few problems were found with minimal aspiration on initial voiceless plosives, and this does not support the LFC proposals which suggest that aspiration on initial voiceless plosives is important. Table 4.47 indicates that variation in vowel quality causes fewer problems than consonant variation, which is consistent with the LFC. However, it seems that absence of a distinction between long and short vowels also does not cause too many problems in the data from South-East Asia, and this does not concur with the LFC suggestions of Jenkins (2000). Table 4.48 shows that fast speaking rate is the biggest suprasegmental issue, though we should remember that most of the tokens involved are from one speaker, MIn, which reminds us that different patterns of misunderstandings would occur in data from a wider range of speakers. Indeed, though only four tokens seem to involve word stress, it is possible that stress placement would be more important for speakers and listeners from elsewhere, though it is also suggested that word stress may not be very salient when there is a tendency for a lack of vowel reduction. Finally, Table 4.49 lists the tokens in which something other than pronunciation appears actually to be the main factor leading to the misunderstanding. These tokens will be discussed further in Chapter 5, together with the 25 tokens that have not been discussed in this chapter. Table 4.49. Tokens where a non-phonological factor seems to be the main issue Factor lexis

grammar misc.

Tokens 14 (terminal degree), 47 (collage), 74 (vinyl), 87 (Berlitz), 101 (inculcate), 103 (until now), 120 (going native), 124 (niche), 170 (cathode),176 (great leap forward), 181 (knuckle), 183 (orchard) 2 (she also in that company), 39 (couple you know father), 100 (then now work), 114 (the living eldest), \RXGRQ¶WKRW), 157 (the cut off will out) 25 (river), 135 (weather) Total

Total 12

6 2 20

Chapter 5 Lexis, grammar and code-switching

Chapter 4 dealt with phonology, as variant pronunciation seems to be the most common source of misunderstanding in ELF interactions. However, misunderstandings may also arise because of unfamiliar lexical items and unexpected grammar. This chapter will consider misunderstandings that occur because of word usage and grammar. In addition, I will briefly consider code-switching, as there are a few instances in the CMACE data in which there is a lack of comprehension of non-English words. In Chapter 4, the focus was on the pronunciation of individual words and short phrases, so reference was made to tokens of misunderstanding. In this chapter, in Section 5.1 on lexis, I will continue to refer to tokens, as the focus is still on words and short phrases; but from then on, for the discussion of grammar in Section 5.2, code-switching in Section 5.3, and also repairs in Chapter 6, we need to consider the wider context, so the analysis will instead refer to instances of misunderstanding rather than tokens.

5.1. Lexis There are various ways that lexical usage can cause misunderstandings: when speakers use words that their listeners do not know; when phrases or idioms are used that are unfamiliar to the listeners; when words occur with unexpected shifts in meaning; and when words are polysemes, so they have two or more distinct meanings, and the listener gets the wrong one. In this section, I will consider each of these in turn. 5.1.1. Words Use of words which the listener is not familiar with can obviously cause a problem. In the CMACE data, there are a number of cases where unfamiliar words occur. Of course, there may be names that are unknown to the listeners, but that does not matter too much so long as the listener knows they are names. However, it becomes more problematic when the listener does not realize that they are names, and three such tokens are listed in Table 5.1.

5.1. Lexis

93

Table 5.1. Misunderstandings involving names No 31 32 87

Spk. MHk MHk MIn

List. FTw FTw FMa

Word Texas Tech Lubbock Berlitz

Heard as Texas there lot of ??

Context study in texas tech you know in lubbock texas. er northern taken from berlitz actually (.)

In fact, in all these tokens, something in the pronunciation contributes to the misunderstanding: in Token 31, Tech has no final [k] (see Section 4.1.9); in Token 32, Lubbock is spoken rather fast (see Section 4.5.3) and also the word northern that follows soon after is pronounced as [l‫ܥ‬ftԥn]; and in Token 87, Berlitz is pronounced as [bælܼd‫( ]ݤ‬see Section 4.3). Quite apart from these tokens involving names, sometimes speakers use words which their interlocutors simply do not know. Fourteen such tokens are listed in Table 5.2. In most cases, the problem is straightforward: the listener is not familiar with the word. But I will elaborate on a few of them. Table 5.2. Misunderstandings involving unfamiliar words No 3 47 62 66 70 74 101 102 116 124 153 169 170 171

Spk. FCh MHk MHk MHk MHk MHk MIn MIn MIn MIn MNg MNg MNg MNg

List. FBr FTw FTw FTw FTw FTw FTw FTw FTw FTw FBr FBr FBr FBr

Word hotpot collage landfill dorm dorm vinyl inculcate disseminate improvise niche cutlass java cathode anode

Context maybe dimsum ... yeah and hotpot (.) paper you know ... collage on the top the dumpster or you know the landfill you the (.) you know (.) dorm (.) the meal plan you know dorm (.) for your kid you know you know still the vinyl plastic (.) is (.) you i inculcate er (.) disseminate er islamic er (.) disseminate er islamic values to my what you can improvise. meaning beautiful so they are giving us niche PHDQLQJWKH\¶UH weed the something with cutlass (.) or you somebody doing java now. or everywhere is electrician putting this wire is cathode this this wire is cathode this one is anode. if

In Token 3, FBr subsequently confirmed that she knows the term steamboat, but she is not familiar with the alternative term hotpot (from the Chinese 䀓 播 KX΅JXǀOLWµILUHSRW¶ for a style of eating where the food is cooked in a pot of boiling broth on the table. She guessed hot plate instead. In Tokens 66 and 70, FTw is familiar with the word dormitory, but she does not know its abbreviation as dorm, so we can say that she does not know this word. In fact, after Token 70, she asks for clarification about the meaning of dorm, and MHk explains it as dormitory.

94

5. Lexis, grammar and code-switching

In Token 74, MHk is talking about vinyl plastic covering for his home in the United States. Not only is FTw not familiar with this concept of using plastic as a protective covering for a house, but in addition she does not know the word vinyl, even when it is used in reference to records. Although the occurrence of unfamiliar words sometimes results in misunderstandings, this is less common than misunderstandings arising from unexpected pronunciation. ELF speakers are generally aware of what their listeners are likely to understand, and they mostly avoid words that are specialist or obscure. Perhaps the most surprising instances from Table 5.2 are the use of vinyl by MHk (Token 74) to refer to the covering of a house in parts of the USA, and cutlass by MNg (Token 153) to refer to a long knife (rather than the equivalent parang that is more usual in Brunei). Finally, under the category of individual words, I will mention Token 183 in which FTw says orchids rather than orchard as shown in Extract 5.1. Extract 5.1. Tw+Ma : 1583 (Token 183) Context: They are talking about places one can visit in Cambridge. 1 2 3 4

FTw

it kind of like like a apple orchids LVQ¶WLW LW¶Ver: (.) y that place the place (.) yes orchard? orch- oh yeah. orchard

FMa FTw

This involves mistaken word selection ± FTw just gets the wrong word. (But she sounds quite happy to be corrected, as I discuss in Section 6.2.2.) 5.1.2. Phrases Sometimes, the listener may know all the words in an utterance, but their use in unfamiliar phrases is problematic. Eight tokens are included in this category. They are listed in Table 5.3, and then each is discussed below. Table 5.3. Misunderstandings involving phrases Number 14 26 67 120 158 159 178 179

Speaker MHk MHk MHk MIn MNg MNg FTw FMa

Listener FMa FMa FTw FTw FBr FBr FMa FTw

Phrase terminal degree flaming hot meal plan going native joint admission and matriculation board test six three three four pepper crab P6

5.1. Lexis

95

Token 14 was discussed in Section 3.2.2 and it is shown again in Extract 5.2. MHk uses terminal degree to refer to RQH¶V final degree. Although this phrase is fairly common in the United States, with 40 tokens in the COCA data (COCA 2013), it seems that it does not occur widely elsewhere, which is why FMa cannot understand it. She subsequently guessed it might be common degree, even though that does not make much sense. Extract 5.2. Hk+Ma : 154 (Token 14) Context: MHk is discussing why he chose to become an artist. 1 2 3

MHk

«ZKDWVXEMHFW\RXFKRVHWR VWXG\ IRU\RXUspel> m a ? your you know f- (.) terminal degree? (.) base on (1.0) your interest

Token 26 has also already been discussed in Section 3.2.2, and the wider context is shown again in Extract 5.3. Extract 5.3. Hk+Ma : 1211 (Token 26) Context: MHk is discussing the problems of moving to a new place. 1 2 3 4 5

MHk FMa MHk

«EHFDXVHHYHU\WLPHZKHQLUHORFDWH\RXNQRZHLWKHUUeally (.) cold freezing cold mm or flaming hot  LW¶VXVHGWREHLW¶VRND\ ORRNOLNH LW¶VEHWWHUWKDQPHSK\VLFDOO\EHFDXVH \RXNQRZOLYLQJLQ«

FMa is able to understand the phrase freezing cold in line 2, even though there is no [r] in freezing, because it is a common collocation; but flaming hot in line 4 eludes her, partly because of the missing [l] (see Section 4.1.4) but also because it is not a common phrase. There are a number of factors that cause FTw to hear meal plan as view pen in Token 67, which is shown in Extract 5.4. Extract 5.4. Hk+Tw : 1969 (Tokens 66 & 67) Context: MHk is discussing the cost of tertiary education in America. 1 2 3 4 5 6

MHk FTw MHk FTw

... you send the kid you know to the university (1.3) tuition is a major part of it you know (.) all the (.) you know (.) dorm (.) the meal plan (.) just killing view pen what is that yeah meal you know eating you know ah meal

96

5. Lexis, grammar and code-switching

In this token, there are a few pronunciation issues: vocalized-L occurs in meal (see Section 4.1.6), and in plan there is no [l] and also a mid vowel that sounds like it might be DRESS (see Sections 4.1.4 and 4.2.2). But a further problem is that use of meal plan to refer to a schedule of payments for meals for university students is only found in the USA. The phrase meal plan occurs 171 times in the COCA data. Although most of these refer to weight-loss regimes, at least eleven are concerned with payment schedules for student meals, and a further ten are in connection with hotels or holiday resorts that include arranged meals in their bookings (COCA 2013). In Token 120, shown in Extract 5.5, FTw hears going native as good to eat. One problem is that native is pronounced with [æ] rather than [eܼ] in the first syllable (see Section 4.2.6). However, in fact FTw is not familiar with the phrase going native. When subsequently asked, she guessed that it might mean continuing to eat your own native food even when you are abroad, which of course is actually the opposite of its intended meaning: adopting the food of the country you are living in. Extract 5.5. In+Tw : 1808 (Token 120) Context: MIn is discussing the kind of food he ate when he was in Cairo. 1 2 3

MIn

«i accept that er we can survive without eating rice @@@@ and then after that then you know is is er going native meaning RIFRXUVHLQHJ\SWWKHUHDUHPDQ\WKLQJEXW«

In Token 158, as discussed in Section 4.5.3, MNg says the phrase joint admission and matriculation board test rather fast, with the twelve syllables taking just 1.62 seconds. The wider context is shown in Extract 5.6. It seems that this is a fixed phrase for MNg, and he fails to realize that others may not be familiar with it so they may be unable to decipher it if he says it at this fast speed. Extract 5.6. Ng+Br : 894 (Token 158) Context: MNg is discussing the university entrance system in his country. 1 2 3

MNg FBr

do you understand yeah @@ so if i now pass that joint admission and matriculation board test LSDVVLWLZLOOQRZHQWHUWKHVFKRRO«

Token 159 occurs less than a minute after Token 158, and it similarly involves a fixed phrase spoken rather fast by MNg, this time six three three

5.1. Lexis

97

four, to refer to the system of education in his country. The wider context is shown in Extract 5.7. Once again, MNg fails to slow down for a fixed phrase which others are not familiar with (see Section 4.5.3). Extract 5.7. Ng+Br : 942 (Token 159) Context: MHk is discussing the education system in his country: six years of primary school, three in lower secondary, three in upper secondary, and four years of university. 1 2

MNg

so you understand. that is the way we we use six three three four (.) six in the primary school «

The wider context for Token 178 is shown in Extract 5.8. In this instance, FMa hears paper instead of pepper (even though paper crab does not sound particularly edible, even with some beer to go with it). This misunderstanding probably occurs because chilli crab is a common dish in the region but pepper crab is a bit more unusual. So we can say that FMa is not familiar with this phrase. Extract 5.8. Tw+Ma : 485 (Token 178) Context: FTw is talking about the food you can eat in Singapore. 1 2 3

FTw FMa

... you can drink beer? (.) you can have eating (.) pepper crab? (2.2) or sing- singapore bi-hoon {rice-noodles} bi-hoon yes

Finally, in Token 179, shown in Extract 5.9, the minimal aspiration on the initial [p] results in FTw hearing P6 as B6 (see Section 4.1.7). In addition, FMa assumes that FTw is familiar with the terms P1 to P6 to refer to the six grades of primary school in Singapore. However, even though FTw spent fifteen years in Singapore, her children actually went to international schools there, so she is not familiar with the term P6 (or, indeed, with P5 said immediately after). Extract 5.9. Tw+Ma : 603 (Token 179) Context: FTw is tDONLQJDERXWKHUFKLOGUHQ¶VHGXFDWLRQLQ6LQJDSRUHSDUWLFXODUO\ the time that they started school when they arrived from England. 1 2 3 4

FTw FMa FTw FMa

late primary school late primary school yeah so p p six p five

98

5. Lexis, grammar and code-switching

Not surprisingly, therefore, use of unfamiliar phrases gives rise to some instances of misunderstanding. Speakers need to be aware about which phrases are commonly used throughout the world and which have more restricted usage. We can also consider the possibility of new phrases being coined in ELF settings. Hülmbauer (2009: 334) gives the example of far-away uncle as an innovative expression that was coined and then shared by an Italian and a German, and this kind of joint creation of new phrases is a clear example of accommodation among ELF interactants. I have not found any such examples in my data, but the possibility of new expressions like this being created may exist even in the kind of brief encounters studied here. I will consider innovative idioms a little further in the next section.

5.1.3. Idioms Somewhat different from the phrases that were discussed in the previous section, idioms convey non-literal meanings for words, and this can be problematic in ELF contexts. Prodromou (2008) observes that the use of pre-fabricated chunks of English enhances fluency, but at the same time they can be troublesome minefields that may cause problems for intelligibility in ELF conversations. Similarly, Jenkins (2007: 41) suggests that use of opaque native-speaker idioms is counterproductive in ELF settings. Seidlhofer and Widdowson (2009) note that onHRIWKHIXQFWLRQVRILGLRPVLVWKHµWHUULWRULDOLPSHUDWLYH¶ the need for speakers to establish and protect their identity as part of a fixed speech community, and as an essential part of this territorial imperative is to exclude outsiders, it is obviously problematic in ELF interactions. However, Seidlhofer and Widdowson (2009: 35) also note that ELF interactants sometimes invent their own idioms as a form of accommodation. So for example (as already discussed in Section 2.1) they report the phrase endangered fields being quite deliberately created and then shared by participants at an international conference. In fact, Pitzl (2009) suggests that ELF speakers are good at unpacking the original metaphorical meanings of idioms and creating new uses, and she finds a number of innovative idioms in the VOICE corpus, including we should not wake up any dogs. Although this certainly deviates from the native-speaker idiom let sleeping dogs lie, it is highly expressive and effective in the ELF context in which it occurs. The adoption of innovative idioms like this may be more likely to occur

5.1. Lexis

99

when people are together over an extended period of time, so it happens less often in the brief encounters investigated here. Quite apart from innovative idioms, I will here consider established idioms that are not familiar to listeners. There are three instances in which such idioms cause a misunderstanding. They are listed in Table 5.4, and then each is discussed in more detail below. Table 5.4. Misunderstandings involving idiomatic usage Number 106 173 176

Speaker MIn MNg FTw

Listener FTw FBr FMa

Phrase opium [of the people] hot cake great leap forward

In Token 106, MIn does not actually say religion is the opium of the people, but the allusion is clearly to this well-known saying by Karl Marx. The wider context is shown in Extract 5.10. Extract 5.10. In+Tw : 962 (Tokens 105 &106) Context: MIn is talking about religion, and whether having no religion might be considered as one kind of religion. 1 2 3

MIn

so meaning the same like er (.) karl marx for example er consider that religion is opium society but (.) communism itself eventually is religion

The first factor that contributes to this instance of misunderstanding is that Marx is pronounced with no [s] on the end (see Section 4.1.8), so FTw is unable to understand it, though she has certainly heard of Karl Marx. Second, MIn says opium society rather than opium of the people, though it is clear that he is alluding to the latter expression. And third, there is no preposition of in opium society, and also no article the before it. FTw hears opium society as helping society, which at least makes sense. This token is classified as an unfamiliar idiom because FTw subsequently confirmed that she is not familiar with the phrase opium of the people, so even if there had been no other problems, she still would probably not have understood it. In Token 173, as discussed in Section 3.1.3, FBr hears out kick instead of hot cake, partly because of the omission of the initial [h] (see Section 4.1.2). The wider context is shown in Extract 5.11. The idiom hot cake is being used to describe someone who is in hot demand from employers, but this usage of the idiom to refer to people rather than things may not be widespread outside Nigeria, and that is why FBr does not understand it.

100

5. Lexis, grammar and code-switching

Extract 5.11. Ng+Br : 2742 (Token 173) Context: MNg is talking about how it is easy for some people to get a job after graduating from university. 1 2

MNg

«they are not looking for anything. that time they will be calling them because they are hot cake (1.3) they need their service.

Finally, in Token 176 (shown in Extract 5.12), FTw uses the expression great leap forward (from the Chinese ⣏嵫徃 dà yuè jìn) as an idiom to GHVFULEHDVXGGHQLQFUHDVHLQRQH¶VNQRZOHGJHRIWKHZRUOG$OWKRXJK)0D can understand all the words fairly well (though perhaps the missing [p] in leap is an issue), she does not understand this idiomatic usage. Extract 5.12. Tw+Ma : 147 (Token 176) Context: FMa has asked FTw about her life in England. 1 2 3 4 5 6 7

FTw FMa FTw FMa FTw FMa

england is (.) s- such a beautiful country (2.2) it was a great (.) leap forward for me in my life sorry it was a great? great leap forward okay forward leap leap forward but okay leap forward

One can regard idioms such as these as troublesome minefields to be avoided. Alternatively, one might suggest that idioms play just as important a role in ELF contexts as in native-speaker settings, allowing for humour and creativity (Pitzl 2009). The conclusion should probably be this: so long as ELF speakers are sensitive to the possibility of idioms not being understood, their use can enrich a conversation immensely. And it seems that misunderstandings arising out of unfamiliar idiomatic usage are quite rare in the CMACE data.

5.1.4. Shifted meaning Words with shifted meaning can cause problems in international communication. Brown (1999: 36) gives the example of how send meaning to accompany someone (as in ,¶OOVHQG\RXRIIDWWKHDLUSRUW) in Singapore English can give rise to confusion for people from places such as Britain who are not familiar with this usage, as when they hear this, they may expect the speaker to put them in a taxi and wave goodbye to them. However, this use

5.1. Lexis

101

of send is so common in places like Singapore and Brunei (Deterding 2007: 81; Deterding and Salbrina 2013: 104), probably influenced both by the Malay word menghantar µVHQG¶ DQGWKH&KLQHVH 復 sòng µVHQG¶ WKDWLW is unlikely to cause a problem, except perhaps for expatriates who have just arrived in the region. And maybe an alternative like accompany is more likely to be misunderstood in South-East Asia. There are nine tokens in the CMACE data in which an unfamiliar meaning of a word contributes to a misunderstanding. They are listed in Table 5.5, then each is discussed below. Table 5.5. Misunderstandings involving shifted meaning No 81 83 85 91 126 151 172 175 181

Spk. MIn MIn MIn MIn FTw MNg MNg MNg FTw

List. FMa FMa FMa FMa MIn FBr FBr FBr FMa

Word elite vagrant cash pattern Hash portion learneds manner knuckle

Context this is of course elite satay actually like my grandfather (.) is vagrant you know consider as (.) non (.) cash subject LW¶VQRWDVDVpattern this and so on to s- stop doing (.) this [MIn: yeah er ...] hash the punishment will be (.) giving you portion and we have more learneds compared to the which is not manner DOVRWKDWRKGRQ¶WFURVV we drank beer? (.) and knuckle eating knuckle

In Token 81, MIn is talking about the satay that originates from his hometown, Ponorogo, and he uses elite WR PHDQ µVSHFLDO¶ )0D LV FRQIXVHG E\ this usage, and she thinks that elite satay refers to the name of one kind of satay. The usage by MIn in this case may be influenced by the use of the word as it is borrowed into Indonesian, as the Malay/Indonesian word elit seems to have a wider meaning than its English equivalent. In Token 83, MIn uses vagrant to refer to someone who moves away from their hometown in pursuit of a career, and he does not imply any of the negative connotation that is generally association with this word. This unexpected usage of vagrant confuses FMa who subsequently transcribed the word as fragrant, even though she knew it was wrong. In Token 85, MIn is talking about language teaching, and he refers to it as a non-cash subject. One problem, as discussed in Section 4.1.8, is that cash has [s] rather than [‫ ]ݕ‬at the end. But the wider issue is the use of noncash subject to describe a subject which does not earn lots of money. In Token 91, MIn is discussing the historical analysis of terrorism, and he uses pattern to refer to something that occurs repeatedly, claiming that terrorism is not so predictable. This unexpected use of the word pattern

102

5. Lexis, grammar and code-switching

confuses FMa, who subsequently transcribed it as part of, partly because of the back quality of the vowel in the first syllable (see Section 4.2.2). In Token 126, MIn and FTw have been talking about engaging in outdoor activities such as hill walking, and MIn has said that he had to stop doing such vigorous exercise. FTw then says that she should advise her husband not to go on the Hash, by which she means the Hash House Harriers, a group of people who go running in the forest once a week. MIn does not pick up this reference, and he hears it as harsh instead. In fact, he appears to be unaware that he has misunderstood the word, something that will be discussed again in Section 6.2.8. In Token 151, MNg uses portion to refer to an instance of punishment that used to be meted out in his high school (as discussed in Section 3.2.1). This use of portion seems to be idiosyncratic of Nigerian English, and FBr is unable to understand it. In her transcript, she just ZURWHµ[[¶WRLQGLFDWH that she was unable to make a guess about the word. In Token 172, MNg says learneds to refer to highly educated people, and he says that there are more learneds in his country than in other countries. This use of the word confuses FBr, who transcribed the word as land rights, though she knew that this was not correct. Again, this word seems to be idiosyncratic in Nigerian English. In Token 175, MNg uses manner DV DQ DGMHFWLYH WR PHDQ µZHOOPDQQHUHG¶7KLVXVDJHFRQIXVHV)%UZKRWUDQVFULEHGLWDVµ[[¶WRVKRZVKH could not even make a guess about the identity of the word. Finally, in Token 181, FTw is talking about what she did when she and her husband visited southern Germany. One of their favourite activities was eating pork knuckle (Schweinshaxe), a common dish in southern Germany, while drinking the beer for which Bavaria is famous. FMa does not understand this use of knuckle. (This token, particularly the repetition of the word knuckle, is discussed again in Extract 6.54 in Section 6.3.2.) Here I have focused on instances in which an unexpected meaning of a word is not understood. However, sometimes in ELF communication we find shifted meanings for words (compared to native-speaker usage) which do not cause a problem. For instance, consider used to. In British English, used to refers to states of affairs in the past (Carter and McCarthy 2006: 661), so he used to live here would mean that he once lived here but he no longer does. In contrast, in some ELF interactions it might mean that he is familiar with this place. See, for example, Extract 5.13, in which MIn is talking about his relatives who live in the Malaysian states of Perak and Johor. In line 6, he says used to and FMa assumes that he means he is familiar with Malaysia, so there is no misunderstanding.

5.1. Lexis

103

Extract 5.13. In+Ma : 86 Context : MIn is talking about his relatives in Malaysia 1 2 3 4 5 6 7 8 9

MIn FMa MIn FMa MIn FMa MIn

«VRZKLle i still have a lot of relative in (.) back in perak in johor very nice right right yeah @@ so i used to basically (.) mm VRPDOD\VLDEDVLFDOO\LW¶VD IDPLOLDUSODFHIRU\RX## right yeah yeah especially of course familiar because we are (.) in asean

A similar occurrence is shown in Extract 5.14, in which MHk says used to WR PHDQ µDFFXVWRPHG WR¶ and FTw appears to have no problem understanding him. While it is true that she misunderstands because the time zone, hearing it as at the time so (see Sections 4.1.7 and 4.1.8), it seems unlikely that this problem arises because of the meaning of used to. Extract 5.14. Hk+Tw : 1182 (Tokens 53 & 54) Context : MHk is talking about the problems of travelling 1 2 3

MHk

«and now you know you are jet lag you know because you used to get up in this time and because the time zone you know LVFKDQJH«

An alternative way to analyse these two occurrences of used to would be to note that there is a missing verb in the utterances, so i am used to malaysia would be standard native-speaker usage in Extract 5.13, and you are used to getting up at this time would be standard in Extract 5.14. Sentences that have no verbs in them will be discussed further in Section 5.2.4. Another word which sometimes has a shifted usage in the South-East Asian context is ever. However, once again people in ELF settings seem to understand it with no problem. In British English, ever can occur in questions and negative statements, but not in affirmative statements (Carter and McCarthy 2006: 81), but this restriction does not always apply in SouthEast Asia. As a result, in Brunei we find an utterance like P\PRWKHU¶VVLster ever wanted to adopt me (Deterding and Salbrina 2013: 63). And this usage is found in the CMACE data. In line 3 of Extract 5.15, MLs uses ever LQDQDIILUPDWLYHVWDWHPHQWWRPHDQµRQFH¶EXWWKLVGRHVQRWVHHPWRFDXVe any problems for FBr.

104

5. Lexis, grammar and code-switching

Extract 5.15. Ls+Br : 2548 Context: They are talking about the person who trains the Laos rugby team. 1 2 3 4

MLs FBr

«WKHO- l- last guy erm:: they had the training in my office? (.) he er:: now he study in the university already maybe (.) he the one who ever come here maybe @@@@ hh you can yeah you can try to ask him

Similarly, in Extract 5.16, MLs uses ever once more in an affirmative statement, but again this does not seem to cause any problems for FBr. Extract 5.16. Ls+Br : 2661 Context: MLs and FBr are talking about sports they like. 1 2 3 4

MIn FBr MLs

i like basketball (1.8) i ever played before but now (1.2) nothing  LGRQ¶WSOD\DQ\WKLQJ \RXGRQ¶WGRDQ\VSRUWVDQ\PRUH" \HDKQR L¶PVFDUHG«

However, there is one instance in which the use of ever does cause a misunderstanding, and this is shown in Extract 5.17. FMa asks MHk about the most beautiful thing he has even seen, and he replies with never instead of ever, which confuses FMa, so she repeats ever. It seems that MHk has probably just selected the wrong word here: maybe he intends to say ever but the wrong word comes out. Or maybe one could say that there is a spurious [n] at the start of the word, so perhaps the issue is actually phonetic. Extract 5.17. Hk+Ma : 1095 (Token 21) 1 2 3 4 5 6

FMa MHk FMa MHk FMa

and what is the most beautiful thing you have ever seen WKHPRVWEHDXWLIXOWKLQJVL¶YHnever seen is you know EVer seen yeah yeah (.) ever seen (3.3) would be you know really simple you know every day (.) sunsets ah @ okay very nice

Finally, I will consider one more expression that often has a shifted meaning in South-East Asia: until now. In Inner-Circle English, if I say he has been busy until now, that would mean he is no longer busy. But in places such as Singapore, it would mean he still is busy (Deterding 2007: 53). And this is how it appears to be used in the ACE recordings. Four instances of until now and one of until today are found in the data, and they

5.1. Lexis

105

are shown in Table 5.6. It seems probable that in all cases, the situation that is referred to continues to be true at the current time (though it is not entirely clear if that is the meaning in the final token, spoken by MLs). Table 5.6. Occurrences of until now (and until today) in the ACE data. Location In+Ma:1085 In+Ma:1499 In+Tw:646 In+Tw:864 Ls+Br:212

Spk. MIn MIn MIn MIn MLs

Context and they are still using as textbook until now DWOHDVWL¶PYLndicated er until now hopefully like until today for example then (.) i continue to bring so until now meaning that he: he is very close to (.) LW¶VHUOD]\WRZDNHXSHDUO\ XQWLOHUP XQWLOQRZ

This use of until now seems to be understood with no problems except in In+Tw:864 (Token 103), which is shown in greater detail in Extract 5.18. In this instance, FTw subsequently wrote as you know instead of until now in her transcription, so it is possible that the shifted meaning of the expression confused her. Although FTw transcribed it with a word that has no [l], there is in fact no L-vocalization for the word-final [l] in this token (as discussed in Section 4.1.6), and this lends support to the suggestion that the shifted meaning of the phrase may in this case have caused the problem. Extract 5.18. In+Tw : 864 (Token 103) Context: MIn is talking about the Chinese friends his son made at school 1 2 3

MIn

«HUWKHLUFKLQHVHIULHQGDOVRVtay with us and so on you know (.) so until now meaning that he: he is very close to (.) many of these er er er what do you call chinese students

In conclusion, there are a few cases in which the non-standard usage of a word (according to Inner-Circle norms) causes a misunderstanding to occur, but there are other cases in which the shifted meaning seems to have become established in the English that is spoken in South-East Asia so it does not cause problems in the ELF recordings that are analysed here. 5.1.5. Polysemes Some words are polysemes: they have two or more distinct but related meanings. For example, head can refer to the top of the body (its basic meaning) or the head of a household (one of its extended meanings), and foot is the lowest part of the body and it can also be the base of a mountain.

106

5. Lexis, grammar and code-switching

If a listener gets the wrong sense of a word, this can give rise to a misunderstanding. There are two clear instances of this in the CMACE data, involving pound and find respectively. Token 69 involves pound misunderstood by FTw, and it is shown in Extract 5.19. Extract 5.19. Hk+Tw : 1990 (Token 69) Context: MHk has been talking about the cost of raising children, claiming that in America it costs one million dollars to put a child through education. He has now started talking about the food students eat in university. 1 2 3 4 5 6 7 8 9 10 11 12 13 14

MHk

FTw MHk FTw MHk FTw MHk FTw

«EUHDNIDVW\RXNQRZOXQFK GLQQHUDOOEXIIHW DOO\RXFDQ eat (.) and then you know the kid also (.) f- freshman the first year they maybe one hundred fifty pound (.) when they fgraduate you know four year later. sometime they will be no matter girl or guys (.) two hundred pounds at least pounds yeah pounds. because they eating every day you know everything you know every you know time all you can eat oh you mean get weight the weight (.) mm EHFDXVHZH¶UHXVLQJQRWNLORJUDP VWLOODPHULFDQ using pounds are using pounds (.) ah

In this extract, MHk uses pound to refer to weight, but having lived for many years in England, FTw thinks instead it refers to money. One contributory factor in this instance is that MHk has just been talking about the cost of raising children, and this may be why FTw thinks of money instead of weight. In fact, of course, 0+N¶VXVDJHLVSHUIHFWO\VWDQGDUGLQWKLVFDVH (apart from the omitted plural -s on the end of the occurrence of pound in line 3). The problem is that pound has two distinct meanings, and FTw gets the wrong one. The second instance of a misunderstood polyseme is shown in Extract 5.20. In line 1 of this extract, FBr asks FJp how she finds the classes in English Language that she is taking, but FJp misunderstands the question, thinking it is about how she found out about the module, and she answers that her friend told her about it. Subsequently, after talking a bit about )-S¶VIULHQGFBr clarifies the question by asking GR\RXWKLQNLW¶VRND\ in lines 19 and 20, and FJp then gives a more appropriate response.

5.1. Lexis

107

Extract 5.20. Jp+Br : 444 (Token 130) Context: They have been talking about the English Language module FJp has been taking while she has been an exchange student in Brunei. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25

FBr FJp FBr FJp FBr FJp FBr FJp FBr FJp FBr FJp FBr FJp FBr FJp FBr FJp FBr

«RND\VR KRZKRZGR\RXfind the (.) classes heh:: (1.3) i think (.) there was one of my senior? from japan? she was in here until march? wait. is her name [name]? yeah yeah yeah yeah yeah yeah yeah yeah yeah yeah yeah yeah @ (.) actually yes i i just did er an interview with her as well er yeah i think so yeah i talked to her as well yeah yeah yeah then (.) her major is english literature so (.) she also take some (.) class about english (.) so yeah yeah i think she she introduce (.) this class to me okay \HDKPD\EHLGRQ¶WUHPHPEHUZKHQ# okay. so er how do you find it do you find it erm do you think LW¶VRND\"FDQ\RXIROORZLW"RU yeah yeah. not so difficult. PPDWILUVWLW¶VDELWGLIILFXOWEXWLWKLQNFRPSDUHGWR RWKHUFODVVLW¶VHDV\WRXQGHUVWDQG okay.

These seem to be the only two instances of misunderstood polysemes in the data analysed here.

5.2. Grammar Innovative patterns of grammatical usage are common among ELF speakers, who are generally more concerned about getting the message across than worrying too much about adhering to traditional notions of standard grammar. This can affect a range of syntactic issues, including singular/plural nouns, article usage before nouns, verbal suffixes including -s and -ed, missing verbs, preposition usage, and word order. But do these issues cause misunderstandings? I will consider each of them here.

108

5. Lexis, grammar and code-switching

5.2.1. Plural nouns The use of plural nouns for logically countable things, like furnitures, equipments, advices, researches and infrastructures, is common in Brunei (Deterding and Salbrina 2013: 52), Singapore (Deterding 2007: 42), and in many new varieties of English around the world (Mesthrie and Bhatt 2008: 53). And Hülmbauer (2013) suggests that plurals such as informations may be preferred in ELF settings, because they convey functional meaning even if they do not adhere to the rules of Inner-Circle Englishes. Does this usage contribute to misunderstandings? In the CMACE data, there are three instances in which the unexpected presence of a plural suffix might have contributed to a misunderstanding. In Token 113, shown in Extract 5.21, FTw is unable to understand vocabularies, and she was subsequently unable to come up with any suggestion for the word. One contributory factor here is that the main stress is on Malay (as discussed in Section 4.4.2), and another factor is that MIn¶V voice fades away on vocabularies. But it is also possible that the non-standard plural contributes to the problem. Extract 5.21. In+Tw : 1134 (Token 113) Context: MIn is talking about the different languages spoken in Brunei. 1 2 3

MIn

«NHGD\DQGRHVQRWVSHDNspeak malay (0.9) of course kedayan has many maLAY vocabularies (.) but LW¶VWRWDOO\VSHDN different language (0.9) yeah like say hakka DQGHUWHRFKHZ«

In line 4 of Extract 5.22, MIn says politician in the singular but then administrators in the plural, and this may have contributed to FTw hearing administrators as ambitious. The main issue in this example is probably the accompanying laughter (see Section 4.5.5); but the variable use of the plural suffix may be a contributory factor. Extract 5.22. In+Tw : 1985 (Token 122) Context: MIn is talking about his career choice in becoming an academic. 1 2 3 4 5 6 7

MIn

FTw MIn

«LGHFLGHGL¶OOL¶OO,¶//GRUHVHDUFKL¶OO:5,7( HU\RX know what what i LOVE more (.) but then in jakarta meaning if: you need to have some (1.1) er decent kind of LIFE then you have to (.) to be politician to be administrators oh so these are things so eventually then these are partly ',6FRXUDJHPHLQGHHGWKHUHZDVDSXVKIRUPHWROHDYH«

5.2. Grammar

109

In Extract 5.23, MHk says kinds of where other people would say kind of. FTw is unable to understand these words, and she transcribed them as counsel. It seems that in this case the spurious -s suffix on kinds may be a substantial factor in contributing to the problem. Extract 5.23. Hk+Tw : 1596 (Token 64) Context: MHk is talking about what you can do to reduce wastage. 1 2 3 4

MHk

MXVWLQGLYLGXDO\RXNQRZHUWRPHLW¶Vkinds of you know (1.2) awareness you know (.) what we can i mean (.) every you know HU WDONRUWKRXJKW\RXNQRZ WRPHLW¶VXVHOHVVZLWKRXW DFWLRQ«

These three instances all involve an added plural suffix. There are also three instances where the absence of a plural suffix might have contributed to the misunderstanding, though in all three cases, there are other factors involved and the missing suffix seems a minor issue. Three tokens from Extract 5.24 have already been discussed in Chapter 4, as hold is misunderstood as old (see Section 4.1.2), seven is heard as same with (see Section 4.1.9), and three is misunderstood as theory (see Sections 4.1.1 and 4.5.4), and there is a phonological explanation for all of these. But what about two being misunderstood as to? There seems to be no obvious issue with the pronunciation, and maybe the absence of an -s suffix on master is a contributory factor. However, this is probably a minor issue. The bigger problem seems to be that the words before and after two master are misunderstood. Extract 5.24. Ng+Br : 2093 (Tokens 162 to 165) Context: MNg is talking about the level of qualifications people in society have, and whether getting extra university degrees is useful. 1 2 3

MNg

if you see so many country you will see somebody hold (.) seven degrees two master three p h d . you have not seen it? yeah sudaQSUHVLGHQW«

Extract 5.25 has already been discussed in Section 5.1.3 in connection with idioms, as the use of hot cake to refer to someone in demand from employers is not familiar to FBr, and furthermore there is no [h] at the start of hot (see section 4.1.2). Although we can suggest that the absence of a plural suffix on cake is unexpected, it seems unlikely that FBr would have understood this idiom even if there had been an -s suffix on the word.

110

5. Lexis, grammar and code-switching

Extract 5.25. Ng+Br : 2742 (Token 173) Context: MNg is talking about how it is easy for some people to get a job after graduating from university. 1 2 3

MNg

«WKH\DUHQRWORRNLQJIRUDQ\WKLQJWKDWWLPHWKH\ZLOOEH calling them because they are hot cake (1.3) they need their service

The third instance involving an absent plural suffix is shown in Extract 5.26. One problem here is that, in an effort to deal with the effects of Lvocalization in tile, MHk pronounces the word as [taܼԥ], and then FTw hears it as tire (see Sections 4.1.6 and 4.5.4). Again, it seems unlikely that the presence of a plural suffix on the word (and also on year and thing in line 2) would have helped. Extract 5.26. Hk+Tw : 2364 (Token 75) Context: MHk is talking about is how a house in the United States needs constant maintenance. 1 2 3

MHk

«HYHU\WKLQJ\RXNQRZDWFHUWDLQWLPHOLNHWKHtile okay ten year you know you have to change because all the thing will be (.) not working «

I have discussed six instances where the presence or absence of a plural suffix may have contributed to misunderstandings; but in reality in none of these tokens does it seem to be a major issue, except perhaps with kinds of in Extract 5.23 (Token 64). And there are literally hundreds of instances where words have extra or omitted suffixes (from the perspective of nativespeaker English), and this does not seem to create any problems. It seems, therefore, that the presence or absence of the plural suffix is rarely an issue in making oneself understood in ELF settings. In fact, quite apart from the use of plurals like furnitures and equipments for logically plural entities, there are other emergent patterns for singular/plural nouns that may be widespread in new varieties of English around the world. Here I will discuss the use of a singular noun after one of, something that is common in Brunei English (Deterding and Salbrina 2013: 54). Table 5.7 lists some examples in which one of is followed by a singular noun in the ACE data, but where no misunderstanding occurs. Although it is true that FTw does not understand Italian in HK+Tw:456 (Token 40), this does not seem to be caused by the use of the singular father following one of; and there appear to be no misunderstandings in the other instances.

5.2. Grammar

111

Table 5.7. Occurrences of one of followed by a singular noun. Location Hk+Tw:456 Hk+Tw:1543 Jp+Br:452 Ng+Br:760 Ng+Br:766 Ng+Br:1533 Ng+Br:2456

Spk. MHk MHk FJp MNg MNg MNg MNg

Context one of you know italian father you know (.) he used the give up one of your t-shirt before you buy another one i think (.) there was one of my senior? from japan? one of my friend is there taking master of teaching LWKLQNKH¶s teaching from one of the school in brunei i ask one of the bruneian (.) you puasa {fasting}? i say i know one of my course mate now

One further issue related to the countability of nouns is the distinction between few (for count nouns such as books) and little (for non-count nouns such as money). This distinction is not always maintained in ELF interactions, but it rarely seems to create problems. For example, in Extract 5.27, MIn says are little to indicate that his children have only a few friends back home in Indonesia, perhaps influenced by the Malay/Indonesian sentence Kawan mereka sedikit saja OLWµTheir friends IHZRQO\¶ EXW)0DXQGHrstands him with no problem. Although some listeners might misunderstand this as saying that their friends are small, there is no evidence that that is the case in this extract. Extract 5.27. In+Ma : 702 Context: MIn is talking about is why his children like to come back to Brunei. 1 2 3 4 5

MIn FMa MIn

«VRPHDQLQJLQP\N- MY children mm for example leave and want to to LIVE in brunei because (.) when we return home and i brought them so meaning (.) their friend are DOORQO\OLWWOHLWKLQNEXWLQEUXQHL HYHU\RQHKHUH«

In conclusion, the variable usage of singular and plural nouns does not seem to cause problems in the ACE recordings. This is consistent with the findings of other researchers (e.g. Seidlhofer 2011: 125). 5.2.2. Articles Dewey (2009: 62) observes that the incidence of the definite article the in a corpus of ELF data is about the same as in a corpus of native speech, but the patterns of occurrence differ. In native speech data, the occurrence of the is largely determined by form, especially whether the noun it modifies

112

5. Lexis, grammar and code-switching

is generic or not; but in ELF interactions, articles carry a much broader discourse function, serving to indicate the salience of an object: important things tend to be preceded by the, while less important things tend not to be. And Dewey (2009: 67) notes that researchers working in applied linguistics too often focus on the form of language, analysing whether learners of English have adopted the normative patterns, but failing to note the functions that the articles may have. In other words, articles may carry some vital nuances of meaning that enhance the intelligibility of an utterance in an ELF setting and which may be overlooked by people engaged in corpus analysis. In addition to this shifted role for articles in indicating the salience of a noun, there are some innovative patterns of article usage observed in ELF discourse. For example, Cogo and Dewey (2012: 64) report that same as often occurs without a preceding article, and we find the same thing happening in the ACE data, particularly in the speech of MHk. Table 5.8 lists nine instances of same as with no preceding article, all by MHk. Table 5.8. Occurrences of same as with no preceding article in the data for MHk. Location Hk+Ma:327 Hk+Ma:541 Hk+Tw:700 Hk+Tw:943 Hk+Tw:1083 Hk+Tw:1700 Hk+Tw:1882 Hk+Tw:2086 Hk+Tw:2105

Context PHDQVWKH\¶UHFRPPXQLVW DQGVDPHDV\RXNQRZFKLQD get the id- the idea you know (.) same as you know american so to me (.) same as you know we (.) er we travel to talent (.) is exist in everybody (.) same as love you know but think about (.) s- same as what we (.) from very beginning to me is same as though after nine eleven (.) everybody you know same as the (.) why you know america spent driving you know (.) i mean same as america you know the is not one. same as america you know everyone get one (.)

It seems rather unlikely that this occurrence of same as with no preceding the causes misunderstandings. Quite simply, from the perspective of ELF intelligibility, the in front of same as is redundant. Or alternatively, the use of the before same as might carry a discourse function indicating heightened importance for the following phrase, so we might say that its absence actually conveys some meaning in the discourse. But are there other instances where the occurrence or omission of articles causes misunderstandings in the CMACE data? There are three instances where the absence of an article may have played some part in causing a misunderstanding. The first is shown in Extract 5.28, in which MHk pronounces next with an initial [l] and FMa does not understand the word the first time it is said.

5.2. Grammar

113

Extract 5.28. Hk+Ma : 1061 (Token 20) Context: FMa has asked MHk about his most exciting art project. 1 2 3 4 5 6 7

MHk FMa MHk FMa MHk FMa MHk

so usually is er is the most important (.) art project i did uh-uh would be next [lekst] hh the next [lekst] (.) one [w‫ݞ‬nt] okay you know would be the most (.) important

The main issue here is pronunciation (see Section 4.1.3), but note the missing article before next in line 3. When MHk repeats next in line 5, still with initial [l] but now including the article the, FMa understands him. It is likely that the inclusion of the article in line 5 has helped, so we might say that its omission in line 3 has contributed to the misunderstanding. (This instance will be discussed again in Sections 6.1.1, 6.3.2, and 7.1.) Another instance involving a missing article is shown in Extract 5.29. Extract 5.29. Hk+Tw : 38 (Token 29) Context: 7KH\DUHWDONLQJDERXW0+N¶VQDPHSDUWLFXODUO\WKHIDFWWKDWKHGRHV not use an English name. 1 2 3 4

MHk

« and all my brother and sisters you know. they have [you] know first name. english first name. my sister is jane? jean? and vivian? my brother is keith. only me. i refuse to have any english you know first name @@@

MHk appears to say that his brother and sisters have no first name, but then he goes on to list the first name of each of them. This leaves FTw somewhat perplexed. In fact, MHk subsequently confirmed that he was trying to say they have you know a first name. (He tends to use you know rather regularly in his speech.) Unfortunately, the you is omitted from you know, which is why it sounds like they have no first name. We can say that the primary cause of this misunderstanding is the omission of you (see Section 4.5.3). However, if MHk had inserted an article, the misunderstanding might have been avoided, even if the you were still omitted: they have [you] know a first name. So perhaps we can say that the omission of the article plays some part in this misunderstanding. The third instance in which a missing article might have contributed to a misunderstanding is shown in Extract 5.30. In line 3, MIn uses pattern to indicate an event whose occurrence is repeated and predictable, so he is

114

5. Lexis, grammar and code-switching

stating that the occurrence of terrorism is unpredictable; but FMa is unable to understand pattern and she subsequently guessed it might be part of. This is partly because of the back vowel in the first syllable of pattern (see Section 4.2.2), and also LW¶V QRW DV SDWWHUQ is a somewhat unusual way of expressing the idea that something is repeated (as discussed in Section 5.1.4). However, the absence of an article before pattern might also be a contributory factor. If MIn had said LW¶V QRW DV D SDWWHUQ, perhaps FMa would have understood him. Extract 5.30. In+Ma : 1499 (Token 91) Context : MIn is talking about the historical analysis of terrorism. 1 2 3

MIn

DWOHDVWL¶PYLQGLFDWHGHUXQWLOQRZKRSHIXOO\QRWHU WKHQ er (.) er but it will come out again and again but but LW¶VQRWDV as pattern this and so on it can come and (.) go

There is in addition one instance in which the presence of an article might have played a part in the misunderstanding. This is shown in Extract 5.31, in which FBr cannot understand Black Swan. Extract 5.31. Ch+Br : 2013 (Token 10) 1 2 3 4 5

FBr FCh FBr FCh FBr

... what about movies. you like to watch movies? yes: i just saw the (.) black swan. i liked it. rex one? yeah the natalie portman OH: RIGHT:

The main problem here is that [r] rather than [l] occurs in black (see Section 4.1.3), and furthermore the vowel in black is not very open (see Section 4.2.2). However, it is also possible that the unexpected occurrence of the before the name of the film contributes to the misunderstanding. In conclusion, the presence or absence of an article may have played a minor role in four instances of misunderstanding, but apart from these, there is little evidence that articles have much impact on intelligibility. 5.2.3. Verb suffixes Cogo and Dewey (2012: 49) observe that the absence of an -s suffix on present tense verbs following a third person singular subject is widespread in their data, and Seidlhofer (2011: 107) similarly reports that this phe-

5.2. Grammar

115

nomenon is commonly found in the VOICE data. In fact, the variable occurrence of the -s suffix on present tense verbs seems to be one of the most widely reported features of ELF grammar that set it apart from most nativespeaker usage. Is there any evidence that it causes misunderstandings? One might claim that the -s suffix on the verb helps listeners to identify what the subject is; and if it is absent, then they may have difficulty in identifying the subject correctly. However, while its omission might cause momentary confusion for Inner-Circle speakers, there seems little evidence that it results in too many problems in ELF conversations. Just two instances where the absence of an -s suffix on the verb might have contributed to a misunderstanding have been found in the CMACE data. In Extract 5.32, MIn is talking about what his son is doing, and he says now work, which FTw subsequently guessed might be no work. It is possible that, if MIn had said now works, there would have been no misunderstanding, though we might note that he is now working or just now working would be more usual, so perhaps we might say that the absence of the -ing progressive suffix on the verb is also a factor. Extract 5.32. In+Tw : 609 (Token 100) Context : MIn is talking about what his children are now doing. 1 2 3 4

MIn FTw MIn

«WKHROGHVWRQHHULQHU ZKDWGR\RXFDOOFanada actually study? yeah er after completing master then now work (.) so PHDQLQJLW¶V#LW¶VYHU\GLIILFXOWWRR#

One further instance involving the -s suffix is shown in Extract 5.33, in which FBr hears LW¶Va post instead of LW¶VKHOSus. One might say that the [s] is misplaced, so it helps us would have been more likely to be understood than LW¶VKHOS us. But probably the most important factor in this instance is actually the absence of [h] at the start of help (see Section 4.1.2). Extract 5.33. Ng+Br : 3383 (Token 174) Context : MNg is talking about getting a job after graduating from university. 1 2 3

MNg

«WKHdegree DOVRL¶PQRWVXUHPD\EH LW¶V on the right one L¶PQRWVXUHLW¶V like (.) LW¶V help us again to know (.) am i right (.) right? when you go to university or somehow  «

I will also consider one instance in which has is used instead of have even though there is a plural subject, as this might be considered an in-

116

5. Lexis, grammar and code-switching

stance of a spurious -s suffix. In Extract 5.34, has rise spoken by MNg is misunderstood as are try by FBr. Extract 5.34. Ng+Br : 2031 (Token 161) Context : MNg is talking about qualifications, and how some people are able to get promotion to a high position without a university degree. 1 2 3

MNg FBr MNg

«GXULQJWKDWWLPHHYHQLI\RXJRWRFKDUWHUEDQN in bandar mm? some of them has rise to (.) supervisor OHYHODOVR«

However, other factors are involved, including the omission of [h] at the start of has (see Section 4.1.2). In addition, rise rather than the expected risen occurs after the perfective auxiliary, and this might be a factor. Let us now analyse the occurrence of past tense (and past participle) suffixes on verbs. There are hundreds of cases in the ACE data in which speakers use a present tense verb to express a past event, and these occurrences almost never seem to cause a problem. Sometimes, it seems that tense selection is simply random. For example, in line 3 of Extract 5.35, FCh says come in line 3, but then in line 4 she says came, and it seems hard to suggest any rationale for the choice. Extract 5.35. Ch+Br : 505 Context : FCh is talking about her last vacation. 1 2 3 4 5 6

FCh FBr FCh FBr

«DQGZHnt back to china for a vacation yeah and when i come back i: (.) transit in bangkok (.) so i spent oh a week in bangkok and chenmai and then (.) came back to brunei again. RND\«

It appears that ELF speakers are able to discount the tense of the verb in interpreting what is said. In fact, I can just find one instance (apart from the absent final -n in has rise that was discussed in Extract 5.34 above) where a missing -ed suffix might have contributed to a misunderstanding, though in this case it is not a tense suffix but rather the derivational -ed suffix that converts the verb relate to the adjective related. In line 5 of Extract 5.36, FTw hears believer of instead of related with. However, as discussed in Section 4.5.4, related with in this instance is pronounced as [riliwiv], so it involves omission of the [t] as well as the -ed suffix .

5.2. Grammar

117

Extract 5.36. Hk+Tw : 845 (Tokens 45 & 46) Context : MHk is talking about Zen painting. 1 2 3 4 5

MHk FTw MHk

«DQGWKHQZK\\RXGRLWDJDLQDJDLQ DQGWKLVLVQRW about you know (.) the result. it is about the process you know mmm VR LW¶VNLQGVRIDOVRWRPH\RXNQRZrelated with arts too

In fact, there is one other token in the same recording where related with spoken by MHk is misunderstood, and it is shown in Extract 5.37. FTw subsequently guessed art is related with to be artists we live with. However, in this instance, related with is pronounced as [wiletiwiv] (see Section 4.1.3), so the -ed suffix is in fact there (even if there is no final [d]), and we cannot analyse this as involving omission of an -ed suffix. Extract 5.37. Hk+Tw : 1816 (Token 65) Context: MHk is discussing the purpose of art. 1 2 3 4

MHk FTw MHk

... the social issue (.) and everything (.) because (.) we said that art is related with truth you know (.) and artist you know sorry the truth? \HDKWKHWUXWK \RXNQRZLIWKLVLVWUXWK\RXNQRZ«

In conclusion, omission of the 3rd person singular -s suffix on a present tense verb rarely results in misunderstandings occurring, and omission of the -ed suffix does not seem to cause a problem.

5.2.4 Missing verbs Many languages do not require a sentence to have a verb, and this is true of the first languages of some of the speakers in the ACE data. For example, in Malay, you can say Buku itu bagus µ7KLVERRNJRRG¶ DQGVLPLODUO\LQ Chinese you can say 征㛔Ḏ⼰⤥ (Zhè-EƟQ VKnj KpQ KăR, µ7KLV ERRN YHU\ JRRG¶ ZLWKQRQHHGIRUDFRSXODr verb in either language. Not surprisingly, utterances with no verbs are also found in the ACE data. But do they cause misunderstandings? I will discuss five instances of misunderstandings in which an absent verb may be one of the factors. In Extract 5.38, FBr hears owns an internet company instead of also in that company, and the absence of a copular verb

118

5. Lexis, grammar and code-switching

may be the main factor. If FCh had said she is also in that company, then there would probably have been no problem. One might note the presence of L-vocalization in also and the pronunciation of in that as [ܼn‫ܭ‬t] (see Sections 4.1.6, 4.1.1, and 4.2.2); but these seem to be minor factors. Extract 5.38. Ch+Br : 390 (Token 2) Context : 7KH\DUHWDONLQJDERXWZKDW)&K¶VIDPLO\GR 1 2 3

FBr FCh FBr

all right. so what about your sister. what is she doing? she also t- er (.) she also in that company DKVKH¶VDOVRZRUNLQJ"

In Extract 5.39, FBr does not understand the cut off will out, and she subsequently guessed the words as they got of way out. One might note that there is minimal aspiration on the [k] in cut which results in it being heard as got (see Section 4.1.7). However, the absence of the copular verb be after the modal will seems to be the main factor in this instance. Alternatively, one might say that out is being used as a verb in this sentence. Extract 5.39. Ng+Br : 868 (Token 157) Context : MNg is talking about education in his country, particularly the selection of students for university. 1 2 3 4 5

MNg

FBr

«HDFKEDVHRQIRXUKXQGUHG.) one hundred percent (.) so after they finish that exam (.) the cut off will out (.) so each university (.) they have your name already (.) they will now look at the cut oh okay:

In Extract 5.40, MHk omits a verb such as have in they no you know snow plough, and this may have contributed to the misunderstanding, though probably the absence of [l] in plough is the main factor that causes FTw to misunderstand him (as discussed in Section 4.1.4), as she subsequently transcribed it as power. Extract 5.40. Hk+Tw : 310 (Token 36) Context : MHk is talking about what happens when it snows in Texas. 1 2 3 4

MHk FTw MHk

«so LW¶VWHUULEOHand then they n:o you know snow plough truck you know mm so they just let you know the (.) snow become ice freeze «

5.2. Grammar

119

Similarly, in Extract 5.41, FMa does not understand afraid, which she hears as are feel, largely because of the missing [r] (see Section 4.1.4), but also perhaps as a result of the absence of a verb in line 3. Extract 5.41. Hk+Ma : 564 (Token 17) Context: MHk is talking about his work as a painter of murals. 1 2 3

MHk FMa MHk

«\RXJHWWKHLGHDLW¶VDKXJH huge painting (.) everybody afraid about the same thing (.) huge

Finally. in Extract 5.42, MLs says \RXGRQ¶WKRW. FBr cannot understand him, partly because the delayed [t] at the end of hot gets lost in the subsequent laughter (see Section 4.5.5), but also because of the absence of a main verb after GRQ¶W. When MLs repeats the phrase adding feel, FBr is able to understand him. Extract 5.42. Ls+Br : 1911 (Token 146) Context : MLs is talking about women wearing Islamic head covering and other women thinking they must feel hot. 1 2 3 4

MLs FBr MLs FBr

«WKHROGZRPHQFRPHDQGWHOO RK\RXGRQ¶Who:t? @@@ sorry? \RXGRQ¶WIHHOKRW" madam oh @ @@@@ okay @

In conclusion, there are a few instances where the absence of a verb seems to contribute to the occurrence of a misunderstanding, but not many.

5.2.5. Prepositions The use of prepositions is a notoriously difficult area of English, particularly because it is so irregular. You can talk about VRPHWKLQJVRZK\FDQ¶W you discuss about something? And speakers in ELF settings do not seem too concerned about such things, as their listeners can usually understand discuss about perfectly well (Cogo and Dewey, 2012: 59). In fact, there seem to be no instances where an added or missing preposition following a verb causes a problem in the ACE data; and there are just two instances in which a missing preposition between two nouns might have contributed to the misunderstanding, both involving an absent of.

120

5. Lexis, grammar and code-switching

In Extract 5.43, there is no of between couple and father(s), and FTw was subsequently unable to come up with a suggestion for either of the tokens of couple in lines 2 and 3. We might note, however, that there is Lvocalization in couple (see Section 4.1.6), and that in addition in line 2 FTw is unable to understand Italy because it is pronounced as [ܼti] (see Section 4.5.4). It seems that this instance of misunderstanding involves multiple issues, and the absence of the preposition is just one of them. Extract 5.43. Hk+Tw : 443 (Tokens 38 & 39) Context : MHk is talking about learning art when he was at school. 1 2 3 4

MHk

«WKHSULPDU\VFKRROLVDOOFDWKROLFVFKRRO VRWKH\KDYH\RX know the father you know from italy (.) you know couple you know father (.) you know couple fathers you know is really (.) er (.) artistic

The second instance involving a missing of is shown in Extract 5.44. The main problem here, as discussed in Section 5.1.3, is that FTw is not familiar with the idiom religion is the opium of the people, so this instance should be classified as lexical, not grammatical. Furthermore, Marx is pronounced with no [s] on the end (see Section 4.1.8). Once again, it seems that there are multiple issues involved in this instance of misunderstanding, and the missing preposition is a minor factor. Extract 5.44. In+Tw : 962 (Tokens 105 & 106) Context : MIn is discussing the claim WKDWµQRUHOLJLRQLVRQHNLQGRIUHOLJLRQ¶ 1 2 3 4

MIn

«VRPHDQLQJWKHVDPHOLNHHU karl marx for example er consider that religion is opium society but (1.0) communism itself eventually is religion PHDQLQJ\RXGRQ¶W yRXGRQ¶W have that concrete idea of god but you have a different kind «

In conclusion, there seem to be almost no instances in which preposition usage causes a misunderstanding in the CMACE data. 5.2.6. Word order Unexpected word order has been reported to disrupt intelligibility more than other features of grammar (Björkman 2009: 242). Just two tokens in the CMACE data might be classified under this category of unusual word order.

5.2. Grammar

121

In Extract 5.45, MHk says the day next rather than the more usual the next day, and this seems to have confused FMa, who subsequently wrote OHW¶Vinstead of next in her attempted transcription. Extract 5.45. Hk+Ma : 1128 (Tokens 22 & 23) Context : MHk is discussing why he finds the sunset every day the most beautiful thing he has ever seen. 1 2 3 4

MHk

«each sunset is different (2.1) different from each day (.) but to 86FRXJKV!ZHWKLQNLW¶VMXVWWKHVDme (1.1) we call today and we call (.) i- the day before yesterday (.) the day next will be tomorrow (1.1) but (.) every day is different «

However, in addition to this unexpected word order, next begins with [l] rather than [n] (see Section 4.1.3). It is not clear which factor is more important in causing this instance of misunderstanding: word order, or the use of [l] instead of [n]. Quite probably the misunderstanding only occurs because of the combination of these two factors. The second instance of misunderstanding arising from unexpected word order is shown in line 4 of Extract 5.46, for which FTw subsequently wrote leading instead of living in her attempted transcription. Extract 5.46. In+Tw : 1253 (Token 114) Context : MIn is talking about his family background. 1 2 3 4

MIn FTw MIn

... er when my father sent me to school of course then i: now i guide my: (.) brother my sister meaning are you the oldest i er i am: er the living old- eldest and such

MIn says living eldest rather than the more usual eldest living, and this seems to confuse FTw. A lack of distinction between long and short vowels may also have contributed to this instance, as discussed in Section 4.2.4; but the unexpected word order is probably the more important factor.

5.2.7. Other grammar issues Here I will deal with two remaining instances where grammar seems to have played a part in the misunderstanding but it is hard to classify the problem using the categories above. Both are from the Hk+Tw recording.

122

5. Lexis, grammar and code-switching

The first part of the instance of misunderstanding shown in Extract 5.47 has already been discussed in Sections 5.1.1 and 5.1.2 above, as FTw is not familiar with dorm as an abbreviation for dormitory, and meal plan is a phrase that is not widely used outside of America. In addition, FTw is unable to understand add in line 7, largely because of the absence of [d] at the end of the word (see Section 4.1.9), and she subsequently guessed that it might be air. But we might also note that meal you know add is a somewhat unusual way of expressing the idea that all the meals are added together. Extract 5.47. Hk+Tw : 1969 (Tokens 66 to 68) Context: MHk is discussing the cost of tertiary education in America. 1 2 3 4 5 6 7 8

MHk FTw MHk FTw MHk

... you send the kid you know to the university (1.3) tuition is a major part of it you know (.) all the (.) you know (.) dorm (.) the meal plan (.) just killing view pen what is that yeah meal you know eating you know ah meal meal you know add (.) because (.) breakfast you know lunch dinner all buffet (.) all you can eat «

The second instance involving miscellaneous unexpected grammar is shown in Extract 5.48, in which FTw hears all full instead of over. One problem is the insertion of two tokens of you know between over and Singapore. But we might also note that it is over Singapore is an idiosyncratic way of saying that people in Brunei once had a higher income than those in Singapore. (In addition, as already noted, over seems to have a medial [f] and to be stressed on the second syllable ± see Sections 4.1.7 and 4.4.1.) Extract 5.48. Hk+Tw : 2240 (Token 73) Context: MHk is suggesting that people in Brunei used to be richer than elsewhere, but maybe they no longer are. 1 2 3

MHk

... brunei is you know (.) they said that you know in the past LW¶VoVER \RXNQRZHU \RXNQRZVLQJDSRUH ZKHQZH¶UH WDONLQJDERXW\RXNQRZWKHSHRSOHLQFRPH\RXNQRZ«

In conclusion, there seem to be few instances in which grammar plays a significant role in causing a misunderstanding to occur. Omitted verbs and unusual word order may occasionally play a part, but misunderstandings caused by other features of grammar are rare.

5.3. Code-switching

123

5.3. Code-switching Code-switching can be regarded as one form of accommodation (Cogo 2009): speakers know that they share languages other than English with their conversational partners, so switching into those other languages is one way of acknowledging their common background and also utilizing all the linguistic resources available to make themselves understood. Various other aspects of accommodation will be discussed in Chapter 6. It generally serves to enhance intelligibility, so it will be dealt with under WKHKHDGLQJRIµUHSDLUV¶%XW,ZLOOFRQVLGHUFRGH-switching in this chapter, as there are a few instances in which it causes misunderstandings to occur. In bilingual societies such as that of Brunei, regular code-switching between English and Malay is the norm in all but the most formal of genres (McLellan 2010), and avoiding code-switching in informal conversations when you know that your interlocutor speaks both English and Malay would be regarded as rather strange behaviour. In fact, sticking resolutely to one language in informal settings is sometimes seen as putting on airs. However, one would expect code-switching to be less common in ELF interactions, as the speakers usually do not have a common first language. For example, Firth (1996: 255) reports that remarkably few instances of code-switching occurred in his corpus of telephone conversations in an international company based in Denmark, largely because most of the interactants in his data did not share a first language. Nevertheless, code-switching does in fact sometimes occur in ELF conversations. For example, Smit (2010) reports that non-German students in an Austrian school started to use German quite often after they had been together for a long period of time; Kalocsai (2011) shows how participants from Turkey, Germany and England on an Erasmus exchange programme in Hungary began to use some Hungarian words in their conversations; and Cogo (2009) gives examples from informal conversations among teachers from different backgrounds at an institute of higher education, showing how code-switching was used to achieve accommodation. In fact, Klimpfinger (2009) reports that 2,184 instances of code-switching can be found in the 700,000 words of the VOICE data, just over half of them involving switching into the first language of the speaker. The most straightforward instances of code-switching are when two participants share a language. In Extract 5.49, FTw probably assumes that MHk speaks some Mandarin Chinese (though Cantonese is his first language) so he should be able to understand 什 (yuán µIDWH¶  DQG LQGHHG

124

5. Lexis, grammar and code-switching

MHk does seem to understand it, himself suggesting fate as an equivalent in line 10. This, then, is an example of the successful use of code-switching to build rapport between two people in an ELF conversation. Extract 5.49. Hk+Tw : 184 Context: 7KH\DUHWDONLQJDERXWKRZRQHILQGVRQH¶VVSRXVH 1 2 3 4 5 6 7 8 9 10 11 12

FTw MHk FTw MHk FTw MHk FTw MHk FTw MHk FTw

but chinese we call it yuán {fate} yeah DQG LW¶V\RXPD\EH\RXUSDVWOLIH yeah we believe in karma yeah and yeah there is maybe something like that you know is kind of fate you know you met somebody @@@

However, in the same Hk+Tw recording, there is one instance of a misunderstanding arising out of code-switching, and it is shown in Extract 5.50. The problem with zen being pronounced as [s‫ܭ‬n] has already been discussed in Section 4.1.7. But the issue with chán (⾵µ]HQ¶ LVDOLWWOHGLIIHrent. MHk pronounces it as [t‫ܭݕ‬n], using an open-mid vowel with the same quality as that in zen, and FTw asks him to confirm if he means [t‫ݕ‬an], using the fully open vowel that would be expected in standard Mandarin (Duanmu 2007: 38). The problem probably arises because MHk¶VILUVWODnguage is Cantonese, so his pronunciation of Mandarin may be a bit nonstandard at times. (As the pronunciation of chán involves a non-English word, it was not discussed in Chapter 4.) Extract 5.50. Hk+Tw : 741 (Tokens 43 & 44) Context: MHk is talking about religion. 1 2 3 4 5 6 7

MHk FTw MHk FTw MHk

... in j- japanese they w- they call zen you know yeah (.) but you know in chinese is chán {zen} you know (.) \RXNQRZLW¶VNLQGV of buddhism chán? \HDKLW¶V\RXNQRZ zen. did you say zen? zen yeah you know

5.3. Code-switching

125

In Extracts 5.49 and 5.50 above, the language involved in the codeswitching is a language that is shared by the two participants. But codeswitching also sometimes takes place into a language that is not shared by both participants. One of the instances discussed by Cogo (2009: 265) involves a teacher from Germany switching into Italian to accommodate to the first language of her colleague. And there are some similar instances of speakers switching into the first language of their conversational partners in the ACE data. In line 7 of Extract 5.51, FJp says one word of Malay, makan µWRHDW¶  to show that she has learned a few words of the language while in Brunei, providing clear evidence of accommodation towards FBr. Extract 5.51. Jp+Br : 746 Context: FBr has mentioned that code-switching between English and Malay is common in Brunei. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

FBr FJp FBr FJp FBr FJp FBr FJp FBr FJp FBr FJp FBr FJp

«LQRQHVHQWHQFHZHFDQVD\ mm we can have malay and english words mm in one sentence (1.0) yeah er erm (.) everyone here can understand it they often say go makan {eat} yeah @@@@ RKWKDW¶VDJRRGZRUGWKDW¶V LVWKDWRQHRIWKHILUVWwords you learn in malay? hhh yeah yeah @@@@@ yeah. makan is erm (.) a quite an important word. mm yeah

In Extract 5.52, MNg uses the Malay word puasa rather than its English equivalent fasting, and this might be seen as him accommodating towards the language of FBr. Alternatively one might regard puasa as a word borrowed from Malay that has become established in Brunei English, so MNg is just using the normal word that is found in the local nativized variety of English. Indeed, there are rather a lot of words that have been borrowed from Malay or Arabic into Brunei English, particularly for referring to religious activities (Deterding and Salbrina 2013: 91), so one could say that MNg is just using one of them.

126

5. Lexis, grammar and code-switching

Extract 5.52. Ng+Br : 1529 1 2 3 4 5 6 7 8 9

MNg

FBr MNg

«GXULQJWKHpuasa {fasting} something right (.) one of the examples (.) i ask one of the bruneian (.) you puasa? i say right. why (.) why are you puasa  LW¶VPDQGDWRU\>MNg:name] (.) another one and oh my mother compulsory for it. the other one told me (.) no. LW¶VDQLVODPLFZD\\RXKDYHWRGRWKDWEHFDXVH of your body of whatever (.) you see different something. that is the way everybody believes WKDW¶VWKHWKLQJLWKLQNZHKDYHWREHSUHSDUHGIRU yeah

The examples above all involve individual words, but there are a few examples of code-switching involving non-English phrases or sentences. Although FMa is ethnically Tamil, she actually speaks some Chinese, and in Extract 5.53 she accommodates towards FTw by saying 孚⋶宕 (ML΁QJ KXi\ǎµVSHDN&KLQHVH¶ )7ZH[SDQGVRQWKLVVD\LQJ 孚⋶宕嶇⤰⤰孚⋶宕 (ML΁QJKXi\ǎ JƝQPƗPDML΁QJKXi\ǎµVSHDN&KLQHVHZLWKPXPVSHDN&KiQHVH¶ DOORIZKLFK)0DDSSHDUVWRXQGHUVWDQGZLWKQRSUREOHP)7ZWKHQ continues using some more Chinese before switching back to English. Extract 5.53. Tw+Ma : 1024 Context: FTw is talking about encouraging her son to speak Chinese. 1 2 3 4 5 6 7 8 9 10 11

FTw FMa FTw FMa FTw FMa FTw FMa FTw

«VRLVDLGWRKLP ZRXOG\RXSOHDVH XVLQJFKLQHVH (.) talk to me ML΁QJKXi\ǎ {speak Chinese} ML΁QJKXi\ǎ JƝQPƗPD ML΁QJ KXi\ǎ {speak chinese with mum speak chinese} KXi\ǎ JjRVPƗPD{tell mum} uh \zQJKXi\ǎ shuǀ{use chinese speak} @ (1.3) and he (.) and he tried. he tried very hard. and after «

The most extensive use of code-switching in the ACE data is in the In+Ma recording, and some instances of misunderstanding occur there. MIn is aware that FMa is from Malaysia, so even though she is Tamil, MIn assumes that she can understand some Malay. In line 2 of Extract 5.54, MIn says baju melayu µ0DOD\FORWKHV¶ LQWKHH[SHFWDWLRQWKDW)0DZLOONQRZ this term. In fact, in this case he is right, as she does understand it. Then in

5.3. Code-switching

127

line 3, MIn seems unable to think of the appropriate term, so after a pause of 1.6 seconds, FMa suggests baju kurung to refer to the close-fitting long dress worn by women in Brunei, and MIn accepts this as appropriate. This instance of code-switching is therefore successful, particularly as there is no straightforward English equivalent for baju kurung. Extract 5.54. In+Ma : 1955 (Token 92) Context: MIn is talking about the inevitable changes taking place in society. 1 2 3 4 5 6 7 8 9

MIn FMa MIn FMa MIn FMa

brunei used to have (.) you know you look at it (.) brunei have dress in baju melayu {Malay dress} beautiful er what is it (1.6) baju kurung? baju kurung @@ and men are using the (.) teluk belanga {traditional style of Malay dress} the what is it eh whatever it IS uh-uh

However, just a few seconds later in this conversation, in line 7 of Extract 5.54, MIn uses another Malay term teluk belanga to refer to the traditional Malay style of clothing for men (named after a place in Singapore called Telok Blangah where a royal Malay court was once located), and FMa does not understand this phrase. It seems likely that MIn realizes this, as teluk belanga is a fairly obscure term for a particular style of clothing, so he says whatever it is to suggest that it is not important; and FMa gives a minimal uh-uh backchannel to encourage him to continue. Although we might say that code-switching has resulted in a misunderstanding occurring in this case, it has not disrupted the conversation too much. In another instance from the In+Ma recording, in Extract 5.55 MIn uses the term pehin menteri ugama µKRQRXUDEOHPLQLVWHURIUHOLJLRQ¶ DQGWKLV is a phrase that FMa cannot understand. She gets ugama µUHOLJLRQ¶  EXW she cannot decipher the rest of it. Extract 5.55. In+Ma : 2029 (Tokens 93 & 94) Context: MIn is talking about the religious authorities in Brunei. 1 2 3 4 5

MIn FMa MIn

for one thing many of the (.) er brunei reLIGious elite you call it (.) pehin menteri ugama {honourable minister of religion} mm-mm VRWKLVLVLW\RX¶UHpehin {honourable title} and (.) awarded title of datuk {honourable title} too

128

5. Lexis, grammar and code-switching

One problem here is that pehin is a noble title in Brunei which is not used in Malaysia. But maybe MIn does not realize this, so he expects FMa to understand the term. In fact he says pehin again in line 4, and once more FMa cannot understand it, though she has no problem with datuk, another honourable title in Malay, because this one does occur in Malaysia. One PLJKW DGGLWLRQDOO\ QRWH WKDW )0D¶V LQDELOLW\ WR GHFLSKHU religious elite in line 1 (see Section 4.4.2) may have contributed to her failure to understand pehin menteri ugama. Finally, there is Extract 5.56, which occurs near the start of the In+Ma recording. MIn makes reference to Sungai Teluk Intan, a place in Perak State in Malaysia. However, FMa does not realize that Sungai µULYHU¶  LV part of the name, partly because it is separated from the rest of the name by what you call, and she subsequently transcribed it as somewhere. Extract 5.56. In+Ma : 119 (Token 80) Context: MIn is discussing visiting some of his relatives in Perak state in Malaysia. Bagan Datuk and Teluk Intan are places in Perak, and Teluk Anson is an alternative name for Teluk Intan. 1 2 3 4 5 6 7 8

MIn FMa MIn FMa MIn FMa MIn FMa

... small bandar (.) bagan datuk something like that okay bagan datuk @@@ right okay so it¶s er (.) sungai {river} what you call teluk intan teluk intan it was teluk anson eventually yes

Given that this is a place name, perhaps it should not really be classified as code-switching. However, it is included here because FMa fails to understand the Malay word sungai and thinks it is an English word instead.

5.4. Miscellaneous Inevitably, there are a few instances of misunderstanding which involve factors that are hard to classify under the categories suggested in this and the previous chapter. Two such instances are discussed here. In line 2 of Extract 5.57, FMa is unable to understand river, and she subsequently transcribed it as mirror. But on listening to it, there does not

5.4. Miscellaneous

129

seem to be much wrong with the pronunciation of river. Maybe the medial consonant has some devoicing (as discussed in Section 4.1.7), but that would seem to be a fairly minor factor. Extract 5.57. Hk+Ma : 1153 (Tokens 24 & 25) Context: MHk is discussing why he finds the sunset every day the most beautiful thing he has ever seen. 1 2 3 4

MHk

«LWLVDSKLORVRSKLFDO \RXNQRZphrase (.) you never step in the same river WZLFHLGRQ¶WNQRZZKHWKHU\RXKHDUG about it you know (.) what does it mean you know the river is nealways there

Two other factors may have contributed to this instance: first, in line 1 FMa is unable to understand phrase because of the omitted [r], and she subsequently transcribed it as phase (see Section 4.1.4); and second, maybe the philosophy is a little unusual. MHk suggests that you never step in the same river twice, because just like the sunset every day, the river is always changing. While this is a profound idea, perhaps when it is combined with a few features of unexpected pronunciation, it confuses FMa. Finally, there is Extract 5.58, in which FBr misunderstands weather as waiter. Certainly, the pronunciation of the medial TH as a plosive is a contributory factor. However, the main problem seems to be that in line 7 MLs indicates that he wants to talk about food, but despite foregrounding food as his new topic, he does not actually talk about it. Under these circumstances, it is not too surprising that FBr hears weather as waiter. I will discuss this instance again in Section 6.2.6, in connection with establishing a topic. Extract 5.58. Ls+Br : 83 (Token 135) Context: MLs is discussing his first impressions of Brunei. 1 2 3 4 5 6 7 8 9 10 11

MLs FBr MLs FBr MLs FBr MLs

«IRUWKHIRUHLJQHUZKRVtay in brunei er:: the transportation  EHFDXVHXVXDOO\LGRQ¶WVHHWKHWD[L yes or public er:: (.) public transportation like the bus (.) or i- in KHUHGRQ¶WKDYHVXEZD\RUWUDLQULJKWEHFDXVH!VPDOO! no and s- so but (.) erm: for the food (2.2) for the weather the weather er: i would say is similar in my country laos (.) you how how is it similar LW¶VKRW

130

5. Lexis, grammar and code-switching

5.5. Summary of lexis, grammar and code-switching It seems that lexical issues, particularly unfamiliar words and phrases, cause most of the misunderstandings that have been discussed in this chapter. Grammar rarely seems to cause a problem, though missing verbs may sometimes be an issue, and also other unusual features of grammar may occasionally combine with unexpected pronunciation to result in misunderstandings occurring. Finally, code-switching generally achieves accommodation between the speakers, but once in a while it causes miscomprehension instead. Having discussed all 147 instances of misunderstanding in the CMACE data, I will now analyse how they are dealt with and also consider some of the strategies that speakers adopt in order to try and ensure that misunderstandings are avoided in ELF conversations in South-East Asia.

Chapter 6 Repairs

This chapter considers how misunderstandings are signalled and repaired. Most of the instances of misunderstanding that have been identified in the ACE data do not actually result in any obvious breakdown in communication, but it is still valuable to consider the ways in which they are dealt with, and whether the repairs are different from native-speaker patterns or not. I will first consider self-initiated repairs and then other-initiated repairs, including those that are not actually signalled. Repairs can be defined to include the avoidance of misunderstandings as well as fixing those that occur (Kaur 2009), so in addition this chapter considers some of the strategies which are employed to ensure that ELF interactions proceed successfully without a breakdown in communication.

6.1. Self-initiated repairs In many self-initiated repairs, it is likely that no misunderstanding actually occurs ± the speakers correct themselves even though there is no need. Such instances of self-initiated repair are of interest as they represent one means of ensuring that misunderstandings are avoided. I will first consider self-initiated self-repairs, either when the speakers correct themselves or when they offer an explanation with no overt prompting from their listeners. I will then discuss self-initiated other-repairs, in which the speaker asks for help in sorting out a problem. 6.1.1. Correcting oneself The most basic kind of self-initiated repair is when the speaker corrects the pronunciation, the word selection, or the grammar of what they have just said. In line 4 of Extract 6.1, MIn pronounces envy as [ܼQ‫ޖ‬YDܼ] but then imPHGLDWHO\ FRUUHFWV LW WR >‫ޖ‬HQYL@ ,W UHPDLQV XQFHUWDLQ ZK\ KH LQLWLDOO\ SUonounces it as [ܼQ‫ޖ‬YDܼ] if he is aware of the standard pronunciation. But this self-initiated repair ensures a misunderstanding is either avoided or immediately resolved, which is why it has not been included in the CMACE data.

132

6. Repairs

Extract 6.1. In+Tw : 2368 Context: They are talking about the importance of exercise. 1 2 3 4 5

FTw MIn FTw MIn

chinese we have a saying (.) resting aha s- s- therefore you can walk (.) f- further right. yeah exactly so i i enVY [ܼQ‫ޖ‬YDܼ ] i ENY\>‫ޖ‬HQYL@P\  what you call er my (.) my friend in er «

However, self-corrections of pronunciation are not very common, partly because many of the speakers are not aware of the ways in which their pronunciation is non-standard and so they do not know how to fix it when misunderstandings occur. In Extract 6.2, MHk pronounces next with an initial [l] (see Section 4.1.3), and this confuses FMa, who subsequently reported that she initially heard it as lext. Extract 6.2. Hk+Ma : 1061 (Token 20) Context: FMa has just asked MHk about the most exciting art project he has been involved in. 1 2 3 4 5 6 7

MHk FMa MHk FMa MHk FMa MHk

so usually is er is the most important (.) art project i did uh-uh would be next [lekst] hh the next [lekst] (.) one [w‫ݞ‬nt] okay you know would be the most (.) important

In this extract, MHk may realize that FMa has not understood him, so he repeats next really carefully, ensuring that all three final consonants [kst] are articulated, and he even adds a spurious [t] on the end of one in line 5. Furthermore, he adds the article the before next in line 5, and this may help to solve the problem (see Section 5.2.2), as FMa later reported that she understood the second token of next correctly. But note that MHk fails to fix the [l] at the start of next. He appears not to know that this is the main issue or how to sort it out. This instance is discussed again in Section 7.1. 6.1.2. Unprompted paraphrase Paraphrasing what one has just said is a regular means of avoiding misunderstandings and also of resolving some that occur. In Extract 6.3, MIn

6.1. Self-initiated repairs

133

describes himself as a vagrant, which he pronounces as [fr‫ݞ‬gr‫ݞ‬n]. His intended meaning is that he has migrated from his hometown in search of work, which is rather an idiosyncratic usage of vagrant (as was discussed in Section 5.1.4). Maybe he is aware that FMa has not understood this word, so he spontaneously provides an explanation, saying that he has moved around rather a lot in his life. Extract 6.3. In+Ma : 444 (Token 83) Context: MIn is talking about his background in Indonesia. 1 2 3 4 5 6

MIn

... i consider myself like my grandfather (.) is vagrant meaning travel from one place so (.) so meaning that er (.) i i er was born in ponorogo study in ponorogo until (.) er lower secondary ups- upper secondary i moved to: (.) the capital surabaya and from there (.) er moved to middle east and eventually to er (8URSHDQGWR1257+DPHULFD«

In fact, this explanation does not enable FMa to identify the word vagrant, and she subsequently transcribed it as fragrant; but this failure to identify the word does not matter too much, as the gist of what MIn is saying seems to be perfectly clear to FMa. One could therefore say that this instance of self-initiated repair is successful even though it does not enable the listener to identify the word. In terms of the categories of intelligibility discussed in Section 2.3, in this instance MIn lacks intelligibility at the word level, but his paraphrase ensures he achieves comprehensibility at the utterance level (Smith 1992: 76; Nelson 2011: 21). Actually, MIn seems to be aware that some aspects of his speech are problematic, and as a result he has a tendency to paraphrase things quite often. One of the phrases he uses most often is meaning that, which occurs 25 times in the In+Ma recording and 55 times in the In+Tw recording, and these figures do not include examples like Extract 6.3 above where meaning is not followed by that. While not all the instances of meaning that or meaning introduce an explanation, as MIn seems to use it as a kind of general-purpose connector, some are followed by an elaboration, and as with Extract 6.3 above, this paraphrasing by MIn generally helps to repair misunderstandings even if it does not always result in the original word being successfully identified. In Extract 6.4, he paraphrases niche. This does not enable FTw to identify the word, as in fact she does not know this word (see Section 5.1.1), and she subsequently transcribed it as needs. But it ensures that the conversation progresses, even if the pause of over two seconds in )7Z¶VUHSO\suggests that the continuation is not totally smooth.

134

6. Repairs

Extract 6.4. In+Tw : 2098 (Token 124) Context: They are talking about living in Brunei. 1 2 3 4

MIn FTw

ZKHQ\RXOLYHKHUHDFWXDOO\ LW¶VYHU\RSHQLQDVHQVHVR they are giving us niche PHDQLQJWKH\¶UHJLYLQJXVRSSRUWXQLW\ to (.) to have your life yeah i think (2.3) er learn to appreciate things

In lines 6 and 7 of Extract 6.5, MIn clarifies the meaning of virgin, a word that FTw does not understand because of the [ܼԥ] in the first syllable (see Section 4.3). Even though once again the explanation does not enable FTw to identify the word, she manages to get the main idea. Extract 6.5. In+Tw : 1359 (Token 115) Context: They are talking about interesting places to visit in Indonesia. 1 2 3 4 5 6 7

FTw MIn FTw MIn FTw MIn

and the other place i like is erm lombok wow yeah lombok right yes !LW¶VDNLQG!RISRRU SRRUFRXVLQRIEDOL correct yeah @ and especially still virgin in the sense then (.) not mu- polQRWPXFK 32/OXWHGE\WKHWRXULVWV«

In Extract 6.6, MNg provides an explanation of hot cake. Although FBr still cannot decipher this phrase, partly because there is no [h] at the start of hot and also because she is not familiar with this idiomatic use of hot cake (see Sections 3.1.3, 4.1.2 and 5.1.3), the paraphrase enables her to get the gist of what MNg is saying. Extract 6.6. Ng+Br : 2742 (Token 173) Context: MNg is talking about how it is easy for some people to get a job after graduating from university. 1 2 3 4 5

MNg FBr MNg

«they are not looking for anything. that time they will be calling them because they are hot cake (1.3) they need their service yeah \HDKQRWOLNHRKLZDQWWRQRQRQRQRQRQRQRZKHQWKH\VHH«

A little different from providing a paraphrase, but equally effective, is giving an illustrative example. In Extract 6.7, FTw cannot understand eitch WKHOHWWHU µK¶  EHFDXVH LW is pronounced as [ܼt] (see Section 4.1.8), so she

6.1. Self-initiated repairs

135

hears it as it. But MIn illustrates what he is saying with French allo as the equivalent of English hello, and then French Henri (with no [h]) compared with English Henry, and these two examples seem to solve the problem. Extract 6.7. In+Tw : 62 (Token 97) Context: They are talking about pronunciation of names. FTw has just said that people sometimes mispronounce her name. 1 2 3 4 5 6 7

MIn

FTw MIn FTw

«LQIUHQFK VReitch LVQRWSURQRXQFHLW¶VWKDWEHFDXVHRI that actualO\\RXGRQ¶WKDYHeitch LQLQIURQWRILW \RXGRQ¶W have (0.8) yeah so meaning in french like allo instead hello allo right @@ !WKDW¶VULJKW! henri henry henri right yeah okay

Another instance of self-initiated repair that is of interest is when the listener does not realize that they have misunderstood something. An example of this is in Extract 6.8 (which has already been discussed in Section 5.1.5). Extract 6.8. Jp+Br : 444 (Token 130) Context: They are talking about the English Language module FJp is taking. 1 2 3 4 5

FBr FJp FBr FJp

«RND\VR KRZKRZGR\RXfind the (.) classes heh:: (1.3) i think (.) there was one of my senior? from japan? she was in here until march? wait. is her name [name]? yeah yeah yeah yeah

LQOLQHVWRWKH\GLVFXVV)-S¶VIULHQG ± see Extract 5.20 in Section 5.1.5) 16 17 18 19 20 21 22 23 24 25

FJp FBr FJp FBr FJp FBr FJp FBr

i think she she introduce (.) this class to me okay \HDKPD\EHLGRQ¶WUHPHPEHUZKHQ# okay. so er how do you find it do you find it erm do you think LW¶VRND\"FDQ\RXIROORZLW"RU yeah yeah. not so difficult. PPDWILUVWLW¶VDELWGLIILFXOWEXWLWKLQNFRPSDUHGWRother FODVVLW¶VHDV\WRXQGHUVWDQG okay

In line 2 of Extract 6.8, FJp gets the wrong meaning for find, and in her answer she explains how she found out about the class rather than giving

136

6. Repairs

her impressions towards it. After a while, FBr repeats the question, this time paraphrasing the intended meaning of find in lines 19 and 20, and FJp then provides an appropriate response. We can therefore say that these cases all involve self-initiated selfrepairs, though we cannot be certain about two things: first, in some of the instances there may have been something in the facial expression of the listener to indicate they had not understood (so it may not actually be selfinitiated); and second, we cannot always be certain about the degree to which the repair is successful, given that in some cases the listeners were still unable to decipher the original word when they subsequently attempted a transcription. 6.1.3. Asking for help Self-initiated other-repairs in which the speaker asks for help in finding a word or explaining a concept are not common in the ACE data, but they do occur. In Extract 6.9, FJp asks for help using the phrase how to say, and FBr offers a possible completion which FJp accepts. Extract 6.9. Jp+Br : 137 Context: FJp is talking about why she came to Brunei as an exchange student. 1 2 3 4 5 6

FJp FBr FJp

«LDSSO\"VRPH XQLYHUVLW\" LQVRXWK-east asia like philippine (.) or malaysia (.) but my english score is not enough so it cannot how to say oh: you cannot go there @@ \HDKEXWLGRQ¶WNQRZZK\ RQO\VSHO!XE d can accept me? (.) so:

In Extract 6.10, MIn seems unable to think of the phrase baju kurung, and he says what is it in an apparent request for help. Or possibly he is just hesitant to use this Malay term, given that FMa is ethnically Tamil, not Malay. However, after a pause of 1.6 seconds and also an alveolar click µWVN¶by MIn suggesting an element of frustration (Laver 1994: 175), FMa offers baju kurung, which MIn immediately accepts as the right phrase. When MIn subsequently uses another rather obscure Malay term teluk belanga in line 7, this gives rise to an instance of misunderstanding, as FMa is not familiar with this phrase; but it does not disrupt the progress of the conversation too much. In fact, the ability of FMa to offer the appropriate term baju kurung in line 4 builds rapport between the two interactants

6.1. Self-initiated repairs

137

(as discussed in Section 5.3), despite the fact that FMa is subsequently not able to understand teluk belanga. Extract 6.10. In+Ma : 1955 (Token 92) Context: MIn is talking about the changes taking place in society. 1 2 3 4 5 6 7 8 9

MIn FMa MIn FMa MIn FMa

brunei used to have (.) you know you look at it (.) brunei have dress in baju melayu {Malay dress} beautiful er what is it (1.6) baju kurung? {close-fitting dress worn by women in Brunei} baju kurung @@ and men are using the (.) teluk belanga {traditional style of Malay dress} the what is it eh whatever it IS uh-uh

However, it is not always clear that a phrase like what is it represents a request for help. In Extract 6.11, FTw seems to assume that what you call is asking for assistance, especially as there are pauses and pause particles before and after it, and she offers a suggestion, but MIn immediately finLVKHVKLVRZQVHQWHQFHZKLFKLVZK\)7Z¶V words in line 3 are unclear). Perhaps MIn is just using what you call as a kind of discourse connector, not a request for help. Maybe we should classify this overlap as one kind of misunderstanding, though there seems to be no disruption of the conversation, so alternatively we could describe it an instance of attempted collaborative completion. Extract 6.11. In+Tw : 834 Context: MIn is talking about bringing up children in a multicultural environment. 1 2 3 4 5

MIn FTw MIn

«P\WZRGDXJKWHUVP\621HU EXWHUPHDQLQJWKH\PLQJOH with many different (.) different er (.) er what you call erm (.) xx er communities yeah meaning that er for example WKHFKLQHVHWKHLQGLDQV HU«

Extract 6.12 is another instance in which MIn may be asking for help. In line 2 he appears to be struggling to think of a word, and in line 3 FTw suggests volcano as a possibility, but it seems unlikely that this really is the word MIn is searching for (partly because the main volcano in Bali is actually in the east). He continues his next turn with improvise (though it is not clear what he means by that, and in fact improvise is not a word that FTw

138

6. Repairs

knows ± see Section 5.1.1), and only then does he echo volcanoes, apparently accepting it as a suitable completion of what he was trying to say. Extract 6.12. In+Tw : 1475 (Token 116) Context: MIn is talking about Ubud in Bali. 1 2 3 4 5 6

MHk FTw MIn FTw

«LWKLQk the west side so meaning ubud is the (.) the (.) er what do you call er: the: volcano oh yeah impo- or what you can improvise. meaning beautiful hotel and so on and volcanoes !PPP!«

Sometimes the request for help may not be explicit, and it may be signalled by means of a pause rather than a direct request (Kaur 2009: 111). In Extract 6.13, MHk seems to start saying material but stops after the first syllable, and FTw completes it for him. Then in line 3 MHk confirms that this is what he was intending to say. Extract 6.13. Hk+Tw : 1552 Context: MHk is talking about wastage in the modern world. 1 2 3 4 5 6 7

MHk FTw MHk FTw MHk

«WKHQZHZLOOOLYHLQWKLVVRFLHW\OHVV(.) ma- (0.5) material material or world create all this you know use all the energy create this kind of thing (.) we were talking about you know er er this (.) yesterday you know how long you know for the (.) nature «

While it is not clear if this really is an instance of self-initiated request for help, the collaborative completion of the utterance seems to be successful in allowing MHk to proceed with what he is talking about. Collaborative completions such as this are discussed further in Section 6.3.4.

6.2. Responses to misunderstandings In Table 6.1, I have attempted to classify the response to all 147 instances of misunderstanding in the CMACE data under one of eight categories. Many of these might be considered other-initiated repairs, though in other cases, especially those involving silence, there may be no initiation for the repair. Furthermore, Table 6.1 includes some cases of self-initiated repairs.

6.2. Responses to misunderstandings

139

Table 6.1. Responses to misunderstandings Response Asking for clarification Correcting Silence Backchannel Selecting part of the utterance Changing the topic Laughter Non-awareness Total :

Number of Instances 15 2 59 46 9 3 11 2 147

Silence is the most common response to misunderstanding, followed closely by providing a backchannel, which is often a minimal utterance like mm. In both these cases, the listener does not overtly signal the misunderVWDQGLQJDGRSWLQJWKHµOHWLWSDVV¶strategy of hoping that it will naturally sort itself out (Firth 1996). In fact, in only two of the categories in Table 6.1, asking for clarification and correcting, are the misunderstandings actually clearly signalled, so I will consider these two categories first. There are too many instances of misunderstanding to deal with each one in detail, particularly for categories such as silence and backchannel in which there are so many examples. So just a few instances will be presented in order to illustrate each of the categories.

6.2.1. Asking for clarification The most obvious example of other-initiated self-repair involves one participant asking another for clarification. This is quite rare in the CMACE data, partly because the participants do not know each other and asking for clarification can be a bit embarrassing, although it might be more common in communities of practice such as the students in the school in Vienna that Smit (2010) studied, as the participants there felt that they knew each other sufficiently well to be quite direct in asking for clarification when it was required. Nevertheless, fifteen such instances have been identified in the CMACE data, which means about one in ten instances of misunderstanding that have been collated in the corpus involve the listener asking for clarification. One example is in line 6 of Extract 6.14 in which FTw asks about pounds, which she has mistakenly assumes refers to money (see Section 5.1.5).

140

6. Repairs

Extract 6.14. Hk+Tw : 1990 (Token 69) Context: MHk has been talking about the cost of raising children, claiming that in America it costs one million dollars to put a child through education. He has now started talking about the food students eat in university. 1 2 3 4 5 6 7 8 9 10 11

MHk

FTw MHk FTw MHk FTw

«EUHDNIDVW\RXNQRZOXQFK GLQQHUDOOEXIIHW all you can eat (.) and then you know the kid also (.) f- freshman the first year they maybe one hundred fifty pound (.) when they fgraduate you know four year later. sometime they will be no matter girl or guys (.) two hundred pounds at least pounds yeah pounds. because they eating every day you know everything you know every you know time all you can eat oh you mean get weight the weight (.) mm

In fact we might note that FTw keeps quiet the first time MHk says pound in line 3 and she only asks for clarification when he says pounds again a few seconds later in line 5. MHk then elaborates, and eventually FTw realizes that pounds refers to weight, not money, so the misunderstanding is resolved. In Extract 6.15, FTw asks for clarification of rain forest, which she hears as wind forced, largely because of the [w] at the start of rain (see Section 4.1.3). Extract 6.15. Hk+Tw : 2799 (Tokens 77 to 79) Context: MHk is comparing living in Brunei with his native Hong Kong. 1 2 3 4 5 6 7 8 9 10

MHk FTw MHk FTw MHk

... i think this you know rain forest (.) is really important. because you know brunei actually is in the heart of you know this rain forest you know wind? (1.2) the you know the rain forest you know? i mean (.) all s- (.) green stuff? around us. that is really help to create you know cooler you know so (.) LW¶VDEHWWHU DLUWKHN- quality is better is it air