Jenkins 1

pfcll 'l l s
0 ^ 5 5
2-0 OO
W \
The Phonology of
English as an
International Language:
New Models, New Norms,
New Goals
Jennifer Jenkins
OXTORD
U N IV E R S IT Y PR ESS
O X fO R D
U N İVE R SI T Y P R t S S
Great Clarendon Street, Oxford 0x2 6 d p

Oxford University Press is a departm ent o f the University o f Oxford.
it furthers the Universıty's objecnve o f exceUence in research, scholarship,
and education by publishîng worldwide in
Oxford N ewYork
Aucldand Cape Town Dar es Salaam Hong Kong Karachi
Kuala Lumpur Madrid Melbourne M exicoC ity Nairobi
New Delhi Shanghai Taipei Toronto
W ith offices in
Argentina Austria Brazil Chile Czech Republic France Greece
Guatemala Hungary Italy lapan Poland Pom jgal Singapore
Stmıh Korea Switzerland Thailatıd Turkey Ukraine VîetnalR
o x f o r d and o x f o r d e n g l i s h are registered trade marks o f

Oxford University Press in the UK and in certain other countries
© Oxford University Press 2000

The moral rights o f the author have been asserted
Database right Oxford University Press (maker)
First published 2000
2017
10
No u n a u th o r iz e d p h o to c o p y in g
Ali rights reserved. No part o f this publication m ay be reproduced,
stored in a retrieval svstem, or transmirted, Ln any form or b y any means,
without the prior permission in ıvriting o f Oxford University Press,
or as expressly permitted by law. or Under terms agreed u itlı the approprıate
reprographics riglıts organization. Rncjuiries con cem in g neprodııction
outside the scope o ftlıe above should he seni to the ELT Rights Department,
Osford University Press, at the address above
You must not circulate this book in any other binding or cover
and you must impose this same condition on any acquirer
Any vvebsites referred to in this publication «te in the public domain and
their addresses are provided by Oxford University Press for inform ation only.
ü \îoro University Press disclnims anv^csponsibiliry for the content
isbn: 978 o 19 442164 5
Printed in China
Contents
Acknowledgements
Introduction
1 The background: Changing patterns in the use of English

The historical shift
Changing ownership; changing terminology
Appropriate pedagogy for an international language
The EIL phonological problem: where do we go next?
2 The variation problem 1: Inter-speaker variation

Inter-speaker variation
Inter-speaker segmental variation and its effects
Inter-speaker suprasegmental variation and its effects
3 The variation problem 2: intra-speaker variation

L l and IL intra-speaker variation: a distinction
Phonological intra-speaker variation and its effects on
interlanguage talk
4 intelligibility in interlanguage talk

What do we mean by intelligibility?
Defining intelligibility in interlanguage talk
Bottom-up and top-down processing
The role of phonology in ILT: miscommunication in
the ILT data
intelligibility and the spread of English
Conclusion
5 The role of transfer in determining the phonological core

The complex process of L l phonological transfer
Conclusions: transfer, intelligibility, and teachability
6 Pedagogic priorities 1: Identifying the phonological core

Establishing the Lingua Franca Core
The origin of the Lingua Franca Core 131
Features of the Lingua Franca Core 134
Redefining phonological error and correctness for EIL 158
7 Pedagogic priorities 2: Negotiating intelligibility in the ELT 165

classroom
Accommodation theory and intra-speaker variation in ILT 167
Communicative efficiency and interlanguage 175
Accommodation and IL repertoire 176
Accommodating classrooms 180
8 Proposals for pronunciation teaching for EIL 195

An overhaul of pronunciation teaching in English language 199
teacher education
An overhaul of pronunciation testing 212
Radical improvement in the status of ‘NNS’ EIL 215
pronunciation teachers
Pronunciation learning for ‘native speakers’ of English 227
Aftenvord: The future of the phonology of EIL 233
Bibliography 237
Index 251
*
Acknowledgements
Many people have contributed in their different ways to this book. First, 1
would like to acknovvledge my debt to ali the learners of English who
provided the data, especiallv learners at Angloschool. Birkbeck College,
Imperial College, and King’s College London. Without them, the book
could not have been vvritten. in addition, the students on my master’s and
undergraduate courses at King’s College London över recent years have
shown a lively interest in the project, asking probing questions and making
perceptive comments vvhich enabled me to refine my core proposals.
I would also like to thank my colleagues at King’s, especially Sara
Garcia-Peralta, John Field, and Neil Murray, for their generous
encouragement. Many colleagues \vorking in the field of phonologv and
phonetics around the vvorld have been helpful and supportive. Specıal
thanks are due to Judv Gilbert, Brvan Jenner, John Levis, Maya Leon Meis,
Jonathan Marks, Larry Morgan, Paul Tench, Michael Vaughan-Rees, and
Robin Walker. Barbara Seidlhofer has been a source of inspiration and a
mine of information, and also provided invaluable comments on a
substantial chunk of the final draft.
Much of the research on vvhich the book is based vvas carried out during
my doctoral studies at the Institute of Education, University of London. My
gratitude goes to John Norrish for his guidance and generous support at
that time, and to fellovv students Koo Yevv Lie and Arifa Rahman for the
insights they gave me during our numerous discussions.
I vvould like to express my sincere thanks to those at OUP who have been
involved in the book, and especially to Cristina Whitecross for her constant
encouragement, to Julia Sallabank for her patience, energy, and dedication
during the preparation of the manuscript, and to Anne Conybeare tor ali
her help in the earlier stages. Above ali, I owe an immcnse debt to Henry
Widdowson. His commitment from the start and his generosity in the
sharing of his vvisdom and time have had a profound influence on the
development of my ideas.
The author and publisher are grateful to the follovving Copyright holders
vvho have given permission to reproduce extracts and adaptations of
Copyright material:
Blackvvell Publishers for an extract from ‘Double Standards: teacher
education in the Expanding Circle’ by B. Seidlhofer from \Vorld Englishes
18/2: 238 (Blackvvell Publishers).
vi The Phonology o f English as an International Language
Oxford University Press for extracts from ‘Sociolinguistics and Second

Language Learning’ by Lesley Milroy from Language and Understanding ©
G. Brown, K. Malmkjaer, A. Pollitt, and J. Williams 1994; Language
Teaching: A Scheme for Teacher Education: Pronunciation by Christiane
Dalton and Barbara Seidlhofer © Oxford University Press 1994; and
Communication in the Language Classroöm by Tony Lynch © Tony Lynch
1996.
The Society of Authors as the Iiterary representative of the estate of
L. P. Hartley for an extract from The Go-Between.
The Times Higher Education Supplement for an extract from ‘English is
as English does’, a review of A. Pennycook: The Cultural Politics o f English
as an International Language (Longman, 1994), published in the THES
Book Review Section 7 July 1999.
Although every effort has been made to trace and contact Copyright
holders before publication, this has not been possible in some cases. We
apologize for any apparent infringement of Copyright and if notified, the
publisher will be pleased to rectify any errors or omissions at the earliest
opportunity.
To Jo h n , N ick, and H arriet
Introduction
For the first time in the history of the English language, second language
speakers outnumber those for whom it is the mother tongue, and
interaction in English increasingly involves no first language speakers
whatsoever. The expansion of the use of English in this direction has been
documented and discussed by applied linguists throughout the past decade
(see, for example, Crystal 1988; Rampton 1990; Phillipson 1992;
Widdowson 1994a; Brumfit 1995; Sridhar 1996; Crystal 1997). Precise
figures vary, according to which authority one consults, how the different
second language users of English are categorized, and in particular, at what
level of proficiency these users are considered to be speakers of English at
ali. Nevertheless, the trend remains clear. Indeed, Crystal suggests that if
one accepts a criterion of ‘reasonable competence’, there may be as many
as 1,350 million second language speakers compared to around 337 million
first language speakers. Even the most conservative estimate based on
‘native-like fluency’ would stili place second language speakers at around
335 million (1997: 60-1). Either way, this trend is likely to continue apace
in the 2 1 st Century.
The only really surprising element, then, is the relative failure of English
Language Teaching (ELT) pedagogy—with the exception of the teaching of
Business English (see Alexander 1996)1—to adjust its methodologies in line
with this changing pattern, in which the goal of learning is more often to be
able to use English as a lingua franca in communication with other ‘non-
native speakers’, i.e. as an international language, than as a foreign
language in communication with its ‘native speakers’. Since it is in their
pronunciation that the existing and emerging second language (L2) varieties
diverge most from each other linguistically, it is arguably this linguistic area
that most threatens intelligibility. This is the area, therefore, that most
demands attention if international communication is to be successfully
promoted through the English language as the trend continues into the new
century. Meanwhile, the majority of EFL teacher training and education
courses, both preservice and inservice, persist with phonology syllabuses
that assume a ‘native-speaker’ interlocutor. They therefore involve elements
that are unnecessary, unrealistic, and, at worst, harmful for preparing
teachers to equip their learners with pronunciation skills appropriate to an
international use of English.
2 The Phonology o f English as an International Language
The purpose of this book is to explore the phonology of English from an

international perspective. This is done in two stages. The first, which
involves descriptıon and analysis, is essentially an examination of how
speakers of English as an International Language (EIL) behave
phonologically, by means of data drawn from lingua franca contexts. The
second stage moves on to a reconsideration of the problems of mutual
phonological intelligibility and acceptability, with the aim of facilitating the
use of EIL. This involves recommendation and, to an extent, prescription in
the discussion of specific new goals for pronunciation teaching.
It will quickly become evident that the concept of mutual intelligibility
for EIL is itself problematic for tvvo main reasons. First, it poses a majör
pedagogic dilemma. Attempts to work towards the promotion and
preservation of mutual international intelligibility will inevitably conflict
with the growing intellectual and socio-psycbological needs to accord the
burgeoning L22 varieties of English parity vvith first language (Ll) varieties
as acceptable norms. The solution proposed here is the establishing of a set
of ‘nuclear norms’ for ali L2 speakers of English (and receptively for L l
speakers also), thus in many ways the phonological counterpart of Quirk’s
(1982) proposals for grammar. Outside this nuclear core, speakers vvould
then be unconstrained in their use of L l features of pronunciation, in other
words, of local phonological norms. Hovvever, the second problem is that
phonological intelligibility is itself extremely difficult to isolate and pin
down. This, in turn, means that the identification of a set of nuclear norms
to guarantee mutual phonological intelligibility in international contexts is
likely to be a complex process.
Nevertheless, once we have identified such a phonological core, we vvill
be able to advocate and implement a far more realistic approach to
phonology vvithin ELT pedagogy. It vvill be possible to establish a clearer
distinction betvveen learners’ productive and receptive phonology, vvith
pronunciation syllabuses no fcnger being required to embrace large
numbers of fine details in an attempt to guide learners to approximate the
speech of ‘native speakers’.
In addition, it vvill become easier to incorporate notions of ‘teachability’
and ‘learnability’ into pronunciation teaching. Some phonological features,
like nuclear stress, vvill be categorized as ‘teachable’ because of the clear-
cut, generative nature of their rules, and others, like pitch movement, as
‘learnable’ (acquirable) outside the classroom—though only after a good
deal of exposure to the language— because of the complexities involved.
This, in turn, vvill enable pronunciation teaching and materials to shift the
focus from ‘vvhat is convenient for teachers to teach’ to ‘vvhat is effective for
learners to learn’ (Seidlhofer and Dalton 1995: 145). We vvill see also that
it is possible for ELT pedagogy to exploit accommodation, itself an area
much overlooked and under-researched in relation to second language
pronunciation teaching, not only for its mediating role in EIL intelligibility,
Introduction 3
but also for its as yet unrecognized contribution to L2 phonological

acquısıtion.
The book begins by discussing the issue of the changed ownership of
English and the implications of the growth of EIL as they relate specifically
to phonological intelligibility. In the second chapter, we briefly consider the
problem of negative attitudes towards language variation in general. The
second chapter goes on to study in greater detail the effects of phonological
variation, here between speakers of different L ls (inter-speaker variation)
and, in Chapter 3, within individual speakers according to sociolinguistic
context and socio-psychological need (intra-speaker variation).
Having examined what L2 speakers actually do, we move on in Chapter
4 to a discussion of intelligibility as the basis for establishing which core
features of pronunciation are likely to promote successful communication
in EIL. Then, in Chapter 5, we consider the complexity and extent of
pronunciation transfer and its implications for the feasibility of teaching the
core. In Chapter 6, bearing in mind the discussions of the two previous
chapters, we settle down to the business of identifying the phonological
core in detail. In Chapter 7, we turn to the problem of devising procedures
for the ELT classroom vvhich will encourage the negotiation of intelligibility
and, thus, the use of the core in both the production and reception of
speech. We conclude, in Chapter 8, with a discussion of the urgent need for
theorists and practitioners alike to take EIL phonology seriously and to
translate its implications into radical proposals for pedagogy, instead of
continuing to promote ways of describing and teaching pronunciation
vvhich have become irrelevant to the changed role of English in the
multilingual vvorld.
It should now be apparent that the issue of phonology for international
English is one vvhich has significance for the vvhole future of the English
language on a vvorldvvide scale. Nevertheless, not only has it been largely
ignored pedagogically, but it has only rarely filtered through into ELT
theoretical literatüre. This is, no doubt, partly the result of the fact that
pronunciation has itself been marginalized during recent years as a result of
difficulties in aligning it vvith and incorporating it into communicative
approaches to language teaching.
Undoubtedly, some important discussion has taken place since the late
1970s among a small circle of vvriters, including Brumfit (1982, 1995);
Kachru (1985, 1992); Quirk (1985); Smith (1983, 1992); Strevens (1980);
and Widdowson (1982, 1984, 1994a), a propos the ovvnership of English
and problems of mutual intelligibility, particularly as they relate to ESL
contexts. Hovvever, even in the vvorks of these authors, the crucial
phonological issue tends to be relegated to a brief footnote or paragraph. It
has thus been left largely to be played out in society newsletters, such as
that of the International Association of Teachers of English as a Foreign
Language (IATEFL) and its Pronunciation Special Interest Group’s
publication Speak Out, which by their very nature reach a relatively small
readership. A further purpose of this book is therefore to rekindle a wider
theoretical interest in phonology and thereby stimulate theoretical
discussion of phonology for EIL.
One final introductory point: the title of this book is an intended echo of
Pennycook’s The Cultural Politics o f English as an International Language
(1994). His book has been profoundly influential on the development of my
own thoughts. However, while I share many of his views, I come to some
different conclusions: English is the international language at present, so
rather than argue in terms of the past why this should not be, I prefer to
look ahead to ways in which we can make the language more cross-
culturally democratic, under the ‘ownership’ (in Widdowsonian terms) of
ali who use it for communication, regardless of who or where they are. This
firmly places me, pedagogically speaking, in the opposite camp to that of
‘real English’ which, I believe, has at present the effect of deflecting
attention away from the nature of EIL.
In attempting to ‘democratize’ the English language, I identify
pronunciation as the area of greatest prej udice and preconception, and the
one most resistant to change on ali sides. So, although many of the issues
at stake are at heart cultural and political, they are embedded in and
symbolized externally by phonology and phonological attitudes. It is,
therefore, to changes in these that we must look if English is to achieve true
integrity as an international language, and thus we have another majör
motivation for giving EIL phonology a high profile.
Notes
1 In this respect note Widdowson’s argument that ‘English as an
international language is English for specific purposes’ (1997: 144).
2 The term ‘L2’ is used th#oughout to contrast with ‘L l ’, i.e. to indicate
English that has not been learnt as a first language or ‘mother tongue’. It
does not distinguish between the ESL (English as a Second Language) and
EFL (English as a Foreign Language) varieties of English.
1 The background:
Changing patterns in the
use of English
English is England’s language but the world’s treasure.

Liu Dailin, Deputy Dean of the Foreign Language Department
at the China Central Radio and TV University, quoted in the
E L Gazette, September 1996
The historical shift

The teaching of English to speakers of other languages has a history
stretching back to the late 15th Century. From that time onwards, English
has been taught, albeit intermittently, to people for whom it was not their
mother tongue, vvhether to further trading and commercial interests,
promote empire, facilitate the everyday survival of refugees and other
migrants, or for a combination of these causes.
Until fairly recently, the goal of such teaching was straightforward:
learners wished primarily to be able to communicate effectively vvith native
speakers of English, vvho vvere considered by ali to be the ovvners of the
language, guardians of its standards, and arbiters of acceptable pedagogic
norms. This vvas the situation regardless of the status of English as a
‘foreign’ or ‘second’ language, respectively EFL (taught to those for whom
English had no internal function in their L l country) and ESL (taught either
to those for vvhom English had an internal function in their L l country or
to those vvho had emigrated to English L l countries). In order to achieve
their goal, it vvas considered essential for these ‘non-native speakers’ to
approximate as closely as possible to the native Standard, particularly vvith
regard to pronunciation and, in the 20th Century, very often vvith respect
to a single prestige accent, Received Pronunciation (RP).
Hovvever, in recent years, vvorldvvide political and commercial
developments, in particular the massive increase in American economic
domination, have begun to present a serious challenge to the traditional
roles of English and consequently to the goals of its pedagogy. This is not
the place for a detailed account of events and, for such, the reader is
referred to the works of Phillipson (1992), Pennycook (1994), and Crystal

(1997). For our present purposes, it will suffice to outline two main phases
of change.
First, the dismantling of the British empire in ESL countries, particularly
in the Indian subcontinent, was accompanied not by the abandoning of the
English language in favour of the L1 (or L ls), but by the more active
promotion of local, and thus Ll-influenced, varieties of English. Inevitably,
tensions have arisen within such countries between the desire, on the one
hand, to accept a range of local English accents from speakers of different
L ls such as Punjabi and Bengali, and, on the other, to retain intelligibility
among them. Talking specifically about the problem in India, Bansal
comments that while ‘the view that the only suitable model is British RP is
not shared by ali educated people in the country ... it is felt that some
reasonable Standard should be insisted upon to ensure effective
communication’. Likewise, on an international scale he considers that ‘it is
desirable to establish certain minimum standards of mutual intelligibility
among the various dialects of English spoken in the world’ (1990: 229-30).
Second, and stili more recently, these same tensions have begun to
emerge in the context of EFL. Here, English has rapidly metamorphosed in
the past half century from a foreign language into an international one. its
most frequent use outside the L1 countries (the ‘English as a Native
Language’, or ENL, countries, like the UK, the USA, Canada, and
Australasia) and the ESL countries is between speakers neither of whom
learnt it as an L l. Thus, it differs crucially from other foreign languages
such as Spanish, Russian, Japanese, and so on, which continue to be learnt
predominantly for communication vvith their L l speakers, usually in the L l
country.
To complicate matters even further, this situation is also becoming more
common among speakers of ESL, several of whose countries (for example,
Malaysia, the Philippines, India, Pakistan) are in the process of switching
from a purely intra-national use of English to an inter-national one,
essentially to facilitate trade vvith countries vvhere English is not spoken as
an L l. Pakistan, for instance, now trades extensively vvith a number of
countries in the Far East, especially Japan, using English as a lingua franca.
Thus, although the ESL countries are not the Central concern of this book,
they play an important part in the development of EIL in terms of both their
own internationalization and their early role in the bid for the recognition
of L2 varieties of English.
Changing ownership; changing terminology

In recognition of the changes outlined above, a number of more forvvard-
thinking and less parochial applied linguists have begun to cali into
question the whole issue of who owns the English language and,
The background: Changing patterns in the use o f English 7
consequently, who is entitled to prescribe the standards against which use

is to be measured. No one denies the ‘rights’ of so-called ‘native speakers’
to establish their own standards for use in interaction with other ‘native
speakers’ (ENL), and even with ‘non-native speakers’ (EFL). However, the
important question is: who should make such decisions for communication
wholly between ‘non-native speakers’, i.e. for English as an International
Language?
The view that ‘native speakers’ do not own EIL has been held for a
number of years. Smith, for example, asserted as long ago as 1976, in
relation to ESL, that ‘English belongs to the world ... . It is yours (no matter
who you are) as much as it is mine (no matter who I am)’ (1983: 2). Brumfit
more recently argues that:
Not only has ‘English’ become international in the last half century, but
scholarship about English has also become international: the ownership
of an interest in English has become international. We are no longer a
language community which is associated with a national community or
even with a family of nations such as the Commonwealth aspired to be.
We are an international community ...
(1995: 16)
This position is expressed most emphatically by Widdowson who, having
explained precisely why English is now an international language (‘it serves
a whole range of different communities and their institutional purposes,
and these transcend traditional communal and cultural boundaries’),
continues by arguing forcefully against native speaker control:
How English develops in the world is no business whatever of native
speakers in England, the United States, or anywhere else. They have no
say in the matter, no right to intervene or pass judgement. They are
irrelevant. The very fact that English is an international language means
that no nation can have custody över it. To grant such custody of the
language is necessarily to arrest its development and so undermine its
international status. It is a matter of considerable pride and satisfaction
for native speakers of English that their language is an international
means of communication. But the point is that it is only international to
the extent that it is not their language. It is not a possession which they
lease out to others, while stili retaining the freehold. Other people
actually own it.
(1994a: 385)
Views of this sort are slowly gaining support among scholars. Within the
EFL community, it is largely ‘native-speaker’ linguists themselves who have
begun to ‘examine critically the construct of the native speaker’ and to
question the latter’s identity, authority, and the appropriacy of its norms
(Kramsch 1998: 19). ‘Non-native speakers’ appear, on the other hand, to be
rather more reticent. (It is noticeable, though, that the same is not true of
the ESL countries, which have a much longer history of English use coupled
vvith linguistic imperialism, and whose members speak out more
vociferously against control by the L l ‘centre’). Many ‘non-native
speakers’, particularly among the teaching fraternity, hold the view that
desirable English is indeed that variety spoken by its natives. This seems to
relate to a lack of confidence for, according to Kramsch, "non-native
teachers and students alike are intimidated by the native-speaker norm and
understandably try to approximate this norm during the course of their
vvork together’ (1993: 9). We ever find expression of the opinion that it is
patronizîng to suggest they should do otherwise and aim to acquire
anything short of ‘native-like’ proficiency:
hı the Expanding Circle1 ... the ideal goal is to imitate the native speaker
of the S ta n d a r d language as closely as possible. Speaking English is
simply not related to cultural identity. Ic is rather an exponent of one’s
academîe and language-learning abilities. It would, therefore, be far from
a compliment to te li a Spanish person that his or her variety is Spanish
English. It vvould imply that his or her acquisition of the language left
something to be desired.
(Andreasson 1994: 402)
While this view is not by anv means universal (see the social-psychological
discussion in the next section), it is one sometimes held by teachers and
prospective teachers. The situation is complex and sensitive, and Ll
speakers of English vvould do vvell to guard against unvvittingly appearing
to patronize L2 speakers. In this concext, Bisong’s (1995) reaction to
Phillipson’s (1992) claims of British linguistic imperialism in Nigeria İs a
case in point. Bisong finds the latter both patronizing and misguided, and
argues that English is simply one of a number of languages available to
Nigerians, a part of their^rich linguistic repertoire'. They learn it ‘for
pragmatic reasons to do with maximizing their chances of success in a
multilingual and multicultural society’, a type of society whose complex
nature is not easily comprehended by the monolingual miııd.
One of the principal areas on vvhich the ovvnership debate has recently
focused is the anachronistic termınologv in use to describe the users and
uses of English. I speak here particularly of the terms ‘native speaker’, ‘non-
native speaker', and ‘English as a Foreign Language’. Taking each in turn,
I will outline the objections that have been made and suggest more realistic
alternatives.
First, ‘native speaker’. Kramsch argues that ‘The notion of a generic
native speaker has become so diversified that it has lost its meaning’ (1993:
49). White this may not necessarily represeııt the situation for ali languages,
it clearly holds good for English. The term, for example, fails to recognize
that many varieties of English in outer circle countries, such as Singapore,
are spoken not only as offlcial languages but also in the home (see Brown
1991). Again, it ignores the fact that English is often one of several
languages available in the repertoires of the multilingual populations of, for
example, India and African countries (see the discussion of Bisong 1995
above). In such contexts, it is often difficult to ascertain which language is
a person’s L l and vvhich their L2. The term perpetuates the idea that
monolingualism is the norm when, in fact, precisely the opposite is true for
the vvorld at large. And, as Rampton (1990) points out, it implies the ethnic
Anglo speaker as a reference point against vvhich ali other Englishes should
be measured, vvhich cannot be acceptable or appropriate for a language that
has passed into vvorld ovvnership.
Similarly, it is entirely inappropriate, indeed offensive, to label as ‘non-
native speakers’ those who have learnt English as a second or .foreign
language and achieved bilingual status as fluent, proficient users. The
perpetuation of the native/non-native, dichotomy causes negative
perceptions and self-perceptions of ‘non-native’ teachers and a lack of
confidence in and of ‘non-native’ theory builders. It leads to ‘non-natives’
being refused places on EFL teacher training courses,2limited publication of
their articles in prestigious international journals (cf. Block 1996), a
simplistic vievv of what constitutes an error, and deficiencies in English
language testing programmes, because speakers are being measured against
an unrealistic and irrelevant Standard.
Rampton (1990) discusses a number of possible alternatives. Hovvever, a
problem vvith these suggestions arises vvith the negative connotations of
their opposites. For example, his ‘expert speaker’ for a fluent speaker
implies the use of ‘non-expert speaker’ for a less fluent one, thus imposing
just as much of a value judgement as does ‘non-native speaker’. My own
suggestion is to reconceptualize the issue, turn the traditional terminology
on its head, and replace the two terms as follovvs.
For those L l speakers of English vvho speak no other language fluently,
I suggest substituting the term ‘native speaker’ vvith ‘monolingual English
speaker’ (MES). L l standards would remain operable here but not be used
to measure L2 varieties of English. This vvould mean, to take a controversial
example, that just because MESs generally use weak forms (thus
pronouncing the vovvel sound in vvords like to, from , and o f as schwa h /),
an L2 speaker would not automatically be expected to do so. Since these
pronunciation features do not appear to contribute to international
intelligibility (see Chapter 6), the speaker vvould not be criticized or
penalized for not doing so.
On the other hand, for both those ‘native speakers’ vvho speak another
language fluently and for ‘non-native speakers’ who speak English fluently,
I suggest ‘bilingual English speaker’ (BES). While these are likely to be
shortened to MS and BS, the intermediate ‘E’, strictly speaking, is necessary.
This is because it distinguishes English from other modern languages, vvhich
are genuinely foreign in that they are learnt and used predominantly for
communication with their L1 speakers, and which, therefore, do indeed
have non-native speakers. Further, it is crucial for the third term, ‘non-
bilingual English speaker’ (NBES), to allow for the fact that a speaker may
be bilingual, but not in English. This term bears none of the negative
ımplications of ‘non-native’ but instead provides a neutral, factual
description. It tacitly acknowledges that many L2 speakers of English may
have no desire to speak it fluently, let alone like a ‘native’, and that their
English may have progressed to the level at which it serves their particular
international communicative purpose.
This categorization is not without its problems. In particular, where
should the line be drawn between BES and NBES and, more importantly,
who should draw it? However, it has two great advantages. Firstly, the term
MES is less favourable than BES, reflecting the fact that monolingualism is
not the world norm nor the preferable condition. Secondly, BES removes the
artificial distinction (in an international context) between speakers of L1
varieties of English and proficient speakers of L2 varieties. In this
connection, there is an urgent need to redefine the term ‘bilingual’ for
English, such that it no longer implies ‘equally competent in two languages’,
an interpretation which itself leads to diffidence on the part of proficient
BESs to describe themselves as bilingual. Rather, the term should mean that
the speaker has attained a specified degree of proficiency in both languages,
although in practice probably going well beyond this level in one of them.
Such a removal of the native/non-native distinction is likely to have a
positive influence on teaching, teacher education, and theory building.
Throughout this book, I will continue to use the terms ‘native speaker’
(or ‘NS’) and :non-native speaker’ (or ‘NNS’) (enclosed by inverted commas)
when discussing my sources. But in an endeavour to practise what I preach,
othenvise, wherever feasible—and bearing in mind that they present new
concepts for the reader—I^vill use the more appropriate terms ‘BES/MES’
and ‘NBES’ when discussing my own work. On those occasions where it is
necessary to distinguish between those bilinguals for whom English is an L1
and those for whom it is an L 2,1 will use the (I hope) more neutral terms
‘L1 speaker’ and ‘L2 speaker’.
Moving now to ‘EFL’, the term does not express the principal purpose of
learning English today, and unless it is genuinely used to describe
native/non-native interaction as the pedagogical aim, the word ‘foreign’ has
a number of negative implications. According to Gika, ‘we teach this
language to help people communicate easily, talk to each other without
linguistic and even cultural barriers, understand each other better ... to
bring people closer’, and the term ‘foreign’ is unhelpful in this context. She
asks, ‘how foreign can it be internationally, since people ali över the world
communicate in English?’ (1996: 15).
One possible alternative to EFL simply involves reversing the second and
third letters to arrive at ELF, or ‘English as a Lingua Franca’. This term

would have a number of immediate advantages: ELF emphasizes the role of
English in communication benveen speakers from different L ls, i.e. the
primary reason for learning English Today; it suggcsts the idea of
communiry as opposed to alienness; it emphasizes that people have
something in common rather than their differences; it implies that ‘mixing’
languages is acceptable (which was, in fact, what the original lingua francas
did) and thus that there is nothing inherently wrong in retaining certain
characteristics of the L l, such as accent; finally, the Latin name symbolically
removes the ownership of English from the Anglos both to no one and, in
effect, to everyone. These outcomes are ali highly appropriate for a
language that performs an international function. However, it remains to be
seen whether ELF ultimately catches on. In the meantime, I will for present
purposes restrictits use to describing the core of pronunciation features that
I identify in Chapter 6 as a model for international English phonology—the
‘Lingua Franca Core’, and will continue to use the more widely-
acknowledged EIL.
English, then, is an international language owned by ali who use it and
in this connection the terminology issue is one of great import. It is not
possible to label someone as a ‘foreigner’ or a ‘non-native’ and believe that
he or she has equal rights to the language. These words both reflect an
underlying attitude and, in turn, feed into it and influence it stili further. As
Alexander argues, ‘Such terms now simply block understanding rather than
allowing English teachers wherever they may originate and wherever they
may work to acknovvledge that English as an International Language (EIL)
is here to stay’ (1996: 35).
Appropriate pedagogy for an international language

In view of these worldwide transitions in the function, contexts of use, and
ownership of English, it was inevitable that people would ultimately begin
to question traditional EFL pronunciation goals. The two main issues at
stake vvere, and stili are, first, the extent to which it is relevant to try to instil
L l pronunciation norms into learners who are rarely likely to communicate
vvith an L l (especially an RP) speaker of English; and second, hovv to
promote international intelligibility in the face of the vast expansion in the
numbers of EFL varieties and their speakers. In other vvords, hovv do vve
identify for pedagogy the ‘minimum standards of mutual intelligibility’, to
vvhich Bansal (1990) refers, vvithout recourse to an L l model?
It vvas, as described above, within the context of ESL that these issues
vvere first raised, vvith vvriters such as Smith (for example, 1983) and Bansal
(for example, 1990) speaking out against the insistence on the teaching of
RP but remaining, at the same time, anxious about the diversification of
English accents vvithin the dialects of the ESL countries. While concerns for
mutual intelligibility have not by any means dissipated, the battle has
largely been won in the ELT literatüre in terms of the acceptance by the
English L l countries of the inappropriacy of UK, USA, and Antipodean
pronunciation norms for the ESL countries. Hovvever, this acceptance
derives in part from a failure to appreciate the international uses of English
for many members of the latter countries: in other words, the myth persists
that their English performs only country-internal functions and that there is
stili a distinct EFL-ESL divide. Moreover, it does not extend to an
acceptance of such L2 pronunciation (or other linguistic) norms for those
ESL speakers vvho have emigrated to the L l countries.
Moving on to the EFL context, we find at present among the teaching
profession little, if any, vvillingness to accept any aspect of L2 phonological
variation as a desirable end-product of teaching. The most vvidely accepted
view among teachers, if not necessarily of their learners or some applied
linguists, is that of Quirk:
... vvhile ‘Japanese English’, ‘German English’, ‘Russian English’ may be
facts of performance linguistics, there is no reason for setting them up as
facts of institutional linguistics or as models for the learners in the
countries (1982: 27).
I shall discuss Quirk’s view in relation to the ESL varieties in the follovving
chapter.
In a much-cited paper, Kachru discusses the sociolinguistic spread of the
vvorld’s users of English in terms of three concentric circles (see also
Phillipson’s (1992) division into the ‘core’ and ‘periphery’). Kachru’s ‘inner
circle’ is made up of the ENL countries (see above) and is ‘norm-providıng’;
his ‘outer circle’ consists of the ESL countries and is ‘norm-developing’; and
his ‘expanding circle’ comprises the EFL countries and is ‘norm-dependent’
in the sense that the eriteria by vvhich usage is judged are imported from the
ENL countries, primarily $ıe UK and USA. This latter circle is ‘currently
expanding rapidly and has resulted in numerous performance (or EFL)
varieties of English’ (1985: 16-17). Their speakers are thus the producers
of Quirk’s ‘performance linguistics’ (see the previous paragraph).
In describing Kachru’s model, Crystal argues that ‘the question of
vvhether autonomous norms can develop in a foreign language situation
(such as Japan) remains unresolved’ (1995: 364). This represents a
significant, if not overly dramatic, theoretical shift from Quirk’s and
Kachru’s positions, vvhich seem to discount any suggestion of making L2
varieties the goal of teaching in the expanding (as opposed to the inner)
circle, to one vvhich tentatively acknowledges the possibility. Indeed, in the
field of cultural studies this sort of theoretical shift in attitude towards
foreign language norms is well advanced, and is even beginning to fîlter
through to the EFL classroom. The vvritings of Byram (for example, 1989);
Prodromou (1988); and Kramsch (for example, 1993, 1998), among
others, are influencing a change in pedagogical attitudes, such that learners

are no longer automatically expected to acquire the English L2 cultural
baggage along with linguistic and communicative competence, but are often
‘permitted’ to apply the language to their own cultural norms. And in the
most forward-thinking classrooms learners are being encouraged to develop
what Kramsch describes as ‘intercultural competence’. This is achieved, for
example, through contrastive work, exposure to a range of cultures, and the
use of literatüre and drama. Such learners develop a receptive awareness of
the fact of difference in cultural norms across Ll/cultural groupings, while
at the same time gaining insight into the nature of the norms of their own
Ll/culture. While there is no suggestion that learners should normally apply
the cultural baggage of other groups to their own English production, this
awareness should not only increase their tolerance of difference but should
also enable them to accommodate mutually, in small but significant ways,
towards members of other groups, as they engage in EIL interaction.
Against ali this, the focus on what the ‘natives’ do, which is so
predominant a feature of EFL pronunciation teaching,3 appears backward-
looking and inappropriate for an international language, not least because
research in social psychology has demonstrated clearly the links between
accent and personal and group identity (for example, Edwards 1982, and
see below). However, popular attitudes to accent are, as the social-
psychological literatüre telis us, firmly entrenched and very slow to alter. In
terms of the English language, this phenomenon has in the past been most
evident in the split between attitudes towards RP and GA and those towards
other accents, whether L1 regional or L2 foreign accents. As regards British
L1 accents, a recent survey by the Institute of Personnel and Development
reveals that job recruiters stili discriminate against people with ‘strong
regional accents— especially those from Liverpool, Glasgow, or
Birmingham’ or with strong ‘working class accents’, and a recruitment
consultant advises such people to ‘upgrade’ (The Guardian, 2 January
1997).4 Such accent-based discrimination occurs in other ENL (English as a
Native Language) countries, most notably the USA, where speakers of Black
English varieties are particularly negatively perceived as compared with
speakers of Standard English or ‘white’ varieties, such as New Yorkese (see,
for example, Eisenstein and Verdi 1983; Lippi-Green 1997).
The British situation has undoubtedly improved regarding much L1
regional variation. As a leader in The Guardian pointed out the day after
reporting the above survey, ‘Fifty years ago the employment as a BBC news
reader of a man with a Yorkshire accent roused middle England to fury.
Today Yorkshire, unlike Scouse or Brummie, is rarely a problem’. Although
intelligibility is often cited as the reason for negative attitudes towards
certain accents today, such as Liverpool, Birmingham, and Glasgow, it is
not always clear which is the chicken and which the egg. Admittedly, some
L1 accents are more difficult to understand relative to others, partly from
lack of exposure, and partly from the extent of their own deviations from
the shared common core of L l phonological features crucial to
intelligibility (see Chapter 4)—note the need for the regional Scottish
dialogue in the film Trainspotting to be subtitled for an American audience
in 1996. However, as Wolff (1959) demonstrated long ago, intelligibility is
not necessarily reciprocal and may be the result rather than the cause of
negative social-psychological attitudes vvhich have, themselves, reduced the
receiver’s motivation to make an effort to understand.
If popular attitudes tovvards L l accents of English have been so slow to
change, it is not surprising, then, that L2 accents are stili regarded
negatively even by the majority of EFL teachers (though British teachers no
longer support en masse the exclusive use of RP in the classroom). They are
seen as unfortunate facts of deviant L2 performance rather than as
acceptable examples of L2 ‘regional’ variation vvith the potential to serve—
at least in some of their features— as teaching models (see Brown 1991: 56).
Meanwhile, those who write pronunciation materials for the EFL market
have a vested interest in preserving the phonological status quo, vvith its
emphasis on the need for learners to acquire ali the features of ‘native
speaker’ (RP or General American) pronunciation, including connected
speech, rhythm, and intonation patterns, regardless of their (ir)relevance to
EIL intelligibility (see Chapter 4 for a discussion of this resistance to change
and of the need to restore the balance betvveen segmental and
suprasegmental phonology for EIL).
Hovvever, in recent years, one of the potentially most significant and
interesting phonological developments has been the increasing antipathy of
a small but grovving number of British phonologists, sociolinguists, and EFL
teachers tovvards the perpetuation of RP as the teaching model for L2
learners of English and the point of reference against vvhich their
pronunciation should be judged. RP, the minority prestige accent (or, more
accurately, group of accen*) vvith its origins in the public school system and
a social elite from London and the Home Counties (though novvadays not
regional in use), has provoked criticism for a number of reasons quite apart
from the irrelevance and, for some, the embarrassment of its social origins
or its stigma in certain intranational contexts. For example, Macaulay
(1988: 115) points out that RP is spoken by only a tiny minority of English
users and argues that it vvould be better if present-day linguists and
phoneticians ‘overcame their fascination vvith the accent of an elite minority
and concerned themselves more vvith the speech of the majority of the
population’. According to Crystal (1995), less than three per cent of the
British population speak RP in its pure form, vvith many educated people
having developed an accent knovvn as ‘modified RP’— a combination of RP
and regional features. Indeed, Daniels (1995) refers to RP speakers as the
‘phantom speakers of English’ because of the unlikelihood of a learner’s
coming into contact vvith one of them. If regional accents have now become
the L l rule and RP the exception, there seems to be little reason to base the
teaching of L2 English on an RP model, other than the fact that ‘even in the
inner circle only a specific elite group is considered as “norm makers”, or
as models for emulation’ (Kachru 1985: 17).
A second argument against the use of RP in teaching is that it is by no
means the easiest accent for an L2 learner to acquire, either productively or
receptively, as compared with certain regional accents, such as Scottish
English (see page 17). It lacks any close relationship with English
orthography, contains a relatively large number of diphthongs, and elides
the /r/ sound after vowels. Weak forms, a feature of both RP and General
American (GA) but (contrary to popular opinion) not common to ali native
varieties of English, present both productive and receptive problems for
learners, as we will see in Chapter 6.
A third argument concerns the fact that RP has altered över time. With
clear distinctions between the speech of older and younger speakers of RP
there is a risk of equipping learners vvith old-fashioned pronunciation. A
further problem relating to this last point is that the most recent changes to
RP are not being incorporated into teaching materials. Examples include:
the loss of the diphthong /üs/ as in the word ‘poor’, which is now more
likely to share the vowel sound /o:/ of the word ‘core’; the lengthening of
the previously clipped final vowel sound N in words ending in ‘y’, such as
‘happy’, ‘very’; the phonetic (but again not phonemic) change to the
diphthong h u / when followed by /M in words like ‘cold’ and ‘goal’ such that
the sound becomes [ou].5 And this is apart from the more extreme changes
in the direction of ‘Estuary English’ such as IV vocalization, where the N
following a vowel in words like ‘beautiful’ and ‘milk’ is pronounced /u/ (see
Brown 1991: 93-5). Interestingly,.Crystal (1996) considers that, regardless
of the wishes of the standard-setting elite, it is possible that the English
language, even ENL, is beginning to be influenced by international
pronunciations. Should this be so and should the process continue, the
implications for the whole English pronunciation system are immense.
Returning to the field of social psychology mentioned above, we find the
attitudes of some teachers and their students unfavourable to the use of RP.
A growing number of ‘native’ EFL teachers vvho speak other varieties of
English are no longer prepared to accept that they should either ‘upgrade’
to an RP accent for teaching purposes or use their own regional accent, but
explain to their students how items should be pronounced ‘properly’, thus
implying that their own speech is in some way inferior. The situation in this
respect as regards ‘non-native’ teachers is rather more complex and,
because it links up vvith so many other issues that I vvill be considering later
in the book, I vvill not discuss it further here except to make tvvo points.
Firstly, these teachers know from the personal experience of learning
English as an L2 how important a role pronunciation plays in both
productive and receptive intelligibility, and therefore tend to focus on it far
more in the classroom than do ‘native’ teachers. Secondly, they tend, for
various reasons again relating to personal experience, to insist on a rather
higher degree of ‘correctness’ than do ‘native’ teachers. Both these points
will be taken up again later.
Moving on to the learners themselves, we cannot escape the dose iinks
between language and ıdentity. According to Wıddowson, the reason why
any attempt to reduce English to ‘an entirely neutral medium for the
conveyance of information’ is doomed to failure is precisely because ‘as
soon as the human factor intrudes, the language grows, changes, varies,
becomes subject to the identifyıng need of speakers to express their own
idennty’ (1982: 11—12). Accent is particularly cioseiv bound up with both
personal and group idenrity, the group in this case being those sharing the
learner’s mother tongue. A survey by Porter and Garvin (1989) reveals that
while some L2 speakers genuinely do desire to acquire an RP accent, many
more, when asked, admit to preferring to retain something of their L1
accent. For the Japanese tertiary level learners of English in Benson’s (1991)
studv, English with a Japanese accent is the second most preferred goal, and
in both Benson’s study and that of Starks and Paltridge (1994), an
American English accent is preferred as a learnİng goal rather than a British
(i.e. RP) one.
Learners who in ali other respects achieve a very high degree of
proficiency in English frequendv retain a number of L1 phonological
features. Although motor control is likely to be an element in this process
(see Chapter 2), identity is probably the more salient issue. As Dalton and
Seidlhofer argue, ‘Pronunciation is so much a matter of self-image that
students may prefer to keep their accent deliberately, in order to retain their
self-respect or to gain the approval of their peers’. Therefore, insisting on
learners conforming to carget-language pronunciation norms and
renounciûg those of their mother tongue Lmay even be seen as forcing them
to reject their own identit> (1994a: 7). Daniels (1995) poetically suggescs
that in retainıng ‘the sounds, the rhythms and the intonation of our mother
tongue’, we avoid cutting ‘the umbilical cord which ties us to our mother’.
It should be added at this point that the context of learning, in terms of
whether the leamer is in a monolingual classroom (usually in their own L1
country) or in a multilingual classroom (more often in an L1 English
country), also plays a part here. This issue will be explored more fully in
later chapters.
The EIL phonological problem: where do we go next?

Thus far, we have established a number of things. English is no w learnt and
spoken most frequently to serve international functions among L2 speakers
in international contexts. its L1 speakers have therefore forfeited the right
to dictate standards of pronunciation for L2 use. Moreover, the main L1
accent on which EFL (as opposed to ESL) pedagogy has hitherto been
centred, i.e. RP, is not even widely used among L l speakers and is therefore
unlikely to be appropriate as the basis for L2 pedagogy. And there are, in
any case, sound social-psychological reasons for not pushing learners of
English to attempt to approximate an L l accent too closely, but any
alternative must, above ali, be capable of promoting mutual intelligibility.
Clearly, then, the optimum situation will be one in which EIL can both
express the identities of its (L2) speakers and be a successful means for
communication, much as any language reconciles, for the greater part
among its L l speakers, the opposing forces of the desire to establish ‘wider
netvvorks of communication’ and to preserve the speaker’s ‘particular -
identity’ (Widdowson 1982: 9).6
Given that RP appears not to satisfy the various criteria that vvould
render it an appropriate pedagogic model for EIL, we need to consider the
alternatives. One option is the use of a Scottish or GA model since, vvith
fewer diphthongs and closer orthographic links, these are probably easier
for L2 learners to acquire and understand. They also lack the negative
connotations of RP (Brovvn 1991). For these two reasons, Abercrombie
proposed Scottish English as an EFL teaching model as long ago as 1956.
Modiano has more recently advocated replacing British English vvith Mid-
Atlantic English as the pedagogic Standard for Europe, since this is ‘a form
of the language in vvhich decidedly British pronunciations have been
neutralized’ (1996: 207). Another option is for learners to focus on any L l
regional variety that appeals to them (Daniels 1995), perhaps including that
of one of their teachers. A third option is that of ‘cloning’, vvhereby learners
model their pronunciation on that of someone they admire, not necessarily
a ‘native’ speaker, but someone. vvhose accent is easily intelligible. This
could include speakers heard only on recorded teaching materials, such as
the journalist Kate Adie, vvho features in the H eadıvay series (see Jenkins
and Kenvvorthy 1998).
An interesting approach to the problem from an ESL rather than EFL
perspective, and referring to the language in general rather than
pronunciation alone, is that of Bhatia, vvho argues:
In the emerging language learning and teaching contexts of variation in
the use of English across the international boundaries, it is necessary to
recognize nativized norms for intranational functions vvithin specific
speech communities, and then to build a norm for international use on
such models, rather than enforcing or creating a different norm in
addition to that.
(1997: 318)
Bhatia’s proposal is taken up again in Chapter 6 belovv, vvhere the Lingua
Franca Core is discussed.
A further possibiiity involves distinguishing betvveen a ‘model’ and a

‘norm’. According to Dalton and Seidlhofer:
If we treat RP and/or General American as a norm, we connect them
strongly vvith ideas of correctness. The norm is invariable and has to be
imitated independently of any considerations of language use. The aim,
hovvever unrealistic, is 100 percent attainment of the norm, vvhich is
regarded as an end in itself.
On the other hand:
If we treat RP and/or General American as a model, we use them as
points of reference and models for guidance. We decide to approximate
to them more or less according to the demands of a specific situation or
a specific purpose. In other vvords, a model is alvvays connected to
language in use and is therefore variable. Pronunciation models are
pedagogic means to achieve the end of effective communication for
specific learners. Ideas of correctness do not really apply— a
pronunciation is simply more or less appropriate to a specific use of
language.
(1994b: 2.6-2.7)
In this sense, RP (or, indeed, GA) vvould serve in the classroom as the
teaching model, providing points of reference to prevent learners from
diverging too far avvay from each other and thus from a common
pronunciation core safeguarding intelligibility. Learners vvould in turn be
expected to adjust their accents further tovvards either the L l or the L2
according to the context of interaction. This is an interesting possibiiity and
ties in closely vvith the direction I vvill be proposing, particularly in its
associations vvith accommodation theory. Hovvever, the fact remains that
despite being the most frequently and fully described accent of British
English, RP is now consideredfby a large majority to be obsolete and, even
if it vvere not, we are soon likely to reach a situation vvhere there are very
fevv teachers able to provide a first-hand demonstration. Meanvvhile, it is
unlikely in the foreseeable future that the British vvould condone a move
tovvards an American, or even a Scottish, accent for the vvorld’s largest
group of English speakers, the NBESs. On the other hand, if vve opt for one
of the multi-model approaches, the safeguarding of mutual intelligibility
becomes stili more problematic.
The conundrum, then, remains and becomes ever more urgent as the
number of new varieties of EIL continues to grovv rapidly, thus increasing
the amount of interaction taking place vvorldvvide among NBESs— so that
they are now the vvorld’s majör group of English speakers. Any solution,
hovvever, must consider carefully the particular nature of the interactions of
this group, and base its proposals on an examination of vvhat its speakers
actually do, rather than on vvhat people think they do.
interaction between NBESs has a number of characteristics that

distinguish it from interaction involving BESs (of either ‘native speaker’ or
‘non-native speaker’ origin). For this reason, I prefer to make a distinction
between interaction among NBESs and ‘EIL’ (which describes purpose
rather than result) and to describe such interaction as ‘interlanguage Talk’
(ILT). The term was first used by Krashen (1981, 1982), and subsequently
by Long and Porter (1985) and Ellis (1994), to describe the simplified
linguistic code in vvhich acquirers of second languages speak to one another.
My own use of the term refers specifically to the speech of NBESs from
different L ls as they engage in interaction. These are members of the
‘expanding circle’ whose English has either fossilized at some point along
the interlanguage continuum before achieving bilingual status, or is stili in
the process of moving in the latter direcrion, probably by means of
continued language instruction. The term is not intended to refer to the
speech of members of the outer circle, i.e. of those groups who speak
established ‘ESL’ varieties of English. Their language is no more fossilized
than that, for example, of L l speakers of English from Australia or
Scotland (but see Chapter 2, pages 3 0 -2 , for Quirk’s and Selinker’s
completely different approach to ESL variation).
ILT is seen as problematic because of the amount of miscommunication
it generates. Schvvartz, for example, argues that while breakdowns in
communication are not uncommon betvveen L l speakers of English,
‘second language learners are faced vvith an additional burden to
interaction—the imperfect command of the language of communication’.
As a result, their conversations are ‘often characterized by errors and
problems of understanding’ (1980: 138-9). Likevvise, Varonis and Gass
point o ut that ILT differs from both ‘native/native’ and ‘non-native/native’
interaction not only in that ILT interlocutors ‘spend more time negotiating
than the other pairs’, but also because ‘their non-understandings involve
more work in the resolution’ (1985a: 83). Lynch (1996: Chapter 6)
provides a couple of good examples of this sort of thing, both of vvhich
interestingly involve loss of intelligibility through pronunciation error.
Indeed, pronunciation appears to play a very majör role in problematic
ILT. This is not surprising when one considers that the different L2 varieties
of English are characterized by the transfer of L l pronunciation features,
which can vary markedly from one variety to another. On the other hand,
these different varieties have far more in common vvith one another in terms
of grammar, the learning of vvhich tends to follow common developmental
routes to a far greater extent than does the learning of phonology (see Ioup
1984). In terms of production, then, ILT is likely to be characterized by a
substantial amount of differential L l phonological transfer, itself leading to
mutual intelligibility problems. This is borne out by a study I carried out
among a multilingual group of postgraduate students at London Universirv
some years ago (reported in Jenkins 1995). Pronunciation emerged as by far
the greatest factor in unintelligibility, and the difficulty tended to increase

with the gap between interlocutors’ first languages.7 This study was
followed up with several further years of research into ILT, some of which
is reported in the next chapter, which confirmed the original finding of
pronunciation as a—probably the—critical factor in unintelligibility in ILT.
The ILT pronunciation problem is compounded on the reception side by
the NBES’s tendency to process information in a bottom-up rather than a
top-down manner (see Chapter 4 for empirical evidence). While most
people would nowadays support an interactive processing model rather
than one which is purely top-down or bottom-up, it seems that NBESs do
not follow the same pattern. Difficulties which they encounter with top-
down skills, particularly in relation to making use of contextual cues, both
linguistic and extra-linguistic, and linguistic redundancy, force them back to
an over-reliance on bottom-up skills which, in turn, leads them to focus too
firmly on the acoustic signal. As Brown argues:
For complex social and psychological reasons, they are less sure that they
have grasped the topic being spoken of, the opinion being expressed
about it, and the reasons for the speaker wanting to talk about it. They
are less sure of the relevance of their own past experience in helping them
to arrive at an interpretation. On top of ali that, they are less sure of the
forms of the language ... For ali these reasons foreign learners are less
able to bring to bear ‘top-down’ processing in forming an interpretation
and, hence, are more reliant on ‘bottom-up’ processing.
(1 9 9 0 :5 9 -6 0 )
This is perhaps something of an oversimplification in that NBESs at different
stages are likely to apply the two types of processing in different
combinations. Nevertheless, they do as a group appear to place an
inordinate amount of attention on the incoming signal, such that when this
signal is distorted by a ı#on-shared pronunciation error, intelligibility
problems are virtually inevitable. Moreover, this is not restricted to
phonological error, since NBESs have a narrower band of allophonic
tolerance than either MESs or BESs, so that phonetic errors may also be
problematic. This is linked closely to the fact that the majority of NBESs
have not been exposed to a range of other NBES accents and the types of
phonological and phonetic transfer they contain. It is precisely this lack of
exposure which causes them to focus on bottom-up information. For, as
Lynch points out, ‘in unusual situations, such as ... dealing with an
unfamiliar accent... we may be obliged to pay greater attention to bottom-
up information’ (1996: 22).
The ILT intelligibility problem thus involves both production and
reception, and in replacing an L1 teaching model, we need to take both fully
into account. If we can identify precisely which phonological and phonetic
features affect intelligibility for ILT hearers (and these will not necessarily
be the same as for L l English speakers— only ILT data can provide this
crucial information)—we can then devise pedagogic measures to facilitate
the accurate production of these by ILT speakers.8 In other words, we can
establish some sort of pronunciation core of intelligibility such that exists
among L l varieties of English (see Chapter 6) and then set about finding
ways to encourage speakers to adjust their speech in its direction as and
vvhen necessary; and here, accommodation processes also have an essential
role.
Indeed, it may well be that Speech Accommodation Theory (SAT)
potentially has the greatest contribution of ali to make to ILT pronunciation
and, thus, to EIL. According to this theory, vvhich, in more recent years, has
become more broadly based and knovvn as Communication Aecommoda-
tion Theory or CAT (see Giles and Coupland 1991), speakers may adjust
their speech either in the direction of that of their interlocutors
(convergence) or away from that of their interlocutors (divergence). The
former process, that of convergence, is the one that is of particular interest
to us in the context of EIL.
The original motivation found to account for convergence vvas the desire
to be liked. This desire can operate both at an interpersonal level, i.e. the
desire to be liked by one’s interlocutor, and at group level, i.e. the desire for
membership of the community represented by an interlocutor. In the
accommodation literatüre, hovvever, group identity is more commonly
associated vvith the process of divergence: a speaker diverges from the
speech of an interlocutor in order to distance himself from the latter and
thus preserve his ovvn sense of group identity. Subsequently, the vvish to be
understood has come to be considered an equally salient motivation for
convergence as the desire to be liked. Both motivations are Central to EIL.
They are also mutually inseparable owing to the fact that, by definition,
interlocutors are bidding for membership of an international community
rather than of one or other’s L l community. Thus, the issue of group
identity coincides vvith the desire for intelligibility in ILT and together they
alter the accommodation conditions (see Chapter 7).
It seems likely that the vvhole future of the English language is to an
extent bound up vvith the process of convergence. On the one hand,
speakers need to be able to adapt and adjust their speech to make it more
comprehensible— and, no doubt, acceptable—to particular interlocutors in
particular settings. This includes the ability to adjust tovvards a more
Standard form of pronunciation and, indeed, it vvould patronize speakers
vvith both L2 and ethnic minority accents such as Ebonics (see Todd 1997),
to suggest othervvise. On the other hand, receivers, including teachers, need
to develop a greater tolerance (in both senses) of difference, and the ability
to adapt and adjust their expectations according to interlocutor and setting.
Bhatia argues in his proposal that we consider international English ‘a kind
of superstructure rather than an entirely new concept’ (see pages 17-18).
The best way to enable this ‘superstructure’ to be added is ‘by making the
learner aware of cross-cultural varıations in the use of English and by
maximizing his or her ability to negotiate, accommodate and accept
plurality of norms’ (1997: 318).
With rhe rapid increase in the number of different varieties of English in
the world, each vvith its own accent, few people, if any, are guaranteed to
be exposed pedagogically to every single accent that they are likely to
encounter. Flexibility is the re fo re the key. Iıı the rest of this book, I will
explore not only the possibility of establishing a phonological core for ETİ-,
but also how pedagogic procedures can help speakers of English to acquire
such flexibility.
Notes
1 This term, taken from Kachru (1985), İooselv refers to those countries
whose people speak Enghsh as a Foreign Language rather than as an
official and/or second language. See page 12 for full details of Kachru’s
representation of the sociolinguistic spread of the world’s users of
English as three concentric circles, the ‘inner’, ‘outer’ and ‘expanding’
circles.
2 The University of Cambridge Local Examinations Svndicate in 1996
combined their schemes for the training of ‘native’ and ‘non-native’
teachers. However, it remains to be seen how the directors of these
training courses respond.
3 This has also become one of the main purposes of the growing number
of corpora of English, such as the British National Corpus, COBU1LD,
and CANCODE. Their automatic application to the teaching of English,
vvelcomed by ‘inner circle’ teachers (possiblv because of the ‘expert’
status it bestows on them and the natural advantage it accords them över
‘non-native’ teachers) *s a questionable step for EIL (cf. Widdowson
1991; Prodromou 1996 a, b).
4 It should be added that the reverse situation sometimes obtains: an RP
accent can be a distinct disadvantage in fields such as popular music and
certain branches of the media, and may provoke ridicule and hostility in
Australia (cf. Bro\vn 1991: 31).
5 Throughout the book, square brackets are used both for phonetic
symbols and to denote actual speech realizations. Slanting brackets are
ıısed only for abstract phonemic distinctions.
6 Nevertheless, vve must not exaggerate the success of L l communication:
as we saw in the previous section, the L l varieties of English do not enjoy
one hundred per cent mutual intelligibility, and more should not be
demanded of L2 than of L l speakers.
7 Germanic languages (for example, Dutch and Danish) and Romance
languages (for example, Spanish and Italian) can be considered to have
smaller gaps within their groupings than the gaps, for example, between
Dutch and Spanish, or Italian and Japanese.
8 This is not to suggest that the intelligibility problem will be identical for
each hearer regardless of their L2. Clearly, this will depend not only on
the NBES’s stage of interlanguage development, but also on their L l.
Nevertheless, it should be possible to isolate a minimum set of features
vvhose correct production will guarantee phonological/phonetic
intelligibility for ali NBESs, regardless of their L l.
2 The variation problem 1:
Variety’s the very spice of life,

That gives it ali its flavour.
Cowper, The Task Book II, ‘The Time-piece’, üne 606 (1785)
Contrary to the view expressed by Covvper, the terms ‘variability’,

‘variation’, and even ‘variety’ have always held certain negative
connotations. In many walks of life they are, or have been for periods of
their history, associated vvith some degree of unreliability, lovvering of
standards, or failing vvide of an accepted norm. Evidence of such is
provided by a large number of dictionary defînitions. The Shorter O xford
English Dictionary (Onions 1973), for example, cites an 1832 biological
defînition of ‘variability’ as: ‘capability in plants or animals of variation or
deviation from a type’. ‘variable’ is defined as: ‘liable or apt to vary or
change; su sceptible of variation; mutable, chan geable, fluctuating,
uncertain’ and, relating to persons, as: ‘Apt to change from one opinion or
course of action to another; inconstant, fickle, unreliable’ (late Middle
English). Again, ‘variation’ is defined as: ‘difference, divergence, or
discrepancy betvveen tvvo or more things or persons’ (1637), and in biology
as: ‘deviation or divergence in the structure, character, or function of an
organism from those typical of or usual in the species or group’ (1859).
Among the OED definitions of ‘variety’ is that of 1579: ‘tendency to change;
fickleness’. Definitions of the verb ‘to vary’ include the vvords ‘diverge,
divergence, disagree, dissent, inconsistent, and discrepancy’, vvhile the late
Middle English defînition of this verb has a very familiar contemporary
ring: ‘to differ, diverge, or depart, in respect of practice or observance (from
some Standard)’. (Emphasis added to items bearing derogatory associations
vvithin definitions.)
Human nature, on this evidence at least, seems to have a strong tendency
to favour conformity to standards, uniformity, and conservatism, and to
disfavour non-conformity, diversity, and change. Because language is so
closely bound up vvith human identity and attitudes, it is inevitable that we
should commonly encounter among speakers of Standard English(es)
strongly held and deeply entrenched convictions as to the superiority of

language which adheres to established linguistic norms and the inferiority
of language which departs from them. When Jean Aitchison was so bold as
to express, in her BBC Reith Lectures in 1996, the view now accepted by
most linguists that language variation and change are natural processes
rather than symptoms of degeneration and decay, she received many
outraged responses from the general public. The following, quoted in an
article by Aitchison herself in The Times H igher Education Supplement, is
a typical example: ‘Madam, how dare you distort, desecrate and defile the
English language as you did in your recent Reith lecture’ (15 March 1996:
19). The following year, John Honey maintained that ‘Standard English is
not merely one variety among many, but instead derives its value from a set
of qualities which are not shared by other, non-standard dialects’ (1997: 5).
He followed this up with an appearance on the BBC2 programme Oldies in
December 1997, where he announced that ‘we’ now know for certain that
some three or four dialects of English are intrinsically superior in quality to
ali the others. This sort of view is evident in the discrimination against
speakers of English with certain regional accents that is stili rife in the UK
and USA in job interview situations (see pages 13-14). And L2 variation is
the target of stili stronger prej udice, even in the usually progressive
Guardian newspaper. For example, in his review of the BBC2 programme
The Ballad o f Yoko and Joh n , Adam Sweeting claims, with reference to the
(admittedly Japanese-) English of Yoko Ono, that ‘the significance of a rare
interview with Yoko was dissipated by the difficulty of deciphering her
mauling o f the English language’ (Supplement, 7 January 1998: 19;
emphasis added).
The problem, though, is not simply that many speakers of Standard
English favour linguistic conservatism and uniformity. It is compounded by
the fact that although they may experience variation as an intra-speaker
phenomenon in their ow» L l use, it is not something they consciously
recognize, particularly in relation to pronunciation. Yet this variation in
pronunciation, morphology, syntax, and lexis is apparently ‘so acute that
the same individual, playing various social roles, may frequently display
forms of Standard and non-standard Englishes in different contexts’ (Brown
et al. 1994: 154). Patterned variation is ‘a ubiquitous characteristic of the
contemporary language ... rather than merely a marginal phenomenon’,
‘language is inherently variable, and variant choices carry clear social
meanings’ (Milroy 1994: 167). This applies, as Milroy points out, especially
to the speech channel and, more particularly, at the phonological level,
where the standardization process is least successful. However, where
pronunciation is concerned, many L l language teachers regard with
suspicion, and ‘lay’ speakers interpret as criticism, any suggestion that they
employ such variation in their own speech.
On the other hand, while the majority of L l English speakers are not
The variation problem 1: Inter-speaker variation 27
sensitive to their own intra-speaker variations, they are quick to notice the
mter-speaker variations of L2 speakers. These they interpret as monolithic
deviations from the ‘rightful’ L l Standard, and do not appreciate the
significance of the variability. Again, this is particularly true of pronuncia
tion where it reflects the majority view of speakers of any L l accent, not
only of those who speak the Standard prestige variety. In the field of second
language acquisition, where authorities are well aware of the synchronic
variation in L2 speech, similar attitudes surface to some extent in the use of
terms like ‘free variation’ and ‘systematic variation’ (cf. Ellis 1994) to
denote variation in grammatical correctness in the target language. Thus,
grammatical variation in interlanguage (IL), both betvveen L2 and L l
speakers (inter-speaker variation) and vvithin the speech of an individual
speaker (intra-speaker variation), is often viewed by the experts as
something to be explained in order to reduce and, preferably, eradicate it.
Hovvever, tvvo concepts hitherto unrecognized in EFL, but fundamental
to international uses of English, vvill emerge from this and the follovving
chapters. Firstly, that of acceptable L2 inter-speaker variation: vvhile inter-
speaker phonological variation can certainly impede communication in EIL,
particularly in interlanguage talk (ILT), it is no longer appropriate to regard
ali such variation from the L l as automatically deviant. Much of it
comprises acceptable regional variation on a par vvith that vvhich vve find
among L l accents of English (vvhere, incidentally, inter-speaker variation
can also present an obstacle to intelligibility). It should go without saying
that an L2 speaker of English who has attained BES status (and vvhose
production cannot therefore be considered an IL) does not exhibit ‘deviant’
pronunciation, but rather L2 regional variation, along vvith some degree of
variation in standardness of the type exhibited by L l speakers. It is at the
NBES level, vvhere speakers stili retain ILs, that the phonological and
phonetic forms that occur can be judged for ‘correctness’. And, from a
study of the types of inter-speaker variation occurring in ILT and their
effects on intelligibility, vve vvill be able to establish later on (Chapter 6)
vvhich variants are acceptable as regional ones and which are not. Thence
vve will be able to redefine the concept of phonological acceptability for EIL
and consider the implications for pedagogy and assessment.
The second, and possibly stili more fundamental— and surprising—
concept for EIL, is that of beneficial intra-speaker variation. By employing
accommodative processes, speakers may vary their pronunciation in such a
way that their phonological variation constitutes a solution rather than a
problem, as vve vvill see both here in relation to same-Ll and different-Ll
interlocutors, and in the follovving chapter in relation to different task
types. But before considering why and how this is so, vve vvill need to look
more closely at these tvvo types of IL variation, inter-speaker (this chapter)
and intra-speaker (Chapter 3), in order to examine the vvays in vvhich they
operate and to assess their effects on intelligibility.
L2 inter-speaker variation is the type of variation with which most people
are familiar. Essentially, it involves the transfer of features of the particular
L l onto the production (and, of course, reception) of the target language.1
It is most noticeable phonologically, where the resulting IL accent tends to
form a stereotype for the whole L2 group of speakers in the collective mind
of the L l community. Meanwhile, L2 speakers are also aware of their inter-
speaker variation, both in their own pronunciation and— generally to a far
greater extent— ın that of speakers from L ls other than their own. This
leads to frequent comments in multilingual settings about the mutual
unintelligibility of other speakers’ IL pronunciations, particularly when
exposure to the ILs concerned has been limited and/or (for reasons which
will later become clear) the interaction takes place in groups rather than
dyads. Worryingly, if Trudgill’s (1998) assessment of the developing
situation is correct, while homogeneity is on the increase among World
Englishes at the lexical and grammatical levels, the gap is widening at the
phonological level.
Hitherto, L2 inter-speaker phonological variation, or phonological
transfer, has b e e n discussed in terms of v a r i a t i o n from a S ta n d a r d n a ti v e
model, with such variation being considered lack of correctness. This
contrasts largely with the situation relating to L l variation, where regional
pronunciation norms may, and often do, cause attitude judgements of
inappropriacy, rather t h a n grammatical judgements of incorrectness
(though noce, the same cannot b e s a id of L l regional morphology a n d
syntax, where non-standardness is in d e e d regarded as inaccuracy).
As soon as we begin discussing the English language in an international
context, however, it becomes crucial to redefine L2— and possibly also L l—
phonological correctness. This means reconsidering the issue in relation to
the wider context of in^rnational phonological norms, rather than
continuing to focus narrowly, and often obsessively, on the norms of a
minority of speakers, RP or GA, within a minority English-speaking
community (Britain or the USA), although the USA represents a more
substantial minority than does Britain. And in view of the close links
between accent and personal and group identity that were discussed in the
previous chapter, the need for such redefinition of phonological correctness
could be said to amount to a moral obligation.
Medgyes argues that ‘there are as many equal varieties of English as
there are countries where English is spoken as a first or second language—
and a lot more, if dialects and sociolects are also taken into account’ (1992:
340). I doubt, though, that he intends us to infer from this that ‘anything
goes\ We are stili faced with the task of distinguishing between an
interlanguage, whose speaker remains at the level of an NBES (non-bilingual
English speaker: see Chapter 1) and an EIL regional varıetv, whose speaker
has progressed to BES (bilingual English speaker) stams, while— in ali
likelihood—retaining numerous phonological and phonetic features of the

L l, which will distinguish her speech from that of BESs from other L2
‘equal varieties’. This is because, as is widely acknowledged, L2 varieties of
English differ most noticeably in their pronunciation. The crucial challenge
for EIL phonological research, then, is to identify those areas in which
differences among the manifold international varieties of English (including
L l varieties) are benign and those in which they pose a potential threat to
international intelligibility, particularly for those vvho remain NBESs, and
whose interactions thus remain at the level of interlanguage talk. Having
accomplished this, we will be in a far stronger position to make claims on
behalf of L2 regional norms.
We may also be able to identify vvithin the L2 varieties common
phonological features necessary for adoption by L l speakers of English
when they themselves engage in international communication. It is
conceivable that such features will ultimately become formalized within a
new international variety, such as the ‘World Standard Spoken English’
(WSSE), predicted by Crystal (1997: 138), and/or lead to a redefinition of
the concept of Standard English, such that it is able to embrace ‘divergent’
L2 features. (See Ahulu 1997, vvho argues strongly and convincingly in
favour of the latter move, though he is admittedly referring to grammar in
general rather than phonology in particular). These developments are
especially plausible vvhere large numbers of L2 varieties share the same non-
L1 phonological features, such as /0/ and /öl substitutions or avoidance of
weak forms. It is important, if English is to have an international future,
that vve find out vvhat the L2 varieties have in common and then build on
this information in pedagogy (for the expression of similar vievvs, see
Ufomata 1990; Sure 1992; Bhatia 1997).
In fact, it seems that the latter process has already begun. Local L2
English norms are emerging for the first time, or at least being offered
professional support, in regions vvhere the language has a strong
international function as well as an intranational one, such as Kenya, Sierra
Leone, South Africa, and, to some extent, Europe. Experts from these and
other L2 English backgrounds have started putting forward strong
arguments for the right of their local variety of English to be norm-setting
rather than norm-dependent: see Sure (1992) for the claim that Kenyan
English should be the Standard against vvhich Kenyan learners’ English
performance is measured; Berns (1995) on the nativizing process among
European varieties of English and the possibility of a new institutionalized
norm-setting variety, ‘European English’; Conteh-Morgan (1997) on the
nativization process of English in Sierra Leone; Hibbert and Makoni (1997)
on arguments against Standard English as the norm for South African
Englishes. Perhaps situations such as the follovving, described by an
Austrian teacher of English in Vienna, will be hard to fînd ten years or so
from now:
I’ve been here at ... very very long and this has been a tradition that
you’re supposed to approximate the native speaker, and unless you
approximate the native British speaker you are sort of regarded as
inferior. And I think that’s vvhere attitudes need to change. I don’t see
why a good EFL teacher, Austrian English teacher, shouldn’t have a trace
of an accent of his local variety of English. We’re talking about
international English, vve’re talking about English as a global language,
and vve’re stili keeping to this idea that the Austrian teacher—I mean I
vvas brought up in that tradition—you must sound m ore British than the
British. (contributor to open discussion on English as an International
Language, British Council/IATEFL ELT Links Symposium, Vienna,
September 1996; emphasis added)
It can only be hoped that the change of attitude this speaker refers to vvill
not be too long in coming, and that L2 teachers of English vvill no longer
be regarded as ‘inferior’ because they retain something of their ‘local variety
of English’.
ESL varieties and the interlanguage controversy

At this point, vve need to make an important distinction betvveen foreign
language (EFL) and second language (ESL) varieties of English. In Chapter
1, vve considered Kachru’s three-concentric circle model of English users.
Kachru (1985, see page 12) distinguishes betvveen English speakers in the
outer circle, for vvhom English performs essential intranational functions,
and those in the expanding circle, vvho learn the language for use
predominantly outside their ovvn country in communication vvith L l
speakers and (more often) L2 speakers from other language backgrounds.
It is undoubtedly true that in a number of ESL countries such as
Bangiadesh, there are mapy learners of English vvho have not yet reached
their target production, and vvho thus stili have interlanguages. However,
vve should be careful not to confuse the inaccuracies of their production
vvith the more stable regional pronunciation varieties used by fluent
speakers of non-British or American varieties of outer circle English.
Unsurprisingly, more conventional authorities such as Quirk (for example,
1990) and Selinker (for example, 1972, 1992) consider varieties like Indian
English to be fossilized versions of S ta n d a r d British or American English.
For example, in an article vvhich elicited outraged responses from vvriters in
both outer and expanding circles, Quirk attacks vvhat he calls ‘liberation
linguistics’ (roughly, the acceptance of non-standard Englishes as legitimate
varieties) and, according to Kachru (1991), implicitly invokes the concept
of ‘deficit linguistics’ (roughly, the assumption that linguistic non-
standardness equals incorrectness). Regardless of proficiency level, Quirk
then bunches the ESL varieties together vvith EFL varieties, such as Russian
English, French English, and Japanese English, labelling them ali ‘non-
native varieties of English’, claiming that these varieties ‘are inherently

unstable, ranged along a qualitative eline, vvith each speaker seeking to
move to a point vvhere the varietal characteristics reach vanishing point,
and vvhere thus, ironically, each variety is best manifest in those vvho by
commonsense measures speak it vvorst’ (1990: 5-6).
On the other hand, a number of less conservative language experts see
the ESL varieties in a very different light. Riney, for example, quoting
Selinker’s (1972) definition, argues that the term ‘interlanguage’ should be
reserved for ‘developing systems of language learners “based on the
observable output vvhich results from a learner’s attempted production of a
TL [target language] norm’” . He himself advocates the introduetion of ‘the
terms (sic) ‘additional language’ to refer to the more stabilized systems of
competent speakers of second or additional languages’ from countries, such
as Singapore, India, or even Holland, ‘vvhose target language norm is, as
time goes on, less and less likely to be a native speaker one based in the UK
or the USA’ (1997: 238).
Nelson condemns Selinker’s (1972) claim ‘that interlanguage (IL)
competence “in vvhole groups of individuals” may become fossilized,
“resulting in the emergence of a nevv dialect ([as] Indian English), vvhere
fossilized IL competences [sic] may be the normal situation”.’ (1995: 273).
Nelson rightly considers this to be ‘a rather devalued use of the
characterization “normal”’ (ibid.) and goes on implicitly to support Y.
Kachru’s argument that:
The question of vvhy a stable system should be characterized as an IL is
not ansvvered. It is also not clear vvhat the difference is betvveen ‘stable’
and ‘fossilized’: that vvhich is fossilized is surely unehanging and
therefore stable! Additionally, 'if ‘an entirely fossilized IL competence’
refers to a community ... it is difficult to see vvhy it is an IL and vvhy it is
‘fossilized’. Presumably American English developed as an IL among a
large portion of the immigrant population from the non-English-
speaking parts of Europe. Does that mean that American English
represents an ‘entirely fossilized IL’? (1993: 266)
This seems to be pure common sense, though two further points need to be
made in connection vvith ESL and EIL. Firstly, fluent English speaker groups
in the expanding circle (my BESs, or bilingual English speakers) deserve to
be treated similarly to competent speaker groups in the outer circle. Their
English is no less stable than that of the latter; nor does it exhibit, any more
than they do, characteristics common to ILs, such as phonological blends
not occurring in either the L l or the L2, for example, the Korean English
IL affricate /dö/ for the phoneme 16/ (Ahn 1998: 302-3).
Secondly, vvhile many vvould agree that the validity of the ESL varieties
for intranational use is now established beyond doubt, it does not follovv
that these varieties are acceptable (or appropriate) in their unadulterated
form for international use. As was discussed in the previous chapter, the gap
is closing between ESL and EFL varieties (though not in the sense implied by
Quirk above) as speakers from both types of community engage
increasingly in international communication. Just as members of the inner
circle will find it necessary to make phonological adjustments to render
themselves intelligible on the international circuit, so will members from the
outer circle be obliged to modify their speech in certain ways if they wish
to be understood by speakers of varieties of English other than their own.
To that extent, some of the suggestions made in this book will be relevant
to their (international) language needs.
Phonological inter-speaker variation and its effects on

interlanguage talk
The time has come for us to examine in greater detail the ways in which
NBES inter-speaker pronunciation may vary from the L l, and the extent to
which such variation can impede intelligibility when NBESs interact with
one another. Looking first at segmental variation and then at
suprasegmental, we will not distinguish at present between acceptable
variation and unacceptable error; but to avoid excessive circumlocution
and political correctness, ali L2 phonological variation from L l forms will
be described indiscriminately as ‘error’, ‘deviation’, and ‘variant’. However,
the reader is advised to bear in mind that in Chapter 4, the status of these
deviations will be reconsidered according to their effects (or lack of them)
on intelligibility in ILT. Hence some are likely to be designated acceptable
L2 regional variants rather than errors.
inter-speaker segmental variation and its effects

It has been widely argued#or some years now that segmental (i.e. phonemic
and phonetic) variation has a rather less serious effect on the intelligibility
of L2 English speech than suprasegmental (i.e. prosodic, including stress,
rhythm, intonation, and features of connected speech).2 This debate will be
taken up in greater detail in later chapters. For the moment, I will simply
point out that to date no serious comprehensive investigation of the relative
contribution to intelligibility of these two areas has been conducted at ali,
let alone within the context of ILT.
Habit formation plays a majör role in the production of L2 sounds with
muscular habits that have always operated to produce the sounds of the L l
being automatically activated in L2 production. Although NBESs often cope
well with the articulation of L2 sounds when they have sufficient planning
time to focus cognitively on them and thus exert control över them, as soon
as they release this control in order to focus their attention on the content
of their message, the old habits tend automatically to return. This situation
The variation problem 1: inter-speaker variation 33
continues until and unless sufficient practice of L2 sounds leads to the

formation of new habits vvhich, for the majority of NBESs, will not occur
throughout the entire L2 phonological system.
Sounds that are phonetically very different from those in the L l are
initially likely to prove most difficult to produce, since the articulators must
be activated in new ways. On the other hand, vvhere there is any degree of
similarity betvveen L l and L2 sounds, there is a tendency to identify the
tvvo, and thence to categorize the nevv sounds in terms of the old. Selinker
considers that the making of such ‘interlingual identifications’ is ‘a basic, if
not the basic, SLA learning strategy’ (1992: 260; emphasis in original).
While the process of interlingual identification makes for greater ease of
articulation at first, ultimately it holds the threat of fossilization. This may
manifest itself simply in a slight degree of foreign accent, but may have
implications for intelligibility, especially in ILT. Examples vvhere this has
proved to be the case are the Japanese pronunciation of the voiceless glottal
fricative /h/ as the voiceless bilabial fricative [<J>] before /u:/ as in the vvord
‘vvho’: and the Spanish pronunciation of bilabial plosive /b/ as the voiced
bilabial fricative []3] as in ‘table’.
In order to master an L2 sound, a learner has to acquire both
physiological knovvledge regarding its articulation and an understanding of
its place in the phonological system to vvhich it belongs. This involves not
only the acquisition of sounds vvhich do not exist in the L l, but also—more
problematically—the recognition and ability to deal with the often complex
overlap betvveen the L l and L2 systems. This complexity manifests itself in
tvvo majör ways. Firstly, tvvo distinct phonemes (i.e. sounds vvhich
distinguish meaning) in the learner’s L l may be reduced to allophonic status
(i.e. different realizations of the same phoneme) in English. For example,
clear [1] and dark [1] are phonemes in Russian but allophones of the same
/V phoneme in English as in, respectively, ‘lip’ and ‘pil!’. If these tvvo such
allophones are distributed vvrongly in a syllable, the effect, if any, is
generally one of oddness rather than unintelligibility. Secondly, tvvo or more
allophones in the learner’s L l may have full phonemic status in English, for
example, [n], [m], and [q] in Japanese, as in the English vvords ‘sun’, ‘sum’,
and ‘sung’. In this case, incorrect distribution of the sounds is very likely to
result in unintelligibility. The latter situation is particularly troublesome,
since learners may not be avvare of any difference in the articulation of
sounds vvhich are purely allophonic in their L l—much as the average L l
speaker of English is not avvare of articulating the phoneme InJ in three
different vvays depending on its phonetic environment (i.e. as a dental, an
alveolar, and a post-alveolar in the vvords ‘tenth’, ‘find’, and ‘range’
respectively). Learners therefore have great difficulty in distinguishing such
sounds perceptively as phonemes of a second language.
In the previous hypothetical example, although production would also
prove difficult prior to extensive instruction, it vvould almost certainly
precede perception. We cannot assume that because NBESs are able to

produce sound contrasts, they can necessarily discriminate aurally betvveen
them; indeed, the opposite is not infrequently true (for example, the often-
cited Japanese-English discrimination problem with IV and 1x1). Even for L l
learners of sound systems, there is evidence that the auditory level is not
necessarily primary (see page 47), so that ‘acoustic differences cannot be
readily perceived until the corresponding articulatory gestures have been
learnt' (Ladefoged 1967: 167). Further problems commonly anse for NBESs
both where t\vo distinct phonemes of English are articulated phonetically
very close together, for example /p/ and Ibl in initial position (the main
difference being that the former is aspirated before a vowel sound while the
latter is not), and where two or more phonetic realizations of one phoneme
are articulated in very different ways, for example clear [1] and dark [1].
It mav be argued that to perceive or produce these distinctıons to any
degree of precision is unnecessary, since the context will compensate for any
defıciency in the form of the message itself. But a striking feature of the data
which are produced below to support the categories of L2 phonological aııd
phonetic deviation is the fact that in the majority of cases, a clear context—
linguistic, extra-linguistic, or both— was available at the time the deviant
form was produced. In spite of the presence of such a context, these forms
regularly İed to intelligibility problems for those receivers who did not share
the speaker’s L l. Those receivers who spoke the same L l as the latter
generally (but not always) found the deviation intelligible. The role of
context in ILT will be discussed at much greater length in Chapter 4.
We can categorize segmental deviation types conveniently into three
groupings:
1 sound substitution and conflation;
2 consonant deletion (or elision);
3 addition.
Deviations involving substitution (for example, Japanese IV for 1x1 and
vice versa) and conflation (for e x a m p l e , both / 0 / and İsi being pronounced
İsi) tend to be caused by the kinds of phonemic and articulatory difficulties
described above. In the case of consonant deletion a sound may be omitted
altogether, such as the omission of the 1x1 in ‘price’, typically by a Taiwanese
speaker of English. Alternacively, the elided sound may be replaced by a
glottal stop /?/ as in the Chinese-English pronunciation of ‘duck’ as /dA?/.
Addition can be of two types: epenthesis and paragoge. Epenthesis refers to
the intrusive sound placed betvveen two other sounds to facilitate a
‘difficult’ articulation. It passes unnoticed in L l speech, for e x a m p l e , the
epenthetic İti in the word ‘mince’ /mınts/ or the İpi in ‘comfy’ /k A m p fı/. In
L2 speech it is most often a vowel that is inserted. For e x a m p l e , an Arab
speaker of English may insert a vowel betvveen the /p/ and İV in the word
‘place’, thus pronouncing it as /pıleıs/. Paragoge involves the addition of a
sound to the end of a vvord. For e x a m p l e , a Korean speaker of English
—
typically pronounces the word ‘luggage’ as [İA g ıd ji]. Deviations involving

consonant deletion and epenthesis or paragoge are closely connected with
differences in permitted syllable structure between the L1 and English, with
an open CV (consonant-vowel) structure tending to be universally preferred
(see below, pages 37 and 101).
An intelligibility problem may result from a unique deviation source,
that is, a deviant sound substitution/conflation, or deletion, or addition
within a single word. More commonly, though, unintelligibility is the result
of multiple such segmental deviations, either within a single word or among
consecutive words.
Sound substitution and conflation

Where deviations involve sound substitutions within single words, the
outcome often depends on vvhether or not a substitution results in a non-
word. For example, the day after his driving test, a Korean student came
into class with a downcast expression and announced ‘I pailed’. On a
different occasion, the same student asked a Japanese student at the start of
the mid-morning break, ‘Do you want a copy?’ In the former type of
situation, where the error ‘pailed’ is a non-word (if we exclude the
homophone ‘paled’, which these students were unlikely to know), a BES
would immediately recognize that an error had occurred and would
probably be able to make the necessary mental adjustment in order to
interpret the word correctly. NBESs, on the other hand, may, and indeed did
on this occasion, have more of a problem. While they were aware that they
had not understood, they did not have sufficient L2 lexical knowledge to
appreciate the non-word status of ‘pailed’, and therefore to search their
mental lexicons for viable alternatives. In fact, one of the group followed
up ‘I pailed’ with the question: ‘Did you pass your driving test?’.
More problematic, though, is the second example, vvhere both a BES and
NBES receiver could (and did) simply misunderstand the intended message
(in this case, a ‘coffee’ rather than a ‘(photo)copy’ was being offered).
Again, though, the BES (or MES) receiver has the advantage, since she is
more likely to make use of contextual cues in arriving at an interpretation.
In fact, even when the context is manifestly clear, NBESs stili seem more
likely to place their trust in an acoustic signal (in other words, the exact
speech sounds which are transmitted, see Chapter 4).
Even where certain deviations occurring in isolation would not normally
lead to non-understanding, when combined with others within the same
word, the combination frequently causes too much distraction. For
example, in isolation the substitution of /t/ for /0/, or schw a paragoge (the
addition of word-final /a/), rarely seem to cause problems of understanding
for the NBES recipient. However, during a discussion between a Korean and
a Taiwanese student of English on the subject of erime, the Korean student
referred to cartheft as /kar tepsta/. His Taivvanese ınteriocutor was

completely unable to grasp the meaning (vvhich she admitted in a foIIow-up
discussion) and simply twice repeated the speaker’s \vord with the same
erroneous pronunciation. (Note that word stress is also involved: see page
39).
Potentially stili more problematic for intelligibility is a word group
containing a range of substitutions and conflations, which, with each
individual deviation further compounding the problem, may be rendered
totally incomprehensible to both NBES and BES interlocutors. For example,
a Japanese student ended a short presentation on the subject of the
European Union with a humorous, unprepared one-liner that sounded very
much like Don lies a fizz off score . Despite manv years of exposure to
Japanese-Englısh ILs, I did not understand the sentence until it had been
repeated foıır rimes. On the other hand, one Japanese student in the group
understood immediately, vvhile the others needed only a single repetition.
Hovvever, the non-Japanese NBESs (European, Iranian, Korean, Taivvanese)
were completely mystified until the sentence had been ‘cranslated’ for them
into: ‘Don’t rise (i.e. raise) the fees of school (i.e. school fees)’.
This is, of course, an extreme example, with a pronunciation error (or
errors) in ever)' vvord along with lexical and grammarical oddity, These
involve a combination of consonant substitution, reduction in vowe!
quantitv, and phonetic approximation (i.e. allophonic rather than
phonemic deviation): ‘donV vvas pronounced vvith a short nasalized [5] and
deletion of /nt/; the [j] in ‘rise’ vvas substituted vvith [I]; ‘fees’ and ‘school’
lost their vowel length, vvhile the vowel of ‘school’ vvas also opened to [o]
and the final [i] vvas deleted; the vovvel of ‘of’ vvas not vveakened, nor vvas
the consonant voiced; and because so much else vvas in doubt, the
conflation of /ö/ vvith İzi in ‘the’ also contributed to the general lack of
intelligibility. Interestingly, though, the grammar errors caused no
intelligibility problems for^these NBESs: they understood the meaning of
‘Don’t rise the fees of school’ instantly. Such lack of effect of grammar error
on ILT İntelligibility is a recurring feature throughout the ILT data that I
have collected över the years. Possible reasons for this phenomenon will be
explored in Chapter 4.
The previous example contains an instance of phonetic approximation
vvhich was difficult to interpret. In fact, the majority of phonetic
approximations seem to lead to foreign accent alone rather than to
unintelligibility, and can therefore be ignored unless a learner vvishes to
acquire ‘native-Üke’ pronunciation (perhaps because he—or more probably
she—intends to teach English, or is an immigrant to an ENL country and
vvants to assimilate as fully as possible to the second language and culture).
On the other hand, in a smail number of cases, phonetic L1 transfer may
cause unintelligibility, as happened in the previous example. Moreover, such
phonetic devıations as do occur seem to cause more of a problem for NBESs
than for BESs/MESs. This is because the latter have a wider band of phonetic
tolerance and range of awareness of the available options than do the
former, and are therefore better able to categorize different phonetic
realizations as belonging to the same English phoneme. By contrast, NBESs
tend to hear phonetic differences as categoıical, i.e. phonemıc. Although
this phenomenon appears to be less of an issue than phonemic substitution
in the ILT data, several comprehension difficulties were caused, for example,
by approximations of word-final /n/ by Japanese speakers (as above), and
by Spanish-English approximations of /b/ with [J3], such that ‘book’ was
pronounced [J3ok] and ‘table’, [teıfâl].
Syllable simplification: consonant deletion and sound addition

These two processes will be discussed together as they are the two methods
used by NBESs to simplify English syllables. NBES simplification of English
syllable structure is, in fact, potentially far more damaging to ILT
interaction than substitution and conflation. As was pointed out above
(page 35), the chief cause of such simplification is the structure of L1
syllables, with the universal preference for an open-syllable CV pattern.
Speakers of CV languages tend to select one of two ways to simplify the
syllables of a non-CV language: consonant deletion (with or without the
insertion of a glottal stop to replace an elided consonant) or addition (by
epenthesis—the addition of a sound word-initially or between sounds, and
by paragoge—the addition of a vowel word-finally, with schwa paragoge
being found only in ILs and not in first languages). Some NBESs employ a
combination of these two strategies, possibly depending on their degree of
proficiency in English (with deletion tending to be an earlier process and
addition a later one) and the internal structure of the syllable in question.
Hovvever, the majority opt for one or the other, depending on the structure
of L1 syllables.
Fluent speakers of English (whether MES or BES) also simplify syllables
by means of the process of elision. Although this process bears some
similarity to that of NBES consonant deletion, the similarity is generally
superficial: L1 English elision is a highly rule-governed process and while
L2 English consonant deletion is also strictly constrained, it is governed by
the rules of the speaker’s L1 rather than those of English (cf. Brown 1990:
76). Epenthesis and paragoge tend to increase intelligibility by reducing
ambiguity and aiding recoverability of the original form (for further
discussion see below, pages 65-6 and 116-19). This is borne out by every
occurrence of these strategies in the ILT field data, other than the example
of ‘cartheft’ cited above (where epenthesis served to emphasize the deviant
conflation of /f/ with /p/ and thus compounded the intelligibility problem).
On the other hand, even in the presence of strong contextual cues such
as the availability of the problem words in written form and the framework
of a clear, familiar topic, the majority of deletions in the data resulted in

non-understanding for ali different-Ll receivers, even those vvho employ the
strategy of deletion in their own ILs (for example, Mandarin and Taivvanese
speakers). Similar findings result from the study of Suenobo et al. (1992) of
the relative effects of errors of consonant deletion, substitution, and
epenthesis by Japanese speakers of English on ‘native speaker’ listeners.
While consonant deletion was found to cause the highest rate of
misperception, epenthesis İed to the highest rate of perception and, in
context, was actually found to increase intelligibility. Such findings lend
support to Gimson’s frequently cited claim (for example, 1978) that it is
consonants rather than vowels which carry the message at segmental level.
Where deletion occurred in the ILT data in isolation and in vvord-final
position, it seemed to have a slightly less serious effect, particularly if it was
substituted by a glottal stop. This type of deviation occurred frequentlv
with final /n/, /nd/, /nt/, and İm] even before a following vowel sound.
Taivvanese speakers, for example, produced the ‘don’t’ in ‘I don’t agree’ as
[do?], ‘kind of’ as [kaı?Dv], and ‘time’ in ‘long time ago’ as [taı?]. On the
other hand, deletion in word-initial position was far more likely to cause a
comprehension problem, even where it was the only error. This is partly
because deletion is not permissible in this position in L l English syllable
structure, and partly because a glottal stop was rarely, if ever, used in this
environment to signal the loss of a sound. When there was more than one
such error within a word or word group, or when errors of deletion
combined with other categories of error, there were always serious
repercussions for the understanding of receivers from different L l
backgrounds.
For example, after being shown a film of the Peking Opera on video as
part of a classroom presentation, a Japanese student asked the Taivvanese
presenter whether the opera was based on a true story. The latter replied
that it vvas just [fı?Jb], a»d although the Japanese student knevv the vvord
‘fiction’, she vvas unable to interpret it vvhen simplified by medial and final
deletion, and merely echoed it as [fıjo]. Again, in a discussion of students’
favourite television programmes, even though ali the different categories of
programmes vvere vvritten on the vvhiteboard, the category ‘children’s
programmes’ vvas rendered completely unintelligible vvhen pronounced (by
another Taivvanese speaker) as [Jüjeî pDgvvae?]. As in many of the previous
examples, I (the teacher) vvas able to interpret the deviant/erroneous
pronunciation well ahead of those students vvho did not share the speaker’s
L l. This vvas no doubt partly because I vvas better able to make use of the
contextual cues, and partly because I had had more extensive prior
exposure to the ILs of the speakers concerned; it can also be attributed to
the fact that, unlike the situation for NBESs, my process of interpretation
vvas not complicated by the presence of a second IL, i.e. the receiver’s as well
as the speaker’s, falling wide of the target in a different direction.
The view that ‘Most segmental errors, though noticeable, do not

interfere with communication’ (Daniels 1995: 8), is thus something of an
overstatement. As Rost points out, even among fluent speakers (he is, in
fact, discussing L1 speakers), ‘most mishearings can be identified as
occurring at a segmental level’, even though higher level phenomena, such
as schematic effects, may also be involved (1990: 52). And as has been
demonstrated above (and see Chapter 4), segmental transfer errors can
prove highly detrimental to successful communication in English,
particularly in ILT. However, some suprasegmental errors hold a similar
potential, and it is to the latter category that we now turn.
Inter-speaker suprasegmental variation and its effects

In this section, we will consider the types of errors and deviations that
NBESs make in the areas of vvord stress, rhythm, and intonation, these being
the areas suggested by the majority of phonology authorities as having the
greatest implications for intelligibility.
W ord stress in ILT

Having dismissed the role of segmental errors, Daniels (ibid.) goes on to
argue that ‘the first and alas, often neglected priority should be to supply
learners of English vvith 10 general and powerful stress rules, because it is
at the level of vvord stress that the errors most damaging to comprehensibil-
ity occur’. But despite Daniels’ fighting vvords, very little research appears
to have been conducted on NBES vvord stress deviations and their effects on
intelligibility, as compared vvith that in the other segmental and
suprasegmental areas. Nevertheless, pronunciation teaching manuals make
frequent reference to the need for correct vvord stress placement in order to
preserve intelligibility. None that I am avvare of, though, provides ‘10
povverful (vvord) stress rules’. This is probably because many of the rules
have multiple exceptions and/or are far too complex for mental storage by
students and teachers alike. This complexity is particularly true of the rules
of so-called ‘simple’ vvord stress (cf. Roach 1991).
Nevertheless, English vvord stress is highly rule-governed, and learners
have problems in acquiring these rules, particularly vvhere the rules of the
L1 are both different and less complex, and thus less ‘marked’ (cf. the
discussion of Eckman’s 1977 M arkedness D ifferential Hypothesis in
Chapter 5). This is the case for L ls such as Finnish, Polish, and Spanish,
vvhich ali have fixed or relatively fixed vvord stress patterns (cf. Dalton and
Seidlhofer 1994b: 3 8 -9 for rules and examples). There is also a problem
vvith cognates and false friends, vvhere the L1 syllable and stress pattern is
likely to be applied in the L2. For example, many European Portuguese
speakers of English pronounce the vvord ‘television’ vvith five syllables, of
which they stress the final one. This is a clear illustration of the differential
effects of similarity on reading and listening comprehension, since
receptively the similarity will be extremely helpful when the word is first
encountered in a written text, but probably unintelligible when heard for
the first time. At the level of production, the L1-L2 syllable and word stress
difference may lead to serious intelligibility problems for both L1 and L2
receiver.
A further difficulty for NBESs with English word stress arises from the
differing L1 cues with which it is signalled cross-linguistically. English tends
to make rather greater use of vowel duration than do the majority of L ls,
which tend to rely more on pitch change and loudness (although, as Dalton
and Seidlhofer point out (1994b), loudness provides less of a guide for the
receiver, as some sounds are intrinsically louder than others). The English
stress system also involves far more vveakening of unstressed syllables than
most other L ls (except European Portuguese), with many L ls making only
a small distinction here between stressed and unstressed syllables. Thus,
although an NBES may place word stress correctly, it may not be perceived
as such, especially by an L1 receiver, who will be accustomed to and
therefore expect the acoustic cues of length and weakness in addition to
that of pitch change.
The importance of correct word stress placement for L1 English receivers
is borne out by recent research (cf. Kenvvorthy 1987; Brown 1990; Dalton
and Seidlhofer 1994b), which suggests that these speakers from childhood
onwards identify words in the first place through their stress patterns, and
are therefore thrown badly off course in interpreting messages with
misplaced stress. Brown illustrates this point with an example in which her
‘instantly preferred interpretation was one that held the stress pattern that
had been produced’ (by an L2 speaker), i.e. ‘anaemia’ rather than one
which made sense in the context of a discussion of King Lear, i.e. ‘animism’,
pronounced /a'mmızm/ (1^90: 51). It is possible, however, that word stress
errors made in context, and in the absence of other error types, are not
automatically damaging for L1 listeners. Word stress patterns differ quite
markedly among L1 varieties of English, most notably RP and GA, with no
great subsequent loss of intelligibility (though admittedly, familiarity with
these accents is likely to have a role in this). In addition, stress patterns may
change över time with the dictates of fashion, while for a small group of
words such as ‘controversy’ and ‘kilometre’, two distinct patterns are
current and intelligible in British English use. Ali this suggests that L1
speakers are capable of a fair degree of flexibility in this area.
Some ILT w ord stress data

The effects of word stress deviations on NBES listeners are at present even
less clear-cut. As was discussed in the previous section on segmental
deviation, NBESs are far less likely than more competent speakers to bring
contextual cues to bear on their interpretation of difficult pronunciations,
and this no doubt extends to their attempts to interpret faulty word stress
placement. However, in the present data, the majority of deviations that İed
to non- or mis-understanding in ILT occurred at the levels of sounds,
syllable structure, nuclear placement, or various combinations of these.
Only rarely did word stress deviations alone present difficulties, although
they did sometimes compound the effects of other deviations, like the
‘cartheft’ example cited above (page 36). Interestingly, one of the few solely
word stress errors to cause an intelligibility problem involved me, an L l
speaker, as listener, in an exchange which lacked any contextual
information. A Turkish student, â propos of nothing that had preceded in
the conversation, asked me for the opposite of the word ‘mature’, but
pronounced it with the stress on the first syllable such that, assuming an
approximation of the final h u / diphthong, I interpreted the word as
‘macho’. The misunderstanding was only cleared up when the student
wrote the word down. On the other hand, the ILT data contain no examples
of word stress deviations alone leading to unintelligibity, even though many
such forms occurred (for example, ‘resort’ and ‘Korean’ stressed on their
first syllables), and there are even examples of such deviations being
corrected by the NBES receiver (for example, ‘sunshine’ with stress on the
second syllable).
The few word stress deviations that did lead to unintelligibility problems
in the ILT data occurred in tandem with other types of deviation. For
example, a French learner of English asked a Hungarian learner, ‘How do
you say hopeless in French?’, but pronounced ‘hopeless’ with stress on the
second syllable, deleting the vvord-initial /h/, and reducing the diphthong
h u/ to the short vowel /d/. Likevvise, in a conversation about ways of
wasting time, an Italian-English speaker asked a Japanese student, ‘Do you
waste your time alone?’, pronouncing ‘alone’ as /'eİDn/, with stress on the
first syllable and two phonemic errors.
In these examples, it is difficult to assess the relative salience to the
listener of the different types of deviation. I suspect, though, that in the first
case, the misplaced word stress would not have caused a problem had there
not also been segmental errors, since the syllable ‘-less’ is familiar in
isolation as well as in suffix form. On the other hand, in the second case,
where a normally completely reduced syllable is given primary stress, and
also because of the consequent misplacement of nuclear stress (see next
page), the opposite may have obtained. Interestingly, in a third case, which
occurred during a classroom exercise on connectives, the correctly
pronounced word ‘also’ (by a Hindi-English speaker) vvas interpreted by a
Brazilian student as ‘although’. Presumably the reason was that he would
have pronounced ‘although’ in this way as a result of L l stress and sound
transfer.
Generally more serious, however, were word stress deviations vvhich

occurred in combination vvith consonant deletions and vvhich, in spite of
contextual cues, rendered vvords totally unintelligible to ali listeners. For
example, the vvords ‘product’ and ‘expenditure’ vvere pronounced by a
Taivvanese-English speaker in a conversation about advertising as [phdAk]
vvith stress on the second syllable, and [epe? ’dıjo] vvith stress on the third
syllable.
However, the most serious vvord stress deviations of ali are probably
those vvhich also affect nuclear placement. The nucleus is the most
prominent syllable in any group of vvords. It is the one vvhich the speaker
has chosen to highlight (by means of extra length and loudness, and a
change in pitch level) as carrying the most salient part of his or her message,
and thus the part on vvhich he or she vvishes to focus the listener’s attention.
This means that deviations in the placement of the nucleus have the
potential to affect the listener’s ability to process entire chunks of the
speaker’s message.
One such example vvas provided on the previous page, vvhere the stress
on the vvord ‘alone’ vvas vvrongly placed on the first syllable. Because the
second syllable of this vvord should have borne not only vvord stress but
also nuclear stress, intelligibility vvas more problematic than vvould have
been the case if the deviation had been one of vvord stress only. Another
instance occurred vvhen a Japanese-English speaker asked a multilingual
group, ‘Does anyone know vvhere is Caribbean?’, using the USA
pronunciation of ‘Caribbean’, vvith stress on the second syllable. There vvas
no response, because none of the students understood the question. The
problem vvas not that they had not heard of the Caribbean, nor that they
could not understand the incorrect grammar (they habitually made the
same mistakes vvith embedded questions and article omission). The cause
vvas the unfamiliar vvord stress, compounded by its co-occurrence vvith
nuclear stress. Another *example of this phenomenon occurred in an
exchange betvveen an Iranian and a Japanese student. The latter had been
speaking about advertising, and the former asked him, ‘Do you think the
advertisers exaggerate subject about something that is unreal?’, placing the
stress in the final vvord on the syllable-initial ‘-un’. Although the question
contained a number of grammatical errors, the listener understood the
intended meaning once the stress on ‘unreal’ had been corrected. On a
further occasion, in a lesson on idiomatic language, a Korean student
defined the idiom ‘to bite off more than you can chevv’ as ‘to undertake
more than you can fulfil’, but pronouncing the final vvord as /'polpıl/. Thus,
the nuclear syllable vvas not only misplaced, but the misplaced nucleus also
contained a phonemic error. Although the Japanese receivers had this
sentence vvritten down on a vvorksheet, they vvere unable to identify it.
Rhythm and intonation

We now turn to the English rhythm and intonation system and consider first
the concept of stress-timing. The original claim for the difference between
so-called ‘stress-timed’ and ‘syllable-timed’ languages is now thought by
many phonologists to be too strong, especially in relation to non-formal
speech (see pages 149-50). Nevertheless, the English system of rhythm and
intonation, with its alternation of strong and weak syllables and the extra
prominence accorded to nuclear stress (common to ali L1 varieties of
English), operates regardless of the degree of formality of the speech
situation, and acts as a ‘guide to the structure of information in the spoken
message’ (Brown 1990: 43). The overall effect may thus be quite different
from that of languages whose important content words are not highlighted
or less important words reduced to the same extent.
This is borne out by Nelson’s (1982) investigation of the differences
between (‘syllable-timed’) Hindi-English and (‘stress-timed’) American
English. Nelson finds that transfer of the former onto the latter leads to the
placement of stress on a syllable next to the one where it would be expected
by a ‘native speaker’, and a lack of unstressed vowel reduction. He
concludes that rhythm is likely to have a significant effect on the
intelligibility of ‘non-native speaker’ varieties of English. Wenk (1986)
likewise observes the transfer of L1 rhythmic patterns of (French) learners
of English on to their English L2. He suggests that learners only overcome
such L1 influence at an advanced stage of proficiency, a view that is also
expressed by Cruz-Ferreira, who refers to intonation as ‘the last stronghold’
of a foreign accent (1989: 24).
This latter claim is well supported in the present data, where advanced
learners of English make frequent'intonation errors, particularly with the
placement of nuclear stress, whether unmarked (on the last content word of
the word group) or contrastive (somewhere earlier). They tend to place it
indiscriminately at or near the ends of sentences. For example, a Japanese
student talking about the students in her previous class, said ‘There were
Spanish, German, French and I could teli the difference between T H E M ’
(erroneous nuclear syllable in Capital letters). Another Japanese student,
discussing her new class, said ‘They aren’t ali Japanese—we’ve got a lot of
other STUDents’. The speakers in both of these examples had been studying
English in the UK for well över a year.
Moving on to tones, although tone universals undoubtedly exist (see, for
example, Brown and Levinson 1987, pages 104-6), it seems that the use of
tones is also to a fairly large extent language-specific. Van Els and De Bot
(1987) demonstrate this in an experiment where they find that the ability of
listeners to identify the L ls of various ‘non-native speakers’ of English is
significantly reduced when the latters’ speech is monotonized. The
avoidance of L1 tone transfer and the correct use of the English tone system
is considered essential by many phonology experts and materials writers,
and the literatüre abounds with references to the risk learners run of
offending ‘native speakers’ (though not necessarily other L2 speakers), if
they do not adhere to L2 politeness norms in their use of tones.
These writers are generally referring to what is often labelled the
‘attitudinal’ function of intonation. However, the use of tones by L l
speakers of English remains ‘elusive’ '(Bradford 1982: 33-4) in relation to
the expression of attitudes and, to a lesser extent, to the expression of
grammar and discourse meaning,4 and is almost always inseparable from
speaker and context. Apart from the pitch direction in a number of
intonational idioms such as ‘You must be joking!’ (Dalton and Seidlhofer
1994b: 45), it is therefore virtually impossible to provide cast-iron rules.
This is evident even among experienced EFL teachers on in-service courses:
agreement rarely results from tasks vvhere, for example, some participants
are asked to answer a question with the word ‘yes’, expressing (unknown
to the other participants) for example, tentativeness, boredom, enthusiasm,
and the like. There are generally almost as many different guesses of the
attitudes being expressed as there are listeners.
Where tones are said to have a grammatical function (for example, a fall-
rise for a yes/no question and a fail for a wh- question), on the other hand,
there seems to be a higher level of (theoretical) agreement as to how these
should be interpreted and also one or two cross-linguistic universals.
Despite this, it is quite probable that intonation does not have a
grammatical function at ali (see Chapter 6, pages 151-2). Moreover,
regardless of the issue of the function of particular tones, a surprising
number of fluent speakers of English, teachers included, while able to
identify the nuclear syllable itself, are unable to perceive consciously the
direction in which its pitch moves. It is presumably for these reasons, and
others which will be discussed in Chapter 6, that pitch direction was very
rarely found to contribute towards unintelligibility and never to be the sole
cause of it in the ILT datft.
Although intonation universals undoubtedly exist because of the
physiological constraints on the vocal apparatus, much intonation
nevertheless consists of highly stereotyped patterns of which L l speakers
are not consciously aware (Nash 1969; Berkovits 1980). It is because
intonation is both fleeting and operates at a subconscious level that NBESs
are rarely aware of transferring their L l patterns onto their English output.
Meanvvhile, although L l speakers of English cannot articulate and explain
intonation errors, they respond to their effects in interpreting meaning
(Bradford 1988: 2), with the intonational message often taking precedence
över the lexical (Nash 1969). The best-known example of this phenomenon
is probably the one described by Gumperz (1982: 173). Recently hired
Indian and Pakistani cafeteria staff working at a British airport were
perceived as ‘surly and uncooperative’ purely on the basis of their
intonation patterns. For instance, when offering gravy, they vvould say the
word ‘gravy’ with a falling tone instead of the rising tone normally adopted
by L1 speakers of English when making offers of this sort. This was
interpreted by the cargo handlers they served as a statement of fact, and so
redundant in the context, and indicative of indifference rather than the
engagement involved in an offer.
It is the area of nuclear placement that seems to present the greatest
suprasegmental threat to intelligibility in ILT, both in the placing of
unmarked nuclear stress on the final content word in a word group and,
more problematically, in the placing of marked or contrastive stress in an
earlier position. One reason why NBESs fail to place nuclear stress correctly
is very probably that they have problems in dividing the stream of speech
into word groups (or ‘tone units’). It then becomes impossible for them ‘to
single out the most important information within a group’ (Kaltenboeck
1994: 17). By grouping vvords thus, fluent speakers of English indicate to
their listeners vvhich vvords should be interpreted together. This is achieved
by structural means: the vvord group is normally bounded by pauses,
contains (at least) one nuclear syllable5 bearing what the speaker selects as
the most important information, and is composed of an alternation of
prominent and reduced syllables. Word groups also tend to coincide vvith
syntactic boundaries, so that failure to divide the speech stream into these
units can result in grammatical ambiguity or misinterpretation.
The claim for the vvord group as the primary structural component of
English speech is supported by research into the use of formulaic
expressions knovvn variously as ‘prefabricated routines and patterns’
(Hakuta 1974), ‘lexical sentence stems’ (Pavvley and Syder 1983), ‘lexical
phrases’ (Nattinger and DeCarrico 1992), and by a host of other terms.
This research, concisely summarized by Widdowson (1990: 92-6),
demonstrates that a significant proportion of vvhat ‘native speakers’ of
English say is composed of ready-made, memorized chunks of language of
varying degrees of fixedness. Although it is only relatively recently that firm
links have been drawn betvveen the vvord group and the lexical phrase
(Seidlhofer and Dalton 1995), the usefulness of the lexical phrase for the
teaching of intonation has previously been noted (Seidlhofer and Dalton
1993; Dalton and Seidlhofer 1994b; Kaltenboeck 1994).
The failure of NBESs to segment their speech into vvord groups results
not only in problems vvith nuclear placement, but also in a lack of pause
vvhich, for the listener, creates a false sense of speed (Nash 1969; Van Els
and De Bot 1987) and reduces the time available for the processing of
information. Inevitably it poses more of a threat in ILT, vvhere the less
competent listener badly needs this ‘pause’ time for processing, vvhile the
less competent speaker needs it for planning. The absence of vvord-grouping
is likely to result in non-fluent speech, vvith pauses occurring in unnatural
places to facilitate the solving of linguistic problems rather than to serve the
purpose of signalling information structure.
However, it is nuclear placement itself which causes the most serious

suprasegmental problem for NBESs at the productive level. English has one
of the most rigid word orders among the world’s languages. This means
that speakers of English, unlike speakers of many other languages, cannot
rely on moving important words to salient positions such as the beginnings
of phrases and clauses in order to emphasize them. Nor can they add
inflections or particles to the endings of words to indicate their relative
importance in relation to each other. To compensate for these restrictions,
English allows free stress placement within the intonation group to an
extent that other languages do not (Creider 1979). What this means in
practice is that any word, regardless of its syntactic position, can be given
nuclear stress if it is the one which the speaker wishes to make the focus of
her or his message. (Though note that in words of more than one syllable,
it is the stressed syllable rather than the whole word which bears nuclear
stress.)
The unmarked position for the nucleus is on the stressed syllable of the
last content word in the word group. But speakers regularly move the
nucleus to earlier positions to indicate some sort of contrast. For example,
the following is simply a statement of fact: ‘My husband drinks a lot of
b e e r ’ , in which ‘beer’ is the nucleus. However, if I say, ‘My Husband
drinks a lot of beer’, I have established a contrast between my husband and

some third party known to both me and the listener. Although some L ls,
for example, German and Russian, share with English the phenomenon of
moving nuclear stress to signal contrastive information focus, many others
achieve this purely by morphosyntactic means, such as word order, clefting,
or topic markers, and retain a fixed position for the nucleus, with a
tendency to prefer the final noun in the intonation group (Cruttenden 1986:
146-50).
A number of studies investigate the extent to which NBESs transfer L1
intonation patterns ont® their English, and the effects this has on their
ability to use English nuclear stress and, consequently, on the intelligibility
of their speech. We will briefly consider three of these studies, which have
particular relevance to the present discussion, before moving on to look at
some ILT data. First, Wennerstrom (1994) examines the use made of pitch
to signal meaningful contrasts among L1 speakers of Spanish, Japanese,
Thai, and (American) English. She finds that while the L1 English speakers
are consistent in their use of contrasts, ali the L2 speakers tend to give equal
prominence to items regardless of their importance in the information
structure. They neither approach the degree of pitch increase produced by
the L1 subjects on new or contrastive information, nor do they reduce pitch
to the same extent on redundant words. She concludes that on the one
hand, the thread of their speech would therefore be difficult for ‘native
speakers’ to follow, while on the other hand, these ‘non-native speakers’
risk missing important aspects of ‘native speaker’ discourse structure.
In this study, the Thai subjects fare the least well and the Spanish subjects
the best. Wennerstrom accounts for this result by the fact that Thai
intonation functions the most differently of the three from English (Thai
being a tonal language in which pitch is used to signal lexical rather than
discourse meaning), while Spanish intonation has certain similarities to that
of English. She also points to the part played by exposure to L1 intonation:
the Thais had been in the USA for only two weeks, while the Spaniards had
been there for över a month and, prior to their visit, had been exposed to
American English via the media. Wennerstrom therefore recommends that
longitudinal studies be conducted to investigate how ‘non-native’
intonation becomes more ‘native-like’ över time. However, as regards
marked— and particularly contrastive—nuclear placement, it appears that
exposure is more likely to benefit reception than production in the short
term. Indeed, it seems to take well in excess of a year to filter through to the
production of speakers of L ls intonationally distant from English, and only
a little less for those who are closer. On the other hand, receptive
competence in marked nuclear placement seems to be reasonably well
acquired after a relatively short period of exposure (Bradford 1982; and see
Chapter 6 below). This is plausible when one considers that the acquisition
of productive intonation competence is a very lengthy process in L1 English
(Cruttenden 1986: 173-4), whereas receptively it is acquired extremely
quickly—in early babyhood or possibly even in utero. So while it may often
(though not always) be the case that pronunciation reception precedes
production, the gap between the two competences seems to be far greater
for intonation than for other phonological areas, in both L1 and L2
English.
In the second study, Lanham (1990) investigates communication
problems between speakers of South African Black English (SABE), whose
mother tongue is one of the group of Bantu languages, and speakers of
South African English (SAE). Bantu phonological rules, and particularly
those governing intonation, are very different from those of SAE, and
because many speakers of SABE are taught English by L2 speakers, there is
inevitably a high degree of transfer. Lanham contrasts the heavy SAE
exploitation of referring tones on nuclear syllables (Brazil et al. 1980; and
see page 49 below) vvith the scant use of these tones made by SABE speakers,
who reserve them for the ends of questions in spontaneous speech. This
fact, along vvith a difference in the way the SABE speech stream is segmented
into vvord groups (not necessarily coinciding vvith syntactic/semantic
boundaries), the greater frequency and unselective placement of SABE
prominent syllables, and the lack of distinction betvveen SABE stressed and
unstressed syllables, leads Lanham to propose that SABE intonation does
not provide a discourse function for SAE speakers. The result, he argues, is
the inability for speakers of the two groups to negotiate interactive
discourse successfully.
In the oldest of the three studies, Nash (1969) investigates the use of
contrastive stress of Spanish (Puerto-Rican) speakers of English and English
speakers of Spanish at three different levels of L2 proficiency. She finds that
the lowest level Spanish speakers of English use too little emphasis to signal
contrasts, while the equivalent English speakers of Spanish use too much.
The former therefore fail to give the necessary emphasis to make contrastive
distinctions, while the latter apply more intensity than is appropriate in the
context. Nash argues that accentual interference has wide-reaching
repercussions on intelligibility because accent is concerned not only with
the identification of meaning-bearing units, but with the identification of
meanings themselves. To this extent, context provides less help (for the L l
receiver) than it does with segmental phonology, because the intonation
pattern is, itself, part of that context. Nash also emphasizes the cumulative
effect of such accentual interference, in which the hearer— unable to tune in
to the speaker—cannot relate the meaning of one utterance to that of the
next, with the result that the utterances become increasingly incoherent
and, ultimately, either the speaker is judged as unintelligible or the message
is misinterpreted.
Nash makes a number of points that are of interest in the context of ILT.
Firstly, she finds that discrepancies between the most and least proficient
Spanish speakers of English are not as great as those between the least
proficient and those with a slight L l accent. This is particularly true of the
segmentation of the stream of speech into tone units, and seems to result
from the fact that the Spanish speakers of English with a slight L l accent
exaggerate those features of English which they think will make them more
intelligible, while the speakers with the heaviest and lightest accents do not.
Presumably the former are not sufficiently competent in the L2 to attempt
such reduction in the transfer of L l intonation patterns, while the latter feel
it unnecessary to exaggerate, since they are already easily intelligible.
Secondly, Nash argue# that, on the one hand, the requirements of
intelligibility in the second language are reduced if both speakers share the
same first language, while on the other hand, where both speakers come
from different L ls, exposure to one another’s imperfect speech will lead to
the modifying of the perceptual apparatus and subsequently to increased
intelligibility. Both these claims are supported by data produced in this
book.
Nash’s investigation is now around 30 years old. Somewhat surprisingly,
considering the importance of the points she raises, there has been little
progress in research or materials production in the area of nuclear stress in
general and contrastive stress in particular. Instead, most intonation
scholars have been attracted towards the work of Brazil and the study of
discourse intonation (see the next paragraph). In Chapter 4, we will
consider why the further investigation of nuclear stress (the old ‘accentual’
function of intonation) holds far more promise for intelligibility in ILT—
though not necessarily for ‘native speaker’/‘non-native speaker’ inter

action—than does the investigation of discourse intonation.
From the 1970s until his death in 1995, David Brazil, together with his
colleagues, worked on his theory of discourse intonation, developed from
the work of Halliday and the Prague School (see Halliday 1970; Brazil et
al. 1980; Coulthard and Montgomery 1982; and the posthumously
published Brazil 1997). Discourse intonation essentially involves the
placing of nuclear stress on the syllable within the tone group that the
speaker sees as most salient in terms of the wider context of the utterance.
In so doing, the speaker is said to select either a ‘proclaiming’ or a
‘referring’ tone, respectively a falling tone ‘p’ to signal that the information
is in some way new from the listener’s point of view, and a fall-rise ‘r’ to
signal that the information is in some way shared, or ‘given’. For example,
the answer to the question ‘When do you want to travel round the world?’
could be the following:
(r) I want to travel round the world//(p) when I leave school
whereas the answer to the question ‘What do you want to do when you
leave school?’ is more likely to be:
(r) When I leave school//(p) I want to travel round the world.
(Adapted from Dalton and Seidlhofer 1994a: 61).
Because of its relationship with parts of the utterance not in the immediate
linguistic vicinity, a misplaced or mis-pitched nucleus may apparently affect
meaning at the global level of the whole interaction. Such an error is
undoubtedly the most difficult of ali types of phonological error to identify
and locate, because so much more than the immediate utterance is involved.
For this reason, along with my strong conviction that discourse intonation
‘errors’ do not affect ILT intelligibility (see the study in Chapter 4), I do not
present any data in the section that follows to exemplify errors in this area
of intonation.
Some IL T intonation data

The intonation data consist of a number of errors in nuclear placement,
both where the nucleus is automatically placed at the end of an intonation
group (either on the last content word or indiscriminately on the last word
whatever its status), and where an attempt to use contrastive stress misfires.
Some of the errors involve nuclear placement alone, while the majority
incorporate a mixture of segmental and/or word stress deviation and
misplaced nuclear stress. As with the deviations described in previous
sections, non- or misunderstanding was in each case the consequence,
despite the presence of linguistic and/or extra-linguistic contextual
information—though it could be argued that intonation is itself an essential
part of the context (see the discussion of Nash 1969 above).
The first examples occurred during a (multilingual) classroom lesson at

upper-intermediate level on idiomatic language connected vvith the body. In
each case, the misplaced nucleus (indicated by Capital letters) vvas the only
error and resulted in non-understanding and a request for repetition:
1 ‘to put your foot in IT’ (Japanese speaker)
2 ‘to pull your finger o u t ’ (Brazilian-Portuguese speaker)
3 ‘to be up to your neck in IT ’ (Brazilian-Portuguese speaker)
On the other hand, vvhere the final item vvas a noun, resulting—probably
by default— in correct unmarked nuclear placement, there vvere no
problems in understanding, for example, ‘to pull someone’s l e g ’ (the same
Japanese speaker), ‘to have your heart in your m o u t h ’ (the same Brazilian-
Portuguese speaker).
Further errors of this type, ali made by Japanese speakers giving short
classroom presentations, and causing non-understanding among the
multilingual listeners, are:
4 ‘How exactly did you come by this painting?’
5 ‘. . . a dirty tricks camPAiGN and allegations surrounding i t .’
6 ‘Richard Branson vvants to expand his COMpanies, but recently he
sold his record COMpany.’
7 ‘Before reading the books I opposed the jury SYStem, but novv I am in
favour of i t .’
Attempts to use contrastive stress, though encouraging as a sign that the

system is beginning to penetrate, are fraught vvith potential intelligibility
problems. This is illustrated by example 4. A further example occurred
during an informal conversation vvhich, although it took place in a
classroom, was not part of the lesson. A group of students and I vvere
making arrangements for a meal together in a local restaurant. A Spanish
(Catalan) student askej me, ‘Will you b o o k the table?’ (making an
unsuccessful attempt at placing contrastive stress). Until she rephrased it as
‘YOU’ RE going to book the table?’, none of us understood her meaning.
The previous example also involved a phonetic deviation, since the
vvords ‘book’ and ‘table’ vvere pronounced [|3ük] and [teıJ31]. The
combination of misplaced nuclear stress and segmental deviation seems to
be a particularly potent one for ILT unintelligibility. In a different
conversation among four students, Brazilian, Svviss-French, Colombian,
and Hungarian, as they sat together making posters for the classroom wall,
the Hungarian student asked the other three: ‘Have you got a blue VUN?’
The three echoed the vvords ‘blue vun’ and ‘vun’ several times until the
Hungarian student, holding up a blue pen, explained ‘Blue vun like t h i s ’,
vvhereupon the vvhole group collapsed in laughter. The point here is that
these students had been vvorking and socializing together for three months
and had therefore gained considerable exposure to each other’s varieties of
English. In addition, the context of this exchange could have been expected
to provide the listeners with clues to meaning, since they were sitting round
a table with paper and coloured pens. I suspect that the segmental error
alone would have presented them with little problem since it occurred on
an unimportant word. Rather, it was the fact that nuclear stress was placed
on this very word, thus presenting it as carrying the most salient
information in the word group, along with the failure to use nuclear stress
contrastively (the ‘blue’ one as opposed to the ‘red’ one), that combined to
destroy the message.
Pirt provides a similar example where Italian speakers of English
negotiate the position of a yellow pencil as contrasted with pencils of other
colours. The yellow pencil is described by one of the speakers as a ‘yellow
p e n ciP, failing to use nuclear stress contrastively. Pirt points out, however,
that the tendency to place the nucleus towards the end of the unit makes
sense in the speaker’s L l Italian, where the corresponding phrase would be
‘matita G iA L İa ’ . She therefore suggests that speakers, rather than simply
transferring L l intonation patterns, have ‘difficulty in adjusting to a new
linguistic paradigm’, in that they fail to recognize the communicative value
of prominence in English (1990: 151).
Similar non-understanding results where there are a number of
segmental errors in conjunction with misplaced contrastive stress, even
vvhen the syllable vvrongly presented as the nucleus is othervvise error-free.
Hovvever, once the non-understanding has occurred, it is generally easier for
interlocutors to clarify vvhat vvas said if the segmental problem is restricted
to a single vvord rather than occurring throughout the word group. An
example of the latter—the combination of segmental errors and misplaced,
though segmentally correct, nucleus vvithin a word group—vvas overheard
during a social conversation. A Taivvanese and Svviss-Italian (though
trilingual) speaker of English, both at a reasonably advanced level of
proficiency, having been banished to the garden as smoking vvas not
permitted indoors, vvere inevitably discussing their smoking habit. During
the course of the conversation, the Taivvanese speaker said to his Svviss
interlocutor, ‘I smoke more than you do ’. Not only did he fail to use
nuclear stress contrastively and instead place it on a non-content vvord, but
he also made segmental errors of both quantity and quality in the vvords
‘smoke’ and ‘more’, vvhich he pronounced respectively as [zmok] and [mo],
After three repetitions had not clarified the meaning for his interlocutor, I
‘translated’ the sentence.
As has already been argued, the number of production errors that L2
speakers of English stili make in relation to contrastive stress at a stage by
vvhich they are able to understand it if correctly produced (vvhether by L l
or L2 speaker), implies that receptive ability well precedes productive
ability in this area. In Chapter 6, as vve come to conclusions about vvhat
should be included in a phonological common core for ILT, I will
substantiate this claim by outlining an experiment I carried out, which

provides clear evidence of a receptive-productive mismatch in the use of
contrastive stress. In the meantime, however, we will move on to examine
the issue of intra-speaker variation and its implications for ILT.
Notes
1 This is not to suggest that L1 phonological transfer is itself a simplistic
unitary process. On the contrary, it interacts with developmental and
universal factors in complex ways, as discussed in detail in Chapter 5.
2 See for example, Cruttenden (1986); Kenworthy (1987); G. Brown
(1990); A. Brown (1991); Dalton and Seidlhofer (1994a); Daniels
(1995).
3 Though this was probably because the topic of this sentence had only a
tenuous link with what had gone before, I was denied contextual help
and, like the NBES students, was forced to rely on the acoustic signal
alone.
4 The term ‘discourse meaning’ is used here to refer to the way speakers
are said to distinguish between a nucleus bearing new information and
one bearing shared or ‘given’ information, by means of a falling tone and
a fall-rise respectively (see Brazil 1997, and page 49).
5 See Brown, Currie, and Kenworthy (1980) for evidence of tone units
containing two nuclear syllables in informal speech.
*
T
3 The variation problem 2:

Intra-speaker variation
‘You have frequently seen the steps which lead up from

the hail to this room ... how many are there?’
‘How many! I don’t know.’
‘Quite so!’ said Sherlock. ‘You have not observed.
And yet you have seen.’
A. Conan Doyle, ‘A scandal in Bohemia’ in The Adventures
o f Sherlock H olm es 1894 (Penguin, 1981)
It was argued at the beginning of Chapter 2 that those who are aware of L2
intra-speaker variation generally interpret it merely as variation in
correctness. In terms of L2 spoken English, variation equals error, and the
entire EFL endeavour is, in a sense, directed towards the standardizing of
learners’ speech to bring it in line with an imagined (see page 27) L1
Standard. Not insignificantly, the verb ‘standardize’ is stili very much in use
on teacher training courses. New’teachers are trained first to elicit, present,
or introduce in some other more student-centred way, a new linguistic item,
be it grammatical or lexical, and then to ‘standardize’ it with each
individual student and/or chorally (i.e. with the whole class in unison). The
purpose of this is to bring the students’ production— by vvhich is chiefly
meant their pronunciation— as close as possible to that of a ‘native
speaker’. Later on, whenever the student deviates from this Standard pro
nunciation, the ‘good’ teacher will attempt to correct this ‘error’, whether
at the precise moment of its making (if the classroom focus at that point is
on accuracy) or later on (if the focus at that point is on fluency). Learners
expect this sort of remedial intervention from their teachers and regularly
complain if they do not receive what they consider to be enough error
correction.
Ali this presupposes, of course, that L2 learners of English should be
aiming for some sort of monochrome, monolithic English, completely free
of any intra-speaker variation in either style or grammatical correctness.
But L1 speakers of English vary their use of the language in both these
areas, however much the majority are unaware that they do so (see page 26
above). In fact, it was only with the publication in 1975 of Crystal aı

Davy’s pioneering work, Advanced Conversational English, that evi
specialists in the field of ELT received their first hint that ‘native speakeı
sometimes speak ungrammatically in addition to employing extensi
stylistic variation. The situation is further complicated in the sense that ti
so-called ‘Standard’ from vvhich these ‘ungrammatical native speakers’ a
deviating is by no means a unitary one. Rather, the term ‘Standard Englis
represents an idealized abstraction far removed from the realities of ‘ti
diversified and variable data of everyday interaction’ (Milroy 1994: 15
and see also page 26 above). In real life, then, ‘“acceptable” English’
‘remarkably variable’ (Brovvn et al. 1994: 153) and this is so particularly
speech, and stili more particularly, at the phonological level.1
Given these facts of L l use of English, Brown argues that ‘we should n
suppose that the second language learner is working tovvards some quı
unattainable (and undesirable) goal—fixed and stable knovvledge of t
target language’ (1996: 191). Far more useful to L2 learners of English
an ability to adjust their language in both style and level of accuracy (t
two often being closely related in both L l and L2 English) in response
the particular set of circumstances in vvhich an interaction is taking plac
And, because learners inevitably engage in intra-speaker variation, it mak
far more sense for pedagogy to build overtly on this automatic process
order to give them more control över it, instead of trying (and failing)
eradicate variation from their ILs altogether.
It could be argued that intra-speaker variation is a matter of natuı
adjustment in use, as speakers converge towards or diverge from the spee
of their interlocutors, and therefore that such accommodative process
cannot be taught— or at least do not need to be taught. Hovvever, I belie
that in this respect, EIL is a special case. In international speech settin
there vvill be an instinctive desire to converge as speakers bid f
membership of th£ international community and simultaneously attempt
make their speech intelligible to interlocutors from a range of L'
Hovvever, the means to converge in this way are unlikely to be wit!
speakers’ L2 repertoires unless there has been pedagogic help in tl
direction.
L l and IL intra-speaker variation: a distinction

The concept of interlanguage as a linguistic system independent of both t
native and target language has long been recognized. Hovvever, the origiı
IL research (for example, Selinker 1972) took precisely the monolitl
approach criticized in the previous section, failing to recognize, let alo
attempting to explain, the potential of IL for internal variation. For ELs £
natural languages and, like ali natural languages, dynamic rather th
static. They are thus subject to various influences vvhich, in turn, lead
systematic language variation.
The variation problem 2: Intra-speaker variation 55
When such variation was first noticed, it was regarded by some scholars
as an indication that ILs were not, after ali, systematic and thus ‘as an
embarrassment for IL theory and its fundamental tenet that IL is a natural
language’ (Kasper 1989: 41). Nevertheless, early studies of IL phonological
and morphological variation, such as those of the Dickersons (for example,
L. Dickerson 1975; W. Dickerson 1976; Dickerson and Dickerson 1977)
and Tarone (for example, 1978; 1979; 1980), were able to demonstrate IL
systematicity, while Corder (1978) brought to the attention of SLA
researchers the fact that first languages themselves vary systematically along
sociological and situational parameters. The focus of IL research thence
shifted to the study of its dynamic character.
Much of the literatüre on IL variation draws parallels with the
sociolinguistic variation of L1 speakers, arguing that the same Labovian
motivations of situation and linguistic context are concerned (for example,
Ellis 1985, 1994; Sharwood Smith 1994).. However, the similarity is to
some extent one of process rather than of product (and even at the level of
process there are a number of differences), since learners sometimes vary
their language in ways not found in either the L1 or the L2 (Cook 1993:
82-92; and see the example of /dö/ given on page 31 above). In addition,
IL variation at ali but advanced levels of proficiency is characterized by
more variability in the production of linguistic ‘error’ than by shifts
between more and less colloquial styles; variability per se is also more
prevalent in ILs than in L ls.
In view of their differences, Tarone distinguishes between L1 and IL
variation by means of the terms ‘style-shifting’ and ‘register-shifting’: ‘in
interlanguage ... style-shifting should be viewed as distinct from the
phenomenon of “register-shifting”.—the sociolinguistic ability to speak
casually in casual situations, or formally in formal situations’. She argues
that ‘the second-language learner may learn only one register of the target
language, and stili style-shift within that register in the sense of paying
greater or lesser attention to speech’ (1982: 73) and, presumably, speaking
in a more grammatically or less grammatically acceptable way. Tarone
(1983) accounts for this situation by claiming that most classroom second-
language learners are likely to be exposed to only a single register.2 Sato also
draws attention to this area of difference between L1 and IL speakers,
pointing out that learners ‘do not have access to the second language norms
about which linguistic forms are associated with which social parameters’
(1985: 195).
Some qualifîcation is called for here. While such differences between L ls
and ILs undoubtedly do exist, for two reasons the situation seems to be less
polarized than these writers suggest. Firstly, as I pointed out earlier, we have
known for many years that educated L1 speakers of English make grammar
‘errors’ in their speech (far less so in their writing). In other words, their
spoken language varies not just in style and register (in the conventional
senses of these terms: see Note 2). It also varies in standardness, that is, in
grammatical acceptability. Such speakers, even in relatively formal
situations such as lectures, make grammatical slips in areas like subject/verb
agreement. Secondly, for the past twenty years or so, second language
classrooms and materials have begun to devote attention to concepts of
formality and social appropriacy. As a result, many learners are nowadays
likely to be exposed to more than one level of formality (i.e. style—or
‘register’ in Tarone’s terms) and to the matching of linguistic forms with
social parameters from a relatively early stage of learning.
It is ali, then, a question of balance. L1 speakers employ variation in style
and register extensively and also, though less so, variation in grammatical
(including phonological) standardness. For IL speakers it is the reverse:
vvhile their speech does shift among different levels of formality, this sort of
variation is not their most frequent. ILs are characterized by a far greater
degree of variation in the form of grammatical error than are L ls. One
contributing factor is the fact that ILs are susceptibile to permeation by both
L1 (transfer) and aberrant L2 forms during— and, through fossilization,
even beyond—the learning process. Another is that learners have different
degrees of control över items according to the status of these items in their
current IL: perhaps their knowledge of the item is incomplete; or perhaps
they have not yet gained full control över the item and encounter problems
of use when faced with a heavy processing load (see Sharvvood Smith 1994:
111 - 2 ).
Phonological intra-speaker variation and its effects on

interlanguage talk
In this section, we will examine some data collected as part of a study into
phonological convergence in ILT to find out what actually happens when
Non Bilingual English* Speakers (NBESs) attempt to converge on one
another’s pronunciation. Because the subjects are not the L1 speakers or
fluent bilinguals of most accommodation research, we cannot assume that,
like the latter types of subjects, they will adjust their pronunciation in the
direction of that of their interlocutors. One reason for this is the repertoire
problem discussed in the previous section. Speakers from different L1
backgrounds frequently depart from target pronunciations in different
ways. They are not necessarily able to produce the various phonological
substitutions and conflations of their different-Ll interlocutors with the
degree of facility and automaticity required in oral interaction, if at ali.
Another reason is that although in multilingual (but not monolingual)
classes, students by definition receive exposure to other IL accents and över
time become accustomed to them, these IL pronunciations are rarely, if ever,
the subject (for reception) and never the goal (for production) of teaching.
A third reason is the psychological (though largely unfounded) fear
The variation problem 2: intra-speaker variation 57
commonly found among students in multilingual classes of acquiring peer

group pronunciation errors, and their consequent resistance to this
outcome.
In an earlier pilot study (reported in Jenkins 1995) whose aim had been
to establish that the intelligibility problem was greater in ILT than in other
interactions involving NBESs, the majority of subjects had claimed to find it
easiest to understand English speakers from their own L l backgrounds and
most difficult to understand those from unrelated L l backgrounds. They
had most frequently cited pronunciation (and hence, by implication, L l
phonological transfer) as the majör source of the intelligibility problem.
Where convergence occurs in ILT it is therefore likely to be similarly
motivated to foreigner talk (FT), vvhich is itself a form of convergence (see
Chapter 7). One of the main motivations of FT is the ‘native speaker’s’
desire to be understood by his ‘non-native’ interlocutor. Because he has
difficulty in understanding the latter, vvith pronunciation being a prime
factor, the ‘native speaker’ assumes the difficulty to be mutual and begins
speaking more slowly, articulating more clearly, and so on. In ILT a similar
(though two-way) process is likely, the main difference being that the
adjustments operate in both directions, and on IL rather than L l.
For ali these reasons, it was predicted that phonological convergence in
ILT vvould take the form of replacement of problematic (for the receiver)
transfer, with interlocutors attempting to converge largely on target-
language forms, rather than on each other’s pronunciation. This, in turn,
vvould lead to variation in target and non-target forms, relating to the
success or othervvise of the transfer replacement process. Such success vvas
thought to be dependent on three main factors. Firstly, the type of transfer
involved, vvith some phonological and phonetic areas being more taxing
than others for speakers from particular L ls. For example, Japanese NBESs
have great difficulty in producing the sound İ3'J. Secondly, the amount of
attention the speaker vvas able to give to replacing transfer rather than
planning content. Thirdly, the salience of interlocutor comprehension at
any given point in the interaction: in other vvords, vvas the intelligibility
pay-off likely to be vvorth the cognitive effort involved? The latter, it vvas
predicted, vvould depend on the type of task being performed. Thus, if
crucial information vvas being exchanged, for example, then phonological
intelligibility vvould be highly prioritized for the speaker.
Having said ali this, hovvever, it vvas not at ali clear hovv the process
vvould vvork in practice. Would transfer replacement affect ali of a speaker’s
habitually-transferred forms or only certain forms thought, for some
reason, to be ‘high risk’ (i.e. more likely to cause intelligibility problems for
the interlocutor)? Or vvould it affect only those forms that the interlocutor
also produced vvrongly but differently from the speaker? Or vvould it
perhaps affect only those forms that the interlocutor produced correctly
(i.e. in the manner of the target language), thus implying convergence on the
interlocutor’s correct pronunciations? Were different combinations of these

alternatives possible, perhaps depending on the State of play at any one
point in the interaction, itself involving factors such as the linguistic
environment of a potentially transferred item and the availability and type
of contextual cues?
The data that follow are drawn from a study in which six learners of
English, two Japanese, three Swiss-German, and one Swiss-French ali of
upper-intermediate/low-advanced level, were recorded completing various
tasks as they practised for the Cambridge Certificate in Advanced English
Speaking examination. They were recorded first in different-Ll pairs
recorded on four separate occasions över a ten-week period, and also
regrouped and recorded in same-Ll pairs (though not in the case of the
single Swiss-French subject). The eight hours of recordings were
transcribed, listened to again, and the transcriptions annotated phonetically
vvherever pronunciation deviated from an L1 target form. The annotated
transcriptions were then examined for evidence of phonological
convergence by means of suppression of L1 transfer errors, or a lack of it,
depending on whether interlocutors shared one another’s L1 or not. The
phonological deviations that did occur were analysed qualitatively and then
selected variables were subjected to statistical tests (Chi-square). The
subjects also took part in interviews and completed questionnaires.
Subsequent analyses, along with a replication using new subjects from a
wide range of L ls (not engaged in examination practice this time), focused
on phonological differences within the different-Ll dyads according to task
type. (For a fuller account of the methodology, see Jenkins 1995.)
We will turn now to the data themselves. The first four extracts are
typical of the data collected in order to compare pronunciation in same-Ll
and different-Ll dyads. To enable valid cross-comparisons to be made, the
following are ali taken from the information exchange task (i.e. Task B) in
each set of recordings. Th# speaker in Extracts 1 and 2 is one of the Swiss-
German subjects, and that in Extracts 3 and 4 is one of the Japanese
subjects. Full discussion of the implications of the data is provided after
Extract 4. For the time being, the important point to notice is the far greater
number of phonological deviations in the same-Ll dyad as contrasted with
the different-Ll dyad in each pair of extracts.
Deviant pronunciations are indicated by means of the words containing
them being transcribed into phonetics immediately beneath them. Word
stress is indicated only where it is misplaced.
rr
First study
Extract 1: SG1 (Swiss-German) in differen t-L l dyad
(J) indicates th at the Japanese in terlocutor spoke briefly a t this point.
Okay, it’s a four storey house with two large balconies and one small
wıv baelkDnis sent
balcon, this the small balcon is on top-is the highest one (J) Balcon (J) I
dıs
think it’s the right word. And in front of the house are is a is a yes it’s a
Eent da
road, and on this road is a a lorry. And and in front of the house too there
j9 ü t dıs sent ds
are is a parking a small parking space with let’s say one, two, three, four,
WIV
five, six, seven, eight parked cars and most of the cars are covered with
kaus da kaovvad
snow. And on the left side of the house there are there are four or five
da deaı
parked cars. Four are co-five are covered with snow and one is is, a red a
kaüvat wıv
red car is not covered with snow. In the back of the hou-of the house you
kaovad wrv da
can see, on the right side of the back of the house is you can see a mountain
with er covered with with trees and snow of course. And there are a f-few
wıv küvad wıv aent
houses behind this main house I described to you.
bıh aın t
Extract 2: SG1 in sam e-L l dyad

(SG 2) indicates th at the Sw iss-G erm an in terlocu tor spoke briefly at this point.
Ali I can see is one square, it’s (unintelligible) first with with two dia-dia-
ıs wıv wıv
diagonals I guess, this is the word, and now in every every corner of your
daıasgDncels W3:rt Df
square is er, is another er, the square is yeah, a small square in every corners
sk v ea j aenüdaı
of your big square is a small one, and the length is about two, two-and-a-
da lerjgs
half, no three centimetres ... (SG2) Yeah. So you have four small squares in
sen tım itaıs haef skveaıs
the big square. Then you have the er a square with the same size in the
da den da skveaı wrv da saıs da
middle where the two diagonals diagonals crosses each other, you have
da daiEegDnsels hsef
another square. (SG2) Same size as the other (unintelligible) (SG2) Yes, you
DÇİ3J
have then (SG2) parallel to the the length of the big square ... Okay, then
hsef da leıjgs da
you have, if you have drawn this er small one in the middle er the four
haef d ıs o n dıs zmo:l dg
corners of this small square er hit the diagonals. (SG2) Then from there you
k o a n a ıs dıs skvear dauegDHiBİs den d e sj
draw a üne to the middle of the white, the length of the big square, so it
d ja o ds Df da leggs ds skv eaj
gives you er (SG2) Four (SG2) Yeah, like arrows ... They ali have the same
grfs seısus hsf
size ... should have the same size,
saıs
Extract 3: J l (Japanese) in different-Ll dyad

(S G l) indicates th at the Sw iss-Germ an in terlocutor spoke briefly at this point.
There is ah three sofas and one is one for three for four persons and other
sao faız
and the rest are for one person. And two cushions on the longer
k ü s jâ z
(unintelligible) and one cushion is on the right hand side of the sofa and ...
sa o fa :
and about the middle orange (unintelligible) is on the table (SGl) Ah,
between mm between the sofa for one people-for one person, and also ahm
sa o fa :
newspapers on the table, square or rectangle I don’t know and the table is
made of wood and glass. Mm ... mm and ah a fruit basket on the cup
<{>fu:t
cupboard and in the fruit basket I can see pine-pineapple and, grape ...
fruit? No, grape and maybe peach, and er there are three pictures on the
sj İ:
wall. Mm on the picture J can see mm I don’t know the name but ali are of
flowers. Mm ...
fjQ Ö W 9 Z
Extract 4: J l in sam e-L l dyad

(J2) indicates th at the Japanese in terlocutor spoke briefly at this point.
I can see a man with his mouth widely open and also his eyes are wide um
ma:n
circle. And mm he o-he opens his eyes wide wide widely and he is wearing
glasses ... and he is wearing hat and he has got whiskers, like, um whiskers
hAt wıska:z
(unintelligible) beard, beard beard here, whiskers on the cheek. (J2) Mm his
bıa:d bıa:d wısk3:z
hands are, he he ... is rise-raise- he is rizing his hands. (J2) Ah, both, över
9üvq:
över his head ... and he is vvearing rings on every every fingers ... He is
şo v a: f ı ı jg 3 :z
wearing neck-er tie. (J2) This one ... and ... (J2) eh ... it’s not round but
square, I think it’s like drop, ah raindrop. (J2) It’s like kind of hat (J2) Hat
sk w ea: je ın d s iD p hA t
hunting hunter (unintelligible) (J2) Yes ... And um ... no, no, no, not beard,
b ıa :d
whiskers ... (J2) Whiskers on the cheek like hairs ... Ah also he is vvearing
w ıs k a :z
watches on both wrist ... Um, I can see two strings from-hanging from
s ts ın ıg z haeggnj
glasses. (J2) Until the side of head, what, I don’t know what it is ... to hold
glasses. If when you take off glasses.
These two sets of extracts have been provided in full in order to exemplify
the contrast observed throughout the data in both amount and type of
phonological deviation between the different-Ll (DL1) and same-Ll (SL1)
condition respectively. Clearly, there is considerably more phonological
deviation in the latter than in the former. And this is ali the more striking
when we consider that these particular DL1 data vvere recorded some six
vveeks ahead of the SL1 data. If no other factors had been involved, vve
vvould have expected phonological error to be more evident in the former
than in the latter, since phonological competence could be expected to have
improved during a six-week period of language learning. More importantly
in a qualitative sense, vvhen vve examine Extract 1 (DL1 dyad) more closely,
vve find that the majority of S G l’s deviations concern non-essential,
grammatical words such as ‘and’, ‘with’, and ‘the’, rather than lexical items
crucial to the information exchange task in hand (the subject’s interlocutor
is trying to identify from six pictures vvhich one is being described). In
addition, ‘with’ is alvvays pronounced vvith a final /v/, vvhich is acoustically
far closer to 16/ than is /d/, making the vvord easier to interpret.
The three exceptions are ‘balconies’/‘balcon’, ‘road’, and ‘covered’ (four
times). The L1 influence in the first can be explained by the fact that the
vvord is a cognate. The vvord ‘road’ is immediately repeated correctly.
Pronunciation of the third item, ‘covered’, hovvever, appears to cause the
speaker some difficulty. Each time he repeats it, he gets a little closer to the
target sound, going from tvvo relatively serious errors (of consonant
substitution and vovvel quantity: see Chapter 6), vvhich his interlocutor did
not understand (she later said that she had eventually guessed the meaning
because of the repetitions of the vvord ‘snovv’) to one minör error (the
wrong short vovvel). Interestingly, vvhile the spelling continues to influence
his pronunciation until the fourth attempt, the subject removes his transfer
error (the substitution of Ivl vvith /w/, in fact an example of overcompensa-
tion) at the second. Ali this suggests strongly that he is attempting to replace
62 The Pbonology o f English as an International Language
those non-target pronunciations which he thinks are likely to cause his

partner comprehension problems, and thus that he has—whether
consciously or subconsciously—identified ‘high risk’ categories.
Moving on to Extract 2, the same speaker in an SL1 dyad, the deviations
from target are not only more frequent, but they also involve more content
words crucial to the successful completion of the information exchange task
(‘square’, ‘diagonals’, ‘length’, and ‘centimetres’) which involves the
subject’s interlocutor in trying to draw the diagram that is being described.
In the feedback session after this interaction, the two Swiss-German
subjects agreed that their pronunciation had seemed to be more accurate
when they were paired with their Japanese partners, but pointed out that
they had found one another very easy to understand, far easier in fact than
they had found their respective Japanese interlocutors.
The next two extracts (3 and 4) come from the speech of one of the two
Japanese subjects. These extracts, like those of the other Japanese speaker,
contain noticeably fewer non-target forms than those of the Swiss-German
subjects. However, this is not the result of higher phonological competence
on the part of the Japanese subjects. Rather it reflects cultural differences in
that they tend to say less, pause more frequently and for longer periods,
speak more slowly, and probably think more carefully before they do so.
This reflects the subjects’ prior educational conditioning, in which making
mistakes involved losing face and was to be avoided if possible (see
Takahashi 1989 on these points). In the SL1 situation, this tendency is
exaggerated: the Japanese subjects each say less in total than they generally
do in their DL dyad recordings, they speak stili more slowly, and pause
more frequently and for longer. Convergence in the DL1 dyads is thus
reflected in these extra-linguistic areas, as well as in the greater number of
transferred forms which nevertheless occur.
Because the Japanese subjects monitor their output so rigorously, they
considerably reduce thefrisk of automatic phonological transfer slipping
through in the way that it does with the Svviss-German subjects, despite the
latters’ attempts to suppress transfer in DL1 dyads. Nevertheless, there are
some phonological differences in the Japanese DL1 and SL1 data: not only
are there rather more phonological transfer errors in the SL1 situation, but
there are also differences in error type. In Extract 3, two of the deviations
are phonetic rather than phonemic. Although phonetic problems may cause
NBESs more problems than they do fluent speakers (see above, pages 36-7),
they are in many cases stili likely to be less serious for NBES
comprehensibility than phonemic errors, such as the substitution of /r/ with
!\1. This is true of the substitution of /fr/ with [<jj>f] in the word ‘fruit’. On
the other hand, her pronunciation of ‘cushion’ contains one of the phonetic
errors that will later be identified as a potential problem for intelligibility,
namely the elision of word-final İn i accompanied by nasalization of the
preceding vowel. However, the subject repeats this word correctly a few
seconds later.
T
In Extract 3, the pronunciation deviation potentially most threatening to

intelligibility for her NBES interlocutor is the substitution of /V with İri in
the vvord ‘flovvers’. The confusion betvveen /r/ and /]/is relatively rare in J l ’s
speech and, in this instance, may, in fact, be a case of overcompensation.
According to my Japanese informants, the single Japanese phoneme that is
somevvhere betvveen these two English phonemes, and of vvhich the latter
are merely allophones, is actually closer to l\J than İri. Japanese speakers of
English, I am told, therefore find this sound the easier of the two to
produce. Early on, they tend to substitute İri with l\J more often than vice
versa. Hovvever, as they increase in profîciency, they become very conscious
of the fact that they produce İri incorrectly and thence for a time err in the
opposite direction, until they finally begin to distribute the tvvo phonemes
correctly. By comparison vvith her DL1 deviations, ali those in this Japanese
subject’s SL1 extract (Extract 4) are phonemic. She repeats the same errors
(in ‘beard’, ‘vvhiskers’, and ‘över’) three times vvithout any apparent attempt
to correct them. The problem that her SL1 interlocutor has vvith the vvord
‘vvhiskers’ (implied in the text and clarified in the follovv-up intervievv) is
because she does not actually know this (somevvhat archaic) vvord.
The overall picture provided by the data of the first study is one of
speakers making considerable effort to replace L l phonological transfer
vvhen they are interacting in English vvith a speaker from another L l as
compared vvith a speaker from their own L l. This vvas confirmed by
statistical tests very strongly in the case of the Svviss-German speakers, less
so in the case of the Japanese speakers, largely because of the lower number
of tokens (examples) obtained.
For various reasons vvhich vvill be discussed later on vvhen we consider
the problematic issue of same-Ll interlocutors in dyads and groups
(Chapter 6), and the role of context in intelligibility (Chapter 4), it vvas
found that the SL1 data probably contained more transfer than these
speakers vvould use in their unmodified vernacular. Thus the SL1 data do
not offer the best possible base-line against vvhich to measure the degree of
replacement of potentially transferred phonological forms. In this
connection, better data vvere found to be provided by the recordings of the
subjects engaged in social conversation, vvhich is probably as near as we can
get to the subjects’ vernaculars. Because of this, the data of the first study,
along vvith those of the replication, vvere examined for evidence of
contrasting amounts of phonological convergence in the social conversation
and information exchange tasks, vvhere the success of the task is dependent
on the mutual intelligibility of the tvvo DL1 interlocutors.
For the moment, we vvill look at extracts from one of the replication
recordings, vvhere a Taivvanese and a Korean interlocutor are engaged first
in social conversation, and then in tvvo different information gap activities.
The reason for including these data at this point is to demonstrate that the
type of convergence (by replacement of transfer) vvhich vve have claimed
and identified in ILT really does involve the making of adjustments

according to the needs of the receiver, and not merely an attempt at
indiscriminate reduction in phonological (transfer) error. This can be clearly
demonstrated where a subject increases the transfer of a particular L1
phonological feature in the information exchange task without affecting ILT
intelligibility. We will look first at two extracts from a Taiwanese subject’s
speech and then at two from her Korean interlocutor.
Replication
Extract 5: Taitvanese subject, social interaction task
(K) indicates th at the K orean in terlocutor spoke briefly at this point.
Middle country, and here I I think I’ve been, I born there for a very long
m ıdu b5 b?
time, I’ve never moved to the big city or the other place. Yes, but I’ve fin-
taı? m ov da
I’ve just finished the senior high school and come to Britain ... London,
fm ıj sınıo sk ö kA b u t®
when I first come here I don’t-I didn’t like London because first I don’t like
kA d5 dıds loda do
the food, yeah, it’s quite terrible in here I think, you know in Taiwan
ds kw aı? sııjk
(unintelligible). (K) Then also I don’t like the weather. (K) But now I’m used
X Z 30 weza nao jo z
to. (K)What do, what do you think? (K) But I think in your country there
sırjk deı
are lot(s) of sunshine. In your country it’s warm. (K) It’s a different way.
b9 dıfs
Extract 6: Taitvanese subject, information exchange task

*
In my picture I think they’re in a garden. The the house, be-er behind the
sıg k b ıh a ı?
house, they have the small garden. And there are one two three four five six,
g a :d s sis
six people in the garden. And I think they er have er one man and with his
g a :d s wız
wife and his mother I think, and they’ve got er three children, two boy, one
mAds sırık
baby. And they are smiling, it seems quite happy and ... er, they’re in the
garden and (unintelligible). I don’t know what else I can say, but the
kae
woman, ah she hold a baby, and ... and, ah, the er old woman she sit
coman hood ood
in the chair in the left my picture, left-hand, and the man sit on the right
ıa ı?
side. And the other people they are standing. (K)
s a ıî
Extract 7: Korean subject, social interaction task

(T) indicates th at the Taiw anese in terlocu tor spoke briefly a t this point.
Yes, the Capital of Korea is Seoul and now, right now, I’m living in Seoul.
IZ 3
But actually I was born in the Southern part of Korea but, I studied in Seoul
cektuasri pAt büt
and after finish at the school and finishing my study I got a job at Seoul so-
a :p ts p m ıj sta sku güd
in Seoul, so I now live in Seoul (T) Ah my family. Yeah. Actually, right now
aek tü sri
I’m living with my wife and my son, but ... the concept of family is very
waıp
different I think here and in my country. In my country when I, when we
sıgk
say about family, uh we think that we have father, I have my father and my
paemıli sııjk p a ıd s
mother and my sisters and my brothers, ali equally, and ... uh I have three
m tıds sıstas sıi:
sisters and (T).
Extract 8: K orean subject, information exchange task
(unintelligible) teli you the six faces I’ve got. The first face I have is smiling
faıst
face (T) Yeah. Actually he’s very happy and his mouth is very big. (T) Yeah.
feıss Eenda maus bıgs
And his eyes is mm, almost closed. Can you imagine? (T) Yes ... And the
ktauzds
other one I have is er maybe feel, she İs very astonished, she is surprised. (T)
astDmJts
and so her eyes is very large, big big. And the other one is oh, now he’s very
la:d3İ big a
unhappy, so his eyebrush, eyebrow eyebrow is ... shortened, eyebrow is
aıbiAjı aıbjaü aıbjau Jütanda
closed.
Both the Taiwanese and Korean languages show a strong preference for CV
syllable structure, and speakers of both therefore have considerable
difficulties in producing the many consonant clusters of English. The
Taiwanese subject deals with the problem by means of consonant deletion,
sometimes replacing the deleted consonants with a weak glottal stop (which
at least indicates to the receiver that something is missing). This is, in fact,
her most frequent type of phonological error. However, in those many areas
where consonant deletion is not permitted according to L1 rules of English,
it is potentially very damaging to intelligibility, because it makes recovery
of the original form difficult, if not impossible. The Taiwanese subject seems
to be aware of the problem— she has identified deletion as a high risk
strategy for ILT intelligibility— but at this stage in her acquisition of English
is only able to suppress deletion by the exertion of great cognitive effort.
This she clearly does in Extract 6, the information exchange task, where
words such as ‘children’ are pronounced without deletion—whereas she
tended in her everyday vernacular classroom speech to produce it as [Juıe?]
(see page 38).
On the other hand, the Korean subject’s addition of paragoge to the ends
of several words (usually schwa paragoge, sometimes [i], depending on the
sound that precedes it) follows the opposite pattern. In the information
exchange task, while reducing his other transfer errors, particularly the
substitution of /p/ for /f/, he actually increases his use of the Korean IL
strategy of paragoge. Although his current classroom vernacular IL contains
little paragoge, as evidenced by its scarcity in Extract 7, he appears to
regard it as an aid to intelligibility for his ILT interlocutor, and therefore
uses it extensively in the information exchange task, particularly on key
items such as ‘closed’, ‘astonished’, and ‘shortened’.
Both subjects admitted reluctantly in follow-up interviews that there had
been occasions when they had not understood one another, and that the
main cause had been the other’s pronunciation. Assuming that the chief
motivation in ILT dyads, and especially those involving the exchange of
crucial information, is communicative efficiency, the opposite behaviour of
these two subjects in simplifying English syllable structure can be explained.
Their mutual difficulty in interpreting one another’s pronunciation can
account for both the Taiwanese subject’s decrease in consonant deletion and
the Korean subject’s increase in paragoge in the information exchange tasks
as compared with the social conversation. This bears similarities to the
motivation underlying foreigner talk (itself a manifestation of convergence),
i.e. the speaker’s difficulty in understanding an NBES interlocutor leading to
an assumption that the latter will have corresponding difficulties in
understanding him. *
We will return to the implications of accommodation theory for ELT
pedagogy in Chapter 7. For the time being, there is an important conclusion
to be drawn relating to attitudes to L2 variation. This takes us back to the
beginning of Chapter 2, where it was made clear that language variation
away from ‘the Standard’ is vievved by many people (including pedagogues)
very negatively. However, we have since observed that L2 variation is very
often the result of an attempt to produce pronunciation that is intelligible
for the particular interlocutor— an attempt which can lead to both more
and less target-like production, according to the interlocutor’s
pronunciation features involved. Phonological variation vievved in this light
is a positive and not a negative characteristic of IL speech. Indeed, it appears
from the data gathered in the first study that it is precisely the lack of such
convergent behaviour which threatens successful communication in ILT.
The main implications for EIL pronunciation pedagogy that emerge from
this chapter are therefore as follows. First, we need to get to grips with the
nature of intelligibility as it relates to these specific L2 contexts, and this we
attempt to do in the next chapter. Second, it will be important to identify
contrived norms based on a subset of core EIL phonological features, which
can then be learnt by ali international speakers of English (including L1
speakers of English). This problem is tackled in Chapter 6. And third, we
must devise methods for classroom pedagogy which build on the instinctive
desire of NBESs to accommodate phonologically to their interlocutors in ILT
speech situations. We return to this subject in Chapter 7.
N otes
1 In fact, with the rise of what I term the ‘corpus syndrome’, the pendulum
has in one respect begun to swing the other way. Corpora without doubt
have important uses, and one of their most salient disclosures for ELT
has, to my mind, been the extent to which L1 speakers of English speak
ungrammatically. This has serious implications for the correction of L2
English learners’ errors though, regrettably, to date little has changed in
this direction in classroom practice. On the other hand, the existence of
the various corpora (British National Corpus, Collins COBUILD, and the
like) has led to an obsession with ‘real’ English, and a feeling that in the
classroom we must predominantly teach language that has been
validated as ‘real’ or ‘authentic’ by virtue of its featuring in a ‘native
English speaker’ corpus. For a fuller discussion of the issues at stake
here, see Widdowson (1991).
2 Note that Tarone’s interpretation of the terms ‘style’ and ‘register’ is not
that generally employed by soc.iolinguists, who usually discuss ‘style’
with reference to level of formality, while reserving ‘register’ to describe
the language specific to particular domains, such as occupational and
interest groups.
'T
4 Intelligibility in interlanguage talk
‘Then you should say what you mean,’ the March Hare
went on. ‘I do,’ Alice hastily replied; ‘at least—I mean
what I say—that’s the same thing, you know.’
‘Not the same thing a bit!’ said the Hatter. ‘Why, you
might just as well say that “I see what I eat” is the
same thing as “I eat what I see!”’
Lewis Carroll, Through the L ookin g Glass, Chapter 1
What do we mean by intelligibility?

As far as the international intelligibility of English is concerned, the view
which once prevailed among applied linguists and ELT professionals was of
‘a one-way process in which non-native speakers are striving to make
themselves understood by native speakers whose prerogative it was to
decide what is intelligible and what is not’ (Bamgbose 1998: 10). This view,
although probably stili widely held by the British and American general
public and even by many teachers of English, is no longer tenable. Not only
does it ignore the ‘non-native’ listener’s perspective (a serious enough
omission in ‘native-non-native speaker’ interaction, a contradiction in
terms in ‘non-native-non-native speaker’ communication), but it also fails
to acknowledge any active role for the receiver. More recently, the focus of
research into intelligibility has tended to shift away from the speaker
towards the listener, and to consider the contribution of factors such as the
listener’s background knowledge and processing skills. This, however, is
stili to simplify matters, for there is as yet no broad agreement on a
definition of the term ‘intelligibility’: it can mean different things to
different people. So, before looking more closely at the intelligibility of
interlanguage talk, it would be as well to clarify how I interpret the term in
the context of communication between Non-Bilingual English Speakers
(NBESs). Let us begin by considering what others have said on the subject.
Bamgbose himself describes intelligibility as ‘a complex of factors
comprising recognizing an expression, knowing its meaning, and knowing
what that meaning signifies in the sociocultural context’ (ibid.: 11). He thus
uses ‘intelligibility’ as a blanket term to cover a range of speaker and
listener factors, arguing that in communication between a speaker and

listener, both contribute to ‘the speech act and its interpretation’ (ibid.).
Bamgboşe’s three aspects correspond closely, as he points out, to Smith and
Nelson’s terms ‘intelligıbilıty’, ‘comprehensibility’, and ‘interpretability’.
The latter authors had earlier noted that in the literatüre on the subject of
the international intelligibility of English, the three terms were often used
interchangeably, with the result that ‘the current State of the art in the study
of the international intelligibility of English is one of confusion regarding
these terms’ (1985: 334). In order to clarify the situation, they suggested
that the term ‘intelligibility’ be reserved for word and utterance recognition,
with ‘comprehensibility’ being used to refer to word and utterance meaning
(i.e. propositional content, or Austin’s (1962) locutionary force), and
‘interpretability’ to refer to the grasping of the speaker’s intention in
producing the utterance (i.e. Austin’s illocutionary force). The term
‘interpretability’ thus replaced for Nelson his previous use of the word
‘intelligibility’ to mean ‘apprehension of the message in the sense intended
by the speaker’ (1982: 63). intelligibility, as defined by Smith and Nelson,
has something in common with Brown’s term ‘identification’, by which she
means the recognizing of items such as proper names and telephone
numbers, and which she contrasts with ‘understanding’—the grasping of
the communicative content of utterances (1995: 10-11).
James, on the other hand, approaches the subject from the speaker’s
standpoint. He uses the word ‘comprehensibility’ ‘as a cover term to refer
to ali aspects of the accessibility of the content—as opposed to the form—
of utterances’, with intelligibility being reserved for ‘the accessibility of the
basic, literal meaning, the propositional content encoded in an utterance’
(1998: 212). Thus, James intends to convey by the term ‘comprehensibility’
the same meaning as Bamgboşe’s ‘intelligibility’, and by the term
‘intelligibility’ the same meaning as Smith and Nelson’s ‘comprehensibility’.
James contrasts intelligibility with ‘communicativity’, which he describes as
‘a more ambitious notion, involving access to pragmatic forces,
implicatures and connotations’. Communicativity is ‘a higher order
achievement’ involving the transmitting of ‘the right social information’
(ibid.: 216-7). Lanham (1990) had earlier pursued the same line of thought,
though with a different allocation of the terminology, in his study of the
intelligibility of ‘errorful English spoken by second- or foreign-language
users’ by proposing a distinction between ‘intelligibility’ and ‘com
prehensibility’. The former he relates to the effect of errors on the
recognition of linguistic form while the latter, he argues, is concerned with
‘the communicative effect of error ... the consequences of error on the com
prehensibility of contextualized discourse’ (ibid.: 243). Brumfit, in a similar
vein, though predating James, contrasts the ‘intelligibility of world English
te x f with what he describes as the ‘richer problem’ of ‘interpretability of
world English com m unication’ (1982: 95; emphasis in original).
intelligibility in interlanguage talk 71
It seems from this small sampling of the literatüre that there is stili no
general consensus in the use of the term ‘intelligibility’, whether viewed
from a speaker or listener perspective. In other words, the terminological
‘confusion’, to which Smith and Nelson drew attention in 1985, is stili with
us. Nelson, in fact, a whole decade later, reiterates his view of the need for
a division of ‘general “intelligibility” or “understanding” into a three-level
system of intelligibility, comprehensibility and interpretability’ (1995: 274).
What is clear, though, is the view, expressed or implicit, in most of the
writings referred to above that, whatever the term actually used for the
concept of formal recognition and recognizability of words and utterances,
matters of form are considered by their writers to be of relatively minör
relevance in spoken communication (and miscommunication) as compared
with matters of meaning. We see this, for example, in James’s <1998)
contrast of ‘intelligibility’ with ‘communicativity’; in Lanham’s (1990)
distinction between ‘intelligibility’ and ‘comprehensibility’; and in Brumfit’s
(1982) claim for the ‘interpretability of ... communication’ as being ‘a
richer problem’ than the ‘intelligibility of ... text’.
The main research interest these days, it appears, is in higher-level
concepts going by names such as ‘interpretability’, ‘communicativity’, and
‘understanding’. The ‘real’ business of imparting and processing messages,
it is frequently said, involves the top-down processing of contextual
phenomena (background knowledge relating to personal and situational
cues) rather than the bottom-up production and reception of linguistic
form. As Brown argues, ‘adequate communication’ is regularly achieved,
despite ‘the pervasive underspecification of meanings of utterances’. This is
because the sheer amount of shared background information enables
interlocutors to establish ‘a structyre of mutual beliefs’. In other words,
speakers are able to construct and interpret utterances in the light of beliefs
about the other’s State of knowledge, and to ascribe to each other the
intentions which they ‘would expect to experience themselves in uttering
the utterance just heard in that particular context’ (1995: 232-3). Of
course, ali this is only possible because, as Brown points out (quoting the
sociologist Cicourel), there is a presumption that interlocutors are ‘playing
the same game’ (ıbid.). The interest is in discovering how this happens.
Although a number of writers (including Brown herself elsewhere) talk
of more interactive models of speech perception, in which higher and lower
level processes interact (see, for example, Flowerdew 1996), Brown is no
doubt quite right about the role of shared belief systems in communication
among fluent speakers of English (her comment was, in fact, made in the
context of research into interaction between L1 speakers of English). For
even leaving aside for one moment the role of extralinguistic contextual
phenomena, there is another compelling reason why recognizability and
recognition of form can play only a relatively minör part in the successful
conveying and receiving of messages among fluent speakers. This relates to
the quality and character of the speech produced and, in particular, to the
way in which the individual speech sounds merge together in an ‘acoustic
blur’ (Brown 1990: 11).
As many authorities have long pointed out (see, for example, Anderson
and Lynch 1988; Brown 1990; Rost 1990), fluent speech perception
involves rather more than the decoding of segments. This is because of the
effects of both co-articulation and of the various assimilatory processes
which take place in connected speech. Looking first at co-articulation, it is
very difficult to isolate individual sounds in the speech stream— and even
within one single word. Sounds change in different phonetic environments
through the influence of neighbouring sounds. For example, in the words
‘tea’ and ‘too’, the İti in the first case will be coloured by the following /i:/,
and in the second by the following /u:/. Thus, the former İti will be
pronounced with spread lips and the latter with rounded lips. Because of
the effects of co-articulation, it is usually difficult to identify where one
sound ends and the next begins. This means that ideal phonemes are
generally no more than ‘reference templates’ for fluent listeners (Rost 1990:
38).
Turning to assimilatory processes or what are more commonly referred
to as the ‘features of connected speech’, we find that there is rarely a one-
to-one match between words said in citation form (i.e. in isolation) and
those uttered within a stream of speech. Assimilatory processes are used by
fluent speakers to facilitate pronounceability by making articulation easier;
that is, they are speaker-oriented. They include processes such as elision (the
omission of sounds), catenation (the linking of sounds across words), and
assimilation (the replacing of sounds to make them in some way closer to
neighbouring sounds) (see Dalton and Seidlhofer 1994a: 24-31 for a fuller
account). These contrast with dissimilatory processes, where speakers
subordinate their speech strategies to their listeners’ needs by articulating
more clearly; that is, they«re listener-oriented. Dissimilatory processes are
marked. Fluent speakers of English do not normally use them, but reserve
them for occasions when clarity is particularly salient, for example, when
dictating an address över the telephone, or when conversing with an
interlocutor who has hearing difficulties, or with one whose English
proficiency is low (see note 3 and Chapter 7 on foreigner talk).
The use of co-articulation and of assimilatory processes of course
presupposes a certain speed of speech, i.e. that of fluent speakers of English.
According to David Crystal, this averages 350 syllables per minute, with
everyday conversation tending towards a range of 400 to 450 syllables per
minute, and with short stretches, such as sequences of auxiliary verbs, being
uttered stili faster. Slower speech made to incorporate features of connected
speech, as Crystal has demonstrated, is reduced to gibberish.1 The use of
such features also assumes that we are dealing with fluent listeners. And in
this connection, note that writers have a tendency to refer generically to ‘the
listener’, as for example, in ‘the ways in which the listener decodes

utterances’ (Rost 1990: 33; emphasis added). Thus they imply that
processes of speech perception are pretty much the same regardless of an
individual listener’s level of proficiency. But as I will demonstrate in the
following sections of this chapter, this is some way from the truth.
Defining intelligibility in interlanguage talk

The majority of the writings cited above relate either to interaction
involving fluent (often L l) speakers of English or to that between ‘native’
and ‘non-native’ speakers of English. In fact, one of the problems in any
discussion of listening in ILT is that the bulk of research on listening skills,
and particularly on the role of context, has been carried out wit-h L l
speakers. Few studies have been conducted from the perspective of L2
listeners, and even fewer with L2 listeners and speakers. Shifting from
communication between fluent speakers of English to that between non-
fluent speakers of English (i.e. NBESs), however, we are presented with a
very different set of circumstances. These exert a strong influence on
speaking and listening, particularly at the level of phonological form.
Communication between fluent speakers takes for granted a certain
amount of shared knowledge. According to Hymes, this knowledge can be
analysed in terms of four parameters:
1 Whether (and to what extent) something is formally possible.
2 Whether (and to what extent) something is feasible.
3 Whether (and to what extent) something is appropriate (adequate,
happy, successful) in relation to a context in which it is used and
evaluated.
4 Whether (and to what extent) something is in fact done, actually
perform ed, and what its doing entails.
(Hymes 1972: 12; emphasis in original)
The third and fourth parameters are of particular relevance to the current
discussion. As regards the third parameter or sociolinguistic competence,
successful communication involves a substantial degree of shared socio-
cultural knowledge. Either through the socialization processes of
childhood (in the case of most L l speakers)2 or through prolonged contact
with the target society (in the case of most L2 speakers in ‘Foreign
Language’ situations), fluent speakers of a language have developed
intuitive knowledge of what are and are not appropriate communicative
behaviours in specific speech contexts, and interpret what is said to them
accordingly.
Fluent speakers also have a high degree of linguistic knovvledge (in the
case of L l English speakers, often purely subconscious) not only in the
Chomskyan sense of grammatical competence—Hymes’s first parameter—
but also in terms of attestedness, his fourth parameter. Their formal

grammatical competence enables them to assess, for example, what can and
cannot be a word and what syntactic and phonotactic permutations are
possible. But they know far more than what is formally possible in an
abstract, or idealized, sense. They know what other fluent speakers actually
do in performance, and this knowledge involves a high degree of phonemic
and phonetic intuition. As regards the former, such speakers are intuitively
aware of the changes which often take place in connected speech (see page
72), and rarely—if ever—notice that words are not produced in full citation
form even if they do not always make the correct grammatical link (for
example, Avudsv/ is often written as ‘would of’ rather than ‘would have’).
For example, they have no problem in interpreting /weaja gooın/ as ‘Where
are you going?’. In terms of the latter (phonetic) intuition, they are able
both to perceive sounds categorially, i.e. to categorize a wide range of
allophones as ‘belonging’ to one particular phoneme, and to make use of
redundant phonetic cues. For example, they know instinctively that a /t/ or
/p/ sound followed by aspiration nevertheless ‘belong’ respectively to the /t/
and /p/ phoneme, and that they are very unlikely to be word-final. Again,
they know that the vowel sound /i/ is slightly longer before a voiced
consonant sound than before a voiceless one (for example, in ‘pig’ as
opposed to ‘pick’), and therefore which sort of sound they can expect to
follow N in either case. Knowledge of this sort, along with the
sociolinguistic competence mentioned in the previous paragraph, enables
speakers to process the ‘acoustic blur’ of which much speech consists.
In EFL situations—interaction between ‘native’ and (non-bilingual) ‘non-
native’ speakers of English—while the latters’ intuitions about the formers’
phonological and phonetic performance are likely to be undeveloped, the
sociolinguistic characteristics described above may stili obtain to some
extent. Although there is not the same degree of shared socio-cultural
background as exists botvveen L1 interlocutors, English socio-cultural
appropriacy is one of the subjects of learning and may even be its goal (for
example, in British Cultural Studies). And outside pedagogic situations,
‘non-natives’ have plenty of access to British, American, Australian, ete.
society and culture through the media. Some degree of shared socio-cultural
background between L1 and L2 speakers of English can therefore be
assumed.3 On the other hand, when used as an international language, that
is, between L2 speakers, English is not tied to any one cultural background4
and thus, unlike EFL, no single English culture can— or should— be the goal
of pedagogy. Culture learning is not part of EIL learning.
However, Davies questions whether ‘a natural language without some
cultural bias is possible’ (1991: 157). Given that, for international
purposes, a bias towards British, American, Australian, ete. culture is
irrelevant, the only other possibility for General English (as opposed to ESP)
classrooms, then, would be to introduce a wide range of L2 cultural
backgrounds. in this regard, Kramsch (1998) discusses the pedagogic use of

cross-cultural awareness-raising activities through which learners may
develop into ‘intercultural speakers’. Nevertheless, huge practical
implications surface when one considers the possibility of reorganizing L2
English classrooms around the world to enable learners to compare and
contrast their own L1 cultures with each other’s and with those of L2
English speakers not represented among their number. Although the
concept is attractive, it therefore remains at present largely an ideal rather
than a pedagogic reality. The consequence for ILT is highlighted in
multilingual classes, such as often feature in ENL (English as a Native
Language) countries. It is learners in such classes who provide the data
which will be presented later in this chapter. And, as we will see, even at
these learners’ relatively high level of English and after a period of exposure
to each other’s English, numerous miscommunications occur. Tarone asks:
In a (sic) ESL classroom composed of 20 learners from seven different
native-language backgrounds, can we really speak of speech
communities and shared qualitative norms in the way we can speak of
these for monolingual speakers? Can we expect learners from different
native-language backgrounds to share the same perceptions of the social
dimensions of English language use ...
(1 9 8 8 :1 1 8 )
My answer is no, we cannot. Each multilingual ELT classroom is a
microcosm of ILT as it occurs in the outside world. In each, we find people
thrown together often in pursuit of common goals but without a common
background to facilitate the achieving of these goals.
And herein lies the crux of the matter. To extend the paradigm of shared
background context to ILT would, in effect, be to deny the role of context
in ILT. Here, there is little if any shared background information available
to interlocutors, be it social, cultural, linguistic, or otherwise. The primary
contextual feature of ILT is, paradoxically, that there are no contextual
features— at least in the sense of shared socio-cultural knowledge. And as
Widdowson argues:
People who have particular knovdedge and experience in common,
whose contextual realities, so to speak, are closely congruent, will
manage to communicate by engaging relevant aspects of contexts with
only sparing use of the linguistic resources at their disposal. Conversely,
of course, those who have little in common have to place greater reliance
on the language.
(1990: 102)
In ILT, speakers ‘have little in common’ apart from their non-bilingual
proficiency in the L2 and the mutual desire to achieve a particular goal in
an interaction.
An illustration will help to demonstrate the point I am making about the

role of shared L l socio-cultural background. In a recent conversation with
my teenage daughter, she was describing the newly built facilities in her
school. She told me that there was a ‘really big new I.T. complex’ and then
added what I heard as ‘there’s a new drummer complex too’. I was initially
surprised (British secondary schools do not normally have such things and,
in any case, this is a girls’ school, where pupils tend to play flutes and
violins rather than drums), but then searched my mental lexicon for an
alternative to ‘drummer’ and soon located ‘drama’. Ali this occurred in a
matter of a couple of seconds, without the need for me even to articulate
the initial miscommunication. My daughter had (apparently) substituted
the short vowel sound /a/ for the long vowel sound /a:/, but my extensive
background knovvledge of both the place and the person (my daughter is
very interested in acting as opposed to, for example, sport), enabled me to
deal with this ‘high risk’ core phonological error (see Chapter 6) instantly.
On the other hand, a similar type of phonological error was not so easily
negotiated when the participants involved were an L l speaker and a (fairly
high level) NBES from Mexico. The interaction took place in a pub. The
male L l speaker (fluent in European Spanish) offered the female Mexican
NBES a drink. She asked for what sounded like /saıös/. This time the listener
was less able to rely on background knowledge to interpret meaning:
although he was familiar with the range of beverages available in English
pubs, he was uncertain of the alcoholic (or non-alcoholic) predilections of
Mexican drinkers. However, he at least had the fluent English listener’s
advantage of knowing that ‘sytha’ is not an English word, let alone the
name of a drink. Even so, and despite being well avvare that Spanish
speakers of ali persuasions regularly transfer the pronunciation of inter-
vocalic /d/ as /ö/ to their English, he had to ask for two repetitions before
finally understanding the speaker’s meaning as ‘cider’ (although the error in
fact remained). This time* then, the relative lack of shared background
meant that greater salience was attached to form and it was thus more
difficult for the listener to interpret meaning when faced with a (‘high risk’
core) phonological error.
However, both these typical examples contrast with ILT, where there is
neither shared socio-cultural knovvledge nor a high level of linguistic
knovvledge such as that of the L l listener in the previous interaction. Nor
is there likely to be much familiarity with an interlocutor’s accent of
English. In Chapter 3 we looked at the types of phonological errors made
by NBESs. Later in this chapter, we will consider a number of examples of
miscommunication drawn from ILT, and will note how often it is the fact
that these listeners are continually forced back on the language itself and,
in particular, on the pronunciation, which leads to the miscommunication.
We will see, furthermore, that miscommunication tends to remain as such, or
to involve more than a couple of simple repetitions to be successfully resolved.
Of course, it is not only in the lack of shared background knowledge

between interlocutors that ILT differs from interaction between fluent
speakers of English. I mentioned earlier the effects of co-articulation and
assimilatory processes on pronunciation and made the point that it is the
speed of speech which motivates these processes. Because NBESs speak far
more slowly than fluent speakers of English, they make little use of co-
articulation and connected speech processes. They acquire some of these
features automatically as their linguistic competence increases beyond that
of NBESs and their speech becomes faster. Other features—notoriously
weak forms— are never acquired for regular use by the majority of L2
speakers, even BESs (see Chapters 2 and 6). What this means then is that
listeners in ILT are receiving speech sounds that are much closer to their
citation form, and thus more amenable to bottom-up processing than- the
sound remnants in fluent English speech.
Another difference between fluent interaction and ILT is that NBESs seem
to have an instinctive reluctance to signal non-comprehension. In ‘NS-NNS’
interaction this reluctance is probably the result of the NBES listener’s desire
not to lose face. In ILT, listeners are concerned also about the face-loss of
their non fluent interlocutors, for the signalling of non-comprehension of
an NBES interlocutor’s speech may serve to draw attention to the latter’s
inaccuracies of phonological form, in other words, to their ‘poor’ English
accent. This was a response that my questionnaires elicited not infrequently,
and particularly from Far Eastern subjects. In fact, some of the latter were
not even prepared to commit comments about their interlocutors’
pronunciation to paper, but instead came to teli me privately.
At the level of phonological form, fluent speakers are generally able to
say what they mean with great regularity and automaticity, while when a
fluent speaker interacts with an NBES, only one of the two interlocutors is
not positioned to do so. In ILT, on the other hand, both interlocutors have
difflculties with phonological form which prevent them from saying what
they mean even though, like Alice, they probably mean what they are trying
to say. In addition, both interlocutors have receptive processing problems as
described above and in the next section, which phonological L1 transfers
(such as the example on page 76) can only exacerbate. A Central dilemma
for ILT, then, is that while in this speech context, probably more than in any
other, successful communication depends to some considerable extent on
the ability of speakers to compensate (particularly at the level of
phonological form) for listeners’ linguistic and extralinguistic inadequacies,
speakers are not well equipped to do so.
And so, finally, to my interpretation of the term ‘intelligibility’. The
reader will be aware from the previous discussion that I attach considerable
importance to the phonomenon of recognition of phonological form in ILT.
Like Smith and Nelson, I belıeve that it is important to standardize the use
of the term ‘intelligibility’, and I have no problem with their restriction of
78 The Pbonology o f English as an International Language
its use to word and utterance recognition. Unlike them, however, I do not
believe that ‘the most serious misunderstandings occur at the level of
comprehensibility and interpretability’ (1985: 335) where ILT is concerned.
This view is not by any means intended to imply that I consider
misunderstandings of a more pragmatic nature to be completely irrelevant
to ILT: ‘comprehensibility’ (recognition of word and utterance meaning)
and ‘interpretability’ (recognition of speaker intention) both have their
place. Rather, it is borne of the conviction, backed by hard evidence (see
examples of miscommunication in the ILT data provided later in this
chapter, pages 84-8), that in ILT such higher-level misunderstandings are
relatively rare because of the regularity with which phonological problems
‘get in the way’, mainly as a result of NBESs’ contextual processing
difficulties. Less able to draw on knowledge of the appropriate and the
attested in order to interpret messages, speakers engaged in ILT have to
focus their attention on meanings inscribed in form to a far greater extent
than do those familiar with the English language and its customary use.
In fact, there is evidence that phonological problems regularly get in the
way of successful communication not only in international contexts, but in
intranational ones also. For example, Bansal (1990) describes how English
is used within India as a lingua franca among speakers from different L1
Indian backgrounds. He argues that the differences in their varieties of
English are ‘much greater in respect of phonological and phonetic patterns’
than in lexis and grammar (ibid.: 219), and goes on to pinpoint a number
of phonetic features as İikely to affect the intelligibility of spoken English
within India’ (ibid.: 229), citing in particular the following: lack of clear
articulation, accent on the wrong syllable, and vowel or consonant
substitution. Similarly, in his discussion of the use of English as a common
language among speakers of the roughly 400 indigenous languages of
Nigeria, Ufomata relates mutual intelligibility specifically to accent and
argues that it is important to ‘study the varieties which have emerged in
second language situations’ in order to ‘identify areas which cause in
telligibility failures within these accents’ (1990: 216).
My own use of the term ‘intelligibility’ is thus, unashamedly, that of
Smith and Nelson (1985), but it is approached more in the spirit of writers
such as Bansal and Ufomata. It concerns the production and recognition of
the formal properties of words and utterances and, in particular, the ability
to produce and receive phonological form, but regards the latter as a
prerequisite (though not a guarantee) of ILT success at the locutionary and
illocutionary level.
This is not to say that the recognition of phonological form is a
straightforward, unilateral matter. It is a commonplace to talk of the
negotiation of meaning as a pragmatic process, but even with such an
apparently limited interpretation of intelligibility, we are stili concerned
with negotiation, with a two-way process involving both speaker and
listener at every stage of the interaction. The negotiation of intelligibility

consists of ‘the establishment and maintenance’ of the necessary conditions
to achieve understanding (Widdowson 1984: 115). In ILT the ‘necessary
conditions’ are, probably above ali, mutually intelligible pronunciation, for
even ‘high adequacy in lexis and grammar can be negated by incompetence
in the signalling phase’ (Gimson 1994: 270). Speakers assess moment by
moment the extent to which their phonological output appears to be
comprehensible to their interlocutor(s), and make adjustments and
corrections as they judge necessary, and insofar as they are able to.
Meanwhile, listeners are as responsible as speakers and, as Bamgbose
(1998) argues, part of a listener’s contribution to intelligibility is in the
making of allowances for an interlocutor’s accent, i.e. in converging
receptively (see Chapter 7 for a fuller description and discussion of the
accommodative process of convergence). Thus, even at the level of
pronunciation, intelligibility is dynamically negotiable between speaker and
listener, rather than statically inherent in a speaker’s linguistic forms, even
though participants find the process of negotiation more problematic than
do fluent speakers.
Lynch provides a number of good examples of this sort of negotiation in
progress. In the following extract, four learners are engaged in a jigsaw
speaking activity. In this type of task, a text is cut up into single sentences,
one sentence is given to each member of a group, they each memorize their
own sentence and hand it back, and the group then work out the original
order by a process of negotiation. In this interaction, A and E are from
Japan, H is from Spain, and J from the Faroe Islands.
E yes + the Queen + Queen Elizabeth
H Queen Elizabeth
E is the largest passenger + largest passenger
J the Queen is the largest passenger?
H passenger?
E the Queen Elizabeth second
A hm?
E sorry + I mistook + Queen Elizabeth the second
A hmhm
H it’s the second?
E is the largest passenger in the
H in the?
E in the world
A yeah in the
H in the?
J in the boat?
E in the world
H in the bar?
E world
J in the w-o-r-l-d (stretching the voıvel)

E in the world yes
H oh w-o-r-l-d
(Lynch 1996: 114)
The main obstacle to successful communication here seems to be E’s typical
Japanese rendering of the vowel sound İ3'J in ‘world’ as /a:/, as shown by
H’s interpretation of ‘world’ as ‘bar’; though the fact that H and J both
consider words beginning with /b/ as possibilities suggests a problem with
the /w/ also. On the other hand, as we would expect, following the
discussion of same-Ll dyads in Chapter 3, the Japanese interlocutor clearly
has no problems with this phonological transfer (‘yeah in the’). Thus the
negotiation in this interaction essentially entails the querying and correcting
of an item of L1 phonological transfer.
The point of ali this, then, has been to emphasize a number of crucial
distinctions between ILT and oral communication either between fluent
speakers of English or between a fluent speaker of English and an NBES.
For, in a number of important ways, the negotiation of intelligibility in ILT
cannot be considered typical of such negotiation in interaction where one
or both interlocutors are fluent speakers of English, not least when it comes
to the controversial issue of the relative roles of bottom-up and top-down
processing in NBESs’ speech perception, which we turn to next.
Bottom-up and top-down processing

A bottom-up model of speech perception assumes that ‘we perceive speech
by building up an interpretation in a series of separate stages, beginning
with the lowest level units (the phonemic segments of words, for example,
/b/, /d/, /g/) and gradually working up to the larger units such as the
utterance, from which we then derive our interpretation of the speaker’s
meaning’ (Anderson and Lynch 1988: 22). This contrasrs with top-down
processing ‘which uses knowledge and expectancies to guess, predict, or fiil
in the perceived event or message’ (Pinker 1994: 474).
Psycholinguists have tended, for the past decade or so, to favour more
interactive models of speech perception rather than purely top-down
models (see, for example, Anderson and Lynch 1988; Flowerdew 1996;
Lynch 1996 for further discussion). Nevertheless, it will be clear from the
previous section that I am suggesting that NBESs, even at relatively high
levels of competence, stili process speech using a predominance of bottom-
up strategies. As listeners they seem to find it difficult to make much use at
ali of the context underlying and surrounding the speech they receive, at
both linguistic and extralinguistic levels. This is not to claim that learners
cannot be taught to apply some (but not, in my view, ali) of the contextual
inferencing strategies they undoubtedly employ in their L l; but simply that
at present they rarely do so (but see, for example, Anderson and Lynch
1988; Field 1998 for a different opinion).
I have already spoken at length about the way in which NBESs’ mutual
lack of shared socio-cultural background forces them into a far greater
reliance on the acoustic signal— on what they actually hear—than is
generally the case between fluent speakers of the language. However, shared
socio-cultural background does not comprise the only type of
accompanying information available to interlocutors. Not only may
information be provided by features of the extralinguistic context, but also
by the co-text, and each of these types of information may or may not be
accessible to interlocutors.
Contextual and co-textual features comprise elements which are in some
way present in, respectively, the extralinguistic and linguistic speech event.
They are thus available to listeners to interact with the speech signal and
compensate for its limitations and deficiencies by confirming or denying
what is unconfidently heard, filling gaps, correcting mishearings and
productive errors, and so on. NBESs’ failure to make use of these features in
processing messages is far more of a threat to successful communication in
ILT than it is in ‘native/non-native’ interaction. This is because, as we have
already noted, in ILT the speech signal received by the listener is likely to
contain considerable L1 phonological transfer, whereas in ‘native-non-
native’ interaction, the speech received by the ‘non-native’ by definition
cannot transfer from an L l.
A couple of examples from my data will serve to illustrate this point (see
Chapter 3 for a full account of the methodology). The two instances of
miscommunication which follow occurred despite the fact that contextual
information was available to guide the listener to more likely meanings than
those of the acoustic signals that he had received.
In the first extract, a Japanese subject (B) was describing an alpine scene
to her Swiss-German interlocutor (A). He had the same set of six pictures,
although in a different order. His task was to identify in his set the picture
being described. He had problems in completing the task successfully
because the speaker had told him that in her picture there were ‘three [led]
cars’. This was confirmed by the following (tape-recorded) exchange which
took place between them immediately afterwards:
A I didn’t understand the let cars.
What do you mean with this?
B Let [let] cars? Three red [red] cars (very sloıvly).
A Ah, red.
B Red.
A Now I understand. I understood car to hire,
to let. Ah, red, yeah I see.
This breakdown in communication occurred even though only one of the
six pictures contained any cars, these cars were red, and there was no
evidence to suggest that they were for hire. As Tench, in his discussion of
my data, puts it, ‘The Japanese phonological fîlter had produced [ied] for
/red/ and the Swiss-German fîlter had perceived the [İed| as [let]’ (1997: 33).
Four weeks later, the same two subjects were engaged in a similar task.
Again, the Japanese speaker was describing a scene which her Swiss-
German interlocutor had to select from a set of six pictures.
B And second picture, the bottom of the bottom
of the picture there’s mm gley [gleı] house,
A (frowns)
B (registers As frown) grey fg.ıeı] and small house,
it’s very s-old?
A Yeah, there’s a grey house, yeah.
B Mm, okay.
So four weeks after the previous interaction, the Japanese subject was stili
having problems with /r/, albeit less frequently. She was more aware of the
problem and quicker to correct and usually to [j], rather than to her earlier
use of the flapped approximation [r]. In the follow-up discussion, her
interlocutor said that he, too, had become aware of her problem in this area
and always listened for it. Nevertheless— and despite the fact that one of his
pictures clearly contained a grey house— he trusted the acoustic
informarion. Influenced also by his Swiss-German filtef, ‘perceiving the [g]
as the Swiss-German unaspirated [k]’ (Tench 1997: 34), he said he had
initially understood his interlocutor to have said ‘clay’. And it was only
when his interlocutor corrected her phonological error that he understood
which picture she was describing.
It seems then that in ILT speakers have very limited access to context to
compensate for the inadequacies of speech perception and production. Nor
do they fare much better in relation to co-text. Compared with fluent
speakers, NBESs are ‘less sure of the forms of the language, the typical
syntactic structures, and the conventional vocabulary’ used in the
discussion of a particular topic (Brown 1990: 60). When something seems
not to make sense, they are unable to decide whether the speaker has used
a word with \vhich they are unfamiliat, whether they have misheard, or
vvhether the speaker has, in fact, made an error. This situation is especially
problematic \vhen the cause of non-sense is speaker error, and thus in ILT
for, as Olsson (1978) points out, a listener needs to know that he has heard
an error before he can begin to seek an alternative meaning. The failure of
NBESs to exploit conrextual and co-textual information and, hence, their
dependency on phonological form, may be either because it is unknown,
and so unavailable, or because it is known but inaccessible for some reason
(for example, the pressure of on-line processing).
What happens in ILT seems to be a circular process. The lack of shared
socio-cultural background between interlocutors, coupled with their
mutual lack of access to aspects of the linguistic context, throws them back
onto a focus on the acoustic signal. This in turn diverts cognitive resources
away from features of the context, which are thence not available to
compensate for any limitations in speech perception or production. And
because ILT is characterized by so much L1 phonological transfer, pro
nunciation is often a first-base obstacle to communication beyond which
the interaction is unable to proceed in a satisfactory manner. In this scheme
of things, there is little scope for the higher level pragmatic factors, which
are so often the culprits where communication fails between fluent
speakers, to enter the equation.
To observe how this happens in practice, we might consider the sentence:
‘It’s very hot in this room’. This may be a simple statement of fact; it could,
on the other hand, be an oblique request for someone to open a window. If
the latter message is intended but only the propositional content of the
sentence is perceived, the result is pragmatic failure. However, a French
NBES is likely to have a problem with the consonant sound fhJ and the
vowel sound /d/: he may omit the Ihl altogether and pronounce the /d/ as
/a/. Thus, the word ‘hot’ /hot/, pragmatically the most crucial one in the
utterance, may well sound more like ‘ut’ /At/. For the reasons already
discussed, an NBES receiver from another L1 will probably not be able to
make use of the available linguistic and extralinguistic cues to compensate
for this mispronunciation. He will therefore be unable to interpret even the
basic propositional meaning of the sentence, let alone its probable
illocutionary force. Instead, he will either let the breakdown in
communication pass vvithout comment, or will signal non-comprehension
and request clarification. If he opts for the latter course, there may well
follow an exchange involving several moves, during the course of which the
first speaker will reformulate and, -most likely, express his original request
more transparently. Whatever the outcome, the effect of the phonological
deviation is to pre-empt the possibility of the emergence of pragmatic
failure: unable to recognize the word ‘hot’, the listener is never in the
position of being able to ponder the speaker’s underlying intention.
It is inconceiveable that pronunciation is a majör cause of problematic
communication in interaction between fluent speakers of English and even,
perhaps, between ‘native’ and ‘non-native’ speakers. But in ILT, given
speakers’ frequent inability to ‘say what (they) mean’ pronunciation-wise,
which is compounded by listeners’ seemingly ubiquitous use of bottom-up
processing strategies, pronunciation is possibly the greatest single barrier to
successful communication. And this seems to be the case well beyond the
beginner and elementary learner levels mooted by most of those who write
on speech perception in L2 English (for example, Anderson and Lynch
1988; Rost 1990, 1996). As we will see in the next section, it is stili much
in evidence when learners are at upper-intermediate level and beyond.
The role of phonology in ILT: miscommunication in the ILT data

In order to demonstrate the extent of the phonological barrier to successful
communication in ILT and thus substantiate the claim made in the previous
paragraph, I will now describe and discuss ali the communication
breakdowns that occurred in one of my studies (see Chapter 3 for the
methodology). In a sense, these breakdowns demonstrate what happens
when the process of convergence, discussed briefly in the last chapter, does
not, for whatever reason, come into play.
Altogether there were forty occasions on which a receiver was unable to
understand the intended meaning of his or her interlocutor. Each one
occurred during interaction in the different-Ll dyads: there was not a single
example of such communication breakdown in the same-Ll interactions.
This does not necessarily mean that the latter interactions contained
absolutely no misunderstandings at the pragmatic level. The occasional
pragmatic failure may have occurred but passed unnoticed or without
comment by the interlocutors, both at the time and during the subsequent
playback of the recordings. However, because of the nature of the
interactions taking place, I can say with confidence that if there were any
such instances of pragmatic failure, they neither disturbed the flow of the
conversation nor affected the final outcome of the task.
I have divided the forty instances of communication breakdown, which
occurred in the different-Ll data in this study, into the following categories:
1 Pronunciation
2 Lexis
3 Grammar
4 World knowledge
5 Ambiguous/miscellaneous cause.
In the data and discussion which follow, the subjects and task types are
coded in the following manner:
J l, J2 Japanese subjects 1 and 2
SG1, SG2, SG3 Swiss-German subjects 1, 2, and 3
SF Swiss-French subject
(A) social conversation task
(B) information-exchange task
(C) joint problem-solving task
The numbers 1—4 before (A), (B), or (C) refer to the specific interaction (the
four separate interactions having taken place regularly över a 10-week
period). The item responsible for each instance of communication
breakdown is shown in bold, regardless of whether or not it was uttered
correctly (or, in other words, whether the problem was receptive only or, as
was more common, productive and receptive). Where the cause is
pronunciation, a phonetic transcript (within square brackets) is provided.
Following each token of breakdown cause, the communicative outcome is

indicated in parentheses.5
1 Pronunciation (Total: 27)

1 (İB) SG1: wood pronounced [wut] (J1 hears ‘water’.)
2 (İB) SF: toys as [taız] (SG3 does not understand until SF repeats
correctly.)
3 (İB) SG3: wool as [wsu] (SF does not understand at ali but says
nothing.)
4 (İA) J2: football match as ['(fmıtabol m;etj] (SG1 does not under
stand until J2 repeats correctly.)
5 (1A )J2: soccer club as [’sAka: kİAv] (again SG1 does' not
understand until J2 repeats correctly.)
6 (İA) SG2: hobbies as [’hDbıs] (J2 is initially not sure what SG2
means, and asks ['hovız] with rising intonation. He says
‘yeah’ at which point J2 understands and the interaction
continues.)
7 (1 B )J2 : wood chair [w] pronounced [©]. (SG2 understands
‘chair’ and ignores the word ‘wood’; see page 89 for
discussion of this example.)
8 (2A) J2: ‘You want work in England?’ as [vvaunt] (SG2 does not
understand until J2 repeats the question correctly.)
9 (2A) J2: ‘after I return to Japan’ as [d3ae'pâe] (SG2 asks ‘after
what?’—J2 repeats with same pronunciation but adds
‘from England’ and SG2 understands.)
10 (2B) J2: ‘three red cars’ as-[İed] (SG2 does not understand and
assumes she means ‘let cars’, that is to say, ‘cars to let’.)
11 (2B) SG2: ‘table is surrounded by chairs’ as [tjeas] (Assumes J2 ’s
problem is ‘surrounded’ and rephrases, but in fact she is
not sure whether he means ‘chairs’ or ‘chess’.)
12 (2B) SG1: balcony as f'baelksn] (J1 does not understand but says
nothing.)
13 (2B) SG1: covered as ['koow3d] f'ksuvad] ['kauvat] and ['kovad]
(J1 eventually guesses from her picture and S G l’s three
repetitions of the word ‘snow’.)
14 (2B) J l : cushions as [’kujo] (SG1 does not understand until J1
repeats correctly.)
15 (2B) J l : scissors as ['sızaız] (SG1 does not understand until J l
repeats correctly.)
16 (2C) J2: ‘it looks more sad, much sadder’ as [sset] and ['saeta]
(SG2 does not understand until J2 repeats correctly.)
17 (3A) J l : ‘Japanese animation fîlms’ with final syllable as [J5] (SG1
understands only Japanese films and does not pursue
the word ‘animation’.)

18 (3A) J l : mail as [ma:l] (SG1 does not understand but says
no thing.)
19 (3A) J2: ‘are they English?’ as ['ıggnf] (SG2 asks ‘Are?’ but then
grasps it before J2 repeats.)
20 (3B) SG3: hat as [hot] (SF only understands after SG3 reformulates
his sentence, though he does not actually improve his
pronunciation of ‘hat’.)
21 (3B) J l : zipper as ['d3İ:pa:] (SG1 does not understand at ali but
says nothing.)
22 (3B) SG1: hat as [hüt] (J2 assumes he said ‘hot’, infers ‘warm
clothes’, and draws a sweater. She is not aware that he
has mentioned a ‘hat’.)
23 (3C) J2: T d take the bird’ as [ba:d] (SG2 asks ‘You’ll take?’ and
understands when she repeats correctly.)
24 (4B )J1: man as [marn] twice (SG1 does not understand until J l
repeats it a third time correctly.)
25 (4B) J2: ‘grey house’ as [gleı] (SG2 initially looks for a ‘clay
house’. J2 registers his frown, corrects her pronunci
ation almost immediately and SG2 understands.)
26 (4B) J2 curtain as ['karıtan] (SG2 does not know at first, but
guesses from the word ‘blind’, although he later claims
not to have been very confident about his guess.)
27 (4C) SG1: ‘The weather is even worse [v3ns] there than here’ (Jl
does not understand until SG1 repeats with correct pro
nunciation.)
2 Lexis (Total: 8)
28 (İB) J l : mantelpitfce (SG1 does not know this word.)
29 (İB) J l : vase (SG1 does not know this word.)
30 (İC) J2: couch potato (SG2 does not know the term—J2
explains.)
31 (2B) SG3: booklets instead of ‘magazines’ (SF does not understand
what he means.)
32 (2B) SF: ashtray (SG3 knows the word but could not place it at
that moment.)
33 (3B) J l : chest long to describe hair length (SG1 understands
when she rephrases as ‘up to her chest’.)
34 (4B) SG1: substitutes the word fumiture for ‘kitchen units’ (Jl
does not understand.)
35 (4B) SG1: substitutes the word plate for ‘kitchen surface’ (Again J l
does not understand.)
3 Gram m ar (Total: 1)
36 (2C) SG1: ‘to use children as the cover is not the baddest idea’ (J1
does not understand ‘baddest’.)
4 World knoıvledge (Total: 1)
37 (2A )J1: flower arranging (SG1 knows the words but not that the
course is available in the school and assumes J1 has
made a mistake.)
5 Ambiguous/miscellaneous cause (Total: 3)

38 (İA) SF: ‘What is your work?’ spoken very fast. (SG3 does not
understand until repeated more slowly.)
39 (İB) SG1: öven pronounced [sofan] then [auvan]; wrong word—he
means ‘fireplace’ (J1 does not understand at ali.)
40 (2A) SF: ‘You need English a lot?’ quietly and with falling
intonation although intended as a question. (SG3 under-
stands when repeated correctly.)
The above comprise ali the examples of communication breakdown (non-
understanding or, less often, misunderstanding) in the different-Ll
interactions that were either apparent when they occurred, or were not
apparent at the time, but were acknowledged by the subjects during the
follow-up interviews. Although there may have been other instances of non-
comprehension, this was unlikely as the pairs of subjects and I listened in
detail to the tapes together, pausing frequently to check whether what had
been meant was what had actually been inferred. And although learners are
often reticent about signalling non-comprehension (see page 77 above),
these subjects appeared comfortable with the idea of doing so in the fo!low-
up stage, if not always during the task itself. This was, I believe, because of
the non-threatening manner in which the follow-up discussion was
conducted, the goal being not to criticize either speaker or listener, but to
help them to identify together the ways in which they could mutually
improve their performance for the forthcoming CAE Speaking examination.
The most noteworthy feature of the list is obviously the high pröportion
of instances of communication breakdown caused by pronunciation errors.
Of the 40 samples, one is the result of speed of delivery; another is the result
of the lack of vrorld knowledge; one is the result of grammatical error (two
if the ‘ambiguous cause’ number 40 is included); eight relate to lexis (nine
if number 39, the ambiguous ‘öven’, is included), either the speaker’s misuse
or the receiver’s lack of lexical knovvledge). But a remarkable 27 of the
breakdowns were the result of pronunciation (29 if we include the two
‘ambiguous causes’ 39 and 40, which involve pronunciation), and ali 27
were caused by the transfer of L l sounds. Thus L l sound transfer is

responsible for more than twice the other causes added together. This
finding is well supported by convictions expressed not only by the six
subjects themselves, but also by many other EFL students who have
experience of ILT, with whom I have spoken. By contrast, it does not fit in
comfortably with current experts’ claims'for suprasegmental phonology as
the prime cause, pronunciation-wise, of learners’ lack of intelligibility (see
Chapters 2 and 6, and Jenkins 1996a). I suspect, however, that the latter
view reflects the fact that many of these people work in EFL and (American)
‘ESL’ rather than ILT situations, and towards a pedagogic goal of
intelligibility for listeners who are ‘native speakers’ of English.
These examples also provide evidence of the minör role of grammar
errors in ILT miscommunication. For, despite countless such errors in the
data, only once did a problem occur (or twice if the lack of inversion in the
final ‘ambiguous cause’ compounded the faulty intonation pattern). The
fact that learners seem to grasp what is intended when their ILT peers make
errors in grammar is probably at least partly a function of the
developmental factor in the acquisition of L2 grammar (see Chapter 5 and
Ellis 1994, Chapter 3). Although learners from different L ls acquire their
L2 English grammar at different rates, taking differential periods of time to
pass through the various developmental stages according to their L l
grammar, they nevertheless pass broadly through the same stages. The
likelihood then is that learners will be able to interpret the grammar errors
of their peers even when their own interlanguage has moved beyond the
developmental phase of which those errors are characteristic. This is
paralleled in L l use, where adults rarely have problems in understanding
the deviant grammar of young children (*M ummy gone and *H e’s tooken
it being typical English examples).6
In fact, it seems that in L l English interaction, syntactic effects have a
very limited role by contrast with lexical effects in the decoding of words.
Rost (1990) notes that in experiments on lexical effects, subjects ‘tended to
identify sounds as English words as long as there was some phonological
evidence that a word could be identified’. On the other hand, ‘data on
syntactic processing suggests that syntactic context plays a very limited role
in constraining word recognition’ in English which, Rost points out, is not
surprising, ‘since it is only rarely possible to predict with certainty the form
class of any given lexical item because each syntactic constituent in English
has optional members’ (ibid.: 4 8 -5 0 ; emphasis in original). In other words,
the syntactic context in English does not provide many cues to facilitate the
prediction of what is to come or the decoding of what has gone before
—what Rost refers to as ‘look-ahead and look-behind assessments of
language as it is heard’ (ibid.). ILT thus has much in common with fluent/Ll
interaction in terms of the role of syntactic (as opposed to phonological and
lexical) context in intelligibility in speech perception.
Turning back to the ILT data above, one other area calls for particular
comment. This is the large number of L1 pronunciation transfers which are
processed bottom-up and result in some sort of communication breakdown,
when top-down listening strategies would probably have avoided this
outcome. In many of the breakdowns where the source can be traced back
to the speaker’s pronunciation, there was extensive extralinguistic
contextual information available. In examples 10, 12, 16, 23, 25, and 27,
the listener had a picture showing this information visually (and remember
that in the case of task C, the two interlocutors were looking at the same
picture). In examples 2, 11, 14, 24, and 26, the listener did not know for
certain they had the item in question in their picture, but they knew at least
(because they had been told at the beginning of the interaction) that their
task B picture showed a very similar physical setting to that of their partner,
with any differences being in the fine detail. Yet in example 11, for instance,
the listener opts for ‘chess’ rather than ‘chairs’ (‘Table is surrounded by
chess’) despite the fact that the former would be extremely unlikely in either
the linguistic or extralinguistic context. In examples 20, 21, and 22, the
listeners were attempting to draw a person from the speakers’ descriptions.
They could therefore be expected to have established a schema which
included words such as ‘hat’ and ‘zipper’, especially considering that in
example 20, the speaker said ‘She wears a hat on her h e a d ’, and the ‘zipper’
in example 21 is mentioned in the context of a raincoat. In examples 4 and
5, although there was no visual information, the subject of football was
already under discussion and, once again, one could have expected a
football match schema to have compensated for pronunciation problems.
Notwithstanding what has just been said, there are signs in the data of
occasional attempts to use top-dowjı, context-based strategies when
difficulties arise. However, not only are these strategies very rare, but the
subjects— despite being of upper-intermediate to low-advanced proficiency
levels— do not seem to use them with any great degree of confidence. In
fact, only in three of these 27 instances did a listener appear to be making
much use of top-down strategies. In example 7, the Japanese speaker
described a wooden chair as a ‘wood chair’, pronouncing ‘wood’ with [oa]
rather than [w], Even though the deviation is only a slight one, the Swiss-
German listener did not understand the word at ali. However, after two
(stili inaccurate) repetitions, he decided to ignore it and, latching on instead
to the word ‘chair’, managed to identify the picture being described. In
example 13, the Swiss-German speaker repeated the word ‘covered’ four
times, each time with a further small improvement in the pronunciation,
but never pronouncing it in such a way as to be identifiable to his Japanese
interlocutor. In the follow-up discussion, she admitted that she had
eventually realized what he meant because she had guessed from her own
set of pictures and several repetitions of the word ‘snow’. Again, in example
26, the Swiss-German listener realized that a ‘carton’ was unlikely in the
context of a window and used this knowledge along with the linguistic
information (the word ‘blind’ co-occurred) to guess ‘curtain’. Nevertheless,
after the Japanese subject had finished her description, he asked: ‘I’m not
sure if I understood it the right way the the window, there’s no there are no
curtains?’, and admitted in the later discussion to having been unconfident
about his guess.
We noted in the previous chapter that in ILT speakers tend to minimize
the use of L1 phonological transfer when they are able to do so and when
intelligibility for an interlocutor is particularly salient. However, there
appears to be something of a tension between the desire to minimize
difficulty for the listener and the desire to minimize difficulty for self. The
avoiding of phonological transfer by summoning up and ‘adding’ the target
form (see Chapter 8 on the concept of ‘accent addition’), demands a
considerable degree of cognitive effort on the part of the speaker, so it
stands to reason that they will not attempt to do so if they do not consider
it to be absolutely essential. This is where context comes into play. When
deciding how much cognitive effort to invest in replacing transfer with a
more target-like pronunciation, speakers may (either consciously or
subconsciously) weigh up the risks involved for their NBES interlocutors. If
clear contextual cues are available, they may assume that it is safe to relax
their ‘controls’ on pronunciation, and thus those sounds which have not yet
become automatically target-like will emerge with L1 transfer. In other
words, like fluent speakers of English in their use of assimilatory features
(see above pages 72-3), they will opt for the easiest route to speech
production if they think they can ‘get away with it’. One explanation for
the phonological transfers in the ILT data may therefore relate to
speaker/listener differences in orientation to context.
In L2, this preference for ease of articulation is particularly true of
sounds in certain phonological environments as compared with others.
Some L2 phonological pro^esses and sound combinations are clearly more
difficult than others for L2 speakers of English, depending on the
interaction of their L1 phonology with phonological and physiological
universals (see Chapter 5). Where such ‘difficult’ processes and sounds are
involved, and in the presence of contextual information, speakers seem less
likely to make the necessary cognitive effort to replace L1 transfer. A
tentative example of this phenomenon is number 25 above (also discussed
on page 82), involving a Japanese speaker’s substitution of /r/ with İV. The
former sound is more marked than the latter, in that if a language contains
İri it will also contain IV, but not vice versa. In addition, although about 75
per cent of languages contain some type of /r/ phoneme, it is most
commonly a variety of trill rather than the (British) English approximant [i]
(Ladefoged and Maddieson 1996), a sound which L1 (British) English
children have difficulty in acquiring. The implication is, then, that it is
universally easier to articulate N than (the (British) English form of) k İ, and
particularly for speakers such as the Japanese, whose first language has only
a single flapped sound somewhere between the two English sounds.
Clusters involving /r/ are even more difficult for them. In fact, clusters with
/r/ are difficult for ali speakers (Maçken and Ferguson 1981)—even L1
speakers of English have articulatory problems with Itri and /dr/, which
they tend to pronounce respectively as /tjr/ and /d3r/.7
The interplay between context and phonological environment may thus
account for forms such as ‘gley’ for ‘grey’ in example 25. The /gl/
combination is easier to produce than /gr/, particularly for those such as
Japanese speakers of English. So on this occasion, because of the presence
of the visual cue, the speaker at first does not make the necessary effort to
select (a cognitive process) and produce (an articulatory process) a [a].
However, the fact that the context does not clarify her meaning for her
interlocutor suggests that speakers and listeners have different agendas in
relation to context and, perhaps oddly, that in the role of speakers they are
not able to keep in mind the difficulties that they have as listeners in dealing
with contextual information.
Even for fluent listeners, language is a guide to context which, itself,
cannot be activated until the utterance has been heard. Listeners have to
select from an ‘explosion of potentially relevant information’ (Brown et al.
1994: 35), whereas a speaker, in referring to features of the context, has by
definition already selected these features. This means, as Brown points out,
that from a listener’s perspective, ‘context alone may not illuminate
language use unless language is first deliberately used to guide listeners to
identifying those features of context which will be relevant to the
interpretation of language’ (1989: 97). But in ILT, the potential for this to
happen is heavily constrained. For here, we are presented with a unique
situation: a combination of the inability of language to ‘guide’ reliably
because of speakers’ phonological difficulties and the inability of listeners
to be guided because of their top-down processing difficulties. The pity is
that, because of the sheer amount of phonological transfer in the speech of
NBESs, an ability to make use of contextual information would be doubly
helpful to NBES listeners. In my view, hovvever, this is not a likely outcome
for the majority, and we should focus the EIL pedagogic effort above ali on
enabling learners, in the words of the Mad Hatter, to ‘say what (they)
mean’. This will best be done, on the one hand, by establishing a core of
intelligible pronunciation which can be made available to ali learners
regardless of their L1 and, on the other, by developing learners’
accommodative skills so that they are able to improve their intelligibility for
specific listeners. These are the subjects respectively of Chapters 6 and 7,
where we move on to consider pedagogic priorities.
intelligibility and the spread of English

No discussion of intelligibility in ILT would be complete without some
comment on the concerns which continue to be voiced över the possibility
that English will ultimately fragment, like Latin, into a number of mutually
unintelligible languages (see, for example, Maley 1985; Quirk 1985;
Crystal 1997; and McArthur 1998 for discussion of this issue). Crystal
points out that ‘prophets have been predicting such doom for some time’.
He cites Henry Sweet’s inaccurate prophecy of 1877 that ‘England,
America, and Australia will be speaking mutually unintelligible languages,
owing to their independent changes of pronunciation’, and a similar one,
made stili one century earlier, by Noah Webster with regard to British and
American English (1997: 134). Although both Crystal and McArthur stress
that ‘this whole topic is so recent that it is difficult to make predictions with
much confîdence’ (Crystal ibid.: 135), they remain positive: ‘despite ali the
diversity and because of ali the recent technological innovations the centre
will after ali probably hold’ (McArthur 1998: 183).
McArthur, in fact, takes the argument one stage further in his conclusion
that the issue can be viewed from three perspectives: the ‘pessimistic’, the
‘optimistic’, and the ‘neutral’/‘pragmatic’. The pessimists, i.e. those people
for whom Latin ‘offers a suitably dire analogue of mutually unintelligible
‘post-English’ languages’, he argues, ‘are usually conservatives in the
Wyldian tradition,8 asserting a ‘Received Standard’ to which ali should
adhere or aspire, and which is for them the true English language’ (ibid.:
181-2). This view, however, is not as thinly spread as McArthur implies.
Not only is it relatively common among the ‘NS’ lay population (see
Chapter 2), but it is also prevalent among teachers in ELT schools,
particularly (though not exclusively) ‘NS’ teachers in the UK and USA, and
is implicit in the bulk of teaching materials produced by publishers in the
‘NS centre’. And such teachers and publishers are not generally people vvho
consider themselves to be ‘conservative’ in the Wyldian or any other
tradition. Widdowson, taking a rather more pragmatic (and realistic)
position, argues that standards of intelligibility in EIL are sustainable, but
in relation to specific purposes rather than to some vague global
intelligibility: ‘English as an international language is English for Specific
Purposes. Otherwise it would not have spread, otherwise it would not
regulate itself as an effective means of global communication. And
otherwise there would, for most people, be little point in learning it at
school or university’ (1997: 144).
Any reference to unintelligibility of course begs the question
‘unintelligible to whom?’ Evidently the many L l speakers of English,
teachers or otherwise, vvho stili regard their role as that of privileged ‘native
speakers’ of the language, mean unintelligible to themselves. In this
connection, we notice that even in the most recent communicative speaking
examinations such as the CAE—which stipulates the examining of
candidates in pairs and minimal input from the two (usually ‘NS’)
examiners—judgements of intelligibility are based on whether these two
examiners understand the candidates, rather than on whether the latter
understand one another. And these judgements are themselves influenced
above ali by the perceived closeness or distance of candidates’ L2 accents
from those of the examiners. (See Chapter 8 for a discussion of the testing
of pronunciation in communicative exams.)
The scenario described in the previous paragraph is unlikely to change,
however, while most of the pronunciation teaching which takes place
around the world remains heavily-biased towards ‘NS’ accents. Willis
(1996), though referring to grammar, describes the outcome of this sort of
bias as ‘conformity’ rather than ‘accuracy’: conformity to ‘NS’ norms rather
than accuracy in the establishing of a relationship between form, and
meaning in the context of the particular interaction. This implies a focus on
acceptability for L1 speakers of English rather than on intelligibility for
NBESs.
In terms of pronunciation, however, interlocutors engaged in ILT, of
necessity, resort to accurate perception of phonological form because they
are unable to access other sources of information in the level-switching
process (from sound to word to co-text to context) which characterizes
fluency. Accuracy is thus critical in Brumfit’s (1984) sense (i.e. as opposed
to fluency). Nevertheless, phonological accuracy in EIL is quite a different
matter from conformity to the norms of L1 speakers. The phonological core
(see Chapter 6) both includes features that would not be considered
acceptable by the latter and omits a number that they would consider
crucial. In this regard, then, I find Willis’s accuracy/conformity distinction
a very useful one, and will take it up again in Chapter 6. Pushing learners
to acquire authentic ‘NS’ norms of course is, and always has been, an
oxymoron in both senses. L2 speakers of English, however high their level
of proficiency, by definition never engage in authentic ‘NS-NS’ interaction.
And even in ‘NS-NNS’ (EFL) conversation, the ‘NNS’ interlocutor is not
receiving authentic ‘NS’ discourse from his addressee because, as Riley
points out, ‘exolinguistic discourse is never a matter of one Native Speaker
discourse plus one Non-Native Speaker discourse, with one speaker
behaving ‘normally’, the other speaking ‘like a foreigner’. Observation has
shown ... that both participants modify their discourse and that this is a
discursive situation sui generis’ (1989: 122).
As far as EIL is concerned, however, we are interested not in intelligibility
for ‘native-speaker’ receivers but for participants in interlanguage talk, i.e.
NBESs. And here, the prospect of mutual unintelligibility does need to be
taken seriously. Above I quoted Crystal’s comment about the recency of EIL
and the consequent difficulty of making sound predictions about mutual
intelligibility. The general consensus, nevertheless, appears to be that
mutual intelligibility in EIL will probably be safeguarded by virtue of
international technology and telecommunications. However, it seems to me

that this view reflects something of an inner circle perspective. While I agree
in general terms that English is unlikely to fragment into the modern-day
equivalents of Italian, Spanish, French, and so on, I do not believe that this
outcome will happen of itself as if by some sort of linguistic magic. Because
EIL is so very different from both EFL and ESL, people will need to work at
it; and they will need pedagogic support to facilitate their efforts.
The starting point of EIL is, of course, that it involves L2 English
speakers from myriad first languages. If we set aside the minority of fluent
speakers of English (whether BES or MES) who engage in EIL, English is
always— and often to some considerable degree— characterized by
phonological transfer from its speakers’ first languages. Indeed,
pronunciation, as we have already noted, is the linguistic level which most
distinguishes one (educated) variety of English from another, whereas the
different L1 groups, even at NBES level, maintain far more common ground
in their grammar and lexis, particularly in written English. We saw earlier
in this chapter the difficulties that result from L1 pronunciation transfer,
not only because of the transfer problem per se, but also on account of
listener factors relating exclusively to interlanguage talk. EIL is therefore at
far greater risk of succumbing to mutual (phonological) unintelligibility
than has ever been the case for EFL or ESL.
Clearly, we need far more EIL-based information than is currently
available as to whose accents are (un)intelligible to whom and, more
specifically, as to precisely which pronunciation features lead to loss of
intelligibility. Most of the small amount of research on the intelligibility of
different L2 accents of English has approached the issue mainly or entirely
from the perspective of ‘NS’ listeners (for example, Nelson 1982 and Bansal
1990 on the intelligibility of Indian English; Lanham 1990 on the
intelligibility of South African Black English; Ufomata 1990 and Tiffen
1992 on the intelligibility of Nigerian English).
Noteworthy exceptions which consider intelligibility from the point of
view of both ‘NNS’ and ‘NS’ listeners are Lane (1963); Smith and Rafiqzad
(1979); Smith and Bisazza (1982); and Smith (1992).9 Lane (1963) looked
at the effects of noise on the intelligibility of foreign accent for both ‘NS’
and ‘NNS’ listeners. He found, predictably, that ‘NNS’ listeners suffered to
a greater extent than ‘NSs’ (63 per cent loss as compared with 55 per cent).
The next two studies compared the intelligibility of different accents of
English for ‘NS’ and ‘NNS’ listeners. Smith and Rafiqzad (1979) found that
the intelligibility of the different accents was consistent for ali listeners; that
only two of the eleven groups of listeners rated their own countrymen as the
most intelligible; and that the ‘NS’ was always one of the least intelligible.
Smith and Bisazza (1982) found that the intelligibility of an accent was
strongly linked to the listeners’ familiarity with that accent, i.e. to the
amount of prior exposure they had had to it. Smith (1992) had similar
concerns to those of the previous two studies. He came to similar

conclusions in relation to ‘native speakers’ of English (they appeared to be
neither the most easily understood nor the best able to understand different
varieties of the language) and the importance of familiarity with a range of
English varieties across cultures. To my knowledge, no other research has
considered phonological intelligibility within a ‘non-native/non-native’
framework until my own work in the 1990s, which focuses on this
interaction type exclusively, and which attempts to identify both which
phonological features impede intelligibility and which are of no
consequence in ILT.10
We noted in Chapter 1 that RP is not necessarily the most easily
understood accent of English, even for L1 speakers who speak other
varieties of English. The studies of Smith, Smith, and Rafîqzad, and Smith
and Bisazza (and see also Smith and Nelson 1985) provide further evidence
that a ‘native speaker’ of a Standard variety of English is not necessarily
either the most intelligible or the best judge of intelligibility in EIL. Their
evidence, along with the distinction between accuracy and conformity
mentioned above, and the findings of my own ILT research, indicates a
pressing need for us to redefine phonological error and correctness within
the context of EIL rather than within the framework of a Standard language
ideology. In the process of doing so, we will be able to redress the balance
between intelligibility and acceptability by giving far more weight to
intelligibility for NBESs and far less to acceptability and appropriacy for L1
speakers. This, in turn, will ensure that judgements of error and correctness
are acceptable and appropriate to EIL.
Conclusion
In terms of pronunciation, what the discussion in this chapter indicates is
the need for some sort of international core for phonological intelligibility:
a set of unifying features which, at the very least, has the potential to
guarantee that pronunciation will not impede successful communication in
EIL settings. This core will be contrived to the extent that its features are
not identical with those of any one L1 or L2 variety of English. As we will
see in Chapter 6, a phonological core of this kind already exists among ali
L1 speakers of English, whatever their variety. A core of sorts also exists
among L2 speakers, insofar as speakers of ali languages share certain
phonological features and processes. However, this shared element is
limited. Indeed, were it not so, there would be no reason for me to write
this book: L2 speakers of English from different parts of the world would
understand each other’s pronunciation with relatively little difficulty. Thus,
while we can build on what L2 speakers already have in common
phonologically, we must take the argument one very large step further by
identifying what they need to have in common and contriving a pedagogic
core that focuses on this need. However, such a core, while necessary, will
not alone be sufficient to achieve the goal of preventing pronunciation from
impeding communication. Bamgbose makes the obvious yet frequently
missed point that ‘it is people, not language codes, that understand one
another’ (1998: 11). Participants in EIL will also need to be able tune into
each other’s accents and adjust both their own phonological output and
their receptive expectations accordingly.
In Chapters 6 and 7 we will look more closely at these two approaches
to EIL communication and consider their pedagogic implications. Chapter
6 is both a discussion of the complex issues involved in the establishing of
a core of phonological intelligibility for EIL, and a presentation of the core
which I am proposing. Then, in Chapter 7, we move on to a consideration
of how best to both promote learners’ productive and receptive use of this
core, and to encourage the development of speaker/listener accommodative
processes which will facilitate mutual intelligibility in EIL. But first, in the
following chapter, we will consider in detail the relationship between L l
phonological transfer and EIL intelligibility.
Notes
1 Lecture at International House London, 4 December 1992 and again at
the 33rd TESOL Convention, New York, 11 March 1999; and see
Chapter 6 page 149 for an example of this phenomenon.
2 This also applies to those in intranational Second Language situations,
most notably—though not exclusively— ESL.
3 In addition, the ‘non-native’ in such speech situations may be addressed
in foreigner talk. In particular, the ‘native’s’ speech will be more slowly
and clearly articulated than normal, to compensate for any gaps in
shared background and linguistic knowledge, which would undermine
the ‘non-native’s’ aMlity to process the ‘acoustic blur’ (see also
Chapter 7).
4 Though EIL is associated with cultural background to the extent that
interlocutors share cultures of specialized knowledge and expertise
within specific ESP genres, and these can— and should— be dealt with in
ESP classrooms.
5 Some of the non- and misunderstandings did not become evident until
the recording of an interaction was played back to the subjects.
6 Asterisks are used to indicate that a grammatically incorrect item is non-
attested.
7 Though I once had a colleague who regularly pronounced the word
‘children’ with an epenthetic (added) schwa sound betvveen the /d/ and
the İri.
8 This refers to Henry Cecil Wylde, the originator of the early 20th
Century term ‘Received Standard English’ (the Standard language
spoken with a public school accent). Wylde described this accent as
‘neither provincial nor vulgar, a type wfaiçh most people would willingly
speak if they could, and desire to speak if they do not’ (1934, quoted
by McArthur 1998: 125).
9 A majör difference between these studies and my own, however, is that
the listener subjects in the former were in the position of
‘eavesdroppers’ whereas those in my studies were also participants in
the discourse.
10 Although not being concerned specifically with accent, Lynch 1996
discusses learner-learner interaction and provides some illuminating
e\amples of phonological problems and their negotiation (see, for
instance, the example taken from Lynch’s data on pages 79-80).

Jenkins 1

Uploaded by

Document Informationclick to expand document information

Document Informationclick to expand document information

Copyright:

Available Formats

Jenkins 1

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Jenkins 1

Uploaded by

Copyright:

Available Formats

pfcll 'l l s

Great Clarendon Street, Oxford 0x2 6 d p

o x f o r d and o x f o r d e n g l i s h are registered trade marks o f

© Oxford University Press 2000

1 The background: Changing patterns in the use of English

2 The variation problem 1: Inter-speaker variation

3 The variation problem 2: intra-speaker variation

4 intelligibility in interlanguage talk

5 The role of transfer in determining the phonological core

6 Pedagogic priorities 1: Identifying the phonological core

7 Pedagogic priorities 2: Negotiating intelligibility in the ELT 165

8 Proposals for pronunciation teaching for EIL 195

Aftenvord: The future of the phonology of EIL 233

Oxford University Press for extracts from ‘Sociolinguistics and Second

The purpose of this book is to explore the phonology of English from an

but also for its as yet unrecognized contribution to L2 phonological

English is England’s language but the world’s treasure.

The historical shift

referred to the works of Phillipson (1992), Pennycook (1994), and Crystal

Changing ownership; changing terminology

consequently, who is entitled to prescribe the standards against which use

third letters to arrive at ELF, or ‘English as a Lingua Franca’. This term

Appropriate pedagogy for an international language

others, are influencing a change in pedagogical attitudes, such that learners

The EIL phonological problem: where do we go next?

A further possibiiity involves distinguishing betvveen a ‘model’ and a

interaction between NBESs has a number of characteristics that

the greatest factor in unintelligibility, and the difficulty tended to increase

Variety’s the very spice of life,

Contrary to the view expressed by Covvper, the terms ‘variability’,

strongly held and deeply entrenched convictions as to the superiority of

likelihood—retaining numerous phonological and phonetic features of the

ESL varieties and the interlanguage controversy

native varieties of English’, claiming that these varieties ‘are inherently

Phonological inter-speaker variation and its effects on

inter-speaker segmental variation and its effects

continues until and unless sufficient practice of L2 sounds leads to the

precede perception. We cannot assume that because NBESs are able to

The variation problem 1: Inter-speaker variation 35

typically pronounces the word ‘luggage’ as [İA g ıd ji]. Deviations involving

Sound substitution and conflation

referred to cartheft as /kar tepsta/. His Taivvanese ınteriocutor was

Syllable simplification: consonant deletion and sound addition

of a clear, familiar topic, the majority of deletions in the data resulted in

The view that ‘Most segmental errors, though noticeable, do not

Inter-speaker suprasegmental variation and its effects

W ord stress in ILT

Some ILT w ord stress data

Generally more serious, however, were word stress deviations vvhich

Rhythm and intonation

However, it is nuclear placement itself which causes the most serious

drinks a lot of beer’, I have established a contrast between my husband and

though not necessarily for ‘native speaker’/‘non-native speaker’ inter­

Some IL T intonation data

The first examples occurred during a (multilingual) classroom lesson at

Attempts to use contrastive stress, though encouraging as a sign that the

substantiate this claim by outlining an experiment I carried out, which

3 The variation problem 2:

though not necessarily for ‘native speaker’/‘non-native speaker’ inter