See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/244511317

The Phonology of English as an International Language

Book · May 2000

CITATIONS READS

1,565 22,757

1 author:

Jennifer Jenkins
University of Southampton
88 PUBLICATIONS   9,429 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

ELF and dis/empowerment View project

ELF and dis/empowerment View project

All content following this page was uploaded by Jennifer Jenkins on 26 March 2018.

The user has requested enhancement of the downloaded file.

 63

7 Pedagogic priorities 2:
Negotiating intelligibility in the
ELT classroom

Certainly, it is important for learners..... above all to have

understood the implications of the well-established sociolinguistic
maxim that linguistic consistency is not normal and indeed would
be dysfunctional in a language used in a real speech community.
(Lesley Milroy 1994, p.25)

In Chapter 4, I argued that listeners engaging in interlanguage talk are generally reluctant
or even unable to compensate for one another's phonological shortcomings by utilising
contextual and cotextual cues to interpret meaning and, consequently, that phonological
intelligibility is essential to successful ILT communication. In Chapter 5 we examined
the influences exerted by the L1 on L2 pronunciation and, by extension, on the design of
any phonological core. Then, in the previous chapter, I presented my core phonology for
EIL, the Lingua Franca Core. The latter prioritises for teaching those features identified
in a corpus of ILT data as being crucial to the preservation of phonological intelligibility
and which, because they also take into account the complex workings of L1 transfer, are
pedagogically feasible.

However, while the teaching of the Lingua Franca Core is necessary, it is not sufficient.
Although the LFC comprises only features which are considered to be teachable, in the
sense that learning is likely to follow teaching (see pp.133-134), we cannot guarantee that
all learners will make the effort to learn the core or, subsequent to learning, will actually
use it at all times in all appropriate interactions. Inevitably there will be occasions when
factors such as processing overload, nerves, strong emotion and the like intrude. These,
in turn, may cause a speaker to revert automatically to earlier and more firmly-established
pronunciation habits and so, very probably, to pronunciation containing L1 transfer in the
core areas.
 64

In Chapter 5 we noted that, generally speaking, there seems to be a one-to-one

correspondence between the relevant (items essential for EIL intelligibility) and the
realistic (items which are teachable) and between the irrelevant and the unrealistic. I
argued, though, that if a particular feature was potentially unteachable because of strong
transfer effects, but was also crucial to EIL intelligibility, then learners' motivation to be
intelligible may override unteachability. This, I suggested, is so in the case of word-initial
consonant clusters. Owing to the interaction of transfer with universal processes,
consonant clusters in any position in a word are very difficult for most adult learners of
English, but in word-initial position, the deletion of sounds threatens EIL intelligibility.

It may be, then, that what is crucial to intelligibility sometimes becomes teachable when
learners discover its contribution to effective communication. But this is unlikely always
to be the case. And if, as a result of future research, some features which I have
designated 'non-core' turn out, after all, to be core, then the outcome may depend on their
degree of unteachability, or difficulty. If, because of the interaction of transfer with
universal or developmental processes, they are at the extreme limit of unteachability, then
the best we can do is to prime learners for future learning by drawing their attention to
these features. Learners will then be better placed to acquire them for habitual use
through any future lengthy exposure outside the classroom. If features are unteachable for
automatic use, but are produceable by means of great conscious effort, learners will need
pedagogic input to enable them to appreciate both why it is sometimes necessary for them
to make this effort, and precisely what is involved.

In essence, there are two issues at stake here, though they are in fact two sides of the
same coin. On the one hand, speakers need to develop the ability to adjust their
pronunciation according to the communicative situation in which they find themselves.
This means that they need to be able to assess the relative necessity of pronunciation
intelligibility for their interlocutor of the moment and, where this is high, to make the
crucial adjustments that will guarantee it for that particular interlocutor. In other words,
they need to be able to 'accommodate' (or more specifically, 'converge') towards their
listeners. On the other hand, listeners also have a role to play. At the end of the first
chapter, I spoke of the need for receivers of English, both L1 and L2, to develop greater
tolerance of accent difference in general, and the ability to adjust their expectations to
accommodate the specific interlocutor in the specific communication event. Listeners
engaged in ILT have to accept that they cannot expect target pronunciation, even in the
core areas, one hundred per cent of the time, and that they must learn to cope with a
certain amount of L1 transfer.
 65

In this chapter, we will consider what these productive and receptive accommodation
skills involve, and how pedagogic means can be contrived to enable learners to develop
them. But because the concept of accommodation is crucial to the discussion of
phonological intelligibility that follows (just as the concept of L1 transfer was crucial to
the discussion of the LFC), and because the term 'accommodation' is used somewhat
loosely to cover a range of (often very general) meanings, we will first establish what
accommodation entails according to accommodation theory.

Throughout the chapter, I will for the most part be using the term 'ILT' in place of the
broader 'EIL'. This is to make clear that our interest in accommodation is grounded in
interactions among non-bilingual English speakers (NBESs), that is, interactions in which
the need for phonological intelligibility is similar for all interlocutors. Bilingual speakers
of English, like L1 speakers, have far greater contextual and cotextual resources at their
disposal, so that where accommodation occurs, it is less likely to be motivated by a desire
for interlocutor intelligibility which, I will argue, is the prime motivation for
accommodation in ILT.

Accommodation theory and intraspeaker variation in ILT

Our concern with accommodation skills relates specifically to their potential to enhance
phonological intelligibility in ILT. In essence, we are dealing here with the question of
intraspeaker interlanguage variation (see Chapter 3), in our case specifically the
variation between correct and incorrect core phonological forms. In accounting for this
variation, accommodation theory offers important pointers for the devising of classroom
procedures that will encourage variation in the direction of correctness, when the
intelligibility of core phonological items is crucial to the success of an interaction. So
before we move on to the pedagogical implications, we will consider precisely why and
how accommodation theory is able to account for phonological variation in ILT and, in
so doing, will identify these pointers.

Accommodation theory (more correctly, Communication Accommodation Theory, or

CAT1) is not the only theory which has been proposed to account for the variation
between correct and incorrect forms in interlanguage. Nor has it been proposed that
accommodation theory provides a unitary model to account for all levels of interlanguage
variation. But several accounts of IL variation concur on the importance of social context
and, in particular, the role of audience factors. Bell (1984), for instance, argues that

1 Originally Speech Accommodation Theory, or SAT.

 66

variation is the result of adjustments, or accommodations made by speakers according to

their interlocutors' personal characteristics, general speech style, and specific linguistic
usage. He thus identifies social factors which are responsible for triggering both
psychological and cognitive processes.

A description of the many other theories and models attempting to account for IL
variation lies well beyond the scope of this book. Anyone wishing to pursue these should
refer to Tarone (1988) for the most comprehensive survey of IL variation to date, as well
as the shorter surveys of Wolfram (1991), Ellis (1994) and Jenkins (1995). However, the
main surveys of IL variation conclude with Bell that despite certain shortcomings,
accommodation theory offers the most satisfactory explanation for IL variation to date.
And, for reasons which we will consider, it is particularly relevant to phonological
variation in ILT.

The accommodation framework

The theory originated in its earlier form of Speech Accommodation Theory (SAT) to
account for the motivations underlying adjustments in people's speech, in particular the
cognitive and affective processes underlying the speech strategies of convergence and
divergence (making one's speech respectively more and less like that of one's
interlocutor). The goals of such speech adjustments were claimed as one or more of the
following: evoking the addressee's social approval, promoting communicative efficiency
between interlocutors, and maintaining a positive social identity (Beebe & Giles 1984).
The many studies conducted within the accommodation framework have tended to place
greatest emphasis on the first and third of these goals, and the majority of them to
investigate L1 interaction. However, studies of IL variation tend to focus on the
motivation of communicative efficiency as being better able to account for adjustments
made by L2 speakers in certain speech situations.

Accommodation theory draws on four social-psychological theories in order to explain

style shifts. Firstly, the theory of similarity attraction, which claims that people are more
attracted to those who share similar beliefs and attitudes than to others. Secondly, social
exchange theory, according to which people weigh up the rewards and costs of
alternatives before they act, usually selecting the alternative which will result in the
greatest reward and smallest cost. Thirdly, the theory of causal attribution, which
suggests that people evaluate one another's behaviour according to their interpretation of
the motives underlying that behaviour. And fourthly, intergroup distinctiveness,
according to which people attempt to maintain their group identity by retaining their
 67

distinctiveness from other groups. Beebe & Zuengler (1983) point out that it is possible
for two or more of these facets to be operating at the same time to cause variation in a
speaker's language and to offer an explanation for the strategies of convergence and
divergence, which form the basis of accommodation theory.

The first work on accommodation was published in the early 1970s, when Giles (1973)
demonstrated the phenomenon of interpersonal accent convergence in an interview
situation, and introduced his 'accent mobility' model. He argued that instead of explaining
situational variation by means of Labov's 'attention to speech' paradigm (Labov 1972)
and the formality or informality of the context, the focus should shift to processes of
interpersonal accommodation and, in particular, to receiver characteristics. These
processes were subsequently incorporated into SAT and led to a plethora of further
research by Giles and others, in which a wide range of speech variables were manipulated
in a number of different settings. In its original form, SAT "aimed to clarify the
motivations underlying speech and intermeshed in it, as well as the constraints operating
upon it and their social consequences" (Giles, Coupland & Coupland 1991:6), by means
of focusing on speech convergence and divergence, and attempting to explain the
cognitive and affective motivations underlying them. Later, the theory was broadened to
include other strategies, such as complementarity (which occurs in situations such as
interviews, where role discrepancies are to be expected) and over-/under-accommodation.
In more recent years, its scope has been further extended to incorporate a whole range of
non-verbal and discursive dimensions of social interaction, which are reflected in its
change of name to Communication Accommodation Theory (Giles & Coupland
1991:63).

In essence, convergence is a strategy by which individuals adapt to one another's speech

(SAT) and other communicative behaviours (CAT) in terms of a wide range of linguistic
and prosodic features, such as speech rate, pauses, utterance length, pronunciation and, in
the case of CAT, non-vocal features such as smiling and gaze. Divergence, on the other
hand, refers to the way in which speakers emphasise speech and non-verbal differences
between themselves and their interlocutors. Divergent strategies range from a few
pronunciation and content differences to abuse and the switch to another language.
Convergence is a strategy of identification with the communication patterns of an
individual and thus internal to the interaction. Divergence, by contrast, is more often a
strategy of identification with the communicative norms of a reference group and thus
external to the interaction, and therefore predicted to occur more frequently in
interactions where speakers have different social identities. Maintenance, a third
accommodation strategy, is in effect a type of divergence in that interactants preserve
 68

their speech patterns and other communicative behaviours across situations, in order to
maintain their group identity.

Within these broad categories are a number of distinctions. Both convergence and
divergence may be upward,by means of a shift towards a prestige variety, or downward
by means of a shift away from it. They may also be uni- or multi-modal (respectively at
one and two or more levels), partial or total (where total would indicate a one hundred
percent matching of the interlocutor's speech on the dimension under consideration),
symmetrical (reciprocal) or asymmetrical (non-reciprocal), large or moderate, and
objective or subjective (convergence/divergence respectively to what is actually heard or
to a belief about the interlocutor's speech, in other words, to a stereotype). There have
also been found to be optimal rates of convergence (Giles & Smith 1979), with the
phenomenon of 'overaccommodation' occurring when a speaker is considered by the
recipient to be making more adjustments than necessary, and thus often leading to
miscommunication despite the speaker's precise intention to produce the opposite effect
(see Coupland et al. 1988 for examples of overaccommodation).

Of the three main motivations found to underlie convergence and divergence, the first
two (to gain the interlocutor's approval and to communicate efficiently) relate primarily
to convergence, and the third (to maintain a positive social identity) to divergence. In its
unmarked form, convergence relates to the theory of similarity attraction, reflecting a
speaker's often unconscious desire for identification with another, and the consequent
adjustments made to his speech in order to sound more similar to the other. Various
studies have shown that through convergence a speaker may increase his attractiveness,
predictability, intelligibility and interpersonal involvement in the eyes of his interlocutor
(Giles et al. 1987). Thakerar et al. point out, though, that "the magnitude of speech
convergence will be a function of the extent of the speakers' repertoires" (1982:218), and
this is something we will need to bear very much in mind in relation to ILT.

The second and third motivations for convergence, i.e. communicative efficiency (also
referred to as 'communicational efficiency', 'communication efficiency' and
'communicative effectiveness') and identity maintenance, are psychological rather than
affective in essence. A desire for communicative efficiency which, as stated above, may
be the principal motivation for convergence, can lead to the cognitive organisation of a
speaker's output. This involves the speaker's organising of his speech to take the
recipient's requirements into account which, in turn, leads to increased intelligibility.
Thakerar et al's seminal study of psychological convergence and divergence (1982) has
demonstrated how a complementary relationship, such as that of interviewer and
interviewee, can increase mutual predictability, which is also likely to facilitate
 69

understanding. More recently, research into discourse attuning has shown how a speaker
takes into consideration the recipient's ability to understand, or his 'interpretive
competence', and has postulated a range of 'interpretability' strategies.

The communicative efficiency motivation is of particular relevance to ILT research.

Speakers who find themselves together in an attempt to accomplish a particular task, the
successful accomplishment of which is to their mutual advantage, will be instrumentally
motivated to facilitate communication in order to achieve a successful outcome. If these
speakers come from different linguistic and cultural backgrounds, then accommodative
processes will have a far greater role in enhancing mutual comprehension than they
would in communication between speakers from similar backgrounds. In particular, when
widely differing accents come into contact in such speech situations, their users will feel
strong internal pressure to converge phonologically in some way, in order to promote
their own intelligibility.

The communicative efficiency motivation will be discussed in this chapter as the primary
motivation for attempts at phonological convergence, and thus for IL phonological
variation in ILT contexts. In other words, variation in these contexts can largely be
viewed not as the negative lapsing into error of intermittently correct items, but as the
positive making of effort (whether consciously or subconsciously)to rectify intermittently
incorrect items, even though the speaker's repertoire and level of competence may
conspire against this intention.

In proposing the updated "reformulated SAT", Giles et al. argued that "SAT presents a
broad and robust basis from which to examine mutual influences in communication,
taking account of social and cognitive factors, and having the scope to cover the social
consequences of speech shifts as well as their determinants and the motivations
underlying them" and has "the flexibility of relevance at both interpersonal and
intergroup levels" (1987:34). For the future, they suggested that, among other things, "an
SAT perspective might have applied relevance for situations rife in potential
miscommunication and misattribution" (op.cit:41). With this in mind, we will move on to
consider the accommodation framework in relation to ILT, a context which - as we have
already seen - is "rife in potential miscommunication"

Accommodation theory: an explanation for phonological variation in ILT

Since L1 pronunciation is to a considerable extent the product of habit, its features are
likely to be transferred automatically in the production of a second language. The transfer
 70

process commonly results in an abundance of L1-specific pronunciation 'errors' of the

kind we observed in Chapter 2, which, by definition vary from speakers of one L1 to
another. Such phonological deviations from the target pronunciation are most likely to
occur when there is a lack of monitoring of pronunciation performance, as contrasted
with syntactic and morphological errors, which often occur despite monitoring, as a result
of learners' processing difficulties relating to degree of knowledge and control of the L2.

Turning to hearer factors, we noted at length in Chapter 4, that NBES receivers have a
tendency to process language with minimal reference to contextual cues and instead to
focus primarily on the acoustic signal. Error-free pronunciation in terms of the subset of
error types (i.e. the LFC) that was proposed in the previous chapter is therefore crucial as
a first stage in facilitating the NBES hearer's ability to interpret spoken interlanguage.
However, interlanguage pronunciation is rarely free even of this subset until a relatively
late stage in the SLA process. And to compound the problem, because of their L1-
specific nature, phonological transfer errors potentially pose a far greater threat to ILT
listeners than do errors from other linguistic areas, where there is more common ground
among NBESs (see Ioup's 1984 study comparing written and spoken ILs). The subjects
who took part in my ILT studies, when interviewed, invariably cited their interlocutor's
pronunciation as the cause of mis- or non-understanding.

So because the problem operates at both speaker and hearer level in ILT, phonological
transfer errors are far more likely to lead to problematic communication here than they
are in so-called 'native/non-native' interaction. However, the principal motivation for
most adult learners of English is their wish to be able to communicate intelligibly with
other speakers of English, and in the Twenty-first Century, these are most likely to be
NBESs from different L1 backgrounds. Thus, we can predict that in those ILT situations
where the conveying of information is highly salient to speakers, they will monitor their
pronunciation and attempt to adjust it by means of some sort of convergence towards that
of their NBES addressees. Where they are successful, the result will be a reduction in the
phonological differences between themselves and their addressees, so rendering their
speech more intelligible to the latter. In this respect, it seems that the motivation
underlying convergence in ILT, that of communicative efficiency, is the same as the
motivation which underlies recoverability. For, as we saw in Chapter 5, it is the desire to
promote addressee comprehension which determines the selection of certain phonological
simplification strategies over others, in particular, epenthesis over consonant deletion (see
pp.117-120 above).

On the other hand, the socio-psychological concepts of intergroup distinctiveness and

identity assertion in which accommodation theory was originally rooted are rather less
 71

relevant to ILT. For while the notion of group dynamics is central to accommodation in
ILT, these dynamics are bound to operate very differently when membership moves from
that of an L1 group to that of a global community. Tarone asks "what is the speech
community of the second-language learner? Where all learners speak the same native
language and share the same culture, this may be clear. But what about the case of the
ESL classroom where nearly every learner may come from a different native-language
background? With what speech community(ies) does a learner in this situation identify?"
(1988:118).

The EIL speech community is, nevertheless, a complicated one. On the one hand
speakers are claiming membership at a global level. They are speaking English as a
lingua franca in order to make themselves intelligible to other members of the EIL
community and would not wish to diverge phonologically, since to do so would be to
jeopardise intelligibility and thus defeat their very purpose in using English. On the other
hand, the majority will not want to lose their L1 identity when they speak their L2
English and this will mean preserving something of their L1 accent. The Lingua Franca
Core solves the dilemma by providing scope for both convergence and maintenance:
NBESs are encouraged to converge on those L2 features which are essential to
intelligibility in ILT, but are at liberty to maintain those features of their L1 where
intelligibility is not at stake.

The promotion of interlocutor comprehension was first mooted as a primary motivation

for speech convergence by Thakerar et al. (1982) and has subsequently been researched
by others such as Shockey (1984) and Takahashi (1989). Giles and Coupland, the
originators of accommodation theory in the 1970s, have more recently claimed for it
"cognitive organization versus identity-maintenance functions", arguing that "increased
intelligibility is a valuable byproduct of convergent acts and may on occasion be the
principal motivation for accommodating" (1991:85; emphasis added). The latter is
undoubtedly true of ILT, whose unique set of speaker and hearer factors mean that
pronunciation errors (i.e. in core areas) at best necessitate much negotiation of meaning
and consequent interruption to the flow of conversation, and at worst cause a total
breakdown in communication. Perhaps most importantly for our present purposes,
accommodation theory not only explains why and how NBESs attempt to adjust their
pronunciation in order to decrease the differences between themselves and their NBES
interlocutors but, in so doing, also provides clues as to how pedagogic intervention may
increase the success rate of these attempts.

A number of problems have, nevertheless, been raised in relation to the application of

accommodation theory to IL variation. For example, Tarone has argued that since most of
 72

the work has been done with dialect speakers or fluent bilinguals rather than with L2
learners, we have not accumulated sufficient second language data to examine IL
variation within this framework. Again, she considers that some style shifts in IL may not
be caused by interlocutor effects but by situational norms and "the communicative
demands of different genres" and, therefore, that "group membership and the individual
identity assertion associated with it may be only one of several possible causes of
interlanguage variation" (1988:50-51). The latter point, though, can be countered with the
argument that the current accommodation (CAT) framework by no means limits the
causes of IL variation to "group membership and individual identity assertion". The
communicative efficiency motivation is also considered important, particularly in
'native/non-native' (as opposed to same-L1 'non-native') communication, and is thought
by some, including myself, to be by far the most salient factor in ILT. To this can be
added that "the communicative demands of different genres" are likely to be inextricably
bound up wih interlocutor effects.

Despite her objections, Tarone summarises her account of the empirical evidence for
interlocutor effect on IL variation by making two points:

... first, there does seem to be clear evidence that second-language learners
produce different variants in response to different interlocutors; but, second,
there are surprisingly few studies documenting this effect. There are a number
of interesting hypotheses which could be explored by further research, but
clearly more data are needed (1988:92).

Tarone nevertheless argues that that the approach provided by accommodation theory
has a number of advantages over the 'attention to speech' model:

...it focuses upon determinable social and social-psychological factors like

multiple group membership and more directly inferrable factors like identity
assertion as causes of IL variation rather than postulating an unobservable
intermediary process of 'attention' ... it allows the researcher to determine the
origin of the variants which make up the different styles of interlanguage ....
to analyse interlanguage as well as foreigner talk and code-switching within
the same framework ... to study shifts in amount of talk, speech rate, duration,
pause and utterance length, stress, pitch, intonation, and even in the content
expressed - none of which are factors easily analysed in terms of grammatical
correctness (op.cit:49).
Zuengler, in similar vein, concludes that "CAT appears promising as a theory of L2
sociolinguistic variation" and that "it is essential that further research be conducted within
a CAT paradigm to determine the extent to which the theory can help us explain the
complexities of nonnative speech" (1991:233-234).
 73

The accommodation framework in previous interlanguage variation research

A problem which we have already identified for research into accommodation in IL and
ILT settings is that most of the work has been conducted in either L1 dialect or L2 fluent
(bilingual) contexts. There has nevertheless been a small but steady growth in the study
of second language learners within the accommodation framework, although almost
entirely in NS-NNS rather than ILTcontexts. And disappointingly, accommodation
theory applied to interlanguage variation - like other sociolinguistic theories of IL
variation - has often been discussed with little by way of empirical support from L2 data,
with researchers tending to rely on L1 data and anecdotal evidence. So although
accommodation theory has been recognised as having the scope to account for synchronic
variation in interlanguage, as Zuengler (1991) points out, there is as yet no large body of
L2 data to give substance to the view.

The majority of studies of IL variation conducted within the accommodation framework

focus on phonological variation.2 Although these studies all find evidence of
accommodative processes at work in IL variation, their results are not all significant.
Zuengler (1989) accounts for this in terms of linguistic proficiency: some forms may
simply not be in the NBES's repertoire. This means that however strong a speaker's
desires to converge, they will be unable to do so where these items are concerned.

We will consider in more detail two areas of research into IL accommodation which have
particular relevance for ILT itself: research into the communicative efficiency
motivation, and that into the role of speakers' L2 repertoires.

Communicative efficiency and interlanguage

Several sources attest to the growing importance that researchers have begun to attach to
communicative efficiency since the early 1980s. At the level of content, Giles & Smith
(1979) had already focused on the issue in the early days of SAT. They had drawn
attention to the fact that speakers take into consideration the listener's knowledge and
converge with him by, for example, using less jargon with an interlocutor who does not
share their expertise, in order to increase mutual intelligibility. The first study to suggest
specifically that increased intelligibility rather than social approval may be, in many
cases, the central motivation for accommodative behaviour was that of Thakerar et al.
(1982). Others have subsequently taken up their claim. For example, Bell argues that "In
concentrating on approval seeking as a reason for style shift, accommodation has often

2These are: Beebe 1977, 1981; Beebe & Zuengler 1983; Berkowitz 1986; Flege 1987; Takahashi
1989; Zuengler 1982, 1987, 1989.
 74

overlooked a more transparent motivation: a speaker's desire to be understood"

(1984:199). Coupland (1984), reporting a personal communication from Thakerar,
suggests that communicative efficiency is more relevant than social approval in his own
study of accommodation in a travel agent setting.

Where interlocutors have different accents, particularly in bilingual situations, mutual

intelligibility becomes an increasing problem and, as Bell points out, "the sharper the
linguistic differences between codes, the larger the issue of intelligibility looms, the
stronger are the pressures to accommodate to the audience" (op.cit:176). Describing the
doctoral research of Shehadeh (1991), a Syrian researcher at Durham University, Lynch
reports that in his investigation of conversation between L2 learners of English, Shehadeh
demonstrates that "certain types of pair and group work create conditions for interaction
in which learners push each other to speak more comprehensibly and more accurately"
(1996:77). In the second half of the chapter, we will return to the topic of pair and group
work and the ways in which they can be manipulated in the classroom to induce
accommodation.

A study carried out by Takahashi (1989) to investigate accommodation within an ILT

setting, lends particularly strong support to the claim that speakers are likely to converge
towards the speech of their interlocutors in ILT contexts chiefly in order to promote
intelligibility. Takahashi examined the speech the speech adjustments of six Japanese
learners of English, three of intermediate level and three of advanced level. All were
interviewed by a high and low proficiency Japanese and a high and low proficiency
Spanish speaker of English Their conversations were analysed with respect to four
quantitative variables: the length of speech and fluency; the amount of talk; the number
of questions; and the use of meaning negotiation. Although Takahashi did not consider
them in her own study, she recommends that similar future research would do well to
investigate qualitative changes to L2 speakers' pronunciation and syntactic complexity.

In analysing her data, Takahashi found the interlocutor to be an important factor in

influencing the subjects' speech at both linguistic and psychological levels. Of particular
interest to us is her discovery that the advanced speakers converged at a statistically
significant level towards the proficiency level of their interviewers: they increased the
amount of speech with their high proficiency interviewers and decreased it with their low
proficiency interviewers. By contrast, the intermediate speakers spoke more than their
low proficiency interlocutors, but less than their high proficiency interlocutors. Thus it
appears that convergence occurred on this variable where it was within the competence of
the L2 speaker to adjust her speech in the direction of that of the interviewer. Where this
was not so, however, convergence did not occur.
 75

Accommodation and IL repertoire

This leads straight into our second central issue: that of the speaker's repertoire. In the
case of Takahashi's lower and intermediate level subjects, they did not increase their level
of speech to converge on that of their interlocutors because it was not within their
repertoires to do so. L2 speakers engaged in ILT are very likely to have the motivation to
adjust their speech in the direction of their receiver, but not to have the ability to do so.
And in this respect, pronunciation provides a particularly strong barrier to successful
convergence, chiefly because "in addition to the social and psychological factors that
affect all language performance, L2 speech is always subject to L1 interference" (Beebe
& Giles 1984:18). And as was discussed earlier at some length, such "interference"
(preferably 'transfer', or better still, 'crosslinguistic influence') is reflected most
extensively at the phonological level,and in ILT tends to involve movement away from a
target form in different directions.

It appears, then, that speakers engaged in ILT have not only competence limitations
regarding the target language, but also repertoire problems in relation to their different-
L1 interlocutors. In particular, they are unlikely to be able to match their partners'
pronunciations, on account of the L1-specific nature of many of these. At the same time,
however, their motivation to adjust their speech in order to promote greater intelligibility
for their NBES interlocutors is in all probability very high. Thus, as Beebe & Giles warn,
when we extend accommodation theory to SLA data, it is important that we consider
repertoire limitations: "For it is the tension between limitations in ability to converge ...
and motivation to converge that makes second-language data unique .... With native
speakers and fluent bilinguals, we assume that the ability to converge is there. With
second-language learners, the capability may not be there" (Beebe & Giles 1984:23;
emphasis in original). In effect, the spirit is willing but the flesh is sometimes weak.

Having considered the relatively small amount of research into the use of accommodation
by IL speakers that is most relevant to our interest in ILT, we will take a brief look at the
phenomenon of foreigner talk. Although this concerns adjustments made in the speech of
an L2 speaker's 'native speaker' interlocutor, foreigner talk is of interest because it
provides useful insights into both the features of L2 speech which elicit such adjustments,
and the role of the communicative efficiency motivation in determining the type and
degree of adjustments.

Foreigner talk
 76

The phenomenon of foreigner talk (henceforth FT) was first documented in the early
1970s, the term being introduced by Ferguson in 1971. It refers to the simplified register
which 'native speakers' use to address 'non-native speakers' and is, Ferguson suggests,
like baby talk, a reflection of the latters' belief about the way the former speak. Arthur et
al. describe FT as arising from language users' "unconscious ability to make a number of
coordinated adjustments in their language that have the net result of simplifying and
facilitating communication" (1980:113), thus making a case for FT as accommodation
motivated by communicative efficiency. Such simplification, they argue, can be elicited
by an interlocutor's 'non-native' accent.

Much information on the linguistic features of FT has been gathered. The most salient
characteristics are generally agreed to be the following: less syntactical complexity, fewer
pronouns, the use of higher frequency vocabulary, more clearly articulated pronunciation
(to the extent of unnaturalness, e.g. by the avoidance of contractions and weak forms),
slower speech rate, more questions (often for the purpose of checking understanding), as
well as the tendency to speak more loudly and to repeat. However, there is still no general
agreement as to exactly what it is that triggers FT, or why it varies across 'native
speakers' or across combinations used by the same speaker on different occasions.

Two studies, Varonis & Gass 1982 and Gass & Varonis 1985a , investigate what features
of ' non-native' speech elicit FT. The 1982 study reveals that when 'native speakers' are
asked for information both by other 'native speakers' and by 'non-native speakers', they
differentiate between the two groups in terms of the forms of response used, even when
the latter group have spoken in grammatically correct language: when replying to the
'non-natives', they repeated the most important part of the message, usually with rising
intonation such as would normally be associated with a question. The study goes on to
investigate why this should be so, and finds that when 'native speakers' evaluate and react
to the speech of 'non-natives', the most salient factor is "the comprehensibility of the total
linguistic input from the non-native to the native", and that such comprehensibility "may
be achieved in a number of different ways through the interaction of various linguistic
and social factors". The authors consider these to be pronunciation (a major factor),
grammar, familiarity with topic, familiarity with person, familiarity with speaker's native
language, fluency and social factors (1982:132).

The 1985a study builds on that of 1982, as does other work of the same authors. For
example, Gass & Varonis 1984 found that a 'native speaker's' familiarity with 'non-native'
speech in general affected positively the 'native speaker's' ability to understand a
particular 'non-native speaker'. This in turn is likely to reduce the amount of FT used (a
point not actually made in the study). Gass & Varonis 1985a demonstrate that 'non-native
 77

speaker' proficiency is determined mainly by pronunciation, fluency and comprehension.

During the course of an interaction, the 'native speaker' subjects appear to reassess their
interlocutors on the basis of their own ability to understand them (this ability at least
partly determined by pronunciation), and to adjust their speech accordingly. FT
modifications are apparently greatest for those 'non-native speakers' who are involved in
a large number of negotiations of meaning during the middle section of the interaction.

The authors account for their findings by suggesting that the 'native speakers' had been
able to determine that their interlocutors were more difficult to understand than they had
originally predicted them to be. This leads the authors to claim that "comprehensibility is
one variable that triggers NS speech modification" or, in other words, FT. They also
suggest that "it is intuitive that NS modifications to NNSs are made for the purpose of
increasing the possibility that the NNS will comprehend" (op.cit:54), though they point
out that this may involve perceived rather than actual comprehension, since evidence of
lack of comprehension is not a precondition for FT. Indeed, such talk was used before the
'non-native' interlocutor had demonstrated either comprehension or the lack of it.

We therefore have some possible explanations for the use of FT: as a response to the
incomprehensibility of 'non-native' speech; as a response to apparent 'non-native speaker'
lack of comprehension; and as a function of 'native speaker' stereotyping of their
interlocutors. However, none of this explains the variation found in FT. Zuengler
suggests that placing FT within the framework of accommodation theory "will clarify
much of the variation.... and bring a coherence to the literature" (1991:235). Zuengler
starts by examining 'native speaker' goals in communicating with 'non-native speakers'.
She argues that many of these coincide with the speaker goals conceptualised in CAT,
particularly the desire for communication efficiency and mutual comprehension.
Likewise, it may be a particularly strong goal in certain types of 'native/non-native'
interaction. For, as Zuengler points out, observers have noted a greater use of FT when
'native speakers' and 'non-native speakers' are involved in a two-way exchange of
information than in one-way communication from the 'native speaker', as a result of
greater concern for (mutual) comprehensibility.

Zuengler claims that for many 'native speakers' interacting with 'non-native speakers', two
dimensions are salient: perceived ethnic and cultural difference, and the 'non-native
speaker's' linguistic/communicative competence. She argues that the 'native speaker'
partner, having one of the goals of CAT (concern for mutual comprehensibility, a desire
to gain the interlocutor's social approval, or a desire to maintain distinctiveness), will
encode the strategies of convergence, divergence or maintenance. Thus, FT adjustments
to speech rate, pronunciation and so on can be explained as the manifestation of
 78

convergence,3 and a lack of FT features as the manifestation of maintenance or

divergence reflecting an interlocutor's desire to maintain distinctiveness from the 'non-
native' interlocutor.

ILT and accommodation: some conclusions

Beebe and Giles argued long ago that accommodation theory has the potential for
"breaking away from its essentially social psychological mold and emerging more
centrally and ... into the interdisciplinary arena" (op.cit:9). This is precisely the situation
as regards ILT, whose variation between correct and incorrect core phonological forms is
fully explicable within an accommodation framework but largely outside the theory' s
original social-psychological parameters. As we saw in the data presented in Chapter 3,
convergence in ILT manifests itself in non-traditional ways - that is, in a closer
approximation to core target forms rather than in adjustments in the direction of an
interlocutor's (incorrect) core forms - in situations where interlocutor comprehension is
particularly salient. This is, perhaps, to be expected, bearing in mind speakers' repertoire
limitations described above and the other phenomena peculiar to ILT discussed in
Chapter 3.

But having established that accommodative processes occur in this manner in ILT, we
need to incorporate the research finding into pedagogy for it to be meaningful. As
Widdowson argues, unless the findings of IL research can be exploited "to enable
language teachers to contrive the most effective conditions possible in classrooms" for
the language learning process to take place, by "providing particular kinds of warrant for
teacher intervention, then they are of little pedagogic value" (1984:324). As far as
accommodation is concerned, instead of leaving it to chance as we have done hitherto
(albeit through ignorance rather than design), we can begin to plan ways of developing
further in our learners the accommodation skills which the vast majority already possess
in their L1 and are clearly motivated - but not always able - to utilise in their L2 English,
particularly in ILT contexts.

3 Though Kasper points out that "convergence strategies used by NS to achieve mutual comprehension
and to facilitate NNS' conversational participation can in fact result in divergence, by creating or
exacerbating the very problems they were designed to forestall or solve. Frequent foreigner talk
adjustments such as completing, rephrasing and restating the NNS' utterances can produce an
asymmetrical distribution of 'interactional authority'" such that "the practice of expert-novice-
convergence....appears more ambiguous in social messages than is usually assumed: instead of
facilitating novice participation, it may accentuate asymmetry and produce psychological divergence"
(1997:354; emphasis in original).
 79

As far as the pedagogic enterprise is concerned, the first step has already been taken. In
order to reduce phonological errors, we must first understand their variable nature. This
we are able to do by recourse to accommodation theory, according to which the greatest
influence on the accuracy or otherwise of pronunciation in ILT is the salience of
interlocutor comprehension. Where L2 phonological items are not fully-acquired (and
even on occasion when they are) speakers are likely to make the effort to adjust in the
direction of correctness, rather than transferring the easier alternative from their L1,
where they perceive a strong need for interlocutor comprehension, but to produce
incorrect forms where they do not. We move now from theory to practice: to a
consideration of how our knowledge of accommodation theory and of its relevance to IL
in general, and to ILT contexts in particular, can be put to good use in teaching.

Accommodating classrooms

So what are the implications for a pedagogy which aims above all to maximise learners'
success in international contexts? To summarise what we have we have established so
far:

1. understanding in ILT contexts is heavily dependent on phonological

intelligibility
2. phonological intelligibility resides in the correct (target-like) production of a
group of core items
3. NBESs desire and may be able, under certain conditions, to accommodate
towards each other by converging on these core items

However, the accommodative ability is embryonic in IL. In the vast majority of cases,
pedagogic intervention is necessary if NBESs are to accommodate their ILT interlocutors
successfully, both productively and receptively, and in larger groups - quite possibly
containing members of their own L1 group - as well as in different-L1 pairs (see below
on the issue of conflicting group membership). We need, as teachers, need to have the
knowledge of precisely what these 'certain conditions' involve and of the extent to which
it is possible for learners to rise above them, as it were, and accommodate their
interlocutors even where the particular linguistic and/or extralinguistic circumstances
render this more difficult.

Even when the conditions are ideal, accommodation in the traditional sense (i.e. speakers
converging on their interlocutors' pronunciations) rarely occurs. In so far as NBESs share
a common purpose, it is to get closer to each other in terms of the L2. They do not want
to acquire one another's pronunciation errors and as long as their motivation for
 80

accommodating their interlocutor remains purely that of communicational efficiency,

they seem to have some sort of inbuilt mechanism (conscious or subconscious) which
safeguards against this outcome. Even if, over a prolonged period of time, the motivation
subsequently broadens to include that of social approval, realisations of traditional
convergence only very occasionally occur. Among the half dozen examples in my data
are the following: a Japanese learner converged on the speed and stress patterns of a
Swiss-French peer; another Japanese learner occasionally used terminal devoicing (e.g.
she pronounced 'sad' as pronounced \sœt\)during the information exchange phase of some
of her interactions with her Swiss-German partner; another Swiss-German learner
occasionally echoed the American pronunciations of his Japanese interlocutor.
These cases of traditional convergence involved friendships which had became very
close, though only the longest one (that between the Japanese and Swiss-French learners,
who were together for six months) resulted in acquisition in the sense that the Japanese
learner's speech continued to manifest the Swiss-French characteristics when not in the
company of her Swiss-French friend.

As a rule, though, interlocutors do not converge on one another's L2 accents even during
the interaction, let alone afterwards. And even where speakers from different L1s actually
share common transferred features in their English pronunciation, the same appears to be
true, at least for those speakers who have not had much previous exposure to one
another's accents. For example, in one interaction in my data, a Japanese speaker was
recorded in extended communication with a Portuguese speaker (from Portugal) for the
first time. Prior to this, the Japanese speaker had never communicated with a Portuguese
L1, and the Portuguese speaker had had very limited exposure to Japanese accents. Both
subjects had a tendency to drop final \n\ and nasalise the previous vowel sound. For
example, the Japanese subject pronounced 'Japan' as [dZœpœ‡] and 'considered' as
[kO‡sIdEd]; the Portuguese subject pronounced 'inland' as [i‡lœnd]. Each makes this
transfer error several times during the social exchange task, but when they move on to the
information exchange task (where intelligibility is far more salient: see Chapter 3), the
Portuguese subject does so only twice and the Japanese not at all. This is very different
from the outcome in the same-L1 pairs, where the number of (shared) transferred features
actually increased from the social exchange phase to the information exchange.

This suggests that in interaction between those from non-familiar L1 backgrounds,

convergence takes the form of an attempt at the blanket replacement of 'high risk' L1
features rather than an attempt to match the interlocutor's similar pronunciations. It is
very likely that both specific pedagogic intervention and repeated (possibly first-hand)
exposure to the other accent are necessary before interlocutors even become aware of
such similarities across L1 accents. The latter point supports Bell's claim that people
 81

naturally converge more to each other in their speech on subsequent occasions (1984:62),
although more research is needed to support the claim for ILT.

The above all runs counter to the fear frequently expressed by teachers and learners alike
in multilingual classrooms that learners will somehow acquire one another's
pronunciation errors. Research in this area suggests that the fear is unfounded (see, for
example, Porter 1986, Lightbown & Spada 1993). My own findings lead to the same
conclusions. Even over time, subjects did not reveal more than a small sprinkling of one
another's phonological transfer errors and, apart from the single example described
above, it is inconceivable that they actually 'acquired' these errors in the sense of
retaining them outside the accommodative situation. Both lack of repertoire and
psychological resistance would militate against such acquisition.

Optimum conditions for accommodation in ILT

Having dismissed the phenomenon of traditional convergence as being largely irrelevant

to ILT contexts, we move now to a consideration of the ideal conditions to elicit
productive and receptive accommodation of the type we have identified in ILT: that is,
productively the replacement of core transferred items with more target-like forms, and
receptively the mental adjustments that render a listener more able to cope where such
transfer replacement fails. I have argued above, in line with accommodation theory, that
where certain conditions prevail, ILT interlocutors have a strong desire to make these
adjustments for the sake of mutual intelligibility. However, successful realisation of this
desire is often another matter. While NBESs clearly do meet with a modicum of success
when they make sufficient effort - as demonstrated by the data presented in Chapter 3 -
pedagogic intervention is needed to enable them to take maximum advantage of the
conditions which elicited the wish to accommodate in the first place.

Below is a list of the conditions most likely to induce speaker and listener attempts to
accommodate phonologically towards one another:

Conditions for productive convergence

• interlocutor intelligibility is the most salient aspect of the interaction to the speaker
• the speaker appreciates the listener's difficulties in making use of extralinguistic context
• the target item is within the speaker's repertoire and can be produced effortlessly
• there is no processing overload to prevent a focus on pronunciation
Conditions for receptive convergence
• the receiver is motivated to understand
 82

• the receiver has had prior exposure to the speaker's accent

• the receiver has had prior exposure to a range of L2 accents and has developed a
tolerance of difference
• the receiver does not have a fear of acquiring the speaker's transfer errors
• the receiver is linguistically and affectively able to signal non-comprehension

These conditions divide into two further categories: interlocutor factors and situational
factors. The former comprises the speaker's phonological repertoire and awareness of
listener difficulty with context, and the receiver's familiarity with other L2 accents,
attitude towards L2 errors, and ability to signal non-comprehension. The latter comprises
those features of the situation which increase the speaker's motivation to be intelligible
and the listener's to understand, and the processing load on the speaker. In all these cases,
it is a relatively straightforward matter for teachers to devise appropriate classroom
activities that will help their learners build on whatever pre-existing motivation to
accommodate they already possess, and develop whatever skills they have already
acquired in this direction.

Before turning to specific ideas for contriving and enhancing these conditions in order to
develop learners' accommodation skills in the EIL classroom, I will discuss these
conditions in greater detail, focusing on how they can be maximised through classroom
teaching, to increase the likelihood that learners will in the future be able to translate their
accommodative desires into linguistic actions. Because there is often a considerable
degree of overlap between productive and receptive conditions, I have regrouped the
items, for the purposes of the discussion which follows, into the interlocutor and
situational factors outlined in the previous paragraph.

Interlocutor factors
Phonological repertoire
In this category,we are concerned above all with speaker and hearer phonological
repertoires though because interlocutors, by definition, are both speakers and receivers, in
practice both sets of repertoires are necessary for all participants engaged in ILT. On the
one hand, speakers need within their productive repertoires the features identified as
crucial to intelligibility in the Lingua Franca Core and, moreover, need to be able to
produce these items with relative ease. This means that most, if not all, NBESs will need
to acquire some new pronunciation habits in the core areas. Without these additions to
their pronunciation repertoires, speakers will lack the essential resources to enable them
to adjust their pronunciation in the direction of more target-like production, no matter
how strong their desire to do so.
 83

On the other hand, receivers need to range far beyond the limits of the LFC in their
receptive repertoires in order to be able to cope with the inevitable failures of their
interlocutors to make core adjustments when other factors intrude and make it difficult
for them to do so. This involves two factors. Firstly, the addition to learners receptive
repertoires of a range of L2 and L1 accents of English. Initially these should be the accent
varieties of those speakers with whom they are most likely to interact, but over time they
will be able to add to this repertoire. The best way for this familiarity to be achieved is
through repeated pedagogic exposure to other accents, with attention in the early stages
of exposure being drawn to areas of difference, especially where these areas are core
items and thus 'high risk'. Secondly, learners need to develop a greater awareness of the
fact of L2 variation in general (particularly in core areas) and a readiness to attempt to
cope with it, especially when faced with a completely 'new' accent. In fact, the second
development should eventually occur automatically as a consequence of the first -
although both can undoubtedly be speeded up by appropriate classroom teaching. L2
speakers of English thus become 'inter-accental' speakers of English (see Chapter 8).

Context
A second interlocutor factor relates to a speaker/receiver difference in the use of
extralinguistic context, something that I touched on earlier (p.91). In the ILT data, there
are occasions when interlocutor comprehension was highly salient to the speaker, and yet
the latter did not make the effort to adjust non-target-like core pronunciations. My
interpretation of these instances is that when deciding how much effort to invest in
replacing phonological transfer, speakers weigh up (whether consciously or
subconsciously) the risks involved for their interlocutors. If clear extralinguistic cues are
available - such as the elicitation pictures used in my studies - speakers may decide that it
is safe to relax their controls on pronunciation. In this case, transfer will only be replaced
if the item is fully within the speaker's repertoire, i.e. is part of procedural as well as
declarative knowledge. And this will be particularly so in cases where a phonological
item is especially problematic for a speaker. For example, Japanese speakers of English
famously have problems in producing \r\ which they regularly conflate with \l\ (far more
so than vice versa) and the problem is inevitably far greater in the production of clusters
such as \gr\ than in that of the single sound (see pp.90-91). This may well account for
exchanges such as that reproduced on p.81, where the Japanese speaker, aware that her
partner had a picture of a grey house before him, did not make the effort to produce the
\gr\ of 'grey'.

It seems, then, that contextual cues are far more salient to L2 speakers than to listeners.
Speakers, whether L1 or L2, tend to opt for the route of greatest ease where they do not
predict negative effects for their listeners. In the case of ILT, speakers seem to lose sight
 84

very temporarily of the context-interpretation difficulties that they themselves have as

listeners, and relax their control in relation to a difficult pronunciation if such context is
available. Some sort of awareness raising is required here to enable NBESs while in the
act of speaking to keep in mind their listener's orientation to context.

The ability to deal with non-comprehension

Again, this is something that I mentioned briefly in an earlier chapter (p.76), where I
pointed out that many NBESs appear to be reluctant to signal their non-comprehension,
particularly in ILT interactions. This reluctance may be the result of a listener's desire not
to lose face by admitting non-understanding. Or, more plausibly in ILT, it may be the
result of their desire not to cause an interlocutor's loss of face by pointing out that their
speech is incomprehensible. This seems to be particularly so where pronunciation is
concerned. As I described earlier, many of my subjects found one another's pronunciation
difficult to understand, but a number were not prepared to mention the fact in their
interlocutor's presence, and either wrote about it on their questionnaire or told me in
private. In this case, their concerns probably relate to the issue of identity, and the fact
that - as these subjects were aware from personal experience - one's identity is very
closely bound up with one's L1 accent in L2.

A third possibility for the reluctance, however, is a purely linguistic one: a lack of
appropriate language with which to signal the non-comprehension. As regards both the
second and third possibilities, specific teaching can both demonstrate the helpfulness (to
speaker as well as listener) of the listener's signalling their nonunderstanding, and provide
linguistic models of ways to do so. As far as accommodation is concerned, in the absence
of such signals, speakers will lack one of the main incentives to adjust their
pronunciation, that is, evidence of their lack of intelligibility for their receivers.

Fear of phonological error acquisition

We considered this issue earlier (pp.179-180). The main point to emphasise here is that
learners generally do not acquire the phonological errors of their peer groups. This only
ever seems to happen in the case of the formation of very strong friendships, where the
motivation for accommodation shifts along the Convergence continuum away from
communicational efficiency and in the direction of solidarity (the more common
motivation for convergence in L1-L1 interaction). As I have already pointed out, in
converging with an interlocutor, a speaker may indeed produce an occasional
phonological error typical of the latter. But the key word here is 'occasional'. The
phenomenon is rare and the retention of these errors as acquired items (i.e. used in the
absence of the specific interlocutor) rarer still.
 85

Convergence continuum: changing motivations for convergence over time

replacement of L1 transfer in core areas->selective replacement of transfer in core areas4

and occasional convergence on interlocutor's
phonological transfer errors
-------------------------------------------------------------------------------------------->
convergence for convergence for
communicative rapport
efficiency

And, in fact, the opposite is far more likely to happen, much to the surprise of learners
when the evidence is presented to them. When I recorded learners in same-L1 pairs at the
end of a ten week period, during which I had recorded the same learners in different-L1
pairs, the results, as we saw earlier (pp.58-64) were extremely revealing. When learners
interacted with a partner from a different L1, they made many fewer pronunciation errors
than they did in interactions with a partner from the same L1. Far from acquiring their
different-L1 partner's errors, the chief influence of this partner was to cause them to make
fewer of their own typical transfer errors. The subjects involved in this particular study
were shocked when they listened to the recordings of their same-L1 dyad interactions,
although they pointed out that they had understood one another rather more comfortably
than they did their different-L1 partners. Nevertheless, they decided to take their CAE
Speaking examination with the latter because they felt their performance would be rated
more highly by the examiners if they made fewer phonological transfer errors.

Situational factors
Motivation to be understood / Motivation to understand
These productive and receptive convergence factors are two aspects of the same
phenomenon: the salience of intelligiblity to both speaker and listener. For whatever
reason, in certain exchanges, interlocutors are particularly concerned that the speaker's
words will be interpreted correctly, and are both aware of the potential for the latter's
pronunciation to impede such intelligibility. One of my studies (see above pp.64-67)
focused on the amount of convergence when interlocutors were engaged in an exchange
of information in order to complete a task as compared with that in undirected social
interaction. The findings were clear: when interlocutor intelligibility is highly salient,
speakers make a much greater effort (and are successful) in replacing L1 transfer in core
areas with closer approximations to the target. This pattern was replicated in various
types of different-L1 dyads: all male, all female, male-female and across a range of L1s,

4 'Selective replacement of transfer in core areas' refers to the following phenomenon: with increasing
familiarity with a particular interlocutor's IL accent, speakers appear gradually switch from the
replacement of all core items (where this is within their competence) to a selectivity based on the
particular interlocutor's receptive needs - which, themselves simultaneously decrease. The same
process is probably true of increasing familiarity with an accent per se, i.e. not only that of an
individual speaker, though future research is required to clarify the extent to which this is the case.
 86

age groups and educational backgrounds. The findings were also confirmed by the
subjects themselves in questionnaires. Most subjects nominated pronunciation as the
main - or even the only - cause of miscommunication in ILT and, in several cases,
claimed to alter their speech, particularly their sounds, as a function of the L1 of their
interlocutor, and/or to try to speak English in a more 'standard' way than they did with
'native speakers'.

Clearly, the salience-of-intelligibility motivation is one that can be exploited in classroom

communication activities in order to give learners useful practice in accommodating their
interlocutors in this way. In the final section of this chapter, we will consider some of the
possibilities.

Processing overload
We have already considered this speaker-oriented factor to some extent in the discussion
of context. The crucial point is that when speakers are simultaneously having to process
language at a number of different levels - lexical, grammatical, and discoursal, if they
encounter processing difficulties, they tend to reserve their conscious efforts for the latter
levels and, in areas where their pronunciation is still variable, to 'allow' it through lack of
focus to return to more established (i.e. L1) habits. Thus, even when interlocutor
intelligibility is salient, there may not be sufficient attention available for phonological
convergence.

Again, pedagogy can help learners to cope better with such situations. Initially, in order
to provide learners with as much opportunity as possible to practise accommodating to
one another, teachers can ensure that their tasks do not cause any such overload. But this,
of course, will not prepare learners for interaction outside their classroom. Gradually,
therefore, the linguistic load can be increased but, at the same time, another factor should
be introduced. That is, through the task design, learners need to be made aware from first-
hand experience that when they have to make linguistic choices, compromising on
pronunciation is likely to be the most damaging one in terms of their intelligibility: they
should not treat it as though it were the last in the queue for conscious attention.

Teaching accommodation skills

Hassall distinguishes between 'explicit EIL' "where informed discussion proceeds in an

open conscious manner with individuals who have the capacity to contribute" and
'implicit EIL' "where models and materials have to be chosen for the students and
presented to them" (1996a:422). The former is the stuff of discussion and debate among
 87

academics and it is the latter, the 'implicit EIL' that is more relevant to the language
teaching we are here concerned with. Elsewhere, Hassell talks thus of implicit EIL:

...its goals are realistic being concerned with communication

between non-native speakers and closely related to the importance
of English in different cultures and communities. The adjustment
in perspective necessary to promote EIL rather than EFL/ESL in
the classroom increases the legitimacy of student speaking to
student in the classroom not just as a preliminary to non-native/
native speaker interaction but as English speaker to English speaker.
(1996b:70)
It is precisely this concept of "student speaking to student" that is so crucial to the
classroom acquisition and development of phonological accommodation skills to prepare
learners for ILT communication outside the classroom. In fact, it is only such interaction
that is able to promote these particular skills. More controlled, teacher-led activities
undoubtedly have their purposes, but without student-student interaction we remain with
EFL or ESL, that is, with students whom we have prepared for communication with
'native speakers'.

The following grid summarises the differences between the two basic types of classroom
activity.
Summary of the effects ofTeacher-led and ILT interaction on IL phonology and SLA
___________________________________________
Teacher-led input/interaction ILT peer group interaction
_________________________________________________________________________

IL phonology • controlled practice of core items • opportunity to practise

accommodation skills
• exposure to non-core NS • exposure to other IL varieties
pronunciation features
_________________________________________________________________________

SLA • grammatical accuracy • comprehensible input ->intake

• L2 pragmatic competence • more opportunities for output
__________________________________________________________________________

I have included the section on Second Language Acquisition in general in order to

reinforce the point that teacher-led activities cannot by themselves constitute EIL. They
provide learners with good L2 grammatical and pragmatic models and opportunities to
practise these in a more controlled way. However, EIL learners do not necessarily require
the full and unmodified NS grammatical inventory, and they certainly do not need to
acquire English NS pragmatic behaviour, particularly at the sociopragmatic level, to
function appropriately in the English language in international contexts.

As regards IL phonology, controlled teacher-led work is of direct benefit to learners in

providing them with models and practice opportunities of the features in the Lingua
 88

Franca Core. This is an essential prerequisite to classroom work on accommodation

skills, as learners will not be able to converge with one another on more target-like
pronunciations if it is not within their capacity to produce these - and this is particularly
the case with nuclear stress. An extensive focus on the LFC sounds, including drilling
and tailor-made minimal pair work (see Brown 1995), will enable learners to develop
new automatic motor plans. Clear demonstrations of the rules of nuclear and contrastive
stress and practice activities, again involving minimal pairs, will help learners move from
receptive to productive competence in this problematic area.

But it is in ILT activities that learners have the opportunity to develop the appropriate use
of these items and, in particular, their accommodation skills. At the same time, they
receive valuable exposure to other IL varieties which helps them develop their receptive
competence for EIL. Less controlled pair and small group work, particularly involving
information exchange, will provide an ideal climate. As Gass & Varonis (1991) point out,
activities involving a two-way exchange of information are better than those which are
only uni-directional, because they involve more negotiation of meaning and thus more
opportunities for learners to adjust their pronunciation, adjust their receptive
expectations, signal non-understanding and the like.

Examples for lower proficiency levels are activities involving directions and diagrams:
both learners have the same map or basic diagram and have to describe to one another
what they need to do in order to complete them. At higher levels, the same basic strategy
can be used, but greater sophistication can be introduced. For example, following the
reading of a text on the psychology involved in supermarket layouts, learners could
design their own layouts, do a standard describe-and-draw activity and then, still without
seeing one another's plans, compare and discuss them, and arrive at an optimum solution.
Although this type of activity is concerned above all with the quality of the interaction
which takes place between the learners, it should not be forgotten that the teacher is still
in ultimate control and has crucial roles both as classroom manager (selecting and setting
up the activity appropriately for the particular learners) and as provider of feedback on
performance. While learners often discover for themselves where their faulty
pronunciation or faulty reception has caused a breakdown in communication, it is
generally the teacher who is able to able to pinpoint precisely what has gone awry, and to
comment also on the suitability of strategies used, such as the asking for repetition or
clarification.

Possibly the most useful ILT activity of all to promote phonological accommodation is
student-student dictation. Brodkey demonstrated the usefulness of dictation in measuring
mutual intelligibility almost three decades ago (1972). Rimmer has more recently argued
 89

that "dictation has a special relationship with pronunciation because of the importance of
listening skills" (1997:36). This is only half the story as far as ILT is concerned, because
ILT dictations involve learners in dictating to one another, which maximises the
possibilities for exposure to different L1s in ILT classrooms. For example, learners could
take turns in dictating a number of statements about themselves, all except one of which
were false, and the task of the listener(s) be to take down only the statement they thought
to be true.

Activities like this involve optimum exposure to one another's accent varieties, as
students will continue to ask for repetition until they think they have understood what
was said. Above all, such activities provide useful insights for both speaker and listener
as to which areas of their respective ILs provide the greatest obstactles to intelligibility in
ILT. Feedback should include a discussion between speaker and listener(s) of who was
responsible for anything that was taken down wrongly, i.e. was it an error of production
or of reception? (See Gilbert 1993 and Rimmer 1997 for other suggestions for dictation
activities).

Receptive competence in ILT is gained through exposure to a wide range of varieties of

accent so that learners develop the ability to interpret other pronunciations than those of
their teachers and co-L1 speakers. Where teaching takes place in multilingual
environments, the competence-through-exposure process will, to some extent, occur
naturally if slowly. But it is unlikely to go far enough in typical short courses. Published
materials are not essential - and as yet, there are, in any case, few recordings of ILT
available. However, learners can, themselves, be recorded speaking together, whether in
naturalistic conversations, interview-type situations, or task-based interactions such as
those in the CAE Speaking examination. The recordings then be exploited as material for
listening texts in the same way that NS-NS interactions are used for listening skills work
in ELT course books.

In addition to such 'covert' exposure to one another's accents, learners can usefully
engage in contrastive work to make their pronunciation differences explicit. By this, I
mean that specific classroom work can focus on different IL accents by making
pronunciation a topic of classroom discussion. Such discussion could highlight learners'
pronunciation differences, and thus make explicit the more subconscious exposure to
different Il accents that occurs naturally in multilingual classrooms and, to some extent,
through listening materials in monolingual ones.

Allwright argues for a more profound pedagogic value of discussions of language itself
within a communicative model of language teaching (though pointing out that teachers
 90

have not fully exploited their potential). He considers that they provide an opportunity for
enhanced learning, since "better understanding is likely to result if learners discuss their
learning and share various understandings... They may learn directly from each other, or,
more likely, they will learn from the very act of attempting to articulate their own
understanding" (1984:158). Thus, discussions of this nature would be of benefit not only
to learners' receptive phonological competence, but also to their overall second language
acquisition. Specific activities could be of the 'How do you pronounce x?' type, where 'x'
is anything from a single sound to a lexical phrase. The process requires handling with
tact, however, so that learners are not made to feel that value judgements are involved
and are aware that in many cases (the non-core areas), their pronunciation differences do
not involve errors, but are perfectly acceptable L2 variants on a par with L1 regional
differences.

In the previous paragraph I referred to a classroom activity that was beneficial not only
for phonological accommodation, but also for SLA in general. In the 'Summary of
Teacher-led and ILT interaction' above, I have suggested that ILT is beneficial to SLA by
providing learners with more comprehensible input, which then becomes intake. This
relates to a body of research produced in the 1980s showing that the speech adjustments
resulting from negotiation in pair work and small group work indirectly benefit the
receiver's language acquisition by virtue of a link between the adjustments made in such
work and the comprehensible input necessary for acquisition. See, for example, Larsen-
Freeman 1985, Long 1985, Long & Porter 1985, and Pica & Doughty 1985. Larsen-
Freeman (op.cit.) argues that peer input is often more accessible to learners than NS
input. In classrooms, apparently 'inaccessible' teacher input may reflect a reluctance on
the part of the learner to negotiate with the teacher because of the power differential,
particularly in whole class situations, where there is the added fear of appearing foolish in
front of the peer group (see Pica & Doughty op.cit; Rulon & McCreary 1986).

Doughty & Pica (1986) claim that the significance lies not in group work itself but in the
nature of the tasks that are carried out within the group framework. Information gap
tasks, particularly two-way tasks, they contend (as do Gass & Varonis: see p.182),
promote far more negotiation arising from comprehensibility problems than do other task
types. The manipulation of output in order to restore comprehensibility is thought to
result in optimally adjusted input for the receiver. Such negotiation of meaning is held to
occur most extensively in groups of mixed-L1 backgrounds, and it is thus ILT that
potentially offers the greatest benefits to SLA for both speaker and receiver: while the
speaker is motivated to adjust his speech and, in particular, to replace his L1 phonological
transfer in core areas, the hearer (as a direct consequence of such adjustments) is
provided with comprehensible input.
 91

One major problem in all that I have said above is that it depends to a large extent on the
availability of ILT classroom interaction and this, by definition, is impossible to organise
in monolingual classrooms. However, more English teaching is carried out around the
world in monolingual than in multilingual classes, and here, as my same-L1 pairs data
demonstrates, L1 transfer does not reduce phonological intelligibility but actually
increases it. Thus, in monolingual classrooms where students are being prepared for EIL,
it is necessary for teachers to spend some time initially on helping students to adjust their
perceptions so that they are aware of the extent of the problem. Although not ideal, this
can be achieved to some extent by playing tapes of L2 English speakers from a variety of
L1s.

The fact that in ILT there is a need for speakers to make phonological adjustments in core
items in the direction of more target-like pronunciation and that the opposite is true of
same-L1 interaction in English, may explain why the accents of learners in monolingual
classrooms tend to fossilize earlier in their development than those of learners in
multilingual classrooms. Apart from the fact that they are exposed only to same-L1
accents and, perhaps, those of 'native speaker' teachers, learners are not motivated to
replace their pronunciation transfers in order to be understood. Allwright refers to the
dangers of "classroom pidgins" developing in homogeneous groups of learners
(1979:178), while Long & Porter (1985) argue that learners in groups composed of
mixed-L1 backgrounds avoid the monolingual group problem of the development of
classroom dialects intelligible only to speakers from the same first language. Bygate
maintains that monolingual groupwork "at least allows and at worst encourages
fossilization and the use of deviant L2 forms" (1988:76-77). Similarly, Aston (1986)
refers to the negotiation of meaning by learners from a monolingual background as
leading to the negotiated acceptance of non-standard forms. The situation is summed up
in the following grid:

Model of predicted phonological outcomes in ILT and same-L1 talk

________________________________________________________________________________________

ILT -> convergence for -> replacement of -> interlocutor comprehension

communicative L1 transfer and enhanced L2 phonological
efficiency acquisition
________________________________________________________________________________________

Same-L1 -> convergence for -> non-replacement -> interlocutor comprehension

talk communicative of L1 transfer and fossilization of IL
efficiency and phonology
solidarity
_________________________________________________________________________________________
 92

One final point that needs making with regard to accommodation in EIL classrooms
relates to the optimum group size. Pica & Doughty argue that because two-way tasks are
likely to be most effective when only two participants are involved, "pair rather than
group work on two-way tasks may ultimately be most conducive to second language
acquisition" (1985:132). Our interest is not so much with SLA - however beneficial a
'side effect' this constitutes - but with developing learners' productive and receptive
accommodation skills for the purposes of international communication in English.
Nevertheless, for different reasons, I have come to the same conclusion as to pairwork
being preferable to group work. One reason is that it is obviously a more complex matter
to converge (albeit on a more target-like pronunciation) with more than one interlocutor.
In my larger group data, I consistently found that speakers seemed to be unable to adjust
their pronunciation. Quite possibly it seemed to difficult a task to approach and they did
not even make the attempt. In a larger group it is, of course, quite possible to abdicate all
responsibility and leave others to do the speaking, and this was always the case with one
or two learners in any group.

However, there is a much more problematic issue at stake here, and this relates to groups
where there is more than one representative of an L1 present. Here, the presence of a
same-L1 interlocutor seems to mutually reinforce their L1 identities. This, in turn, leads
to phonological convergence which, as we noted earlier, at first takes the form of
subjective convergence, but gradually becomes objective convergence as the interaction
progresses. And identity is not the only factor involved: embarrassment also plays a part.
Many of my subjects admitted to feelings of embarrassment in situations where they had
to speak English with members of their own L1 group. They found it unnatural to use a
lingua franca when there was no authentic need for them to do so and therefore increased
L1 phonological transfer not only for the sake of mutual comprehension and the
expression of group identity, but also as a response to their embarrassment.

Obviously much thought will have to be given to the problem of accommodation in

groups containing members of the same L1. This is a reasonably frequent situation
internationally, and the EIL enterprise will to some degree be threatened if research is
unable to identify and pedagogy to implement a solution. Somehow the feelings of group
identity that remain grounded in the same-L1 contingent within a larger EIL group need
to be encouraged to shift outwards, to identify with the EIL group as a whole.
Phonological accommodation should then follow automatically, since the primary
motivation will become EIL intelligibility rather than L1 identity. For the time being,
however, it seems that different-L1 pairs provide the best context for the classroom
development of EIL accommodation skills.
 93

At this point, we have come full circle in that ILT is both problematic discourse and the
solution to the problem, to the extent that the difficulties encountered by speakers
engaged in ILT can be solved by appropriate pedagogy in multilingual classrooms. There
are still a number of unanswered questions, some of which will be dealt with in the final
chapter. Others, such as the problems in relation to monolingual classrooms and larger
groups will remain for another day. But whatever developments take place in the next
few years, one thing is almost guaranteed: that accommodation will play a major role in
international uses of English. As Nelson says, "the community of speakers will by sheer
numbers and geographical distribution require active accommodation from all
participants to retain a high degree of intelligibility across varieties" (1992:337). The
need to eschew the "dysfunctional" linguistic consistency to which Milroy refers (p.164
above) is as critical in the EIL speech community as in any other, and probably more so
as far as pronunciation is concerned. NBESs must be prepared both to cope with major
pronunciation differences in the speech of their different-L1 partners and to adjust their
own pronunciation radically for the benefit of their different-L1 hearers. And this sort of
preparation can only be achieved through pedagogy.
 94

View publication stats