Abstract
If EIL is to be regarded as an efficient medium for communicating information to all interlocutors from widely diverging language backgrounds, rather than as a means of emphasising speakers’ own linguistic identities, this should be reflected in EIL pronunciation norms. The need to preserve intelligibility is, of course, accepted even by those who see EIL fundamentally as a medium for communication between non-natives (Jenkins, 2006). If, like Jenkins, one then goes on to exclude native speakers when setting EIL standards, it is clearly difficult to define intelligibility in terms of the widely variant standards of non-native speech while eschewing recourse to native-speaker English. This could be why, despite its explicitly non-native orientation, Jenkins’s EIL pronunciation model is essentially based on British RP and GA (Network American). Nevertheless, Jenkins eliminates certain phonological features of these two accents from her model, purportedly since these are either irrelevant, or ‘unteachable’, or because they are considered to be lacking in a number of native Englishes.
In fact, many non-natives might object to being denied access to a type of English that can be used easily with non-native and native speakers alike. Furthermore, a recent study (Van den Doel, 2006), based on a large-scale survey of well over 500 native-speaker judges from throughout the English-speaking world, shows that if students of English were indeed to follow Jenkins’s recommendations, their resultant pronunciation could expose serious issues of intelligibility and acceptability. This paper will examine these findings and enlarge upon them, suggesting that (1) while intelligibility is obviously of prime significance for native speakers, they also regard acceptability as a major concern; (2) pronunciation features in non-native speech are likely to be notably downgraded when they are also stigmatised in local native Englishes. These findings indicate limitations in Jenkins’s current position, and suggest that a truly international English pronunciation model requires account to be taken of the attitudes and needs of native speakers.
Key words: English as an International Language, EIL pronunciation norms, Jenkins’s EIL pronunciation model
In discussions about the status of English as an international language, it is frequently claimed that most communication in English does not involve any native speakers. While this may be disputed (cf. Trudgill, 2005b, pp. 77–78 and Bruthiaux, 2003, p. 168), the idea of non-native communication in English outstripping interaction with native has given rise to the notion of English as a Lingua Franca (ELF). Jennifer Jenkins (2000) and her followers assert that in communication between non-native speakers, the English used is a clearly definable and describable entity in its own right.
As has been noted before (see Gibbon, 2005; Scheuer, 2005; Trudgill, 2005b; Wells, 2005), there are a number of paradoxical features about ELF. ELF is supposedly based on description of non-native interaction, but it also implicitly prescribes to non-natives how this interaction should take place. Moreover, the model sets out to ignore the concerns of native speakers, yet it is essentially “grounded”, by Jenkins’s (2000, p. 131) own admission, in native-speaker varieties. Finally, despite the non-native emphasis, it is ironical that some of its most famous supporters have been native speakers. Is it possibly an attempt on the part of English-speaking Western educators to control “what sort of English people should speak” – as Holliday (2005, p. 164) seems to imply?
In fact, there is a growing body of evidence (cf. Major et al., 2002; Major et al., 2005; Scheuer, 2005; Van den Doel, 2006) which suggests that a great many learners of English may be biased against non-native English and consequently do not appreciate being taught non-native models. Holliday (2005, p. 9) quotes a Taiwanese teacher, Kuo, who makes the following point:
Although I did feel comfortable to be told that I did not have to be native-speaker like, I would definitely feel upset if I could not reach my own expectation in pronunciation. … If we take Jenkins’s view and tell them to stay where you are …. At some point we would terribly upset the learners because they might want to … It’s been clear that I’m a language learner from the periphery and – listen to this – I prefer to speak for myself!
This non-native resistance presents a challenge to lingua franca researchers, some of whom have claimed that learners following native-speaker models are no different from people who want “to change the colour of their skin, the straightness of their hair, or the shape of their eyes to conform to other groups” (Cook, 1999, p. 196, as quoted in Jenkins 2006, p. 154). Non-native speakers (like me) are not necessarily motivated by linguistic insecurity; nor does choosing a native-like model of English obliterate speakers’ L1 identities, pace Jenkins (2000, p. 211). For most learners of English, their L1 is not an endangered heritage language (like Welsh or Basque) which they need to protect against the onslaught of English, but a system in which they habitually express an identity of their own outside of English. While Jenkins ascribes the wish to sound more native-like to the pressures of a native-speakerist language ideology (2006, pp. 154–155), she ignores the fact that non-natives may just want to make themselves as clear as possible not only to native speakers, but also to those millions of non-natives who have made the same choice.
For English as an International Language to function as an efficient medium, it would be advisable to take a broad rather than a narrow view of intelligibility. This means that EIL speakers attempt to make themselves understood not only to other non-natives, but also to native speakers and also to those non-natives who favour a native-like model. This clearly goes beyond teaching non-natives to sound non-native (a target which, by definition, they must surely already have attained). The extent to which worldwide intelligibility can be achieved depends on the needs, attitudes and demographic profile of the individual learner; however, it cannot be stressed enough that pursuing this goal is certainly in the learner’s own interest. Speech perception research shows (as cited in Trudgill, 2005a, p. 219) that non-natives find it harder than natives to understand other speakers of English – especially non-native speech containing far less of the crucial phonological information. Native speakers (like Jenkins) are better able to use contextual information, whereas non-native speakers of English (like me) find it tougher to process another speaker merging minimal pairs. When Dutch businessmen talk about their earning their celery rather than their salary, this may be harder for Japanese non-natives than for Americans, whereas I myself find it more difficult to deal with the confusion of pork and fork in Korean English.
Thus, learning to pronounce all crucial English phoneme contrasts is essential for improving the learner’s ability to understand other speakers. Pitted against such a broad view of International English is a narrower view, which concentrates on non-native interaction only, and ignores the gravitational pull of native-speaker varieties. This kind of English as a Lingua Franca can afford to disregard native speakers, because it is hoped that, once it has drawn in enough adherents, the native speakers will simply follow suit and learn this slimmed-down version themselves. To quote Jenkins (2000, p. 227): “The perhaps unpalatable truth for ‘N[ative] S[peaker]s’ is that if they wish to participate in international communication in the 21st Century, they too will have to learn EIL.”
It is actually unclear why native speakers have to acquire a model which is, in fact, grounded in native-speaker speech. After all, if any of Jenkins’s recommendations lead to increased intelligibility among non-native speakers, this is because many of the features of the Lingua Franca core are derived from native-speaker models. One example is Jenkins’s recommendation to preserve most English consonant sounds (Jenkins, 2000, p. 132). Doubtless this improves intelligibility – but some of the details are unclear, like the questionable insistence on aspiration of initial fortis plosives. Jenkins (2000, p. 140) claims incorrectly that it is “particularly important” for non-proficient non-native speakers to distinguish between [pæt] and [phæt], but this is simply not true. This surely holds only true for those whose L1s employ aspiration as an acoustic cue – speakers of Mandarin Chinese, for instance, as opposed to speakers of Malay, Indonesian, and the Dravidian languages of Southern Asia (Narasimhan, 2001, p. 245).
Certain other features recommended for exclusion are actually far more important, either because they could lead to phoneme confusion or because they are subject to stigmatisation. Examples include the dental fricatives /θ ð/, which Jenkins excludes from the Lingua France core, recommending their replacement by /f v/ or /t d/. As far as vowel sounds are concerned, the Lingua Franca core allows “L2 regional qualities ... if consistent” (Jenkins, 2000, p. 159), provided distinctions of vowel length are maintained. This disregards the potentially devastating effect of such common and consistent substitutions as the /æ/ vowel in trap by the /e/ vowel in dress, found in countless varieties of L2 English stretching all the way from Korea to Turkey and beyond. Other examples of non-native phoneme confusion sanctioned in the LFC are the failure to distinguish strut-lot and foot-goose. (See Wells, 1982, pp. 118–124 for these reference vowels.) These are certainly consistent “regional L2 qualities” in many parts of Europe.
An important rationale, Jenkins (2000, p. 27) suggests, for including or excluding certain features from the Lingua Franca core is whether or not these are also to be found in certain local native varieties of English. If, for instance, certain phonemic realisations also occur in local native varieties, it would seem “unreasonable”, as Jenkins (2000, p. 139) puts it, “to have ‘higher’ expectations of L2 speakers”. Nevertheless, if native-speaker norms were actually completely irrelevant, it should not matter whether some non-native variation is similar to native variation. Additionally, some of those features may be subject to considerable stigmatisation, not just from speakers of other varieties, but surprisingly also from those who are likely to use the stigmatised forms themselves. A good example is the use of schwa epenthesis, resulting in film being pronounced fillum. My recent research (Van den Doel, 2006) shows that, while this feature is characteristically attributed to speakers of Irish English, it is fact strongly stigmatised even in the Irish Republic itself.
A curious suggestion of Jenkins’s is to give low priority given to weak forms, that is, to allow /tə/ (‘to’) to be pronounced as /tuː/ (‘to’), even in unstressed positions, as in: Are you travelling to Korea? She claims, incorrectly, that weak forms are also absent in the speech of native speakers from as far afield as South Africa and Scotland (Jenkins, 2000, pp. 146–147; cf. Trudgill, 2005b, p. 82). Even if this were the case, a strong realisation of to would still run the risk, in this context, of being interpreted contrastively.
While Jenkins may invoke empirical evidence to show that her recommendations for the Lingua Franca core nonetheless help to streamline non-native interaction, my own research (Van den Doel, 2006) proves that certain of these actually impede interaction with native speakers - even though such interaction is, in my view, an important aspect of any communication in English as an International Language.
My study employed a large-scale Internet survey to investigate priorities in English pronunciation training. Well over 500 native speakers from throughout the English-speaking world, including North America, the British Isles, Australia and New Zealand, were asked to detect and evaluate non-native pronunciation features in sentences read by bilingual actors. Respondents were free to choose which of the two models of English supplied (RP and GA) they felt most competent to judge, and were subsequently provided with 32 recordings of Dutch pronunciation errors in what was otherwise native-sounding English. Respondents were able to access each recording from a separate Internet page, presented to them in random order, from which they could listen to the recording and evaluate the error. A sample page is provided in Fig. 1. By clicking on hyperlinked letters or letter combinations, judges could select the phoneme which they thought contained the error, without needing any knowledge of phonetics or linguistics. There was also a facility to indicate suprasegmental problems. Respondents were then required to indicate its gravity, before being allowed to go on to the next Web page. They were also encouraged to provide additional comments. Thus, for every pronunciation feature it was possible to record (a) whether or not it had been detected and (b) how it had been assessed.
A number of interesting results emerged. Firstly, it turned out that participants’ success in detecting an error is not necessarily linked to their assessment of its severity. For instance, British, Irish and Australian judges report more errors than North Americans, but diagnose fewer of those as serious. Conversely, North American respondents detect fewer errors, but evaluate those errors as being more serious. This was noticeably true of the replacement of dental fricatives by dental stops, which was considerably more stigmatised by North Americans than by other natives. Secondly, it was possible to construct a hierarchy of error for all groups by collating the data on detection and evaluation. This revealed that stress errors were generally considered by far the most significant.
Figure 1. A sample page from the Internet survey held by Van den Doel (2006).
The results also suggested that, while native speakers tended to prioritise errors that demonstrably impeded intelligibility, in some cases they also attached great importance to those errors which did not lead to an actual breakdown in communication but nevertheless caused irritation or amusement. The fact that many participants volunteered additional information on the errors in question reinforces the notion that such foreign pronunciation errors are severely stigmatised, especially if they were similar to pronunciations heard in native-speaker varieties. This suggests that, where non-native realisations are “on a par” with native realisations (as Jenkins, 2000, p. 27 herself puts it), this may reinforce rather than reduce the negative evaluations associated with them. An informal assessment of the severity of schwa epenthesis as an error by Irish respondents is provided in Figs. 2 and 3 (cf. Van den Doel, 2006, pp. 182–183).
Figure 3. Negative comments on schwa epenthesis volunteered by respondents other than from Northern Ireland and the Irish Republic.
A comparison of Jenkins’s recommendations with the effect they may have on native-speaker audiences is certainly instructive. If the pronunciation features discussed earlier are listed in a native-speaker hierarchy of error, there are certain striking differences, as may be seen in Fig. 4. Interestingly, th-substitutions tend to rank among the more serious errors, especially those involving /θ/ in thin and author – but with the notable exception of /ð ~ d/ confusion in the high-frequency function word that, where it was often not detected. This clearly suggests that stop realisations of /θ/ and to a lesser extent /ð/ are evaluated negatively by native speakers, especially, as it turns out, by North Americans. Vowel errors tend also to rank highly, especially the widespread confusion of trap and dress. Note how the apparently trivial error of schwa epenthesis greatly outranks lack of aspiration (considered a vital error by Jenkins) and also the error in vowel length. It should be realised, however, that it was mainly the Americans and Canadians who considered aspiration insignificant.
Another interesting result is the disparity between the two different cases of weak form avoidance. In the second instance, the strong form cannot be taken contrastively, whereas this interpretation is more likely in the first instance.
(1) We’re going to Wales for a long relaxing holiday.
(2) They all said that that may be done very differently.
This suggests giving a higher priority to the role of weak and strong forms in EIL, particularly in those instances where this affects stress patterning. Finally, the very low priority given to overdark /l/ should be noted. This was in fact entirely in keeping with Jenkins’s recommendations – although, again, this error was considered much more serious by North Americans.
Figure 4. Bar chart with error bars showing overall error severity and standard errors for a selected number of tokens.
There seems to be a noticeable British bias to Jenkins’s English as a Lingua Franca – especially her prioritisation of th-stopping, aspiration and overdark-l. This would mean that the priorities given to these features by native-speaker judges from Britain have been given undue significance. Any such bias would be most significant, especially in view of Holliday’s (2005, p. 164) reference to the Lingua Franca project as being seen as a plot from the centre “to control what sort of English people should speak” – in this case a variety that is acceptable in the main to British judges. In any event, it is clear that Jenkins’s prioritisations do not factor in the attitudes which different groups of native speakers may have to potentially stigmatised pronunciation features.
Setting the standards for a truly international English model should not be based on local or parochial concerns, no matter whether these are native or non-native, but on a broad view of communication that takes in the needs of both L1 and L2 speakers, and on the widest possible range of communicative situations. This ambitious objective can only be realised if a pronunciation model is adopted which results in maximum intelligibility and acceptability worldwide. The results of my investigation suggest that Jenkins’s Lingua Franca falls short of this ideal.
References
Bruthiaux, P. (2003). Squaring the circles: issues in modeling English worldwide. International Journal of Applied Linguistics, 13, 159-178.
Cook, V. (1999). Going beyond the native speaker in language teaching. TESOL Quarterly, 33, 185-209.
Dziubalska-Kołaczyk, K. & Przedlacka, J. (Eds.) (2005). English pronunciation models: A changing scene. Berne: Peter Lang.
Holliday, A. (2005). The struggle to teach English as an international language. Oxford: Oxford University Press.
Jenkins, J. (2000). The phonology of English as an international language: new models, new norms, new goals. Oxford: Oxford University Press.
Jenkins, J. (2006). Points of view and blind spots: ELF and SLA. International Journal of Applied Linguistics 16, 137-162.
Major, R., Fitzmaurice, S. F., Bunta, F. & Balasubramanian, C. (2002). The effects of nonnative accents on listening comprehension: Implications for ESL assessment. TESOL Quarterly, 36, 173-190.
Major, R., Fitzmaurice, S. F., Bunta, F. & Balasubramanian, C. (2005). Testing the effects of regional, ethnic, and international dialects of English on listening comprehension. Language Learning, 55, 37-69.
Narasimhan, S. (2001). Speakers of Dravidian languages: Tamil, Malayalam, Kannada, Telugu. In M. Swan & B. Smith (Eds.), Learner English: A teacher’s guide to interference and other problems (pp.244-250). Cambridge: Cambridge University Press.
Scheuer, S. (2005). Why native speakers are (still) relevant. In K. Dziubalska-Kołaczyk & J. Przedlacka (Eds.), (pp. 111-130).
Trudgill, P. (2005a). Finding the speaker-listener equilibrium: segmental phonological models in EFL. In K. Dziubalska-Kołaczyk & J. Przedlacka (Eds.), (pp. 213-228).
Trudgill, P. (2005b). Native speaker segmental phonological models. In K. Dziubalska-Kołaczyk & J. Przedlacka (Eds.), (pp. 77-98).
Van den Doel, R. (2006). How friendly are the natives? An evaluation of native-speaker judgements of foreign-accented British and American English. Utrecht: LOT.
Wells, J. C. (1982). Accents of English. Cambridge: Cambridge University Press.
