subject: The Intelligibility Filter [print this page] It is possible to evaluate the specific features of the Hong Kong accent by passing them through the 'intelligibility filter' provided by ELF research: this procedure resembles the technique outlined in Brown (1991), but with reference to empirical data on intelligibility. Of the features identified by Hung and Deterding et al., the following would at first sight seem to be unproblematic for international intelligibility: the dental fricative substitutions (usually [f] for [0] and [d] for [d]), dark [1] vocalisations and the merging of the [se] and [e] vowels. While I do not wish to examine in detail the Juicy Couture Necklaces arguments regarding the relative importance of these features, it is worth considering each of them briefly.
Dental fricative substitutions were also noted in the study of ASEAN English carried out by Deterding and Kirkpatrick (2006), and they too were found not to affect mutual intelligibility. They are also present in many LI English accents, as noted by Jenkins (2000, 2007). Although it is debatable how far LI accent features should be allowed to inform an assessment of international intelligibility -especially if it is accepted that the number of non-native speakers of English exceeds the number of native speakers it does seem that there is something in language universals that militates against an insistence on LI English norms. Features such as these are termed 'weak links' of the target language by Williams (1987), who believes they may be important factors in language acquisition and change. Psycholinguistic principles also suggest the relative unimportance of the dental fricatives, as the small number of minimal pairs involving these sounds results in a low probability of lexical confusion: 'in terms of minimal pairs, all pairings involving [6, d] are unimportant' (Brown, 1991:121).
Similar arguments can be raised in favour of ignoring dark [1] vocalisation, which also occurs frequently in LI accents (as well as in NVEs such as Singaporean English). Bauer (1986) found that in New Zealand English, vocalised [1] was so prevalent that many people could not make a dark [1] preconsonantally (in Shockey 2003:35). Interestingly, such substitutions may also be indications of language change on a wider scale. Shockey (2003:112) notes that modern Portuguese shows final [1] vocalisation to be 'a living process': Spanish mal, Portuguese mau.
Although merging the [ae] and [e] vowels would not appear to cause intelligibility problems according to Jenkins's Lingua Franca Core, this is a slightly more complex issue. Firstly, it is not entirely clear what Jenkins (2000:159) means Juicy Couture Watch by saying that 'L2 regional [vowel] qualities are permissible if consistent'. Of course, there is considerable variation in vowel quality across different speakers and different accents. But while individual vowel qualities may vary, what is important for international intelligibility is that most vowel contrasts are maintained. Thus while a speaker's actual vowel qualities will show some divergence from the idealized versions implied by a vowel quadrilateral, if they are merged then what Brown (1991:77) calls 'the maintenance of perceptual distance' will be lost. This will almost certainly affect intelligibility.
