A Study Of Intelligibility Of Chengdu-Accented English At Word Level

Effects of non-native pronunciation features on communication are investigated quantitatively in the study. An intelligibility experiment was carried out to examine the intelligibility of Chengdu-accented English. In the experiment, 8 native English-speaking listeners and 8 Chinese college teachers of English were presented with recordings of 20 Semantically Unpredictable Sentences (SUS) and a diagnostic passage of strictly controlled context, spoken by 9 Chengdu speakers and 2 native English-speaking controls.Follow-up interviews were also carried out with the native listeners questing for their experience during the transcription. The interviews reveal that Chengdu speakers of low-level pronunciation are barely intelligible in the experimental condition, a low-level context. However, more than 80% of the words in the passage, according to the statistics afterwards, are recognized after repeated listening.A feature-based analysis of the recordings concerning the words unidentified by the native listeners is performed, from which an error inventory of 9 categories is compiled. Based on a thorough discussion of the research findings in relation to other researchers' proposals home and abroad, lists of "core" and "non-core" features affecting intelligibility are proposed for Chengdu speakers as the intelligibility of their English pronunciation is concerned. The deleterious errors are sorted by the speakers' intelligibility level, which produces another list of the errors sorted by fossilization degree.Sentence and passage intelligibility are scored as percent of words correctly transcribed by the native listeners and the non-native listeners after repeated listening. A series of statistic analyses are conducted afterwards, probing into the relationship between context and intelligibility, listener's familiarity to the speaker and intelligibility, as well as native listener's language background and intelligibility. The statistic analyses indicate that context affects the intelligibility of Chengdu-accented English significantly, but exerts very little influence upon the intelligibility of native English. In addition, the accented English under zero-context is more intelligible to the Chinese teachers who are supposed to be more familiar to the accented-English than to the native listeners. However, the native listeners can understand the accented English better in the discourse of a low-level context. Furthermore, it is found that the intelligibility of a same speaker vary with native listeners of different language backgrounds. The Americans and the British can understand the Chengdu-accented English better in the context-free SUS than the Australians do. However, the intelligibility difference becomes insignificant when the passage of low-level context is concerned.The study finds that most errors significant to intelligibility found among Chengdu speakers are shared by most Chinese students, which indicates that future studies in a national scale are likely to find a common pronunciation "core" that are significant to intelligibility for Chinese speakers.