Perceptual training on Hindi dental and retroflex consonants by native English and Japanese speakers

Abstract

Previous research has shown that both native English and native Japanese speakers have great difficulty distinguishing the dental and retroflex stop-consonants of the Hindi language. However, Japanese speakers perform much better than English speakers—particularly on voiced-unaspirated consonants. This differential performance is thought to be due to contrastive experience that Japanese have with /d/ versus flapped /r/. (While English does contain a flapped /r/, it is not contrasted with /d/.) This research further compared English and Japanese speakers’ ability to distinguish the Hindi contrast by providing laboratory training and generalization tests for both language groups. Twenty native speakers from each language group were pretested and post-tested on voiced-aspirated, voiced-unaspirated, voiceless-aspirated, and voiceless-unaspirated dental and retroflex consonant-vowel syllables produced by 2 native Hindi speakers in 3 vowel contexts (/a/, /e/, /o/). Training consisted of 12, 30-min sessions in which only voiced-aspirated tokens produced by 6 Hindi speakers in 2 vowel contexts (/a/ and /o/) were presented in a self-paced, self-selected fashion. While Japanese speakers performed at much higher levels than English speakers, both groups improved markedly and showed generalization to new speakers, new vowel contexts, and new voicing/manner contexts. [Work supported by NIDCD and Applied Telecommunications Research, Kyoto, Japan.]