Perception of coarticulation with Japanese voiced and voiceless vowels

Abstract

Listen

Translate article

Two experiments investigate whether coarticulatory information of voiced and devoiced vowels in their preceding segments can be exploited by speakers of different dialects and/or by second language learners. In standard Tokyo Japanese, the high vowels [i, u] are devoiced between voiceless segments while vowel devoicing is not typical in Osaka dialect. This study examines how vowel’s voicing status affects listeners’ ability to exploit coarticulatory information in initial consonants. In both experiments, listeners consisted of native speakers of Tokyo and Osaka dialects and native speakers of English learning Japanese as a second language. In the first experiment, the post-alveolar fricative /(esh)/ followed by a vowel, either /i/ or /u/, was produced in real words by native speakers of Tokyo and Osaka dialects. The listeners were presented with a series of excised /(esh)/ to determine the original vowel identity. The second experiment replicated the first experiment testing with the velar stop /k/. The preliminary results suggest vowel devoicing may cause perceptual difficulties in the fricative-vowel context, but not in the stop-vowel context. The results also indicate a difference in cue weighting among 3 groups. Explanation for this finding will also be explored by way of acoustic measurements.