Second International Conference on Spoken Language Processing (ICSLP'92)
Banff, Alberta, Canada
A joint Japanese and Dutch experiment studied the effect of language on the perceptual contribution of temporally distributed cues (i.e. VC transitions and CV transitions) in identifying intervocalic voiceless stop consonants. Japanese and Dutch listeners identified Japanese stop consonants from isolated VCV, CV, VC, or VC1-C2V syllables (all extracted from natural Japanese VCV utterances) under varying noise conditions, in which up to 70 ms of the release burst and the vocalic transition are replaced by noise. The two language groups perform very similar for the VCV and CV stimuli; the average consonant identification scores of both groups are higher for the VCV stimuli than for the CV stimuli when the CV cues are largely eliminated by noise replacement. The difference between the results of the two types of stimuli indicates that Japanese as well as Dutch listeners can use the VC cues (present in VCV stimuli) in identifying the stop consonants. However, the identification scores for the VC and VC1-C2V stimuli are lower for Japanese than for Dutch listeners, suggesting that the phonotactic system influences identification. Due to the absence of syllable-final stop consonants in the Japanese language, Japanese listeners are less inclined to use the VC transitions as cues to a syllable-final stop consonant than Dutch listeners.
Bibliographic reference. Kashino, Makio / Wieringen, Astrid van / Pols, Louis C. W. (1992): "Cross-languages differences in the identification of intervocalic stop consonants by Japanese and dutch listeners", In ICSLP-1992, 1079-1082.