Perception of short-term spectral cues for stop consonant place by normal and hearing-impaired subjects.

The purpose of the experiments described in this paper was to investigate the use by normal and hearing-impaired subjects of the acoustic information at syllable onset for identification of the place feature of synthetic voiced stop consonants. Two synthetic speech continua served as stimuli. The full-length continuum comprised 14 syllables that varied from /ba/to/da/to/ga/. The 14 stimuli of the short-stimulus continuum were truncated versions of the full-length syllables; each contained a burst portion plus two glottal pulses. All normal subjects categorized the full-length stimuli accurately and consistently, but only two of seven normals performed equally well with the short-stimulus continuum. The short-stimulus performance of the remaining five normals improved after training, but remained significantly poorer than their performance with the full-length stimuli, indicating that the information at stimulus onset did not provide all the necessary place information for those subjects. Hearing-impaired subjects' identification of the full-length stimuli was only slightly less consistent than the normals', and was highly correlated with their identification of stop consonant place in natural syllables. Their post-training performance with the short stimuli did not differ from that of the trained normal group. These results indicate that various forms of signal distortion imposed by sensorineural hearing loss may, under certain conditions, have negligible effects on subjects' use of acoustic place cues when those cues are at suprathreshold levels.