Relative amplitude of aspiration noise as a voicing cue for syllable-initial stop consonants.

Abstract

The present experiments demonstrate that amplitude of aspiration noise (relative to the following periodic portion of the vowel) is a cue for the distinction between voiced and voiceless syllable-initial stop consonants in English, and that it can be traded for voice onset time (VOT). In Experiment I, the category boundary on a synthetic VOT continuum (/da/-/ta/) was found to be a linear function of the amplitude ratio between the aspirated and unaspirated vocalic portions over a 24-dB range. A 1-dB increase in the ratio led to a shortening of the VOT boundary by 0.43 msec., on the average. In Experiment II, the synthetic stimuli were prefixed with a natural 10 msec. release burst, and burst and aspiration amplitudes were varied orthogonally. Both factors affected the voicing boundary in the expected direction but not independently; their interaction was ascribed to backward masking of weak bursts by strong aspiration noise. After accounting for this interaction, the effect of burst amplitude seemed very small compared to that of aspiration amplitude. These results suggest that the amount of aspiration noise is the primary voicing cue in the situation investigated, and that the perception of the noise follows psychoacoustic laws of time-intensity tradeoff. The methodological significance of amplitude parameters in speech synthesis is pointed out.