Aspiration and Voice Onset Time in Word-Initial Consonant Clusters in English

Abstract

A list of 224 English words beginning with prestressed consonant clusters or single consonants was recorded in the frame “Say … instead” by three talkers. The timing of glottal gestures relative to supraglottal events was studied by spectrographic analysis of the data. Voice onset time (VOT) defined in this case by the onset of visible voicing striations in the spectrogram at frequencies above the first formant (relative to stop release) was measured for all words involving plosive consonants. In voiced plosives, this interval is occupied by frication noise of about 18-msec duration before a vowel and 25 msec before a sonorant consonant. In voiceless plosives, the frication noise burst is followed by an interval of aspiration noise. The sum of these two noise intervals, i.e., the VOT, was about 60 msec preceding vowels and 75 msec preceding sonorant consonants. If a voiceless plosive was preceded by [s], the VOT was much less, i.e., about 5 msec greater than the data for the analogous voiced plosive. The VOT was always about 10 msec less for labial plosives and 10 msec greater for velar plosives. The VOT was about 20% longer preceding the high vowels [i] and [u]. In [s]-sonorant clusters, an aspiration interval of 30–50 msec was observed following the disappearance of visible frication noise. [Work supported in part by National Institutes of Health and by Office of Naval Research.]