Silence as a phonetic cue: Evidence from a study of duplex perception

Using a method contrived by Rand [J. Acoust. Soc. Am. 55, 678–680 (1974)], we earlier showed that the transition cues that distinguish [spa] and [sta] can be perceived simultaneously in two phenomenally different ways: as integral parts of those syllables, and as nonspeech chirps [D. Isenberg and A. M. Liberman, J. Acoust. Soc. Am. 65, S79(A) (1979)]. We further showed that although a silent interval (between the fricative noise and the vocalic portion of the syllables) was critical to the perception of the transitions when, on the speech side, they supported identification of the stop consonants ([p] and [t]), it had no measurable effect on the discrimination of those same transitions when, on the nonspeech side, they were heard as chirps. This result was taken to imply that the effect of silence is phonetic rather than auditory. This implication is, however, open to the reservation that different tasks were used on the two sides of the duplex percept: identification for speech, discrimination for nonspeech. The purpose of the present experiment is to speak to that reservation. Accordingly, a discrimination task was used on both sides of the duplex percept. Listeners were presented pairs of stimuli that comprised all combinations of the silence and formant transition cues. They were asked on each trial, to attend to either the speech or the nonspeech and to report their confidence that the members of a pair could be discriminated. The pattern of responses differed markedly for the speech and nonspeech sides of the duplex percept. It indicated that silence greatly affected discrimination of the speech but had little effect on nonspeech. This is taken as further evidence that the perceptual effect of silence as a cue for stop consonants is phonetic (rather than auditory).