Perception of the duration of a silent interval in nonspeech stimuli: a test of the motor theory of speech perception.

Abstract

The categorical perception of synthetically produced speech stimuli varying along a single dimension (duration of silent interval) has been used as evidence for a motor theory of speech perception. The suggestion has been made that before language learning, discrimination along the silent-interval-duration continuum is comparable to discrimination along any unidimensional continuum, that is, 2.3 bits of information transmitted. To test this hypothesis, two groups of four subjects each listened to 1200 presentations of a set of 10 stimuli consisting of silent intervals of from 10 to 100 msec bounded by noise bursts. The subjects rated the stimuli on a 10-point scale of stimulus duration. Group I was instructed to make a judgment of long or short before rating. Group II was not instructed to make this categorical judgment. An informational analysis of the resulting confusion matrixes resulted in less than one bit of information transmitted for both groups. It is concluded that discrimination of nonspeech stimuli varying along the continuum of silent interval results in fewer (rather than more) discriminated categories than does the discrimination of speech stimuli varying along the same continuum.