Sound-silence patterns and the perception of speech rate

Abstract

An earlier study [S. Feldstein, J. Acoust. Soc. Am. Suppl. 1 60, S46 (1976)] found that the sound-silence sequence of monologues yielded a measure of speech rate highly correlated with words-per-minute. The present exploration asked whether listeners use the sound-silence sequence in their perceptions of speech rate. Thirty 1-min speech segments were presented in three forms: as originally recorded; in pitch-scrambled form that kept stress and intensity contours intact but not words; in “tone-silence” form in which all speech sounds were changed to 1200 Hz tones and only the sound-silence pattern remained intact. Of the 30 segments, ten were spoken at an ordinary rate of speed, ten rapidly, and ten slowly. After each segment, listeners rated the speed at which it was spoken on a ten-point scale. Analyses indicated that the magnitudes of the mean ratings increased appropriately from “slow” to “ordinary” to “fast” segments for the three segment forms, but differences among the ratings were not significant for the “tone-silence” form. Apparently, listeners were unable to reliably use only sound-silence patterns to detect speech rate.