Acoustic Measures of Temporal Intervals Across Speaking Rates

Abstract

It has been suggested previously that at least some levels of the temporal organization for speech production are characterized by proportional timing. The proportional timing model maintains that the duration of temporal intervals within a sequence would remain proportionally invariant across changes in overall duration of the sequence. In order to test this hypothesis for the acoustic level of speech production, 18 women produced three trials of the utterance " Buy Bobby a poppy " at each of three speaking rates (i.e., slow, normal, fast). Acoustically derived temporal intervals were paired to form ratios reflecting either syllablelevel or phrase-level relative timing. Findings indicated that ratios of temporal intervals at both the syllable-level and phrase-level did not remain invariant across speaking rates. Rather, statistically significant changes in the relative duration of both types of intervals were observed as a function of overall rate of production. For most of the obtained ratios, the direction of these changes was highly consistent across individual subjects.