In read sentences and paragraphs, as well as simulated man‐computer interactions, time intervals between onsets of stressed vowels (“disjunctures”) clustered near mean values around 0.4 to 0.5 sec, with standard deviations of about 0.15 sec. Contrary to published hypotheses, durations of disjunctures tended to increase about linearly with the number of intervening unstressed syllables. Mean
... [Show full abstract] disjuncture durations doubled when spanning clause boundaries, and tripled when spanning sentence boundaries. Mean pause durations. as measured by durations of unvoicing, tended to be equal to or twice the mean interstress interval, for clause and sentence boundaries, respectively. Syntactically dictated pauses thus appear to be one‐ or two‐unit interruptions of rhythm. Long disjunctures also accompanied some perceived boundaries between syntactic phrases, and were found useful in determining which of several minimally contrastive syntactic structures had been spoken. Mean disjuncture durations, adjusted to take account of the number of unstressed syllables between stresses, appear to correlate well with differences between phonemic, phonetic, and computer‐determined transcriptions of utterances, so that mean disjuncture durations, as acoustic measures of speech rate, might predict some phonological rules that apply to an utterance. [Research supported by the Advanced Research Projects Agency, Department of Defense.]