Abstract:

This pilot study examines the influence of several factors on time intervals between syllables judged to be isochronous, including number of intervening segments and number of intervening syllables. Earlier work [Fant et al. (1991); Lehiste (1977)] has primarily been concerned with interstress intervals, rather than perceptual ``beats,'' though more recent work has examined beats per se [Couper--Kuhlen (1993)]. For our study, five labelers with musical backgrounds applied a system for labeling perceptual isochrony to 1 min each of spontaneous, read, and radio news style speech [Dilley, forthcoming]. For strings indicated as isochronous by three or more labelers we determined the intervals between the ``rhythmic beats,'' i.e., perceptually strong syllables which formed the perceived isochronous sequences. For simplicity, the ``beat length'' was taken to be the time between successive vowel midpoints (or syllabic sonorant segments) for these perceptually strong syllables. Preliminary results for 1.5 min of speech indicate that beat lengths range from 200 to 800 ms, and that there is considerable variability for successive beat lengths in a perceptually isochronous sequence. This variability can be reduced somewhat by adjusting for the number of intervening segments.