How does a listener resolve disparate acoustic elements into a single perceptual stream? The evident coherence and the continuity of speech are a challenge to accounts of perceptual organization. Vocally produced sound is distributed across seven octaves, and the perceptual coherence of a speech stream therefore entails integration of spectral attributes spread over a wide frequency range.

Speech signal components that are desynchronized from the veridical temporal pattern lose intelligibility. In contrast, audiovisual presentations with large desynchrony in visible and audible speech streams are perceived without loss of integration. Under such conditions, the limit of desynchrony that permits audiovisual integration is also adaptable. A new project directly investigated the potential for adaptation to consistent desynchrony with unimodal auditory sine-wave speech. Listeners transcribed sentences that are highly intelligible, with veridical temporal properties. Desynchronized variants were created by leading or lagging the tone analog of the second formant relative to the rest of the tones composing the sentences, in 50-msec steps, ranging from 250-msec lead to 250-msec lag. In blocked trials, listeners only tolerated desynchronies <50 msec, and exhibited no gain in intelligibility to consistent desynchrony. Unimodal auditory and bimodal audiovisual forms of perceptual integration evidently exhibit different temporal characteristics, an indication of distinct perceptual functions.

Talkers differ along several dimensions that listeners can resolve. These include acoustic effects of native variation in the scale and shape of the articulatory anatomy, and the effects of age and use on the tissues of the vocal tract. Linguistically, talkers differ in phonetic habits occasioned by dialect and idiolect, and differ paralinguistically in manner of affective expression conveyed vocally. However, listeners are themselves likely to vary in sensitivity to intertalker variation. In this study we aimed to identify differences in sensitivity to talker variation as a function of linguistic experience. Speech samples were produced by female talkers 15–17 years old drawn from two dialect groups, one from Brooklyn, NY, and one from Bloomington, IN. Each talker produced sentences in a list-reading task. Listeners in our tests were also native either to Brooklyn or to Bloomington, and each was far more familiar with one dialect than the other. Tests of apparent similarity of talkers in each set were conducted with listeners from the same and different dialect group. The results of similarity scaling analyses calibrate the contribution of sensitivity to idiolectal contrast within and across dialect in the perception of a talker’s characteristics. [Research supported by NIDCD.]

Speech signal components that are desynchronized from the veridical temporal pattern lose intelligibility. In contrast, audiovisual presentations with large desynchrony in visible and audible speech streams are perceived without loss of integration. Under such conditions, the limit of desynchrony that permits audiovisual integration is also adaptable. A new project directly investigated the potential for adaptation to consistent desynchrony with unimodal auditory sine-wave speech. Listeners transcribed sentences that are highly intelligible, with veridical temporal properties. Desynchronized variants were created by leading or lagging the tone analog of the second formant relative to the rest of the tones composing the sentences, in 50-msec steps, ranging from 250-msec lead to 250-msec lag. In blocked trials, listeners only tolerated desynchronies <50 msec, and exhibited no gain in intelligibility to consistent desynchrony. Unimodal auditory and bimodal audiovisual forms of perceptual integration evidently exhibit different temporal characteristics, an indication of distinct perceptual functions.