The influence of sentence articulation rate on the internal structure of phonetic categories

Abstract

Numerous experiments have shown that listeners perceive phonetic distinctions in a rate-dependent manner. A case in point is the /b/–/p/ voicing distinction, specified by voice onset time (VOT). One standard rate effect is a shift in the/b/–/p/ boundary toward longer VOT values as speech becomes slower. Both a slowing of the target syllable itself and a slowing of the sentence containing the target syllable produce the effect. A recent investigation [J. L. Miller and L. E. Volaitis, Percept. Psychophys. 46, 505–512 (1989)] demonstrated that when speaking rate is specified by the target syllable, it alters not only the location of the category boundary, but also which stimuli within a category are judged to be the best exemplars of that category. In the present investigation, it was asked whether the rate at which a sentence is articulated can have the same comprehensive effect. A series of 215-ms syllables were created that ranged from /bi/ through /pi/ to a breathy, exaggerated version of /pi/; VOT varied from 10 to 205 ms. Each syllable was inserted into a fast and slow version of the sentence frame ‘‘She said she heard −− here.’’ A preliminary two-choice study (with VOT values between 0 and 60 ms), confirmed the standard effect of sentence rate at the /b/–/p/ category boundary. In the main experiment, listeners heard sentences of the same rate in random order. They rated each target syllable for its goodness as /pi/ on a scale of 1 to 10; the higher the number, the better the exemplar. It was expected that for each sentence rate, only a limited range of target stimuli would be given high ratings. At issue was the location of this range of best exemplars. If sentence rate, like target syllable rate, alters the internal structure of the category, then as the sentence becomes slower, the best-exemplar range should shift toward longer VOT values. This is precisely what happened. This finding underscores the comprehensive nature of the listener’s adjustment for speaking rate during phonetic perception. [Work supported by NIH.]