ASA 125th Meeting Ottawa 1993 May

4aSP1. Speech perception as a talker-contingent process.

Lynne C. Nygaard

Mitchell S. Sommers

David B. Pisoni

Speech Res. Lab., Dept. of Psychol., Indiana Univ., Bloomington, IN 47405

The present study reports an experiment designed to investigate the nature of perceptual adaptation and memory representation for spoken words produced by familiar and unfamiliar talkers. To determine how familiarity with a talker's voice affects perception of spoken words, two groups of subjects were trained to recognize the names of ten voices (five male; five female) over a 9-day training period. One group of subjects then identified words presented at four signal-to-noise ratios that were produced by the same set of talkers that they had learned to recognize during training. Control subjects identified the same words at the same signal-to-noise ratios but the words were produced by a set of new talkers that these subjects had not heard during training. The results showed that the ability to explicitly identify a talker's voice improved intelligibility of novel words produced by the same talkers. Subjects who heard familiar voices in the word intelligibility task were better at identifying novel words in noise than control subjects who heard unfamiliar voices. The results suggest that speech perception may be a talker-contingent process whereby familiarity with aspects of the talker's vocal source facilitates the subsequent phonetic analysis of the acoustic signal. [Work supported by NIH grant to Indiana University.]