Abstract:

Sawusch and Gagnon [J. Exp. Psychol., HPP 21, 635--652 (1995)] obtained categorical responses to /la/--/ra/ sine wave analogs when listeners who heard the stimuli as nonspeech were frequently reminded of F3's range or extremes. This result contrasts with responses to sine wave /la/--/ra/ obtained by Best et al. [Percept. Psychophys. 45, 237--250 (1989)] and responses to sine wave /wa/--/ja/ obtained by Johnson and Ralston [Phonetica 51, 195--209 (1994)]. Only those listeners who heard the analogs as (distorted) speech responded categorically. Instead of comparing speech and nonspeech listeners, the experiments tried to create an implicit expectation that sine wave analogs are (distorted) speech by playing listeners speech first. Listeners identified and discriminated members of sine wave /la/--/ra/ after identifying and discriminating either members of the original /la/--/ra/ speech continuum or of the same sine wave continuum. Other groups of listeners heard speech or sine wave /wa/--/ja/ before the sine wave /la/--/ra/; their performance allows us to separate acoustic from type similarity. If after hearing speech, listeners hear the analogues as implicit speech, they should perceive them more categorically than listeners who hear analogues first. Results show instead that listeners responses are as strongly categorical to the sine wave /la/--/ra/ in the second phase after all preceding stimulus types, even though the analogues are not heard as speech. Categorical perception thus does not depend on an implicit speech percept, nor are frequent reminders required to elicit it to sine wave analogues. [Work supported by NIH.]