Rediscovering auditory coherence in phonetic patterns
Deborah A. Gagnon, James R. Sawusch · The Journal of the Acoustical Society of America · 1990
It has been suggested that phonetic percepts are achieved by invoking the use of a specialized speech mode in processing the acoustic signal [A. M. Liberman and I. G. Mattingly, Cognition 21, 1–36 (1985)]. In support of this contention are demonstrations of differences between speech and nonspeech perception of the same set of stimuli. For example, subjects provided with speech labels (/ra/ and /la/) for categorizing a three-formant tone analog continuum that varied in only the F3 transition dimension demonstrated categorical perception, while those subjects provided with nonspeech labels did not [C. T. Best et al., Percept. Psychophys. 45, 237–250 (1989)]. However, an alternative explanation is that the nonspeech group's poor categorization performance may have resulted from a failure to provide adequate training. In the present study, providing more extensive training resulted in steep identification functions for both speech and nonspeech groups. In addition, peaked discrimination functions were found for nonspeech subjects. Indicating that the stimuli were being perceived categorically. The data support a methodological explanation of dissociations in categorization performance between speech and nonspeech instructed groups and suggest that a “speech is special” claim remains tentative. [Work supported by NIDCD Grant DC00219 to SUNY at Buffalo.]