Toscano, J. C., & McMurray, B. (2012, November). Paper presented at the 11th Auditory Perception, Cognition, and Action Meeting, Minneapolis, MN.

Abstract:  A number of researchers have become interested in the effects of indexical information on speech perception. A commonly studied indexical cue is talker gender, which produces systematic changes in acoustic cues that are relevant for speech. However, it has been difficult to manipulate listeners’ perception of gender, making controlled experiments examining its effects difficult to run. Here, we present a novel approach for manipulating the perceived gender of a talker’s voice that involves changing fundamental frequency, formant frequencies, and formant bandwidths. We generated stimuli using this approach and demonstrate that listeners correctly categorize the sounds at both endpoints of a male/female continuum, approximating the likelihood of male and female responses seen with natural stimuli. In contrast, a simpler method that involves only manipulating fundamental frequency does not produce correct responses at both ends of the continuum. We also used this approach to create stimuli that had an ambiguous gender and found that preceding talker context (i.e., whether the preceding sentence was spoken by a man or woman) had an effect on listeners’ categorization of the stimuli. Thus, the approach presented here provides a way for researchers to manipulate perceived talker gender in order to examine the effects of contextual variation on speech perception.