Research

Ohio State Dissertations in Linguistics (OSDL)

Elizabeth Strand (2000)

Gender Stereotype Effects in Speech Processing

Advisor: Keith Johnson

Download .pdf file [PDF]

Abstract:

In this dissertation, I investigate the influence of gender stereotypes on low-level speech processing. Based on the notion that stereotypes serve as cognitive shortcuts in cognitive processing (i.e., stereotypical things are processed faster), I test both auditory-only and audiovisual speech perception to determine whether speech produced by speakers with stereotypical voices and faces is processed faster than speech produced by speakers with nonstereotypical voices and faces. My results indicate that there is indeed a processing deficit associated with perceiving speech from nonstereotypical voices and faces, and thatgender information does, therefore, influence speech perception.

The first four experiments assessed the levels of stereotypicality of a set of voices and faces, such that a number of faces and voices could be identified for use in stimuli construction for a set of speech perception experiments. We identified two stereotypical voices (male and female), two nonstereotypical voices (male and female), two stereotypical faces (male and female) and two nonstereotypical faces (male and female). To select these voices and faces, we combined measures from multidimensional scaling solutions for voice and face similarity ratings with accuracy and reaction times from speeded voice and face gender classification tasks. Combining results from these two methods lets us assess notions about gender stereotypes implicitly, rather than relying on more explicit measures as has been done in thepast.

These faces and voices were then used in two speech perception experiments. Participants were presented with words uttered by various speakers, and were asked to repeat the targets as quickly as possible in an auditory naming task paradigm. Response time was measured. In one set of trials, participants were presented with the words only, while in a second set of trials, each word was preceded by a prime in the form of a picture of a face of the supposed speaker of the word. Levels of stereotypicality as well as gender of the faces and voices were mixed and matched such that audiovisual pairings of all possible crosses of face and voice gender, and face and voice stereotypicality,were presented to participants.

Reaction time results indicated that there is a significant effect of voice stereotypicality on reaction times, such that words uttered by speakers with stereotypical voices are faster to be processed and repeated in the speeded auditory naming task. Results from the face-primed auditory naming task replicated the voice stereotypicality effect, and also showed that for female voices, there is a significant priming effect of face stereotypicality on the processing and repetition of the target words. Results also indicated an interaction of face gender and voice stereotypicality in perception of the face-primed female voices. Interestingly, however, we found a null effect of face priming for the male voices. Our results indicated that at least for some voices, gendered face information significantly facilitates the processing of speech information.

These results support the hypothesis that gender stereotypes do influence speech processing, and that they do it very automatically at a low level of cognition.