At its core, my dissertation addresses one primary question: What does it mean to sound like a lesbian? On the surface, this may seem a relatively simple question, but my work takes a broad perspective, approaching this single question from a multitude of perspectives. To do so I carried out a combination of production and perception experiments and interpreted the results through the exemplar-based Attention Weighted Schema Abstraction model that I developed.

Following the introduction Chapter 2 lays out the AWSA model in the context of previous literature on stereotype conception and speech and sexuality. Chapter 3 presents the production experiment, which recorded speakers reading a series of single words and sentences and interviews discussing stereotypes of sexuality. Phonetic analysis showed that though speech did not vary categorically by sexual orientation, familiarity with Queer culture played a significant role in variation of speech rate and mean pitch. This pattern was only seen for straight and bisexual speakers, suggesting that lesbian stereotypes are used to present an affinity with Queer culture, which was further supported by the decrease or loss of their significance in interview speech.

Chapter 4 describes three perception experiments. First, listeners judged speakers on attributes including likelihood of being a lesbian. Despite production results showing no variation by sexual orientation, listeners correctly identified lesbian speakers as more likely to be lesbians. Though not accurately predicting self-identification, four phonetic features contributed directly to lesbian ratings. A Principal Components Analysis of the ratings identified four recognizable social personae, including an ‘angry butch lesbian’ persona, showing a complex lesbian speech stereotype. The second experiment tested how listeners would judge sexuality without prompting. Listeners provided five adjectives to describe each speaker. Of 3480 adjectives, only one referenced sexual orientation. Finally, listeners judged only sexual orientation. Listeners were highly consistent in lesbian and straight ratings, but not at all for bisexual. In interpreting my results through the AWSA model, I posit a process of perception of lesbian speech, in which, rather than cueing orientation directly, phonetic features prompt perception of gender expression. Non-linguistic stereotypes then link these to sexual orientation.