Authentic self, incongruent acoustics : a corpus-based sociophonetic analysis of nonbinary speech.

This thesis examines the ways six nonbinary speakers in Christchurch, New Zealand present their gender identity via speech. It examines their productions in reference to both established trends in the literature, as well as speech collected from ten binary speakers (5M, 5F) at the same time. It seeks to examine whether, in addition to encoding binary gender, speech also encodes nonbinary gender. Three hypotheses are proposed and tested across multiple linguistic variables. The first hypothesis regards acoustic incongruence, and posits that nonbinary speakers may assert their nonbinary identities via speech that utilises particular combinations of variables which create either ambiguity or dissonance in regards to established binary-gender norms. Ambiguous gender incongruence arises from the use of speech that is neither reliably perceived as female, nor reliably perceived as male. Dissonant gender incongruence arises from the use of speech that is reliably perceived as both male and female. The second hypothesis predicts that nonbinary speakers will show greater variation in speech based on immediate contextual factors, compared to binary speakers. This difference is hypothesised to be due to to nonbinary speakers paying greater attention to production, and the greater degree of variation in their own speech over time compared to binary speakers. Hypothesis 3 predicts that nonbinary speakers are not a uniform population, and that their use of incongruence will be influenced extensively by their individual condition, including their professed speech goals, history, and gender identity. The hypotheses are tested quantitatively in regards to five linguistic variables: Pitch, pitch range, monophthong production, Vowel Space Area (VSA), and intervocalic /t/ frication rates. The interaction between multiple variables together is also considered. In-depth examinations of the variation utilised by a single speaker in the form of "Spotlights" address the hypotheses from a qualitative perspective. Overall, the thesis finds some evidence for Hypothesis 1. In every linguistic variable examined, nonbinary speakers show some distinction from binary speakers that is not explained fully via speaker Assigned Sex at Birth (ASAB). Some binary speakers also seem to produce incongruence, particularly binary women and particularly within single variables. The small scale of the study presents a limitation in addressing Hypothesis 2, but avenues for future work are identified. The qualitative evidence provides strong support for Hypothesis 3, in the examination of individual nonbinary speakers and the way their measured productions support their professed speech goals and identities. Overall, this dissertation presents one of the first comparative analyses of nonbinary speech, and presents a number of novel approaches to examining phonetic data from a statistical perspective that still accommodates an analysis of individual agency and goals in identity building