Revisiting the sociophonetics of sexuality in spontaneous speech

This study surveys sociophonetic variation in sibilant and vowel production for 44 Bay Area English speakers. I focus on (i) whether sexual orientation is an independent predictor, or interacts with gender, and (ii) what the patterning of bi + (bisexual, pansexual, queer, etc.) and non-binary speakers adds to previous claims (e.g., Gaudio 1994, Pierrehumbert 2004, Willis 2023). All participants completed a Map Task in pairs; 37 participated in a sociolinguistic interview. Recordings were auto-transcribed, forced-aligned, and hand-checked. For /s ʃ z ʒ/, tokens under 0.05 sec, in /str/ clusters, and/or adjacent to another sibilant were excluded. Before getting COG and skew, 0.02 sec were subtracted from token edges and voicing was Hann band-filtered. For all vowels and sibilants, mean and midpoint f0 were taken, and F1-F4 for vowels (using the median of values at midpoint plus six surrounding timepoints). Results show sexual orientation alone is not a robust predictor of /s/ frontedness. Both gender and sexuality interacted with duration for /s/ frontedness and vowel f0; gender and sexuality also interacted to influence /ʃ/ frontedness and /z/ devoicing, as well as vowel space shape. These results imply that, when bi + and non-binary identities are incorporated, the phonetic indexation of sexual identity is not easily generalizable within orientation labels, at least for informal speech.