Cross-contextual consistency of /s/ length and spectral quality in gay men’s speech

Abstract

Previous literature on gay(-sounding) speech has shown that gay men tend to produce longer /s/ tokens with higher spectral centers of gravity (Munson & Babel (2010), Levon (2006, 2007), and Linville (1998)). The existing body of work, however, has only shown this difference using word-initial, non-cluster /s/ tokens or aggregate average values across all contexts. This study specifically investigates these questions in /sC/ clusters, both word-initally and -finally. Participants produced words in a carrier phrase containing either /s/ or an /s/-stop cluster, in either word-initial or word-final position (N = 16). To control for speech rate variation, /s/ tokens were measured with respect to participants’ syllable length. Preliminary findings show that gay speakers did in fact lengthen /s/ tokens in all contexts, at similar rates across cluster and singleton contexts. Significant differences between gay and straight men were found for both /s/-to-syllable ratio and spectral center of gravity at /s/ midpoints. These findings contribute to the depth of gay speech acoustics research and provide a basis for context-specific sociophonetic perceptual studies.