Gender, sexuality, and fricative variation in a post-spectral moments world

Spectral moments are widely used to characterize variation in fricative spectra (Forrest et al., J. Acoust. Soc. Am. 84, 115–123 [1988]). One piece of evidence for their utility comes from studies showing that spectral moments for the fricative /s/ predict listeners’ judgments of the gender and sexual orientation of the person talking (Munson et al., J. Phon 34, 202–240 [2006]; Munson, Lang. Speech 50, 125–142 [2007]). Recent work has shown that spectral moments are suboptimal, as each moment indexes multiple underlying articulatory parameters. Shadle recently proposed a series of alternative measures that are more closely related to articulation than spectral moments (J. Acoust. Soc. Am. 153, 1412–1426 [2023]). In this presentation, the analyses originally presented in Munson et al. [2006] and Munson [2007] are reanalyzed using Shadle’s proposed measures. Those studies examined the acoustic characteristics of /s/ in gay, lesbian and heterosexual men and women, and listener ratings of the talkers’ masculinity–femininity and sexual orientation. Preliminary results show that the relationship between fricative acoustics and perceptual judgments is qualitatively different for the new measures and for spectral moments, consistent with Shadle [2023]. More generally, these findings call for future sociophonetic studies of fricative variation to use these newer measures.