Disentangling acoustic and social biases in creaky voice perception: The effects of f0 and face gender on creakiness ratings

Creaky voice has historically been associated with men's speech (e.g., Henton & Bladon, 1988), supported by acoustic studies (e.g., Gittelson et al., 2021; Klatt & Klatt, 1990). Since around 2010, sociolinguistic work alongside public discourse has perpetuated greater creaky voice use by women instead, typically implementing impressionistic coding (e.g., Podesva, 2013; Yuasa, 2010). This study investigates whether this recent shift can be attributed to perceptual social and acoustic biases related to (perceived) speaker gender and pitch (f0), respectively. Using a matched-guise paradigm, 40 Canadian English listeners rated the perceived creakiness of the same modal and creaky voices—altered to have ambiguously gendered formants and median f0s (115, 135, 155 Hz)—paired with female and male faces. Bayesian regression analyses revealed strong effects of voice quality and moderate effects of f0: creaky and lower-f0 stimuli were rated as creakier. No overall effect of face gender was found. However, a weak interaction between face gender and f0 suggests a possible gender prototypicality bias: female faces were rated as slightly creakier at lower f0s and male faces at higher f0s. These findings show that neither acoustic nor gender-based biases alone can account for widespread reports of women-led creaky voice use.