Gender differences in the acoustic realization of creaky voice: Evidence from conversational data collected in Northern California

Abstract

Although several sociophonetic studies report greater breathiness among female speakers, a pattern often attributed to sexual dimorphism in vocal fold physiology (Södersten and Lindestad, 1990), recent studies in North America report that young women use more creaky voice than men (Yuasa, 2011; Podesva, 2013). While these recent studies examine conversational data, they rely on auditory techniques to identify creaky phonation, leaving its acoustic realization in conversational speech largely unexplored. The present study investigates the acoustic properties of creaky voice in hour-long sociolinguistic interviews with 30 speakers (15 females, 15 males; age 18–86) from Northern California. Measures of spectral tilt were taken at the midpoint of all vowels in the corpus (N = 362,429), and data were fitted to a mixed effects linear regression model. As expected, several linguistic factors influence H1-H2 values (previous and following segment, intensity, F0, F1, vowel duration, stress, phrase position, phrase duration), alone and in interaction. With regard to social factors, H1-H2 is significantly lower for female speakers, indicating greater creak, even though the H1-H2 measure under-captures creakiness for female speakers (Simpson, 2012), and no age effect was observed. In sum, while females are generally creakier, apparent time data do not indicate that this is a recent trend.