Exploring the contribution of voice quality to the perception of gender in Scottish English

This study investigates how voice quality, here phonation, affects listener perception of speaker gender, and how voice quality interacts with pitch, a major cue to speaker gender, when cueing gender perceptions. Gender differences in voice quality have been identified in both Scottish (Beck and Schaeffler 2015; Stuart-Smith 1999) and American English (Abdelli-Beruh et al. 2014; D. Klatt and L. Klatt 1990; Podesva 2013; Syrdal 1996; Wolk et al. 2012; Yuasa 2010). There is evidence from previous research that suggest gender differences in voice quality may also influence listener perception of speaker gender, with breathy voice being perceived as feminine or female characteristic by listeners (Addington 1968; Andrews and Schmidt 1997; Bishop and Keating 2012; Holmberg et al. 2010; Porter 2012; Skuk and Schweinberger 2014; Van Borsel et al. 2009) and creaky voice being perceived as masculine characteristic (Greer 2015; Lee 2016). However, some studies have found that voice quality has little effect (Booz and Ferguson 2016; King et al. 2012; Owen and Hancock 2010). The present study seeks to investigate the contribution of voice quality, taking into account the various methods of producing voice quality differences in stimuli, cultural differences in gendered meanings of voice quality, and different methods of quantifying ‘perceived gender’, which may contribute to the conflicting results of previous studies. To investigate the contribution of voice quality to perceptions of speaker gender, a perception experiment was be carried out where 32 Scottish listeners and 40 North American listeners heard stimuli with different voice qualities (modal, breathy, creaky) and at different pitch levels (120Hz, 165Hz, 210Hz), and were asked to make judgements about the gender of the speaker. Differences in voice quality were produced by a speaker with the ability to create voice quality distinctions, as well as created through copy synthesis from the speaker’s voice. Listeners were asked to indicate whether they thought the voice belonged to a man or a woman and rate how masculine and feminine the voice sounded. Relative to modal voice, I predicted that listeners would be more likely to categorise breathy voices as women, and would rate them as more feminine and less masculine, and that listeners would be less likely to categorise creaky voices as women, and would rate them as more masculine and less feminine. I also predicted that there might be differences in how Scottish listeners and North American listeners perceived voice quality, given that the gender differences in voice quality in these two varieties of English have been found to differ in previous research. Consistent with my predictions, I found that relative to modal voice, listeners were more likely to categorise breathy voice stimuli as women, and rated breathy voice stimuli as more feminine and less masculine. However, in contrast with my predictions, I found that relative to modal voice, listeners were more likely to categorise creaky voice stimuli as women, and rated them as less masculine, but not more feminine. Furthermore, contrary to predictions, I did not identify differences between Scottish and North American listeners in terms of voice quality perception. Differences were also found in how breathy and creaky voice influence gender perception at different pitch levels. Overall, these results show that voice quality has an important influence on listener perception of speaker gender, and that the gendered meanings of creaky voice are changing and have disassociated from its low pitch. Future research should consider whether this evaluation among Scottish listeners this may reflect a wider change in the gender differences in production