This semester, the Montreal Computational and Quantitative Linguistics Lab (MCQLL) will hold lab meetings on alternating Tuesdays from 15h00 – 16h00, beginning this Tuesday, September 5th.
Meetings will be hybrid in person and on Zoom. We’ll be meeting in person in room 117 of the McGill Linguistics Department at 1085 Docteur Penfield. If you would like to attend virtually, Zoom meetings will be held here.
In our first meeting, Jeanne Brown will be presenting “Creaky Voice in Canadian English: An acoustics-focused method”. The abstract is below.
Title: Creaky voice in Canadian English: An acoustics-focused method
Abstract: Creaky voice is a voice quality attributed to vocal fold compression without complete glottal closure. Acoustically, it is characterized by three key properties: a low pitch, irregular vocal pulses, and decreased transglottal airflow. Previous work provides competing evidence of gender differences with regard to creaky voice in English: some studies find more creak in men’s speech (Henton & Bladon, 1988; Klatt & Klatt, 1990) whereas more recent studies frequently report increased creakiness among women (Podesva, 2013; Yuasa, 2010). Many of the existing sociophonetic studies of creaky voice rely on impressionistic (auditory or visual) coding which are vulnerable to important perceptual biases (Davidson, 2019; White et al., 2022). Moreover, while creaky voice is attested in many varieties of English, comprehensive studies of creak in Canadian English are rare. This study is an acoustic analysis of creaky voice across gender in Canadian English. Three acoustic correlates of creak were extracted at vowel mid-points: a spectral slope measure, H1*-H2*, and two Harmonics-to-Noise Ratios (HNRs), cepstral peak prominence (CPP) and HNR at the lowest frequency band (0-500Hz). Spectral slopes are acoustic indicators of glottal constriction, lower spectral slope measures correlating with a more constricted glottis (reduced airflow) and therefore more creakiness. HNRs are measures of waveform periodicity with lower HNRs corresponding to higher levels of noise/aperiodic vibration which is also a typical property of creak. Linear mixed-effect models tested differences within each acoustic measure of creak in relation to social and linguistic factors. Results reveal that men’s vowels have less reliable f0 tracks, lower H1*-H2* values and lower HNRs (CPP and HNR05) than women’s vowels, presenting substantial evidence for more creakiness in men’s speech. Overall, this study highlights the importance of acoustic measures in quantifying creak and provides new insight into the relation between creaky voice and gender.