Acoustic and auditory evidence of sociophonetic variation in Vancouver: LTAS and vocalic data.

Abstract

Formant frequencies of ten Canadian English vowels from identical environments in reading style were calculated for 128 subjects in the middle and older age groups of the sociolinguistic Survey of Vancouver English. LPC-based F1, F2 distributions were derived using a computerized speech lab (CSL) program and evaluated statistically to identify differences in vowel quality across four socioeconomic-status (SES) classes for both sexes. Long-term average spectra (LTAS) of 60 s of reading text were also computed and compared with vowel formant results. The SDDD LTAS dissimilarity index (standard deviation of the differences distribution) was also applied to the 192 subjects in all (including young) age groups. Auditory evaluations of each subject’s production of each vowel were compared with the results of formant analysis and both LTAS analyses. Here, F1 values were more uniformly differentiated than F2, and some groups’ vowels were more vulnerable to rejection of F2 on bandwidth criteria than others. Differentiation is significant and consistent for a high proportion of vowel formant distributions for some SES groups, notably between working-class (WC) and middle-class (MC) middle-aged women. Some F1, F2 patterns are compatible with auditory predictions of a hierarchy of vocalic susceptibility to contrasting voice quality settings. [Work supported by SSHRCC.]