Acoustic classification of velar fricatives in Assamese

Abstract

Previous work on the acoustic discrimination of fricative contrasts has commonly studied categories which are not held to vary phonemically; instead the focus is placed on accounting for variation in the salience and stability of the acoustic cues mapping onto those categories (cf. Forrest et al., 1988; Jongman et al., 2000; McMurray and Jongman, 2011). The present study, like Jannedy et al. (2015), addresses a set of contrasts—the voiceless posterior obstruents /x, h, kh/ in Assamese—where /h/-voicing and /kh/-lenition processes interact with the discriminability of the velar fricative differentially according to positional (CV, VCV, and VC) and speaker (gender, dialect) variables. Results of principal components logistic regression (PCLR) models applied at both fixed and cumulative time windows over consonant and transition intervals are presented. Overall, /x/-/h/ discrimination was significantly more accurate in intervocalic position (96%) relative to word-initial position (84%). Data from intervocalic productions further suggest a significant dialectal difference in the /x/-/kh/ contrast, with discrimination poorer in Nalbari (68%) than in Jorhat (86%) data, in line with Sarma’s (2012) analysis. Implications for the integration of lexical information by means of phonotactic priors derived from a corpus (Baker et al., 2002) are also discussed.