Articulatory and acoustic characteristics of occlusivized sonorants in Shakou Hakka

Abstract

This study examines a pattern of sonorant occlusivization in a dialect of Hakka Chinese spoken in Shakou Township, Guangdong, China. This variety exhibits variable occlusivization of /l/ (realized as centralized [ld ∼ ld ∼ d]) and denasalization of initial /m, ŋ/ (realized as prenasalized stops [?b, ?g]). Mid-sagittal and coronal ultrasound with synchronized audio, EGG, and nasalance recordings were collected from one speaker, along with acoustic recordings from 5 speakers. Partial occlusivization of /l/ occurred in 74.8% of words with high /i, y, u/ but does not occur in words containing low or mid vowels. Occlusivization of /l/ has an average duration of 42ms and is characterized by a significant decrease in mid-frequency acoustic energy (−6 dB, p < 0.001). Ultrasound data reveal tongue body raising and parasagittal contact during occlusivization, indicative of overlap between the tongue gestures required for /l/ (with lateral airflow) vs. /i, y, u/ (with lateral bracing). Denasalization, which has a similar mean duration of 39.5ms, occurs in 42.9% of words containing non-high vowels and 86.25% of words containing high vowels. Although these patterns differ with regard to their phonological conditioning, gestural interaction during sonorant-vowel sequences may provide a common underlying mechanism for both processes.