Tone as a primary perceptual cue in cross-linguistic speech perception: A comparison of Cantonese and Mandarin second-language speech of English clusters

Abstract

The study examined the patterns of production and perception of L1 Cantonese and Mandarin speakers for English consonant-/l/ clusters in five consonant conditions (/ph, th, kh, f, s/) and three vowel conditions (/i, a, u/). Five Cantonese, 5 Mandarin, and 5 native English speakers were assigned to read aloud words in C[l]V and C[l]VC structure in carrier sentences. Results showed that Cantonese speakers’ /l/ was often reduced, indicated by a shorter average duration of the CV transition compared with native English speakers. However, Mandarin speakers showed a longer duration in the same measurement. Despite the identical segmental and syllable structures of these two languages in the involved words, realizations were different. Nevertheless, we found pitch patterns of Mandarin speech of L2 English featured falling tone, whereas Cantonese speakers utilized level tones. To further examine the tonal effect, we normalized the tone from the production results to a level tone at 200 Hz and presented them, together with productions with real inserted vowels in between C and /l/, to another group of Mandarin and Cantonese speakers to discriminate in an ABX paradigm. Mandarin speakers scored a significantly lower accuracy rate, indicating that their perception of duration was influenced by tone structure.