Perception of non-native quantity contrasts by Cantonese and Japanese speakers

Abstract

This paper reports on the findings of a pilot study which investigated Cantonese and Japanese speakers’ ability to perceive non-native quantity contrasts. Our goal was to examine whether L1 phonology affects the perception of non-native sound contrasts at a discrete, categorical level or a gradient, “featural” level. We recruited native Cantonese and Japanese speakers for a series of AXB discrimination and identification tasks. Cantonese has partial quantity contrasts (i.e., where duration is not the primary acoustic cue) in a few vowel pairs, whereas Japanese has systematic two-way quantity contrasts in both vowels and consonants. The synthesized stimuli were Japanese (non-native for Cantonese speakers) and Estonian (non-native for both groups) nonce words contrasting in vowel and consonant quantity. For Estonian, quantity is a three-way distinction (short, long, overlong). The results showed that while Japanese speakers outperformed their Cantonese counterparts in discrimination, their identification accuracy for overlong Estonian vowels was anomalously low. These findings are discussed with reference to the “feature hypothesis” in L2 phonological acquisition.