Effects of word length and presentation order in the perception of non-native sequences

Abstract

Previous research in the perception of non‐native phonotactics indicates that listeners have difficulty distinguishing certain sequences like [ebzo]‐[ebuzo] (Japanese listeners) or [zgamo]‐[z■gamo] (English listeners). Whereas discrimination of phonemes in monosyllables (e.g. [k’a]‐[q’a]) is diminished but better than chance [J. Werker and R. Tees J. Acoust. Soc. Am. 75, 1984], discrimination of non‐native sequences in longer words is often below chance. In this study, the effect of length and presentation order on the discrimination of non‐native phonotactics is investigated. English speakers (n=30) were presented with pairs of the type CCVCV‐C■CVCV (e.g., /tkapu‐t■kapu/) and CCV‐C■CV (/tka‐t■ka/) in an AX discrimination task. Half of the ‘‘different’’ pairs were presented with the ■ word first, and half with the cluster word first. Results indicate that English listeners are significantly less accurate on the discrimination of longer words both in ‘‘same’’ and ‘‘different’’ trials. Presentation order was also significant: listeners are more accurate when the cluster word is presented first than when the ■ word is. These results suggest that while adults can discriminate non‐native sounds or sequences under optimal contexts, this ability is not necessarily helpful in situations with more realistic input, such as language acquisition. [Research supported by NSF.]