The disambiguation of Australian English vowels in lateral-final syllables

Discriminability of Australian English vowel pairs followed by coda /d/ or coda /l/ was examined using a lexical decision task. 30 native listeners categorized 10,800 Australian English words either with /hVd/ or with/hVl/ structure in a binary forced choice task. Both in the coda /d/ and the coda /l/ condition, 16 target words differing only in the nuclear vowel were presented as auditory stimuli, paired with each of the remaining 15 words as competitors. Accuracy and reaction time were measured. Results of the /d/ condition show that vowels are intrinsically similar in perception when vowels of the target and competitor pair share spectral similarity but differ in length. Comparison of the coda /d/ and /l/ condition shows that coda /l/ makes vowel disambiguation harder across the board and it also makes the disambiguation of intrinsically similar target-competitor pairs even harder. The results also show that coda /l/ might reduce the contrast between /ʉ:-ʊ/, /æɔ-ae/, and /əʉ-ɔ/ pairs to such an extent that contextually conditioned mergers may occur in perception. The potential context-based mergers suggest that the rimes containing /l/ might only allowa subset of Australian English vowels to appear in the nucleus, therefore the relationship between a nucleus and coda /l/ might be stronger than the relationship between a nucleus and a coda /d/.