Perceptual assimilation, discrimination, and acquisition of non-native and second-language vowels assimilated as uncategorised

Abstract

Non-native and second-language (L2) phones are perceived in terms of their similarities and differences to the listener's native (L1) phonemes. A non-native phone that is reliably identified as similar to a single L1 phoneme is described as being categorised according to PAM/PAM-L2 (Best, 1995; Best & Tyler, 2007), and such categorised assimilations have been the focus of much of the research on non-native and L2 speech perception. It is also possible that non-native/L2 phones are perceived as being speech-like, but are not identified with any one particular L1 phoneme. In such instances, they are said to be assimilated as uncategorised. This thesis presents experiments designed to address how these uncategorised phones are perceived, discriminated, and acquired by adult L2 learners. The findings have theoretical implications for models of cross-language and L2 speech perception, and contribute to our understanding of the perception and acquisition of uncategorised phones. The first question addressed in this thesis was whether there were any systematic differences in the way in which uncategorised non-native phones are assimilated within the L1 phonological system. In the first experiment, native Egyptian Arabic speakers residing in Egypt perceptually assimilated and rated all of the Australian English vowels in relation to their L1 vowels. Results revealed new assimilation types for uncategorised phones. They may be perceived as being moderately similar to just a single L1 phoneme (focalised), to two or more L1 phonemes (clustered), or unlike any of the L1 phonemes (dispersed). This suggests that not all uncategorised phones are perceived in the same way, but rather, they vary in the extent to which they are perceptually identified with L1 phonemes. The second set of experiments addressed the issue of how well pairs of non-native phones might be discriminated when one or both phones are assimilated as uncategorised (i.e., Uncategorised-Uncategorised and Uncategorised-Categorised assimilations, respectively). Native Australian English speakers discriminated Danish monophthongal and diphthongal vowel contrasts varying in assimilation type, including Uncategorised-Uncategorised and Uncategorised-Categorised. Discrimination accuracy was modulated by the presence of perceived phonological overlap in the categorisations to L1 phonemes, with partially overlapping contrasts discriminated less accurately than non-overlapping contrasts. By considering the different uncategorised assimilations and the presence of perceived phonological overlap to L1 categories, it was possible to fine-tune predictions for Uncategorised- Uncategorised and Uncategorised-Categorised assimilations much better than if overlap were not considered. The final aim of this thesis was to examine the acquisition of uncategorised L2 phones in adults. It was predicted that new category formation would be more likely to occur for uncategorised, than for categorised, L2 phones. Egyptian Arabic speakers acquiring Australian English in an immersion setting were recruited for a 1-year longitudinal study. They were assessed on their perceptual assimilation of the English vowels, and also on their discrimination accuracy of English vowels that formed Uncategorised-Uncategorised and Uncategorised-Categorised assimilations, as it is these assimilations that should be easily acquired according to the PAM (Best, 1995) and SLM (Flege, 1995) theoretical models. The learners were not absolute beginners, but they differed on six factors related to L2 experience (i.e., length of residence, age of foreign language acquisition, age of immersion, proportion of L2 use, L2 vocabulary size, and duration of English as a foreign language training). There was no evidence of new category acquisition, perhaps because they had already reached a plateau in L2 learning at the start of the study, or possibly because a longer period of immersion may be needed to determine whether they show improvements over time. However, by considering variability among individual learners, the six factors were shown to predict discrimination accuracy to a certain extent, predominantly for Uncategorised-Uncategorised and Uncategorised-Categorised assimilations. Overall, the experiments reported in this thesis provided a much-needed systematic and thorough investigation into the perceptual assimilation, discrimination, and acquisition of uncategorised non-native/L2 phones. The experiments demonstrated that uncategorised phones vary in their perceptual assimilation to the L1, and that discrimination accuracy is dependent upon the presence of perceived phonological overlap for contrasts involving uncategorised phones. This has important theoretical implications for both cross-language and L2 speech perception models. A number of interesting questions are also raised for L2 phonological category formation, answers to which have the potential to provide a step forward in our understanding of L2 acquisition.

Date of Award	2017
Original language	English