Phoneme awareness and the perception of native phonological categories

Abstract

Recent research on native language speech perception showed that some native Australian English (AusE) listeners performed poorly at identifying their native vowel categories on a categorisation task (Faris et al., 2018; Shaw et al., 2018). This was surprising, because native language (L1) speech perception begins in infancy and improves robustly as children develop into expert speaker-hearers of their L1(s). Thus, they should be able to identify their L1 speech categories with relative ease. However, considering that the task requires metalinguistic abilities to be able to make judgements about what is perceived in speech, it is possible that poor performance on the task was as a result of interference from phoneme awareness. This thesis aimed to test that specific hypothesis by asking L1 AusE listeners to categorise and discriminate between vowels and consonants in their L1, as well as to complete a set of phoneme awareness tasks. First, it was hypothesised that both vowel and consonant identification (through categorisation) would be significantly correlated with phoneme awareness, which would significantly predict speech categorisation accuracy. This is because the categorisation task requires the listeners to determine how written word (for vowels, e.g., heed - /i/) and letter (for consonants, e.g., “B” - /b/) labels in English correspond with the various phonological vowels and consonant categories. Thus, if participants possessed weak phoneme awareness abilities, it should show up as poor categorisation performance. This hypothesis was tested in two experiments: Experiment 1a (vowels) and Experiment 1b (consonants). To test phoneme awareness, participants completed three tasks in which they were asked to either count (e.g., belt = 4 sounds), delete (e.g., bend minus /b/ = end), or reverse (e.g., bomb reversed = mob) the sounds in some English words. The second hypothesis was that participants’ ability to discriminate between pairs of L1 vowel and consonant contrasts would not be significantly correlated with their phoneme awareness accuracy. This is because the discrimination task, which does not require the overt use of orthographic labels, should not require phoneme awareness; thus, it should not interfere with participants’ discrimination abilities. This was tested in Experiment 1b (vowels) and Experiment 2b (consonants). The data were analysed using a series of hierarchical regression analyses that assessed model fit across several different steps in which both covariate and phoneme awareness predictor factors were entered. The results confirmed the hypotheses as measures of phoneme awareness was significantly correlated with vowel and consonant categorisation accuracy in the respective experiments. An interesting finding contrary to the hypothesis was that measures of phoneme awareness significantly predicted discrimination accuracy but only up until categorisation accuracy was entered as a subsequent predictor variable. Overall, the results suggest that phoneme awareness abilities modulate performance on a categorisation task and should be taken into consideration in the design of perceptual assimilation tasks that aim to assess L1 and especially L2 speech perception. Implications for psycholinguistic research as well as L2 speech learning are discussed, with directions suggested for further research.

Date of Award	2023
Original language	English
Awarding Institution	Western Sydney University
Supervisor	Tamara Watson (Supervisor)