Non-linear dynamics of adult non-native phoneme acquisition perception and production

A number of formal models of non-native speech perception make predictions regarding ease of discrimination based on the phonetic similarity of non-native phonemes to phonemes from the learner’s native phonology (Best, McRoberts, & Goodell, 2001; J. Flege, 1995; Kuhl, et al., 2008). Recently, individual differences in non-native sound learning have been examined from a dynamical perspective, focusing on initial perceptual abilities of individuals, and how the structure of these perceptions influences learning (Tuller, Jantzen, & Jirsa, 2008). Here, monolingual speakers of American English were trained in the perception of the Spanish tap and trill rhotics to examine if different types of native/non-native speech contrasts exhibit different structure of initial perception and learning dynamics. In addition, production of the target phonemes by the learners was recorded analyzed to viii determine the effects of perceptual training on production and examine the relation between initial perceptual structure and production learning dynamics. Results showed that initial perceptual structure for the rhotics was predictive of learning dynamics, though the types and distribution of initial patterns were different from those seen in prior work with different phoneme contrasts. Production improvement was seen in a number of participants, corresponding to changes in perceptual constraints over the course of training. However, articulatory (motor) constraints mitigated considerably the relation between perceptual ability and production ability. These results are consistent with current models of non-native speech perception, and provide further evidence of non-linear perceptual learning processes. Production results suggest variable interaction between perceptual and articulatory constraints on speech production learning.