Use this URL to cite or link to this record in EThOS: https://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.819843
Title: The effect of high variability and individual differences on phonetic training of Mandarin tones
Author: Dong, Hanyu
ISNI:       0000 0004 9359 6981
Awarding Body: UCL (University College London)
Current Institution: University College London (University of London)
Date of Award: 2020
Availability of Full Text:
Access from EThOS:
Full text unavailable from EThOS. Please try the link below.
Access from Institution:
Abstract:
High variability phonetic training (HVPT) has been found to be more effective than low variability phonetic training (LVPT) in learning various non-native phonetic contrasts. However, little research has considered whether this applies to the learning of tone contrasts. Two relevant studies suggested that the effect of high variability training depends on the perceptual aptitude of participants (Perrachione, Lee, Ha, & Wong, 2011; Sadakata & McQueen, 2014). It is also unclear how different types of individual difference measures interact with the learning of tonal language. What work there is, suggests that musical ability is related to discriminating tonal information and in general attention and working memory are linked to language learning. The present study extends these findings by examining the interaction between individual aptitude and input variability and between learning outcomes and individual measures using natural, meaningful L2 input (both previous studies used pseudowords). In Study 1, forty English speakers took part in an eight-session phonetic training paradigm. They were assigned to high/low variability training groups. High variability used four speakers during the training sessions while low variability used one. All participants learned real Mandarin tones and words. Individual aptitude was measured using an identification and a categorisation task. Learning was measured using a categorical discrimination task, an identification task and two production tasks. Overall, all groups improved in both production and perception of tones which transferred to novel voices and items, demonstrating the effectiveness of training despite the increased complexity of the training material compared with previous research. Although the low variability group exhibited better learning during training than the high variability group, there was no evidence that the different variability training conditions led to different performances in any of the tests of generalisation. Moreover, although performance on one of the aptitude tasks significantly predicted overall performance in categorical discrimination, identification and training tasks, it did not predict improvement from pre- to post- test. Critically, there was also no interaction between individual aptitude and variability-condition, contradicting with previous findings. One possibility was that the high variability condition was too difficult as speakers were randomly presented during training, resulting in low trial-by-trial consistency. This greater difficulty might block any advantage of variability for generalisation. In order to examine this, Study 2 recruited additional 20 native English speakers and tested them in a further condition, identical to the previous high variability condition except that each speaker was presented in their own block during the training. Although participants performed better in training compared with the high variability group from study 1, there was again no difference in generalisation compared with the previous conditions, and again no interaction between individual aptitude and variability-condition was found. Bayes Factors were also used to assess the null results. There was evidence for the null for the benefits of high variability for generalisation but only ambiguous evidence regarding whether there was interaction between variability and individual aptitude. The HPVT used in Study 1 and Study 2 did not replicate the interaction between variability-condition and aptitude found in previous studies. Moreover, although one of the measures of aptitude did correlate with the baseline measures of performance, there was no evidence that it predicted learning due to training. Additionally, the two individual aptitude measures used in Study 1 and 2 – taken from Perrachione, et al. (2011) and Sadakata and McQueen (2013) – are not comprehensive. They are natural language-related tasks which directly measure tone perception itself, rather than the underlying cognitive factors which could underpin this ability. Another interesting question is whether these different cognitive factors might contribute to learners at different stages differently, particularly since language training studies vary as to whether they use current learners of the language or naïve participants, a factor may contribute towards differing findings in the literature. To explore these issues, Study 3 investigated the relationship between a battery of cognitive individual difference measures and Mandarin tone learning. Sixty native English speakers (forty of whom were currently studying Mandarin at undergraduate level, twenty of whom were naïve learners) took part in a six-session training paradigm. With high-variability training stimuli similar to that used in Study 2 (four speakers blocked), their learning outcomes were assessed by identification, categorical discrimination and production tasks similar to Study 1. Their working memory, attention and musical ability were also measured. Overall, both groups showed improvements during training and in the generalisation tasks. Although Mandarin learner participants performed better than naïve participants overall, the improvements were not generally greater than naïve participants. Each of the individual difference measures was used to predict participant’s performance at pre-test and their improvement due to training. Bayes Factors were used as the key method of inference. For Mandarin learner participants, both performances at pre-test and pre- to- post improvement were strongly predicted by attention measures while for naïve speakers, musical ability was the dominant predictor for pre- to- post improvement. This series of studies demonstrates that Mandarin lexical tones can be trained using natural stimuli embedded in a word learning task and learning generalises to untrained voices and items as well as to production. Although there is no evidence in the current data that the type of training materials affected learning outcomes, tone learning is indeed affected by individual cognitive factors, such as attention and musical ability, with these playing a different role for learners at different stages.
Supervisor: Not available Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID: uk.bl.ethos.819843  DOI: Not available
Share: