Exploring the Effects of Automated Pronunciation Evaluation on L2 Students in Thailand

Download (PDF, 713KB)

Author: Simon Moxon, Walailak University, Thailand
Email: [email protected]
Published: June 11, 2021
https://doi.org/10.22492/ije.9.3.03

Citation: Moxon, S. (2021). Exploring the Effects of Automated Pronunciation Evaluation on L2 Students in Thailand. IAFOR Journal of Education: Language Learning in Education, 9(3). https://doi.org/10.22492/ije.9.3.03


Abstract

A significant barrier to effective communication in a second language is the awareness and accurate reproduction of phonetic sounds absent in the mother tongue. This study investigated whether the automated evaluation of phonetic accuracy using speech recognition technology could improve the pronunciation skills of 105 (88 female, 17 male) Thai undergraduate students studying English in Thailand. A pre-test, post-test design was employed using treatment and control sample groups, reversed over two six-week periods. Treatment group students were given access to an online platform on which they could record and submit their speech for automated evaluation and feedback via SpeechAce, a speech recognition interface designed to evaluate pronunciation and fluency. Independent samples t-test analysis of the results showed statistically significant improvement in pronunciation accuracy of students in the treatment group when compared to those in the control group (t (89) = 2.086, p = .040, 95% CI [.083, 3.423]), (t (89) = -4.692, p < .001, 95% CI [-5.157, -2.089]). Pearson’s correlation analysis indicated a weak to moderate, but statistically significant correlation between frequency of practise and pronunciation test score (r =.508, p < .001), (r = .384, p = .021). The study has limitations as the sample group was predominantly female, and time constraints limited students’ use of the software. Future studies should investigate possible gender differences and experiment with different forms of visual feedback.

Keywords

phonetics, pronunciation, speech recognition, SpeechAce