• Shazia Khanum Yunus

  Author

• Neha Dey

  Author

Pronunciation continues to be a problem for EFL learners since conventional classrooms are unable to provide immediate and personalized feedback in these cases. This research explores the application of Automated Speech Recognition (ASR) technology as an additional tool in pronunciation teaching, based on Second Language Acquisition (SLA) theories, which include Noticing, Output, and Learner Autonomy Hypotheses. Using a descriptive quantitative approach, 30 EFL undergraduate learners at an intermediate level (N=30) underwent a 6-week intervention, where practiced using ASR guidance (about 20 min. 3 times a week). The performance of learners before and after the intervention was assessed based on a 100-point rating system in five aspects of pronunciation: segmental correctness, word stress, sentence intonation, speech fluency, and speaker confidence. The results of data analysis indicate uniform improvement of mean scores in all analyzed metrics. The greatest increase was found in the metric of self-reported confidence of speakers (+24 points, from 54 to 78), thus indicating the affective value of low-pressure automated assessment. The metrics of word stress (55 to 74) and sentence intonation (52 to 71), along with speech fluency (60 to 79), have increased in unison by +19 points. The segmental accuracy score gained by +18 points (58 to 76). ASR tools are effective in fostering independent phonological development and confidence in communication. Although ASR technology is unable to fully substitute for human-based instruction because of the difficulties with assessment of suprasegmental components, it is an extremely valuable supplementary tool in language education curriculum.

• PDF