Asr-dictation on smartphones for vowel pronunciation practice

Abstract

This study aims to explore mobile-assisted Automated Speech Recognition (ASR) dictation systems for vowel pronunciation practice by examining whether ARS can be useful for pronunciation improvement and speech recognition accuracy. Additionally, learners’ attitudes towards using these systems were explored. Twenty-one Macedonian EFL learners practiced pronouncing 26 words with the following minimal pairs: /i/, /ɪ/; /æ/, /ɛ/; /u/, /ʊ/; /ɑ/, /ʌ/. The participants were di vided into an experimental group (n=11) and a control group (n=10). This study used a mixed methods approach including qualitative and quantitative analysis. Results demonstrated that while the control group did not show any improvement, the experimental group improved their accuracy. ASR written output and human judgment was also found to be within an acceptable agreement for most vowels. Furthermore, while occasional inaccurate feedback sometimes caused frustration, ASR training was generally enjoyed and considered as a practical and safe environment for practice. The findings provide some support for the use of ASR in EFL classrooms with careful planning and direction from the teacher. Using ASR as a tool for controlled and structured practice with individual words is particularly applicable when the focus is to raise learners’ phonological awareness and perception of English vowel sounds.