음성음운형태론연구 32집 1호 이미리내
2026.05.08 12:14
ChatGPT 기반 음성 상호작용에서 나타난 한국인 EFL 학습자의 영어 모음 /i/–/ɪ/ 발화에 대한 음향 분석
Abstract
An acoustic analysis of Korean EFL learners’ production of the English vowel contrast /i/–/ɪ/ in ChatGPT-based interaction. The study explores whether interaction with ChatGPT’s speech-recognition feedback can improve Korean EFL learners’ production of the tense–lax vowel contrast /i/–/ɪ/. Korean university learners were divided into high- and low-proficiency groups and produced target words containing the two vowels up to five times during interaction with ChatGPT. Pronunciation changes were analyzed using recognition rate, acoustic measures (F1, F2), and Euclidean distance from native speaker vowel centroids. For the tense vowel /i/, both high- and low-proficiency groups showed significant F1 lowering, indicating a higher tongue position, while the low-proficiency group additionally exhibited F2 increase, reflecting greater tongue fronting toward the target vowel. For the lax vowel /ɪ/, however, acoustic changes were limited: F1 values showed minimal change, and clear convergence toward the native centroid was observed only in the high-proficiency group. These findings suggest that ChatGPT’s speech-recognition feedback can support articulatory adjustment and self-regulated pronunciation practice, particularly for L2 vowels close to learners’ L1 categories, whereas vowels without a direct L1 counterpart may require additional explicit instruction beyond ASR-based feedback.
Keywords
ChatGPT, pronunciation, acoustic analysis, English vowels, feedback, Korean EFL learners
댓글 0
2sisstore
