探討運用自動語音辨識為基礎的電腦輔助發音教學系統於技術型高中學生英語發音學習之研究

Abstract

本研究探討以自動語音辨識（ASR）技術為基礎的電腦輔助發音訓練（CAPT）系統，特別是「酷英語發音評量平台」，對臺灣技術型高中學生（英語能力為 CEFR A1 至 A2 級）發音表現的影響。本研究同時探討學生對使用 ASR 發音學習工具的態度。本研究共招募35名學生參與。其中，27名國中教育會考英語科成績為C等級的學生被納入主要的量化分析，以評估其發音進步情形；而35名學生全數參與質性資料分析，針對平台使用經驗提供意見與建議。本研究歷時六週，包含前測、十次練習課程、後測，以及以科技接受模式（Technology Acceptance Model, TAM）為基礎之態度問卷調查。本研究透過分析量化資料，包括整體發音、準確度、流暢度與完整度的前測與後測成績，以及根據科技接受模型（TAM）設計之問卷項目作答結果，以評估學生的學習成效與對系統的態度。另亦蒐集開放式問答的質性資料，以進一步了解學生的看法與建議。研究結果顯示，許多學生在發音表現上有明顯進步，特別是在「完整度」與「準確度」兩個面向。三個評量面向中，以「完整度」的平均進步幅度最大（+12.83 分，+18.00%），其次為「準確度」（+6.24 分，+8.51%）與整體發音表現（+6.48 分，+8.54%）。「流暢度」的進步幅度較小（+0.38 分，+0.45%），且在不同學生之間變異較大。在學習態度方面，學生普遍對ASR平台持正面看法，肯定其操作簡便、即時回饋及自主練習的特性。然而，許多學生亦指出希望平台能新增可調整音檔播放速度、重播自己錄音的功能，並認為仍需要教師輔助，以協助釐清不易辨識的發音並提供個別化指導。研究結果顯示，ASR為基礎的電腦輔助發音教學系統有助於提升CEFR A1 至A2等級英語學習者的發音表現，尤其在「完整度」與「準確度」兩個面向。然而，「流暢度」的提升仍具挑戰性，顯示未來在運用ASR系統時，應結合教師指導與其他教學策略，以達到更全面且穩定的學習成效。本研究亦針對教學應用、研究限制及未來研究方向提出相關建議。This study investigates the effects of using an Automatic Speech Recognition (ASR)-based Computer-Assisted Pronunciation Training (CAPT) system, specifically the Cool English Pronunciation Assessment platform, on the pronunciation development of vocational high school students in Taiwan with CEFR A1 to A2 level English proficiency. It also explores students’ attitudes toward using ASR-based pronunciation tools.A total of 35 students participated in this study. Of these, 27 students with a CAP English score of C were included in the primary quantitative analysis of pronunciation improvement, while all 35 students contributed to the qualitative analysis of feedback and suggestions. The six-week intervention involved pretests, ten practice sessions, posttests, and a Technology Acceptance Model (TAM)-based attitude questionnaire. Quantitative data, including pretest and posttest scores in overall pronunciation, accuracy, fluency, and completeness, as well as responses to the TAM-based questionnaire items, were analyzed to measure improvement and attitudes toward the system. Qualitative data were collected through open-ended responses to further understand students’ perceptions and suggestions.Results indicated that many students demonstrated measurable improvements, particularly in completeness and accuracy. Among the three dimensions, completeness showed the greatest average gain (+12.83 points, +18.00%), followed by accuracy (+6.24 points, +8.51%) and overall pronunciation (+6.48 points, +8.54%). Gains in fluency were smaller (+0.38 points, +0.45%) and more variable across individuals.Students generally reported positive attitudes toward the ASR platform. They appreciated the platform’s ease of use, immediate feedback, and autonomy for self-paced practice. However, many emphasized the need for adjustable audio playback speed, the ability to replay their own recordings, and teacher support to supplement the ASR system, particularly for addressing unclear pronunciations and providing individualized correction.The findings suggest that ASR-based CAPT systems may help support pronunciation learning for EFL students at the CEFR A1–A2 level, particularly in the areas of completeness and accuracy. Nevertheless, fluency development remains challenging, highlighting the need for integrating ASR tools with teacher-guided support and complementary pedagogical strategies. Implications for practice, limitations, and directions for future research are discussed.