在資訊快速的時代，科技融入教育已經是進行式，在傳統的英語教學課堂中，教師通常難以即時根據每位學生的程度或需求進行教學調整，也很難長時間提供符合學生情境的模擬練習，使部分學生在表達的過程中容易陷入焦慮。本研究嘗試透過生成式 AI 所具備的自然語言與互動能力，設計「英語學習同伴系統」以因應上述困境。讓 AI透過結構化的提問與提示的機制，引導學生針對英文文章內容進行理解與回應。系統同時設計教師端儀表板，用以呈現學生的回應互動歷程，協助教師掌握學生參與情形，並作為後續教學調整的依據。 本研究於臺灣北部某國小六年級進行為期八週的準實驗設計，總樣本數為 74 人。其中32 名實驗組學生透過平板使用本系統完成每週學習單，另有 42 名對照組學生則由教師以傳統授課方式帶領閱讀相同主題之內容，並搭配紙本學習單進行討論。研究結果指出，學生英語能力與第二語言意願皆有顯著提升，科技接受模式顯示學生對於本系統的學習輔助價值表示肯定。教師訪談指出，儀表板操作方便，有助於即時瞭解學生的參與情形。部分平時在課堂上較少主動發言的學生，在本次活動中展現出相對積極的互動行為，顯示本系統在促進學生參與方面具有一定潛力。 綜上所述，本研究證實「英語學習同伴系統」的可行性與成效，能在一定程度上補足即時回饋與個別化支持的不足，進而促進學生的語言表達。未來建議擴大樣本並延長追蹤週期，同時導入語音辨識與獎勵式遊戲設計，並透過混合方法，進一步分析 AI 對話內容的質量與促進學習增益的潛在機制，以持續優化生成式 AI 在基礎語言教育中的應用。;In an era of rapid technological advancement, the integration of technology into education has become a prevailing trend. In traditional English classrooms, however, teachers often struggle to adjust instruction in real time based on individual students′ needs and abilities. Moreover, it is challenging to provide extended, contextually meaningful practice opportunities, which can lead some students to experience anxiety during language expression. To address these challenges, this study developed an “AI-based English Learning Companion System” that leverages the natural language understanding and interactive capabilities of generative artificial intelligence. The system guides students in understanding and responding to English texts through structured questioning and prompting strategies. In addition, a teacher-facing dashboard was designed to visualize students’ response histories, helping educators monitor participation and inform instructional adjustments. This eight-week quasi-experimental study involved 74 sixth-grade students from an elementary school in northern Taiwan. Among them, 32 students in the experimental group used the system on tablets to complete weekly learning tasks, while 42 students in the control group received traditional instruction with paper-based worksheets. The results showed significant improvements in students’ English proficiency and willingness to communicate in a second language. The technology acceptance analysis further indicated that students perceived the system as a valuable learning aid. Teacher interviews revealed that the dashboard was easy to use and helpful in identifying students’ participation levels. The teacher also observed that some students who typically remained quiet in class showed more active engagement during the activity, suggesting that the system may have potential in enhancing student participation. In summary, the study provides initial evidence supporting the feasibility and effectiveness of the AI-based learning companion system. The system appears to partially address the limitations of real-time feedback and individualized support in traditional instruction, and may help foster students’ language expression. Future research is recommended to expand the sample size and extend the study duration. It is also suggested to integrate speech recognition and gamified reward features, and to employ mixed-methods approaches to further examine the quality of AI-generated dialogue and its potential mechanisms for supporting learning gains, thereby refining the application of generative AI in foundational language education.
