| 摘要: | 生成式 AI 的普及與興起，不僅帶來更高的便利性，也大大地影響語言學習場域，顛覆了過往學習模式。許多研究將生成式 AI 投入到英語學習中，證明生成式 AI 對於英語學習場域的有效性。另一方面，受益於生成式 AI 所帶來的技術革命，具有擬人與具身化特徵的代理 ( Embodied Agents ) ，透過模擬真人的語言和行為互動提升學習體驗，其對學生的教學成效也受到關注。然而目前對於應用具身化於英語學習場域的研究相對較少，因此透過結合生成式 AI，搭配具身化代理概念，本研究設計一具身化互動系統，探討具身化代理對於學生英語口語能力等影響。
本研究以159名台灣大學生為受測對象，分為三組：有具身化代理組60人、無具身化代理組69人以及對照組30人，共計12週。在實驗過程中，有具身化代理組別能夠與具身化代理互動，並觀察到代理的面部變化；無具身化組別無法看到代理，僅能夠進行對話互動；對照組則是傳統教師面授教學。實驗結束後，收集學生的英語口語能力、溝通意願、動機以及學習過程中的變化並進行分析。
研究結果指出，有具身化代理組別在溝通意願、英語學習動機、英語口語能力測驗以及學習過程中的關係探究，皆顯著優於無具身化代理以及對照組，顯示出具身化能夠顯著提升學生英語口語能力。本研究分別探討學生在課堂外以及課堂內的使用情況，發現有具身化組別的學生在課堂外的平均使用次數顯著高於無具身化組別的學生，證實有具身化代理對於學生使用情況有顯著影響，能夠增加學生使用的意願，進而促進語言學習的持續性與深度。另外也分析了學生對於各代理形象的切換頻率，發現在使用過後，學生傾向使用男性代理進行對話練習，反映出具身化代理的形象選擇能提供個人化學習體驗，根據個別偏好調整互動風格，從而提升學生的參與度和練習成效。
本研究證明有具身化代理於英語口語學習的影響，包括顯著提升英語口語能力、溝通意願、學習中的關係探究，以及英語學習動機。此外，學生對於使用具身化互動系統持正面態度，為具身化代理於英語教學環境提供實證支持，可做為未來相關研究的參考。;The rise in popularity of generative AI not only enhances convenience but also dramatically affects the field of language learning, challenging traditional learning models. Numerous studies have integrated generative AI into English language education, demonstrating its effectiveness in this domain. On the other hand, leveraging the technological revolution introduced by generative AI, agents with anthropomorphic and embodied characteristics (Embodied Agents) enhance the learning experience by simulating the language and behavioral interactions of real individuals. Their effectiveness in teaching has also garnered significant attention. However, there is a relative scarcity of research focused on the application of embodiment in the field of English learning. Therefore, by combining generative AI and the concept of embodied agents, this study designs an Embodied English Learning System to explore the impact of embodied agents on students′ English speaking proficiency.
This study involved 159 Taiwanese college students, divided into three groups: 60 participants in the embodied agent group, 69 participants in the non-embodied agent group, and 30 participants in the control group, for a total of 12 weeks. During the experiment, the embodied agent group was able to interact with the agent and observe its facial expressions, while the non-embodied agent group could only engage in conversational interactions without visual access to the agents. The control group received instruction from traditional teachers. Following the experiment, data on the students′ English speaking proficiency, willingness to communicate, motivation, and changes in their learning processes were collected and analyzed.
The results of the study show that the embodied agent group outperforms the non-embodied agent group and the control group in terms of Willingness to Communicate, English Learning Motivation, English Speaking Proficiency, and Relationship of Inquiry during the learning process, demonstrating the effectiveness of this system. This study examined the use of the agent both outside and inside the classroom. The result found that the average number of times the embodied agent group outside the classroom was significantly higher than that of the non-embodied agent group. This confirmed that the embodied agent had a significant impact on the students′ usage, increasing their willingness to use the agent, and thus promoting the continuity and depth of language learning. In addition, the frequency with which students switched between agent images was analyzed. It was found that after using the agent, students tended to prefer male agents for dialogue practice. This result reflects the idea that the selection of embodied agent images can offer a personalized learning experience, tailor the interaction style to individual preferences, and consequently enhance students′ engagement and the effectiveness of their practice.
This study demonstrates the impact of embodied agency on oral English learning, highlighting significant improvements in English speaking proficiency, willingness to communicate, relationship of inquiry in learning, and motivation to learn English. Additionally, students demonstrate a positive attitude toward using the embodied interactive system, providing empirical evidence for embodied agency in English teaching environments. This can serve as a reference for future research related to this topic. |
