| 摘要: | 由於真實情境中的日常生活練習可以激發學習興趣並鞏固EFL學習中的英語技能，如何在真實情境中有效地提供EFL學習是一個不斷興起的問題，尤其是在創新科技的推動下。在這種情況下，具有辨識技術的智慧學習環境和雲端運算具有前瞻性且技術成熟，適合應用在真實環境的 EFL 學習中。因此，我們提出Smart UEnglish+應用程式，以幫助學生利用圖像分析技術（image-to-text recognition, ITR）及語音辨識技術（speech-to-text recognition, STR）等智慧機制進行 EFL 口說及對話練習。Smart UEnglish+旨在引導學生在真實情境中進行有意義且持續的口說及對話練習。該系統將學習內容融入日常生活，鼓勵學生探索身邊的事物，使學生無所不在地進行學習。此外，透過 Google Dialogflow與聊天機器人和同儕的設計對話，幫助學生在真實環境中進行英語對話。本研究共有49名學生分為兩組參與實驗，實驗組具有ITR和STR輔助，控制組具有STR輔助。此外，實驗組具有聊天機器人和同儕對話練習，但是控制組只有聊天機器人對話練習。使用可持續性和可擴展性的真實情境學習（Sustainable and Scalable Authentic Contextual Learning , SSACL）問卷進行定量和定性分析，調查兩組學生的學習行為、作業表現及學習成效。結果表明，實驗組在後測口試中明顯優於控制組。此外，實驗組中，ITR生成的詞彙和聊天機器人對話之間存在顯著相關性，表明ITR生成的詞彙可以幫助學生建構與聊天機器人的對話內容，並進行有意義的情景式對話練習。關於STR，兩組學生皆可以透過即時反饋糾正發音，有效地提升準確度和流暢度。接受問卷調查的實驗學生表達在真實情境中使用 Smart UEnglish+ 的積極性和強烈意願，這有助於他們在學習過程中習得更多的英語短句及新詞彙並多走動。建議在真實情境中使用Smart English+進行口說和對話練習，有效提高EFL學習。;Because daily life practices in authentic contexts can motivate learning interests and consolidate English skills in EFL learning, the rising question is how to support EFL learning effectively and smartly in authentic contexts. In this circumstance, smart learning environment with recognition technology and cloud service become promising and mature for EFL learning in authentic contexts. Therefore, we proposed Smart UEnglish+ app to help students practice EFL speaking and conversation with smart mechanisms like image-to-text recognition (ITR) and speech-to-text recognition (STR) technology. Smart UEnglish+ aims to guide students to conduct meaningful and continuous speaking and conversation practices in authentic contexts. This system connected EFL learning with daily life, encouraged students to apply English to their surroundings, and enabled them to learn ubiquitously. Furthermore, conversation with chatbot and peers were also designed to help English conversation in authentic contexts with Google Dialogflow. In this study, there were 49 students divided into two groups, the experimental group with ITR and STR and the control group with STR, and furthermore, the experimental group had chatbot and peer conversation, while the control group only had chatbot conversation. Quantitative and qualitative analysis with Sustainable and Scalable Authentic Contextual Learning (SSACL) questionnaire were used to investigate their learning behaviors, assignment performance, and learning achievements between two groups. The results found the experimental group outperformed the control group significantly in the oral test. Furthermore, there is significant correlation between ITR-generated and chatbot conversation in the experimental group, which indicated that ITR-generated words could help students construct conversation content with chatbot and conduct meaningful and situational conversation practices. Regarding STR, it was found both groups could effectively improve their pronunciation accuracy and fluency through feedback. Experimental students with the questionnaire survey had a positive and strong willingness to use Smart UEnglish+ in authentic contexts, which helped them learn more English phrases and new vocabulary and walk more during the learning process. It is suggested to use Smart English+ for speaking and conversation in authentic contexts and improve EFL learning effectively. |
