當人工智慧觸發語句嵌入情緒與文化線索：軍校生的英語溝通行為路徑如何被改變？

When AI Trigger Sentences Embed Emotional and Cultural Cues: How Do They Reshape Cadets’ English Communication Pathways?

本研究探討如何運用人工智慧指令所設計的情緒與文化觸發語句，在為期10週的環景情境教室英語課程中，形塑軍校生的社會情緒學習、跨文化溝通能力與語言行為發展。研究以ChatGPT指令建置具固定互動回合的口說機器人，並在特定互動序列中嵌入情緒與跨文化觸發語句，以引發可追蹤且一致的行為反應。透過序列分析比較實驗中期與末期的互動結果顯示：社會情緒行為由情緒標記逐步發展至情緒解釋與調節；跨文化能力從文化察覺提升至詮釋與協商；語言產出則由迴避逐漸轉向較高流暢度。量化分析亦顯示跨文化溝通能力顯著提升（t(29) = 4.28, p < .001,d = .78），與行為軌跡相互支持。研究指出，促進高層次英語溝通行為的關鍵並非沉浸式科技本身，而是人工智慧觸發語句的精準設計與控制。理論上，本研究提供軍事語境下社會情緒學習與跨文化溝通能力發展的行為證據；實務上則提出人工智慧語言學習系統的設計原則。未來建議人工智慧口說系統加入基於二語習得理論的鷹架支持與精確度回饋，以更全面促進學習者的語言與跨文化能力發展。

This study examines how AI-designed emotional and cultural trigger sentences shape military cadets’ socio-emotional learning, intercultural communicative competence, and linguistic behavior during a ten-week panoramic classroom-based English program. A ChatGPT-driven speaking robot was constructed using fixed command prompts, embedding emotional and intercultural triggers at predetermined sequential positions to elicit traceable and consistent behavioral responses. Generalized sequential querier (GSEQ) sequential comparisons between the mid- and end-points of the intervention revealed clear developmental patterns: socio-emotional behaviors progressed from labeling to explanation and regulation; intercultural competence advanced from cultural noticing to interpretation and negotiation; and linguistic production shifted from avoidance to greater fluency. Quantitative analysis further confirmed significant improvement in intercultural competence (t(29) = 4.28, p < .001, d = .78), aligning with observed behavioral trajectories. Findings indicate that the driver of higher-order communicative development is not immersive technology itself, but the precision and control of AI-generated trigger design. Theoretically, the study provides behavioral evidence for socio-emotional and intercultural development in military learning contexts; practically, it offers design principles for AI-mediated language learning systems. Future AI speaking systems should incorporate scaffolding and accuracy-oriented feedback grounded in second language acquisition theory to further support learners’ linguistic and intercultural growth.