The cross-cultural adaptation and psychometric evaluation of the Chinese version of the stigma scale for caring for individuals with sexually transmitted infections (STISS)-student version: a translat

Consider a first-year university student in Lanzhou who, the night before an oral examination, turns not to a textbook but to a conversational AI assistant. She rehearses answers, receives instant corrective feedback, and gradually feels something unexpected: the fear that had paralyzed her for three semesters begins to recede, replaced by a cautious confidence that she might, after all, have something worth saying in English. Whether this anecdotal shift—from AI-assisted practice to genuine willingness to speak—reflects a systematic psychological process is the central question motivating the present research. AI-powered tools—including large language model chatbots, intelligent grammar checkers, and automated pronunciation coaches—have entered mainstream EFL classrooms with remarkable speed. Surveys of Chinese university students consistently report that the majority use at least one AI tool for English study, with ChatGPT-type applications and AI translation assistants among the most common (; ). Yet adoption statistics tell us little about whether, and through what mechanisms, AI tool use translates into meaningful communicative engagement. Willingness to communicate (WTC), defined as the learner's readiness to initiate second-language (L2) communication given the choice (), is widely regarded as a proximal determinant of actual language use and long-term proficiency growth. Understanding how AI tool use is linked to WTC is therefore both theoretically important and practically urgent. Unlike research grounded in the technology acceptance model (TAM) or the unified theory of acceptance and use of technology (UTAUT), which seek to explain the antecedents of technology adoption, the present study does not examine acceptance ; instead, it asks how AI tool use shapes communicative behavior through intervening affective mechanisms. Accordingly, two research questions guide the study. RQ1: Does AI tool use predict WTC through sequential mediation via AI self-efficacy and foreign language enjoyment? RQ2: Does foreign language anxiety moderate the enjoyment-to-WTC pathway such that the beneficial effect of enjoyment is weaker for high-anxiety learners? Two theoretical frameworks guide the present investigation. Social cognitive theory (SCT; ) posits that self-efficacy beliefs—individuals' confidence in their capacity to perform specific tasks—are critical mediators between environmental affordances and goal-directed behavior. In the L2 domain, learners who perceive AI tools as effective scaffolds are likely to develop stronger beliefs in their own language ability, which should, in turn, expand their willingness to engage communicatively. Alongside SCT, control-value theory (CVT; ) proposes that academic emotions—including enjoyment—are generated by learners' appraisals of control over and value of learning activities. Because AI tools may heighten both perceived control (instant feedback, adjustable difficulty) and value (authentic communicative practice), they are well positioned to elicit foreign language enjoyment (FLE). FLE, in turn, has been identified as a robust positive predictor of WTC (; ). The present model therefore proposes a serial mediation pathway: AI Tool Use → AI Self-Efficacy → FLE → WTC.