Abstract
The rapid integration of artificial intelligence (AI) into language learning environments is reshaping translation pedagogy while raising important ethical and behavioral considerations. This research examines the structural relationship between students' ethical perceptions of AI-assisted translation and their engagement in language learning activities. Data were gathered from 525 undergraduate English learners through a structured questionnaire addressing AI usage patterns, ethical perceptions (ethical awareness, academic integrity concern, algorithmic bias perception), and engagement behaviors. The dataset was analyzed using SPSS, applying descriptive statistics, ANOVA, correlation analysis, regression analysis, and cluster analysis. The results indicate that: (1) Ethical perceptions drive engagement: Ethical awareness and algorithmic bias perception are significant positive predictors of critical engagement, suggesting that recognizing system limitations stimulates deeper cognitive participation and "informed trust" rather than disengagement; (2) Integrity regulates reliance: Academic integrity concern is significantly and negatively associated with AI reliance, acting as a regulatory mechanism against over-dependence; and (3) Distinct learner profiles: Cluster analysis identified three distinct learner profiles: "Integrated Adopters" (High-Engagement/High-Trust), "Passive Dependents" (Low-Engagement/High-Reliance), and "Cautious Skeptics" (Low-Engagement/Low-Reliance). This research concludes that ethical literacy is a critical cognitive determinant of learning behavior. It offers empirical guidelines for shifting from prohibitive ethics to competency-based instruction that fosters responsible and engaged AI use in translation education.