VietTESOL International Convention 2026

The integration of generative AI into translator education has introduced a productive tension: although AI tools increase access to translation support, interaction designs that provide translations on demand may bypass the communicative reasoning central to translation competence. This study examines translation technique use and functional reasoning among 20 Vietnamese undergraduates completing seven English–Vietnamese IT translation tasks with a purpose-built Skopos-based chatbot designed to withhold translation assistance until learners articulated communicative purpose and target audience. Reflective journals (140 total; 2,591 coded extracts) were analysed using directed content analysis (Hsieh & Shannon, 2005) with a theory-driven codebook derived from Molina and Hurtado Albir’s (2002) translation technique taxonomy and Nord’s (1997) skopos dimensions. Frequency, co-occurrence, and Shannon diversity analyses were applied to examine the distribution, alignment, and breadth of learners’ technique and reason repertoires. Three findings emerged. First, learners produced substantially more functional reason codes than technique codes per extract (1.841 vs. 1.245), suggesting that the chatbot scaffolded communicative reasoning beyond simple technique identification. Second, co-occurrence analysis revealed systematic alignment between techniques and functional motivations. Third, Shannon diversity analysis showed that learner repertoires remained broad and balanced across participants and tasks (normalized H = 0.851 for techniques; 0.887 for reasons), inconsistent with the strategic uniformity often associated with unreflective AI use. The findings support human-centered AI design in translator education and provide a rare empirical account of technique–reason alignment in AI-mediated English–Vietnamese IT translation.