<![CDATA[Mobile-assisted language learning (MALL) apps often present generic activities that ignore the semantic meaning of real-world places and provide limited skill-specific, mastery-based progression. This pilot feasibility study introduces the Adaptive Contextual Task Engine (ACTE), a lightweight on-device recommender that personalizes tasks using location semantics, CEFR-aligned modules, mastery status, and performance timing. ACTE was evaluated with 10 university students aged 18–23 in a simulated café environment to balance ecological validity and experimental control. Participants completed three A2 speaking tasks, the System Usability Scale (SUS), and a five-item relevance questionnaire. Results showed a mean SUS score of 72.0, exceeding the benchmark of 68. Participants rated task appropriateness for the café at 4.3/5 and real-life usability at 4.7/5, while 90% agreed that the tasks reflected authentic language use. Qualitative feedback confirmed contextual authenticity but indicated the need for clearer scoring explanations. These findings suggest that ACTE offers a practical, privacy-conscious, and replicable framework for situated MALL by linking semantic place affordances with mastery-based progression in controlled real-world simulations.]]>
Copyrights © 2026
