DuoTest Flex: an AI-driven alternative assessment approach for beginner Spanish instruction

As artificial intelligence (AI) continues to influence language education, DuoTest Flex (DTF), an original Spanish assessment model, presents an alternative assessment model designed to balance learner engagement, real-world relevance, and assessment validity. The model combines three components: gamified micro-assessments delivered through Duolingo for Schools, a reduced formal test, and flexible options for learners. It is grounded in Communicative Language Testing, Self-Determination Theory, and Microlearning principles to enhance learning outcomes, promote autonomy, and reduce performance pressure. This study implemented DTF over a semester in CEFR-aligned beginner Spanish courses at a Malaysian technical university with 57 participants. A convergent parallel mixed-methods design was used to analyze learner performance and perceptions. Quantitative data from engagement metrics, test scores, and questionnaires were complemented with a thematic analysis of open-ended qualitative feedback. Results showed greater engagement, higher enjoyment, and improved listening proficiency, alongside more varied outcomes in writing performance among participants. The study positions DTF as a data-informed alternative to traditional testing and supports the integration of AI-enabled tools into language assessment in higher education.