Engagement without attainment? Exploring AI-assisted language learning through SLA theories

Artificial Intelligence (AI) is increasingly used in second language acquisition (SLA), yet its pedagogical value remains contested. While most research focuses on mainstream chatbots, less is known about interactive AI tools that could be used for active, task-based learning. This study examined four experimental Google AI tools (Quick Draw, XYZ Toy, Say What You See, and GenChess) through core SLA theories. Sixty-four Japanese university students used the tools over two semesters, and their reflections were analyzed through a hybrid deductive-inductive coding approach. The findings suggest that the tools supported learner engagement and vocabulary awareness. They also created opportunities for peer interaction, although such interactions were often predictable and remained close to the surface of language learning. Deeper linguistic processing was also limited, as the tools rarely generated adaptive feedback or spontaneous negotiation of meaning. The study highlights the need for careful teacher mediation and learner agency when using such tools. It also shows that an AI-assisted approach requires careful pedagogical design if it is to support meaningful language learning.