ACADEMIC FLUENCY AND COGNITIVE INSTABILITY IN MULTILINGUAL UNIVERSITY CLASSROOMS UNDER AI-SUPPORTED CONDITIONS

Abstract

The quick growth of the scope of generative AI’s impact on higher education not only affects information search procedures but is also expected to change the interaction dynamics in multilingual classrooms. Responses written with generative AI assistance typically sound fluent and structured and look academically consistent even though their conceptual stability may be in question in many instances. Such a paradox became one of the main factors that motivated the current research.

The present study examines the relationship between external fluency in academic discourse and cognitive stability in spontaneous interactions in multilingual university classrooms. It was conducted with first-year students from the faculty of “Cultural and Art” during the seven-week course of English at the B1 level with the participation of Kazakh, Russian, and Turkmen-speaking learners. The methodology comprised classroom observation, discourse analysis, and interactive methods applied under low AI use conditions. AI-based information could be employed by learners only in the initial segment of each class devoted to topic introduction and information search.

The results obtained from classroom observations revealed that preliminary preparation through AI assistance increased the level of structure fluency and vocabulary skills in response delivery in its early phases. However, with increasing spontaneity of clarification, contradiction, and argumentative dialogue, it becomes challenging to forecast cognitive stability in many cases. Indicators of hesitancy, correction processes, silencing, and translanguaging become common at those stages where independent conceptual development is required.