Generative Artificial Intelligence and Cognitive Load in Second Language Learning: A Narrative Review

Abstract

Generative artificial intelligence is rapidly reshaping the landscape of second language learning by offering immediate linguistic support, automated feedback, and opportunities for interactive practice. While research has documented its pedagogical potential and its influence on learner affect, far less attention has been paid to the cognitive mechanisms through which AI affects learning processes. This article adopts Cognitive Load Theory as an analytical framework to examine how generative AI modifies the distribution of intrinsic, extraneous, and germane cognitive load during L2 learning tasks. By synthesizing findings from studies on conversational AI, human–AI interaction, digital communication, and AI-enhanced learning environments, the review shows that AI can reduce unnecessary cognitive burden, make complex tasks more manageable, and facilitate deeper processing under appropriate conditions. At the same time, AI may introduce new sources of cognitive strain or encourage superficial engagement when learners rely uncritically on its output. The review further discusses the instructional implications of these patterns, emphasizing the roles of task design, teacher orchestration, and emotional factors in regulating cognitive load. It concludes by identifying key directions for future research, including longitudinal analyses, individual differences, and the integration of cognitive and social perspectives. Overall, the cognitive load lens provides a productive foundation for understanding both the opportunities and the constraints of generative AI in L2 learning.