Comparing the Impact of Synchronous and Asynchronous Generative AI -Assisted Learning in Language Education

• In Event: AI and Immersive Technologies in Language Learning and Communication

Abstract

Generative AI (GAI) is transforming language education by enabling scalable, personalized learning. Yet its impact in K–12 contexts remains underexplored, particularly across online modalities. Guided by Cognitive Load Theory and Bloom’s Taxonomy, we conducted a quasi-experiment with 161 Chinese students (ages 8–14) randomly assigned to synchronous or asynchronous classes using structured or unstructured GAI chatbots over ten weeks. Results showed asynchronous with structured GAI yielded the largest gains in vocabulary and listening with reduced cognitive load, whereas synchronous with structured GAI maximized critical-thinking gains, especially in deductive reasoning. Structured scaffolds proved decisive regardless of modality, underscoring the joint influence of AI design and learning mode on cognitive engagement and outcomes.

Authors

• Zhihui Zhang, Chinese University of Hong Kong

• Thomas K.F. Chiu, The Chinese University of Hong Kong

• Scott Aubrey, The Chinese University of Hong Kong