Brendan Flanagan

Japanese EFL students struggle with specific grammatical structures, revealing a critical need for targeted instructional strategies that address both accuracy and idiomaticity.

High-proficiency EFL learners engage in richer dialogues with AI chatbots, while low-proficiency learners often get stuck in repetitive feedback cycles.

LLM-generated EFL exercises reveal that question type significantly influences cognitive load and learning outcomes, challenging traditional CEFR assumptions.

Despite the push for open science, a comprehensive survey reveals significant gaps in the availability and use of open datasets within learning analytics research, highlighting the need for better data sharing practices.