Work

Long-Term Effects of LLM Tutors on EFL Learners’ Thinking Habits: A Systematic Review of Longitudinal and Follow-Up Evidence

Year: 2026
Type: preprint
Abstract: Large language model (LLM) tutors are increasingly used to provide individualized practice, feedback, and strategy coaching for English as a foreign language (EFL) learners. However, most evaluations focus on short-term performance gains, leaving uncertainty about whether these tools produce durable changes in learners’ thinking habits, such as reflective thinking, metacognitive monitoring, and critical reasoning, that persist beyond initial exposure. This systematic review will synthesize longitudinal and follow-up evidence on the long-term effects of LLM-based tutoring on EFL learners’ thinking habits. Using a PICO-T framework, we will include EFL learners (Population) who engage with LLM tutors or LLM-mediated tutoring systems (Intervention/Exposure), compared with non-LLM instruction, human tutoring, conventional digital tools, or baseline conditions (Comparators), and report validated or clearly operationalized measures of thinking habits (Outcomes) at two or more time points, including delayed post-tests or follow-up assessments (Timing). We will search Scopus, Web of Science Core Collection, ERIC, PsycINFO, and LLBA, alongside targeted grey literature sources. Two reviewers will independently screen, extract data, and appraise risk of bias using design-appropriate tools. Where outcomes are sufficiently comparable, random-effects meta-analysis will estimate pooled effects; otherwise, we will conduct structured narrative synthesis. The review will clarify whether LLM tutors support sustained thinking habit development in EFL contexts, identify design features associated with lasting benefits, and highlight gaps for future longitudinal research.
Language: English
Open Access status: green
OpenAlex ID: W7130678330
Cites:
Cited by:
Related to: 10