Artificial Intelligence in English Language Teaching: A Systematic Review of Emerging Practices

Abstract

Since 2020, large language models and intelligent tutoring systems have transformed English language teaching. This systematic review examines empirical studies from 2020 to 2026 to identify emerging practices, learner outcomes, and ethical challenges. Using PRISMA 2020 guidelines, we selected 63 peer-reviewed studies from Web of Science and Scopus and assessed them with the Mixed Methods Appraisal Tool. Research increased sharply after 2022, mainly in Asian EFL contexts with quantitative methods. Thematic analysis shows that AI tools, especially ChatGPT and conversational agents, improve writing and speaking skills. ChatGPT-assisted writing yields large effects (d=0.78) on grammatical accuracy and lexical diversity. AI also lowers foreign language anxiety and promotes learner autonomy via instant, non-judgmental feedback. Teachers' roles have shifted from delivering knowledge to "AI orchestration," requiring digital literacy. However, ethical issues persist, including academic integrity risks, bias against non-standard varieties, and a "digital divide" in low-resource settings. AI holds promise for personalised ELT but requires teacher support and institutional guidelines. Future studies should use longitudinal designs to test long-term gains.