aceModerate evidence
More
Meta-analysis of 23 studies finds AI tools tied to large gains in English learning
Study title: Exploring the impact of artificial intelligence on English language teaching: A meta-analysis
- Evidence type
- Systematic review or meta-analysis
- Sample
- n = 23
- Sources
- 1 sourcePeer-reviewed
- Published
- 2 min read
- Truvace record
- TRV-2026-2831cite or check it
Study title: Exploring the impact of artificial intelligence on English language teaching: A meta-analysis
A meta-analysis pooled 23 peer-reviewed experimental and quasi-experimental studies published between 2019 and 2025 on AI in English as a Foreign Language teaching. It found a large overall effect on learning outcomes (Hedges' g = 1.10, 95% CI 0.75 to 1.44).
Read the study abstract
The proposed meta-analysis study examines the pedagogical role of artificial intelligence (AI)-based interventions in English as a Foreign Language (EFL) education through a synthesis of empirical evidence of 23 peer-reviewed experimental and quasi-experimental experiments published in 2019–2025. Scheduling a strict systematic review process in accordance with the PRISMA principles, the work investigates AI solutions, including chatbots, automated writing assessment platforms and virtual reality applications, as…
The gain
A meta-analysis of 23 experimental and quasi-experimental studies from 2019–2025 found AI tools were linked to a large pooled improvement in English-as-a-foreign-language learning outcomes (g = 1.10, 95% CI 0.75 to 1.44).
In brief
The gains showed up in writing accuracy, speaking fluency and learner motivation, along with lower anxiety and greater enjoyment. Measures of motivation and engagement rose more than countable measures such as word count.
Results varied widely between studies (I2 = 92.66%), so the average effect hides large differences in tools, learners and measurement methods.
Main points
- The review followed PRISMA procedures and covered 23 peer-reviewed experimental and quasi-experimental studies from 2019 to 2025.
- The AI tools studied included chatbots, automated writing assessment platforms and virtual reality applications.
- Effect sizes were calculated as Hedges' g in a random-effects model using Comprehensive Meta-Analysis software.
- The pooled effect was g = 1.10 (SE = 0.18, 95% CI 0.75 to 1.44).
- Affective measures such as motivation and engagement showed stronger increases than purely quantitative measures such as word count.
- Heterogeneity was high (I2 = 92.66%), pointing to contextual and methodological differences.
- The authors recommend combining AI with teacher input for ethical, learner-centred teaching.
The full story
The authors searched for experimental and quasi-experimental studies on AI in English as a Foreign Language classrooms and screened them using PRISMA principles, ending with 23 studies from 2019 to 2025. The interventions included chatbots, automated writing assessment platforms and virtual reality applications.
Pooling the studies in a random-effects model gave a large overall effect of g = 1.10 with a confidence interval of 0.75 to 1.44. The improvements covered writing accuracy, speaking fluency and learner motivation, and learners also reported less anxiety and more enjoyment.
A subgroup analysis found that the type of measurement tool moderated the effect: affective variables such as motivation and engagement rose more than quantitative ones such as word count. The authors recommend teachers and AI working together.
Sources
1 on file- Acta Psychologicadoi:10.1016/j.actpsy.2025.105649
Talk it over
Who gets this benefit, and who was left out of the study or story?
Meta-analysis of 23 studies finds AI tools tied to large gains in English learninghttps://truvace.com/item/exploring-the-impact-of-artificial-intelligence-on-english-language-teaching-a-m
Who gets this benefit, and who was left out of the study or story?
The gain A meta-analysis of 23 experimental and quasi-experimental studies from 2019–2025 found AI tools were linked to a large pooled improvement in English-as-a-foreign-language learning outcomes (g = 1.10, 95% CI 0.75 to 1.44).
Answer in the threadHow do we know? What can this kind of evidence show, and what can’t it?
The evidence This rests on a systematic review or meta-analysis with a sample of 23, from Acta Psychologica.
Answer in the threadWhat does it leave open? What else would you want to know before acting on it?
The record’s own caveat Heterogeneity across the 23 studies was very high (I2 = 92.66%), so results differ substantially by tool, learner group and measurement method.
Answer in the threadHave you seen AI used this way where you study, work or live? Did it go the same way?
Answer in the thread
The questions are the same on every record. The details under them are taken from this record as written.
What readers say