aceModerate evidence
More
Meta-analysis finds AI-assisted language classes ease learners' anxiety and boredom
Study title: The effects of generative AI on foreign language emotions: A meta-analysis
- Evidence type
- Systematic review or meta-analysis
- Sample
- n = 3,461
- Sources
- 1 sourcePeer-reviewed
- Published
- 2 min read
- Truvace record
- TRV-2026-2778cite or check it
Study title: The effects of generative AI on foreign language emotions: A meta-analysis
A meta-analysis pooled 43 effect sizes from 23 experimental and quasi-experimental studies covering 3,461 learners. It found generative AI-assisted instruction produced a large overall improvement in foreign language emotions compared with traditional instruction (g = 0.788).
Read the study abstract
Despite increasing attention to generative artificial intelligence (GenAI) in second language education, its effects on foreign language (FL) emotions remain unclear due to the lack of systematic evidence. In the current meta-analysis study, 43 effect sizes from 23 experimental and quasi-experimental studies (N = 3461) were synthesized to explore the effects of GenAI-assisted instruction on FL emotions. A three-level random-effects model revealed a large and statistically significant overall effect (g = 0.788, p…
The gain
A meta-analysis of 23 studies with 3,461 learners found generative AI-assisted language instruction improved emotions more than traditional teaching (g = 0.788), with the largest gains in reduced anxiety and boredom.
In brief
The gains were not even. Reductions in negative emotions such as anxiety and boredom were larger than increases in positive emotion such as enjoyment, which suggests AI tools may work best at easing the discomfort of learning a language.
Differences by technological, pedagogical and social features of the tools were mostly limited. An early signal favouring one or two sessions per week over more frequent use disappeared in sensitivity analyses, so it should be read cautiously.
Main points
- The analysis synthesised 43 effect sizes from 23 experimental and quasi-experimental studies with 3,461 participants in total.
- A three-level random-effects model gave an overall effect of g = 0.788 (p < .001), described as large.
- Emotion type significantly moderated results: anxiety and boredom improved more than enjoyment.
- Moderator analyses based on the usefulness framework found limited evidence of stable differences across technological, pedagogical and social affordances.
- Lower session frequency (1-2 per week) first outperformed higher frequency, but the difference became non-significant once studies with unspecified frequency were excluded.
The full story
The study, published in Acta Psychologica, set out to fill a gap in systematic evidence on how generative AI in second language education affects learners' feelings. It pooled studies that compared AI-assisted instruction with traditional teaching.
Using a three-level random-effects model, the authors found a large overall benefit (g = 0.788). They interpret this as AI reducing negative emotions and raising positive ones, with a stronger effect on anxiety and boredom than on enjoyment.
Moderator analyses grounded in the usefulness framework generally did not show stable differences across technological, pedagogical or social affordances. The session frequency finding lost significance in sensitivity tests, and the authors say it should be interpreted cautiously.
Sources
1 on file- Acta Psychologicadoi:10.1016/j.actpsy.2026.107336
Talk it over
Who gets this benefit, and who was left out of the study or story?
Meta-analysis finds AI-assisted language classes ease learners' anxiety and boredomhttps://truvace.com/item/the-effects-of-generative-ai-on-foreign-language-emotions-a-meta-analysis
Who gets this benefit, and who was left out of the study or story?
The gain A meta-analysis of 23 studies with 3,461 learners found generative AI-assisted language instruction improved emotions more than traditional teaching (g = 0.788), with the largest gains in reduced anxiety and boredom.
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 3,461, 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 The apparent advantage of 1-2 sessions per week became non-significant when studies with unspecified frequency information were excluded, so that finding should be interpreted cautiously.
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