Truace
Subscribe
70/100

Moderate evidence

Cite
More
Export CSVExport JSONEmailLinkedInX

Review of 15 studies shows AI helped second-language learners, most with vocabulary

Study title: A meta-analysis examining AI-assisted L2 learning

Evidence type
Systematic review or meta-analysis
Sample
n = 2,156
Sources
1 sourcePeer-reviewed
Published
1 min read
Truvace record
TRV-2026-2840cite or check it

Study title: A meta-analysis examining AI-assisted L2 learning

A meta-analysis pooled 15 quantitative studies covering 2,156 participants and 53 effect sizes to test whether AI-assisted instruction improves second-language learning. After correcting for measurement and sampling error, it found a large positive effect of d = 1.167.

Read the study abstract

Abstract Numerous quantitative studies have investigated how artificial intelligence (AI) impacts the development of second language (L2). While individual studies delve into the effects of AI interventions on L2 learning, a meta-analysis provides a comprehensive evaluation of AI’s effectiveness in second language acquisition (SLA). Despite the growing body of meta-analytical research in AI-assisted language learning, several potential moderators have not been thoroughly investigated in previous meta-analyses. T…

The gain

A meta-analysis of 15 studies with 2,156 participants found AI-assisted second-language learning had a large positive effect (d = 1.167) after correcting for measurement and sampling error.

In brief

The gains were uneven across skills and settings. AI worked better for receptive skills than productive ones, and for vocabulary more than other language skills. It was also more effective in classrooms than outside them.

The pooled results showed no significant difference between K-12 and college learners, and the type of AI-assisted interaction did not significantly change effectiveness. The authors ran moderator analyses because true effects varied significantly from study to study.

Main points

  1. The analysis pooled 15 studies, 2,156 participants and 53 effect sizes on AI-assisted second-language learning.
  2. The corrected overall effect was large and positive, d = 1.167.
  3. AI-assisted learning was more beneficial for receptive skills than for productive skills.
  4. AI technologies were especially strong at building vocabulary compared with other language skills.
  5. In-class use was more effective than out-of-class use, and IMALL outperformed ICALL; K-12 and college learners did not differ significantly.

The full story

The authors set out to fill a gap in earlier meta-analyses of AI-assisted language learning, which had not thoroughly examined several potential moderators. They pooled quantitative studies and corrected effect sizes for measurement and sampling error.

The Q statistic indicated that true effect sizes varied significantly across studies, which led to a theory-based moderator analysis. Type of AI-assisted interaction was not a significant moderator.

Differences did appear by skill and setting. Receptive skills and vocabulary benefited most, in-class contexts beat out-of-class contexts, and IMALL was more impactful than ICALL. Effects for K-12 and college learners were not significantly different.

Sources

1 on file
  1. International Review of Applied Linguistics in Language Teachingdoi:10.1515/iral-2024-0213
    Peer-reviewed2025-03-17

Talk it over

Four questions to start onWho gets this benefit, and who was left out of the study or story?

Start with a question

Review of 15 studies shows AI helped second-language learners, most with vocabularyhttps://truvace.com/item/a-meta-analysis-examining-ai-assisted-l2-learning

  1. Who gets this benefit, and who was left out of the study or story?

    The gain A meta-analysis of 15 studies with 2,156 participants found AI-assisted second-language learning had a large positive effect (d = 1.167) after correcting for measurement and sampling error.

    Answer in the thread

  2. How 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 2,156, from International Review of Applied Linguistics in Language Teaching.

    Answer in the thread

  3. What does it leave open? What else would you want to know before acting on it?

    The record’s own caveat The Q statistic showed that true effect sizes varied significantly across the included studies, so the single pooled figure of d = 1.167 does not apply uniformly to every learning context.

    Answer in the thread

  4. Have 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

More from Education