aceModerate evidence
More
Language-learning robots boosted results in a review of 11 studies, with group lessons faring best
Study title: Multimodality and social interactions in AI-Enhanced embodied robot-assisted language learning: A meta-analysis
- Evidence type
- Systematic review or meta-analysis
- Sample
- n = 595
- Sources
- 1 sourcePeer-reviewed
- Published
- 2 min read
- Truvace record
- TRV-2026-2767cite or check it
Study title: Multimodality and social interactions in AI-Enhanced embodied robot-assisted language learning: A meta-analysis
A meta-analysis of 11 empirical studies covering 595 learners examined physically embodied, AI-enhanced robots in second language settings. It found a positive overall effect on learning outcomes, with Hedges' g of 0.83 (95% CI 0.46 to 1.21).
Read the study abstract
Embodied robots are increasingly used to support second language (L2) learning, yet evidence on when and why they help remains mixed. This meta-analysis examines how robot design features and instructional conditions relate to L2 learning outcomes, drawing on multimodal learning theory and sociocultural theory. Following PRISMA 2020 guidelines, Web of Science and Scopus were searched for empirical studies that used physically embodied, AI-enhanced robots in L2 settings and reported sufficient statistics to compu…
The gain
A meta-analysis of 11 studies (595 learners) found AI-enhanced embodied robots were linked to better second language outcomes, with a pooled effect of Hedges' g = 0.83 (95% CI 0.46 to 1.21).
In brief
Of six design and instruction factors tested, only the robot-learner interaction format reached significance (p = .049): group-based formats showed larger effects than one-on-one interaction. Robots used as teaching assistants had the largest subgroup effect, though social role did not significantly moderate outcomes.
Richer multimodal and more autonomous designs were not consistently linked to stronger learning. The authors suggest outcomes related more to how robots are positioned within instruction than to technical sophistication, though heterogeneity was high and the evidence base is small.
Main points
- The review followed PRISMA 2020 guidelines and searched Web of Science and Scopus for studies using physically embodied, AI-enhanced robots in second language settings.
- Eleven studies with 17 effect sizes and 595 participants met the inclusion criteria.
- A three-level random-effects model with cluster-robust variance estimation gave a pooled Hedges' g of 0.83 (95% CI 0.46 to 1.21, p < .001).
- Heterogeneity was high (I2 = 84.4%), and the authors say the estimate should be read with caution.
- Group-based interaction showed larger effects than one-on-one interaction (p = .049), the only significant moderator of six.
- Robot morphology, communicative modality, autonomy, social role and intervention duration did not significantly moderate outcomes.
What the review found when it combined the studies
These are the review’s own pooled results, copied from its abstract. Truvace did not recalculate them.
Second language learning outcomes with embodied AI-enhanced robots
Hedges' g 0.83 (95% CI 0.46 to 1.21)
595 people
The review’s own words
yielded a positive overall effect (Hedges’ g = 0.83, 95% CI [0.46, 1.21], p < .001)
Code checks every number above against the review’s text before it is shown. The short labels are an AI model’s reading of that text, so the review’s own sentence sits under each one. A pooled figure is only as strong as the studies behind it. How Truvace uses pooled results
The full story
The analysis was framed by multimodal learning theory and sociocultural theory, and aimed to clarify when and why embodied robots help second language learners, given mixed earlier evidence.
Only studies that used physically embodied, AI-enhanced robots and reported enough statistics to compute effect sizes were included. The final set was 11 studies yielding 17 effect sizes among 595 learners.
The pooled effect was positive and statistically significant, but heterogeneity was high at 84.4%. Among six moderators, only interaction format reached significance, with group formats outperforming one-on-one use.
Robots in teaching-assistant roles showed the largest subgroup effect, but social role was not a significant moderator. More multimodal and more autonomous designs were not consistently tied to stronger learning, which the authors read as pointing to instructional positioning over technical sophistication.
Sources
1 on file- Educational Research Reviewdoi:10.1016/j.edurev.2026.100833
Talk it over
Who gets this benefit, and who was left out of the study or story?
Language-learning robots boosted results in a review of 11 studies, with group lessons faring besthttps://truvace.com/item/multimodality-and-social-interactions-in-ai-enhanced-embodied-robot-assisted-lan
Who gets this benefit, and who was left out of the study or story?
The gain A meta-analysis of 11 studies (595 learners) found AI-enhanced embodied robots were linked to better second language outcomes, with a pooled effect of Hedges' g = 0.83 (95% CI 0.46 to 1.21).
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 595, from Educational Research Review.
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 was high (I2 = 84.4%) and the evidence base was small, so the authors say the pooled estimate should be read with caution.
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