aceModerate evidence
More
AI voice chatbots boosted language learners' results, a review of 19 studies finds
Study title: Effects of speech-enabled LLM-based chatbots on L2 learning outcomes and affective domains: a multilevel meta-analysis
- Evidence type
- Systematic review or meta-analysis
- Sample
- n = 2,643
- Sources
- 1 sourcePeer-reviewed
- Published
- 1 min read
- Truvace record
- TRV-2026-2743cite or check it
Study title: Effects of speech-enabled LLM-based chatbots on L2 learning outcomes and affective domains: a multilevel meta-analysis
A multilevel meta-analysis pooled 19 studies, covering 32 samples and 2,643 second-language learners, on chatbots that use voice mode and large language models for speaking practice. The overall effect size was 0.68 for learning outcomes.
Read the study abstract
Abstract Recent developments in natural language processing and automatic speech recognition have significantly advanced chatbot communication using the voice mode, establishing these tools as viable resources for L2 speaking practice. More recently, the L2 chatbot domain has leveraged large language models (LLMs) to enhance contextual understanding and facilitate meaningful dialogues, prompting research into the effects of speech- and LLM-based chatbots on L2 learning outcomes and affective domains. This study…
The gain
A meta-analysis of 19 studies (32 samples, 2,643 learners) found speech-enabled LLM chatbots were linked to a pooled effect size of 0.68 on second-language learning outcomes and 0.64 on affective measures.
In brief
For affective domains, the overall effect size was 0.64, and both results were statistically significant (p < 0.001). Affective domains cover the emotional side of learning.
Moderator analyses showed that learner proficiency, educational level, interaction format and the context of chatbot use significantly influenced how effective the chatbots were, so the average gain does not apply equally to every learner group.
Main points
- The analysis pooled 19 studies with 32 samples and 2,643 learners in total.
- The overall effect size for second-language learning outcomes was 0.68 (p < 0.001).
- The overall effect size for affective domains was 0.64 (p < 0.001).
- Learner proficiency significantly moderated how effective the chatbots were.
- Educational level, interaction format and context of chatbot use also significantly influenced effectiveness.
The full story
The study is a multilevel meta-analysis, a method that handles multiple samples drawn from the same study. It looked at chatbots that combine automatic speech recognition with large language models to support spoken practice in a second language.
Two outcome families were examined: learning outcomes and affective domains. Pooled effects were 0.68 and 0.64 respectively, both significant at p < 0.001.
Moderator analyses found that learner proficiency, educational level, interaction format and the context in which chatbots were used shaped the size of the effect.
Sources
1 on file- International Review of Applied Linguistics in Language Teachingdoi:10.1515/iral-2026-0172
Talk it over
Who gets this benefit, and who was left out of the study or story?
AI voice chatbots boosted language learners' results, a review of 19 studies findshttps://truvace.com/item/effects-of-speech-enabled-llm-based-chatbots-on-l2-learning-outcomes-and-affecti
Who gets this benefit, and who was left out of the study or story?
The gain A meta-analysis of 19 studies (32 samples, 2,643 learners) found speech-enabled LLM chatbots were linked to a pooled effect size of 0.68 on second-language learning outcomes and 0.64 on affective measures.
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 2,643, from International Review of Applied Linguistics in Language Teaching.
Answer in the threadWhat does it leave open? What else would you want to know before acting on it?
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