Effectiveness of Artificial Intelligence (AI) in language teaching

 

BY AUNG KHANT KYAW

---

Authors

Torres, P. J., & Kahveci, Y. E. (2025) (ScienceDirect)

---

Published In

Computers & Education: Artificial Intelligence, Volume 9, December 2025, United States. (ScienceDirect)

---

1. Brief Research Background

Artificial Intelligence (AI) has seen rapid adoption in English as a Foreign Language (EFL) instruction, especially after the global shift to online and blended learning following the COVID-19 pandemic. While many individual studies have explored AI tools for specific language skills (e.g., writing, speaking, vocabulary), there has been no comprehensive evidence synthesis examining AI’s overall effectiveness across multiple language learning domains.

Torres and Kahveci (2025) conducted a multilevel meta-analysis of empirical studies published between 2022 and 2025 to determine the overall impact of AI on language learning outcomes. The study analyzed a wide range of contexts, tools, and instructional settings to provide a clear understanding of how AI supports EFL learning. (ScienceDirect)

---

2. Literature Review (Brief)

Prior research on educational technology and language learning indicates that technology can enhance student learning when it facilitates active engagement and individualized feedback. AI tools — including chatbots, adaptive learning platforms, and intelligent tutors — can personalize learning experiences and offer real-time feedback, which may improve language skills such as vocabulary, writing, listening, and reading.

However, results across studies have been inconsistent, with some showing strong benefits and others showing weak or context-dependent effects. This prompted the need for a comprehensive meta-analysis to synthesize those results and draw broader conclusions about the effectiveness of AI in language instruction. (ScienceDirect)

---

3. Research Keywords

Artificial intelligence (AI)
Language teaching
English as a Foreign Language (EFL)
Meta-analysis
Learning outcomes
Instructional effectiveness
Educational technology (ScienceDirect)

---

4. Research Scope

The study focused on evaluating the effectiveness of AI tools in language teaching across a wide range of empirical research. The analysis included:

• 46 empirical studies from 2022–2025

• 117 effect sizes measuring outcomes across five major language skills: vocabulary, reading, writing, listening, and speaking

• Multiple instructional settings (face-to-face, blended, online)

• Different learner groups, including K-12 and higher education

The meta-analysis did not involve new experimental data; instead, it synthesized existing research findings to assess overall impact and moderator variables. (ScienceDirect)

---

5. Related Literature Topics

AI-enhanced language learning
Meta-analysis in education
Constructivist learning theories
Adaptive learning systems
Cognitive load and technology use
Blended and online learning environments
Language learning outcomes and engagement (ScienceDirect)

---

6. Overall Research Framework

The meta-analysis framework is based on several theoretical foundations:

• Constructivist Theory: Learning occurs through active interaction with tools and environments.

• Adaptive Learning Theory: AI can tailor instructional content to learners’ proficiency and needs.

• Cognitive Load Theory: Immediate feedback and guided practice can reduce unnecessary cognitive burden and support skill acquisition.

In this model, AI tools function as supplementary instructional supports that provide adaptive practice, feedback, and engagement opportunities. Higher engagement and personalized feedback are expected to lead to better language learning outcomes. (ScienceDirect)

---

7. Key Findings

The meta-analysis revealed several important results:

Overall Effectiveness

• AI tools have a significant positive impact on language learning outcomes (medium-to-large effect size). (ScienceDirect)

Skill-Specific Effects

• Vocabulary showed the strongest positive effect.

• Reading and writing also showed substantial improvements.

• Listening and speaking exhibited positive but relatively smaller effects. (ScienceDirect)

Contextual Moderators

• AI was more effective in face-to-face and blended instructional settings than in fully online environments.

• Younger learners (K-12) benefited more from AI tools than college students.

• Effectiveness was similar across different AI platforms, indicating that how AI is implemented matters more than which tool is used. (ScienceDirect)

Limitations in Certain Areas

• AI support did not significantly improve long-term learner self-regulation.

• Some complex language tasks requiring deep cognitive and metacognitive skills showed weaker benefits from AI alone. (ScienceDirect)

---

Reference

Torres, P. J., & Kahveci, Y. E. (2025). Effectiveness of artificial intelligence (AI) in language teaching: A multilevel meta-analysis across major language skills. Computers & Education: Artificial Intelligence, 9, 100522. https://doi.org/10.1016/j.caeai.2025.100522 (ScienceDirect)

---