The effectiveness of artificial intelligence in English instruction for speaking and listening skills - A meta-analysis

Q: What does the meta-analysis assess about AI in English education?

It assesses the extent to which AI contributes to students’ English speaking and listening skills through aggregated evidence across studies.

Q: What is the overall reported effect size for AI-enhanced instruction?

The analysis reports a substantial overall effect (SMD=0.981, 95% CI [0.571, 1.391], p<.001).

Q: How do speaking and listening outcomes differ in the results?

Speaking shows particularly notable improvement (SMD=1.033), while listening outcomes trend positively (SMD=0.714) but do not reach statistical significance.

Oct 15, 2026

Uploaded by OmBimo

The meta-analysis evaluates how artificial intelligence contributes to English speaking and listening skill development in educational contexts. A systematic review and meta-analytic synthesis followed PRISMA guidelines, drawing on 2017–2024 peer-reviewed experimental and quasi-experimental studies indexed in Scopus, ERIC, and EBSCOhost. Nineteen studies met criteria, showing a substantial overall effect of AI-enhanced instruction (SMD=0.981; 95% CI [0.571, 1.391], p<.001), with stronger improvements in speaking (SMD=1.033). Listening results were positive but not statistically significant. Study heterogeneity reflected differences in populations, settings, and AI applications. Risk-of-bias assessment indicated predominantly low-to-moderate bias, while publication bias checks supported reliability.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages20

Reading Time50 min

AccessFree to download

Article info

Abstract

Introduction

Rationale for AI in English education

AI-enabled pedagogical shift

Challenges in speaking and listening practice

Need for meta-analytic synthesis

Contemporary Educational Technology, 2025, 17(4), ep596 ISSN: 1309-517X (Online) OPEN ACCESS Research Article The effectiveness of artificial intelligence in English instruction for speaking and listening skills: A meta-analysis Thada Jantakoon 1 0000-0001-5935-5875 ThitiJantakun 2* 0009-0007-4250-7589 Kitsadaporn Jantakun 2 0009-0004-0856-6524 Weerapa Pongpanich 3 0009-0005-4540-5719 Rungfa Pasmala 4 0009-0009-7710-9593 Panita Wannapiroon 4 0000-0001-8633-5781 Prachyanun Nilsook 4 0000-0003-3019-3635 1 Rajabhat Maha Sarakham University, Mahasarakham, THAILAND 2 Roi Et Rajabhat University, Roi Et, THAILAND 3 Rangsit University, Pathum Thani, THAILAND 4 King Mongkut’s University of Technology North Bangkok, Bangkok, THAILAND * Corresponding author: [thiti100@reru.ac.th](thiti100@reru.ac.th) Citation: Jantakoon, T., Jantakun, T., Jantakun, K., Pongpanich, W., Pasmala, R., Wannapiroon, P., & Nilsook, P. (2025) . The effectiveness of artificial intelligence in English instruction for speaking and listening skills: A meta-analysis. Contemporary Educational Technology, 17(4), ep596. [https://doi.org/10.30935/cedtech/17310](https://doi.org/10.30935/cedtech/17310) ARTICLE INFO ABSTRACT Received: 24 Sep 2024 Accepted: 26 Jun 2025 The increasing integration of artificial intelligence (AI) in education has raised significant questions about its pedagogical value, especially in language learning. This meta-analysis examines the extent to which AI contributes to the development of English-speaking and listening skills. A systematic review of the literature was conducted by the preferred reporting items for systematic reviews and meta-analyses guidelines, utilizing peer-reviewed studies indexed in Scopus, ERIC, and EBSCOhost from 2017 to 2024. Nineteen studies met the inclusion criteria, all of which utilized experimental or quasi-experimental designs with measurable learning outcomes. The analysis reveals a substantial overall effect of AI-enhanced instruction (standardized mean difference [SMD] = 0.981, 95% confidence interval [0.571, 1.391], p < .001), with particularly notable improvements in speaking proficiency (SMD = 1.033). Although listening outcomes showed a positive trend (SMD = 0.714), the effect did not attain statistical significance. Considerable heterogeneity was noted across the studies, reflecting variations in learner populations, instructional contexts, and AI applications. Quality appraisal using the risk of bias in non-randomized studies of interventions framework indicated a predominantly low to moderate risk of bias. Publication bias analysis, including funnel plot symmetry and fail-safe N, further confirmed the reliability of the results. These findings highlight the advantages of AI in enhancing speaking skills within English instruction and underscore the need for further empirical studies to investigate its impact on listening comprehension. Collectively, the results provide timely, evidence-based guidance for educators and policymakers aiming to integrate AI effectively into language education. Highlight the advantages of AI in enhancing speaking skills within English instruction and underscore the need. Keywords: artificial intelligence, English instruction, speaking skill, listening skill, meta-analysis INTRODUCTION In recent years, the integration of artificial intelligence (AI) into educational contexts has initiated a paradigm shift in instructional methodologies, particularly in the domain of English language education (Ling, 2023) . As English continues to serve as the dominant global lingua franca, the demand for innovative and effective pedagogical approaches has intensified, prompting educators and researchers to explore the potential of AI in fostering language acquisition outcomes (Duisenova & Zhorabekova, 2023) . Advancementsin AI have enabled the development of adaptive learning platforms, real-time feedback systems, and int

FAQs

What does the meta-analysis assess about AI in English education?

It assesses the extent to which AI contributes to students’ English speaking and listening skills through aggregated evidence across studies.

What is the overall reported effect size for AI-enhanced instruction?

The analysis reports a substantial overall effect (SMD=0.981, 95% CI [0.571, 1.391], p<.001).

How do speaking and listening outcomes differ in the results?

Speaking shows particularly notable improvement (SMD=1.033), while listening outcomes trend positively (SMD=0.714) but do not reach statistical significance.

Oct 15, 2026

Uploaded by OmBimo