The cognitive-affective impact of generative AI on EFL/ESL learning: A meta-analysis

Repository record updated on:

Oct. 17, 2026, 1:04 p.m.

Database record updated on:

Oct. 18, 2026, 6:04 p.m.

Last Harvested on:

Oct. 18, 2026, 6:04 p.m.

access-status:

open.access

bitstream:

[['https://scholarspace.manoa.hawaii.edu/bitstreams/c17aa4cc-7594-4cf3-b175-87d14cd801fa/download', '30_01_10125-73714.pdf']]

bitstream_txt:

[['https://scholarspace.manoa.hawaii.edu/bitstreams/0724bea2-8dc5-4a20-8f7f-15666fb2fcd1/download', 'full_text']]

collection:

Volume 30 Number 1, 2026

contributor.author:

Hsu, Liwei Wang, Yu-Chun

date.accessioned:

2026-10-16T21:42:28Z

date.copyright:

2026

date.issued:

2026-10-12

description.abstract:

Since ChatGPT’s public release in November 2022, generative artificial intelligence (GenAI) has rapidly transformed English as a Foreign/Second Language (EFL/ESL) education, yet its precise impact on learning outcomes remains subject to debate. This systematic review and meta-analysis synthesizes evidence from 33 pooled empirical studies and 6 narrative-synthesis studies examining GenAI’s effects on language learning outcomes from November 2022 to March 2026. Following PRISMA 2020 guidelines, we searched multiple databases, applied a pre-specified effect-size verification rule to screen for reporting anomalies and calculated random-effects models for three outcome domains. Results revealed a large overall effect (Hedges’ g = 1.253, 95% CI [0.954, 1.542], I² = 94.4%). Language performance outcomes showed strong effects (g = 1.197) with substantial heterogeneity (I² = 88.4%). Affective outcomes demonstrated a large but highly heterogeneous effect (g = 1.3347, I² = 98.3%) while self-regulation outcomes yielded a large but non-significant preliminary estimate (g = 1.400, p = .187, I² = 99.1%). Research design and intervention duration emerged as significant moderators with experimental designs producing the strongest effects. The literature is overwhelmingly concentrated at the tertiary level (87.9%), and a tertiary-only sensitivity analysis confirmed that pooled estimates are not driven by aggregation across developmental stages. Egger’s test indicated no evidence of publication bias in the corrected dataset (p = .49). Evidence certainty was rated MODERATE using GRADE criteria. These findings support GenAI’s effectiveness in EFL/ESL learning while emphasizing the importance of implementation design, the context-sensitivity of affective outcomes, and the need for sustained metacognitive scaffolding to promote self-regulation.

endingpage:

25

format.extent:

25

identifier.citation:

Hsu, L., & Wang, Y.-C. (2026). The cognitive-affective impact of generative AI on EFL/ESL learning: A meta-analysis. Language Learning & Technology, 30(1), 1–25. https://doi.org/10.64152/10125/73714

identifier.doi:

https://doi.org/10.64152/10125/73714

identifier.issn:

1094-3501

identifier.uri:

https://hdl.handle.net/10125/73714

language:

en

llt.topic:

Review Articles

number:

1

publicationname:

Language Learning & Technology

publisher:

University of Hawaii National Foreign Language Resource Center Center for Language & Technology

rights:

http://rightsstatements.org/vocab/InC/1.0/

rights.license:

Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License

rights.uri:

https://creativecommons.org/licenses/by-nc-nd/4.0/

site_url:

/item/10125-73714/

startingpage:

1

subject:

Generative AI, meta-analysis, EFL/ESL learning, language performance outcomes, affective outcomes, self-regulation

title:

The cognitive-affective impact of generative AI on EFL/ESL learning: A meta-analysis

type:

Article Text

volume:

30