12,445 papers · continuously updated · last export: 10 Aug 2026livingmeta.ai
← Browse all papers
AI evidence extraction

ChatGPT in ESL Writing: L2 Learners’ Practices and Perspectives

Inyoung Na, Mahdi Duris, Volker Hegelheimer · 2025

AI-generated evidence extraction, verified across multiple analytical personas. Not a substitute for the peer-reviewed original.

7/10
Relevance
1/4
Quality (LMQS)
E
Evidence
0
Citations
0.00
FWCI

This is an AI analysis. Read the peer-reviewed original at the publisher: https://doi.org/10.31274/isudp.2025.211.04

Methodology & findings

Study design

Qualitative case study with mixed methods data collection.

Sample

N = 8, 3 groups

Primary method

The abstract indicates qualitative analysis of interaction patterns but does not specify statistical software or formal statistical methods. Analysis involved examination of learner prompt types, ChatGPT feedback types, learner response types, and assessment of modified output through case narratives.

Main result

The study found that "the participant who critically evaluated and selectively integrated ChatGPT's feedback (the Adapter type) showed the most beneficial patterns for autonomy and meaningful L2 development, while over-reliance on ChatGPT's generative capabilities (as observed for the Controller type) posed challenges for deeper engagement with the writing process."

Reports effect sizes.

Research paradigm

Qualitative/interpretivist

Author conclusions

The authors conclude that "This study contributes to the emerging study of AI-assisted writing by identifying specific interaction patterns that enhance or inhibit language development, providing insights for tailoring instructional strategies to maximize ChatGPT's potential in L2 writing pedagogy."

Risk of bias

Small sample size (n=8) limiting generalizability; Selection bias - participants self-selected into ESL writing program using ChatGPT; No comparison/control group; Potential observer effect from screen recording; Case narrative focus on three participants may not represent all eight participants equally; Single institution setting (U.S. university ESL program); Self-selection bias in participant recruitment; Potential observer effect from screen recordings and interviews; Limited temporal scope (3-week span); Small sample size (n=8) limits generalizability; Self-selection bias: participants volunteered for an ESL program; Potential Hawthorne effect: awareness of being recorded may influence behavior; Lack of control group for comparison

Open questions raised

  • The authors identify gaps in understanding AI-assisted writing by focusing on specific interaction patterns that enhance or inhibit language development, suggesting a need for future research on how to tailor instructional strategies to maximize ChatGPT's potential in L2 writing pedagogy.
  • The study addresses emerging gaps in understanding AI-assisted writing and the need for research on specific interaction patterns between L2 learners and ChatGPT that support language development.
Data: not_statedCode: not_statedExtracted from: pdfAgreement 62%

Explore related topics

AI Manuscript WritingAcademic WritingCase Study

Related papers

← Browse all papers