Original Research
Checkpointed Evaluation in AI-Assisted Integrated Reading-to-Write Revision: Conscious-Access Routines and No-AI Transfer
- Abstract
- Full text
- Metrics
Integrated reading-to-write (RTW) tasks in English for Academic Purposes (EAP) ask students to work carefully with sources. Writers must keep the original meaning, connect claims to evidence, and manage stance and voice when using several texts. Generative AI can make revision easier, but it may also produce wording that sounds convincing without enough evidence. This qualitative longitudinal multiple-case study examined how EFL university writers used conscious-access routines during AI-assisted revision and whether these routines later appeared in no-AI writing. Twenty-four third- and fourth-year English majors in a 16-week EAP course completed four RTW cycles. Each cycle included no-AI drafting and a 60-minute Gemini-supported revision session. Two revision conditions kept tool access and time-on-task the same. The checkpointed condition required uptake decisions, short reasons, and confidence ratings. The non-checkpointed condition allowed students to evaluate suggestions as they wished. Six data sources produced 384 coded revision episodes. The findings also showed clear differences between the two conditions. Evidence-first routines increased across cycles, while fluency-first routines decreased. No-AI milestone writing showed more evidence-first instances in the checkpointed condition at Weeks 8 and 12. Structured checkpointing may therefore help students use AI more responsibly by checking sources and evidence before accepting fluent suggestions.
Page 1 of
Download Count : 9
Visit Count : 181
Keywords
Integrated Reading-to-Write; Generative AI; Metacognition; Conscious Access; English for Academic Purposes
Author(s) Information
Publisher’s Note
The claims, arguments, and counter-arguments made in this article are exclusively those of the contributing authors. Hence, they do not necessarily represent the viewpoints of the authors’ affiliated institutions, or EUROKD as the publisher, the editors and the reviewers of the article.
Acknowledgements
The authors thank the students who participated in the study and contributed their time and experiences.
Funding
This research received no external funding.
CRediT Authorship Contribution Statement
Hesham Aldamen: Investigation, Data Curation, Writing – Original Draft, Writing – Review & Editing
Mohamad Almashour: Conceptualization, Methodology, Formal Analysis, Writing – Review & Editing
Marwan Jarrah: Conceptualization, Methodology, Supervision, Project Administration, Writing – Review & Editing
Generative AI Use Disclosure Statement
ChatGPT was used during revision for language editing, condensation, and formatting. It was not used for data collection, coding, analysis, or analytic decision-making. The authors reviewed all AI-assisted changes and take full responsibility for the manuscript.
Ethics Declarations
The study was conducted in accordance with the Declaration of Helsinki. The Institutional Review Board of the University of Jordan approved the study (Decision No. 437/2025, 13 May 2025). All participants provided informed consent, participation was voluntary, and the data were anonymized.
Consent for Publication
All participants provided informed consent for their anonymized data to be used in this research and for the findings to be published. No identifying personal information is included in the manuscript.
Competing Interests
The authors declare no competing interests.
Data Availability
Participant traces, screen recordings, and interviews are not publicly available because of ethics and consent restrictions. Non-identifying materials may be requested from the corresponding author, subject to institutional approval. Study instruments will be deposited in IRIS where permitted.