The DOI matches its Crossref record; 37 of the 37 references it cites were found in a registry or an index; none of the 55 sources its claims cite is checked for support yet.
Identity, Agency, and Human–AI Mediation in L2 Academic Writing: An Interactional Study
Open accessAbstract
Generative language models are increasingly used in academic writing, yet research on AI-assisted writing often reduces them to instructional tools or feedback providers and relies mainly on outcomes or self-reports. This study instead examines human–AI coordination as the analytic focus and conceptualizes composing as a distributed activity involving writers, prompts, system outputs, and evolving drafts. Drawing on human–computer interaction and distributed cognition, it reports a 12-week longitudinal qualitative study of 25 multilingual L2 academic writers. Data include AI interaction logs, time-stamped draft histories, screen recordings, and stimulated recall interviews anchored in observable interactional moments. Analysis of episodes linking prompts, AI output, writer uptake, and draft change shows that agency emerges through cycles of proposal, evaluation, constraint tightening, and consolidation, while identity is negotiated through selective alignment with AI-generated academic voice and disciplinary norms.
Overview
CoChat Summary
TL;DR
This 12-week qualitative study of 25 multilingual L2 academic writers treats human–AI composing as the analytic object, showing that agency and identity are enacted through observable prompt–output–uptake–revision sequences rather than residing in writers or tools. Writers cycled through proposal, evaluation, constraint tightening, and consolidation, selectively aligning with or distancing from AI's standardized academic voice.
Key findings
- Agency emerged as an interactional accomplishment across composing stages: writers rarely accepted AI output wholesale, instead cycling through provisional acceptance, evaluation, modification, and rejection. In one episode (E01), P05 requested a stronger topic sentence, asked the system to "make it slightly more cautious," then adopted the sentence and added a human-authored transition.
- Resistance was enacted through repeated deletion and constraint tightening rather than explicit refusal; P14 issued three successive narrowing prompts within a two-minute span and deleted each AI output without insertion.
- Identity was negotiated through micro-level revision decisions: P08 changed the AI's categorical "reduces" to the modalized "may reduce" and planned a personal example; P03 kept an AI transition's syntactic frame but replaced abstract nominalizations and "textbook" vocabulary; P19 deleted a polished AI paragraph entirely, stating "If I kept it, the paper would not feel like mine anymore."
- Longitudinally, early interactions showed higher prompting density and greater tolerance for provisional AI uptake, while later stages emphasized revision, consolidation, and selective suppression, with no participant relinquishing control over authorial consolidation.
Methods
- Interpretive qualitative design grounded in distributed cognition and situated action; 25 third- and fourth-year undergraduates (18 female, 7 male; ages 20–23; CEFR B2–C1) in one Advanced Writing course at a public university in Jordan, completing three tasks (argumentative essay, reflective narrative, short research paper) over 12 weeks using GPT-4o-mini.
- Data: complete ChatGPT interaction logs for all 25 participants, time-stamped draft version histories, 24 screen-recorded sessions from a purposively selected subset of 8 participants, and stimulated recall interviews anchored to specific time-stamped moments (used only to contextualize, not as independent analytic evidence).
- Analysis segmented synchronized data into "interactional episodes" (prompt → AI output → writer uptake/resistance → draft change), coded with an eight-move interactional codebook, then subjected to cross-episode/cross-case trajectory analysis and negative-case analysis; delayed recoding of a random 15% of episodes yielded 94% agreement.
Limitations
- Stated: conducted in a single Advanced Writing course at one public university in Jordan, so findings support analytical transferability rather than statistical generalizability; patterns may differ across L2 settings, disciplines, proficiency levels, genres, or AI platforms; results are tied to GPT-4o-mini, whose output variability, prompt sensitivity, and tendency toward generic or unsupported suggestions may partly explain writers' resistance and reformulation; the study does not evaluate learning outcomes, improvement, or instructional effectiveness.
- Evident from the text: screen recordings and recall interviews covered only 8 of the 25 participants (24 sessions), so the fine-grained real-time evidence rests on a subset; stimulated recall data were deliberately excluded as independent analytic evidence, leaving interpretations of writer intent dependent on observable actions alone.
Model-written summary. Check it against the paper before relying on it.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
0 Valid · 1 Not valid · 41 Not checked · 142 Own results and interpretations
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Rather, they choose when to follow AI suggestions and when to reject them so that the text still sounds like their own work and fits the expectations of their academic field.”
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Rests on 2 assumptions not shown here.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“The study shows that writing with AI is not a simple process in which students just accept suggestions from a tool.”
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“while identity is negotiated through selective alignment with AI-generated academic voice and disciplinary norms.”
plainstates an association- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Rests on an assumption not shown here.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“They do not simply copy an ''academic'' style from AI.”
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“The findings also show that students think carefully about voice and identity.”
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
88 more claims
Create a free account to see all 184 claims and to verify the 42 references still open.
- It may state two separate claims as one. 44
- Probably not a substantial claim. 62
Citations
Cites
Lincoln, Y. S., Guba, E. G., & Pilotta, J. J. (1985) · International Journal of Intercultural Relations · doi:10.1016/0147-1767(85)90062-8 · Cited by 45,787
Modeling and remodeling writing
Hayes, J. R. (2012) · Written Communication · doi:10.1177/0741088312451260 · Cited by 893
Hyland, K. (2002) · Journal of Pragmatics · doi:10.1016/s0378-2166(02)00035-8 · Cited by 855
AI-Mediated Communication: Definition, Research Agenda, and Ethical Considerations
Hancock, J. T., Naaman, M., & Levy, K. (2020) · Journal of Computer-Mediated Communication · doi:10.1093/jcmc/zmz022 · Cited by 658
Proceedings of the 2018 CHI Conference on Human Factors in Computing Systems
(2018) · doi:10.1145/3173574 · Cited by 569
Artificial intelligence for education: Knowledge and its assessment in AI-enabled learning ecologies
Cope, B., Kalantzis, M., & Searsmith, D. (2021) · Educational Philosophy and Theory · doi:10.1080/00131857.2020.1728732 · Cited by 518 · cited 2×
“Such trajectories are consistent with distributed cognition accounts emphasizing that tools participate differently across stages of practice and that writing should be understood as temporally extended activity rather than as a single act (Cope et al., 2020; Hayes, 2012).”
Human-AI collaboration patterns in AI-assisted academic writing
Nguyen, A., Hong, Y., Dang, B., & Huang, X. (2024) · Studies in Higher Education · doi:10.1080/03075079.2024.2323593 · Cited by 449 · cited 3×
“Writing, in this sense, is constituted through coordination between human and nonhuman actors whose contributions are asymmetrical but interdependent (Nguyen et al., 2024).”
InfluentialPosthumanist Applied Linguistics
Pennycook, A. (2017) · Routledge eBooks · doi:10.4324/9781315457574 · Cited by 406
Case Study Research in Applied Linguistics
Duff, P. (2018) · Routledge eBooks · doi:10.4324/9780203827147 · Cited by 359
Voice in academic writing: The rhetorical construction of author identity in blind manuscript review
Matsuda, P. K., & Tardy, C. M. (2007) · English for Specific Purposes · doi:10.1016/j.esp.2006.10.001 · Cited by 316
Cited by
Unpacking EFL students’ engagement with GAI in digital multimodal composing
Zhang, Y., Zhang, L. J., Li, J., & Yang, Y. (2026) · Language learning & technology · doi:10.64152/10125/73712
Metadata
- Authors
- Aldamen, H., Almashour, M., & Jarrah, M.
- Affiliations
- University of Jordan (JO)
- Corresponding author
- Jarrah, Marwan
- Venue
- Sage Open
- Publisher
- SAGE Publications
- Published
- 2026-04
- Volume
- 16
- Issue
- 2
- ISSN
- 2158-2440 (ISSN-L)
- OpenAlex
- W7166466978
- Semantic Scholar
- 3caa6ea114e92cb8e82476cc09400ef2faa1f18d
- Cited by
- 1
- References
- 34
- Citations by year
- 2026: 1
- FWCI
- 6.07
- Type
- journal article
- Language
- en
- Open access
- gold
- doi.org
- Licence
- cc-by-nc
- creativecommons.org/licenses/by-nc/4.0/ (vor)
- journals.sagepub.com/page/policies/text-and-data-mining-license (tdm)
- Locations
- SAGE Open (publishedVersion, cc-by-nc)
- DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion)
- Topics
- Social Sciences › Linguistics and Language › Discourse Analysis in Language Studies
- Social Sciences › Arts and Humanities › Language and Linguistics › EFL/ESL Teaching and Learning
- Social Sciences › Education › English Language Learning and Teaching
- Keywords
- AI-assisted writing, generative language models, L2 academic writing, human-AI coordination, writer agency, academic identity, distributed cognition, multilingual writers, draft revision, disciplinary norms, academic voice
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
A language model drew the claims from the paper on 2026-11-18, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).
A second model labelled each claim on 2026-11-18 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
How to cite
Cite CoChat for the checks on this page, and the paper for its own findings.
- This page
- CoChat. (2026). Citation check: Identity, Agency, and Human–AI Mediation in L2 Academic Writing: An Interactional Study. https://app.cochat.ai/papers/10.1177/21582440261463001
- The paper
- Aldamen, H., Almashour, M., & Jarrah, M. (2026). Identity, Agency, and Human–AI Mediation in L2 Academic Writing: An Interactional Study. Sage Open. https://doi.org/10.1177/21582440261463001
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.