The Cognitive Reconstruction of L2 Writing:An Empirical Investigation of Error Correction Processes in Human GPT-4 Collaborative Writing
Abstract
correction. Thirty intermediate Chinese EFL learners were assigned to an Experimental Group (EG) using GPT-4 or a Control Group (CG)
using Grammarly. They composed an essay while their eye movements were tracked. Subjective cognitive load and stimulated recall data
were also collected. Results showed the EG produced texts with significantly higher accuracy and syntactic complexity than the CG, with
no fluency difference. Eye-tracking revealed the CG had prolonged, localized fixations on errors, indicating high monitoring load, while the
EG showed a more holistic, integrative attention pattern. The EG also reported lower mental demand and frustration, perceiving GPT-4 as a
"collaborative consultant" versus Grammarly as an "automated examiner." Findings support a model of cognitive reconstruction where LLM
collaboration offloads lower-level error processing, freeing working memory for higher-order concerns and mitigating traditional fluency
accuracy-complexity trade-offs.
Keywords
Full Text:
PDFReferences
[1] Kellogg, R. T. (1996). A model of working memory in writing. In C. M. Levy & S. Ransdell (Eds.), The science of writing: Theories,
methods, individual differences, and applications(pp. 5771). Lawrence Erlbaum Associates.
[2] Kohnke, L., Moorhouse, B. L., & Zou, D. (2023). ChatGPT for language teaching and learning. RELC Journal, 54(2), 537550.
[3] Lantolf, J. P., Thorne, S. L., & Poehner, M. E. (2020). Sociocultural theory and second language development. In B. VanPatten, G. D.
Keating, & S.Wulff (Eds.), Theories in second language acquisition(3rd ed.,pp. 223247). Routledge.
[4] Schmidt, R. (1990). The role of consciousness in second language learning.Applied Linguistics, 11(2), 129158.
[5] Skehan, P. (1998).A cognitive approach to language learning. Oxford University Press.
[6] Sweller, J. (1988). Cognitive load during problem solving: Effects on learning. Cognitive Science, 12(2), 257285.
DOI: http://dx.doi.org/10.70711/aitr.v3i9.9025
Refbacks
- There are currently no refbacks.
