This study investigated the feasibility of using artificial intelligence (AI)-generated reformulations as written feedback in second language (L2) writing contexts. A comparative analysis was conducted to assess the degree of lexical resemblance between reformulated texts by human individuals and those by AI. Following Cohen's definition of reformulation, seven first language (L1) English as a foreign language (EFL) university instructors and three types of AI chatbots − ChatGPT-3.5, Claude2 and Bing Chat Creative Mode − reformulated picture description narratives written by five L1 Japanese EFL university students. Utilizing corpus analysis tools, the study examined text lengths, lexical sophistication, lexical diversity and incorporation of new vocabulary and multiword units. The findings revealed that among all the reformulator types, human-generated reformulations most closely resembled the original drafts in text length and vocabulary usage. Furthermore, among the AI systems, Bing Chat Creative Mode was found to be the most comparable to human reformulators in these aspects. In contrast, ChatGPT-3.5 and Claude2, showcased more advanced vocabulary and greater lexical diversity. The study concludes with pedagogical implications.
AckermannKChenY (2013) Developing the Academic Collocation List (ACL): a corpus-driven and expert-judged approach. Journal of English for Academic Purposes12(4): 235–247.
2.
AdamsR (2003) L2 output, reformulation and noticing: implications for IL development. Language Teaching Research7(3): 347–376.
3.
AmirjaliliFNeysaniMNikbakhtA (2024) Exploring the boundaries of authorship: a comparative analysis of AI-generated text and human academic writing in English literature. Frontiers in Education9: 1–11.
4.
AnthonyL (2022) TagAnt (Version 2.5) [Computer software]. Tokyo: Waseda University.
Berber SardinhaTB (2024) AI-generated vs human-authored texts: a multidimensional comparison. Applied Corpus Linguistics4(1): 100083.
8.
BiberD (2004) If you look at…: lexical bundles in university teaching and textbooks. Applied Linguistics25(3): 371–405.
9.
ChenYLiuC (2021) Reformulation on Chinese EFL learners’ email literacy. Pragmatics & Beyond New Series328(1): 15–40.
10.
ChenY-SWuH-J (2022) Developing sustainable email pragmatic competence for EFL learners through reformulation. Sustainability14(24): 16868.
11.
CohenAD (1982) Writing like a native: The process of reformulation. ERIC Document No. ED224338. Available at: https://eric.ed.gov/?id=ED224338 (accessed 19 November 2025).
12.
CohenAD (1983) Reformulating second-language compositions: A potential source of input for the learner. ERIC Document No. ED228866. Available at: https://eric.ed.gov/?id=ED228866 (accessed 19 November 2025).
13.
CoyleYFérez MoraPASolís BecerraJ (2020) Improving reference cohesion in young EFL learners’ collaboratively written narratives: is there a role for reformulation?System94: 102333.
14.
Di San PietroCBFrauFMangiaterraV, et al. (2023) The pragmatic profile of ChatGPT: assessing the communicative skills of a conversational agent. Sistemi Intelligenti35(2): 379–399.
ErmanBWarrenB (2000) The idiom principle and the open choice principle. Text & Talk20(1): 29–62.
17.
FredrickDRCravenL (2025) Lexical diversity, syntactic complexity, and readability: a corpus-based analysis of ChatGPT and L2 student essays. Frontiers in Education10: 1–8. 10.3389/feduc.2025.1616935.
18.
FujisawaYShintaniN (2025) Comparative effects of direct written corrective feedback and AI-generated reformulation on second language learners’ depth of processing and grammatical accuracy. System133: 103765.
19.
GeorgiouGP (2024) Differentiating between human-written and AI-generated texts using linguistic features automatically extracted from an online computational tool. arXiv. Available at: https://doi.org/10.48550/arxiv.2407.03646(accessed 19 November 2025).
20.
GlassGVPeckhamPDSandersJR (1972) Consequences of failure to meet assumptions underlying the fixed effects analyses of variance and covariance. Review of Educational Research42(3): 237–288.
21.
HanaokaO (2007) Output, noticing, and learning: an investigation into the role of spontaneous attention to form in a four-stage writing task. Language Teaching Research11(4): 459–479.
22.
HanaokaOIzumiS (2012) Noticing and uptake: addressing pre-articulated covert problems in L2 writing. Journal of Second Language Writing21(4): 332–347.
23.
HeatonJB (1996) Composition Through Pictures. London: Longman.
24.
HerboldSHautli-JaniszAHeuerU, et al. (2023) A large-scale comparison of human-written versus ChatGPT-generated essays. Scientific Reports13(1): 1–11. 10.1038/s41598-023-45644-9.
HuMNationISP (2000) Unknown vocabulary density and reading comprehension. Reading in a Foreign Language23(1): 403–430.
27.
HylandF (2000) ESL Writers and feedback: giving more autonomy to students. Language Teaching Research4(1): 33–54.
28.
IzumiS (2002) Output, input enhancement, and the noticing hypothesis: an experimental study on ESL relativization. Studies in Second Language Acquisition24(4): 541–577.
29.
KangEY (2020) Using model texts as a form of feedback in L2 writing. System89: 102196.
30.
KimHRBowlesM (2019) How deeply do second language learners process written corrective feedback? Insights gained from think-alouds. TESOL Quarterly53(4): 913–938.
31.
KyleKCrossleySAJarvisS (2021) Assessing the validity of lexical diversity using direct judgements. Language Assessment Quarterly18(2): 154–170.
32.
LapkinSSwainMSmithM (2002) Reformulation and the learning of French pronominal verbs in a Canadian French immersion context. The Modern Language Journal86(4): 485–507.
33.
LauferB (1989) What percentage of text-lexis is essential for comprehension? In: LaurenCNordmanM (eds) Special Language: From Humans to Thinking Machines. Bristol: Multilingual Matters, 316–323.
34.
LiuDDembergV (2023) ChatGPT vs human-authored text: insights into controllable text summarization and sentence style transfer. arXiv. Available at: https://doi.org/10.48550/arxiv.2306.07799(accessed 19 November 2025 ).
35.
MartinezRSchmittN (2012) A phrasal expressions list. Applied Linguistics33(3): 299–321.
36.
MizumotoA (2015) Langtest (Version 1.0) [Web application]. Available at: http://langtest.jp (19 November 2025 accessed ).
MizumotoAPinchbeckGGMcLeanS (2021) Comparisons of word lists on new word level checker. Vocabulary Learning and Instruction10(2): 30–41.
39.
MizumotoAYasudaSTamuraY (2024) Identifying ChatGPT-generated texts in EFL students’ writing: through comparative analysis of linguistic fingerprints. Applied Corpus Linguistics4(3): 100106.
40.
PawleyASyderFH (1983) Two puzzles for linguistic theory: nativelike selection and nativelike fluency. In: RichardsJCSchmidtRW (eds) Language and Communication. London: Longman, 191–225.
41.
QiDSLapkinS (2001) Exploring the role of noticing in a three-stage second language writing task. Journal of Second Language Writing10(4): 277–303.
42.
SantosMLópez SerranoSManchónRM (2010) The differential effect of two types of direct written corrective feedback on noticing and uptake: reformulation vs. error correction. International Journal of English Studies10(1): 131–154.
43.
SasakiMMizumotoAMatsudaPK (2024) Machine translation as a form of feedback on L2 writing. International Review of Applied Linguistics in Language Teaching63(4) https://doi.org/10.1515/iral-2023-0223.
44.
SchmiderEZieglerMDanayEBeyerLBühnerM (2010) Is it really robust? Reinvestigating the robustness of ANOVA against the normal distribution assumption. Methodology: European Journal of Research Methods for the Behavioral and Social Sciences6(4): 147–151. 10.1027/1614-2241/a000016.
45.
Simpson-VlachREllisNC (2010) An academic formulas list: new methods in phraseology research. Applied Linguistics31(4): 487–512.
46.
SwainM (2005) The output hypothesis: theory and research. In: HinkelE (ed.) Handbook of Research in Second Language Teaching and Learning. Mahwah, NJ: Lawrence Erlbaum Associates, 471–483.
47.
Tocalli-BellerASwainM (2005) Reformulation: the cognitive conflict and L2 learning it generates. International Journal of Applied Linguistics15(1): 5–28. 10.1111/j.1473-4192.2005.00078.x.
48.
VygotskyLS (1978) Mind in Society: Development of Higher Psychological Processes. Cambridge, MA: Harvard University Press.
49.
WuJ (2025) A corpus-based multidimensional analysis of linguistic features between human-authored and ChatGPT-generated compositions. International Journal of Linguistics Literature & Translation8(5): 102–110.
50.
YangLZhangL (2010) Exploring the role of reformulations and a model text in EFL students’ writing performance. Language Teaching Research14(4): 464–484.
51.
ZenkerFKyleK (2021) Investigating minimum text lengths for lexical diversity indices. Assessing Writing47: 100505.
52.
ZhangMCrosthwaiteP (2025) More human than human? Differences in lexis and collocation within academic essays produced by ChatGPT-3.5 and human L2 writers. International Review of Applied Linguistics in Language Teaching: 1–28https://doi.org/10.1515/iral-2024-0196.
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
