Mapping research on ChatGPT and LLMs in EFL writing: A scoping review with bibliometric analysis

AN ANALYSIS ON THE UTILIZATION OF CHATGPT IN ENHANCING STUDENTS’ ENGLISH LANGUAGE KNOWLEDGE FOR THESIS WRITING: A CASE STUDY OF TWO BEST GRADUATES

LLM-Generated Feedback in L2 Writing: A Scoping Review

The release of ChatGPT in November 2022 transformed second language (L2) writing instruction and led to rapid growth in research on large language model (LLM)-generated feedback; however, no synthesis has mapped this literature in terms of feedback quality, learner uptake, and pedagogical integration. This scoping review examines 185 empirical studies published between November 2022 and March 2026 that were identified through a Scopus search (n = 283 screened) and analysed using a systematic keyword-based charting framework applied to full abstracts, with full-text analysis of 35 studies. The review identifies four major patterns: (1) comparative AI–human feedback research dominates the literature (27.6%); (2) content-level feedback remains underexplored (13.5% of studies); (3) learner uptake is rarely measured as a primary outcome; and (4) LLM feedback is broadly comparable to teacher feedback for surface-level errors but weaker for content and argumentation, while learner perceptions often exceed demonstrated performance outcomes. Hybrid AI–teacher models show promising but underexamined potential, accounting for only 12.4% of the literature. The field shows a focus on perceptions rather than learning outcomes, an apparent tendency toward positive-results reporting, and no clear teaching models. This study proposes a typology of LLM feedback functions and outlines a research agenda focused on uptake, longitudinal outcomes, and hybrid AI–teacher integration.

Generative AI in University EFL Writing Instruction: A Systematic Literature Review

GenAI is most educationally defensible when integrated as a guided formative-feedback resource rather than as a substitute writer or replacement for teacher expertise, as well as for assignment design, AI-use disclosure, feedback literacy, prompt literacy, and process-based assessment.

Evaluating the Ability of ChatGPT as an AI-Powered Chatbot: Assisting EFL Students' English Scientific Paper Writing

ChatGPT is an AI-powered chatbot capable of generating human-like text based on the input provided through conversations like humans in natural language. This study aims to evaluate the capability of ChatGPT as an AI-powered chatbot in assisting EFL students in English scientific paper writing, with a particular focus on undergraduate theses. This research applies a content analysis approach to examine the textual outputs generated by ChatGPT in response to academic thesis-related prompts. The researcher conducted 13 trials by inputting various thesis-related prompts into ChatGPT. The queries covered multiple components of a thesis, including the title, background (research gap, research question, and significance), literature review, methodology (design, data collection, and analysis), findings, discussion, conclusion, and references. The results indicate that ChatGPT was able to generate coherent responses for most thesis components. However, its output for the findings, discussion, and conclusion was inadequate due to the absence of actual research data and context, which are essential for producing valid results. Additionally, when given a differently formulated research objective, ChatGPT occasionally provided inconsistent or inappropriate responses. While the tool demonstrates a strong contextual understanding of text and generates relevant content, its limitations highlight potential issues related to academic integrity, including plagiarism and ethical concerns. This study emphasizes that while ChatGPT can effectively support various stages of English scientific paper writing, its outputs require careful human evaluation, verification, and refinement before being incorporated into academic work.

Artificial Intelligence Tools and English Writing Development in Higher Education: A Scopus-Based Bibliometric Analysis Using Biblioshiny and VOSviewer

The use of generative AI has transformed the planning, drafting, revising, and assessment of English writing in higher education. Although research on the area has expended rapidly, there is remains a lack of a coherent understanding of the fields intellectual structure, publications patterns, key researchers, collaboration networks, and the thematic development. This study examines the evolution of research on AI tools and English writing in higher education and draws inferences for BS English programmes in Pakistani universities. A Scopus-based dataset was filtered based on year, subject-area, document-type, language, publication-stage, and source-type. From the 2,458 records initially identified, 1,034 English-language journal articles published in the Social Sciences at the final publication stage were included for the period 2020-2026. Performance analysis and science mapping were performed using Biblioshiny/bibliometrix and VOSviewer, respectively. The analysis examined annual scientific production and citations; productive sources and authors; locally cited documents; Bradford’s and Lotka’s laws; h-index indicators; country collaboration; keyword co-occurrence; thematic mapping; trend topics; Reference Publication Year Spectroscopy (RPYS); and a logistic model of the publication life cycle. Scientific production increased from 1 article in 2020 to 431 in 2025 and 354 in 2026; however, the 2026 figure should be interpreted with caution because the year may be unfinished. The conceptual structure was dominated by ChatGPT (257 occurrences), artificial intelligence (223 occurrences), generative AI (114 occurrences), and EFL writing (59 occurrences). Arab World English Journal was the most productive source, while Guo K and Wang Y were the most prolific authors. Bradford’s Law analysis displayed that 18 journals accounted for 34.3% of the literature, while Lotka’s Law showed a highly attentive authorship pattern. Thematic and temporal studies revealed a shift from general research on AI tools and chatbots toward generative AI, large language models, and EFL writing. Overall, the field is rising swiftly but remains uneven across methodological and geographical dimensions. The outcomes highlight the importance of curriculum-integrated AI literacy, process-oriented writing assessment, transparent AI-use disclosure, faculty development, and locally grounded empirical research in Pakistani universities.

A Qualitative Evaluation of ChatGPT-Generated Academic Writing in Higher Education

The purpose of this study is to evaluate academic texts generated by ChatGPT, in terms of academic language use, structural integrity, content quality, and source reliability. Within this scope, ChatGPT-4 was prompted to produce an academic text in Turkish on a specific topic, and the resulting text was assessed using a rubric developed by the researchers. Subsequently, the text was qualitatively analyzed through detailed document analysis. The findings reveal that while ChatGPT generally demonstrates adequate performance in academic language use (such as objective tone, academic style, and proper application of academic grammatical structures), certain shortcomings were observed in critical analysis and adherence to APA 7 formatting rules. Although the organizational structure of the text presents a largely coherent and logical flow, minor structural weaknesses (such as the lack of a roadmap in the introduction) were identified. In terms of content, arguments remained mostly superficial, the literature was not examined in depth, and some references could not be verified. Furthermore, instances of information being inconsistent with or misinterpreted from the cited sources were detected. These results indicate that generative AI tools can be supportive in academic writing, but their outputs require careful evaluation.