Single Paper Session 12

Session Page

Session Information

This page shows the session details and the presentations assigned to this session.

Back to programme overview
Add this session to calendar

Can Algorithm-based Feedback Help Students to Write Better? A Meta-analysis

Keywords
AI writing;EFL and ESL writing;L1;meta-analysis
Domain
Assessment and Evaluation

Abstract

Against the backdrop of rapid developments of algorithm-based feedback tools - from older tools mainly providing feedback on grammar and spelling to more advanced tools based on generative artificial intelligence offering more comprehensive writing support - our meta-analysis examines to what extent algorithm-based feedback improves not only surface- (e.g., grammar and spelling) but also deep-level (e.g., structure, content, coherence) writing outcomes for different (language) learners (first, second, and foreign language learners) at secondary school and university. Algorithm-based feedback tools may be very useful for language learners as they can provide timely feedback and help with revision (Escalante et al., 2023), which can be particularly relevant for foreign language (FL) learners who often have limited contact with first language (L1) speakers outside the language classroom, as opposed to second language (L2) learners.For this meta-analysis, we reviewed experimental and quasi-experimental studies published between 2011 and the end of 2024, covering five European languages in four different databases. Results from the 33 included studies indicated that algorithm-based feedback was beneficial for improving writing in general (g = 0.36). Specifically, positive effects were observed for surface-level outcomes at post-test (g = 0.31), though no lasting effects were found at maintenance (g = -0.02). In contrast, deep-level writing outcomes showed sustained improvement, with positive effects both at post-test (g = 0.31) and maintenance (g = 0.54). No significant differences between secondary and university students were observed. However, L2 learners, in general, seemed to profit most from algorithm-based feedback, showing gains in surface- (g = 0.77, bordering on significance), and deep-level outcomes (g = 0.46). While no significant differences were found between the effects of specific types of algorithm-based feedback tools in moderator analyses, feedback from Grammarly and Pigai statistically enhanced students’ writing but effects of ChatGPT feedback were non-significant. We discuss implications for future research and educational practice, also in light of the small transfer of learning from algorithm-based feedback to new writing tasks.ReferencesEscalante, J., Pack, A., & Barrett, A. (2023). AI-generated feedback on writing: insights into efficacy and ENL student preference. International Journal of Educational Technology in Higher Education, 20(1). https://doi.org/10.1186/s41239-023-00425-2

Strong but Not Static: Reading-Writing Connections in Primary Grades

Keywords
developmental aspects of writing ability;L1;reading and writing;writing models
Domain
Cognitive Science

Abstract

Strong but Not Static: Reading-Writing Connections in Primary Grades Aim: This study examines the relations between reading comprehension and written composition and the predictors of these relations, using longitudinal data from U.S. primary grade children.Theoretical Framework: The Interactive Literacy Model (Kim, 2020, 2025) posits that reading and writing are related through shared underlying skills (shared skills hypothesis). However, the magnitude of this relation is not fixed; rather, it varies as a function of multiple factors (dynamic relations hypothesis). We investigated three research questions: (1) What is the relation between reading comprehension and writing quality? (2) Does this relation vary by grade level (a proxy for development)? (3) What shared predictors explain reading comprehension and writing quality?Methods: We analyzed longitudinal data from 263 children across grades 1 and 2. Reading comprehension and written composition were assessed using multiple tasks. Shared predictors included oral discourse skills and lexical literacy skills (word reading and spelling), also measured by multiple tasks.Findings: Reading comprehension and writing quality were strongly related across grade levels, though the correlation was stronger in grade 1 (.81) than grade 2 (.70), supporting the dynamic relations hypothesis. Both oral discourse skills and lexical literacy skills explained the reading-writing relation. Furthermore, the relative contributions of these predictors to reading comprehension and writing quality differed between grades 1 and 2.Relevance: Writing is not an isolated skill. Many writing tasks involve reading source materials, and effective revision requires reading proficiency. Understanding the nature of reading-writing relations has important implications for both writing theory development and instructional practice. This study contributes empirical evidence for the dynamic nature of literacy connections during early development.Keywords: Reading-writing relations, shared skills, dynamic relations, interactive dynamic literacy model ReferencesKim, Y.-S. G. (2020). Interactive dynamic literacy model: An integrative theoretical framework for reading and writing relations. In R. Alves, T. Limpo, & M. Joshi (Eds.), Reading-writing connections: Towards integrative literacy science (pp. 11-34). Springer.Kim, Y.-S. G. (2025). The science of reading-writing connections. In C. MacArthur, S. Graham, & J. Fitzgerald (Eds.), The Handbook of Writing Research, 3rd Edition (pp. 109-124). Guilford Press.

The linguistic impacts of generative AI on L2 writing output

Presenting author
Meier, Ramona , Ghent University, Belgium
Keywords
acquisition;EFL and ESL writing;Generative AI;writing and linguistics
Domain
Learning and Instructional Technology

Abstract

In recent years, research on generative AI (GenAI) and its use for language learning has proliferated, highlighting affordances of the tools, while remaining conscious of potential limitations (Warschauer et al., 2023). Previous work on the use of GenAI tools for L2 English writing has explored the roles ChatGPT can fulfil by employing mainly (quasi-)experimental designs where AI training was provided (e.g. Fang & Han, 2025). However, there is a lack of work focusing on preexisting GenAI usage patterns in EFL students and their effect on L2 writing outcomes. While previous studies focus on the role of GenAI and its potentials, the impacts of such tools on linguistic factors, specifically in synthesis writing, remain underexplored (Yoo, 2025). This study aims to broaden our understanding of students’ preexisting GenAI practices and their impacts on synthesis writing. Participants in this cross-sectional study will complete a synthesis writing task twice (with and without GenAI). Screen recordings, semi-structured interviews, and measures of complexity, accuracy, and fluency (CAF) will be used to analyze their practices, engagement, and language. We expect to find improved performance on the GenAI-assisted task, potentially dependent on the methodical use of GenAI throughout the process, leading to more complex, accurate, and fluent texts. Theoretical and pedagogical implications of the study will also be discussed during the presentation. Keywords: GenAI, EFL learning, L2 writing development, CAF References Fang, S., & Han, Z. H. (2025). On the nascency of ChatGPT in foreign language teaching and learning. Annual Review of Applied Linguistics, 45, 253-273. Warschauer, M., Tseng, W., Yim, S., Webster, T., Jacob. S, Du, Q., Tate, T. (2023). The affordances and contradictions of AI-generated text for writers of English as a second or foreign language. Journal of Second Language Writing 62, Article 101071. Yoo J. (2025). Reading-Writing Connections: A Systematic Review of Second Language Synthesis Writing. L2 Journal: An Open Access refereed Journal for World Language Educators, 17(1), 1-55.