Direct and indirect data-driven learning

An experimental study of that-complementation

Bibliographic Data

ID21636256
AuthorsKar Kee Ann Yeung (0009-0004-1264-5663, Yulin University), G Hu (0000-0002-2297-4784, Hong Kong Polytechnic University, corresponding author)
Year2026
Volume72
Pages101301
Publication date2026-06-01
Open AccessYes
TypeARTICLE
VenueJournal of Second Language Writing (JOURNAL)
Journal identifiersISSN: 1060-3743 • E-ISSN: 1873-1422
PublisherElsevier BV (PUBLISHER)
DOI10.1016/j.jslw.2026.101301
OpenAlexW7148702175
LanguageEN
References cited41

While previous studies have demonstrated the pedagogical utility of data-driven learning (DDL), little research has compared direct (computer-based) and indirect (paper-based) DDL in English-as-a-second-language writing instruction. To address this gap, this study examined the effectiveness of the two DDL approaches in improving first-year college students’ knowledge and use of that -clauses in an English-for-academic-purposes course. Using a pre-post-delayed quasi-experimental design, two experimental groups received either direct or indirect DDL interventions, while a control group received traditional teacher-fronted instruction. Both DDL interventions led to short-term gains in the frequency of that -clause use. While the indirect DDL intervention was effective in sustaining such gains beyond the post-test, the gains for the direct DDL group disappeared on the delayed test. In terms of variety of use, only the direct DDL group showed improvement from the pre-test to the post-test and maintained this improvement on the delayed test. The direct DDL intervention was also somewhat more effective than the indirect DDL intervention in improving participants’ variety scores on the post-test. Finally, neither the direct nor the indirect DDL intervention had any significant effect on the accuracy with which that -clauses were used. Implications for second language writing instruction and future DDL research are discussed

Child and Animal Learning Development · Domain Adaptation and Few-Shot Learning · Machine Learning and Algorithms

Tools

Open DOIOpen Access