Learning Sequences and Events in Individual, Pair and Group Data-Driven Learning
TL;DR
Investigation of how different DDL configurations shape both learning outcomes and processes in academic reading comprehension indicates that DDL led to overall gains in academic reading comprehension, with the magnitude of these gains varying by configuration.
Abstract
Abstract Previous research has shown that the learning process in data-driven learning (DDL) is cyclical, iterative, and non-linear, even when tasks are designed around seemingly linear pedagogical frameworks such as Illustration–Interaction–Intervention–Induction (Carter, Ronald, and Michael McCarthy. 1995. “Grammar and the Spoken Language.” Applied Linguistics 16 (2): 141–58; Flowerdew, Lynne. 2009. “Applying Corpus Linguistics to Pedagogy: a Critical Evaluation.” International Journal of Corpus Linguistics 14 (3): 393–417). Building on this work, the present study investigates how different DDL configurations shape both learning outcomes and processes in academic reading comprehension. Drawing on constructivism, sociocultural theory, and the noticing hypothesis, the study adopts an established analytical framework distinguishing between seven macro-level learning sequences and twelve micro-level learning events in DDL. French-speaking Master’s students enrolled in a linguistics programme participated in a six-week hands-on DDL intervention targeting move analysis of English research article introductions (Swales, John M. [1981] {2011}. Aspects of Article Introductions. Michigan: Michigan Classics Edition). Learners worked individually, in pairs, or in groups while consulting the British Academic Written English corpus via Sketch Engine. Learning outcomes were evaluated using pre- and post-tests analysed with Bayesian linear mixed-effects regression, while learning processes were examined through video-based observation and annotation in ELAN. Results indicate that DDL led to overall gains in academic reading comprehension, with the magnitude of these gains varying by configuration: Pair DDL was most likely to support improvement, whereas Group showed weaker and more variable outcomes. Behavioural analyses reveal Pair offered a balance between scaffolding and efficiency, whereas Individual relied heavily on self-regulation and Group incurred substantial coordination cost. These findings suggest that balancing learner autonomy and peer collaboration may be a relevant consideration in DDL design.