Enhancing extensive reading with data-driven learning

Abstract

This paper investigates using data-driven learning (DDL) as a means of stimulating greater lexicogrammatical knowledge and reading speed among lower proficiency learners in an extensive reading program. For 16 weekly 90-minute sessions, an experimental group (12 students) used DDL materials created from a corpus developed from the Oxford Bookworms Graded Readers, while a control group (10 students) had no DDL input. Both classes were required to read a minimum of 200,000 words during the course. An embedded-experiment design (Edmonds & Kennedy, 2017) was adopted consisting of both qualitative and quantitative forms of investigation. Quantitative data from the Vocabulary Levels Test by Nation and Beglar (2007) and a C-test (Klein-Braley & Raatz, 1984) constructed from an upper-level Bookworms reader found statistically significant lexicogrammatical improvements for both groups, but greater improvement took place within the control group. Qualitative data derived from a repertory grid analysis of student constructs revealed several possible reasons for the experimental group’s lack of engagement with DDL. The study concludes that careful attention to students’ learning preferences and a softening of the DDL approach may ensure better results with lower proficiency learners.