How useful are corpus tools for error correction? Insights from learner data

Abstract

This study contributes to the growing literature on data-driven learning in ESL/EFL instruction, examining how advanced English learners choose to use corpus tools for error correction on their own following classroom training. Controlled L2 production and survey data from U.S.-based EAP students were collected over the course of four consecutive semesters and then analyzed quantitatively and qualitatively. The research examined how learners approach the correction of different type of errors using corpus tools and which types of errors could be more successfully corrected with these tools. The results of a Chi Square Test indicated that participants’ success at error correction depended on the type of error being addressed (p < .001). While local lexicogrammatical errors were chosen by learners to be corrected in over half of all participant revisions, these corrections were inappropriate more often than corrections of register errors. Survey data and qualitative analysis of errors further revealed that participants often struggle with identifying the context of the concordance lines and sorting through multiple lines to identify patterns. The paper concludes with a discussion of implications and pedagogical recommendations.