Corpus Use in Language Learning: A Meta‐Analysis

Abstract

This study applied systematic meta‐analytic procedures to summarize findings from experimental and quasi‐experimental investigations into the effectiveness of using the tools and techniques of corpus linguistics for second language learning or use, here referred to as data‐driven learning (DDL). Analysis of 64 separate studies representing 88 unique samples reporting sufficient data indicated that DDL approaches result in large overall effects for both control/experimental group comparisons (d = 0.95) and for pre/posttest designs (d = 1.50). Further investigation of moderator variables revealed that small effect sizes were generally tied to small sample sizes. Research has barely begun in some key areas, and durability/transfer of learning through delayed posttesting remains an area in need of further investigation. Although DDL research demonstrably improved over the period investigated, further changes in practice and reporting are recommended. Open Practices This article has been awarded Open Materials and Open Data badges. All materials and data are publicly accessible via the Open Science Framework at https://osf.io/jkktw. Learn more about the Open Practices badges from the Center for Open Science: https://osf.io/tvyxz/wiki.

Applying the Bundle–Move Connection Approach to the Development of an Online Writing Support Tool for Research Articles

“…Data‐driven learning refers to direct applications of corpora in which learners themselves acquire hands‐on experience of using a corpus for learning purposes, often with guided tasks or materials (see Mizumoto & Chujo, , for different types of data‐driven learning). Its overall effectiveness (i.e., resulting in positive outcomes) for language learning has been reported in meta‐analyses (Boulton & Cobb, ; Mizumoto & Chujo, ). In L2 academic writing studies, corpus consultation as a reference source has been found effective, among other reported benefits (Yoon, ), in correcting writing errors (Gaskell & Cobb, ) and raising awareness about the writing conventions of a specific discourse community (Chang, ; Friginal, ), particularly by using a specialized research article corpus (Lee & Swales, ).…”

A Meta‐Analysis and Meta‐Regression of Incidental Second Language Word Learning from Spoken Input

“…Nowadays, better software is available. Boulton and Cobb (, p. 382) argued that they did not use meta‐regression because that would “mainly [be] suited to continuous [predictor variables].” However, categorical predictors can easily be included in a regression model through dummy coding or other forms of contrast coding. After all, an analysis of variance model is also a regression model with categorical predictors (Field, ).…”

“…The only difference is that, as in metaanalysis, the dependent variable is not the measurement originally used in the Goldschneider &DeKeyser, 2001, andLi, 2010). Instead, researchers often study predictor variables by splitting their data set by the levels of their predictor(s) and calculating separate effect sizes for all these subsets (e.g., Boulton & Cobb, 2017;Mackey & Goo, 2007;Montero Perez et al, 2013). Significance can be determined by considering whether the confidence intervals of the effect sizes of the subsets overlap (Mackey & Goo, 2007) or through Q tests (Montero Perez et al, 2013).…”

Evaluating lexical coverage in Simple English Wikipedia articles: a corpus-driven study

“…For our analysis, we used the corpus profiling program AntConc to make word lists based on frequency, and to measure coverage for OBEL and Baumann and Culligan's (1995) version of West's (1953, see www.lextutor.ca/freq/lists_ download) General Service List (GSL). To estimate coverage, we created word lists that exclude OBEL and GSL word families from the SEW texts.…”