Last week, I wrote about the ambiguity of research on written corrective feedback (WCF), with some scholars arguing it works while others finding little or even negative effects of it. Ferris (2004) identified a number or reasons, mostly owing to a lack of vigorous experimental research that compares a treatment (corrective feedback) group to a control (no feedback group). Kang and Han (2015) directly address this issue in their meta-analysis of 22 studies, looking at whether WCF is effective in general, what type, and what factors influence the effects.
Kang and Han, in recognizing both the lack of agreement on and the continued interest in WCF, recognize the lack of consensus on WCF, especially Truscott’s claims that it has no role in language instruction. They also recognize a lack of meta-analyses in this area, citing only one directly related, conducted by Truscott (2007) with 12 published studies (6 experimental, 6 not), and finding a negative effect size for the experimental studies. The authors have directly responded to the calls for more research by focusing on a greater number of published studies in which comparisons between a treatment (WCF provided) and control (no WCF) can be made. They included 22 such studies in their meta-analysis, attempting to answer the questions below.
RQ1. Is written corrective feedback generically effective for improving L2 written accuracy?
From their analysis, they found that a there is a small to moderate effect of WCF on L2 written accuracy, which is quite different from Truscott’s negative effect. This is likely due to the inclusion of more studies and the inclusion of unpublished PhD dissertations. It also matches a previous study that found a similar effect for WCF for both L1 and L2 settings.
RQ2. If so, which type of written corrective feedback is more effective?
The authors looked at several types of written corrective feedback: indirect (e.g. a highlight with a proofreading code) or direct (highlight + correction), and focused (only certain errors) or unfocused (any error). The current analysis found no statistical differences here. This may be due to type interacting with a variable such as proficiency, with some research suggesting lower proficiency learners benefit from direct feedback. Other research has also suggested that indirect correction might have longer-term effects. No research has looked at this relationship. The raw data shows direct WCF has a slightly higher effect, but the difference was not significant. Likewise, focused correction had a much higher effect than unfocused, but there was no significant difference.
RQ3. What factors mitigate the efficacy of written corrective feedback?
A number of factors were also looked at, including proficiency, setting, length of treatment, and genre of writing task. Proficiency was the strongest variable with significant results. The greater the proficiency, the greater the effects. The effect was actually negative for beginners. According to the authors, this “reinforces the notion that developmental readiness should be considered when providing feedback (p. 11). However, only one “beginner” study was included, likely skewing the results. Next, students in an ESL context may fare better with WCF, possibly because ESL classes have more of a focus on writing process and revision [to me, a bold generalization based on some pretty old research from the 90s]. They interestingly found that one-off WCF treatments produced large effects than longer treatments. This may be moderated by type, intensity, language features, etc. Although interesting, none of the results here were statistically significant. Finally, returning to statistically significant results, the authors found that genre type had an influence on effectiveness. They studies included compositions, letter writing, and journal tasks. Effects were higher for compositions, with letters and journals have less of an effect, likely due to their personal nature.
The meta-analysis serves as more evidence that WCF is a worthwhile instructional technique if a certain number of variables are met. First, WCF should be used with higher proficiency learners (intermediate and advanced). However, if it is to be used with lower-proficiency students, direct correction may be better than indirect correction. Research also suggests that focused correction may be better, but that was not indicated by this meta-analysis. Second, genre should be considered, with composition-like genres being more conducive to WCF, especially if it is included as an integral part of the overall writing process.
Ferris, D. R. (2004). The “grammar correction” debate in L2 writing: Where are we, and where do we go from here? (and what do we do in the meantime?). Journal of Second Language Writing, 13, 49–62. Retrieved from http://onlinelibrary.wiley.com/doi/10.1111/modl.12189/full.
Kang, E., & Han, Z. (2015). The efficacy of written corrective feedback in improving L2 written accuracy: A meta‐analysis. The Modern Language Journal, 99(1), 1-18.
Truscott, J. (2007). The effect of error correction on learners’ ability to write accurately. Journal of Second Language Writing, 16, 255–272.