Effectiveness of coded corrective feedback in the development of linguistic accuracy in L2 writing: Impact of error types and learner attitudes

Corrective feedback (CF) has been a contentious topic among scholars who are interested in second language acquisition (SLA) and L2 writing. Since early 1990s, oral CF researchers have shifted focus from comparing the relative effectiveness of various types of CF strategies to examining the role played by various moderating factors such as learners' individual differences (IDs) and target linguistic features (e.g., Dekeyser, 1993; Li, 2010; Sheen, 2011). In contrast, to date the bulk of written CF research still concentrates on examining the two fundamental questions (i.e., whether CF facilitates L2 written accuracy and, if so, which type of CF strategy is more effective) without paying due attention to learner-, context-, and linguistic target-related factors (Ferris, 2004; Liu & Brown, 2015).;Driven by this gap, the current study investigated the efficacy of error coding in relation to error types and learner attitudes. Three research questions were addressed: (a) how effective is error coding in promoting L2 written accuracy in editing and in new writings, (b) to what extent is the effectiveness of error coding moderated by error type, and (c) to what extent is the effectiveness of error coding associated with L2 learners' attitudes. Forty-one intermediate ESL learners in an intensive English program in the U.S. participated in the experiment. They wrote three persuasive essays with two drafts for each. Sixteen types of linguistic errors were identified and marked in the first drafts, and the participants were required to edit the coded errors. Editing outcomes were categorized as "successful correction", "unsuccessful correction", or "ignored in correction". Overall accuracy was measured through error rate (total number of errors per 100 words). Upon the completion of the three writing tasks, a forty-item Written Corrective Feedback (WCF) Attitude Questionnaire was administered to elicit participants' attitudes towards English academic writing, linguistic accuracy, and coded CF.;Results from data analyses revealed that: (a) overall, intermediate ESL learners were able to successfully correct 64% of all coded errors and attempted to correct another 15% but with limited success. The remaining 21% of coded errors were ignored; (b) over a seven-week period of time, there was no statistically significant reduction in error rate although a consistent decrease in error rate was seen across the three essays; (c) in both editing and new writings, error coding appeared to be more effective in reducing word- and phrase- than clause-based errors; and (d) no statistically significant correlation was found between intermediate ESL learners' overall attitudes and their development of accuracy in editing and in new writing tasks. However, a closer examination of the individual items in the WCF Attitude Questionnaire revealed a significant correlation between learners' perceptions of the usefulness of the error codes and their improvement in accuracy in new writings. A number of theoretical and pedagogical implications for CF research and L2 writing instruction are discussed.