The purpose of this study was to investigate the relative effects of two types of written corrective feedback (WCF), i.e., direct corrective feedback (DCF) and metalinguistic corrective feedback (MCF), on selfcorrection of the same text (revision) and on correct use in new pieces of writing as to conditionals. The students with two different levels of proficiency were randomly assigned to three groups: DCF, MCF, and Control Groups, and received different feedback after performing a Japanese-English translation task. In the experiment, the ratio of correctly revised errors to total errors in the same text and the test scores in new writing tasks were statistically analyzed. The main results were: (a) in the higher proficiency group, both DCF and MCF had a positive effect on revision, but a clear effect of them was not found in new pieces of writing, (b) in the lower proficiency group, DCF immediately had a positive effect on revision, but no effect on improvement of correct use of the target structure in new pieces of writing, and (c) MCF gradually and positively influenced the students’ revision, leading to almost the same ratio of self-correction as DCF, and to improvement in new pieces of writing accordingly. Implications for pedagogy mainly relating to L2 writing are: (i) for higher proficiency students we have only to give an opportunity to write and revise, giving WCF of any kind; (ii) for lower proficiency students we should continuously conduct an activity of writing and revising, offering metalinguistic information (MCF), which imposes a great cognitive load on reflection.
View full abstract