What types of feedback are most effective?

Bitchener, J. (2008). Evidence in support of written corrective feedback. Journal of                       Second Language Writing. 17, 102-118.

In “Evidence in support of written corrective feedback,” Bitchener does a good job of synthesizing previous research and schools of thought on written corrective feedback. The main focus of his research is to investigate whether written corrective feedback helps improve the accuracy of L2 writing in new pieces of writing. Bitchener explains that it needs to be new pieces, not corrections of old pieces, to truly test whether anything is getting transferred to long-term knowledge.

He has suggestions for study design, chiefly that there needs to be a control group, that a pre- and post-test are necessary and that the study needs to narrow its focus to one error category (or at least a very small number). Bitchener acknowledges the possible ethical issues of a control group that have prevented better designed studies. I was glad of this, as it was my first thought upon reading that there was a control group. It is not right to deny a student population of instruction that could possibly benefit them. Bitchener suggests several ways of getting around the ethical issues of denial, such as not instructing or correcting a group on a particular construction during a semester and reversing it the next (given that the students are taught for a long enough amount of time).

Bitchener also wants to differentiate between forms of corrective feedback. His three groups, besides the control who receive no feedback, are direct corrective feedback, direct corrective feedback with written meta-linguistic explanation, and direct corrective feedback with written and oral meta-linguistic explanation. The direct corrective feedback was written above each targeted error, while the meta-linguistic feedback was in the form of a written explanation at the end of the writing sample or orally in the form of a mini-group lesson or one-on-one instruction.

In discussing his results, Bitchener’s study seems to show that all three forms of feedback in conjunction are the most effective, but that direct corrective feedback on its own was more successful than direct corrective feedback with just written meta-linguistic explanation. He points out that there could be various amounts of previous instruction in the targeted function between groups prior to the study as indicated by the pre-test. However, the statistical difference was minimal. It’s also possible that written meta-linguistic explanations are too confusing without any other kind of instruction. We know that students have to recognize and come to understand an error before they can be expected to change it.

I found this study to be well-designed and liked Bitchener’s suggestions for limiting the scope of research so that you can actually process the results. If there is too much information to focus on, it is difficult to determine which factor affects the others.