What does research say about error correction?

Imagine you are observing a capable teacher, teaching a group of students. At one point, after a discussion task, the teacher initiates open class feedback and elicits a few answers. One student offers up a suggestion, which is almost completely incomprehensible, not because of pronunciation but because of a large number of language problems. The tutor smiles and nods, and says ‘Yes, good, anyone else?’ Now imagine you are giving post observation feedback; would you bring it up?

This situation happened to me many years ago and I did bring it up and suggested that if communication was essentially not happening because of errors, then it might be prudent to correct those errors. The tutor in question disagreed. They believed that there really wasn’t much point in correcting errors because students learn in their own time anyway and error correction is not really very effective. Prior to this encounter, I had taken it as an article of faith that error correction was effective, but what evidence was there to back that assumption up? I decided to find out.

Anyone who starts looking into the research on error correction will quickly realise that it is not in short supply. In fact, it is probably one of the most researched areas in language teaching research. This perhaps shouldn’t come as a surprise because error correction is a direct and contained form of language teaching, which makes it a good model for studying. A student says something incorrect, the teacher corrects it, and then what? Do they continue to make the mistake or do they start using the form correctly? Teachers presumably hope for the former situation as it would indicate we are having some kind of effect. After many decades of research, surely there would be a clear answer.

However, things are rarely that simple and on top of finding that there were huge numbers of papers on error correction, I also discovered that no one seemed to agree very much on very much of anything. For example, by the time I started looking into the effectiveness of written error correction, there was already a long debate between researchers. On one side Truscott claimed that the evidence that error correction was useful was weak. He even suggested correction may even be harmful as it wastes times that could be spent on more useful things (Ferris, 2011: 7). Ferris, however, sought to argue otherwise. Their debate went on for over 15 years.

“Research to date seems to suggest that direct correction is more effective for written feedback.”

Even those who do believe it is effective, seemingly couldn’t agree on what type of correction was effective (errors codes anyone?). Similarly, in the area of oral error correction debate continues over whether it is more effective to elicit corrections from students or just directly tell them the correct version.

However, more recently, trends have started to emerge and it does seem that there are things we can say with relative certainty. For example, it will come as relief to the vast majority of teachers to learn that research into error correction overwhelmingly shows that error correction, in any form, is more effective than no correction (Bitchener & Ferris, 2012; Plonsky and Brown, 2015). The ‘harm’ argument also seems to be overstated. No one has, to date, produced evidence that correcting students harms them or leads to their English getting worse. If anything, there are studies that show that students claim to want more correction than their teachers currently give them (Weaver, 2006).

There are also other fairly solid findings that might surprise teachers. For example, many tutors use correction codes and favor a ‘let students think about the right answer’ approach to written error correction, assuming it is more effective than telling them the answer. This is a reasonable assumption to make since more effort could lead to better learning. However, there is always the possibility that students will not be able to arrive at the right answer by themselves or may arrive at the wrong answer. Research to date seems to suggest that direct correction is more effective for written feedback (Bitchener & Ferris, 2012; Shintani & Ellis, 2015).

In terms of oral correction, it is less clear which method is best. It is certainly the case that there is evidence (Li 2010, in Shintani & Ellis 2015) that explicit forms of correction (for example, asking students to correct their sentences, or questioning ‘he go?’) is more effective than more subtle forms of feedback such as recasts (that is, just repeating the sentence with the error corrected). Though the case is not as clear as with written corrective feedback. What does seem clear is that studies either find an advantage for explicit feedback, or they find no difference. From this, it could be argued that the safest bet is explicit feedback. Though as Long notes (2007)’ recasts can be a fairly unobtrusive way to give feedback.

The argument over recasts versus other forms of oral feedback is important because recasts seem to be the default method of oral correction for teachers. A study by Kartchava (2020) indicates that teachers’ beliefs about oral correction and their actions may be quite different. When interviewed, teachers claimed that recasts, prompts and other forms of explicit correction were all effective strategies. Yet their lessons showed an almost exclusive reliance on recasts (86%) and none of their feedback was explicit.

A final piece of research that surprised me was in relation to the idea that error correction should not be done during fluency activities. I had heard this mantra for most of my teaching career and never questioned its wisdom. It sounds a little bit like the idea that you shouldn’t swim until an hour after eating – plausible, but somewhat dubious on closer inspection. There is, in fact, no evidence to support the idea that correcting students during a fluency activity will have some kind of detrimental effect. You may still choose not to do it: after all, you may not want to interrupt students mid flow. But if you do want to students to improve, giving some error correction, even if it is delayed, will likely lead to positive outcomes.

In general, don’t worry too much about upsetting students or giving feedback at the wrong time, but do perhaps choose your approach with an eye on the research literature.

References

Bitchener J & Ferris DR (2012) Written Corrective Feedback in Second Language Acquisition and Writing. New York: Routledge.

Ellis R & Shintani N (2014) Exploring Language Pedagogy Through Second Language Acquisition Research. London: Routledge.

Ferris DR (2011) Treatment of Error in Second Language Student Writing. Ann Arbor: University of Michigan Press.

Kartchava E (2020) Pre-service English as a second language teachers need training on corrective feedback to reconcile their beliefs and practices. OASIS Summary of Kartchava E, Gatbonton E, Ammar A & Trofimovich P (2020) in Language Teaching Research. Available at:https://oasis-database.org/

Long MH (2007) Problems in SLA. Second Language Acquisition Research Series. Mahwah, NJ: Lawrence Erlbaum Associates.

Plonsky L & Brown D (2015) Domain definition and search techniques in meta-analyses of L2 research (Or why 18 meta-analyses of feedback have different results). Second Language Research 31 (2) 267–278.

Shintani N & Ellis R (2015) Does language analytical ability mediate the effect of written feedback on grammatical accuracy in second language writing? System 49 110–119.

Truscott J (1996) The case against grammar correction in L2 writing classes. Language Learning 46 327-369

Weaver MR (2006) Do students value feedback? Student perceptions of tutors’ written responses. Assessment & Evaluation in Higher Education 31 (3) 379–394.