Friday, 27 May 2016

Try this it works! Error correction for speaking

I first met Chris Smith at IATEFL 2014. I was drawn, like a moth to a flame, to his talk entitled "error correction for speaking: An evidence based approach" (write up here) How could I resist? I didn't agree with all of Chris' conclusions but I did enjoy his talk and when I ran into him in the pub the other day I invited him to write a guest blog post. Here it is!

Error correction for speaking: an evidence-based approach - See more at:
Error correction for speaking: an evidence-based approach - See more at:
Error correction for speaking: an evidence-based approach - See more at:
Error correction for speaking: an evidence-based approach - See more at:

(Chris is an EAP tutor in the ELTC at the University of Sheffield You can follow him here.)

There are lots of ever-present arguments and controversies in EFL, but few are as persistent as whether error correction for speaking actually has any effect. One of the principles of Communicative Language Teaching (CLT) is that students should be allowed to speak, communicate and develop fluency. Concurrent with that have been ideas to reduce error correction (EC) or even avoid it all together. Are these good ideas? What evidence is there about EC?

A short history of EC in EFL

(Although I wasn’t there, so feel free to tell me it wasn’t that way!)

Tracing a rough history of error correction in ELT, back in 1960s, a behaviourist influenced Audiolingual approach dominated. This argued that errors should not be tolerated, with correction being immediate and direct (Richards and Rodgers, 1986, p58) because they would propagate bad language behaviour.

The pendulum began to swing the other way with Communicative Language Teaching prevailing in the 1970s until we find the Natural Approach in the 1980s, stating: “Our view is that overt error correction of speech, even in the best of circumstances is likely to have a negative effect on the students’ willingness to try to express themselves” (Krashen and Terrell, 1988, p177). So the argument here is that EC is worse than useless! Firstly it doesn’t work and secondly it will kill any desire to communicate in the student. Krashen’s position creates two separate points although in this post I’ll focus mainly on the first: whether EC is actually effective in terms of acquisition or learning.
Krashen’s ideas were very influential in EFL literature. Harmer (1991, p49) warns against intervention during communicative activities. Ur (1996, p247) recommends correcting for accurate production but not for fluency exercises. Edge (1989) argues that EC should only be given on recently taught items and that learners need uninterrupted communication. Hedge (2000, p290) reports that trainee teachers are often advised to avoid correcting insensitively and causing anxiety or embarrassment.  

Types of corrective feedback

A previous post, from Leo Selivan, talked about the way applied linguistics does not use the same terminology as teachers, and this is true in error correction literature too. This is perhaps understandable since EFL literature is aimed at training teachers, while applied linguistics research is required to be peer-reviewed. Nevertheless it makes it confusing when different terminology is used to describe the same topics.

Numerous taxonomies of error correction techniques can be found (e.g. Harmer, 1991; Lyster and Ranta, 1997; Ur, 1996), all describing the same things, but often doing so in different ways. For example, Ur uses “explanation”, while Lyster and Ranta use “metalinguistic feedback” and Harmer uses “echoing” whereas Lyster and Ranta use “recast” (albeit with a slightly different definition). Most journal articles seem to follow Lyster and Ranta’s terms now, but this may be difficult terminology to grasp for teachers trained with teaching manuals.

In broad terms, we can divide EC into 3 groups: implicit correction, which can involve repeating in correct English (recasts) and negotiating meaning, but where the discourse is not stopped to highlight an error; explicit correction, which can use a variety of techniques, but crucially, where the teacher ensures the error and correction are noticed; and delayed correction, where the teacher allows conversation to continue but then later picks up on errors made, perhaps writing several on the board and eliciting corrections and explanations.

Evidence on the effectiveness of error correction

There have been a number of observational or experimental studies in which two or more groups of students are given instruction, with one control group receiving no EC, while the other group(s) receive(s) (different types of) EC. Here are a few, which are all describing spoken EC studies.

Lightbown and Spada (1990) analysed 4 different classes of 10-12 year olds over a 5 month period. They did not intervene in the teacher’s styles, but by observing and noticing the differences between teachers, they concluded that fluency, accuracy and communication could be developed best by a teaching approach that includes EC.

Carroll, Roberge and Swain (1992) compared adult learners at 2 different levels, one group getting EC and the other not, for instruction on particular vocabulary and grammar points. They found positive results, stating “correction clearly had an effect on learning in all the conditions tested” (p.185).
Carroll and Swain (1993) examined the effects of 4 different types of EC in separate groups against a control group of no EC. All 4 groups significantly outperformed the control group when tested on the target grammar point, with the group receiving explicit correction with an explanation performing the best.

Lyster and Ranta (1997) asked whether all types of feedback are equally effective. Their experiment yielded data which suggested explicit EC was more effective than implicit EC. This suggests it is important for teachers to make sure students realise that a correction has been provided.

Loewen (2005) analysed 17 hours of classroom interaction, counted 491 instances of explicit correction of non-target language and devised individualised tests to check recall of this. He found positive results, concluding “incidental focus on form does have some effect on L2 learning” (p381). This contradicts what was recommended in some earlier teaching manuals, which recommended only correcting target language, and not correcting at all during fluency activities. Loewen’s evidence suggests correcting at any time can produce improvements in language development.

Ellis, Loewen and Erlam (2006) found that explicit feedback was more effective than implicit feedback and that the benefits became more evident over time, suggesting explicit feedback aids long term acquisition, so this supports Lyster and Ranta (1997).

Some studies have been less conclusive than those described; however, the general trend is for research to demonstrate that EC has a positive effect on language development. There has been controversy over implicit corrections, including recasts, as learners may not notice them. However, the research shows them to have some beneficial effect (it’s worth remembering that although the speaker may not notice the correction, other learners might). In an overview of recasts,  Long (2007, p76) stated: “There is mounting support from research in both first and second language acquisition for the claim that [negative evidence] does affect competence, facilitating language development when it occurs.”

There is even stronger support for explicit EC where the teacher makes sure the learner notices they have made an error. The studies above found it to be more effective than implicit EC. In a comprehensive overview of studies into corrective feedback, Ellis (2008, p885) states: “There is clear evidence that corrective feedback contributes to learning.”

So, when it comes to evidence based EFL, we can conclude that the evidence shows that error correction works. I would also assert that if people want to argue that it does not work, they cannot merely cherry pick one or two articles that did not find a link. They would need to show why all the clear evidence mentioned above (and more) is wrong. So going back to Krashen and Terrell, they asserted that EC is useless, and this idea has been dogmatically perpetuated. However, this is demonstrably wrong. The evidence shows that EC clearly is effective.

Another aspect of Krashen and Terrell’s argument was that EC will raise an affective filter, discourage communication and prevent learning. This is an idea that keeps being brought up and I presented about this at IATEFL 2015. The recording of that is available here, so I won’t repeat the content of that at length. Suffice to say that when I investigated my students’ (EAP pre-sessional) attitudes to EC, they overwhelmingly said they believed it to be effective, they did not find it embarrassing and they wanted more of it than they were getting, which flies in the face of the affective filter concept.

Implications for classroom practice

EC works, students know this and want more of it, particularly explicit corrections with explanations. If students are making mistakes, they want to know, want to be told why it’s wrong and want the correct form provided. This is how they can improve the accuracy of their speech.

So more class time should be given over to EC, form-focused instruction, feedback on production, working with what students are saying and helping them to say it better. If you plan a stage where the students speak for 5-10 minutes related to a language point, you can include a post speaking EC stage, telling them you are going to correct any mistakes or try and improve their language, by asking a display question to each student.

If you are listening to a student and the focus is meaning, you may not want to stop them in the flow of speech but you can make a note and come back to it later. If your students are involved in a discussion task, let them get on it with it, but make notes, and once the task is finished, do some language work. All of these things need time, so they need to be considered at the planning stage.

And if you are unsure whether your own students would respond as positively to more error correction, ask them. You can include it on a needs analysis form, or as a separate survey, so that you can adopt an evidence-based approach


CARROLL, Susanne and SWAIN, Merrill (1993) Explicit and Implicit Negative Feedback: An Empirical Study of the Learning of Linguistic Generalizations. Studies in Second Language Acquisition 15, 357-386.

CARROLL, Susanne, SWAIN, Merrill and ROBERGE, Yves (1992). The role of feedback in adult second language acquisition: Error correction and morphological generalizations. Applied Psycholinguistics 13, no. 2 173-198.

EDGE, Julian (1989) Mistakes and Correction. London, Longman.

ELLIS, Rod (2008). The Study of Second Language Acquisition. Oxford, OUP.

ELLIS, Rod, LOEWEN, Shawn and ERLAM, Rosemary (2006). Implicit and Explicit Corrective Feedback  and the Acquisition of L2 Grammar.

HARMER, Jeremy (1991). The Practice of English Language Teaching (New Edition). Harlow, Longman.

HEDGE, T (2000). Teaching and Learning in the Second Language Classroom. Oxford, OUP.

KRASHEN, Stephen D. and TERRELL, Tracy D. (1988) The Natural Approach: Language acquisition in the classroom. Hemel Hempstead, Prentice Hall.

LIGHTBOWN, Patsy M. and SPADA, Nina (1990). Focus on Form and Corrective Feedback in Communicative Language Teaching. Studies in Second Language Acquisition 12, 429-448.

LOEWEN, Shawn (2005). Incidental focus on form and second language learning. Studies in Second Language Acquisition, 27(03), 361-386.

LONG, Michael H. (2007). Problems in SLA. London, Lawrence Erlbaum.

LYSTER, Roy and RANTA, Leila (1997). Corrective Feedback and Learner Uptake: Negotiation of form in communicative classrooms. Studies in Second Language Acquisition 20, 37-66.

RICHARDS, Jack C. and RODGERS, Theodore S. (1986) Approaches and Methods in Language Teaching. Cambridge, CUP.

UR, Penny (1996). A Course in Language Teaching. Cambridge, CUP.


  1. In other words, yet again, the behaviourists were right and the cognitivists were wrong.

    Vis a vis that, see this too:

  2. I tried to upload an answer to this on my phone - it clearly hasn't worked. I wanted to point out that in recasts - the uptake was far greater in pron and lexis when learners made mistakes, and therefore was more effective. But in grammar errors there is no point in a recast.
    Also I deeply question the issues around error correction and motivation. Some extrovert learners (of which we have all had some) really do benefit from being held back and corrected a little, in their extravagant attempts to use the language. Others, especially those coming from South East Asia, are really inhibited by any form of correction, and already feel over corrected. Therefore I think that a good deal of attention needs to be paid to the learner motivation.
    Personally, I like to make a note of errors and then address them in a new language point one day when the learner has almost forgotten about his/her error. I am trying to facilitate noticing. I didn't know about noticing until I started my masters - but I have been a language learner myself for millions of years and recognised the concept immediately.

    1. Hi Genevieve. There is the danger of over-correction leading to a loss of motivation. I don't disagree.

      I think the reason my students were so overwhelmingly pro-correction was that their typical profile was a pre-masters university student, adult, upper-intermediate, experienced in learning and therefore probably confident about language learning and not afraid of a bit correction.

      It would be unwise to assume those results could be generalised to all learners, although I think it is best to try and find out.

  3. Really enjoyed this post. Good to see where we stand when it comes to research evidence on this.
    As you point out above, we definitely don't want to overdo the correction. I used to really hate those Ts who corrected you so much that you could never actually say anything in class. But in general, I'm glad that the evidence seems to conform that EC works. I've always thought that sts are much more robust than what Krashen would have us believe.
    Is there any research showing whether delayed or on the spot explicit feedback is more effective?

  4. Hi,

    Really interesting post - thanks for sharing. Nice to have an overview of research. You touched on it in the post but I wonder was more found about the effectiveness of explicit correction of different errors (i.e. errors in pronunciation, grammar, vocab). While I agree that there may be a bit too much made of the "loss of face" argument when talking about explicit correction, I have noticed that learners shift a little more uncomfortably when a pronunciation error is pointed out as opposed to a vocabulary/grammar one. Obviously, the way the error is dealt with and the sensitivity of the teacher is crucial, but I wonder if there has been much research specifically on explicit error correction of pronunciation.


  5. Interesting article and blog that's new to me.

    I've never liked the term error correction and prefer to refer to it and think about it using the word feedback. The reason I don't the term is that it only describes what the teacher is doing or has done and says nothing about student engagement. Errors were corrected. Well, that may be true but how about error correction retention? Teachers may wonder why Student A repeats mistake B over and over again despite the error having been corrected. Something wrong in the telling, perhaps?

    For me, it's feedback rather than error correction all the way. It's the same thing of course but a different way to conceptualise it. Feedback can mean and involve dialogue, examples, analogies, humour, comparisons and so on as part of a process of engaging learners in the process of noticing and understanding why and how things went wrong in their language production. For me, error correction is about telling, whereas feedback is about dialogue.

    When we say error correction to me it's the wrong way to describe what the process of sharing post-production information with learners should look like. I know it might sound only like a question of terminology but I do think the term error correction has a very old-fashioned, simplistic ring to it that needs to be ditched in favour of feedback.

    Anyway, this blog challenges me to think in terms of evidence-based notions and that's always a good idea.