Introduction
This paper evaluates and synthesizes research on the question of how error correction affects learners’ ability to write accurately, combining qualitative analysis of the relevant studies with quantitative meta- analysis of their findings. The conclusions are that, based on existing research: (a) the best estimate is that correction has a small negative effect on learners’ ability to write accurately, and (b) we can be 95% confident that if it has any actual benefits, they are very small. This analysis is followed by discussion of factors that have probably biased the findings in favor of correction groups, the implication being that the conclusions of the meta-analysis probably underestimate the failure of correction.
Error treatment is one of the key issues in second language writing faced by both teachers and researchers. There has been controversy as to whether error feedback helps L2 students to improve the accuracy and overall quality of their writing (Kepner, 1991; Truscott, 1999; Ferris, 1999). Truscott (1996, 1999, 2007) held a strong view against error correction. He argued that all forms of error correction of L2 student writing are not only ineffective but also harmful and should be abandoned. He further emphasized that although most L2 students clearly desire grammar correction, teachers should not give it to them. Ferris (1999) rebutted this claim by arguing that Truscott had overlooked some positive research evidence on the effects of grammar correction. With the existing data (Kepner, 1991; Chandler; Hyland, 2003; Bitchener, 2008), it is still too early to have a conclusive answer to the question of whether error correction is effective in improving the accuracy of L2 writing in the long term for learners of all levels. Therefore, L2 writing teachers simply cannot dismiss students’ strong desire for error feedback.
The overall goal of establishing Meta-analysis the ineffectiveness of correction entails two sub-goals, each requiring quantification of the research findings. The first is (1) to find the best estimate of the overall effect of correction on accuracy. This estimate will be an average based on all the relevant research. Because any estimate is just that, an estimate, the second requirement is. (2) to determine an upper limit on how helpful correction might be.
Confidence that any benefits of correction are no greater than that number, based on existing research. Following standard practice, I will set this confidence level at 95%. The overall goal, then, is a conclusion of this type:
‘‘Based on existing research, the best estimate of the effect of correction on students’ ability to write accurately is X, and we can be 95% certain that any benefits produced by correction are no greater than Y’’.
The primary conclusion, based on the controlled experiments, is (a) the best estimate is that correction has a small harmful effect on students’ ability to write accurately, and (b) we can be 95% confident that if it actually has any benefits, they are very small. This conclusion receives support from the finding that absolute gains made by corrected students are quite limited, even when beneficial extraneous factors are not controlled. And these negative results probably overestimate the success of corrected groups, especially in regard to grammar errors.
These conclusions must be contrasted with those reported by the authors of another meta-analysis of research on the effects of correction, Russell and Spada (2006). After synthesizing a number of studies, they reported that correction is quite effective, stating that their findings were inconsistent with my conclusion that research has found it ineffective (Truscott, 1996, 1999b). But in fact their findings are entirely consistent with that conclusion. They obtained a different answer than I did because they asked a different question.
The main evidence comes from controlled experiments comparing the effects of correcting with those of not correcting. This distinction should not be confused with that between correcting errors and providing no feedback at all. No one, to my knowledge, recommends the latter policy. Provision of comments on content and clarity, for instance, appears to be universally accepted regardless of one’s position on error correction. A question of pedagogical interest, therefore, is how classes that use only these other forms of feedback compare to those that use error correction.
The findings of the controlled experiments, the average effect size was calculated by the three methods described above. All produced negative means, so the best estimate is that correction has (very small) harmful effects on students’ ability to write accurately. The confidence limits are .094, .289, and .162. In other words, one calculation method allows 95% confidence that correction has no better than a small beneficial effect on accuracy, while the other two allow 95% confidence that any beneficial effects are too small to even qualify as small effects. Given these results, the issue appears to be not whether correction is effective, but whether it is merely ineffective or (more likely) mildly harmful.
Some studies have not included control groups, instead looking only at absolute gains made by groups receiving correction. The limits of such evidence are clear: in the absence of a control group, one cannot determine whether observed gains resulted from correction or from other factors. Thus, even if corrected students consistently showed significant improvement in their accuracy, this finding in itself would tell us nothing about the value of correction. The way to draw implications from uncontrolled studies is to quantify the gains they find in a way that does allow comparison with general standards and with gains expected in the absence of correction
Ferris (2003, 2004) pointed out various problems with the studies, problems that are standard in research on language teaching. But no one has suggested that these flaws introduced any systematic bias against correction groups. So they cannot begin to account for the overall pattern of failure for correction. In addition, two factors probably have systematically biased the findings in favor of correction groups, making them look better than they actually are
One biasing factor is the reliance on measurements done as part of the class, in the same setting in which the correction was done and with no break between the correction period and the testing. Students are prone to forgetting over time, especially when they no longer have the same teacher, context, and learning tasks reminding them.
Corrected students tend to shorten and simplify their writing (Kepner, 1991; Semke, 1980, 1984; Sheppard, 1992), apparently to avoid situations in which they might make errors. The observation is that learners who find a construction difficult tend to avoid it, using it only when especially confident that they can get it right, or when they have no choice. For grammar, the possibility of paraphrase means they very often do have a choice (Schachter, 1974).
This observation has clear relevance to correction, the immediate goal of which is to make learners aware of their errors. Learners are often confused about the corrections they receive (e.g. Lee, 2004; see also Truscott, 1996). When they do understand, this does not mean they have gained mastery of the corrected form, especially of how to apply it to other contexts (see Odlin,1994; Truscott, 1996). So in many cases, a natural reaction to a correction is ‘‘I have a problem, but I’m not really clear about it’’. This awareness creates a clear motivation for avoiding the type of construction corrected. The frequent uncertainty about the exact nature of the problem should also produce the broader type of avoidance found in the research, as learners cannot be sure when they are stepping into dangerous territory.
In terms of gains/losses, the no-correction group was slightly better than the correction-only group, despite the latter’s huge advantage in hours of instruction, suggesting that the correction may well have been harmful. The performance of the control group was also somewhat better than that of the correction–conference group on past tense and was not dramatically lower on prepositions. It was only on articles that a clear superiority showed for the combination of correction plus conferences plus an additional 16 hours per week of instruction. And this advantage resulted entirely from a dramatic and unexplained reversal in this group’s performance on the final task. They declined in accuracy from the first to the second and from the second to the third task but then made a huge improvement from the third to the fourth (of four), yielding a stunning effect size of 1.52 for that final portion of the study. A similar pattern appears with this group on past tense and (to a much lesser extent) on prepositions. With such peculiar numbers, one must ask if the findings for the last assignment for this group were somehow contaminated by an additional, unknown variable (cf. the discussion by Robb et al., 1986, of a similar but less dramatic case in their own findings).
Comment
Learning to write is like learning to read. Both follow a sequential process. Writing requires and combines more basic skills than any other subject area. Writing is a method of represent language in visual or tactile form. Complete writing should fulfill all the following criteria:
1. it must have a purpose of communication
2. it must consist of artificial graphic marks on a durable or electronic surface
3. it must use mark that relate conventionally to articulated speech
Most students prefer underlining and description of the errors in their writing because they wanted to know what kind of errors they made.. Teacher’s direct correction for all errors is the best correction. Direct correction is one of the easiest ways to correct errors because the correct forms are provided, but students may not understand why they made those errors and tended to make the same errors when they wrote different sentences. Underlining and description can provide information on that, so the students can figure out the correct forms by themselves. Pure underlining is sometimes confusing because students may make wrong guesses about their errors. In the process of marking errors, students sometimes took a semantic error as a morphological error, or a morphological error as a syntactic error. Compared with L2 writers in secondary school, college students in this study showed more responsibility in their own writing. They knew that they would be independent writers after finishing this writing course and would, therefore, have to rely on themselves to fix their errors in their future writings.
Doing writing is not easily as we think, moreover for the students. They have difficulty in writing skill. They often face some difficulty in writing. So the teachers have to teach correctly. Teacher’s feedback is usually needed by the students to make their writing better. After doing writing, teachers always give correction to the students, so the students will know the mistake and error in their writing. But, many researchers disagree with it.
Two recent studies comparing different strategies on specific types of errors have provided more evidence in support of written corrective feedback. Bitchener et al (2005) compared two types of feedback groups (a combination of direct written feedback and oral conference feedback and direct written feedback only) with the control group (no corrective feedback) on three types of errors (prepositions, the past simple tense, and the definite article). The study found a significant effect of the combination of written and oral feedback in the use of the past simple tense and the definite article in new pieces of writing. However, no effect was found in the use of prepositions. The findings were confirmed by a recent study by Bitchener (2008) who compared three types of direct corrective feedback: a combination of direct feedback, written and oral meta-linguistic explanation; direct feedback and written meta-linguistic explanation; and direct feedback only. It was found that the accuracy of students who received feedback in the immediate post-test outperformed those in the control group who received no corrective feedback in the use of the referential indefinite “a” and referential definite “the”. More importantly, this level of performance was retained 2 months later. Results of the two studies indicate positive effects of written corrective feedback on particular linguistic subjects.
One biasing factor is the reliance on measurements done as part of the class, in the same setting in which the correction was done large and with no break between the correction period and the testing. This sentence gives us a wide view how to measure and when will teacher assessed error correction. The author of research did error correction in one chance. I think it has advantage and disadvantage. If teacher does the error correction and testing it directly without breaking time it makes the students feel so tired. The error correction did not run well. But, if teacher does the error correction and testing it with break time, I think it is better for the students in enrich their writing skills.
Teacher is not only as educator but also as evaluator. Inherent in the profession of teaching is the need to make corrections, but teachers are often unsure as how much to correct, or even how to go about it. For instance, Tarmuji is worried how he is to deal with error correction with his beginning English classes. He wants to correct his students and thereby improve the quality of their language, but is afraid that if he corrects the students too much, they will become discouraged and stop taking risks in the language. So, to comment on the journal and giving some tips about error correction, below I will add some additional idea to add the theory in error correction.
I start with “what does the process of error correction consist of?” Error correction sequences consist of four steps:
1. A student error
2. The teacher’s feedback which may take the form of explicit correction, recast, clarification request, metalinguistic feedback, elicitation or error correction.
3. The students’ response, which may or may not still need repair
4. Reinforcement of a correct response by the teacher (on occasion)
Students respond more successfully when the correct form is not supplied for them and there is negotiation of form. i.e. with clarification requests, metalinguistic feedback, elicitation or error correction.
Secondly, “what is the difference between intake and uptake?” Intake occurs during the process of giving it and getting it, where students are taking in new information and processing it. Uptake occurs during the using it stage and describes the process of students retrieving information that is already part of their consciousness. It describes the learner’s responses to the teacher’s feedback following either an erroneous utterance or a query about a linguistic item. Some researchers argue that uptake may contribute to second language acquisition by facilitating noticing and pushing learners to produce more accurate linguistic forms.
Third, “what is the difference between a recast and a repair?” A recast is what the teacher says purpose of helping a student notice his or her mistakes and repair it on his or her own. Several different types of recasts are listed below. A repair is the student’s correction after the recast.
Fourth, “what are the types of recasts a teacher can use?” Explicit recast – this is clear and very direct on what has to be corrected. It helps the student notice one thing in particular which need to be corrected. Implicit recast – this is more subtle and often employs gestures. It suggests to the students that there is something to correct rather than directing the student to error immediately. Short recast – this type of recast refers only to the length of the recast, so that only a short cue is given. Long recast – this is a longer length of recast, like a longer cue or perhaps an explanation.
Other recasts that can be used are pronunciation, grammar, vocabulary, substitution request, addition request, declarative, interrogative, one repair needed, multiple repairs needed, clarification requests, metalinguistics feedback, elicitation, and error repetition. Retrieved on www.educ.ualberta.ca/staffolenka.bilash/best%20of%20bilash/error20%correction.html August 07, 2015.
These recasts are not always used individually; quite often, a recast can belong to several categories at the same time. For example, a grammar recast can also be a long recast as well as a substitution recast. Or, an elicitation recast can be a vocabulary recast as well as an interrogative recast.
Based on the explanation before, there are many ways to do error correction. So, teacher does not be afraid to make error correction in order that the student’s language ability can increase especially in writing skill.