Correct I or I Don’t Correct Myself: Corrective Feedback on EFL Students Writing

The hot debate is in progress between giving corrective feedback (CF) to students to promote their writing ability and accuracy or not giving CF. This kind of dilemma that might be caused by the controversial researches that published over the last 20 years baffled teachers about giving or not giving the CF. After Truscott, J. (1996) gave a case against grammar correction in L2 writing classes and Ferris, D.R. (1999) gave a case for grammar correction the question got tougher. The essence of this experimental investigation is to explain giving written CF to the EFL students writing. For the purpose of this study, 160 EFL students of Azad University participated in an extracurricular course; there were three experimental groups and a control group. After the results were manifested, it was cleared that giving corrective feedback should be considered as a productive way in improving EFL students writing.


INTRODUCTION
In which way should teachers correct their students' written productions? This is a fundamental question for second language teachers trying to help their students develop their ability and accuracy in their second language writing. Since 1996 and Truscott first published of the article "the case against grammar correction in L2 writing classes" debate about whether to give L2 students feedback on their written grammatical errors has earned a great interest to researchers (Ferris, 1999(Ferris, , 2002(Ferris, , 2004Truscott, 1996Truscott, , 1999. On several grounds, Truscott (1996) claimed that grammar correction has no effect in writing accuracy promotion and should be abandoned. From an analysis of studies by Kepner (1991), Semke (1984) and Sheppard (1992), Truscott concluded that there is no convincing research evidence that error correction ever helps student writers improve the accuracy of their evidence that error correction ever helps student writers improve the accuracy of their writing. He explained that this finding should not be surprising. On the one hand, he argued that error correction, as it is typically practiced, ignores SLA insights about the gradual and complex process of acquiring the forms and structures of a second language (Bitchener & Young, 2005).
On the other hand, he outlined a range of practical problems related to the ability and willingness of teachers to give and students to receive error correction (Truscott, 1999).

R E T R A C T E D R E T R A C T E D R E T R A C T E D R E T R A C T E D
Moreover, he claimed that error correction has harmful effect because it puts away time and energy from productive aspects of a writing program. Not surprisingly, these claims have since generated considerable amount of vigorous debate at international conferences and in published articles (Ellis, 1998;Ferris, 1999;Ferris & Hedgcock, 1998;Truscott, 1999). On the other part of the bridge, Ferris (1999) claimed that his arguments were premature and overly strong given the rapidly growing research evidence pointing to ways in which effective error correction can and does help at least some student writers. While acknowledging that Truscott had made several compelling points concerning the nature of the SLA process and practical problems with providing corrective feedback, Ferris maintained that the evidence he cited in support of his argument was not always complete (Bitchener & Young, 2005). As Chandler (2003) also points out, Truscott did not always take into account the fact that reported differences need to be supported with statistically significant evidence.
In addition, Ferris maintained that there were equally strong reasons for teachers to continue giving feedback, among them is the belief that students have considered its value. However, Ferris did accept that it is necessary to consider ways of improving the practical issues highlighted by Truscott. Despite his call for the abandonment of error correction, Truscott (1999), in his response to Ferris, acknowledged that many interesting questions remain open and that it would be premature to claim that research has proven error correction can never be beneficial under any circumstances. However, he suggested that researchers and teachers should acknowledge that grammar correction is, in general, a bad idea until future research demonstrates that there are specific cases in which it might not be a totally misguided practice. Being agree with the future research focus proposed by Ferris (1999), he suggested that attention be given to investigating which methods, techniques, or approaches to error correction lead to short-term or long-term improvement and whether students make better progress in monitoring for certain types of errors than others. The following section surveys some of the major findings from studies that have sought to examine these issues.

Review of Literature
Much interest has been expressed in recent years in literature on the effectiveness of giving CF for helping L2 writers to improve the accuracy of their writing (see Bitchener & Knoch, 2010;Sheen, 2010). Besides, there is evidence that written CF can help writers improve their written accuracy when asked to revise their texts (Ferris, 1999(Ferris, , 2006. Moreover, there is large amount of evidence on long-term effectiveness of written CF on accuracy improvement, revealed in the writing of new texts (Bitchener, , 2009 There seems to be some doubt, however, about the extent to which accuracy improvement in text revisions can be seen as a predictor of improved accuracy in new text writing over time (Sachs & Polio, 2007;Truscott & Hsu, 2008). Although there is growing evidence of the relationship between written CF and accuracy improvement over time, the research base has so far been limited to testing its effectiveness with certain linguistic error domains and categories. Thus, the extent to which written CF is effective for different domains and categories of linguistic knowledge has yet to be more fully explored. Truscott (1996) has argued that no single form of correction can be expected to help learners acquire knowledge of all linguistic forms and structures because the acquisition, for example, of syntax, morphology, and lexis requires an understanding not only of form but also of meaning and use in relation to other words and other parts of the language system. Referring to syntactic knowledge, for instance, he argues that written CF cannot be expected to facilitate the acquisition of such knowledge because it comprises more than a collection of discreet items. The researches done by Mackey and Oliver (2002), Mackey and Philp (1998), and Mackey, Philp, Egi, Fujii, and Tatsumi (2002) on syntactic structures like question forms and one study by McDonough (2006) on the use of dative constructions have revealed positive effects when oral CF is provided. If the hypothesized advantages of written CF over some forms of oral CF prove to be true (see Sheen, 2010, for a discussion of differences between the two modalities), it may be that written CF is able to target complex forms and structures (e.g. syntax) as well as, and maybe better than, oral CF. So far, limited research has focused on the role of written CF for helping learners acquire specific linguistic forms and structures. Those that have been published have tended to investigate the effectiveness of written CF for treating discrete, rule-based items. One study (Bitchener, Young, & Cameron, 2005) investigated the effect of written CF on three linguistic error categories (use of the English article system. The past simple tense, and prepositions)and found it to be effective for helping L2 writers improve the inaccuracy in the first two, rule-based, categories, but not in the more idiosyncratic use of prepositions.
That study did not examine which particular functional uses of the article system were most effectively targeted with the feedback (Bitchener, Young, & Cameron, 2005). Given the range of functional use of the article system in English, it is important to know whether written CF is more effective in targeting certain uses rather than all uses. In a series of more focused studies on the extent to which written CF can effectively treat specific functional uses of the article system, Bitchener (2008) 2009) investigated its effect on two particular functional uses of the English article system (the use of the indefinite article for first mention and of the definite article for subsequent or anaphoric mentions) and found that writers who received written CF outperformed those who did not receive CF. Gains were evident in both the immediate and delayed post-tests, thereby demonstrating that written CF can have an immediate effect on the writing of a new text and that the level of improvement can be retained over time. Several researchers (Bitchener et al., 2005;Butler, 2002;Ferris, 2002Ferris, , 2006Master, 1995) have noted that advanced writers continue to experience difficulty in their use of the English article system.
A range of studies has investigated the extent to which different types of written CF may have an effect on helping L2 writers improve the accuracy of their writing. These studies have categorized written CF as either direct or indirect. Direct CF has typically been defined as that which provides some form of explicit correction of linguistic form or structure above or near the linguistic error. It may consist of the crossing out of an unnecessary word/phrase/ morpheme, the insertion of a missing word/phrase/morpheme, and the provision of the correct form or structure. More recently, direct CF has included written meta-linguistic explanation and sometimes oral form-focused instruction. On the other hand, indirect CF has been defined as that which indicates that in some way an error has been made but it does not provide a correction. It is typically provided in one of two ways: (1) underlining or circling an error and (2) recording in the margin the number of errors in a given line. Rather than the teacher providing an explicit correction, writers are left to resolve and correct the problem that has been drawn to their attention (Bitchener et al., 2005). In earlier research (Ferris, 2003), the provision of a code to show the category of error also tended to be included within the indirect category. Theoretical arguments have been advanced for both the direct and indirect approaches. Those supporting indirect feedback suggest that this approach is best because it invites L2 writers to engage in guided learning and problem solving and, as a result, promotes the type of reflection on existing knowledge that is more likely to foster long-term Volume 7 acquisition and written accuracy. Those more in favor of direct feedback suggest that it is more helpful to writers because it (1) reduces the type of confusion that they may experience if they fail to understand or remember the feedback they have been given (for example, the meaning of error codes used by teachers); (2) provides them with information to help them resolve more complex errors (for example, syntactic structure and idiomatic usage). Studies that have investigated the relative merits of different types of feedback have tended to be grouped according to those that have compared (1) direct and indirect types of feedback; (2) different types of indirect feedback; and (3) different types of direct feedback. Considering, first, those that have compared direct and indirect types, Lalande (1982) reported an advantage for indirect feedback; Robb, Ross, and Shortreed (1986) and Semke (1984) reported no difference between the two approaches; and Chandler (2003) reported positive findings for both direct and indirect feedback. Limitations in the design and execution of these studies (see Ferris, 2003;Guenette, 2007 for discussion of these issues) and differences in their contexts and in the proficiency level of their participants make it difficult to assess the value of the claims that are made. It should also be noted that most of these studies did not look at new pieces of writing, so they provide no information about the long-term effectiveness on written accuracy.
Studies by Ferris and Roberts (2001) (4) no corrective feedback. They found that those in group one who received direct error correction and oral meta-linguistic explanation outperformed both groups two and three for the past simple tense and the definite article but found no such effect for prepositions. They suggested that the addition of oral meta-linguistic explanation may have been the crucial factor in facilitating increased accuracy.  investigated the effectiveness of other direct feedback combinations: (1) direct error correction with written meta-linguistic explanation (of the rule and an example of its use) and oral meta-linguistic explanation (in which discussion and clarification occurred); (2) direct error correction with written meta-linguistic explanation (of the rule and an example of its use); (3) direct error correction; and (4) no corrective feedback. Feedback was provided on only two functional uses of the English article system (the indefinite article ''a'' for first mention and the definite article ''the'' for subsequent or anaphoric mentions). Groups one and three outperformed the control group while group two only just failed to do so. When the study was extended (Bitchener & Knoch, 2008a, 2009) to include an additional 69 learners, no difference was observed between the same three treatment combinations. Thus, it is possible that the larger sample size eliminated the difference in effect between group two and the other two treatment groups in the first study . Another study by Bitchener and Knoch (2010), investigating over a 10 month period the relative effectiveness of the same four different feedback approaches, found that each of the groups who received one of the treatment options outperformed the control group and that there was no difference in effectiveness between the three treatment groups, suggesting therefore that none of the

International Letters of Social and Humanistic Sciences Vol. 7
written CF options was any more effective than another. The special significance of this finding was its investigation over a 10 month period and therefore its longitudinal measurement of the effectiveness of different types of CF on accuracy retention.
Sheen's (2007) study of the relative effectiveness of two types of direct feedback (error correction and written meta-linguistic explanation) also found no difference between the two feedback options in the immediate post-test, but in the delayed post-test conducted 2 months later found an advantage for written meta-linguistic explanation over direct error correction. Sheen suggests that the passage of time may have been the critical factor in facilitating this delayed effect for meta-linguistic explanation.

1. Aims
The essence of this investigation is to study the controversial question of whether giving CF can helps low advanced L2 writers to achieve a greater proficiency over linguistic errors, measured by promotion and improving in accuracy when L2 writers use verb patterns after model verbs in writing in new written texts over an eight week period.

Research Question
Does using CF for low advanced EFL writers improve their accuracy in correct verb pattern after model verbs?

3. Participants
A total of 220 EFL students of Azad University Branch of Ghorveh, all male and ranging in age from 21 to 24 years participated in this experimental study. It should be stated that they are all originally Iranian and Persian was their first language. The participant were recruited to voluntary participate in a writing skill training program as a summer extracurricular course at Azad University. Prior to this program a test had been taken among 211 students and according to result 160 students were ranked as low advanced EFL writer.
There was also a survey questionnaire on how these 160 students liked to be provided with feedback and the interesting thing was almost 95 % liked to be provided feedback by direct instruction of the teachers and only 5 % liked to be provided feedback with their own investigations and studies.

4. Instrument
A pretest-posttest design was run. For collecting data three pieces of writing were required. The subject of these three writings was a-250-essay on "where do you want to go for your holiday?" this topic was selected because of some reasons: (1) the topic requires writers to use verb pattern after model verbs,(2)the program was held in summer so the topic was been like real one because students were heading for holiday. It should be mentioned that four teachers were responsible to run pretest-posttest design. A Pretest was conducted at the beginning of the instruction; an immediate posttest was conducted 3 days after the pretest and 45 minutes immediately after giving feedback, and delayed posttest at the last week (week eight). Three different types of written CF were used in this investigation: (1) direct written CF (DCF) in the form of written meta-linguistic (2) indirect written CF (ICF) in the form of underling (3) form focused written CF (FFCF) (4) and group four without any CF (NoCF). For analyzing data tests of one way ANOVA and two-way-repeated-measure ANOVA were used.

5. Procedure
160 EFL students of Azad University attended in four classes in the beginning of the course. They were divided into four classes in order to be suited with four types of treatments; (1) Direct written CF, (2) Indirect written CF, (3) Form focused CF, and (4) No giving CF. Group one were received DCF in the form of meta-linguistic, group two were received ICF in the form of underling, group three were received FFCF instruction, and group four were received no CF. On the day of registration they were told about the pretest in the first day of the course. On day one the pretest was administrated.
Three days elapsed and on fourth day the written texts were returned for three treatment groups with three different types of CF. Once again should be stated that there was no type of CF for the control group. After students considered their feedback about 45 minutes, the immediate post test was administered. Four days later the immediate post test results with different types of CF were returned. During the time between the immediate posttest and delayed posttest, teachers were required not to give instruction on verb patterns after model verbs. There was set a criterion for the scoring procedure: for each wrong use of verb patterns after modals there were -4 scores for students and the score was based on 100. It should be stated that in these three tests teachers just examined verb patterns after model verbs and nothing else. For assuring the ratter reliability, four teachers rated each piece of paper and then the average score was computed. At the last week (week eight) the delayed posttest was administered and its results were computed. Table 1 shows the descriptive statistics for three treatment classes given CF and the control group without giving CF. By analyzing the means, it is crystal clear that only three treatment groups were able to improve their mean accuracy at the time of immediate posttest and to make use of it in the delayed posttest. Decreasing in means of control group is obvious that indicates the importance and the effect of CF in accuracy promotion. DCF with the promotion of 5.50 in mean score might be show the importance of giving written CF in the form of meta-linguistic. All in all the promotion in mean scores of the three treatment group and the drop in the mean score of the control group crystal clearly show the importance and effect of giving different types of CF and based on them the answer to the research question is verifiably positive. Fig. 1 provides a visual representation of the mean score of three treatment group and the control group during the pretest, immediate post test, and delayed posttest.   1 illustrate while the three treatment groups increased in the delayed posttest, the rate of DCF is a little bit more than the other two other types of CF. it is also illustrating that the control group mean score in the delayed posttest dropped dramatically.

1)DCF 2)ICF 3)FFCF 4)NoCF
Because of the problem of one way ANOVA that indicate no statistically significant differences between the three groups at the pretest time, a two way repeated ANOVA was used. Three level CF types were entered as independent variable and scores plus time in three levels were entered as dependent variables. Table 2 clearly indicates the result that there is no significant interaction between types of CF and time p = .092. Knowing this result, the interpretation of the main effect became easier. Now CF types (one of the main effect) were significant; showing significant differences among the three groups p = 000. Time the other main effect isn't significant shows there is no significant difference over three testing time p = .519 Tukey's post hoc pair wise comparison was run to show the three groups were significantly different with each other.
Volume 7 The research question studied whether giving written CF on using correct verb patterns after model verbs helped low advanced EFL students to improve their writing accuracy. For addressing the question the results of the study clearly showed that all three treatment groups -Direct CF, Indirect CF, and Form focus CF-outperformed the control group in the immediate posttest and immediately after they were provided with different types of CF, like what happened in the study of use of English article done by Bitchener & Knoch,2010, treatment groups outperformed the control group. To say it in another way all three treatment groups' accuracy improved after they were provided with different types of CF. it should be considered that although treatment groups accuracy improved, the share of Direct CF is a little bit more than the other two types of CF and it is a good confirmation to what Bitchener & Knoch, 2010 investigated on the use of English articles.
These results support recent studies on correct use of verb patterns after model verbs with low advanced proficiency students (Sheen, 2007;Sheen et al., 2009). In each of these studies, the effectiveness of giving written CF was evident immediately after it had been provided in the immediate post -test pieces of writing and its effect was retained in the delayed post -test pieces of writing.
The consideration for the dramatically drop in control group in the delayed posttest is very interesting, yet questionable. It was revealed in the questionnaire which was administered at the first of the program that students have the habit of being provided with feedback directly and by their teachers so the rational reason for this drop is quite clear, they liked to be provided with feedback with the teacher and had no interest in obtaining it by themselves as they stated in the questionnaire that they liked to be prepared with direct feedback and when this providing feedback weren't done the scores dropped and the mean score dropped respectively.
The findings of this study is restricted to give corrective feedback to EFL low advanced students on correct use of verb patterns after model verbs so further research examining other linguistic features would be a useful follow up to this study.

International Letters of Social and Humanistic Sciences Vol. 7
A pedagogical implication of the findings of the study for giving corrective feedback on correct use of verb patterns after model verbs in EFL contexts is that giving direct form of CF in the form of meta-linguistic provides students with an excellent mean to improve their writing accuracy. For doing this, teachers can give a report on the students writings in which indicate the problematic part for the students.