Instructor ’ s Direct and Indirect Feedback : How do they Impact Learners ’ Written Performance ?

Written corrective feedback (WCF) is widely used among writing teachers in attempt to improve students’ language accuracy in their written tasks. Experts of second language (L2) writing stated that WCF plays a significant role in language accuracy development because it is related to the development and improvement of students’ accuracy in second language writing. Corrective feedback can be grouped into direct WCF and indirect WCF. However, results in many past empirical studies were inconclusive because of variation in how the effectiveness of WCF was measured. Additionally, there is a lack of investigation pertaining to the students’ perceptions toward the feedback they received, which may reveal important factors that influence the students’ reception of the feedback that resulted in performance changes. Hence, the purpose of this study is to explore factors that contributed to the accurate use of the past tenses resulting from direct WCF and indirect WCF. Interviews were conducted with 12 students and the qualitative findings revealed the following factors: 1) learner attitudes towards feedback provided, 2) learner beliefs towards what corrections entailed, and 3) types of scaffolding. Additionally, the findings revealed the importance of teacher-student conference, the need for students to be trained over a longer period of time in relation to the use of error codes and the need for instructor to understand and implement direct and indirect feedback accordingly in their teaching.


INTRODUCTION
The role of English language is significant in many Asian countries because it serves as a second language for international communication and telecommunications.For example, in the Malaysian context, English language has been incorporated into the Malaysian bilingual education system for the past fifteen years (Darmi & Albion 2013).However, mastering the language is still an issue for many students, especially from rural areas, despite learning it since the beginning of schooling.Furthermore, as stated by Hiew (2012), a portion of students fail to master English upon completing secondary school because of distinctive life background and different levels of English proficiency.When students' performance in English declines as early as at primary/elementary level, it is not surprising that their performance will continue deteriorating as they move to high schools and beyond.Even though all language skills are emphasized and taught extensively in schools, writing skills seem to be the most difficult skill for many learners.Experts like Han (2002), Havranek (2002), and Swain (1991) strongly believe that it is important for writing teacher to emphasize on language related focused feedback so as to improve students' written performance (as cited in van Beuningen 2010).These researchers believe that written corrective feedback (WCF) helps students in distinguishing their written errors and aids them in producing the correct forms.In terms of learners' reaction to instructor's feedback, researchers like Diab (2005) believed that the more comments learners read, the more useful those comments became to them.These students revealed that teachers' feedback acts as a "security blanket", whereby the comments helped them become better in writing; hence, they clearly believed in the effectiveness of such corrective feedback.While Ahangaria, Hejazib and Razmjouc (2014) state that feedback by instructors acts as scaffolding for students, especially for the less proficient ones.Based on the findings from past studies, it is evident that instructors and learners indicated that WCF plays an important role in aiding learners in enhancing students' writing skills.Yet, there aren't enough studies that focus on how students made use of the WCF feedback provided in their subsequent revision of their writings, which may reveal critical insights for teachers as well as L2 students.

PEDAGOGICAL APPROACHES LINKED TO ERROR CORRECRTION
In the early years, error correction was generally approached based on intuition about what seemed to be effective in practice.In other words, the approach to method does not rely on experimentation alone.According to Krashen (1982), it relies rather, on the insights, introspections and observations of experienced language teachers.Krashen further noted that although the results of WCF research were frequently published in professional journals, there was a lack of studies that looked into insights or perceptions pertaining to WCF.Furthermore, according to Krashen (1982), "mini-conferences" were often arranged by language teaching organizations so that experienced teachers could share their insights and techniques with others, yet its coverage was limited only to the words of the teachers on the techniques to be tried out in different classes.Moreover, empirical support for new techniques related to WCF is scarce.Studies in WCF in the past were based on the five fundamental questions listed by Hendrickson (1978, in Bitchener & Ferris 2012), and they relate to whether errors should be corrected, when errors should be corrected, which errors should be corrected, how they should be corrected, and who should do the correcting.
The first question listed by Hendrickson (1978) emphasizes on whether learner errors should be corrected.The main focus of this pedagogical approach is on reasons for correcting the errors.According to Corder (1973), George (1972) and Kennedy (1973) (as cited in Bitchener & Ferris 2012), correction was important because it was expected to help learners identify their own errors and discover functions and limitations of the syntactical and lexical forms of the target language.For instance in a survey conducted by Cathcart and Olsen (1976) on college students' attitudes toward error correction, the results revealed that students wanted their errors to be corrected and they wanted errors to be corrected more than what teachers believed was necessary.However Harmer (1983( , as cited in Ellis, 2013) ) argues that WCF has a place in ensuring 'accuracy' of written work but not in 'fluency' of written work.Harmer (1983) proposes that when students are involved in a communicative activity, the teacher should not interrupt by 'telling students that they are making mistakes' or 'insisting on accuracy.' Additionally, Harmer (1983) recommends that instead of interfering the students while they are still engaging in a communicative activity, the teacher should write down a list of errors that the students committed and highlight the errors to them when the activity ends.
In terms of when errors should be corrected, Ellis (2013) states that in the case of oral WCF, teachers have the choice of either correcting right away an error or making a note of the errors and delaying correction till later.The reason being correction interferes with students' effort to communicate and also makes students nervous.As a result, students are less ready to take risks.Another researcher, Gattegno (1972, in Ellis 2013) suggests that teachers to not rush into correcting student errors even in accuracy work, so that students have time to make sense of their mistakes.Similarly, Hendrickson (1978, in Bitchener andFerris, 2012) states that teachers generally reject the compulsive concern with error avoidance, but willingly accept a wide range of errors and they only consider errors which they think are the most problematic.
The third question highlighted by Hendrickson (1978) concerns which learner errors should be corrected.Hendrickson (1978) states that in terms of errors, teachers tend to consider three broad categories of errors which are worth of correcting, i.e., errors that substantially impair communication, errors that have significantly stigmatizing effects on the listener or reader, and errors that occur repeatedly in learners' speech and writing.Ellis (2013) warns against over-correction and proposes that teachers should be selective in the errors they correct.However, Ellis (2013) does not highlight which errors teachers should correct and which ones they should ignore.In relation to this, researchers like Corder (1967) makes distinction between 'mistakes' (i.e.performance slips) and 'errors' (i.e.deviations resulting from gaps in competence).Corder (1967) recommends that only the latter needs to be corrected.In addition to the distinction between 'mistakes ' and'errors', Burt (1975, in Ellis 2013) states that teachers should address 'global' errors but ignore 'local'.However, despite what the previous researchers wrote about the importance of correcting errors, these researchers seemed to fail to take into account the problems teachers face in determining whether a deviation is an error or a mistake, or whether it is global or local.
The fourth question posed by Hendrickson (1978) concerns how errors should be corrected.In the past, there was inadequate empirical support on how errors should be corrected, although various error correction methods were being advocated.James (1998) lists three principles in error correction that can be used to tackle students' errors.First, it involves techniques in error correction which allow students to improve their accuracy in expression.Secondly, it involves considering students' affective factors and ensures it is not face-threatening to them when correcting their errors.Lastly, it involves indirect correction as it encourages students to self-correct their errors using the heuristic method.It also involves presenting the correct form to students without embarrassing them.
Finally, the fifth question listed by Hendrickson (1978) relates to who should correct learner errors.Even though there are suggestions made about the value of implementing teacher correction, peer correction, and self-correction, there is still a lack of conclusive findings that provide convincing arguments pertaining to who should be tasked with correcting learners' errors.Hendrickson, 1978(as cited in Bitchener & Ferris 2012), points out that it is important for the outcomes of any corrective feedback approaches be supported or disproved by results of empirical studies.
Overall, it can be seen that studies in the past have shown that teachers and researchers tacitly carry out error correction in the language classrooms.Since providing written corrective feedback forms an important aspect in instruction in teaching writing, researcher have conducted many studies on error correction that focus on providing feedback in the written form.

PAST STUDIES RELATED TO WRITTEN CORRECTIVE FEEDBACK (WCF)
Findings from a longitudinal study carried out by Lalande (1982) suggested that "students who were given indirect WCF outperformed students that received direct WCF" (van Beuningen, De Jong & Kuiken 2008, p. 282).Findings from another study by van Beuningen, De Jong and Kuiken (2012) revealed that indirect group was involved in more form-focused activities than the direct group, because the two treatments differed in more aspects than just the provision of corrective feedback.In the Frantzen (1995) and Rob, Ross and Shortreed, (1986) studies, the findings exhibited that both direct and indirect WCF worked effectively.However findings from a study conducted by Ferris et al. (2000) revealed a dissimilar pattern, whereby indirect WCF worked effectively on students' accuracy gain in subsequent writing and direct WCF improved students' accuracy in revisions.In the study by Robb et al. (1986), the results showed no accuracy gains for any of their four feedback types (direct error correction; coded feedback; highlighting; marginal error counts).
In recent studies on direct vs. indirect WCF, findings from Rahimi and Asadi (2014) revealed that both direct and indirect groups improved significantly in revisions with students in the indirect group wrote more accurately in their essays over time.In Ghandi and Maghsoudi's (2014) study, the results indicated that students who received direct feedback did not show any improvement in spelling accuracy.Findings from another study reported that direct and indirect WCF had equal short-term effect in developing learners' accuracy; however direct WCF showed a more significant long-term effect as compared to indirect WCF (Salimi & Ahmadpour 2015).
Other studies have also examined the efficacy of different types of feedback involving direct and indirect WCF.Studies by Bitchener, Young and Cameron (2005); Bitchener (2008) and Sheen (2007), which explored the effectiveness of different types of direct WCF, revealed that direct meta-linguistic explanation resulted in a more positive effect than other types of direct feedback.Whilst studies from Ferris and Roberts (2001) and Kassim and Ng (2014) investigated different types of indirect WCF, specifically on coded vs. uncoded feedback and focused indirect vs. unfocused indirect WCF respectively.
Due to conflicting results, the specific effects of direct and indirect feedback remain unclear.While some studies suggest that indirect WCF may be more beneficial in particular contexts (e.g., Lalande 1982, van Beuningen et al. 2012, Ferris et al. 2012, Rahimi & Asadi 2014, Ghandi & Maghsoudi 2014), other studies (e.g., Chandler 2003, Bitchener & Knoch 2010, Salimi & Ahmadpour 2015) provide evidence that direct WCF may be more effective.There are studies (e.g., Rob et al. 1986, Frantzen 1995, Ferris et al. 2000) which also provide no difference across various types of direct and indirect WCF.Studies of the effect of WCF on written accuracy have, to date, yielded conflicting findings.Despite the abundant research on WCF, the findings on the effectiveness of different types of WCF remain inconclusive.Thus, there is a need for additional studies that compare the effectiveness of the two major forms of WCF: direct WCF and indirect WCF.

CONTEXTUALIZING THE OUTPUT HYPOTHESIS
The current study is guided by the framework based on the Output Hypothesis by Swain. Swain (1993) contends that comprehensible output (i.e.second language production) ensures mental grammatical processing and plays a significant role in L2 acquisition.According to Ellis (1998), Swain lists out three functions of the Output Hypothesis: the noticing function, the hypothesis-testing function, and the reflective function.The first function denotes learners' awareness towards certain linguistic forms which takes place in a language production.With the help from this function, learners are able to realize the linguistic "gap" in their interlanguage system and subsequently, noticing the "gap" pushes them to seek for sufficient knowledge to fill this "gap".The second function suggests learners to use a form of trial-and-error to test their comprehensibility of certain linguistic forms.Learners will notice what they do not know or what they partially know when they encounter linguistic gaps between what they want to write and what they are able to write.The process of testing gives them the opportunities to modify or reprocess their output when WCF invokes their awareness.The third function refers to learners' metalinguistic knowledge.In other words, learners reflect on the language they learn, and thus, the output enables them to control and internalize the linguistic knowledge.
In line with the Output Hypothesis, it is deemed important to explore what are the elements that are connected to direct and indirect feedback which affect students' responses towards the provided feedback.Thus the objective of this study to explore factors that contributed to the accurate use of the past tenses resulting from direct WCF and indirect WCF.Specifically, the present study aims to answer the following research question: 1. What are the factors that influenced the performance of the students in the use of past tenses in relation to direct WCF and indirect WCF?

METHOD
The data of the current study is extracted from a bigger study that employed a mixed methods research design in comparing the effectiveness of direct and indirect WCF.Semi-structured interviews that were conducted with selected participants forms the qualitative data used in this study.The qualitative inquiry is to explore the factors that influenced the performance of the students in the use of past tenses when direct and indirect WCF were provided.According to Mason (as cited in Edwards & Holland 2013), the advantage of semi-structured interview is that researchers can use a thematic, topic-centred, biographical or narrative approach.The topics, themes or issues can then be extracted according to what they aim to uncover, but with a fluid and flexible structure.Semi-structured interviews also enable researchers to ensure that relevant contexts are brought into focus in order to produce situated and contextual knowledge.
DATA COLLECTION PROCEDURE 12 high school students from a public school were involved in the current study.They consisted of 6 students who received direct feedback and 6 students who received indirect feedback.From the 12 students who were interviewed, the data from 6 students were selected and they were from those who had performed well, those who had demonstrated no progress and those who had shown a decline in their performance after receiving the direct or indirect WCF.Prior to the interviews, these participants had been exposed to direct and indirect feedback over a period of three weeks during their English lessons.Four-open ended questions were used to elicit insight about factors that improved students' accuracy in writing and students' perceptions on the types of feedback they received.Each interview session lasted between 10 minutes to 15 minutes.The interview questions consisted of: 1. What are the problems you face when you receive correction based on the error symbols or when your errors are corrected directly?(direct and indirect WCF groups) 2. How does corrective feedback help you in your writing?(direct and indirect WCF groups) 3. What did you do when you read your work marked with symbols?(indirect WCF group) 4. Will you recommend this feedback to be practised by teachers?Why?Why not?
(direct and indirect WCF groups) From the students' responses, the researcher attempted to identify factors that affect the students' accuracy in the use of past tenses, which resulted from the direct and indirect WCF received during the lessons.

ANALYSIS OF DATA
In order to analyse the interview data, a thematic analysis approach was employed to explore the relevant factors that influence the learners' perception and reaction toward the feedback received.In the context of the present study, the transcribed interviews were analysed manually in accordance with what is proposed by Braun & Clarke (2006), whereby it involves familiarization with the data, coding, searching for themes, reviewing themes, defining and naming themes, before producing the report.A total of 12 interviews were transcribed and analysed.

RESULTS AND DISCUSSION
Base on the transcribed data, as shown in Table 1, three main factors emerged as a result of the thematic analysis that had been carried out.The three main factors are 1) learner attitudes towards the two WCF, 2) learner beliefs about what correction entailed, and 3) three types of scaffolding.Table 1 shows the overall codings detected from individual students who were interviewed.Base on the table, in terms of learner attitudes towards the two WCF, for students who received direct feedback, the sub-codlings could be divided into 2 areas, one that relate to the fact that the students viewed Direct WCF as clear and straightforward, and another that relate to the fact that Direct WCF did not pose any challenges.Whereas in terms of learner beliefs towards what corrections entailed, it can be sub-divided into learners who preferred to have their errors pointed out, and those who preferred teacher to explain what their mistakes were.For the students who received indirect feedback, apart from their beliefs towards what corrections entailed and attitudes towards feedback received, the types of scaffolding seem to impact how they view the functions of WCF.In order to delineate the factors related to students' reception of the received feedback, the first part of the next section presents the responses related to direct feedback.The subsequent part reveals the responses related to indirect feedback.

PART ONE: RESPONSES RELATED TO DIRECT FEEDBACK
The results revealed that there were two major factors that influenced direct WCF in students' performance in the use of the past tenses and they are: 1) learner attitudes towards direct feedback and 2 The first factor that influenced direct WCF in students' performance is learner attitudes towards the feedback.The findings suggest that students perceived direct feedback as very straightforward and clear.Figure 1  In general, Student A and Student B from the direct WCF group seemed to favour direct feedback.They agreed that corrective feedback was helpful in improving their next piece of written work.Student A favoured direct WCF because there were correct answers written on top of the sentences.Similarly, Student B preferred direct WCF because the correct answers were given by the researcher.Since the direct feedback was easy and clear, both Students A and B indicated that they were able to write better.
However, Student C from the direct WCF group expressed her concern about the benefit of the direct feedback she received.At the revision stage of the draft, she highlighted the problems linked to the teacher's feedback.Figure 2 below outlines the excerpt.

Researcher
: What are the problems you have to cope when your mistakes are corrected directly like this?
Student A : There was no problem at all because there are correct answers written on top of the sentences.That's why I don't think there are any problems with me doing the correction.
Student B : I didn't have any problems, teacher because the corrections are all given by you.I find it very helpful.

Direct WCF
Researcher : What are the problems you have to cope when your mistakes are corrected directly like this?
Student C : First, I think the correction is really good in general because I know nothing about grammar correctly.But the thing is, I think this kind of correction will make me feel lazy.Like, I don't have to think.I just have to write and put whatever that is already written.Second, I don't know exactly know the difference between 'brought' and 'took.'It's basically the same thing.
Student C stated that although the direct feedback she received was good, she was concerned that the direct feedback did not require her to do much thinking.Student C also stated that the researcher's feedback changed the intended meaning.However she did not entirely reject the feedback because she was willing to incorporate the corrected answers.Student C felt that she still needed the feedback because her grammar was weak.According to Swain and Lapkin (2002), in relation to corrective feedback and learner attitudes, learners may reject teacher feedback because it is perceived as violating their own beliefs about language conventions or as altering their intended meaning.In the context of this study, Student C did not totally disagree with the corrective feedback, as a whole, she still agreed that direct WCF was beneficial.
DIRECT FEEDBACK DID NOT POSE ANY CHALLENGES Some students from the direct group revealed that direct feedback was very straightforward and clear.It did not pose any challenges in terms of figuring out the correct form of language.The excerpt related to this is shown in Figure 3. Student A indicated that not only the direct feedback was more straightforward, it also helped her notice her spelling and grammar mistakes easily.She did not have to work out the answers on her own.According to Bitchener (2008), direct feedback is easy and straightforward because there is no additional delay in knowing whether learners' own hypothesized correction is correct.

LEARNER BELIEFS ABOUT WHAT CORRECTION ENTAILED IN RELATION TO DIRECT FEEDBACK
The second factor that influenced direct WCF in students' performance is learner beliefs about what correction entailed in relation to direct feedback.It was discovered that students rather have their errors pointed out by the teacher.Figure 4  In Figure 4, there was a clear recognition that the students from the direct WCF group benefited from the direct feedback they received.They expressed an understanding what the corrections entailed, indicated preferences to have their errors pointed out rather than to have them merely underlined.The reason being by simply underlining the errors, they did not know what kind of errors have been committed.In other words, the students had a firm belief that direct WCF helped them notice the errors.If not being provided with direct feedback, they would fail to notice the errors and subsequently improve their performance.In Figure 5, Student C suggested that providing written feedback solely is insufficient; instead, the writing teacher needs to explain to the students what their mistakes are.In other words, an additional form of direct WCF, which is a one-on-one individual conference between teacher and student, is essential.This seemed to relate to what is deemed as writing conference by Hyland (2006), who states it as a two-way interaction between teacher and student(s), whereby meaning and interpretation are constantly being negotiated by participants, and subsequently it provides both teaching and learning benefits.
According to Saito (1994), in the first language (L1) setting, teacher-student conferences that involve a teacher and a student interacting individually about the students' writing, have become increasingly popular in writing instruction.Findings of a past research on L1 writing carried out by Carnicelli (1980, in Saito 1994) found that two-way communication in a writing conference appeared to be more effective than written comments because it allowed students to convey their opinions and needs, and to get clarification from the teacher.In the case of L2 setting, teacher-student conferences do work as effective as teacher-student conferences in the L1 setting and these are demonstrated by various studies like Goldstein andConrad (1990, in Saito 1994), Amrhein and Nassaji (2010), and Tootkaboni and Khatib (2014).
Similarly, a study carried out by Mansourizadeh and Abdullah (2014) compared the effects of oral and written meta-linguistic feedback in improving students' accuracy in writing, whereby students were grouped into those who received direct WCF with written meta-linguistic feedback, those who received direct WCF with oral meta-linguistic feedback and those who received direct WCF with oral meta-linguistic feedback and collaborative interactional activity.The findings revealed that learning occurred in all three conditions; therefore, it can be concluded that meta-linguistic feedback can enhance L2 students' writing skills as the feedback improves awareness of language rules and noticing, which is crucial in language learning.

Researcher
: Will you recommend direct feedback to be practised by your writing teacher?If yes, why?If no, why not?
Student C : Yes, but only if it's used to set up the base for the students, but not to use it in a long term.Set up the base first, and then let them use their own words and let them make the mistakes and learn from their mistakes so that teachers can guide them from there.And explain to them what are their wrongs and where are their mistakes.I don't think instead of just write it on the students' paper is enough.The teacher should explain why this is wrong and this word should be used instead of this word.
Students who received feedback in the form of teacher-student conference or even meta-linguistic feedback considered the feedback as opportunities to individually interact with their teacher.Hyland andHyland (2006, in Abdollahifam 2014) emphasize that when feedback is contextualized and personalized, students like it and they tend to pay more attention to it.All three students (D, E and F) in the indirect WCF group seemed to face difficulties in the beginning of receiving the feedback.The reason being they were not familiar with the error codes.Student F, for example, admitted that when she was first introduced to indirect feedback, she corrected her errors by trial and error because she was not familiar with the error codes.Although all three students in the indirect WCF group experienced difficulties when first introduced to indirect WCF by the researcher, they were determined to understand the error codes and to correct their errors.The students persisted to attempt to understand the error codes and they did not reject the feedback given. .Moreover, in the process of understanding the error codes, students from the indirect group were challenged to find the correct answers according to the codes assigned.Even though indirect WCF can cause confusion in understanding the error codes (Bitchener and Knoch, 2008), this did not impede Student D, Student E and Student F from trying to attend (Indirect WCF)

Researcher
: What are the problems you have to cope when your mistakes are corrected indirectly like this?
Student D : I have a problem with my past tense.Like, I can't put the past tense in the correct way.I can't put it correctly because I didn't understand the symbols at first, even with the guideline paper for the symbols.After you explained and wrote other examples on the whiteboard, then I understood.The symbols are not too confusing.
Student E : At first, when teacher show the symbols, I don't understand, because a bit confused and um, it looks like weird and uh, all symbols my teacher didn't use before.And then, after teacher explain and she write on the board, I understand better.After that, when teacher used again, I don't feel confused anymore.Uh, and then, when I see the symbols, I have to try to guess.I need to know what I need to put in.
Student F : Okay, I have to face, like when I first got back the paper with the correction, I really didn't know what the symbols meant.You also gave me a guideline paper to the symbols and I referred these symbols based on the guideline.I saw the examples in the guideline, so I tried to do the correction by trial and error.When I first received the symbol guideline, I didn't really understand.I needed your explanation help me because I've never seen these symbols before.It's my first time.So, as weeks passed by, whenever I saw the symbols, I managed to understand them.
to their correction.Students are likely to feel satisfied and motivated when they successfully figure out the meaning of the error codes; thus, it encourages them to try to produce the correct answers.Evidently when students have a positive attitude towards error correction, their attitude will positively influence the scope of engagement in learning a language.Although students in the indirect group were likely to accept the feedback, the feedback itself consumes more cognitive effort.The reason being when students make their own corrections, the process also involved more cognitive processing and additional delay in knowing whether their own hypothesized correction is correct.The following excerpt illustrates the situation.In Figure 7, Student D, who is from the indirect WCF group, indicated that she learned more from finding her own errors and making her own corrections.The additional delay occurred when Student D had to guess, indicating that she had to test out whether her hypothesis on the error codes was correct.Nevertheless, her main concern was on her linguistic errors.In relation to this, the findings in the Enginarlar (1993) and the Diab (2005) studies, which were based on their respective surveys of English-as-a-foreign-language (EFL) learner attitudes toward feedback techniques, concluded that students were concerned with accuracy and perceived attention to linguistic errors as effective feedback from teachers.Furthermore, Cumming (1995, in Hyland 2003) states that ESL students in academic contexts place a particularly high value on form-focused feedback, which resulted in them paying attention to linguistic forms.Moreover, students perceive having error-free work as highly desirable.Therefore, despite the additional delay due to hypothesis testing, Student D perceived the feedback as beneficial because it helped her focused on her linguistic errors.

LEARNER BELIEFS ABOUT WHAT CORRECTION ENTAILED DESPITE RECEIVING INDIRECT FEEDBACK
The second factor that relates to how indirect WCF impact students' performance is that, students preferred to have their error types identified when it comes to receiving indirect feedback.Figure 8  Students in the indirect WCF group expressed their preference for having the error types identified via the error codes as compared to having the errors merely underlined.When their errors were identified through the error codes, the correction helped them notice the errors.Without the error codes, they might fail to recognize the errors.

TYPES OF SCAFFOLDING THAT INFLUENCE STUDENTS' PERFORMANCE
Scaffolding is a process whereby learners are given support until they can apply the new skills and strategies independently (Rosenshine & Meister 1992, as cited in Larkin 2002).Bitchener and Ferris (2012) noted that when learners get the appropriate amount of scaffolding from teachers and more advanced peers, learners can eventually be self-regulated (i.e., able to use the L2 autonomously).In particular, it is believed to be most effective within the learner's zone of proximal development (ZPD).ZPD is derived from a socio-cultural theory of human mental processing based on the work of Vygotsky.It is also related to the formation of procedural knowledge, a domain or skill that happens when the learner is not yet capable of using the L2 autonomously (Bitchener & Ferris 2012, p. 18).However, with the scaffolded assistance of the more proficient partner, the learner's level of performance can be raised.In relation to WCF, scaffolding offers instructors means to evaluate learner's learning of literacy skills, and is helpful to students with less proficiency in English as it improve their writing skills (Ahangaria, Hejazib & Razmjouc 2014).Thus, scaffolding in writing allows a shift in students' language development to the ZPD.In connection with ZPD, scaffolding can be incorporated at various stages like the Teacher stage, the Class stage, the Group stage, and the Individual stage.

SCAFFOLDING BY INSTRUCTOR
Scaffolding by instructor is one of the scaffolding that influenced students' performance.Scaffolding by instructor means that it involved the instructor demonstrating the way to perform a new or difficult task.When students could not get the answers, the instructor and students would work together to find the answers.An example of this is illustrated in the excerpt below.In Figure 9, Student D of the indirect WCF group stated that she needed the guideline symbols (i.e.error chart) in order for her to understand the error codes.Additionally, the aid from the teacher via providing examples also facilitated her correction.As mentioned by Ellis and Larkin (1998), during the initial Teacher stage of the scaffolding process, the instructor usually need to introduce and model the task for students.In this case, the researcher introduced the error codes and modelled the codes via explaining the error chart and examples, which eventually helped Student D in figuring out the answers.10 explained when she did not understand the error codes, she worked together with her friend before correcting her work.Working together cooperatively with her friend was facilitative towards attending to her corrections.According to Ellis and Larkin (1998), the Group stage is a form of guided practice or peer-mediated practice.Peer-mediated practice is important because students may learn from their peers as much as they do from the teachers.Additionally, this process is crucial because the practice offers opportunities for students to interact and dialogue among themselves about various aspects of performing the task.In this case, Student E opted for peer review, allowing her to discuss her texts and discover other's interpretations of her errors with her classmates.
Similar findings related to scaffolding in WCF were also revealed in the study by Aljaafreh and Lantolf (1994).They conducted a longitudinal study of adult L2 learners receiving one-to-one written feedback from their language tutor on weekly writing assignments.The Aljaafreh and Lantolf (1994) study incorporated the Vygotsky's notion of ZPD to analyze the interaction between learner and tutor during the error correction process.The findings of the study revealed that effective error correction and language learning depend essentially on mediation supported by other individuals.Learners who engage with other individuals dialogically in constructive feedback are able to co-construct a ZPD because the feedback serves as scaffolded guidance.When scaffolding becomes relevant to learners, it can therefore be appropriated by learners to modify their interlanguage systems.Aljaafreh and Lantolf (1994) concluded that learning is not something an individual does alone, but is a collaborative effort that necessitate the involvement of other individuals.

SCAFFOLDING RELATED TO SELF-INITIATIVE
Students' self-initiative in finding the correct answer is another factor that influenced their performance.In other words, students tried to solve the problem by themselves without any assistance from the instructor or peers.An example is shown below.

Researcher
: What did you do when you read your work marked with these symbols?
Student E : At first, I have to read the guideline.I have to understand the guideline and then, I checked with my friend when I don't understand before I made the correction.The second time for the correction, it is quite easier for me because um, I really understand the guideline and er, yeah.It's easier for me to make correction.

Indirect WCF
Researcher : What did you do when you read your work marked with these symbols?
Student F : After I read the first draft, I did my correction.The symbols are not confusing.The correction was not difficult because I had to rewrite the first draft with the help from the guideline paper.Also, I didn't feel worried that much about making more mistakes.
In Figure 11, Student F of the indirect WCF group stated that with the assistance provided by the guideline paper, she was able to do the corrections by herself.She found the error codes not confusing because the explanation and examples provided in the error chart were clear.According to Ellis and Larkin (1998), when students work at Individual stage, this form of student-mediated practice gives them the opportunities to practice and to build fluency, and both the overt and covert behaviours associated with the task can be performed automatically and quickly.In other words, when students have mastered the task given, they become independent.In the case of Student F in this study, she was able to self-regulate her learning with the feedback received; thus, she was capable of independent problem-solving.

IMPLICATIONS OF STUDY
The findings of this study revealed that in terms of both direct and indirect feedback, students' beliefs and attitudes towards what written corrective feedback entailed affect how they perceive and respond to the feedback.However, for students who received indirect feedback, scaffolding by instructors, and scaffolding by collaborating with fellow students as well as self-initiative influence how these students act on the feedback in their subsequent revision of their writing.Hence, teachers should be made aware of the impact of these two types of WCF.Yet in general, neither direct nor indirect WCF was employed by teachers, thus many students were not aware of the benefits of these two feedback options.Furthermore, it seems that students in the indirect group need more time to learn to use the symbol guideline (i.e. the error chart).Students from the indirect group were initially given two weeks of treatment (week 4 to week 6) to enable them to be familiarized with the error chart.Nevertheless the students stated in the interviews that they still faced difficulties in understanding the feedback that contains error codes, which suggests that students need to be given longer period of training to enable them to understand what the error codes entail.In addition, one student from the direct group pointed out the needs of having a one-on-one conference feedback.In this study, a combination of WCF and one-on-one feedback would enable students to use the past tenses with improved accuracy, as the combined feedback option allows teachers to discuss with their students which linguistic errors should be focused on.

LIMITATIONS AND CONCLUSION
Overall, the findings of the present study revealed that there were three factors that influenced the performance of the students in the use of past tenses in relation to direct WCF and indirect WCF: 1) learner attitudes towards feedback provided, 2) learner beliefs towards what corrections entailed, and 3) types of scaffolding.Furthermore, the findings evidently show that both direct and indirect WCF had positive effects in improving students' accuracy use of past tenses.The qualitative findings revealed that students favored both direct and indirect WCF.It seems that students in the indirect WCF group need more time to adapt to the use of the error chart.The findings also show that students need to be trained over a longer period of time before they could practice on their written tasks.The guidance from the class teacher on the error codes can enhance students' awareness towards their writing weaknesses.
Additionally, the findings also highlighted the importance of one-on-one feedback.The one-on-one discussion between instructors and students may potentially enhance the student's language learning experience and subsequently improve their writing abilities.
Future research may want to consider one-on-one discussion between teachers and students as a research variable.Finally, there are some limitations pertaining to this study.The present study only focuses on direct and indirect corrective feedback, as well as on one targeted linguistic feature, that is the past tenses, and not other linguistic features.Additionally, the present study focuses on one genre of writing task, which is descriptive writing.Future researchers may want to review these limitations before finalizing their research design.
) learner beliefs about what correction entailed in relation to direct feedback.The excerpts of the 3 students (Student A, Student B and Student C) from the direct WCF group are presented below.(Note: excerpts have not been corrected for grammatical errors.)LEARNER ATTITUDES TOWARDS DIRECT FEEDBACK FIGURE 1. Students' responses towards the clarity of direct WCF

FIGURE 2 .
FIGURE 2. Student's response towards the disadvantage of direct WCF

FIGURE 3 .
FIGURE 3. Student's response on acceptance towards direct WCF LEARNERS PREFERRED TEACHER TO EXPLAIN WHAT THEIR MISTAKES WEREInterestingly, Student C of the direct WCF group had a different belief: she preferred the teacher to explain what her writing mistakes were.Her belief is further discussed below.

FIGURE 5 .
FIGURE 5. Student's response on her suggestion towards direct WCF PART TWO: RESPONSES RELATED TOINDIRECT FEEDBACK    In terms of how students reacted to indirect feedback, the results revealed that there were three major factors that influenced the indirect WCF students' performance in the use of the past tenses, they are 1) learner attitudes towards indirect feedback, 2) learner beliefs about what correction entailed despite receiving indirect feedback, and 3) types of scaffolding related to indirect feedback.The excerpts of the 3 students (Student D, Student E and Student F) from the indirect WCF group are outlined below.LEARNER ATTITUDES TOWARDS INDIRECT FEEDBACKIn the case of the indirect group, most of them seemed to face problems in the beginning of receiving indirect feedback.The first factor that influenced indirect WCF in students' performance is shown in Figure6below.

FIGURE 6 .
FIGURE 6. Students' responses towards the disadvantage of indirect WCF

FIGURE 7 .
FIGURE 7. Student's response on her acceptance towards indirect WCF

FIGURE 9 .
FIGURE 9. Student's response towards the first type of scaffolding

FIGURE 11 .
FIGURE 11.Student's response towards the third type of scaffolding

TABLE 1 .
Codings related to factors linked to students' perceptions of WCF below illustrates the situation.The guideline paper helps me to, er, find correct examples.When I see the symbols, I don't know my mistakes.Then, I look at the guideline paper, there got examples.I see examples and I look at mistakes.After that, I guess and guess.It takes time, but the symbols actually help me to see what mistakes I make.For example, if I see symbol 'vt', I know that is present tense or past tense mistake…or if I see symbol 'art', I know I make mistake with 'a' or 'the'.I think this correction helps me to improve my grammar.I learn to write more.
so confused, okay.I had no idea what you wrote on my paper.After you gave me the guideline and showed some simple examples on the whiteboard and explained them to class, then only I understood.After that, I just did my correction.Scaffolding by collaborating with fellow students is another type of scaffolding that influenced students' performance.It involves the student working with a partner or a small cooperative group to complete a task.An example of the situation is elucidated below.FIGURE 10.Student's response towards the second type of scaffolding Student E of the indirect WCF group in Figure