Multiple-Draft/Multiple-Party Feedback Practices in an EFL Tertiary Writing Course: Teachers’ and Students’ Perspectives

Based on various sources of data collection for a qualitative research project, the study reported in this paper set out to examine four teachers’ and sixteen students’ perceptions of a multiple-draft/multiple-party feedback approach to English as a Foreign Language (EFL) student writing. This approach had been implemented as a trial in a tertiary setting in Vietnam. Three sources of feedback at three phases were provided. These included (1) peer/group written and oral feedback on the students’ first drafts, (2) a teaching assistant’s written and oral feedback on their second drafts and (3) the lecturer’s written feedback on their final drafts. Content analysis of the data revealed that all participants valued this multiple feedback approach because of its practicality and the quality of the feedback which participants believed contributed to writing improvement. Based on the participants’ reactions, the study highlights the potential of multiple-draft/multiple-party feedback practices for improving English language writing in a tertiary context.


Introduction
There is little doubt that providing feedback on student writing is regarded as one of the most important tasks for English language teachers (K.Hyland, 2003).However, in spite of teachers' concerted efforts to give feedback, students continue to make the same errors and depend on the teacher to provide feedback (Lee, 2011b).One possible reason is the quality and delivery of the feedback.For example, a common practice in the Asian context is for teachers to highlight errors and allocate marks to students' final writing products (Lee, 2011b(Lee, , 2011d)).Such a practice can overwhelmingly result in students keeping a greater eye on grades or scores and less on teachers' comments as Black, Harrison, Lee, Marshall, and Wiliam (2007) and Lee (2011bLee ( , 2011d) ) contend.Accordingly, opportunities to focus on the learning process are lost.In other words, assessment for learning (AFL) is not effectively activated in such feedback practices.
In Vietnam, teaching and learning about writing are normally believed to be challenging for both teachers and learners since they make "a lot of demands on any teacher and learner of EFL" (H.H. T. Nguyen, 2009, p. 61).Consequently, research has focused on examining teachers' and learners' motivation, attitudes and perceptions towards teaching and learning EFL writing in Vietnam (for example, in studies by T. M. H. Nguyen & Hudson, 2010;Tran, 2007) rather than on the application of AFL in feedback practices in general and/or in multiple-draft feedback practices in particular, the research interest of the study reported in this paper.The study, which is part of a larger project that described practice, then developed and trialled an alternative assessment feedback approach (see T. H. Nguyen, 2016b), attempts to foreground the ten principles of AFL introduced by the Assessment Reform Group (ARG) in 2002.It will do this through (1) its evaluation of a trialled multiple-draft/multiple-party feedback (Filipi, 2017) approach and (2) its analysis of the teachers' and students' perceptions and reactions to the approach in two classes of an EFL tertiary writing course in Vietnam.Specifically, the study seeks to answer the following research questions: 1) In what ways does the trialled multiple-draft/multiple-party feedback approach in an EFL tertiary writing course in Vietnam align with the identified principles of AFL? approach?
We start by presenting an overview of the concept of AFL and a brief summary of previous research with regard to teachers' and learners' perceptions of feedback practices in English writing classes before turning to the details about the study.

The Concept of AFL and Its Links to Feedback
The Assessment Reform Group (ARG) (2002, p. 2) defines assessment for learning (AFL) as: "the process of seeking and interpreting evidence for use by learners and their teachers, to identify where the learners are in their learning, where they need to go to and how best to get there".It has developed a list of ten principles which we believe are workable in a classroom context from the perspectives of both teachers and learners.These principles can inform the development of tasks that promote AFL so that they occur as ongoing tasks in the classroom rather than as separate tasks that are tacked on or additional to them (Black et al., 2007).The ten principles characterize AFL as: 1) being part of effective planning 2) focusing on how students learn 3) being central to classroom practice 4) being a key professional skill 5) being sensitive and constructive 6) fostering motivation 7) promoting understanding of goals and criteria 8) helping learners know how to improve 9) developing the capacity for self-assessment 10) recognising all educational achievement A number of the above principles (2, 5, 6, 7, 8, 9 and 10) refer to the centrality of feedback and to how it contributes to raising learners' achievement (Swaffield, 2008).Indeed, various scholars (e.g., Black et al., 2007;Brookhart, 2008;Lee, 2011a;Swaffield, 2008) maintain that there is a close and supportive relationship between quality feedback which should be both "descriptive and diagnostic" (Lee, 2017, p. 16) and AFL.For example, Swaffield (2008, p. 70) maintains that: "feedback is a direct way of making learning explicit, and it has a large part to play in helping pupils become more autonomous learners and supporting their development of agency" while Brookhart (2008) maintains that good feedback has the potential to strengthen AFL by stimulating students' cognition and motivation.
The role of feedback in AFL contributes significantly to the current investigation on feedback practices with its specific focus on how to stimulate the implementation of AFL in EFL tertiary writing classes.As noted in the definition of AFL, both teachers and learners play significant roles in the process of ongoing learning and assessment.Therefore, in order to understand how feedback can be built on to have a significant role in improvement, we first need to understand from previous research how teachers and students have reacted to various feedback practices.

Teachers' Perceptions
Research on teacher feedback on writing in English as a Second Language (ESL) and EFL classrooms reveals that teachers believe the purposes of feedback are to: (1) help students increase their language awareness as well as their writing skills (Evans, Hartshorn, & Tuioti, 2010;Lee, 2011aLee, , 2011d)); (2) meet students' demands and expectations (Evans et al., 2010;Ferris, 2014;Lee, 2008;Montgomery & Baker, 2007); (3) meet the requirements of their schools and/or programs (Evans et al., 2010;Lee, 2008Lee, , 2011d)); and (4) develop learners' motivation (Evans et al., 2010;Lee, 2011d).Research has also been conducted on the patterns of feedback in written and/or oral teacher feedback on different drafts in: ESL contexts (Ferris, 2006(Ferris, , 2014;;Paulus, 1999), EFL secondary settings (Lee, 2011a(Lee, , 2011c(Lee, , 2011d) ) and Turkish (Ekşi, 2012) and Chinese tertiary contexts (Yang, Badger, & Yu, 2006).While most of these previous studies have focused on the purposes of feedback in English writing classes in general and on the benefits of teacher and peer feedback in developing students' writing through a multi-draft feedback approach in ESL contexts or EFL secondary schools in particular, only a very small number of studies (e.g., Ekşi, 2012 andYang et al., 2006) have investigated the multiple-draft feedback approach in EFL higher education.The current study seeks to extend the literature on teachers' perceptions about various sources of feedback in a specific EFL tertiary writing course in Vietnam.

Students' Perceptions
In studies of students' perceptions, prior research has examined students' opinions about teacher and peer feedback in English writing classes.A substantial number of studies indicates that ESL/EFL students favour teacher feedback provided either directly or indirectly.For example, students believe that direct feedback/correction is helpful for effective revision (Chandler, 2003;F. Hyland, 2003F. Hyland, , 2011) ) and time efficient both for students and teachers in revising and correcting drafts (Chandler, 2003).Teachers' indirect correction is also valued by students for its perceived positive impact on their learning; for example, in developing their language acquisition (Chandler, 2003;Leki, 2006) and in enhancing their learning autonomy (F.Hyland, 2011).
English language learners have also been shown to value peer feedback for the following reasons.First, it can strengthen their autonomy in multiple-draft feedback practices (Ekşi, 2012;Yang et al., 2006).Second, engaging in peer feedback by reviewing their writing helps to develop their vocabulary and understanding of sentence structure (Ekşi, 2012;Sato, 2013;Yang et al., 2006;Zhao, 2014).Third, opportunities for peer interaction help students establish mutual understanding by limiting misinterpretation or miscommunication, and in so doing they believe they can develop their speaking skills (Sato, 2013;Yang et al., 2006;Yu & Lee, 2014;Zhao, 2014).Thus, peer feedback is believed to benefit students' language acquisition.
In summary, while this brief review illustrates that students appreciate both teacher and peer feedback, there appears to have been less attention paid to systematic classroom approaches that combine both multiple-party and multiple-draft phases, particularly in the Vietnamese context.This suggests a fruitful area for research.

Context and Participants
The University of Success (a pseudonym) is a state university whose primary objective is the professional development of teachers for all educational levels in Vietnam.One of the major objectives of the Faculty of English in this university is to educate English majors in teaching English as a Foreign Language (TEFL) as part of a four-year Bachelor of Arts (BA) program.The research reported in this paper was focused on the third-year English writing program in which there were fifteen class hours dedicated to writing skills per semester.Each class lasted fifty minutes.In the first semester of this course, a key focus of learning was to be able to produce a written task of at least 150 words that summarised the information illustrated in a table, a chart or a process -a task similar to Writing Task 1 in the IELTS academic writing module.In the second semester, students were taught to produce argumentative and discursive tasks of at least 250 words -similar to Writing Task 2 in the IELTS academic writing.
The participants, from two writing classes, included sixteen third-year English majors, two lecturers and two teaching assistants.Pseudonyms were used for all the participants to ensure ethical research practices.

Data Collection and Analysis
Data for this study was derived from twenty individually audio-recorded, semi-structured interviews of 30-45minute duration, classroom observations of two 50-minute writing sessions in each class, and five students' writing samples from each class.Class observation data was gathered in week 6 and week 9 of the second semester.In these writing classes students sat in groups of four to provide written feedback on the first drafts written by their peers in a different group.Interview and document data was collected at the end of the second semester.Interview questions for student participants differed slightly from those for teacher participants because of their different roles in the writing classes (see Appendix A).Responses to the interview questions were given in Vietnamese by the participants and then translated into English by the first author.
In this qualitative research, content analysis (Lichtman, 2014) was adopted as the primary method for data interpretation and analysis using the "Three Cs strategy" of coding, categorizing and conceptualizing to seek major themes from the interview and observation data.The three-step strategy was conducted in the following ways.First, different symbols for coding purposes were assigned to each participant group.Second, all the data was interpreted, summarized and categorized into sub-themes and major recurring themes across the data sample to address the research questions.Third, the first author's own reflexivity in her role as a teacher of English writing skills in Vietnam was used for interpretation and analysis.Such an approach accords with Patton (2002) who stresses that the researcher should pay attention to reflexivity to strengthen the credibility of the findings.
In analysing and interpreting the data, a "thick" description (Denzin, 1989) was conducted to enable a "thick" or deeper interpretation of the data.In order to achieve the former, actual quotations from the participants, researcher notes from class observations, examples from students' writing drafts and peer and teacher feedback were used to formulate findings.In providing a thick interpretation of the data, analysis of the similarities and differences between the teachers' and students' perspectives about the multiple-feedback approach was made with reference to the ten principles of AFL discussed above and by drawing on the literature on feedback practices in L2 writing classes.

How the Multiple-Draft/Multiple-Party Feedback Approach in Two EFL Tertiary Writing Classes in Vietnam
Aligns with the AFL Principles

The Multiple-Draft/Multiple-Party Feedback Approach
The development of the multiple-draft/multiple-party feedback approach to EFL students' writing was a major component of a doctoral research project which was piloted in the second semester of the third-year writing program in an undergraduate course in TEFL at the University of Success, Vietnam (see T. H. Nguyen, 2016b).Description about this trialled approach is illustrated in Figure 1, followed by an analysis and discussion about the ways it links to the AFL principles.
As can be first draft f receive the receive the which alig for self-ed

B. By dback
Analysis of the ten samples of student writing in both classes revealed that the teaching assistants delivered feedback on all four criteria in the checklist by ticking and providing additional written feedback in the relevant column for each criterion (see Appendix B).Consistent with Lee's (2017) claim that quality feedback needs to be both descriptive and diagnostic, the feedback provided both direct correction and detailed comments on each of the criteria in the checklist and pointed out where improvements were needed.In this respect, the teaching assistants were potentially enacting the eighth principle of AFL which facilitates learners in making improvements with teachers' constructive guidance.Hoa's statement illustrates this point: When giving written feedback, I focused on errors in all four criteria in the checklist and noted common errors that most students had made.I even noted down serious mistakes that a few of the students had made to mention in the general oral feedback session.
In this phase, the teaching assistants also made notes for a subsequent whole class "clinic" that occurred directly after papers were returned to the students.They gave oral feedback on common errors against each of the criteria in the checklist by drawing on relevant examples from the student drafts.Some analysis of good or poor sample answers was also provided.Throughout the session, they asked learners for their input as demonstrated in the following: Regarding oral feedback, I usually wrote some of the students' sentences on the board and asked the whole class to indicate errors and suggest ways to improve them.When their suggestions for correction were not similar to mine, I would suggest my own correction.(Thu) These clinics provided an opportunity for the students to develop their critical thinking about error analysis and improvement because they were actively engaged through invited interaction to make suggestions and analyse errors.It allowed students to further enhance their language awareness and extend their experience of giving feedback during the first draft process.Therefore, the potential for three principles (7, 8 and 9) of AFL were activated because students were able to more explicitly understand the criteria for assessment and the purpose and goals of their writing.This in turn enabled them to know how and where to improve, and at the same time, develop their capacity for self-assessment while reviewing by better understanding how to work with feedback received.
A further point about the importance of making feedback comprehensible by promoting understanding about goals and criteria (AFL principle 7) and providing feedback that is constructive (AFL principle 5) is the medium chosen for delivering it.The classroom observations indicated that the teaching assistants used their first language, Vietnamese, in giving their oral feedback.The choice of L1 in oral feedback delivery was reported to exert an immediate benefit to the participants.Lan, a student in Class A, stated that "the teaching assistant usually gave oral feedback in Vietnamese.In doing so, we could understand her feedback easily".
The support of L1 in the L2 setting in this study is in line with the literature about the role of the L1 in L2 learning and teaching (e.g., Filipi, 2018;Storch & Aldosari, 2010).For instance, Storch and Aldosari (2010, p. 371) argue that using the L1 could give learners "timely assistance about word meaning" to ease L2 learning, while Filipi (2018) proposes that teacher language alternation practices are bound up with expectations that teachers have about what a student should understand in the respective languages.Switching to the L1 is therefore a practice that is launched to establish and facilitate understanding.The latter certainly seems to have relevance with respect to the teaching assistants' practices because they did not expect the students to have sufficient proficiency in English to understand the metalanguage necessary to explain grammatical errors.This finding certainly points to the need for further studies that examine language alternation and the distribution of each language in delivering feedback in an EFL tertiary context.

Final Drafts: Lecturer's Feedback
Lecturer written feedback is recorded in the last phase of the trialled feedback approach.The analysis has shown that the lecturers' written feedback focused on providing general comments in addition to allocating a final numerical mark on the students' final drafts.Analysis also showed that in addition to a total mark, the lecturers provided a numerical mark for each criterion; i.e. ideas, organization, vocabulary and grammatical range (see Appendix B).Research has shown that marks can help to confirm the writers' level and foster the students' motivation in their subsequent writing of papers (Ferris, 2006), promoting the sixth principle of AFL (fostering motivation).
Despite the common assumption that the final drafts were the best of the three drafts, the lecturers reported having to spend time giving further explicit comments and making indirect error correction on student writing.One reason given for the need to do this, was their belief that the students had neither the skill nor the experience to provide feedback.Oanh's statement is illustrative: Due to the students' limited capacity in providing feedback, the teaching assistants' and lecturers' commitment to feedback was more noticeable.It seemed that the students and teachers spent equal amounts of time giving feedback, but the latter's commitment in providing feedback was more than that of the former.Some errors in the mechanics of language were sometimes found in the final drafts.
The above suggests a need to better prepare students for effective peer assessment.It also points to the ways in which existing gaps in student language knowledge, provided by rich assessment processes, can be used to inform teaching.
So far, we have been discussing data (derived from the classroom observations, the student drafts and interview data) to highlight the alignment of each feedback phase with the principles of AFL.Next, we turn to the participants' perspectives by examining the interview data.

Teachers' and Students' Perceptions of the Multiple-Draft/Multiple-Party Feedback Practices
Analysis of the interview data showed that participants held shared views about three strengths of the multiple-draft/multiple-party feedback approach: quality of the feedback, perceived improvement in students' writing and students' active participation in the feedback approach.A recurring weakness that was reported was the participants' skepticism about the accuracy of peer feedback.Discussion begins with an examination of the three perceived strengths.

Perceived Strengths
 Quality of the feedback Student participants appreciated the opportunity to receive three sources of feedback at three stages of their writing.Teacher participants expressed their satisfaction with the process of providing feedback on the students' second and final drafts which had been substantially revised by the students themselves after the peer review.
Ten student participants agreed with Duong's view below while three teacher participants agreed with Yen about feedback quality.
We were very interested in this approach because we could receive feedback for two drafts before writing the final draft to receive the lecturer's comments and marks.Another interesting thing was that there was the teaching assistant's feedback.(Student Duong) I realised that this practice was totally new compared to previous feedback approaches.The teaching assistant worked very effectively in the second draft correction, so it was easier for me to give feedback on the final drafts.(Lecturer Yen) The participants' reports about the quality of multiple-draft/multiple-party correction revealed that this practice activated the first, third and sixth principles of AFL: being part of effective planning, being central to classroom practice and fostering motivation.Notably, group feedback and the teaching assistant's oral feedback occurred during class time and did not disrupt the normal lesson.It involved effective planning for teaching and learning through: the specific staging of processes and planning of three separate phases, the organization and grouping of different participants, and provision of checklists and appropriate timing.Furthermore, the first draft feedback activity was learner-centred and afforded the students opportunities to participate actively in feedback practices, and to receive informed, valued and targeted feedback on more than one draft and from various sources.The latter concurs with the recommendations for best feedback practices put forward by Bitchener and Ferris (2012, p. 134) who suggest that instructors are generally believed to "have the required knowledge" and "sufficient training and experience to know what works best for individual learners".Similarly, the lecturers believed that the multiple-draft approach was an advantage for them since their task of providing feedback on the final drafts was considerably lessened as illustrated in Yen's statement above.
Additionally, findings about feedback quality in this feedback practice was documented in the perceived growth of the teacher's professional development which aligns with the fourth principle that characterizes AFL as being a key professional skill.In the words of Hoa, the teaching assistant: I realised that I was more confident in providing feedback on the students' writing in this approach.Although I had to spend much time giving feedback, I felt happy in locating and correcting errors on the students' drafts.I think it was a time-consuming task, but it was really rewarding and effective.
Hoa's satisfaction with the process accords with the belief that in AFL "teachers come to enjoy their work more and to find it more satisfying because it resonates with their professional values" (Black et al., 2007, p.19).
The above findings about feedback quality in this multiple-draft/multiple-party feedback approach adds to the literature on the positive impacts of multiple-draft feedback in L2 writing (Ekşi, 2012;Ferris, 2014;Lee, 2011aLee, , 2011cLee, , 2011d;;Paulus, 1999;Yang et al., 2006).This research highlights various patterns of teachers' feedback on different drafts in ESL contexts or in EFL secondary education.However, missing from this research is the EFL higher education context especially the Vietnamese one.As far as we are aware, only two empirical studies have proposed the application of a two-pronged approach to feedback: paired peer feedback on first draft correction followed by teachers' feedback on the final drafts (Ekşi, 2012;Paulus, 1999).The current study extends this to a three-pronged feedback approach (Filipi, 2017) and supports Bitchener and Ferris' (2012) recommendation about the need for and usefulness of various instructors' feedback in L2 writing by extending it to include teacher assistant feedback.It thus draws attention to the fact that multiple-party feedback approaches have been under-explored in L2 writing classrooms.

 Students' writing improvement
The next strength of the multiple-draft/multiple-party feedback approach identified by the participants was improvement in the students' writing.As seen in the table below, a greater number of participants expressed the view that language use (such as vocabulary and grammatical structures) improved while fewer participants believed that improvements to content (including organization, and generating, expressing and developing ideas) accrued.The last principle of AFL speaks to the need to "recognize the full range of achievements" (ARG, 2002, p.2). Within the writing task in the course under study, this includes both the mechanics of writing and effective ideas for development of topic.More than half of student participants believed that they achieved progress in both control of language and content in their writing.The following excerpt illustrates the students' perspectives on various areas in which they perceived themselves to have improved.
I realise that I've had more ideas for my writing.Before, I had often been stuck in developing my ideas, so it took a long time to complete a paper.Also, I've made some progress in my vocabulary such as word use, and in combining simple sentences into compound, complex ones.I have been aware of using academic words and of avoiding informal words in my writing.I have made more progress than in previous semesters.

(Hang)
Further examples about the students' advances in control of language relative to the development of main ideas to strengthen their arguments are illustrated in Appendix C.
All the teacher participants believed that the students had improved their capacity to use academic vocabulary and sentence structures.The following is one typical statement to illustrate: I realised that in this semester the students have made much more progress in language use, especially in the essays about advantages/disadvantages and about problems/solutions.They could use a variety of academic words and sentence structures such as inversion.The students used many interesting words as well.(Lecturer Yen) However, only the lecturer Oanh reported the students' progress in developing ideas and organization of the essay: The students have made some progress in the following aspects.First, they were aware about the specific audience of their writing, so they were always careful in planning their writing such as how to write, how to develop an idea.Second, the students have made so much progress in organizing an essay.Almost 100% of the students in my class were aware that they had to produce a complete essay (including the introduction, the body and the conclusion in 40 minutes).They were also aware about how their writing papers were assessed such as how well they addressed the task and how well they expressed their ideas.Their vocabulary and grammar were also better than before.
These findings are consistent with prior research on the effects of multiple draft correction in L2 writing (Ferris, 2006;Paulus, 1999) which report that students made more gains in their language accuracy than in content.On a related point, Black et al. (2007, p.19) argued that in formative assessment "the changes can be made step by step -a big "leap in the dark" is not necessary", so the above aspects of the students' improvements were an encouraging sign for the efficacy of the multiple-draft/multiple-party feedback practices in student writing development that could be enhanced over time.

 Students' engagement in feedback practices
The final perceived strength of the multiple-draft/multiple-party feedback was the students' active participation in this process, evident in their engagement during class and the development of their learner autonomy.
Nine students and all four teachers expressed similar viewpoints in the interviews about students taking the task seriously and their interest in the feedback activities.The following excerpts are representative of the students' perceptions: In my group, we were so serious in doing group feedback.Every member was aware of the need to give feedback on both strengths and weaknesses to their peers.We took it seriously not because of the teachers' request, but because we all thought that the others were giving serious feedback on our drafts, so we should do it as well as possible.(Phan) The students were very serious in doing group feedback.They even brought materials and dictionaries with them to refer to when giving feedback to their peers.(Hai) Group feedback was very effective because it created a comfortable atmosphere in writing classes.(Dieu) Student perceptions also suggest that group feedback helped to increase their sense of comfort and respect for their peers during their engagement which in turn promoted the students' serious participation in this activity.The result supports the principle of creating "a comfortable environment for students" for effective peer feedback as suggested by K. Hyland (2003).
The students' opinions about their active engagement in this multiple-draft feedback practice were consistent with the teacher participants' perspectives.All of the teacher participants praised the students' positive attitudes.For example, Hoa said: In this feedback practice, the students understood the criteria about the qualities of a well written paper and about how to give feedback on their peers' writing.They were also aware of what they were doing and what they could improve after being involved in group feedback, so they were serious in participating in this activity.This contrasts with the previous semester when the students had been required to produce only one draft; they didn't know how to improve their writing upon receiving a mark.They didn't make any effort to revise their writing.In this practice, the students were better motivated than before… They wanted to know about the reasons for their errors, and they asked immediately about what to do for improvement.In general, they were very active in asking questions.
Additionally, students' active participation in this feedback approach could also be found in their own awareness of the need to develop their autonomy to improve their writing.For instance, Bich said: When revising my drafts, I myself found a proper way to improve my writing.Sometimes I asked for help from my friends based on the teachers' suggestions to find out the most appropriate way to improve my writing.
Hong spoke of her awareness of cultivating her autonomy: Before I didn't have a habit of noting down my strengths and weaknesses when I received feedback.But in this feedback practice, I pushed myself to do this, to find out what was useful for me later.This was the first time I had done this task.
In summary, the participants' perceptions about the strengths and pedagogical benefits of the multiple-draft/multiple-party feedback showed how they were enacting the AFL principles as listed above with particular reference to principles 1, 2 and 8 (i.e.being part of effective planning, focusing on how students learn and helping learners know how to improve), highlighting the benefits of interaction and the diagnostic nature of the feedback so that the learners could act on advice for their writing improvement.

Limitations
Participants highlighted one weakness of the multiple-draft/multiple-party feedback approach: concerns about the accuracy and lack of confidence in their peers' feedback.Ten students and three teachers expressed this view.Lan's comment illustrates: It was clear that in peer responses some didn't know how to correct errors despite the guidelines in the checklist.We didn't trust much of the group feedback because we were at the same level and couldn't be sure that our friends were better than us.We still waited for the teaching assistant's feedback.
Students also maintained that their own limited knowledge impacted negatively on the effectiveness of peer feedback which may have led to the students' distrust of their peers' feedback.
Group feedback was only partially effective because our knowledge about writing was not good enough.Many of us were not able to give feedback on some items in the criteria, especially ideas and vocabulary.I could only give feedback on simple vocabulary errors; there were errors which I couldn't locate or suggest for improvement.(Le) The teacher participants' views accorded with those of the students.Oanh, for example, confirmed the students' lack of confidence in their peers' feedback: I realised that there were some students who were not confident in giving feedback on their friends' writing and some students who didn't trust their friends' feedback.Therefore, there was some feedback focusing on very simple errors like spelling mistakes.In general, there was an imbalance in giving feedback on the four criteria.
Thu, the teaching assistant, added that the students' peer feedback was limited as it focused on isolating errors.
I realised that most of the students' feedback on the first drafts could not be used in my process of giving feedback.I usually recorrected many of the errors on the second drafts.Some students could locate errors but couldn't give any suggestions, especially about expression of ideas.
We can conclude that, notwithstanding the benefits of peer feedback, the students nonetheless expressed a distrust of their own and their peers' advice, most likely because of perceived or actual insufficient English language proficiency or English language knowledge.A number of suggestions to overcome such a potential obstacle to what has been considered to be an otherwise highly valued and beneficial experience in peer feedback will be provided next in the conclusion.

Conclusion
This study sought to show how a multiple-draft/multiple-party feedback approach (which encompassed peer/group feedback on the first drafts, teaching assistant feedback on the second drafts and lecturer feedback on the final drafts), aligned with the set of 10 principles that characterize AFL as proposed by the Assessment Reform Group (2002), and to understand the participating teachers' and students' perceived benefits of the approach.Using content analysis, interpretation of the data showed that the approach to feedback aligned with principles 1 to 9 and contributed to the tenth principle.Analysis also uncovered a process approach to writing in which students were given agency to improve their writing in collaborative, student-centred ways.The "three-pronged", multiple party feedback (Filipi, 2017) approach developed is an extension of the two-party feedback practice examined in previous research about multiple-draft feedback in English writing classes.It supports Bitchener and Ferris' (2012) recommendation about the need for and the usefulness of various instructors' feedback in L2 writing.
The findings about teachers and learners' perceptions about the multiple-draft/multiple-party feedback approach revealed that the strengths and benefits of the approach greatly outweighed the limitations.While the limitations identified are serious, they can nonetheless be overcome through attention to organization and careful training.To this end we propose following: -First, the rotation of members' roles in group/peer feedback can occur every four weeks over the academic year instead of the same role throughout the semester.This means students will have opportunities to provide feedback in all of the four criteria in the checklist so that their knowledge about a wide range of linguistic elements will be then enhanced.
-Second, the grouping of students and the organisation of groups needs to be considered carefully.Each group should have at least one student with a high level of English proficiency who can assist the others as needed.
As claimed by Pham (2013), groups of high-and low-achieving students could exert positive impacts on assessment as a group project.
-Third, there should be frequent training sessions to ensure peer feedback accuracy.The training sessions should be organized by writing instructors at least twice per writing semester.Explicit instructions about, and modelling of peer feedback for each of the criteria in the checklist, should be illustrated and worked through by instructors and students through interaction and in collaborative ways.
This qualitative case study is obviously limited by its small sample size.However, the aim was not to make generalizations based on the findings.Rather, through the detailed, "thick" description (Denzin, 1989) of the findings, the study extends the literature about teachers' and students' perceptions about feedback practices in general and the multiple-draft/multiple-party feedback approach in particular.Additionally, the study attempts to shed light on the potential of AFL for developing EFL university learners' writing skills through feedback practices that incorporate a student-centred component.The use of ten AFL principles (ARG, 2002) was particularly useful as a guide for analysing and understanding the practices around feedback and for implementing change in practice.Further research is needed to explore how multiple-draft/multiple-party feedback practices develop students' metacognition and self-regulation, an important aspect of AFL, and how the L1 and L2 are implicated in this work.

Table 3 .
Participants' perspectives about the students' writing improvement