A CASE STUDY OF COMBINED PEER-TEACHER FEEDBACK ON PARAGRAPH WRITING AT A UNIVERSITY IN THAILAND Nguyen

Writing in English is challenging for ESL writers, so feedback is crucial in assisting them. Although several studies have been conducted on the effectiveness of peer and teacher-feedback in ESL writing, studies on the combined peer-teacher feedback model tend to be scarce. This study thus reported on the combined feedback model in two paragraph-writing classes of sixty students at a university in Thailand where English is taught as a foreign language, students are reported to be passive in class activities and most writing programs are still taught using the traditional method. Students’ peer comments (both valid and invalid ones), their revisions based on both their peers’ and teacher’s feedback (correct and incorrect revisions) and their grades on each paragraph were recorded, and a five-point Likert scale survey and a focus group interview were conducted. The findings indicated its success in terms of students’ positive attitudes towards this feedback model, the usefulness of peer comments, high percentages of feedback incorporations and the high overall writing scores. This paper is thus expected to shed some light on how Thai university students with their passive style of learning English positively react to this interactive activity and partly reflect how in-service teachers adjust feedback strategies in their actual teaching situations.

2009), and it was also claimed to be more informative than teacher-feedback since it is pitched more at students' level of development.It can thus contribute to their learning development and increase their motivation (Hyland & Hyland, 2006).Additionally, with the help of supportive peers, students' attitudes towards writing can be enhanced and their apprehension can be lowered.Moreover, being assigned the role of a teacher through peerreview, students are actively engaged in their own learning and assume responsibility of their own learning progress (Liu & Hansen, 2002).Finally, by reading each other's drafts critically students can learn more about writing and revision and at the same time they are able to identify the weak and strong points in their writing, and hence improve their writing proficiency and become autonomous learners (Hansen & Liu, 2005).
Despite the theoretical support and empirical evidence in support of its facilitative role in L2/FL writing, peer-feedback has not been widely used in L2/FL writing classrooms and the most favored type of feedback in L2/FL writing classes at universities is still teacher-feedback (Yu & Lee, 2016).This could be due to various issues associated with the use of peer-review, such as time constraint, teacher roles and student characteristics (Rollinson, 2005).Furthermore, Hu (2005) indicates that students' limited knowledge of the target language and its rhetorical conventions, the "surface" nature of students' comments, and students' various inappropriate attitudes towards peer-review are likely to hinder the implementation of peer-feedback in L2/EFL writing classes.Moreover, Zhang (1995) states that cultural background was presumed to render peer-feedback ineffective, especially for Asian students who were used to teacher-dominated pedagogies and preferred to incorporate teacherfeedback because the teacher was considered as the expert and the only source of authority (Carson & Nelson, 1996).Fei (2006), for example, found that her Chinese students felt doubtful about the quality of peer suggestions, hesitated to use peer comments in revision, and had very negative perceptions of the helpfulness of peer-review.Similarly, a couple of studies conducted on peer-feedback with Thai university students also indicated that peer-review was not appreciated (Chamcharatsri, 2010) and they preferred teacher-feedback (Srichanyachon, 2011(Srichanyachon, , 2012)).However, as stated by previous researchers (Min, 2006;Paulus, 1999;Rollinson, 2005), the effectiveness of peer-feedback largely depends upon the way in which it is implemented in the writing classroom.Berg, Admiral, and Pilot (2006) outline several optimal design features for peer-feedback to be successful, including a manageable length requirement (five to eight pages) and sufficient time for the review task.However, reported research on trained peer-response following their suggestions tends to be scarce in the literature, so one of the aims of the current study is to fill in this gap.

Combined peer-teacher feedback
Recently, a combined use of teacher-and peerfeedback was also found to be welcomed by students in L2/FL settings (Demirel & Enginarlar, 2016;Maarof et al., 2011).In particular, Maarof et al. (2011) state that "teacher-feedback can assist learners to notice a target structure, to compare it with their existing knowledge and to integrate it into that knowledge.Peer-feedback, on the other hand, can also help learners to notice the target structure while reconfirming its use and providing additional input via the learners ' input." (p. 33) These two forms of feedback should be therefore best seen as complementary (Zamel, 1985), and combining them systematically could thus provide students additional benefits, such as decreasing writing anxiety, improving writing ability and being more confident in their abilities to make decisions about their own writing and revision choices (Paulus, 1999).Furthermore, Demirel and Enginarlar (2016) also found that the combined peer-teacher feedback model helped students make useful revision in form, content and organization, resulting in an increase in their writing scores.
Writing in English poses several challenges for L2/FL students as they have to get used to new conventions of writing other than their own culture's, express themselves in a new language and cope with the multifaceted nature of writing.These challenges make writing one of the most difficult skills to develop for students and cause an overreliance on the teacher for all kinds of corrections and guidance.Such overdependence on teachers tends to be much greater for writing teachers in Thailand, where English has been taught as a foreign language and as a separate subject rather than being used as the medium of communication for decades.Although English is a compulsory subject for Thai students from primary to tertiary levels, it is taught more in Thai than in English (Bennui, 2008).Thai university students' English proficiency is reported to be less than satisfactory (Boonpattanaporn, 2008;Komin, 1998), and their writing is of particular concern as writing is not systematically taught as a subject (Chamcharatsri, 2010;Puengpipattrakul, 2013;Srichanyachon, 2011;Wongsothorn, 1994).Furthermore, most writing programs are still taught using the traditional model, emphasizing the accuracy of grammatical structures and vocabulary (Chamcharatsri, 2010;McDonough, 2004;Siriphan, 1988), and the formative tests in most writing programs stress objective-type questions, which require sentence completion, reordering sentences, reordering words and error correction (Wongsothorn, 1994).Students thus have very few actual opportunities to represent their ideas and knowledge through the written mode.Moreover, in Thai educational contexts, students have not been required to engage actively in class activities, so working cooperatively leads to students' resistance and confusion (Kongpetch, 2006;McDonough, 2004).This passive learning style could be partly due to Thai traditional belief of "silence denotes wisdom" and Thai cultural constraint about the need to avoid criticism (Puengpipattrakul, 2013;Root, 2016).Such practices in writing classrooms in Thailand are likely to make it difficult for Thai students to develop their writing abilities.
The need for improvement of Thai students' writing skills and the possibility of a contribution of feedback to such an improvement made it necessary to develop a working model of feedback to be used in writing classes in Thailand.Rather than using peer-feedback occasionally, this study made feedback a natural component of the paragraph writing class in a structured way.A combined feedback model in which students and teacher commented on the same writing was therefore developed, implemented and evaluated in the present study.The main aim of the study was to examine the effectiveness of the combined use of peer-teacher feedback in paragraph writing classes at a university in Thailand.Hence, the research questions posited for the study were 1) Is the combined feedback model successful in the educational setting in Thailand?, 2) What are the students' attitudes to this feedback model?, 3) Is students' feedback useful and to what extent students incorporate peer-feedback into their revision as compared with the teacher's?, and 4) How much do they benefit from this interactive activity in terms of their overall improvement?.The results of this study are expected to shed some light on how Thai university students with a reported passive style of learning English react to this new interactive learning activity and partly reflect how in-service teachers adjust feedback strategies in their actual teaching situations.

METHOD Context and participants
The English curriculum at this university has three obligatory writing courses, namely Writing 1 (paragraph writing), Writing 2 (short compositions) and Writing 3 (five-paragraph academic essays), and each of which is taught in three successive terms of fourteen weeks each, starting from their third year of study.Besides two obligatory English grammar courses mainly taught in Thai by a Thai teacher of English, what English-majored students at this university studied in their first two years are general subjects in Thai language, and English communication courses 1 and 2 are also considered as their general subjects.
Combined peer-teacher feedback was conducted in the Writing 1 course with third-year English-majored students who met once a week for a 14-week semester with 150 minutes each, using the selected course book (Writers at Work-From Sentence to Paragraph by Laurie Blass and Deborah Gordon, 1 st Edition, 2010).This book consists of ten chapters with ten different writing topics, and the target vocabulary and grammatical points for each topic are also presented in each chapter.Although the objective of this course is to help students develop their skill in writing an academic paragraph, very little information about paragraph writing is given in this book.That is why the chair of English division at this university supported the teacher's innovation in her writing course for the improvement of students' writing abilities.This year, the researcher taught this course to two intact classes of 60 students (32 and 28 students each) who were at the age of 20 and 21 (for the ease of reference, G1 refers to the group with 32 students while the other is G2).These students' English proficiency level was around upper-elementary or pre-intermediate, and this batch consisted of a few male students (four in each group).The score for this course includes 5% of their class-attendance, 45% of assignments allocated by the teacher, and the other 50% is from midterm and final tests (20% and 30%, respectively).

Procedures
In order to both meet the course requirements and implement the feedback activities in this Writing 1 course, the author revised its curriculum instead of teaching the book chapter by chapter.In the first five weeks of the course, a genre-based approach was employed to teach the students the generic structure of an academic paragraph.During this time, students were familiarized with the basic components of an academic paragraph (topic sentence, supporting sentences and concluding sentence) and how to compose each through stepby-step instructions as well as thorough practice with the materials developed by the researcher.From weeks 6 to 14, the students were asked to write seven complete paragraphs of 150 words each for seven topics chosen from the course book ( i.e. 1) "All about me", 2) Daily activities, 3) Your family, 4) Your favorite book/movie/TV show (choose 1), 5) Your idol, 6) Your future plans and 7) Your memorable trip) at home, and feedback activities were implemented in class.While the first two writings (W1 and W2) on topics 1 and 2 were employed in the training stage (weeks 6-8), the other five were graded for 45%.However, to see whether or not grading the overall writing influenced the effectiveness of this activity, only three of G1's writings (W3, W5, and W7) on topics 3, 5 and 7 were marked (15% each), and 9% was given to each of all G2' five writings.Following Rollison's (2005) suggestions, in the setting-up stage the teacher decided to have students write three drafts, work in groups of four, selected by themselves but encouraged to work with different peers over the course, and use indirect written feedback (using provided correction-symbols to indicate the mistakes instead of providing corrections).The first draft was checked by their peers and the writer, first independently and then in a consensus group for clarifications and suggestions for revision, using the responding guidelines (Appendix A), and their second and third drafts were checked by the teacher using the same guidelines and correction-symbols.Their first language was employed in this interactive activity.When they submitted their second and third drafts, a summary of their responses including explanations for their choice of not incorporating any suggested comment was required (Appendix B).Peers' comments and writers' revisions were graded with a deduction of 1% from their obtained assignment score (45%) for irresponsible comments and ignoring the given feedback.In three-week training, class discussion on the benefits of peerfeedback and appropriate attitudes in peer-feedback activities and non-threatening practice on Topics 1 and 2 were conducted with the above-set-up criteria.Generally, the class procedures in the last nine weeks were (a) follow-up activities (returning students' last assignment, asking them to read the comments and ask friends or teacher for suggestions or clarifications of coded errors, summarizing commonly-made mistakes, and explaining the comments to those who asked for help) (45 minutes), (b) peer-feedback on the new writing (1 hour), and (c) lessons in the book and preparation for the following writing topic (45 minutes).Because the considerable class time was dedicated to peer-feedback activities, students were asked to check new vocabulary and do grammatical exercises in the course book at home, and in the last 45 minutes of every class they were corrected and ideas for their following writing topic were also discussed.
To understand these students' attitudes towards this new learning activity, a five-point Likert scale survey (strongly disagree = 1, disagree = 2, not sure = 3, agree = 4, strongly agree = 5) and a focus-group interview with eighteen volunteering students from both groups (9 each) were conducted.Furthermore, to learn about the usefulness of peer-feedback and the employment of provided feedback in the revised drafts, peer comments (both valid and invalid ones) (the former refers to the accurate and useful feedback while the inaccurate and useless comments are considered invalid) and how they incorporated peer-and teacher-feedback (correct and incorrect revisions) were recorded.Their writing scores were also examined to evaluate their overall improvement.The grading process was conducted independently by the researcher and an inter-rater, and the reliability of the two raters was assessed by using Pearson's Product Moment Correlation Coefficient.The correlation values between the two raters for G2's five writings (W3, W4, W5, W6, and W7) were 0.82, 0.91, 0.84, 0.81 and 0.89, respectively while 0.81, 0.85 and 0.87 were for G1's W3, W5 and W7, respectively.Discussions between the two raters on grading disagreements were finally conducted until the agreement was reached.

FINDINGS
Although the answers to research questions 2, 3, and 4 are presented in this section, their detailed discussion is provided in the Discussion section altogether with the answer to research question 1 on the effectiveness of this combined feedback model in the educational context in Thailand.

Student attitudes towards feedback activities
To learn about these students' opinions on this interactive activity, the 5-point Likert survey was administered at the end of the semester.As suggested by Sullivan and Artino (2013) that a mean score is not a very helpful measure of central tendency of Likert-scale data, the percentages of students' positive (strongly agree and agree), neutral and negative (strongly disagree and disagree) attitudes are also included in Table 1.
As suggested by Wiboolsri (2008) for the mean score of 3.5 as the acceptable value representing a positive attitude, it can be concluded that these students were very positive towards this activity as means of most surveyed items are much higher than 3.5, except for negatively-worded items (8, 11, 12, 14, 18 and 21).However, by giving low mean scores to the negative items, these students showed their active participation (Items 14 and 18) and their interest in and acknowledgement of the usefulness of this activity (Item 8).This finding was reaffirmed with the very high means and percentages of their positive attitudes in Items 4, 5, 6, 7, 9, 13 and 17.For Items 11, 12 and 21, these students revealed that they valued teacher-feedback more highly than their peers' (Items 11 and 12) although they expected to have feedback from both (Items 21 and 22).Furthermore, although it was their first time to experience this interactive activity (Item 1), all Thai students in this study did not show their resistance as reported by previous research with students in the deeply rooted teacher-centered pedagogies (Carson & Nelson, 1996;Chamcharatsri, 2010;Fei, 2006).As claimed by previous scholars (Min, 2006;Rahimi, 2013;Rollinson, 2005), the sufficient training (Items 2 and 3) and the way this activity was implemented in class (Items 19, 0, 23 and 24) could be deemed for this success.

Students' feedback usefulness
To learn about the usefulness of peers' comments, both students' valid and invalid feedback on their friends' first draft was recorded.As seen in Table 2, most feedback on all writing aspects (format, organization-format and language-mechanics) provided by peers was accurate (95% and 96% for G1 and G2, respectively).The high percentages of valid comments by students in this study are in line with those in the literature (Crookes, Davis, & Caulk, 1994).Furthermore, because the response guidelines were provided (Appendix A), these students commented on both global (contentorganization) and surface (language-mechanics) levels although their focus was centered more on the latter, which was similar to Chinese students in Hu (2005).Despite a very small percentage, selfdiscovered feedback was present though most was on content and organization, like adding, deleting or rearranging details, but all was found to be valid in this study.By reviewing peers' writing and rereading their own, these students identified their own errors and made changes to their writing for the better (Min, 2006)

Comparison of peer and teacher-feedback incorporations
As can be seen in Table 3, all feedback provided by the teacher was incorporated into their third drafts while an extremely high proportion of their first revisions made were incorporations of peer comments (99.7% and 94.8% for G1 and G2, respectively).Although G1 incorporated almost all peer-feedback into their second drafts (99.7%), their invalid incorporations were marginally higher than that of G2 (42.5% versus 40%, respectively).
Regarding revisions on the global level, the higher percentages of damaging revisions were also found from both groups, and G1 students tended to make more invalid incorporations from both peer and teacher suggestions.In terms of language-mechanics teacher-feedback was likely to be incorporated more successfully.In general, these students tended to make slightly more relevant revisions from teacherfeedback (62% and 68% as compared with 57.6% and 59% of that provided by peers for G1 and G2, respectively).

Overall improvements
Table 4 shows the high average writing scores (83% and 81.4% for G1 and G2, respectively) for their 45% assignments (W3-W5-W7 for G1 and W3-W4-W5-W6-W7 for G2).While students in G1 showed their gradual improvements, G2 students' scores were fluctuating during the course, resulting in their lower overall grades.However, the interview with G2 students revealed that the difficulty of writing topics (4) (Your favorite book/movie/TV show) and ( 6) (Your future plans) resulted in their poor ideas and hence lower grades.This is likely to suggest that writing topics might influence on students' writing scores.A closer look at the average errors and the incorporations of feedback from both peers and teacher in W4 and W6 for which G1 was not given grades (Table 5) showed the potential effect of grading on this activity.While both groups incorporated all teacher-feedback into their revisions (100%), G2 students tended to work harder in these two writings with more errors identified by peers and higher percentages of peer-feedback incorporations.This finding displayed a slight contrast to G1 students' general trend of intensively employing provided feedback into their revisions as shown in Table 3 (99.7%as compared with 94.8% for G2).Furthermore, a few instances of selfdiscovered errors and corrections were also found in G2 students' second drafts of W4 and W6, representing 100.5% in Table 5.Therefore, grading tends to influence these Thai students' attention and engagement in the provisions and incorporations of feedback to some extent.

DISCUSSION
The answers to research questions 2, 3 and 4 in this study tended to show the effectiveness of the combined feedback model for this group of university students in Thailand where students were reported to be passive in learning English and teacher-centered pedagogies exist.Despite their big class, their low level of English proficiency, their inexperience with group work and their culturally reported "passive" learning styles (Kongpetch, 2006;McDonough, 2004;Puengpipattrakul, 2013;Root, 2016), the students in this study showed their great interest in working with their peers, their satisfaction with their peers' comments, their intensive incorporations of feedback from both peers and the teacher into their revisions, and most importantly the improvements in their writing in English.Furthermore, as seen in the questionnaire (Item 22), these students wanted to have feedback from both the teacher and peers, and as revealed in the interview, it was known that the former assisted them with language while the latter helped them with ideas.These results are thus different from those of previous studies (Fei, 2006;Nelson & Carson, 1998;Tsui & Ng, 2000) which found the resistance of students with entrenched teachercentered learning experiences to peer-feedback.The success of this study, however, could be due to many factors, such as (1) the step-by-step training on peer response, (2) the way this activity was implemented in class, (3) grading and (4) the school's permission to revise the curriculum.As stated by previous researchers (Min, 2006;Rahimi, 2013;Rollinson, 2005), sufficient training on peer-feedback would lead to its effectiveness as students understand the rationale behind this activity and how to do the task.The information in the questionnaire also revealed these students' acknowledging the benefits of training in doing this task (Items 2 and 3, Table 1).In fact, understanding the rationale of this new practice is essential for these Thai university students whose cultural norms 'may be antithetical to the pedagogical principles' of peer-feedback (Hu, 2005, p. 332), and with such understanding their positive attitude towards working with others as a fruitful way of acquiring the language tended to be cultivated.Moreover, as revealed in the focus-group interview with these students, having them comment first, then their teacher's reviewing the following drafts and followups provided in each class not only reduced their pressure in doing the task but also maintained their enthusiasm in this activity.It was because they knew that their writing and comments would be checked and backed up by the teacher, and this process gave them a chance to think over their initial work and develop it in subsequent drafts.Hence, their language knowledge was gradually consolidated and updated.In fact, such consistent feedback provisions and follow-ups helped them review their language used and gradually build up their confidence in English writing since these students have hardly had chances to write in English.Another possible explanation for the effectiveness of the feedback activity in this study is the grading of their comments and their paragraphs.As for these students who will become English teachers, gaining good scores in all subjects is what they aimed at in order to secure a job in the future.In fact, in order to improve the country general education, Thailand has given teacher-students with a GPA of 3.0 and higher some favorable conditions after their graduation.The interview also revealed that grading the comments made these students more responsible in giving feedback and revising their writing, and this information was triangulated with the examination of W4 and W6 in which G1 was not given grades.Additionally, some even showed their dislike for being marked five times as they believed that the more times the teacher checked their writing, the higher probability that the teacher found their weakness, which could result in their lower scores.Last but not least, for the effectiveness of any innovative pedagogy it is imperative to have the supportive environment from the school as well as other community members (Hyland & Wong, 2013).In fact, the success of this study was because the researcher was granted the right to adjust the curriculum for enhanced student writing.Lee, Mak, and Burns (2016) also state that despite the teachers' relevant subject knowledge, their attempts will be impeded by the unsupportive environment of their school.School leaders therefore need to be sufficiently open-minded to allow for the bending of the rules of the system, where appropriate and necessary, as change does not occur at the individual level, but supportive or stimulating conditions are necessary to foster real change in practice (Fullan, 2007).
Besides these main contributing factors to the success of this combined feedback activity at a university in Thailand, some considerations need to be taken into considerations in applying this model with L2/FL low-level students at a similar educational setting.First, because of their low level of English proficiency, students' mother tongue (L1) should be employed in peer and teacher interactions to assist them in understanding and being understood (Lantolf & Thorne, 2006).Second, written-feedback tended to be effective in this interactive activity with students whose daily exposure to English communication is limited, L1 interference is deeply-rooted and pronunciation, intonation and vocabulary was in need of improvement.Indeed, listening to peers' reading their writing, attending to both global and local errors and at the same time providing oral comments if oral feedback was conducted would be strikingly challenging and consequently could create a disastrous confusion in these students.In addition to the written mode, the use of indirect or codedfeedback was believed to help facilitate these students' writing development than the direct ones.Because low-level students are still at their developmental stage of learning English, providing them with corrected forms would not produce the reflection and cognitive engagement that helps them acquire linguistic structures and reduce errors over time.As stated by Ferris and Roberts (2001), consistently marking the error types, paired with mini-lessons which build students' knowledge, would yield more long-term growth in student accuracy.Furthermore, to maximize the benefits of indirect-written-feedback and to engage these students into the activity, the required responsesummary for their revised drafts tended to serve as a useful tool.Moreover, the teacher's quick returning their writing with feedback (on weekly basis) was believed to maintain their interest and their enthusiasm in writing as what they wrote and revised in the previous draft was still fresh in their mind.Receiving the teacher's feedback timely, their knowledge was consolidated in an uninterrupted manner.However, with the great effort and time required to check students' comments and provide further feedback (two times for each writing, and with a large class in this study), the teacher's time devotion, strong commitment, and patience was absolutely necessary for the success of this activity.

CONCLUSIONS
To sum up, this combined feedback model was successful in paragraph writing classes at a university in Thailand in terms of the usefulness of peer comments, students' positive attitudes, high percentages of feedback incorporations and the high overall scores.This success tended to confirm Rollinson's (2005) statement that "if the class is adequately set up and trained can the benefits of the feedback activity be fully realized" (p.29).In fact, despite their reported passive-learning styles and entrenched teacher-centered pedagogies, these Thai university students showed their active participation and positive engagement in this interactive learning activity.The combined peer-teacher feedback is indeed a time-consuming process, but its benefits were undeniable and applicable to these Thai university students.Although students' cultural backgrounds and the target language levels have been claimed to render peer-feedback ineffective, the success of this model tended to result mainly from sufficient training with the teacher's adequate awareness of contextual differences and her students' own characteristics, and then adjusting feedback strategies accordingly.It is generally accepted that students' reviewing peers' writing makes them cognizant of the assessment criteria and the requirements of the writing, and in the long run helps them become more self-reliant writers who have the skill to self-edit and revise their writing (Lee et al., 2016;Rollinson, 2005).However, feedback from various sources makes a positive contribution to their approach to writing by transforming the writing class from being an extension of a grammar course where language structures are practiced to a platform where they share, discuss and develop their ideas and motivates them to make revisions to improve their writing skills (Demirel & Enginarlar, 2016).

Table 3 .
Comparison of students' incorporations of peer and teacher-feedback

Table 4 .
Writing scores

Table 5 .
Average errors and incorporations of provided feedback in W4 and W6