Effect of teacher and peer written corrective feedback on writing components in EFL classrooms

This study aimed to investigate the impact of teacher and peer written corrective feedback (WCF) on Indonesian senior high school students' writing performance. A total of 71 Indonesian senior high school students from Grade X participated in this study; 36 were provided teacher WCF and 35 peers WCF. To collect pre and posttest data, the participants were asked to write a legend essay. Using qualitative data analysis, we aimed to reveal the effectiveness of teacher and peer WCF in improving students' writing performance. Adapted scoring rubric was employed to measure students' overall writing performance, and competencies in relation to writing components such as content, organization, grammar, vocabulary, and mechanics. The results revealed that peer WCF can better enhance students' writing abilities compared to teacher WCF. Furthermore, students who received teacher WCF showed substantial improvement in performance relating to all writing components except mechanics. In contrast, peer WCF enhanced students' organization and vocabulary related performance.


INTRODUCTION
In Indonesia, senior high school students have to master all four language skills (reading, writing, speaking, and listening) as prescribed in the Curriculum 2013 (K-13). Based on the curriculum, senior high school students in Grade X are required to learn different types of writing styles, such as recount, narrative, analytical exposition, expository, procedure, and news items (Kemendikbud, 2013). Though writing is considered a daunting skill to be mastered by students compared to other language skills, it is most important for educational success (Tillema, 2012). Ariyanti and Fitriana (2017) also confirm that writing is a challenging skill for Indonesian students to master. Nunan (2001) says that writing is the process of thinking to invent ideas and organize them into various written forms. Additionally, learning writing entails basic familiarity with higher-level subskills of planning and organizing and lower subskills of spelling, word choice, and mechanics (Richards and Renandya, 2002).
The focus of teaching writing is on students' cognitive development: how to write what they know. It also deals with their low motivation in writing. If the teaching-learning process lacks a variety of methodologies, it will affect students' writing achievement. Myles (2002) states that writing practice encourages the development of writing skills; consequently, students need lots of practice to improve their ability to compose written texts.
As a result, many techniques have been adopted to help Indonesian senior high school students improve their writing skills as learners of English as a foreign language (EFL). One such technique is to provide written corrective feedback (WCF) to students. Many studies have investigated the nature of WCF and its role, especially in second language (L2) and EFL contexts. There have been debates on whether WCF is beneficial for L2 and EFL students' writing ability (Ferris, 1999;Truscott, 1996). However, there are studies that show WCF could have a powerful impact on enhancing students' writing abilities (Hattie and Timperley, 2007).
To help students improve their writing abilities, providing teacher and peer WCF could be an alternative technique that can be adopted in the classroom. Therefore, this study aims to determine whether teacher and peer WCF can improve Indonesian senior high school students' performance in writing narrative texts, especially legends, and competencies in relation to writing components (content, organization, vocabulary, language, and mechanics). Bitchener and Storch (2016) define WCF as a written response to a linguistic error that has been made in the writing of a text by L2 students. WCF attempts to either fix the incorrect usage or provide information about where the error has occurred, the cause of the error, and how it may be corrected. According to Williams (2003), the goal of WCF is to teach skills that help students improve their writing proficiency and produce written texts with minimal errors and maximum clarity. Hattie and Timperley (2007) conceptualize WCF as information given by an agent (teacher, peer, parents, selfexperience) concerning aspects of one's writing performance. According to Saville-Troike and Barto (2016), WCF is a type of interaction that can enhance L2 acquisition by making nonnative speakers aware that their usage is not acceptable in some way, and provides a model for "correctness. "

Teacher Written Corrective Feedback
A teacher plays an essential role in contributing feedback to students, establishing realistic goals, and evaluating students' development during teaching and learning activities in a classroom. Teacher provides guidance and support to students to generate and apply evaluation criteria, reflect on their learning, set goals, and organize samples of their work in their portfolios. A teacher periodically evaluates students' work by making corrections, and giving ideas, suggestions, and feedback concerning the strengths and weaknesses of their work. According to Bitchener and Ferris (2012), teachers' WCF such as edge comments, request for clarification, and comments on grammatical issues has proven to help EFL writers. Students require their teachers to give feedback and feel unsatisfied when they do not receive comments about their writing. In addition, students also wish for their teachers to provide feedback on global issues (i.e., content, organization, and purposes) as well as local issues (i.e., sentence structure, word choice, and grammar) (Straub, 1997).
In her study on teacher WCF, Zamel (1985) found that teachers mostly concentrated on language errors in student writing. She believed that it is because writing is mainly seen as a product. As a result, teachers tend to regard themselves as language teachers rather than writing instructors. In particular, when the process approach is applied to teaching writing, many teachers shift their focus from form to other issues such as content and organization (Conrad and Goldstein, 1999). Additionally, Ferris (1997) found that 85% of teachers' comments focus on ideas and rhetorical development, while the rest discuss grammar and mechanics. It is advised that teachers give equal attention to writing components (content, structure, organization, language, and style) in their WCF (Ferris, 2003;Hyland and Hyland, 2006).

Peer Written Corrective Feedback
Peer WCF is one way to help students improve their writing by providing comments, corrections, opinions, ideas, and suggestions. It provides students with more opportunities to learn from each other. Peer WCF is a form of assessment made by learners of similar status. Peer feedback is not concerned with the assessor's final score and has a qualitative output. The assessor examines the strengths and weaknesses of a particular performance at length and recommends further enhancement (Gielen et al., 2010). Double et al. (2020) found that peer feedback, that is, assessment, evaluation, or grading by peers could improve students' academic performance. According to Elfiyanto (2020), peer assessment or peer feedback can enhance Indonesian students' ability to write narrative essays. In addition, he found that peer feedback can make students become more active during teaching and learning.
Prior research on teacher and peer WCF identifies the valuable impacts of peer feedback. Peterson (2013) emphasized that peer feedback benefits not only the students who receive suggestions for improving their writing but also the feedback providers. The students become more aware of good writing qualities by assessing and commenting on peers' writing. Furthermore, peer WCF also develops students' self-assessment abilities as they gain experience using the criteria to evaluate their own writing (Lundstrom and Baker, 2009). Tsui and Ng (2000) found that peer feedback convinces them to have a more prominent impact than teacher feedback.
Applying peer WCF in classroom is beneficial for many reasons. First, according to Gielen et al. (2010), peer WCF can increase social pressure on students to perform well on an assignment. Instead of the actual output of peer WCF, even an announcement that it will occur might boost students' performance. Second, higher education studies indicate that students frequently recognize peer WCF as understandable and helpful because they feel that their fellow students are in the same situation and condition. Lastly, peer WCF is faster to reach the students. As teacher WCF is often delayed after submitting a task or test, and sometimes is not provided until the topic has changed, imperfect feedback from a fellow student presented almost directly may have much more influence than perfect feedback from a tutor four weeks later (Gibbs and Simpson, 2004).

Legend as Genre
A legend is a very old story or set of stories from ancient times. The stories are not always true that people tell about a famous event or person. Legendary stories in Indonesia, such as Malin Kundang and Sangkuriang, are a tantalizing mix of realism and fantasy. Since the legend genre belongs to narrative text, Hartono (2005) states that a legend's social function is to amuse, entertain, and deal with the actual or vicarious experience in different ways. Legend deals with problematic events that lead to a crisis or turning point of some kind, which in turn finds a resolution. He also says that language features of a legend focus on specific participants and use past tense, temporal conjunctions, and temporal circumstances. Besides, the legend is one of the text types that need to be taught to senior high school students in Indonesia (Kemendikbud, 2013). Widiati et al. (2017) states that a legend's generic structure consists of orientation (gives the reader information about who was involved, what happened, where it happened, and when it happened) -complication (shows the beginning of the conflict) -a sequence of events (tells how the story develops after the conflict) -resolution (provides the solution to the conflict) -coda (explains how the characters have changed and what they learned from the experience or the moral message of the story). Therefore, learning how to compose a legend could help the students practice simple past tense and indirect speech, as they are needed to be mastered by the students. Also, the students could gain the moral messages of the story to be implemented in real life.
Conflicting findings on teacher and peer WCF indicate that existing research in this area has been inadequate. Therefore, it is meaningful to examine the effect of teacher and peer WCF on Indonesian senior high school students. In addition, this study takes Indonesian senior high school students' writing proficiency into account. The study results will provide other insights into enhancing Indonesian senior high school students' writing skills and improving their competencies related to writing components by applying WCF. With these goals in mind, the study attempted to answer the following questions: 1. Is teacher WCF more effective than peer WCF in improving Indonesian senior high school students essay writing skills?
2. What is the effect of teacher and peer WCF on Indonesian senior high school students' writing performance concerning components of writing competence (content, organization, vocabulary, language, and mechanics)?

Participants
This study was conducted at a public senior high school in East Java, Indonesia. Seventy-one senior high school students participated in this study; 36 students were provided teacher WCF and 35 students were assisted by peer WCF. The reason behind selecting Grade X students as participants was that they were learning how to write a narrative text (legend) in English class. First, a pretest was conducted to compare the two student groups in terms of language proficiency. The Shapiro-Wilk test found no significant differences between Teacher WCF group (0.310) and Peer WCF group (0.092), as both had p-value higher than 0.05. Furthermore, the pretest results showed that the total mean scores for both groups were equal. Teacher WCF group had mean 53.00, while the peer WCF group scored 54.26.
In Tribble (1996) scoring rubric, if a student scores 80 or above, he/she gets an A. One gets a B for scores between 60-79, C for scores between 40-59, D for scores between 20-39, and F for scores between 0-19. The pretest results of both groups were determined to belong to the C level.
Moreover, to ensure reliability, two scorers who were teachers from two different schools scored the pre and posttest papers separately. The analysis showed a high agreement rate (88%) between the two scorers.

Procedure
This study employed a quasi-experimental study approach of pretest treatment and posttest design. To answer the research questions, data related to students' written essays were gathered through pre and posttest. The two student groups received WCF from two different sources: One group received peer WCF and the other teacher WCF.
The data were collected over five meetings conducted during three weeks. In the first meeting, a pretest was conducted and participants were asked to compose a 100-word essay in 30 minutes. In the next three meetings, the treatment process was performed. Teachers trained the students on composing suitable compositions, especially a legend, and how to receive and provide WCF by explaining the usage of the feedback checklist. During the meetings, both student groups learned how to edit and revise their drafts. The difference between the two groups only differed in terms of the feedback source. It took 90 minutes per meeting to conduct the treatment process.
Finally, in the fifth meeting, a posttest was conducted. Students were asked to revise their drafts prepared during the pretest. For the teacher WCF group, students revised their drafts based on the feedback provided by their teacher. For the peer WCF group, the teacher distributed the pretests randomly among the students with a peer feedback checklist attached. The peer WCF process lasted for 20 minutes, after which the peers returned their friends' compositions along with their feedback. Students were given 15 minutes to read and understand the feedback provided by the two feedback sources, and were asked to revise their drafts in 30 minutes.

Data Collection and Analysis
The researchers applied statistical analysis methods to analyze quantitative data, including the comparison of teacher and peer WCF groups and systematic measurement of quantities, such as the numerical values assigned to participants' writing ability and its different components.
The pre and posttests were analyzed to prove whether there was a significant difference in the writing abilities of the students. Here, the writing components being scored were content, organization, vocabulary, language, and mechanics, before and after applying treatment in the classroom. The students' writing skills were considered as dependent variable, while the independent variable was the WCF sources (teacher and peer). The data analysis was conducted using SPSS Version 23 to measure the descriptive and inferential statistics. Descriptive statistics analyzed the pre and posttests by measuring the mean, standard deviation, and percentage for each component. Inferential statistics analyzed the writing components in both the pre and posttests by using independent sample tests.

FINDINGS AND DISCUSSION
To answer the first research question (RQ1) regarding the differences in the effectiveness between teacher and peer WCF in improving Indonesian senior high school students' essay writing skills, descriptive statistics analysis was conducted. Table 1 provides pre and posttest results for the two groups (teacher and peer).
Based on Table 1, the lowest score of the pretest for the teacher WCF group is 26, while that of the posttest is 34. Furthermore, the highest score of the pretest is 80 and that of the posttest is 81. For the peer WCF group, the pretest's lowest score is 25 and the highest is 80. For the posttest, the lowest score is 42 and the highest is 81. Additionally, the mean score of the teacher WCF group is 53.00 in the pretest and 57.64 in the posttest. For the peer WCF group, the man score is 54.26 in the pretest and 62.31 in the posttest. Though the highest scores for both groups are equal, there are differences in lowest and mean scores. Moreover, both groups make some improvement from pre to posttest. However, the posttest result of the peer WCF group is more significant in terms of improving students' performance in writing narrative essays, especially Legend, than the teacher WCF group.
This result corroborates with Hedgcock and Lefkowitz (1994) study which reported that peer feedback had more pos-itive effect on improving students' essay writing skills. The finding was also supported by Ruegg (2015) who said that peer WCF was useful in improving the global aspects of writing. Homayounzadeh et al. (2016) stated that peer WCF effectively enhances students' long-term acquisition of linguistic structures, writing accuracy, and recognition of grammatical errors.
The second research question (RQ2) investigated the effect of teacher and peer WCF on the writing components (content, organization, vocabulary, language, and mechanics) of Indonesian senior high school students. Here, the researchers compared the results of the posttest of teacher and peer WCF groups using an independent sample t-test. The results are shown inTable 2.

Teacher Written Corrective Feedback
The homogeneity test is performed to determine whether the sample groups from the population have similar variance. The data are homogeneous if the significance value (2-tailed) is higher than 0.05.
Based on Table 2, teacher WCF had significant impact on content, organization, vocabulary, and language, since the significance values are higher than 0.05. Thus, the data gathered from the pre and posttest are considered homogeneous. Furthermore, the significance values (2-tailed) are also higher than 0.05. Thus, it can be concluded that the data are normally distributed. Additionally, the mean difference between pre and posttest is 0.50 for content, 1.14 for organization, 1.67 for vocabulary, and 0.89 for language.
As for the writing component mechanics, Levene's test significance score is 0.003 which is less than 0.05. Consequently, the data obtained from the pre and posttest are considered in homogeneous. Furthermore, the significance value (2-tailed) is 0.008, which is also less than 0.05. Thus, it can be said that there is a significant difference in the average score between pre and posttest. Additionally, there is a mean difference of 1.14 between the pre and posttest.
It is thus concluded that teacher WCF increased students' competencies relating to four writing components: content, organization, vocabulary, and language. However, it did not have any effect on mechanics component. This result is supported by other studies (Biber et al., 2011;Paulus, 1999) which stated that teacher WCF led to an improvement in content writing. Furthermore, it was found that teacher WCF had greater impact on improving content, grammar, and language components of writing than peer WCF. This implies that teacher WCF mainly focuses on content, which leads to an increase in grammatical accuracy in student writing.

Peer Written Corrective Feedback
As for Peer WCF, the significance values for content, language, and mechanics are lower than 0.05. Thus, the data gathered from pre and posttest are not considered homogeneous. Fur- thermore, the significance value (2-tailed) for content is 0.003, which is also lower than 0.05. Thus, it can be concluded that there is a significant difference in the average score between pre and posttest. Additionally, the mean difference between pre and posttest is 2.00 for content, 2.69 for language, and 1.31 for mechanics.
In contrast, Levene's test significance value for organization and vocabulary components is higher than 0.05. Thus, the data from pre and posttest are considered homogeneous. Furthermore, the significance value (2-tailed) for the two components is also higher than 0.05. Thus, it can be assumed that the data are normally distributed. Additionally, the mean differ-ence between pre and posttest is 1.11 for organization and 0.94 for vocabulary.
The findings show that peer WCF could not improve students' competencies relating to three writing components: content, language, and mechanics. However, it did increase students' writing skills relating to organization and vocabulary components. This finding was supported by Paulus (1999) which stated that peer WCF generated more comments on organization and vocabulary.
However, benefit of peer WCF mainly depends on how this kind of feedback is provided in the classroom. If it is not delivered efficiently, peer WCF can be neglected. Moreover, it is not to be said that peer WCF is not a beneficial activity. Holt (1992) agrees that the problem is not the peer WCF itself, but its application in the classroom. Peer WCF can be more beneficial if students provide more positive opinions than just assessing their peers' written compositions. Moreover, many studies agree that students require training on how to provide WCF, making peer WCF more effective (Berg, 1999;Min, 2006). As teachers recognize the advantages of peer WCF, their beliefs regarding peer WCF also develop positively.

CONCLUSIONS
This study's findings highlight that peer WCF is more effective than teacher WCF in enhancing Indonesian senior high school students' writing performance, especially in writing legend texts. This is evident from the mean scores of peer WCF group being higher than those of teacher WCF group. Moreover, it was also recognized that several components of writing competence (content, organization, vocabulary, and language) improved significantly among the Indonesian senior high school students after receiving teacher WCF, while peer WCF had a positive impact on organization and vocabulary components.
This study recommends combining teacher and peer WCF techniques as the two complement each other in enhancing students' writing performance. Moreover, rather than only relying on teacher WCF, combining both feedback sources could provide additional benefits, for instance, making students more confident in their own writing and deciding the revision options. In addition, it could decrease their writing anxiety and enhance their writing abilities (Kurt and Atay, 2007). Furthermore, the present study suggests that it may be useful to allow the use of L1 in EFL writing classrooms, especially when peer WCF is involved. L1 could help students face difficulties in writing tasks (Saeli and Cheng, 2019). Thus, L1 could be used in peer response training programs. Additionally, writing teachers could consider encouraging students to apply L1 to enhance their feedback practices.
For future research, it would be appropriate to assign more time to the treatment process to help the students better understand how to receive and provide feedback to their peers, comprehend the feedback given by their teacher, and check their own writing before submitting it to the teacher. Moreover, further studies could examine how L1 can be employed to boost the efficacy of peer WCF in EFL writing classrooms.
This study also has some limitations. First, the study only used a quantitative method to gather the data, which may be inadequate to respond to the research questions and interpret the findings. Since the number of participants in this study was limited, the results could not be generalized. However, the findings provided us with more in-depth understanding of the application of teacher and peer WCF in EFL classrooms. If these two kinds of feedback techniques are employed in schools, a detailed picture of WCF usage could be drawn. Finally, this study did not include the students' perception of the teacher and peer WCF; if it could be included, the results would be more comprehensive.