Assessing the effect of giving and receiving written corrective feedback on improving L2 writing accuracy: does giving and receiving feedback have fair mutual benefit?

Abstract

Although the findings of second language (L2) studies conducted to date have provided evidence for the positive effect of written corrective feedback (hereafter feedback) on improving L2 learners’ writing grammatical accuracy, there is no conclusive evidence regarding which kind of feedback is more beneficial for enhancing L2 writing. This study compares the differential effects of giving and receiving unfocused direct feedback on improving Iranian EFL learners’ writing accuracy. To this end, 61 learners of English were randomly assigned to three groups, including a feedback giver group (n = 19), a feedback receiver group (n = 22), and a control group (n = 20). The participating groups took Cambridge English Preliminary Test (PET), completed four translation tasks as the treatment, and took two tests, namely a translation test and a picture description test. The data analysis indicated the effectiveness of the feedback provided by peers. Additionally, further data analysis revealed that the participants in the feedback giver group outperformed the participants in the feedback receiver and the control groups in translation and picture description tests, underscoring the positive effect of giving feedback on learners themselves rather than learners who receive feedback. The findings and potential pedagogical implications of the study are discussed in detail.

Introduction

The issue of providing written corrective feedback (hereafter feedback) has been an ongoing debate in second language acquisition (SLA) for almost three decades (for a review, study Reinders & Mohebbi, 2018). Yet, there is still no conclusive evidence concerning the efficacy of specific feedback strategies in improving second language (L2) learners’ writing accuracy. L2 researchers, to date, have investigated the effect of feedback on L2 learners’ writing ability from different perspectives, including focused and unfocused feedback (e.g., Bitchener & Knoch, 2009; Nemati, Alavi, & Mohebbi, 2019), direct and indirect feedback (Karim & Endley, 2019; Kim & Bowles, 2019), the feedback medium (Zhang, 2020), L2 learners and teachers’ perceptions, preferences, needs, and practices in writing tasks (Bahrouni & Tuzlukova, 2019; Motallebzadeh, Kondori, & Kazemi, 2020; Nemati, Alavi, Mohebbi, & Masjedlou, 2017; Raza, 2019), and the source of feedback, namely teacher and/or peer (Ferris, 1995; Hyland, 2000; Miao, Badger, & Zhen, 2006; Peterson & Portier, 2014; Zhang & McEneaney, 2020). These studies and their inconclusive and somehow contradictory findings have given rise to a great deal of discussion on the value of feedback in developing writing accuracy. Surprisingly enough, despite the importance attached to the L2 learners’ active role in L2 learning, little research has examined the potential positive effect of learners’ active role in the form of correcting their peers’ writing in L2 writing pedagogy.

The experimental studies conducted to date have yielded contradictory results regarding the impact of peer feedback compared to teacher corrective feedback (hereafter teacher feedback). Evidently, there is a gap in the literature concerning the effect of peer feedback, in particular, the potential benefit of giving feedback for the learners themselves rather than their peers. Therefore, this study might shed some light on the effects of peer feedback on enhancing L2 learners’ ability to write more accurately, especially for the learners who give peer feedback.

In the next section of the article, we outline the arguments for and against the teacher and peer feedback and review the most recent research investigating the effect of these feedback strategies on improving the accuracy of L2 learners’ writing.

The arguments for and against teacher and peer feedback

There have been many arguments for and against the efficacy of teacher and peer feedback on L2 writing. Despite the importance attached to teacher feedback, L2 research findings cast doubt on the perceived effect of teacher feedback on prompting L2 writing (Ferris, 1995). It is argued that teacher feedback might have detrimental effects on L2 learners’ writing (Hyland, 2000).

When a learner receives teacher feedback, they are usually expected to include all points provided by the teacher from A to Z. This degree of authority runs the risk of undermining the authenticity of the writing task (Berg, 1999), leaving no choice for learners in revising the composition (Muncie, 2000). Reliance on teacher feedback might deprive learners of taking responsibility for their writing tasks (Lee, 2009), which is in sharp contrast with the tenets of fostering learner autonomy. With such a profile, teacher feedback might not seem to serve any promising purpose with respect to L2 writing (Kozlova, 2010). Mahfoodh (2017) investigated learners’ emotional responses to teacher feedback. The results showed that the students felt frustrated after receiving teacher feedback. Some students were in favor of the teacher’s feedback, while others rejected it, some expressed satisfaction and some were dissatisfied with teacher feedback.

Unlike teacher feedback, peer feedback can help L2 learners build skills that can be transferred to their editing tasks and enhance their confidence in L2 writing (Byrd, 2003). The emerging process-oriented approach to L2 writing stresses the value of peer feedback in improving L2 writing (Moon, 2008). Process writing theory, collaborative learning theory, interactionist theory, and sociocultural theory supports peer feedback (Yu & Lee, 2016). Peer feedback may have the potential to provide invaluable opportunities for L2 learners to negotiate meaning; such feedback might create a facilitative socio-interactive environment in which L2 learners are encouraged to scaffold each other; and it might foster autonomy making L2 learners less dependent on the teacher (Hu, 2005; Hu & Lam, 2010).

Indeed, peer feedback can give the feedback receiver the option of casting doubt on the feedback provided, analyzing it against their knowledge, and discarding the feedback in their writing (Berg, 1999). L2 learners should experience learning opportunities through peer feedback, which might not be easily gained from teacher feedback (Berg, 1999). Likewise, peer feedback may reduce L2 learners’ writing apprehension and may be beneficial for reviewers as well as writers (Chaudron, 1984). Lundstorm and Baker (2009) claimed that peer feedback is helpful to L2 learners who give feedback because such learners become familiar with more global aspects of good writing and gain useful experience in critical reading of a manuscript. It also seems likely that peer feedback might help L2 learners come up with an evaluative checklist in providing feedback that might be taken into account in feedback givers’ manuscripts as well.

Similarly, peer feedback may help L2 learners develop general writing skills and enable them to self-evaluate their compositions. Hyland (2010) urged L2 teachers to view L2 learners as active participants in L2 writing through employing peer feedback, which involves receiving feedback, giving feedback, analyzing peer feedback, and deciding whether to include peer feedback or not. She claimed that active student participation and engagement are of overriding importance in L2 writing pedagogy if the L2 learning potential of feedback is to be fully achieved. Keen (2010) encouraged L2 teachers to take advantage of implicit learning through engaging learners in peer feedback, which has the potential to help learners internalize effective strategies, techniques, and structures that can be activated in their own writing. Miao et al. (2006) found that most of the teacher feedback and more than half of peer feedback was involved in the revised drafts of the learners. Interestingly, peer feedback led to meaning-changes while teacher feedback resulted in changes in the surface level.

On the other hand, some issues have been raised which might overshadow the effectiveness of peer feedback in improving the writing accuracy of L2 learners, in particular the need for training learners, structuring and monitoring the peer feedback sessions, potential problems about social roles and cross-cultural dynamics within pairs or groups in peer feedback, and the need for an appropriate balance between peer feedback, teacher peer feedback, self-editing, and other sources (Ferris, 2003). As Chong (2017) observed, there is a strong and positive relationship between learners’ writing proficiency and the quality of feedback they give on their peers’ writing. In a recent study, Sánchez-Naranjo (2019) reported a positive effect of peer review, especially the role of training learners to give feedback on enhancing text quality.

When an L2 learner gives feedback, they are only able to provide feedback on a certain number of errors already mastered (Kozlova, 2010). Likewise, L2 learners focus mainly on surface aspects of writing and turn a blind eye to more important revising concerns like semantic or textual ones; they provide ambiguous feedback; and they take a critical and biased stance towards their peer’s composition (Liu & Sadler, 2003). Similarly, Diab (2010) pointed out that peer feedback might not be as useful as expected because L2 learners might not be able to indicate the errors in their peer’s compositions, may not trust their peer’s WCF, and might be mainly inclined towards TWCF.

To date, L2 researchers (e.g., Liu & Sadler, 2003; Ma, 2010; Miao et al., 2006; Min, 2006; Nelson & Carson, 1998; Paulus, 1999; Peterson & McClay, 2010; Suzuki, 2008;Tsui & Ng, 2000; Villamil & De Guerrero, 1996, 1998; Wigglesworth & Storch, 2009; Zhu, 2001) have examined different aspects of peer feedback in improving L2 writing. The studies conducted so far have reported mixed findings with regard to the effectiveness of peer feedback in improving L2 learners’ writing accuracy. Two recent studies provided evidence for the effect of peer feedback: Wu and Schunn (2020) investigated the relationship between peer feedback features, student perceptions and the potential chance of implementing the feedback and provided support for the effectiveness of peer feedback. Pham, Huyen, and Nguyen (2020) reported an 11-week case study inquiring into the effect of the trained peer feedback of 92 English-major students. They concluded the positive impact of peer feedback on revising their writing. However, these are two studies in different contexts which makes the findings less comparative and conclusive. Evidently, what we need is more research projects investigating this issue in various educational contexts.

The study

The studies conducted to date in this field have yielded conflicting findings. Lee (2011) underscored the fact that although teacher feedback is a dominant technique in writing pedagogy and L2 learners rely more and more on their teachers’ comments, their writing does not improve significantly as a result of teacher feedback. In contrast, Peterson and Portier (2014) reported an advantage for peer feedback. They observed that the learners gave feedback on the content and conventions in their peers’ writing. They concluded that “Written compositions improved and writing development was enhanced through the giving and receiving of peer feedback” (p. 20). As a result, our study may extend the research on the efficacy of giving and receiving feedback on the writing accuracy of L2 learners in an EFL context in Iran.

Research question

The following main research question guided the present study:

Is there any significant difference between peer feedback giver and receiver groups in terms of writing accuracy?

Method

Participants

Sixty-one Iranian learners of English (male and female, ranging in age from 18 to 21) from two state universities participated in this study. They were first and second-year university students majoring in Translation Studies and English Language Teaching (ELT). The participants had Azari-Turkish as their first language and Persian as a formal language of instruction in Iran. The participants were randomly divided into three groups, namely a feedback giver group (n = 19), a feedback receiver group (n = 22), and a control group (n = 20). The homogeneity of the participating groups in terms of L2 proficiency was assured through Cambridge English Preliminary Test (PET).

Instruments

Proficiency test

To get assurance as to the homogeneity of the participating groups in terms of L2 proficiency, the participants took PET. The test has four main sections: reading, writing, listening, and speaking. In this study, we excluded the listening section as it was irrelevant to the present study. Total marks for reading, writing, listening, and speaking components are 25, 25, 25, and 25. PET has two cut-scores, pass and pass with merit. Those who get 70 pass the test and are judged to be at an intermediate level of proficiency.

Translation tasks

We chose translation task because this task, similar to tasks based on visual materials such as picture description task, assesses only writing, not the content knowledge that is assessed in argumentative writing tasks (Hughes, 2003). Also, this task provides an obligatory context for learners to produce the targeted structure(s). We selected five reading texts from the Active Reading 2 book. The readability indices of the texts were computed by readability software 1.0 and the Flesch-Kincaid readability (FKRT). The FKRT readability indices calculated for text 1, text 2, text 3, text 4, and text 5 appeared to be 57 (Grade 9), 56 (Grade 9), 51 (Grade 10), 48 (Grade 10), and 42 (Grade 11), respectively. In the Flesch-Kincaid readability test, the higher numbers (90.5–100.0) imply that the passages are easier to read, and lower numbers (0–30) mark passages that are more difficult to read. The numbers given in parentheses stand for the readability of the passages. The indices computed revealed that the passages were appropriate for the participants of the study.

The selected passages were translated into Persian. To get assurance as to the reliability of the translations, two of our colleagues checked the translations. Additionally, a list of keywords for each passage was given in English. Four passages were used in the treatment sessions.

Posttests

To examine the effect of different levels of the independent variable, i.e., giving and receiving peer feedback on the participants’ L2 writing accuracy, they were asked to translate a text from Persian into English and perform a picture description task. In this task, a series of five pictures were provided which depicted a story. The participants were required to narrate the story in 150 words or more.

Procedures

Before the main study, we conducted a pilot study to figure out any potential problem in the data collection procedure and the designed tasks and also come up with the time which participants needed to complete the tasks. Based on the findings of the pilot study, the tasks were adapted, and the time of task completion was set as 30 min.

On the day before the treatment sessions of the main study started, the participants took the PET. They proved to be homogeneous in terms of second language proficiency, F (2, 58) = 2.81, p = .068. Then, they were randomly assigned to three groups, namely a feedback giver group, a feedback receiver group, and a control group. They completed the translation tasks for four sessions. After completing the first task, the feedback giver group provided feedback on the manuscripts of the feedback receiver group. The participants in the feedback giver group could use a dictionary in commenting on the manuscripts. At the beginning of the second session, the participants in the feedback receiver group were offered the chance of reviewing the comments given by the feedback giver group and revising their writing with the comments incorporated. Then, they completed the second translation task. For four sessions, the feedback giver group completed their translation tasks and provided feedback on the manuscripts’ of participants in the feedback receiver group. In fact, the participants in the feedback giver group provided direct unfocused feedback by providing the correct form of all errors. Likewise, the feedback receiver group was required to review the feedback provided on their previous manuscripts, reflect on them, and complete a new translation task. The participants in the control group only completed the translation tasks without providing or receiving any feedback. When the treatment sessions were over, all participant groups took a translation task, as a testing instrument. After 2 weeks, they completed a picture description task, as the second testing instrument, to assess the effect of treatment conditions on a different task.

Every single participant’s writing was coded for the number of clauses and error-free clauses. Following Foster and Skehan (1996), the accuracy of L2 writing was measured by calculating the number of error-free clauses as a percentage of the total number of clauses. The rater who was one of our colleagues who had 10 years of teaching experience corrected the manuscripts, and the number of error-free clauses was divided by the total number of clauses in each manuscript and was multiplied by 100. One-third of the manuscripts in each task was randomly scored by another rater to ensure inter-rater reliability. The inter-rater reliability for the translation task and the picture description task were .95 and 1, respectively.

Results

Cambridge English preliminary test

As already mentioned, PET was used to check the homogeneity of the participating groups. The data obtained were submitted to SPSS for running statistical analyses. Firstly, the Levene’s test of equality of error variances was run to check whether the ANOVA could be run or not. The results of the Levene’s test of equality of error variances, p = .43, indicated that ANOVA could be used. Table 1 shows the results of the Levene’s test of equality of error variances for the proficiency test.

Table 1 Levene’s test for the proficiency test

Secondly, an ANOVA was run to ensure the homogeneity of the participating groups. The results of the ANOVA run revealed that there was no statistically significant difference among the three participating groups in terms of L2 proficiency. Table 2 shows the ANOVA results for the proficiency test.

Table 2 ANOVA results for the proficiency test

Translation task

The first testing instrument which was used to examine the effectiveness of different conditions of the study was the translation task. The data obtained was put into SPSS and a One-way ANOVA was run. Table 3 shows the descriptive statistics for the translation task, and Table 4 presents the results of the ANOVA run for the translation task.

Table 3 Descriptive statistics for the translation task

Table 4 ANOVA results for the translation task

As Table 3 presents, the mean for the feedback giver group (M = 72.06) is higher than the mean for the feedback receiver group (M = 50.58), which in turn is higher than the mean for the control group (M = 46.19).

The difference across the participating groups reached significance, F = 8.52, p=. 001. The values observed for the F and p indicated that the three groups involved did not perform equally in the translation task because of the different conditions to which they were exposed during the treatment period (Table 4).

To statistically determine where the significant differences lay between the participating groups, a post-hoc (LSD) test was also run. As Table 5 represents, there was a statistically significant difference between the feedback giver group and the feedback receiver group, in favor of the feedback giver group, p = .002. Likewise, the difference between the feedback giver group and the control group reached statistical significance in favor of the feedback giver group, p = .001. However, the difference between the feedback receiver group and the control group did not reach statistical significance, p = .50.

Table 5 Post-hoc Analysis (LSD test) for the translation task

In brief, the data analysis revealed that the participants in the feedback giver group outperformed the participants in the feedback receiver group and the control group in the translation task. Although the difference between the feedback receiver group and the control group was not significant, the descriptive data showed that the participants in the feedback receiver group performed better than the participants in the control group in the translation task.

Picture description task

In the second task, the members of the participating groups were required to narrate a story based on a sequence of pictures given to them. Since the Levene’s test for equality of error variances appeared to be significant, p = .001, the non-parametric alternative to ANOVA, i.e., the Kruskal-Wallis H Test was run for the picture description task. Table 4 shows the results of the Kruskal-Wallis H Test for the picture description task.

As it is clear in Table 6, the feedback giver group obtained the highest overall mean rank, the control group did the lowest, and the feedback receiver group ranked in between.

Table 6 Kruskal-Wallis H Test for the picture description task

The output presented in Table 7 for the Kruskal-Wallis H Test demonstrated that there was a statistically significant difference in feedback across the three groups involved in the present study (Gp1, n = 19: Gp2, n = 22:, Gp3, n = 20), x² (2, n = 61) = 22.91, p < .001.

Table 7 Test statistics for the picture description task

The feedback giver group obtained a higher median score (Md = 95.65) than the feedback receiver and control groups, which recorded median values of 75.00 and 36.67, respectively. Since the difference across the three groups reached statistical significance in the Kruskal-Wallis H Test, there was a need to carry out three follow-ups Mann-Whitney U tests between pairs of groups (i.e., between the feedback giver group and the feedback receiver group, the feedback giver group and the control group, and the feedback receiver group and the control group) to locate the exact difference among the three groups of participants.

To control Type I errors, a Bonferroni adjustment was applied to the alpha values (i.e., .05 ÷ 3 = .0167). The descriptive statistics for the Mann-Whitney U for the feedback giver group and feedback receiver group were presented in Table 8. As it is shown, the mean rank for the first group (MR = 26.05) was higher than the mean rank for the second group (MR = 16.64).

Table 8 Feedback giver group and feedback receiver group ranks in the picture description task

The Mann-Whitney U run for the feedback giver group and feedback receiver group showed that there was a statistically significant difference between these two groups (Md = 95.65, n = 19, Md = 75.00, n = 20), U = 113, z = − 2.51, p = .1012 (Table 9).

Table 9 Test statistics for the feedback giver group and feedback receiver group in the picture description task

Considering the pieces of information presented in Tables 8 and 9, it can be claimed that the feedback giver group outperformed the feedback receiver group significantly in the picture description task.

The second Mann-Whitney U was run between the feedback giver group and the control group. The descriptive statistics presented in Table 10 demonstrated that the mean rank for the feedback receiver group (MR = 27.82) was higher than the mean rank for the control group (MR = 12.58).

Table 10 Feedback giver group and control group ranks in the picture description task

The test also showed that there was statistically significant difference between these two groups too (Md = 95.65, n = 19, Md = 36.07, n = 22), U = 41.50, z = − 4.18, p = .001 (Tables 9, 11 and 12).

Table 11 Test statistics for the feedback giver group and control group in the picture description task

Table 12 Feedback receiver group and control group ranks in the picture description task

The descriptive data gained from the third Mann-Whitney U run revealed that the mean rank for the feedback receiver group (MR = 27.59) was higher than the control group (MR = 14.80). As it is clear in Table 13, there was statistically significant difference between the feedback giver group and control group (Md = 95.65, n = 19, Md = 36.67, n = 20), U = 86, z = − 3.37, p = .001. Simply put, the feedback receiver group outperformed the control group in describing the pictures given as the posttest.

Table 13 Test statistics for the feedback receiver group and control group in the picture description task

Discussion

The present study examined the effect of giving and receiving peer feedback on L2 learners’ writing accuracy in an EFL context in Iran. More specifically, it aimed to investigate whether the participants who provided feedback could take advantage of feedback more than the participants who received feedback. First, it is worth mentioning that although there has been plenty of arguments for and against feedback in the last three decades (e.g., Bitchener & Knoch, 2010; Ferris, 2004; Truscott, 2004), the results of this study provided support for the effectiveness of feedback, namely unfocused direct feedback on improving L2 learners’ ability to write accurately.

The result of the study implies the positive effect of peer feedback. Peer feedback might have the potential to enhance learner autonomy and encourage learners to take an active role in L2 learning. Peer feedback might raise learners’ awareness of their strengths and weaknesses in L2 writing, help learners become familiar with global aspects of L2 writing and critical reading, foster reflection, facilitate collaborative learning and assume an active role for learners in L2 learning.

More specifically, viewed in light of the involvement load hypothesis (Hulstijn & Laufer, 2001), it seems reasonable to argue that peer feedback engages L2 learners more in performing L2 writing tasks, which in turn leads to improved L2 writing ability. The involvement load hypothesis argues that learning is dependent on the amount of mental effort or involvement that a task imposes. Three factors play a crucial role in task-induced involvement, namely need, search, and evaluation. “Need” implies the obligation to complete the task. “Search” indicates the attempts which an L2 learner makes to resolve the task, and “evaluation” involves passing judgment about the outcome of the task. Therefore, it can be argued that the participants in the feedback giver group were engaged in a writing task that created the need to complete the task, i.e., providing the feedback, so they searched for the errors in the manuscripts, evaluated them, and finally delivered the correct form. Peer feedback might involve learners deeply in an authentic task that provides them an invaluable opportunity to read their peers’ manuscripts critically, locate the errors in the manuscripts, and employ their explicit knowledge to give the correct form. Then, based on the feedback provided for their peers’ manuscripts, they would be able to avoid making the same errors in their manuscripts.

As the literature indicates, L2 teachers need to be encouraged to take advantage of peer feedback compared with teacher feedback. As already mentioned, teacher feedback is undermined because when L2 learners are provided with teacher feedback, they are expected to include all of the teacher feedbacks in their revised draft. In contrast, peer feedback has a positive effect on L2 writings of both the learners who provide and those who receive feedback even if they decide not to respond to all the peer feedback they receive. Furthermore, learners who provide feedback might take more advantage than learners who receive feedback. In sum, as Yu (2016) summarizes, we need to focus on seven issues about peer feedback: effectiveness of peer feedback compared with teacher feedback; benefits of peer feedback for givers; computer-mediated peer feedback; peer feedback training; learners’ viewpoints, preferences, needs, and motives; peer interaction and group dynamics; and contextual and cultural issues. Each of these issues needs further research to provide useful guidelines and instructive suggestions for teachers to practice peer feedback in language learning classrooms. Besides, as Yu and Hu (2017) highlight, we need to consider individual differences that play a vital role in the quality of peer feedback too.

Concluding remarks

In conclusion, the findings of the present study lend support to the effectiveness of peer feedback, in particular, the unfocused direct peer feedback in enhancing the writing accuracy of L2 learners in the EFL context. Moreover, the results of this study revealed that L2 learners who provide feedback on their peer’s manuscripts gain more advantages in their own L2 writing tasks than the learners who receive peer feedback. Hence, L2 teachers should be urged to give L2 learners the chance of providing feedback on their peer’s manuscripts. Unfortunately, teacher feedback is the dominant strategy in writing instruction in L2 learning classrooms, in particular EFL contexts. L2 learners are given little opportunity to take a more active role in L2 learning and merely revise their manuscripts, if they are asked, and include all of the teacher feedback given without reflecting on them. In brief, it seems imperative that L2 researchers and practitioners need to conduct more studies in different language learning contexts with a more significant number of participants at different language proficiency levels, in this field and inform L2 teachers of the effectiveness of implementing peer feedback in L2 writing pedagogy. However, as Reichelt (2019) rightly argues, we do need to take into account the contextual, sociolinguistic, and educational factors in L2 writing instruction, in particular when it comes to adopting a specific feedback strategy. Also, as Sánchez-Naranjo (2019) highlights, systematic training of learners to give feedback on their partner’s writing should be taken into account too.