A Comparison of the Effects of Revision- Mediated and Attention-mediated Indirect Coded Feedback on EFL Learners’ Written Syntactic Accuracy

This study investigated the immediate and sustained effects of revision-mediated (RICF) and attention-mediated (AICF) indirect coded corrective feedback on the written syntactic accuracy development of 50 upper-intermediate Iranian EFL learners assigned to two RICF and AICF groups. They received eight-session treatments, followed by an immediate posttest and then a delayed posttest after a four-week time interval. The collected written scripts were co-rated for syntactic accuracy by the researcher and her colleague. Descriptive and inferential statistical analyses were conducted with SPSS 21. The Wilcoxon Signed Rank Test and the paired-samples t-tests revealed that each treatment separately (i.e., revision-mediated and attention-mediated CF) had a statistically significant impact on EFL learners’ written syntactic accuracy both in the short and long term. Moreover, the independent samples t-tests indicated no significant difference between the effects of revision-mediated and attention-mediated CF on EFL learners’ written syntactic accuracy both in the short and long term. The paper provides discussion and implications.


Introduction
Corrective feedback (CF) is an important part of the second language (L2) writing programs (Karim & Nassaji, 2020;Mao & Lee, 2020). Although many teachers believe that CF helps L2 learners improve their writing accuracy (Ferris, 2004), the effectiveness of CF has been a controversial topic in the field of second language acquisition (SLA). This debate started with Truscott's arguments in 1996 that WCF is ineffective and even harmful. Since then, other researchers have responded to his argument and said that CF is effective (e.g., Bruton, 2009Bruton, , 2010Chandler, 2003Chandler, , 2004Ferris, 1999Ferris, , 2004Ferris, , 2006. A substantial amount of research since then has been conducted to examine whether and to what extent WCF assists L2 writing. Nonetheless, studies have shown that even if teachers provide the learners with clear and useful feedback, learners also must work with it to brush up their writing (Elwood & Bode, 2014). If learners are not required to respond to the received WCF, they may ignore it or attend to it only partially (Ellis, 2009;Liu & Brown, 2015). In other words, WCF can be effective and learners can notice corrections as long as they are required to notice and process the received corrections (Shintani & Ellis, 2015). Therefore, learners' involvement plays a crucial role in learning outcomes and helps teachers better provide WCF (Ellis, 2010). As a result, teachers need to improve their WCF by understanding their engagement (Han & Hyland, 2015). Some researchers believe that such revision can be a helpful and important step towards the long-term acquisition of a language feature (Ferris, 2004(Ferris, , 2010Guénette, 2012;Sachs & Polio, 2007). As Sachs and Polio (2007) stated, "reports of noticing during the processing of written feedback were related to subsequent revision changes" (p. 85). Furthermore, Truscott and Hsu (2008) claimed that revision has an important role in good writing, so that if the learners do not receive the feedback and do not revise their essays, they may assume "that they have nothing to learn from their written assignments" (Hyland, 2013, p. 184). Moreover, according to scholars, to engage the learners with the feedback and also hold them accountable for their learning, they should be required to revise their essays based on the teacher's feedback (Ferris, 2006;Guénette, 2012;Hyland, 2003;Shintani & Ellis, 2015;Storch & Wigglesworth, 2010). Therefore, the revision is incorporated by several scholars in their studies and positive results have been found (Chandler, 2003;Diab, 2015;Frear, 2012;Shintani, Ellis, & Suzuki, 2014;Suzuki, 2012;Van Beuningen, De Jong, & Kuiken, 2012).
On the other hand, some scholars stated that even under the condition of no revision opportunity, WCF could be effective (Shintani & Ellis, 2015). Learners can succeed in noticing corrections (Ellis, 2009) provided that learners are asked to notice and correct their mistakes based on the received feedback (Shintani & Ellis, 2015). Some claimed that the revision requirement is neither necessary to notice nor the same as normal WCF practice given in real classrooms because the learners are normally not required to revise their written text based on teacher's feedback (Stefanou & Révész, 2015). It is also discussed that the students may revise their texts passively without noticing their errors and the provided WCF. Therefore, it is crucial that teachers draw learners' attention to the target of the provided WCF (Polio, 2012;Stefanou & Révész, 2015). Attracting learners' attention can be achieved by asking them to review the received feedback and carefully examine their errors (Ellis, 2009;Polio, 2012). Consequently, some researchers applied this methodology (Ellis et al., 2008;Sheen, 2007;Shintani & Ellis, 2015;Stefanou & Révész, 2015) and found promising results.
Given the mentioned points, it can be hypothesised that providing indirect coded written feedback, as a way of error correction, followed by either revision requirement or attention requirement, may help L2 learners promote their written syntactic accuracy.
The findings of such a study inform researchers, practitioners, and teacher trainers to develop teachers' knowledge in this area. Learners may also be motivated and use discovery learning and become autonomous by revising the received indirect coded written feedback attentively. Considering the significance of such practical consideration of WFC, very few studies have addressed systematic exploration of the mentioned issues. Therefore, in this study, the following research question was addressed to serve the objectives of the research: 1. Does revision-mediated indirect coded feedback have any significant instructed effect on EFL learners' syntactic accuracy in writing? 2. Does attention-mediated indirect coded feedback have any significant instructed effect on EFL learners' syntactic accuracy in writing in the short and long term? 3. Is there any significant difference between the effect of revision-mediated indirect coded feedback and attention-mediated indirect coded feedback on EFL learners' syntactic accuracy in writing in the short and long term?

Review of Literature
Chandler (2003), Frear (2012), andVan Beuningen et al. (2012) included the mediating factor of revision and found promising results. Some other researchers have added the revision in their studies and realised positive results. Suzuki (2012) explored the effects of WCF by asking 24 Japanese learners of English to write out their reflection in Japanese of the CF they received. The effects of the type (e.g., grammar-based vs. lexis based) of WCF were then evaluated by investigating the success of immediate subsequent text revisions. Firstly, learners successfully corrected their errors during immediate revision in the first essay by receiving direct feedback on linguistic errors. Secondly, not only lexis-but grammar-based WCF were accompanied with improved accuracy. These results support arguments that furnishing learners with the opportunity to language about or reflect on their developing linguistic knowledge in the course of L2 learning bring about L2 learning and development.
Then, Shintani et al. (2014) compared the effects of DCF and metalinguistic explanation (ME) on 214 Japanese university students' accurate use of the indefinite article and the hypothetical conditional sentences. The participants were considered preintermediate learners. Both types of feedback were provided with and without a revision opportunity in the form of rewriting. The feedback improved the accuracy of the hypothetical conditionals but not of the indefinite article. It was also proved that the lasting effect of the DCF was longer than one for the ME. Further, the revision opportunity positively affected the efficacy of the feedback. Shintani and Ellis (2015) examined whether language analytical ability (LAA) mediated the extent to which 118 Japanese university students of English improved in their accurate use of the indefinite article and the past unreal conditional in new writing depending on the type of feedback (direct feedback or syntactic explanation) and on whether they could revise. LAA played an increasingly significant role for those learners who had revised their original writing using the feedback. It is suggested that the extent that LAA was involved depended on a complicated interplay involving the type of feedback, the possibility to revise, and the target structure, which affected the depth of processing the learners engaged in.
Focusing on the mediating role of attention, Sheen (2007) first compared the effects of two types of WCF (direct-only and direct metalinguistic) on the acquisition of articles. The participants were 91 adult intermediate ESL learners. There was also a control (i.e., no-correction) group. After the participants received feedback, they were required "to look over their errors and the corrections carefully for at least 5 minutes" (p. 264). Eventually, it was revealed that both interventions significantly improved the immediate posttests in comparison to the control group.
Ellis et al (2008) compared the effects of focused and unfocused WCF on the accuracy after three treatment sessions with which Japanese university students used the English definite and indefinite articles to give first and anaphoric reference in written narratives. Both groups were asked to write three narratives and presented them with WCF. After receiving the feedbacks, the participants were required to look over their errors and the corrections carefully for at least five minutes. The focused and unfocused groups learned from pre-test to posttests while taking a test involving a new piece of narrative writing and an error correction test. Both groups outperformed a control group that received no correction on the second posttest.
Stefanou and Révész (2015) reported on a classroom-based study that investigated the effectiveness of direct WCF with two treatment sessions. Eighty-nine Greek EFL learners were randomly divided into 3 groups: direct feedback plus metalinguistic feedback, direct feedback only, and comparison. The use of the article for specific and generic plural references was the linguistic target. After receiving the feedback, the participants were given five minutes to look over their errors and attend to the respective feedback carefully. Finally, an advantage for receiving direct feedback over no feedback was revealed, but the results supplied no clear evidence for the benefit of contributing metalinguistic information.
Lately, Soltanpour and Valizadeh (2018a) studied the effects of attention mediation versus revision mediation comprehensive direct CF (DCF) on EFL learners' syntactic accuracy of their argumentative essays of 83 Iranian EFL learners, studying at an upper-intermediate level. There were three groups: one who received DCF plus a time to pay careful attention to and study the errors, received DCF plus a revision requirement, and the control group that received only the DCF (without any additional assignment). The results revealed that both careful attention requirement and revision requirement significantly outperformed the group that only received the feedback. It was also demonstrated that the group required to pay careful attention to and study the feedback significantly outperformed the one involved in the revision requirement.
In short, to help teachers enhance their WCF, a more thorough understanding of learner engagement with WCF is needed (Han & Hyland, 2015). Thus, revision studies are interesting and provide significant evidence helping teachers clarify their practice (Ferris, 2010).

Method Design
The study was carried out in real classrooms "within the context of an instructional program, with ecologically valid writing tasks", which was recommended by Storch (2010, p. 42). Because the non-random convenience sampling (i.e., intact groups) was utilised, the study had a quasiexperimental design. The study had two groups and included pre-test, immediate post-test, and delayed posttest. The independent variable in this study was divided into two categories of revision-mediated and attention-meditated indirect coded feedback types. The dependent variable in this study was the syntactic accuracy of the EFL learners' writing performance.

Participants
The participants were 50 Iranian EFL Persian native speakers. Learners' age ranged from 15 to 20 years old. They studied Summit 1 and Summit 2 (Saslow & Ascher, 2012) at the upper-intermediate proficiency level at a Language Institute in Iran. Their homogeneity in terms of their language proficiency level was checked by an Oxford Placement Test. Then the students were assigned to two experimental groups: 20 women and five men in Revisionmediated Indirect Coded Feedback (henceforth, RICF), and 23 women and two men in the Attentionmediated Indirect Coded Feedback (henceforth, AICF) groups. In this study, the participants' written assignments were rated by two raters: the researcher of this research and another experienced English teacher. They held master's degrees in Teaching English as a Foreign Language.

Instruments
Several instruments and materials were administered in this study to set up the experiments, collect data and analyse the obtained data: Oxford Quick Placement Test (OPT), writing elicitation tasks, pretest, immediate posttest, and delayed posttest, as well as syntactic accuracy measurement formula.

Writing Elicitation Tasks
Eleven writing elicitation tasks were selected from various IELTS (International English Language Testing System) and TOEFL (Test of English as a Foreign Language) sites on the Internet. The argumentative essay writing genre was chosen as the intended writing assignment for the students. Argumentative essays are, at their core, pieces of writing that aim to convince the reader of the writer's own opinion. This genre of essay writing provides an opportunity for the learners to practice the skill of persuasion. After briefing and administering the OPT, every session, the participants were provided with a topic. They were required to write an essay of 150-200 words in a maximum of 25 minutes and turn it in timely.

Pretest, Immediate Posttest, and Delayed Post test
At the outset and the end of the eight weeks of treatment, a pretest and an immediate posttest of writing were administered. After a four-week time gap, a delayed posttest was conducted to examine the sustained effect of the treatment on the students' writing development. The topics of the tests were also of argumentative types.

Writing Syntactic Accuracy Measurement Formula
To assess the writing syntactic accuracy, a wellpractised formula was used: Accuracy Index = [total number of syntactic errors/total number of words] × 100. The formula was already used by Chandler (2003), Truscott and Hsu (2008), Soltanpour and Valizadeh (2018a) as well as Valizadeh (2020). Syntactic errors in this research referred to the writers' errors related to the grammatical arrangement of words in a sentence and errors about how words change their form and combine to make sentences. The indirect coded feedback to the syntactic errors committed by both AICF and RICF groups were written above in corrected words by the researcher. The coded feedback system is summarised in Table 3. As Table 3 shows, the syntactic errors made by the learners including punctuation, capitalisation, spelling, and grammatical points, which involve singular/plural forms, verb tense, missing words, connectors, prepositions, pronouns, articles, word choices, subject-verb agreement, and word order were addressed by coded CF.

Procedure
The whole course was comprised of 16 sessions of 90 minutes. At the outset, a paper-and-pencil version of OPT was administered to remove the initial differences in general English proficiency among the participants in this study. A Few of the participants was omitted because of the nonhomogeneity. The participants whose scores ranged between 40 and 47 (upper-intermediate) (Geranpayeh, 2003) were selected and assigned into two experimental groups of AICF and RICF (each included 25 participants). In Session 2, the format and elements of argumentative essays were worked on in a writing workshop. In the same session, both groups sat for a pretest of writing, which was used to assess the learners' initial level of syntactic accuracy before entering the intervention phase. The collected written scripts were rated for syntactic accuracy by the researcher and her colleague. Next, as brief tutoring to the indirect coded feedback system, the participants were given a handout accompanied by the researcher's oral instructions and examples on how to memorise, interpret and understand the codes while revising their assignment (RICF) or attending to their errors (AICF).
The treatment period lasted for eight sessions of 90 minutes. In every session, the RICF and AICF groups were assigned the same topic to write about in a maximum of 25 minutes. The required word limit was between 150-200 words. The researcher's feedback was given to the syntactic errors in the form of the annotated coding system. The participants were required to figure out what the received codes refer to and correct their errors based on the coding system. Then they turned in their writings to the researcher to check learners' corrections again. Afterwards, the researcher returned their writings. The only difference between the two experimental groups was that the participants in the RICF group were required to revise (i.e., rewrite) their initial drafts on a separate sheet of paper according to the received WCF in 20 minutes, while the participants in AICF had similar 20 minutes to look over their errors and pay close attention to the received feedback. Therefore, the two experimental groups were not treated differently except for their reactions to the received feedback.
After the treatment was over, an immediate posttest was administered. The participants' writings were rated for syntactic accuracy before a delayed posttest was conducted in Session 16. The time limit for each test was a maximum of 25 minutes and the word count was between 150 and 200. The same rating formula for measuring syntactic accuracy was conducted for the collected papers in the delayed posttest.

Inter-Rater Reliability Indices
The Cronbach α indices were calculated as interrater reliability coefficients for all the conducted tests in this study. The indices ranged from a high measure of α = .999 for the immediate posttest in the AICF group to α = .981 for the pretest of the RICF group.

Normality Distribution of Test Scores
The assumption of normality was initially examined through the histograms followed by the Kolomogrov-Smirnov and Shapiro-Wilks test of normality. The only tests that did not prove to be normal were the pretest of both groups. Their histograms did not show normality and the Sig. The value of the tests was below .05, suggesting a violation of the assumption of normality. The data for the other tests proved to be normal.

Homogeneity of the Groups
To ensure the groups were homogeneous, an independent-samples t-test was conducted to compare the OPT scores for the AICF and RICF groups. There was no significant difference in scores of the AICF group (M = 43.36, SD = 1.890, N = 25) and the RICF group (M = 43.68, SD = 2.015, N = 25); t (48) = .579, p = .565.
Moreover, a Mann-Whitney U Test was done to compare the pretest scores of the AICF and RICF groups. It also revealed no significant difference in the scores of the AICF group (Md = 11.3700, N = 25) and the RICF group (Md = 11.5600, N = 25), U = 310.500, z = -.039, p = .969.
It was concluded that the two groups were homogeneous before the administration of the treatments.

Answer to 1 st Research Question
The first research question examined whether RICF has any significant effect on EFL learners' syntactic accuracy in writing in the short and long term. To investigate the immediate effects of the RICF, a Wilcoxon Signed Rank Test was conducted. The descriptive statistics score (Table 2) show a reduction in values from the pretest to the posttest. It should be noted that as the formula [total number of syntactic errors/total number of words] × 100 was utilised for scoring the syntactic accuracy of the essays, the fewer errors the essays included, the smaller value (mathematical quantity) they were given, so the lower values reveal the existence of fewer errors and better performance. As a result, it can be concluded that the RICF group showed improvement in performance. The Wilcoxon Signed-Rank Test revealed this improvement was statistically significant, z = -3.942, p = .000, r = .55. The effect size was large, too, based on Cohen's (1988) guidelines (as cited in Pallant, 2013).
Then to investigate the delayed effect of RICF on learners' performance in terms of the syntactic accuracy, the participants' immediate posttest scores were compared to their delayed posttest scores with a paired samples t-test. There was a statistically significant difference in the syntactic accuracy writing in the RICF group from the immediate posttest (M = 8.6240, SD = 2.38909) to the delayed posttest (M = 5.9444, SD = 2.46765), t (24) = 3.514, p < .002 (two-tailed). The calculated eta squared statistic (.20) indicated a below medium effect size, based on the guidelines proposed by Cohen (1988, cited in Pallant, 2013.

Answer to the 2 nd Research Question
The 2nd research question examined whether AICF has any significant instructed effect on EFL learners' syntactic accuracy in writing in the short and long term. To investigate the immediate effect of the AICF, a Wilcoxon Signed Rank Test was conducted. The descriptive statistics (Table 3) show a reduction in values from the pretest to the posttest.
The lower values reveal the existence of fewer errors and better performance. In this study, the Wilcoxon Signed Rank Test revealed this improvement was statistically significant, z = -4.157, p = .000, r = .58.
The effect size was large, too, based on Cohen's (1988) guidelines (as cited in Pallant, 2013).
Then to investigate the delayed effect of AICF on learners' performance in terms of the syntactic accuracy, the participants' immediate posttest scores were compared to their delayed posttest scores with a paired samples t-test. There was a statistically significant difference in the syntactic accuracy writing from the immediate posttest (M = 8.4952, SD = 2.78365) to the delayed posttest (M = 7.0124, SD = 2.74088), t (24) = 7.092, p < .000 (two-tailed). The calculated eta squared statistic (.51) indicated a large effect size, based on the guidelines proposed by Cohen (1988, cited in Pallant, 2013.

The Answer to 3 rd Research Question
The third research question in this study examined whether there is a significant difference between the instructed effect of RICF and AICF on EFL learners' instructed syntactic accuracy in writing. An independent samples t-test was run to compare the mean scores of both RICF and AICF groups on the immediate posttest to examine the immediate effect of the two types of mediation on the short-term improvement of their syntactic accuracy in writing.
Then, this study examined whether there is any significant difference between the effects of RICF and AICF on EFL learners' sustained syntactic accuracy in writing. Another parametric independent samples t-test was run to compare the mean scores of both RICF and AICF groups on the delayed posttest to examine the sustained effect of the two types of mediation on the long-term improvement of their syntactic accuracy in writing. The assumption of equality of variances was met (Levene's F = .51, p = .47 > .05, 95% Confidence Interval = -2.55107, .41507). Based on the results, it was concluded that the measure of t (48) = -1.448, p = .15>.05 was not significant and no statistical difference could be reported in mean scores for the RICF (M = 5.94, SD = 2.46) and AICF (M = 7.01, SD = 2.74).

Discussion and Conclusion
The present study attempted to investigate the likely immediate and sustained effects of two separate strategies of AICF and RICF types on EFL learners' written syntactic accuracy. Additionally, the two mentioned treatments were compared to discover the differences between their effects.
Statistical analysis in this study revealed that each of the treatments separately (i.e., AICF & RICF) developed the syntactic accuracy in the writing performance of the EFL learners. Moreover, the statistical analysis in this study reported no significant difference between the effects of RICF and AICF on the immediate and sustained improvement of syntactic accuracy of EFL learners' writing.
Theoretically, comprehensive WCF needs a high level of attentional control and conscious thought, increasing the attentional load to automatise the L2 information sent via feedback (Frear & Chiu, 2015) and reduces the likelihood of learners' awareness. Therefore, comprehensive WCF might not be useful for less proficient learners because they may not attend to the feedback and recognise a mismatch or gap between what they can produce and what they need to produce (Gass, 1997;Schmidt, 2001). By contrast, comprehensive WCF can benefit advanced learners who can probably deal with a high cognitive load required for processing feedback targeted at a wide range of error categories (Mao & Lee, 2020). As the learners in this study were at the upperintermediate level of English proficiency, they benefited from the provided WCF type.
Moreover, because the learners in this study were required to revise their written papers based on the feedback they received, this could have made them attend to their errors and increase their level of attention and reflection; therefore, the finding of this study also support the researchers who demonstrated the effectiveness of reflection in their studies (e.g., Hemmati & Soltanpour, 2012;Soltanpour & Valizadeh, 2017). In addition, the finding of this study supports the previous studies which supported the effectiveness of indirect WCF (Karm & Nassaji, 2020). Some studies have claimed that the provision of codes in indirect WCF can promote scaffolding that results in reflection and negotiation of linguistic items (Buckingham & Aktuğ-Ekinci, 2017;Ferris & Roberts, 2001;Valentín-Rivera, 2016).
Furthermore, the results of the present study support Soltanpour and Valizadeh's (2018a) finding in terms of the effectiveness of the revision-mediated and attention-mediated WCF. However, the current study did not find any significant difference between the effects of the two variables. At the same time, Soltanpour and Valizadeh's (2018a) demonstrated that attention-mediated WCF was significantly better than revision-mediated WCF in terms of improving the learners' written syntactic accuracy.
Further, the results of this study are in line with second language acquisition theories. It is now widely accepted that effective L2 pedagogy should encourage the learners' attention to linguistic form. In the absence of such attention, L2 learning could become slower, more difficult, and less successful (Doughty, 2003). An instructional involvement that has received considerable attention and has been prescribed in the SLA field (see Norris & Ortega, 2000 for a review) is Long's focus-on-form approach (Long, 1996;Long & Robinson, 1998). Another crucial role accompanied by such attention is its ability to make learners aware of "a mismatch or gap between what they can create and what they need to create, as well as between what they create and what target language speakers create" (Schmidt, 2001, p. 6). This concept has been commonly referred to as noticing the gap (e.g. Schmidt, 1990). Ellis (1995) has used the term cognitive comparison instead because, in his view, learners also need to notice whether their output is the same as the input.
In brief, the results of this study are in favour of teacher's corrective feedback, confirming that teacher's RICF and AICF can be effective in improving students' writing accuracy in ESL/EFL contexts.