Abstract
Despite the proliferation of research on how second language (L2) learners engage with feedback on L2 writing in recent years, little is known about how young and low-proficiency L2 learners process teacher feedback. The present study investigated how Chinese lower-secondary school learners of English as a foreign language (EFL) cognitively and behaviourally engaged with teacher feedback in two conditions: individual written languaging and collaborative oral languaging. Eighty-one students (aged 13–14 years, A1–A2 English proficiency) from two classes and two English teachers at a public lower-secondary school in China participated in this study. Comprehensive teacher feedback (focusing on language errors, content, and organization) was provided to students on three writing tasks completed over six weeks. Findings showed that collaborative processing of teacher feedback elicited students’ deeper cognitive processing, drew their attention to issues beyond linguistic errors and encouraged learner autonomy. On the other hand, individual written languaging promoted students’ noticing of teacher feedback in their languaging process, although with a primary focus on grammar and mechanics. Written languaging also enabled students to act on more teacher feedback points in their revisions than the collaborative processing condition. Pedagogical implications from the comparison of the two feedback processing conditions are discussed.
I Introduction
The debate regarding the effects of written corrective feedback (WCF) on writing accuracy has received substantive scholarly attention in the past two decades. Findings from systematic meta-analyses (Brown et al., 2023; Kang & Han, 2015; Lim & Renandya, 2020; Liu & Brown, 2015) generally support the positive role of WCF in improving second language (L2) learners’ writing accuracy. A more recent shift of research interest focuses not only on how teachers give the most effective feedback but also on how students process or engage with WCF. Zhang and Hyland (2018) argue that students’ engagement with feedback crucially impacts to what extent students retain and internalize the feedback into their interlanguage system. As an umbrella term, engagement with feedback refers to students’ investment and commitment to learning shown in their responses and attitudes to the feedback (see Zhang & Hyland, 2018). Following the tripartite engagement framework conceptualized by Ellis (2010) and later Han and Hyland (2015), cognitive engagement refers to how L2 learners allocate their attentional resources in responding to feedback; behavioural engagement looks at how students incorporate feedback into their revisions, while affective engagement is concerned with students’ attitudes and emotional responses to feedback. Despite this growing research interest in learners’ engagement with WCF, very few studies have explored young and low-proficiency L2 learners’ engagement with teacher feedback, nor compared their engagement with feedback in different conditions, for example, individual vs. collaborative processing (see Y. Kim & Emeliyanova, 2021; Manchón et al., 2020; Mujtaba et al., 2021). The current study examines this under-represented group of L2 learners’ cognitive and behavioural engagement with teacher feedback.
Tools utilized to analyse students’ engagement with teacher feedback in the literature mainly include verbal reports, such as think-aloud protocols, stimulated recall interviews, and semi-structured interviews. More recently, two additional tools, namely written and oral languaging, have been implemented to collect data on feedback processing, and these tools draw on Swain’s (2006) construct of languaging. Swain defined languaging as ‘the process of making meaning and shaping knowledge and experience through language’ (p. 98). Swain et al. (2015) maintain that languaging not only externalizes but also transforms thoughts. In other words, languaging does not just convey existing meaning but, more importantly, makes meaning and brings thoughts into existence, especially when dealing with cognitively challenging tasks.
Languaging can be written or oral. Written languaging when engaging with WCF refers to the externalization of cognitive thoughts, namely the understanding of WCF in the written form (Suzuki, 2012). Research findings have shown that written languaging serves not only as a tool to elicit cognitive thoughts but also facilitates L2 learners’ noticing of feedback and helps to improve L2 learners’ writing accuracy in the revised texts (see Cerezo et al., 2019; Ishikawa & Révész, 2020; Suzuki, 2012, 2017). Oral languaging can be either self-directed or other-directed (collaborative dialogue). Collaborative dialogue enables novice learners to pool linguistic resources in the paired discussion (Storch & Wigglesworth, 2010; Wigglesworth & Storch, 2012b). Empirical evidence suggests that collaborative dialogue/oral languaging enhances learners’ noticing of feedback, especially feedback beyond the word level, and leads to more accurate revisions (see Brooks & Swain, 2009; Manchón et al., 2020; Storch & Wigglesworth, 2010; Wigglesworth & Storch, 2012a, 2012b; Zhang, 2022). However, previous studies mainly reported on the revision quality after students’ collaborative processing of teacher feedback without probing learners’ cognitive depth of processing when students responded to feedback collaboratively. Furthermore, there is inadequate research comparing different feedback processing conditions (individual vs. collaborative) (Y. Kim & Emeliyanova, 2021; Manchón et al., 2020; Mujtaba et al., 2021), and no study has looked into the impact of different modalities of languaging (written vs. oral) on L2 learners’ feedback processing.
Another crucial research gap is that studies reviewed above on learners’ engagement with WCF were all conducted in tertiary-level educational contexts. L2 learners at the secondary school level, who are less cognitively mature and with probably lower L2 proficiency, are under-researched in the literature. The present study addresses this research gap by examining Chinese lower-secondary school EFL learners’ feedback processing. With the emphasis on learner autonomy and student-centred classrooms following the issue of Core Competencies and Values for Chinese Students’ Development (Core Competencies and Values Research Team, 2016) in China, teachers are encouraged to give students more autonomy in the classroom. In response to this pedagogical policy change, it is hoped that comparing the two different feedback processing conditions in this study can help shed light on how to promote the effectiveness of teacher feedback on young EFL learners’ writing and cultivate Chinese lower-secondary school students’ learner autonomy.
II Literature review
This section reviews the theoretical perspectives and related empirical studies in more detail.
1 Theoretical underpinnings
The effectiveness of feedback, as Wigglesworth and Storch (2012b) emphasized, depends on whether L2 learners process and act upon it. The assumption is that the deeper students process the feedback, the greater the gains in short-term feedback uptake and long-term learning. This assumption can be supported by both cognitive perspectives and socio-cultural theory (SCT).
Viewing feedback as a type of input for L2 learners to further process, the cognitive perspective highlights the role of attention and noticing (e.g. Robinson, 1995; Schmidt, 1990). Leow’s (2015) model of the L2 learning process in instructed second language acquisition (SLA) outlines three distinct cognitive processing stages: input processing, intake processing and knowledge processing. Leow (2020) later explains in his feedback processing framework that feedback input needs to be first attended to, detected, or noticed in the input processing stage to become feedback intake that can be further processed. Leow (2015) also underscores the central role of the depth of processing (DoP) during the feedback processing stage, referring to the cognitive effort individuals deploy to process the input via activating prior knowledge, testing hypotheses and forming rules. Thus, from the cognitive point of view, whether the feedback can be further processed and stored in the internal system depends on the amount of attention the L2 learner pays to it, the DoP, and the capacity of the attentional resources (i.e. working memory).
Unlike the cognitive perspective, SCT views that humans’ cognitive development does not reside within the brain but first emerges in social interactions (Lantolf, 2000; Vygotsky, 1978). In other words, learning is not regarded as something an individual does alone. Instead, it is an inherently socially constructed endeavour involving (usually more capable) others scaffolding novice learners within their zone of proximal development (ZPD) (Lantolf, 2000). Vygotsky (1978) defines ZPD as ‘the difference between the learner’s independent problem-solving and the potential development, via solving problems under experts’ guidance or in collaboration with more capable peers’ (p. 85). Feedback is then deemed as ‘a form of assistance’ (Bitchener & Storch, 2016, p. 73), which provides potential scaffolding opportunities, depending on whether such assistance is attuned to the novice’s ZPD. One interesting postulation to enhance the effectiveness of feedback processing is to pool multiple resources of expertise (Brooks & Swain, 2009) in dealing with complex problems, for example, by collaboratively processing teacher feedback in pairs. Storch and Wigglesworth (2010) argue that dyadic collaborative processing enables two novices to pool their linguistic resources and offers students the opportunity to provide scaffolding to each other by responding to the peer’s questions and concerns immediately in their paired discussion (see also Wigglesworth & Storch, 2012b).
In brief, the current study draws on both theoretical perspectives to compare individual and collaborative processing of teacher feedback.
2 Individual processing of teacher feedback
Previous studies on the individual processing of teacher feedback mainly centred on the effects of different feedback types and the quality of students’ revisions and have produced mixed results. Research has shown that different types of feedback elicit different levels of feedback processing from L2 learners (Caras, 2019; H.R. Kim & Bowles, 2019; Shintani & Ellis, 2013). For example, studies by Shintani and Ellis (2013) using eye-tracking and stimulated recall interviews and by Caras (2019) using think-aloud protocols found that metalinguistic feedback elicited deeper cognitive processing than direct corrections. H.R. Kim and Bowles (2019) analysed L2 learners’ think-aloud protocols and found that they were inclined to pay more attention to sentential and paragraph-level issues on receiving reformulation. However, when L2 learners received direct corrections, they had higher DoP for surface-level errors (H.R. Kim & Bowles, 2019). Cerezo et al. (2019) reported that indirect WCF led to more noticing, but direct WCF prompted L2 learners to engage with feedback more deeply in their written languaging. Therefore, feedback with varying degrees of explicitness may elicit different levels of cognitive engagement, particularly when responding to different error types (Caras, 2019; H.R. Kim & Bowles, 2019; Shintani & Ellis, 2013).
A number of studies have also reported L2 learners’ difficulties in processing teacher feedback individually. For instance, Park and Kim (2019), using think-aloud protocols, investigated how 24 Korean university intermediate EFL learners engaged with WCF. Their study found that with indirect WCF, Korean EFL learners tended to either quickly abandon difficult errors (which were not within their ZPD from the SCT perspective) or failed to accurately correct their errors despite spending considerable effort, given their partially developed knowledge of the target language. Zheng and Yu’s (2018) study reported similar concerns. The researchers examined 12 lower-proficiency EFL students’ engagement with comprehensive WCF at a Chinese university. Analysis of students’ immediate oral reports showed that students only superficially noticed indirect WCF and demonstrated a limited understanding of the metalinguistic rules underlying their errors. Zheng and Yu (2018) conclude that low English proficiency may create obstacles for students’ revisions, and more explicit explanations may be needed to help such students understand teacher feedback. However, Cerezo et al.’s (2019) study of how 46 Spanish undergraduates (B1 level) processed direct and indirect feedback (using written languaging), found that regardless of the different feedback types, students seemed unable to understand all feedback.
Previous studies investigating students’ engagement with different types of teacher feedback have mainly used verbal reports rather than written languaging as a data collecting tool. Findings from a few studies utilizing written languaging suggest that it can be applied as both a concurrent data elicitation tool and a facilitator to enhance students’ noticing of feedback and improve their accuracy gains (Ishikawa & Révész, 2020; Suzuki, 2012, 2017). Suzuki (2012, 2017), the first studies to investigate the use of written languaging, involved 24 Japanese university students with intermediate English proficiency. Students were given direct teacher feedback on their first drafts and asked to complete a written languaging task, which required them to write down their understandings of the error and teacher feedback. Findings suggested that written languaging helped improve writing accuracy in the revised texts and stimulated participants to reflect on their errors by noticing with reasons, considered as a higher processing level than mere noticing. Ishikawa and Révész (2020) investigated 82 Japanese university EFL learners’ written languaging in response to focused feedback on the structure of present counterfactual conditional. Their results showed a positive correlation between the frequencies and high DoP evident in written languaging episodes with later accuracy in a grammatical recognition task.
These studies collectively demonstrate the positive effects of written languaging and the difficulties students experience when processing teacher feedback individually. However, due to the lack of research on written languaging in contexts other than university L2 learning settings, whether and how young and low-proficiency L2 learners benefit from processing teacher feedback via written languaging warrants further investigation.
3 Collaborative processing of teacher feedback
The effects of collaborative feedback processing have been primarily investigated in collaborative writing contexts where students form pairs in discussing and responding to the teacher feedback received on their collaboratively written texts. Researchers have reported a number of benefits associated with the collaborative processing of feedback, including contribution to accuracy gains (Storch & Wigglesworth, 2010; Wigglesworth & Storch, 2012a, 2012b; Yang & Zhang, 2010), and enabling students to solve complex linguistic problems (Brooks & Swain, 2009). In particular, Zhang (2022) found that the joint effort of L2 learners revising in groups of three stimulated a higher level of cognitive engagement with comprehensive teacher feedback, on all aspects of writing. The comparison of students’ first and revised drafts showed that students applied rewriting and re-organization extensively. Therefore, a collaborative approach to writing and revising seems to encourage L2 learners’ engagement with teacher feedback beyond the word level by focusing also on idea development and text reconstruction (Zhang, 2022).
Worthy of note is that less research has focused on collaborative processing in individual writing, where students collaboratively discuss the feedback provided on their individually written texts. A few relevant studies on the effects of collaborative processing on students’ accuracy gains in individually revised drafts have yielded encouraging results (Hanjani & Li, 2014; Y. Kim & Emeliyanova, 2021; Mujtaba et al., 2021). Y. Kim and Emeliyanova’s (2021) and Mujtaba et al.’s (2021) studies specifically compared university L2 learners’ individual vs. collaborative processing of indirect teacher feedback. Findings from these two studies suggested that collaborative processing enabled L2 writers to resolve significantly more errors in the revision than individual feedback processing. Mujtaba et al. (2021) further uncovered that the collaborative processing group outperformed students in the individual processing condition regarding the accurate use of verb and word choices. Both Y. Kim and Emeliyanova’s (2021) and Mujtaba et al.’s (2021) studies drew on socio-cultural theory and explained the advantage of joint cognitive effort in collaboratively processing cognitively demanding linguistic errors compared to individually to account for the findings.
While the above studies have suggested the potential effects of collaborative oral languaging on the quality of students’ revisions, Manchón et al.’s (2020) study is the only one to date that has compared the cognitive processing of teacher feedback (direct corrections) in individual and collaborative processing conditions. A total of 118 intermediate Spanish university EFL learners were divided into two feedback processing (individual and collaborative) and two control groups. Students in the two feedback processing conditions were instructed to reflect on their errors by filling out a written languaging form, noting down the original error, the error correction, and explanations for each error. The control groups had students who completed writing either individually or collaboratively and revised their texts without receiving teacher feedback. Manchón et al. (2020) reported no significant difference between the two feedback processing conditions regarding students’ noticed errors, DoP, or the accuracy of the rewritten texts. On the other hand, the control groups left significantly more errors unrevised than the two feedback processing groups. Therefore, the researchers concluded that it was whether students had access to teacher feedback, rather than the individual or collaborative written languaging conditions, that mediated students’ cognitive processing and revision accuracy.
In addition to the scarcity of studies examining L2 learners’ cognitive engagement with feedback in the collaborative processing condition, there are several other aspects of students’ engagement with feedback that necessitate further research. First, no study has compared different modalities of languaging (written vs. oral) in individual and collaborative feedback processing conditions. Many previous studies only looked at students’ engagement with direct or indirect feedback, which may not reflect L2 teachers’ usual feedback practices in real language classrooms (see Furneaux et al., 2007; Lee, 2004), thus affecting the ecological validity of their findings (Liu & Brown, 2015). In addition, the investigation of students’ engagement with WCF has dominated the existing literature, omitting engagement with teacher feedback comments on content and structure (see Mao & Lee, 2023). Importantly, most previous research on learners’ engagement with feedback recruited university L2 learners as the participants, who are more cognitively mature and probably at higher L2 proficiency levels than secondary school EFL learners. It thus remains to be seen how young and low-proficiency EFL learners cognitively and behaviourally engage with teacher feedback individually and collaboratively in written and oral languaging conditions. The current study addresses these research gaps and is guided by the following two research questions.
• Research question 1: How do Chinese lower-secondary school EFL learners cognitively engage with comprehensive teacher feedback individually and collaboratively?
• Research question 2: How do different feedback processing conditions influence students’ behavioural engagement with teacher feedback?
III The study
This study, part of a larger research project, was conducted in China. Participants were eighty-one grade eight students (aged 13–14 years) from two Chinese public lower-secondary school classes. Their English proficiency was A1–A2, according to the Common European Framework of Reference (CEFR). Two experienced English teachers of the two classes were invited to provide comprehensive feedback, namely commenting on language use, content, and organization as they usually do on three writing tasks completed by students over six weeks.
1 Data collection
At the beginning of the school semester (18 weeks), consent was obtained from students, students’ parents, and two English teachers of the two classes. The two classes were then randomly assigned as the individual written languaging condition (41 students) and collaborative processing condition (40 students). Considering that students at this school had no prior experience doing individual written languaging or collaborative oral languaging, training on written languaging and collaborative processing was carried out by the first-named researcher in both classes to familiarize students with the in-class activities they were required to do in the two conditions, as suggested by Y. Kim and Emeliyanova (2021). The training procedures followed Hu’s (2005) four steps designed for peer review training, including awareness raising, demonstrating, practising, and reflection. The training was carried out in two school sessions over two weeks, which was deemed appropriate given the practical time constraints for the larger research project data collection schedule. Specifically, students in the two languaging conditions were introduced to the benefits of individual written/collaborative oral languaging reported in the literature to raise students’ interest and motivation. They were given a demonstration of the two types of languaging in class before languaging about feedback provided on two writing samples from previous students. Students then had the opportunity to raise questions and concerns at the end of the two training sessions. The formal data collection started in week five and lasted for six weeks.
Students in both conditions were required to write two drafts of three short texts (70–100 words). These texts formed part of the regular class work but were not formally assessed (see the example writing task in Appendix A). The first draft was completed in a test-like condition in class where students were not allowed to use dictionaries or consult textbooks or notebooks. Teachers then provided comprehensive feedback and returned the drafts with the feedback to students before the second session. Comprehensive teacher feedback commenting on most/all errors in students’ writing is what learners usually receive in this instructional context, and hence this study maintained high ecological validity by not intervening in how teachers provide feedback on students’ writing.
In the following week, students in the individual processing condition were given the written languaging sheets to reflect on their errors individually. Following Suzuki (2012), the written languaging sheet included a few questions (written in Chinese): ‘Why is this linguistic form incorrect/wrong? Why did the teacher give feedback on this form? Please write your understanding and explanations in Chinese or English.’ The sheet attempted to elicit the students’ cognitive thoughts while processing teacher feedback individually. Students in the collaborative processing condition had to discuss the teacher feedback they received on their individually written texts in self-selected pairs. Student pairs needed to swap their roles as feedback receivers and co-processors as they discussed the feedback on each of their texts together. At the beginning of each treatment session, the first-named researcher set up the task, and then the English teacher gave some general oral comments as what they usually do on returning written feedback to students. After that, it should be noted that students in the two conditions then had equal time (20–25 minutes) to complete either individual written or collaborative oral languaging of written teacher feedback in class. Four randomly selected pairs’ discussions were recorded in each collaborative languaging session. After collaboratively processing the feedback, students were asked to revise their first drafts after class and submit them within the same week. The research procedures are visualized in Figure 1. These procedures were repeated twice in the subsequent weeks.
Research procedures.
A summary of data collected in this study can be found in Table 1. To analyse students’ cognitive engagement with teacher feedback, 123 written languaging sheets were collected from 41 students over three writing tasks in the individual processing condition. In the collaborative processing condition, 12 audio recordings over the three writing tasks were collected from 24 students. To analyse students’ behavioural engagement with teacher feedback, all 81 students’ first and revised drafts in both conditions were collected in all three writing tasks, totalling 486 writing drafts.
Data collection.
2 Data analysis
The basic analytic unit was languaging episodes, including written languaging episodes (WLEs) and oral languaging episodes. The definition of a languaging episode followed the concept of language-related episodes (LREs), meaning an instance where learners deliberate ‘about their use of language, question their language use or correct themselves or others’ (Swain & Lapkin, 1998, p. 326). Since comprehensive teacher feedback in this research included not only feedback on linguistic errors, a languaging episode in this study refers to students’ deliberation about their language use, content and organization of their writing or verbalization of the revisions they made to their written text when processing the teacher feedback (see also Suzuki, 2012, 2017). WLEs were analysed by importing students’ written languaging sheets into NVivo (QSR International Pty Ltd., 2022). Verbatim transcriptions of all the audio recordings of students’ collaborative dialogues in the collaborative processing conditions were also analysed on NVivo and separated into different oral languaging episodes (see transcription conventions in Appendix B). The examination of the data showed that the triggers for oral languaging episodes were not solely teacher feedback points but also valid peer feedback and self-initiated feedback, as well as occasions where students sought clarification or made confirmation checks to each other.
To analyse students’ cognitive engagement with teacher feedback, the written and oral languaging episodes were firstly analysed for their languaging coverage, namely what proportion of feedback points was covered in their languaging episodes and for the foci of these episodes (Suzuki, 2012; Williams, 1999). WLEs were coded as vocabulary-WLEs, grammar-WLEs, mechanics-WLEs, content-WLEs, and organization-WLEs. Oral languaging episodes were coded as language-related episodes, including vocabulary-LREs, grammar-LREs, mechanics-LREs, and content-related (CREs) and organization-related episodes (OREs).
Students’ cognitive engagement with teacher feedback was then operationalized as the amount of cognitive effort students spent, for instance, whether they drew on prior knowledge, formed or tested hypotheses, in responding to teacher feedback drawing on the concept of DoP (Leow, 2015). The detailed coding scheme for DoP built on Cerezo et al.’s (2019) five DoP levels and was informed by an iterative process of applying the rules of different DoP levels to the cognitive efforts shown in students’ languaging episodes. The full descriptors for low DoP, medium DoP, and high DoP can be found in Table 2.
Depth of processing coding rules.
Three coding examples for the three levels of DoP are provided in Table 3. A low DoP was coded when the learner only repeated/restated the teacher feedback provided, which suggested the student’s mere noticing of the error/feedback. When the error category (e.g. identifying the subject of the sentence as shown in the example) or translation of the error/feedback was provided, a medium DoP was labelled because it showed the learner’s attempted analysis of the feedback. Finally, a languaging episode with high DoP represented instances where the learner provided metalinguistic explanations or possible solutions to the specific error. As the example of high DoP shows, a direct correction was provided in brackets, and the subsequent WLE demonstrated Wang’s cognitive effort in analysing the parts of speech by drawing on her prior metalinguistic knowledge. She also provided an alternative revision to the error instead of only repeating the teacher feedback. Therefore, this episode was coded as showing high DoP.
Depth of processing coding examples.
Notes. DoP = depth of processing. WLE = written languaging episodes.
Since the collaborative processing involved two students at the same time, the DoP shown in the oral languaging episodes was firstly coded based on each individual’s DoP. As a result, six types of DoP combinations were identified: Low-Low DoP (L-L), Medium-Low DoP (M-L), Medium-Medium DoP (M-M), High-Low DoP (H-L), High-Medium DoP (H-M), and High-High DoP (H-H). In the second stage of the DoP analysis, each pair’s overall DoP was coded (see Table 4). Students’ overall cognitive efforts as a pair were coded according to either the same DoP level reached by both students (e.g. Low-Low was coded as an overall low DoP) or the higher level that either of the two students in a pair showed (e.g. an overall medium DoP for Medium-Low). Coding each pair’s overall DoP adopted the SCT perspective of how collaborative processing of teacher feedback allowed both participants in a pair to pool their linguistic resources. The outcome and effectiveness of a pair’s discussion consider both interlocutors’ efforts instead of merely focusing on individual contributions to the discussion. Only by examining a pair’s overall DoP in responding to teacher feedback is their collaborative effort reflected.
Collaborative processing individual and overall depth of processing (DoP).
The analysis of students’ behavioural engagement with teacher feedback looked at students’ revision behaviours, revision outcomes and revision operations in the two conditions. Revision behaviours focused on analysing whether students took actions to revise the error or not, distinguishing between acted-on and unchanged errors. An interesting third category also emerged where students did not directly act upon the commented error but instead deleted the error completely or rewrote the whole sentence without addressing the initial error. This third category was coded as circumvention, including subcategories deletion only and deletion + rewriting. Two examples of circumvention can be found below.
Example 1. Circumvention by deletion only
Original sentence with teacher correction: Doing chores can make our home (very) clean very much.
Revision: Doing chores can make our home clean.
Example 2. Circumvention by deletion + rewriting
Original sentence with teacher correction: Here
Revision: I have some advice for you.
For revision outcomes, the codes successful and unsuccessful were used to categorize students’ revisions in response to feedback on language errors. Revisions for feedback on content and organization were coded as appropriate or inappropriate according to whether the revision reasonably addressed the feedback. Revision operations focused on analysing students’ detailed operations taken in response to feedback on content and organization, for example, addition, substitution, and re-organization. Coding categories for revision operations were adapted from Zhang and Hyland (2018, 2022) (see Appendix C).
IV Results
Findings for students’ cognitive and behavioural engagement with teacher feedback are presented in this section. Comparison of oral and written languaging coverage, foci, and DoP are illustrated first, followed by students’ revision behaviours, outcomes, and operations in the individual and collaborative processing conditions.
1 Cognitive engagement: Languaging coverage, foci, and DoP
The 123 writing scripts from 41 students in the individual written languaging condition received a total of 1,015 teacher feedback points, while the 24 scripts written by the 12 focused pairs in the collaborative processing condition had 201 teacher feedback points in total (see Table 5). The analysis of students’ coverage of the feedback points in their languaging episodes found that individual processing elicited a greater feedback coverage than the collaborative processing condition (85.6% vs. 75.1%, see Table 5). Chi-square test results suggested that the difference between the two conditions was significant with a small effect size: (1, n = 1216) = 13.66, p < .001, ϕ = .11. Students in the individual written languaging condition covered significantly more feedback points on grammar with a small effect size: (1, n = 809) = 25.15, p < .001, ϕ = .18, and mechanics with a medium to large effect size: (1, n = 141) = 31.64, p < .001, ϕ = .47. In contrast, the collaborative processing condition prompted students to cover significantly more feedback on content with a small to medium effect size: (1, n = 110) = 7.16, p = .007, ϕ = −.26. Coverage for lexis and organization did not show statistically significant variations between the two conditions (lexis: (1, n = 66) = .080, p = .78, ϕ = −.021; organization: (1, n = 90) = 3.62, p = .057, ϕ = −.20).
Languaging coverage and foci.
The collaborative processing condition promoted students’ deliberations about issues beyond those receiving teacher feedback. Table 6 summarizes the distributions of the triggers for oral languaging in pairs. The 151 teacher feedback points covered in students’ collaborative oral languaging, generated 158 oral languaging episodes (there were occasions where students discussed the same feedback point in multiple episodes), accounting for 78.6% of the total number of oral languaging episodes. Notably, more than 20% of the oral languaging episodes were elicited by peer feedback, clarification re-questions, confirmation checks, and self-initiated feedback. In contrast, in the individual written languaging condition, 100% of the episodes focused on processing the feedback provided by the teacher.
Triggers of collaborative oral languaging episodes.
Regarding the level of cognitive engagement (see Table 7), this study found that students demonstrated high DoP in significantly more collaborative oral languaging episodes than in individual written languaging episodes with a small effect size: (1, n = 1070) = 4.74, p = .029, ϕ = .07. Furthermore, participants in the collaborative processing condition also produced significantly fewer low DoP oral languaging episodes compared to those in the individual written languaging condition with a small effect size: (1, n = 1070) = 13.10, p < .001, ϕ = .11.
Languaging depth of processing (DoP) comparison (percentages in parentheses).
The analysis of students’ individual DoP in the collaborative processing episodes suggested that students tended to spend different amounts of cognitive effort when they assumed different roles, namely the feedback receivers and the feedback co-processors who helped process the feedback received by their peers. Figure 2 compares students’ individual DoP in the two roles. When learners were in the role of a feedback co-processor looking at the feedback their peers received, they demonstrated more instances of high DoP than when they were looking at their own texts (44.3% vs. 24.7%). Chi-square tests found that the difference was statistically significant with a small to medium effect size: (1, n = 316) = 13.46, p < .001, ϕ = −.21. Conversely, the actual feedback receivers showed low DoP in nearly half of the languaging episodes (47.4%), which was significantly more than their feedback co-processors with a small effect size: (1, n = 316) = 8.27, p = .004, ϕ = .16.
Collaborative processing roles and individual depth of processing (DoP).
Example 3 gives an example of a passive feedback receiver in a collaborative oral languaging episode. The episode came from the beginning of Guo and Yuan’s (pseudonyms) pair talk. Guo (feedback co-processor) initiated the question about why the teacher changed the word form in turn 1. Then from turn 2 to turn 7, Yuan (feedback receiver) emphasized that Guo should be responsible for explaining the teacher feedback to her, although Guo tried to discuss it with Yuan. It may be that when collaboratively processing teacher feedback, students regarded their interlocutor as a peer tutor to help them understand the teacher feedback, so they tended to wait for their peers to decode the meaning of the feedback, while they assumed relatively passive roles in the discussion. This exemplifies feedback receivers’ lower level of cognitive processing than their co-processors’ DoP shown in Figure 2.
Example 3. A passive feedback receiver
Original sentence with teacher correction: Our school is holding all kinds of volunteer Guo (feedback co-processor): 你这个单词为什么要把 Yuan (feedback receiver): 是你要跟我讲 [You should tell me the reason.] Guo: 对啊我就跟你讲诶 [Yeah, I am discussing with you] Yuan: 为什么啊 [So why?] Guo: 我就问你诶，为什么 [I am asking you, what do you think?] Yuan: 你跟我讲 [You are supposed to explain this to me] Guo: 哈啊哈，要加复数形式嘛，变
2 Behavioural engagement: Revision behaviours, outcomes, and operations
The analysis of students’ revision behaviours found that participants in both conditions acted on most teacher feedback points, 91.5% and 84.3% for the written languaging and collaborative processing conditions, respectively (see Figure 3). Despite involving another peer in the feedback processing, students in the collaborative processing condition responded to less feedback than their counterparts who processed teacher feedback individually via written languaging, and the difference was statistically significant with a small effect size: (1, n = 1673) = 20.65, p < .001, ϕ = .11. Furthermore, it was found that collaborative oral languaging made students circumvent significantly more errors than the individual written languaging condition with a small effect size: (1, n = 1673) = 16.74, p < .001, ϕ = −.10.
Revision behaviours.
A closer look at the proportion of feedback points circumvented by students showed that it included two categories: deletion only and deletion + rewriting. Zhang (2022) observes that rewriting as a revision operation usually involves higher levels of deliberation about the texts, which presumably requires deeper cognitive engagement. However, and has been noted by one of the reviewers, rewriting may not always lead to better revised draft compared to deleting only. Figure 4 shows that while over 82% of the written languaging circumventions fell into the category of deletion only, where students only deleted the whole problematic sentence, nearly a third of the circumventions in the collaborative processing condition went further and involved a rewrite of the original sentence. The Chi-square test results suggested that the difference in occurrences of rewriting was not statistically significant between the two conditions: (1, n = 161) = 3.68, p = .055, ϕ = .15.
Circumventions.
Regarding revision outcomes, successful and unsuccessful revisions in response to feedback on language errors and the appropriateness of the revisions addressing issues on content and organization were coded. This study found that both conditions had high accuracy rates (more than 90%) in students’ revised drafts in response to feedback on language, content and organization (see Figure 5). Collaborative processing led to a significantly higher rate of appropriate revision following feedback on content and organization with a small effect size: (1, n = 234) = 5.57, p = .018, ϕ = −.15. However, no significant difference was found between the two conditions’ revision outcomes in response to feedback on language errors: (1, n = 1473) = 1.89, p = .169, ϕ = −.04.
Revision outcomes.
Figure 6 summarizes the different revision operations students took when addressing feedback on content and organization in the two conditions. The bar chart shows that both conditions demonstrated generally similar revision operations. In more than half of the revisions in both conditions, students opted for ‘addition’, while ‘re-organization’ was the second most frequent operation.
Revision operations.
V Discussion
Table 8 summarizes the main findings in response to the two research questions, regarding these secondary school students’ cognitive and behavioural engagement with teacher feedback found in the individual and collaborative processing conditions. The positive and negative effects of the two conditions on students’ engagement with teacher feedback are intertwined and discussed in relation to previous empirical studies and theoretical underpinnings: cognitive perspectives and SCT.
Written languaging and collaborative processing comparison.
Notes. DoP = depth of processing.
This study found that in the individual processing condition, written languaging was an engaging and motivating task because students covered most of the teacher feedback points in their WLEs (cognitive engagement) and responded to most of them in the revised texts (behavioural engagement). These positive effects of written languaging aligned with what was found by Suzuki (2017) and Ishikawa and Révész (2020). Nonetheless, students’ greater percentages of feedback coverage and acted-on feedback points in the individual processing condition compared to the collaborative processing condition identified in this study are not consistent with previous research findings (Y. Kim & Emeliyanova, 2021; Manchón et al., 2020; Mujtaba et al., 2021). While both studies by Y. Kim and Emeliyanova (2021) and Mujtaba et al. (2021) reported that the collaborative processing group resolved more and abandoned fewer errors than the individual processing group, Manchón et al. (2020) found no significant difference between the individual and collaborative WCF processing conditions in terms of the percentages of errors noticed.
The discrepancy between the current study and previous research may be ascribed to the facilitating effects of written languaging on students’ individual processing of teacher feedback. Previous empirical studies (Ishikawa, 2018; Ishikawa & Révész, 2020; Manchón et al., 2020; Moradian et al., 2017; Suzuki, 2012, 2017) have shown that written languaging can enhance L2 learners’ noticing of feedback. However, participants in Y. Kim and Emeliyanova’s (2021) and Mujtaba et al.’s (2021) studies were not asked to engage with written languaging in the individual processing condition. Therefore, by requiring students to externalize their cognitive processing of the teacher feedback via written languaging, students in the present study might have been able to language about more errors and act on more feedback points than if they had not been asked to complete the written languaging task.
In addition to the individual vs. collaborative processing conditions, the nature of the errors, along with the different modalities of the feedback processing in the two conditions, namely written vs. oral languaging, might have also influenced students’ engagement with teacher feedback. While participants in Manchón et al.’s (2020) study did written languaging in both individual and collaborative feedback processing conditions, the current study had students engage with written vs. oral languaging tasks. With the explicit instruction to write down the reflections for each feedback point, the written languaging condition might have made students feel more motivated and committed to addressing each feedback point (Manchón et al., 2020). Students in the collaborative oral languaging condition, on the other hand, were likely to feel less obliged to discuss each feedback point, especially in response to direct corrections (which accounted for more than 75% of all feedback) on some overt errors. As seen in Table 5 above, only 50% of mechanical problems were discussed in the oral languaging episodes, which was the least frequently deliberated feedback focus, while up to 94% of mechanical issues were covered in the WLEs (the category with the highest coverage). Mechanics-related LREs were also found to receive the least attention in the collaborative feedback processing when students wrote collaboratively (e.g. Fernández Dobao, 2012; Storch, 2007). Hence, differences between the two conditions may be attributed to the languaging modalities and error types.
One important practical concern that could have affected students’ feedback coverage was the time constraints given by the school timetable, especially for the collaborative processing condition. Previous studies (Y. Kim & Emeliyanova, 2021; Mujtaba et al., 2021) gave participants in the collaborative processing condition double the time as learners in the individual processing used. Students in the current study had equal time (20–25 minutes) to complete their individual or collaborative processing after receiving teacher feedback. However, as students in the collaborative processing condition needed to talk through two pieces of writing within the same time as their individual processing counterparts, it is possible that some pairs may have run out of time, so they did not manage to cover every feedback point in their discussion, which led to the lower feedback coverage in the collaborative processing condition. This also corroborates what was reported in studies comparing individual and collaborative writing: collaboration usually takes longer than individual work (Storch, 2005).
Despite the smaller proportion of feedback coverage in the collaborative processing condition, students demonstrated high DoP in significantly more languaging episodes, namely showing more extensive cognitive engagement when working collaboratively than individually. This finding can be explained from the SCT perspective of learning. Students in both conditions received comprehensive teacher feedback, which potentially provided them with expert-novice scaffolding, depending on whether the feedback contingently responded to students’ needs and whether it fell within the particular student’s development level (ZPD) (Lantolf et al., 2015). The difference between the two conditions was that the collaborative oral languaging allowed students to discuss their concerns with another peer after receiving comprehensive teacher feedback. This task helped to form an integrated form of scaffolding opportunities from both the expert and novice, namely the (expert + novice) – novice scaffolding. A higher percentage of the high DoP found in the collaborative processing condition compared to the individual processing indicated that the co-construction of ZPD between pairs in processing teacher feedback collaboratively may have expanded students’ overall ZPD capacity. This expansion of ZPD enabled students to become more capable of understanding teacher feedback (see also Mujtaba et al., 2021). In contrast, a higher proportion of low DoP shown in students’ WLEs suggested that students had more difficulties processing feedback individually, because some feedback was more likely to fall beyond their individual ZPD than the co-constructed ZPD in the collaborative processing condition.
Although a higher proportion of high DoP was found in the collaborative processing condition, it was observed that students in both conditions were not engaging deeply with all feedback points provided, as around half of their languaging episodes only demonstrated medium or low DoP. This may be attributed to the difficulties for lower proficiency L2 learners to fully understand teacher feedback, as noted in previous studies (e.g. Cerezo et al., 2019; Park & Kim, 2019; Suzuki, 2017; Zheng & Yu, 2018). These school-age learners’ lack of familiarity with the individual written and collaborative oral languaging tasks employed in this research might have also affected their cognitive engagement with feedback, since they had little experience in processing teacher feedback via written or oral languaging in their traditional English classrooms. Previous research has shown that when facing unfamiliar tasks, adolescent L2 learners can be disadvantaged and thus not perform well (see Pinter, 2014; Vandergrift, 2005).
Another plausible explanation for the proportions of low DoP found in this study is that verbalizing cognitive processes may be challenging for adolescent L2 learners. In other words, students could not verbalize explanations or grammar rules, even though their attention was drawn to the form. Similar difficulties in justifying revisions in response to teacher feedback were also observed among adult Chinese university EFL learners by Man et al. (2021). Young L2 learners, in particular, have been reported to experience difficulties verbalizing thoughts and providing introspective data (Gu et al., 2005; Pinter, 2014; Pinter & Zandian, 2015). Therefore, it is possible that deeper cognitive processing did occur, but these teenage students were unable to externalize it in the written or oral form (see also Suzuki, 2017). It is also possible that when facing novel verbalization tasks, L2 learners need more time to gain familiarity with the task (Man et al., 2021). Previous studies have shown that L2 writers tend to draw on intuitions over rules in response to error corrections, so they seldom articulate the underlying metalinguistic knowledge (see also Armengol & Cots, 2009; Ferris et al., 2013; Y. Kim & Emeliyanova, 2021; Polio, 2012; van Weijen et al., 2009).
Regarding what students focused on in their languaging episodes, this study found that Chinese lower-secondary school EFL learners were inclined to orally discuss more feedback points on content in the collaborative processing condition, while learners doing written languaging appeared to focus predominantly on grammar and mechanics. Students’ revised texts in the collaborative processing condition showed more rewriting and more appropriate revisions in response to feedback on content and organization. Similarly, Zhang (2022) also found that collaborative revising prompted Chinese university students’ more profound engagement with feedback beyond the word level by generating and re-organizing ideas, which were more meaning-focused. The importance of learners’ focus on content and its potential contribution to improved writing performance has been highlighted in studies on collaborative writing (Abrams, 2019) and peer feedback (Hu & Lam, 2010; Min, 2006). Hence, students’ greater focus on meaning instead of mere linguistic errors when processing feedback collaboratively found in the current study may potentially better facilitate L2 writing development.
Possible explanations for the different languaging foci in the two conditions can be found in the cognitivists’ perspective of feedback processing. Given that young adolescent and low English language proficiency EFL learners tend to have more limited attentional resources (Williams, 2001), comprehensive teacher feedback provided to students in both conditions was likely to pose a high demand on their cognitive load (see also Shintani et al., 2014; van Beuningen et al., 2012). Learners’ predominant focus on grammatical and mechanical issues in the written languaging condition might have drawn their attention away from other writing aspects, for example, idea development (see also Coyle & Larios, 2021; Polio, 2012). On the other hand, collaborative processing may have played a crucial role in reducing this overload (see also Mujtaba et al., 2021) and enabled students to discuss higher-order issues like the content of their writing with peers. Brooks and Swain (2009) suggest that more sources of expertise are needed when the problem learners face is more demanding. Therefore, the opportunity to discuss comprehensive teacher feedback with peers distributes and alleviates the cognitive burden and enables the learners to better allocate their attentional resources to different aspects of writing. This greater cognitive capacity in the collaborative processing pairs further helped them to focus on issues beyond teacher feedback, namely generating peer feedback, self-initiated feedback, clarification requests and confirmation checks. Since the collaborative processing condition had languaging episodes triggered by feedback other than teacher feedback, it suggests that discussing feedback in dyads might boost students’ learner autonomy in their pair work.
Albeit the students’ different levels of cognitive engagement with feedback in the two feedback processing conditions shown in their feedback coverage, languaging foci, and languaging DoP, no significant difference in the revision outcomes (behavioural engagement) in response to feedback on language errors was found. One possible reason is that higher percentages of higher DoP episodes do not equal higher accuracy in revisions. The greater languaging DoP found in the collaborative processing condition did not mean that students were more able to resolve errors accurately. This is because, as the DoP coding scheme suggested, languaging episodes with high DoP only meant that students demonstrated a tremendous cognitive effort in processing and responding to the feedback, not necessarily leading to accurate interpretations or responses to feedback.
This result does not align with what was reported by Y. Kim and Emeliyanova (2021) and Mujtaba et al. (2021) who found that in comparison to individual processing, collaborative processing of feedback contributed to significantly more accurately resolved errors in the revised drafts. One possible explanation for the differences in the findings between the current research and previous studies is the participants’ level of language proficiency. Chinese lower-secondary school EFL learners involved in the present study had low proficiency, whereas the university ESL/EFL students in Y. Kim and Emeliyanova (2021) and Mujtaba et al. (2021) had intermediate proficiency levels. As Shehadeh (2011) notes in the collaborative writing setting, lower-proficiency L2 learners lack the linguistic resources to assist each other, and hence they may benefit less from collaboration.
Another issue worth noting is that in the collaborative processing condition, it was observed that the feedback co-processors, rather than the writers, seemed to have assumed more responsibilities in explaining what the teacher feedback meant and how to resolve the problem. In contrast, the actual feedback receivers played more passive roles in the discussion. As Y. Kim and Emeliyanova (2021) caution, learners’ writing performance may be largely influenced by the interaction patterns they form during the collaborative discussion of teacher feedback. The writers’ non-collaborative orientation (Storch, 2002) may have compromised the potential (expert + novice) - novice scaffolding advantages in expanding student pairs’ co-constructed ZPD in the collaborative processing condition over the expert-novice scaffolding formed in the individual written languaging condition. For this reason, revision outcomes between the two processing conditions did not show significant differences.
Finally, the teachers in this study provided mostly direct teacher feedback on all students’ first drafts, which was not finely attuned to learners’ developmental levels, and this may also partially account for the similar revision outcomes found in the two conditions. In studies conducted by Y. Kim and Emeliyanova (2021) and Mujtaba et al. (2021), participants received indirect feedback (editing symbols), which required students to resolve the problems during the languaging process. In response to the mostly direct provision of teacher feedback, learners in the present study could superficially or procedurally engage with direct feedback (Nystrand & Gamoran, 1991) by incorporating all overt corrections without any deep or accurate understanding of them. Consequently, students’ writing development may benefit less from direct error corrections.
VI Conclusion, caveats, and implications
This study examined how young and low-proficiency EFL learners cognitively and behaviourally engage with teacher feedback individually via written languaging and collaboratively through pair talk. Findings revealed that individual written languaging enhanced students’ noticing of teacher feedback as they covered more feedback points in their WLEs and acted on a greater proportion of feedback in the revisions. However, collaborative oral languaging promoted deeper cognitive processing of teacher feedback. It encouraged students to pay attention to more global aspects of their writing instead of predominantly focusing on grammatical and mechanical issues. Collaborative processing also helped to boost students’ autonomy by triggering oral languaging of issues beyond the teacher feedback they received.
Admittedly, this study has a number of limitations. It did not adopt a counter-balanced research design or include a control group due to the practical constraints associated with conducting a study in a Chinese lower-secondary school. Equal in-class time was given to the two conditions, which might have disadvantaged students in the collaborative oral languaging condition, because they had more limited time to deliberate about the teacher feedback given on two pieces of writing. Another limitation of the current study relates to the short-term languaging training in the two classes. Students seemed to experience difficulties verbalizing their cognition, which may have limited the available evidence of the cognitive effort they spent in the languaging process. Future studies should attempt to address these limitations, by including more training sessions to provide more empirical evidence regarding young L2 learners’ cognitive processing of teacher feedback in different languaging conditions. Further research comparing the long-term effects of individual vs. collaborative processing conditions on L2 writing development and the perceived benefits of different feedback processing conditions from teachers’ and students’ perspectives is also deemed necessary.
Findings from this study can shed light on L2 writing instruction in similar contexts. Both written languaging and collaborative processing can be utilized as effective pedagogical activities to facilitate young and low-proficiency L2 learners’ engagement with teacher feedback. Written languaging may be helpful in encouraging students’ active processing and responses to teacher feedback, with a primary focus on language use. Collaborative processing, on the other hand, may help students process feedback at a deeper level, better boost students’ learner autonomy, and stimulate deliberations on writing issues beyond linguistic errors. L2 writing teachers are thus recommended to choose appropriate feedback processing activities according to their different teaching objectives.
Footnotes
Appendix A
Appendix B
Appendix C
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The author(s) received the Graduate Research Publication Grant from the School of Languages and Linguistics at the University of Melbourne for the publication of this article.
