The Collaborative Discussion Model: Developing Writing Skills through Prewriting Discussion

This study aims to investigate the effect of peer-assisted prewriting discussion on second language (L2) academic writing and its benefi ts for students with different profi ciency levels. While there is a signifi cant body of research exploring the positive impact of collaboration on L2 writers’ written performance and the ways it could be organised, there is little practical consideration on how to formulate explicit instruction. The rationale for this research lies in designing and arranging explicit instruction that could lead to L2 learners producing a higher quality writing output. Based on both qualitative and quantitative methods, and drawn on students’ written texts and data analysis, the current study was conducted to devise and test a proposed model, which the author will term the ‘collaborative discussion model’ (the CDM). The control and experimental groups of Russian EFL students (n = 48) were engaged in written assignments after naturally occurring discussions and then the latter group was involved in an instructor-led discussion. The practice writing tasks were rated with the analytic rubric used in IELTS, assessing task response, coherence and cohesion, lexical resource, and grammatical range. The fi ndings suggest that collaborative prewriting tasks, accomplished in the experimental group of students with different levels of L2 profi ciency, may encourage students to engage more in refl ection about the content and language of the text. As the texts produced after introducing the CDM were scored higher, especially on the criteria of task response and lexical resource, it is suggested that scaffolding prewriting discussions can potentially augment the writing skills of learners and the CDM can be used as a complementary activity to address the challenges associated with academic writing. The results of the questionnaire can imply that there are benefi ts of explicit instruction for students with different levels of L2 profi ciency, although in nuanced ways and different degrees.


Introduction
Academic writing purports to be an integral aspect of students' academic and professional training across the disciplines in the majority of higher institutions all over the world. University curricula are intensively using academic writing in English as a second language as a compulsory component of students' education. By and large, writing is seen as a complex language-focused process that consists of multiple interactive stages based on strategic functions (e.g., idea generation, outlining, drafting, revising, and editing) and, thus, requires from the user a range of embedded skills that must be combined in order to contribute to high-quality output. Not only does it embrace language-related elements, but it represents a complex social and cognitive process accompanied by an emotional component that affects learners' motivation and engagement (TEAL Center, 2012).According to the Common European Reference Framework for languages (CEFR), an advanced learner can write clear, well-structured expositions on complex subjects, underlining the relevant salient issues, expand and support points of view at some length with subsidiary points, reasons, and relevant examples (Council of Europe, 2011). Thus, academic English teaching and learning at the tertiary education level can be acknowledged as one of the most challenging tasks for both teachers and students.
In order to facilitate the process of academic writing acquisition, classroom practitioners have started to incorporate a variety of collaborative activities based on the sociocultural theory positing that language and cognitive skills develop through interactions with others. In the multifaceted process of academic writing, THE COLLABORATIVE DISCUSSION MODEL collaborative learning classroom practices, with learners interacting in order to solve academic problems, have manifested a number of benefits. Besides enhancing the overall writing skills of learners, such activities can help them become more flexible, adapt to various language contexts and studying environments, and synergise literacy development in both speaking and writing in conjunction with listening and reading (Fernandez Dobao, 2012;Kim, 2008;Neumann & McDonough, 2014, 2015. However, it seems necessary to pinpoint that although collaborative activities appear to be a widely utilised instructional technique in the learning environment, an array of factors can circumscribe the desired effect of collaboration. Not all interactions can be implicitly effective in achieving the goals of mastering a variety of linguistic features (Chen & Yu, 2019;McDonough & De Vleeschauwerb, 2019;Rahimpour et al, 2011;Zambrano et al, 2019). In content-focused collaborations, for example, language forms necessary for the language acquisition can be undervalued and overlooked while learners should be encouraged to reflect on their language usage, such as lexical choice, grammar, mechanics, etc. (Swain, 2010;Swain & Lapkin, 2001;Yazdi Amirkhiz et al, 2013). Therefore, this limitation necessitates designing and arranging collaborative classroom activities that could lead to L2 learners producing high-quality writing output. Some researchers have emphasised the vital importance of explicit instruction to scaffold developing writers in the process of planning, drafting, and revising the text as well as to boost their self-efficacy and motivation for writing (De Smedt et al, 2020;McDonough & De Vleeschauwerb, 2019). According to Mascolo (2009), it seems a rather challenging task for the teacher to devise and then apply in the classroom the model of instruction with pedagogically meaningful aims. It is also paramount to organise effective collaborative groups so that both learners of lower and higher proficiency levels could benefit from collaboration (Zambrano et al, 2019).
The present study corroborates and extends the body of research on collaborative prewriting discussions by supplementing it with a proposed instructional model, which the author will term 'the collaborative discussion model' (the CDM). It can scaffold learners while they acquire needed writing strategies and skills as well as confidence and positive motivation for writing. The main aim of the study is to investigate the effect of peerassisted collaboration scaffolded by the CDM on Russian students. Based on both qualitative and quantitative methods, and drawn from the students' written texts and data analysis, this study addresses the following research questions: (1) How does the proposed CDM affect the writing performance of Russian students?
(2) What stages of the discussion based on the CDM, if any, contribute the most to high-quality writing output?
(3) Do students with higher and lower L2 proficiency levels benefit equally from collaboration?

Literature Review
To date, scaffolding collaborative discussions as a prewriting strategy has been substantially implemented and studied in the field of academic writing teaching. Guided participation of the learners in the process of language acquisition implies not only their active involvement but interactive participation, which means that the learner appropriates skills and constructs knowledge from cooperation with the teacher. Only through experience conflicting with their own comprehension and in conflicting situations can a learner fill in the gaps in their knowledge (Mascolo, 2009). Recent studies have analysed effective educational situations when learners can support each other through actively partaking in the process of critical thinking during the discussion, with some of them being assigned the role of being more knowledgeable others (Kim, 2008(Kim, , 2009Leeser, 2004;Swain & Lapkin, 2001;Suzuki & Itagaki, 2007;Watanabe & Swain, 2007. While communicative tasks are highly valued as they provide opportunities to communicate in the L2, a host of prerequisites should be taken into consideration to make interactions more successful. Effective collaboration can be administered in deliberately developed collaborative groups. In recent research, there has been a narrower focus on how scaffolding occurs when learners collaborate, in dyads, triads, or other small-size groups, in order to co-construct written texts (Di Camilla & Anton, 2012;Edstrom, 2015;Storch, 2005;Wigglesworth & Storch, 2009, 2012, where the benefits of such collaboration were well-documented, providing persuasive arguments in favour of this. Fernandez Dobao (2014) compared pairs and groups of four learners and found that the latter were more successful at solving language-related problems. Zambrano et al. (2019) concluded that learners can take advantage of collaborative work when they receive guidance on how to collaborate but not when collaborative groups develop naturally. Previous research has also indicated that it is particularly explicit instruction, organized in various ways, that translates into better learner performance after peer-assisted collaboration (De Smedt et al., 2020;McDonough & De Vleeschauwer, 2019;Rahimpour et al., 2011;Zambrano et al., 2019). In a study by Ellis and Yuan (2003), the effects of different types of prewriting conditions, namely pre-task planning as a form of instruction, unpressured on-line planning, and no planning, on the writing performance of Chinese learners were examined, with pre-task planning resulting in greater fluency and syntactic variety of the output. Another study to report is the one by McDonough and Neumann (2014), who found that when students were given explicit instructions and visual tools facilitative of brainstorming ideas during preparation for writing they had more time to focus on linguistic features while writing. Teng and Zhang (2019) went further and proved that self-regulated learning strategies-based writing instruction contributed to increased levels of both linguistic and performance selfefficacy in students.
In order to elaborate on the findings of the above studies, it is necessary to highlight how collaborative discussions are organized. They are mainly presented in the form of 'languaging', described by Swain as 'the process of making meaning and shaping knowledge and experience through language' (Swain, 2006, p. 96). That means that while discussing the task, learners produce meaningful and comprehensible output in terms of the language. The concept of languaging is usually implemented into practice through language-related episodes (LREs), defined as segments of collaborative dialogue where the students reflect on what language forms to use (lexical choices, grammatical constructions, and mechanics) or correct themselves and others while performing the task (Swain & Lapkin, 1998, 2001Yazdi Amirkhiz et al, 2013). It has been concluded that the participants' higher proficiency levels result in a greater number of LREs produced in the course of collaboration (Williams, 2001(Williams, , 2008. Consequently, the L2 proficiency levels of learners and the way how they are organised in groups can be considered as imperative factors that can influence the procedure of LREs, their intensity, quality, and eventually the written outcome (Fernandez Dobao, 2012, 2014Kim, 2008Kim, , 2009Leeser, 2004;Neumann & McDonough, 2014, 2015Watanabe & Swain, 2007. In some cases, empirical studies have provided contradictory findings concerning the form-meaning construct of writing. Given that concentrating on every aspect of writing has been proved to be demanding due to the limitation of human attention, there could be trade-offs between content and form (Rahimpour et al., 2011). This means that, depending on the structure of the prewriting task, learners could either focus on the meaning of the ideas or how to formulate them, and this can have an effect on the written output. Thus far, some researchers have concluded that LREs have contributed to enhanced fluency and syntactic complexity, detecting no significant effect of collaborative planning on accuracy and giving little credit to the grammatical aspect (Ellis & Yuan, 2003;Kang & Lee, 2019;Rahimpour et al, 2011). However, a cluster of studies has come to the opposite conclusion, that collaboration has a positive impact on linguistic accuracy. Learners working together fulfilled the tasks more competently, in some cases producing shorter but grammatically more accurate texts (McDonough & Neumann, 2014, 2015Storch & Wigglesworth, 2009, 2012Williams, 2001Williams, , 2008. Storch (2005) revealed that collaboratively produced texts are not only linguistically more accurate and complex but also the quality of content improved; when students wrote a text collaboratively in dyads, the theses of their texts were more accurate and relevant.
To this end, an array of studies has explored what students take into account in their discussions while preparing for writing. Due to their nature, LREs imply that students tend to focus on form in terms of grammar, lexicon, mechanics, or discourse (Edstrom, 2015;Kim, 2008). Williams (2001) found that in meaning-focused classrooms, learners showcase the tendency to concentrate on lexis and, as a result, lexical LREs are far more frequent than grammatical LREs. At the same time, specifically devised grammar-focused tasks can enhance attention to grammar prior to writing (Fernandez Dobao, 2012;Leeser, 2004;Swain & Lapkin, 1998). Yet, it has been revealed that in LREs learners also discuss such components of written texts as their content or organisation (Elola & Oskoz, 2010;Storch, 2005). Students feel more confident expressing their ideas and analysing the ideas of their peers about the content and organisation of texts in the course of such prewriting sessions.
The current study builds on the findings of the previous research, positing the positive correlation between learner-learner interaction before writing and the quality of the produced texts. It aims to look into the THE COLLABORATIVE DISCUSSION MODEL collaborative discussion method as a pre-writing strategy for the individual writing tasks of Russian students in an EAP course and, most importantly, focuses on practical considerations in the design and implementation of the collaborative discussion model. To achieve this, the study investigated what students tend to discuss during spontaneous collaborative activities before writing and what stages of the scaffolding discussions are considered the most or least useful by the students, which can further be a benchmark for improving prewriting instruction.

Participants
The focal participants were 48 EFL students (26 males, 22 females) with the same linguistic background, from different academic groups enrolled in the bachelor degree programme in the Department of Economics at the National Research University Higher School of Economics, Moscow, Russia. Their mean age constituted 19.2 years (SD = 1.6) and they had studied English previously for a mean of 10.1 years (SD = 1.9), with some of them starting in primary school, the others in secondary school, and all of them continuing in their first year at university. Upon entering the university, the students had been tested on their level of English language proficiency by the English Department. At the time of the experiment, they fell within the B1-B2 levels according to the Common European Framework of Reference. The participants were enrolled in the EAP course, which was required for their degree, and aimed to develop all four academic English language skills with no special emphasis on writing skills. At the end of the course, the students were supposed to take an IELTS mock examination. In terms of writing, the course targeted essay-level writing skills. The focus was on different essay types, i.e., opinion, discussion, advantages and disadvantages, problem and solution, and mixed question essays. The classes were conducted for two class periods per week, which also included Business English classes, in a 16-week semester. The participants were divided equally into the control group (CG) and the experimental group (EG). All the participants were engaged in the experiment on a voluntary basis and were guaranteed that in no case would the results of the experiment have any impact on their grades for the course.

Materials and Procedure
At the outset of the study, the students were involved in a naturally occurring face-to-face discussion, as one of the pre-writing activities. Before this, they had read and listened to texts on the corresponding topics in the students' theme-based EFL textbook. Then they were asked to accomplish a writing task (pre-test) in the form of an essay corresponding to the standard of the IELTS examination. In the pre-test stage, the students of both the experimental (EG) and the control (CG) groups were organised into smaller subgroups of four students according to their L2 proficiency level and each group included a more knowledgeable student to scaffold the peers with lower levels. The discussions in each group, which were audio-recorded, lasted for approximately 20-25 minutes before the students proceeded to the classroom-based individual writing. They were given 30 minutes to accomplish the task, a shorter time than that allocated in IELTS because the stage of planning had been done collaboratively. The submitted assignments were assessed independently by two raters. Inter-rater reliability constituted .79, which was calculated as the two-way mixed average measures intraclass correlation coefficient (in SPSS). The raters were specifically trained to evaluate essays according to the analytic rubric in the form of IELTS writing band descriptors for Task 2 (public version). The rubric included such components as task response with the requirements from candidates to formulate and develop a position in relation to a given prompt in the form of a question or statement; coherence and cohesion with the overall clarity and fluency of the message, how the response organises and links information, ideas, and language; lexical resource with the range of vocabulary the candidate has used and its accuracy and appropriateness; grammatical range and accuracy at the sentence level (Appendix C). The four components in each text were scored separately and constituted the average total score.
The post-test took place eight weeks after the pre-test. Based on the findings from the previous part of the experiment, the experimenter designed a structured prewriting discussion task, the collaborative discussion model (the CDM). Its implementation and effectiveness were tested further (Appendix A). The instructions and materials used for the discussions were devised in the way to encourage student reflection on content, organization, and language, particularly, understanding the task, brainstorming main ideas, generating supporting details, providing examples, evaluating and organising ideas, together with eliciting useful vocabulary and grammar. This time, at the post-test stage, the students in the same experimental subgroups were given the CDM in the form of handouts. However, the control subgroups participated again in naturally occurring discussions. The same amounts of time of approximately 20-25 minutes and 30 minutes were allocated for discussion and writing, respectively.
At the final stage of the experiment, the EG students were asked to express their perceptions on the usefulness of organising instruction-led scaffolding activities, the CDM, in a specially designed questionnaire (Appendix B). The questionnaire asked the students to evaluate the different components of the prewriting session in terms of their practicality: understanding task requirements, brainstorming main ideas, generating supporting details, providing examples, evaluating ideas, organising ideas, and eliciting useful vocabulary and grammar. It also included questions about students' preferences for individual or collaborative types of prewriting activities and open questions about reasons why this type of activity (individual or collaborative) is preferred and how students felt while collaborating with the others and fulfilling the assignment. During the final stage of the study, all data and results were processed and described.

Data Analysis
The students' individually written texts were rated using the analytic rubric applied in IELTS, with equally weighted subscores for each component, which were composed average total scores for overall quality. The subscores included task response, coherence and cohesion, lexical resource, and grammatical range and accuracy and could be scored from 1 (poor) to 9 (excellent). The analysis of the written assignments was based on the number of scores from 4 (as the minimum given the proficiency level of the participants) to 9 (as the maximum) received by the students in both the control and the experimental groups as a total and also separately for each component of the analytic rubric before and after using the CDM. Task response was checked with a focus on the quality of addressing the task, expressing a clear position on the topic, developing ideas and supplementing them with details and examples, and the relevance of these ideas to the topic. Coherence and cohesion were gauged for the layout of the texts, presenting the central idea within a paragraph, sequencing the information with an overall progression, and using cohesive devices within and between sentences. The assessment of lexical resource included linguistic accuracy, the control of lexical features, word choice, collocations, word formation, spelling, and register. Grammatical accuracy was rated for the usage of the tenses, subject/verb agreement, subordinate clauses, word order, articles, prepositions, and punctuation. The collected data were afterwards subjected to statistical analysis, namely to a dependent t-test for paired samples, to check whether there were any statistically significant differences between the two tasks completed after naturally occurring discussions and after instruction-led discussions in both groups. The students' t-distribution under the null hypothesis, implying no differences between pre-test and post-test results, was calculated for both groups.The effects of the introduced CDM on the total scores and subscores for each component of the writing tasks were examined. Since the sample size was relatively small, the data collected via the questionnaire were analysed manually. The results of the study are presented in the next section.

Results
The first research question addressed the effect of the devised collaborative discussion model on the writing performance of students. As can be seen from Figure 1, at the pre-test stage of the study the most frequent scores for essays were 6 out of 9 (n = 11; n = 9), according to the IELTS assessment system, in both the control and experimental groups, which corresponded with the proficiency level of the participants. At the post-test stage, the results of the writing tasks changed in both groups. The majority of final essays in the EG earned a score of 7 (n = 11), while in the CG the most frequent score remained 6 (n = 9), with the number of scores 7 and 8 increasing by 1.
In order to determine if the pre-test and post-test results differed statistically significantly, they were compared using Student'st-values for the experimental and control groups. In the EG (Table 1), the differences between the pre-test and post-test total scores could be determined as highly significant (t = 4.412, p < .001), with the THE COLLABORATIVE DISCUSSION MODEL critical values for t-tests on 23 degrees of freedom (n-1) being 3.768 (p = .001). The total average essay scores changed positively with a mean difference of .45 (SD = .228).

Figure 1
Number of scores for the students' essays after pre-test and post-test discussions for the experimental group (EG) and the control group (CG)

Pre-test and Post-test (CG)
Pre-test Post-test In the CG (Table 2), marginally significant differences were found between the pre-test and post-test total results (t = 2.145, p < .05), with the critical values from the t distribution on 23 degrees of freedom (n-1) being 2.069 (p = .05). The average total score alongside the average subscores for the four components of the essays increased in their mean value, with the mean difference varying from .33 to .21 (SD = .062), which showcases consistent improvement. Thus, the total scores represent larger effect sizes as the t-values move further away from zero. In sum, in relation to the first research question, the post-test texts written by both the experimental and control groups were of higher quality. In the EG, the total scores for the final essays were significantly higher than those obtained during the pre-test stage. In the CG, the results also improved but less significantly.
The second research question asked what stages of discussion supported by the collaborative discussion model contributed most to the quality of the writing output, provided that it improved. Based on the previous analysis, it could be suggested that different parts of the CDM had a different impact on the results of the essays. In the experimental group, the differences across the subscores for task response and lexical resource could be viewed as highly significant (t = 7.474, t = 4.796). The subscores for task response improved the most, with a more noticeable mean difference (MD = .75), which implies that more students received higher scores for this part of the task. The results of the pre-test and post-test differed significantly with a confidence level of 95 percent in relation to the subscores for coherence and cohesion (t = 2.769, p = .011) and also showcased a marginal significance for grammatical range and accuracy (t = 2.145, p = .043). In the control group, concerning the criterion of task response, the difference was significant with 99 percent confidence (t = 3.391, p < .01). However, across the criteria of coherence and cohesion, lexical resource, and grammatical range and accuracy, marginally significant differences were found between the pre-test and post-test results (t = 2.304, 2.769, 2.460, correspondingly), where the calculated t-emp did not exceed the critical values (p < .05).
In the questionnaire addressing the learners' attitudes toward the components of the CDM and the usefulness of working with peers in class (Appendix B), the students reported a positive perception of the structural prewriting discussions, with the highest rankings given to brainstorming, eliciting supporting details, and eliciting useful vocabulary. They believed the tasks were helpful for understanding the assignment and for selecting relevant examples. They found evaluating which ideas were appropriate for the assignment requirements and organizing ideas into a prewriting plan to be less useful. Their lowest rating was for choosing necessary grammatical structures. When asked if they preferred to work alone or with peers to carry out prewriting activities, the majority of the students claimed that they preferred to work in groups because they could generate more ideas, exchange different viewpoints, clarify information they had not understood, and get suggestions from their peers. Those five students who preferred to work alone mentioned that if they performed individually, they would have greater concentration and less dependence on others as well as less stress without being in the focus of their peers.

Figure 2 Students' opinions on useful components of prewriting discussions (where 1 = not helpful; 5 = very helpful)
When elaborating on the aspects of the prewriting discussions that were helpful, 21 of 24 students mentioned the general topics of ideas and content most often, specifically gathering more ideas to write about, selecting examples, developing a better understanding of the topic, and being exposed to critical views that lead them to find more relevant examples. Only eight students mentioned organisation, with five of them stating that the prewriting discussions helped them improve the organisation of their ideas. However, two students claimed that the organisation section of the prewriting discussions was not helpful and that this component of prewriting could be best done individually. Responding to the question about feelings, all the students pointed to the pleasant atmosphere of the classroom collaborative activity and positive feelings. In sum, based on higher subscores for task response and lexical resource in the post-test essays and on the students' opinions, it could be concluded that the parts of the CDM that focus on brainstorming, eliciting supporting details, and eliciting useful vocabulary produced a larger effect on the final result.

THE COLLABORATIVE DISCUSSION MODEL
The third research question asked if students with higher and lower L2 proficiency levels benefited equally from collaboration. The results of the questionnaire suggest that practically all of the learners benefited somewhat from the task. Most of the students showed a positive attitude towards the collaborative discussions and felt a positive influence from this on their written assignments. They claimed a greater variety of ideas, knowledge, and creativity to share, and therefore more possibilities for language development. Almost half of them felt the positive impact of collaboration on the lexical variability for their texts. Five of these respondents found the discussions helpful for vocabulary learning, explaining that they were able to memorise a number of words they had learned from their more advanced peers. Four of the weaker students also pointed out that working in small groups provided motivation and a fun atmosphere, which made them feel comfortable. It might seem that the weaker students contributed little to the collaborative discussions but they were in fact actively involved in the process, not only as listeners and observers. Enjoying the comfortable atmosphere, they participated in the collaborations by asking questions, thus facilitating the discussion, and eliciting new ideas. Yet, it is also necessary to highlight that six of 24 learners felt the collaborative writing activities could not help them enhance their lexical or grammatical knowledge because of the similar proficiency level of their peers. These more knowledgeable learners saw opportunities to improve their fluency and overall speaking skills but did not think they could learn grammar or vocabulary while working with learners of the same proficiency level. Practically all the respondents mentioned the favourable and pleasant atmosphere while working with their peers, which allowed for arousing positive emotions accompanying the learning process. Therefore, it can be concluded that scaffolding activities proved to be beneficial for students with different levels of L2 proficiency although in nuanced ways and to different degrees, which is supported by the results of the study.

Discussion
The findings of this study are in line with those of previous research studies that established a positive relationship between the patterns of collaborative discussion and learners' performance in writing (McDonough & Neumann, 2015;Storch, 2005;Wigglesworth & Storch, 2009). As a whole, with the collaborative discussion model used as an explicitly instructed prewriting strategy inside small groups of peers with different levels of English language proficiency, the quality of essays of the students improved. The findings indicate that the texts produced following the instruction-led prewriting tasks were scored higher than the texts produced following the naturally occurring discussions. At the beginning of the experiment, during the naturally occurring discussions the learners spent time allocated for discussion chatting on the topic and content in general and even digressing from the topic. It was clear that the students did not make the best use of the time given for planning and ignored the criteria used in assessing their written assignments in the format of IELTS, although they were fairly aware of them. After the CDM was introduced and applied in the experimental subgroups, the learners tended to be more focused on the components of the assignments using the writing prompts and bearing in mind what they should discuss. As a result, the quality of the essays improved, especially relative to some particular criteria. Juxtaposing the initial essays and the final ones, it could be said that the most impactful part of the CDM appeared to be understanding the task, brainstorming main ideas, generating supporting details, providing examples, and also eliciting useful vocabulary since the quality of the essays concerning the corresponding criteria of task response and lexical resource improved significantly. Cohesion and coherence and grammatical range and accuracy improved marginally. This study has shown that during the instruction-led discussions the students discussed the content, organization, and vocabulary, which is in line with the previous research where the learners' higher degree of attention to the structure of the text and lexical choices rather than accuracy was attributed to the meaning-focused nature of the tasks (Storch & Wiglesworth, 2009;Williams, 2001Williams, , 2008. But these results contradict those of Storch (2005) and DiCamilla and Anton (2012), whose students focused their efforts on inventing content and also on addressing both grammatical and lexical issues. In the second experiment, the section of the CDM eliciting necessary grammar proved to be the least helpful, which could be explained by time constraints and difficulty for struggling students to advance the aspect of grammar mastering in such restricted conditions. Research into grammar instruction in the planning stage (TEAL, 2012) also showcases its negative results in terms of students' overall writing quality. Although teaching grammar purports to be an essential part of the language acquiring process, it seems more productive to use practice-oriented grammar approaches and their integration with other writing activities. Likewise, coherence and cohesion are attributed to such constructs of academic writing that tend to be rather problematic to be achieved comprehensively in a short period of time, although understanding the proper organisation of a text and using appropriate linking devices contribute much to the process.
The questionnaire data largely confirm the results of the study, as students reported that they talked about content more frequently during the prewriting discussions and that the prewriting discussions were the most helpful for brainstorming the main ideas, choosing the most relevant ones, and generating supporting details alongside eliciting the necessary task vocabulary. The students talked about the organisation of the ideas less frequently during the prewriting discussions, and fewer students reported that the prewriting discussions were helpful for eliciting the necessary grammar, although two of the stronger students mentioned that this section prompted them to revise and modify the grammatical structures in terms of complexity and variability. Given the fact that students' weak points concerning the formal aspect of academic writing appeared to be the organisation of the text, distinguishing between the topic sentence and supporting ideas, and well-balanced argumentation (Pospelova, 2016), the CDM can be an effective tool for fostering academic skills in writing. This refers both to macro-skills development like accomplishing the communicative function of the text, organising ideas logically, soliciting and using peer feedback for producing a written output, and micro-skills like using appropriate words and word combinations, grammatical patterns, and cohesive devices.
Notwithstanding the obvious benefits of instruction-led practices based on scaffolding, it seems necessary to elucidate the limitations of such scaffolding. The present study was implemented in groups of four students. When discussing the role of the learner-learner interactions in collaborative writing, we must consider the fact that not all small groups work equally effectively. These differences arise due to a plethora of factors such as topics for discussion, the size of the group, the personality types of participants, and their learning styles, confidence, experience, and goals (Storch, 2005), which could have influenced the results of this study. Identifying students' individual abilities and background knowledge can be a challenge to the organisation of successful scaffolding. Therefore, it is crucial that teachers be able to determine a more-knowledgeable-other (MKO) and allocate students in groups in a way so as to achieve the effect of matching learners with higher and lower proficiency levels without depriving them of the motivation to communicate (Fernandez Dobao, 2014;Leeser, 2004;Watanabe & Swain, 2007. In order to foster collaborative interactions, teachers must monitor group activities and use their observations to place passive or disengaged students effectively in future groups. The optimal number of participants in a group is yet to be explored.
Although it was not the main focus of the study, the students' perceptions about the collaborative discussions were elicited through the questionnaire administered at the end of the experiment. Practically all the respondents mentioned the favourable and pleasant atmosphere while working with the peers, which allowed for arousing positive emotions accompanying the learning process.

Conclusion
The present study attempted to investigate the effects of explicit instruction on the quality of writing, what students focus their attention on in the process of structured prewriting collaborative activities, and to design and test a collaborative discussion model that could scaffold effective learner-learner interactions based on the practicality of the components of the discussion. The study also addressed the question of benefits to students with higher and lower L2 proficiency levels from collaboration as a prewriting strategy. The analysis of the data confirmed that active-learning classroom practices aimed at articulating the steps in the complex process of academic writing were beneficial. In the course of prewriting collaborative discussions, with more advanced students scaffolding the less accomplished peers, learning occurs through a gradual process of the internalisation and employment of regulatory language used by others, as the weaker learner moves from assisted to independent performance (Vygotsky, 1978). Research into collaborative writing has shown that learners discuss content as well as the organisation of the text and language, and their interaction is positively associated with the text quality (McDonough & Neumann, 2015).
The current study indicates that collaborative prewriting tasks encouraged students to engage in reflection about their own and their peers' ideas. As the texts produced following collaborative prewriting tasks were scored higher, it is suggested that prewriting discussions can potentially augment learners' knowledge of the language (Wigglesworth & Storch, 2012).Inquiry learning engages students in critical thinking via a focused investigation into the assigned topic. The findings suggest that scaffolding may enhance inquiry and performance (Simons & Klein, 2007), especially when students are intentionally instructed on how to accomplish the task. Additionally, encouraging students to reflect on their own performance is the key to perceiving progress not only in terms of the language, but in terms of communicative objectives. Active engagement in prewriting discussions may also contribute to learners' motivation and confidence.
The results of the study, although limited with the size of the sample, further extend the body of research into the learner-learner interactions in language-related episodes during prewriting discussions by supplementing existing data with the practical collaborative discussion model. The CDM was applied as an instrument to guide a reflective, language-focused conversation by defining a specific focus on the components of academic writing assignments.Being explicitly instructed, the learners were able to concentrate better on the task and ultimately were more successful in solving both conceptual and languagerelated problems. However, the grammatical aspect and cohesion improved marginally, thus leaving the door open for further research. Hence, it is proposed that the designed collaborative discussion model can be used as a tool to enhance in-class active discussions and as an alternative or a complementary activity to address the challenges associated with academic writing. The CDM suggests tasks that focus on the necessary elements of writing. Based on peer-peer scaffolding, it can engage students to be actively involved in the process of learning, by constructing knowledge rather than passively receiving it. It can develop the ability to comprehend and analyse issues, articulate thoughts, produce a higher quality of written output, and help students be more responsible for the learning process as a whole. The CDM can also be used as a sample for teachers to design similar models depending on the purposes of the task, the needs of learners, their individual characteristics, their motivation and readiness to cooperate, and other factors. Thus, teachers can adjust the instruction to reflect specific class learning conditions.
One of the essential constraints that collaborative discussions may face is the time required, which can explain why these types of activities are not commonly used in L2 writing classrooms. It should also be noted that the teacher's role in organising effective concurrent discussions and providing valuable feedback to students is paramount (Swain & Lapkin, 1998).The results have shown that unless students are given a specifying instruction, they tend to only express their ideas on the assigned topic without evaluating them in terms of the quality, considering alternatives, justifying them, and organising them into a writing plan. Learners may also need assistance establishing appropriate roles and expectations when working together.
This study can help teachers better analyse their students' writing and guide them on how to better instruct students in order to improve their formal writing competence. To this end, teachers need to focus on the parts of the language that are significant for the writing process and present some difficulties for students when evaluating the effectiveness of teaching materials and teaching techniques, revealing what parts of the syllabus have been inadequately learned or taught and require further attention. uses a wide range of structures with full flexibility and accuracy; rare minor errors occur only as 'slips' 8 sufficiently addresses all parts of the task presents a well-developed response to the question with relevant, extended and supported ideas sequences information and ideas logically manages all aspects of cohesion well uses paragraphing sufficiently and appropriately uses a wide range of vocabulary fluently and flexibly to convey precise meanings skillfully uses uncommon lexical items but there may be occasional inaccuracies in word choice and collocation produces rare errors in spelling and/or word formation uses a wide range of structures the majority of sentences are error-free makes only very occasional errors or inappropriacies 7 addresses all parts of the task presents a clear position throughout the response presents, extends and supports main ideas, but there may be a tendency to overgeneralise and/ or supporting ideas may lack focus logically organises information and ideas; there is clear progression throughout uses a range of cohesive devices appropriately although there may be some under-/over-use presents a clear central topic within each paragraph uses a sufficient range of vocabulary to allow some flexibility and precision uses less common lexical items with some awareness of style and collocation may produce occasional errors in word choice, spelling and/or word formation uses a variety of complex structures produces frequent error-free sentences has good control of grammar and punctuation but may make a few errors 6 addresses all parts of the task although some parts may be more fully covered than others presents a relevant position although the conclusions may become unclear or repetitive presents relevant main ideas but some may be inadequately developed/unclear arranges information and ideas coherently and there is a clear overall progression uses cohesive devices effectively, but cohesion within and/or between sentences may be faulty or mechanical may not always use referencing clearly or appropriately uses paragraphing, but not always logically uses an adequate range of vocabulary for the task attempts to use less common vocabulary but with some inaccuracy makes some errors in spelling and/or word formation, but they do not impede communication uses a mix of simple and complex sentence forms makes some errors in grammar and punctuation but they rarely reduce communication 5 addresses the task only partially; the format may be inappropriate in places expresses a position but the development is not always clear and there may be no conclusions drawn presents some main ideas but these are limited and not sufficiently developed; there may be irrelevant detail presents information with some organisation but there may be a lack of overall progression makes inadequate, inaccurate or over use of cohesive devices may be repetitive because of lack of referencing and substitution may not write in paragraphs, or paragraphing may be inadequate uses a limited range of vocabulary, but this is minimally adequate for the task may make noticeable errors in spelling and/or word formation that may cause some difficulty for the reader uses only a limited range of structures attempts complex sentences but these tend to be less accurate than simple sentences may make frequent grammatical errors and punctuation may be faulty; errors can cause some difficulty for the reader 4 responds to the task only in a minimal way or the answer is tangential; the format may be inappropriate presents a position but this is unclear presents some main ideas but these are difficult to identify and may be repetitive, irrelevant or not well supported presents information and ideas but these are not arranged coherently and there is no clear progression in the response uses some basic cohesive devices but these may be inaccurate or repetitive may not write in paragraphs or their use may be confusing uses only basic vocabulary which may be used repetitively or which may be inappropriate for the task has limited control of word formation and/or spelling; errors may cause strain for the reader uses only a very limited range of structures with only rare use of subordinate clauses some structures are accurate but errors predominate, and punctuation is often faulty

Appendix A THE COLLABORATIVE DISCUSSION MODEL
Note. The table does not include data for bands 3-0. Adapted from https://takeielts.britishcouncil.org/sites/default/fi les/ielts_task_2_writing_band_descriptors.pdf. In the public domain.