Implementing Digital Multimodal Composing in L2 Writing: A Focus on English Learner Engagement With Feedback

L2 writing has been embedded in digital technology and broadened into digital multimodal composing. Recent research has emphasized the composing process and the effects of feedback, but few on L2 writers’ engagement with feedback. Therefore, with the multi-draft approach, the research investigated two English majors’ responses to peer and video feedback. First, from their first drafts and peer checklists, the author analyzed the salient problems of their multimodal texts. Then, the researcher explored how they respond to feedback form their drafts and interview responses. The findings suggest some problems including overuse of texts, visual disharmony, and “disconnection” from the audience. Besides, the study reveals a complex relationship among three engagement dimensions, manifested in interconnectedness and inconsistencies. Meanwhile, learner engagement with feedback is mediated by individual and contextual factors. The findings contribute insights into understanding L2 writers’ engagement with feedback.


INTRoDUCTIoN
The impetus for this change can be found in recent advances in communication technology and social media, which have resulted in L2 writing no longer being confined to alphabetic text-based writing (Belcher, 2017).Instead, L2 writing has already been embedded in digital technology and broadened into multimodal text composing.By definition, digital multimodal composing asks L2 writers to produce texts with multiple semiotic modes, such as language, image and sound, using digital tools (Hafner & Miller, 2011).Multimodal texts mainly include digital stories, PPT slides, videos, or posters (Yi et al., 2020).The novel writing genre is designed for wider audiences on the internet, which exhibits interactivity and multimodality.However, there is a dissonance between language-centred writing activities in school and students' practices of constructing meaning with multimodal resources outside the classroom.To bridge the gap, literacy educators and L2 writing experts at home and abroad are increasingly interested in the pedagogical use of DMC in educational settings.They have incorporated DMC into L2 classrooms (Yi et al., 2020).They have shown that this digital genre does not diminish the L2 writers' focus on language mode but also enhances their autonomy, interest and enjoyment in L2 writing as well as their metalanguage development (Hafner, 2020;Zhang & Yu, 2022).
Most studies on DMC have focused on the planning and drafting process (Wang, 2021), discussing how L2 writers use different modal resources and strategies to make meaning from a social semiotic perspective.Nevertheless, scant attention has been given to the sharing and reflection phases.Moreover, limited research has focused on summative evaluation, which means feedback is one-way.In contrast, the multi-draft approach emphasizes the negotiation and interaction between writers and readers (Zhao, 2021).Although previous research has probed into how L2 writers incorporate feedback into their revision by comparing their multiple drafts (Zhang, 2019), behaviour engagement is merely one of the dimensions of student engagement.In fact, it is interrelated with the affective and cognitive dimensions, and could be mediated by individual or contextual factors (Han & Hyland, 2015).Consequently, to bridge these gaps, the present exploratory study will examine DMC from a multi-draft process perspective, exploring salient problems in their first drafts.More importantly, it will delve into how they respond to feedback.

Previous Research on Digital Multimodal Composing in L2 Classrooms
In the 21st century, L2 writers are increasingly interacting with various multimodal texts in digital environments due to multimedia and new technologies (Yi et al., 2020).Traditional monomodal language-based text communication cannot sufficiently assist students' writing in the rhetorical contexts they encounter in the digital era (Li, 2020).Therefore, in response to the multimodal realities and radical changes in digital environments, L2 experts and scholars have incorporated DMC into L2 classrooms.Teachers' and students' perceptions of DMC, and the impacts of DMC have been the focus of previous research on DMC in L2 classrooms.
Most research concerns students' perceptions, including the perceived benefits and challenges of DMC and students' attitudes toward peer interaction.In detail, multiple benefits of multimodal composition were recognized by L2 writers, including investment, motivation, and authenticity (Zhang & Yu, 2022), genre and audience awareness (Oskoz & Ilola, 2016b).Specifically, Zhang and Yu (2022) have reported that the developing L2 student writers positively perceived the affordances of DMC.They have access to multiple modes to fully express creative ideas, feeling less anxious in L2 writing, being more motivated to interact with authentic audiences, and forming the habit of constantly revising scripts.Researchers have also looked at teacher views.Such initiatives have not only addressed the challenges of teachers' investment and participation in DMC.Hafner and Ho (2020) investigated how ESP teachers perceived the evaluation of a digital video documentary.Based on their findings, the researchers proposed a process-based model for assessing DMC that includes formative or summative tactics and orchestration of multimodal affordances.
As to the effects of DMC, research has discussed how participation in multimodal composing led to gains in aspects of students' English proficiency, mainly including the increase in vocabulary knowledge and metalinguistic awareness (Unsworth & Mills, 2020).For example, Shin (2018) used a case study to examine the argumentative multimodal writing of a bilingual sixth-grade kid in an English language arts class.It investigated how his argumentative writing process and metalanguage development were impacted by multimodal writing activities and was informed by social semiotics.The findings showed that the student was developing a knowledge of the metafunctions, metalanguages, and intermodal connections of different semiotic modes, which further enabled him to identify the register of the argument in his work.Additionally, earlier studies looked examined how DMC affected students' writing abilities.For instance, Yang and Wu (2012) investigated the impact of digital storytelling (DST) on the academic performance of senior high school students in Taiwan's EFL environment.The participants' level of linguistic competence ranged from beginner to intermediate.This study demonstrated that students who participated in DST activities outperformed those who did not in terms of English learning performance based on a pretest and posttest quasi-experimental approach.In the pretest and posttest, the participating students were required to produce a 100-word narrative essay as part of the writing evaluation.The study provided evidence for the impact of DST on writing ability.However, it only paid attention to grammatical accuracy and ignored other crucial factors, including complexity, fluency, and functional adequacy.Recently, the study evaluated students' growth as L2 writers using two writing assignments, an argumentative essay and a narrative essay while considering both linguistic components (lexical and syntactic complexity, accuracy, and text length) and functional components (content and communicative effectiveness).The study's results demonstrated that the experimental group's writing skills had improved more than those of the control group in terms of complexity, lexical diversity, text length, communicative efficiency, and content.
However, for most existing studies, DMC has been conceived more broadly as an L2 instructional strategy or L2 learning task than a novel L2 writing genre or task.Only a handful of studies have applied DMC as a writing genre in second-language writing classrooms to discuss and compare the potential effectiveness of this digital genre with traditional writing genres.However, these studies still need to consider using traditional writing scoring criteria such as complexity, fluency, and accuracy to assess multimodal writing texts, ignoring the fact that DMC draws on a much more extensive range of semiotic resources than traditional print-based writing.

Research Questions
This study explores L2 writers' digital multimodal composing practices in the L2 writing classroom from a multiple-draft perspective.Specifically, the present study investigates L2 writers' salient problems in composing their first drafts.Meanwhile, it also delves into how they respond to peer and video feedback from a multi-dimensional perspective.On this basis, the following research questions were explored in this study: Q1: What are the salient problems with L2 writers' multimodal texts identified by peer and video feedback?Q2: How do L2 writers respond affectively, cognitively, and behaviorally to peer and video feedback?

Context and Participants
The study will be conducted at a top-notch university located in southern China.The reasons for selecting this university as a research site are: (a)the school has a computer lab equipped with multimedia equipment, and all classes could use it for their instructional activities throughout the academic year; (b)most students have their own smartphones and laptops, and they use these electronic devices frequently outside of class.Additionally, the researcher implemented the study in an L2 writing classroom.The study is conducted with 38 second-year English majors with at least ten years of experience in English.Through purposeful sampling (Patton, 1990), two focus students-Pan and Zhao were selected from this class to conduct an in-depth analysis of their responses to the feedback:1) They made some significant changes in their second drafts, which helped to explore their reflection activities in depth; 2) They had experience in PPT making and presentation before.

Digital Multimodal Composing Project
The writing tasks correspond to the contents of the curriculum.The students had learned and understood about the expository genre in the course, so the multimodal composing task for this study was to create an expository PPT, and to give a PPT presentation in class.Specifically, students began to write their expository texts through Microsoft Word, where were required students to introduce or recommend their favourite things, places, songs, people, etc., to their classmates.And then, based on their expository texts, they composed multimodal expository texts with Microsoft PowerPoint, and prepared for their PPT presentation in the class.

Checklists
To support students' self-reflection and revisions for their first drafts, the researcher developed a preliminary evaluation checklist by combining the assessment elements from previous studies, which mainly includes Jiang (2022), Hafner (2020), Yang (2016), Han (2013) and Hung (2012).This evaluation checklist consisted of three main aspects-students' explanatory essays, PPT production, and PPT presentations.Specifically, this study drew on the peer assessment scale for English expository writing in Bai ( 2013), and assessed students' expository writing in terms of three dimensions: content, structure, and language expression.In evaluating students' PPT production, this study drew on Yang (2016) and Hafner (2020)'s evaluation criteria to assess students' multiliteracy in terms of language mode, non-language modes, and PPT overall effects (e.g., logic, creativity, coherence).Meanwhile, to assess students' PPT presentations, the researcher referred to Han (2013) and Yang (2016)'s evaluation elements for PPT presentations, which mainly involved oral expression (e.g., pronunciation, speech speed) and interaction with the audience (e.g., body language, eye contact).

Research Procedure
The whole study lasted for 14 weeks in total, and was divided into 7 stages, which are pre-writing, drafting, presentation, reflection and revision, re-presentation, reflection and interview.Specifically, at the pre-writing stage, the teacher explicitly provide knowledge about the expository genre, and assigned writing tasks.At the drafting stage, students drafted their expository texts in Microsoft Word and then composed PPT slides based on their expository texts.Then, they prepared for their subsequent PPT presentation in class.During the presentation stage, participants gave an oral presentation according to their student number.Meanwhile, the researcher recorded each participant's oral presentation using video equipment.During the reflection and revision phase, participants evaluated each other's PPT presentations according to the evaluation checklists.Afterwards, participants received the evaluation checklists from their peers and the video recordings of their own PPT presentation.Based on the multiple feedback, they self-reflected their first drafts and revised their drafts in preparation for the next PPT presentation in class.After the second oral presentation, students were asked to reflect their performance.After that, the researcher selected Pan and Zhao as the two focal participants for 20-min interviews respectively.

Salient Problems Identified by Feedback
Through reviewing peer assessment checklists and video recordings, it was found that both students shared some salient problems in producing multimodal texts and oral presentations.Specifically, language has been used as the principle carrier of constructing ideational meaning, while non-language modes including image and video serve primarily as an interactive hook for engaging the target audience.Pan even simply used the pictures to achieve the decorative effects.This issue is consistent with Shin's (2020) finding that participates can not distribute ideational meanings across language and non-language modes.This may confirm that they transformed traditional compositional norms into the novel digital genre, since their previous school writing privileged linguistic resources as the primary carriers of information and visual modes as an interactive hook.In addition, both students showed a lack of thought when selecting colors and fonts, therefore, problems appeared such as font colors overlapping with background colors, all fonts being bolded, and inconsistent fonts.This may be due to their lack of understanding of the affordances of these sign systems, and could not appropriate modal resources for various purposes of multimodal writing (Shin, 2018).As to language design, there are still some grammatical errors in expression, but language accuracy could reflect English writers' pure linguistic proficiency to some extent.It has been evidenced that English learners tend to make less errors if they are proficient in English (Wu, 2016).Besides, although both students have employed a variety of modal resources in their first drafts, scant attention was given to the coherence of the whole text and the thematic consistency of the modal resources, which to some extent disrupted the visual harmony, and break up the coordination of the whole slides.It may prove that the degree of success or effectiveness in achieving cohesion is linked to the writers' cognitive ability, technological facility, and aesthetic sense (Wang, 2021).As to their PPT presentations, both students chose to read directly from their prepared scripts or from the text on their slides, indicating that they relied excessively on existing information and made little effort to do extensions based on their slides due to their limited cognitive efforts.In addition, when conducting PPT presentation, their posture and facial expressions were very rigid, and they made little eye contact with the audience.This is mainly because they regarded the spoken word as a key factor in the success of oral presentation and the main mode of conveying meaning.Instead, they were not conscious of the importance of non-verbal modes such as body language in enhancing audience engagement and achieving interactive meaning.

Students' Engagement with Feedback
By analyzing two students' interviews and their two drafts, the researcher has discussed their emotional, cognitive, and behavioral engagement with the peer and video feedback below.

Pan: Under Engagement, but Facilitated by Peer Interaction
Pan shows little confidence and motivation in PPT production and presentation, as evidenced by her responses, "I just download the template every time" "I'm not good at computers and hyperlink".Meanwhile, her responses also indicate her limited e-proficiency and multimodal literacy.When receiving video feedback, Pan immediately felt embarrassed and anxiety.This could be explained in her interview: Teacher: How did you feel when you watched your first presentation video?Pan: So embarrassed.Teacher: Why did you feel so awkward?Pan: I just glanced at it and felt that my body language was so unnatural.It's so embarrassing.It's not even close to what I thought it would be.
It has demonstrated that Pan showed her unwillingness to engage with video feedback.However, Pan did not dismiss the value of the video feedback.She was aware of the stiffness of her body language during the presentation and noticed the gap between the real performance and the desired effects.Although she was aware of this problem, Pan did not adopt any effective strategy to alleviate her embarrassment, but chose to the avoidance strategy.It is lack of motivation that makes her fail to use cognitive and behavioral strategies to refine her oral presentation (Ellis, 2010).As seen in her second presentation recording, despite being consciously aware of her posture, Pan still relied much on her prepared script, which resulted in infrequent eye contact with the audience (see Figure 1).
Likewise, her engagement with peer feedback in the research period was also minimal, which can be evidenced by the final interview: Teacher: Li commented that there are too much text on your last few slides, have you considered cutting it down?Pan: I have barely cut the text.I made PPT and did presentations many times, so I thought that the audience wouldn't understand what I was trying to express with little text on the PPT slides(see Figure 2).
It is clear that due to her lack of metalanguage knowledge, she failed to understand this feedback accurately.This is in line with Han and Hyland's (2015) finding that students could not understand the feedback and make the correct revision without grasping the metalanguage knowledge, regardless of whether they attempted to process feedback at a deeper level or not.
Nonetheless, Pan's cognitive engagement was enhanced through peer discussion and interaction, which can be evidenced by her behavioral operations on the sub-headings (see Figure 3): Teacher: I see that you made changes to the subheadings on your slide 2? Why did you make the change?Pan: Yes, because Li commented that the wording of my subheading was inappropriate.Teacher: How do you understand this evaluation?Pan: In fact, I was confused by his comment at first, so I argued with him.Then he asked me about do you think your three subheadings are consistent in structure?Do you want to command your audience by using the imperative structure?Teacher: Did Li's reply assist your understanding of this feedback?Pan: Definitely.I realized I did not word them improperly, because I wanted to inform the audience, not command them.And I also used Baidu to search for the format of the heading.I realized my three subtitles are inconsistent in form, with noun phrases and imperative sentences.So I rephrased them all to noun phrases.From Pan's responses above, her cognitive engagement has showed a dynamic process.She felt confused about the feedback at the beginning, and then understood and acknowledged this evaluation.To facilitate her revision, Pan also utilized external resources, including seeking help from peers and search background knowledge on the website.Meanwhile, she also searched for background knowledge related to the heading with the help of Baidu.

Zhao: Deep Engagement, and Effective Revision
Zhao was one of the most motivated students, and was very serious about this writing task, "I spent a long time making my PPT and thought over the details".She also showed a positive engagement with feedback, when Mrs. Rachel asked how she felt about receiving the video feedback: Zhao: In fact, in the first few minutes, I felt a little upset and disappointed.But later, I felt it was not a big deal.I just felt that I still have much room for progress.The video recording showed me my shortcomings.So I watched the video from beginning to end and found some salient problems.
Her response explains the dynamic nature of emotional engagement.When receiving video feedback, Zhao first experienced negative emotions, but she quickly converted the disappointment into motivation by taking a future-oriented perspective.Besides, this positive affective engagement also facilitated her deep cognitive engagement with feedback, "I couldn't believe how stiff my body language was during my own presentation".It can be seen that Zhao was clearly aware of her salient issues, and resorted to meta-cognitive strategies to consciously plan and monitor her revision behaviors in her second draft.From her second oral presentation, Zhao started her speech with the greeting "Good afternoon, everyone!"which aroused the audience's attention and helped her establish a positive connection with the audience by making the eye contact (see Figure 4).
When talking about the movie's theme and inspiration, the speaker used this offer act to make the interactive meaning realized, which strengthens the audience's impression and understanding.Similarly, Zhao also had a positive attitude towards peer feedback and verbally acknowledged its value: "I basically made changes with reference to the problems pointed out by Tang.These evaluations were very useful and helped me to better revise and refine my first "draft".When asked about the peer evaluation that PPT has too many words, she explained, "Yes.If there are too many words on the PPT, it may reduce the audience's attention to other elements on the PPT, such as images and videos".It is evident that she was aware of the equal importance of language and non-language modes in the construction of multimodal texts.Aside from noticing and understanding feedback, she also employed various meta-cognitive operations when reflecting her first draft: Zhao: When watching other students' presentations in class, I felt that they put too much text in their slides, and then I was just so focused on reading the text that I couldn't care about other elements on their slides.
It can be seen that Zhao self-reflected her own problem by linking her past experiences as a audience with her current experience of making presentations, which made her deeply understand her peer's comments.Besides, she also adopted monitoring strategies.When revising slides, she identified and corrected some extra errors enlightened by peer feedback.In her revised draft, not only did she cut the text on slide 4 following Tang's evaluation, but she also reduced the lyrics on slides 5 and 6 by herself (see Figure 5).She responded to this change in the interview, " Tang's comment on slide 4 reminded me of the problems here.I wanted to convey the thematic meaning through the lyrics, music and pictures together.If the lyrics were placed too much, it would weaken the audience's attention to the pictures and music".
Zhao's explanation not only revealed her alertness and sensitiveness to the overuse of text spurred by peer feedback, but also her noticing and deep understanding of meta-linguistic rules.
Likewise, Zhao's modification of the images on slide 4 also reflected her deep cognitive engagement: Teacher: I notice you have deleted some images on slide 4. How did you consider that?Zhao: I originally wanted to use pictures to illustrate the text, but Tang said too many images would pose a visual burden to the audience, and some of the images were thematically repetitive.So I deleted some pictures that were duplicated on the theme.I also rescaled them to look coordinated and aesthetically pleasing.Teacher: Why did you adjust the position of the picture?Zhao: Because I wanted the images and text to be unified in theme.The pictures on the left depict his fancy of close encounters with ordinary people; the theme of those on the right is his fancy of dramatic adventures.It is clear that in processing peer feedback, Zhao performed metacognitive operations to regulate her mental effort and make changes.After she deleted the redundant images based on peer assessment, she further considered text-image relation, and thematic harmony between the images and the text (see Figure 6).In her second draft, the language and image represented primarily a concurrence relationship.It is evident that Zhao utilized images as a primary mode to create the ideational meaning of the text beyond the interpersonal one.As such, she did not privilege linguistic resources as the carriers of ideational meaning over visual modes, avoiding uses of images mainly as an interactive hook for interpersonal meaning.Besides, she also demonstrated a deepening awareness of the semiotics of linguistic and visual modes, which might lead to the development of her L2 writing and serve as her metalanguage for the communicative potentials of semiotic systems (Belcher, 2017).

CoNCLUSIoN Findings
Through an in-depth analysis of the data, the major findings are summarized as follows.First of all, from checklists and video recordings, some salience problems in students' first drafts have been summarized, involving their PPT production and presentation.In producing multimodal texts, these two students overused the language mode, which led to the imbalance within various modes.It revealed that they could not distribute ideational meanings across language and non-language modes.This may be due to the fact that their previous school writing privileged linguistic resources as the primary carriers of information and visual modes as an interactive hook.Additionally, scant attention was given to the coherence of the whole text and the visual harmony, which may be linked to their cognitive ability, technological facility and aesthetic sense.Moreover, the linguistic and visual modes provided more affordances for L2 writers than other modes.It is evident that their past habitual use of media and modes could determine how they avail of the new, showing that "new media has dimensions of old media within" (Leander 2009).For PPT presentations, both of them overly rely on their prepared scripts or slides, and made little effort to do extensions based on their slides.Besides, they were not conscious of enhancing eye contact to engage the audience, which resulted in their disconnection with the audience.
Additionally, the current study demonstrates the interconnections and inconsistencies among the dimensions of their engagement with feedback.In terms of interconnectedness of the three dimensions, the present study shows that students had positive affective engagement and extensive cognitive and behavioral engagement.This may imply that students are on task, thinking, and enjoying the learning process when they are most engaged and motivated in their studies.Moreover, learner engagement with feedback was also mediated by individual factors, such as motivation, confidence, and metalanguage knowledge.Students a lack of motivation tend to adopt the avoidance strategy, and fail to use cognitive and behavioral strategies or operations to refine their drafts.Besides, students could not make the correct revision without grasping the existing metalanguage knowledge, regardless of whether they attempted to process feedback at a deeper level or not (Han & Hyland, 2015).Another important influence on learner engagement was contextual factors.Students' deeper engagement or effective revision operations can be viewed from the sociocultural perspective, which reveals how feedback contingent to a learner's Zone of Proximal Development (ZPD) can enhance engagement (Aljaafreh & Lantolf, 1994).Through negotiation with peers as the effective scaffolding, Pan developed her semiotic awareness of how modes and language function in the context of designing multimodal explanatory texts.And her clear awareness of interpersonal functions led Pan to make appropriate rhetorical decision.On the other hand, student engagement with feedback also displays the inconsistencies within and across the three dimensions.Students might still feel embarrassed upon receiving video feedback in spite of their appreciation of its value, and might not conduct some cognitive operations to make revisions despite their claim of understanding the feedback.In addition, students might value the worth of peer feedback and incorporate it into their revision with low cognitive engagement.These conflicts might be attributed to some individual factors (metalanguage knowledge, confidence and motivation) and contextual factor (peer interaction).

Limitations
Although this research has enlightening significance for L2 writers' multimodal composing research and teaching, there are still some limitations.First of all, as these two participants were average students, the current findings cannot be generalized to students of other proficiency levels or in other pedagogical contexts.In addition, the retrospective interview data was collected after L2 learners revised their texts, and they might have been vulnerable to memory elapses over time.Therefore, further studies need to be conducted over a longer period, with more participants of differing proficiency levels, and in different pedagogical and social contexts.

Implications
It is critical for teachers to keep curricular and instructional practices that promote L2 writers' development of metalanguage for multimodal composing.The explicit instruction on multimodality will bridge old and new composing practices within various mediums across multiple contexts, and support students in representing and communicating ideas in a strategic way with various semiotic systems.Besides, given that multimodal composing knowledge is a prerequisite for high-quality multiliteracy education, teachers should become knowledgeable about well-designed curriculum and have an open mind toward novel writing techniques.
Figure 1.Pan's second presentation recording Figure 3. Pan's two drafts of slide 2 Figure 5. Zhao's two drafts of slide 5 Figure 6.Zhao's two drafts of slide 4