We are grateful to the Newcastle University (Faculty of Humanities and Social Sciences Research Fund, 2017 Spring call) and Université Clermont Auvergne (Foreign researchers–short research visit, 2018 call) for funding towards the project titled An examination of experienced online language teachers' multimodal instruction-giving practices. We would also like to thank our teacher and learner participants, Alsic editors and reviewers, and Professor Paul Seedhouse and Professor Nicolas Guichon in particular for their insightful comments on earlier versions of this manuscript or presentations.
1Online language learning is an area of growing interest and one-to-one or small-group classes are gaining popularity due to the flexibility of time and space they offer. Research in this area is also gaining momentum (eg Develotte et al., 2008; Hampel & Stickler, 2012; Kozar, 2016; Guichon, 2017), especially in investigating online language teachers' semio-pedagogical activity (Guichon, 2013), which involves skills and competencies in using "various semiotic and technological resources" (Guichon, 2017, p. 57). One such skill is giving task instructions. However, without careful consideration and planning, teachers' task explanations can be problematic due to distinct affordances and challenges of the online teaching platform. This paper investigates small group multimodal synchronous online language teaching via videoconferencing (Skype). We explore how experienced teachers deliver instructions for the same task.
2Research on teachers' instruction-giving practices is scarce. Emerging research has begun to explore written instructions in materials (Tomlinson & Masuhara, 2017) and verbal instructions in face-to-face (Markee, 2015) and online language teaching (Cappellini & Combe, 2017; Satar & Wigham, 2017). While much research has investigated trainee teachers' practices, the present study is unique in identifying how experienced online language teachers give instructions in multimodal synchronous online language teaching and the higher- and lower-level actions that constitute the instructions.
3Tasks engage learners in language use and elicit linguistic output (Ellis, 2000). Tasks typically involve "(1) some input (ie information that learners are required to process and use); and (2) some instructions relating to what outcome the learners are supposed to achieve" (Ellis, 2000, p. 195). Being an essential component of tasks, the delivery of instructions is important for several reasons. First, successful task completion "is often predicated on the effectiveness of [the] instructions" (Watson Todd et al., 2008, p. 26). Second, language learning happens in meaning-focused interaction during task completion (Nunan, 2004). Instructions offer opportunities for authentic communication (Watson Todd et al., 2008) and foster "immediate situational feedback" (Tomlinson & Masuhara, 2017, p. 343). Third, instruction-giving is part of task-based teaching competencies (Raith & Hegelheimer, 2010). Fourth, instructions may constitute a significant amount of teacher talk time (Ha & Wanphet, 2016). Finally, just as task-as-workplan may differ from task-as-process (Breen, 1987), spoken instructions may differ from planned instructions, ie task instructions-as-workplan may differ from task instructions-as-process. By understanding how experienced online teachers deliver task instructions in videoconferencing, we can guide teacher training, improve the quality of online language teaching, and establish best practice. Yet we discuss the challenges for the latter in our discussion and conclusions.
4In this paper, although we focus on what the teacher does, we do not intend to imply that instructions are a one-way process. We acknowledge that instructions can be collaboratively co-constructed with learners (eg Markee, 2015; St John & Cromdal, 2016) as an interactive process (Somuncu & Sert, 2019). We document data in relation to such interactivity in our analyses where relevant, but do not carry out a conversation analysis of the data. Our analytical method, multimodal interaction analysis (Norris, 2004) takes as its unit of analysis the mediated action. This is defined as a "social actor acting with/through mediational means" (Norris & Pirini, 2016, p. 21). Given multimodal interaction analysis' primary interest in revealing mediated actions of social actors, it enables a sole focus on teachers' actions and does not require a turn-by-turn analysis to demonstrate the sequential organisation of interaction, which would be the case when using a conversation analytic method (for more information on conversation analysis see Seedhouse, 2005).
5Using grounded theory (Strauss & Corbin, 1998) and multimodal interaction analysis (Norris, 2004), we address the following research questions:
1. What higher-level actions comprise experienced online language teachers' task instructions-as-process?
2. Which multimodal elements operate in the same higher-level actions employed by different teachers in their task instructions-as-process?
6In the following section, we introduce instructions within Task-Based Language Teaching (TBLT) and elaborate on multimodality in task instructions. We discuss the ways in which our focus on instruction-giving differs from instructional conversations (Meskill & Anthony, 2007; Tharp & Gallimore, 1991). We introduce Markee's (2015) task instruction fragments that form our theoretical framework and explain why we refer to them as higher-level actions, which is guided by our methodological approach.
7In this section, we review literature on teachers' instruction-giving practices, including studies that offer a multimodal perspective, much of which, however, is grounded in face-to-face classroom settings.
8Instructions can be written (Tomlinson & Masuhara, 2017), spoken (Markee, 2015; Seedhouse, 2008) or a spoken instantiation of written instructions (Ha & Wanphet, 2016). Regarding written instructions, Tomlinson and Masuhara (2017, p. 345-351) identify 11 criteria including succinctness, ie presentation of instructions "in the briefest and most concise way" (p. 348); specificity of instructions clarifying what to do and how to do it; and unambiguity avoiding pronouns, synonyms and using clear referrals. While written instructions are "static, pre-arranged and planned," spoken instructions are "dynamic, spontaneous, and unplanned" and enhance written instructions in the classroom (Ha & Wanphet, 2016, p. 152). Investigating two English-as-a-Foreign language (EFL) classrooms in which teachers provide instructions for the same task, Ha and Wanphet (2016) describe how verbal instructions complement instructions written on the materials and facilitate learner understanding of the task requirements. Furthermore, they demonstrate that teachers' spoken reformulation of written instructions serves a variety of functions, including calling students' attention, checking understanding, giving options and ideas, emphasising important information, helping learners process the instructions, and creating interaction (Ha & Wanphet, 2016). Some of these functions might be considered as essential task instruction fragments, which we demonstrate in our analyses.
9However, the word instruction can be misleading as it can be used to refer to different concepts. Lindwall, Lymer and Greiffenhagen (2015) distinguish between three types of instructions: instructions as education, as directives, and as written texts, such as user guides. Instructions as directives are procedural information targeted at "setting up tasks and making them followable" (St. John & Cromdal, 2016, p. 253) and this is the type of instruction we refer to in this paper. Instructions as directives can also be considered within instructional conversations (Meskill & Anthony, 2007; Tharp & Gallimore, 1991). Instructional conversations are dialogues "between teacher and learners in which the teacher listens carefully to groups of students' communicative intent and tailors the dialogue to meet the emerging understanding of the learners" (Tharp & Gallimore, 1991, p. 1). They are social and pleasurable interactive strategies which engage learners in thinking, meaning-negotiation and consequently learning. This understanding of instructional conversation aligns with the socio-cultural approach to language learning wherein learning occurs through interaction (Lantolf, 2001) and interaction is the content in language learning. Similarly, as Thornbury (2000, p. 2) puts forth: "Teaching–like talk–should centre on the local and relevant concerns of the people in the room, not on the remote world of coursebook characters, nor the contrived world of grammatical structures." Although instructional conversations highlight the quality of interaction or talk between learners and teachers, the practice contradicts guidance provided in most teacher training courses to minimise teacher-talk time.
10Exploring instructional conversation practices of teachers in an asynchronous online language course, Meskill and Anthony (2007, p. 11) demonstrate how an instructor "set[s] up the language learning task and orchestrat[es] instructional conversation around that task." In this paper, we set out to investigate the former, ie teachers' higher-level actions in "setting up the language learning task" and exclude the latter, ie those in "orchestrating instructional conversation around that task" (Meskill & Anthony, 2007, p. 11). As such, we do not address the role of the teacher or learners during the task interaction, nor the impact of how the task is set up on learner or interactional outcomes (eg learning, interactional dynamics, or ensuing instructional conversation), which are beyond the focus of this paper. However, we have observed elements of instructional conversation even in this initial setting-up-the-task stage (eg Extract 3). While this could be due to different teaching styles of the participating teachers (see Section 4), the videoconferencing context with participant images being constantly present on the screen might be generating pressure for teachers to be genuine participants in the ongoing social interaction.
11Previous research in face-to-face classrooms has explored spoken task instructions. Seedhouse (2008) demonstrates how successful experienced teachers "create a pedagogical focus, that is, to get students to do what they want, in an apparently effortless manner" (2008, p. 42) using "instructions … full and explicit as possible whilst presenting a single, undiluted focus" (2008, p. 55). In another study that investigated an experienced teacher's classroom teaching, Markee (2015, p. 120-121) identified six task instruction fragments, which inform the learners about: (1) how they will be working (in dyads or small groups), (2) what resources they will need, (3) what tasks they have to accomplish, (4) how they will accomplish the task, (5) how much time they have to accomplish these tasks, (6) why they should do something. In this paper, we build on Markee's (2015) fragments. In this paper, in line with our analytical lens–multimodal interaction analysis (Norris, 2004) which scrutinises mediated higher- and lower-level actions–Markee's (2015) fragments constitute various higher-level actions in which the teachers engage while giving instructions. On the other hand, the modes they employ in doing so form the lower-level actions. Thus, this study demonstrates the higher-level and lower-level actions that comprise task instructions-as-process.
12Despite the key role spoken instructions play in language teaching, research focusing on instructions in online audiovisual language teaching is rare. Indeed, we identified only three studies. The first, by Codreanu and Combe Celik (2012) touches upon online tutors' use of pre-prepared instructions not only as a quick means to launch activities, using a paste-and-copy technique, but as memos-to-self to help them refocus interactions on the pedagogical objectives. The second, by Cappellini and Combe (2017), explores teachers' techno-semio-pedagogical competence in two online environments: an asynchronous learning management system (Moodle) and a synchronous webconferencing platform (Adobe Connect). The authors show how trainee teachers modify instructions according to mode as they negotiate instructions in conversation with the learners in the webconferencing platform. They also underline challenges of the online environment and how trainees may unexpectedly need to deliver technical instructions concerning task resources (downloading a document). However, they do not investigate experienced teachers' practices, explore the different fragments constituting instructions, nor analyse instructions' multimodal nature.
13The third study is our previous work on the use of multimodal resources in instruction-giving by trainee teachers, which we explain in the next section.
14An increasing number of studies explore multimodality in face-to-face classroom teaching (eg de Silva Joyce & Feez, 2018). Studies of language classrooms have explored how posture shifts and mutual posture alignment signal task launch (Hellermann & Pekarek Doehler, 2010) and the ways in which participants "collaboratively converge on courses of action" (Markee, 2015, p. 127) through embodied actions such as eye gaze, gestures, and orientation to cultural artefacts. In a recent study, Somuncu and Sert (2019) focus on trainee teachers' orientation to learners' non-understanding of instructions and demonstrate how trainees successfully manage non-understanding by employing physical and digital visual artefacts.
15Despite an interest in multimodality in online language learning, including a focus on gaze (Satar, 2013; Shi, Stickler, & Lloyd, 2017), gestures (Lee et al., 2019; Satar, 2016; Wigham, 2017), silence (Kozar, 2016), and webcam impact (Cohen & Wigham, 2018; Guichon & Wigham, 2016; Hampel & Stickler, 2012), very little attention has been paid to the multimodal nature of task instructions in online language teaching. Due to restrictions imposed on the interactional space by the two-dimensional webcam frame–with limited access to gaze direction, visible gestures within the webcam frame, establishment of mutual gaze, and differences in proximity (distance)–teachers' multimodal practices in instruction giving can differ significantly from their face-to-face practices. Our earlier work (Satar & Wigham, 2017), in which we observed trainee teachers' instructions for a role-playing task remains the prime example in the area. We showed how trainee teachers capitalise on non-linguistic semiotic resources including word stress, gaze, gestures, proximity to the webcam, and text-chat to mark different instruction-giving stages, allocate roles, and introduce key vocabulary. We observed deployment of text-chat and digital talk-external artefacts such as online pictures, videos, and websites during task accomplishment and demonstrated how trainee teachers appropriated such resources when performing pedagogical actions during instruction-giving.
16In short, few studies have explored the multimodal aspects of language task instructions either in face-to-face or online teaching via audiovisual platforms. Moreover, research on task instructions is almost non-existent in online language teaching and only involves an examination of trainee teachers' practices. This study sets out to examine how experienced teachers deliver task instructions online.
17Following a qualitative research design, we draw on qualitative and micro-analytic analyses of screen-recorded lessons to understand the complexity of instruction-giving. A semi-controlled corpus (Tellier, 2013) was constructed from online lessons, the basis of which was the same task. Thus, the corpus allowed the researchers to control for variation in task type and participant characteristics, which enabled a qualitative comparative analysis.
18Three experienced online English teachers (one male, two female) were recruited in Spring 2018 from two online language teaching providers: iTalki (Chen & Jiang, 2007) and SpeakPlus (Paniza & Barry, 2014). All volunteer teachers had a teaching qualification. They had a minimum of two years online teaching experience (see Table 1) and regularly taught lessons using the videoconferencing platform Skype (Skype, n.d.).
Table 1–Teacher profiles.
- 1 As part of the project, a further three learners participated in the lessons where one of the teach (...)
19Six1volunteer language learners studying a foundation-level English course at a higher education institution in Turkey were recruited. They had a B1-B2 CEFR (Common European Framework of Reference for Languages: Learning, Teaching, Assessment; Council of Europe, 2001) level and their motivation to participate was the opportunity to practise speaking skills because they did not have many opportunities for such practice in their face-to-face classes. Indeed, the call for participation emphasised this opportunity. The online classes were completed outside of their institutional settings and were not graded. Two learners were allocated to each teacher. Group composition was based on learners' availability.
20Each teacher held three 60-minute lessons, conducting the introductory lesson with their preferred activities. For the subsequent lessons, they were provided with exactly the same task resources (task-as-workplan) for a convergent and a divergent task (Ellis, 2003). No additional instructions to those presented on the resource sheets were shared with the teachers (Appendix A); they could introduce the tasks in the way(s) that suited their own practices.
- 2 Sarah completed this lesson over two interactions. During the first interaction, there were many te (...)
21Data for this paper come from the second lesson2, where the convergent task (Appendix A) was used. The task encourages learners to reach a consensus in order for a reasonable solution to be produced (Wegerif et al., 1999). We chose to study this task because in convergent tasks learners need to reach a single outcome collectively, meaning that they both need to understand the task information and the instructions correctly. The task is divided into two micro-tasks: an information exchange activity during which learners must compare two gift-package deals for a colleague's leaving present and decide upon one; and a collaborative email writing activity asking other colleagues for financial contributions towards the purchase of the gift package.
- 3 In later images, the image of the researcher is her profile picture and not a still shot of her web (...)
22Primary data sources were screen recordings of the online lessons using Snagit (Snagit, n.d.) by the teachers and, to minimise potential data loss, by a researcher who participated as a silent observer with muted microphone and camera3. Table 2 presents the details of the semi-controlled corpus.
Table 2–Screen recordings collected for lesson 2 and lesson length.
- 4 Sarah completed this lesson over two interactions, with a total lesson length of 87 minutes. See No (...)
Teacher
|
Screen recording by:
|
Length of the recordings
|
Total amount of data
|
Craig
|
The teacher
The researcher
|
60 mins
60 mins
|
120 mins
|
Karen
|
The teacher
The researcher
|
60 mins
60 mins
|
120 mins
|
Sarah
|
The researcher
|
50 mins + 37 mins4
|
87 mins
|
23Ethical approval was obtained from the university ethics committee and all participants gave informed consent. Pseudonyms are used for personal information. Some images are blurred according to preferences stated in participants' consent forms.
24To answer the research questions, first, we used grounded theory analysis (Strauss & Corbin, 1998) to identify which fragments comprised teachers' instructions. Informed by Markee's (2015) instruction fragments classification, but staying close to the data via multiple iterations of watching the screen recordings, we employed a bottom-up approach. We tagged the data in ELAN (Sloetjes & Wittenburg, 2008) with descriptors of the higher-level actions that emerged. During the open-coding stage, we generated categories by grouping similar items and defined and developed them until we reached theoretical saturation where we did not observe any new categories. Through constant comparison, we then related the categories to refine our higher-level actions and searched for variation within and between categories. In our analysis of instruction-giving as-process, we observed that teachers decided to divide the macro-task into smaller steps. Our analysis covers the instructions for all steps. However, we excluded teachers' task facilitation guidance through directives and questions once the learners began engaging with the task. For Craig and Karen, our analysis was largely based on the teachers' own screen recordings but, as this was not available for Sarah, we used the researcher's screen recording. Although this does not present any issues for the data presented in this paper, it means that we did not have access to all changing screens and actions that might be taking place on Sarah's end of the online interaction (eg accessing other documents, switching to another window) and the data might be representing a semiotic lag (Wigham & Satar, in press) regarding time difference between Sarah's actual actions and when those actions were received on the researcher's screen.
25Second, to examine which multimodal elements operate within the same instruction-giving fragments, we offer a detailed micro-analysis of data extracts using multimodal interaction analysis (Norris, 2004), which scrutinises mediated actions (unit of analysis) within social practices comprising lower- and higher-level actions. Lower-level actions are "the smallest interactional meaning unit" (Norris, 2004, p. 1), or modes, and higher-level actions are "a chain of lower-level actions, with an opening and a closing" (Norris & Pirini, 2016, p. 5). Within this method, we considered each instruction-giving fragment as a higher-level action, while the lower-level actions constituted various modal units, for instance, gaze shifts, spoken utterances, head movements, etc. (See Norris, 2004, for a full explanation of the different modes and Satar & Wigham, 2017, for an explanation of different types of gestures, gaze, etc.). Extract selection was based on several criteria. First, we focused on "sequences in which the teacher produces instructions" (Seedhouse, 2008, p. 6). We also looked for sequences that illustrated the new fragments that emerged in our grounded theory analysis. Then, we refined our selection to sequences that were typical examples of the chosen fragments, but also multimodally salient to demonstrate how various modes operate together. It was also important to demonstrate variety in all teachers' practices for the same fragments. Three fragments revealed to be most appropriate for these purposes: communicating key task information, suggesting ways into task, and launching the task. Once the segments for micro-analysis were selected, they were transcribed (Appendix B) and analysed using multimodal interaction analysis focusing on the lower-level actions observed in each higher-level action.
26Measures were taken to augment the credibility of our qualitative analyses including seminar presentations to receive feedback on potential different interpretations of data. Following independent tagging and discussion, collaborative data coding represented full agreement between the researchers. For triangulation and respondent validation (Hammersley & Atkinson, 2007), we conducted post-lesson interviews with the participants and sought feedback from participating teachers.
27Before addressing our research questions regarding the higher-level actions that comprise experienced online teachers' task instructions (section 4.1) and the multimodal elements that operate in the same higher-level actions of instruction-giving (section 4.2), we feel it is important to give an overview of our observations regarding the teachers' (Craig, Sarah, and Karen's) differing instruction-giving styles and instruction management practices to contextualise our analyses.
28Craig dedicated about half the task time to instructions and approached instruction delivery as a venue to foster authentic communication. He employed the strategy of using instruction-giving actions for task contextualisation, personalisation, and humour, thus developing an interpersonal relationship with the learners. During instructions, he was highly present socially with expressive facial features, brisk pace of speech, and little silence. We observed that the instruction-giving phase was characterised by teacher talk and interactivity largely constituted requests for minimal responses from learners to confirm understanding either as minimal response tokens or head nods.
29Karen adopted the technique of eliciting instructions from the learners and approached instruction delivery as an opportunity for learning to learn by focusing on study skills (see 4.4.1). For example, she asked learners to identify different sections of the written instructions to facilitate work with any written task instructions. This approach allowed the learners to occupy nearly one third of the floor space during instruction-giving and also allowed Karen to check learners' understanding of the task instructions, be responsive to misunderstandings, and, when necessary, offer immediate feedback.
30We observed that Sarah's strategy was to ask the learners to read the resource sheet, confirm their understanding, and facilitate task completion once the task was underway rather than explaining all requirements upfront. Sarah conducted the lesson over two separate occasions. In the first attempt, one learner, Demet, was late for the lesson, had technical problems, and refused to read the resource sheet, hence asked for more time to focus on the resource sheet individually. Although the other learner, Sevil, had understood the task, her attempt to explain the task to Demet also failed. The teacher, Sarah, decided to reschedule the lesson and approached the instructions as written directives in the task resource to be read beforehand by the learners, and then orally confirmed learners' understanding of the instructions during the lesson. Indeed, at the beginning of the second attempt, both learners had read and understood the resource sheet and it did not take them long to begin the task following a recap of the instructions that Sarah elicited. Instructions in the first attempt were more teacher-talk centred and predominantly involved getting learners to access and read the resource sheet. Student talk in the rescheduled lesson was greater and the instruction phase was more dialogical: as both learners had read the task information, Sarah asked one learner to summarise it and then briefly recapped the key points which appeared to prove efficient.
31We now turn to our analysis of higher-level actions that comprise the three teachers' task instructions-as-process and the multimodal elements that operate as lower-level actions within these.
32Following grounded theory analysis, 13 categories of higher-level actions in task instructions-as-process emerged from our data. Figure 1 demonstrates these actions, six of which (represented in beige) existed in Markee's (2015) instruction-fragments, whereas seven of these (represented in blue) are new in our study. We further observed sub-categories for three of the existing actions (represented in green), some of which, within the managing resources action, related specifically to online teaching (sending the resource, receiving the resource, opening the resource).
Figure 1–Higher-level actions in task instructions-as-process.
33While instruction fragments specified by Markee (2015) include an identification of task resources, online teachers in our study needed to go beyond identification and actively manage the task resources. This comprised a range of further higher-level actions such as allocating different documents to different learners and instructing them on how to access the resource sheets (Wigham & Satar, in press). We also observed that online teachers made different task stages salient through summarising previous task steps as well as forward-organising subsequent steps. Other higher-level actions included clarification of key task information (such as different learner roles and vocabulary); a focus on study skills during task instructions, for instance by explicitly stating the task type; activating schemata for the task by contextualising and personalising the task; and checking learner understanding of the task instructions before launching the task. The micro-analysis presented in the next section demonstrates how online teachers communicated key task information, suggested ways in which learners can approach the task, and launched the task. Examples of all higher-level actions presented in Figure 1 can be found in Appendix C.
34We present three micro-analyses to illustrate the multimodal construction of three higher-level actions that comprise task instructions-as-process: communicating key task information, suggesting ways into task and launching the task. In the analyses, the multimodal elements represent the lower-level actions as defined by Norris (2004). Figure numbers refer to image numbers in the corresponding extract.
Extract 1–Student A is different to Student B.
35Extract 1 illustrates Karen's use of the spoken language, gestures, gaze, posture shifts, and proximity for the higher-level actions of communicating key task information, followed by suggesting ways into task, and launching the task. In this extract, the male learner is Erol. Figure 5 (line 4) presents the teacher's screen layout.
36Key task information is that student A and B have different information on their resource sheets. Karen communicates this linguistically and para-linguistically. In lines 1 and 2, she utters "different" three times and accentuates this through gestures. Figures 1 and 2 (line 1) show Karen's iconic gestures (McNeill, 1992), ie hands with palms turned towards the webcam which represent students A and B. This information is foregrounded in line 2; Karen's static iconic gestures are repeated and become dynamic as she moves her hands forwards and backwards. Thus, key task information is foregrounded in learners' awareness/attention and becomes the focal point of attention. This is achieved by the joint employment of language and the static and dynamic iconic gestures, resulting in high modal density.
37In line 3, in the spoken language mode, the discourse marker "so" indicates the teacher is moving into a different higher-level action: suggesting ways into task. Language, gestures, and posture shifts operate within this higher-level action. Karen manages learner interaction in lines 4 and 5, by using learner names, ie vocatives in the language mode; she nominates Gonca to take the first turn, followed by Erol. Other lower-level actions employed for turn-management include Karen's deictic gestures (Figures 5 and 6; McNeill, 1992) and posture shifts (Figures 8 and 10), which are directed towards the learners' images on Karen's screen (Figure 5). Figure 7 shows her iconic gesture to illustrate the task resource ("piece of paper") on which Gonca can find the information to explain to Erol. Using lower-level actions in the modes of gesture, posture, and language, Karen suggests potential ways into task: in line 5, Karen's language use tells Erol what he can do in response to the information Gonca provided. By shifting her posture and directing her metaphoric gestures for "yes" and "no" (Figures 9 and 10) towards different areas of the screen, Karen foregrounds one way to accomplish the task, ie comparing and contrasting information. Therefore, in lines 4 and 5, Karen emphasises important elements of the higher-level action through high modal density; ie who ("Gonca", "Erol") is going to do what (explain, compare, contrast).
38In line 6, Karen confirms (Figure 11) student understanding through an emblem (Kendon, 1982), ie a thumbs up gesture. The linguistic output in line 7, "okay", can be considered as both confirmation of understanding and moving into a new higher-level action of task launch through floor withdrawal. Karen leaves the interactional space to the learners via her language unit ("off you go") and proxemically by moving away from the webcam, thus increasing the interpersonal space (Andersen, 2008) (Figure 12). Spoken language and posture shifts are the multimodal elements that operate within the higher-level action of launching the task.
39Extract 1 presented how Karen employs the modes of spoken language, gestures, gaze, posture shifts, and proximity for the higher-level actions of communicating key task information, followed by suggesting ways into task, and launching the task. Extract 2 illustrates how Sarah communicates key task information, focuses on task accomplishment, explains how learners will work, and launches the task. The extract is from Sarah's rescheduled lesson and begins with Sarah asking Sevil to read the resource sheet. It evidences how instructions are elicited from the learner and summarised by the teacher.
Extract 2–Can you read the beginning of the task?
40In Extract 2, lines 1 and 2, Sarah asks the learner, Sevil, to read the task out loud (from the resource sheet in the print mode). What follows (lines 3-21) is a series of short turns by Sevil explaining (rather than reading) sub-task outcome, key task information, and task accomplishment, which are supported by Sarah's backchannels (continuers) comprising either a single lower-level action (eg a head nod) or multiple lower-level actions (eg language: yes and facial expression: smile). In line 22, Sarah summarises the information Sevil gave, emphasises how the learners will be working ("together"), and simultaneously monitors both learners with frequent gaze changes (line 22, Figures 1-6). In line 22, Figure 7 is an image of Sarah's screen during this interaction, which demonstrates her gaze directed towards Sevil (Figures 1, 3 and 5) and Demet (Figures 2, 4, and 6). Thus, the higher-level action for the instruction-giving fragment defining roles: explaining how students will be working is emphasised through high modal density achieved via a combination of the modes of language and gaze. Sarah repeats this higher-level action in line 36 employing the same modal units of language and brisk alternation in gaze direction (Figures 13-16) and head movement by tilting her head towards the learners.
41Between lines 23 and 31, Sarah repeats key task information. The fact that learners have different information is not explicitly stated in the language mode but we again observe Sarah's brisk gaze alternation between the learners (lines 24-25, Figures 8-11). Sarah then clarifies how the learners will accomplish the task: "talk about it" (line 27) and "discuss it" (line 28), which is followed directly by stating the task outcome (lines 29, 30, 31). This is accompanied by head nods (lines 28-30), serving as beat gestures (McNeill, 1992) that increase the modal density for this higher-level action.
42The following higher-level action in Extract 2 is checking understanding (lines 32-33). We observe the modes of spoken language, gaze, and head movement operating in this higher-level action. In the spoken language, Sarah says "yes" with a rising intonation and pauses briefly after each word. Sarah's gaze direction shifts between the learners (line 32, Figure 12). Sarah's slow head nod (lower-level action) supports the higher-level action of checking understanding (line 32, Figure 12).
43The instruction-giving phase concludes with the next higher-level action: launching the task predominantly in the mode of spoken language (line 34). It also comprises a repetition of other higher-level actions including focusing on task accomplishment, communicating key task information, formulating task stages, and defining roles.
44Similar to Extract 1, we also observe a confirmation check (checking understanding) in Extract 2 preceding a clear task launch. However, while in Extract 1 Karen foregrounds her gestures, in Extract 2 Sarah's gestures are not visible. Instead, her alternation of gaze direction and head movements operate in the background. In the next extract (Extract 3), we focus on the multimodal elements of Craig's instructions for the same higher-level actions.
Extract 3–Anne Watson is changing jobs.
45Extract 3, Figure 2 demonstrates Craig's screen layout. The two learners are Didem (positioned in the top right) and Eda (positioned below Didem's image). The researcher's profile picture is on the top left. Prior to Extract 3, Craig had sent two links via text-chat which pointed to online versions of the task resource sheets. The extract begins with Craig describing which part of the resource he would like the learners to look at as he reads the task context from Student A's resource sheet in a Google Document. In line 1, Figure 1, we see Craig highlighting a section of this resource, thus the print mode operates within this higher-level action of managing resources.
- 5 We adopt this term following Norris (2004) who explains that a person is aware of something s/he is (...)
- 6 Due to Craig's screen layout, Craig's gaze shifts are subtle in his recording, but the upward movem (...)
46In lines 1-3, the print mode is in the foreground of Craig's attention/awareness5 while he manages resources, identifies the task rationale, and states the task outcome. In line 4, his attention shifts from the print mode to the interactional space (Figure 2) when he moves to the next higher-level action: communicating key task information (lines 4-7). Several modes bring this information to the foreground within the learners' awareness/attention through increased modal density, which are (a) in the spoken language mode: key adjectives (line 5: "similar but slightly different''), vocatives (lines 6-7: addressing the learners with their names), and intonation (underlined words in lines 4-7), (b) in the gesture mode: iconic gesture to highlight "two" (Figure 4), a beat gesture while the iconic gesture for "two" is sustained by moving the hand forwards and backwards (Figure 5), and an iconic small and closed gesture used to accompany "slightly" (Figure 6), (c) in the mode of facial expression: an animated, expressive face with eyebrows moving upwards and downwards (Figures 3-9), (d) in the gaze mode6: shifts in gaze direction (Figures 7-8 towards Eda, Figure 9 towards Didem).
47Lines 1 and 2 also illustrate how Craig adds comments during instructions: a personal comment ("my goodness") and a comment that situates the learners already within their task roles (through the use of the second-person singular pronoun) to explain why learners will do the task ("because you're good colleagues yeah you're all good friends"). We observe Craig introducing further personalisation in lines 7-8 when he embarks on an off-topic sequence about his use of the retail website Amazon. While the extra information Craig offers in lines 2, 6, and 7 is task related and achieves the higher-level actions of identifying task rationale, ie explaining why the learners will do the task and serves to communicate key task information; comments in lines 1 and 8 are personal.
48During lines 9-12, Craig repeats two higher-level actions (1) stating task outcome and (2) communicating key task information. Lines 13-15 demonstrate an attempt to announce next task stage, how the learners will accomplish the task, and a paraphrase of task outcome using head-tilt (a directional shift) in the mode of head movement which accompany the stressed words "your" (Figures 10, 11). The head tilt functions as a deictic gesture referring to each learner.
49In line 17, Craig suggests a potential way into task. Here, the lower-level actions are the words in spoken language, some of which are stressed through intonation, facial expressions, an iconic hand gesture (Figure 12) to represent "together," and a beat facial gesture (Figure 13) with raised eyebrows to stress "about."
50Lines 19-21 illustrate how the higher-level action: launching the task is achieved. Similar to Extract 2, this sequence incorporates other higher-level actions. First, Craig refers to the print mode he employed at the beginning of the extract and initiates this stage with a description of the resource (lines 19-20). Next, he allocates time for a silent period for task preparation and repeats his suggestion for one way into the task (line 21). In this extract, spoken language is the predominant lower-level action for the higher-level action: launching the task and a shift in modal density, which was high throughout the extract and reduced for task launch, leads to silence for task preparation.
51Instructions are an important aspect of task-based language teaching as they may predict success in task accomplishment (Watson Todd et al., 2008). This study explored instruction-giving practices of three experienced online teachers' task-based language lessons. We now discuss our findings in relation to the research questions.
52Overall, our study demonstrated different approaches to instruction delivery; (1) instructions as a venue to foster interaction (Ha & Wanphet, 2016), authentic communication, and opportunities for learning to learn, (2) instructions as directives for task completion, which is the main venue for learning, or (3) instructions presented as written directives in the task resource to be read beforehand by the learners and orally confirmed during the lesson. Here, we also need to question the criteria for interactive instructions. Should teachers provide rehearsed, full, explicit (Seedhouse, 2008), succinct and unambiguous (Tomlinson & Masuhara, 2017) directives and ask for minimal responses to confirm understanding, or aim to engage learners fully in instructional conversations (Meskill & Anthony, 2007) and even elicit the instructions from learners? We suggest if instructions are only intended to be directives, ie getting the learners to do something, then teachers can perhaps plan them carefully, spend as little time as possible, and offer simple, unambiguous instructions, or ask learners to engage with them prior to the lesson (eg Sarah). However, if instructions are deemed as opportunities for interaction and learning, they can occupy more time and be designed to be interactive through elicitation (eg Karen) or capitalised on as venues for establishing interpersonal relationships by exchange of personal information, anecdotes, or humour (eg Craig) through instructional conversations.
53Building on the six instruction-giving fragments in face-to-face classrooms identified by Markee (2015), our data illustrated 13 higher-level actions in task instructions-as-process. We provided explanations and examples of these in Appendix 3 and explained how several of these higher-level actions unfolded in interaction in our multimodal micro-analyses. We indicated that managing resources stood out as a highly important higher-level action for the online context. Similarly, Cappellini and Combe (2017) emphasised the semio-techno-pedagogical skills of trainee language teachers in guiding learners to access task resources. We further explore this higher-level action in a forthcoming publication (Wigham & Satar, in press).
54Three extracts in section 4 demonstrated that each teacher capitalised on the affordances of different modes given their preferred visual framing (Guichon & Wigham, 2016). While Karen's visual space projected through the webcam was a head-and-torso shot, which allowed her to effectively use her hand gestures, as well as her posture in relation to sideway shifts and changes in proximity, her gaze shifts were less observable. In contrast, Craig's visual frame was a head-and-shoulders shot, which foregrounded his facial expressions allowing him to employ various features of his face, including gaze, but especially his eyebrows for emphasis. However, being closer to the webcam, his gestures were not always visible. Craig's practices illustrated how teachers can benefit from the print mode by asking learners to refer to documents shared online. Both Karen and Craig illustrated effective ways of combining multiple modes with spoken language to achieve emphasis. Finally, Sarah's visual frame was a close-up shot whereby her shoulders and hands were never visible. Within this framing, she predominantly employed the spoken language mode accompanied by gaze shifts, head nods, head tilts, and smiles. In the spoken language mode, Sarah's pace was slower with salient pauses, especially compared to Craig and Karen's brisk pace. In our previous work (Satar & Wigham, 2017), we observed similar employment of multimodal elements by trainee teachers who used vocatives, gaze shifts, and head tilts to allocate roles and postural shifts to demonstrate withdrawal from the interactional space to launch the task.
Table 3–Teachers' framing and predominant multimodal elements for instruction-giving.
Teacher
|
Framing category
(Guichon & Wigham, 2016)
|
Predominant multimodal elements
|
Karen (Extract 1)
|
Head-and-torso shot
|
spoken language,
hand gestures,
posture shifts,
proximity shifts
|
Craig (Extract 3)
|
Head-and-shoulders shot
|
spoken language (strong intonation),
print (reading the task resource),
facial expressions (especially eyebrows),
hand gestures,
gaze shifts
|
Sarah (Extract 2)
|
Close-up shot
|
spoken language,
gaze shifts,
head tilts,
head nods,
some facial expressions (smiles),
print (asking learners to read the task resource)
|
55Overall, by employing various modes, the teachers achieved increased modal density in their instructions and thus presented the instructions as the focal point of attention by foregrounding them in the learners' awareness/attention (Norris, 2004). This could be one reason why all learners successfully completed the task regardless of differences in how instructions were delivered.
56Online teaching offers increased access to language learning, especially for learners from remote areas, who have limited mobility, who do not have teaching services nearby, or busy learners who need flexibility regarding lesson time and place. Therefore, high-quality, effective online language teaching has the potential to have a large socio-economic impact, yet we do not know much about the practices and skills of experienced online language teachers. This study bridges this gap by investigating experienced online language teachers' delivery of task instructions in an online multimodal space using a qualitative research design.
57Our study demonstrated variety in teachers' multimodal composition of higher-level actions linked to task instructions-as-process within respective visual framing. All learners in our data successfully completed the task: They all exchanged information, made a decision, and composed an email albeit within various lengths of time or quality. Regarding theory, although research and pedagogical advice for instructions focuses on brevity and succinctness, we questioned whether instructions could be exploited as a venue for authentic, interpersonal, and pedagogical interaction (eg in the form of instructional conversations); designed interactively and elicited from the learners; or implemented in task facilitation through interaction with the learners during task completion. Our data come from an online context; however, such questions have not yet been answered in face-to-face contexts, either. Future research attempting to answer such questions would need to identify success criteria before drawing conclusions, ie whether the success of instructions will be based on an evaluation of task outcomes (cognitive, social, affective, or task-based), or, following Wigham and Guichon (2019), it will be related to time (eg task completion or teacher-talk time). Existing best practice recommendations seem overly simplistic given the complexity of instruction-giving.
58Our findings are limited to the small data set generated as a semi-controlled corpus. Data from a larger set of fully naturally-occurring data could help generalise the results. In line with Shi, Stickler, and Lloyd's work (2017) amongst others, other data collection methods, such as eye-tracking, would also be useful in multimodal analysis to track where learners' gaze is directed during and after the instructions. Concerning the number of students in our setting, we incorporated two learners and one teacher in each lesson to allow for comparisons with our previous research, yet we have also collected further data to investigate differences in relation to learner numbers, which will be reported in future publications. Moreover, our participating teachers had a minimum of two years' online teaching experience. However, none had received any specific training to teach online and in one case (Sarah) their previous experience was largely based on teaching without the webcam as it was more convenient due to slow Internet speeds. Thus, our findings emphasise the importance of teacher training especially regarding semio-pedagogical competence, which relates to teaching competencies and skills in employing "various semiotic and technological resources" (Guichon, 2017, p. 7). Our analysis of lower-level actions that operate in task instructions-as-process shows that each teacher employed different levels of multimodality when giving instructions in online classes. As Chun, Kern, and Smith (2016, p. 5) remind us each communication tool "brings its own material properties, feel and techniques of use, affordances and limitations," thus pedagogical skills developed in face-to-face training are not always transferable to online teaching contexts which have different affordances and limitations. Finally, our participating teachers' understanding of their roles varied slightly from keeping learners happy, providing grammatical and study skills support, to task completion, which could have impacted on their instruction-giving practices. These views were expressed in the interviews conducted, which we have not reported in this paper due to space restrictions.
59Instruction-giving in synchronous online language teaching is an area ripe for future research. Beyond our previous suggestions, other potential directions include exploring (1) affordances and challenges of online environments for instruction-giving, eg managing resources, (2) the interactive, collaborative nature of instructions within a recipient design focusing on how misunderstandings are resolved, (3) identifying successful ways of providing instructions against predetermined success criteria, (4) impact of task-repetition by the teacher or different number of learners in the lesson on higher-level actions employed in instructions-as-process, (5) task instructions in completely naturally-occurring settings, (6) instructions for task facilitation once the learners are engaged in the task, (7) personalisation and contextualisation during instructions, (8) managing time in relation to instructions, and (9) the impact of task design on the higher-level actions in instructions-as-process required for task completion.
60By focusing on one aspect of online language teaching, this paper has demonstrated the multimodally complex nature of higher-level actions comprising instruction-giving practices of experienced online teachers. An investigation of different teachers' instructions for the same task enabled us to present variety in task instructions-as-process while illustrating certain patterns in higher- and lower-level actions observed across different teachers' practices. We have proposed several future research directions, which will significantly enhance our understanding of instruction-giving as one aspect of online language teachers' semio-pedagogical skills.
61Both authors contributed equally to the manuscript.