Creating a response space in multiparty classroom settings for students using eye-gaze accessed speech-generating devices

Abstract

Conversation Analysis was used to explore how teachers, personal care assistants, and students organized inclusive multiparty classroom interaction when one of the students in the classroom used an eye-gaze accessed speech-generating device (SGD). Scaffolding and collaborative practices that created a response space for the construction of the eye-gaze accessed SGD-mediated turn were identified and analyzed. The participants were two adolescent students with severe cerebral palsy and intellectual disability who relied on eye-gaze accessed SGDs, and their teachers, personal care assistants, and classmates with intellectual disabilities. The data consisted of 2 hr and 40 min of video recordings collected in the participants’ classrooms. Three practices were identified (a) the practice of explicit turn allocation organization and the use of display questions, (b) the practice of locally contingent on-screen scaffolding activities, and (c) the practice of dealing with turn competition by classmates. Teacher and assistant practices differed with regard to the student’s access to the vocabulary relevant to answering the teacher’s question. The practices were found to create a response space for students using SGDs accessed via eye gaze, thereby ensuring their educational inclusion in the classroom.

Keywords:

• Cerebral palsy
• conversation analysis
• eye-gaze
• inclusive classrooms
• speech-generating device

UNESCO (1994) and the United Nations Human Rights (1989) agreed that all children, regardless of their socioeconomic circumstances, ethnicity, gender, or functionality, have the right to express their opinions and to participate in inclusive and equitable education. Yet, inclusive social interaction involving students with complex communication needs depends heavily on augmentative and alternative communication (AAC), for example, the use of speech-generating devices (SGDs) (Beukelman & Mirenda, 2013); however, research on how SDG-mediated interaction is achieved in the context of classrooms is limited. Teachers, teacher assistants, and personal care assistants (henceforth “assistants”) reported on the challenges in terms of social and pedagogical inclusion of students with complex communication needs (Andzik, Chung, & Kranak, 2016; Tönsing & Dada, 2016). In addition, students who rely on SGDs described the difficulties in achieving an interactional space (i.e., enough time to produce an SGD-mediated initiative or response) in the classroom (Howery, 2018).

Producing an SGD-mediated contribution is time consuming. In their detailed analysis of SGD-mediated interaction in home settings, Savolainen, Klippi, Tykkyläinen, Higginbotham, and Launonen (2020) found that the process of constructing an SGD contribution took up to 81 s. A few studies have analyzed in detail how the participants handle this lengthy production process. On the one hand, Savolainen et al. (2020) found that the communication partners (i.e., mothers) used collaborate practices; they waited in silence or participated in non-competing activities while their children constructed their SGD contributions. On the other hand, previous research also demonstrated how the prolonged time producing an SGD-mediated contribution could lead to temporal misalignments (i.e., sequential collapse of shared reference and mutual understanding) between the communication partner and the participant relying on the SGD, which resulted in misunderstandings (Clarke & Wilkinson, 2009; Higginbotham & Wilkins, 1999). In order to manage the progression difficulties and to locate the SGD-mediated contribution in a recognizable and sequential position, Clarke and Wilkinson, (2007) found that the communication partners (i.e., peers) framed their contributions within adjacency pairs (e.g., question-answer sequences) and produced meta-interactional prompts that were explicit pre-invitations of a next action, such as “Ask me a question about X.”

Although research using detailed analysis has provided new insights into the use of SGDs in dyadic interaction, there is a lack of studies investigating participants’ practices of turn allocation in multiparty (i.e., four-party or more) interaction. Turn allocation practices in ordinary conversation were first described in a seminal paper by Sacks, Schegloff, and Jefferson (1974), in which they demonstrated that the number of participants affected how turn allocation was accomplished. In dyadic conversation, the task of identifying a recipient was unambiguous; in triadic (i.e., three-party) interaction, one of the recipients could be excluded; and in multiparty conversation, the participants were faced with more options: all co-participants were possible recipients, or only one. Furthermore, Jefferson (1972) demonstrated that multiparty interaction could split up in side sequences, and Egbert (1997) illustrated that non-competing parallel conversations could develop (i.e., schisming) within multiparty interactions. Furthermore, regardless of the number of participants in an interaction, participants without disabilities use syntax, prosody, and gaze in a systematic way to hold the floor (i.e., to continue to talk), to select a next speaker and to disambiguate recipiency (i.e., readiness to take the turn) (Goodwin, 1980, 2000).

To date, only a few studies using detailed interaction analysis have included SGDs accessed by eye-gaze technology (Engelke & Higginbotham, 2013; Hörmeyer & Renner, 2013). Eye-gaze technology can enable activity and participation in social interaction for participants with severe physical impairments (Perfect, Hoskin, Noyek, & Davies, 2020). However, technology uptake depends on the communication partner’s attitudes, technological skills, access to support, and opportunities for training in combination with preconditions (e.g., vision, cognition, and motivation) of the participant who relies on the SGD (Perfect et al., 2020). Previous research on eye-gaze accessed SGD-mediated interaction focused on adult dyadic or triadic interaction, including participant without intellectual disability in home settings (Engelke & Higginbotham, 2013; Hörmeyer & Renner, 2013). These studies revealed that the communication partners (i.e., staff or relatives) positioned themselves beside the participant who relied on the SGD to oversee the screen, the face, and the gaze practices. Research has not yet included participants with intellectual disability, and it has not been examined how eye-gaze accessed SGD-mediated interaction work in multiparty interaction, for example in classroom settings where it is essential to provide the student with a possibility to interact and to demonstrate knowledge.

A large body of research has demonstrated that classroom interaction involving students without disabilities is often organized in dyads between the teacher and the whole student group, or between the teacher and one student at a time (Heritage & Clayman, 2010). Heritage and Clayman (2010) found that this pre-allocated turn-taking organization emerged within context-specific and goal-oriented activities in which participants aligned themselves with their institutional identity, such as being a teacher or a student. Typically, classroom interaction has been found to be organized in terms of IRE sequences in which the teacher Initiates the interaction and the student provides a Response that the teacher then Evaluates (Mehan, 1979). Classroom interaction is often characterized by a power hierarchy in which the teacher asks display questions (i.e., questions with known and predictable response alternatives) (Lee, 2006; Kathard et al., 2015) or polar questions (i.e., yes/no questions) (Lee, 2008) that the student responds to. Macbeth (2000) found that the practice of asking display questions guided students to demonstrate their knowledge.

Students without disability display recipiency (i.e., readiness to take the turn) in the classroom through a combination of linguistic resources (i.e., spoken turns) and embodied resources such as gaze orientation, body movements (e.g., hand raising), and silence (Sahlström, 1999). Usually, they wait for next-selection from the teacher, and they seldom self-select during the teacher’s turn. Radford, Bosanquet, Webster, and Blatchford (2015) found that teachers and assistants working with students with special education needs used responsive actions that stimulated and facilitated the learning of students with special education needs. They adopted different scaffolding roles (repair, supportive, and heuristic) that engaged students in self-repair, supported students’ engagement and motivation, and encouraged students to use their own learning strategies and solve problems independently. In addition, they used scaffolding techniques such as well-timed responsiveness, deictic gestures, verbal direction, and correction. Koole and Elbers (2014) argued that teachers’ contingent responses to students’ turns within dyadic structures in classroom talk were essential for achieving interactions that scaffolded students’ learning processes. Research found that social and pedagogical inclusion of students with, for example, complex communication needs was demanding (Andzik et al., 2016; Radford et al., 2015; Tönsing & Dada, 2016). Moreover, implementing eye-gaze accessed SGDs in classroom settings was difficult (Rytterström, Borgestig, & Hemmingsson, 2016; Tegler, Pless, Blom Johansson, & Sonnander, 2019; van Niekerk & Tönsing, 2015). In a study by Rytterström et al. (2016), teachers expressed a conflict between the need to follow the process on the SGD screen, the student’s eye movements and facial expressions, and the need to face the rest of the class.

Given the importance of including students with complex communication needs in the social and pedagogical activities of classrooms, and the lack of research on how to achieve inclusive education for students using eye-gaze accessed SGDs, the aim of this study was to explore how inclusive multiparty classroom interaction was organized when one of the students used an eye-gaze accessed SGD. The research questions were: which scaffolding and collaborative practices used by teachers, assistants, and classmates created an interactional space for the production of an eye-gaze accessed SGD-mediated turn in multiparty classroom interaction? Did the identified practices differ between the student cases?

Method

Participants

The participants using eye-gaze technology were two students with severe cerebral palsy and intellectual disability (Anna and Steve, pseudonyms), their teachers, assistants, and classmates. The inclusion criteria for the participants were (a) reliance on eye-gaze technology to access the SGD; (b) use of the SGD in multiparty classroom interaction several times per week; (c) aged 7 to 21 years, which corresponded to the age of school attendance for children and adolescents with intellectual disability in Sweden; (d) severe cerebral palsy; (e) intellectual disability, and (f) unintelligible speech. The criterion of severe cerebral palsy corresponded to Levels 4–5 on the Gross Motor Functional Classification System (GMFCS) and to Levels 3–5 on the Manual Ability Classification System (MACS) (Eliasson et al., 2006; Palisano et al., 1997). Hence, they were transported by wheelchairs, had difficulties with head and trunk control, and were unable to use their hands. The criterion of unintelligible speech corresponded to Levels 3–4 on the Viking scale (Pennington et al., 2013). The criterion of intellectual disability was achieved by only including participants participating in the special school system, which in Sweden demands that intellectual disability has been diagnosed by a psychologist. Anna and Steve attended two different schools within the special school system for students with learning disabilities, which is part of the elementary school system. The degree of intellectual disability among the students within the special school system for students with learning disabilities range from mild to severe; the teachers and assistants are familiar with AAC-mediated interaction; and the staff/student ratio is higher compared to regular schools.

Professionals at pediatric habilitation centers in the central and southern parts of Sweden were asked to identify potential participants who met all study inclusion criteria. Eight eligible participants were identified; their caregivers were informed about the study and asked to inform their child or adolescent about the study. Caregivers of three participants provided written consent; two participants—Anna and Steve—met all six inclusion criteria. At the time of data collection, the third participant did not use his SGD in the classroom. The teachers, assistants, and classmates of Anna and Steve were informed about the study and asked to participate. The caregivers of the classmates were informed about the study and asked if their adolescents could participate. Written consent was obtained from the teachers and assistants and from the caregivers of the classmates. Some of the caregivers of the classmates only consented to audio recording of their children. Hence, these students were in the classroom during the data collection but were not video recorded.

Anna

This participant was a 14-year old adolescent with physical and intellectual disabilities due to dyskinetic cerebral palsy. Her teacher and assistant reported good language comprehension, typical hearing and vision. At the time of the study, she had used her SGD (Tobii C12 with Communicator^{TM 1}) with synthetic voice output for 3 years and 8 months. The vocabulary of the SGD consisted of 20–24 symbols per page, in total more than 400 symbols. The symbols were organized into general and specific situational categories. There were core and fringe vocabularies organized into single words and prerecorded phrases. The software interface did not include a sentence bar. Anna was not literate. She used her SGD independently and had access to it all day as it was attached to her wheelchair. In addition to the SGD, she looked up for “yes” and down for “no.” Anna’s facial expressions and gaze behavior were well-timed and expressive. A paper-based communication board had been used prior to the prescription for the SGD, but this board was no longer used because Anna was faster and more independent using the SGD instead. Anna had known her assistant for 6 months and her teacher in science and mathematics for 2 years. The assistant had a high school diploma, and the teacher had an elementary teaching degree. Neither the teacher nor the assistant had experience with eye-gaze accessed SGD-mediated interaction prior to working with Anna. Anna’s six classmates all had an intellectual disability: five used natural speech, and one used an Apple iPad^{TM 2} accessed by finger pointing. Three were able to walk and three used wheelchairs, which they operated independently. In addition, there were three assistants to the classmates in the classroom.

Steve

This participant was an 18-year-old adolescent with physical and intellectual disabilities due to diplegic spastic cerebral palsy. He had used his SGD (tablet computer with PCEye with Communicator^{TM 3}) with eye-gaze technology and synthetic voice output for 6 months. The vocabulary consisted of 6–14 symbols (single words and prerecorded phrases) per page, adding up to about 200 symbols. The software interface did not include a sentence bar. Steve was not literate and he had some difficulty operating the eye-gaze technology. Prior to his prescription for eye-gaze technology, his communication partners had practiced partner scanning (i.e., they operated the SGD by asking Steve questions, which Steve confirmed or denied by gaze practices). Steve’s facial expressions and gaze behavior were minimalistic and often delayed. According to both the teacher and the assistant, Steve had typical hearing and vision, but they were unsure about his language comprehension. Steve had known his assistant for 1 year and 4 months and his teacher for 1 year. The assistant had a high school diploma, and the teacher had a preschool teaching and elementary teaching degree. Neither the teacher nor the assistant had experience of eye-gaze accessed SGD-mediated interaction prior to working with Steve. Steve had four classmates, all with intellectual disabilities: two used natural speech, one used natural speech and sign language, and one used natural speech and a communication board with graphic symbols. In addition, there were three assistants to the classmates in the classroom.

Setting

All video recordings took place in classrooms at Anna’s and Steve’s respective schools. The first author and the teachers agreed upon suitable time points for video recording. In the classrooms, just prior to the start of recording, the first author reminded Anna and Steve, their classmates, and the staff that participation was voluntary and that they could withdraw from the study at any time.

Research design

The study used a descriptive observational design. Conversation Analysis (CA) was used to analyze the video sequences because it allowed for detailed analysis, which was required to identify interactional phenomena and patterns that otherwise might have gone unnoticed. The methodological principles and procedures of CA were deployed to investigate participants’ methods of organizing turn-taking during eye-gaze accessed SGD-mediated multiparty interaction. The analysis of turn-taking proceeded in the following steps: (a) identifying the participant’s methods of making recipiency understandable; (b) identifying the linguistic and prosodic design of turns at talk (Åhlund & Aronsson, 2015; Kendon, 1967; Lee, 2006, 2008); and (c) analyzing embodied actions, such as gaze behavior, hand gestures, body posture, and movement within the turn (Goodwin, 2000; Lerner, 2003; Sikveland & Ogden, 2012). The notion of adjacency pair (Clarke & Wilkinson, 2007; Sacks et al., 1974; Schegloff, Jefferson, & Sacks, 1977) was used to account for the temporal and sequential aspects of SGD turn organization: initiatives (or first pair parts) made certain sets of responses (or second pair parts) relevant and recognizable (e.g., request-granting, question-answer). Previous research on ordinary talk-in-interaction found that linguistic and prosodic features as well as embodied resources are important resources to project an upcoming possible point of turn completion and turn allocation, namely transition relevance places (TRP). Yet, there is a lack of knowledge on how these processes are accomplished in eye-gaze accessed SGD mediated interaction in general, and in classroom contexts in particular. TRPs are used in the analyses to identify how and when it is relevant for the participating children to respond to the teacher’s initiatives, and to analyze how the participants deal with actions occurring in that interactional place. TRPs are legitimate junctions in the ongoing conversation occurring after a possible completion of turn construction units (TCU, i.e., the smallest interactional unit in a given context). The notion of a three-part action trajectory (Schegloff et al., 1977) was used to analyze repair or correction of troubles that appeared during SGD turn construction: the first action identified the source of trouble; the second action initiated repair; and the third action effectuated repair of the trouble.

The study was approved by the appropriate school authorities. The data was part of a larger corpus of eye-gaze accessed SGD-mediated interactions recorded in six Swedish preschools and special schools from May to the Swedish Ethical Review Authority and the November 2018.

Materials

Two cameras were used to record interactions: one recorded the SGD screen, and the other recorded the participants who used the SGDs and their communication partners. In all, about 5 hr and 30 min were video recorded (Table 1). Approximately 2 hr and 40 min consisted of multiparty classroom interaction including access to the SGD and this data was used in the analysis (Anna 01:30 and Steve 01:09) (Table 1).

Table 1. Video recordings of classroom interactions, by hours and minutes.

Numbers of participants	Class/activity	Anna	Steve
Multiparty interaction	Morning assembly	–	00:26
	English lesson	00:42	–
	Science lesson	00:16	00:43
	Mathematics lesson	00:19	–
	First language	00:13	–
	Summary	01:30	01:09
Four-party interaction	Lesson with one classmate	–	00:26
Dyadic interaction with assistant	Desk job	00:42	00:26
	Talk during break	00:41
	Totals	02:53	02:36

Procedures

A systematic five-step work process (Higginbotham & Engelke, 2013) was used to choose illustrative and representative examples of scaffolding and collaborative practices used by teachers, assistants, and classmates that constructed an interactional space for the production of an eye-gaze accessed SGD-mediated turn in multiparty classroom interaction. The process was iterative: the video recordings and the transcripts were reviewed repeatedly in every step. First, the first author scanned all data to identify SGD-mediated turns initiated by Anna or Steve (self-initiated) and SGD-mediated turns initiated by the teacher (other-initiated) (Table 2). Second, the first author screened all SGD-mediated turns for conversational topics and transcribed them verbatim, including silences and embodied actions (Table 2). The methodological principles and procedures of CA were used to transcribe data. The transcription conventions suggested by Jefferson (2004) were combined with those of Von Tetzchner and Basil (2011) to represent the participants’ detailed and multimodal practices including spoken, embodied, and SGD-mediated actions (see Appendix). Third, the first and the last authors analyzed and discussed the transcripts and the video recordings together and in data sessions involving a research group knowledgeable in CA to assure reliability and validity. The research group viewed sequences of transcribed video recordings. The transcripts were checked for reliability and the group proposed and discussed different ways of analyzing the sequences from a CA point of view. Fourth, SGD-mediated turns were selected and analyzed in their local interactional context. Fifth, the selection of illustrative sequences was achieved purposefully to illustrate the scaffolding and collaborative practices used by teachers, assistants, and classmates that constructed a response space.

Table 2. Number of eye-gaze accessed SGD-mediated turns in the multiparty classroom interaction.

Design/practice	Teacher-initiated turns with participant responses		Participant-initiated turns with teacher responses
	Anna	Steve	Anna	Steve
Initiation was a display question	23	9	0	0
Initiation was an open-ended question	0	0	0	0
Initiation was a single word	0	0	0	5
Summoning the recipient by name	22	9	0	0
Responded without scaffolding	12	1	0	5
Responded with teacher scaffolding	9	0	0	0
Responded with assistant scaffolding	2	8	0	0

Results

Three scaffolding and collaborative practices were found to create a response space for eye-gaze accessed SGD-mediated turns in multiparty classroom interactions: (a) the organization and design of turn taking in recognizable IRE sequences, (b), the locally contingent on-screen and multimodal scaffolding activities, and (c) teachers’ practices defending the student’s response space from classmates’ competing responses. The scaffolding practices differed in one aspect: availability of required vocabulary on the students’ SGDs in order to answer the teacher’s question.

Organization and design of an interactional response space

The participants organized the classroom interaction in terms of initiation, response, and evaluation (IRE) sequences, where the response action was extensively collaboratively constructed due to the use of eye-gaze accessed SGDs. Teachers systematically designed their initiatives as display questions (Lee, 2006) and summoned the student by name (Mehan, 1979). This practice publicly identified the student as the recipient but also positioned the SGD-mediated response in a sequentially recognizable position (Clarke & Wilkinson, 2007) in the ongoing classroom interactions. Addressing the next speaker with both gaze direction and naming the first name is, in the terms of Lerner (2003), an explicit and context-free address practice. Lerner found that explicit naming was the strongest form of addressing a recipient in talk-in interaction, but not the most widely used practice in ordinary conversation. The practice of naming is common in classroom settings. University students in a study by Cooper, Haney, Krieg, and Brownell (2017) reported that their attitudes and behaviors were affected in a positive way when the teacher practiced naming. In the data of the present study, this strong address practice was the most common when addressing SGD users, which indicate that the teacher both selected a specific next speaker within the multiparty setting and oriented to the challenges of participants depending on SGD to self-select as recipients.

Extract 1 (Supplementary Material) was from a lesson in science about the planets in the solar system at Anna’s school. The class had jointly recalled that Mercury was closest to the sun when the teacher addressed and selected Anna as recipient by gazing at her, summoning her by her name and asking a display question Anna can you say something (Line 1) about (Line 3). The teacher knew the content of the prerecorded planet on the SGD. Anna started to respond in overlap with the teacher’s initiative. She looked at the symbol TIME (Line 2) on her search for SCHOOL PAGES (Line 5); however, the teacher’s question was cut off by a self-selecting classmate “I’m finished” (Line 4). After dealing with the interruption (analyzed in the third section of the results), the teacher renewed the initiative with a specification, Anna can you say something (Line 12) about Mercury (Line 13). After additional page changes in the SGD system, SCHOOL PAGES (Line 5), NATURAL SCIENCE (Line 11) and PLANETS (Line 14), Anna finally activated the symbol MARS and responded “Mars is a terrestrial planet” (Line 15). At the end of the SGD turn, she shifted gaze orientation from the SGD to the teacher, constructing a possible TRP. Her response, however, was rejected as the teacher evaluated her answer and said Now you were on Mars (Line 17), and then renewed the initiative a third time by adding (while lowering her voice) ^OMercury^O (Line 17). After additional scaffolding his (analyzed in the second section of the results), Anna produced a correct answer, Mercury is closest to the sun (Line 47), which the teacher evaluated exactly (Line 48).

Extract 2 (Supplementary Material) was from a science lesson on human anatomy at Steve’s school. In the beginning of the extract, the assistant had just selected and programed a picture page in the SGD’s software with eleven body parts without consulting the teacher. The teacher was standing in front of the class without visual access to the SGD screen when he selected Steve as recipient by gazing at him, summoning him by his name and asking Steve what’s in the last picture (Line 1) while pointing at the symbol HAND on the worksheet, which was not among the eleven options on the SGD. The initiative was successful in terms of recipiency; the classmates waited in silence for 5 s before one of them self-selected and indicated that she wanted to respond to the question (Line 8). Steve’s later but incorrect response “sternum” (Line 20) was preceded by the teacher’s and assistant’s scaffolding actions (analyzed in the second and third section of the results). In all, Steve produced three incorrect responses “sternum” (Line 20), “sternum” (Line 23), and “elbow” (Line 28) without gazing at the teacher (and hence avoiding the construction of a TRP with his gaze), indicating that the turn had not been brought to a completion. During these efforts to answer, the assistant (not the teacher) evaluated Steve’s answers (Line 21, 24, 25, 27, and 29) (analyzed in the second section of the results).

To summarize, the two extracts showed that the organization of IRE sequences by way of the teachers’ summoning of the student by name and asking display questions clarified recipiency and prolonged the response space in which the student could respond on the SGD irrespective of whether the answer was correct or not. The two extracts also illustrated a difference regarding the way the participants displayed recipiency to the initial question. Anna, but not Steve, had the required vocabulary on the SGD, which probably helped her to: (1) accept recipiency immediately and start to respond without scaffolding actions, and (2) construct a TRP in the end of the SGD-mediated turn by shifting gaze from the SGD to the teacher, indicating that her answer turn was complete.

Locally contingent support: on-screen scaffolding activities

The process of producing an SGD-mediated contribution was often scaffolded by the teacher or the assistant in close and detailed orientation to the students’ gaze actions on-screen. In Extract 1 (Supplementary Material), the teacher walked over to Anna when she produced the incorrect answer “Mars is a terrestrial planet” (Line 15). He positioned himself behind her while pointing at the symbol MERCURY and whispering Up there, () there (Line 30, Figure 1). The indexical spoken turn was closely coordinated with Anna’s moving of the eye-gaze cursor on the screen, and the deictic gesture specified the verbal instruction and stressed its relevance. Deictic gestures are typically coordinated with indexical spoken turns and sustained until a display of shared understanding has been achieved (Sikveland & Ogden, 2012). The multimodal indexical directives also provided Anna with support without explicitly mentioning what the target symbol represented. This demonstrated a delicate balance between providing Anna with sufficient support but not doing the answering for her. During the complete scaffolded process (Lines 17–45) of response construction, the classmates collaborated to sustain Anna’s right to the interaction floor by keeping quiet. The teacher acknowledged Anna’s attempts to activate MERCURY by whispering there yes (Line 37), works? (Line 39), and by saying I see- Anna that you are speaking but (Lines 44–45). He also made a gesture of turning his palms up (Line 44), which contributed to make the SGD accountable for the troubles rather than Anna. The teacher’s comments and gestures all worked to make Anna’s turn construction process public for the other students, helping them to refrain from interfering activities.

Figure 1. The teacher points at MERCURY in response to Anna’s Gaze at VENUS.

Extract 2 (Supplementary Material) illustrated how the assistant’s scaffolding actions renewed and secured the response space for Steve. There was 2 s of silence after the teacher’s initiative Steve what’s in the last picture (Line 1), during which Steve did not display recipiency. Hence, the assistant reformulated and renewed the teacher’s initiative by a combination of verbal and embodied clues. She said that picture (.) what is that (.) what am I pointing with (Line 4) while pointing at the symbol HAND on the worksheet (Line 5). Steve looked at the worksheet and after an additional 3 s of silence, without displaying recipiency, one of the classmates self-selected by saying “>så<” (okay/right) (Line 8) (analyzed in the third section of the results). The assistant once again reformulated and renewed the teacher’s initiative by saying yes what is that (Line 9) and what is that (Line 14). Steve finally displayed recipiency and started to respond FOOT KNEE CAP (Line 15). The assistant followed the movements of the eye-gaze cursor on the screen and commented no have to search a bit more I think (Line 16). After some additional attempts at answering, the assistant also commented on Steve’s incorrect answer “sternum” (Line 20) with a response particle nä::e:j? (no…?) (Line 21). The curled prosodic contour of nä::e:j? worked to soften the action of rejection when initiating repair of Steve’s responses with the SGD, working in reverse of the particle’s semantic meaning. Steve treated the curled response particle nä::e:j? as an other-initiations of repair, but also as go-aheads, that he was on the right track and could continue to try eye-gaze click, as he systematically responded with new attempts. Steve repeated” sternum” (Line 23), and this time the assistant provided him with a new combination of verbal and embodied clues: what am I using here (Line 24) while waving with her fingers in front of Steve in response to his gaze at the symbol of FOOT. In Line 27, still waving, she produced an unfinished turn What is it I’m- in response to Steve’s gaze at the symbol on elbow (Figure 2). In Line 28, Steve produced a third incorrect answer, “elbow”, which the assistant again evaluated with a curled nä:::?.The combination of verbal and embodied clues provided locally contingent support for Steve’s on-screen eye-gaze activities that directed him toward eye-gaze clicking on the symbol FINGER on the screen, but without doing the answering for him, hence providing Steve with a voice. The assistant’s feedback compensated for the lack of the symbol HAND in the on-screen vocabulary.

Figure 2. Steve Gazes at ELBOW and the assistant illustrates fingers with her hand.

To summarize, the two extracts showed that the production process of the SGD-mediated response was scaffolded by a facilitator who had visual access to the screen and used a combination of verbal and embodied clues in response to the participant’s movements of the eye-gaze cursor on the screen and the sound of the synthetic voice. The scaffolding practice prolonged the response space, supported Anna or Steve in their search for a corresponding symbol, and guided (as well as controlled) the classmates during the construction process. The preconditions across data differed. Anna had the required symbol on the SGD. Hence, her teacher used deictic gestures at the correct symbol and prosodically styled and mitigating spoken response turns. Steve did not have the required symbol on his SGD, thus, Steve’s assistant used modeling deictic gestures to guide Steve to respond with an existing symbol with a similar meaning (i.e., FINGER instead of HAND) and prosodically styled and encouraging response particles.

Defending the response space: Teachers dealing with classmates’ competing responses

There were occasions where classmates self-selected during the teacher’s initiative or during Anna’s or Steve’s SGD-mediated turn. In Extract 1 (Supplementary Material), for example, one of Anna’s classmates commented in overlap with the teacher’s initiative that “I’m finished” (Line 4), which is an uncommon action in classroom settings (Sahlström, 1999). A side sequence developed (Lines 5–7), where the teacher first told Anna to wait (Line 6) and then informed the interrupting classmate with a whispering and smiling voice, followed by laughter, that he was going to remove the classmate’s SGD let’s do like this Erik (Line 6) and I’m moving this a bit (Line 9).

In Extract 2 (Supplementary Material), one of Steve’s classmates self-selected during Steve’s turn by saying “>så<” (okay/right) (Line 8). The så was produced with stress and increased speed as a unit in its own right, marking the termination of a parallel activity that the student was involved in (Ottesjö & LindströM, 2005). The teacher rejected the classmate’s attempt to gain access to Steve’s response space in two steps. First, he turned to the student and said we will wait we will let (Line 10). The classmate immediately acknowledged by doing a thumb-up (Line 11). Second, the teacher added in a whispering voice maybe Steve can answer (Line 12), thereby resuming the main activity and renewing the allocation of Steve as next speaker.

The two extracts demonstrated a locally contingent affinity between the teacher and the competing classmate. Side-sequences developed in which the teacher and the competing classmate agreed on waiting for Steve or Anna to respond on the SGD. By whispering, practicing smiley voice, and laughter, the participants built local participation frameworks (Åhlund & Aronsson, 2015; Norén & Sigurd Pilesjö, 2016) where they could negotiate meanings (i.e., they agreed on not interrupting during the process of constructing an SGD contribution) before returning to the public framework that also involved Steve and Anna. The prosodic package of the correction (increased speed and decreased volume) and the laughter mitigated the corrective actions.

Scaffolding practices: Availability of required vocabulary

The analysis revealed that Anna, but not Steve, had access to the relevant vocabulary on the SGD. This difference affected the interaction in two ways. First, the fact that Anna had access to the required vocabulary on-screen may have helped her to display recipiency and start to respond without scaffolding actions. Steve did not have access to the required vocabulary on-screen and may have oriented to this when not displaying recipiency. Second, Anna’s teacher guided Anna toward the correct answer by using a combination of deictic gestures at the correct symbol on-screen and prosodically mitigated spoken turns. Steve’s assistant, however, compensated for the absent on-screen symbol by guiding Steve toward an alternative present symbol with a similar meaning (i.e., FINGER instead of HAND) by using a combination of modeling deictic gestures, prosodically styled spoken turns, and encouraging response particles.

Discussion

The present study illustrated that a number of scaffolding and collaborative practices used by teachers, assistants, and classmates designed a response space in which the student could produce an eye-gaze accessed SGD-mediated contribution. Three key findings across data were revealed: (a) the organization and design of turn taking in recognizable IRE sequences, (b) the locally contingent on-screen and multimodal scaffolding activities, and (c) the practices dealing with classmates competing for response space. Furthermore, accessibility to the relevant vocabulary affected the teachers’ and assistants’ scaffolding and collaborative practices.

Organization and design of an interactional response space within IRE sequences

The organization of IRE sequences initiated with display questions that summoned the student by name was found to clarify recipiency and secure the student’s response space for answering on the SGD. These findings are similar to previous work on classroom interactions, where summoning by name and display questions have been described to clarify recipiency in multiparty talk (Heritage & Clayman, 2010; Lee, 2006; Lerner, 2003; Mehan, 1979). The practice of organizing the interaction in IRE sequences also resembled the sequential organization of adjacency pairs in SGD-mediated interaction as described by Clarke and Wilkinson, (2007, 2008), since it facilitated intersubjective understanding of recipiency by restricting the response alternatives in the second part pair of a question-answer sequence. The finding that the teachers in the present study primarily asked Anna or Steve display questions in the multiparty classroom interaction resembles previous findings (Kathard et al., 2015). In their study, Kathard et al. (2015) cautioned that the format of display questions could create recitation scripts in which the student played a passive role. However, in the current study, the organization of IRE sequences and display questions were found to facilitate the interactional and pedagogical inclusion in multiparty classroom interaction when one of the students used an eye-gaze accessed SGD.

Locally contingent support: on-screen scaffolding activities

The present study also demonstrated that a facilitator (i.e., either the teacher or the assistant) could assist the student during the construction process of the SGD-mediated turn by a combination of locally contingent verbal and embodied on-screen activities. These activities presumed that the facilitator had visual access to the SGD screen and timed their response (i.e., initiated repair and guided toward correct answers) in relation to the movements of the eye-gaze cursor on the SGD screen and the synthetic voice. In their study, Koole and Elbers (2014) also demonstrated that local contingency is a key aspect of scaffolding students’ learning in typical classroom interactions.

The locally contingent support had two functions. First, it encouraged the student to continue searching for an answer on the SGD even though the support also displayed the facilitator’s pedagogical stance on incipient symbol choices (as correct or incorrect answers) before they had been effectuated as an SGD-mediated turn with the synthetic voice. The practice guided and maintained the student’s activity when eye-gaze click attempts failed or when students’ actions predicted that they were likely to select incorrect symbols. Second, the facilitators’ contingent scaffolding practices made the SGD production process public to the other students, enabling them to monitor for initiation and completion of SGD-mediated responses. This practice secured the response space for the student using the SGD, as it guided the classmates in their collaborative achievement of a response space in the classroom talk, and hence established a context less vulnerable to classmates’ attempts to enter that space. Based on these findings, facilitators were found to practice a combination of repair, supportive, and heuristic scaffolding roles as described by Radford et al. (2015). Hence, they encouraged the students who relied on their SGDs to carry on their process of constructing a response and they motivated the classmates to collaborate to sustain Anna’s or Steve’s interaction floor by not speaking. In conclusion, the findings confirmed the necessity of visual access to the SGD screen during the construction process as described by Rytterström et al. (2016).

Defending the response space

The present study revealed that, on a few occasions, one of the classmates self-selected during the IRE sequence that addressed Anna or Steve as the recipient. The teacher in both Anna’s and Steve’s class treated these self-selections as sources of trouble and initiated corrective actions that restored the threatened response space. In line with Åhlund and Aronsson (2015), the analysis showed that the teachers attuned their voices (e.g., prosodic variations in duration, amplitude, and pitch), whispered, and laughed, which mitigated the corrective action. The prosodic and voice stylizations built local alliances and participation frameworks during side sequences in the dyadic interactions with the interfering student and worked to clarify and secure the response space for students with SGDs.

Interactional gaze practices using Eye-Gaze technology

The present study found that Anna and Steve by necessity were visually tied to the SGD screen during the prolonged turn construction process, which mirrored previous research on this (Engelke &Higginbotham, 2013) and partly impeded the typical gaze behavior (Goodwin, 1980). The teachers, assistants, and classmates in this study accepted the student’s lack of typical gaze practices during the turn-construction process. They treated Anna’s or Steve’s gaze orientation toward the SGD screen as a meaningful action that claimed and held the turn. A possible explanation for this might be explained by the work of Rossano, Brown, and Levinson (2009) who found three exceptions for not practicing typical gaze practices in face-to-face speech interaction. The participant (a) looked at nothing in particular, (b) followed a deictic gesture, or (c) was involved with a competing activity. In the present study, gaze orientation toward the SGD during the construction process may constitute a fourth exception from typical gaze practices: the use of a communication resource that occupies one of the participant’s gaze. Anna and Steve were occupied with accessing their SGD with eye-gaze technology, which did not permit typical gaze practices at the recipient until arriving at a possible TRP. At other occasions however, at the end of SGD-mediated turns, Anna’s and Steve’s gaze practices resembled typical gaze behavior. For example, Anna shifted gaze from the SGD to the teacher at the end of the correct SGD-mediated turn, displaying a possible TRP and Steve continued to gaze at the SGD after providing wrong answers, indicating that the turn was still under construction.

Implications for practice

Students who rely on eye-gaze accessed SGDs may be included in multiparty classroom interaction provided that the recognizable organization of IRE sequences is applied and that either the teacher or the assistant have visual access to the SGD screen and can provide locally contingent on-screen and multimodal scaffolding activities that guide the student as well as the classmates.

Limitations and future directions

The data collection covered 2 days of video recordings from two Swedish schools in two different cities. The students who relied on the SGDs were approximately the same age and attended the same curriculum for students with learning disabilities. Hence, the results might not be generalizable to other contexts, to other ages or to participants without intellectual disability. In addition, the data were restricted to teacher-led classroom interaction in science and mathematics classes, leaving out for example interaction in smaller groups and other school subjects. Accordingly, neither the video recordings nor the selection of extracts are representative for all multiparty interaction in classroom settings. However, illustrative examples of practices provide useful knowledge on scaffolding techniques that may potentially facilitate inclusion of students who rely on eye-gaze accessed SGDs. Additional research is needed to investigate SGD-mediated turn-taking practices when turns are initiated by students with or without intellectual disabilities in other classroom activities and school subjects.

Conclusion

Traditional classroom interaction with practice of IRE sequences, summoning the student by name, and asking display questions in combination with a facilitator’s on-screen scaffolding activities and defense of the response space provided and prolonged the response space for SGD-mediated turn construction in multiparty classroom interaction across data. The practices promoted a degree of social and pedagogical inclusion, offering students using SGDs an opportunity to display knowledge or demonstrate understanding in the ongoing interaction with teachers and classmates, either contingent upon support or independently.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

1 Tobii C12 with Communicator is a product of Tobii Dynavox, Pittsburgh, Pennsylvania, www.tobiidynavox.com

2 Apple iPad© is a product of the Apple Computers Inc., Cupertino, California, www.apple.com

3 PCEye with Communicator is a product of Tobii Dynavox, Pittsburgh, Pennsylvania, www.tobiidynavox.com

Appendix 1

Appendix

Transcription conventions.

Statement	Convention
how are you	Naturally spoken turns are italicized
“four”	Synthetically spoken turns are italicized and placed in quotation marks
FOUR	Italic capital letters indicate a symbol on the SGD
#FOUR	One hash symbol before a word in italic capital letters indicates an un successful attempt at eye-gaze clicking on a symbol
##FOUR	Two hash symbols before a word in italic capital letters indicates a successful eye-gaze clicking that activates the synthetic voice or changes the page
( )	Empty parentheses indicate speech that is inaudible for the transcriber
xxx⌈xxx ⌊xxx xxx	Left-aligned brackets indicate the onset of simultaneous talk, or embodied action at the point where it begins
xxx⌉ xxx⌋xxx xxx	Right-hand brackets indicate where simultaneous actions ends.
{(nods) yes}	Curly braces indicate a combination of two semiotic modalities
(0.6)	Parenthesis surrounding a number indicate (in tenth of a second) the duration of a longer pause during or between turns.
.hh	A period preceding the letter h indicates an inhale.
two:::	A colon indicates an extension of the syllable or sound it follows. The number of colons marks the length of the duration.
two,	A comma indicates a slight rise at the end of a prosodic unit.
two?	A questioning mark indicates a high rise at the end of a prosodic unit.
xxx↑ xxx↓ xxx	Up or down pointing arrows indicate rises and falls in a turn’s pitch contour, immediately prior the pitch change.
^otwo^o	Degree signs indicate talk at a lower volume than surrounding talk.
^ootwo^oo	Double degree signs indicate whispering.
>two< >two<	Signs for more than (>) and less than (<) indicate speech that is faster or slower than surrounding speech.
do you mean that	Underlining of words or parts of words indicate emphasis.
(nods)	Parentheses contain a description of embodied actions (gaze, gesture, body posture, etc.).
((creaky voice))	Double parentheses contain a description of some aspect of spoken or embodied actions.