Implementing Stay-Play-Talk With Children Who Use AAC

Abstract

Withdrawal and multitreatment single subject research designs were used to evaluate the effectiveness of stay-play-talk (SPT) interventions on social behaviors of preschool-aged peers to children with disabilities. Each group included at least one socially competent peer and one child with Down syndrome who used an augmentative and alternative communication (AAC) device as a primary mode of communication. Peers were trained to use SPT strategies during free play sessions, and a modified reinforcement system and modified peer arrangement were introduced for one group. For one group, results indicate a functional relation between the original SPT intervention and increased stay and play behaviors. For the other group, results indicate a functional relation between SPT with modified arrangement and increased stay and play behaviors. Future research is needed to determine effectiveness of SPT interventions for children with disabilities who have more sophisticated functional play skills, as well as utility of creating peer dyads compared with peer triads.

Keywords

single-subject designs experimental studies research methodologies social skills intervention strategies augmentative and alternative communication

Peer social interactions are particularly important for young children’s social-emotional competence, language development, and social inclusion (Guralnick, Hammond, Connor, & Neville, 2006; Joseph, Strain, Olszewski, & Goldstein, 2016). Adequate social communication assists children in forming and sustaining these social relationships. Although there is a clear link between social interactions and positive outcomes, interaction rates of children with disabilities are generally lower than those of typically developing children (Honig & McCarron, 1988). Moreover, children who use augmentative and alternative communication (AAC) devices are at a particularly greater risk for social isolation (Clarke & Kirton, 2003).

High quality inclusive early childhood environments provide an avenue for these important social interactions between typically developing children and children with disabilities (Strain & Bovey, 2011; Wolery & Hemmeter, 2011); however, placing children in inclusive environments is not sufficient for increasing positive interactions (Odom, Buysse, & Soukakou, 2011). In inclusive environments, typically developing children tend to interact with their typically developing classmates without adult intervention or interference (Guralnick, Connor, Hammond, Gottman, & Kinnish, 1996). Thus, to improve rates of interactions between children with and without disabilities, adults may need to provide prompting and reinforcement.

Stay-play-talk (SPT) is a peer-mediated intervention designed to increase the social interactions of children with and without disabilities in inclusive preschool classrooms (Goldstein, Kaczmarek, Pennington, & Shafer, 1992). Typically developing peers are taught specific strategies to stay, play, and talk with their peers with disabilities, and teachers prompt and reinforce the peers for engaging in SPT behaviors in free play or other settings or activities throughout the day (English, Goldstein, Kaczmarek, & Shafer, 1997; Goldstein & English, 1997; Robertson, Green, Alper, Schloss, & Kohler, 2003). Research has shown that SPT interventions are successful in increasing the rate (Goldstein et al., 1992) and the number, length, and reciprocity of social interactions between typically developing children and children with disabilities (Hughett, Kohler, & Raschke, 2013; Kohler, Greteman, Raschke, & Highnam, 2007). A recent study (Thiemann-Bourque, Brady, McGuff, Stump, & Naylor, 2016) incorporated Picture Exchange Communication System (PECS) training into an SPT intervention; however, these authors used a series of A-B designs rather than an experimental single case design, which precludes identification of functional relations (Horner et al., 2005).

To date, SPT interventions have not explicitly included a high-tech AAC training component for preschool-aged participants who use high-tech AAC devices. This may be problematic because many children with disabilities have communication impairments, and AAC devices may increase the quality of their social and functional communication (Cosbey & Johnston, 2006). Clarke and Kirton (2003) found that children who use AAC took fewer communicative turns, made fewer initiations, and generally used more nonverbal gestures than AAC device activations during communicative interactions with their peers. However, a recent review of interventions designed to increase peer interactions for children who use AAC found only one study with positive effects with preschool children (Therrien, Light, & Pope, 2016). This study, and many others that explicitly include a high-tech AAC teaching and modeling component, focused on the children’s functional, rather than social, communication. In this study, Cosbey and Johnston (2006) found that teaching the use of a single-switch voice output communication aid and a peer mediated social skills intervention increased the frequency of requests of children with severe disabilities. Similarly, Bock, Stoner, Beck, Hanley, and Prochnow (2005) investigated whether PECS or voice output systems led to a greater number of requests in children with communication deficits. The outcome measures in these studies were specifically the number of request statements, rather than social communicative statements. Although these interventions are important for promoting functional communication, they lack the social communicative aspect that is important for supporting social interactions and friendships in an inclusive setting.

Additional rigorous research on the training of the social use of other AAC systems with peers is needed given their potential to increase opportunities for social interactions. Moreover, when peers model appropriate AAC usage during play, the number of social communicative behaviors of children with autism who use AAC devices increases (Trembath, Balandin, Togher, & Stancliffe, 2009). The current study extends research on existing SPT interventions by investigating whether SPT was effective for teaching typically developing children to stay, play, and talk with children who use a high-tech AAC device. The research questions were as follows:

Research Question 1: When compared with a baseline condition, will SPT with an AAC training component increase the percentage of intervals in which a child with a disability and his or her peers stay near each other, play with similar toys, and talk to each other?

Research Question 2: When compared with a baseline condition, will SPT with an AAC training component increase the number of social interactions between peers and the children with disabilities?

Method

Participants

This study included two target children and four typically developing peers between the ages of 36 and 63 months. Prior to beginning the study, written parental consent was obtained for all children in accordance with approved Institutional Review Board procedures. During the study, participants’ verbal assent was assessed daily. One of the peers was withdrawn in the first week of the study due to a combination of school absences and failing to provide verbal assent. To be considered for the study, the target participants met the following criteria as indicated by teacher report or direct observation: (a) engaged in low levels of peer social interaction but were still responsive to peers’ bids for attention, (b) engaged in pretend play, (c) answered questions with at least one intelligible word (e.g., spoken, AAC), and (d) used an AAC device. To be considered for study inclusion, the peer participants met the following criteria: (a) engaged in high rates of peer social interaction, (b) engaged in high rates of compliance when given directions from a teacher, (c) regularly engaged in pretend play, (d) receptively understood questions from peers, (e) answered questions with at least one intelligible word, and (f) were likely to interact with peers during free play. Pretend play was included to ensure that children would likely have the skills to adequately play with their peers.

The target participants were Skylar, a 38-month-old Caucasian female child with Down syndrome, and Trent, a 63-month-old Caucasian male with Down syndrome. Skylar rarely interacted with her peers, preferred adult attention and interactions, and communicated via speech-generating software, sign language, and unintelligible vocalizations. Trent rarely engaged in play with peers, often exhibited noncompliant behaviors, and communicated via speech-generating software and vocalizations. Both children had been taught to use an AAC device and could use it independently to request (Skylar) and to request and comment (Trent) to adults, although they did not have regular access to devices and had not been previously taught to use them with peers. The peer participants paired with Skylar were a 36-month-old Caucasian male, Jaden, and a 38-month-old Caucasian male, Mason. Jaden and Mason had high social skills and often engaged in play with multiple peers although, based on observation, they spent the majority of their free time playing with each other. Trent’s peer was a 48-month-old Caucasian male, Michael, who typically engaged in play with a different peer from the same classroom. The peer participants were enrolled in the same inclusive preschool as Skylar and Trent; one peer had a diagnosed disability. The children’s teachers reported that all three of the peer participants interacted with other children often or very often, though they did not interact with the target child.

Settings

This study was conducted in two classrooms in a university-based inclusive early childhood program in southeastern United States. At study onset, Skylar’s classroom consisted of 10 children (including three children with disabilities) between the ages of 36 and 51 months; Trent’s classroom consisted of 12 children (including four children with disabilities) between the ages of 46 and 64 months. Each classroom had a lead teacher, an assistant teacher, and a graduate teaching assistant, and sometimes additional staff were present. The classrooms were approximately 8 m × 9 m and included a child-sized table and chairs, a small gross motor area, a sensory table, a bookshelf, and toys on child-level shelves that allowed for exploration (e.g., manipulatives, props, blocks, musical instruments).

The primary researcher and videographer implemented all intervention sessions inside the child’s classroom during a typically occurring or researcher-contrived free play activity (i.e., participants were brought to the classroom while nonparticipants were on the playground or in the gym). The videographer used a handheld camera and moved around the classroom to ensure all participants remained in frame. When participants were in the same center, the videographer adjusted the camera to ensure participant’s hands were visible and distinguishable (to increase the accuracy of coding play). During all sessions, the children were free to play in any part of the classroom. During intervention sessions, the other children and teachers in the classroom continued to follow their daily classroom routine and the other children could interact with the participants at all times. The primary researcher conducted all training sessions in a separate room down the hall from the classroom or in the target child’s classroom when nonparticipants (i.e., all other children) were not present. The separate room was approximately 5 m x 6 m and contained a child-sized kidney-shaped table, child-sized chairs, and various developmentally appropriate toys (e.g., puzzles, animal agents, books).

Materials

During baseline and intervention sessions, the participants played with toys that were in their classrooms (e.g., puzzles, blocks, trucks). The intervention materials included a token board, tokens, edibles (e.g., Goldfish^®, pretzels) and tangible items (e.g., stickers) reported to be preferred and that were offered as reinforcement, and an iPad^® with Proloquo2Go^™ software, a speech-generating AAC application. Skylar and Trent each had access to screen pages with buttons corresponding to the items and actions relevant to the centers in their classrooms. All sessions were recorded using a Vixia mini video camera. Data collectors used ProcoderDV to collect all data for baseline and intervention sessions (Tapp, 2003).

Response Definitions and Measurement Systems

The primary dependent variables were the peers’ behaviors: staying, playing, and talking with the target child. Stay was defined as being within 1.2 m (i.e., the length of three classroom tiles) of the target child and one of the following indicators: (a) in the same defined classroom center as the target child (e.g., book center, construction zone), (b) oriented toward the target child, or (c) facing the same direction as the target child (e.g., looking at the same toy on top of a shelf while standing in different centers). Play was defined as the peer manipulating the same or related materials (i.e., toys belonging to the same center) as the target child. Talk was defined as using language to communicate with the target child either with speech, sign language, or with an AAC device. Communication with an AAC device included pressing one or more buttons in conjunction with a second indicator to suggest communicative intent; these second indicators included the following: (a) joint attention gestures (e.g., point, show, give), (b) eye gaze shift in the direction of toy or peer, or (c) vocalization. See Table 1 for examples and non-examples of SPT behaviors. Intervals were coded as stay (S), stay and play (SP), stay and talk (ST), stay, play, talk (SPT), or none. For play or talk to be coded, the interval must have also been coded as stay, which aligns with Parten’s (1932) social play taxonomy. Research decisions were made based on the total percentage of intervals that peers were staying with the child with a disability, in isolation or in conjunction with playing and talking.

Table 1.

Operational Definitions, Examples, and Non-Examples for SPT Behaviors.

Behavior	Definition	Example	Non-example	Not codable
Stay	Any body part of target peer within 1.2 m of any body part of target child, and the target peer and target child are in the same center of the classroom, or the target peer and target child are oriented toward each other or in the same direction	Target peer stands next to target child; target peer sits in chair next to target child, who is sitting or standing; target peer has one foot/arm in the space the target child is in (or vice-versa)	Target peer is more than 1.2 m away; target peer is in a different space than the target child; target child has both feet outside of the space the target peer is in	Target child and/or target peer are off camera and could be within stay range
Play	Target peer manipulates materials while target child manipulates the same materials or materials that have a related function; target peer shows or gives materials to target child; target peer engages in pretend play while target child engages in related pretend play	Target peer hands target child a block; target peer drinks from a cup while target child picks up a plate	Target peer builds a bridge with blocks while target child picks up a cup; target peer taps target child on the shoulder	Unable to see target peer and/or target child’s hands and cannot reasonably assume both are manipulating objects
Talk	Target peer vocalizes language or activates AAC device directed to target child or to both peers about shared materials	Target peer says “Can I have that?” to target child; target peer presses “I like that” on AAC and looks at peer	Target peer screams or cries; target peer presses multiple buttons on AAC with his head down; peer is talking about the target child

Note. Holding objects counts as “play” within the first interval. The child must manipulate the object within the next interval for it to count as “play.” Manipulating AAC device can count as “play” if both children are manipulating the device. SPT = stay-play-talk; AAC = augmentative and alternative communication.

Data collectors recorded data using fixed interval momentary time sampling to code the SPT behaviors (Gast & Ledford, 2014). Intervals were 5 s in duration for a total of 60 intervals for each 5-min session. At the end of each interval, the observer recorded whether or not each target peer was demonstrating each of the three behaviors defined above. Sessions were also coded for a secondary variable: number of social interactions, a social interaction was defined as a verbal (e.g., vocalization, sign) or nonverbal (e.g., point, give) communicative interaction between the target child and target peers. Data for social interactions were collected using a count-based measurement system; data collectors recorded an occurrence at the onset of the behavior. Social interactions directed from peers to the target child and those directed from the target child to peers were coded; social interactions between peers were not coded. If a verbal and nonverbal social interaction occurred simultaneously (e.g., the peer hands the target child a toy and says “here you go!”), this was coded as two separate social interactions (i.e., one verbal and one nonverbal interaction).

Experimental Design

The researcher used an A-B-A-B withdrawal design replicated for two participants (Gast & Ledford, 2014). A withdrawal design is a demonstration design that uses sequential and repeated introduction and withdrawal of the intervention. This design allowed the researcher to evaluate causality between the SPT intervention (i.e., training, materials, and reinforcement) and the peers’ social behaviors (i.e., STP behaviors) compared with their behaviors in the baseline condition.

For Mason, Jaden, and Skylar, the researcher expanded to a multitreatment design (C-D-C-D; Gast & Ledford, 2014) to compare the effects of two different versions of the SPT intervention (SPT + modified reinforcement and SPT + modified arrangement). A multitreatment design was added because after implementing the SPT with modified reinforcement condition, levels of play and talk were functionally equivalent to the original SPT intervention. Thus, SPT with a modified arrangement was introduced to further target increasing levels of play and talk compared with SPT with modified reinforcement condition.

Withdrawal designs control for maturation effects because behavior change is demonstrated bidirectionally (i.e., from baseline to intervention and from intervention to baseline conditions). In addition, in the withdrawal design, each replication of effect (i.e., each introduction and withdrawal of the intervention) increases the internal validity of results by increasing demonstrations of experimental control. Replicating this design for similar participants allowed for inter-participant replication (i.e., external validity).

Procedures

Screening sessions

Before conducting the first baseline condition, the researcher, a graduate student in special education, selected four peers based on teacher report. To ensure target children met inclusion criteria, the implementer conducted probe screening sessions with participants who met initial screening criteria after receiving parental consent. Probe sessions for the target child were 5-min sessions during free play in the classroom with one of the peer participants. During these sessions, the peer participant (a) said the target child’s name and offered a preferred item and (b) said the target child’s name and showed the child a toy. For both types of trials, data were collected on whether the target child looked at the peer and/or toy, and to qualify for study participation, each participant had to respond to three out of the four peer bids for attention. Prior to these probe sessions, the implementer role-played both of the scenarios with the peer.

Baseline

Baseline sessions occurred once or twice a day for 5 min during typical or researcher-contrived free play in the target child’s classroom. To begin the session, the implementer gathered the three participants, and said, for example, “Mason, Jaden, and Skylar, it’s time to play!” The implementer did not interact with or prompt the participants during the session. The participants were able to play anywhere and with any toys in the classroom. The adults in the classroom were asked to not interact with the participants during these sessions unless a participant’s behavior might result in harm.

SPT training sessions

After baseline sessions were conducted, three (Trent and Michael) or four (Skylar, Jaden, and Mason) 15-min training sessions were conducted in a separate room down the hall or in the child’s classroom at a time when nonparticipants were absent. The training sessions occurred in the morning or afternoon on days that intervention sessions were implemented; for both groups, the last two training sessions were implemented in the morning and afternoon of the same day. During each session, one or more strategies (e.g., play with the same toys as your buddy; give your buddy a toy) were introduced, explained, and demonstrated via role-play with a second implementer, a doctoral student in Early Childhood Special Education.

During the first training session for both groups, the implementer taught the participants to stay with their buddy and do what their buddy is doing. At the end of this training session, the implementer explained the contingency of staying, playing, and talking to receive tokens to trade in for a reward. During the second session, the implementer taught the participants to give their buddy a toy and to manipulate the same materials. During the third session for Mason, Jaden, and Skylar, the implementer taught the participants to say their buddy’s name, ask their buddy questions, and comment on what their buddy is doing both verbally and with an AAC device. During the fourth training session for Mason, Jaden, and Skylar, the researcher introduced the AAC device to the peers and allowed them to explore the relevant pages. The peers were then taught how to appropriately communicate with the AAC device using prompting, and they practiced with the second implementer. For Michael and Trent, the third and fourth training sessions (described above) were condensed into a single session because there was only one peer. In all training sessions, the implementer provided each child with three behavior-specific praise comments related to strategy use. During the role-play practice sessions, the implementer praised each child approximately once per minute or prompted the child to engage in a certain behavior and then praised the child.

SPT

SPT sessions occurred for 5 min in the free play area of the target child’s classroom. To begin the session, the implementer gathered the three children in an area of the classroom and said, “For 5 minutes, we are going to be buddies! We are going to stay, play, and talk with our buddy. Mason and Jaden, you are buddies with Skylar, and Skylar, you are buddies with Mason and Jaden. Time to play!” During the session, the participants played with toys available in the classroom. Adults other than the implementer did not interact with the participants except to prevent physical harm. During the session, the implementer provided descriptive feedback every minute to each child on a rotating schedule (e.g., every 20 s to a different child). If the peers and target children were next to their identified “buddy,” they received a token on his or her column of the token board. Although the participants received tokens for staying, descriptive feedback served as a reminder for the next behavior in the STP hierarchy. For example, if the peer were staying next to the target child but not playing, the peer received a token/praise for staying and a reminder to play. At the end of the session, the implementer gathered the children again and provided brief social praise. The participants were then permitted to choose reinforcers from an array of preferred edible and tangible items.

Withdrawal to baseline

After a clear level change in the target behaviors between the baseline and SPT conditions, the withdrawal to baseline phase was implemented with identical procedures to the baseline sessions described above. The session began with the implementer delivering the task direction, “Mason, Jaden, and Skylar, it’s time to play!” SPT materials (i.e., strategies poster, token board, reinforcers) were not available.

SPT with modified reinforcement (SPT-C)

After the second phase of intervention was implemented for Mason, Jaden, and Skylar, a modified condition was implemented in an attempt to elicit more complex interaction behaviors. In this condition, the general procedures from the original SPT condition were followed (e.g., contingency review, task direction). However, the reinforcement system consisted of 10 (rather than five) total opportunities to receive tokens. In this condition, the participants received tokens for playing, not just staying, with the target child. This reinforcement was provided on the same 20-s rotating schedule. In addition, every minute (e.g., 0:00–1:00, 1:00–2:00) each participant was provided a token for the first time they talked to the target child during the minute. Thus, at the end of each participant’s 1-min interval, they were provided (a) a token for playing or feedback on playing if they were not already and (b) a model for talk if they had not already received reinforcement for talk during the previous minute. Talk was modeled on the AAC device for Skylar, coupled with a verbal model, and Mason’s and Jaden’s models for talk were either speech only or via AAC device plus speech.

SPT with modified arrangement (SPT-D)

This condition was implemented for Mason, Jaden, and Skylar after there was no clear level change in the peer’s levels of play behaviors between the SPT and SPT-C conditions. The procedures for this condition consisted of the same procedures as the C condition, but only one peer was present for each session (i.e., sessions included either Mason and Skylar or Jaden and Skylar). This decision was made because the peers were playing with and talking to each other a considerable amount; this was consistent with their typical classroom behavior, but it may have interfered with rates of interaction with Skylar. Thus, instead of providing feedback every 20 s to one of the participants, the experimenter provided feedback every 30 s on the rotating schedule.

Interobserver Agreement (IOA)

Reliability data were collected for at least 33% of all sessions and conditions using ProcoderDV (Tapp, 2003) for each participant; sessions were randomly selected and videos were viewed and coded independently. Before beginning data collection, the second coder was trained to criterion on the coding system. The second coder first reviewed the operational definitions, examples, and non-examples. Then, the primary researcher explained the data collection system, and the researcher and second coder watched a video of a sample session, while the researcher explained the rationale for each code. The second coder coded previously recorded practice sessions until criterion (90% agreement) was achieved across two consecutive sessions.

IOA was calculated on a point-by-point basis separately for each dependent variable. An agreement was recorded if both observers recorded the occurrence or nonoccurrence of each code (e.g., if the primary researcher recorded “stay” for one interval, and the second observer recorded “stay play,” there would be agreement for stay and talk, and disagreement for play). For count-based variables, an agreement was recorded for occurrence of social interactions if both data collectors recorded the occurrence of a social interaction within 5 s of each other. IOA was calculated as follows: number of agreements divided by the sum of agreements plus disagreements, multiplied by 100 to result in a percentage of agreement ([agreements] / [agreements + disagreements] × 100; Gast & Ledford, 2014). IOA for SPT data fell below the criterion of at least 80% for one session; the researcher and second coder reviewed the discrepant video together and discussed disagreements.

Average IOA data in baseline sessions across SPT behaviors (see Table 2) were 99% (range = 92%–100%) for Mason, Jaden, and Skylar and 99% (range = 95%–100%) for Michael and Trent. Low agreement for Mason and Jaden was due to a discrepancy in data collectors’ coding regarding the functional relatedness of toys; since stay was coded based on participants’ orientation toward functionally related toys, this discrepancy affected the stay IOA data.

Table 2.

Average SPT IOA Data Across Sessions.

Dependent Variable	Mason, Jaden, & Skylar			Michael & Trent
	Mason (%)	Jaden (%)	Skylar (%)	Michael (%)	Trent (%)
Baseline
Stay	97(92–100)	99(98–100)	—	98(95–100)	—
Play	100	100	—	100	—
Talk	100	100	—	99(97–100)	—
SI	100	100	83(50–100)	89(80–100)	89(67–100)
SPT
Stay	88(55–100)	98(95–100)	—	96(81–96)	—
Play	97(93–100)	95(85–100)	—	91(90–100)	—
Talk	98(92–100)	100	—	97(90–100)	—
SI	82(33–100)	93(67–100)	100	87(81–90)	92(84–97)
SPT C
Stay	96(88–100)	96(92–100)	—	—	—
Play	95(86–100)	95(93–98)	—	—	—
Talk	99(87–100)	99(98–100)	—	—	—
SI	81(29–100)	84(50–100)	81(0–100)	—	—
SPT D
Stay	100	99(98–100)	—	—	—
Play	92(87–92)	92(87–97)	—	—	—
Talk	99(98–100)	96(93–100)	—	—	—
SI	92(75–100)	86(84–100)	100	—	—

Note. Low SI agreement was due to low occurrences. SI = social interactions; SPT = stay-play-talk; IOA = Interobserver agreement.

Procedural Fidelity

Procedural fidelity data were collected via video for at least 33% of sessions across all conditions, implementers, and participants by a non-implementing data collector. After data collection, the implementer reviewed the procedural fidelity data before beginning the next session. Procedural fidelity was measured separately for each participant, behavior, and session, and was calculated on a point-by-point basis, (correct behaviors / [correct + incorrect behaviors]) × 100, for each behavior (Gast & Ledford, 2014).

For the 5-min baseline sessions, the implementer did not interact with the participants after delivering the task direction “It’s time to play.” SPT training sessions lasted for approximately 15 min, and the implementer behaviors were to (a) ensure all correct materials (i.e., SPT strategies visuals, reinforcers, toys) were present before beginning, (b) introduce the strategies for the day, (c) check with each participant to make sure he or she knows who his or her buddy is, (d) praise each child with three behavior-specific praise statements related to any SPT strategy, and (e) provide the children with two reinforcers at the end of the session.

SPT classroom sessions were designed to last for 5 min. During SPT sessions, the implementer behaviors were to (a) remind the participants to stay, play, and talk with his or her assigned buddy before the session begins, (b) begin the session upon delivery of the task direction “it’s time to play,” (c) avoid interaction with any of the participants during the session unless a child is in physical danger, (d) post SPT strategies visuals at children’s eye-level in the free play area of the classroom, (e) provide descriptive feedback to each child every minute, (f) provide a token to each child every minute if he or she has been staying with the target peer, and (g) provide the corresponding number of reinforcers (based on number of tokens earned) to each child at the end of the session. For the SPT-C and SPT-D conditions, steps (e) and (f) were modified as described above.

Procedural fidelity data were collected for 40% of all sessions for Mason, Jaden, and Skylar, ranging between 33% and 50% of sessions for each condition. The average procedural fidelity across all participants and conditions was 99%; data averaged 100% in the baseline conditions, 93% in training sessions, 98% in SPT conditions, 99% in SPT-C conditions, and 100% in the SPT-D conditions. Procedural fidelity data were collected for 37.5% of all sessions for Michael and Trent, ranging between 33% and 40% of sessions for each condition. The average procedural fidelity across all conditions was 98%; data averaged 99% in the baseline conditions, 93% in training sessions, and 99% in SPT conditions.

Results

Videos were coded after every session and data were graphed and visually analyzed each day. Data are presented as the percentage of intervals that peers engaged in SPT and the number of social interactions; verbal and nonverbal interactions are plotted together and descriptive data on verbal and nonverbal interactions are shown in Table 3. Experimental decisions regarding condition changes were determined based on the presence of at least three consistent stay data points in a condition, or a sufficient number of data points to conclude data were likely to remain variable but without a detectable trend.

Table 3.

Average Number of Verbal and Nonverbal Social Interactions Across Conditions.

Interaction Type	Mason, Jaden, & Skylar			Michael & Trent
	Mason	Jaden	Skylar	Michael	Trent
Baseline
Verbal	0.1(0–1)	0	1.6(0–6)	8.2(0–18)	12.6(1–26)
Nonverbal	0.1(0–1)	0	0	6.5(0–24)	8.7(0–20)
SPT
Verbal	0.4(0–3)	1(0–3)	0.6(0–4)	14.5(2–21)	15(7–22)
Nonverbal	0.2(0–1)	0.9(0–3)	0.1(0–1)	8.7(0–16)	6.9(4–13)
SPT-C
Verbal	2.4(0–6)	2.6(0–5)	0.5(0–2)	—	—
Nonverbal	0.4(0–1)	2.1(0–4)	0	—	—
SPT-D
Verbal	5.8(1–11)	13.9(5–24)	0.8(0–3)	—	—
Nonverbal	3(0–11)	2.1(0–4)	0.1(0–1)	—	—

Note. SPT = stay-play-talk.

Baseline to SPT Comparisons for Trent and Michael

Stay

Michael’s levels of stay during the first baseline condition were highly variable, ranging from 0% to 100% of intervals (see top panel of Figure 1). In the first SPT condition, there was an immediate and consistent level change to 85%; data stabilized at or near 100% during the next four sessions. In the second baseline condition, there was a slight decrease in level from SPT, although the data were variable. During the second SPT condition, there was an immediate level change to 95%, with all data points consistently between 78% and 95% of intervals. Thus, the data were highly variable and generally at a lower level in baseline (though there was some overlap with the SPT conditions); the data were slightly variable though generally higher in the SPT condition. Although the data in the second baseline condition were only slightly lower than the SPT condition, this is not unusual in A-B-A-B designs (Gast & Ledford, 2014).

Figure 1.

First three panels depict the percentage of intervals Michael engaged in target behaviors; The bottom graph depicts the number of social interactions of Michael and Trent.

Play

Michael’s play data were at or near 0% of intervals during the first baseline condition (see second panel of Figure 1). In the first SPT condition, there was a clear and consistent level change, with data ranging from 45% to 75% of intervals. In the second baseline condition, data were highly variable, ranging from 0% to 52% of intervals. In the second SPT condition, data were considerably less variable and returned to levels similar to the first SPT condition. Thus, Michael’s levels of play were low and/or highly variable in the baseline conditions and higher and more stable in the SPT conditions.

Talk

Michael’s levels of talk were at or near 0% during the first baseline condition. In the first SPT condition, levels of talk increased, though were variable, ranging from 0% to 23% of intervals. In the second baseline condition, levels of talk generally decreased and ranged from 0% to 12%. During the second SPT condition, talk levels were more consistently above 0%, though they did not increase above 10%. Thus, Michael’s level of talk increased during the first SPT condition, and all following conditions were generally, though variably, higher than the initial baseline condition. Thus, as shown in the third panel of Figure 1, no functional relation was demonstrated for this dependent variable.

Social interactions

Data were also collected on the number of social interactions of Michael (the target peer) and Trent (the target child; see bottom panel of Figure 1). These data were variable across conditions, and no changes in level, trend, or variability were temporally associated with condition changes.

Baseline to SPT Comparisons for Skylar, Jaden, and Mason

Stay

Jaden’s and Mason’s levels of stay behaviors during both baseline conditions were consistently at or near 0% of intervals. In the first session of the SPT condition, there was an immediate level change to 100% for Jaden and 93% for Mason. However, data were variable for the remaining sessions in this condition (minimum values of 25% for Jaden and 45% for Mason). During the second SPT condition, data were also variable, though the last four sessions consistently ranged between 77% and 100% for both participants. Thus, as shown in the first panel of Figure 2, levels of stay increased in variability and level in the SPT conditions relative to baseline.

Figure 2.

First three panels depict the percentage of intervals Mason and Jaden engaged in target behaviors; The bottom graph depicts the number of social interactions of Mason, Jaden, and Skylar.

Play

Jaden’s and Mason’s play behaviors were stable at 0% of intervals in both baseline conditions (see second panel of Figure 2). In the first SPT condition, there was a slight increase in level, which was larger for Mason than for Jaden. In the second SPT condition, Jaden’s and Mason’s play levels were highly variable. Across participants, play data were at 0% in baseline conditions; data were slightly higher and highly variable in the SPT condition. Although data in the SPT condition were variable, a functional relation was present because there were clear level differences between the conditions.

Talk

Jaden’s and Mason’s levels of talk were at 0% of intervals for both baseline conditions and the first SPT condition (see third panel of Figure 2). In the second SPT condition, there was a slight increase in level, which was larger for Mason (13%) than for Jaden (2%). Given the considerable amount of overlap between conditions, no functional relation was present for talk.

Social interactions

Data were also collected on the number of Mason’s, Jaden’s, and Skylar’s social interactions. Mason’s and Skylar’s number of social interactions were generally at or near 0 across both baseline and SPT conditions. Jaden’s social interactions were slightly higher and more variable in the SPT conditions.

SPT Modification Comparisons

Stay

During the first SPT-C condition, data for Mason were consistently high between 90% and 100%; data for Jaden initially dropped to 38% but were then stable between 88% and 95%. In the first SPT-D condition, Mason’s data were consistently at 100%; Jaden’s data also were high but more variable. In the second SPT-C condition, both participants’ data initially dropped to 45% and 50% and then increased to 82% to 100% of intervals for the rest of the condition. In the second SPT-D condition, Mason’s and Jaden’s levels of stay were high yet slightly variable. In these four modified conditions (i.e., C-D-C-D), levels of stay were generally higher and slightly less variable in the SPT-D conditions relative to the SPT-C conditions.

Play

In the first SPT-C condition, Mason’s data continued to be variable though decreased slightly (range = 2%–45% of intervals); Jaden’s data were still highly variable. In the first SPT-D condition, there was a clear level change and decrease in variability in both Mason’s and Jaden’s data; the percentage of intervals in which the participants engaged in play behaviors was 73% to 87% for Mason and 58% to 85% for Jaden. In the second SPT-C condition, data for Mason and Jaden were highly variable. In the second SPT-D condition, Mason’s data were initially similar to the previous SPT-C condition, with an increasing trend, and Jaden’s data were similar to the first SPT-D condition. Due to the trend and relative dearth of data points, it is difficult to draw conclusion about play levels in this condition. Thus, across participants, data were highly variable in the SPT-C conditions, and data were higher and less variable in the SPT-D conditions (see second panel of Figure 2).

Talk

In both SPT-C conditions, Mason’s and Jaden’s talk data were consistently at or slightly above 0% of intervals (see bottom panel of Figure 2). During the first and second SPT-D conditions, Mason’s data slightly increased in level and were more variable; Jaden’s data increased in level and were less variable. Thus, data were at 0 or very close to 0% in the SPT-C conditions and were higher though more variable in the SPT-D conditions.

Social interactions

Social interactions in the SPT-C conditions were low and slightly variable (range = 0–7) for Mason and Jaden and at or near 0 for Skylar. In the SPT-D conditions, social interactions increased in level and variability for Mason and Jaden (range = 1–26; see bottom panel of Figure 2); Skylar’s remained at or near 0. Verbal and nonverbal social interactions were coded separately; the averages across conditions are displayed in Table 3. Nonverbal social interactions were low in the SPT-C conditions; in the SPT-D conditions, nonverbal interactions increased in variability for Mason (outliers of 8 and 11 interactions) and slightly in level for Jaden (range = 2–4). However, verbal social interactions data varied by condition. In the first SPT-C condition, there was a small level change with little overlap with the previous condition. In the first SPT-D condition, there was an immediate and consistent level change, ranging between 4 and 9 for Mason and 5 and 19 verbal interactions for Jaden. In the second SPT-C condition, verbal interactions immediately decreased in level, though were variable. In the second SPT-D condition, an increasing trend was present for Mason, and there was an immediate and clear level change for Jaden, ranging from 14 to 24 social interactions. For Skylar, these data were low and variable across conditions, and no changes in level, trend, or variability were temporally associated with condition changes. Thus, the SPT-D conditions resulted in increased levels of verbal social interactions for both Mason and Jaden, and no changes in interactions for Skylar.

Discussion

The findings of this study suggest two primary conclusions: (a) for Mason, Jaden, and Skylar, SPT with modified arrangement resulted in peers more consistently staying in proximity to, playing with, and talking more to the target child; and (b) for Michael and Trent, SPT resulted in the peer staying and playing with the target child more consistently. Results indicate a functional relation between the original SPT intervention and increased stay and play behaviors for Michael and Trent. For Mason, Jaden, and Skylar, results indicate a functional relation between SPT with a modified arrangement and increased stay and play behaviors.

Implications

There are several key implications from this study. First, characteristics of participants with disabilities may be important to consider when designing SPT interventions. Skylar exhibited a low frequency of engagement in functional play behaviors, which likely decreased the ease with which she and her peers could engage in play together. Comparatively, Trent’s engagement in functional play was significantly higher than Skylar’s; this may have contributed to Trent’s and Michael’s higher levels of play in the SPT intervention compared with Mason’s and Jaden’s play with Skylar. Interestingly, Skylar had previously demonstrated functional and pretend play and met inclusion criteria for study participation (i.e., nonfunctional play resulted from a performance rather than an acquisition deficit). It is likely easier for children to join in play that is functional, rather than initiating play with the target child or joining play that is repetitive or manipulative in nature. Similarly, the target children’s receptivity to peer initiations may be an important component of the differences in social interactions between groups. Trent was more interested in responding to and initiating play with his peer, and many social interactions occurred because of Trent’s initiations. Moreover, the number of interactions initiated by Trent is significantly higher than those initiated by Skylar; this likely affected the number of social interactions throughout the study.

The composition of peer dyads or triads may be important when designing an SPT intervention. The results from the SPT with modified arrangement condition for Mason, Jaden, and Skylar indicate that creating peer buddy dyads, rather than triads, may result in more staying, playing, and talking between the typically developing child and the child with disabilities. Moreover, teachers reported that Mason and Jaden were best friends and tended to play with each other in free play, which may have affected their motivation to initiate play with Skylar. Similarly, differing play interests between the target child and peers may have affected the levels of play. Several times Mason and Jaden played together in a different center than Skylar; when prompted to play with Skylar, the peers explicitly stated that they would rather stay where they were because they preferred the toys in that center and they were not allowed to move the toys (this stated rule was not one learned or enforced during the study). After the researcher explained that the children could move toys during the SPT sessions, Skylar was not necessarily interested in the toys that the peers brought from other centers. Thus, teachers and researchers should choose groups carefully and be cognizant of possible drawbacks of choosing certain peer buddy groups based on friendships and play scheme preferences.

The SPT intervention was effective in increasing the levels of stay for both peer buddy groups and play for Michael and Trent. This suggests that the intervention is effective in increasing the amount of time peers remain in proximity to a peer with a disability, but more nuanced strategies may be necessary to increase socially valid interactions. For example, SPT with modified arrangement was more effective in increasing the play and talk of Skylar’s peers. Because there were many differences between the two buddy groups (e.g., child with disability levels of functional play and receptivity to peers; dyad vs. triad buddy groups), it is not possible to determine which factors affected the different levels of SPT behaviors between groups.

Children’s proficiency and frequency of using an AAC device may be an important consideration as it may affect a child’s level of communication. Although Skylar and Trent did not have access to an AAC device in the classroom, they had both been taught to use the devices to request and make comments in the context of play, and they were functional users of the device. On a typical day, however, Skylar communicated more often and more proficiently via sign language, though her signs were not widely acknowledged by her teachers or peers. Throughout the study, Skylar’s communicative use of the device was limited; her functional use of the device consisted primarily of saying her peer’s names. She typically repeatedly pressed multiple buttons, without shifting her attention or eye gaze to indicate communicative intent. Informal teacher report indicated that she used her device appropriately with adult support, which may suggest that talking with peers is a more sophisticated skill than requesting or commenting with adults. However, adult modeling was a component of the SPT with modified reinforcement and modified arrangement conditions, and her verbal interactions did not increase in these conditions.

At study onset, Trent spent the majority of the session exploring his device, either in a center by himself or with Michael. Although the researcher provided access to the device prior to starting the study, this is likely an artifact of his lack of access to one in the classroom, despite his ability to effectively communicate via AAC. Once Trent started playing during the sessions, he used his device to appropriately communicate with Michael, typically during seated play. Michael adequately verbally responded to this communication, which may have influenced Trent’s use. However, anecdotally, Trent used vocalizations and gestures more frequently than his AAC device to communicate with Michael. Although he did not have a large repertoire of intelligible vocalizations, Michael appeared to understand and respond to these communications in the context of active play. A recurring play scheme during the study included putting food into a shopping cart and driving the shopping cart around the classroom; perhaps the active nature of this play scheme was less conducive to using the device than others (i.e., it is more difficult to carry the device around when pushing a shopping cart). However, when Michael and Trent were engaged in seated play (e.g., cars, building towers), Trent consistently used his device. Overall, Trent’s level of communication was higher than Skylar’s, and Michael’s responsiveness was higher than that of Mason’s and Jaden’s.

One component of the training sessions consisted of teaching peers to use the AAC device. The researcher modeled appropriate use and used prompting to teach the peers to comment with the device in the context of play with a confederate adult. Although Mason and Jaden were initially interested in talking with the device, their button activations were primarily noncommunicative (e.g., continually pressing multiple buttons). When Michael used Trent’s device in the beginning of the study, Trent often pushed him, verbally said “no” or “mine,” or grabbed the device back from him. This may have decreased Michael’s motivation to activate the device as the study progressed. These observations suggest that more substantial peer training involving AAC device use and more supported interactions between the child with and without disabilities may be necessary.

The addition of the AAC component to the SPT intervention resulted in increases in peers’ prosocial skills, consistent with previous research on SPT interventions without an AAC component (e.g., Hughett et al., 2013; Kohler et al., 2007). However, the variable target child outcome data (i.e., Trent’s social skills increased, but Skylar’s did not) in this study may not be consistent with previous research that indicates SPT results in increases in target child social skills (Barber, Saffo, Gilpin, Craft, & Goldstein, 2016; English et al., 1997; Kohler et al., 2007).

Limitations and Future Research

The operational definitions for play may have influenced the results of this study. Our definitions were based on previous SPT research, which aligned with Parten’s (1932) social play taxonomy. She defined play relative to the child’s interactions with peers across the following categories: (a) onlooker, (b) solitary, (c) parallel, (d) associative, and (e) cooperative; the latter three categories are generally considered appropriate social play goals for preschool children and the focus of SPT. Thus, the definition for play was such that the criteria for stay had to first be satisfied (consistent with parallel play; Parten, 1932); if the participants were engaging in a joint engagement interaction farther than 4 ft apart, this was not coded as play. For example, Michael and Trent’s grocery store play scheme was not conducive to being within 4 ft of each other; although they were clearly engaging with each other (e.g., presence of joint attention), the definition for play did not capture these and similar instances of typical child play. In addition, participants were reinforced for playing next to the target child, not necessarily with, which often resulted in parallel play rather than associative or cooperative play; however, parallel play is an appropriate initial social play goal for preschoolers. The operational definition for play may have underestimated more complex forms of social play (i.e., associative or cooperative play). In addition, although SPT sessions were brief, the level of teacher support was high, which may not be feasible when teacher to child ratios are high. Future research should include systems that specifically result in reinforcement for the peer when he or she is engaging in more complex types of social play with the target child, systematically fade the reinforcement to reduce prompt dependency, and increase feasibility of classroom implementation.

Similarly, the coding system may have unduly influenced the levels of coded talk. The use of momentary time sampling likely underestimated the coded occurrences of talk; it is probable that some utterances between the target child and his or her peers were not captured by the coding system. However, nonverbal and verbal social interactions (the secondary variable) were coded separately; because this was a count-based measure, verbal social interactions may be a more accurate value for the amount of talk between the children. However, social interactions were coded after the study was complete (i.e., post hoc), which did not allow those data to be used formatively (i.e., the data were not used to make decisions during the course of the study).

Second, the target children’s receptivity to peers and engagement in functional play may have decreased the impact of this intervention. There were repeated instances of unanswered questions or comments from Skylar’s peers, and Skylar’s lack of responding to Mason and Jaden may have decreased their motivation and/or willingness to continue to initiate meaningful interactions with her. Screening procedures included probe sessions which were designed to determine whether the children with disabilities were receptive to peer’s bids for attention. These probe sessions consisted of the peers saying the target child’s name while offering a preferred item. This may have overestimated the extent to which Skylar would be responsive to other peer initiations; a more comprehensive intervention package (i.e., one that first teaches children to attend to their peers and encourages functional play) may be beneficial for children who do not typically attend to more complex initiations from peers and have nonfunctional play behaviors.

This study was originally designed to be implemented in the classroom during typically occurring free play activities alongside the other children in the classroom. After implementing several sessions with Mason, Jaden, and Skylar this way, sessions were then implemented in researcher-contrived free play settings. This was done after a parent of a nonparticipant demonstrated concern that rewarding some children for being near others might discourage them from including other, nonparticipating children. Thus, some sessions occurred while nonparticipating children were in the classroom and some while only the participants were in the classroom. This latter context may limit the likelihood of generalizing these SPT behaviors to other contexts that typically occur during the school day. Implementing this intervention on a class-wide basis during typically occurring classroom activities may increase the acceptability and generalizability of SPT. Researchers might consider variations of SPT that could be conducted in a class-wide manner (e.g., each child is assigned a buddy, and SPT behaviors are reinforced for all children).

Finally, social validity data were not collected, and additional data for Mason and Jaden were not collected because neither participant returned to school after winter break. This lack of sufficient data limits the ability to draw conclusions regarding the effectiveness of the SPT with modified reinforcement compared with the SPT with modified arrangement conditions. Future research should further analyze the efficacy of creating buddy interventions with peer dyads compared with triads.

Conclusion

The purpose of this study was to investigate the effectiveness of an SPT intervention with an AAC training component on related peer behaviors (staying, playing, talking, and social interactions with a child with a disability). Despite the limitations, the findings of this study are of considerable importance because they suggest SPT interventions with peer dyads may increase the amount of time peers stay and play with children with disabilities who use AAC. Further research on this topic is warranted to address the ideal social skills intervention package for children with varied levels of AAC usage and other communication modes.

Footnotes

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

References

Barber

A. B.

Saffo

R. W.

Gilpin

A. T.

Craft

L. D.

Goldstein

(2016). Peers as clinicians: Examining the impact of stay play talk on social communication in young preschoolers with autism. Journal of Communication Disorders, 59, 1–15.

Bock

S. J.

Stoner

J. B.

Beck

A. R.

Hanley

Prochnow

(2005). Increasing functional communication in non-speaking preschool children: Comparison of PECS and VOCA. Education and Training in Developmental Disabilities, 40, 264–278.

Clarke

Kirton

(2003). Patterns of interaction between children with physical disabilities using augmentative and alternative communication systems and their peers. Child Language Teaching and Therapy, 19, 135–151.

Cosbey

J. E.

Johnston

(2006). Using a single-switch voice output communication aid to increase social access for children with severe disabilities in inclusive classrooms. Research and Practice for Persons with Severe Disabilities, 31, 144–156.

English

Goldstein

Kaczmarek

Shafer

(1997). Promoting interactions among preschoolers with and without disabilities: Effects of a buddy skills-training program. Exceptional Children, 63, 229–243.

Gast

D. L.

Ledford

J. R.

(Eds.). (2014). Single case research methodology: Applications in special education and behavioral sciences (2nd ed.). New York, NY: Routledge.

Goldstein

English

Shafer

Kaczmarek

(1997). Interaction among preschoolers with and without disabilities: Effects of across-the-day peer intervention. Journal of Speech, Language, and Hearing Research, 40, 33–48.

Goldstein

Kaczmarek

Pennington

Shafer

(1992). Peer-mediated intervention: Attending to, commenting on, and acknowledging the behavior of preschoolers with autism. Journal of Applied Behavior Analysis, 25, 289–305.

Guralnick

M. J.

Connor

R. T.

Hammond

Gottman

Kinnish

(1996). Immediate effects of mainstreamed settings on the social interactions and social integration of preschool children. American Journal on Mental Retardation, 100, 359–377.

10.

Guralnick

M. J.

Hammond

M. A.

Connor

R. T.

Neville

(2006). Stability, change, and correlates of the peer relationships of young children with mild developmental delays. Child Development, 77, 312–324.

11.

Honig

A. S.

McCarron

P. A.

(1988). Prosocial behaviors of handicapped and typical peers in an integrated preschool. Early Child Development and Care, 33, 113–125.

12.

Horner

R. H.

Carr

E. G.

Halle

McGee

Odom

Wolery

(2005). The use of single-subject research to identify evidence-based practice in special education. Exceptional Children, 71, 165–179.

13.

Hughett

Kohler

F. W.

Raschke

(2013). The effects of a buddy skills package on preschool children’s social interactions and play. Topics in Early Childhood Special Education, 32, 246–254.

14.

Joseph

J. D.

Strain

Olszewski

Goldstein

(2016). A consumer reports-like review of the empirical literature specific to preschool children’s peer-related social skills. In Reichow

Boyd

Barton

Odom

(Eds.), Handbook of early childhood special education (pp. 179–197). London, England: Springer.

15.

Kohler

F. W.

Greteman

Raschke

Highnam

(2007). Using a buddy skills package to increase the social interactions between a preschooler with autism and her peers. Topics in Early Childhood Special Education, 27, 155–163.

16.

Odom

S. L.

Buysse

Soukakou

(2011). Inclusion for young children with disabilities: A quarter century of research perspectives. Journal of Early Intervention, 33, 344–356.

17.

Parten

M. B.

(1932). Social participation among pre-school children. The Journal of Abnormal and Social Psychology, 27, 243–269.

18.

Robertson

Green

Alper

Schloss

P. J.

Kohler

(2003). Using a peer-mediated intervention to facilitate children’s participation in inclusive childcare activities. Education and Treatment of Children, 26, 182–197.

19.

Strain

P. S.

Bovey

E. H.

(2011). Randomized, controlled trial of the leap model of early intervention for young children with autism spectrum disorders. Topics in Early Childhood Special Education, 31, 133–154.

20.

Tapp

(2003). ProcoderDV. Nashville, TN: Vanderbilt Kennedy Center.

21.

Therrien

M. C.

Light

Pope

(2016). Systematic review of the effects of interventions to promote peer interactions for children who use aided AAC. Augmentative and Alternative Communication, 32, 81–93.

22.

Thiemann-Bourque

Brady

McGuff

Stump

Naylor

(2016). Picture exchange communication system and pals: A peer-mediated augmentative and alternative communication intervention for minimally verbal preschoolers with autism. Journal of Speech, Language, and Hearing Research, 59, 1133–1145.

23.

Trembath

Balandin

Togher

Stancliffe

R. J.

(2009). Peer-mediated teaching and augmentative and alternative communication for preschool-aged children with autism. Journal of Intellectual & Developmental Disability, 34, 173–186.

24.

Wolery

Hemmeter

M. L.

(2011). Classroom instruction: Background, assumptions, and challenges. Journal of Early Intervention, 33, 371–380.