Condition-related differences in verbalization patterns

Within each group, we were interested in identifying specific conditions that facilitated greater verbalization compared to other conditions. We hypothesized that children would engage in greater

verbalization with social partners during conditions that promoted social interactions without added motor demands. In contrast, we expected children to verbalize less during conditions in which they engaged in movement games involving imitation and synchrony. Our hypotheses were based on previous evidence on executive function deficits and motor impairments commonly seen in children with ASDs. We expected that conditions which required dual tasking i.e. conditions in which children were required to talk or sing songs as they were engaging in gross and fine motor activities, would elicit lower

verbalization, given children’s impairments in executive function including problems in attention shifting and working memory (Griffith, Pennington, Wehner, & Rogers, 1999; Hill, 2004; Ozonoff & Jensen, 1999). In addition, training activities involved imitation-based motor games, and given the motor impairments in this population, we expected children to demonstrate difficulty in speaking while

simultaneously engaging in movement-based activities. Children with ASDs demonstrate impairments in gross motor skills including bilateral coordination (Fournier et al., 2010; Green et al., 2009), imitation/ praxis (Dewey et al., 2007; Mostofsky et al., 2006), postural control (Minshew et al., 2004), balance, and gait (Hallett et al., 1993; Vilensky et al., 1981). Similarly, children also demonstrate impairments in fine motor control including poor visuomotor coordination, bimanual coordination, manual dexterity, and handwriting skills (Bhat et al., 2011; Fuentes, Mostofsky, & Bastian, 2009; Ghaziuddin et al., 1994; Ghaziuddin & Butler, 1998; Provost, Heimerl, & Lopez, 2007). Hence, we expected that conditions which involved practice of imitation and/or synchrony games in all 3 groups would be challenging for children and would therefore afford lower amounts of socially-directed verbalizations compared to conditions without any concurrent motor demands. In line with our hypothesis, all three groups demonstrated greater verbalization during conditions that promoted social interactions with minimal movement demands (see Figures 4A, 4B, & 4C). In the music group, children verbalized most during the social interaction phase and the action song condition compared to all other conditions. During the social interaction phase, children were encouraged to sing hello and farewell songs. Similarly, in the action song condition, where we used common nursery rhymes, children chose to spontaneously sing these songs with their social partners (see Figure 5-4A). In contrast, children verbalized less during the more complex conditions that

required them to synchronize with their social partners (see Figure 5-4A). Children seemed to focus on the motor demands of the task at the cost of verbal communication. Similarly in the robot group, children verbalized most with their social partners during the social interaction phase compared to other conditions (see Figure 5-4B). Lastly, in the academic group, children verbalized most during the social interaction phase and the reading condition. During the building and art-craft conditions, children focused on the fine motor aspects of the task and thereby demonstrated a lower quantity of verbalization compared to the first two conditions (see Figure 5-4C).

Children in the robot group also engaged in maximum self-directed verbalization during the drumming game compared to the warm up and social interaction conditions (see Figure 5-4B). As discussed previously, children associated quarter-eighth drumming patterns with words such as “Pal-Buddy” or “Grape-Lemon”. Children chose to spontaneously use these verbal cues as they practiced the drumming patterns. This might have led to the higher levels of self-directed verbalization during this condition compared to other conditions. Lastly, compared to all other verbalization targets, children verbalized the least to the robot across all conditions. However, comparing between conditions, children verbalized most to the robot during the social interaction phase and least during the walking game. During the social interaction phase, the robot greeted the child and also engaged in general conversation with the child such as asking children about their day, asking them about their family, favorite foods etc., which might have elicited responses from children. In contrast, we used the mobile non-verbal Rovio robot during the walking game, which might explain lower levels of verbalization to the robot during this game.

5.5. Clinical Implications

We compared the effects of novel movement-based therapies with a stationary, standard-of-care treatment on the spontaneous and responsive communication skills of children with ASDs. We were interested in examining what each type of context – music, robotic, and academic- affords in terms of verbal

contexts, in which children and their social partners were relatively confined at the table, naturally afforded trainers and models with multiple opportunities to initiate communicative exchanges with children. In contrast, the movement-based contexts required children and their social partners to engage in gross motor imitation and joint action to background music; both the movement and the music naturally restricted the number of opportunities available to trainers and models for initiating conversation with children. Secondly, the activities practiced in the standard-of-care intervention were highly familiar to children, since they often engaged in these activities in special education settings. In contrast, the

activities practiced in the movement groups were highly novel and challenging for children, since they are not conventionally included within the treatment plan for autism. Our data suggest that children in the movement groups might have found it harder to communicate while moving to the beat of music, leading to lower amounts of socially-directed verbalization in the music and robot groups in the early session compared to the academic group. However, it was very encouraging to see that over time, children in the music group showed a large magnitude of increase in the duration of social verbalization that reached levels comparable to those in the academic group. The robot group demonstrated improvements in verbalization to the model following training but the levels were lower than that seen in the music and academic groups even at the end of training. The academic group did not show any training-related increase in socially-directed verbalizations possibly due to a ceiling effect. We also found that the nature of the movement contexts, music and robotic, promoted predominantly spontaneous communication, whereas the academic context promoted greater amounts of responsive communication compared to the other 2 groups. In the music group, children were very engaged in the context and over sessions learned songs and began to spontaneously sing songs. The enjoyable context also fostered greater communication between children and their social partners. Although children increased their conversation with the model across sessions, contrary to our hypothesis, children did not find the robots very engaging and the robots failed to maintain children’s interests across sessions. As a result, children engaged in behaviors

conditions that focused on communication without any added motor demands afforded greater verbal communication compared to conditions that placed dual task demands on children.

Overall, we think that movement-based contexts have great potential in facilitating social communication skills. Specifically, children find music and movement activities enjoyable and engaging and these contexts can be used to promote reciprocal communication between children and their social partners. Other data from our lab also suggest that children in the music group demonstrated greater social

attention, an increase in positive affect, and a reduction in negative behaviors across training sessions. In contrast, the robot group showed greater boredom, more negative affect, and no reduction in negative behaviors across sessions. Lastly, the academic group directed greater attention to objects and engaged in greater sensory behaviors with objects. However, this group showed the greatest compliance with

training, probably due to the familiarity of the practiced activities. Our study suggested that the current robotic technology has limited utility as an adjunct tool in the treatment of children with ASDs. Future attempts at using robotics for autism will need to address the limitations of the current technology by making robots more autonomous and dynamically contingent to children and by providing robots with human-like capacities to adapt to the needs of each individual child. The verbal repertoire of the robot needs to be broadened, and efforts must be made to make the robot’s speech comprehensible to children with ASDs. To summarize, our work suggests that movement-based interventions, especially music-based activities, are engaging for children with ASDs, and by capitalizing on their inherent strengths can serve as promising tools to promote social communication and motor skills in this population.