Parent-child nonverbal engagement during spoken versus sung book-sharing in preschoolers with and without autism

Musical activities and shared attention

Book reading activities and shared attention

Shared book reading is a common joint parent–child activity that involves attention to one’s partner and a common object (book). During shared book reading activities, preschoolers with ASD generally show greater passive engagement or non-engagement compared to same-aged peers with TD (Bean et al., 2020; Fleury & Hugh, 2018; Lanter et al., 2013; Westerveld et al., 2020). However, children’s engagement can be moderated by the specific book context and parents’ behavior (Fleury & Hugh, 2018). Both children with ASD and children with TD were more actively engaged (verbally/nonverbally communicating with their parents about the book) during familiar books compared to novel books (Fleury & Hugh, 2018). Increased quality of parents’ book reading (e.g., supporting child’s attention to the text and reading comprehension) also increased active engagement in children with ASD and TD during a book sharing activity (Fleury & Hugh, 2018). However, to the knowledge of the authors, no studies have considered joint book sharing in a musical compared to a non-musical context.

In the current study, we investigated how a musical context may modulate child and parent behavior during book sharing. We considered two measures of nonverbal engagement during parent–child musical and non-musical book sharing: Visual attention (gaze) and interpersonal movement coordination. We focused on these nonverbal measures of engagement due to their involvement in both musical and non-musical interactions (e.g., Kim et al., 2008; LaGasse, 2014; Paul et al., 2015; Tunçgenç & Cohen, 2018) and because they can be used with children with varying language levels (Kim et al., 2008; Pfeiffer et al., 2018). These measures also allowed us to consider the behaviors of each partner individually as well as within an interactional, dyadic framework.

With three aims, we address whether musical contexts increase visual attention and interpersonal coordination for parent–child dyads of children with TD and ASD. Our first aim was to compare caregiver and child nonverbal engagement during book sharing with versus without a musical context. In order to create these contexts, caregiver–child dyads shared children’s picture (non-musical) books and song (musical) books. We hypothesized that, for both ASD and TD dyads, the musical context of song books would increase child and caregiver engagement as measured via increased visual attention to the activity and increased interpersonal movement coordination. The second aim was to compare caregiver and child nonverbal engagement during dyadic book sharing for dyads with children with ASD and dyads with children with TD. We hypothesized that children with ASD would show less engagement than children with TD during the book sharing activity. The third aim was to examine individual differences in engagement during these book sharing activities. We hypothesized that caregivers’ and children’s visual attention would positively correlate with each other, reflecting caregivers’ attunement to their children’s engagement in the task. We also hypothesized that children’s engagement would relate to their language level.

Method

Participants

Participants included 29 preschool-aged children with ASD (male n = 10, female n = 3) aged between 21 and 69 months and with TD (male n = 10, female n = 6) aged between 20 and 52 months and one of their caregivers. In the ASD group, caregivers included 9 mothers, 2 fathers, 1 grandmother, and 1 grandfather; in the TD group, caregivers included 12 mothers, 3 fathers, and 1 grandmother. To encompass the grandparents, the term “caregiver” (rather than parent) will be used throughout the rest of the manuscript. Two additional participants (one ASD, one TD) were excluded from the study because the caregivers did not follow the task instructions (below; i.e., did not sing during the song book condition). Participants were recruited through medical center clinics, preschools, community events, and social media. No musical training was required for participation. The children with ASD and with TD were matched on nonverbal developmental level (Mullen Scales of Early Learning [MSEL] Visual Reception age equivalences (see Table 1). Children with ASD were older than children with TD and had lower language levels (MSEL Receptive/Expressive Language age equivalences (see Table 1).

OPEN IN VIEWER

Table 1.

Child Demographics and Clinical Scores by Group.

	ASD	TD	Test statistic(χ² or t)	Significance (p)
n (M: F)	13 (10:3)	16 (10:6)	0.70	.40
Chronological Age (months)	46.6 (10.3)	35.9 (8.5)	3.06	.005**
Race (% Caucasian)	69.2%	87.5%	1.46	.23
Household Income (% Over $70,000)	69.2%	67.8%	0.001	.98
MSEL VR T Score	42.0 (16.3)	61.8 (10.9)	-3.85	<.001***
MSEL VR AE (months)	38.4 (14.7)	43.6 (13.9)	-0.94	.35
MSEL RL/EL T Score (average)	33.6 (11.4)	57.3 (8.8)	-6.10	<.001***
MSEL RL/EL AE (average; months)	29.6 (14.0)	42.1 (13.0)	-2.40	.02*
ADOS-2 Comparison Score	6.8 (2.8)	—	—

Mean (SD); ASD: Autism spectrum disorder; TD: typical development; MSEL: Mullen Scales of Early Learning; VR: Visual Reception (nonverbal problem solving); RL/EL: Receptive Language/Expressive Language; AE: age equivalence (months); ADOS-2: Autism Diagnostic Observation Schedule.

t-scores have M = 50, SD = 10.

Significance: *p < .05. **p < .01. ***p < .001.

All children with ASD were diagnosed prior to study enrollment by licensed clinicians with research reliability on the Autism Diagnostic Observation Schedule (ADOS-2; Lord et al., 2012). Twelve children with ASD and all children with TD ² were administered the MSEL, a standardized developmental assessment for infants and children ages birth to 68 months (Mullen, 1995). The MSEL measures skills in five domains: gross motor, fine motor, visual reception (nonverbal problem solving), receptive language, and expressive language. MSEL Language Composite (averages of MSEL Receptive Language and Expressive Language scores) was used as measures of children’s language functioning.

The university Institutional Review Board approved the study protocol. Written, informed consent was obtained by a parent or guardian of the children. Children above the age of three years and with appropriate language level provided verbal assent.

Procedure

Book sharing activity

The book sharing activity was completed as part of a larger task battery and was administered following other study activities; thus, all children and caregivers were familiar with the assessment room for the task. The caregiver–child dyad sat across from each other at a table in an assessment room at the university research center. Caregivers were given two bags of books and were verbally instructed to share books in the first bag for 5 min and then, following hearing a knock on the door, to share books in the second bag for 5 min. Caregivers were instructed to have only one book out on the table at a time but were not informed about the content of the books in each bag in advance. For video purposes (due to movement analyses), caregivers were instructed to try to keep their children and themselves on their respective sides of the table and to avoid touching their children when possible. When caregivers opened each bag, they saw an information sheet, which instructed them to “read and share” (picture book bag) or to “sing and share” (song book bag) and also provided reminders about the activity length and positioning at the table. The first bag contained five wordless picture books (e.g., Good Dog Carl [Day, 2009]; Mine! [Crum, 2012]), and the second bag contained five song books (e.g., Wheels On The Bus [Raffi, 1988]; You Are My Sunshine [Davis, 2011]). Due to concerns about losing children’s attention and transitioning if the song book activity were completed first, the picture book activity was completed first for all participants. Following completion of the task, caregivers were asked if they knew any of the books or songs. Caregivers reported that they were not familiar with the exact picture or song books but that they knew the associated songs for the song books they shared. This was also evidenced through observation of the caregivers singing the songs in the song book condition as caregivers could not sing songs they did not know.

The book sharing interaction was recorded using two video cameras recording at 30 Hz. The primary camera angle captured the profile view of both the caregiver and child, and a secondary camera angle captured the frontal view of the child.

Visual attention coding (VAC)

Caregivers’ and children’s visual attention (gaze) during the book sharing activity were manually coded from videos using a five-second, hierarchical partial interval coding schema adapted from Klimenko (2007). Coding began when the title of the book was said or the book was opened, whichever came first. Each five-second interval was coded as either Book sharing or Transition (changing between books); VAC means were calculated only from the Book sharing intervals. On average, 4.70 min (SD = 0.27) were coded for the picture book condition and 4.37 min (SD = 0.44) were coded for the song book condition for VAC scores. The codes for visual attentional behavior for the child ranged from 0 (looking away from book and caregiver) to 4 (looking at caregiver for at least two seconds, i.e., sustained attention to caregiver) for each 5-s interval. For caregivers, the codes ranged from 0 (looking away from book and child) to 5 (looking at child for at least two seconds, i.e., sustained attention to child). The coding scheme takes into account that during a triadic interaction, it is appropriate for an individual to direct their attention toward both their partner and the shared object, and also considers whether the visual attention is for a short moment (glance) or if it is sustained. The different coding schema (and ranges) for child and caregiver capture differences in ways young children and adult caregivers engage in this interactive task, reflecting caregivers’ more advanced and varied approaches to coordinating gaze and shifting attention during book sharing. See Table 2 for a full description of the coding schema. Mean scores for each book condition were calculated for each child and caregiver. Coding occurred using the primary camera angle (capturing profile view of caregiver and child). If a child’s gaze was unclear, coders could defer to the secondary camera angle. Videos were coded using ProcoderDV™ (Tapp & Walden, 1993) and were divided among three trained Research Analysts. A second coder co-coded 17% of the videos for reliability (child visual attention ICC = 0.91; caregiver visual attention ICC = 0.92).

OPEN IN VIEWER

Table 2.

Visual Attention Coding (VAC) Schema (adapted from Klimenko, 2007).

Child’s attentional behaviors		Caregiver’s attentional behaviors
Score	Description	Score	Description
0	Looking away from book or caregiver	0	Looking away from book and child
1	Glancing at the book one or more times (less than 2 s)	1	Exhibiting a functional look to comment on or direct the child’s behavior (e.g., to sit and/or to attend to book reading)—not to communicate about the book
2	Looking at the book (for at least 2 s or more; i.e., sustained attention to book)	2	Looking at the book (i.e., sustained attention to book)
3	Glancing at the caregiver one or more times (less than 2 s)	3	Glancing at the child once (for less than 2 s)
4	Looking at the caregiver (for at least 2 s or more; i.e., sustained attention to caregiver)	4	Glancing at the child more than once (each less than 2 s)
		5	Looking at the child (for at least 2 s; i.e., sustained attention to child)

Interpersonal movement coordination

Using the profile camera view of the dyad, separate regions of interest (ROIs) were defined for the caregiver and child based on their positioning across from each other. In order to be used for analysis of interpersonal coordination, excerpts could not contain any ROI crossings (i.e., participants had to be on their respective sides of the table) or automatic camera refocusing. For these reasons, 1-min excerpts of each participant’s picture book and song book activities were extracted for analysis starting at least 15 s after the book sharing began. The first 1-min excerpt that met these criteria was extracted. If no section with a 1-min duration met the criteria existed, the next longest possible excerpt (at least 30 s) was extracted. Video excerpts only included time when the dyad was continuously sharing a book; they did not include any time when the dyad was transitioning between books. The excerpts for movement analyses averaged 58.1 s (SD = 6.8, range = 31.8–60). Of the 58 excerpts, 50 reached the full 1-min duration.

To analyze coordinated movements within the dyad, we used time series analysis methods that have previously been used to examine coordination of speech and movement in musical and nonmusical tasks (e.g., Schultz et al., 2016, 2021; Schultz & Palmer, 2019). We extracted the magnitude of change between consecutive frames within the two ROIs containing only the caregiver and only the child (see Ramseyer, 2020). Videos were converted to grayscale then filtered in the temporal domain using a bandpass Butterworth filter between 1 and 14 Hz to remove artifacts relating to frame changes, interlacing, and slow drifts. The magnitude of change between consecutive frames was extracted for each ROI. Each time series was range normalized and subjected to time series analysis. We used dynamic time warping on the two time series to temporally align movement patterns within a window of 5 s. Data were pre-whitened to remove autoregressive components following Dean and Dunsmuir (2016). This conservative approach reduces the risk of spurious significant cross-correlation coefficients, although the observed coefficients may underestimate movement coordination. The cross-correlation coefficient at lag 0 was calculated to measure the strength of the relationship between the movement of the caregiver and child. To allow cross-correlation coefficients to be compared between conditions, these values were Fisher transformed prior to statistical analysis.

Data analysis

Separate linear mixed-effects models (LMEM) were used to analyze children’s mean visual attention, caregivers’ mean visual attention, and interpersonal movement coordination (Fisher transformed cross-correlation coefficients) during the book sharing activities. All LMEMs contained the fixed factors Book (2; Picture, Song; within-subjects) and Group (2; ASD, TD; between subjects), and the random factor Dyad where Book was nested within Dyad and allowed unequal variance between ASD and TD groups. Effect sizes were measured using generalized eta squared (η² _G ) for repeated-measures effects, where 0.02 = small, 0.13 = medium, and 0.26 = large (Bakeman, 2005). Spearman correlations (r_s) were computed to assess children’s and caregiver’s mean visual attention across the picture and song book sharing activities. Spearman correlations were also used to characterize relationships between children’s visual attention scores and MSEL Language age equivalents. Correlation coefficients were interpreted using Cohen’s guidelines where 0.10 = small, 0.30 = medium, and 0.50 = large (Cohen, 2013).

Videos were analyzed using custom-made scripts in MATLAB. Analyses were conducted using R Software (R Core Team, 2019). Correlations were performed using the cor function, LMEMs were performed using the lmer function from the lme4 library (Bates et al., 2015), and within-subject effect sizes were obtained using the aov_car function from the afex library (Singmann et al., 2020). Time series analysis and dynamic time warping were performed using the TSA (Chan et al., 2020; Cryer & Chan, 2008) and dtw (Giorgino, 2009) libraries, respectively.

Results

Child visual attention

For children’s visual attention, there were significant main effects of Book, F(1, 27) = 8.68, p = .007, η² _G  = 0.05 and Group, F(1, 27) = 4.34, p = .047, η² _G  = 0.12, indicating small and medium effect sizes, respectively. There was no significant interaction between Book and Group, F(1, 27) = 0.48, p = .50, η² _G  = 0.003. As shown in Figure 1, children mostly visually attended to the books with slightly higher visual attention scores during the song book activity (M = 2.19, SEM = 0.059) than during the picture book activity (M = 2.07, SEM = 0.06). Children with TD (M = 2.23, SEM = 0.037) exhibited slightly higher visual attention scores than children with ASD (M = 2.01, SEM = 0.078).

OPEN IN VIEWER

Figure 1.

Child Mean Visual Attention Coding (VAC) Scores by Diagnostic Group and Book Activity.

We also assessed the relationship between children’s visual attention levels and children’s language level for each group. Individual differences in children’s visual attention were generally consistent across the picture and song book activities in children with ASD, r_s(11) = 0.68, p = .013, and children with TD, r_s(14) = 0.60, p = .014, both indicating large effect sizes. We therefore averaged children’s visual attention scores across the picture and song book activities. For children with ASD, visual attention scores were significantly correlated with their MSEL Language age equivalents, r_s(10) = 0.67, p = .018, suggesting greater visual attention to the activity/caregiver in children with ASD with higher language skills. There was no significant relationship between children with TD’s visual attention scores and language skills, r_s(13) = -0.2, p = .92.

Caregiver visual attention

For caregivers’ visual attention, there was a significant main effect of Book, F(1, 27) = 21.69, p < .001, η² _G  = 0.16, a medium effect size. As shown in Figure 2, caregivers of children with ASD and TD demonstrated greater visual attention scores, suggesting somewhat increased gaze toward their children, during song books (M = 3.73, SEM = 0.096) than during picture books (M = 3.33, SEM = 0.075). However, there was no significant main effect of Group, F(1, 27) = 0.05, p = .82, η² _G  = 0.002, or significant interaction between Book and Group, F(1, 27) = 0.19, p = .67, η² _G  = 0.002.

OPEN IN VIEWER

Figure 2.

Caregiver Mean Visual Attention Coding (VAC) Scores by Diagnostic Group and Book Activity.

Visual attention within dyads

We also examined the relationships between child and caregiver visual attention for the song and picture book activities. There was a significant correlation with a large effect size between caregiver and child mean visual attention scores in dyads with children with ASD during the song book activity, r_s(11) = 0.72, p = .007 (Figure 3(a)). This relationship was not observed during the picture book activity, r_s(11) = -0.14, p = .66. For the dyads with children with TD, caregivers’ and children’s mean visual attention scores were not significantly correlated during the song book activity, r_s(14) = 0.25, p = .35, or during the picture book activity, r_s(14) = 0.30, p = .26 (Figure 3(b)).

OPEN IN VIEWER

Figure 3.

Relationship Between Caregiver and Child Visual Attention Coding (VAC) Scores for Each Book Activity in Dyads with Children with ASD (3(a)) and with TD (3(b)).

Interpersonal movement coordination

For movement coordination, there was a significant main effect of Group, F(1, 54) = 8.05, p = .006, η² _G  = 0.13 (medium effect size), but no significant main effect of Book, F(1, 54) = 0.78, p = .38, η² _G  = 0.01, or interaction between Group and Book, F(1, 54) = 0.08, p = .78, η² _G  = 0.001. As shown in Figure 4, while overall magnitude of coordination was moderate, movement within TD dyads (mean r = 0.37, SEM = 0.017) was more coupled than within ASD dyads (mean r = 0.31, SEM = 0.013).

OPEN IN VIEWER

Figure 4.

Interpersonal Movement Coordination Cross-Correlations by Diagnostic Group and Book Activity.

Discussion

The current study examined caregiver–child engagement during shared book activities across multiple levels of analysis (visual attention, movement coordination) and contexts (picture and song books). Visual attention scores were higher in children with TD than in children with ASD. While all children predominantly focused their attention on the books, children with TD spent somewhat more time engaged in the activity than children with ASD by glancing at their caregivers. This supports our second hypothesis and is consistent with evidence that children with ASD show reduced preference for social stimuli and attend to faces less than children with TD (Sasson & Touchstone, 2014). We observed no difference in caregivers’ visual attention by children’s diagnostic status (ASD, TD) during the book sharing activity. This finding is in line with previous observations showing the quality of caregivers’ book reading does not differ between caregivers of children with ASD and TD (Fleury & Hugh, 2018).

For both caregivers and children, visual attention scores were significantly greater during the song book activity than during the picture book activity, supporting our first hypothesis that musical activities create a context that scaffolds caregiver–child engagement. Overall, children’s visual attention scores in both groups suggested that children’s gaze was primarily directed toward the books. This is unsurprising given that the books were generally directly in front of the children and the task focus was on book sharing. Although the effect size was small, both children with ASD and TD showed slightly greater visual attention scores (i.e., more occasional gazing toward their caregiver and/or less looking away from the task) during the song books compared to picture books. As depicted in Figure 1, the range of visual attention scores between children with ASD and TD were similar in the song book condition while more children with ASD had lower visual attention scores than their TD peers in the picture book condition. Another recent study also reported that infants with and without ASD displayed similar amounts of gaze toward an adults’ face when the adult engaged them with song but that infants with ASD gazed less at the adult when she engaged them with speech (Macari et al., 2021). Song and music contexts may naturally support children’s engagement across a range of child functioning levels. Indeed, there is some suggestion that music therapy may be particularly appropriate for children with lower cognitive skills (Crawford et al., 2017).

Caregivers also exhibited more visual gaze toward their children during the song book activity than the picture book activity. Additionally, caregiver–child visual attention scores were correlated in the song book activity for the ASD group. Attributes of song activities, such as their familiarity and predictability, may be particularly relevant for modulating dyadic interactions in dyads with children with ASD. The slight increases in both child and parent attention for the song book context may suggest that caregivers are somewhat more available for moments of social engagement (e.g., for times when children do direct their gaze toward their caregivers) and to reinforce and support their children’s social and on-task behavior. Future studies could examine if song contexts support engagement between caregiver–child dyads with ASD by modulating both partners’ gaze behavior on a moment-to-moment basis in an interactive, contingent manner (i.e., via a transactional feedback loop), as well as examine if these effects would compound over time (e.g., with repeated sharing of the same song book).

Our findings are consistent with prior studies of increased social attention in children with TD or ASD during musical activities (e.g., Beck & Reiser, 2020; Kim et al., 2008; LaGasse, 2014; Paul et al., 2015; Simpson et al., 2013; Tunçgenç & Cohen, 2018) and also extend such findings to the behavior of caregivers engaging in these activities. Dyadic musical activities naturally involve familiarity, predictability, and emotionality, all of which may facilitate engagement (Cirelli et al., 2018; Hernandez-Ruiz, 2020; Lense & Camarata, 2020). Previous studies indicate that familiarity with specific books increases engagement in both children with ASD and TD during shared book reading (Fleury & Hugh, 2018). In the current study, although caregivers and children did not know the specific books, caregivers had to know the songs in order to sing the song books and caregivers indicated that their children knew the songs. Familiarity with the songs may have allowed caregivers to direct their gaze at their children more regularly. During the wordless picture books, caregivers may have needed to look at the pictures more to describe the pictures and narrate a story. This suggests that even in an activity with novel components (i.e., the specific books), known songs provide a context that may help structure caregivers’ and their children’s behavior. The predictable rhythmic patterns, melodies, and lyrics of the song books may help dyads track how the song unfolds over time and contribute to increased attention and behavior modulation (Hardy & LaGasse, 2013; Miller et al., 2013). Song activities may also support attention and engagement because they naturally involve predictable and salient multimodal cues (e.g., head movements, gestures) that are coordinated with the rhythm of the song (Lense & Jones, 2016; Trehub et al., 2016) and because song activities modulate emotion and arousal levels in both partners in the dyad (Cirelli et al., 2020; Cirelli & Trehub, 2020; Fancourt & Perkins, 2018; Shenfield et al., 2003). These attributes of the song books might have made them more accessible for children and caregivers.

This is the first study to examine caregiver–child body movement coordination in preschoolers during song and picture book sharing tasks. Our finding that movement of dyads with children with TD were more coordinated than dyads of children with ASD supports our second hypothesis and is in line with a growing body of research into reduced spontaneous interpersonal movement coordination in older children and adults with ASD (Fitzpatrick et al., 2017; Fournier et al., 2010) as well as reduced in-phase rocking behavior by children with ASD versus children with TD (Marsh et al., 2013). Decreased interpersonal coordination may reflect reduced attunement to the social interaction in dyads with children with ASD and may involve a combination of attentional, social, and motor processes (e.g., Fitzpatrick et al., 2017). While the overall magnitude of the movement coordination was small, this may in part reflect the free context of the activity (sharing books while sitting at a table) as prior studies of movement synchrony in children have generally used more explicit movement or imitation tasks (e.g., Fitzpatrick et al., 2017; Marsh et al., 2013). Moreover, our study used conservative estimates of coordination using pre-whitening in order to circumvent spurious correlations (following Dean & Dunsmuir, 2016).

In contrast to our findings with visual attention coding, we did not observe any differences in movement coordination between the song and picture book sharing conditions. Although we are surprised by this finding, this may be explained by multiple factors. First, caregivers and children were seated at a table for these activities, potentially limiting the range of movements across the song and picture book contexts. Second, due to potential ROI crossings, only a shorter portion (up to one minute) of the interaction was analyzed for the movement analyses versus longer periods of time for the visual attention coding. Third, while we hypothesized increased coordination in the musical context due to the opportunity to utilize pantomime gestures and move to rhythmic cues, dyads were not required to engage in song-associated gestures during the activity. This is consistent with findings that young children may not spontaneously sustain rhythmic behavior during musical activities, especially in laboratory settings (Cirelli & Trehub, 2019; Eerola et al., 2006; Endedijk et al., 2015). Nevertheless, our findings of diagnostic group differences suggest such interpersonal movement coordination metrics may be a promising approach for further analysis of dyadic caregiver–child interaction in very young children with and without ASD. These results further demonstrate the feasibility of applying video-derived movement analysis with young children (e.g., Pfeiffer et al., 2018). Future studies could examine interpersonal coordination during other types of play activities in young children, including during more natural settings (e.g., during a circle time activity on the floor or when explicitly instructed to use song-associated gestures), to further probe how context impacts interpersonal coordination in children with and without ASD.

The current study is novel in its consideration of both child and caregiver behavior in matched musical and non-musical natural interaction contexts. To our knowledge, this is the first study to compare caregiver–child engagement in a book sharing task across musical and non-musical contexts and to consider how such contexts might impact both child and caregiver behavior in dyads with children with and without ASD. However, there are several limitations to consider. Although study participation did not require any musical skills, participants responded to study recruitment materials about musical activities and therefore may have a predisposition for music. Both the ASD and TD groups were heterogeneous in age and language level, and while matched on nonverbal developmental level, the ASD group had lower language skills than the TD group. There were individual differences in children’s attention to the activities, which were related to language level of the ASD group. The small sample size may have precluded detection of significant effects; however, diagnostic differences were generally similar to those in a recent study of parent–child book sharing with a slightly larger sample size (Fleury & Hugh, 2018). Future studies could include a larger sample size including a language age-matched group to further assess individual differences across contexts. Another key limitation is that dyads completed the picture book activity prior to completing the song book activity. Future studies could counterbalance tasks to control for any order effects. Nevertheless, as one might expect engagement and attention to decrease over time (e.g., Cirelli et al., 2020), our finding of increased visual attention during the song book activity (which was administered second), is consistent with the idea that attributes of the song book activity supported caregiver and child engagement in the activity and with each other as participants were already comfortable in the assessment settings prior to completing the book sharing activity. Finally, the nature of our movement analyses required that the caregiver and child sit across from each other at a table. Future studies may explore other measures of movement coordination that would allow for more natural positioning and touch during a caregiver–child book sharing activity, such as motion capture.

We examined child and caregiver attention and interaction during a book sharing activity. We found differences between young children with TD and ASD in their visual gaze behavior and interpersonal movement coordination, as well as differences in the visual gaze behavior of both children and parents during a musical context. Although prior studies have focused on how musical activities affect behavior of children with ASD, the current study highlights the importance of considering both partners in the interaction. Future studies can further investigate mechanisms by which musical contexts incidentally support caregiver–child engagement and how such activities are incorporated into therapeutic activities for families of children with ASD. Our findings indicate that musical activities can strengthen engagement between caregivers and children with ASD and TD.