Social interaction links active musical rhythm engagement and expressive communication in autistic toddlers

Abstract

Rhythm is implicated in both social and linguistic development. Rhythm perception and production skills are also key vulnerabilities in neurodevelopmental disorders such as autism which impact social communication. However, direct links between musical rhythm engagement and expressive communication in autism is not clearly evident. This absence of a direct connection between rhythm and expressive communication indicates that the mechanism of action between rhythm and expressive communication may recruit other cognitive or developmental factors. We hypothesized that social interactions, including general interpersonal relationships and interactive music-making involving children and caregivers, were a significant such factor, particularly in autism. To test this, we collected data from parents of autistic and nonautistic children 14–36 months of age, including parent reports of their children’s rhythmic musical engagement, general social skills, parent-child musical interactions, and expressive communication skills. Path analysis revealed a system of independent, indirect pathways from rhythmic musical engagement to expressive communication via social skills and parent-child musical interactions in autistic toddlers. Such a system implies both that social and musical interactions represent crucial links between rhythm and language and that different kinds of social interactions play parallel, independent roles linking rhythmic musical engagement with expressive communication skills.

Lay summary:

Musical rhythm has been linked to multiple social and communication abilities, and atypical rhythm abilities are implicated in autism. We found that for autistic toddlers, rhythm is connected to expressive communication skills through both general social skills and parent-child musical interactions. This pattern of relationships has implications for both how rhythm and language are linked in autism and the development of novel intervention approaches building on music.

Introduction

Emerging scholarship increasingly links music, social, and communication skills across behavioral, genetic, and neurocognitive domains (Nayak et al., 2022). As music is a common interest and sometimes skill in autistic individuals (Dickie et al., 2009; Steinberg, Shivers, et al., 2021), there is substantial interest in harnessing music engagement, or the observable behavioral instantiation of an underlying capacity for participating in musical activities, as a motivating and strengths-based approach to support social and communication skills in this population (Cheever et al., 2018; Geretsegger et al., 2022; LaGasse, 2014, 2017; Lense & Camarata, 2020; Thompson et al., 2019). Yet despite the ubiquity of music engagement in autism (Steinberg, Liu, et al., 2021), little research has considered individual differences in musical and social communication behaviors in young autistic children. Identifying potential paths among musical experiences, social skills, and communication skills early in development in autism is a necessary first step for establishing potential intervention ingredients and targets, as well as laying the foundation for future work aimed at identifying for whom music-based interventions may be appropriate (Cheever et al., 2018; Lense & Camarata, 2020).

Musical engagement may be directly and indirectly related to social and communication skills in both autistic and nonautistic populations. Communication and language development occurs within a social context both in neurotypical and autistic development (Carpenter et al., 1998; Hoff, 2006; Mundy et al., 2007; Tomasello, 1992; Tomasello & Farrar, 1986). In autism, differences in the quantity and quality of social interaction (Clements et al., 2018; Kasari et al., 2010; Locke et al., 2010; Petrina et al., 2014) have cascading effects on the later development of language and communication skills (Kasari et al., 2008; Mundy & Bullen, 2022; Thurm et al., 2007). For example, individual differences in early social motivation, as well as early social communication skills such as reciprocal imitation and coordinated joint attention are linked to later expressive and receptive language and communication skills in autism (Adamson et al., 2009; Bottema-Beutel, 2016; Bruinsma et al., 2004).

Outside autism, music represents a common avenue for social interactions. Early childhood musical experiences, particularly those involving active participation such as joint caregiver-child musical play (Cirelli, Jurewicz, et al., 2020; Politimou et al., 2018) or caregiver-child music classes (Fancourt & Perkins, 2018; Lense et al., 2020; Perkins et al., 2018), are highly social interactions present in most young children’s lives (Ilari, 2016; Lamont, 2008). Among nonautistic children ranging from young toddlers and preschoolers to school-aged individuals, social musical engagement promotes children’s prosocial behaviors (Beck & Rieser, 2022; Cirelli et al., 2014; Cirelli, Peiris, et al., 2020; Rabinowitch & Meltzoff, 2017), social attention (Kirschner & Tomasello, 2010), empathy (Rabinowitch et al., 2013), and perceived similarity with others (Rabinowitch & Knafo-Noam, 2015), as well as caregiver’s positive social behaviors like smiling (Trehub et al., 2016) and gaze toward their child (Liu et al., 2022). Increased musical engagement between young children and caregivers is also associated with increased child communication and language skills in typical development (Franco et al., 2022; Papadimitriou et al., 2021; Politimou et al., 2019).

Child-Caregiver Musical Engagement in Autism

Shared musical engagement may also support social interactions in autism. As with nonautistic children, music and song capture and maintain autistic children’s social attention more effectively than speech from infancy through the school-aged period (Macari et al., 2021; Paul et al., 2015; Thompson & Abel, 2018). Autistic school-aged children and adolescents exhibit increased neural activation and connectivity for song versus speech (Lai et al., 2012; Sharda et al., 2015). Some small-scale empirical studies suggest music intervention may be a vehicle for supporting social interaction skills such as increased incidental eye contact and turn-taking durations during musical versus non-musical play in young autistic children (Carpente, 2017; Kim et al., 2008; Thompson et al., 2014). Music-based interventions in autistic school-aged children are also linked to several communication outcomes such as the use of structural language and pragmatics (Sharda et al., 2018). However, not all studies report positive effects of musical interventions on social and communication skills outside of intervention sessions (i.e. transfer of skills; Simpson & Keen, 2011), which may in part relate to measurement issues as well as heterogeneity within autism (Bieleninik et al., 2017; Broder-Fingert et al., 2017; Sharda et al., 2019), and highlights the need to examine individual differences in musical and social communication skills.

Musical experiences can also promote caregivers’ social engagement with their autistic children. For example, during a book sharing activity, caregivers looked at their preschool-aged children more during song versus picture book sharing, suggesting song supports caregivers’ availability for moments of social engagement; parent and child visual attention was associated only in the song book condition (Liu et al., 2022). Musical play also supports caregivers’ use of nonverbal physical responsiveness with their autistic preschoolers (e.g. imitating musical instrument play; Boorom et al., 2020); these types of mirroring or contingent imitation behaviors are associated with children’s development of joint attention and language skills (Gulsrud et al., 2016; Ingersoll & Schreibman, 2006). Shared parent-child musical activities promote social bonding in families of autistic toddlers and preschoolers as they do in the nonautistic population (Lense et al., 2020; Lense, Liu, et al., 2022; Steinberg, Liu, et al., 2021; Thompson et al., 2014), and parents frequently report using music to practice communication skills or for social interaction games with their young autistic children, though there is substantial variability across families (Steinberg, Shivers, et al., 2021). However, despite this potential platform of parent-child musical interaction for supporting child communication skills, relationships between children’s musical engagement and their parents’ use of and involvement of their children in musical activities have yet to be examined in autism.

Rhythm’s Relevance to Social Communication

Musical engagement requires the capacity to perceive and comprehend musical stimuli in combination with the motivation to act on these capacities. This musical capacity is comprised of a set of interrelated cognitive and biological mechanisms (Honing et al., 2015). When considering specific components of music engagement that may link to social and communication behaviors, there is increasing attention to the role of rhythm processing. Rhythm perception, including the ability to perceive a steady, underlying beat, emerges very early in typical development, as evidenced both by behavioral (Ilari, 2015; Zentner & Eerola, 2010) and neural measures (Edalati et al., 2023; Háden et al., 2015; Honing et al., 2009; Winkler et al., 2009). Rhythm skills are associated with social connection (Marsh et al., 2009), interpersonal coordination (Keller et al., 2014; Lang et al., 2016), and individuals’ capacities for linguistic communication (Nayak et al., 2022; Nitin et al., 2023) in nonautistic children and adults.

These correlations are bolstered by rhythm’s mechanistic role in several forms of social and communicative behaviors as studied in nonautistic individuals. Beat perception, a fundamental component of rhythm perception, recruits a network of brain regions involving not only auditory processing but also motor skills and reward processing (Cannon & Patel, 2021; Fiveash et al., 2023; Patel & Iversen, 2014). Beginning in infancy, predictable rhythms enable synchronization of infant-caregiver eye gaze (Lense, Shultz, et al., 2022) and increase infants’ neural sensitivity to variations in auditory patterns to support learning (Suppanen et al., 2019). Rhythm processing is implicated in neurotypical adults’ ability to perceive and entrain to timing patterns in naturalistic conversation (Wynn et al., 2022), to synchronize social gaze behaviors like mutual eye contact (Wohltjen & Wheatley, 2021), and to coordinate many other physical actions (Levitin et al., 2018). Such synchrony and coordination between nonautistic individuals leads, in turn, to increased interpersonal rapport (Miles et al., 2009) and mechanistically support general prosocial behaviors and attitudes (Michael et al., 2020). In addition, rhythm is central to large-scale entrainment of groups (Launay et al., 2016; Merker et al., 2009), which drives group cohesion (Cross, 2001). Across individual and group levels, the evidence from nonautistic populations provides a strong rationale for a causal theory whereby rhythm contributes to social and communication behaviors.

Such a causal approach is evident in recent work arguing that rhythm-related skills are a vulnerability in many neurodevelopmental disorders characterized by challenges with social, communication, and language skills (Ladányi et al., 2020; Lense et al., 2021). In autism, difficulties in the rhythm and timing of interpersonal skills have long been observed across the lifespan (Lord et al., 2020) such as the coordination of eye contact, vocalizations, or smiling, as well as the synchronization of speech or movement behavior (Liu et al., 2022; Ochi et al., 2019; Wan et al., 2013; Zwaigenbaum et al., 2005). Autistic children and adolescents exhibit reduced interpersonal synchrony on rhythmic movement tasks such as reduced phase alignment when rocking in a rocking chair (preschoolers) or swinging a pendulum (adolescents) with another person (Fitzpatrick et al., 2016; Marsh et al., 2009, 2013), with degree of synchrony associated with specific social skills during the school-aged period (Fitzpatrick et al., 2017; Romero et al., 2018). Among autistic adults, there is some evidence for links between nonsocial rhythm synchronization (e.g., tapping to the beat of an auditory stimulus) and social skills (Vishne et al., 2021).

The Current Study: Musical/Rhythm Engagement, Social Engagement, and Communication in Autism

Given the ubiquity of musical activities and their social impacts for children and parents, clarifying the connections across young autistic children’s and their parents’ musical engagement and children’s social and communication skills will be informative for identifying potential intervention processes, targets, and candidates. In the current study, we first characterize relationships between child and parent everyday home musical behaviors during early childhood across autistic and nonautistic development. Next, we test the hypothesis that autistic children’s musical engagement, with an emphasis on rhythm as an early emerging and foundational component for coordinating social interactions (D’Ausilio et al., 2015; Lense & Camarata, 2020), is pivotal for supporting their social and communicative development. We propose that social engagement -- considering both children’s general interpersonal skills and specific interactive musical participation -- critically link autistic children’s musical/rhythm engagement and their communicative development. Consistent with developmental cascade frameworks in which skills in one domain (e.g., social/musical interaction) have downstream effects on another domain (e.g., communication), (Bradshaw et al., 2022), we call this mechanism the Sociomusical Mediation of Child Communicative Development in Autism (henceforth Sociomusical Mediation; see Figure 1). A summary of the key components of this model, and the measures used to operationalize them, can be found in Table 1.

Figure 1.

Schematic of the Sociomusical Mediation of Child Communicative Development Model

Note. The Sociomusical Mediation model demonstrates potential indirect pathways of influence among a child’s musical, particularly rhythmic, engagement (far left); their general social skills and motivation (top center); their musical interactions with their parents (bottom center); and their expressive communication skills (far right).

Table 1.

Index of Terminology

Term	Definition	Relevant Measure in Current Study
Musical ability/capacity	The ability or capacity to perceive, produce or interact actively with musical signals using any mode of production (e.g. singing, drumming, strumming). This refers to a set of biological capabilities rather than interest or motivation.
Musical engagement	Observable behavioral instantiation of an underlying capacity for participating in musical activities, including music-making, music listening, dance, etc (i.e. musicality), in conjunction with the motivation to use these capacities. This may or may not involve social interactions, requires some degree of musical capacity, and (in early childhood) involves musical stimuli containing strong rhythmicity. Musical engagement is directly observable in a naturalistic context.	Music@Home Child Active Engagement (CAE)
Parent-child music-making	Parents making music, engaging in musical play, or engaging in a musical activity with their child. This is distinct from other forms of parental musical activity by its child-directed nature and the child’s participation in the activity (i.e. “I make music WITH my child” rather than “I sing TO my child” or “I sing in a choir with other adults”).	Music@Home Parent-Child Music-Making (PCMM)
Parent initiation of singing	Parents singing to their children that does not require or rely on reciprocation, such as singing lullabies.	Music@Home Parent Initiation of Singing (PIS)
Social interactions	Activities or behaviors that involve an individual engaging with or relating to another person, and which often involves verbal/vocal and non-verbal communication.	Vineland Interpersonal Relationships (IPR)
Expressive communication	The use of vocal, verbal, and gestural skills to communicate	Vineland Expressive Communication (EC)

Methods

Participants

Participants included 74 (60m, 14f) autistic toddlers and 49 (28m, 21f) nonautisic toddlers between 14 and 36 months (Autistic: M = 28.3, SD = 5.3; Nonautistic: M = 22.9, SD = 5.6), recruited in two large metropolitan areas in the Southeastern United States (see Table 2 for sample summaries). Children who were diagnosed with a genetic syndrome (e.g., Down syndrome), non-autistic developmental disorder, or hearing impairment were excluded from the study. In the autistic group, parent respondents were 66 mothers and 8 fathers; in the nonautistic group, parents were 47 mothers and 2 fathers. Diagnoses for autistic toddlers were determined via clinical best estimate based on assessments (including the Autism Diagnostic Observation Schedule, Second Edition (ADOS-2) (Lord et al., 2012)) by licensed, research-reliable clinical psychologists. All toddlers were administered the Mullen Scales of Early Learning (MSEL; hereafter, Mullen) as a standardized assessment of their developmental (visual reception, i.e., nonverbal problem solving), fine motor, and receptive and expressive language skills (Mullen, 1995). We computed the Nonverbal Developmental quotient (NV-DQ) as the ratio of the arithmetic mean of visual reception and fine motor skills age equivalents (AE) to chronological age:

$$\text{NV-DQ}=100\times \frac{\left(V{R}_{AE}+F{M}_{AE}\right)/2}{\mathit{\text{chronological age}}}.$$

Table 2.

Children’s Demographic Information and Clinical Scores by Diagnostic Group

	Autistic	Nonautistic	Test Statistic (χ2 or t)	Significance (p)
n (M:F)	74 (60:14)	49 (28:21)	8.30	.004**
Chronological Age (months)	28.3 (5.3)	22.9 (5.6)	5.39	< .001***
Race			10.7	.01*
Caucasian	60.8%	87.8%
Black	14.9%	4.1%
Asian	1.4%	0.0%
Mixed Race	23.0%	8.2%
Ethnicity (% Not Hispanic)	81.9%	93.6%	3.33	.07
Household Income (% Over $70,000)	51.6%	97.9%	38.6	< .001***
Maternal Education			46.0	< .001***
Up to High School/GED	21%	0%
Vocational/2-Year College/Some College	35%	4%
Bachelor’s Degree	31%	29%
Graduate Degree	14%	67%
MSEL NV-DQ	72.3 (20.8)	112.5 (12.5)	13.2	< .001***
MSEL RL AE (months)	12.1 (8.0)	25.6 (7.3)	9.4	< .001***
MSEL EL AE (months)	13.8 (6.8)	25.0 (8.0)	7.9	< .001***
VABS-3 Expressive Communication v-scale score	5.6 (3.3)	15.3 (1.7)	21.3	< .001***
VABS-3 Interpersonal Relationships v-scale score	8.1 (2.3)	15.6 (1.6)	20.7	< .001***
ADOS-2 Comparison Score	8.4 (1.8)	—	—	—

Note. Mean (SD); MSEL = Mullen Scales of Early Learning; NV-DQ = Nonverbal Developmental Quotient; RL = Receptive Language; EL = Expressive Language; AE = age equivalence (months). v-Scale scores have mean = 15, standard deviation = 3. ADOS-2 Comparison Scores range from 1–10. See Supplement A for more precise breakdowns of household income (Table A1) and maternal education (Table A2).

Significance:

^***p < 0.001.

^**p < 0.01,

^*p < 0.05,

All nonautistic children had developmental and language scores no more than 1.5 standard deviations below the population mean on the Mullen. The study was approved by the university institutional review boards.

Measures

Child Communication and Social Skills

Parents completed the Vineland Adaptive Behavior Scales, Third Edition Comprehensive Interview (VABS-3; hereafter, Vineland) (Sparrow et al., 2016) to characterize children’s adaptive functioning in the Communication and Socialization domains. We focused on children’s adaptive behavior for our analyses because we wanted to consider their generalized, everyday performance of their communication and social skills (consistent with our parallel focus on children’s everyday musical experiences; see below). The Vineland Expressive Communication subdomain, which includes children’s use of speech, language, and gestures, was used to characterize children’s expressive communication abilities, while the Interpersonal Relationships subdomain, which assesses children’s inclination and ability to respond and relate to others, was used to characterize their social skills.

Child and Parent Music Engagement

We used Music@Home-Infant (henceforth, Music@Home), a validated and reliable parent-report measure of the home musical environment, to capture parent and child musical engagement and activity (Politimou et al., 2018). The Infant form was initially validated for nonautistic children up to two years of age, with another version of the Music@Home validated for nonautistic children two to four years of age (Music@Home-Preschool; Politimou et al., 2018). Though our samples’ chronological age spanned across the two form versions, the Infant version was considered the most appropriate for our sample given the developmental and language age of the autistic participants (consistent with prior work with this measure; Steinberg, Shivers, et al., 2021). As well, the Music@Home-Infant’s child-centered items focus on rhythm, while the Preschool version emphasizes musical object use.

The Music@Home scale consists of 18 items, six of which are reverse-coded, and yields an overall Music@Home score as well as scores for the measure’s four subscales: Child Active Engagement with Music (how much children respond to music, particularly through rhythmic movement; e.g., “I have noticed my child moving in time with the beat of the music”), Parent Initiation of Singing (how frequently parents sing to their children; e.g., “I sing in playful contexts to/with my child at least once a day”), Parent Initiation of Music-Making (how much parents engage their children in joint musical activity; e.g., “I make music with my child (including toy instruments) almost every day”), and Parental Beliefs (what parents think about music and development; e.g., “I believe that music is part of a well-rounded education”). Since the questions for Parent Initiation of Music-Making focus on interactive parent-child music-making (“with my child”), rather than on actions only initiated by the parent, we refer to this subscale as Parent-Child Music-Making to reflect the active involvement of parents and children and that these activities may be parent- or child-led (N. Politimou, personal communication, September 13, 2023). All items are rated on a 7-point Likert scale ranging from one (completely disagree) to seven (completely agree).

In this study, we focused on the Child Active Engagement, Parent Initiation of Singing, and Parent-Child Music-Making subscales, since they directly reflect musical actions. (The Parental Beliefs subscale does not assess either child or parent activities, so we did not analyze it in the current study.) Child Active Engagement scores can range from 6 to 42, Parent Initiation of Singing scores range from 5 to 35, and Parent-Child Music-making scores ranges from 3 to 21; thus, the total Music@Home range combining across these three subscales is from 14 to 98. The Music@Home measure was reliable in all three relevant subscales across both autistic and nonautistic samples (⍺ ranges from 0.81 to 0.87). Parents completed the Music@Home via REDCap (Research Electronic Data Capture), a secure online survey and database web application (Harris et al., 2009, 2019).

Analyses

All group differences and correlation analyses controlled for NV-DQ. To contextualize the home musical experiences of the families of autistic children, we first described and compared child and parent musical activity on the Music@Home between the nonautistic and autistic samples using linear regression with NV-DQ as a covariate. To examine relationships between child and parent musical engagement, we computed intercorrelations among Music@Home subscales within each diagnostic group and used regressions to test whether diagnosis moderated these intercorrelations.

To assess how the home musical environment was related to measures of social and communicative development in autism, we computed partial Pearson correlations between Music@Home subscales and the Vineland Expressive Communication and Interpersonal Relationships subdomain v-scores, controlling for NV-DQ. Partial correlations involving Vineland Expressive Communication were also replicated with Mullen Expressive Language for a clinician-administered measure of child expressive communication (see Supplement B). We focused on standardized scores from the Vineland inventory over age-equivalents from either the Vineland or Mullen inventories due to their focus on adaptive function relative to chronological age, which makes them an ideal index of development/delay. In addition, since the Vineland inventory captures children’s performance in everyday environments while the Mullen emphasizes performance of specific skills during a standardized clinical assessment, and since standardized Mullen t-scores tend to have limited variability particularly in autistic toddlers (e.g. Akshoomoff, 2006), we relied on the Vineland standardized indices in the main analysis.

We then conducted a structural model to test our proposed hypothesis of a pivotal role of autistic children’s musical/rhythmic engagement for supporting their social interactions and communication skills. Specifically, we computed a path model using robust maximum-likelihood estimators reflecting the proposed structure of how Music@Home Child Active Engagement is tied to Vineland Expressive Communication, including paths via both Vineland Interpersonal Relationships (i.e., general child social interaction skills) and Music@Home Parent-Child Music-Making (i.e., parent-child social musical interactions) (Figure 1). This path structure mirrors the hypothesized Sociomusical Mediation model. We also computed an analogous path model involving the weighted mean of Parent-Child Music-Making and Parent Initiation of Singing as a combined measure of parent-involved musical activity (Supplement C, Figures C1 and C2). However, since Parent-Child Music-Making more directly assesses musical interactions between parents and children (see Table 1 for a summary of this distinction), we focus on the Parent-Child Music-Making model here. Both path models were also reproduced with NV-DQ as an exogenous variable. All analyses were conducted in R (R Core Team, 2022), using the lavaan (Rosseel, 2012) and psych (Revelle, 2022) packages.

Relationships among music, social, and communication skills were only assessed in autism. This is consistent with the aims of the paper and the proposed model for Sociomusical Mediation of Child Communicative Development in Autism, which seeks to explain the heterogeneity in adaptive social and communication skills in autism (in contrast, there is limited variability of such adaptive skills in typical development).

Results

Musical Engagement in Autistic and Nonautistic Children and their Parents

Nonautistic children exhibited higher Child Active Engagement with Music (M = 37.8, SD = 3.8) than autistic children (M = 31.8, SD = 7.0) [t(116) = 3.5, p < .001], but neither of the parent-involved subscales (Parent Initiation of Singing; Parent-Child Music Making) significantly differed between the two diagnostic samples (see Table 3). The intercorrelations between the two parent-involved Music@Home subscales were largely similar in nonautistic and autistic samples, indicating similar tendencies in parents’ propensities to engage in these different types of musical behaviors regardless of their child’s diagnostic status (Table 4). However, though the correlations between the child-based (Child Active Engagement with Music) and the two parent-involved subscales were statistically significant in both the autistic and nonautistic groups, the relationships were noticeably weaker in the autistic than in the nonautistic sample, suggesting a moderating effect of child diagnosis. Indeed, there was a significant interaction between Parent Initiation of Singing and child diagnostic status when predicting Child Active Engagement (p<.01; Figure 2). Given the sex and age differences between the autistic and nonautistic groups, we repeated analyses controlling for these covariates: all patterns remained the same. Though the moderating effect of diagnosis on the relationship between Child Active Engagement and Parent-Child Music-Making increased, it did not reach conventional levels for statistical significance (without covariates, p = 0.13; with age and sex covariates, p = 0.11). As well, within the autistic group, there was substantial heterogeneity in family income and maternal education (Table 2; see also Supplement Tables A1, A2); there were no relationships between these variables and any Music@Home subscale (p’s > .8, adjusting for multiple comparisons).

Table 3.

Music@Home Subscale and Total Scores by Diagnostic Group

	Autistic	Nonautistic	t	p
Child Active Engagement	31.5 (6.8)	37.9 (3.8)	3.5	< .001***
Parent Initiation of Singing	26.7 (4.8)	27.9 (4.9)	0.7	.46
Parent-Child Music-Making	15.1 (3.8)	16.4 (3.5)	0.6	.54
Total†	73.4 (12.0)	82.2 (10.3)	2.2	0.03*

Note:

^†This Total Music@Home score reflects the subscales used in the current study as reported in the table. The full Music@Home also includes a Parental Beliefs subscale.

Significance:

^***p < .001,

^*p < .05.

Table 4.

Intercorrelations Among Music@Home Subscale Scores by Diagnostic Group

	Autistic		Nonautistic
	PIS	PCMM	PIS	PCMM
CAE	0.30*	0.39**	0.58***	0.41**
PIS		0.54***		0.51**

Note: CAE = Child Active Engagement; PIS Parent Initiation of Singing; PCMM = Parent-Child Music-Making. All correlations are partial Pearson correlations, controlling for NV-DQ, and p-values are corrected for multiple comparisons using the Holm method.

Significance:

^***p < .001,

^**p < .01,

^*p < .05.

Figure 2.

Relationship Between Child Active Engagement and Parent Musical Behaviors

Note: Diagnosis moderates relationships between Child Active Engagement and Parent Initiation of Singing (left panel; p<.01), but not between Child Active Engagement and Parent-Child Music-Making (right panel; p=.13).

Musical Engagement of Children and Parents Relate to Children’s Social and Communication Skills in Autism

We next examined relationships among child and parent music engagement with children’s social and communication skills in autism. There were significant positive correlations between Music@Home subscale scores and Vineland subdomain scores in autistic toddlers, though relations differed for specific child and parent music engagement subscales (Table 5). Child Active Engagement in Music was positively correlated with children’s social skills as measured via the Vineland Interpersonal Relationships v-scale scores (r = .31, p < .05). Neither Parent-Child Music-Making (r = .16, p = .36) nor Parent Initiation of Singing (r = −0.01, p = .94) were associated with children’s interpersonal relationship skills after correcting for multiple comparisons (using the Holm-Bonferroni method). In contrast, Parent-Child Music-Making (r = .31, p < .05), but not Child Active Engagement (r = .19, p = .30), was positively correlated with children’s Expressive Communication subdomain scores on the Vineland. Similar patterns were seen with the Mullen Expressive Language scale, which was marginally associated with Parent-Child Music Making, as well as Parent Initiation of Singing, when controlling for NV-DQ (see Supplement B). This may reflect construct differences between the two measures of expressive language and communication, though given the marginal strength of the correlation between these scales, the differences in statistical significance across these measures when applying the conventional threshold may be a spurious result. In either case, replication studies with larger samples would be necessary to arrive at a firm interpretation.

Table 5.

Correlations Between Music@Home Subscale Scores and Vineland Subdomain Scores in the Autistic Sample

	IPR VSC	EC VSC
CAE	0.31*	0.19
PIS	−0.01	0.15
PCMM	0.16	0.31*

Note: CAE = Child Active Engagement; PIS Parent Initiation of Singing; PCMM = Parent-Child Music-Making; IPR VSC = Vineland Interpersonal Relationships v-scale score, EC VSC = Vineland Expressive Communication v-scale score. All correlations are partial Pearson correlations, controlling for NV-DQ, and p-values are corrected for multiple comparisons using the Holm method.

Significance:

^*p < .05.

Independent Paths in Sociomusical Mediation System

To assess how this pattern of bivariate correlations mapped onto the Sociomusical Mediation framework proposed in Figure 1, we implemented this framework as a path model using robust maximum-likelihood estimators. The resulting model (Figure 3) was an excellent fit for the data (TLI = 1.1; SRMR = 0.01; RMSEA = 0.0 [0.0, 0.3]; sample size-adjusted BIC = 1065.6) and consists of significant indirect effects of Child Active Engagement in Music on Expressive Communication, mediated through both Parent-Child Music-Making and Interpersonal Relationships. These mediator variables did not have a significant covariance when accounting for the rest of the path model.

Figure 3.

Path Analysis of Sociomusical Mediation Model

Note. The path analysis shows independent paths of influence from children’s musical rhythm skills (left) to their expressive communication skills (right) through different kinds of social interaction. The potential covariance between children’s general social skills (top, middle) and children’s musical interactions with their parents (bottom, middle) is not significant when accounting for the full proposed model.

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

To investigate the extent to which NV-DQ impacts this model, we implemented a further path model including NV-DQ as an exogenous variable. The resulting model (Figure 4) was also a strong fit for our data (TLI = 1.1; SRMR = 0.002; RMSEA = 0.0 [0.0, 0.2]; sample size-adjusted BIC = 1527.952). While the largest single effects were for NV-DQ predicting Expressive Communication and Interpersonal Relationships, the pattern of significant, independent pathways linking Child Active Engagement in Music with Expressive Communication through both Parent-Child Music-Making and Interpersonal Relationships remained unchanged from the model without NV-DQ. Lastly, to consider a broader range of parent-involved musical behaviors toward and with their child, we implemented the path model (both with and without NV-DQ) with the weighted sum of Parent-Child Music-Making and Parent Initiation of Singing to verify structural similarity; the resulting models are structurally identical to those solely involving Parent-Child Music-Making (see Supplement C).

Figure 4.

Path Analysis of Sociomusical Mediation Model With Nonverbal Developmental Quotient

Note. This path analysis includes nonverbal developmental quotient (NV-DQ) as an exogenous variable, but is otherwise identical to the model in Figure 3, and demonstrates the same independent paths of influence from children’s musical rhythm skills to their expressive communication skills through different kinds of social interactions. NV-DQ significantly predicts both general social skills and expressive communication but does not predict either children’s musical rhythm or parent-child musical interaction.

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

Discussion

Connections across musical engagement, social engagement, and communication exist across typical and clinical populations from both individual difference and intervention perspectives (Lense et al., 2021; Lense & Camarata, 2020; Nayak et al., 2022), with particular emphases on the role of rhythm in linking across these domains (Fiveash et al., 2023; Ladányi et al., 2020; Lense et al., 2021). We report similarities and differences in children’s and parents’ everyday musical behaviors in autistic and nonautistic toddlers, including relationships between children’s and parents’ musical engagement and notable reductions in children’s active music engagement in autism. We found evidence of independent indirect pathways linking autistic children’s capacity for active musical engagement with their expressive communication skills via their broad interpersonal engagement and specific parent-child music-making experiences. Our findings implicate a system of child and caregiver social and musical behaviors in supporting children’s communication skills.

Reduced Active Engagement with Music and Relationships with Parent Singing in Autism

The current study emphasizes connections between children’s musical engagement and their social development. Autistic children exhibited reduced child active engagement in music versus nonautistic children, and autistic children’s active engagement with music was associated with their social skills as assessed via the Vineland parent interview, including when controlling for children’s non-verbal developmental level. This replicates and extends a prior survey study in somewhat older preschool-aged children that also reported reduced child active music engagement in autistic versus nonautistic children (including children with other neurodevelopmental disabilities such as Williams syndrome or Down syndrome, further suggesting findings are not due only to, for example, developmental or motor skills; Steinberg, Shivers, et al., 2021). Nonautistic (neurotypical) children’s active engagement in music (but not parent-involved musical behaviors) on the Music@Home has previously been linked with parent-child joint engagement in other activities (e.g., book reading, developmental activities) (Politimou et al., 2018). Combining results across studies, children’s capacities for and motivation to participate in musical activities, as quantified via the Music@Home Child Active Engagement subscale, may relate to a child’s readiness and motivation to engage with their caregivers.

In contrast to the reductions in autistic children’s active music engagement, parent-involved music engagement including singing and interactive musical play with their child occurred at similar levels with both autistic and nonautistic toddlers. The two parent-involved measures were highly correlated in both diagnostic groups, perhaps reflecting parents’ general tendency to musically engage with their child. However, the relationships between parent and child measures, though still significant, were somewhat attenuated in autism, particularly for parent initiation of singing. In addition, while neither parent measure was significantly correlated with children’s general social skills, Parent-Child Music-Making had a stronger relationship with general social skills and was significantly correlated with children’s expressive communication. Singing may serve different functionalities, as it is most often used for arousal regulation (Cirelli, Jurewicz, et al., 2020; Shenfield et al., 2003; Trehub et al., 2015). Alternatively, perhaps this distinction between Parent Initiation of Singing and Parent-Child Music-Making reflects that parent singing does not necessarily involve any active participation from the child, while joint music-making does. Parent-Child Music-Making may also be combining parent-led and child-led activities, as this subscale on the Music@Home measures parent-reported frequency of parent-child music-making but does not distinguish between parent-led and child-led activities. Given the importance of parent responsiveness to child-led activities for supporting social communication in autism (Gibson et al., 2021), an important direction for future studies may be to examine the impact of parent-initiated versus child-initiated shared music activities on child communication development.

Independent Causal Pathways Link Rhythm and Expressive Communication in Autism

Our path model reveals that these two forms of social interaction (generalized child interpersonal skills and parent-child music-making) independently mediated autistic children’s active music engagement and their expressive communication skills. That is, once accounting for the full model hypothesized by the Sociomusical Mediation hypothesis linking children’s musical/rhythmic engagement and their expressive communication skills through social engagement, the covariance between child general interpersonal skills and parent-child music-making remains insignificant, and the magnitude of this relationship is reduced. This suggests dynamic, transactional relationships among musical and communicative capacities of autistic children and their sociomusical ecosystems. This model is robust to the inclusion of nonverbal developmental quotient as an exogenous variable, suggesting these relationships across rhythmic/musical engagement, social and musical interaction skills, and communication skills are at least somewhat independent of children’s general developmental skills.

With its emphasis on children’s physical, rhythmic responsiveness to music, the Music@Home Child Active Engagement subscale findings provide converging streams of evidence in support of rhythmic entrainment and synchronization as a core aspect of social interaction (Keller et al., 2014; Redcay & Schilbach, 2019; Yun et al., 2012). Rhythm synchronization abilities and rhythm-based musical interventions in autism are connected to several socio-cognitive and pragmatic abilities in school-aged children and adults (Sharda et al., 2018; Vishne et al., 2021) with the current study extending the relationships between these constructs to autistic toddlers. Rhythm’s relationship with communication skills, suggested by a wide range of theoretical and empirical results (e.g. Gustavson, Nayak, et al., 2023; Nayak et al., 2022; Nitin et al., 2023), appears to be largely mediated through social interactions in autism.

Moreover, beyond the role of children’s general interpersonal engagement in supporting their expressive communication, there is a significant, independent path between children’s active musical engagement and expressive communication skills through Parent-Child Music-Making. This mediation indicates a specific role for a musical environment component, integrating our results with transactional theories of child development (e.g. Sameroff, 1975, 2010) and therapy design (e.g. Frost et al., 2020). These theories, which argue that children develop through bidirectional interactions with their environment (including parent-mediated activities), have been directly implicated in a variety of domains including language development (Blume et al., 2021), and have also been proposed as a mediating process for effects of parent-mediated music interventions (PRESS-Play framework; Lense & Camarata, 2020). Together, these results represent converging evidence for integration of musical engagement with transactional development of social and communication skills.

Translational Implications for Intervention Design and Implementation

The mechanisms revealed by our analysis have possible translational implications. Given primary relationships between children’s musical engagement and their interpersonal skills, targeting social skills may be a necessary primary step for music-based interventions that ultimately aim to support children’s communicative development. The presence of an independent pathway through Parent-Child Music-Making indicates that parent-mediated musical approaches may be especially fruitful, particularly for children with greater levels of musical/rhythmic engagement or skills (i.e., strength-based approach). These approaches are feasible, especially when integrated with existing parent-mediated models such as the Parent-Early Start Denver Model (Hernandez-Ruiz, 2018, 2020) or parent-mediated Reciprocal Imitation Training (Liu et al., 2023), and have shown potentially positive outcomes in small studies (Hernandez-Ruiz & Braden, 2021) though more research with larger sample sizes is needed.

Alternatively, better social skills may simply enhance musical activities’ ability to impact communication skills; in this case, social skills would act as a constraint on the efficacy of music-based interventions targeting expressive communication. Here, social motivation may be among the criteria making an individual a good candidate for music-based interventions. In this case, the Sociomusical Mediation model offers a possible explanation for why music-based interventions are not consistently found to have positive effects above and beyond other intervention approaches (Marquez-Garcia et al., 2022). These possibilities are not mutually exclusive, so both may be accurate to some extent. This cannot be distinguished by our non-causal data, especially since we are not directly assessing an intervention or change over time. However, our results do offer a compelling argument for considering children’s social skills in conjunction with music-based interventions aimed at improving expressive communication, whether as an intermediate outcome or a constraint on the treatment’s efficacy.

Limitations and Future Directions

Our results are derived from cross-sectional rather than longitudinal observations. To properly assess any causal relationships among musicality, social skills, and expressive communication, we would need to employ a longitudinal, cross-lagged paradigm (Lense et al., 2021). Indeed, path analyses rely on causal assumptions with respect to the variables involved, but since they are computed using correlations, they are agnostic as to the direction of each causal relationship, so each path could conceivably go the opposite direction. However, we have a strong rationale for the causal direction proposed in our model. The pattern of bivariate correlations and their mediations in the path model suggests Sociomusical Mediation is true regardless of the directionality of the path model effects, given the lack of a significant direct correlation between musical engagement and expressive communication in autism. Overall, our results can effectively reveal underlying mechanisms and provide a foundation for future causal analyses, such as a cross-lagged intervention-based assessment paradigm including medium-to-long-term follow-up observations.

The Music@Home and Vineland inventories are parent-report and interview measures, respectively, which may involve response bias (Wojcik et al., 2022). Yet, as parents are key social partners for their children, parents’ perceptions of their children’s behavior are important for continually driving parent-child social feedback loops (e.g. Warlaumont et al., 2014). However, the parent interview measure of child communication skills was highly correlated with clinician-administered, direct-observation measures (i.e., Mullen). In addition, these measures are reliable, validated indicators of everyday activities and behaviors, making them highly applicable to our core questions. Using the Vineland, a parent-report measure of adaptive function, rather than the direct-observation Mullen assessment of early learning, allowed us to focus on everyday use rather than observed behavior in an unfamiliar environment, and to investigate the link between musical engagement and everyday behavior. However, these relationships between home musical engagement and adaptive social communication skills are bolstered by similar patterns with directly assessed communication skills, as measured by the Mullen inventory (see Supplement). This implies that the relationships we observed here reflect constructs that are observable both by clinicians and caregivers, while slight disparities between the magnitude of the correlations may reflect both the role that familiarity plays in children’s behavior and the differences between clinician and caregiver observations of that behavior. Future work could probe the sources of these distinctions by incorporating multiple forms of measurement, including direct observation of everyday activity or ecological momentary assessment. Both Music@Home and the Vineland are widely used in the literature, allowing our analysis to build effectively on prior results while offering bridges to other forms of observation. Future studies could include additional measures of rhythmic entrainment and social function such as eye-tracking (e.g. Lense, Shultz, et al., 2022) and neuroimaging (e.g. Bouwer et al., 2023) or more naturalistic observations of musical activities such as via home recordings (e.g. Mendoza & Fausey, 2021).

Our hypothesized model here implies that child music rhythm engagement predicts parent-child musical interactions. However, other aspects of the broader sociomusical ecosystem may also influence how much parents make music with their children. Some such factors, such as shared genetic and environmental contributions to joint musical behaviors at this developmental stage (Gustavson, Coleman, et al., 2023; Niarchou et al., 2022), are already implicated in the feedback loops that make up the Sociomusical Mediation model, while others, such as parents’ own musical and social inclinations, are not explicitly accounted for. Incorporating these factors into the Sociomusical Mediation framework could be a fruitful avenue for future investigations, particularly in studies aimed at further elucidating the mechanisms underpinning the pathways described in this paper.

Lastly, this paper focuses on testing this model, with its emphasis on the mediating role of social interaction, in autism. Multiple studies have assessed direct connections between rhythm and specific types of language and communication skills (e.g., grammar skills) in nonautistic children (Chern et al., 2018; Gordon et al., 2015; Lee et al., 2020; Nitin et al., 2023). The model tested here in autism may not characterize the relationships between these domains outside of autism. The Vineland adaptive behavior measures of social and communicative skills exhibit reduced variability in nonautistic populations, limiting our ability to statistically test such a cognitive model. In addition, our nonautistic sample was constrained by limited diversity with respect to household income and maternal education when compared to our autistic sample, limiting our ability to differentiate effects of diagnosis from other socioeconomic factors. Investigating this multiple pathways model in a larger, more diverse nonautistic population, along with identifying robust measures of social and communicative function that are sufficiently precise to provide meaningful variability in such a population, represents a fruitful avenue for future research.

Conclusions

We found a system of independent causal pathways through which different forms of social interaction mediate the effect of musical engagement, which comprises the observable behavioral manifestation of underlying musical capacities such as rhythm perception and entrainment skills and the motivation to act on those capacities, on expressive communication in autism. These results extend prior theories and empirical models to autism and to a younger developmental period than previously studied. In addition, our focus on musical activity extends these theories into more naturalistic, ecologically-valid behaviors. This model of independent sociomusical mediation pathways can be assessed in future longitudinal work, including within the development of naturalistic behavioral interventions.

Furthermore, our results reflect a potential role of music in developing social relationships, a principle which has been central to the recent social bonding hypothesis of the evolutionary provenance of musicality (Savage et al., 2021). This theory posits that musicality, or the set of capacities enabling humans to engage in musical behaviors, co-evolved alongside the development of human society to support humans’ ability to form and strengthen interpersonal bonds. Our findings here linking children’s musical engagement with both musical interactions and broader interpersonal skills, which in turn connect with communication skills, complement evolutionary perspectives and emphasize music’s social and interactive nature that manifests in a variety of contexts and across neurodiverse populations (Kasdan et al., 2021).

Data Availability Statement

The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.