Verbal Ability, Behavior Problems, and Mother-Child Relationship Quality in Autism Spectrum Disorder

Abstract

This study examined differences in mother-child relationship quality and parent-rated child behavior problems based on child verbal status (i.e., minimally verbal versus verbal) in mothers and their adolescent and adult children with autism spectrum disorder (n=219 dyads; child M_age=25.38 years, SD=10.22). Relationship quality was assessed via parent-reported maternal burden and mother-child closeness, and coded speech samples ascertaining maternal critical and positive remarks regarding the child. Groups did not differ in relationship quality. The verbal group was more likely to display disruptive and socially inappropriate behaviors, but otherwise the groups did not differ in behavior problems. Verbal status moderated the relationship between behavior problems and negative (maternal burden, critical remarks) but not positive (mother-child closeness, positive remarks) aspects of relationship quality.

Autism spectrum disorder (ASD) is a lifelong disorder characterized by social communication deficits and restricted and repetitive patterns of behaviors and interests (American Psychiatric Association 2013). ASD not only affects the individual with the diagnosis, but also impacts the family, including the mother-child relationship (Cadman et al. 2012; Hastings et al. 2005; Smith, Greenberg, & Seltzer 2012). In fact, the bidirectional association between mother-child relationship quality and child behavior problems is a well-established phenomenon in parent-child dyads including children with ASD (Bradshaw et al. 2020; Greenberg, Seltzer, Hong, & Orsmond 2006; Hastings, Daley, Burns, & Beck 2006; Smith, Greenberg, Seltzer, & Hong 2008). These studies represent an important line of work given the heightened levels of behavior problems present in individuals with ASD (Estes et al. 2009; Shattuck et al. 2007) and the higher levels of stress and lower levels of well-being in their parents (Abbeduto et al. 2004; Estes et al. 2009). What is less well-known is the contribution of other child characteristics, such as child verbal status (i.e., minimally verbal versus verbal), which may provide new insights into factors that moderate the association between child behavior problems and mother-child relationship quality. Using a combination of parent report measures, ratings by trained interviewers, and coded speech samples, this study sought to examine 1) differences in perceptions and blinded ratings of mother-child relationship quality (i.e., maternal burden, mother-child closeness, maternal critical and positive remarks regarding the child) based on child verbal status, 2) differences in parent-reported child behavior problems based on child verbal status, and 3) whether child verbal status moderated the association between mother-child relationship quality and child behavior problems.

Minimally Verbal Individuals with ASD

Although the definition of minimally verbal has varied over time, there is a general consensus that individuals with few to no functional spoken words can be classified as minimally verbal (Bal, Katz, Bishop, & Krasileva 2016; Koegel, Bryan, Su, Vaidya, & Camarata 2020; Tager-Flusberg and Kasari 2013). Prior to a certain age, often around five years, children with ASD without spoken language are frequently referred to as preverbal rather than minimally verbal. Preverbal indicates that spoken expressive language is delayed but imminent, whereas as minimally verbal indicates that the individual is not expected to begin speaking verbally in the future (Tager-Flusberg and Kasari 2013). Once an individual with ASD reaches adolescence, it is not expected that they will progress to be verbal if they are currently minimally verbal (for a review, see Magiati, Tay, & Howlin 2014). Approximately 30% of individuals with ASD remain minimally verbal into adulthood though the percentage is higher among individuals with ASD and intellectual disability (ID; Luyster, Kadlec, Carter, & Tager-Flusberg 2008; Rose, Trembath, Keen, & Paynter 2016; Tager-Flusberg and Kasari 2013). Despite the substantial number of individuals with ASD who remain minimally verbal, the majority of research has focused on verbal individuals with ASD, and often with IQs within the typical range (Jack and Pelphrey 2017; Tager-Flusberg and Kasari 2013). Of the studies including minimally verbal individuals with ASD and ID, the majority focus on preschool or school-age children (Chang, Shih, Landa, Kaiser, & Kasari 2018; Jack and Pelphrey 2017; Tager-Flusberg and Kasari 2013; Thiemann-Bourque, Brady, McGuff, Stump, & Naylor 2016). Even fewer studies have included minimally verbal adolescents and adults with ASD (Maltman et al. 2020) who also may be at the greatest risk for poor outcomes (e.g., fewer friendships and less vocational success).

Though few studies have explicitly examined minimally verbal individuals with ASD, studies to date suggest that lower spoken language abilities may be associated with greater overall impairments and worse outcomes (Orsmond, Seltzer, Greenberg, & Krauss 2006; Shattuck et al. 2007; Tager-Flusberg and Kasari 2013; Williams et al. 2018; Woodman et al. 2016). One study found a trend where lower spoken language ability between ages four and five was associated with a less desirable profile of change in ASD symptoms and adaptive behaviors over a 10-year period in adolescence and adulthood in ASD (Woodman, Smith, Greenberg, & Mailick 2016). Other studies have found that poorer spoken language abilities negatively impacted social relationships and vocational independence (Friedman, Sterling, DaWalt, & Mailick 2019; Shattuck et al. 2007). It may be that greater speech and language impairments intensify the barriers to building relationships and independence, such as through employment opportunities. When an individual is not able to speak, they may utilize problem behaviors to communicate. The inability to talk with a parent may also negatively impact aspects of the parent-child relationship. Specifically, caregiver burden may be higher because the child with ASD cannot verbally communicate their needs, and mother-child closeness may be reduced due to the inability to have spoken conversations, although whether this is the case has not been studied.

Mother-Child Relationship Quality in ASD

Parenting an adolescent or adult with ASD presents unique challenges. The social deficits characteristic of ASD may hinder mother-child relationship quality. Mother-child relationship quality is a multidimensional construct of the parent-child experience, including the trust, respect, warmth, positivity and negativity of the relationship (Farooqi, 2014; Fielding-Gebhardt, Warren, & Brady 2020; Smith et al. 2008). Mother-child relationship quality can be assessed through a number of indicators, such as caregiver or maternal burden, mother-child closeness, and maternal positive and critical remarks about their child (Cadman et al. 2012; Fielding-Gebhardt, Warren, & Brady 2020; Hastings et al. 2006; Smith et al. 2008; Smith, Hong, Greenberg, & Mailick 2016). Given that relationship quality is inherently a broad construct, it is important to use a multi-method approach to measurement (Farooqi 2014). In studies of adolescence and adulthood, these aspects of relationship quality are commonly assessed through parent report and coded speech samples provided by the mother (Bussanich, Hartley, & Bolt 2017; Cadman et al. 2012; Fielding-Gebhardt et al. 2020; Smith et al. 2016). In younger children, relationship quality is often measured through a combination of observations from parent-child interactions and parent ratings (Foran, Fraude, Kubb, & Wamboldt 2020; Newton, Laible, Carlo, Steele, & McGinley 2014; Smith, Romski, Sevcik, Adamson, & Barker 2014). While parent-child relationship quality can represent a number of constructs (e.g., maternal warmth, criticism, caregiving burden), this study utilized two measures of maternal perceptions (i.e., maternal burden and mother-child closeness) and two blinded ratings (i.e., maternal critical and positive remarks) from coded speech samples to examine mother-child relationship quality (see the Methods for additional information regarding these measures).

Previous research found that mother-child relationship quality was negatively impacted in families of children with ASD, although a wide range of mother-child relationship quality was reported (Baker, Smith, Greenberg, Seltzer, & Taylor 2011; Hastings et al. 2005; Orsmond et al. 2006). One study on mothers of children with ASD found higher levels of parent-reported caregiver burden compared to mothers of children with other developmental disabilities (Cadman et al. 2012). In a study comparing mother-child relationship quality among mothers of adolescents and adults with Down syndrome, fragile X syndrome, and ASD, mothers of children with ASD reported lower levels of mother-child closeness and higher levels of pessimism about their child’s future (Abbeduto et al. 2004). Research on the potential impact of child verbal status on mother-child relationship quality could help to explain the range of relationship quality reported among mother-child dyads with ASD.

Behavior Problems in ASD

Individuals with ASD often have clinical levels of behavior problems (Shattuck et al. 2007; Smith, Barker, Seltzer, Abbeduto, & Greenberg 2012), which at times exceed levels of behavior problems observed in individuals with other developmental disabilities (Anderson, Maye, & Lord 2011; Estes et al. 2009). A number of factors may impact the presence and severity of behavior problems in individuals with ASD. Research examining the association between spoken language ability and child behavior problems is mixed. Studies have found no differences, more severe behaviors in less verbal children, and more severe behaviors in more verbal children (Baghdadli, Pascal, Grisi, & Aussilloux 2003; Matson, Boisjoli, & Mahan 2009; Williams et al. 2018). The limited research largely focuses on preschool and school-age children, and it is unclear if findings from young children apply to adolescents and adults. Interestingly, research has found that children with lower spoken language abilities demonstrate more frequent self-injurious behaviors but less aggression per caregiver report, suggesting the importance of examining behavior problems individually (Baghdadli et al. 2003; Matson et al. 2009). Matson et al. (2009) found that lower language abilities were associated with lower levels of parent-reported aggressive/disruptive behaviors but higher levels of stereotyped behaviors in toddlers with ASD. Another study on children and adolescents with ASD found that the verbal group demonstrated more severe externalizing behaviors, but the verbal and minimally verbal groups were similar on overall behavior severity, stereotyped behaviors, and internalizing behaviors per parent report (Williams et al. 2018). The profile of behavior problems present in minimally verbal compared to verbal adolescents and adults remains an important unanswered question.

Mother-Child Relationship Quality, Child Behavior Problems, & Child Verbal Status

Core impairments in social communication and restricted and repetitive patterns of behaviors and interests in addition to co-occurring behavior problems in individuals with ASD can lead to increased caregiver stress and lower levels of mother-child relationship quality between mothers and their children with ASD (Bradshaw et al. 2020; Greenberg et al. 2006; Hastings et al. 2005; Orsmond et al. 2006). Child behavior problems are particularly important to study given previous research suggesting that these behaviors have a larger impact on maternal stress than ASD symptom severity (Hastings et al. 2005). In one study using parent report, high levels of mother-child closeness predicted decreases in externalizing and internalizing behavior problems in sons and daughters with ASD 18 months later (Smith et al. 2008). In a study including adults with ASD, more maternal positive remarks and less maternal criticism were associated with less severe externalizing and total problem behaviors (Greenberg et al. 2012). Further, higher levels of self-reported maternal burden have been associated with more severe child behavior problems in ASD (Bradshaw et al. 2020).

Despite the clear associations between child behavior problems and mother-child relationship quality, the role of child verbal status has been relatively unexplored. Having a minimally verbal adolescent or adult with ASD may alter the association between mother-child relationship quality and child behavior problems, particularly if either the quality of the mother-child relationship and/or severity of behavior problems differs based on child verbal status. Moreover, child verbal status may moderate the association between these two variables based on the results of previous research indicating a unique profile of behavior problems present in minimally verbal individuals with ASD (Baghdadli et al. 2003; Matson et al. 2009; Williams et al. 2018). This study asked the following questions:

1. Does the relationship quality between mothers and their adolescent and adult child with ASD differ based on child verbal ability? We hypothesized a lower overall quality of the mother-child relationship in the minimally verbal group compared to the verbal group.

2. Do child behavior problems differ for minimally verbal versus verbal individuals with ASD? We hypothesized a different profile of child behavior problems in the two groups such that the minimally verbal group would be less likely to display disruptive behaviors but more likely to display unusual/repetitive behaviors.

3. Does child verbal status moderate the association between mother-child relationship quality and child behavior problems? We hypothesized that verbal status would moderate the association between mother-child relationship quality and child behavior problems such that there would be a stronger association between mother-child relationship quality and child behavior problems in the minimally verbal group due to a compounding impact of being minimally verbal and displaying behavior problems on mother-child relationship quality.

Methods

Procedure

Mothers and their adolescent and adult children with ASD who participated in this study were part of a larger longitudinal study on adolescents and adults with ASD (n = 406; Seltzer et al., 2003). Participants in the larger study were recruited through schools, clinics, agencies, and media announcements. Approximately half of the longitudinal study participants lived in Wisconsin and half lived in Massachusetts at the start of the study; recruitment and study procedures were consistent across locations. Inclusionary criteria for the larger study included the following: 1) the child was at least 10 years of age at the study start, 2) the child had a diagnosis of ASD from a medical professional, psychologist, or educational professional prior to study entry, and 3) Autism Diagnostic Interview - Revised (ADI-R; Lord et al. 1994) administration as part of the larger study confirmed the child’s ASD diagnosis (see more about the ADI-R in the Measures section below). A small subset of the larger sample (< 6%) did not meet criteria for ASD on the ADI-R but demonstrated patterns of impairments consistent with ASD, and thus were included in the larger sample.

In order to be included in the current study, the child also had to have an intellectual disability (ID), which was characteristic of approximately 70% of the larger study. Determination of ID status included multiple sources of information. Trained examiners administered the Wide Range Intelligence Test (Glutting, Adams, & Sheslow 2000) and the Vineland Screener (Sparrow, Carter, & Cicchetti 1993). Participants with standard scores of 70 or below on both measures were classified as having ID and those with scores above 75 were classified as not having ID. When individuals had missing data or scored between 71 and 75 on one or both measures, a clinical consensus process was followed. Specifically, available records were reviewed (standardized assessments; parent report of prior diagnoses, intellectual functioning, and adaptive behavior; and clinical and school records) and three psychologists independently rated these cases. In instances where all three individual ratings did not align, the psychologists discussed until 100% consensus was reached. Data from the current study were collected at the second time point of the larger study.

Participants

Based on the inclusionary criteria outlined above, the current study included 219 adolescent and adult children with ASD and their mothers. Participants were grouped based on child verbal status (minimally verbal, n = 76; verbal, n = 143). See Table 1 for descriptive statistics of the two groups and combined sample.

Table 1.

Participant Characteristics

Variable	Verbal (n = 143)	Minimally Verbal (n = 76)	Full Sample (n = 219)	p-value
Child Variables
aChild age (years)	M = 25.63, SD = 10.44	M = 24.91, SD = 9.83	M = 25.38, SD = 10.22	.622
bChild sex (male)	n = 106, 74.1%	n = 55, 72.4%	n = 161, 73.5%	.779
bResidential status (co-residing)	n = 81, 56.6%	n = 36, 47.4%	n = 117, 53.4%	.190
Mother Variables
aMaternal age	M = 54.75, SD = 10.75	M = 54.40, SD = 11.37	M = 54.63, SD = 10.95	.825
a,cMaternal education	M = 4.49, SD = 1.95	M = 3.59, SD = 1.93	M = 4.18, SD = 1.98	.001**
bMarital status (married)	n = 98, 70.5%	n = 53, 69.7%	n = 151, 70.2%	.906
bMaternal ethnicity (white)	n = 136, 95.1%	n = 66, 86.8%	n = 202, 92.2%	.030*

Note. Group assignment based on ADI-R Overall Level of Language, Current.

^aIndependent samples t-tests used to examine differences between verbal and minimally verbal groups.

^bChi-squared tests used to examine whether demographic characteristics were more strongly associated with verbal or minimally verbal groups.

^cMaternal education was measured as follows: 0 = less than 8th grade, 1 = 8^th – 11^th grade, 2 = high school graduate or GED, 3 = 1 – 3 years of college but no associate degree, 4 = associate degree, 5 = bachelor’s degree, 6 = post-bachelor’s education, but no graduate degree, 7 = graduate degree.

^*p < .050

^**p < .010.

Full sample.

The adolescents and adults ranged from 11 to 53 years (M = 25.38 years, SD = 10.22 years) at the point of data collection for the current study. The majority were male (n = 161, 73.5%) and approximately half of the sample co-resided with their mother (n = 117, 53.4%). Their mothers ranged in age from 33 to 83 years (M = 54.63, SD = 10.95). The majority of mothers had completed at least some college (n = 155, 70.8%), although educational attainment ranged from having less than a high school degree to obtaining a graduate degree.

Group comparisons.

All adolescents and adults with ASD in the current sample had ID in order to account for previously identified differences in behavior problems and parent-child relationship quality associated with ID status (Al-Krenawi, Graham, & Al Gharaibeh 2011; Shattuck et al. 2007; Smith, Greenberg, Seltzer, & Hong 2008). On average, mothers of verbal children had higher levels of maternal education compared to the mothers of minimally verbal children, t(216) = −3.27, p = .001. Mothers of verbal children were also more likely to be white compared to mothers of minimally verbal children, X²(1) = 4.73, p = .030. See Table 1.

Measures

Child verbal status.

During home visits, a research reliable examiner administered an abbreviated form of the ADI-R to all mothers, which included all algorithm items. The ADI-R is a semi-structured parent interview in which a trained interviewer assesses the target child’s current and previous autism symptoms. Higher scores are indicative of more severe symptoms. The ADI-R is considered one component of the gold standard for ASD diagnosis, and demonstrates good reliability and validity (Hill et al. 2001; Lord et al. 1994). We utilized one item from the ADI-R (item #33) to classify each child’s current verbal status and group the mother-child dyads. In order to rate this item, a standard procedure as part of the ADI-R protocol was utilized. Specifically, the examiner asked the mother to describe her child’s typical spoken language and communication at the time of the interview, and examples were requested. Specific probes were used to obtain additional information. Probes included questions regarding the length of utterances the child typically used (single words, short phrases, or full sentences), whether the phrases included a verb, whether others could understand the child, and for those with limited spoken language, how many words the child used on a daily and monthly basis. Once again, specific examples were requested in relation to each question. When sufficient information was obtained, the examiner rated the child’s overall level of functional spoken language on a three-point scale based on the following definitions:

0 = Functional communication used on a daily basis that involves phrases of three or more words that at least sometimes include a verb and are comprehensible to others.

1 = No functional use of three-word phrases in spontaneous, echoed, or stereotyped speech, but uses speech on a daily basis with at least five different words in the last month.

2 = Fewer than five words total or speech not used on a daily basis.

If the individual received a score of “0,” they were classified as verbal. If the individual received a score of “1” or “2,” they were classified as minimally verbal. Classifying verbal status using the ADI-R is consistent with previous work (Baghdadli et al. 2003; Bal et al. 2016; Maltman et al. 2020). It is important to note that this item of the ADI-R probes verbal status in terms of speaking ability and does not account for augmentative or alternative forms of communication.

Child behavior problems.

Child behavior problems were measured using the Problem Behavior subscale of the Scales of Independent Behavior – Revised (SIB-R; Bruininks, Woodcock, Weatherman, & Hill 1996). The Problem Behavior subscale probes eight types of problem behaviors, including (1) hurtful to self, (2) unusual or repetitive, (3) withdrawn or inattentive, (4) destructive to property, (5) hurtful to others, (6) disruptive, (7) socially inappropriate, and (8) uncooperative. Mothers report whether their child has displayed each type of problem behavior in the last six months and if so, they rate the frequency (1–5, where 1 = less than once a month and 5 = one or more times an hour) and severity (1–5, where 1 = not serious and 5 = extremely serious) of the behavior. A general maladaptive behavior score is computed using a standardized algorithm (Bruininks et al. 1996) where higher scores are indicative of more frequent and severe behavior problems. The SIB-R adjusts for child age, and values range from 90 to 150, with scores at or above 110 considered above the clinical cutoff for significant behavior problems. Previously established test-retest reliability, inter-rater reliability, and construct validity of the Problem Behavior subscale of the SIB-R were good to high (Bruininks et al. 1996). Using parent report to measure behavior problems is consistent with previous research and the SIB-R has been found to be sensitive to change into adolescence and adulthood (Hauser, Kover, & Abbeduto 2014; Smith, Hong, Greenberg, & Mailick 2016; Usher, DaWalt, Hong, Greenberg, & Mailick 2020; Woodman, Smith, Greenberg, & Mailick 2015).

Mother-Child Relationship Quality.

This study utilized a multi-method approach to examine both positive and negative aspects of mother-child relationship quality through parent report measures (i.e., parent perceptions) and coding of parental speech samples. Maternal burden and maternal critical remarks represented negative aspects of mother-child relationship quality while mother-child closeness and maternal positive remarks measured positive aspects of the mother-child relationship (Bengtson and Schrader 1982; Magaña et al. 1986; Zarit, Reever, & Bach-Peterson 1980). Maternal burden and mother-child closeness captured the mothers’ perceptions of relationship quality and maternal positive and critical remarks utilized expert coding to characterize relationship quality. Significant positive correlations between the two positive measures (maternal positive remarks and mother-child closeness) as well as the two negative measures (maternal burden and maternal critical remarks) of mother-child relationship quality provided construct validity for conceptualizing positive and negative mother-child relationship quality through these dimensions (see Table 2).

Table 2.

Pearson Correlations Among Study Variables

Variable	1	2	3	4	5	6	7	8
1.Child Age	-
2.Verbal Status	−.03	-
3.Maternal Education	−.17*	−.22**	-
4.Co-Residence	.42***	.09	−.01	-
5.SIB-R General Maladaptive Score	−.38***	.05	.21**	−.03	-
6.Mother-Child Closeness	.21**	−.03	−.14+	.17*	−.37***	-
7.Maternal Burden	−.33***	−.02	.14*	−.27***	.45***	−.51***	-
8.Maternal Positive Remarksa	−.18**	−.03	−.06	−.06	−.13+	.27***	−.08	-
9.Maternal Critical Remarksa	−.14*	−.10	−.05	−.12+	.29***	−.23**	.26***	−.13+

⁺p < .100

^*p < .050

^**p < .010

^***p < .001.

^aSpearman correlations were utilized for correlations including positive and critical remarks.

Maternal burden.

Mothers completed the Burden Interview (Zarit et al. 1980), which assesses subjective burden associated with caregiving demands. The measure includes 29 items rated on a three-point scale (0–2, where 0 = not at all true and 2 = extremely true), and higher scores are indicative of greater maternal burden. As an example, one item reads: “Because of my involvement with my son/daughter, I don’t have enough time for myself.” This measure has been used in previous work to capture parenting burden associated with caring for a child with a disability (Hartley, Barker, Seltzer, Greenberg, & Floyd 2011; Kring, Greenberg, & Seltzer 2010; Orsmond et al. 2006). Reliability and concurrent validity for this measure are strong (Vitaliano, Young, & Russo 1991). Cronbach’s alpha from the current study was .85.

Mother-child closeness.

The Bengtson Positive Affect Index was used to measure mother-child closeness (Bengtson and Schrader 1982). Each mother rated her feelings of trust, affection, understanding, fairness, and respect toward her child as well as the extent to which she perceived her child exhibited each feeling toward her. This 10-item measure utilizes a six-point scale (1–6, where 1 = not at all and 6 = extremely) with higher numbers indicative of higher levels of mother-child closeness. This measure has been used extensively in previous work and found to be a reliable measure of mother-child closeness in ASD (Bussanich, Hartley, & Bolt 2017; Orsmond et al. 2006; Woodman, Smith, Greenberg, & Mailick 2015). Cronbach’s alpha was .78 for the 10 items.

Maternal critical and positive remarks.

All mothers provided a Five Minute Speech Sample (FMSS; Magaña et al. 1986). This measure begins with a specific, standardized prompt instructing the mother to describe her relationship with her child with the disability and her thoughts and feelings about that individual for five minutes without interruption. The FMSS was recorded, transcribed, and coded to measure the mother-child relationship. Coding procedures were consistent with a previously developed coding scheme by Magaña et al. (1986), and a coder with over 30 years of experience identified both the number of critical and positive remarks. A positive remark was defined as a positive statement or praise of the child’s characteristics or behaviors. Conversely, a critical remark was defined as a negative comment regarding the child’s behaviors or characteristics. Interrater reliability for the number of positive and negative remarks was previously calculated between the primary rater and another trained rater based on the larger sample and reliability was within the acceptable range (see Baker et al. 2011; Smith et al. 2008). The FMSS was derived from the longer Camberwell Family Interview and has demonstrated good interrater reliability and concurrent validity (Beck et al. 2004; Moore and Kuipers 1999). Further, this measure has been reliably used to measure parent-child relationship quality in families of children with neurodevelopmental disorders including ASD (Beck, Daley, Hastings, & Stevenson 2004; Greenberg et al. 2012; Hastings, Daley, Burns, & Beck 2006; Smith et al. 2008; Woodman, Smith, Greenberg, & Mailick 2015).

Data Analysis Plan

Prior to analyses, we checked the assumptions for parametric analyses, and both maternal critical remarks and maternal positive remarks violated the normality assumption. Thus, all analyses including these two variables utilized nonparametric statistical analyses.

Research Question 1.

We estimated independent samples t-tests to examine whether there were group differences in mother-child closeness and maternal burden. Specifically, we ran two separate tests with mother-child closeness and maternal burden as the dependent variables, and verbal status as the independent variable in each model. We estimated Mann-Whitney U tests to investigate group differences in maternal critical remarks and maternal positive remarks. Maternal critical remarks and maternal positive remarks served as the dependent variables and verbal status served as the independent variable in each model.

Research Question 2.

To investigate differences in child behavior problems based on child verbal status, we estimated chi-squared tests separately for each of the eight child behavior problems (coded as zero-absent or one-present). To examine whether there was a difference in the overall general maladaptive behavior score by group, we utilized an independent samples t-test predicting child general maladaptive behavior score from child verbal status.

Research Question 3.

All analyses in research question 3 controlled for maternal education, mother-child co-residence (0 = co-residing, 1 = living separately), and child age. This decision was based on group differences in maternal education, descriptive differences in residential status, the wide age range present in this sample, and correlations between these variables and the dependent variables in the current study (see Table 2).

We utilized multiple linear regression models to predict mother-child closeness and maternal burden. When predicting the number of maternal positive remarks, we estimated Poisson regression models. Poisson regression was selected based on the distribution of the data and the fact that maternal positive remarks was a count variable. To predict the number of maternal critical remarks, we utilized zero-inflated Poisson regression models. We elected to use zero-inflated Poisson regression models to predict maternal critical remarks due to the excess of zeros and count nature of maternal critical remarks. Zero-inflated Poisson models allow for estimating the count data and excess zeros separately within a single model (Long 1997). In each of the zero-inflated models, we included the same variables in the zero-inflation model and the count model except in the case of the interaction term, as we did not hypothesize that an interaction between verbal status and child general maladaptive behavior score would predict excess zeros.

We utilized three-step hierarchical models to examine whether verbal status moderated the association between child general maladaptive behavior score and both positive aspects of the mother-child relationship (i.e., mother-child closeness and maternal positive remarks) as well as the negative aspects of the mother-child relationship (i.e., maternal burden and maternal critical remarks). In the first step, we examined the role of the demographic variables (i.e., maternal education, mother-child co-residence, and child age) in predicting each dependent variable (i.e., mother-child closeness, maternal positive remarks, maternal burden, and maternal critical remarks). In the second step, we added child verbal status and child general maladaptive behavior score. In the third step, we tested for moderation by adding the interaction between child verbal status and child general maladaptive behavior score. Model comparisons were used and the best fit models were retained. Child verbal status was dummy coded with zero representing the verbal group and one representing the minimally verbal group. Child general maladaptive behavior scores were mean-centered.

Results

Correlations among study variables are presented in Table 2.

Descriptive Findings

In Research Question 1, we found that the verbal and minimally verbal groups were not significantly different on any measure of mother-child relationship quality (see Table 3).

Table 3.

Parent-Child Relationship Variables by Group

Variable	Verbal M (SD)	Minimally Verbal M (SD)	Test Statistic	p-value
aMother-child closeness	46.13 (6.96)	45.67 (7.49)	.430	.668
aMaternal burden	35.25 (8.90)	34.81 (8.33)	.348	.728
bMaternal critical remarks	0.35 (0.91)	0.17 (0.48)	4861.50	.149
bMaternal positive remarks	2.01 (1.95)	1.81 (1.75)	5073.50	.617

^aIndependent samples t-tests with test statistic t were used to examine group differences in mother-child closeness and maternal burden.

^bMann-Whitney U tests with test statistic U were used to examine group differences in maternal critical remarks and maternal positive remarks.

For Research Question 2, when examining group differences in the eight child problem behaviors, verbal status was not significantly associated with the presence of the following behaviors: hurtful to self, hurtful to others, destructive to property, unusual or repetitive behavior, withdrawn or inattentive, or uncooperative. Child verbal status was significantly associated with the presence of disruptive behavior, p = .020, as well as socially inappropriate behavior, p = .009. Compared to the minimally verbal group, the verbal adolescents and adults were more likely to display both disruptive and socially inappropriate behaviors. Groups were not significantly different in terms of general maladaptive behavior scores (see Table 4).

Table 4.

Child Behavior Problems by Group

Variable	Verbal n (%)	Minimally Verbal n (%)	Test Statistic	p-value
aHurtful to self	67 (46.9%)	44 (57.9%)	2.42	.120
aHurtful to others	58 (40.6%)	29 (38.2%)	0.12	.730
aDestructive to property	52 (36.4%)	28 (36.8%)	0.01	.944
aDisruptive behavior	82 (57.3%)	31 (40.8%)	5.45	.020*
aUnusual/Repetitive behavior	114 (79.7%)	66 (86.8%)	1.72	.190
aSocially inappropriate behavior	99 (69.2%)	39 (51.3%)	6.83	.009**
aWithdrawn or inattentive	96 (67.1%)	50 (65.8%)	0.04	.841
aUncooperative	81 (56.6%)	36 (47.4%)	1.72	.190
bGeneral maladaptive behavior score	M = 113.43, SD = 9.43	M = 114.37, SD = 10.98	−0.66	.507

Note.

^aChi-squared tests with test statistic X² were used to examine whether individual behavior problems were more likely to be present in one group over the other.

^bAn independent samples t-test with test statistic t was used to examine group differences in general maladaptive behavior scores.

^*p < .050

^**p < .010.

Multivariate Findings

Research Question 3 explored whether child verbal status moderated the association between child general maladaptive behavior score and positive and negative aspects of mother-child relationship quality.

Positive aspects of mother-child relationship quality.

Step 1 of the model predicting mother-child closeness did not show a significant effect of co-residence, child age, or maternal education. In step 2, adding verbal status and child general maladaptive behavior score significantly improved the model, F (2, 189) = 11.24, p < .001. Co-residence, p = .026, and child general maladaptive behavior score, p < .001, significantly predicted mother-child closeness. Specifically, mothers of co-residing sons and daughters rated their relationship as less close than mothers of adolescent or adult children living in separate residences. Similarly, the mother-child relationship was less close when the child with ASD had more severe behavior problems. Adding the interaction between verbal status and child maladaptive behaviors in step 3 did not significantly improve the model, F (1, 188) = 0.13, p = .715; thus, the model in step 2 was retained (see Table 5).

Table 5.

Linear Regression Models Predicting Mother-Child Closeness and Maternal Burden

Variable	Estimate (S.E.)	p-value	Multiple R2	F
Mother-Child Closeness
Step 1			.06	4.20**
Intercept	44.61 (1.86)	<.001***
Co-Residence	1.52 (1.11)	.173
Child Age	0.10 (0.06)	.078
Maternal Education	−0.42 (0.26)	.108
Step 2			.16	7.31***
Intercept	46.62 (1.91)	<.001***
Co-Residence	2.41 (1.07)	.026*
Child Age	−0.02 (0.06)	.793
Maternal Education	−0.25 (0.26)	.336
Verbal Status	−0.90 (1.03)	.383
Behavior Problems	−0.25 (0.05)	<.001***
Maternal Burden
Step 1			.14	11.37***
Intercept	40.07 (2.11)	<.001***
Co-Residence	−3.01 (1.24)	.016*
Child Age	−0.22 (0.06)	<.001***
Maternal Education	0.42 (0.29)	.146
Step 2
Intercept	37.99 (2.11)	<.001***	.28	16.02***
Co-Residence	−4.20 (1.16)	<.001***
Child Age	−0.06 (0.06)	.359
Maternal Education	0.13 (0.28)	.644
Verbal Status	−0.35 (1.13)	.754
Behavior Problems	0.36 (0.06)	<.001***
Step 3			.30	14.21***
Intercept	38.08 (2.10)	<.001***
Co-Residence	−3.73 (1.18)	.002**
Child Age	−0.06 (0.06)	.313
Maternal Education	0.10 (0.28)	.713
Verbal Status	−0.35 (1.12)	.753
Behavior Problems	0.45 (0.07)	<.001***
Verbal Status*Behavior Problems	−0.21 (0.11)	.047*

^*p < .050

^**p < .010

^***p < .001.

When predicting maternal positive remarks, step 1 of the model revealed a significant effect of child age, p = .001. On average, mothers produced fewer positive remarks when children were older. Adding verbal status and child general maladaptive behavior score in step 2 significantly improved the model, p < .001. Child age, p = .001, and child general maladaptive behavior score, p < .001, significantly predicted maternal positive remarks; on average, mothers produced more positive remarks when children displayed fewer behavior problems. In step 3, adding the interaction between verbal status and child general maladaptive behavior score did not significantly improve the model, p =.246; for that reason, the model without the interaction was retained. See Table 6 for the results of the models predicting maternal positive remarks.

Table 6.

Poisson Regression Models Predicting Maternal Positive Remarks

Variable	Estimate (S.E.)	p-value	aPseudo R2	Residual Deviance
Step 1			.02	406.13***
Intercept	1.29 (0.18)	<.001***
Co-Residence	0.02 (0.11)	.818
Child Age	−0.02 (0.01)	.001**
Maternal Education	−0.04 (0.03)	.097
Step 2			.04	390.03***
Intercept	1.48 (0.19)	<.001***
Co-Residence	0.10 (0.11)	.360
Child Age	−0.03 (0.01)	.001**
Maternal Education	−0.03 (0.03)	.238
Verbal Status	−0.15 (0.11)	.162
Behavior Problems	−0.02 (0.01)	<.001***

^aMcFadden’s Pseudo R².

^*p < .050

^**p < .010

^***p < .001.

Negative aspects of mother-child relationship quality.

When predicting maternal burden, the first step of the model revealed a significant effect of co-residence, p = .016, and child age, p < .001; across groups, mothers reported less maternal burden when children were older and were not co-residing. In step 2, adding verbal status and child general maladaptive behavior score significantly improved the model, F (2, 204) = 20.17, p <.001. Co-residence remained a significant predictor of maternal burden, p < .001, and child general maladaptive behavior score was also positively associated with maternal burden, p < .001. Maternal burden was higher for mothers of children who displayed more severe behavior problems. In step 3, adding the interaction between verbal status and child general maladaptive behavior score significantly improved the model, F (1, 203) = 4.00, p = .047 (see Table 5). As visually depicted in Figure 1, the strength of the association between maternal burden and child behavior problems was stronger for mothers of verbal adolescents and adults than for mothers of minimally verbal adolescents and adults.

Figure 1.

Predicted marginal means plotted for maternal burden by group at the mean level of general maladaptive scores and at one standard deviation above and below the mean. The dashed vertical line represents the SIB-R clinical cutoff score.

When predicting maternal critical remarks, the zero-inflation model in step 1 revealed a significant effect of child age, p = .026, and maternal education, p = .047, indicating that there was an increased probability of excess zeros in maternal critical remarks for younger children and mothers with lower educational attainment. In addition, in the count model, child age, p < .001, and maternal education, p = .009, were significant predictors of maternal critical remarks. On average, mothers produced more critical remarks when they attained lower levels of education and when their children were older. Adding verbal status and child general maladaptive behavior score in step 2 significantly improved the model, p = .006. Regarding the zero-inflation model, child general maladaptive behavior score was a significant predictor of excess zeros, p = .027, indicating that there was an increased probability of excess zeros in maternal critical remarks at low levels of child behavior problems. Maternal education, p < .001, and verbal status, p = .015, significantly predicted maternal critical remarks in the count model. On average, mothers of verbal adolescents and adults produced more critical remarks compared to mothers of minimally verbal adolescents and adults. In step 3, adding the interaction between verbal status and child general maladaptive behavior score significantly improved the model, p = .029. As visually depicted in Figure 2, at low levels of child behavior problems, the number of maternal critical remarks was similar between groups; however, at higher levels of child problem behaviors, mothers of verbal adolescents and adults produced more critical remarks on average compared to mothers of minimally verbal adolescents and adults. (see Table 7).

Figure 2.

Predicted marginal means plotted for maternal critical remarks by group at the mean level of general maladaptive scores and at one standard deviation above and below the mean. The dashed vertical line represents the SIB-R clinical cutoff score.

Table 7.

Zero-Inflated Poisson Regression Models Predicting Maternal Critical Remarks

Variable	Estimate (S.E.)	p-value	aPseudo R2	Log-likelihood
Step 1
Count Model			.09	−130.4***
Intercept	3.68 (0.87)	<.001***
Co-Residence	−0.28 (0.37)	.463
Child Age	−0.15 (0.03)	<.001***
Maternal Education	−0.23 (0.09)	.009**
Zero-Inflation Model
Intercept	11.83 (4.51)	.009**
Co-Residence	0.10 (0.96)	.918
Child Age	−0.54 (0.24)	.026*
Maternal Education	−0.44 (0.22)	.047*
Step 2
Count Model			.14	−123.1***
Intercept	1.74 (1.01)	.086
Co-Residence	0.05 (0.39)	.899
Child Age	−0.02 (0.04)	.523
Maternal Education	−0.36 (0.09)	<.001***
Verbal Status	−1.18 (0.48)	.015*
Behavior Problems	0.01 (0.02)	.412
Zero-Inflation Model
Intercept	1.40 (1.93)	.468
Co-Residence	1.32 (1.13)	.241
Child Age	0.00 (0.07)	.941
Maternal Education	−0.46 (0.26)	.083
Verbal Status	−0.72 (1.22)	.559
Behavior Problems	−0.17 (0.08)	.027*
Step 3
Count Model			.16	−120.7***
Intercept	1.27 (1.04)	.219
Co-Residence	0.25 (0.39)	.525
Child Age	−0.01 (0.03)	.727
Maternal Education	−0.36 (0.09)	<.001***
Verbal Status	−0.36 (0.52)	.488
Behavior Problems	0.03 (0.02)	.083
Verbal Status*Behavior Problems	−0.09 (0.04)	.029*
Zero-Inflation Model
Intercept	0.68 (1.93)	.724
Co-Residence	1.68 (1.08)	.120
Child Age	0.02 (0.06)	.685
Maternal Education	−0.49 (0.25)	.051
Verbal Status	0.26 (0.95)	.782
Behavior Problems	−0.19 (0.07)	.006**

^aMcFadden’s Pseudo R².

^*p < .050

^**p < .010

^***p < .001.

Discussion

This study provides new insights into the nuanced association between parent report of child behavior problems and aspects of mother-child relationship quality (measured using both parent questionnaires and coding of speech samples) for a subgroup of individuals with ASD. Research to date has largely focused on verbal individuals with ASD, and studies on minimally verbal individuals have primarily included preschool and school-age children. Studying minimally verbal adolescents and adults with ASD is crucial in order to understand and support individuals with ASD and their parents across the lifespan. In the present study, the associations between child behavior problems and positive aspects of mother-child relationship quality (i.e., mother-child closeness and maternal positive remarks) were similar regardless of child verbal status. However, child verbal status moderated the association between child behavior problems and negative aspects of mother-child relationship quality (i.e., maternal burden and maternal critical remarks) such that high levels of child behavior problems appeared to erode the mother-child relationship only when the child was verbal.

Mother-Child Relationship Quality & Child Behavior Problems

Lower verbal abilities have been associated with slower trajectories of change, more significant intellectual disability, and poorer daily living skills (Luyster et al. 2008; Orsmond et al. 2006; Shattuck et al. 2007; Woodman et al. 2016). Despite these challenges, we found that mothers of minimally verbal individuals maintained levels of relationship quality comparable to mothers of verbal adolescents and adults. The lack of differences based on child verbal status largely aligns with findings from Orsmond et al. (2006) which included individuals with ASD with and without ID and found that child language ability was not a significant predictor of mother-child closeness, maternal criticism, or maternal warmth. This finding may also reflect the fact that a lack of speaking ability is not equivalent to the absence of communication more generally; minimally verbal adolescents and adults often communicate through other means, such as through gestures, sign language, or speech-generating devices (Maljaars, Noens, Jansen, Scholte, & van Berckelaer-Onnes 2011).

Of note, when controlling for child behavior problems, groups were different in maternal critical remarks; mothers of verbal children were more likely to produce critical remarks compared to mothers of minimally verbal adolescents and adults. Previous research found that maternal critical remarks were intimately linked to child behavior problems (Baker et al. 2011; Greenberg et al. 2006; Hastings et al. 2006). While the groups were not different on overall behavior problem severity, we found that the verbal group was more likely to display socially inappropriate and disruptive behaviors. The specific types of behaviors characteristic of the verbal group may have contributed to modest increases in maternal critical remarks by mothers of verbal individuals. Mothers of verbal adolescents may need professional support in order to decrease criticism as past research has demonstrated that change in maternal criticism predicts later child behavior problems in adolescents and adults with ASD (Baker et al. 2011).

Moderating Effects of Child Verbal Status

Among the most important findings of this study was a significant interaction between child verbal status and behavior problems predicting maternal critical remarks; probing further, groups were similar in maternal critical remarks at low levels of behavior problems, but differences grew at higher levels of child behavior problems (see Figure 2 for reference). It is worth noting that predicted differences indicated approximately one critical remark difference within the five minute speech sample between groups at high levels of maternal critical remarks. However, previous research suggests that maternal criticism is linked to child behavior problems in ASD (Baker et al. 2011; Greenberg et al. 2006), suggesting that any presence of maternal critical remarks may be clinically meaningful in this population.

Contrary to our hypothesis, there was a stronger association between maternal burden and child behavior problems for the verbal group than for the minimally verbal group. These findings indicate that mother-child relationship quality may not be affected so much by impairment, but rather by interpersonal disruption that arises from the presence of behavior problems. The stronger link between child behavior problems and maternal critical remarks in the verbal group may reflect the specific types of behaviors present in the verbal group. Specifically, verbal adolescents and adults in this sample were more likely to display socially inappropriate and disruptive behaviors compared to the minimally verbal group. Socially inappropriate and disruptive behaviors may be particularly difficult for mothers to deal with and lead to increases in criticism from mothers. This argument aligns with findings from Hastings et al. (2006) which found that maternal criticism and maternal distress related to parenting a child with ID were associated specifically with child externalizing behavior problems. Of the two behavior problems more frequently present in the verbal group was disruptive behaviors, which is considered an externalizing behavior problem. The increased presence of socially inappropriate behaviors in the verbal group may also contribute to the study findings. Specifically, literature on marital interactions revealed that when one parent was negative in a marital interaction, the other parent was also more likely to be negative (Goetz, Rodriguez, & Hartley 2019). In the context of the present study, there may be a similar process whereby socially inappropriate behaviors, which were defined as “any socially offensive behaviors such as talking too loudly, swearing, touching others too often, or belching,” may lead to increases in maternal critical remarks, which in turn may increase socially inappropriate behaviors. Thus, the different association between negative aspects of parent-child relationship quality and child behavior problems in the two groups may be a reflection of the specific behavioral profiles differentiating verbal versus minimally verbal adolescents and adults with ASD.

The lack of moderating effects between child behavior problems and child verbal status on positive aspects of mother-child relationship quality suggests that there is a robust association between child behavior problems and both parent-child closeness and maternal positive remarks that is not influenced by child verbal status. Future research should examine whether interventions that address maternal positivity and child behavior problems show similar results among mothers and verbal versus minimally verbal individuals with ASD.

Study Implications

This study offers important new insights into the well-established association between child behavior problems and mother-child relationship quality during adolescence and adulthood. Child verbal status acts as a moderator between child behavior problems and negative aspects of mother-child relationship quality. Notably, there were no overall group differences in mother-child relationship quality, suggesting that the measures of relationship quality used in this study were not negatively impacted by child verbal status. However, high levels of child behavior problems appeared to erode the mother-child relationship between mothers and verbal children. Conversely, mother-child relationship quality was less impacted by high levels of behavior problems in dyads including minimally verbal adolescents and adults. These results are encouraging for families of minimally verbal adolescents and adults with ASD and may suggest a need to specifically target disruptive and socially inappropriate behaviors in verbal adolescents and adults with ASD.

Limitations and Future Directions

Although there were a number of strengths to the current study, there were also limitations. The study was cross-sectional; examining how the parent-child relationship unfolds over time and whether the trajectories differ based on child verbal status represent important future directions for longitudinal studies. Mother-child relationship quality was measured through speech samples and parent report, and child behavior problems were measured through parent report. While these measures are consistent with the literature in families of adolescents and adults with ASD, future studies should continue to refine measures of relationship quality and child behavior in this population. Specifically, when possible and especially during earlier stages of the life course, researchers should consider obtaining observations of parent-child interactions in addition to parental perspectives gathered through questionnaires or by coding parental speech about their relationship with their child. Future research should also explore novel measures that probe relationship quality from the perspective of the child; development of these measures should utilize novel approaches that take into account the subgroup of adolescent and adult children with minimal speaking abilities.

The current study used one item in a parent interview to measure speaking ability. Obtaining observational measures of verbal ability and characterizing expressive language through all modalities (e.g., sign language, speech-generating devices, etc.) may provide more insight into differences in child behaviors and parent-child relationship quality across expressive language profiles. This is particularly important given that not all individuals with ASD communicate through spoken means and not all speech reflects functional communication. Furthermore, including individuals with a range of verbal abilities and intellectual functioning such as those without intellectual disability represents an important next step. While the sample size was relatively large, the mothers were primarily well-educated and white; it remains important to recruit and include families from a range of ethnic and socioeconomic backgrounds. The lack of fathers and other family members or caregivers in the current sample also represents a limitation and necessary future direction.

Completing a functional analysis (Iwata et al. 1994; Lerman et al. 2005) of both behavior problems and verbal behaviors simultaneously in the context of parent-child interactions also represents an important future direction. This information may help researchers, clinicians, and parents mitigate problem behaviors, support functional communication, and improve mother-child relationship quality. Evidence-based interventions for adolescents and adults with ASD are limited, although previous research indicates positive results of multi-family group interventions to support both individuals with ASD and their parents (DaWalt, Greenberg, & Mailick 2018). However, to date, there are no evidence-based interventions specifically targeting families of minimally verbal adolescents and adults with ASD, and in fact, those most severely affected with ASD have largely been left out of intervention studies (Stedman, Taylor, Erard, Peura, & Siegel 2019). Family-based interventions that can support parents and their minimally verbal adolescents and adults with ASD represent a critical future direction, particularly since parents may need more support when they have more severely affected adolescent and adult children with ASD.

Conclusions

Minimally verbal adolescents and adults with ASD represent an understudied group. While they face a number of challenges compared to verbal peers, mothers of minimally verbal individuals report similar levels of relationship quality and similar overall levels of child behavior problems, although mothers report that minimally verbal individuals are less likely to demonstrate disruptive and socially inappropriate behaviors. Moreover, heightened behavior problems may have less adverse effects on negative aspects of mother-child relationship quality when the child is minimally verbal compared to verbal. This study provides new insights into the mother-child relationship in ASD and highlights the importance of considering child verbal status in the context of individual behavioral profiles as well as family experiences.