This study explored the relationship between Turkish mothers' style of interaction and the engagement of their preschool-aged children with autism. Data were collected from fifty mother–child dyads in which all children had diagnoses of autism. Video recordings of mother–child interaction were analyzed using the Turkish versions of the Maternal Behavior Rating Scale and the Child Behavior Rating Scale (O. Diken, 2009). Similar to mothers from Western countries, Turkish mothers tended to engage in highly directive interactions with their children. However, a cluster analysis revealed considerable variability in mothers' style of interaction. This included a directive nonengaged style, a directive/achievement-oriented style, and a responsive style of interaction. Children's level of engagement was associated with differences in mothers' style of interaction. Children were least engaged with directive/nonengaged mothers and most engaged with responsive mothers. However, children's engagement was only associated with their mothers' responsiveness, not with their directiveness. Results are discussed in terms of their implications for early intervention.
For more than 30 years there has been interest in understanding how mothers and other caregivers interact with young children with autism (Mahoney & Nam, 2011). This research has been stimulated partly by concerns that the difficulties children with autism have in engaging in reciprocal interaction might interfere with the ability of their parents and other interactive partners to provide the types of stimulation necessary to maximize the children's developmental potential (see, e.g., Adamson, Bakeman, & Deckner, 2004; Siller & Sigman, 2008).
Early investigations compared patterns of interaction between caregivers and preschool-aged children with autism to those between caregivers and developmentally matched samples of typically developing children and/or children with other types of disabilities. In general, findings from these studies have indicated few differences in the way parents interact with children with autism, with the exception that the parents are more directive. For example, Kasari, Sigman, Mundy, and Yirmiya (1988) compared interactions between mothers and their preschool aged children with autism with mother–child dyads in which children either had intellectual disabilities or were typically developing. Caregivers of children with autism were similar to other caregivers in their level of responsiveness to their children. However, caregivers of both children with autism and children with intellectual disabilities were more directive than caregivers of typically developing children. Similar findings were reported by Watson (1998), who compared the degree to which mothers of children with autism and mothers of typically developing children used language that was supportive of their children's focus of attention. While there were no group differences in mothers' responses to their children's focus of attention, mothers of children with autism produced 85% more utterances that directed their children to attend to objects or toys not related to their focus of attention.
Findings from these studies are consistent with studies that have compared interactions between parents and their children with disabilities other than autism to those between parents and their developmentally matched samples of typically developing children (Cunningham, Reuler, Blackwell, & Deck, 1981; Peterson & Sherrod, 1982; Rondal, 1978). Most of these studies reported that parents of children with disabilities engage in similar patterns of interaction as do parents of typically developing children, particularly as indicated by the quality and complexity of their language (Peterson & Sherrod, 1982; Rondal, 1978) and level of responsiveness (Adamson et al., 2004). Yet the one parenting feature that consistently differentiates interactions with children with disabilities is elevated directiveness. Many have interpreted this finding as indicating that parents accommodate the social-interactive deficits of children with disabilities by using directive behaviors to offset some of the children's social inadequacies, such as low levels of interaction, failure to engage in meaningful play or communication (Adamson et al., 2004; Watson, 1998), and lack of responsiveness to adults (Kasari et al., 1988; Watson, 1998). High parental directiveness has been described as a logical, and perhaps necessary, interactive strategy to help children engage in the types of social, communicative, and play activities that are necessary for developmental learning (Doussard-Roosevelt Joe, Bazhenova, & Porges, 2003; Landry, Taylor, Guttentag, & Smith, 2008; Spiker, Boyce, & Boyce, 2002).
Recent investigations have focused less on determining how parents and their children with autism differ from other groups of dyads and more on how variability in parents' style of interaction is associated with the behavior and development of their children. Results from these studies have indicated considerable variability in the manner that parents interact with children with autism, particularly along dimensions of responsiveness, affect, and directiveness. Results indicate that generalizations from comparative studies do not accurately reflect the complex patterns of interaction that transpire within individual parent–child dyads (Ruble, McDuffie, King, & Lorenz, 2008; Siller & Sigman, 2002, 2008).
In addition, studies examining the association of parents' style of interaction with children's social-communicative behavior have provided little support for the assumption that directive parenting enhances social interaction. Adamson et al. (2004) assessed the association of parents' use of directives, including both behavioral- and attention-regulation strategies, with children's engagement. While parents of children with autism used a higher frequency of behavioral-regulation strategies than parents of typically developing children, directiveness was not associated with the quality of children's social interaction with their parents. Ruble et al. (2008) assessed the interactive behavior of parents of preschool-aged children along six global dimensions. As in the work of Adamson et al. (2004), parental directiveness was not associated with the level of children's social initiation, while parental responsiveness was.
Perhaps due to the increasing evidence that parental responsiveness is positively associated with the development and social-emotional functioning of a wide range of children (Mahoney & Nam, 2011), studies investigating the associations of parents' interaction with the development of children with autism have focused on responsiveness. At least three studies have indicated that variations in the rate at which parents respond to their children are associated with children's communication development. Siller and Sigman (2002, 2008) reported two longitudinal studies of the language development of children with autism. Results from both studies showed robust associations between the frequency of mothers' contingent responsiveness when their children were 4 years old to the children's subsequent language development at 6 (Siller & Sigman, 2008), 14, and 20 years of age (Siller & Sigman, 2002). McDuffie and Yoder (2010) reported that the frequency of both parents' responses to their children's current focus of attention and to their verbal communication predicted increases in vocabulary for children with autism six months later.
In the ecological model of child development, Bronfenbrenner (1979) postulated that parents' interactions with their children are influenced not only by children's social characteristics and the nature of their disabilities but also by prevailing cultural, political, religious, and child-rearing beliefs of the community and family. The majority of research on parent–child interaction involving children with disabilities has been conducted in the United States and Western Europe. As a result, findings from this literature may be partly influenced by the common cultural, religious, and child-rearing beliefs of parents in these countries and may not be fully representative of the manner that parents interact with children in non-Western countries. This study will investigate patterns of interaction between Turkish mothers and their young children with autism and examine how variations in mothers' style of interaction are associated with their children's interactive engagement.
Anecdotal reports suggest that Turkish parents tend to be highly directive and authoritarian with their children (Yaman & Mesman, 2010). These qualities are thought to be partially associated with Turkey's being a “collectivist” culture, in which interdependence is emphasized and the expression of children's own wants and needs is inhibited while attention to the needs of others is encouraged (Markus & Kitayama, 1991). In order to achieve these outcomes, parents have been reported to be more authoritarian, using more restraining behaviors during social play and expecting more obedience (Kağıtçıbaşı, 2007). In addition, other accounts suggest that less than 50% of mothers play with their children, which is attributed to Turkish mothers' having limited formal education including training in child development (Bekman, 2004; Kağıtçıbaşı, 1989).
We are unaware of any studies that have examined interactions between Turkish mothers and children with autism. However, a few studies have examined interactions between Turkish mothers and children with other disabilities. Results from these studies have been consistent with findings from studies with Western parents. Turkish mothers tend to be highly dominant and directive (Ceber-Bakkaloğlu & Sucuoğlu, 2000; I. H. Diken, 2009, 2012), as well as less responsive (Ceber-Bakkaloğlu & Sucuoğlu, 2000) than parents of typically developing children. In addition, these reports indicate that children with disabilities are less actively involved in interaction, respond less to their parents, and display higher levels of inappropriate behavior than typically developing children (Ceber-Bakkaloğlu & Sucuoğlu, 2000). Yet how children's interactive behavior is associated with their mothers' style of interaction has yet to be investigated.
This study is a descriptive analysis of the patterns of interaction among a large sample of Turkish mothers and their preschool-aged children with autism. The purpose is to explore the relationship between Turkish mothers' style of interaction and the interactive engagement of their children. The following questions were addressed: (1) How do mothers of children with autism interact with their children; and (2) is the interactive behavior of children with autism associated with their mother's style of interaction? Consistent with findings reported from Western countries, we hypothesized that although mothers would tend to be highly directive, there would be considerable variability in the manner in which they interacted with their children. We also hypothesized that children's participation in interaction would be associated with their mothers' style of interaction. In particular, we expected that children's interactive engagement would be positively associated with their mothers' responsiveness and negatively associated with their directiveness.
Participants were 50 mothers and their children with autism. Thirty-five children were male and 15 were female. The age of the children ranged from 24 to 72 months, with a mean of 54.9 months (SD = 11.7). Children came from various educational institutions including state-affiliated special-education schools, private special-education and rehabilitation centers, and university-based centers in the city of Eskisehir in Turkey. All had received medical diagnoses of autism from psychiatrists or neurologists that were based upon the DSM-IV or ICD-10 criteria. The mothers' age ranged from 25 to 45 years, with a mean of 34.5 years (SD = 5.30). Eighteen had a university degree, 8 had a high-school degree, 18 had completed elementary school, and 6 were nonliterate. All of the mothers were married. They signed a letter of consent to participate after the purpose and procedures of the study were explained.
Procedures used for observing mother–child interaction were similar to those that have been used in several published studies (Mahoney, Finger, & Powell, 1985; Masur & Turner, 2001; Siller & Sigman, 2002). Mothers and their children were video recorded for 20 min while they played with their children with a set of developmentally appropriate toys. The first 5 min were excluded from the analyses to give mothers and children some time to adjust to the camera and setting.
Data were collected in one of the classrooms of the centers where children received special-education services. All mothers were familiar with the context, since they had had the opportunity to observe their children receiving services in these classrooms. Classrooms were carpeted and furnished with a table and two chairs. Developmentally appropriate toys consisted of picture books, crayons, puzzles, cars, blocks, dolls, plastic animals, stacking rings, a phone, and kitchen articles. Mothers were instructed to play with their children as they normally did using only the toys provided.
Two observational scales were used to assess mothers' interactive style with their children and children's engagement in the interaction with their mothers.
Turkish version of the Maternal Behavior Rating Scale (TV-MBRS)
Mothers' style of interaction was assessed by TV-MBRS (O. Diken, 2009). The Maternal Behavior Rating Scale (MBRS) was developed by Mahoney (2008) to assess mothers' interactional behavior with their children. It consists of 12 items: responsivity, sensitivity, effectiveness, acceptance, enjoyment, expressiveness, inventiveness, warmth, achievement, praise, directiveness, and pace. Items are rated on a 5-point Likert scale with ratings of 1 reflecting a low incidence of the quality being assessed and ratings of 5 indicating a high incidence. O. Diken (2009) translated the MBRS and investigated its validity and reliability with Turkish mothers of children with various delays and disabilities. The Turkish version yielded three factors: responsivity (sensitivity, responsivity, effectiveness, and inventiveness), affect (acceptance, enjoyment, expressiveness, warmth, and praise), and directiveness (achievement, directiveness and pace). These factors accounted for 73.4% of the variance of this scale. The Kaiser-Meyer-Olkin measure was .83. Cronbach's α was .87 for the responsivity subscale, .86 for affect, and .61 for directiveness.
Turkish version of the Child Behavior Rating Scale (TV-CBRS)
Children's interactive engagement with their mothers was assessed with the TV-CBRS (O. Diken, 2009). The Child Behavior Rating Scale (CBRS) was developed by Mahoney and Wheeden (1999) to assess the interactive engagement of preschool-aged children with disabilities. It consists of seven global items: attention, persistence, interest, cooperation, initiation, joint attention, and affect. O. Diken (2009) translated the CBRS and investigated its validity and reliability with Turkish children with various developmental disabilities. The TV-CBRS was found to have two factors: attention (interest, attention, persistence, and cooperation) and initiation (affect, joint attention, and initiation). These factors explained 63.10% of the total variance. The Kaiser-Meyer-Olkin measure was .82. Cronbach's α was .91 for initiation and .79 for attention.
Video-recorded observations were coded by two independent coders, first with the TV-MBRS and second with the TV-CBRS. Coders completed approximately 30 hr of training, until interrater agreement of 80% was attained on each of the two scales. Reliability was computed based on interrater agreement for 25% of the observations used for this analysis. Interrater reliability was computed using the formula
for both the TV-CBRS and TV-MBRS (Richards, Taylor, Ramasamy, & Richards, 1999). Overall exact agreement was 90%, ranging from 85% to 95% for both the MBRS (.90 for the responsivity subscale, .95 for the affect subscale, and .85 for the directiveness subscale) and the CBRS (.90 for the initiation subscale and .90 for the attention subscale).
Mothers' Interactive Style
Table 1 reports the mean and range for each of the items and scales on the TV-MBRS. The highest-rated items were associated with the directiveness subscale, with the directiveness item having the highest rating of all 12 items. Ratings for most of the items in the responsiveness subscale had average scores that were slightly below the midpoint. Ratings for the affect subscale also clustered around the midpoint, with the exception of praise, which was more than 0.5 points below the midpoint. In general, these data depict the interactive style of this sample of mothers as being in the average range for responsiveness and affect but well above the midpoint for directiveness.
The minimum and maximum scores for TV-MBRS ratings were distributed across all 5 Likert ratings for 11 of the 12 items and 4 of the 5 ratings for the directiveness item, indicating substantial variability in the mothers' style of interacting with their children. Correlations examining the association of mothers' age and education as well as children's age and gender to the three TV-MBRS factors were only significant for relationships between responsiveness and mother's education (r = .27, p < .05) and responsiveness and child's gender (r = .28, p < .05), indicating that mothers were more responsive with girls than with boys.
Children's Interactive Engagement
Table 2 reports the mean and range for each of the TV-CBRS items and scales. The mean ratings for six of the seven items were within 0.2 point of the midpoint, indicating that children displayed average levels of interactive engagement. The one exception was joint attention, which had an average rating 0.4 point below the midpoint. Yet there was considerable variability in the level of children's interactive engagement, as reflected by the relatively high standard deviations (MSD = 1.36) and the fact that the full range of ratings were used for all items. Correlational analyses indicated that children's TV-CBRS subscale scores were not significantly associated with children's age or gender.
Maternal Interactive Style Clusters
SPSS version 20 was used to classify mothers in terms of the 12 MBRS variables for interactive style. A hierarchical cluster analysis suggested a four-cluster solution. However, a k-means cluster analysis using the squared Euclidian distance procedure with four clusters yielded three clear clusters and a fourth that included only 8% of the mothers. However, a three-cluster solution yielded a clear profile of dimensions of maternal interactive style. The means (cluster centers) for the MBRS items are shown in Table 3. ANOVAs indicated that significantly more variation between clusters than within clusters for all variables.
Cluster 1 included 38% of the sample (n = 19). Mothers in this group displayed high levels of directiveness, but very low ratings on responsiveness and affect items. This cluster was identified as “Low affect/responsive.” Cluster 2 included 24% of the sample (n = 12); average ratings for most directiveness items were considerably above the midpoint, while ratings on responsiveness and affect items were in the average range. This cluster was identified as “Directive.” The third cluster included 38% SD of the sample (n = 19). Ratings for most responsiveness items in this group were well above the average range, while affect and directiveness items were in the average range. This cluster was identified as “Responsive.”
Relationship Between Maternal Interactive-Style Clusters With Children's Interactive Engagement
To validate the clusters, a multivariate analysis of variance (MANOVA) was computed to determine whether children's interactive engagement as assessed by the two TV-CBRS subscales was related to the three maternal clusters. Results are shown in Table 4. There was a significant effect of maternal cluster on children's interactive engagement (p < .01). Univariate analyses indicated that this effect was significant for both children's attention (p < .01) and initiation (p < .05). As indicated in Figure 1, the mean TV-CBRS ratings demonstrated an increase in children's interactive engagement from the Low affect/responsive cluster to the Directive and Responsive clusters. Yet Bonferroni t tests indicated that differences in children's TV-CBRS scores were only significant between the Low affect/responsive and Responsive clusters.
Association of Mothers' Style of Interaction to Children's Engagement
A multiple regression was computed to examine the association of TV-MBRS subscales to children's total engagement. Affect was not included in this analysis because its high correlation with responsiveness (r = .75) raised the likelihood of collinearity. Results indicated that mothers' interactive style accounted for 19% of the variability in children's engagement. Directiveness had a negative, nonsignificant association (β = −.08, T = 0.62, p = .550), while responsivity had a positive, significant association with children's total engagement (β = .45, T = 3.36, p = .002).
Summary of Findings
Findings from this descriptive investigation of Turkish mothers' interactions with their young children with autism were supportive of most of the hypotheses for this study. As reported in numerous studies, the most dominant interactive characteristic of these mothers was their directiveness with their children. Directiveness was the TV-MBRS item that received the highest rating, with 64% of the mothers receiving ratings that were above the midpoint.
Yet despite the tendency of these mothers to be highly directive, there was considerable variability in the patterns of interaction they displayed with their children. Cluster analysis indicated that mothers could be grouped into three distinct patterns of interaction. The Low affect/responsive cluster was characterized not only by mothers using moderately high levels of directiveness, but perhaps more notably by extremely low levels of responsiveness and affect. The low ratings these mothers attained on effectiveness indicate that they had difficulty engaging their children, despite the fact that their interactive pace ratings suggested they exerted moderate level efforts to interact with them.
Mothers in the Directive cluster displayed even higher levels of directiveness than those in the previous group. This group also displayed average levels of responsivity and above-average ratings in effectiveness and pace, indicating that they were moderately successful at engaging their children. Their moderately high ratings in achievement suggest that these mothers were directive not only to engage their children but also to encourage their children to produce higher-level, or more appropriate, play and communicative behaviors.
Mothers in the Responsive cluster displayed moderately high levels of responsiveness and only average levels of directiveness. In addition, their effectiveness ratings indicate that they were more successful than the other two groups at engaging their children. While these mothers had ratings in achievement that were slightly above the midpoint, their ratings on the affect subscale suggest that they were as focused on enjoying their children as they were on promoting higher-level skills or behaviors.
Association of Mothers' Interaction to Children's Engagement
The major focus of the parent–child research literature is on determining whether the manner in which parents interact with children with autism impacts their children's social functioning and developmental learning. To investigate this issue, this study focused on the association between mothers' style of interaction and their children's engagement. This was predicated on the assumptions that early developmental learning occurs in the context of parent–child interaction and that children benefit from these experiences to the degree to which they are active participants.
Results supported our hypothesis that children's engagement was positively associated with their mothers' responsiveness. While these results do not refute the possibility that this resulted from more-engaged children eliciting higher levels of responsiveness from their mothers, they are nonetheless consistent with results from the descriptive study reported by Ruble et al. (2008) as well as findings from developmental interventions which reported that increases in parental responsiveness were causally related to increases in children's engagement for children with general disabilities including autism (Kim & Mahoney, 2005; Mahoney & Perales, 2005).
Although we had hypothesized a negative relationship between parental directiveness and children's engagement, results indicated that these behaviors are not associated with each other. This finding may be attributable to the fact that directiveness played at least two different functions for the mothers in this study. On the one hand, some mothers appeared to use directives to engage their children. This was most evident for Low affect/responsive mothers whose children displayed extremely low levels of interactive engagement. These mothers' low ratings on effectiveness suggest that their use of directives was not effective at engaging their children.
On the other hand, several mothers appeared to use directives to teach higher-level behaviors and communications to their children. This was evident for mothers in the Directive group whose children displayed average levels of engagement. While these mothers may have used directives partly to promote their children's engagement, their moderately high ratings on achievement indicate that they also used directives to encourage their children to produce higher-level developmental behaviors. Even though mothers in this group were more directive than the Low affect/responsive cluster of mothers, they were also more effective at engaging their children, pointing to the possibility that the types of directives they used played a role in promoting their children's engagement.
Perhaps the role of parental directives cannot be understood by focusing on directives as an isolated class of interactive behavior. It may be that directives play a significant role in supporting children's engagement but this can only be understood in terms of the purpose, or pragmatic function, of parents' directives (e.g., engage, teach, show information related to child's activity). In addition, the pattern of findings observed for Directive mothers points to a possible interaction between directiveness and responsiveness, such that certain types of directives enhance the effects of responsiveness on children's engagement. Clearly, future research is needed to investigate parental directiveness in terms of the interactive context and pragmatic function of this behavior.
The main purpose of this study was to determine whether Turkish mothers interact with young children with autism in a manner that is similar to that of mothers from Western countries. The majority of mothers in this sample engaged in highly directive interactions, a parenting style that is associated with the general child-rearing practices ascribed to Turkish parents. Due to the lack of a non-Turkish comparison group, this study could not determine whether the interactive qualities displayed by Turkish mothers were influenced by their cultural child-rearing practices. Furthermore, as described previously, the parallels between the patterns of interaction observed in this sample and those reported for Western mothers were striking. This suggests that, as is the case with parents and children from Western countries, children's diagnosis of autism along with its unique social interactive and developmental characteristics is likely a major influence on mothers' style of interaction.
Yet despite the apparent influence of culture and autism, it is noteworthy that more than one third of the mothers in this study engaged in a responsive style of interaction with their children, a finding that parallels research with parents from Western countries (Ruble et al., 2008; Siller & Sigman, 2002, 2008). Not only does this point to the fact that parents can engage in a responsive style of parenting even when their children have autism, it also indicates that responsive interaction is culturally compatible even in countries where responsive parenting may not be the norm. Moreover, it underscores the likelihood that a complex set of factors affect parents' interactions across cultures. Perhaps some of the most important of these are parents' knowledge of child development and understanding of disabilities, as well as the support they receive in caring for their children (Dunst & Hanby, 2010).
One of the major limitations of this study was the lack of standardized test data. This reflects the current state of developmental assessment as well as the procedures that are used to diagnose autism in Turkey. Standardized tests are not commonly used for diagnosis, primarily because they have not yet been translated and validated for Turkish children. Similarly, there is a lack of standardized instruments for assessing autism in young children. As a result, physicians diagnose autism almost exclusively by using the clinical criteria outlined in the DSM-IV or ICD-10. After this, children are referred to state-affiliated guidance and research centers, where they receive informal educational assessments. However, since children with autism tend to be underdiagnosed during the preschool years because of the limited amount of intervention resources for this age level, the children who do receive diagnoses tend to have moderate to severe levels of autism. As a result, the children who participated in this study are representative of the types of preschool-aged children who are diagnosed with autism in Turkey. Given the reliance on DSM-IV and ICD-10 clinical criteria, many would likely meet the diagnostic criteria for autism that are used in Western countries as well.
In the absence of standardized test data, this study was not able to examine associations between children's developmental and social characteristics with the way mothers interacted with their children. However, the fact that children's age and gender were neither different across maternal style clusters nor associated with variability in mothers' style of interaction supports the notion that much of the variability in parenting observed in this study can be attributed to characteristics of parents such as their child-rearing practices, psychological status, and personal beliefs related to their role in promoting their children's development.
Implications for Early Intervention
Insofar as responsiveness is the primary parenting characteristic associated with positive developmental outcomes for children with autism (Mahoney & Nam, 2011), this study points to the likelihood that large numbers of parents of children with autism in Turkey may not be engaging in the types of interactions that are likely to maximize their children's developmental potential. As a result, findings from this study suggest a need for early-intervention services that encourage parents to maximize their responsiveness with their children.
Over the past 25 years there has been an increasing interest in relationship-focused intervention for children with a variety of disabilities including autism; this attempts to promote children's development by coaching parents to engage in highly responsive interactions with their children (see reviews in Mahoney & Nam, 2011; McCollum & Hemmeter, 1997; Trivette, 2003). Recently, Karaaslan, Diken, and Mahoney (2013) evaluated the effectiveness of a relationship-focused intervention called Responsive Teaching (RT; Mahoney & MacDonald, 2007) with a sample of 19 preschool-aged Turkish children with a range of disabilities, including autism spectrum disorders, and their mothers over a 6-month period of time. Subjects were randomly assigned to treatment conditions: The control group consisted of standard preschool classroom services, whereas the RT group received biweekly RT parent–child sessions in addition to standard services. Compared to the control group, mothers in the RT group made significantly greater increases in responsiveness and their children made greater changes in interactive engagement. There were also significant group differences in children's developmental outcomes. Children in the RT group improved their developmental quotient scores by an average of 42%, compared to 7% for children in the control group.
Paralleling findings reported in this study, changes in mothers' responsiveness accounted for a major proportion of the variability of changes in children's interactive engagement; changes in children's interactive engagement were strongly associated with the developmental improvements children made during the intervention. These findings highlight the importance of responsive parent–child interaction and point to relationship-focused intervention as a viable method for addressing this issue both with Turkish and with Western parents.
Editor-in-Charge: Glenn Fujiura