Interventions Aimed at Improving Child Language by Improving Maternal Responsivity

Abstract

Maternal responsivity, or the ways in which mothers provide for, interact with, and respond to their children, helps to shape their children’s development, including language development. In this chapter, we describe maternal responsivity as a multilevel construct with different measures appropriate for each level. Molar responsivity refers to aspects of interaction style such as affect that can best be measured with rating scales. Molecular responsivity refers to contingent maternal behaviors that occur in response to child behaviors; and are best reflected by the frequencies of occurrence of these contingent behaviors. Results of many studies have demonstrated that both molar and molecular responsivity are related to important child outcomes such as language development. Children of more responsive mothers tend to have better outcomes. Based on these findings, interventions aimed at improving maternal responsivity and thereby child language outcomes have been developed and investigated through a number of studies. Results have shown positive outcomes for maternal responsivity and lesser secondary benefits to child language outcomes. Some of the qualities that appear associated with better outcomes include timing the interventions to co-occur with specific developments in child behaviors, teaching over a span of approximately 10–12 sessions, and designing lessons to be culturally sensitive to individual families.

A highly responsive mother reacts to her infant’s crying by offering soothing vocalizations, warmth, and affection. The infant soon stops crying and begins making cooing sounds, which in turn elicit soft vocalizations by the mother. So goes one of countless episodes in the all-important relationship between a young child and a responsive caregiver. The interactions that characterize this type of highly responsive relationship change over time in accordance with developments in the child’s behavioral repertoires. For example, as the infant begins to smile and produce a greater variety of sounds, the responsive mother is likely to react contingently and differentially to these behaviors. Such interactions between infant and primary caregiver play a foundational role in optimal communication and language development. In this chapter, we use the term responsivity to refer to how a parent responds to and provides for a child (most often mothers in our examples).

The quantity and quality of maternal interactions with infants and young children has been shown to impact children’s communication and language development, as well as aspects of cognition and emotional development (Landry, Smith, Swank, Assel, & Vellet, 2001). Mothers who talk more have children who also talk more (Hart, 1991; Hart & Risley, 1995; Paavola, Kunnari, Moilanen, & Lehtihalmes, 2005). In contrast, children who experience long periods of relatively low maternal responsivity tend to show lower language development (Landry et al., 2001). In fact, the (thankfully) few cases of extreme language impoverishment have shown that very low levels of responsivity during early sensitive periods of development may permanently arrest some aspects of language development (Curtiss, 1977; Farran, 2001). Similarly, the quality of mother’s speech as well as nonspeech behaviors has been linked to some aspects of child communication development. For example, responding more often to a child’s initiations as opposed to redirecting the child to the adults’ focus of interest has been reported to facilitate early language development (McCathren, Yoder, & Warren, 1995). Other examples of language enhancing parent behaviors (such as “motherese” or “parentese”) are at the heart of social interactionist accounts of early language development (Bates, O’Connell, & Shore, 1987; Bruner, 1977; Hoff-Ginsberg, 1990; Snow, 1991). These language enhancing parent behaviors have been documented across numerous cultures (Bornstein et al., 2008; Kartner et al., 2008). However, there is cultural variability in these behaviors and the variability reflects differences in underlying values and beliefs such as the value of child talkativeness or initiations (van Kleek, 1994).

Relative use of language enhancing behaviors also reflects socioeconomic status, frequently indicated by parent (especially mother) educational attainment (Hoff, Laursen, & Tardif, 2008). That is, higher levels of educational attainment are linked to high levels of parent responsiveness. Similar observations had been made in terms of the impact of parental responsivity on children “at risk” for developmental disabilities (Warren & Brady, 2007).

It can be admittedly difficult to disentangle the directionality of effect between maternal and child behaviors given the evidence of bidirectional effects within the dyad. Nevertheless, in this chapter we have chosen to focus on the maternal contribution to the dyadic equation because this serves as the vehicle for interventions intended to optimize responsivity. In this chapter, we discuss various key components of maternal responsivity drawing in part on our research on the effects of maternal responsivity on the development of children with fragile X syndrome (FXS). We then focus the remainder of the chapter on interventions aimed at enhancing responsivity to facilitate child development. We conclude by summarizing the common aspects of effective interventions and offer suggestions for further research on improving child communication development through improved maternal–child interactions.

1. Responsivity is a Multilevel Construct

Responsivity refers to how a parent responds to and provides for a child. Throughout the remainder of this chapter we will refer specifically to maternal responsivity because the bulk of research and intervention has been focused on maternal–child interactions. However, we can think of no reason why this literature would not apply to fathers, other relatives, and caregivers in general who spend time interacting with young language learning children.

Maternal responsivity can operate at different levels of analysis. Warren and Brady (2007) described three levels: general responsivity, molar responsivity, and molecular responsivity. General responsivity refers to basic caregiving. Mothers may be viewed as generally responsive if they provide for the basic needs of a child. For example, if the child is well fed, kept safe, and comfortable, the mother is being responsive to basic biological needs. In addition, at this most general level, the mother may seek out appropriate services for her child and advocate for her child’s needs. Although general responsivity is clearly important for child welfare, much of the research about the links between responsivity and child language attainments has focused on more specific aspects of caregiver–child interaction. We have used the terms molar and molecular responsivity to describe these levels of caregiver–child interaction. These levels are defined and discussed in the following sections.

1.1. Molar responsivity

Molar responsivity describes the affective aspects and verbal interaction style components of mother–child interactions. For example, a mother may be viewed as responsive when she displays qualities such as “warmth” or “positive affect” toward her child (cf. Landry, Smith, Swank, & Miller-Lonear, 2000). These qualities of interaction can be described and measured through the use of rating scales. For example, rating scales were developed and used by Mahoney and colleagues to describe maternal responsiveness/sensitivity, affect, achievement orientation, and directiveness (Kim & Mahoney, 2004, 2005). Studies have reported a positive relationship between these qualities of maternal interaction and child outcomes. Using a Likert-type scale called the Maternal Behavior Rating Scale, mothers of children with disabilities were found to be less responsive, have lower affect scores and to be more directive than mothers of the typically developing children (Kim & Mahoney, 2004). In addition, these authors reported significantly less engagement in young children with developmental disabilities, compared to typically developing peers. The researchers performed both correlations and regression analyses to test for significant relationships between maternal responsivity and child engagement. Their results indicated that more responsive and affective mothers had children with higher engagement scores. Directiveness was not significantly related to child engagement. Earlier research by Mahoney, conducted with mothers of children with Down syndrome, had reported that directive maternal behaviors actually had positive effects on children’s behaviors, if the directiveness was related to a topic of interest and engagement by the child (Mahoney & Neville-Smith, 1996).

Additional examples of molar responsivity are demonstrated by sensitivity to specific changes in child development. Maternal adjustments in reaction to child behaviors are examples of responsivity because they demonstrate sensitivity to and adjustment toward the child’s perceived level of development. For example, the terms “motherese” or “infant/child-directed speech” describe use of exaggerated prosody, higher pitch, slower rate, and repetitious speech toward infants (Fernald, 1993), as well as older individuals perceived as having cognitive disabilities (Nind, Kellet, & Hopkins, 2001). However, these distinctive speech patterns typically dissipate as the child grows older (D’Odorico, Salerni, Cassibba, & Jacob, 1999). That is, mothers appear to adjust to developmental changes in the child. D’Odorico et al. (1999) found that mothers decreased their use of conversational devices, or utterances without referents, used to maintain a child’s attention (e.g., “no”?) and increased their use of object-referenced speech (e.g., “that’s a telephone.”) between the ages of 11 and 16 months. This is the age span during which most babies transition from babbling to first words. Mothers’ utterances were relatively stable before and after this age period, suggesting that mothers may have been fine-tuning their speech to reflect child changes. Highly negative correlations between child vocabularies and mothers use of conversational devices at 21 months suggests that poor maternal adjustments to child developments may be detrimental to vocabulary production.

Use of infant-directed speech is often not a conscious adjustment on the part of the mother, but it may be part of a style that promotes language development by increasing interest in speech itself. Singh (2008), for example, reported improved infant word recognition when mothers spoke with infant-directed speech (compared to a more adult-directed speech pattern). Another example of maternal responsivity is the fact that observant mothers are much more of developmental progressions such as sounds and words during the first year of life, than they are to changes that occur later in development (Warren & Brady, 2007).

In addition to use of different speech patterns, responsive parents adjust attention to child behaviors. Legerstee, Varghese, and van Beek (2002) found that mothers of typically developing children and children with Down syndrome redirected younger infants (mean CA 8.6 months) more often than older infants (mean CA 16.5 months), whereas the mothers of the older infants maintained the child’s focus of attention more often. It is possible that mothers of the younger infants redirected more because their infants were less skilled in maintaining attention on their own.

1.2. Molecular responsivity

We have used the term molecular responsivity to refer to contingent maternal actions that can be directly linked to changes in child behaviors (Warren & Brady, 2007). For example, if the child vocalizes and the mother immediately responds by speaking to the child, this would be considered a contingent response by the mother. Or if a young child attempts to say a word and their mother recasts the child’s attempt by saying the word more clearly, this would also be considered a molecular example of responsivity because the mother’s action is in direct response to a notable change in the child’s behavior. Researchers such as Bornstein, Tamis-LaMonda, Hahn, and Haynes (2008) and Spiker, Boyce, and Boyce (2002) have pointed to contingent maternal responsivity as particularly important for child developmental outcomes including emotional security, social relationships, cognitive development, and language outcomes. Contingency leads to predictability and this predictability is thought to enable the child’s development of self-regulation and emotional well-being (Landry, Smith, & Swank, 2006). These types of responses support and encourage the child’s vocalizations and verbal development.

Molecular responses may be contingent upon changes in child attention or play behaviors. For example, maternal responses to shifts in a child’s attention (demonstrated by looking in a particular direction) or initiating play with a particular object are often analyzed (Legerstee et al., 2002; Paavola et al., 2005). One of the key factors in determining contingency is often the lag in time between child behavior and maternal responses (Bornstein & Tamis-LeMonda, 1989; Kartner et al., 2008). Maternal responses that occur within several seconds of the child’s change in behavior are typically considered contingent and would follow under our description of a molecular analysis of maternal responsivity.

The following example, drawn from the authors’ research on maternal interactions in families that have a child with FXS, will provide an illustrative summary of the different levels of responsivity research.

1.2.1. Research example

In our work with families who have a child with FXS, all children live in what we would describe as a generally responsive interaction. That is, all participants were adequately fed, clothed and basic child needs have been met. (Note that this may not be the case in studies that include children who live in extreme poverty or in neglectful situations.) We have employed both molar and molecular measures to further describe maternal responsivity and to the potential relationships between responsivity and child developments (Warren, Brady, Sterling, Fleming, & Marquis, 2009).

Global ratings have been employed by watching samples of videotaped mother–child interactions and then assigning a value between 1 and 5 for molar qualities of maternal interaction. These ratings were modeled after those used by Landry and colleagues (Landry et al., 2001, 2006). A 30-min observation of mother–child interaction is divided into 10-min segments. The three 10-min segments are scored separately on each of the following components using a five-point rating scale: positive affect, warmth, flexibility/responsiveness, physical control, verbal discipline, and punitive tone. Scores for each 10-min segment are averaged for one single score per component, per dyad, per observation. This procedure yields an indication of maternal responsivity, relative to other mothers in our sample and relative to participants in other studies that have used this scale.

In addition to this molar level of coding, we have coded molecular responsivity. Molecular coding is measured by coding each maternal behavior toward the child to determine, for example, how often the mothers respond contingently to the child’s changes in behavior, and how often mothers use language to interpret their child’s communication attempts. For this level of coding, research assistants observe the videotaped interactions along with a transcript of all child communication behaviors. We are interested in specific types of maternal responses; therefore, observers record redirects, recodes, requests for verbal complies, comments, and admonishments directed to the child. Data derived from these molecular scales enable us to analyze the degree to which frequencies and proportions of specific maternal responses following child behaviors are related to child outcomes.

Not surprisingly, several aspects of the molar and molecular ratings are correlated with each other. Nevertheless, we have found that the molecular ratings that capture the contingent nature of maternal responsivity are most closely aligned with child cognitive and communication outcomes (Sterling, Brady, Warren, Fleming, & Marquis, 2006). One reason for this difference is probably due to the greater variability in the molecular scores, compared to the molar scores. When considered together, data from the two scales provide complimentary information about overall maternal interaction style and specific behaviors that have been linked to child language developments. A mother who scored high on the flexibility/responsiveness scale would routinely follow the child’s lead in interactions. If this same mother also provided vocabulary input while following the child’s lead, her child would likely show gains in vocabulary development. For example, saying “night night bear” as the child is covering a bear with a blanket is an example of combined flexibility/responsiveness and contingent comments. Interactions such as these are often suggested for promoting language development (Tamis-LeMonda, Cristofaro, Rodriguez, & Bornstein, 2006).

2. Responsivity Relates to Child Outcomes

Responsivity correlates significantly with, and in many cases predicts, certain child outcomes, in typically developing children, in children at risk for developmental delays and in children with identified disabilities such as Down syndrome, autism, and FXS. Differences in maternal responses to child behaviors have been observed within just a few hours of birth. For example, babies who experienced increased maternal talk when being picked up were found to be less fussy and to spend more time in a quiet awake state compared to infants in a control group who did not have extra exposure to maternal talk (Thoman, Korner, & Beason-Williams, 1977). Thus, differences in child behaviors can be observed to follow responsive caregiving right from birth. Changes in fussiness or attentiveness may show up at these very early stages, setting the stage for later changes in communication and language that appear attributable to responsivity.

2.1. Language and communication outcomes

In terms of language outcomes, maternal responsivity has been linked to several important aspects of early language development, such as productive and receptive vocabulary sizes (Bornstein & Tamis-LeMonda, 1989; Tamis-LeMonda, Bornstein, Baumwell, & Melstein Damast, 1996). Children with more responsive mothers during their first year of life had higher vocabulary scores once they began to talk (around 13 months in one study, and later in toddlerhood in another). In these studies with typically developing children and their mothers, responsivity not only relates to cumulative vocabulary size, but also to important milestones such as first words and the onset of the vocabulary burst (first 50 words) (Tamis-LeMonda et al., 2006). That is, children with more responsive mothers said their first words and hit the 50-word marker at significantly younger ages than children with less responsive mothers. The importance of the vocabulary spurt is that it signals not only increased vocabulary size, but also coincides with the onset of beginning sentence productions.

Responsivity has also been found to be an important variable for families who have children at risk of developmental delays. Risk may be associated with poverty or other adverse environmental circumstances and also with prematurity. In a series of studies, Landry and colleagues investigated the role of responsivity in mothers of children who were born preterm. Some children also had very low-birth weight, and additional medical risks. Sustained responsive caregiving was found to be positively associated with a number of child outcomes in these children, including vocabulary development. Responsivity remains important throughout the preschool years. In a study with over 500 low-birth weight children, responsivity at 30 months predicted verbal IQ measures obtained when children were 5 years old (Fewell & Deutscher, 2002). Responsivity was a significant factor even after differences in initial vocabulary comprehension were considered.

Although research has often indicated that early maternal responsivity is particularly important for child development, Landry and colleagues reported that the amount of responsivity sustained over time, that is over many years, is an important factor in child language outcomes as well as important social, emotional, and cognitive outcomes (Landry et al., 2001; Landry, Smith, Swank, & Guttentag, 2008). This study measured maternal responsivity with global rating scales, and concurrent child language development at child ages of 6, 12, 24, 36, and 48 months of age. Children who experienced highly responsive parenting early in development, but not later, or later but not earlier, scored significantly lower on measures of language, cognitive and social development than children who experienced sustained, consistently high levels of responsiveness.

Additional evidence for the importance of sustained responsivity is provided by the work of Hart and Risley (1995). This frequently cited study reported on the cumulative effects of variations in children’s language experiences between 1 and 3 years of age. The authors calculated differences in language input over multiple observations across children who eventually had relatively low versus high vocabularies. Children who ended up with higher productive vocabularies had experienced vastly more utterances spoken to them over time, compared to children with relatively lower vocabularies. In addition, the authors point to differences in types of utterances such as responses to child initiations and parent utterances containing nouns as further contributing to positive child language outcomes.

2.2. Responsivity in families who have a child with a disability

Children who have various disabilities are likely to have delayed language development and for these children it may be beneficial to provide a responsive learning environment. Research by Warren, Yoder, and colleagues has found significant relationships between maternal responsivity and language outcomes in children with developmental disabilities. Yoder and Warren (1999b) analyzed relationships between early child communication attempts, maternal responsiveness to these attempts and later expressive and receptive language scores. Early child communication significantly correlated with these language outcomes, but the effects were mediated by maternal responsivity. Children who communicated more and whose mothers responded more to these attempts had better outcomes.

In addition, these authors reported that effects of an intervention package, prelinguistic milieu teaching (PMT), were dependent on initial levels of maternal responsivity (Yoder & Warren, 1999a). Children who had highly responsive mothers at the outset of intervention showed better effects of the PMT intervention than did children who had less responsive mothers. In subsequent research Yoder and Warren added parent responsivity training to their intervention studies to complement direct practitioner provided PMT. They termed this approach responsivity education/prelinguistic milieu teaching (RE/PMT). Fey, Warren, Brady, and colleagues have reported the effects of a randomized clinical trial of this approach (Fey, Warren, Brady, Finestack, Bredin-Oja, et al., 2006; Warren, Fey, Finestack, Brady, Bredin-Oja, et al., 2008) and it is described later in this chapter.

Children with disabilities such as Down syndrome, autism, or fragile X may experience lower levels of maternal responsivity due to low rates of child initiations and/or difficulty sustaining interactional routine compared to children without disabilities. Delayed onsets of important prelinguistic and early linguistic skills are generally associated with these disabilities (Chapman, 2003; Franco & Wishart, 1995; Roberts et al., 2005; Rondal, 2003). Delayed onset of certain communicative behaviors such as pointing may delay and decrease the amount of maternal responsivity because the lack of intentional communication gives parents few child behaviors to contingently respond to. For example, Tomasello and colleagues have shown that mothers often respond to infant pointing gestures by providing the verbal labels for the objects pointed to by the infant (Tomasello, 1995, 1999). Thus, when a child is delayed in onset of pointing, they may experience cumulatively less vocabulary input, particularly during bouts of joint attention. This type of vocabulary input has been found to predict child vocabulary acquisition over time.

There is evidence that many children with developmental disabilities experience potentially important differences in responsivity from early infancy onward. For example, Slonims and McConachie (2006) reported very early differences in maternal responsivity between mothers of infants with Down syndrome compared to mothers with typically developing infants. Maternal responsivity and child social and communication behaviors were measured when children were 8 weeks old and again at 20 weeks old. The 8-week measures were not significantly different, but by 20 weeks of age the mothers of children with Down syndrome were judged as being more remote and less sensitive than the mothers of the typically developing children. Correlations between infant and mother behaviors at the two time points suggested that maternal responsivity for the Down syndrome group were related to child behaviors, whereas behaviors of mothers of typically developing children were primarily associated with nonchild variables such as maternal mental health.

Some interactional characteristics associated with developmental disabilities may be disruptive to maternal responsivity. These include low child initiation rates, slower response times, gaze avoidance or atypical eye gaze patterns, hypersensitivity to environmental stimuli, social anxiety or shyness, and poor speech intelligibility (Warren & Brady, 2007). Put another way, children with disabilities may have difficulty providing clear signals about their needs (Landry et al., 2008). A mother may respond to these child behaviors by directing their child to interact in particular ways. For example, mothers may become less likely to wait for their child to respond over time if the child’s response latency is excessively long. In the case of children with ASD or FXS, it may be that problem behaviors including impaired social and cognitive engagement could suppress parental responsiveness that affects parental engagement and responsiveness over time. The resulting interaction pattern is one of low child engagement and high parent directiveness. To summarize, the interaction patterns observed in dyads that include a child with disabilities may develop some characteristic features such as maternal directiveness, but these characteristics are likely to be associated with specific child behaviors that may make it difficult for some mothers to be highly responsive.

FXS provides an interesting context for studying maternal responsivity. Children with FXS display a social, cognitive, and linguistic profile similar to that of autism, including the aforementioned behaviors that are disruptive to maternal responsivity. In addition, women who are carriers of FXS are more susceptible to depression and anxiety, which can also impact responsivity. Warren et al. (2009) examined the relationship between maternal responsivity and later communication and language development in a cohort of young children with FXS. Responsivity was measured at the molecular level, and both maternal and child communicative behaviors were coded. The results indicated that maternal responsivity predicted child language variables at 36 months of age, including number of different words, rate of child communication, as well as scores on expressive and receptive language subscales on the Mullen Early Learning Scales (Mullen, 1995). In other words, children with FXS who were exposed to a highly responsive style of parenting scored higher on language outcomes than children who received less responsive interactions. This finding complements the literature on responsivity within typical development; both children with developmental disabilities and children with typical development demonstrate a measurable advantage when exposed to high levels of maternal responsivity early in life (Landry, Smith, Miller-Lonear, & Swank, 1998; Landry et al., 2001; Tamis-LaMonda, Bornstein, & Baumwell, 2001; Warren et al., 2009).

In addition to varying responsivity in accordance with child developments, mothers may be more or less responsive depending on whether the child is first born or latter born (Furman & Lanthier, 2002). There are many possible explanations for possible differences in responsivity within a family. It could be that mothers of latter-born children are more responsive because they are experienced, more assured in caregiving, and better at reading their child’s behavioral and emotional signals. Alternatively, mothers of latter born children may be more influenced by the competing demands of caring for multiple children. Although there have been many studies conducted on the influences of these family dynamics on parenting style, it is difficult to disentangle the effects of birth order, age of children, spacing of children, size of family, family support variable, and so on. Nevertheless, it is important to remain mindful of parenting within the complex family structure when trying to evaluate maternal responsivity and its role on child outcomes.

The literature reviewed above provides examples of different responsive techniques repeatedly observed in typical mother–child interaction that may play important roles in early language development. This literature has been steadily accumulating for nearly a half century and is central to social-interactional theories of language development. We have provided only selected examples above. Taken as a whole, whether one is focused on typical child development or atypical development, the research supports the contention that a consistent, contingent, and adaptive interaction style is the most facilitative of early language development. Not surprisingly then, a number of early intervention programs have been developed with the goal of enhancing various pivotal aspects of maternal responsivity that may have a cumulative impact on development. In the following section, we briefly describe several interventions aimed at primarily improving maternal responsivity, with a primary goal of thus improving children’s language outcomes.

3. Interventions Aimed at Improving Responsivity

Thus far we have made a case for the important contribution that cumulative exposure to highly responsive maternal interactions can provide for child development in general and language development in particular. One of the most salient and promising outcomes of the research on maternal responsivity is the recognition that intervention might attempt to establish and enhance this parenting style as a means of facilitating young children’s language development.

The goal for the intervention approaches we review next is to teach mothers an interaction style that the descriptive literature indicates to be closely associated with optimal child development. Central to this style is the notion that it should be responsive to the child’s interests, not the adults, and to the extent possible be responsive to child initiations. Questions, models, and imitation prompts may be used, but only within conversational routines based on the child’s interests. Furthermore, the aim is to utilize this style throughout the day whenever opportunities naturally arise. This style may be especially effective in the context of structured routines such as dialogic book reading (Whitehurst, Falco, Lonigan, & Fischel, 1988; Zevenberger & Whitehurst, 2003). The overall aim is to emulate the consistent ubiquitous responsivity that has been associated with relatively advanced child language outcomes.

Relative to other behavioral interventions, a substantial amount of research supports the conclusion that a highly responsive child-focused interaction style can be taught to a parent or practitioner with a relatively modest effort. Bakermans-Kranenburg, van IJzendoorn, and Juffer (2003) examined the effectiveness of various parental interventions on parental sensitivity, attachment, and responsivity by reviewing a large number of studies and using meta-analytical procedures to analyze cumulative results. Effects across 70 different parental intervention studies that were aimed at improving infant attachment and security were compared. Although this child outcome (sensitivity/attachment) is not the same as the language outcomes focused on in the current chapter, the literature is relevant because many of the interventions were aimed at improving maternal responsivity. Results of this meta-analysis indicated that relatively short interventions that lasted less than 16 sessions were found to be more effective than longer interventions in establishing more responsive interaction styles. An additional finding was that mothers increased their responsivity significantly more in studies that used nonprofessional (e.g., experienced mothers) interveners, compared to professional (e.g., graduate student) interveners.

Based on this large set of data, Bakermans-Kranenburg and colleagues concluded that interventions that began at around 6 months after birth or later were slightly more effective than those that began prenatally or at birth. Further, well-timed, focused interventions that lasted “long enough to lead to sustained changes in mother–child interactions, but not so long as to become burdensome” were optimal. The issue of timing of intervention is important and in need of further research. It may be that there is something specific about child development at the age of 6 months that provide an optimum social context for interventions. Or, it may be that by the time her child is 6 months of age, mothers are comfortable with being a mother, and are ready to learn new interaction strategies.

Recent research by Landry and colleagues investigated the timing of interventions. Mother–child dyads who experienced the PALS intervention (described below) during both infancy and the toddler period were compared to dyads that experienced the intervention only in infancy. They found that mothers who experienced both phases of intervention had higher levels of contingent responsiveness. Increases in maternal warmth were associated with early interventions, and changes in mothers’ language input were associated with interventions during the toddler stage. Similar timing differences were found for child language outcomes. Greater effects on receptive and expressive language scores were seen following the toddler intervention. Thus, optimal timing of interventions differed depending on the outcome variable and “the degree to which the behaviors were linked to a child’s changing developmental needs” (Landry et al., 2008, p. 1350).

Changes in maternal responsivity have been documented following as little intervention as a single viewing of a developmental assessment session (Anderson & Sawin, 1983), but sustained changes in maternal interactions may require participation in an ongoing intervention that provides weeks or months of regular learning opportunities. Such is the nature of the programs briefly described below. These intervention approaches are generally based on the same underlying principles and many of the same general set of adult behaviors and child skills although sometimes different terms are used for the same underlying concept.

A number of interventions have been developed that address parent responsivity within everyday interactions. Interventions such as It Takes Two to Talk and RFI (each described in following sections) encourage parents to implement strategies throughout the day, by embedding strategies within everyday routines. Other interventions have created routines that provide a context for learning new skills (Snyder-McLean, Solomonson, McLean, & Sack, 1984). Research by Woods and colleagues has emphasized the importance of parent input into the selection of teaching routines (Kashinath, Woods, & Goldstein, 2006; Woods-Cripe & Venn, 1997). Kashinath, Woods, and Goldstein taught mothers to interact in specific ways during individually selected everyday contexts, such as dressing or mealtimes. A unique feature of this intervention was that teaching contexts were identified by first querying families about the routines they already engaged in, such as dressing, music, and play. Next, the different routines were evaluated by each mother and the interventionist to identify the routines that were most representative, occurred most frequently, and were most preferred by the mothers. Mothers learned interaction strategies such as waiting for their child to initiate and contingently imitating their child’s communication, within the contexts of the selected routines. A potential benefit of this approach is that individualization is likely to increase parent implementation of the intervention. In addition, generalization of these strategies to nontraining contexts was also observed.

Results from the routine-based intervention implemented by Kashinath et al. (2006) were evaluated using a single-subject design. Five mothers each showed increases in various aspects of responsivity including waiting for child communication and contingent imitation of child responses. The five children of the mothers who participated also showed improved communication. Increases were observed in children’s use of gestures, single words, and multiword utterances.

3.1. Interventions based on materials from the Hanen Center

One of the most developed and marketed interventions aimed at improving parental responsivity and child language outcomes was developed at the Hanen Center. A number of intervention programs are now available that promote language learning opportunities for children with a range of language disorders including Autism and Late Talkers. “It Takes Two to Talk” (Girolametto & Wieitzman, 2006) was developed to teach parents of young children to apply language facilitation strategies across everyday contexts. The curriculum for “It Takes Two to Talk” is intended to be delivered by a certified speech-language pathologist (SLP). It includes information on the development of nonverbal and early verbal communication, how to recognize, respond to, and provide opportunities for children’s early communication attempts, and how to incorporate these communication opportunities throughout the day. One of the theoretical underpinnings of the Hanen Center approach is based on Vygotsky’s concept of a zone of proximal development. Parents are encouraged to provide input that matches the child’s input. For example if the child is producing mostly one-word responses, parents are told to reduce the length of their utterances to single words or short phrases.

The Hanen Center approach is “family centered” in that it recognizes the family as the most important element in a young child’s life (Girolametto & Wieitzman, 2006). Parents help each other through group interactions and each parent actively participates in selecting appropriate intervention goals for their child. The curriculum is delivered primarily through 11 weekly sessions between a trained SLP and small groups of parents. Parents are provided feedback on their interactions through the use of videotaping and counseling. It is worth noting that Bakermans-Kranenburg et al. (2003) reported the use of videotaping as a significantly effective intervention strategy.

A number of research studies have found that parents participating in “It Takes Two to Talk” have changed their interaction styles and became more responsive to their child’s communication after participating in the intervention. Girolametto (1988) found that, compared to a control group, mothers who participated in “It Takes Two to Talk” used fewer turns, and maintained longer conversational exchanges. A follow-up study by Tannock, Girolametto, and Siegel (1992) showed similarly positive results in terms of mothers’ responsivity. Unfortunately, across these two studies, no significant effects were found for children’s language development. Based on these outcomes, the authors added additional material to the curriculum of “It Takes Two to Talk” that focused more on strategies aimed at stimulating word production (Girolametto, Perarce, & Weitzman, 1996).

Results from the version of the intervention that included focused word stimulation indicated positive results for both parents and children. In focused stimulation, targeted words are spoken more often during the intervention, and opportunities for the child to use these words may be provided (Ellis Weismer & Robertson, 2006). Children in the intervention group had a significant increase in their number of target words learned and had significantly larger vocabularies according to parent report.

Another program developed at the Hanen Center, “More than Words” is an offshoot of the program “It Takes Two to Talk” and was developed specifically for children with ASD and their mothers. This intervention emphasizes many of the same principles of reciprocity as “It Takes Two to Talk,” but also includes information on the importance of affect, predictability, structure and the use of visual supports to facilitate language learning in children with ASD. Like the “It Takes Two” program, intervention is provided through individual sessions and parent groups over an 11-week time period. Girolametto, Sussman, and Weitzman (2007) found that three mothers of children with ASD increased in the following ways after participating: following their child’s lead, promoting children’s participation in routines and modeling language at the children’s level (matching). The participating mothers’ children also showed increased vocabulary size and numbers of different words after intervention. Larger studies with a control group will allow a more definitive evaluation of the role of “More than Words” in promoting maternal responsivity and child language outcomes.

An intervention based on “More than Words” was implemented in a pilot study of responsivity in mothers of children with FXS conducted by the current authors (Brady, Sterling, & Warren, 2008). Four mothers participated in ten weekly intervention sessions that utilized the teaching materials provided by the Hanen Center. It should be noted, however, that, although the intervention was implemented by a Hanen certified clinician, it did not include a group component. That is, all aspects of the intervention were provided in 1:1 intervention sessions. As in program, other applications of this participating mothers were taught to wait for their child to initiate interactions, recognize communication attempts, follow the child’s lead, and provide simplified input to their child.

A multiple baseline design was used to evaluate responses by each participating dyad. Videotaped interactions were collected before, during and after interventions and maternal and child behaviors scored from these videotapes served as primary data sources. Three of the four mothers increased their proportion of responsive interactions and decreased proportions of directive-style behaviors during the intervention sessions. The fourth mother did not show any changes in her behavior. It may be noteworthy that this fourth mother also had the full mutation of FXS; hence, the instructional materials may not have been explicit enough for this mother.

The children of the three mothers who responded to the intervention also showed positive changes in language outcomes during intervention. Specifically, increased vocabulary size and rates of intentional communication were observed. Positive language changes were not observed in the fourth child.

3.2. Relationship focused intervention (RFI)

This intervention, developed by Mahoney and colleagues, is designed to help parents learn and incorporate responsive interactions in their daily routines with their children and is conceptually similar to the Hanen approach. Topics addressed by RFI include interacting with children in play, turn taking, following the child’s lead, increasing the number of responses, and decreasing the number of redirectives issued to the child. As described in two recent studies by Kim and Mahoney (2004, 2005) intervention includes four components: classroom-based instruction; home-based instruction; feedback; and evaluation. Children were between 3 and 8 years of age, with developmental age equivalents around 2–4 years, based on the Vineland Adaptive Behavior Scale. Intervention is typically provided in weekly sessions lasting from 1 to 2 h over the course of 3 months. The majority of instruction (e.g., eight sessions) is provided in a classroom setting and then home-based instruction sessions (e.g., two sessions) offer opportunities for interventionists to observe and provide feedback to mothers during daily routines. Videotaped observations of mother–child interactions provide an additional source of feedback.

Results from the Kim and Mahoney studies were similar to those reported for earlier versions of the Hanen programs—mothers changed but only modest changes in child behaviors were observed. Effects on mothers’ responsiveness, affect and achievement orientation were significant, but only child affect significantly increased in the intervention group. However, small sample sizes may again limit the significance of their findings because only 18 dyads participated in this study, 10 in intervention and 8 as control dyads. Participants were not randomly assigned to groups; rather participants were matched according to maternal age, years of education and level of stress of the mothers, age of the child and functioning level of the child. Post hoc analyses indicated that children whose mothers made larger gains in responsiveness did show substantial changes in their interactive behaviors.

3.3. The playing and learning strategies (PALS) program

The PALS program was developed by Landry and colleagues (Landry et al., 2006, 2008) to address maternal responsivity in families at risk of low social engagement due to biological risk associated with prematurity and very low-birth weight. PALS is a home visiting program designed to teach mothers to engage in a highly responsive style that is similar to those targeted by It Takes Two to Talk and RFI. The goal of PALS is to establish a style of interaction that includes four different aspects of responsiveness: contingent responding, emotional-affective support, support for infant foci of attention, and language input matched to infant’s developmental level. Ten weekly 90-min home visits provide maternal instruction. The format for the visits includes (a) asking mothers to review their experiences across the last week; (b) describing the current visit’s targeted behavior; (c) watching and discussing videos of other mothers modeling targeted behaviors with coaching; (d) videotaping mothers with their infants engaged in toy play, feeding, bathing, and so forth with coaching; (e) allowing mothers to critique their behaviors and infant responses while viewing the videotape; and (f) planning for practice during the coming week.

Landry et al. (2006) evaluated the effectiveness of PALS by comparing performance of a group of mothers who participated in PALS to another group of mothers who participated in a control group condition. The control group experienced the same number of visits as did mothers in the PALS group, but the focus of the control group was developmental milestones, rather than responsivity. A strength of this study was the large number of participants—264 mother–infant pairs in total. Measures were obtained longitudinally during four assessments conducted when infants were between 6 and 13 months of age. Positive outcomes were obtained for both mothers and infants. Increased maternal responsiveness facilitated greater growth in infants’ social and emotional skills and early communication development including vocabulary. Additional assessments of infant interactions with an experimenter also showed growth in infant skills. That is, the infants generalized their interaction and communication skills to interactions outside of mother–child interactions.

Results from Landry and colleagues’ studies also feature changes in more distal measures of child outcomes, including standardized language measures. Improvements in receptive and expressive scores on the Preschool Language Scale (Zimmerman, Steiner, & Evatt Pond, 2003) and the Peabody Picture Vocabulary Test (Dunn & Dunn, 1997) demonstrated that changes in child language behaviors over time were also associated with changes in mothers responsivity following PALS interventions.

3.4. Responsivity components in language intervention studies

A focus on maternal responsivity is often one component of a complex intervention than includes child goals in addition to maternal interaction goals. An example of such a complex approach is responsivity education/prelinguistic milieu teaching (Warren, Bredin-Oja, Fairchild Escalante, Finestack, Fey, et al., 2006). RE/PMT evolved over many years of research (Fey et al., 2006; Warren et al., 2008; Yoder & Stone, 2006; Yoder & Warren, 2001, 2002). The child-focused component of the intervention is aimed at improving prelinguistic gestures and vocalizations in children who have significant delays in language development and who are currently infrequent communicators. The parent-focused component follows many of the same themes addressed by It Takes Two to Talk, such as following the child’s lead, observing child communication and waiting for child initiations.

Yoder and Warren (2002) included parent responsivity education as part of their comprehensive early language intervention program. Parents were provided individualized treatment focused on increasing responsivity and providing ideal language models and feedback, in a series of 11 weekly sessions. The authors found significant increases in parents’ use of optimal contingent responses for parents in the intervention group, compared to children in a control group. Optimal responses included responding to child initiations and linguistic mapping of child communication attempts. For example, if a child pointed to a spinning top and said “that” the parent might linguistically map the child’s communication by saying “yes, that’s a top.”

In a study that replicated many of the components of Yoder and Warren’s (2002) study, Fey et al. (2006) also reported an increase in linguistic mapping following RE/PMT intervention. In contrast, parents who did not receive RE/PMT (because they were in the control group) did not show increases in linguistic mapping. These increases in linguistic mapping are encouraging because they reflect learning of one of the key contingent responses targeted in the intervention, and because linguistic mapping is part of a general approach called language stimulation that has been shown to be effective in improving child language outcomes (Ellis Weismer & Robertson, 2006).

One concern regarding parent-focused interventions is that the intervention may increase parental stress levels—levels that may already be high due to having a child with a disability (Kim & Mahoney, 2004). The study by Fey et al. (2006) monitored stress levels of the mothers who participated. Mothers completed the parental stress index (PSI) (Abidin, 1995) before and after intervention. Potential differences in stress that could be attributed to participating in the intervention were evaluated with an ANOVA. There were no significant differences for treatment versus nontreatment groups, and the PSI scores were within normal limits. The authors concluded that RE/PMT did not increase parental stress as measured by the PSI.

4. Summary and Conclusions

Responsive interactions between infants and young children and their primary caregivers are essential for optimal language development and this basic finding has been well established in the literature. Importantly, a growing body of research is becoming available regarding interventions aimed at improving maternal responsivity and thereby also impacting child language attainments. A number of effective aspects of intervention such as length and timing can be identified that are common to the interventions reported in the literature and summarized in the previous sections. The intervention programs that were described in previous paragraphs were generally effective following 10–12 weekly sessions. This amount of time appears sufficient to achieve sustained change in maternal responsivity and modest changes in child language outcomes associated with these maternal attainments.

Interventions appear most effective when they are timed to co-occur with expected child developments. For example, the studies by Landry et al. (2008) reported optimal child outcomes when intervention was provided during both infant and toddler years, but specific gains in distal language outcomes were tied to the toddler phase of intervention. For example, greater use of multiword utterances was observed in children following responsivity intervention delivered during the toddler years. The authors concluded that optimal timing of intervention depended on skill complexity and the age at which the targeted skill would naturally be observed. However, most of the children in the Landry et al. studies did not have identified developmental disabilities, and further research regarding these timing issues in children with varying degrees of delayed development are needed.

To date, most of the research studies have provided outcomes based on using a curriculum that teaches multiple aspects of responsivity, such as following the child’s lead, being contingent, waiting for child communication, and providing linguistic input. These components are based on studies documenting their relationships to child language outcomes. Use of videotaped feedback was also frequently identified as a significant feedback component.

Although the research summarized in this chapter indicates that interventions can be effective in increasing maternal responsivity, potential interventionists and researchers must remain mindful of the sensitive, complex nature of family interactions when planning and conducting interventions. In previous sections, we noted the need to monitor stress levels that could be elevated by intervention. Families who are likely to participate in these interventions are at risk for increased stress due to the challenges faced in raising a child with a disability, financial burdens associated with disability, and potential cognitive limitations by the mother or other members of the family. Interventions aimed at improving maternal responsivity may often be improved, therefore, by individualizing curriculum and goals to reflect individual family concerns.

In addition to individualizing interventions to reflect family dynamics, concern for a family’s culture should also be maintained. As mentioned in the introduction of this chapter, responsivity may indeed vary across cultures, in accordance with different values toward child communication. Interventionists are faced with the challenge of how to address interaction styles and practices that may sometimes be at odds with a family’s cultural background. van Kleek (1994) suggests that interventionists may alter existing programs to fit a particular family or develop different materials based on a study of successful and unsuccessful interactions within the target family.

In terms of affecting both maternal responsivity and child language outcomes, interventions that combine both responsivity education and language interventions, such as the RE/PMT approach may be most likely to generate positive results. The combined approach has several advantages. Mothers learn a facilitative interaction style that is likely to foster child language at current and future levels of child language development. Children will benefit from individualized intervention aimed at promoting growth toward an advanced level of language and communication. In addition, parents place extreme value on the child language component and may be more likely to “buy-in” to an intervention that includes child goals, as opposed to an intervention that is exclusively focused on maternal responsivity.

Although the research to date generally supports interventions aimed at improving child language by increasing responsivity, further research is needed in this area. Specifically, further studies are needed to help identify components within treatment packages that are optimal and to identify the optimal timing of intervention components (relative to both child chronological and developmental ages).