Reinforcement Contingencies in Language Acquisition

Tweet

Reinforcement-based language intervention procedures amplify learning processes that operate in children’s natural language acquisition.

Key Points

Introduction

Young children’s failures to meet early language milestones often result in the diagnosis of a neurodevelopmental disorder. According to the Diagnostic and Statistical Manual of Mental Disorders (5th ed.; DSM-5; American Psychiatric Association, 2013), autism spectrum disorder (ASD) is diagnosed on the basis of deficits in social communication and interaction coupled with the presence of restricted or repetitive behavior or interests, whereas persistent difficulties with social communication without restricted or repetitive behavior patterns may lead to a diagnosis of social (pragmatic) communication disorder. By definition, these disorders involve deficits in the social use of language (i.e., pragmatics). However, additional language impairment is often present in ASD, such as delayed acquisition of vocabulary and grammatical functions (Gernsbacher, Morson, & Grace, 2015). Other children whose language acquisition is delayed or impaired include most diagnosed with intellectual disability of known or unknown origin, and children who are considered to have specific language impairment (Language Disorder in the DSM-5) in the absence of other major developmental issues. Currently, ASD afflicts almost 1.5% of all children in the United States (Christensen et al., 2016), an additional 1% to 2% are diagnosed with intellectual disability (Maulik, Mascarenhas, Mathers, Dua, & Saxena, 2011), and as many of 7% of children may show specific language impairment at school entry (Tomblin et al., 1997). Language-related difficulties overlap substantially across these diagnostic categories (Leyfer, Tager-Flusberg, Dowd, Tomblin, & Folstein, 2008), and response to intervention (RTI) may follow a similar trajectory regardless of a child’s diagnosis (Davis, Lancaster, & Camarata, 2015).

Early language competence predicts a variety of educational outcomes (e.g., Rowe, Raudenbush, & Goldin-Meadow, 2012), and early impairments are associated with later academic difficulties (Catts, Fey, Tomblin, & Zhang, 2002). Among individuals with more severe disabilities, language competence impacts adaptive functioning in adulthood (Su, Chen, Wuang, Lin, & Wu, 2008). Thus, services to address language and communication skills are warranted for the large number of children who have difficulties in this area. In the United States, access to services is impacted by multiple aspects of federal and state-level education and health care policy; for example, by implementation of the Individuals With Disabilities Education Act (2004), and by state-level insurance mandates that affect coverage of services to children with ASD diagnoses.

What are the crucial elements of language intervention for children who fail to acquire age-typical language skills through naturally occurring interactions? This article describes the role of reinforcement contingencies in language acquisition and their use in language intervention for children.

Language Acquisition and Reinforcement Contingencies

Two obvious truths characterize nature and nurture in human language acquisition. The first is that we learn language from our social environment, as evidenced our acquiring the phonology, vocabulary, and grammar only of the specific language(s) to which we are exposed. The second is that we are equipped a mechanism that permits language learning to occur. Humans, like other species, are equipped to learn through the immediate consequences of their actions; a learning process known as operant reinforcement. Early suggestions that reinforcement contingencies played a major role in the acquisition of all aspects of language (Skinner, 1957) were met with skepticism (e.g., Chomsky, 1959) that may have stemmed, in part, from failure to consider the full range of consequences that can potentially serve to reinforce the behavior that produces them (Moerk, 1983; Schoneberger, 2010). These include social reinforcement by even the subtlest reactions of listeners, which affect who we talk to and how (e.g., Borrero et al., 2007; Salzinger & Pisoni, 1960; Salzinger, Portnoy, Zlotogura, & Keisner, 1963) and potentially, “automatic” reinforcement by auditory feedback from our own speech (e.g., Palmer, 1998; Vaughan & Michael, 1982). Evidence now exist that the contingent consequences of communicating are, in fact, an important driving force in language acquisition. A comprehensive review is beyond the scope of this article, but a few relevant findings are worth mentioning.

Language Acquisition and Contingent Responses of Listeners

Infants’ prelinguistic vocalizations are reinforceable by response-contingent attention (e.g., Goldstein, King, & West, 2003; Rheingold, Gewirtz, & Ross, 1959). When attention is provided only following specific forms of vocalizations, such as vowel sounds, those forms increase selectively (e.g., Routh, 1969), whereas no increase occurs when attention is independent of the infants’ vocalizations (e.g., Goldstein et al., 2003). These short-term effects observed under laboratory conditions mirror a natural patterns of interaction in which infants’ vocalizations are affected in the long run by contingent parental responses. Noncry vocalizations of infants increase temporarily when parental attention is unavailable (Goldstein, Schwade, & Bornstein, 2009; Gros-Louis & Wu, 2012), suggesting that these vocalizations have been reinforced with attention in the past (i.e., they possess an operant function). Early presence of an operant function may enhance later language learning (Wu & Gros-Louis, 2014). Maternal responsiveness to mother-directed infant vocalizations also predicts increases in the frequency and quality of communicative vocalizations over time (Gros-Louis, West, & King, 2016). Reinforcement by parental attention promotes learning to use sounds to communicate.

Adult density of contingent attention to infants’ and toddlers’ communicative responses predicts later language outcomes, extending at least into early school age (e.g., Tamis-LeMonda, Bornstein, & Baumwell, 2001), and does so better than the content of language input (Girolametto, Weitzman, Wiigs, & Pearce, 1999). Increases in contingent attention over time predict increases in child vocabulary (Bornstein, Tamis-LeMonda, & Haynes, 1999).

Exactly how does reinforcing early vocalizations translate into acquiring vocabulary, syntax, and other advanced language skills? Perhaps developmentally advanced vocalizations facilitate further reinforcement, resulting in earlier acquisition of the fundamental building blocks of language. Older children’s acquisition of more complex aspects of language shows, for example, that parents respond differentially to children’s utterances depending on grammatical correctness (Hirsh-Pasek, Treiman, & Schneiderman, 1984; Moerk, 1983); such feedback influences the grammaticality of the child’s speech (e.g., Saxton, Kulcsar, Marshall, & Rupra, 1998). Contingent responses of conversation partners powerfully influence children’s language development.

Language Acquisition and Auditory Feedback

Effects of contingent attention can go beyond simple reinforcement. In one study (Goldstein & Schwade, 2008), mothers in an experimental condition responded vocally to all of their infants’ noncry vocalizations. One group of mothers always responded by vocalizing vowels, and another group by vocalizing consonant-vowel syllables. Mothers in corresponding yoked control groups provided the same density and type of vocal attention, but independent of the infants’ vocalizations. The infants in the contingent attention groups began to model their own vocalizations (i.e., vowels vs. consonant-vowel syllables) after those heard during contingent feedback, without directly imitating their mothers. The contingency may have promoted infants’ attention to the structure of the mothers’ vocalizations. Alternatively, infants may have vocalized when motivated to receive the mother’s attention. Both explanations fit automatic reinforcement by auditory feedback in vocal learning (e.g., Mowrer, 1950; Skinner, 1957). This hypothesis is difficult to test experimentally; however, tentative support comes from research on the babbling of hearing-impaired infants (e.g., Stoel-Gammon & Otomo, 1986), by computer simulations of vocal learning based on auditory feedback (e.g., Callan, Kent, Guenther, & Vorperian, 2000), and by research on the role of auditory feedback in the vocal learning of songbirds (Brainard & Doupe, 2000). Thus, as first proposed by Skinner (1957), both social and automatic reinforcement operate in language acquisition.

Applications

The use of reinforcement-based procedures in language intervention for children goes back at least half a century. Early studies showed how children could be taught vocal imitation, use of vocabulary words, morphological inflection, and syntax through contingent reinforcement (e.g., Garcia, Guess, & Byrnes, 1973; Guess, Sailor, Rutherford, & Baer, 1968; Lovaas, Berberich, Perloff, & Schaeffer, 1966; Schumaker & Sherman, 1970). Intervention typically consisted of arranging repeated opportunities to reinforce a response under carefully selected conditions. Naturally occurring response-contingent events were often supplemented with more powerful ones, such as exaggerated forms of attention or access to preferred items. The procedures were designed to amplify reinforcement contingencies present in the natural environment so as to accommodate learners who might show reduced sensitivity to other people’s social responses, fail to pay attention to relevant features of the environment, or simply need more learning opportunities to acquire new skills. Since then, decades of research on reinforcement-based language interventions for children with ASD and other diagnoses have focused on refining procedures and addressing challenges. Reinforcement contingencies embedded into naturalistic teaching environments and interactions (e.g., Koegel, O’Dell, & Koegel, 1987; McGee, Krantz, & McClannahan, 1985; Rogers-Warren & Warren, 1980) aim to increase spontaneity, generalization, and treatment acceptability. Functional vocabulary use has been promoted by paying careful attention to a learner’s motivation to obtain an outcome (e.g., Hall & Sundberg, 1987). Strategies have been developed to overcome discrimination difficulties (e.g., Ingvarsson, Kramer, Carp, Petursdottir, & Macias, 2016), and enhance generalization (e.g., Williams, Carnerero, & Pérez-González, 2006). Individualized assessment strategies have identified effective prompts (e.g., Ingvarsson & Hollobaugh, 2011) and error correction procedures (e.g., Kodak et al., 2016), and prompt-fading procedures have been refined (e.g., Charlop, Schreibman, & Thibodeau, 1985). Instructional spacing and temporal arrangement of learning opportunities have been investigated (e.g., Haq et al., 2015; Majdalany, Wilder, Greif, Mathisen, & Saini, 2014). Procedures have developed for teaching complex verbal skills (e.g., metonymy; McKeel, Rowsey, Belisle, Dixon, & Szekely, 2015). Finally, reinforcement contingencies may affect acquisition of linguistic information in ways that go beyond simply strengthening the behavior that participates in the reinforcement contingency (e.g., May, Hawkins, & Dymond, 2013; Vladescu & Kodak, 2013). Theoretical accounts that address such effects within the framework of operant reinforcement theory have implications for intervention strategies (Barnes & Rehfeldt, 2013; Miguel & Petursdottir, 2009).

Two examples illustrate how language intervention may involve amplification of naturally occurring reinforcement contingencies.

Implications for Service Provision

To summarize, response-contingent events play a large role in typically developing children’s language acquisition through their natural interactions with caregivers. Taking advantage of this learning process, reinforcement-based approaches to language intervention have a long history of successful use, as documented through decades of research that show how specific aspects of language can be taught through the deliberate arrangement and amplification of reinforcement contingencies under carefully selected stimulus conditions. Contrary to the contention of early critics (e.g., Weiss & Born, 1967), such procedures do not represent an artificial approach to establishing language skills, instead, mirroring processes that operate in typical language acquisition. Using these procedures systematically, within a comprehensive early intervention program, promotes substantial language gains for children with ASD (Virues-Ortega, 2010; Warren et al., 2011). However, even smaller gains can make a large difference in the life of a child whose language impairment is severe. Regardless of impairment and a child’s general prognosis given other developmental issues, progress in language and communication skills enhances independence and quality of life.

The need for providing evidence-based services to address children’s language- and communication-related difficulties is already widely recognized. Although the use of reinforcement-based procedures for teaching language is often associated with the practice of applied behavior analysis (ABA), effective arrangement of reinforcement contingencies generally plays a large role in evidence-based intervention approaches to teaching language and communication (e.g., Rogers, 2006). The challenge for policy makers, however, is to ensure that existing intervention technologies are delivered to all children who need them.

Ideally, interventions should be delivered within comprehensive programs where reinforcement contingencies supporting language development operate throughout the day. Everything known about operant reinforcement suggests that a reinforcement contingency only during limited periods of time (e.g., in brief, infrequent therapy sessions) is unlikely to result in major behavior change. In typical language acquisition, children learn from interactions with communication partners that occur throughout their waking hours. When the natural reinforcement contingencies embedded in such interactions fail to support language acquisition, supplementation and amplification should occur throughout the day as well. Language intervention programs need to be designed and overseen by professionals who, regardless of their professional credentials (e.g., speech–language pathologist, behavior analyst), possess expertise in identifying appropriate individual treatment goals and designing reinforcement-based procedures appropriate to these goals. However, service delivery need to emphasize the training of those who interact with a child on a regular basis (e.g., parents, teachers, behavior technicians) to implement these interventions during the course of daily routine.

When a young toddler falls behind in meeting language milestones, a reasonable first response may be to promote an environment rich in natural language interactions where caregivers maximally respond to communication efforts. In this context, a number of public and private initiatives have recently begun to address young children’s early language experiences (e.g., Providence Talks, 2015). These programs focus on raising public awareness of the importance of talking to children and providing parents with resources for stimulating their children’s language development, the primary aim being reduce the socioeconomic status (SES)-related “word gap” in the first few years of children’s lives (Hart & Risley, 1995). Such programs may help prevent later difficulties. However, nonresponders must quickly be identified for further support, with an intensity that is appropriate to the severity of the impairment, as well as the presence of other issues, such as ASD symptoms.

Application of a RTI framework to early childhood education may be relevant (Division for Early Childhood of the Council for Exceptional Children, National Association for the Education of Young Children, & National Head Start Association, 2013). Within an RTI framework, universal screening with curriculum-based assessment procedures identifies children in need of developmental support. Frequent progress monitoring then continuously updates decisions regarding appropriate levels of support for each individual, within a hierarchy of evidence-based interventions of increasing intensity, ideally delivered within ongoing instruction. In such a system, access to intervention is not tied to a specific diagnosis, but rather to each child’s individual strengths and weaknesses. At present, barriers to implementation include the absence of a universal system of early childhood public education (in the United States) for applying an RTI framework, as well a need for developing comprehensive language curricula for use within such a model (Greenwood et al., 2011).

Many parents currently rely on their private or public health insurance plans to access services for young children in need of language intervention, including speech-language therapy and early intervention based on ABA. Although ABA-based services have long been recommended as the treatment of choice for young children with ASD by the American Academy of Pediatricians (Myers & Johnson, 2007), access to these services has historically been limited in both educational and private health care settings (Bilaver, Cushing, & Cutler, 2016). Consistent with the notion that effective reinforcement contingencies must operate for a substantial portion of a child’s day, time in ABA-based intervention predicts response (Virues-Ortega, 2010, 2013), which has implications for needed coverage. In recent years, most states have enacted insurance laws that require coverage of services for children diagnosed with ASD (see National Conference of State Legislatures, 2015); however, scope and coverage varies across states and may fall short of meeting all children’s needs (Mandell et al., 2016). Moving forward, implementation and impact of insurance reform laws should monitor children’s access to services at the level of intensity needed.

In conclusion, we recommend directing resources toward the development of service delivery models that support early identification of language difficulties and the effective reinforcement contingencies in a child’s daily life to support language acquisition. Along with continued development and refinement of technology through research and theory, doing so will allow children with language delays to derive maximum benefit from what we know about the role of reinforcement contingencies in language acquisition.