Voices from the Field

Mark Wolery is a professor in special education. He is a longtime member of DEC and currently one of the co-editors of the Journal of Early Intervention. Mark is a recognized leader in the field and in 1998, he was honored by DEC with the Mary McEvoy Service to the Field Award.

Fields such as early intervention and early childhood special education are similar in some ways to the fashion industry in consumer societies. Trends flourish and excitement exists for what is new, and a disregard occurs for what is old—or is considered old even when an item’s utility remains. Last season’s sweater may be warm, comfortable, and an ideal fit, but it is eschewed nonetheless for the new. A couple decades ago, a major shift occurred in the field. We moved from purposeful direct instruction sessions to using naturalistic approaches. The extent of this shift is so great that the words “teaching” and “instruction” sound strange to some ears.

Before proceeding, I must define the context of the remarks in this essay. The statements here focus on early childhood classrooms in which young children with disabilities are enrolled, ideally inclusive classrooms. I am not talking about intervention for infants and toddlers who receive services in homes from professionals who mediate the interventions through responsive and caring family members or other adults with whom children spend time. I am writing about classrooms, because I have thought of them more deeply and studied them more carefully.

A foundational assumption about which most early intervention professionals would agree is that young children with disabilities learn from (a) their experiences (interactions) with their physical environments (toys, materials, etc.), (b) their experiences (interactions) with their social environments (parents, teachers, siblings, classmates, etc.), and (c) observing events in their environments, usually others’ behavior. They learn little, if anything, from what others “tell” them; lectures are highly ineffective with young children. We and older children may learn from being told, but young children do not. Two questions arise from this assumption: (a) “What skills do we want children to learn?” and (b) “How should we organize their experiences and observations so they will learn those skills?” Some may argue that early childhood special education is not about teaching and learning skills but about promoting development. I counter, as Donald Baer did decades ago, that “to develop” is a transitive verb requiring an object to complete its meaning. If we “promote development,” we must ask, what is being developed? The object of that transitive verb, to develop, is behavior, skills, abilities (all synonyms). I recognize young children’s biology and physiology are rapidly changing, but their biological and physiological changes and status place opportunities and limitations on their experiences and behaviors. Experiences, not the passage of time or mystic unseen forces, account for children’s adaptation to their environments. Our interventions must target their experiences.

“Teaching is the process of organizing and structuring children’s experiences and observations to maximize the likelihood they will learn the desired behaviors.”

Back to the first question, what skills do we want children to learn? Traditionally, we identified developmental domains (social, communication, motor, cognitive, and self-care), but this potentially ignores the interactions among domains and poses false skill categories. For example, is waving and saying “bye-bye” a social, communicative, motor, or cognitive (recognizing departure) skill? Functionally, it is all four. An alternative is to think about three types of outcomes. First, we want learning of positive dispositions, which are behavior patterns characterizing children’s interactions with the world. Examples are being curious, initiators of action, flexible, persistent, creative, altruistic, and many others. These behavior patterns are acquired from their give and take experiences with the environment, and we should be careful to ensure children acquire and use them (i.e., we should reinforce them when we see examples). Second, we want children to learn chains of behaviors—sequences of behaviors that when put together form complex skills. Examples of chains range from as simple as putting on a coat and washing hands to as complex as carrying on conversations and engaging in multiple-turn play interactions. Chains of behavior are terribly important because they allow children to be more independent and less discrepant from their age mates. Third, we want children to learn discrete skills—single behaviors of relatively short duration. Examples are putting a piece in a puzzle, saying “hi,” naming objects and events, naming letters, reading words, and so forth.

The second question, how should we organize children’s experiences and observations so they will learn those skills, is where instructional approaches come into play. Teaching is the process of organizing and structuring children’s experiences and observations to maximize the likelihood they will learn the desired behaviors. When our work is seen as a teaching and learning endeavor, the research on establishing and transferring stimulus control takes precedent. This sounds technical, but it is not, and it occurs in two primary ways: response shaping or response prompting. In response shaping, students are systematically reinforced for doing successive approximations of the desired behavior. Response shaping is a powerful and underused tool, but it is inefficient for some skills.

With response prompting, children are taught skills through the use and remove of prompts (help, assistance). When children are faced with a new stimulus (a situation in which a behavior is needed that they do not currently do), prompts (assistance, help) should be delivered usually by adult so they can do the needed behavior. When that new behavior occurs, then reinforcement is delivered. Almost always more than one opportunity is needed. After repeated instances of getting assistance when faced with the need for a behavior they do not have and getting reinforced for doing the behavior, prompts can be removed systematically. Several procedures, called response prompting procedures, exist, including constant and progressive time delay, simultaneous prompting, system of least prompts, and most-to-least prompting (Wolery, Ault, & Doyle, 1992). When prompts are systematically removed, children begin to do the behavior without assistance—meaning they now have a behavior to do when faced with the situation for which only recently they did not have a behavior. Control has been transferred from the prompt to the situation in which behavior is needed.

Regardless of how the teaching is done, directly or naturalistically, the processes of establishing and transferring stimulus control are in play whether acknowledged or not. Individuals who argue for naturalistic approaches say that (a) skills have to be taught in context, (b) embedding instruction into ongoing activities is sufficient, and (c) skills taught through direct instruction do not generalize. I do not agree.

The notion that teaching in context is necessary has an appealing logic. For self-feeding, self-dressing/undressing, social, and communication skills, teaching in context takes advantage of naturally occurring needs for new behaviors, which means children should be primed for instruction. This assumption is challenged by two facts. The first is practical. Many observations in many classrooms indicate the naturally occurring times result in little to no teaching. Missed opportunities for instruction abound. Perhaps this is poor teaching, but I believe it reveals an underlying flaw of naturalistic approaches: Adults simply do not, perhaps cannot because of other demands and contingencies, take advantage of teachable moments at sufficient frequency. The second fact rests in research; some data suggest isolated direct instruction results in superior learning and better generalization. Two approaches were evaluated for teaching children to spear with a fork. One taught during meals, and the other was done having sessions in which children speared Styrofoam^ΤΜ pieces (Richman, Sonderby, & Kahn, 1980). The latter resulted in more spearing in actual meals. Ledford and Wolery (in press) used small group direct instruction to teach preacademic skills, and when a child was correct, they received three tokens, one for themselves and one for each group mate. This resulted in frequent sharing in generalization contexts, which did not occur prior to this specific, out-of-context sharing experience. Likewise, examples exist of teaching preacademic skills when children initiate transitions by giving one trial per transition; this out-of-context instruction resulted in rapid learning (Wolery, Doyle, Gast, Ault, & Simpson, 1993).

Another problem with naturalistic approaches in classrooms is that relatively few “trials” exist in naturally occurring activities. The embedding studies rarely teach anywhere near the number of skills most children need to learn. Thus, little evidence suggests children’s Individualized Education Programs (IEPs) can be addressed adequately by naturalistic approaches.

“Use naturalistic teaching approaches, but you must also use multiple direct instructional sessions per child per day.”

The assumption that direct instruction does not result in generalization is false. A review of response-prompting strategies in small group direct instruction found 47 studies, 26 of which measured generalization. We evaluated children’s accurate generalization at the most distal measurement point from instruction and found participants generalized at 89.5% correct (Ledford, Lane, Elam, & Wolery, 2012). Perhaps, old direct instruction did not promote generalization; however, when direct instruction teaches multiple behaviors simultaneously rather than one behavior at a time, uses short (3 to 8 min) sessions, is made interesting, uses near errorless response-prompting procedures, and uses multiple exemplars, then generalization is likely.

My point is simple: Use naturalistic teaching approaches, but you must also use multiple direct instructional sessions per child per day. A couple generations of young children with disabilities have exited their preschool classrooms with fewer skills than were possible because their interventionists only used naturalistic approaches. This state of practice is neither acceptable nor recommended. The procedures exist for doing better by our children.