In 2000, the British Medical Journal published an important clinical trial of preschool speech and language services in 16 community clinics in Bristol, England (Glogowska et al., 2000). This study was the largest trial of its kind to address the communication problems of preschool children. Unfortunately, its findings were largely negative; after 12 months, the treatment group showed a significant advantage on only one of the five primary outcome variables, auditory comprehension, over a control group that had received only “watchful waiting” over the same time period. This lone treatment effect was small and, arguably, not clinically significant. Most children in both groups were still eligible for preschool speech and language services at the end of the 12-month study period.

As disappointing as these results are, it is important to note that the study was designed to study the effectiveness of early communication services as typically provided in one community in the United Kingdom (Law and Conti-Ramsden, 2000). Effectiveness studies evaluate treatment effects under relatively typical clinical conditions. As such, the investigators learned that, on average, the children in the treatment group received only 6.2 hours of intervention, or 30 minutes per month. In fact, the most intervention provided to any child was only 15 hours over a 12-month period! This study demonstrates most clearly that small, insignificant doses of early language intervention are not effective in eliminating or reducing the broad range of problems associated with preschool language impairment (see specific language impairment in children; social development and language impairment; preschool language intervention).

Unlike studies of effectiveness, which monitor client change under typical clinical conditions, efficacy studies are designed to determine, under more idealized, laboratory conditions, whether an intervention is directly responsible for positive outcomes. There is ample evidence that when preschool language interventions are applied regularly with reasonable intensity, they are efficacious, leading to clinically significant improvement in the children's language and early literacy skills. There are many varieties of preschool interventions, however, and clinicians must carefully consider the options available to them.

The principal differences between different preschool intervention approaches are best captured by determining where the interventions fall on a continuum of intrusiveness. Approaches that are highly intrusive use direct teaching methods in clinical settings, usually with the clinician as the intervention agent, to address predetermined treatment objectives, such as specific words or grammatical structures. In contrast to the prescriptive character of highly intrusive approaches, minimally intrusive approaches have goals that are stated more broadly, with less focus on specific targets. Example targets include the use of longer and more complex sentences, personal reports, and stories, with the child using an increasingly varied vocabulary. This gives the child latitude to learn from the rich set of linguistic options available in intervention contexts. In general, the clinician exerts limited control over the child's agenda. Descriptions and examples of approaches at each end of the intrusiveness continuum follow.

In protocols that are maximally intrusive, the child may examine a picture, object, or event presented by a clinician, who then presents a linguistic model and a request for the child to imitate. If the child imitates correctly, the clinician provides social or other reinforcement and then presents another stimulus set. If the child imitates incorrectly, the stimulus is repeated or simplified, and the child is prompted to imitate again. This procedure originally stems from stimulus-response psychology, but in contemporary versions, goals may be attacked in ways that are based on linguistic principles to encourage generalization to targets not directly trained (Connell, 1986).

These types of intrusive approaches were popular in the 1960s, 1970s, and 1980s, and experimental evidence indicates that they can be used to teach productive use of words, grammatical forms, and even conversational behaviors to preschool children with language impairments (Cole and Dale, 1986; Yoder, Kaiser, and Alpert, 1991). In contrast to the interventions studied by Glogowska et al. (2001), however, in these successful programs, intervention is provided intensively, often for 10 minutes to 2 hours daily, with outcomes measured after periods of several months of intervention.

Maximally intrusive intervention options have fallen out of favor because of evidence that language forms learned in this manner in the clinic do not transfer well to typical communicative contexts. Furthermore, because teaching focuses on discrete language acts, and there are no planned opportunities for the child to learn language incidentally, success depends to a large extent on the clinician's ability to identify the most appropriate communication targets for each child.

The minimally intrusive approach described by Norris and Hoffman (1990) is based on whole-language principles and differs dramatically from maximally intrusive methods. There are three general steps to this approach. The first involves selection of a theme around which the therapy room or preschool classroom is organized. This theme typically is repeated across sessions, as the children engage in dramatic play, shared book reading, art projects, and other theme-oriented activities. This thematic repetition provides greater familiarity, thus enabling children to become active participants in the activities, with reduced guidance from adults. It also provides for a natural repetition of language forms, such as words, grammatical structures, and story structures, making it easier for children to learn and use them. Second, the clinician follows the child's lead, waiting for the child to communicate rather than guiding the child's attention. Third, the clinician evaluates the child's communicative efforts and provides appropriate consequences. If the child's efforts are unclear, the clinician may ask for clarification (e.g., “You want what?” “Do you want a cookie or a pencil?”), use the cloze procedure by providing a model utterance for the child to complete (e.g., “Tell Sandy you want a “), or otherwise help the child to repair the communicative attempt. If the child communicates adequately, the child's message is affirmed with an appropriate verbal or nonverbal act. In addition, after the child's attempt (e.g., “Me eat cookie”), the clinician can recast the child's utterance by correcting its form (e.g., “Oh, you ate your cookie”) or by altering its form in some way (“Can I have a cookie now?”). In interventions such as this, it is easy and appropriate to focus on early literacy skills, such as letter knowledge, rhyming, and phonological awareness. In keeping with the limited clinician intrusiveness, however, the clinician does not directly teach specific words, language structures, story structure, or early literacy targets, nor are efforts made to get the child to imitate or to produce language out of context.

As appealing as these child-oriented approaches are, there is only limited empirical evidence that they are efficacious in facilitating language use among children with language impairments. Furthermore, it has not been adequately demonstrated that focusing broadly on the communication of meaning leads to gains in the specific areas of grammatical, phonological, and discourse weakness exhibited by preschoolers with language impairment. Techniques such as following the child's lead, recasting the child's utterances, and following the child's utterances with open-ended questions can be efficiently taught to parents or paraprofessionals, and this is an important feature. For example, Dale et al. (1996) taught parents to use these procedures during shared book reading with their children over two relatively brief sessions. Parents made more changes as a result of the intervention than the children did, but outcomes were measured after only 2 months. A longer intervention period may have resulted in greater effects on the children's performance.

Contemporary language interventions typically are hybrids that fall somewhere between the extremes in intrusiveness. For example, so-called milieu interventions blend the identification of discrete intervention targets and direct teaching using imitation and other prompts (i.e., more intrusive components) with the principles of creating natural contexts for communication, following the child's lead, and recasting the child's utterances (i.e., less intrusive components). These approaches appear to be especially efficacious for children at the single-word or early multiword stages (Yoder, Kaiser, and Alpert, 1991). Gibbard (1994) demonstrated that parents can be taught to use milieu procedures in as few as 11 sessions over a 6-month period, yielding effects commensurate with those of clinician-administered treatment. When they are applied with moderate intensity, milieu approaches increase not only the length and complexity of children's utterances, but also the children's conversational assertiveness and responsiveness (Warren, McQuarter, and Rogers-Warren, 1984).

Another popular hybrid intervention is called focused stimulation. Most focused stimulation approaches create contexts within which the interventionist produces frequent models of the child's social and linguistic targets and creates numerous opportunities for the child to produce them. Interventionists follow the child's lead, recasting the child's utterances and using the child's language targets, but they do not prompt the child to imitate. Fey et al. (1993) used this type of approach over a 5-month period to facilitate the grammatical abilities of a group of 4-to 6-year-old children with impairments of grammatical production. They also trained parents to use these techniques over a 12-session parent intervention. The children who received intervention exclusively from their parents made gains that were, on average, equivalent to the gains of the children who received 3 hours of weekly individual and group intervention from a clinician. Observed gains in the parent group were not as consistent across children as were the gains of the children in the clinician group, however.

In sum, preschool language intervention of several different types can be efficacious. Although individual clinicians have their strong personal preferences, there are few experimental indications that any one approach is dramatically superior to the others. To achieve clinically meaningful effects, however, these interventions must be presented rigorously over periods of at least several months. Furthermore, it remains unclear whether existing approaches are sufficient to minimize the risks for later social, behavioral, and academic problems preschoolers with language impairments typically experience once they reach school. To this end, promising hybrid preschool classroom interventions have been developed that aim to enhance not only the children's spoken language, but their problems in social adaptation and early literacy as well (Rice and Wilcox, 1995; van Kleeck, Gillam, and McFadden, 1998).