This paper presents a critical review of studies designed to teach language production skills to children with specific language impairment. The evidence reviewed suggests that a number of training approaches are effective, often resulting in gains that exceed the rate seen in normal development. Close analysis of children’s post-treatment usage reveals that these gains go beyond rote learning, and may even result in response classes that are different from those seen in the adult linguistic system. The chief limitations in these training studies are that children seem to make use of the target linguistic form only in untrained utterances that share certain topographical features with the utterances used in training, and that there is little evidence that children’s use of the target form will extend to speaking situations that bear little resemblance to the training/testing situation.

Cluster Reduction is purported to be a general phonological process applied by misarticulating children to simplify production of the two consonants in a cluster. In this study the generality of the process was explored in six misarticulating children. The children produced errors on all /s/ and /r/ or /l/ clusters in baseline. In a counterbalanced order three children were trained on /s/ clusters first and /r/ or /l/ clusters second in a multiple baseline design; whereas three children received training on /r/ clusters first and /s/ clusters second. Generalization to other forms of /s/ and /r/ or /l/ clusters was tested throughout training. Results showed that with one exception children generalized to within class clusters but not to across class clusters. It was concluded that if cluster error patterns reflect processes, the processes are not general enough to operate across classes, but rather, are confined to the cluster receiving training.

This study explored novel ways of using the media for education, especially second language instruction. Various combinations of visual and auditory presentations of messages were compared, for example, both written script and spoken dialogue in subjects’ first language (L1), or script and/or dialogue in a second laguage (L2), and so on. Subjects were elementary pupils with advanced training in L2. The dialogues of radio programs were transcribed, permitting such combinations as: dialogue in L2, script in L1 (the normal subtitling format); dialogue in L1, script in L2 (reversed subtitling); both dialogue and script in L1 or in L2; and so forth. On L1 and L2 tests of memory, certain combinations (e.g., reversed subtitling) were much more promising for the development or maintenance of second language skills, or for literacy training, than was conventional subtitling. Theoretical and practical inplications are discussed.

Basic cognitive psychology can provide guidelines for how to design computer systems that are easy for people to use. Prior knowledge of the real world and natural language affect the way people use computers, so computer systems that are consistent with this prior knowledge are easier for people to use. Providing users with mental models for a computer system will allow them to solve more complex problems and understand sequences of commands better. Computer systems should be flexible to accommodate the individual differences between users, but the complexity of understandable computer systems is constrained by the limited information processing capacity of users. The domain of human–computer communication promises to be a fertile domain for applied tests of cognitive theories.