Retarded subjects were taught generative pluralization rules concurrently in both the receptive and productive modalities of language. Receptive training established correct pointing to either one or a pair of objects, in response to a spoken singular or plural label of the object(s); productive training established correct spoken labels of one or a pair of objects presented visually. However, these pluralization rules were established in each modality only for a specific class of plurals: those ending in -s for one modality, those ending in -es for the other modality. This training was successful in establishing generative, or rule-governed behaviors, such that untrained examples of singulars and plurals were usually responded to correctly. Nevertheless, despite this concurrent, generative behavior, probes revealed little generalization between modalities: three of four subjects did not generalize clearly from receptive training with one class of plurals to correct productive use of that class, nor did they generalize from productive training of the other class of plurals to correct receptive response to that class. The fourth subject, however, did show strong generalization of both these types. It was concluded that automatic generalization between receptive and productive language is not necessarily an inevitable result of language training in such subjects, and therefore may require explicit, if temporary, programming, such as by direct reinforcement.
展开▼
