Effects of differential reinforcement on the generalization of a replacement mand in three children with severe language delays

We investigated variables that may influence the generalization of a replacement mand in 3 young children with severe language delays. A multiple baseline design consisting of one stimulus class of manding opportunities that we arbitrarily divided into three categories (i.e., food, toys, and events) was used for each child. During baseline probes, ail children manded mainly by reaching, grabbing, or leading. We then taught each child a replacement mand using a single member of the stimulus class. Acquisition of the replacement mand occurred under highly restricted conditions in a setting that was completely isolated from the generalization settings. Postacquisition probes revealed almost exclusive use of old manding forms. Subsequently, extinction of the old forms and reinforcement of the replacement mand were introduced in a sequential fashion. Two children manifested a substantial increase, and 1 child displayed a moderate increase in the occurrence of the replacement mand (i.e., generalization occurred). These results suggest that a differential reinforcement procedure can alter the probability of the occurrence of response class members across a variety of stimulus conditions.