APPLICATION OF THE BEST-EXAMPLE THEORY OF CATEGORIZATION TO CONCEPT DEVELOPMENT AND COMMUNICATION TRAINING WITH SEVERELY HANDICAPPED STUDENTS

The effect of object selection on the development of generalized object categories, and generalized labeling of objects within categories, was investigated with severely handicapped students. Development of object categories was monitored through manual sign training. The effectiveness of three strategies to select objects for use during sign training was explored. Selection was conducted according to the tenets of the best-example theory of categorization. The strategies for selection were defined according to the number of examples and the level of goodness-of-example of examples presented during training. The specific strategies were as follows: presentation of one good example (GE1); presentation of three good examples (GE3); and presentation of one good, one moderate, and one poor example (ALL). Acquisition and generalization of manual signs were tested in both receptive and expressive response modalities.;As predicted by the best-example theory, results indicated that training based on good examples (GE3 or GE1) produced significantly more accurate generalization at all three levels of goodness-of-example than did training with a range of examples (ALL). Presentation of three good examples (GE3) tended to result in more accurate generalization than did presentation of one good example (GE1). These differences could not be accounted for by the differences in amount of acquisition during training, since there was no significant difference across conditions on this measure. In addition, receptive learning was more advanced than expressive learning. Because there was virtually no expressive performance during posttest conditions, the relationship between receptive and expressive learning could not be determined.;The results of the study have important implications for efficient training of concepts and labels for concepts with severely handicapped students. Several good examples of categories should be presented. Procedures a teacher may use to determine goodness-of-example of materials to be used during training are suggested. These include ratings of goodness-of-example and descriptions or illustrations of best examples, each of which could be elicited from adults with whom a teacher has contact.;In addition, the results suggest that revisions be made in some previously suggested or practiced teaching procedures. The training of generalized concept learning should be conducted in a systematic manner by selecting only good examples of the concept to be learned. Generalization should be planned prior to initial training, as opposed to after training has been conducted. To increase efficiency, training should be planned so that a minimum number of examples of a concept are presented to result in a maximum amount of accurate spontaneous generalization. Finally, positive examples of a concept are sufficient to highlight category boundaries; it appears to be unnecessary to present negative examples to define what is not a member of a category.;Three pairs of students were given training to produce manual signs within each of the three training conditions, according to a Latin square design. Three phases of training were conducted, one for each condition. Two signs were taught during each phase. Training categories for which the students did not have functional object use and which the students could not label receptively were selected. Spontaneous generalization to novel members of the training categories was assessed for all three levels of goodness-of-example (good, moderate, poor).