A comparison of three approaches to semantic memory: Network, feature comparison, and schema theory

Abstract

During any interaction, interactants continually draw upon semantic memory. Semantic memory is the matrix of concepts to which incoming language production is compared for meaning, and it contains the rules for language use. This study compared the adequacy of the network, feature comparison, and schema theory approaches to semantic memory in predicting recall reaction times for new semantic concepts.

This study's methods controlled for subjects' associative history with stimulus concepts, so that all subjects had essentially the same exposure to stimulus concepts. Past semantic memory research failed to achieve this degree of control (McCloskey, 1980). A unique world of creatures was created so that experimenters could control the relationships between the creatures and their defining and characteristic features. Subjects learned the stimulus concepts and responded to recall test sentences from which their recall reaction times were generated.

Results supported schema theory predictions. Recommendations were made about schema theory's role in future semantic memory research. All three approaches were discussed in terms of their potential usefulness in developing a comprehensive semantic memory theory.