Abstract
A schema formation analysis was applied to listeners' abstraction of themes from melodic variations. A set of transformations of a prototypical melody was generated by systematic application of five transformational rules. The prototype represented the structural central tendency of the set. In Experiment 1 participants listened to the transformations and were tested for abstraction of the prototype in two ways; (a) they were asked to draw the melodic contour of 10 notes best describing the central tendency of the set of transformations, and (b) they were given a false recognition task in which they were asked to recognize the prototype and other transformations, none of which were in the original set, in a procedure analogous to that used by Franks and Bransford (1971). Both measures indicated that participants abstracted the prototype while listening to the original set of transformations. Experiment 2 replicated these findings with participants fully informed about the purpose and design of the experiment, a different task for recognizing the prototype (judging the similarity of transformations to the abstracted prototype), and a different prototype. In Experiment 3 participants who listened to the prototype produced data on the contour drawing and recognition tasks that were indistinguishable from those produced by participants who abstracted the theme from its variations.