Two experiements were run to examine the nature of communication between performer and listener. The aim of experimente 1 was to test the generative hypothesis, according to which boundaries between more important hierarchical groupings are emphasized through the use of timing and dynamics. Subjects listened to a performance and were asked to segment it twice according borh to the highest and lowest hierarchical level of grouping structure. Results indicate that timing and dynamics alone do not determine perceived segmentation as predicted by the generative hypothesis. In experiment 2, listenes assessed different performances using a set of adjectives. Adjectives were grouped into four categories alluding to (1) music structer; (2) emotions; (3) dramatic structue, and (4) kinetics. Results suggested that aspects of each performance could be represented by adjectives in all four categories, but the category of emotions was the most salient. Results are discussed accoridng to an interpretative approach to musical performance