This study is a comparison of the effectiveness of three approaches used to elicit expressivity in music students' performances: (a) aural modeling, (b) verbal instruction addressing concrete musical properties, and (c) verbal instruction using imagery and metaphor. Thirty-six college pianists worked with three melodies, one in each instructional condition. With each, subjects first gave a baseline performance, then received instruction for performing more expressively, and then gave a final performance. Subjects also verbally reported their thoughts during the process. Results confirmed that musicians can accommodate all three types of instruction used in the study and that each has strengths and weaknesses related to the characteristics of the music being performed and the musicians themselves. Additionally, analysis of the verbal reports suggested that musicians may use a cognitive translation process whereby they convert metaphor/imagery information into more explicit plans for changing the expressive musical properties of their performance (e.g., loudness, tempo, articulation). August 22, 2005 January 30, 2006