A Comparison of Practice and Self-Report as Sources of Information About the Goals of Expert Practice

Abstract

A concert pianist recorded her practice as she learned the Italian Concerto (Presto) by J. S. Bach for performance, commenting on what she was doing as she practised. After the performance the pianist reported decisions made during practice on three basic dimensions (fingering, technical difficulties, familiar patterns of notes), four interpretative dimensions (phrasing, dynamics, tempo, pedal), and three performance dimensions representing features of the music attended to during performance (basic, interpretative, expressive). Number of features per bar served as predictor variables in regression analyses in which the predicted variables were number of starts, stops, and repetitions. Practice was divided into three separate learning periods. Practice was affected by basic dimensions in the first two periods and by interpretative dimensions in the last two periods, while performance dimensions affected practice throughout. The frequency of comments made while practising showed similar patterns, providing converging evidence for changes in the pianist's goals across the learning process. Practice and self-reports did not, however, entirely agree. Self-reports failed to mention practice of dynamics and indicated that selection of performance features occurred late in the learning process whereas practice data showed that both dynamics and performance features were practised right from the start. Practice sometimes provides information not available in self-reports.