Abstract
Expression is examined from the viewpoint of communication theory and it is argued that a proper understanding of expression involves an integrated description of both performance and perception. A framework is developed in which to couch a general theory of expression. As an example, a number of algorithms, implemented in Lisp are described which model the performance and perception of rubato. The model is based on two factors: 1) the use of “phrase final lengthening” to signal a group boundary and 2) the ability of the listener to track a variable tempo. The study shows that rubato is a rich source of information for the listener and that any realistic music parser must take this into account. On the other hand any performance model must take into account the constraints of perception.