ABSTRACT
Returning to Figure 1, this combination of piece-specific and more general knowledge can be regarded as the basis for two parallel paths: one is the construction of a motor program, which contains the abstract specification for all the motor activities required to perform the piece. This clearly draws principally on the piece-specific component of the knowledge representation, though general knowledge of musical structure is involved here too (Shaffer, 1981). The second path from the knowledge representation leads to a set of generative principles which take this structural, stylistic and aesthetic knowledge as their input and give a set of expressive modifiers as their output. These modifiers act on the first two motor programs which are conceptually distinct though fused in reality. This first form, labelled ‘canonical’ in the figure, contains all the information necessary for an accurate rendition of the piece, but not sufficient for an expressive performance. This can be thought of as the motor programming equivalent of the information contained within the score, but structurally parsed, and represented in terms of the requirements for action. The expressive modifiers, which are the output from the generative expressive principles, act on this canonical program so as to give rise to a second version of the program, labelled ‘expressive’ in the figure, which contains the necessary and sufficient information for an expressive performance of the music. As noted above, these two versions of the motor program are only conceptually distinct: a number of authors, going right back to Seashore (1938/67), have shown that performers are unable to produce performances that are devoid of expression, demonstrating either that a canonical program does not exist as an independent step in the process, or at the very least that it has no access to the later stages of the production system other than via an expressively modified version.
