Art works, ontology of

2. Works as types

Let us consider works of literary and musical art. There are physical objects significantly associated with both genres and these are analogous to authentic objects in painting and sculpture: autograph copies of the text or score. But neither Austen’s autograph nor any other copy of the text can be regarded as identical with the work Emma, since no particular copy need survive in order for the work to survive. (Word processing makes the ontological irrelevance of an autograph particularly obvious.) Nor can the work be identified with the class of copies of its text or score. Classes are so defined that a class could not have had members different from those it does have. But there could have been more or fewer or different copies of Emma than there are, without the identity of the novel being threatened. With music, drama and other ‘performance’ works, there are performances to consider as well as copies of the score. But a sonata cannot be identified with its original performance by the composer – there may be no such performance – nor with the class of its performances: a symphony might have had different performances from the ones it did, or might never have been performed, without being a different or non-existent work.

For these reasons we may choose to identify literary and musical works with ‘types’, of which copies of their text or score are ‘tokens’. It is common to make a distinction between word- (or letter-) types and tokens, since the same word (or letter) may be inscribed many times; these inscriptions are tokens of the type. The same distinction applies to items of musical notation; we say there are a number of D♯ semiquavers on the page. Sentence-types are sequences of word-types (which are sequences of letter-types), and texts are sequences of sentence-types. On the view we are now considering, literary works are texts. The closest parallel to this idea for musical works is that they are scores, which are similarly defined as sequences of note-types, the tokens of which are particular inscriptions of notes. Since texts and scores as defined here are abstract structures, we may call an approach of the kind just described a version of structuralism.

One objection to this proposal is that it does not allow us to say that works are created by their authors or composers, since types are abstract objects not capable of being affected by human action; instead, what we normally think of as the act of composition would, according to the structuralist, be an act of discovery. This objection is not decisive. The structuralist need not deny any of the evident facts about composition: that hard work and talent are required for the composition of significant works; that without that talent and effort these works would not be available to us. We admire those who prove difficult and important mathematical theorems, and our admiration does not dissipate with the thought that these theorems are not created by the people that prove them. Whether the work of the artist is to be described as creation or as discovery can be counted as spoils to the victor in this debate.

Another, more serious objection, but one answerable within a generally structuralist outlook, is that the proposal does not make any provision for the obvious difference between ‘performance’ and ‘non-performance’ works, since it treats plays, symphonies and novels alike as having tokens which are particular inscriptions. We can rectify this by saying that musical works are sequences of sound-types rather than note-types. While note-types have as their tokens particular inscriptions, sound-types have as their tokens particular sounds. A sound-type might be identified by specifying a pitch, a duration and a degree of loudness. (To conform with our ordinary ways of individuating sounds we would need to preserve some vagueness in these specifications.) Thus defined, an instance of the work would consist in the actual production of tokens of the sound-types constitutive of the work; it would be a performance of the work. A comparable stipulation can be made for plays and other non-musical performance works.

There are facts in addition to mere performability which seem equally to demand reflection in our theory of the work. Musical works, at least typically, are intended to be performed on certain instruments, and the specification of the work in terms of sound-types alone fails to accommodate this: a sound-type can be produced in ways that would be inappropriate for many works. We could meet the performance-means objection by specifying that the work is a sequence of sound-types-as-performed-on-certain-instruments. But we must go further still. A person who produces the appropriate sequence of sounds on the appropriate instruments by accident, or while improvising, is not, strictly speaking, performing the work. We need to specify further that the work is a sequence of sound-types-as-performed-on-certain-instruments-as-a-result-of-intentionally-following-a-certain-score.