Philosophy Looks at the Arts
ed. Joseph Margolis
(Third Edition, 1987)
[229] hared ating nakes ch is ristic ivity the is it. de of the jeen ¡st It is • nO be tist be er- are est W- Id of 1-Toward an Ontology of Artworks
NICHOLAS WOLTERSTORFF
What sort of entity is a symphony? A drama? A dance? A graphic art print? A sculpture? A poem? A film? A painting?
Are works of art all fundamentally alike in their ontological status?
These are the questions to be discussed in this paper.
I A Phenomenology of the Distinctions Among Works of Art
In several of the arts there is application for the distinction between a performance of something and that which is performed. In music, for example, one can distinguish between a performance of Verklaerte Nacht and that which is thereby performed, namely, Arnold Schoenberg's work Verklaerte Nacht. Similarly, in dance one can distinguish between a performance of Swan Lake and that which is thereby performed, namely, the ballet Swan Lake.
Some people will be skeptical as to whether, in the cases cited and others of the same sort, we really do have two distinct entities--a performance and that which is performed. But assuming it to be true that the concept of a performance of something and the concept of something performed both have application to the arts, there are two sorts of considerations which force one to the conclusion that that which is performed on a given occasion is distinct from the performance of it.
In the first place, a thing performed and a performance thereof will always diverge in certain of their properties. For example, having been composed by Schoenberg is a property of Verklaerte Nacht but not of any performance of Verklaerte Nacht. On the other hand, taking place at a certain time and place is a property of every performance of Verklaerte Nacht but not of Verklaerte Nacht itself. It is worth noting that a work pertormed may diverge from performances thereof not only in 'ontological'
[230]
properties but also in 'aesthetic' properties. For example, it may be that having the voice part begin on A natural is not a property of any performance of Schoenberg's Pierrot Lunaire, though it is a property of the work itself, indeed, an essential one.
A second sort of consideration, one which is actually a specific applica- tion of the first, also leads to the conclusion that in certain of the arts one must distinguish between those entities which are performances and those entities which are works performed. This second sort of consideration hinges on applications of the concepts of identity and diversity. That which is performed on one occasion may be identical with that which is per- formed on another; George Sell, for example, may twice over have conducted a performance of Verklaerte Nacht. Thus, there may be two distinct performances of one single musical work. But two distinct things cannot each be identical with some one thing. Thus, the two distinct performances cannot both be identical with the work performed. But if one of them, call it A, was identified with the work performed, then the other, call it B, would, by virtue of being a performance of the work performed, be a performance of performance A. Not only that, but performance A would be capable of being performed on many other occasions as well. Both of these consequences, however, seem impossible.
Let us henceforth call a work of art which can be performed, a performance-work. Most if not all performance-works are universals, in that they can be multiply performed.
It would seem that performances in the arts are as correctly called "works of art" as are performance-works. The ontological status of per- formances is relatively clear, however, while that of performance-works is immensely perplexing. Performances are occurrences or events. They take place at a certain time and place, begin at a certain time and end at a certain time, last for a certain stretch of time, and have temporal parts in the sense that each performance is half over at a certain time, three-quar- ters over at a certain time, one-eighth over at a certain time, etc. But what sort of entity is a performance-work? That is something which we shall have to discuss in considerable detail. What should already be clear, though, is that performance-works are not occurrences (events). Thus, already we can answer one of our opening questions. Works of art are not all alike in their ontological status.
In certain of the nonperforming arts distinctions similar to the perfor- mance/ performance-work distinction have application. Consider, for ex- ample, graphic-art prints. Here, a commonly applied distinction is that between a particular impression and the work of which it is an impression; between, for example, the tenth impression of Obedient unto Death and the print of which it is an impression, namely, George Rouault's Obedient
[231]
unto Death. And consider those cases in which sculpture is produced from a mold. Here, a commonly applied distinction is that between a particular casting of, say, The Thinker and the sculptural work of which it is a casting, namely, Rodin's The Thinker. And consider thirdly those cases in the field of architecture in which many different buildings are produced according to one set of specifications. Here, a commonly applied distinc- ton is that between a given example of, say, the Tech-Bilt House No. and that of which it is an example, namely, the Tech-Bilt House No. 1.
It may be noticed that an impression of a work of graphic art, a casting of a work of sculptural art, and an example of a work of architectural art particular arts together. In order to have a convenient terminology, let us call the entities of which there can be impressions, castings, or ex- amples, object-works. And let us say that impressions, castings, and examples are objects of object-works. Thus, as a counterpart to the performance-performance-work distinction, we have the distinction be- tween impressions, castings, and examples on the one hand and object- works on the other.
The considerations which impel us to distinguish between an object- work and those entities which are objects thereof are parallel to those which impel us to distinguish between an entity which is a performance- work and those entities which are performances thereof. One consideration is again that of divergence in properties. For example, having a thumb- print in the lower left corner may be a property of a given impression of Rouault's print Obedient unto Death, or even of all impressions thereof, though it is not a property of the print Obedient unto Death. A second consideration is again to be derived from applications of the concepts of identity and diversity. For example, there can be two different castings of the same sculptural work; and neither both of these castings together nor either one singly can be identified with the work. In the case of object- works there is yet a third sort of consideration which may be adduced. one hinging on applications of the concepts of existence and nonexistence. Any one of the several objects of an object-work can be destroyed without the object-work thereby being destroyed. I could, for example, perform the horrifying operation of burning my impression of Rouault's Obedient unto Death, but I would not thereby put the print itself out of existence. Nor could I put the print out of existence by destroying any one of the other impressions, nor even by destroying the original etched plate.
It would seem that both object-works and the objects thereof are entitled to being called "works of art." Further, the ontological status of the latter is relatively unproblematic: They are physical objects. Of course, plenty of things about the nature of physical objects remains unclear. Yet we know
[232]
what they are, and it is clear that impressions, castings, and examples are to be numbered among them. But what is an object-work? What is its ontological status? That is something which we shall have to discuss in detail.
There remain literary works, films, and paintings to consider. A literary work can be both written down and sounded out. There can be both copies of it and utterances of it. Now, a copy is a physical object, whereas an uttering of something is a certain sort of event. Further, the copy of relation seems closely similar to the example of, the impression of, and the casting of relations. Accordingly, I shall say that a copy of a literary work is an object of it; and I shall add literary works to the group of entities to be called object-works. Furthermore, an utterance of a literary work is an event, very much like a performance. Accordingly, in the class of things to be called performance-works I shall include also literary works. Literary works, then, are both performance-works and object- works.
Saying this, however, makes one want to look back to see whether we do not have good ground for saying that works of music and drama are also both performance-works and object-works. In the case of dramatic works I think it is clear that we must say "No." A dramatic performance is a pattern of actions. The actions will in all but the most unusual cases include speech actions. But in all but the most unusual cases they will include other sorts of actions as well. More importantly, that pattern of actions which is a dramatic performance will always include actions of role-playing. For these reasons, a reading aloud or a recitation of the script of a drama is not yet a performance of the drama. A copy of the script for a drama is not a copy of the drama but instructions for proper performances thereof. The script may of course be a literary work in its own right. And that work can have both readings aloud and copies. But the drama is not the script. And a copy of the script is not a copy of the drama. The drama has no copies. All it has is performances. Dramas are only performance-works.
Music presents a somewhat less clear situation. The crucial question is this: Does a copy of the score stand to a work of music in a relationship similar enough to that in which a copy stands to a work of literature to justify us in calling the score-copy an object of the work? It seems to me not decisively clear one way or the other. What does seem clear is that a word can be both written down and uttered aloud, whereas a sound cannot be written down but only sounded out. The marks in a copy of a score are not instances of sounds but rather instructions for producing sounds. Of course, an instance of some sequence of words can also be treated as instructions for the utterance of that sequence; yet at the same time it is
[233]
genuinely an instance of those worlds. Some words, especially those in primitive cultures, are never written down; some, especially those in technical languages, are never sounded out. Yet most words have a dual manifestation. The same is not true of sounds. But suppose someone suggests that music should be thought of as being composed of notes rather than sounds, and then goes on to argue that notes, like words, can he both sounded out and written down. Obviously, this is a suggestion worthy of further investigation. Whether it is true or false is not at once clear. But nothing that is said hereafter will depend essentially on whether or not it is true. So I shall continue to suppose that music consists of sounds.
The film seems to have a dual status similar to that of words. One and the same film may have many copies, a copy being a physical object; and it may also have many showings, a showing being an occurrence (event). Thus, a film, like a literary work, has claim to being regarded as both an object-work and a performance-work. There is this difference worth not- ing, though: A showing of a film will always occur by way of the showing of a certain copy of the film, whereas the utterance of a literary work need not occur by way of the reading of some copy of the work. One can recite it from memory.
As for paintings, it seems that neither the object/ object-work distinction nor the performance/performance-work distinction has application, nor does it seem that any close counterpart to these distinctions has applica- tion. There is, of course, the distinction between the work and reproduc- tions of the work. But this is a quite different distinction, as can be seen from the fact that one can also have reproductions of each of the various impressions of a print. What is lacking in painting is any counterpart to the print/impression distinction. All one has is a counterpart to the impression/ reproduction distinction. The point may be put by saying that all the impressions of a print are originals, none is a reproduction. The conclusion must be that a painting is a physical object. But more will be said on this matter later in our discussion.
To say this is not, of course, to deny that reproductions of paintings along with reproductions of sculpture are, in some cases at least, entitled to being called "works of art" in their own right. So too are films, though they are for the most part 'reproductive' of performances and of visible events and objects. And so too are recordings, though most recordings are 'reproductive' of sounds and of audible performances. It is interesting to note, however, that in the case of visual-art reproductions and sculpture reproductions one again often has application for the print/ impression or the work/casting distinction, and that in the case of recordings (records) one can distinguish between the recording on the one hand and the various
[234]
discs of the recording on the other and, in turn, between a given disc on the one hand and various playings of the disc on the other.
Though I have called what we have done thus far phenomenology, what I have said is of course not free from ontological commitment. In saying that the distinction between performances and that which is performed can be applied to the arts, I said something which entails that there are performance-works. And I said that of most if not all of these it is true that they can be multiply performed. A thorough nominalist would deny that there are any multiply performable entities. Similarly, he would deny the existence of multiply-objectible*entities. I think it would be worthwhile to consider how the nominalist conviction that there are no such entities might most plausibly be developed; and I also think it would be worthwhile to consider whether any decisive arguments against such nominalism can be offered. But I shall not on this occasion attempt either of these. Rather, the question which I wish to discuss in detail is this; What is the ontological status of performance-works and object-works?
To simplify our terminology, I shall henceforth in this paper call only performance-works and object-works art works. And both performances of art works and objects of art works will be called examples of art works. I shall continue to use "work of art" to cover both art works and their examples, along with such things as paintings which are neither. Perhaps here is also a good place to remark that the fact that the perfor- mance/ performance-work distinction or the object/ object-work distinction applies to a certain art does not imply that it applies throughout that art. There may be works of that art which are neither. Those works of music, for example, which are total improvisations (as distinguished from those which are improvisations on a theme) are neither performances nor performance-works.
II The Sharing of Predicates and Properties Between Art Works and Examples
We cannot here discuss all competitors to the theory proposed in the following pages. But for an understanding of the theory, it will be usetul briefly to consider and put behind us the view that performance-works and object-works are sets of their examples. The untenability of this suggestion can be seen by noticing that whatever members a set has it has necessarily, whereas a performance-work or object-work might always have had different and more or fewer performances or objects than it does have; and by noticing that if set a has no members and set B has no members, then a is identical with B, whereas it is not the case that if art work y has no instances and art work & has no instances, then y is identical with 8.
[235]
That there is but one null set is clear enough. But that a set cannot have had a different membership from what it does have is a fact apt to be confused with related but different facts. The property, having been a disciple of Saint Francis, is a property shared in common by all and only the members of a certain set, that one, namely, whose members are all and only the disciples of Saint Francis. Let us for convenience name this set D. Now, whoever has the property of having been a disciple of Saint Francis has it only contingently. Accordingly, that set which is D might have been such that some of its members lacked this property; indeed, all might have lacked it. Alternatively, persons who are not members of D might have had this property. Thus, some other set than D might have been such that all and only its members have the property of having been a disciple of Saint Francis. But all these facts pertaining to what might have been in place of what is, are thoroughly compatible with the fact that D has its membership essentially.
To begin, consider some logical predicate which in normal usage can be predicated of two different things in such a way as to assert something true in both cases. Let us say that in such a case those two things share that predicate. One striking feature of the relationship between an art work and its examples is the pervasive sharing of predicates between the art work on the one hand and its examples on the other. "Is in the key of C minor" can be predicated truly of Beethoven's Opus 111 and also of most if not all performances of Beethoven's Opus III. "Has the figure slightly off-center to the right" can be predicated truly of Rouault's Obedient unto Death and likewise of most if not all impressions of Obedient unto Death. And so on, and on.
Of course, not every predicate which can be predicated truly of an art work or which can be predicated truly of examples of some art work, is shared by the work and its examples. "Is a performance" and "is an occurrence" are never shared, nor are "can be repeatedly performed" and "can repeatedly occur." Nor is "composed by Hindemith" ever shared. "Is thought about by me" is in some cases shared between a certain art work and all its examples, in other cases it is shared between a certain artwork and only some of its examples, while in yet other cases it is not shared between an art work and any of its examples. And "has 'no' as its third word" is unshared between the poem Sailing to Byzantium and my particular copy of it, whereas it is shared between the poem and most copies of it.
One naturally wonders, at this point, whether when a predicate is shared between an art work and some one or more of its examples, there is normally also a sharing of some property for which the predicate stands. If so, then the predicate is used univocally . On the other hand, if the predicate stands for two different properties but if there is some systematic
[236]
relation between these, then the predicate is used analogically . If not even this is true, the predicate is used equivocally . Shortly, we shall discuss this issue pertaining to properties. Meanwhile, without yet committing our. selves on it. let us see whether we can find some pattern in this pervasive sharing of predicates. (See also Wolterstorff [9]: 250-254.)
From the start, one feels that there is some connection between a predicate's being true of the examples of an art work and its being true or the work. Can this feeling be substantiated? The example we have already used provides us with evidence for concluding that the following formula will not do: A predicate P is true of some art work W if P is true of every example of W. For "is a performance" is true of all the examples of performance-works but cannot be true, in its normal sense, of any of those art works themselves.
So suppose that from here on we discard from consideration those predicates which are true of one or more of the examples of some art work but which, in their normal meaning, cannot be true of the work itself. (When a predicate P used with normal meaning cannot be true of W, P will be said to be excluded by W. Likewise, when a property p cannot be possessed by W, P will be said to be excluded by W.) What then about the formula: For any predicate P which is not excluded by W, P is true of W if P is true of every example of W? One objection to this formula is that it is far more constricted in its application than what we were looking for. For we saw that "has a G sharp in its seventh measure" may be true of Bartok's First Quartet even though of many of its performances it is not true. Indeed, it may be true of none of the performances.
A clue to a better formula can be gotten by looking more closely at this example. Is it not the case that "has a G sharp in its seventh measure" is true of Bartok's First Quartet in case it is impossible that something should be a correct performance of Bartok's First and lack the property of having a G sharp in its seventh measure? Is it not the case that has 'no' as its third word" is true of Sailing to Byzantium in case it is impossible that something should be a correct copy of Sailing to Byzantium and lack the property of having "no" as its third word?
These examples naturally suggest to us the following formula: For any predicate P which is not excluded by W, if there is some property being P which P expresses in normal usage and is such that it is impossible that something should be a correctly formed example of W and lack being P, then P is true of W.
But to this general formula as well, there are counterexamples. Consider for instance the predicate "is a performance or was highly thought of by Beethoven." There will be many works such that this predicate used in its
[237]
normal sense will not be excluded by the work. Likewise, it is impossible that something should be a correctly formed example of some such work and lack the property of being either a performance or highly thought of by Beethoven; for it is impossible that that thing should lack the property of being a performance. Yet the predicate in question may very well not he true of the work. For the work cannot be a performance, and it may not have been highly thought of by Beethoven. And in general, take a predicate of the form "is either A or anti-W," where "is anti-W» represents à predicate such that (i) it is excluded by W and (ii) when predicated of examples of W it stands for a property such that necessarily if something is an example of W, then it has that property, and where "is A" represents any predicate whatsoever which is not excluded by W. Then the disjunc- tive predicate' represented by "is either A or anti-W» is itself not excluded by W and is itself such that when predicated of some example of W, it stands for a property such that it is impossible that something should be a correct example of W and lack it. Yet obviously the predicate may very well not be true of W.
The essence of the difficulty here would seem to be that some predicates stand for properties such that it is impossible that something should be an example of W at all, correct or incorrect, and lack the property. Such properties might be said to be necessary to examples of W. If we could eliminate from consideration predicates standing for such properties, then counterexamples of the sort suggested will be forestalled. So let us say that a predicate P is acceptable with respect to W if and only if P is neither excluded by W nor is such that any property for which it stands when truly predicated of examples of W is one which is necessary to examples of W. Then our proposed formula becomes this: For any predicate P which is acceptable with respect to W, if there is some property being P which P expresses in normal usage and is such that it is impossible that something should be a correctly formed example of W and lack being P, then P is true of W. (It should be noticed that the claim here is not if and only if, but just if.)
The core feature of this proposal is the suggestion that what is true of correctly formed examples of an art work plays a decisive role in deter- mining what can be predicated truly of the work. Or, to put it yet more indefinitely, the core feature is the suggestion that the concept of an art work is intimately connected with the concept of a correctly formed example of the work.
Perhaps if we considered the matter in detail, we could find still more pattern to the sharing of predicates between artworks and examples than what we have thus far uncovered. But enough has been uncovered for our subsequent purposes. So let us now move from the level of language to
[238]
the level of ontology and consider whether, when predicates are shared according to the general pattern uncovered, there is also a sharing of properties designated by those predicates.
One is naturally inclined to think that there is. Our dictionaries do not, after all, tell us that a certain word standardly ineans one thing when truly predicated of an art work and something else when truly predicated of an example of the work. Yet I think that we must in fact come to the conclusion that predicates shared between art works and their examples do not function univocally when the sharing follows the general pattern we have uncovered. For what one means, in truthfully predicating "has 'no' as its third word" of some copy of Sailing to Byzantium is that the third word-occurrence is "no." But when one truthfully predicates "has no as its third word" of Sailing to Byzantium itself, one cannot mean this. For the poem does not consist of word-occurrences. Similarly, what one means in truthfully predicating "has a G sharp in its seventh measure? of some performance of Bartok's Fifth is that in its seventh measure there was an occurrence of the G-sharp pitch. But the Quartet itself does not consist of sound-occurrences. So I think it must be admitted that we have not discovered a systematic identity but only a systematic relation between the property designated by some predicate when it is truthfully predicated of some art examples and the property designated by that same predicate when truthfully predicated of the art work. Our conclusion must be that the sharing of predicates between art works and their examples pervasively exhibits analogical predication.
The situation is as follows. Suppose that P is a predicate which can be shared between an art work W and its examples, and suppose further that a property for which P stands when truthfully predicated of examples of W is being P. Then for those cases in which the sharing of P fits the general pattern which we formulated, P when truthfully predicated of W stands for the property of being such that something cannot be a correct example of it without having the property of being P.
III Art Works Are Kinds
We have seen some of the fundamental relations which hold between an art work and its examples. But we have not yet gained much insight into the ontological status of art works. We are left so far without any satisfying answer to our question: What is an art work? We must take a next step.
The proposal I wish to make is that performance-works and object- works are kinds (types, sorts)-kinds whose examples are the performances performance; an object-work is a certain kind of object.
[239]
A phenomenon which tends at once to confirm us in the suggestion that art works are kinds whose examples are the examples of those works is the fact that kinds which are not art works are like art works in just the ways that (as we saw earlier) sets of their examples are unlike art works. Just as an art work might have had different and more or fewer perfor- mances and objects than it does have, so too the kind Man, for example, might have had different and more or fewer examples than it does have. If Napoleon had not existed, it would not then have been the case that Man did not exist. Rather, Man would then have lacked one of the examples which in fact it had. And secondly, just as there may be two distinct unperformed symphonies, so too may there be two distinct unex- ampled kinds e.g., the Unicorn and the Hippogriff.
Not only does it seem that art works are kinds. What is even more striking is their many close similarities to those special kinds of kinds familiarly known as natural kinds.
It has long been noticed by philosophers that in the case of natural kinds there is a pervasive sharing of predicates and/ or properties between kinds and their examples. Let us look at the pattern of such sharing, beginning with a proposal made by Richard Wollheim. Having excluded from consideration those properties which cannot be shared between kinds, and examples, his suggestion is that the following is necessarily true: The K shares a certain property with all Ks if and only if it is impossible that something should be an example of the K and lack the property ([8]:64-65).
What must be clearly perceived about this formula is that it speaks of properties, not of predicates. And in many if not most cases, a sharing of a predicate does not have, underlying it, a sharing of a property for which the predicate stands. That property which a grizzly possesses, of being something that growls, is not a property which the Grizzly could possess. Once one sees this, it becomes clear that the formula has an extremely limited application. Cases of shared predicates are common. Cases in which those predicates stand for properties which can be shared are relatively uncommon. Thus, the formula gives very little insight into the relation between kinds and their examples. Perhaps it's true that all grizzlies growl. And certainly it is true that the Grizzly growls (though at the same time it's true that something can be an example of the Grizzly while being mute). Yet this is not a counterexample to the formula; because what the Grizzly's growling consists of cannot be identical with what a grizzly's growling consists of. The tenability of the formula with respect to such cases is bought at the price of giving us no illumination with respect to them.
Even so, however, there are rather obvious counterexamples of other
[240]
sorts to the formula, It could happen that a certain kind would share with al its examples the property of having been referred to by someone or other. Yet most kinds are such that something could be an example o them and still lack this property.
But now consider once again the sentence The Grizzly growls." Is it not the case that "growls" is true of the Grizzly if it is impossible that something should be a properly formed grizzly and not grow? A grizay muted is a malformed grizzly, and so also is grizzly born without a growi What makes "growls" true of the Grizzly is that something cannot be a properly formed grizzly unless it growls. In botanical and zoological taxonomy books, one is not told about the features shared by all examples of a certain kind, nor about the essential features shared by all examples of a certain kind, but about the features which a thing cannot lack if it is to be a properly formed example of the kind. So already we have for natural kinds the same pattern which we earlier uncovered for art works- For any predicate P which is acceptable with respect to the K, if there is some property being P which P expresses in normal usage and which is such that it is impossible that something should be a properly formed example of the K and lack being P, then P is true of the K.
As in the case of art works, we must raise the question whether when we have a sharing of some predicate between a kind and its examples, we also have a sharing of some property for which that predicate stands. With respect to those cases which fit our general formula, I think the answer must be, "No, we do not." Predications in such cases are not univocal. But neither are they equivocal. They are analogical. When grizzlies growl, they emit from their throats certain characteristic sound-patterns which we English-speaking people call "growling." But the kind, Grizzly, does not do that. Its sound-emission cannot be caught on some record. Yet-the Grizzly growls. "Growls," when truly predicated of the Grizzly, would seem to stand for the property of being such that something cannot be a properly formed example of its unless it growls. Thus, there is a systematic non-univocality about "growls." The predicate "growls" stands naturally for two quite different properties, one holding of the Grizzly and one holding of at least every properly formed example thereof. In general, for those cases in which the sharing of predicates between a kind and its examples follows the general pattern which we have formulated, the predicates are used analogically in exactly the way in which they were seen to be used analogically in the corresponding cases for art works.
In concluding this section of our discussion, let us articulate an impor tant assumption which we have been making throughout. Consider the kind: Red Thing. This does seem to be a genuine kind; it differs from the class of all red things in that it might have had different and more of
[241]
fewer members than it does have. But now notice that there cannot be a distinction, among examples of this kind, between improperly formed examples and properly formed examples of the kind. For it is not possible that some of the examples of the Red Thing should be improperly formed examples of this kind (i.e., things improperly red), nor is it possible that some should be properly formed examples of the kind (i.e., things properly red). Or consider the two kinds: Properly Formed Orchid and Malformed Orchid. There seems no reason to doubt that there are such kinds as these. But neither of these can have properly formed examples, nor can either have improperly formed examples.
When a kind, the K, is such that it is possible that it should have properly formed examples and also possible that it should have improperly formed examples, let us say that the K is a norm-kind. We have assumed throughout our discussion that art works and natural kinds are both norm-kinds.
IV What Kind of Kinds Are Art Works?
Having suggested that art works are kinds, we must now take the next step of considering which kind a given art work is to be identified with. For the sake of convenience, I shall conduct the discussion by referring exclusively to music. But I shall have an eye throughout on the application to artworks generally.
In performing a musical work, one produces an occurrence of a certain sound-sequence, the sound-sequence itself being capable of multiple occur- rences. Accordingly, that particular kind which is some musical work has as its examples various occurrences of sound-sequences. It is a kind whose examples are sound-sequence-occurrences.
But now more specifically, with which of those many kinds whose examples are sound-sequence-occurrences is a given musical work to be identified? An answer which comes immediately to mind is this:
(1) A musical work W is identical with that kind whose examples are occurrences of that sound-sequence which correct performances of Ware occurrences.?
About this it must be said, however, that for most if not all musical works there is no such sound-sequence. This is so, for one reason, because standards of correctness by no means wholly determine which sound-se- quence must occur if W is to be performed correctly. Performances of a given work can all be correct in every detail--yet differ significantly. The musical works which come closest to permitting no divergence among their
[242]
correct performances are those "totally serialized' works of the last quarter century. Yet even these permit some divergence.
So 1) must be discarded. And the revised suggestion we should consider is this:
(2) A musical work W is identical with that kind whose examples are occurrences of members of that set of sound-sequences which correct performances of W are occurrences of.
But this suggestion, though it is better than (1), is also not satisfactory. On this view, an incorrect performance of some work W would not be an example thereof. Yet an incorrect performance of some work is, in spite of its incorrectness, a performance thereof. And is it not on that account an example of the work? Are not all performances of W to be counted among the examples of the kind with which W is identical?
So (2) must be revised by dropping the reference to correct performance, this:
(3) A musical work W is identical with that kind whose examples are occurrences of members of that set of sound-sequences which performances of W are occurrences of.3
But now we must in turn raise a question concerning the satisfactoriness of (3), a question which plunges us into a whole nest of subtle matters. In performing a musical work one produces an occurrence of a certain sound- sequence. It is clear, however, that that very same sound-sequence can in principle occur in other ways than by way of someone performing that work-or indeed, any work. It can be made to occur, for example, by the blowing of the wind, or by someone's doodling on a piano, or by an elec- tronic organ's going berserk. Performing is one way of producing occur- rences of sound-sequences. But the very same sound-sequence which can be produced by the activity of performing can be produced in other ways as well. So a performance is not merely an occurrence of a certain sound- sequence. It is an occurrence produced by the activity of performing.
The question to consider now is whether a musical work is just a certain kind of sound-sequence-occurrence, no matter how produced, or whether a musical work is a certain kind of performance. Is (3) the formula we want, or the following:
(4) A musical work W is identical with that kind whose examples are performances of W (or, more simply stated, with the kind: Per- formance of W)?
[243]
To answer this question we shall have to look a bit into the nature of performing.
Performing is clearly an intentional act. But what is the nature of the intention involved? When someone performs Beethoven's Opus 111, what is it that he intends? Does he perhaps intend to follow the directions of the score in producing a sound-sequence-occurrence? Well, often he does indeed have this intent. But even success in this intent is seldom a sufficient condition for having performed the work. For though scores do, among their other functions, provide specifications for producing examples of the work, seldom are all the matters pertaining to correct performance specified in a score. Naturally, many are simply presupposed by the composer as part of the style and tradition in which he is working, and others are suggested without ever being specified. If the performer limits himself to following the specifications in the score, not even attempting in other respects to produce a correct example of the work, it is at the very least doubtful that he has performed the work.
An even more decisive objection is that one can perform some work without at all intending to follow the specifications of the score for the work. For there may be no score. If Beethoven had composed his Opus 111 before scoring it, it would have been possible for him to perform the work without being guided by the score for that or any other work. That is, it would have been possible for him not merely to produce an occur- rence of a sound-sequence which could also occur as a performance of Opus 111, but actually to perform the work. In fact, of course, there now is a score for Opus 111. But the vast bulk of indigenous folk music remains unscored. So in that case performers are never guided by the scores for the works. Yet those works can be performed.
It is true that specifications can be laid down to performers by means other than scores. The folk-music performer can be told verbally how some passage is to be performed. Or the rhythm can be stomped out for him by foot. But quite clearly it is no significant improvement over our first thought to say that the intention involved in performing a work is the intention to follow the specifications for producing examples of that work. For of most works it is true that even when we include all specifications, whether expressed in score notation or otherwise, these specifications are woefully insufficient for determining correct examples. Bartok, in his early career, set about scoring various Hungarian folk songs. His work did not consist of taking specifications which were expressed in something other than the Western scoring system and expressing them in score notation. On the contrary, his work consisted of providing those works, for the first time in their careers, with specifications for performance. And by and large our Western scores contain all the specifications we have for our musical
[244]
works. But these are not enough. Correct performance requires knowledge of more matters than these.
What emerges from all this is that to perform a work one must have knowledge of what is required for a correct example of the work; and one must then try to act on such knowledge in producing an occurrence of a sound-sequence. It is the producer's acting on his knowledge of the requirements that makes of some sound-sequence-occurrences a perfor. mance of the work, instead of merely an occurrence of a sound-sequence that might also have occurred in a performance of the work. Such knowledge may be gained in many ways--for example, from scores or other specifications. But seldom will it be gained wholly from specifications which have been expressed. And sometimes it may not be gained from these at all.
It is not necessary, though, if one is to perform a work, that one succeed in one's attempt to act on one's knowledge of what is required of a sound-sequence-occurrence if it is to be a correct example of the work Even if one makes mistakes and so does not actually produce a correct example, still one may have performed the work. Of course there are limits, albeit rather indefinite ones, on how seriously one can fail and still have performed the work.
So my suggestion concerning the nature of performing is this: To perform a musical work W is to aim to produce a sound-sequence-occur- rence in accord with one's knowledge of what is required of something if it is to be a correct example of W, and to succeed at least to the extent of producing an ample of W.5
An implication of this understanding of the nature of performing is that when a performer deliberately departs from the requirements for a correct performance of some work W, he is then not performing W. Sometimes such departures are motivated by the performer's inability to negotiate some passage. In other cases they are motivated by the performer's belief that he can thereby produce an aesthetically better performance. But whatever the motive, if Anthony Newman, say, deliberately departs trom what is necessary for a correct example of Bach's Prelude and Fugue in D minor, then (strictly speaking) he is not performing that work. He is probably instead performing Newman's variation on Bach's Prelude and Fugue in D minor.
We have been discussing at some length the concept of performing. It is time now to return to the question which led us into this discussion, namely, which is the correct view as to the nature of the musical work, (3) or (4)? Is a musical work just a certain kind of sound-sequence-occur- rence, or is it a certain kind of performance?
What is worth noticing first is that on both (3) and (4), art works can
[245]
be viewed as norm-kinds. That is evident in (4), but probably not so on (3). For on (3), W is just a kind whose examples are the occurrences of a certain set of sound-sequences. And sound-sequence-occurrences just are or are not occurrences of that sequence of which they are occurrences. It makes no sense to speak of some as correct occurrences thereof and some as incorrect occurrences. However, there seems to be nothing against thinking of certain of the 'member' sound-sequences as correct ones, and the rest as incorrect ones. Then correct examples of the work will be those which are occurrences of the former; incorrect examples will be those which are occurrences of the latter. Thus, the fact that art works are norm-kinds is as compatible with (3) as with (4).
What is also worth noting is that on both (3) and (4) those properties which a work has by virtue of what counts as correctness in examples belong to it essentially. A work which has a G natural in its seventh measure cannot fail to have had a G natural there, on pain of not being that work.
But in addition to the similarities, there are significant differences between the kinds which (3) proposes to identify with art works and those which (4) proposes to identify with them. So let us contrast some of the implications of each of these views. If (3) were true, then one would have to distinguish between examples and performances of works. As a matter of contingent fact, it might be that only performances of W were examples of W. But there would be no necessity in this. And as a corollary, on (3) one could hear Bartok's Second Sonata without anyone performing it. For, presumably, by hearing an example of the work one can hear the work; and on (3) a work can have nonperformance examples. Also, on (3) one could hear a work W without hearing a performance of W, by hearing a performance (correct or incorrect) of a distinct work W'. Further, one could hear several different musical works by listening to a performance of just one work. On (4), however, none of these results obtain.
Furthermore, if (3) were true, it would be possible for W to have among its examples sound-occurrences which are also examples of W'. For a sound-sequence which could occur as a correct (or indeed incorrect) per- formance of W might also occur as an incorrect performance of W'. One's first inclination is to say that on this point too, (3) and (4) differ. On (4), it Would seem, distinct works cannot share any examples. But in fact this is false. On either view, musical works W and W' are identical if and only if whatever is necessary for something to be a correct example of W is also necessary for something to be a correct example of W', and vice versa. And on either view, something is a performance of a work W only if its pro- duction is guided by W's requirements for correct examples. But it seems clear that a given performance can be a performance of two distinct works;
[246]
and so also it seems clear that the production of a given performance can be guided by the requirements for correct examples of two works at once. This can happen in two sorts of ways. If works W and W' are related in such a way that whatever W requires for correct performance W' also does, but not vice versa (there being matters which W settles in terms of cor- rectness but which W' leaves optional), then, though W and W' are distinct works, whatever is a performance of W will also be one of W'. But even when works do not 'overlap' in this way, a group of performers, or even a single performer, can perform two works at once. This indeed is required by some of the works of Charles Ives. To perform them, hymns, folk songs, and patriotic songs must all be performed concurrently. Thus, on (4) as well as on (3), distinct works can share some, and even all, examples. However, on (4) the only cases of shared examples between two works W and W' will be those in which a performance is guided both by the criteria for correct examples of W and by those for correct examples of W'. And this limitation does not hold on (3).
But though there are a number of significant differences between the conception of a musical work offered us by (3) and that offered us by (4), none of the differences thus far pointed out seem to provide a solid reason for preferring either of (3) or (4) to the other. On some issues, (3) may give us a more 'natural' understanding; on others, (4). But on none of these issues does either seem to give us a clearly mistaken understanding. And though the implications of the two views diverge on more issues than we have cited. I do not think that these other divergent implications yield any more decisive reason for preferring one view to the other. Though the kinds with which each view proposes to identify a given musical work both exist and are definitely distinct from each other, yet it is simply not clear which of these different kinds is identical with the work. The situation seems to be that when we refer to and speak of what we regard as art works, we, in all likelihood, do not with definiteness mean to pick out entities of either sort as opposed to those of the other. For on most matters entities of these two sorts do not differ. And the matters where they do differ are so far off on the edge of our normal concern in the arts that we have never had to make up our minds as to which of these sorts of entities we intend to be dealing with.
It is possible, of course, that future developments in the arts will force us to make up our minds. And perhaps those developments are already upon us. For example, John Cage's work 4 feet 33 inches requires no particular sorts of sounds, or sound-sequences whatsoever. There are requirements for a correct performance of the work; namely that the pianist keep his hands poised above a piano keyboard for 4 minutes and 33 seconds. And there are sounds to be listened to, namely, the sounds
[247]
produced by the audience as they gradually realize what is being per- petrated. But the requirements include no specifications concerning sounds whatsoever. Thus, it is obvious that (3) simply lacks application to this case. (4), though, is still relevant. And if works such as this are eventually regarded as works of music, a decisive shift away from (3) toward (4) will have occurred.6
V What Is It to Compose?
On either of the two views we have been considering, a composer of a musical work can be thought of as one who determines what constitutes correctness of performances of the work. And such determination of correctness has in turn two phases. The composer must think of, or consider, the correctness-conditions in question. And in addition, since in the course of composing a work he normally considers a great many more such than he actually settles on, he must certify these as those he wants. This much is essential to being a composer.
But normally a composer does more than determine what constitutes correctness of performance. Normally he also produces a score. Now, as we remarked earlier, a typical function of a score is to provide specifica- tions for producing (correct) examples of the work associated with the score. Yet it does seem possible for a composer to determine a set of correctness-conditions which he knows to be impossible of being followed by any present performers on any extant or anticipated instruments. And it does seem possible for the composer to score such a work. Of course, he would not expect to hear it, and neither would he think of his score as providing specifications for performances. The score would just be the composer's record of his determination. It seems in fact that this is what every score is. Most scores function and are meant to function to guide performances. But what is true of every score is that it is a record of the artist's determination of correctness-conditions. If the record is in addition publicly legible, then it can also serve to communicate to others a knowledge of the conditions, and thus of the work. Musical afficionados can then become acquainted with the work by reading the score, and some may even thereby get some enjoyment from it.
Typically, then, there are at least these two activities involved in com- posing a work of music: The artist determines what constitutes correctness of performance, and he makes a record of his determination. Normally, of course, these two activities do not take place in neat separation. But sometimes they do. Mozart said that he imagined whole symphonies in his head. Housman said that he imagined poems while shaving. And concerning such cases, a question to consider is whether thereby an art
[248]
work has been composed. Can one compose "in one's head'? Well, one can determine correctness-conditions in one's head, but one cannot in one's head record the determination. Thus, depending on whether one regards recording one's determination as necessary for composing a musical or literary work, different answers will be given.
A consequence of what we have been saying is that two people car compose the same work. For surely it is possible for two people to determine and record the same correctness-conditions. Thus, Beethoven's Opus 111 is not necessarily just an opus of Beethoven. Indeed, it is not necessarily an opus of Beethoven at all. So also, the same musical work can in principle be known and performed in two different and independent cultures.
Does the artist, by composing his musical work, thereby also create it That is, does he bring it into existence? If (4) is the correct theory as to the nature of the musical work, so that a musical work is a performance- kind, then it certainly seems plausible to hold that he does, at least if composing is understood as consisting just in the determination of cor. rectness-conditions. For on (4) there can be no examples of a work which are not performances thereof. And to perform the work, one must know what is required of something if it is to be a correct example. But no one can know what these correctness-conditions are until they have been determined by someone or other. And for someone to determine the correctness-conditions is just to compose the work. On the other hand, composing the work would seem sufficient for bringing it into existence. Certainly it does not seem necessary that it also be performed. For there seems nothing contradictory in the notion of unperformed musical works.
But if (3) is the correct theory as to the nature of a musical work, it is not plausible to hold that in composing one creates. For on (3) a work may have examples which are not performances. And so there is nothing to prevent its having examples before performances have been made possible by the determination of correctness-conditions. But, surely, if there are examples of a work at a given time, then the work exists at that time. Thus, if (3) is correct, composing a work cannot in general be viewed as bringing it into existence.
But what, then, on (3), are the existence criteria for musical works? One possible view is that there exists such a work as W at time t if and only if W is being exemplified at t. But this view has the implausible conse- quence that musical works not only come into existence and go out, but that most of them exist intermittently. For on this view, the work exists when and only when it is being exemplified. And rare is the musical work which has no pair of exemplifications such that there is some time, between the occurrence of members of the pair, when the work is not being exemplified.
[249]
An alternative view would be that there exists such a work as W if and only if W is being exemplified or has been exemplified. But this view has the consequence that there cannot be a musical work which has not been exemplified. In fact, however, the contemporary literature concerning music is filled with the laments of composers whose works go unex- emplified.
So perhaps the best view is that a musical work W exists just in case it is possible that there be an exemplification of W. For this view has none of the untoward consequences of the other views. On this view, it would not be possible for a work to be composed but not exist; on this view, a work would not cease to exist when all exemplifications ceased; and on this view, there could be unexemplified works.
It should be noticed, though, that on this view musical works exist everlastingly. For if it is ever possible that there be something which is an example of Opus 111, then it is always possible. Neither by composing nor by any other activity on his part does a composer bring his work into existence. Rather, if (3) is correct, the composer should be thought of as a selector rather than as a creator. To compose would be to select a certain kind of sound-occurrence. The only thing a composer would normally bring into existence would be a token (copy) of his score. Creation would be confined to token creation. Furthermore, since the selection of the work occurs in the process of determining its correctness-conditions, such deter- mination would have to be viewed as consisting in discovering the condi- tions. By contrast, on the view that the artist brings his work into existence by composing it, determination of the conditions for correctness can best be thought of as consisting in devising the conditions.
It must be admitted that there is something odd in thinking of musical works as existing everlastingly, waiting to be selected and recorded. But perhaps the correct view is that though the entity which is a musical work exists everlastingly, it is not a musical work until some composer does something to it. If so, then in answer to the question, "What must be done to a kind in order to make it a musical work?" one can take one's stand at at least two different points. One can hold that it is not a musical work until someone has determined its correctness-conditions, or one can hold that it is not a musical work until its correctness-conditions have been recorded as well as determined.
VI How to Tell Correctness
Our discussion concerning the ontological status of art works has con- centrated on music. The detailed application to the other arts of the points we have made can be left to the reader. But two final matters must be considered. One is this: Why can paintings and sculptures not be viewed
[250]
as single-exampled kinds rather than as physical objects, thereby giving us as sinied theory" of art works? P. F. Strawson, after saying that in"i Pertain sense, paintings and works of sculpture" are types, adds Thi footnote:
The mention of paintings and works of sculpture may seem absurd Are they not particulars? But this is a superficial point. The things the dealers buy and sell are particulars. But it is only because of the empirical deficiencies of reproductive techniques that we identify these with the works of art. Were it not for these deficiencies, the original of a painting would have only the interest which belongs to the original manuscript of a poem. Different people could look at exactly the same painting in different places at the same time, just as different people can listen to exactly the same quartet at different times in the same place ([6];231.)The situation is not quite as Strawson represents it, however. Of course there is nothing impossible in a certain object-work's having but one object. But object-works are norm-kinds, and being such they have as- sociated with them certain requirements for something's being a correct example of the work. What is different in the case of paintings is that there are no such associated requirements. There simply are no require- ments for something's being a correct example of some kind of which The Odalesque is the premier example. Of course, one can pick out things which to a certain close degree resemble this painting. There is a kind corresponding to them, and the painting is an example of it. But this is not a norm-kind, and none of our names of paintings are names of such entities.
Secondly, a question which has been pressing for a long time is this: How do we tell what constitutes a correct example of some art work? By now, however, the question has almost answered itself. In the case of works produced by some artist, the answer is that we try to discover the relevant features of that artifact which the artist produced (or which he arranged to have produced) as a record of his selection and as a guide or production-item for the making of examples. Of course, we will often discover that we cannot find out with any surety what the relevant features of that artifact were (are). We may no longer have the poet's original copy of the poem nor any very reliable evidence as to what it was like in crucial respects. Or we may have several copies from the poet's hand and not know which he authenticated. Or we may have an original authenticated copy but it may contain mistakes made by the poet, and we may find it impossible to determine which of various possibilities he had in mind. Or
[251]
we may have an original, authenticated, and correct copy, but we may no longer know how to interpret all the symbols. In all such cases and many others we simply have to acknowledge that we are to some extent uncer- tain as to what constitutes a correct example. To that extent, we are also uncertain as to the character of the work. Yet it is clear what we must look for: the features of that original artifact.
In the case of those art works sustained in the memory of a culture and for which there is no artifact functioning as guide or production-item, we simply have to find out what the culture would regard as a correct and what it would regard as an incorrect example of the work-which is the same as finding out what the culture takes the art work to be like in those respects.?
References
...
Notes
...