Excerpted from The Aesthetic Understanding, Methuen (1983). Reprinted by permission of Cambridge University Press.
Music may be used to express emotion, to heighten a drama, to emphasize the meaning of a ceremony; but it is nevertheless an abstract art, with no power to represent the world. Representation, as I understand it, is a property that does not belong to music.
The word ‘representation’ has many uses, and may often be applied to music. Therefore I shall discuss not the word, but the phenomenon, as it occurs in poetry, drama, sculpture and painting. Being common to both painting and poetry, this phenomenon cannot be identified with the semantic properties of a linguistic system, for painting, unlike poetry, does not belong to such a system. How, then, is it to be characterized? I suggest the following five conditions, not as an analysis, but as a partial description of the aesthetic significance of representation:
1 A man understands a representational work of art only if he gains some awareness of what it represents. His awareness may be incomplete, but it must be adequate. He may not see Masaccio's Tribute Money as a representation of the scene from the Gospel; but to understand it as a representation he should at least see the fresco as a group of gesturing men. If a man does not see the fresco in some such way—say because he can appreciate it only as an abstract arrangement of colours and lines—then he does not understand it.
2 Representation requires a medium, and is understood only when the distinction between subject and medium has been recognized. Merely to mistake a painting for its subject is to misunderstand it; so too is there misunderstanding when a man is unable to extract the features of the subject from the peculiarities and conventions of the medium. (A varnished painting of a man is not a painting of a varnished man, however much it may look as though it were.)
3 Interest in a representation requires an interest in its subject. If an interest in the Masaccio depends in no way upon an interest in the scene portrayed, then the fresco is being treated not as a representation but as a work of abstract art.
4 A representational work of art must express thoughts about its subject, and an interest in the work should involve an understanding of those thoughts. (This is an ingredient in condition 3.) I mean by ‘thought’ roughly what [Gottlob] Frege meant by ‘Gedanke’: the sense or content of a declarative sentence. In this sense thoughts may be spoken of as true or false, although of course it is not always the truth-value of a thought that is of interest in aesthetic understanding. It is clear that a representational work of art always conveys thoughts, in this Fregean sense, about its subject. Among the thoughts that give rise to my interest in King Lear, and which give a reason for that interest, are thoughts about Lear. These thoughts are communicated by the play, and are common property among all who understand it. Something similar occurs in the appreciation of a painting. Even in the most minimal depiction—say, of an apple on a cloth—appreciation depends on determinate thoughts that could be expressed in language without reference to the picture; for example: ‘Here is an apple; the apple rests on a cloth; the cloth is chequered and folded at the edge.’ Representation, in other words, is essentially propositional.
5 Sometimes we feel that a work of art is filled with thought, but that the thought cannot be detached from the work. It is impossible to put it into words (or into other words). Such cases, I should like to say, are cases not of representation but of expression. Why I should make such a distinction, and why I should make it in that way, will be apparent later.
6 Interest in representation may involve an interest in its lifelike quality; but it is not, for all that, an interest in literal truth. It is irrelevant that the depiction be inaccurate; what matters is that it be convincing. To require accuracy is to ask for a report rather than a representation.
I shall rely on an intuitive understanding of these conditions: they tell us what it is to treat something as a representation, rather than as a report, a copy, or a mere inarticulate sign. On this account, what makes a passage of prose into a representation is not so much its semantic structure as the specific intention with which it is composed. The semantic structure is relevant only because it provides the means whereby that intention is fulfilled. Representational literature is literature written with the intention that conditions 1–5 should be satisfied. Thus one may treat as a representation something that is not a representation; one may achieve representation by novel means; one may create a representation that is never understood, and so on.
Now some philosophers—those who think that music is a language—will give an account of musical representation on the model of description in prose or verse. But such an approach is surely most implausible. Anything that we could envisage as a semantic interpretation of music (a theory of ‘musical truth’) would deprive music of precisely the aesthetic aims for which we admire it, turning it instead into a clumsy code. Furthermore, all attempts to explain music in such terms end by giving rules of reference without rules of truth. We are told that a certain passage carries a reference to love; but we are not told what the passage is supposed to say about love. And to speak of language where there is ‘reference’, but no predication, is simply to misuse a word. We are in fact leaving the realm of representation altogether and entering into that of expression. But there is no need to prove that music is a language in order to assign to it the expressive properties that are mentioned, for example, by Deryck Cooke.
A better attempt to prove that music is a representational medium begins by comparing music to painting. It can be said with some truth that music, like painting, may deliberately ‘imitate’, or ‘copy’ features of an object. Is this not, then, a kind of representation? Examples are familiar: Saint-Saëns' Carnaval des Animaux, ‘Gretchen am Spinnrade’, La Mer. And it is natural to consider such pieces as attempts to ‘depict’ the objects referred to in their titles. But perhaps what is meant here by ‘depiction’ is not what is meant when we refer to the visual arts. A few observations about painting will therefore be appropriate.
It is a commonplace that depiction is not simply a matter of resemblance. Nor is it enough that the resemblance be intended; nor even that the artist should intend the resemblance to be noticed. No doubt Manet intended us to notice the resemblance between his Olympia and Titian's Venus of Urbino. But that is certainly not a case of one painting representing another. It is for such reasons that we might wish to lay the burden of our analysis of depiction on the notion of an ‘aspect’. The artist intends that the spectator should see the painting as its subject, not merely that the spectator should notice a resemblance between the two. In other words, the painter intends that we should have the experience of a certain aspect—that we should feel that seeing his painting is importantly like seeing its subject—and not merely that we should notice a resemblance. Thus a painter may intend to copy the Mona Lisa, but he does not (as a rule) intend that his painting should be seen as Leonardo's; rather, he intends that it should be seen as the woman in Leonardo's painting. On this view, the intention in depicting is not to ‘copy’ an object, but rather to create a certain visual impression. And surely, it will be argued, precisely the same process, and the same intention, may exist in writing music. Sounds are created which are meant to be heard as other things, as the babbling of brooks, the warbling of birds, the roaring and plodding of animals. Ö
However, a difficulty now arises. …
Representation can be begun … only where it can also be completed. If music is to be representational, then its subject must be not only picked out, but also characterized. But that requires a context, and in music the context seems to add no further precision to the ‘representational’ parts. A certain passage in Der Rosenkavalier ‘imitates’ the glitter of a silver rose. But what more does this passage say about the glitter, except that it is a glitter (and even that may go unnoticed)? The context adds nothing to the thought, and while there is musical development, the development of a description seems scarcely to be in point. So too, when the imitation of birdsong in [Olivier] Messiaen is given musical development, there is no thought about the birdsong which is made more determinate by that process. The birdsong is absorbed into the musical structure and takes on a meaning that is purely musical. But, it might be said, does not the music none the less convey a thought about the birdsong, in the sense of a purely musical thought? Why should it matter that the thought cannot be put into words? Such a retort gets us nowhere. For whatever is meant by a purely musical thought, we can envisage also a purely painterly thought—a thought that finds its only expression in lines and colours, but which cannot be put into words, and which consequently cannot be regarded as true or false. And it is part of the point of calling painting a representational art that the thoughts involved in its appreciation are not all purely painterly, that, on the contrary, an experience of a painting will involve thoughts about its subject, thoughts that could be put into words. This ‘narrative’ element is an essential feature of the phenomenon of representation. If we insist none the less that there is a type of ‘representation’ that is purely symbolic (which contains ostension but no description) then we are simply denying the role of representation in aesthetic interest.
It is true, all the same, that I may hear a passage of music as something that I know it not to be. I may hear a passage as forest murmurs, for example, as rushing water, as an approaching or receding horse. Should we lay any emphasis on this phenomenon? One problem is that a man may hear and appreciate ‘representational’ music without hearing the aspect. And while it is true that I may also hear poetry without knowing what it says (as when I listen to the reading of a poem in Chinese), to do so is not to appreciate the poem as poetry. An interest in poetry is not an interest in pure sound; a genuine interest in music, on the other hand, may by-pass its representational pretensions altogether. Therefore we cannot assume that a composer may sit down with the honest intention of creating a piece to be heard as, say, the quarrel between Mr Pickwick and Mrs Bardell's lawyers. For he cannot be sure that it will be heard in that way; his intention is vitiated, and must be replaced by what is at best a hope or a wish. If the intention endures none the less, it is because there is available to the composer some independent way of specifying his subject: for instance, through the words of a song, or through action on the stage. Thus, in the more adventurous attempts at representation, such as we find in the symphonic poem, the composer is apt to depend on a specific literary reference in order to secure the hearer's complicity in what is better described as an imaginative endeavour than as an inevitable perception. It is thus with Don Juan and Don Quixote, with Taras Bulba and the anecdotal works of Charles Ives.
The argument is of course by no means conclusive. But certain facts are significant all the same. It is significant, for example, that, while a man may look at an untitled picture and know immediately what it represents, it is most unlikely that he should do the same with an untitled symphonic poem. Significant too is the indefiniteness of the relation between music and its ‘subject’: the music does not determine some one natural class of interpretations, and can usually be fitted to widely contrasting themes. A quarrel between Mr Pickwick and a lawyer may be ‘represented’ by music that serves equally well the purpose of ‘depicting’ a forest fire. We see this ambiguity evidenced in the ballet, where the action is usually left so far indeterminate by the music that several incompatible choreographies may exist side by side as accepted members of the repertoire; as in The Rite of Spring. Hence, while the aspect of a painting, and the meaning of a sentence, are publicly recognized facts, which make possible the intention characteristic of representational art, there are no similar facts to enable the intention to be carried over into the realm of music. Ö
When we learn of a piece of music that it is supposed to represent something, then its ‘auditory aspect’ (the way it sounds) may change for us, even when what is ‘depicted’ is not a sound. On learning of its subject we may come to ‘hear it differently’, despite the fact that the subject is not something audible. Consider [Claude] Debussy's prelude, Voiles, which may be said to depict the slow drift of sails in a summer breeze. Learning that, I may begin to hear in the musical line a leisurely and day-dreaming quality that I did not hear before, as though I were watching the to-ing and fro-ing of sails on a calm bright sea. But here, of course, what is ‘depicted’ is not something heard. May we not say, all the same, that we hear the music as the drifting of sails?
Even if we grant the force of those remarks, however, we find ourselves facing another, and yet more serious objection to the view that there is ‘representation’ in music. The objection is that one can understand a ‘representational’ piece of music without treating it as a representation, indeed, without being aware that it is supposed to have such a status. On the other hand, the very suggestion that one might understand—say—Raphael's St George (National Gallery of Art, Washington) while being indifferent to, or ignorant of, its representational quality, is absurd. To suggest such a thing is to suggest treating the Raphael as a work of abstract art; it is to ignore the feature of representation altogether, because it is thought to be insignificant, or because it is thought to play no part in aesthetic interest. But to take such a view is simply to dismiss the problem. If I recognize the existence of a problem about music it is partly because I think that there is an aesthetically significant notion of representation employed in the discussion and enjoyment of painting.
Now someone might object to the view that one cannot both understand the Raphael and also have no knowledge of its subject. He might claim that at least a partial understanding of the painting could be achieved by studying it as a piece of abstract art. One may understand the composition of the painting, he will say, the balance of tensions between ascending and descending lines, the sequence of spatial planes, and so on, and in none of this need one have an awareness of the subject. But such a reply is wholly misguided. For it seems to suggest that these important aesthetic properties of the Raphael—composition, balance, spatial rhythm—are quite independent of the representation; whereas that is clearly not so. For example we perceive the balance between the upward thrust of the horse's hind legs and the downward pressure of the lance only because we see the two lines as filled with the forces of the things depicted—of the horse's muscles and the horseman's lance. Take away the representation and the balance too would dissolve. And the same goes for the composition. Alter the representational meaning of the horse (close its eye, for example, or attach a bangle to its hoof) and the composition would be utterly destroyed. Nothing here is comprehensible until the representation is grasped.
Let us return, then, to our example. When a passage from Voiles reminds me of drifting sails I do indeed hear an aspect of the music. But the important part of this aspect—the part that seems essential to a full musical understanding—can be perceived by someone who is deaf to the ‘representation’. It is possible to hear the relaxed and leisurely quality of the musical line while being unaware that it depicts the movement of sails. The ‘reference’ to sails does not determine our understanding of the music in the way that representation determines our understanding of the visual arts. Ö
In search for examples of genuine musical representation we may be led by this argument back to the suggestion that the true subject-matter of music is sound. Sounds have properties which music, being itself sound, may share; so music ought to be able to depict sounds. For there will be no difficulty here in explaining how it is that the music may lead us inevitably to the thought of what is represented. Thoughts of a subject will therefore form an integral part of musical appreciation. But again there is a peculiarity that deserves mention, since it seems to suggest that even here, in the most plausible examples, there is yet another of our five features of understanding representation that fails to belong to music: feature 2. When music attempts the direct ‘representation’ of sounds it has a tendency to become transparent, as it were, to its subject. Representation gives way to reproduction, and the musical medium drops out of consideration altogether as superfluous. In a sense the first scene of Die Meistersinger contains an excellent representation of a Lutheran chorale. But then it is a Lutheran chorale. Similarly, the tinkling of teaspoons in [Richard] Strauss's Sinfonia Domestica, and the striking of anvils in Rheingold, are not so much sounds represented as sounds reproduced, which in consequence detach themselves from the musical structure and stand out on their own. Nor is this an accident. On the contrary, it is an inevitable consequence of the logical properties of sounds. For sounds … may be identified as individuals independently of the objects that possess them. In attempting to represent them, therefore, one need have no regard to the object that produces them: one represents the sound alone. But since there is nothing to music except sound, there ceases to be any essential difference between the medium of representation and the subject represented.