Starting from theories of expression from the 17th and 18th century (the Camerata Fiorentina, Reid, Mattheson...), Peter Kivy wants to demonstrate that 'some emotive predicates are applicable to music, and why these are applicable intersubjectively' (p.11).
Peter Kivy thereby rejects the theory that music would be the expression of some subject. He therefore makes a distinction between 'to express' and 'to be expressive of' (p.12): the face of a Saint-Bernard is 'expressive of sadness', but that does not mean that the dog itself feels sad. Also music can be 'expressive of', without being the expression of the composer. Peter Kivy also rejects the theory that music would be expressive in the sense of 'eliciting emotions in the listener': 'Sadness is a quality of the music, not of the power of the music to do things to the listener' (p.23). He rather endorses the 'cognitive speech theory' (Reid) which contends that 'music can resemble the passionate speaking voice' so that we recognise the corollary emotion (p.23). But he wants to extend this theory in two respects. To begin with, he states with Mattheson that music 'in its structure, bears some resemblance to the "emotive life" ' (p.39). Resemblance must hence be extended to 'the way we express (feelings) in gesture, facial configuration, posture, and so forth' (p. 40). Music is not so much a 'speech icon', as rather an 'emotive icon' 'resembling not the vocal expression of sadness, but its expression in bodily gesture and posture' (p.53). The 'resemblance' may be literal, but also 'synaesthetic' or 'metaphorical' (p.54). It does not suffice, however, to extend the theory from expressive speech to expressive behaviour as such: next to resemblance ('contour'), there is also a 'conventional' relation between music and expressive behaviour'. Only seemingly is there thus introduced a second factor in the 'contour-convention' theory (p. 113): 'all expressiveness by convention was originally expressiveness by contour' (p.83). Such is the central thesis of Peter Kivy's book. From this general perspective, Peter Kivy deals with particular problems like that of the precision of the expression: according to him the 'expressiveness' is not general as with Hiller/Langer, nor precise, but 'moderately indeterminate' (p. 198), whereby the text can provide full determination.
Although Peter Kivy starts from old mimetic theories, he is profoundly indebted to the nowadays prevailing semiotic-linguistic model, especially the version of Susanne Langer (Philosophy in a new Key, 1942) and Nelson Goodman (Languages of Art, 1976)
That is already apparent form the distinction between 'to express' and 'to be expressive of'. Peter Kivy cannot accept that 'music expresses sadness', because 'only sentient beings can have emotions' (p.6) - so that, if we nevertheless want to contend that 'this music expresses sadness', 'the obvious candidate is the composer' (p.13), which, according to Peter Kivy, is not the case, were it alone for the fact that the composer need not be in a sad mood when composing sad music. But, in my view, that is not a reason to deny that 'music expresses sadness'. It suffices to replace the example of the Saint-Bernard with that of the Mona Lisa. It is true that the smile of the Mona Lisa is not the expression of da Vinci. But that does not imply that the smile would be nobody's expression - that it is merely ''expressive of ...': it is rather the expression of the mood of the Mona Lisa. Contending that the painting of the Mona Lisa is merely 'expressive of' because it is not a 'sentient being', comes down to claiming that there is only paint to be seen on the canvas, and not the Mona Lisa herself. As soon as we accept that an absent being is presentified before our eyes - as soon as we adopt the mimetic paradigm - it is no longer nonsensical to contend that the Mona Lisa 'expresses' rather than that the painting is 'expressive is of'. And, what applies to the painting, applies even more to music (in so far as it is mimetic, see below): also in music, we are dealing with the imitation of beings that express their feelings in auditory expressions - just think of Peter Kivy's own example: in Monteverdi's Lamento d'Arianna, it is not just sounds that 'are expressive of', but Arianne who expresses her grief. The lament thus only differs qua sensory modality from the smile; and both are an expression of a 'sentient being' - the Mona Lisa and Arianna.
As soon as we leave the semiotic paradigm for what it is, and begin to think within the framework of the theory of mimesis, there is no longer a problem with understanding music as an expression of feelings. That becomes only a problem when we drop the idea of imitation, and get the wrong impression that we only hear a lament without the concomitant suggestion of a sentient being that is expressing itself through it. The question then is no longer how a smile on a canvas or a lament sung by a singer can be the expression of beings that are nowhere to be seen, but rather how it is that a visual or audible configuration can be expressive of an emotion. That question has nothing to do with painting or with (mimetic) music, but rather with expressiveness as such: it obtains to the lament of a real lover as well as to that of an imaginary Arianna. Not otherwise than Nelson Goodman, Peter Kivy unawares transforms an art philosophical question - how it is that music is experienced as the expression of a being whose presence is merely suggested by the performance of a lament - into a problem of semiotics - how it is that a descending interval is experienced as a lament. Assuming that - apart from the question whether the descending interval is sung by a real person or by a singer that imitates the lamenting Arianna - a semiotic theory could answer that question, that would be a step forward in the development of a semiotic theory of expression, but it would contribute nothing to our understanding of the difference between a real woman in distress and Arianna, whose presence is only suggested through the singing of her lament. To phrase it differently: the art philosophical problem of (mimetic) music is not to understand why a given interval expresses a given emotion, but rather how it is that when hearing that interval from the mouth of a singer, we nevertheless have the impression of hearing Arianna lament, and not the singer.
Therewith, the whole drama of every semiotic approach to the theory of the image is summarised in a nutshell - and my criticism completed. But it will prove illuminating to have a closer look at the false problems that arise when the mimetic paradigm is replaced with the semiotic.
Let us first examine Peter Kivy's answer to the question why a particular sequence of tones sounds like a lament. In a first phase, Peter Kivy finds the explanation in the resemblance with speaking: music as a 'speech icon'. This in principle correct insight - that a sung or played lament is identical with a real lament - cannot explain music that is not reducible to aural expression. As mentioned above, Peter Kivy tries to solve the problem by extending 'resemblance' in general. But such extension creates a new problem: 'We don't hear human posture and movement' (p.54). Peter Kivy hence has to use the term 'resemblance' no longer in its literal sense, but rather in an analog or metaphorical sense: as 'emotive icon'. But, according to Peter Kivy, even such 'speech icon' extended to 'emotive icon' cannot explain the effect of chromaticism and major and minor scales. To explain these phenomena, he contrasts 'emotive icon' (also called 'contour') with 'convention'. He thus constructs a continuum icon/contour (motivated signs) and 'convention' (see: mimesis and semiosis, motivation), whith, on the pole of the icon/contour, a subcontinuum of ''literal resemblance' (speech icon) and 'metaphoric' or 'analog' 'resemblance' (emotive icon).
It is not difficult to see how this theory is a compromise between two conflicting tendencies: the tendency to distance himself of mimetic theories (speech icon) like those of the Camerata, and the tendency to distance himself from representatives of the semiotic paradigm who do not sufficient justice to the 'mimetic'. That is betrayed in the choice of the term 'resemblance' (which, to Peter Kivy, can be used literally as well as metaphorically). The term allows him to distance himself from Langer's 'isomorphism' and from Goodman's 'move from resemblance to symbol' (p.61) on the one hand, but, on the other hand, also from the mimetic theory, that he dismisses as 'representation'. Music can be 'representational' ('resembling' in the literal sense, as with the lament of Arianna), but that is not always the case (p.64). On top of that, he rejects (in the wake of Wolheim's criticism of mimesis) representation, because that implies the intention to imitate (p.64): there can be resemblance without (the intention of) representation (p. 66). The contradictory tendencies that lie at the roots of the choice of the term 'resemblance' explain the arbitrariness in its use: now Peter Kivy is talking about 'the musical line' 'as a kind of musical icon, resembling a piece of human emotive expression' (p.20), then about the 'analogy to the impassioned speaking voice' (p.27), again about ''analogue to rapid bodily movement' (p.69), 'formal analogy' (p.40), 'congruence' (77), and finally about metaphor and synaesthesia (p.55-56). He would welcome the term 'isomorphism' as well, were it not that Susanne Langer already used it in the sense of a 'general symbol', which he rejects (p.60-61). It speaks volumes that Peter Kivy refuses to provide an 'philosophical analysis' of the term resemblance, and that he just sweeps the problem under the carpet by stating that he will use the term 'in the ordinary way' (p.66).
Be that as it may, Peter Kivy is constructing a variant of the familiar 'continuum between image and sign'. As I argued in 'Semiosis and mimesis', there is no such thing as a continuum between image and sign, as rather an unbridgeable gap. On the other hand, there is something like a continuum between different degrees of motivated and unmotivated signs - a continuum of motivation of signs. Nevertheless, that continuum is readily invoked by all those who want to construct a semiotic theory of the image, and who can thereby not overlook the literal 'resemblance' of many an image with the real world. It speaks volumes that Peter Kivy wants to minimise the role of convention by stating that: 'all expressiveness by convention was originally expressiveness by contour' (p.83). Here as elsewhere, we witness the return of the repressed mimesis under the guise of the 'motivated sign'.
Of course, the question as to the relation between a real complaint and the music of Monteverdi should be answered. But it should be clear that the answer to that question - that the music is a monosensorial duplicate of the imaginary complaint - is not an answer to the totally different question how it is that the music of Monteverdi is a 'representation' of the lamenting Arianna: for it is not the lament that is being 'represented', but rather the lamenting Arianna. The question is not what the relation is between a real bird song and a recording of it - which is a plain monosensorial duplicate - but rather how it is that when hearing that bird song, we have the certain impression of a bird singing, where there is only a loudspeaker emitting the required vibration.
SEMIOSIS AND MIMESIS
Apart from that, it matters to point to the fact that the answer to the question whether music is expressive depends on the kind of music we are dealing with. For there are many kinds of music, and a distinction between them is only possible when we make a clear distinction between mimesis and semiosis.
To begin with, there are the many kinds of mimetic music (see: 'Auditory mimesis and music'). Monteverdi's 'Lamento d'Arianna' is an example: by duplicating the lament of an imaginary Arianna, the singer imitates a lamenting Arianna. Next to the many kinds of mimetic music, there is also music which is a pure sign. For, it seems to have escaped Peter Kivy's attention that many movements are audible indeed: just think of the movements of working people, and especially of the movements of musicians who produce the signs that command, structure and synchronise the movement of workers or dancers (see 'Movement conjuring signs'). To these signs, the continuum of different kinds of motivation is fully applicable: the signs are sometimes motivated by auditory identity (think of the drone of soldiers boots), but more often with analogy (as when the sergeant uses 'high' and 'low' as an analog sign added to the words 'left' and 'right'). How much Peter Kivy thinks in terms of dance music, is apparent from his suggestion to read the aria from the second part of the first Brandenburg concerto 'as a choreography of expression. Think of the motions a player or conductor might make in performing it; or the way a dancer might move to it' (p.52). Next to dance music, there is also music in which mimesis and semiosis are combined: image conjuring music (think of some forms of 'tone painting' and of program music') whereby the movement conjuring signs are used to conjure images in the mind, as when ascending and descending scales are used to 'represent' waves. These three kinds of music may be combined in diverse ways.
It is rather impossible to make statements that apply equally to these three kinds of music. Whoever wants to contend something about 'musical expression', has first to make it clear about what kind of music he is talking. With dance music, it is not so much the music which is expressive, as rather the - by the way: visual - expression of the movement of the dancers. Only with what we call speech music and with many kinds of mimetic music (the Lamento d'Arianna as well as many absolute music) are we dealing with sounds that are the expression of (mostly imaginary) beings.
MUSIC AND EMOTION
Only when the necessary distinctions are made, can we properly tackle the problem of the relation between music and emotion.
To begin with, it should be stressed that not only music, but practically the whole world is 'expressive'. Granted, not of every of these realities can it be said that they 'express' something: it might be more appropriate to speak of 'being expressive of'. The problem of the Saint-Bernard is, hence, just like the question why a particular musical motive expresses grief, not a problem of music theory, but - as is already apparent from the example of the Saint-Bernard, a problem of the real world as such.
A problem that is a genuine problem of music theory - or of art philosophy in general - is the question whose emotions are expressed in music. It is apparent, then, that in the Lamento d'Arianna, it is the feelings of Arianna that are expressed, not those of Monteverdi or those of the singer. And, in the 'Vorspiel' to 'Das Rheingold', it is the 'mood' of primordial waters that is rendered, not those of the composer or the musicians. It is interesting to note that this question, that continues to haunt the philosophy of music, is not at all a problem in the visual arts: everybody will agree that it is the Mona Lisa who expresses her feelings and not da Vinci, or that it is from the landscape that the impression of the sublime arises, and not from Kaspar David Friedrich. It should be noted that the problem can only arise in art: in the real world the emotions are always those of whoever shows them. In non-mimetic speech music, it is undoubtedly the lamenters who express their feelings in the music. And that goes also for the visual expressions of (non-mimetic) dancers. But, precisely because Peter Kivy does not distinguish mimetic from non-mimetic music, he cannot possibly conceive of cases where a lament would be expressive of the feelings of the lamenter himself.
Next, there is a problem that has to do with (mimetic or non-mimetic) music in particular. As a rule, music is combined with movement conjuring signs. These signs invite the dancers to communally execute the same movements, or the singers to communally sing the same song, so that all of them come to share the same mood (see: sympathetic mimesis). The same holds true for mimetic music. That is why our reaction to the lament of Arianna is fundamentally different from our reaction to the smile of the Mona Lisa: when hearing Arianna's lament, we cannot but feel compelled to join her singing and hence to feel ourselves equally miserable as Arianna, whereas nobody feels inclined to take the posture and the expression of the Mona Lisa. In combination with the theory that music is the expression of the singer or the composer, this leads to the erroneous impression that music would be able to directly mediate feelings from one soul to another.
Let us, finally, remark that the problem of 'feelings without object' that continues to haunt music philosophy ever since Hanslick introduced it, is a false problem. From the Mona Lisa we do not know what prompted her expression, and we will never know what prompted the Scream on Munch's famous picture. Nevertheless, nobody has ever experienced this as a problem, quite the contrary. Whoever would concoct a story that would explain these expressions, would rather destroy their expressiveness - for the same reason that the weeping of child moves us all the more when we do not know what the cause of the weeping is. Hence, full determination by a text is possible (as in songs or arias), but not at all necessary.
© Stefan Beyst, june 2012, translated june 2012
Your reaction: mail to stefan beyst
Stay informed about new texts: mailinglist
Background to this text: stefan beyst: theory on art