Stephen Davies, University of Auckland
I want to express my gratitude to Anton Killin, Matteo Ravasio, and Nemesio Puy for engaging so generously with my work. Their consideration of some of my earliest writings, as well as more recent ones, remind me of how many decades I have spent beating around the same bush…
In January 1980 my first philosophy paper was published in the journal Mind. Fortunately for me, it was subsequently collected for The Philosophers’ Annual. Impressed by this, my seniors at the University of Auckland converted what was a temporary job into a permanent one. At last, after some frustrations and delays, my academic career was underway.
The paper derived from my Ph.D., which was awarded in 1978. In it I outlined a resemblance theory of musical expressiveness and also discussed the case in which the listener responds by feeling what the music expresses. According to my view, music is expressive in virtue of a resemblance between its dynamic pattern and behaviours that can be recognised as giving expression to an emotion where one does not know the emoter’s beliefs or the intentional object of her feeling. This view has been termed the “contour theory” or “appearance emotionalism”. And as regards the listener’s mirroring response, this depends on emotional contagion and provides an exception to the cognitive theory of the emotions. (This was before psychologists began to study human-to-human emotional contagion and mirror neurons.) These and other views were developed and elaborated in Musical Meaning and Expression of 1994.
Much more recently I have written on the evolutionary status of art and on its ancient origins.  I speculate on the basis of physiological and behavioural data that recognisably musical behaviours might go back as far as 500,000 years to our predecessor Homo heidelbergensis, and in any case should pre-date the permanent departure of some Homo sapiens from Africa about 60,000 years ago.
1 Reply to Killin
Anton Killin invites me to draw these two strands of my work together and thinks that doing so makes clear that musical expression is more about the communication of affect than the recognition of contour. He writes: “In my view, beginning at the beginning of music, some kind of hybrid expression-arousal theory of Pleistocene musical expressivity is not only salvageable but it is very plausible […]. Later developments in music—for example, those that render music’s participatory status as one of trained specialists performing for passive audiences—utilise our capacities to intentionally express emotion through music and be emotionally stirred by music, and are thus candidates for contour theory. But not all later developments are plausible candidates for contour theory (e.g., Inuit song duels)”.
I think these comments tend to skate over a distinction that we need to bear in mind. The first part concerns whether or not composers/performers express their emotions in the music they make and whether or not listeners respond emotionally to the music’s expressive communication. Both claims have been denied by philosophers, the first by Zangwill, for instance, and the second by Kivy. But I am not so inclined. Composers most certainly can put their feelings into their music by appropriating the expressive potential of musical materials. (We should recall how sophisticated this mode of self-expression is, though. It is one thing to burst into tears, but quite another to burst into a song duel.) And the fact that musical expressiveness is deliberately contrived, unlike the shape of a willow tree or the basset hound’s physiognomy, primes the listener to hear a communicative element in the music and to be moved by that. Moreover, I think that the expressiveness is heard directly in the music, not deduced from a self-conscious comparison of its dynamics with human behaviours. So I see my theory as consistent with the importance of self-expression and emotional communication in music’s earliest forms. However, and now we get to the second part that should be seen as distinct from the first, the expression and arousal theories claim more than that music can express its maker’s feelings and can arouse the listener by doing so. They are theories purporting to explain how music is expressive. The truthmaker for “the music is expressive of e” is alleged to be “its maker feels e” in the one case and “the listener feels e” in the other. As theories attempting to analyse music’s expressiveness, I think both fail.
When Killin writes of a “hybrid expression-arousal theory of Pleistocene musical expressivity,” I think that, rather than appealing to those theories of musical expression, he is better understood as drawing attention to the facts of self-expression and arousal and to their occasional importance in the musical encounter. And the importance of those facts is not inconsistent with appearance emotionalism in my view.
I do have a concession to make, however. When I thought about the resemblances between human emotions and music, I discounted some that have been claimed for shared forms, shared phenomenological profiles, and shared sonic qualities. To take the last, the sounds that are wrung from us under the force of powerful emotions—sobs, shrieks, roars, groans—are not very musical and are not usually mimicked in expressive music. I did not consider a comparison with prosodic features of emotionally heightened speech, however, even though some seventeenth- and eighteenth-century theories tried to account for musical expressiveness as recalling these. In recent years there has been empirical confirmation of a correlation between expressive inflections in music and emotion-laden speech in the work of the psychologist Patrik Juslin and his colleagues. They asked musicians to play a single, short piece happily, angrily, sadly, and so on and matched differences in the renditions with differences in the prosodic features of speech expressive of the same emotions. (Notice, by the way, that by keeping the piece always the same, they are in no position to tell whether the overall dynamic contour contributes to musical expressiveness, so they are not warranted in claiming to have accounted for musical expressiveness in its entirety.
I accept that this empirical programme is successful in accounting for at least some aspects of music’s expressiveness. Moreover, and here is the concession that brings my view nearer to Killin’s, I think that prosodic expressiveness rather than dynamic profile might play the larger role in short, repetitive songs that are generally simple and similar in character. The expressive contribution of the music’s dynamic profile might become more prominent as musical pieces become longer and more complex, which presumably first happened after the earliest music in the history of music.
But there is one caution to register. The assumption is that music imitates expressive speech. If it were the other way round, we would have no explanation of how music is expressive. Yet there is controversy over whether language preceded music and over whether they both descended from some quasi-musical protolanguage. So it is simply not clear how far back in musical history we could trace a resemblance between music’s expressiveness and emotional verbal prosody.
2 Reply to Ravasio
I have always regarded the attribution of expressive predicates to music as literal. When we say “the music expresses sadness” (or “is sad” for short), of course this is not to be glossed as “the music feels sad.” In my view, it is instead equivalent to “the music presents in its dynamic pattern the appearance characteristic of sadness as it might be displayed in a person’s comportment,” which is literally true of the music. In my copy of the Concise Oxford Dictionary, the second meaning listed for “sad” is causing or suggesting sorrow (a sad story) and the sixth for “expression” is conveying of feeling in the performance of a piece of music. Dictionaries deal in literal meanings, not lively metaphors.
I think the same about talk of musical space—notes being higher and lower, for instance—and musical movement. The movement we hear in music is that of temporal process by change—like fluctuations in the value of a currency—rather than that of a re-identifiable individual that moves from place to place. I guess that the experience of musical space and temporal process is a by-product of our experience of octave equivalence; that is, of hearing notes at the octave as equivalent but as occupying different locations within the dimensions of a sonic world generated by music. Octaves are “the same note but higher or lower, bigger or smaller.”
I suggested (in 2011) that all these usages are polysemous and that they are cross-cultural. All cultures describe music in spatial or size terms and attribute expressiveness to at least some of their music, just as they all recognise octave equivalence even if the scales they divide octaves into might differ. This universality is important, because it suggests that these terminologies, rather than being eccentric, highly creative neologisms, instead respond to something deep and fundamental in our experience of music, something that is grounded in our evolved auditory capacities.
In discussing these cases, Matteo Ravasio draws a distinction between two kinds of polysemy. One sort might be grounded in perceived (or perceivable) resemblances, which is how I explain the expression of emotion in music. But others, such as the darkness of certain timbres, do not rest on resemblances that we could point to, despite our sense of the appropriateness of this cross-modal use of the term “dark.” And these apparently ungrounded cases also may have an expressive import. Therefore, the contour theory does not account for all the kinds of musical expressiveness.
The first example is the clarinet glissando that begins Gershwin’s Rhapsody in Blue. Ravasio maintains that this equally resembles emotion characteristics of positive and negative emotions, so appearance emotionalism cannot account for its expressive character (the nature of which he does not specify). I don’t find this example troubling. It seems to me that in the few seconds it occupies, the glissando is expressively ambiguous at best, and that the relaxed, calm descent over the next four measures resolves the ambiguity to the positive side.
The second example is of a saturated, dark timbre, rich in overtones and using a low register, that has a menacing character. Here, I guess Ravasio intends the weight to fall on the menacing character of the timbre rather than on its darkness, to which I return below. But as with the glissando, I think I would need to hear more, to hear how the timbre features in the wider musical process, before applying the term “menacing” to it.
A third case is that of the sad character of an isolated minor triad, for example. This is a limb I have already gone out on. I don’t think decontextualized minor chords are expressive of sadness. We should recall that the “happy” major scale allows for minor chords (rooted on the supertonic, mediant, and submediant) and that, in Medieval music, the major third was a discord. If we hear sadness in the minor chord, as music psychologists regularly suggest that we do, I think this is because we contextualise it in imagination as a tonic triad that brings with it associations with sad-sounding dynamic processes.
Ravasio refers to my talk of dark timbres and unresolved tensions, as well as to the tension of the tritone. All of these do seem to be examples of Ravasio’s second kind of polysemy. None of these qualities appears to depend on an enumerable resemblance across domains. They fall under the category of synaesthesia, perhaps. But though these features can combine with others to contribute to the emergence of musical expressiveness, I am not sure I would count them as examples of emotional expression in their raw form.
I should add that I had not thought of making the distinction between the two kinds of polysemy that Ravasio draws, though I am now convinced by his account. I also agree that his musical examples point to some of the first-order properties on which music’s expressiveness supervenes. What I question is the idea that, on their own, tense intervals, dark timbres, and the like are unambiguously expressive. So I also deny that the argument here establishes the subsequent conclusion that music involves modes of expressiveness not covered by appearance emotionalism.
I want to re-emphasise a point made earlier. Ravasio thinks that appearance emotionalism is committed to a particular phenomenology of the listening experience, one in which we are aware of a resemblance between human expressive behaviour and the dynamics and structures of musical processes. This isn’t quite how I think of it. Our immediate awareness is likely to be of the music as expressive on its own terms. But when I think about why it is so, or how I can justify attributing one expressive character as opposed to another to the music, the resemblance with humanly expressive comportments comes to mind. (Or, in other cases, perhaps it is the resemblance to the prosodic features of expressive speech that comes to mind.) By contrast, I have no ready-to-hand explanation of why I hear notes as high or low, or timbres as bright or dark, or chordal progressions as involving the waxing and waning of tensions. If there is an explanation, I would expect it to be physiological, with talk of linked or shared neural circuits and the like.
3 Reply to Puy
“Are these two similar works or two versions of a single work?” This is the kind of question that might give philosophy a bad name with ordinary folk. And it might not matter how we answer, so long as the path we take in getting there provides clarifying insights. But it is a question I have considered in several contexts: first, when I was thinking about the transcription of musical works against the background of issues of authentic performance, and later, having written at length on the varied ontology of musical works.
I argued that, where musical works are instrument-specific and the transcriber adapts the original to a different instrumental medium while preserving its sound-structural outline, the resulting transcription is a different but derivative work. Nemesio Puy regards this view as counterintuitive and argues against it.
One counterexample Puy offers is that in which Astor Piazzolla’s Primavera Porteña is transcribed by Jorge Valdano and premiered as a new work under the title of Bonaerensis. According to Puy, “[m]ost of the audience […] and especially those who bought their tickets expecting the premiere of a new work, would feel cheated” or would take the piece for an ironic conceptual work. I am inclined to agree. But this is an unusual case in which Valdano’s piece flouts the usual convention for transcriptions by not publicly announcing the source from which his is derived.
There is a more or less standard way of titling one’s work as a transcription when it is intended for concert presentation. (This was well-observed for transcriptions intended for fee-paying audiences from at least the nineteenth century.) It acknowledges the primacy of the original by giving the name of its composer first and the name of the transcriber second. In addition, the transcription retains the name or title of the original. So we get Beethoven–Liszt, Symphony No. 5 in C minor, for instance, or Mussorgsky–Ravel, Pictures at an Exhibition. It is not correct to say, as Puy does, that “[w]hen Ravel’s transcription is performed, programmed, or recorded, what it is said to be performed, announced, or recorded is a work by Mussorgsky, not by Ravel.” What is normally said to be performed is a piece by Mussorgsky–Ravel, this being shorthand for saying the source piano work is by Mussorgsky and the orchestral transcription that is to be played is by Ravel. (Pictures at an Exhibition is fairly unusual in that the transcription is played more often than the original and I would not be surprised if many music listeners are unaware that the original was not an orchestral piece.)
Turning now to the counterexample that is introduced first: there is the case of piano versions of ballets and operas created for use in rehearsals. These are more often characterised as “piano reductions” than as “piano transcriptions” and the name of the reducer might not be indicated. In public performance, the intended orchestra usually replaces the piano. But in the absence of suitable orchestral resources, the performance may be accompanied by a piano playing the reduction in a public concert. (Riccardo Muti played the whole piano reduction of La Traviata one night, when the musicians were on strike.) In my view, this can be represented as a performance of the original work, though in a compressed form. We might compare this with a concert (that is, unstaged) rendition of an opera. So I accept that Lucia should agree with Maria that she is singing ‘Anch’io dischiuso un giorno’ from Verdi’s Nabucco in the rehearsal room, but I disagree that this is a transcription and as such a counterexample to my view. In most cases, transcriptions are based on the entirety of a work and are presented as independent of the original, whilst acknowledging their derivative status. That is not what happens when the orchestral score of an opera or ballet is reduced for piano for the purposes of facilitating rehearsal with singers or dancers.
In sum, I don’t accept the proposed counterexamples as supporting the objection that my position is unintuitive and seriously revisionist.
Puy then criticizes the argument that a change of medium for a medium-specific work results in a new work. The first objection is to my treating “virtuosic” as applying to musical works, as against only to performances of those works. (The point here is that I claim this can be an essential property of a musical work that depends on the medium prescribed by the composer.) Puy’s claims are that it would sound “strange to claim that a musical work is virtuosic per se,” that we could make a work virtuosic by having the musician play it while lying down, that improvements to musical instruments can render works that were originally virtuosic easy to play, and that an easy piece could be virtuosic when well-played by a novice.
All I can say about the first point is that my intuitions are different. Would we say that it is performances of The Rite of Spring but not the work that is noisy? I don’t see why. It surely was not the case that Stravinsky discovered that it was noisy only at the first performance. To say that the work is noisy is to say that accurate performances of it will be noisy, because it is determinatively specified as such by the composer. Similarly, to say that a work is virtuosic is to say that, even if it is played with seeming ease, it should be apparent that it is difficult to play on the specified instrument under standard circumstances, including relativisation to the time of composition (to allow for general improvements in technique).
The assumption about “standard conditions” also deals with the prostrate horn player. It would surely be more challenging to play a concerto with the instrument behind one’s back—and such musical tricks are not unknown—but I agree that we would not attribute the property of being hard to play to the work when the usual playing conditions are not observed. What of improvements to instruments that make the performance easier? If there are radical changes then there might be questions to be asked about the authenticity of the rendition, precisely because the playing comes easier. But as Stan Godlovitch has argued, musical “guilds” typically control the kinds of innovations that are acceptable in order to prevent the removal of challenges to skilful instrumental playing, which is why the pre-programmed synthesizer has not replaced all the other instruments. And I disagree that the neophyte’s playing is virtuosic as a result of being both flawless and very difficult for her to bring off. The relevant standards and degrees of difficulty are relative to professional-level performances.
As Puy notes, I allow (and have argued) that not all musical works are instrument-specific, especially in earlier times when orchestras were not standardised. Arranging such a work for instrumental resources that might not have been available to its composer is then consistent with performing the original, as against creating a transcription. So, I agree with Puy that Giovanni Gabrieli’s Canzona per sonare primo (which is specified by the composer as “for playing on all sorts of instruments with the organ bass”) could be authentically instanced by different ensembles. I would add this small caveat, however. We can often know a work’s general instrumental type. For instance, it is for keyboard, though it could be performed equally on a clavichord, harpsichord, or organ; or it is for brass, or strings, or woodwind, or voices. A work specified at that more general level—either by convention or through the composer’s publicly expressed intention—could be transcribable in principle.
Puy challenges me to come up with further properties that belong to the work as a result of the specified instrumentation. I don’t think that is difficult. See, for instance, the discussion in Levinson 1990, where the way we describe passages of the work depends on how the sounds are elicited from the indicated instruments. In an organ work, a passage for the pedals (feet) might be rushed whereas, when it is transferred to the keyboard (hands), it isn’t.
The next argument repeats the claims about orchestral transcriptions of Mussorgsky’s Pictures at an Exhibition to which I have replied above. Puy says these show that a work can be instrument-specific yet be instanced in performances on other instruments. I think that imperfect renditions can count as performances of their target work, though they involve departures from maximal authenticity and (like the piano reduction rendition of an operatic aria at a public concert) may require careful advertising if an audience is expected to pay. It is hard to find an ophicleide these days, so we usually substitute the tuba for performances of Mendelssohn’s Overture to A Midsummer Night’s Dream. But I do not think of transcriptions as similarly imperfect. It would be odd to fault Liszt’s piano transcriptions of Beethoven symphonies for being for the piano.
Finally, Puy challenges the analogy I drew between colour in painting and instrumental “colour” or timbre. This analogy, which is not my invention but is entrenched in many music cultures, is “essentially mistaken,” he says. A closer, more precise analogy would be between painterly colour and musical harmony, though this better analogy does not help my argument.
Sound structures thus would depend on heights of sounds and not on timbre. It follows, then, that the form or structure of a musical work would not depend on timbre. Since a change of medium entails a change of timbre, and since the structure of a musical work would not depend on timbre, a change of medium of a work W would not entail a change in W’s structure, and hence in W’s identity.
This strikes me as missing the point altogether. Analogies are just that, not identities. A more extended or closer analogy is not thereby a better one for a given purpose. The analogy I drew was perfectly apt for my purpose, which was to highlight a similarity between how we experience colour and timbre, not to seek an underlying structural similarity between them. To repeat: just as colour can contribute to the form, content, and expressiveness, and thereby to the identity, of a representational painting, so timbre (commonly called instrumental colour) can contribute to the form, content, and expressiveness, and thereby to the identity, of a musical work. Instrumentation regularly contributes to the delineation of form, to the articulation of melody, and to the presentation of expressive nuance. Instrumental colour can contribute to the work’s sound structure, therefore, and thereby to a work’s identity. Imagine the note sequence of Ravel’s Bolero rendered by a computer in pure sine wave tones and hear how the work fades away…
About the Author Stephen Davies is a Distinguished Professor of philosophy at the University of Auckland, New Zealand. He mainly writes on aesthetics, particularly the philosophy of music but also works on political philosophy.
Davies, Stephen. (1988). “Transcription, Authenticity and Performance”. British Journal of Aesthetics 28, pp. 216–27.
—. (1994). Musical Meaning and Expression. Ithaca: Cornell University Press
—. (2001). Musical Works and Performances: A Philosophical Exploration. Oxford: Clarendon Press.
—. (1999). “Response to Robert Stecker”. British Journal of Aesthetics 39, pp. 282–87.
—. (2003). Themes in the Philosophy of Music. Oxford: Oxford University Press.
—. (2006). “Artistic Expression and the Hard Case of Pure Music”. In: Contemporary Debates in Aesthetics and the Philosophy of Art. Ed. M. Kieran. Oxford: Blackwell, pp. 179–91.
—. (2007). “Versions of Musical Works and Literary Translations”. In: Philosophers on Music: Experience, Meaning and Work. Ed. K. Stock. Oxford: Oxford University Press, pp. 79–92.
—. (2008). “Musical Works and Orchestral Colour”. British Journal of Aesthetics 48, pp. 363–75.
—. (2010). “Emotions Expressed and Aroused by Music: Philosophical Perspectives”. In: Oxford Handbook of Music and Emotion: Theory, Research, Applications. Ed. P. N. Juslin and J. Sloboda. Oxford: Oxford University Press, pp. 15–43.
—. (2011a). “Infectious Music: Music-Listener Emotional Contagion”. In: Empathy: Philosophical and Psychological Perspectives. Ed. P. Goldie and A. Coplan. Oxford: Oxford University Press, pp. 134–48.
—. (2011b). “Cross-cultural Musical Expressiveness: Theory and the Empirical Programme”. In: The Aesthetic Mind: Philosophy and Psychology. Ed. E. Schellekens and P. Goldie. Oxford: Oxford University Press, pp. 376–88.
—. (2011c). “Music and Metaphor”. In: Musical Understandings and Other Essays on the Philosophy of Music. Oxford: Oxford University Press.
—. (2012). The Artful Species: Aesthetics, Art, and Evolution. Oxford: Oxford University Press.
—. (2014). “Protolanguage”. In: Music in the Social and Behavioral Sciences: An Encyclopedia. Vol. 2. Ed. W. F. Thompson. London: Sage Reference, pp. 914–16.
—. (2015). “How Ancient is Art?” Evental Aesthetics 4 (2), pp. 22–45.
—. Forthcoming. “Evolution of Music”. In: The Oxford Handbook of Western Music and Philosophy. Ed. N. Nielsen. Oxford: Oxford University.
Godlovitch, Stan. (1998). Musical Performance: A Philosophical Study. London: Routledge.
Higgins, Kathleen Marie. (2012). The Music Between Us, Chicago: University of Chicago Press.
Juslin, Patrik N., and Laukka, Petri. (2003). “Communication of Emotion in Vocal Expression and Music Performance: Different Channels, Same Code?” Psychological Bulletin 129 (5), pp. 770–814.
Killin, Anton. (2017). “Hominin Musicality and Musical Expressivity: Revisiting Davies’ Contour Account”. Debates in Aesthetics 13 (1), pp. 18–27.
Kivy, Peter. (1989). Sound Sentiment. Philadelphia: Temple University Press.
Levinson, Jerrold. (1990). “Authentic Performance and Performance Means”. In: Music, Art, and Metaphysics. Ithaca: Cornell University Press.
Mark, Thomas Carson. (1980). “On Works of Virtuosity”. Journal of Philosophy 77, pp. 28–45.
Puy, Nemesio. (2017). “Musical Versions and Transcriptions: Disputing Stephen Davies”. Debates in Aesthetics 13 (1), pp. 53-74.
Ravasio, M. (2017). “Stephen Davies on the Issue of Literalism”. Debates in Aesthetics 13 (1), pp. 28-52.
Zangwill, Nick. (2004). “Against Emotion: Hanslick was Right about Music”. British Journal of Aesthetics 44, pp. 29–43.
 Davies 2011a.
 Davies 2012, 2015, forthcoming.
 Killin 2017.
 Ibid., p. 16.
 Zangwill 2004; Kivy 1989.
 Davies 1999.
 Davies 2006.
 Davies 1994.
 Killin, op. cit., p. 16.
 For instance, Juslin and Laukka 2003.
 Davies 2011b.
 Davies 2014.
 Davies 2011c.
 Davies 1994.
 Ravasio 2017.
 Davies 1994.
 For a philosophically and empirically informed discussion of musical synaesthesia, see Chapter 6 of Higgins 2012.
 Davies 1988.
 Davies 2007.
 Puy 2017.
 Ibid., pp. 37-38.
 Ibid., p. 38.
 See http://www.apnewsarchive.com/1995/Muti-Plays-Piano-for-La-Scala-During-Orchestra-Strike/id-7cbd2e98db5ef7b3e19acff90d33428b.
 Mark 1980.
 For discussion of the move from natural to valved horns, see Davies 2001, pp. 219–22.
 Godlovitch 1998.
 Davies 2008.
 Puy, op.cit., p. 48.
 Ibid., p. 49.