Philosophers on Music

Philosophers on Music

MIND ASSOCIATION OCCASIONAL SERIES

This series consists of occasional volumes of original papers on predefinedthemes. The Mind Association nominates an editor or editors for eachcollection, and may cooperate with other bodies in promoting conferencesor other scholarly activities in connection with the preparation of particularvolumes.

Publications OfficerM. A. Stewart

SecretaryR. D. Hopkins

recently published in the series:

Desert and JusticeEdited by Serena Olsaretti

Leviathan after 350 yearsEdited by Tom Sorell and Luc Foisneau

Strawson and KantEdited by Hans-Johann Glock

Identity and ModalityEdited by Fraser MacBride

Impressions of HumeEdited by Marina Frasca-Spada and P. J. E. Kail

Ramsey’s LegacyEdited by Hallvard Lillehammer and D. H. Mellor

Transcendental ArgumentsProblems and ProspectsEdited by Robert Stern

Reason and NatureEssays in the Theory of Rationality

Edited by Jose Luis Bermudez and Alan Millar

Values and VirtuesAristotelianism in Contemporary Ethics

Edited by Timothy Chappell

Philosopherson MusicExperience, Meaning, and Work

Edited by

Kathleen Stock

1

1Great Clarendon Street, Oxford ox2 6dpOxford University Press is a department of the University of Oxford.It furthers the University’s objective of excellence in research, scholarship,and education by publishing worldwide inOxford New YorkAuckland Cape Town Dar es Salaam Hong Kong KarachiKuala Lumpur Madrid Melbourne Mexico City NairobiNew Delhi Shanghai Taipei TorontoWith offices inArgentina Austria Brazil Chile Czech Republic France GreeceGuatemala Hungary Italy Japan Poland Portugal SingaporeSouth Korea Switzerland Thailand Turkey Ukraine Vietnam

Oxford is a registered trade mark of Oxford University Pressin the UK and in certain other countries

Published in the United Statesby Oxford University Press Inc., New York

The Several Contributors 2007

The moral rights of the authors have been assertedDatabase right Oxford University Press (maker)

First published 2007

All rights reserved. No part of this publication may be reproduced,stored in a retrieval system, or transmitted, in any form or by any means,without the prior permission in writing of Oxford University Press,or as expressly permitted by law, or under terms agreed with the appropriatereprographics rights organization. Enquiries concerning reproductionoutside the scope of the above should be sent to the Rights Department,Oxford University Press, at the address above

You must not circulate this book in any other binding or coverand you must impose the same condition on any acquirer

British Library Cataloguing in Publication DataData available

Library of Congress Cataloging in Publication DataPhilosophers on music : experience, meaning, and work / edited byKathleen Stock.

p. cm.Includes bibliographical reference and index.ISBN-13: 978-0-19-921334-4 (alk. paper) 1. Music--Philosphy and

aesthetics. I. Stock, Kathleen.ML3800.P46 2007781’.1--dc22

2007012568

Typeset by Laserwords Private Limited, Chennai, IndiaPrinted in Great Britainon acid-free paper byBiddles Ltd., King’s Lynn, Norfolk

ISBN 978–0–19–921334–4

10 9 8 7 6 5 4 3 2 1

Preface

K AT HL E E N S TOCK

The idea for a volume of new work on issues in the Philosophy ofMusic emerged after a conference entitled ‘Aesthetics from an AnalyticPoint of View’, which I co-organized with Julian Dodd. This was heldat the University of Manchester in May 2003, and kindly sponsored bythe Mind Association, the British Society of Aesthetics, the Analysis Trust,and the University of Manchester. Three of the contributions here (thoseof Paul Boghossian, Julian Dodd and Michael Morris) are versions ofpapers delivered at that conference. The rest have been commissionedspecially, with the exception of Jenefer Robinson’s paper, which is alargely revised version of a paper which first appeared in a special issue ofthe Revue Francaise d’Etudes Americaines 86 (2000): 77–89. Together thesepapers constitute some of the best new work in what is an exciting fieldof research, and one which has much to engage, not just aestheticians,musicologists and music practitioners, but metaphysicians and philosophersof language as well.

I wish to thank all of the contributors for getting involved in the project,for their readiness to read and comment upon the work of fellow authors,and for producing such excellent work themselves. Particular thanks aredue to Julian Dodd for motivating me to take the volume on. I would alsolike to thank Sandy Stewart for his helpful guidance and encouragementon editorial matters.

Kathleen StockUniversity of Sussex

This page intentionally left blank

Contents

Preface v

Notes on Contributors ix

List of Musical Examples xi

Introduction 1Kathleen Stock

Part I. Musical Ontology 21

1. Sounds, Instruments, and Works of Music 23Julian Dodd

2. Doing Justice to Musical Works 52Michael Morris

3. Versions of Musical Works and Literary Translations 79Stephen Davies

Part II. Musical Expression 93

4. Expression in Music 95Derek Matravers

5. Explaining Musical Experience 117Paul Boghossian

6. Persona Sometimes Grata: On the Appreciation of ExpressiveMusic 130Aaron Ridley

Part III. Musical Meaning 147

7. Can Music Function as a Metaphor of Emotional Life? 149Jenefer Robinson

viii contents

8. The Structure of Irony and How it Functions in Music 178Eddy Zemach and Tamara Balter

Part IV. New Issues 207

9. Music and Electro-sonic Art 209Gordon Graham

10. Thoughts on Rhythm 226Roger Scruton

Index 257

Notes on Contributors

Tamara Balter is a doctoral student in Music Theory at Indiana University,Bloomington.

Paul Boghossian is Professor of Philosophy at New York University. He haspublished many articles on a variety of topics, including colour, rule-following,eliminativism, naturalism, self-knowledge, a priori knowledge, analytic truth,realism, and the aesthetics of music. He is currently at work on a book on thenotion of objectivity.

Stephen Davies is Associate Professor in Philosophy at the University of Auck-land. His most recent books are Themes in the Philosophy of Music (Oxford UniversityPress, 2003) and The Philosophy of Art (Blackwell, 2005).

Julian Dodd is Professor in Philosophy at the University of Manchester. He worksmainly in metaphysics, the philosophy of language and the ontology of art, and hisrecent publications include Truthmakers: The Contemporary Debate (co-edited withHelen Beebee, Oxford University Press, 2005), and Works of Music: An Essay inOntology (Oxford University Press, 2006).

Gordon Graham is Luce Professor of Philosophy and the Arts at PrincetonTheological Seminary. Art and Re-enchantment, his 2005 Stanton Lectures at theUniversity of Cambridge, will be published by Clarendon Press (Oxford) late 2007.

Derek Matravers is Senior Lecturer in Philosophy at the Open University, anda Fellow of Jesus College, Cambridge. He is currently working on a book on thenature of value (forthcoming with Acumen Press).

Michael Morris is Professor in Philosophy at the University of Sussex. Apartfrom his work on aesthetics, his principal interests are in the philosophy oflanguage. Currently, he is completing an introduction to philosophy of languagefor Cambridge University Press, as well as working on a monograph on describingthe world. He is also collaborating with Julian Dodd on the Routledge PhilosophyGuidebook to Wittgenstein’s Tractatus.

Aaron Ridley is Professor of Philosophy at the University of Southampton. Hehas recently published The Philosophy of Music: Theme and Variations (Edinburgh

x notes on contributors

University Press, 2004), and is currently completing a book, Nietzsche on Art andLiterature, to be published by Routledge.

Jenefer Robinson is Professor of Philosophy at the University of Cincinnati. Sheis the co-editor of a 2004 special issue of the Journal of Aesthetics and Art Criticism onArt, Mind, and Cognitive Science, and the author of Deeper than Reason: Emotionand its Role in Literature, Music and Art (Oxford University Press, 2005).

Roger Scruton is visiting Professor in Philosophy at the University of Bucking-ham. He has published more than thirty books, including philosophy, politicaland cultural commentary, criticism, and novels. His most recent work on musicis Death-Devoted Heart: Sex and the Sacred in Wagner’s Tristan and Isolde (OxfordUniversity Press, 2003).

Kathleen Stock is Lecturer in Philosophy at the University of Sussex. Her mainarea of research interest is the imagination. She has published several articles inaesthetics, and is currently writing a monograph on imagination and fiction.

Eddy Zemach is Ahad Ha’am Professor Emeritus of Philosophy at the HebrewUniversity of Jerusalem. His areas of research interest include metaphysics, aes-thetics, epistemology, and philosophy of language, and his most recent book inaesthetics is Real Beauty (Penn State Press, 1997).

List of Musical Examples

1. Haydn Quartet in E Major Op. 54 No. 3, opening bars (mm. 1–14). 169

2. Beethoven: String Quartet in F Major, Op. 135, i, mm. 1–10. 184

3. Beethoven: String Quartet in C-sharp Minor, Op. 131, v,mm. 30–48. 186

4. Mozart: Le nozze di Figaro, Act IV, Scene I, No. 23, Cavatina, L’hoperduta, mm. 29–36. 188

5. Mozart: Le nozze di Figaro, Act III, Scene VIII, No. 19, Recitative,mm. 24–25. 191

6. Beethoven: 33 Veranderungen uber einen Walzer von A. Diabelli,Op. 120, variation 1, mm. 1–5. 197

7. Beethoven: 33 Veranderungen uber einen Walzer von A. Diabelli,Op. 120, variation 13, mm. 1–16. 198

8. Haydn: String Quartet in G Major, Op. 76, No. 1, iv, mm. 179–88. 200

9. Beethoven: String Quartet in F Major, Op. 135, iv, mm. 240–58. 201

10. Diana Deutsch: pattern hearing. 228

11. Wagner: Prelude to Parsifal. 230

12. Wagner: Prelude to Parsifal. 231

13. Bartok: Music for Strings, Percussion and Celesta, 1st movement. 232

14. Boulez: Le marteau sans maıtre. 232

15. Az-Zeloub: Le maıtre sans marteau. 233

16. Elgar: Violin Sonata, 1st movement. 235

17. Dvorak: New World Symphony, 3rd movement. 236

18. Sibelius: Violin Concerto, 3rd movement. 237

19. Rhythmic augmentation. 245

20. (a) Delirio (Peruvian folk song). (b) Messiaen: Harawi no. 2‘Bonjour toi, colombe verte’. 245

21. Non-retrogradable rhythms. 245

xii list of musical examples

22. Messiaen: Canteyodjaya. 246

23. Bulgarian Christmas carol. 247

24. Wagner: Tristan and Isolde, motive of ‘The Look’. 249

25. Wagner: Tristan and Isolde, elaboration of ‘The Look’ motive. 249

Introduction

K AT HL E E N S TOCK

This volume presents ten contributions to the philosophy of music, nearlyall wholly new, from leading thinkers in the analytic tradition.¹ Some arewell known for work in aesthetics, while others are better known for workelsewhere in philosophy. Indeed, it is striking that the art form of musicseems like no other to attract philosophers from diverse areas of analyticphilosophy. One part of the explanation is undoubtedly that, given thelove many philosophers have for music, and in particular for classical music,their attention is specially drawn to its philosophical possibilities. Perhapsequally significantly, though, analytic philosophers in general have an urgeto keep things tidy, and are drawn to music because of the suspicion thatit cannot be easily tidied away. That is, it seems to present some specialphilosophical problems not posed by other art forms.

Often these overlap with or are germane to problems in more generalphilosophical areas. For instance: musical works are thought to presentspecial ontological problems, and are often of interest as such to themetaphysician; the issue of what it is to experience music as expressivetends to interest those working on issues in the philosophy of emotion; thequestion of musical meaning tends to attract those active in the philosophyof language; and so on. At the same time, perhaps as a corrective to thetendency of certain philosophers to aim, apparently exclusively, at theillumination of problems in the philosophy of music by reference to moregeneral philosophical views, others have stressed the need to attend tomusic as music, and, if not to treat it as entirely sui generis, then at least toavoid those generalizations which do not do full justice to its nature, and,

¹ An exception is the contribution of Jenefer Robinson, a much modified version of a paper whichfirst appeared in a special issue of the Revue Francaise d’Etudes Americaines 86 (2000: 77–89).

2 kathleen stock

in particular, to the experience of individual works.² Of course, ideally,these two goals need not be in conflict. The best works in the philosophyof music combine, as the contributions here do, a deep understanding ofmusic with a keen sense of the relevance of more general philosophicalissues to the problems it raises.

1. Musical Ontology

One of the most prominent of the problems facing philosophers of musicis that of identifying what sort of thing a musical work is. On the face of it,musical works present some puzzling features which makes their ontologicalcategorization less than straightforward. Namely, a musical work apparentlycan exist as a performed work, or as a score, or perhaps even (just) as anintentional object, thought of by a composer. It can be performed, or arecording played of it, on several different occasions, and even in severalplaces simultaneously, or need never be performed at all. Similarly, thescore of a work can be multiply reproduced or never produced in the firstplace. Meanwhile, two performances of the same work can sound verydifferent to one another.

A historically dominant approach to facts such as these is to see musicalworks as norm-types, repeatably instantiable as particular spatiotemporaltokens,³ thereby categorizing them in terms already familiar from otherdebates in philosophical ontology. Opinions differ as to what shouldcount as the relevant tokens: they may be construed as particular dateable,locatable performances, or sound sequence-events (that is, sequences ofsound, located spatiotemporally), and perhaps scores too. That a work maybe performed or played on multiple occasions, but need not be, nor evenever written down, supposedly reflects the fact that a type can have manytokens existing consecutively or simultaneously, but need not have any.Meanwhile, that two performances of the same work can sound differentlyis supposedly explained by their being more or less properly formed tokensof the same type.

² This need is stressed explicitly by Michael Morris and Aaron Ridley in their respective contribu-tions: Chapters 2 and 6.

³ See, for instance, Wolterstorff (1980).

introduction 3

Julian Dodd defends a version of this view, arguing that a work is anorm-type, whose tokens are sound sequence-events (2000, 2002, 2004). In‘Sounds, Instruments and Works of Music’ he defends, as part of this view,sonicism: the claim that one work is identical to another if and only if thetwo are sonically indistinguishable. He claims that sonic properties are theonly kind of property normative for a musical work, understood as a type,in that they alone can determine what count as properly or improperlyformed tokens of it; this amounts to sonicism, in effect. In embracingsonicism, Dodd rejects ‘performance-means essentialism’—the claim that,as well as or instead of certain sonic properties, a particular instrumentationis essential to a work’s identity.

The version of sonicism that Dodd favours is timbral sonicism, accordingto which the relevant sonic properties normative within a musical workinclude timbral ones. Dodd agrees with a claim made elsewhere byJerrold Levinson that to classify timbral properties as inessential to awork’s proper formation is potentially to allow its aesthetic character tobe compromised (Levinson 1980: 73–8). He denies, however, that thissupports performance-means essentialism, since, he argues, the connectionbetween timbral properties and instrumentation is contingent rather thannecessary; thus, though an intended timbral aspect might be indicatedby a particular instrumentation, it could in principle be produced viaother means. In fact, Dodd suggests, the employment of a particularinstrumentation in a work should not be understood as an end in itself fora composer, but rather as a means of specifying a particular kind of soundwith particular timbral properties as normative for that work, based onwhat instruments are available to the composer at the time.

Amongst the objections to his position that Dodd rejects is Levinson’sclaim that some aesthetic and expressive properties normative for a givenwork depend on performance means, thereby supposedly showing that insuch cases the performance means are likewise normative for the work(Levinson 1980: 76–7, 1990a: 396–401). Dodd responds by insisting thatthe aesthetic and expressive content of a work is a product of its sound ratherthan its connection to a performance means. He argues that it is possible,with imagination, to hear a piece as played on a given instrument withoutit being played on that instrument; and that the aesthetic and expressiveproperties which, Levinson claims, depend on performance means dependinstead on imaginatively hearing the work as performed in a certain way.

4 kathleen stock

In the rest of the paper, Dodd defends this move against some furtheranticipated objections from Levinson.

The type theory of musical works, of which Dodd is an exponent,comes under powerful attack in the second contribution to this section.In ‘Doing Justice to Musical Works’ Michael Morris makes two powerfulobjections to the view that musical works are types; and, indeed, withcertain adjustments, to any view which classes musical works as generalentities.

First, the type theory cannot account for the fact that works of art,generally, are essentially meaningful; which is to say what they are ‘therefor’ is to be understood. This is not just to say that meaning can be foundin them, as it is in the natural world, for instance, but something stronger:that, in some sense, their nature prescribes that one try to understandthem, so that to fail to do so is to fail ‘to do justice to’ them. The typetheory is naturally understood as entailing that musical works cannot becreated, but are instead discovered; indeed, this is acknowledged by itsadherents (Dodd, for example). Yet, Morris argues, if musical works areeternal existents which cannot be created, then they cannot be essentiallymeaningful in the stronger sense just articulated. The type theorist cannotacknowledge that there is anything in a work to be understood in thissense; though she may acknowledge that musical works have characteristicpsychological effects on listeners, she cannot think of them as products ofunderstanding. Nor may she account for the meaning in a musical work bythinking of it as arising subsequent to the work’s ‘discovery’ by a composer;for this would be inconsistent with the point, entailed by the claim thatmusical works are essentially meaningful, that the meaning of a work isconstitutive of the entity that it is.

Morris’s second objection is to a particular version of the type theory:that musical works are types whose tokens are performances (though withslight adjustments he claims that it will apply to any conception of musicalworks as general entities). Performances, insofar as they let works be heard,allow them to be understood. But more than that, a performance onlycounts as of a given work if the performer understands the work, to someextent. This creates a problem for the type theory, insofar as it purports tobe a reductive theory, aiming to specify the relation between performance(token) and work (type) in terms of some common property shared byboth, without any reference to the work as such.

introduction 5

Morris concludes by casting some doubts on certain common assump-tions of type theorists and others: that musical works must in some sensebe ‘repeatable’; and that an explanation of what it is for a thing to berepeatable in the way a musical work is must lie in the assimilation of worksand performances to more familiar ontological kinds, such as generals andparticulars.

Not all questions of musical ontology concern the nature of musicalworks per se. There are equally interesting questions to be raised abouthow we should think of those musical objects generated by the variousstages of the process of composition. In ‘Versions of Musical Works andLiterary Translations’, Stephen Davies is concerned with the ‘ontologicallyprovocative’ issue of works, musical and otherwise, whose completionapparently precedes aspects of their composition. He urges that, in manycases, we should categorize a musical work of which it is true that aspectsof composition succeed its completion, not as distinct from that workcompleted earlier, but as a ‘version’ of it. A work version, understood inDavies’s special sense, is produced wherever features of the work, whichotherwise would be constitutive of its identity as such, are intentionally and‘moderately’ altered, either by the author or, in the case of works whosecomposition is unfinished or ambiguous, by someone else.

Davies carefully distinguishes the class of work versions from otherswith which it might be confused. First, work versions are distinct fromdrafts, which predate completion of a work, and pose no particularontological puzzle; presumably because, at the time of their production,one cannot yet identify what will turn out to be identity-constitutivefeatures of the work, and so there can be no question of such featuresbeing altered in production of the draft. Second, work versions are distinctfrom performance interpretations of a work: performances of a work which(pace Morris) reproduce its identity-constitutive features though which maydeviate from what is prescribed by the composer in other ways. Third,they are distinct from ‘interpretation versions’: performance interpretationsof a work, produced by a performer in order to be performed repeatedly,by her or others. A performance interpretation of a work, like a workversion, may post-date the work in question’s moment of completion,but, unlike it, is not intended as a recomposition of that work, and—ifit is accurate—nor does it alter any of its identity-constitutive features.Fourth, work versions are distinct from transcriptions: that is, adaptations of

6 kathleen stock

a medium-specific work to some new medium. These, like work versions,alter identity-constitutive features of a work: specifically, they change awork’s (identity-constitutive) instrumentation, such that the medium ofthe work, and perhaps also its contents, are then changed. Here, unlike inthe case of a work version, the alterations wrought are sufficiently radicalto license our calling their product a new, derivative work, rather than aversion or interpretation of the original one.

Having delineated the work version as a distinct entity, Davies examinesits relevance as a category to other art forms, noting that work versions canbe found elsewhere in the arts. One such place is in the visual arts, insofar as,for instance, an artist might moderately rework a painting after its officialcompletion time. From this he draws the interesting consequence that,in one sense, paintings and other handmade visual artworks, traditionallythought of as ontologically singular, can exist in multiples. He denies,however, that no meaningful distinction can be preserved between worksin these forms and those which are more usually thought of as multiplyinstantiable: for, while the former can exist only in successive versions (sincea change in one version signals the end of it and the production of another),the latter can exist in multiple versions or instances simultaneously. Finally,Davies looks to the literary arts, urging us to think of translations as workversions rather than transcriptions of the original work, at least insofar asartistically relevant features of the original work are preserved in translation.

2. Musical Expression

Another central philosophical problem raised by music, apparently separatefrom any ontological puzzles it raises, is that of saying what is meant byclaims that attribute expressive or emotional properties to music, such as(rather, basically) ‘the music is sad’. This, broadly speaking, is the problemwith which contributors in this section of the volume are concerned. Asa question about the language we use to describe music, it tends to attractphilosophers actively concerned with questions of meaning. Meanwhile,since many think the question is best tackled by giving an account ofwhat it is to hear music as expressive, it also attracts those interested inanalysing musical experience, often in terms of other more general sorts ofexperience (for instance, emotional experience of some sort).

introduction 7

One philosopher who takes the latter approach is Derek Matravers.In ‘Expression in Music’ he stresses that a relevant explanation of whatit means, for instance, to say that music is sad (taken as a representativecase), should describe what it is to hear music as sad, phenomenologicallyspeaking; and, moreover, should make clear what such an experience has incommon with cases in which the epithet of ‘sad’ is literally applied. It neednot take the form of an account of the properties that cause the experience,since, generally, an account of the cause of a given experience need notilluminate the nature of that experience. Better, he argues, to characterizethe experience directly. For his own part, he advocates a ‘dispositional’theory, according to which music is heard as sad if it tends to induce in thelistener some non-cognitive feeling(s) associated with a typical reaction tothe expression of sadness (Matravers 1998). In his paper here, he indirectlydefends this view by comparing it to other leading accounts, and showingthat, despite some well-known objections, it is not obviously inferior tothem. Along the way, he provides a helpful map of the territory.

Broadly speaking, Matravers categorizes prominent attempts to charac-terize the experience of sad music into two kinds. The first has it that musicis experienced as apt for description as ‘sad’ because it is heard in terms ofbehaviour expressive of sadness. This can be cashed out in several ways.One such way, roundly rejected by Matravers, is to say that music is sad invirtue of an experienced resemblance to sad behaviour. Another, proposedby Jerrold Levinson, is that sad music is heard by a suitably informed listeneras the sui generic expression of sadness by a musical persona (1996: 107 etpassim). Matravers initially objects that, whether this is understood simply asa redescription of the experience of sadness or as a positing of an imaginativeactivity supposedly explanatory of it, it is unhelpful, since in both cases,the supposedly antecedent notion of music being heard as the sui genericexpression of a persona is dependent upon one’s grasp of what it is to hearmusic as expressive (but see Ridley: Chapter 6, this volume). In addition tocritical discussion of Levinson’s responses on this point, Matravers launchesa more general objection against his account: not everyone who hearsmusic as sad would agree that they hear it as the expression of a musicalpersona, and in the absence of any consensus, it is unclear what resourcesLevinson may call upon to persuade dissenters that this is the right account.

The second sort of answer considered by Matravers focuses, not onthinking of sad music in terms of behaviour expressive of sadness, but

8 kathleen stock

on the sad feelings produced in the viewer on listening. Matravers’spreferred dispositional account falls into this category. In defence ofhis view, Matravers sketches out the lines he takes against some well-known objections to arousal theories. To the objection that feeling sad,understood as a full-blooded emotional state with cognitive components, isnot necessary to the experience of hearing music as sad, he agrees, stressingthat it is a non-cognitive component of a typical reaction to sadness whichis necessary to such an experience, and that, moreover, it need not be as‘strongly present’ in the experience as full-blooded sadness would be. Tothe complaint that such feelings are not sufficient for the experience ofmusic as sad, he responds that the relevant feelings are those which closely‘track’ the nature of the music, and hence which cannot be experiencedin other contexts. To the objection that, even so, the posited feelings areinsufficiently integrated with one’s aural experience of the music to beequivalent to the experience of expression, he denies this, arguing that thefact that two co-occurring mental states can be logically distinguished fromone another does not show that they cannot be experienced as one unifiedstate, phenomenologically.

Towards the end of his rich paper, Matravers raises an issue affectingboth his account and others; namely, that hearing music as sad can involveeither hearing the sadness, not as a product of oneself but emanating‘from without’, or, alternatively, hearing it ‘from within’, as the expressionof one’s own sadness. Levinson’s persona account can accommodate theformer fact, but needs to cite something like empathy on the listener’spart to accommodate the latter. Meanwhile, to accommodate such facts,Matravers suggests that his own view be supplemented by the claim that thefeelings sad music induces in a listener can be empathetic or sympathetic.

Assuming, like Matravers, that certain pieces of absolute music character-istically provoke emotion in the listener, in ‘Explaining Musical Experience’Paul Boghossian reiterates that what we are after is not a mere causal expla-nation of this fact. Rather, he urges, we should want an explanation of whyit is rational for us so to respond. To him, the most promising-looking suchaccount is one which attributes expressive meaning to music. The problemof expressive meaning he then characterizes as that of identifying thoseproperties of sounds which enable one to hear a musical work as expressiveof some emotion. The relevant properties are musical properties: those ofpitch, harmony, melody, rhythm, and so on. Hearing a work as having

introduction 9

certain musical properties (yet to be fully specified) makes the attributionof certain expressive meanings apt.

He examines a potential problem for this claim. Roger Scruton hasargued that musical properties are not literally possessed by sounds, claimingthat they are metaphorically possessed instead ((1999); see Chapter 10,this volume, for an articulation of this view with respect to rhythm inparticular). It might seem to follow, against Boghossian, that we cannotexplain the presence of an expressive property in terms of the presenceof musical properties. Indeed, Scruton argues that both musical propertiesand expressive properties are metaphorically rather than literally possessedby musical works, and concludes on this basis that there is no explanationavailable of why a particular musical or expressive description is apt for agiven piece of music.

In response, Boghossian denies that the attribution of such properties tomusic must be metaphorical. For one thing, he claims, the use of metaphorto describe music must be intentional, whereas the propensity to hearsound as having musical or expressive properties need not be. For another,for a metaphorical description to be applicable to an experience, theexperience must possess some literal content, which the former is supposedto illuminate; yet musical experience does not possess such content. Ifexpressive properties are not metaphorically possessed, then we need notyet deny that the expressive meaning of a work is potentially explicable interms of some other set of properties. Meanwhile, Boghossian claims, evenif musical properties turned out not to be real properties of sounds, thiswould not compel us to reject his claim about expressive meaning: namely,that a work is heard as having certain expressive properties in virtue of beingheard as having certain musical properties. For to hear a work as having acertain property is compatible with its not having it.

Though he does not defend it directly here, Boghossian is most attractedto the sort of view which says that a work is heard as having certainexpressive properties in virtue of being heard as having those musicalproperties which sound ‘the way a person would sound who was expressingE vocally’ or ‘the way a person would look who was expressing E gesturally’.(He acknowledges, but leaves to one side the difficulties, raised by Matraversin Chapter 4, about identifying resemblances across sensory modalities.) Aswe have seen, an alternative to resemblance views has been presentedby Jerrold Levinson. The rest of Boghossian’s paper is spent addressing,

10 kathleen stock

first, aspects of Levinson’s positive account of expression, and second anobjection of Levinson’s to the resemblance view.

I shall concentrate here on the first set of points. Boghossian focuses ona claim of Levinson’s presented in a recent version of his view (Levinson2005): a work is expressive of some property E iff the work is disposed to beheard, by a suitably informed listener, as if it were an expression of E. (Thisis referred to here as a ‘dispositional account’ but is to be distinguishedfrom the similarly-named sort of arousal view endorsed by Matravers.) Itapparently would follow from this claim that hearing a work as expressiveof E is equivalent to hearing it as being disposed to be heard by a suitablyinformed listener as if it were an expression of E. Yet, Boghossian objects,it is not clear that the equivalence in fact obtains; indeed, it is not clearhow the latter property could be audibly detected at all. To this Levinsonhas responded that, at least for suitably qualified listeners, reference to sucha complicated and non-audible dispositional property is not required in ananalysis of what it is to hear a work as expressive, since for such listenershearing a work as if it were an expression of E just is equivalent to hearing itas expressive of E (2005). Boghossian worries that this response does not yetshow how a qualified listener (and so nor any other listener, presumably)can be justified in finding a work expressive of some particular emotion,which, as we have seen, is what he thinks an account of expressive meaningshould demonstrate. In addition, he expresses two further worries. Oneconcerns the potential vacuity of Levinson’s reference to a suitably qualifiedlistener. The other is about his analysis of what it is for a qualified listenerto hear a work as an expression of E in terms of her imagining something,given the relatively unconstrained nature of imagining generally.

In ‘Persona Sometimes Grata: On the Appreciation of Expressive Music’Aaron Ridley defends a version of a position rejected by Matraversin Chapter 4: that the experience of expressive music may involve theimaginative postulation of a musical persona in the relevant emotionalstate. Unlike Levinson, Ridley is not concerned to argue that such apostulation is necessary to any experience of expressive music whatsoever,and, unlike other commentators, for instance Stephen Davies (1997),nor is he concerned to argue directly that it is not necessary for such anexperience. Rather he argues that sometimes it is necessary for full appreciationof a particular piece of expressive music. This is not simply a compromiseposition, but a principled stance emerging from his view that any conclusion

introduction 11

about what is involved in the experience of expressive music should bemotivated primarily by considerations drawn from critical practice in musicappreciation, rather than from facts about expressive properties in otherobjects, or generally.

Ridley identifies the following as a possible issue between members ofthe pro- and anti-persona camps. In the most general version of each view,a member of the pro-camp might think that perception of a feature of abearer (whether musical or otherwise) as expressive of an emotion E is aptiff the bearer is imaginatively perceived as a person; while a member ofthe opposing camp might think that perception of a feature of a beareras expressive of E is apt iff the feature would be expressive of E wereit a feature of a person. In the former case, the postulation of a personais supposedly prior to the experience of expression; in the latter, it issupposedly posterior to it. Ridley notes that both positions, in such generalforms, have little or nothing to say about the nature of the bearer: it mightbe a willow tree, or a musical work, or something else entirely. Yet thismatters, because, although the dispute is usually treated as a conceptualissue, Ridley argues that, for many possible bearers, in fact its resolution isan empirical matter which cannot be decided in advance, since for manypossible bearers (willow trees, for instance, that might be seen as sad) thereare no normative constraints governing the precise grounds on which weshould see them as expressive, even when we all agree that they can indeedbe seen as expressive.

He then turns to music, where, obviously, there are normative constraintsgoverning the perception of musical works, and so where the discussionabout whether a persona is involved or not is, to some extent, a conceptualone. Even so, he notes, participants in the debate tend to talk about musicalworks at their most general level as bearers of expressive properties, withoutpaying attention to the distinctive nature of the bearer in particular cases.This, he thinks, makes it the case that the debate is bound, eventually,to become an impasse between those who think that the postulation ofa persona is prior to the perception of expressive features in a work, andthose who think it posterior to such perception. Instead, Ridley advocatespaying attention to the nature of a given musical work in decidingwhether postulation of a persona is required to properly understand itsexpressive character. He notes that this move is consistent with some ofthe objections made against the claim that involvement of a persona is

12 kathleen stock

necessary for perception of expression: for instance, even if Davies is rightthat perception of a feature of a musical work as expressive of E is apt iff thefeature would be expressive of E were it a feature of a person (something hetakes to be prior to the postulation of any persona), this would be consistentwith Ridley’s own claim that a particular piece of music may require thepostulation of a persona in order to fully appreciate its expressive nature;that is, to respond to it in a way which goes beyond what is simply ‘apt’. Insupport of the latter claim, and against generalist accounts, Ridley discussestwo pieces of music, one for which the postulation of a persona enhancesappreciation, and one where it does not.

3. Musical Meaning

The third section of the volume contains two papers each concernedwith musical meaning, in some sense. The first contribution, by JeneferRobinson, is concerned with the relation between music and metaphor.Meanwhile, in their joint contribution, Tamara Balter and Eddy Zemachaddress the issue of music which apparently has ironic meaning by presentinga semantic theory of irony, understood as an aesthetic property which maybe possessed by many kinds of music, and in many different guises.

In ‘Can Music Function as a Metaphor of Emotional Life?’ Robinsonexamines whether music can function as metaphor: while she eventuallyconcludes that strictly speaking, it cannot, she also urges that we seepieces of music, potentially, not just as having aspects apt for metaphoricaldescription in expressive terms, but, more radically, as metaphoricallyexemplifying expressive qualities, by referring to extra-musical aspects ofthe world via certain expressive aspects possessed by them. (FollowingGoodman (1976), ‘expressive’ is used here in a generous sense to include,not only emotional qualities, but also those picked out by adjectives suchas ‘weighty’, ‘watery’, ‘sunny’, ‘twittering’ and so on.)

Robinson accepts a point rejected by Boghossian in Chapter 5: namely,to say that a piece of music has expressive properties is to make ametaphorical claim. In this, she follows Nelson Goodman. Goodman hasargued that musical expression is a kind of exemplification, where (roughly,but accurately enough for our purposes) the exemplified qualities of anobject are those which are possessed by the object; and to which the

introduction 13

object refers. Specifically, expression is metaphorical exemplification: theexpressive qualities of a musical work are those which are metaphoricallypossessed by the work; and to which the work refers (Goodman 1976: 95).Though critical of aspects of this account, Robinson is sympathetic to itinsofar as it draws attention to the important role played by metaphors inthe description of music’s expressive qualities. However, she thinks thatGoodman does not fully exploit the idea of metaphorical exemplification,restricting his attention to the metaphorical possession of mere qualitiessuch as sadness or heaviness. To remedy this, Robinson makes a distinctionbetween two kinds of metaphorically possessed qualities: those that belongto the ‘musical surface’ of a work, such as sadness and heaviness, and thosedeeper metaphors that characterize the unfolding of extended musicalpassages, perception of which structures one’s experience of the work in amore radical way.

Further examples of the former include breeziness and fieriness; examplesof the latter are ‘perseverance through adversity’ and ‘struggle leading tovictory’. Goodman is concerned only with the former sort in his discussionof expression; yet, Robinson notes, if surface features of a work are apt fordescription in metaphorical terms it does not yet mean that the work refersto qualities of breeziness or fieriness, understood as extra-musical phenom-ena. So breeziness and fieriness need not be exemplified by the work, in astrong sense. In contrast, she suggests, where a work, understood as a whole,has expressive qualities such as perseverance through adversity, this phe-nomenon, understood extra-musically, is genuinely referred to by the work.In this sense, we can say that perseverance through adversity is (part of) theexpressive meaning of the work, and is metaphorically exemplified by it.

This is not, however, to say that the listener’s attention is distractedaway from the work, Robinson argues, because where a work refers tosome extra-musical phenomenon in this way, it has a structure isomorphicto it, and so reflection upon the relevant phenomenon leads one backtowards the relevant aspect of the work and to enhanced understandingof it. A well-known objection to this sort of view is that the structureof a work must underdetermine what extra-musical features it refers to.Robinson thinks this objection can be seen off if we acknowledge thatmetaphorical talk of music is guided by ‘umbrella’ conceptual metaphors,which determine the course of our experience of a given work and whatfurther metaphors we think of as appropriate to apply to it. Hence, we

14 kathleen stock

should say in response to the objection, not that a given work refers tosome extra-musical phenomenon absolutely, but that it does so relative tosome conceptual metaphor applied either to the work, or to the musicof which the work is an example. Meanwhile, what conceptual metaphorit is appropriate to apply to a work will depend upon which conceptualmetaphors it was conceived under at the time of composition; this isillustrated by Robinson with an extended discussion of the metaphor ofabsolute music as ‘discourse’, current in the eighteenth century as theform emerged. She goes on to suggest that the noticing of the way apiece of music metaphorically exemplifies, say, a conversation amongstinterlocutors can enable us to hear emotional expression in the voices, andhence, the piece. None of this, however, forces us to admit that musicalworks themselves can be metaphors, in a strict sense. They cannot be, sincethey have no literal meaning to start with.

In ‘The Structure of Irony and How it Functions in Music’, EddyZemach and Tamara Balter are concerned with musical meaning of adifferent kind. According to them, irony, whether verbal, situational,dramatic, general, parodic, or romantic, has certain necessary features.Namely, where a situation, whether in art or reality, is perceived asironic it (i) causes some person, usually the perceiver, to project (thatis, think of) a possible situation, where the possible situation thought of(ii) is thought of as ‘genidentical’ to the actual one considered (i.e. thetwo constitute versions of a single ‘transworld individual’ and as such are‘counterparts’ of one another). In this sense, irony is essentially modal. Afurther characteristic feature of most varieties of irony is (iii) that the personconcerned is caused to think unfavourably of the actual situation whichconfronts him, in comparison to its projected counterpart. Hence, on thisaccount, the use of words is not essential to irony: what is essential is theprojection of a possible situation against which reality may be compared,usually unfavourably, which is something which might be accomplishedby music as well as words.

According to Balter and Zemach, many forms of irony, including musicalirony, fit this basic template. Much of the argumentative weight of theauthors’ claims depends on their close readings of particular musical works,to which I am unable to do justice here. Instead I shall focus on the basicclaims they make about different varieties of irony, and about how eachmay be manifested in music of a certain character.

introduction 15

The first sort of irony discussed is verbal irony, which involves under-standing a sentence as ironic. Understanding a sentence ‘p’ as ironic, weare told, is equivalent to its causing one to project a possible situation thatsatisfies ‘p’, where this situation (i) is understood to contrast ‘sharply’ withits ‘real-world counterpart’, i.e. the relevant situation which obtains in theactual world, and with which it is thought of as genidentical; and (ii) isunderstood as superior in nature to that counterpart, so that it ‘mocks’it. Since music does not have the kind of content that a sentence does,unless it includes lyrics, verbal irony is rarely found in music. However, itis argued, other types of irony, understanding of which does not dependon understanding sentences, are more frequently found in music, and it isto these that the authors turn next.

Broadly speaking, these can be divided into two categories: on theone hand, those in which a possible situation is projected, and which isunderstood as superior to the actual situation of which it is a counterpart;and on the other, those in which a situation is projected on to the actualone and which is understood as unsatisfactory or deformed in some way.

As well as verbal irony, situational and dramatic irony fall into the firstcategory. Situational irony, occurs, it is claimed, when an actual situation isviewed by an onlooker with the thought of an idealized possible situationin mind, against which the former is unfavourably compared. The idealizedsituation comes to mind as the result of the onlooker’s apprehensionof the actual situation, in conjunction with internalized norms whichbring to mind how the situation should be, ideally. The category of‘situations’ is here as elsewhere construed broadly, as potentially includingmusical events: hence, music can convey situational irony by includingan actual compositional feature which is ‘mocked’ or ‘contradicted’ bythe feature one expects to find, according to familiar norms governingcomposition.

In the same category is dramatic irony, which can occur in fictions,including musical fictions such as opera, as well as in ‘real life’. In dramaticirony, a superior counterpart to the actual situation (or actual situation asdepicted by the fiction) is projected, not, this time, by the onlooker whoperceives the situation as ironic, but by some other (fictional) protagonistwho believes the counterpart to be actual, whereas the onlooker sees thatin fact this is not the case. The effect of this is to make the projectedsituation, and so too the projector, look ridiculous. Dramatic irony can be

16 kathleen stock

signalled in music, not only in the fictional narrative of an opera, but alsoinsofar as music can signal to the listener, via her understanding of featuresof the work’s composition, that a fictional world is not as a protagonisttakes it to be.

Meanwhile, into the second category falls parody, and general andromantic irony. A parodic work is one intended to cause its spectator torealize some unpleasant truths about a particular actual situation, by causingher to project an only slightly deformed counterpart which she initiallymistakes for the actual situation in question. Realization of her mistakethen causes her to identify the deformation in the projected situation aspresent in the actual one. In this way, the work powerfully conveys somecritical point about actuality. Likewise, in music, a work which is a parodyof some other musical work or feature presents itself as a deformed versionof its target, in order to persuade the listener that some of the unflatteringaspects of the parody actually belong to the target as well.

Romantic irony is similar to parody, in that a romantically ironicwork is intended to cause its spectator to project a deformed versionof some actual entity, so as to mock the latter; the difference beingthat in this sort of irony, the actual entity which is the target of themockery is the work itself, or at least, some aspect of it. A piece of musicdisplays romantic irony where it contains some aspect which is presentedas genidentical to some other aspect of the same work, to which it isjuxtaposed or in close proximity, and to which it contrasts sharply in orderto mock it.

Finally, general irony occurs wherever a situation causes an onlooker toproject a number of counterpart situations, all of which she considers to beas deformed as the actual one. An art work manifests general irony whenits narrative takes the audience to a number of ‘worlds’, each of which issuccessively or simultaneously ‘rejected’ without any accompanying senseof being rejected for something better. In the musical sphere, generalirony occurs, for instance, in the simultaneous presentation, in a singlepiece, of diverse ‘levels, styles and topics’ which cannot be made to coheremelodically, harmonically, rhythmically, thematically or stylistically; as, forinstance, in Stravinsky’s Three Pieces for String Quartet, which is illuminatinglydiscussed.

introduction 17

4. New Issues

The two papers in this final section are distinguished by their carefulattention to important issues which tend to be somewhat left in the philo-sophical shade besides those of musical ontology, expression and meaning.In his contribution, Gordon Graham addresses himself to developments incontemporary composition, enquiring, by way of a comparison with ourexperience of classical music, whether electro-sonic art can be called musicat all, properly speaking. Meanwhile, in his paper, Roger Scruton offers usa detailed analysis of a crucial yet neglected aspect of musical experience:musical rhythm.

In ‘Music and Electro-sonic Art’ Graham is concerned with the provoca-tive question of whether a particular genre of composition should becategorized as music at all. Namely, he asks whether electro-sonic artworks—digitally produced sequences of atonal sounds, by composers suchas Lutoslawski and Varese—can be called music, properly speaking; and,furthermore, whether they afford the same opportunities for value as domusical works, either in terms of listening experience, and in terms ofactive engagement in their production.

His answer depends on a prior understanding of the value of music.Music, according to Graham, is ‘intentionally organized sound’, and, assuch, is necessarily to be listened to: not simply as a means to some non-auralexperience only contingently available via aural experience (say, that of anexpressed or felt emotion, or represented image or idea), but rather as an endin itself. His is not the claim that the content of music is essentially sui generis;rather, it is the claim that, whatever the content of music, it is necessarilyaccessed aurally, so that its sonic character is part of its value. In this respect, itcontrasts with most speech tokens, whose valuable properties are only con-tingently aural ones, usually to do with the communication of information.Meanwhile, another part of music’s value which issues from its essentiallyaural nature are the multiple opportunities it affords to engage in its produc-tion: most obviously, by composing or performing, but also by the inventionand production of instruments, the management of acoustics, and so on.

Graham then moves to discuss electro-sonic art works, understood(roughly) as composed sound sequences which mainly employ as their basic

18 kathleen stock

constituents sounds other than the twelve tones of the chromatic scale.He notes that works in this form are, like musical works, intentionallyorganized sound sequences necessarily orientated towards the enrichmentof aural experience, and not simply intended as a contingent means tovaluable non-aural experience: however, he denies that such works countas music, since, in their wholesale abandonment of tones, they differ soradically from existing music. Such a denial is not meant to entail anyparticular evaluative stance towards electro-sonic art. This is made clearby subsequent discussion, where Graham gives a partly positive responseto the question of whether electro-sonic art can enrich aural experienceto the extent that music can, granting that it may share many structuralfeatures with music, and may turn out to share expressive features withit too. However, he notes a further difference between electro-sonic artand music which, as well as underlining the difference between the two,has implications for the comparative value of the former against the latter:namely, that electro-sonic art does not provide as many opportunitiesas music does for active engagement in its production, and hence cannotmatch music’s value in this respect. In particular, he suggests, the productionof electro-sonic art allows little conceptual space for performance of it,distinct from composition.

The second contribution to this section is that of Roger Scruton. In it,he focuses on musical rhythm, analysing it first and foremost as a feature ofmusical experience and distinguishing it from any purely structural featuresof composition, such as metre. He grants that the rhythm of a work isoften established via its metre, which organizes a work into repeatablesegments, and can lead us to hear certain notes within each segment asstressed or unstressed. However, the existence of metre in a work is notyet the experience of rhythm, Scruton argues.

That it is insufficient for rhythm is demonstrated by works without adiscernible rhythm from non-classical metric traditions, both non-Westernand modern; meanwhile, that it is unnecessary is shown by rhythmicnon-musical phenomena, such as speech and dancing. Furthermore, withinmusic, Scruton notes, rhythm can be generated or at least enhanced, notjust by accents and stress, but by melody and harmony as well. Such factslead Scruton to distinguish between the ‘beat’ of a work—the patternof time values and accents it contains—and its ‘rhythm’. Unlike beat,rhythm can emerge from melodic and harmonic factors as well as metric

introduction 19

ones, and the interplay between them, and is a way of hearing a work asmoving in a certain way. Whereas the beat of a work can be conveyedvia symbols, experience of musical rhythm can only be characterizedphenomenologically, and is identified by the ear, and perhaps also bythe body, in dance. Such claims reflect Scruton’s wider view that theexperience of music is an ‘acousmatic’ one, in the sense that it is heard asgoverned by a ‘virtual causality’, linking one sound to the next via suchmetaphorical aspects as movement, position and distance, and not whollydependent on the organization of causal features of its constituent sounds.

An important distinction is drawn by Scruton between ostinato rhythm,in which a rhythm is imposed via factors ‘external’ to the melodic andharmonic lines, and rhythm which is internal to, and emerges from,melodic and harmonic movement. An example of ostinato rhythm is thattypically found in pop and dance music, where the rhythm is imposed viaa percussive beat wholly distinct from melodic or harmonic constituents.In contrast, in the predecessors of pop music, jazz and early rock, as inmost classical music, percussion is typically used to enhance a pre-existingrhythm generated from ‘within’, either by instrumental or vocal aspects ofmelody or harmony. In atonal music, Scruton suggests, rhythm of this kindis difficult to produce, since usually it relies on an impression of movementand return, which in turn usually depends on the use of harmony andmelody. Often composers of atonal music are forced to compensate byintroducing rhythm from outside, as it were, via ostinato, or some metricaldevice. An exception here is Messiaen, and Scruton discusses his techniquesfor generating rhythm internally in some depth. He identifies Wagner asa precursor to Messiaen in this respect, as an innovator in his ability togenerate rhythm as an internal feature without reliance on techniquesfamiliar from the classical tradition. Another way in which the atonalcomposer can generate rhythm is by incorporating speech rhythms into awork, and in this regard, also, Wagner is identified as an innovator.

Meanwhile, from the distinction between rhythm which is generated aspart of the ‘musical surface’ of a work, and ostinato imposed from without,Scruton extracts some practical and even moral consequences for the natureof dances associated with each. For one thing, he suggests, the prevalence ofmodern dance music employing ostinato has perniciously narrowed musicalpreferences towards music of a similar kind. Perhaps more seriously, heargues that, whereas dance music which generates rhythm from within

20 kathleen stock

guides people as to how they should move, thereby facilitating socialinteraction and directing their attention towards other people, dance musicwhich uses ostinato, in contrast, does not facilitate and even inhibits suchinteraction, encouraging the individual to become isolated from others,and ‘lost’ in the music in a morally detrimental way.⁴

References

Davies, S.(1997) ‘Contra the Hypothetical Persona in Music’, in M. Hjortand S. Laver (eds), Emotion and the Arts, Oxford: Oxford UniversityPress, 95–109; reprinted in his Themes in the Philosophy of Music, Oxford:Oxford University Press (2003), 152–68.

Dodd, J. (2000). ‘Musical Works as Eternal Types’, British Journal of Aesthetics40: 424–40.

(2002) ‘Defending Musical Platonism’, British Journal of Aesthetics 42:380–402.

(2004). ‘Types, Continuants and the Ontology of Music’, BritishJournal of Aesthetics 44: 342–60.

Goodman, N. (1976) Languages of Art, Indianapolis: Hackett.Levinson, J. (1980). ‘What a Musical Work Is’, The Journal of Philosophy 77:

5–28; reprinted in his 1990b, 63–88.(1990a) ‘Authentic Performance and Performance Means’, reprinted

in his 1990b, 393–408.(1990b) Music, Art and Metaphysic, Ithaca, NY: Cornell University

Press.(1996) ‘Musical Expressiveness’, in his The Pleasures of Aesthetics, Ithaca,

NY: Cornell University Press, 90–125.(2005). ‘Musical Expressiveness as Hearability-as-Expression’, in

M. Kieran (ed.), Contemporary Debates in Aesthetics and the Philosophyof Art, Oxford: Blackwell.

Matravers, D. (1998) Art and Emotion, Oxford: Oxford University Press.Scruton, R. (1999) The Aesthetics of Music, Oxford: Oxford University Press.Wolterstorff, N. (1980) Works and Worlds of Art, Oxford: Clarendon Press.

⁴ Thanks to those contributors who offered helpful criticism of an earlier draft of this Introduction.

PA RT I

Musical Ontology


1Sounds, Instruments, and Worksof MusicJ UL IA N D OD D

1

A satisfying account of the ontology of works of music must do two things.¹First of all, it must enlighten us as to musical works’ ontological nature: that is,tell us to which ontological category such works belong. Building on this,it must then explain how musical works are individuated: that is, provide aninformative account of when we have one and the same work of musicand when we have numerically distinct such works. What I shall termthe simple view does just this.² Its first component—what I shall call thetype/token theory—has it that works of music are types whose tokens aresound sequence-events: datable, locatable patterns of sounds.³ Its secondcomponent—sonicism—has it that work W = work W ∗ if and only if W

¹ By ‘works of music’ I mean works of pure, instrumental music. This restriction should beunderstood to apply throughout this paper.

² The simple view (or, at least, something approaching it) is defended in my 2000, 2002, 2004 and2007.

³ Such sound sequence-events will tend to be either performances: sound sequence-events producedby someone performing the work; or else playings: sound sequence-events that involve either thereproduction of such performances via CDs and the like, or the use of a playback artefact (such as aplayer piano) in the production of the sounds. However, I see no reason to rule out the possibility thata work could have tokens that were neither performances nor playings. The wind blowing through thetrees could produce a token of ‘Greensleeves’. (This example is Wolterstorff ’s (1980: 86), though he isless happy about accepting it as a consequence than I am. According to Wolterstorff, whilst there is noconclusive reason to restrict a work’s tokens to performances and playings, such a restriction is desirablesince it accords with our pre-theoretical intuitions about what kinds of thing do, and do not, count asinstances of works. I am unpersuaded by his discussion (1980: 86–8), but will say no more about thislocal dispute.)

24 julian dodd

and W ∗ are sonically indistinguishable. However, both of these claims needto be further clarified and rendered more precise.

To begin with, the type/token theorist should treat the types that areworks of music as norm-types: types that admit of properly and improperlyformed tokens. In this respect, works of music resemble types such as ThePolar Bear and The Piano:⁴ something can be a polar bear and yet bemissing an ear; something can be a piano and yet be out of tune; andsomething can genuinely count as a performance of a work and yet containmistakes here and there. As Nicholas Wolterstorff has explained (1980:54–8), granting this latter point is essential, if we are to allow for the factthat a work can be improperly performed. Such incorrect performances areflawed but performances of the piece nonetheless.

With this in mind, let us follow Wolterstorff (1980: 58) in saying thata property F is normative within a type φ just in case (i) φ is a norm-type;and (ii) it is impossible for there to be something that is a properly formedtoken of φ and which lacks F. Consequently, for any work of music W ,there is a set � that comprises the properties normative within W . In orderfor a performance to be a properly formed token of W , it must have everymember of �; to count as a performance of the piece at all, a sound-eventmust not lack too many of �’s members.

This claim is fine as far as it goes, but merely serves to raise the question ofhow such types are individuated, albeit in a new form. For the question nowbecomes: what is the nature of the properties that are members of �? Thesecond component of the simple view—sonicism—has a straightforwardanswer: the members of � are all purely acoustic in character, and, hencework W = work W ∗ if and only if W and W ∗ have the same—purelyacoustic—properties normative within them. According to a sonicist,whether a sound sequence-event is a well-formed instance of W is solelydetermined by how it sounds. Nothing else matters.

2

Having distinguished the two components of the simple view, I shall now putthe first to one side and focus squarely on the second. In particular, I want

⁴ I capitalize names of types.

sounds, instruments, and works of music 25

to consider whether the sonicist is right in thinking that the properties nor-mative within works of music comprise merely acoustic properties and not,in addition to these, performance-means properties: properties concerning howa performance’s constituent sounds are produced. Boiled down to its bareessentials, my question is this: can a properly formed token of a work W onlybe produced by means of the particular instrumentation specified in W ’s score?

According to the sonicist, of course, the answer to this question is aresounding ‘No’. Since the only members of � are acoustic properties,a properly formed instance of W may be produced by other means ofsound production than those specified by the composer. A properly formedperformance of the Hammerklavier Sonata, for example, need not haveits constituent sounds produced by the playing of a piano; the soundsconstituting such a performance could also be produced by the operationsof a space-age Perfect Timbral Synthesizer: a machine that emitted acousticfacsimiles of piano-sounds. For the sonicist, whether a sound sequence-event faithfully represents the piece is determined by its acoustic appearancealone. How these sounds are actually produced is by the by.

Swimming against a strong intellectual current, I wish to defend a versionof this sonicist claim. And in so doing, I reject what I shall call instrumentalism:the view that a properly formed token of W must be produced by theinstrumentation called for by W ’s score.⁵ The instrumentalist insists thata synthesized performance of the Hammerklavier—even if acousticallyindistinguishable from a performance she acknowledges to be properlyformed—would nonetheless be a defective presentation of the piece. Andher saying this amounts to a claim that the sonicist’s description of theproperties that constitute � is incomplete. For the instrumentalist has it thatthe properties normative within W include not only acoustic properties,but precisely those performance means-properties that the sonicist ignores.Clearly, if the instrumentalist is right about this, then the sonicist hasgiven, at best, only a necessary condition for work-identity. For once itis granted that performance means-properties are normative within works,we have no choice but to admit that W and W ∗ —even if acousticallyindistinguishable—are distinct, if their respective scores specify differentmeans of performance.

⁵ Those who argue vigorously and ingeniously against sonicism from the perspective of instrumen-talism include Stephen Davies (2001: 60–71), Stan Godlovitch (1998), Jerrold Levinson (1980, 1990a,1990b) and Kendall Walton (1988).

26 julian dodd

In what follows I shall argue that a suitably nuanced sonicism canrepel the challenge presented by instrumentalism. And my route to thisconclusion will be via two claims. First, I shall present one particular variantof sonicism— timbral sonicism—as the face-value theory when it comes tothe question of musical works’ instrumentation. It is, I shall claim, thetheory that we should accept unless it is defeated. Second, and crucially, Ishall argue that the arguments offered by instrumentalists are insufficient todefeat it. To be sure, sonicism is an unfashionable doctrine. But then again,aestheticians, of all people, should not be fashion victims.

3

Let us treat � as the set of properties normative within the Hammerklavier:the set of properties a sequence of sounds must have to be a properly formedtoken of the piece. The sonicist, we have seen, takes these properties tobe purely acoustic in nature: features concerning the pitch and duration ofnotes, melodic, harmonic and articulational features, and such like. Sucha view, I have claimed is prima facie correct. Why is this so? For tworeasons. First of all, sonicism—or a version thereof—is the default positionwhen it comes to musical works’ individuation. There is something deeplyintuitive about the idea that works of music are pure types of sound-event:in other words, that what makes a certain work that work is simply thatits performances should sound like that. And I am not alone in thinkingthis. Even Jerrold Levinson, one of sonicism’s most articulate and ingeniousopponents, accepts that, when it comes to the types that are musical works,‘the most natural and common proposal ... is that a musical work is a soundstructure—a structure, sequence, or pattern of sounds, pure and simple’(1980: 64).

Ultimately, I doubt whether adequate sense can be made of the claimthat a type has parts or constituents,⁶ which is why I have set up thesonicist thesis without any talk of ‘sound structures’ or the like. Butwhatever view we take on the ontological nature of types, Levinson’ssuggestion that sonicism is intuitive and natural is surely correct. Someoneunencumbered by developments in analytical aesthetics will most likely

⁶ For an attack on the idea that types are structured, see my 2007: Ch. 2.


accept that a performance of the Hammerklavier on a Perfect TimbralSynthesizer, indistinguishable to the ears from a performance on a piano,would be a no less satisfactory presentation of the piece. True enough,if such a synthesizer were easier to play than a piano, we might thinkless of the performer’s achievement, but the respective patterns of sounds theperformers produced—qua representations of the piece—would surely beregarded as being on a par. Intuitively, the vehicles of musical meaningare the sounds we hear—to appreciate a work of music we need only useour ears—and in the example we are considering, the performances inquestion are acoustic doppelgangers.

Here, then, we have a bare intuition pointing towards sonicism. Butfurther weight may be lent to the sonicist cause by considering how wellit coheres with a plausible account of the nature of musical experience.Roger Scruton describes our musical experience as acousmatic (1997: 2–3);and what he means by this is that, in hearing sounds as music, we attendpurely to the sounds themselves, and not to their causal origin. Scrutonbelieves that our experience of sounds as music sees us detach these soundsfrom the circumstances of their production, and hear them as organizedaccording to pitch, rhythm, melody and harmony (1997: 20). We hearthe sequence of sounds as an organized whole, something that developsand progresses, so that, rather than focusing on the fact that a middle C isproduced by an oboe, for example, we hear it as the response to the B thatpreceded it, and as calling for the E that follows it (Scruton 1997: 19).

Obviously, if this view of our musical experience is correct, instrumen-talism takes on a hugely uncompulsory air. For if hearing the music as itought to be heard requires us to attend merely to its qualitative nature, andnot to the ways in which the sounds are produced, then nothing aboutthe nature of our musical experience threatens to dislodge us from thedefault position that is sonicism. Having said this, Scruton’s account ofour experience of music is flawed. For, as we have seen, Scruton believesthe only essential features of our musical experience to be organizationalfeatures: pitch, rhythm, harmony and melody. As a result, he comes toregard timbre as an inessential feature of our musical experience, eventhough he accepts that it can contribute to musical meaning (1997: 77). Inmy view this is a mistake: timbral properties, as we shall see in a moment,are normative within works, for the reason that their presence determinesmany of a work’s aesthetic properties; so even though timbre is not itself

28 julian dodd

an organizational feature of sound sequences, an appreciation of timbre isessential to our experience of sounds as music.

What follows, if we depart from Scruton’s position this way? As Scrutonhimself notes (1997: 77), timbral features are specified by reference to theircustomary origin: if we are to avoid recourse to metaphor, we have nochoice but to conceive of a certain timbre as ‘the sound of a middle Cplayed on an oboe’, for example. Consequently, it follows that Scrutonis wrong in thinking that our musical experience can always be describedwithout making any reference to sounds’ usual origin. But, as it happens, thisturns out to be a relatively small amendment, for what we might terma modified acousmatic account of our musical experience is still compatiblewith a denial of instrumentalism. Whilst it is true that we must makereference to certain sounds’ customary causal origins in order to put ourfinger on their timbral aspects, what does not follow from this is that weneed be cognizant of the sounds’ actual origins; and this leaves it open forthese sounds to have been produced by non-standard means. We do ‘nothave to identify [the sounds’] cause in order to hear them as they should beheard’ (Scruton 1997: 3); we need only characterize them in terms of theircustomary means of production. And if this is right, and if we grant thehighly plausible thesis that how a musical work should be heard is revelatoryof its ontological character, then our modified acousmatic account of thenature of musical experience strongly suggests that we need not departfrom sonicism. In order for an audience to hear a performance as it oughtto be heard, the performers need not use certain specific instruments; theyneed only produce sounds with the sonic appearance of sounds producedby such instruments.

Be this as it may, the sonicist is obliged to explain what she takes the‘auditory character’ definitive of a work’s identity to consist in. Whichacoustic properties comprise set �: the set of properties normative withinthe Hammerklavier? According to the view that Stephen Davies calls ‘puresonicism’ (2001: 60), the members of � are the sorts of properties one wouldexpect—melodic, rhythmic, harmonic, articulational, tempic properties,and the like—but with one notable exception: timbral properties. On sucha view, a performance of the Hammerklavier would be in no way defective,if produced on a Hammond organ, as long as the resultant sound sequencehad the tonal structure and other (non-timbral) acoustic properties indicatedby its score. In my opinion, though, such a position is too extreme to


be acceptable. Indeed, as I shall now explain, it is implausible in itself,inadequately motivated, and subject to a serious objection. However, anappreciation of pure sonicism’s demerits opens up the possibility of therebeing a version of the doctrine that avoids these pitfalls.

That the pure form of sonicism is counter-intuitive (almost) goes withoutsaying. A performance of the Hammerklavier on a Hammond organ or,perhaps, by a flute quartet,⁷ might capture the piece’s tonal make-up, butsuch a performance would entirely ignore the composer’s specificationof a piano in the score. As I shall argue presently, the fact that a scoreinstructs performers to use specific instrumentation does not entail that aproperly formed performance must use the instrumentation specified. Suchan instruction is essentially context-bound and provisional, its purposebeing that of instructing performers how to produce a properly formedsound sequence given the instruments available at the time. The goal of suchan instruction—the production of sound sequences of a certain qualitativecharacter—would be met even if such a sound sequence were producedwithout such instrumental instructions having been followed. Having saidthis, Beethoven’s specification of a piano on the score of the Hammerklavierdoes at least demonstrate that a correctly formed performance of the pieceshould be constituted by piano-like sounds. His scoring of the piece forpiano was not completely arbitrary; in specifying a piano he was makingit clear that performances of it would be inadequate, if constituted by thekinds of sounds made by organs, violins or flutes. The fact that Beethovenspecified that a piano be used, whilst not entailing that being produced by apiano is normative within the work, does serve to characterize the qualitativenature of the sounds that must constitute a proper instance. Such soundsshould have the timbral quality typical of sounds produced by such aninstrument.

Saying this, of course, serves merely to invite the sonicist to retreat fromthe pure version of the doctrine to its timbral counterpart: the version thattakes timbral properties to figure in �. As Levinson has noted, it is a deci-sive objection against pure sonicism that it renders some of a composer’schoices—for instance, trumpet or oboe, or bassoon or cello—motivelessand arbitrary (Levinson 1990a: 244). But the timbral sonicist preciselyavoids this charge by ensuring that timbral properties are normative within

⁷ This example is Levinson’s (1990a: 240).

30 julian dodd

musical works. Elsewhere, Levinson recognizes that timbral properties,though generally specified via performance means properties, are in prin-ciple ‘physically and logically separable from them’ (1990a: 220); and anappreciation of this possibility enables us to formulate the weaker, timbralversion of sonicism that can account for the importance of the choices thatcomposers, in practice, make.

It is true, of course, that a performance of the Hammerklavier on aHammond organ would still be recognizable as a performance of the piece;and this, perhaps, is the source of the pull that pure sonicism exerts oversome philosophers.⁸ But, having set out the type/token theory in the wayin which I have, such a motivation for this version of sonicism is swiftlyrevealed as misconceived. A performance using the Hammond organ could,indeed, count as a performance of the piece; but acknowledging that this isso is quite compatible with recognizing that a properly formed performancemust be made up of piano-like sounds. A work of music is a norm-type,remember: its identity is determined by the properties that are normativewithin it. Consequently, the mere fact that there could be a performanceof the Hammerklavier on a Hammond organ (if such a performance had asufficient number of the other properties normative within the work) doesnot demonstrate that timbral qualities are not members of �. It is a familiarfact that a work may have instances that are improperly formed. After all,that there can be performances (albeit imperfect ones) of the piece that missout certain passages, play wrong notes, or misconstrue the rhythm does notshow that a properly formed token of the work need not obey the score inthese respects.

Pure sonicism is thus an under-motivated doctrine. But it also faces adecisive objection whose origin lies in the work of Levinson (1980: 73–8).For Levinson draws our attention to the fact that many of a work’s aestheticand expressive properties depend upon their timbral, in addition to theirmore broadly acoustic, features. The Hammerklavier, Levinson notes, is a‘sublime, craggy and heaven-storming piece of music’ (1980: 76), but asound sequence would fail to represent it as such, if it were producedby a Hammond organ or (worse still) by a recorder ensemble.⁹ Likewise,John Cage’s In a Landscape would lose much of its gentle, exploratory,

⁸ Over Peter Kivy (1988), for example.⁹ The latter example is Stephen Davies’s (2001: 64).


contemplative character, if played on a trombone. What examples such asthese reveal is that a properly formed performance of a piece—one that,among other things, does justice to the work’s aesthetic content—musthave a specific timbral character. It is insufficient merely for such aperformance to be properly formed with respect to its other audiblefeatures.

This being so, it is my contention that a philosopher appreciative ofsonicism’s status as the face-value theory, and yet sensitive to the problemsbesetting sonicism in its purest form, should adopt timbral sonicism. But bethis as it may, the instrumentalist insists that a proper understanding of therole played by scores, and of the features determinative of a work’s aestheticcontent, reveals that performance means-properties, and not merely timbralproperties, are normative within the types that are works of music. In thenext two sections I explain why such a conclusion is unwarranted.

4

As we have seen, the crux of dispute between the timbral sonicist andthe instrumentalist concerns whether performance means-properties arenormative within works. If they are, then, pace timbral sonicism, W andW ∗ may yet be numerically distinct even if sonically indistinguishable (ifthey differ with respect to the performance means-properties normativewithin them). Consequently, given timbral sonicism’s status as the face-value theory of musical works’ individuation, we need to consider whetherthe instrumentalist’s arguments for performance means-properties beingnormative within works are sufficiently strong to warrant our giving upwhat amounts to the default position.

The first such argument concerns the nature of a work’s score. In §3 Icharged the pure sonicist with disregarding the fact that composers specifythe use of certain instruments to produce sounds. Could it not be arguedthat this manoeuvre equally demonstrates the falsehood of even timbralsonicism? Does not such an appeal to the specification of instruments inscores throw one directly into the arms of instrumentalism? Levinson, forone, thinks that it does. In his view, scores are definitive of musical works,and so, if Beethoven’s Quintet Op. 15 calls for a clarinet to be played,this means that ‘the Quintet Op. 15 without a clarinet is not the same

32 julian dodd

piece—even if all sound-structural characteristics (including timbre) arepreserved’ (1980: 75). Levinson’s point, in other words, is that Beethoven’sspecification of a clarinet means that the work essentially involves theclarinet, and hence that a sound-event cannot be a properly formedinstance of the piece unless a clarinet has been used in the production ofthe sounds.¹⁰ If Levinson is right, being produced by a clarinet is normativewithin the work.

However, as I briefly explained in §3, this way of understanding theinstrumental instructions in scores is not obligatory. Indeed, once oneaccepts timbral sonicism as the prima facie, natural way of individuatingworks of music, one will interpret scores’ instrumental specifications inanother way entirely. This alternative reading has it that the way tomake sense of the fact, for example, that the Hammerklavier’s score callsfor a piano is to see this apparently exceptionless instruction as havingan altogether more provisional nature. For the philosopher impressed bytimbral sonicism’s intuitiveness, a work’s score specifies how a soundsequence-event must sound, if it is to be a properly formed instance of thework; and the score’s specification of certain instrumentation should be seenin this context: as exclusively serving the aim of facilitating the production ofsound sequence-events with certain specific timbral features. True enough,the Hammerklavier’s score contains an instruction to produce such sounds incertain specific ways, i.e. by means of a piano. But this does not entail that aproperly formed performance must involve the specified instrumentation, asopposed to some other aurally indistinguishable means of sound production.For this instruction is aimed at a particular constituency—namely, potentialperformers of the piece at the time the piece was composed—and iswholly in the service of the production of sound sequence-events of acertain qualitative nature. In short, the score was instructing performers in1818 to use a piano simply because no other means of producing soundswith the requisite timbral features existed at that time. However, since thepurpose of such an instruction is merely to specify the tone colour thatmust be had by a well-formed instance of the work, it is an instructionthat can, in principle, be overruled as long as sounds with the right timbralqualities are produced.

¹⁰ This position is endorsed by Currie, who claims that ‘[a] performance that violates the composer’sdirections as to how the sounds are to be produced is not a correct performance of it’ (1989: 49).


A predictable objection to this is to point out that composers often believetheir own instrumental instructions to have the status of exceptionlessinstructions for performers. It may well be the case, for instance, thatBeethoven, had he been given the choice, would have preferred, forwhatever reason, a performance of the Hammerklavier on the piano ratherthan one that made use of a Perfect Timbral Synthesizer. But the crucialpoint is that such a consideration cuts little ice against the timbral sonicist’sarguments. Convinced by her doctrine’s status as the default position, andimpressed by the way in which it coheres with the modified acousmaticaccount of our musical experience, the timbral sonicist concludes thatperformance means-properties cannot be normative within works, andhence that instrumental specifications in scores should be interpreted in thealternative way suggested.

Indeed, it seems to me that our objector may well have been misled bya misunderstanding of the sonicist’s claim about the instrumental specifi-cations in scores. That such specifications serve solely to characterize thequalitative nature of the sounds constituting a properly formed occurrenceof works is not a psychological claim concerning the nature of composers’intentions; it is a claim concerning how such specifications should beinterpreted in the light of mature philosophical theory concerning both theindividuation of works of music and the nature of our musical experience.The timbral sonicist denies that the proper way to perform a work cansimply be read off from the composer’s actual (or hypothetical) preferences,and hence allows that composers can misinterpret their own works.

Naturally, saying this will not of itself placate the instrumentalist. Davies,for example, would be tempted to reply by drawing our attention tokinds of scores that seem especially concerned with the production ofsounds on specific kinds of instrument. One such example is provided bysixteenth-century lute tablature, in which the notation tells the performerwhere to place his hands on the neck of the lute. Such a score, Daviessuggests can only be ‘[i]nstrument specific. Rather than thinking of hispiece as an abstract sound structure and being indifferent to the mannerof realizing it, [the composer] is addressing himself expressly to a musicianwho is holding a lute in his hands’ (Davies 2001: 62). But such an exampleprovides no new challenge to the timbral sonicist. Right enough, theproducer of such a score addresses himself to someone holding a lute, butit does not follow from this that a properly formed token of the piece must

34 julian dodd

be produced by a lute rather than by an artefact that produces lute-likesounds. The same kind of response as before suffices to make the point:although the composer addresses himself to lute players, his instructionsare (dare one say it) to be interpreted instrumentally;¹¹ they functionmerely as a means of specifying the qualitative nature of the sounds thatperformers must produce. So, given that sounds with such a qualitativenature could also be produced by non-lutes, a performer may ignore theletter of the score’s instrument-specific instructions, just as long as she is ableto produce sounds indistinguishable from those produced by a lute player.Whether he recognizes it or not, how the sounds are produced is onlyof practical, de facto interest to the composer: an interest in performancemeans subserves the aim of specifying the acoustic appearance of the soundsthemselves.

Levinson makes a similar objection to Davies’s, but, ultimately, to nogreater effect. Levinson too objects to the sonicist’s conception of a score asa recipe for providing an instance of a type of sound-event for its own sake.As he sees it, ‘[w]hen Beethoven writes a middle C for the oboe, he hasdone more than require an oboelike sound at a certain pitch—he has calledfor such a sound as emanating from that quaint reed we call an ‘‘oboe’’’(1980: 74). And one reason why Levinson takes this line is that he believesit to be the only position that makes sense of the way in which composersactually think of the tone colours that they wish to see produced. As hehimself puts it:

[c]omposers are familiar with tone colors only insofar as they are familiar withinstruments that possess them. We do not find composers creating pure combina-tions of tone color, and then later searching about for instruments that can realizeor approximate these aural canvasses; it would obviously be pointless or at leastfrustrating to do so.

(Levinson 1980: 74)

To my mind, though, this objection sees Levinson mistake an epistemolog-ical fact for a metaphysical one. For the fact that a work’s composer describesthe kinds of sounds he wants produced by referring to their customaryorigins does not entail that only sounds with such origins can figure in acorrect performance. A timbral sonicist, we noted in §3, accepts that sounds

¹¹ This joke is Kivy’s (1988: 85).


must be thought of in terms of their customary origin, if their timbralaspects are to be characterized. What she denies is that this connection inthought between sound and customary origin need actually obtain for a per-formance to be properly formed. As we have just seen, there is a defensibleinterpretation of scores available to the timbral sonicist, according to whichinstrumental specifications serve merely as a means for characterizing howcorrect performances should sound. In the light of this, pointing out whatwe all know—namely, that we think of tone colours by means of thinkingof their customary origin—is beside the point.

Once this observation has been made, Levinson’s attempts to bolsterwhat we might call the argument from scores against timbral sonicism aredoomed. No doubt, he is right to point out that composers often conceiveof a use of instrumentation prior to conceiving of pure types of sound:double-stopping is an example of this, as is pizzicato. We may agree withLevinson that ‘ideas for pizzicato passages do not first occur to composersand then, only later, what will serve, for the nonce, to realize them’ (1990a:244). But, to reiterate, this evident fact does not have the implications forthe conditions of correct performance that Levinson takes it to have. Factsabout how composers conceive of sounds do not determine which kindsof properties are normative within works.

What, then, is left of the instrumentalist’s appeal to the authority of scores?Not much. One might try to bolster it by adopting what Stan Godlovitchdescribes as a ‘socio-historical perspective’ on the question (1998: 56).According to Godlovitch, once one understands that musical communitiesare akin to guilds—defining their own conditions for membership andrank—one comes to appreciate that such communities will automatically behostile to the idea of performing the Hammerklavier upon a Perfect TimbralSynthesizer, even if such a performance were an exact acoustic replicaof a performance on a piano. Guilds, reports Godlovitch, ‘systematicallyresist instrumental innovations in order to preserve their own structurewhich requires the establishment and maintenance of a skill hierarchybased on handicaps legitimised by the Guild’ (1998: 53). According to themusic-making establishment, then, the value of a performance is not simplydetermined by how it sounds, but by how it is achieved. In particular, theperformer must demonstrate skill in overcoming the handicaps presentedby playing the piece on the kind of instrument sanctioned by the musicalcommunity (1998: 57).

36 julian dodd

Now, I see no need to disagree with any of this. Godlovitch, it seems tome, makes a convincing case for his sociological account of the values helddear by musical communities. But the crucial point is that such a descriptiveaccount of how music-making communities come to the opinions that theydo on matters of instrumentation can in no way support a philosophicalthesis concerning the individuation of works of music. That the practice ofmusicians embodies a commitment to instrumentalism does not entail thatthis doctrine is correct. To make good this latter thesis we need to comeup with sound philosophical arguments, not a sociology of belief. Merelyreporting the beliefs embedded in musical practice does not help us decidewhether these beliefs are true.

Having abandoned such a sociological approach, it might be temptingto try to buttress the argument from scores by arguing that the norm ofauthenticity demands that performers use the instruments specified by thecomposer. As Davies sees it, ‘[i]f the composer determinatively instructs thatcertain instruments should be used, they must be played if the performanceis to be fully authentic’ (2001: 69). And Davies thinks this because hebelieves that there exist conventions in music that necessarily connect thescored sounds with the specified instrumentation:

[I]mprovisation, the creation of works, the public specification of works, theperformance of works, the reception of works—all these assume, involve, and relyon practices and conventions, some of which are predominantly musical and otherswhich have more to do with social structures and purposes. As a result, connectionsthat otherwise might be merely contingent take on a different status because theybecome normative within the relevant practice ... . Within the relevant practices,playing the appropriate instruments is not merely a useful means to the productionof the desired result, which is supposedly the creation of an abstract sound structure.Instead, the use of those instruments is part of the end, which is to articulate thespecified sound structure on the required instruments. To do otherwise is to ‘cheat’and thereby to undermine the playing’s status as a performance.

(Davies 2001: 64–5)

I remain unconvinced, however. For one thing, and as I have explainedalready, the composer’s instruction to use certain instruments can be viewedas provisional. To be true to the piece, a performer need only producesounds with the same timbral features as sounds produced using the specifiedinstruments. Authenticity in performance is, indeed, a desideratum, but what


I am suggesting is that the features required for a performance to beauthentic cannot be simply read off from the score in the way in whichDavies suggests. The simple specification that a piano be used, for example,must be interpreted in the light of the nature of our musical experienceand the fact that sonicism is the face-value theory when it comes to thequestion of the relation between works of music and the instrumentationspecified in their scores.

Davies will remain unsatisfied by this, no doubt. Indeed, at this pointhe will surely seek to press his claim that the conventions and practicesunderlying the making and appreciation of music give composers, musiciansand listeners an essential interest in how sounds are produced, not just theirqualitative appearance. But the case that Davies presents as an example ofsuch a convention is nothing of the kind, as we shall now see. Daviesclaims that we regard a performer who uses alternative instrumentation tothat specified in the score as having cheated, and that such a decision onthe part of the performer undermines the playing’s status as a performance(2001: 65). This claim, though, needs careful handling. To begin with,the timbral sonicist, by contrast with his pure sibling, will agree that aperformance of a piece on instruments that produce sounds of a differenttone colour to those produced by the specified instruments is inauthenticand deficient. So let us focus on a case in which the instrumentalist’s andtimbral sonicist’s positions diverge: our imagined example in which theHammerklavier is performed on Perfect Timbral Synthesizer. Davies arguesthat such a performance would be a less adequate performance of the piecethan a sonically indistinguishable performance making use of a piano; andhe takes this to be demonstrated by the fact that we would regard theproducer of the synthesized sounds as having cheated in some way.

As a matter of fact, it seems to me that we can distinguish two kinds ofcase here, neither of which have the implications that Davies believes themto have. In the first kind of case, the synthesized performance is producedwith the aim of deceiving the audience into thinking that a piano has beenused. Clearly, in this case, if we discovered the deception we would,indeed, regard the performer as having cheated; but, equally, the source ofour resentment would not be that the performer had misrepresented thepiece by playing it on a synthesizer, but that the performer had, in effect,lied to us about what she was doing. In the second kind of case, by contrast,the audience is well aware that the performer is using a synthesizer, but

38 julian dodd

is also aware that such an instrument is easier to play than the piano, thusmaking the performer’s task less demanding than is usual. Here too wemight regard the performer as having cheated, but, once more, our sense ofhaving been short-changed does not have its origin in such a performance’sbeing inauthentic; this time, the audience would feel cheated because theperformer had not been tested to the degree that other performers of thepiece, using conventional instrumentation, have been. In this instance,the performer’s decision to use a synthesizer does, indeed, ‘underminethe playing’s status as a performance’ (Davies 2001: 65); but, once more,this is not because such a performance lacks authenticity, but because theperformer’s feat is less impressive than it would have been had traditionalinstrumentation been employed.

So my point is this: although we could feel cheated by synthesizedperformances, the fact that this is so is explicable in ways that do notentail the truth of instrumentalism. At this juncture, Davies may seek toreply by insisting that there is a possible case of our feeling cheated by asynthesized performance that I have not considered, and that does entailthe falsehood of timbral sonicism. He might, in other words, claim thatwe would still feel cheated by a synthesized sonic doppelganger even if wewere aware of the sounds’ causal origin and even if such an instrumentwere as difficult to play as a piano. But I simply deny that we would feelin any sense cheated by such a performance. The performance would be asound-sequence indistinguishable from a performance using a piano; and itwould make demands of the performer that are on a par with the demandsfaced by pianists. What would there be for an audience to feel resentfulabout? Short of simply begging the question, I cannot see a reply for Davieshere. There is no case in which our sense of having been cheated by asynthesized performance tells against timbral sonicism.

5

So much for variants of the argument from scores. Such arguments by nomeans force us to admit that performance means-properties are normativewithin works, and hence do nothing to undermine the timbral sonicist’sidentity criterion. The timbral sonicist is not yet home and dry, however.For as I noted in §3, an instrumentalist may also argue for her position by


claiming that a performance’s transmission of a work’s aesthetic contentand artistic import requires not only that it sound a certain way, but that thesounds be produced in a certain way. Arguing along these lines, Levinsonclaims that a work’s aesthetic content and artistic properties are ‘determinednot only by its sound structure ... but also in part by the actual means ofproduction chosen for making that structure audible’ (Levinson 1980: 76).According to Levinson, the kind of thinking about the determination ofaesthetic content that prompted my move from pure to timbral sonicismultimately commits us to instrumentalism; and so, if he is right, the linecannot be held where the timbral sonicist wishes to hold it. A considerationof what is required of a performance, if it is to transmit a work’s aestheticand artistic content, reveals that performance means-properties—and notjust acoustic properties—are normative within works. And this can onlymean that—contrary to the sonicist’s identity criterion—there can benumerically distinct works that are sonically indistinguishable.

Some examples should help Levinson’s style of argumentation to stick.First, consider Paganini’s Caprice Op. 1, No. 17. This piece, Levinsonclaims, has the artistic property of being virtuosic and, hence, a correctperformance of it should have a virtuosic quality too. But, argues Levinson,such a performance could only have this quality, if it made use of a violin:

if we did not conceive of the Caprice No. 17 as essentially for the violin, asinherently a violin piece (and not just a violin sounding piece) then it would not meritthat attribution. For, as executed by a computer or by some novel string instrumentusing nonviolinistic technique, its sound structure might not be particularly difficultto get through.

(Levinson 1980: 77)

For Levinson, then, the Caprice Op. 1, No. 17 is virtuosic, but a perfor-mance could only transmit this artistic quality, if the sounds were actuallyproduced by a violin rather than by some kind of violin-substitute.

As for artistic properties, so for aesthetic properties. For, according toLevinson:

[t]he aesthetic qualities of the Hammerklavier Sonata depend in part upon the strainthat its sound structure imposes on the sonic capabilities of the piano; if we are nothearing its sound structure as produced by a piano, then we are not sensing thisstrain, and thus our assessment of aesthetic content is altered.

(Levinson 1980: 76–7)

40 julian dodd

And the same goes, if Levinson is right, for works’ expressive content too.Mozart’s Serenade in E Flat, K. 375 begins with a passage which has an‘assertive, attention-getting quality’ (Levinson 1990b: 396); and the reasonwhy the piece’s opening has this expressive content is that the soundsproduced by the winds have a honking aspect: that is to say, air is forcedthrough narrow openings in tubes in the same way in which a goose—aliteral honker—forces air through its windpipe (Levinson 1990b: 396).Now, according to Levinson, in order to appreciate the passage’s assertivecontent, the audience must hear the sounds as honkings; but they can onlydo the latter, if they hear these sounds as produced in the way in whichhonkings are produced; which is to admit that the passage’s expressivecontent is dependent upon its being played by wind instruments and not,for example, by Perfect Timbral Synthesizers:

Would those sounds— i.e. those produced on the Perfect Timbral Synthesizer—bea honking in the quasi-literal sense in which the accented outputs of oboes andclarinets are such? I suggest not. If it is aesthetically appropriate to take sounds inperformance for what they are, and not for what they aren’t, and if the opening’sparticular assertiveness is partly a matter of its properly coming across as a honking,in the sense described, then part of the aesthetic character of that opening isdistorted or undercut if the performing means that Mozart directly specifies arebypassed and only the resulting sound (or timbral complex) that he indirectlyspecifies is adhered to.

(Levinson 1990b: 396–7)

So, if Levinson is to be believed, it turns out that being performed usingwind instruments (and not simply being performed by means that exactly mimicthe sounds made by wind instruments) is normative within the piece becausea performance on Perfect Timbral Synthesizers would fail to producegenuine honking sounds at the piece’s beginning, and hence would fail todeliver the assertive content Mozart intended.

Needless to say, further examples of this kind are easily constructed.The appreciation of a piece’s expressive content, so it is claimed, requiresus to hear passages as making musical gestures akin to human expressionsof emotion: we must hear phrases as sighing, crashing, booming, or suchlike (Levinson 1990b: 399–401). But in Levinson’s view, in order tohear gestures in music, one must hear the sounds produced as madewith the relevant gestures; and this, in turn, requires that the sounds


really are produced by means of an instrument with which such gesturescan be made.¹² Hence, if a work is to be performed properly, sucha performance must not merely perfectly ape the sounds made by thespecified instrumentation; it must use those very instruments to producethose sounds.

So how should a timbral sonicist reply? Two points need to be made atthis stage. First, and contrary to the spin that Levinson himself occasionallyplaces on these examples, they cannot of themselves demonstrate a work’sinstrumentation to be essential to it. Even if it is shown, for example, thata performance of Paganini’s Caprice No. 17 can only transmit the work’saesthetic content, if its constituent sounds are produced on a violin, itwould only follow that the work essentially involves the violin, if thework’s aesthetic content were essential to it, and this is a thesis thatLevinson does not defend (1990a: 221–2). At best, then, the examples canonly demonstrate that performance means-properties are normative withinworks, not that they are so normative in each possible world in which theworks in question exist.

Having made this point, such examples are still a threat to the sonicist’saccount of the identity conditions of musical works, of course. Thesonicist’s distinctive claim is that acoustic properties alone are normativewithin works, so she repudiates the suggestion that a properly formedperformance of the Caprice, for example, must make use of a violin.However, and this is my second point, Levinson’s examples do not evenshow this much, as we shall now see.

To begin with, it would be as well to focus on Levinson’s argumentativestrategy. Levinson takes his discussion of his various examples to demon-strate the truth of the following claim: that it is a necessary conditionof a work W ’s having certain aesthetic, expressive or artistic propertiesthat certain performance means-properties are normative within W . Inother words, he points to certain features that works are taken to haveindisputably, and then claims that works can only have these features, ifinstrumental properties are normative within such works. However, inwhat follows I argue that Levinson’s examples are insufficient to justify

¹² One example of this putative phenomenon should suffice. Drums can contribute a pounding orbattering content to passages of music; and one might be tempted to follow Levinson (1990b: 400) insupposing that this expressive content can only be fully transmitted if the sounds are actually made witha pounding or battering gesture.

42 julian dodd

this claim. To this end, my strategy will be disjunctive. First of all, I shallargue that the kinds of artistic properties to which Levinson appeals are notgenuinely possessed by works at all. When it comes to artistic properties,I thus deny the thesis whose presumed truth supposedly requires the truthof instrumentalism. Aesthetic and expressive properties, on the other hand,certainly are genuinely possessed by works of music; but in this case, mycomplaint against Levinson is that his examples do not demonstrate that awork’s having such a property is genuinely contingent upon instrumental(as opposed to merely timbral) properties being normative within it.

Let us start by considering artistic properties and, in particular, byreturning to Levinson’s own example. Paganini’s Caprice Op. 1, No. 17,he says, deserves the attribution ‘virtuosic’, and, hence, a performance—ifit is do justice to the piece’s virtuosic quality—must make use of the violinrather than, say, an acoustically indistinguishable instrument that, perhaps,demanded less of a performer. But at this point, I think that the startingpoint of Levinson’s discussion of this example—namely, his presumptionthat the work literally possesses the property being virtuosic—will simply bedenied by a sonicist. True enough, a performance may be virtuosic (if theperformer plays a violin rather than a space-age sound-alike that is easier toplay); but a sonicist will not accept that being virtuosic is normative within thework: i.e. that any well formed performance must be virtuosic. For, since thesonicist takes the Caprice to be a pure type of sound-event—a type whosenormative properties are all acoustic in nature—she precisely repudiates thesuggestion that a property concerning how sounds are produced, as opposedto their qualitative nature, can be normative within it. Of course, if itwere true that the piece itself were virtuosic, it would follow that it couldonly be played on an instrument that required technical excellence of theperformer. But it is the antecedent of this conditional that no self-respectingsonicist will accept. It is performances of works that can be virtuosic, not theworks themselves.

A similar response can be made to other putative examples of howa work’s artistic properties determine performance means-properties asnormative. Levinson asks us to imagine a piece written for violin to beplayed in such a way as to make certain passages sound as if they wereplayed by a flute. Would not this piece, by contrast with a sonicallyindistinguishable piece written for violin and flute, be original and unusual?And would not this demonstrate that a correct performance of the former


piece—one that preserved its artistic import—would have to be playedon the violin alone (Levinson 1980: 77–8)? But to this we can makewhat is essentially the same reply as before, albeit with a slightly differenttwist. Once again, Levinson’s mistake is to suppose that a sonicist willagree that works themselves possess artistic properties such as unusualness andoriginality. Since the sonicist takes works of music to be pure types ofsound-event, she denies that they can be analytically tied to instruments,and hence denies that such works can themselves be unusual by virtueof involving surprising instrumentation. By contrast, a sonicist will insistthat it is composers and their compositional acts that, strictly speaking, have theartistic properties concerned. What is unusual and original in Levinson’sexample, is not the work itself, but a particular composer’s scoring of sucha work for solo violin rather than violin and flute. The work itself is a puresound event-type, and hence cannot have properties such as these.

One thing is clear, however: the same sonicist strategy will not workfor examples taken to illustrate the fact that performance means-propertiesare normative within works by virtue of determining such works’ aestheticor expressive qualities. Works certainly can be sublime, craggy and assertive,so a timbral sonicist cannot reply by denying this. Nevertheless, replies areavailable, and we shall now see that Levinson is mistaken in thinking thata work’s having such an aesthetic or expressive quality may be dependentupon performance means-properties being normative within it.

Consider, once more, the Hammerklavier’s sublimity and cragginess. Itis Levinson’s contention that unless we hear the piece as played on thepiano, we will have no sense of the strain that the piece places on theinstrument, and hence will fail to pick up on its true awesomeness. In thissense, according to Levinson, ‘[t]he aesthetic qualities of the Hammerklaviersonata depend in part on the strain that its structure imposes on the soniccapabilities of the piano’ (1980: 76–7). But this conclusion is resistible.Whilst it is true that we must hear a performance of the piece as playedon the piano, if we are to grasp the full gamut of its aesthetic qualities,it does not follow that a properly formed performance must actually makeuse of the piano. A truth about aspect perception is surely this: to see orhear o as (an) F does not entail that o really is (an) F. I can see a smile asfriendly, and yet the smile be contemptuous; I can hear what is really a cryof surprise as a cry of pain. So, taking her cue from facts such as these, atimbral sonicist will insist that all that is required for a performance of the

44 julian dodd

Hammerklavier to transmit the piece’s cragginess and awesomeness is that itsconstituent sounds sound as if they originate from a piano. As long as thesounds produced sound just like those that a piano would produce, it is ofno consequence whether they are actually produced by a piano or not. Allthat is required is that sounds seem to have, rather than actually have, thatorigin.

Exactly the same response is available in the case of works’ expressivequalities. Let us return to the assertive character of the opening of Mozart’sSerenade in E Flat, K. 375. Levinson’s claim, remember, is that this passagewould not have this content, were it not for the fact that it has a honkingquality; and that the sounds would not have this honking quality— i.e.would not be a honking in the required ‘quasi-literal sense’—unless theywere actually made by oboes and clarinets (Levinson 1990b: 396–7).However, this second claim is false since a sound can have a honking qualityand yet not be a literal—or even quasi-literal—honking. Just as long asa sound seems for all the world to be a genuine honking sound, it has ahonking quality; it is not necessary for the sound to be the result of airhaving been forced through tubes. And such a reply to this case yields ageneralizable moral. We may grant Levinson the fact that understanding apiece’s expressive content requires us to hear gesture in music (1990b: 398);but we should deny that ‘expressive content in music is not detachablefrom the means of performance that are written into musical compositions’(1990b: 399). As long as we hear the music as embodying a certain gesture,it does not matter whether the sounds have actually been produced withthat gesture. So, for example, if we hear a run of tones as a genuine glissandoon the piano, it will succeed in conveying the insouciance associated withsuch a flicking gesture, even if the sounds are produced by pressing buttonson a Perfect Timbral Synthesizer. The music’s insouciant quality requiresonly that we imaginatively construe the sounds to have been produced bya flicking or sweeping gesture; these sounds need not have been genuinelyproduced by a glissando-proper.

Levinson’s reply to such a sonicist move ultimately cuts no ice. Envisaginga case in which a sound sequence exactly duplicates a glissando but isproduced by other means of performance, he says this:

If this imaginative construal [of the sounds as having been produced by a glissando]is in fact unsupported by performance in the appropriate manner, that is, on a


keyboard, and we are aware of this, the resultant effect is not that proper to themusic, but rather some degree of cognitive dissonance.

(Levinson 1990b: 399)

But the crucial clause here is ‘and we are aware of this’. Once this clause isinserted, the sonicist can acknowledge that such cognitive dissonance mayoccur. Granted, if the sounds are glissando-like, and yet we know them notto have been produced by such a gesture, the music might—though neednot¹³—fail to have the abandon for us that it would have had otherwise. Butsuch a difference in perceived expressiveness admits of an explanation thatavoids any commitment to instrumentalism. Rather than being explainedby the fact that the sounds are not produced by a glissando, an audience’sfailure to hear the passage as expressive of abandonment is caused by afailure of imagination on their part: a failure to hear the sounds as if madeby a glissando. Putting things this way, sonicism is unscathed. For the factremains that a sound sequence produced on a Perfect Timbral Synthesizercould be heard as a glissando, and hence could have the same expressivecontent as a sonically identical sequence produced in the characteristic way.All Levinson has shown is that we might fail to hear a sound sequence asmaking a certain gesture, if we know it not to have been produced bymeans of such a gesture; and this need not be denied by a sonicist.

Ultimately, then, Levinson faces a dilemma. If, in the glissando exampleand others like it, he argues that the music would fail to have theappropriate expressive content, if we knew it to be produced by non-standard performance means, the sonicist can grant such a possibilitywithout damaging her position. If, on the other hand, Levinson claims thatthe music would fail to have the appropriate content even if we took thesounds to have been produced by a genuine glissando, this claim wouldseem to be false. For as we have seen already, sounds can be heard asembodying a certain gesture, and hence have a certain expressive content,even if they are not made with that gesture. All that is required is thatthe sounds are qualitatively indistinguishable from sounds produced with thatgesture. Levinson has failed to make a case for the claim that the performers

¹³ This qualification is crucial. Levinson seems to regard it as being extremely difficult for an audienceto imagine sounds to have a certain source whilst knowing these sounds to have another origin (1990b:403). As we shall see in §6, Levinson overestimates the difficulty in what is an everyday phenomenon.

46 julian dodd

must actually be doing what we imagine them to be doing, if we are tohear such gestures in the sounds made.

6

As we have just seen, and as Levinson himself recognizes (1990b: 402–3), thesonicist responds to his argument by claiming that a performance of a piececan transmit its aesthetic and expressive content, just as long as the soundsproduced are imagined to be produced in certain specific ways. To insist thatthe sounds should be so produced is to insist on more than suffices. Levinson,though, is dissatisfied with this reply for three reasons. First, he claims thatits cogency depends upon the audience’s not learning the real origin of thesounds they are hearing, or else being willing to be fooled concerning theirsource, ‘neither of which is a viable stance vis-a-vis a performance’ (1990b:403). Second, he states that a sonic doppelganger of, say, a glissando willpresent merely a simulacrum of the expressive gestures that belong to the pas-sage, rather than the gestures that really belong to it (1990b: 404). Finally, hecontends that such a sonic doppelganger cannot count as authentic becauseit would effectively stymie our ability to compare and evaluate performances(1990b: 404–5). I shall consider these objections in reverse order.

Let us, then, start by examining Levinson’s third objection. A perfor-mance of the Hammerklavier on a Perfect Timbral Synthesizer, even ifindistinguishable from a properly formed performance on a piano, wouldbe inauthentic, Levinson claims, because it would present us with a severecritical quandary. In short, we would not know how to evaluate it. AsLevinson puts it, if a piano were not involved:

we would be pretty much completely at sea in regard to assessing the particularexpressiveness of the performance ... , its particular manner of bodying forth [thework’s] inherent expression. For an important dimension of assessment wouldhave been removed: how have the instrumentalists, given their control over andway of internalising the gestural capacities of their instruments, related themselves,at each turn, to the demands of this music, which is conceived for and referred tothose capacities?

(Levinson 1990b: 405)

But inasmuch as there is a genuine worry here, it would seem to beirrelevant to the point at issue, which is, remember, whether a performance


of the work on a Perfect Timbral Synthesizer could transmit the full rangeof the piece’s aesthetic content. It may well be difficult to evaluate such aperformer’s achievement in transmitting such content, given that we tend tomake such evaluations according to how well the performer has overcomethe limitations of the instruments for which the work was written. Itwould, indeed, be easier said than done to compare the achievementsof an Ashkenazy with those of someone who produced a qualitativelyindistinguishable sound sequence on a synthesizer: we are accustomed tothe problems posed by playing the piece on a piano—our familiarity withboth the instrument itself and the attempts made by pianists down theyears give us a reasonably clear sense of this—but until we had a clearidea of the problems and constraints faced by an operator of a PerfectTimbral Synthesizer, any such comparison with performances by pianistswould be obstructed. Having said this, it is far from clear why the existenceof such a puzzle would demonstrate the synthesized performance to beinauthentic, i.e. to fail to present the full extent of the performed work’saesthetic content. For the puzzle to which Levinson draws our attentionin fact concerns how we would evaluate a performer’s achievement intransmitting a work’s aesthetic content; it is irrelevant to the questionof whether a synthesized performance could deliver this content. And, ofcourse, to this latter question, the timbral sonicist has a clear answer: suchcontent is determined wholly by how a performance sounds, and to noextent by how these sounds are produced.

What of Levinson’s second objection to the sonicist’s reply to hisargument? Must a synthesized performance inevitably fail to convey a work’saesthetic substance? I don’t think so. Naturally, a timbral sonicist agrees withLevinson that ‘an authentic performance should seek to present a piece’saesthetic substance as it is’ (1990b: 404). But her claim is precisely that thisaesthetic substance is not dependent upon specific means of performancein the way that Levinson supposes: as long as a sound sequence soundsexactly like a glissando on a piano, so that the audience imagines the soundsto have been produced with a sweeping gesture along a piano keyboard,the resultant sounds will have exactly the same aesthetic substance as agenuine glissando. The glissando-like sounds express the sweeping gesture,and hence give the music an insouciant quality, even though they were notproduced with such a gesture. Aesthetic substance is determined by how themusic sounds, not how it is produced. To put it another way, the vehicle

48 julian dodd

of a work’s musical meaning is not, as Levinson assumes (1990b: 404),sounds-as-produced-by-specific-instruments, but those sounds themselves,whatever their origin. His claim that a sonic doppelganger of a glissando‘gives the appearance, but not the reality of the complex of soundings thatare the true vehicle of a piece’s musical meaning’ (1990b: 404) sees him,once more, beg the question against the timbral sonicist.

Nonetheless, Levinson will insist that a synthesized sonic doppel-ganger—even if it manages to convey the right expressivity to an audi-ence—inevitably ‘falsifies or disguises the basis for that expressivity’ (1990b:404), and, as a consequence, cannot be authentic. And this, in essence,is his one remaining objection to the kind of sonicist response to hisoriginal argument that I outlined at the end of §5. This objection emergesonce the following kind of case is considered. Imagine that an audienceattends a performance of Mozart’s Serenade in E Flat, K. 375, well awarethat it is performed using a Perfect Timbral Synthesizer rather than windinstruments. As a result, in order to hear the assertive quality of the piece’sopening passage, an audience must hear the sounds produced as a sequenceof honkings whilst at the same time knowing the sounds not to be honk-ings, i.e. not to have been produced by forcing air through narrow holes intubes. But this, Levinson suggests, is none too easy a thing to do: we mustimagine the sounds to be honkings while knowing them not to be (1990b:403); that is, we must allow ourselves to be fooled concerning the sounds’origin. And, in any case, even if an audience could succeed in doing this, aperformance that required such a thing of its auditors could not be countedas authentic, and hence could not be properly formed:

A performance that enforces mental acrobatics on listeners in order that an intendedexpressiveness should emerge, which expressiveness should emerge effortlessly andunconsciously, can hardly be thought to further authenticity. On the contrary, toensure that a listener’s experience of a performance be informed by the thought ofcertain performing actions in the right way, the performance should actually involvethose very actions and the very instruments that make them possible.

(Levinson 1990b: 403–4)

But I have two replies to this reasoning. First of all, an audience’s hearingsounds it knows to be synthesized sounds as honkings is not thereby allowingitself to be fooled as to the sounds’ origin. On the contrary, the audience isimagining these sounds to have an alternative source to that which it knows


them to have actually. The audience knows the sounds to be synthesized butimagines them to be produced by clarinets and oboes. And before we aretempted to look askance at such a phenomenon, we should remember thatimaginative engagement of this kind is not unfamiliar to audiences of theperforming arts: the way in which an audience imaginatively engages with aplay—imagining that the people on stage are in real situations, but knowingthe participants to be actors—is another example of the same genus.

Second, such a reflective exercise of the imagination is not an instance of‘mental acrobatics’. As I have just explained, knowing that p but imaginingthat not-p is an everyday phenomenon par excellence. True enough, for theexpressiveness of the music to emerge in a synthesized performance, the audi-ence must employ greater imagination than if the sounds are produced by thewind instruments that the composer had in mind. But the exercise of such acapacity is both effortless and perhaps even unconscious: one just sits back inone’s seat, shuts one’s eyes, and lets the sounds, what one knows about suchsounds’ customary origin, and one’s imagination do the rest. I fail to see whya synthesized performance of a piece that required this of its audience wouldnot be authentic. After all, this requirement, so easy to meet, ensures that sucha performance would succeed in transmitting the piece’s aesthetic content.

7

Given timbral sonicism’s status as the face-value theory of how musicalworks are individuated, we should only give it up once it has been shownto be untenable. But there is nothing in the instrumentalist’s arsenal ofarguments that forces us to take such drastic action. The fact that a work’sscore instructs performers to use certain instruments does not entail thatthere could not be a properly formed performance of the piece producedby other means. As long as these other means of sound production wereused to produce sounds aurally indistinguishable from sounds produced bythe instruments specified in the score, the demands of authenticity wouldbe satisfactorily met. Equally, the attempt to argue that a work’s possessionof certain aesthetic and artistic properties is dependent upon its havingperformance-means properties normative within it falls short of its target.All that is required for works to have the content they have is that theysound as they do. How these sounds are produced is incidental.

50 julian dodd

So can we now embrace timbral sonicism—and, with it, the simpleview—with a clear conscience? Not just yet. Some philosophers take thesonicist version of the type/token theory to have Platonic consequencesthat require us to identify musical works with types of a more complexkind.¹⁴ Others, self-styled contextualists, argue that sonicism should berejected because composers occupying differing musico-historical contextsinevitably compose distinct works, even if the said works are acousticallyindistinguishable.¹⁵ Finally, the other component of the simple view—thetype/token theory—by no means commands universal acceptance. Worksof music have been claimed to be continuants (Rohrbaugh 2003), action-types (Currie 1988), and even action-tokens (Davies 2004). Ultimately,I doubt whether the arguments for these various positions add up, butmaking the case for this claim is beyond the scope of this paper.¹⁶

References

Currie, G. (1989) An Ontology of Art, Basingstoke: Macmillan.Davies, D. (2004) Art as Performance, Oxford: Blackwell.Davies, S. (2001) Musical Works and Performances: A Philosophical Exploration,

Oxford: Oxford University Press.Dodd, J. (2000) ‘Musical Works as Eternal Types’, British Journal of Aesthetics

40: 424–40.Dodd. J. (2002) ‘Defending Musical Platonism’, British Journal of Aesthetics

42: 380–402.Dodd. J (2004) ‘Types, Continuants and the Ontology of Music’, British

Journal of Aesthetics 44: 342–60.Dodd, J. (2007) Works of Music: An Essay in Ontology, Oxford: Oxford

University Press.Godlovitch, S. (1998) Musical Performance: A Philosophical Study, Lon-

don: Routledge.

¹⁴ See, for example, Levinson’s justly celebrated construal of musical works as indicated types(1980, 1990a).

¹⁵ Levinson, once more, is in the vanguard of this camp (1980, 1990a), though versions of this viewhave become more widely accepted than his own positive ontological proposal. See, for example,Currie (1989) and Davies (2001).

¹⁶ An early version of this paper was given at the British Society of Aesthetics conference in Oxfordin September, 2004. Many thanks to all those who attended. Special thanks are due to David Daviesand Kathleen Stock, who gave me extremely helpful written comments.


Kivy, P. (1983) ‘Platonism in Music: A Kind of Defence’, Grazer Philosophis-che Studien 19: 109–29.

Kivy, P. (1987) ‘Platonism in Music: Another Kind of Defence’, AmericanPhilosophical Quarterly 24: 245–52.

Kivy, P. (1988) ‘Orchestrating Platonism’, reprinted in his 1993, 75–94.Kivy, P. (1993) The Fine Art of Repetition, Cambridge: Cambridge University

Press.Levinson, J. (1980), ‘What a Musical Work Is’, reprinted in his 1990c,

63–88.Levinson, J (1990a) ‘What a Musical Work Is, Again’, reprinted in his

1990c, 215–66.Levinson, J (1990b) ‘Authentic Performance and Performance Means’,

reprinted in his 1990c, 393–408.Levinson, J (1990c) Music, Art and Metaphysics: Essays in Philosophical

Aesthetics, Ithaca, NY: Cornell University Press.Rohrbaugh, G. (2003) ‘Artworks as Historical Individuals’, European Journal

of Philosophy 11: 177–205.Scruton, R. (1997) The Aesthetics of Music, Oxford: Oxford University Press.Walton, K. (1988) ‘The Presentation and Portrayal of Sound Patterns’, in

Jonathan Dancy, J. M. E. Moravcsik and C. C. W. Taylor (eds), HumanAgency: Language, Duty and Value, Stanford: Stanford University Press,237–57.

Wolterstorff, N. (1980) Works and Worlds of Art, Oxford: ClarendonPress.

Further Reading

When it comes to the ontological nature of works of music, competingversions of the type/token theory are propounded by Levinson (1980,1990b) and Kivy (1983, 1987) respectively. I take Kivy’s Platonist sidein my 2000, 2002 and 2007. Alternatives to the type/token theory aresuggested by Currie (1989), David Davies (2004) and Rohrbaugh (2003).

The second aspect of the simple view—sonicism—is criticized byLevinson in his 1990b, and by Stephen Davies in his 2001: 60–71. Kivyresponds to such arguments in his 1988.

2Doing Justice to Musical WorksMICHA E L MORRIS

1. Introduction

There is something puzzling about musical works. The puzzlement isontological in a broad sense: we find it difficult to understand what kind ofthing musical works are. We note, for example, that a musical work cannotbe stored in a cellar,¹ though copies of a score can be. Musical works can beperformed on several occasions, though they can exist unperformed. Somemusical works are written, but not all are.

The leading response to this ontological puzzlement is to attemptto remove it by assimilating musical works to other, supposedly morefamiliar, kinds of thing. I shall call this the assimilating response, andthose who adopt it assimilators. The dominant version of the assimilatingresponse aims to remove our puzzlement by denying that musical works areparticulars. Instead, they are types (Wollheim 1980, §35; Dodd 2000) or kinds(Wolterstorff 1975) or universals.² If they are types or kinds, what are theytypes or kinds of? If they are universals, what are their instances?³ Thereare more and less reductive answers to these questions within the dominanttradition. According to the more reductive answers, musical works aretypes or kinds of sound sequence; or else they have particular sequences ofsounds as their instances. According to the less reductive, musical works

¹ Compare Heidegger: ‘Beethoven’s quartets lie in the storerooms of the publishing house likepotatoes in cellar’ (Heidegger 1956: 145). But this looks like a rare case where Heidegger seems toaccept uncritically the kind of ontology which elsewhere (including elsewhere in this work) he isconcerned to make us aware of.

² Levinson (1980 and 1990b) is standardly cited as someone who holds that musical works are types,but there are features of his view which make that classification a little complicated.

³ Although it is often uncritically said that musical works have instances, we should recall that theword ‘instance’ is correlative with ‘universal’ or ‘property’, just as ‘token’ is correlative with ‘type’.

doing justice to musical works 53

are types or kinds of performance; or else they have particular performancesas their instances.

The assimilating response is motivated by a certain ontological con-servatism. It takes certain kinds of things—generally things with clearspatio-temporal boundaries—to be familiar and unpuzzling; and it aims toacknowledge the existence of as few things other than these supposedlyfamiliar items as possible. This general ontological conservatism is appliedat two points to produce the view that musical works are universals, types,or kinds. In the first place, it allows us to acknowledge only those partic-ulars which have relatively clear spatio-temporal boundaries: the relevantparticulars here are therefore sound sequences or performances. And thenit lets us accept just those non-particular things which we have generaltheoretical reasons for acknowledging. At this point, assimilators see them-selves as having to argue for the acknowledgement of universals, types orkinds, against even more parsimonious views, which only accept classes inaddition to spatio-temporally well-defined particulars (see Goodman 1981).

I shall argue that this approach is altogether wrong. Musical works arenot types or kinds or universals, and the assimilating response in general isthe wrong response to our puzzlement about the nature of musical works.I will take as my principal target the specific version of the common viewwhich holds that they are types of performance. I choose the idea that theyare types (rather than kinds or universals) because that seems to be thebest worked out. And I choose the idea that they are types of performance(rather than sound sequence) because that is less reductionist, and the basicobjection I will make to the general view is, in effect, that it is impossiblyreductionist.

This is how I’ll proceed. First, I’ll lay out what seems to me to be thebasic argument for the common view. My objection to this view dependson claiming that it cannot accommodate the fact that musical works, likeall works of art, are essentially meaningful, so next I’ll explain that. I willthen argue that there are two ways in which the common view cannotaccommodate the meaningfulness of works of art. Finally, I will look againat the basic argument for the common view, in order to see where it goeswrong, and I will say something about how we should respond to ourpuzzlement about the nature of musical works.

54 michael morris

2. The Basic Argument for the Type-Token View

I will call the view that musical works are types, of which their performancesare tokens, the type-token view. The basic argument for it is that it providesthe best explanation of the following facts:

(i) musical works can be heard;(ii) the existence of a musical work does not depend on any particular

performance;(iii) the same musical work can be performed many times.

Fact (i) seems undeniable. We might doubt the truth of (ii), if there areany performances which can be regarded as musical works in their ownright, or of (ii) and (iii), if there are any musical works which are essentiallyimprovisations, but (ii) and (iii) remain plausible for the large bulk of whatwe ordinarily regard as musical works, and might be thought to be essentialto anything which can properly count as being both musical and a work.⁴

Fact (i) requires an intimate relation between musical works and perfor-mance, because it is only in being performed that a musical work can beheard. But that relation cannot be identity with any particular performance,because of facts (ii) and (iii). Facts (ii) and (iii) make it difficult to regardmusical works as spatio-temporally located objects of any familiar kind, inwhich case it is natural to describe them as ‘abstract’.⁵ There is a naturalaversion (among many philosophers, including assimilators) to admittingthe existence of abstract particulars, so facts (ii) and (iii) encourage us to denythat musical works are particulars. Once we deny that musical works areparticulars, it is natural to suggest that they are types. And when we suggestthat they are types of performance, we seem to be able to explain howthey can be heard—by being tokened in a particular performance; howtheir existence is independent of any particular performance—because theexistence of a type is independent of the existence of any particular oneof its tokens; and how the same musical work can be performed manytimes—because it is of the nature of types to be capable of having manytokens. And the explanation is ontologically conservative, which is always

⁴ In fact, I think this thought is wrong, since I think that performances of musical works are themselvesboth musical (with luck) and works; but there is no inconsistency in making the performable work,and not the performances of it, the centre of our interest here.

⁵ Note that ‘abstract’ is not an innocent word: abstraction is most famous as a theory of generality.


a virtue in the eyes of an assimilator: we do not need to introduce anyradically new categories specially for the case of musical works, and we donot need to admit any surprising combinations (such as abstract particulars).

Types are individuated by the conditions which things have to meet tocount as tokens of them. The claim that musical works are types wouldbe empty (at best), if no such condition could be found to unify thethings which count as performances of a given work. What might such acondition be? There is just one natural candidate. In order for a musicalwork to be capable of being performed repeatedly, it must in some sensebring with it a prescription which indicates how it is to be performed.The prescription can be more or less precise in various ways: the notesand their values can be prescribed more or less precisely, as can tempi andinstrumentation—and so on. Let me gather together all the various respectsin which ways of performing a work might be prescribed under the crudephrase the notes. Then we can say that a performance is note-perfect if it meetsall the conditions (however precise) which are prescribed. It seems that thevery fact that a work can be performed repeatedly requires there to be sucha thing (as it were) as a note-perfect performance.

The concept of a note-perfect performance can be used in one of twoways to define the condition which performances must meet in order tocount as performances of the same work. First, the type might be definednon-normatively, simply as the type of note-perfect performances: this is anon-extensionalist version of Goodman’s famous proposal (Goodman 1981:186). If we follow this line, it will turn out that only those performanceswhich are note-perfect are really performances of the relevant work. Thatwill have the surprising consequence that some familiar, but technicallydifficult works may never actually have been performed. We could justbrazen this out—claiming, perhaps (with an eye to fact (i)), that suchworks have never really been heard—but we are unlikely to win manypeople over. The other alternative is to define the type normatively, as thetype of performances for which being note-perfect is a requirement oran ideal. (This is a variant of the view proposed in Wolterstorff (1975and 1980).) On this view, to be a performance of a given work is to bea performance of which it can strictly be said either that it ought to benote-perfect or else that if it is not note-perfect that is some kind of defectin it. This is surely the natural way for someone to flesh out the type-tokenview.

56 michael morris

We seem in this way to be able to substantiate the claim that musicalworks are types of performance. We have also—apparently incidentally,though we may claim it as a virtue of the view—found ourselves endorsinga particular approach to the criticism of performances. We will be temptedto think that good performances will be at least close to note-perfect;if we are inclined to include instrumentation, tempi and pitch amongthe things which are prescribed of a note-perfect performance—whichcertainly seems natural in the case of many musical works—we will findourselves giving high value to authenticity in performance. (This line ismost explicitly adopted by Davies (2001: Ch. 5); Levinson (1990a) clearlyvalues authenticity, though he does allow that a good performance maynot be authentic in Levinson (1987).)

I think this is the principal argument for the type-token view (see Dodd2004). It can be adapted, mutatis mutandis, to suit any version of the viewthat musical works are not particulars but in some way general. All versionsof the view, however, are subject to the same basic problem: they cannotproperly accommodate the fact that musical works, like all works of art,are essentially meaningful.

3. Musical Works are Essentially Meaningful

Some things are true of musical works just in virtue of their being worksof art. Among these is the claim that musical works are meaningful: thisought to be a truism, since it seems bound to be a truism that works ofart are meaningful. I will offer a brief consideration in support of the lattertruism, before dealing briefly with a simple objection.

First of all, it is clear that works of art are meaningful because it is clearthat one cannot properly consider a work of art as a work of art withoutconsidering it as meaningful. To treat a work of art as not being meaningfulwould be to treat it as some kind of found object, a curious feature of thenatural world, and this is precisely not to treat it as a work of art. When weconsider works of art as works of art, we regard them as things which callfor interpretation—or at the very least, as things of which interpretation islegitimate—and an interpretation is a presentation of meaning. When wereflect on this, we can see that no objection to the claim that works of artare meaningful can be provided by cases where we are inclined to say that


what is presented as a work of art is meaningless. For example, someonemight declare, in a reactionary spirit, that a painting was a ‘meaninglesssmear’, or a piece of music was nothing but ‘meaningless noise’. But, first,the very fact that what we take to be a meaningless smear is exhibited asa painting, or that what we take to be meaningless noise is presented as apiece of music, seems to give meaning to what might antecedently havebeen meaningless. And, second, even if we finally conclude that what wehave before us really is nothing but a meaningless smear or meaninglessnoise, all this will mean is that what we have before us is not really a workof art. So the examples cannot threaten the claim that works of art must bemeaningful.

We might feel some unease, however, because of a certain kind ofpreconception about the notion of meaning. The problem can be expressedquite roughly and intuitively as follows: nothing could be meaningfulwithout having a meaning, but any attempt to say what a work of artmeans seems inevitably to trivialize it; art always seems to transcend anyinterpretation.

There is indeed a strong temptation to associate the notion of meaningwith the possibility of a complete, explicit, and decisive interpretation, butwe need to be careful about the argument involved here. It has two crucialpremises:

(1) If something is meaningful, it is possible to say what it means.(2) To say what a work of art means is to trivialize it.

Neither of these premises is obviously true; (1) seems to depend on asignificant claim about language, which we might express like this:

(1a) Everything is describable.

(After all, if the strong general claim of (1a) were not true, why should onethink that meaning in particular can always be specified?) And (2) dependsupon a claim about works of art, which we might express like this:

(2a) Works of art transcend all interpretations.

In fact, I think both (1a) and (2a) are true. The real problem lies withsomething else which (2) depends on:

(2b) To say what a work of art means is to capture its meaning (finallyand completely).

58 michael morris

We can understand the link with (2a) if we paraphrase (2a) as saying justthis:

(2a∗) No interpretation (no paraphrased saying, for example) cancapture the meaning of a work of art (finally and completely).

If this is how the argument works, there are two problems with it. First,it depends on certain assumptions ((1) and (2b)) which are fundamentallyassumptions about language, rather than about either art or the generalconcept of meaning. And, secondly, it seems impossible to formulate theargument without presupposing that works of art are meaningful: (2a) hasto be interpreted in the manner of (2a∗), for example.

Works of art are not only meaningful, but essentially meaningful: that isto say, those things which are in fact works of art could not exist withoutbeing meaningful (or, indeed, having the meaning they have). This canbe seen by considering what we naturally take to be essential to a workof art. In any work of art (including works of music), we are able todistinguish between essential and non-essential parts and features. In thecase of a painting, for example, it might be essential that a brush was usedin a particular way on a certain part of the canvas, but inessential that thebrush in question was made by a particular person in his workshop oneMonday morning. In the case of a poem, it will be essential that thosevery words are used, but it will probably be inessential that the poem wasfirst written with a fountain pen. What this suggests is that the essentialparts and features of a work of art include those which are relevant to itsmeaning, or at least central to its meaning. We will happily tolerate quiteextensive variations while still regarding ourselves as having the same workof art, provided that the variations do not affect the meaning of the work,or don’t affect it centrally. (There are tricky questions about when we areto count revised editions of a musical or literary work as being differentversions of the same work, and when we should think of them as differentworks.)

Incidentally, this fits nicely with an explanation of why the meaningof a work of art cannot be captured in paraphrase. A paraphrase willalways differ from the original work in a significant number of its essentialfeatures: otherwise it wouldn’t be a paraphrase. If those features are relevantto meaning, we can understand why the paraphrase cannot capture themeaning.


If it is true that the essential features of a work of art include in particularthose which are relevant, or central, to its meaning, it would be miraculousif works of art were not essentially meaningful. For it seems that for worksof art not to be essentially meaningful, it would have to be the case that ascheme of individuation which made no reference to artistic meaning couldnevertheless, quite coincidentally, select as essential parts and features of anobject all those parts and features which are counted as essential when weare concerned with meaning. But the idea of such a miraculous coincidenceis itself incoherent, on most conceptions of possibility and necessity, whichallow that what is possible or necessary is necessarily possible or necessary.Within such a conception of possibility and necessity, the idea of this kind ofmiraculous coincidence between schemes of individuation becomes the ideaof the mere coincidence of the essential properties identified by the differentschemes across all possible worlds; and nothing could merely coincide acrossall possible worlds. So a natural conception of what is essential to works ofart forces us to accept that works of art are essentially meaningful. And thisapplies to musical works just as much as to works of art of other kinds.

4. To be Meaningful is to be there to be Understood

There is a familiar slogan in the philosophy of language, that a theory ofmeaning is a theory of understanding (see, e.g., Dummett 1975: 99). Thetruth at the heart of this slogan is this: the meaning of words is what is to beunderstood in them. The notion of meaning is correlative with the notionof understanding. We can use this point to explain what it is for a musicalwork (or a work of art more generally) to be meaningful.

A mere correlation with understanding is not enough, however. Some-one can understand how weather systems work, but this fact on its owndoesn’t make weather systems meaningful; nor is there any obvious sensein which what such a person understands is the meaning of weather sys-tems. This point seems to apply to the natural sciences in general: theunderstanding provided by the natural sciences is not an understanding ofthe meaning of their objects, so there is no direct requirement that anobject of natural-scientific enquiry be meaningful. At the very least, somefurther argument is required to show that natural sciences presuppose themeaningfulness of their objects.

60 michael morris

What might be missing here? The obvious thought is that there is tooindirect a connection between natural science and teleology. It might bethat the natural sciences presuppose some teleology, general or specific.That is, it might be that the natural sciences in general presuppose thegeneral purposive orderliness of nature; or it might be that certain naturalsciences presuppose that their particular objects have purposes of somekind—perhaps the notion of a biological species presupposes the notion ofa proper development of a member of a species, and the notion of a bodilyorgan presupposes some idea of function. But insofar as these things arestudied by natural sciences, the teleology is not the object of our interest.And this seems to explain why it is that the objects of natural sciences arenot thought of as meaningful by those sciences, nor is it their meaning (ifthey have any) which those sciences aim to understand.

So how does something have to be for a correlation with understandingto make room for the notion of meaning? Here is a simple suggestion.Something is meaningful if it is, in some sense, there to be understood. Thegerundive construction shows the teleology: being understood is what it isfor. And the meaning of a meaningful thing is what is to be understood init: that is to say, it is that part or aspect of the thing the understanding ofwhich is the understanding which the thing is for. This simple suggestionseems to be confirmed by the two contrasting cases of words and the naturalworld. Words are there to be understood: that’s what they’re for. And themeaning of a word is what is to be understood about a word. On the otherhand, consider the natural world. Many may doubt that the natural world ismeaningful, but even they can surely agree with this: to regard the naturalworld as meaningful is to think it is there to be understood; and to look forthe meaning of the natural world is to look for what it is about the naturalworld the understanding of which would be understanding what is thereto be understood about it.

This analysis of meaningfulness is confirmed by the case of works of art:works of art are there to be understood. I shall now offer an independentargument for this claim.

Here are three ways of approaching what is, I think, the same featureof the concept of a work of art. First, we think that works of art demand,or deserve, a certain kind of attentiveness. In some cases (in galleries, or atconcerts, for example), the careful adoption of an appropriately attentiveattitude can be comic; but it nevertheless shows that at least those who


so carefully adopt the attitude believe that attentiveness is demanded,and that they take themselves to be following a common assumption.Second, it often seems natural to talk about the ‘appreciation’ of art. Theetymology of this word—which suggests putting the proper price onsomething—reveals something both inadequate and ugly about this naturaltalk (as if price were what matters). But the word itself, as I understand it,and the idea that appreciation is appropriate to art, both again suggest thatworks of art demand or deserve something. And third, to adapt a Kantianthought about morality,⁶ we want to say that to think of something as awork of art is to think of it not merely as a means. We can use a paintingas a duckboard, or a sculpture as a hat-stand, but in these uses we are nottreating the works as works of art. That implies that to think of somethingas a work of art is to think of there being something which is due to thething itself, and not merely as a consequence of any use it might have. Sofor something to be a work of art is for it to be right to think that there issomething which is due to the thing itself.

This adaptation of what is originally a Kantian thought about moraldemands makes it natural to introduce the concept of justice in explainingwhat is important about works of art. It is natural to say that there is such athing as doing justice to works of art. Not only that: there is a sense in whichjustice to works of art is demanded of us; in some sense, it seems, we oughtto do them justice. But this looks at first sight to be a very peculiar sense of‘ought’. There is something of the precious aesthete in the attitude whichtakes us to have a moral obligation to attend to works of art. And on theother hand too much is lost if the imperative amounts to no more than this:‘You ought to attend to works of art if you like that kind of thing.’ This failsto distinguish works of art, and other essentially meaningful things, fromanything else. The natural solution is to understand this ‘ought’ as relatingto the demands of the work itself: you ought to attend to the work, if youare to do what it asks. And this idea of the demands of the work itself isnaturally explained teleologically: it is to do with what the work of art is for.What a work of art is for is being done justice to: it is there to be done justiceto. This is the sense in which it asks to be done justice to. The feature of the

⁶ Note that the Kant I make use of is the Kant of the Groundwork and the second Critique, not theKant of the third Critique. In fact, I think that the view of works of art which I arrive at is in tune withthe view of the third Critique, but reaching it by means of the third Critique would have been morecomplicated and controversial.

62 michael morris

concept of art which we understand in those three ways I have mentionedis just this, then: to think of something as a work of art is to think of it asbeing there to be done justice to. From which it seems to follow that ifsomething really is a work of art, it really is there to be done justice to.

The formulation in terms of justice is derived from that adaptation ofKant’s conception of moral obligation, but I think it is also true to theexperience of considering works of art. Think again of the experience ofbeing in a gallery or at a concert, for example. It seems to me impossibleto see a picture in a gallery, and understand even the tiniest part of thesignificance of its location there, without thinking that the picture is there(in the world, not just in the gallery) to be given its due. Similarly, it isimpossible to hear a piece of music in a concert without thinking that it isthere (in the world, not just in the concert) to be given a proper and justattention. It is just this kind of ‘due attention’ (Hume 1757: 232) whichthe comic self-styled aesthete wants to be thought to be giving when shefrowns so elaborately in the gallery; and it is in order to find out what suchattention should be directed at that the earnest apprentice goes to classes inmusical appreciation.

What is actually involved in doing justice to a musical work (or a workof art, more generally)? It is surely at least a necessary condition of doingjustice to a work of art that one understand it, at least partially. Considerwhat a response to a work of art would have to be like, if it did not involveunderstanding. We would have to count it as a merely causal, or merelypsychological response. Such a response is one which could, in principle,have been produced by another means—by a drug, for example. Just asit could only be a joke to say that someone did justice to a drug bysuccumbing to it, so it could only be a joke to say that someone did justiceto a work of art by responding to it in a way which could equally well havebeen a response to a drug. (Compare Wittgenstein (1966: 29): ‘Would asyringe which produces these effects on you do just as well as the picture?’)An important part of denying that a work of art is to be treated merely asa means is to insist that to take it seriously as a work of art is to attend towhat is peculiar to it.

Nor would matters be much improved if we required merely thatunderstanding be a precondition for any response which could count as doingjustice to a work of art. This would, in effect, be to treat the work ofart as like a drug which could only work if taken after a meal: it would


be saying that the merely psychological response induced by a work ofart is only available to someone who has understood it. On this view, itis a contingent and empirical fact that the relevant response can only beproduced in someone who has understood the work in question, although itis a conceptual truth—definitional, in effect—that something only countsas a work of art if the response depends on such an empirical precondition.But it is surely not merely a contingent and empirical fact that doing justiceto a particular work of art requires understanding it: that is to say, for eachparticular work of art, nothing could count as doing justice to it which didnot involve understanding it. The proper response to a work of art mustbe one which is not merely preceded or accompanied by understanding: itmust be a response made with understanding, an understanding response.

This seems to be enough to show that if works of art are thereto be done justice to, they are there to be understood. Even if morethan understanding is required for one to do justice to a work of art,understanding is so intimately involved in what it is to do justice to a workof art that we can say that works of art in general, and musical works inparticular, are there to be understood. (It is even tempting to suggest that,given a proper recognition of what is really involved in understanding, awork of art can ask for nothing more than to be understood: that doingjustice to a work of art just is understanding it. But this is more than weneed for our purposes.)

5. Musical Works must be Created

If musical works are types, or any kind of non-particular, it is natural tothink that they cannot be created.⁷ A type is individuated by the conditionwhich things have to meet to count as tokens of it; it is natural to think that

⁷ This conditional is a central feature of an argument presented by Levinson (1980: 7). It is elaborated,in order to draw the conclusion that musical works are not created, by Dodd (2002); that conclusion isanticipated, e.g., by Wolterstorff (1980) and Kivy (1983 and 1987). Levinson himself uses the conditionalto argue in the other direction, against the view that musical works are ‘pure sound structures’; buthe thinks there are such things as ‘initiated types’: see Levinson (1980 and 1990b). I suspect that theapparent plausibility of this idea depends upon an ambiguity of construal: if we say that a work is a typeof performance initiated by a composer, do we mean that the composer initiated the type (surely nota natural thing to say), or do we mean that the work is a type of performance in a tradition or practiceinitiated by the composer (a more natural thing to say—though, of course, I don’t think that worksare types at all)?

64 michael morris

conditions exist atemporally (some say ‘eternally’); so it is natural to thinkthat types also exist atemporally.⁸ The same applies, mutatis mutandis, touniversals: for redness to exist, for example, is for there to be such a thingas being red, a condition which things have to meet to count as red. Someadopt an ‘Aristotelian’ view of universals, according to which universalscannot exist without being instantiated (see, e.g., Armstrong 1989: Ch. 5),and a similar view could in principle be adopted in the case of types. But,for one thing, being instantiated or tokened is not the same as being created;so this ‘Aristotelian’ proposal would not on its own entitle us to think thatmusical works are created. And, for another, it is implausible to think thatwritten music depends for its existence on being performed. The pointremains: if musical works are types, the natural thing to think is that theyare not created but discovered. If this is the natural thing to think, it hasa special significance in relation to the basic argument for the type-tokenview. For that argument works by suggesting that a certain assimilatingexplanation is the natural response to a puzzlement we feel about musicalworks. That explanation is only natural if it is itself understood in thenatural way. It may be possible to understand types as created entities, butif we do this, we lose the clarity which the type-token view promised tobring to the issues.

Whether the view that types are eternal represents the core of the type-token theory, or only a dominant form of it, it faces a simple objection.This is that nothing can be meaningful, in anything like the way worksof art are, without having been created.⁹ The point is intuitive enough,but is underlined by my analysis of meaningfulness as being there to beunderstood. Nothing can have that specific teleological purpose withouthaving been created.

It’s important to note that this does not involve denying natural teleology.It is reasonable to think, for example, that, as a matter of natural fact, hearts

⁸ Those who think it is natural to think of types as coming into existence tend to understand thenotion of a type by example, rather than on the basis of any explanation or definition. Thus Wollheim(1980, §35) explains the notion of a type by means of the examples of the Red Flag and the word ‘red’.But this kind of explanation is not only unhelpful because it doesn’t actually tell us what types are; theexamples are themselves all open to question (it is not obvious to me, for example, that the Red Flagand the word ‘red’ are not particulars). The moment you attempt to explain what types are, in termsof conditions which things have to meet to count as tokens of them, you immediately make it naturalto take types to be eternal entities.

⁹ Note the contrast between this argument, which is in essence a transcendental argument, and thekind of appeal to intuition and common practice which is found in Levinson (1980 and 1990b).


are there to pump blood. That’s what they are for, and their being forthat is quite independent of the hypothesis that they have been created.The problem is rather with the specific teleology involved in works ofart. Being there to be understood cannot be a natural purpose, unless wesuppose that nature itself is created. In fact, considering what would beinvolved in thinking that the world as a whole is there to be understoodlends further support to the claim that nothing can be meaningful withouthaving been created.

Of course, every natural scientist (perhaps also every artist) must thinkthat the world is there to be understood in some sense, and this need notinvolve thinking that the world was created. But the sense in which thenatural scientist or artist, just as such, must think of the world as beingthere to be understood is similar to that in which a cleaner must think ofdirt as being there to be cleaned up: the attitude is induced by the job ofbeing a natural scientist or artist—their job just is to understand the world,we might say—and is not due to the nature of the world itself (as its raisond’etre, perhaps). On the other hand, to think that the world itself is thereto be understood, in virtue of the nature of the world and not because ofthe special job of the natural scientist or artist—to think, perhaps, that itis part of the raison d’etre of the world that it be understood—really doesseem unintelligible without thinking of the world as created. Interestingly,the creation in question is naturally thought of as creation as by an artist:think of the God of the opening pages of Genesis who after each day stepsback, as it were from an easel, and sees that it is good.

Might someone respond on behalf of those who deny that works ofart are created, and say that works of art themselves are not there to beunderstood in virtue of their own nature? Such a person might say that theidea of works of art being there to be understood is really just a projectionof the demands of a particular job, just as it is only as an object of thedemands of the job of being a natural scientist or an artist that we saythat the natural world is there to be understood. This suggestion is barelyintelligible, I think. What is the job whose business is to understand worksof art, and for which works of art are there to be understood? The job ofartist or critic? But these jobs themselves are only intelligible if we have anidea of what works of art are: after all, the artist is someone who producesworks of art, and the critic is someone who interprets and appraises them.Which, then, of the things there are, are the works of art? At least they

66 michael morris

are things which are essentially meaningful. But that means that they arethings which are there to be understood, in some sense which is morefundamental than the sense in which anything is there to be understood bythe person whose job is to understand it.

This abstract point, that meaningfulness of the kind which works ofart possess requires creation by something like an artist, is reinforced byconsidering what it would be to understand a work of art, if works ofart were not created. We would need to suppose that artistic media havea special kind of property, which I shall call being intrinsically affective. Tobe intrinsically affective a medium has to have an effect on a hearer orviewer which is entirely independent of the hearer or viewer assumingthat anything that partakes of that medium was produced by an artist. Soit might be thought that particular kinds of sequences of sound tend tomake people feel sad, or uplifted, or relaxed; or that colours occurring inparticular kinds of combination with other colours tend to make peopleuneasy, or calm, or cheerful. We might suppose that intrinsic affectivenesswas dependent upon historical, cultural, and social factors: people broughtup in different conditions might be susceptible to artistic media in differentways, independently of their assuming that anything that partakes of themedium was produced by an artist. And this might even allow us to grantthat languages can be intrinsically affective: certain phonetic combinationsmight gradually tend to have particular effects on people of particularcultures.

What is there to understand in a work of art, if we concentrate just onthis kind of intrinsic affectivity and ignore any assumption that the work isproduced by an artist? We can, of course, as psychologists, understand andpredict the effect produced in viewers and hearers by viewing and hearingthe relevant works; in this way, and to this extent, we understand themedium. And as hearers and viewers, we might become psychologicallyinformed, and notice and predict these effects on us; in a certain sense weend up understanding ourselves. But neither of these kinds of understandingis understanding the work of art. And neither kind of understanding can berequired to do justice to a work of art. After all, we were led to introducethe notion of understanding precisely to distinguish what works of art dofrom merely psychological effects; but appeal to what I’ve called intrinsicaffectivity seems capable of explaining nothing but merely psychologicaleffects.


There is a broader version of what is at root the same problem: if wedeny that works of art are created we are forced into a suspect conceptionof artistic media. There is, at best, a very limited range of effects of anyartistic medium which are felt quite independently of the assumption thatsomeone put those parts or features there. And artistic media are, in general,precisely media in which artists work to produce works of art: it is unclearthat we could have a conception of an artistic medium as such which doesnot assume that it is a medium in which works of art are made.

The type-token view cannot be defended against these objections byclaiming that although the things which are works of art are eternal entities,they become open to aesthetic understanding only after the interventionof an artist. For I have argued that works of art are not only meaningful,but essentially meaningful. We could not have the very same things withouttheir being meaningful. Consequently, we could not have had the verysame things before the intervention of an artist. (This means that so-called‘ready-mades’ are not properly so-called; but that is hardly a counter-intuitive claim: it is quite natural to think that taking an already-madeobject and placing it in a gallery creates a work of art which did not existbefore.)

Insisting that the meaningfulness of a work of art depends on its havingbeen created exposes us to a possible objection. Surely, it may be suggested,the meaningfulness of a work of art can only depend on its having beencreated if what is there to be understood in a work of art is the reason whyit has been created as it is. That is, we will be driven to suppose that themeaning of a work of art is to be explained in one or both of two ways:by explaining the intentions of the artist, or by explaining the historicalcircumstances of various kinds which gave rise to the work, and which thework may be understood to express. It may be very useful for understandingthe life and mind of the artist to work out her intentions, and it may beimportant for understanding the history of a period to understand what isshown in a work about the circumstances of its production, but neither ofthese things—the objection goes—is understanding the work of art itself.

As it happens, I agree with the objector that neither historical norbiographical understanding counts as understanding the work itself.¹⁰ But

¹⁰ But biographical and historical information can help us to understand the medium, and, as will beclear in a moment, I think understanding the medium is crucial to understanding the work.

68 michael morris

it is not true that the meaningfulness of a work of art can only dependon its having been created if what is there to be understood in a work isthe reason why it has been created as it is. So insisting that works of artbe created does not force us into the unacceptable position we are hereoffered.

But if what is there to be understood in a work of art is not just thereason why it has been created, what difference does creation make? Howcan creation be essential to a work of art’s being meaningful? I will offeronly the beginning—indeed, the caricature of a beginning—of an answer.For the simplest (a ludicrously simple) kind of case, let’s begin with theconception of an artistic medium which is available to someone who thinksthat works of art are not created: artistic media are intrinsically affective, in thesense which was introduced before. Suppose—to keep things ludicrouslysimple—that we have a sequence of sounds which tends to make us sadfollowed by one which tends to make us cheerful. What is the differencebetween such a sequence of sequences (the sad-making sequence followedby the cheerful-making sequence) appearing as a natural occurrence, andone appearing as the result of the work of an artist who understands thisintrinsic affectiveness? Of the latter, but not the former, we can say this:this sequence of sequences is there because it is intrinsically affective inthis way. That looks as if it is enough to ensure that we have somethingwhich is there to be understood; we have something which is meaningful.

Nothing in any real work of art presents itself like this, of course. Itis even doubtful that we can isolate a level of intrinsic affectivity whichunderlies whatever is not intrinsically affective—which underlies, that is,whatever depends on assuming that what we have was created. The crucialparts and features of a work of art strike us, at first sight or hearing, as meantthrough and through: we never react to them as if to something which hasoccurred naturally. Moreover, the medium of any work of art is alwaysformed within a tradition over time, and is always understood within thattradition. The tradition itself is a tradition of using the medium intentionally,given the properties the medium has acquired as an intentionally exploitedmedium within a tradition of using it intentionally. The whole thing is soshot-through with intentionality that we cannot begin to make sense ofthe intrinsically affective qualities of that medium.

For all that, the caricature case does enough to show what we needto do to make sense of the contribution of creation to the meaning of


works, without lapsing into thinking that meaning is to be found in theartists’ intentions or the historical circumstances of the works’ creation. Themedium of the work of art, whatever it is, has properties of its own. Theseproperties are not the intrinsically affective properties of the caricature:they are properties of a medium whose very nature involves their beingexploited intentionally. They may (surely will) include qualities which arecommonly described as expressive and representational. The crucial pointis that at the time of the creation of any given work of art, there areproperties of the medium which are there independently of the intentionsand beliefs of the would-be creator. It is the artist’s business to understandthese independent properties and to know how to work with them. If shedoes understand them, and can handle them, she can produce a work of artwithin that medium of which the following can be said: its essential partsand features are there (in the work of art) because of the properties of themedium.

If this is what the artist does, we have the following account ofthe difference creation makes, which is compatible with denying thatunderstanding a work of art is a matter of understanding the artist’sintention. I claimed in Section 3 that the essential parts and features of awork of art are those which are essential or central to its meaning. If theycan be said to be there (in the work of art) because of the properties of themedium, then it seems that the meaning of a work of art must be due tothose properties of the medium. More precisely: the meaning of the workof art is due to the intentional exploitation of the independent properties ofthe medium. That there is something there to exploit is quite independentof the artist’s intentions: no examination of the artist’s intentions will showus what lies in the medium, ready for intentional exploitation. And thecontribution of the artist’s intention is just this: it brings it about that whatlies ready for intentional exploitation is intentionally exploited. The artist’sintention produces something meaningful, but it does not determine theparticular meaning it has—that is due, rather, to what lies in the medium,independently of her intentions. So we can see why understanding a workof art is not a matter of understanding the artist’s intentions; rather, it is amatter of understanding the medium which is intentionally exploited.

But the difference made by the artist’s intention is still crucial. Withoutthat intention, we could not say that the features of the work were there(in the work) because of the properties of the medium. It is only when

70 michael morris

we can say this that we can say that there is something which is thereto be understood, something which would not exist without existing inorder to be understood. It is only when an artist intentionally exploits thoseindependent properties of the medium that we have something which ismeaningful at all.

6. Musical Works cannot be Types of Performance

Musical works are distinctive (if not unique) among works of art in theirrelation to performance: they are there to be performed. What can we sayabout the relation between performance and work, if musical works aremeaningful in the way all works of art must be?

First of all, there cannot be a merely accidental connection betweenthe two fundamental teleological facts about musical works, that (as worksof art) they are there to be understood, and that (as works of music inparticular) they are there to be performed. Nor is the connection withhearing incidental here: musical works are there to be heard, and theyare only heard in performance. The first thing we need to say about therelation between musical works and performance is just this: performancelets a musical work be understood—it opens it to understanding, we mightsay—by letting it be heard.

But this doesn’t mark a fundamental distinction between musical worksand mechanically reproducible works. Take the case of an etching, forexample. The process of pressing the copper plate on paper lets the workbe seen as it is supposed to be seen, and hence opens it to the kind ofunderstanding which it was designed for. This seems entirely analogous towhat I have said about the relation between musical works and performance,allowing for the fact that one is a visual, and the other an aural, medium.But pressing the copper plate on paper is not performing the etching: etchingsare not performed. What is the difference? The natural suggestion is this: inthe case of a performable art, the process by which the work is opened tounderstanding itself depends on understanding. Performers themselves haveto understand the work, even if only in performing it. What a performerdoes is guided all the time by her understanding of the work.

A performance seems to be connected by understanding to the work ofwhich it is a performance—and not once, but twice. First, its nature as a


performance is defined functionally, as being there to open the work tounderstanding. And secondly, the performance only counts as a performance(rather than a reproduction) by itself depending on an understanding ofthe work. We may say that a performer doesn’t understand the work she’sperforming, but we surely don’t mean that she doesn’t understand it at all: allwe mean is that she has a poor or superficial understanding. But (someonemay object) mightn’t someone just play the notes in a score, withoutany conception of how they should be played—just as, perhaps, someonecould read out loud words in a language she could not understand—andwouldn’t this count as performance without understanding? I think not:we cannot really understand playing as uncomprehending as this, except asa kind of mentally disengaged automatic behaviour; and then it cannot bethought of as a performance of a work.

It seems to me, then, that the notion of a performance cannot beexplained without the notion of understanding; and the object of under-standing is nothing other than the work itself. The most fundamentalobjection to the type-token view of musical works is simply that a tokencannot be related to a type of which it is a token in this kind of way.

This objection can be elaborated in a number of ways, given theconception of what performances are which I’ve just spelled out. Here isone of them. The claim that the relation between works and performancesis the relation between types and their tokens is a reductive claim: the‘of’-relation which holds between performances and the works they areperformances of is supposed to be nothing but the antecedently understood‘of’-relation which holds between tokens and the types of which they aretokens. But the reduction which is demanded, if performances are involvedwith understanding in the way I have claimed, looks impossible.

A type is individuated by a common property which every one ofits tokens must share in order to count as tokens of that type. Thefundamental thing which makes tokens of a given type tokens of thattype is their possession of the relevant property. The type-token viewclaims, then, that the fundamental thing which makes performances of agiven work performances of that work is their possession of some commonproperty. And since the view is a reductive view—aiming to specify therelation between performances and a work without using the notion of awork itself—the relevant common property must be specifiable withoutreference to the work. What performances of a given work have in

72 michael morris

common, if my account of the nature of performance is even roughlyright, is that they are doubly involved in understanding of that work. Sothe type-token view has to suppose that an account of understanding thatwork can be given without reference to the work. But it is just not crediblethat a suitable reductive account of such understanding can be given. Tosuppose that it can is just to enter the familiar minefield which surroundsall forms of intentional reductionism.

Someone might object that this argument depends on my having chosento attack the particular version of the type-token view which holds thatmusical works are types of performance, but in fact the argument will applyin any case, given that musical works are there to be heard and are onlyheard in performance. Suppose, for example, that it were suggested that amusical work is a type of sound sequence. Let us ask: accepting the view forthe moment, for the sake of argument, which sound sequences count astokens of a given work? Given that musical works need to be performedto be heard, the answer has to be: all and only those which are realizationsof performances of that work. Thinking of the relation between a workand realizations of performances of it as the relation between a type and itstokens faces at least the same difficulties as those which face the attempt totreat the relation between a work and its performances in the same way.

Nor do the problems depend on my having chosen to attack the type-token view, as opposed to a universal-instance view or a kind-member view ofthe relation between musical works and their performances. The relationbetween a universal and its instances cannot be doubly dependent uponunderstanding in the way in which the relation between a work and itsperformance is;¹¹ nor can the relation between a kind and its members.

We may also note that once we take account of the double link betweenperformance and understanding, we have space to introduce a plausibleliberalization in our approach to the criticism of performances. Once werecognize that the connection between a performance and the work ofwhich it is a performance depends on understanding, we can accept that itneed be no defect in a performance that it is not a note-perfect performance,in the sense introduced in Section 2. This is helpfully intuitive: we have

¹¹ We should therefore be suspicious of Levinson’s choice of the word ‘instance’ as a quasi-technicalterm for ‘a sound event, intentionally produced in accord with the determination of the work by thecomposer, which completely conforms to the work’s sound and instrumental structure as so determined’:Levinson (1987: 377).


all heard great performances which pull the music about a little, or disobeycertain performance directions, or change the original instrumentation, andit is intuitively just silly to regard these deviations from the letter (as it were)of the score as defects. But more importantly, it allows room for whatcan be seen a priori to be a possibility: a composer can, intentionally andperfectly properly, write a piece of music which is literally unperformable;the whole point of a piece of music may depend on the fact that there canbe no actual note-perfect performance of it. Given that, it can be no defectin any performance of it that it fails to be note-perfect.

7. Puzzlement and the Assimilating Response

If the type-token view is wrong, there must have been something wrongwith the basic argument for it. I suggested that the basic argument is that itoffers (so it is claimed) the best explanation of those three facts:

(i) musical works can be heard;(ii) the existence of a musical work does not depend on any particular

performance;(iii) the same musical work can be performed many times.

The obvious weakness of arguments to the best explanation is that theymay not have considered all of the possible explanations. Are there otherexplanations available? There are. Much more needs to be said about theseother explanations than I can say here, but their very existence is enoughto undermine the initial case for the dominant version of the assimilatingresponse.

The assimilating response to the puzzlement we feel about musical worksdepends on two assumptions about these three facts:

(a) the basic problem manifested in the three facts is the possibility of asingle thing being ‘repeatable’;

(b) the best explanation of the possibility of a single thing being ‘repeat-able’ is provided by assimilating such things to things of a familiarkind.

I have used the word ‘repeatable’ in this interpretation of the diagnosisoffered by the assimilating response, because it is used by some of itschampions (Rohrbaugh 2003; Dodd 2004; Wolterstorff 1975: 122); but I

74 michael morris

leave it in scarequotes, because there is some uncertainty about what itmeans (the dominant version of the assimilating response wants it to beheard as meaning something like multiply instantiable, but without quitesaying that it does mean that).

It is still possible to deny (a), on any natural interpretation of the word‘repeatable’. For just because a performance depends on the performerunderstanding the work, a performance cannot be regarded as a mere‘repetition’ of the work. And thinking a little about the importance ofthe notion of understanding here in fact suggests an alternative explanationof our three facts. That is simply that a musical work is an objective,and objectively meaningful, thing, whose existence and meaningfulness isindependent of any understanding of it. It can be performed again andagain for the same reason as that for which it is possible to look at the sameobjectively existing flower (for example) on more than one occasion. Asan explanation of facts (i)–(iii), this looks as plausible as that provided bythe type-token view.

What assumption (a) does is assimilate musical works to other kinds ofworks of art which might be more simply regarded as ‘repeatable’:¹² forexample, novels (because there can be many copies of the same novel),etchings (because there can be many offprints from the same etching-plate), or photographs (because there can be many prints from the samenegative). Does assumption (b) hold even of these? This is far from clear.An alternative explanation is suggested by the fact that it is not eveninitially intuitive to regard works of art of all kinds as types, or universals,or kinds: it is counter-intuitive (even if bravely defensible) to say thatpaintings and stone sculptures are types, for example. We might then thinkthat the fundamental account of the ‘repeatability’ of ‘repeatable’ workslies in an understanding of what is special about the media in which theyare produced. What is distinctive about the media involved in ‘repeatable’works is that a single act of creation is the source of a number of things (thecopies or prints, for example) which have the same meaning. Why shouldwe count this as one thing? Just because we have a single act of creationand the same meaning.

It is important not to be misled here. We never need to say that thedifferent copies of the same novel just are the same novel: we need only

¹² The assimilation is clearly at work in Wolterstorff (1975).


say that they are copies of the same novel. We do not need to say thatthe different prints of the same photograph just are the same photograph:we can always insist on saying simply that they are prints of the samephotograph. The assimilator may try to press us on this, by asking how twopeople can literally be reading the same novel unless there is some sense (tobe explained, for example, by a type-token view) in which the copies arethe novel. But we can always resist this pressure: what it is for two peopleliterally to be reading the same novel just is for them to be reading copies ofthe same novel. We can always resist the attempt to assimilate the relationbetween a copy and what it is a copy of to that between particular andgeneral. And it looks as if we will have some version of the argument of thelast section to support us in this: the relation between a copy and what it isa copy of must always be an intentional relation (however mechanical themeans of reproduction), which the relation between particular and general(token and type, instance and universal, member and kind) cannot be.

I have claimed that it is far from obvious that assumption (b) is true,that, even where we have ‘repeatable’ works of art, the best explanation oftheir ‘repeatability’ is the one provided by the assimilating response. So farI have only illustrated this with versions of the assimilating response whichtake ‘repeatable’ works to be in some way general. But an assimilatingresponse is one which aims to resolve the puzzlement we feel about theseworks of art by assigning the works in question to some familiar ontologicalcategory, and the candidate categories are not all general. Someone mightclaim, for example, that ‘repeatable’ works are continuants, of which thevarious copies are ‘stages’, perhaps, or ‘embodiments’ (see Rohrbaugh(2003), who takes the copies to be ‘embodiments’). This proposal ineffect assimilates ‘repeatable’ works to familiar things like tables, ships, andpeople. But we should ask why this is a better explanation of unity despite‘repeatability’ than the simple one I have offered already, in terms of amedium which allows a single act to generate multiple copies which havethe same meaning. And as before, it seems extremely doubtful that theintentional relation which must hold between a copy and what it is a copyof can be modelled intelligibly in terms of the relation between the stagesor embodiments of a continuant and that continuant itself.

No doubt assimilators, of whatever kind, will have something to say inresponse to these points: after all, I have offered no more than sketches ofalternative explanations of facts (i)–(iii), rather than any finished theory.

76 michael morris

Nevertheless, even these brief remarks seem to me enough to take theshine off the assimilators’ arguments to the best explanation. And given thefundamental defects of the views which the assimilators end up proposing, Ithink we should feel no temptation to adopt any version of the assimilatingresponse.

In the face of the problems which confront the assimilating response, itis tempting to say that we should just regard musical works (or performableworks in general, perhaps) as sui generis. This may be right, but I think it isnot enough just to say this and have done with it. For saying that musicalworks (or performable works in general) are sui generis may be understoodas trying to do something which the assimilating response also aims to do:it may be understood as trying simply to remove the puzzlement which wefeel about the nature of musical works. Accept that musical works are suigeneris, it might be taken to say, and stop worrying about them.

There are two ways in which this seems inadequate. First, it seemsto forestall the general overhaul of our everyday ontological assumptionswhich acknowledging the true nature of musical works seems to demand.The supposedly familiar, spatio-temporally well-defined items like soundsequences or performances turn out not to be more basic than musicalworks: we need, then, to revise our usual assumptions about what shouldbe taken as familiar and unpuzzling in ontology, and to examine our reasonsfor holding them.

And, second, it is not at all obvious that the right response to anontological puzzlement, like the one we feel about musical works, is toattempt to remove it. It seems to me that to get a serious sense of thenature of musical works, we don’t need just to acknowledge and recordthe fact that they are sui generis: we need to keep their peculiarity alive. Itis tempting here to draw a parallel between the task of the philosopher ofmusic and that of the musical performer. The philosopher of music, wemight say, needs to be attentive to the true nature of musical works, assuch and in general, just as a performer needs to be attentive to the truecharacter of a particular work she wants to perform. And in both casesthat attentiveness cannot exist without something which is at least akin topuzzlement: the philosopher’s attentiveness to the nature of musical worksdepends on keeping alive a sense that musical works are not just any oldthings, just as the performer’s attentiveness to the character of a particularpiece of music depends on not letting any feature of it seem merely routine.


This seems to be required for the philosopher to do justice to the nature ofmusical works, just as it is for a performer to do justice to the piece she isplaying.¹³

References

Armstrong, D. M. (1989) Universals: An Opinionated Introduction, Boulder,CO: Westview Press.

Davies, S. (2001) Musical Works and Performances, Oxford: Clarendon Press.Dodd, J. (2000) ‘Musical Works as Eternal Types’, British Journal of Aesthetics

40: 424–60.Dodd, J. (2002) ‘Defending Musical Platonism’, British Journal of Aesthetics,

42: 380–402.Dodd, J. (2004) ‘Types, Continuants, and the Ontology of Music’, British

Journal of Aesthetics 44: 342–60.Dummett, M., (1975) ‘What is a Theory of Meaning?’, in S. Guttenplan

(ed.), Mind and Language, Oxford: Clarendon Press.Goodman, N. (1981) Languages of Art, 3rd edn, Brighton: Harvester Press.Heidegger, M. (1956) ‘The Origin of the Work of Art’, trans. A. Hofstadter;

reprinted in M. Heidegger, Basic Writings, ed. D. F. Krell, 2nd edn,London: Routledge (references to the reprint).

Hume, D. (1757) ‘Of the Standard of Taste’, reprinted in D. Hume, EssaysMoral, Political, and Literary, ed. E. Miller, Indianapolis: Liberty Fund.

Kivy, P. (1983) ‘Platonism in Music: A Kind of Defence’, Grazer Philosophis-che Studien 19: 109–29.

Kivy, P. (1987) ‘Platonism in Music: Another Kind of Defence’, AmericanPhilosophical Quarterly 24: 245–52.

Levinson, J. (1980) ‘What a Musical Work Is’, Journal of Philosophy 77:5–28.

Levinson, J. (1987) ‘Evaluating Musical Performance’, Journal of AestheticEducation 21: 75–88; reprinted in Levinson (1990c) (references to thereprint).

Levinson, J. (1990a) ‘Authentic Performance and Performance Means’, inLevinson (1990c).

¹³ I am very grateful to Julian Dodd for provoking me to think about the issues of this paper, to PaulDavies for many conversations about them, and to Stephen Davies, Terry Diffey, Jerrold Levinson,Stefano Predelli, Roger Scruton and Kathleen Stock for their comments on an earlier draft.

78 michael morris

Levinson, J. (1990b) ‘What a Musical Work Is, Again’, in Levinson (1990c).Levinson, J. (1990c) Music, Art, and Metaphysics, Ithaca, NY: Cornell

University Press.Rohrbaugh, G. (2003) ‘Artworks as Historical Individuals’, European Journal

of Philosophy 11: 177–205.Wittgenstein, L. (1966) Lectures and Conversations on Aesthetics, Psychology

and Religious Belief, ed. C. Barrett, Oxford: Blackwell.Wollheim, R. (1980) Art and Its Objects, 2nd edn, Cambridge: Cambridge

University Press.Wolterstorff, N. (1975) ‘Toward an Ontology of Art’, Nous 9: 115–42.Wolterstorff, N. (1980) Works and Worlds of Art, Oxford: Clarendon Press.

Further Reading

A clear statement of the kind of view which is my target is Wolterstorff(1975), where musical works are said to be kinds. The view that musicalworks are types is often associated with the groundbreaking work of Levin-son (1980), although his version has a number of awkward complexities.A bold, clear, and austere version of that kind of view (nearer perhaps toWolterstorff ) is Dodd (2000).

3

Versions of Musical Worksand Literary TranslationsS T E P HE N DAV IE S

The composition of a musical piece can fall short of the work’s completion.In 1821, Schubert sketched an E major symphony he did not finish.The following year, he abandoned another symphony after writing twomovements and outlining its scherzo. We call it the ‘‘Unfinished.’’ Mahlernotated much of his Tenth Symphony in short score, but only the firstmovement was fully orchestrated before his death.

A less often remarked fact is that a work’s composition can overshoot itscompletion. It is the description apt for these cases that is the topic of thischapter. But before I get to that, it is useful to describe some of the signsthat show a work to be finished.

1. What Signifies the Completion of a MusicalWork?

There are a number of indicators that a musical work is complete. (1) Thecomposer declares that it is so. In a letter to his father dated April 10 of 1784,Mozart writes of his piano concerto in G, K. 453: ‘‘I have finished todayanother new concerto for Fraulein Ployer’’ (Anderson 1966: 874). (2) Orthe composer marks the score or a catalogue. In the case of K. 453, Mozartentered the opening theme into the thematic catalogue he had started in

80 stephen davies

February 1784.¹ (3) Or the score is copied for performance. Along with oth-er manuscripts, that of K. 453 was dispatched to Mozart’s father in Salzburgon May 15, 1784. Mozart wrote: ‘‘I am not particular about the symphony,but I do ask you to have the four concertos copied at home, for the Salzburgcopyists are as little to be trusted as the Viennese’’ (Anderson 1966: 876).(4) Or the work receives a public performance designated as its premier.K. 453 was played publicly for the first time with Fraulein Ployer as thesoloist on June 13, 1784 (Deutsch 1966: 225). (5) Or an authorized versionof the score is printed and sold. The published score of K. 453 was advertisedfor sale in Vienna in August and September of 1785 (Deutsch 1966: 249–52).

None of these signs of a work’s completion is necessary. The unsigned butfinalized manuscript of many an unperformed and unpublished symphonyhas languished in its composer’s bottom drawer. Also, because they aretemporally spread, the combination of these markers can leave the precisemoment of completion somewhat uncertain. Writing to his father on June12 of 1784, Mozart observes of the autograph score of his piano concertoin D, K. 451, that there is ‘‘something missing’’ in the solo passage in C inthe Andante. He continues: ‘‘I will supply the deficiency as soon as possibleand send it with the cadenzas’’ (Anderson 1966: 880). Yet the concerto was‘‘finished’’ about three months earlier, on March 22 of 1784, and premieredwith Mozart as soloist on March 31 at a subscription concert. It was fairlycommon for Mozart not to write out the detail of the soloist’s part if hewas its player. He would later record the part in full for the concerto’spublication.

Nevertheless, if all these signs are in place, we can have a high levelof confidence that the work is finished. (Cases in which an unfinishedwork bears all these markers are conceivable but, in practice, more or lessnon-existent.)

2. Versions by the Composer

As I indicated at the outset, the process of composition sometimes outlivesthe work’s completion. In other words, the signs of completion, including

¹ Bach penned the letters ‘‘S. D. G.’’ in the margins of his scores, which translates as ‘‘solely tothe glory of God.’’ At the end of his scores, Haydn wrote: ‘‘Fine. Laus Deo.’’ Other composers havesigned and dated their scores.

versions of musical works/literary translations 81

public premier and publication, are plainly in place, and the composerdesignates that the work is finished, yet he later changes it in waysthat should affect its identity. As I discuss below when considering (self-authored) transcriptions, sometimes the result is a new, derivative work. Myconcern, though, is with the case in which the composer’s post-completionefforts result in a change or addition to the finished work, not in a newone. I call the result a version.

Here are some examples: Bruckner revised and altered his early sym-phonies after their first publication. For instance, in 1891 he recomposed hisFirst Symphony of 1866.² Published editions and manuscripts of Chopin’sworks differ, and not only as a result of editorial error. Chopin wasresponsible for most of the variants. This has led some musicologists toclaim that Chopin did not share our modern concept of musical works ascompleted, fixed, re-identifiable individuals. I disagree (see Davies 2001:92–7, 119–23). Whereas composers of earlier eras quickly moved on tonew pieces, even if they frequently inserted into these material borrowedfrom prior works written by themselves or others, Chopin and his contem-poraries became more reluctant to relinquish control of their compositions,even after publication. I do not interpret this as showing that the workconcept had yet to take shape. Instead, I see it as reflecting a pragmaticconcern within a tough but unregulated market, both to do what wasnecessary to cater to the taste of the purchasing public and to exploit anywork for all that could be wrung from it.³

Hardheaded pragmatism can be seen also to lie behind the creation ofwork versions in other situations. Mozart had a great success with DonGiovanni in the provincial center of Prague, but when he took it to Viennathe court singers expected him to add numbers tailored for them, whichhe did. Throughout the nineteenth century, composers who exportedtheir operas to Paris had to conform to the French passion for ballet byinserting dances. When Stravinsky lost access to royalties for Firebird and

² To complicate matters, the status of many editions of Bruckner’s symphonies is controversial.Other composers and editors mangled the early editions and the twentieth-century scholarly editionsof Robert Haas and Leopold Nowak by no means agree. My comments do not rely on theseuncertainties, however.

³ The same phenomenon is evident with new music that is posted on the Internet and frequentlyremixed. Composers of such pieces seem to have a robust idea (the standard one) of what a work isand of when it gets released to the community, but they take advantage of the chance to play aroundwith it after ‘‘completion.’’

82 stephen davies

Petrushka, following the revolution in Russia, he re-scored them and hadthe new versions published in Europe and America. The alterations aremodest—the notations of certain rhythms are simplified and there is areduction in the numbers of flutes and such like. As was plainly intended,the changes did not depart so far from the originals that they counted asnew pieces, but they did make it more likely that the new incarnationswould be played and that the revenue could again go to the composer. Inall these cases, what resulted were versions of old works, not new ones.

By contrast, consider Stravinsky’s composition of Les Noces. With thesound structure more or less intact, he struggled with the orchestration.He scored the piece for a large orchestra and, again, for mechanical pianos,before settling on the instrumentation that is now familiar, which usespercussion and four pianos played live. On the account presented here, thework does not have three versions because the various changes antedatedthe work’s completion. Instead, it had several drafts. They were phasesin its initial composition, not alterations made after its finalization. Moregenerally, it is not unusual for a composer to tinker with the work duringrehearsals or after its premier, as it is prepared for publication. Thoughthe completion date is not always clear-cut, these changes usually concernthe work’s finalization. Authors’ versions, if there are any, come after thatpoint.

As I use the term, a version is produced by the work’s composer if hechanges elements that should be constitutive of the work’s identity after thepiece’s completion, and where the alterations intentionally and moderatelyalter identity-relevant features of the original, but without resulting in theproduction of a new but derivative work. Drafts are like versions, exceptthey are made prior to the work’s completion.⁴ Multiple versions of thesame musical work can co-exist. Though the final version sometimes hasa special authority because it indicates the composer’s definitive thoughtsabout the work, this is by no means always the case. For instance, the operacomposers who adapted their works for Paris productions often foundthe process distasteful even if pragmatically desirable. The order in which

⁴ Where one studies the psychology or history of the work’s creation, drafts may be more interestingthan versions, but versions are more philosophically intriguing, which is why I focus on them. Theproduction of succeeding drafts is a predictable part of the piece’s development and completion. Bycontrast, versions are ontologically provocative, since, with them, the identity of the work apparentlysurvives alteration in the kinds of features we would normally think of as work-constitutive.


versions are produced implies nothing on its own about their aestheticimportance.

The introduction of work versions to our musical ontology is messy,but so what? It respects a lack of neatness shown by composers themselvesin the labeling and dissemination of their music. And, apart from the dis-comfort that untidiness among our classifications can bring, this ontologicalprofligacy is supportable. All that is required is that the performers (analysts,historians) make clear that they are dealing with the version published inParis, or the version of 1863, or the version for Prague as against the one forVienna, or whatever. In other words, even if works with multiple versionsare equivocal, we can usually individuate their versions clearly enough.

3. Versions by Others

A version of a work can be produced by a person other than its composer.This happens, for example, when an unfinished piece is completed by athird party. Deryck Cooke produced a performing version of Mahler’sTenth Symphony. Each of the many composers who continued the last,unfinished fugue of Bach’s Art of Fugue produced a different version of thework. Mozart’s Requiem exists only in a version finished by three of hispupils. For these cases, there is no definitive, finalized composition. Thereis the piece left unfinished by the composer and one or more versionsindicating a way of completing it.

Versions of a work are also produced by music editors if the composer’stext, or extant copies of it, are ambiguous or conflicting. If the editorchooses between several possibilities—for instance, she decides that theC in measure 40 should be sharpened, though the sharp is indicated inonly some sources for the work and is not mandated by the performancepractice or conventions—the result is a version.

4. Work Versions versus Performance Interpretations

Work versions should be distinguished from performance interpretations.Because the prescriptions addressed to the work’s performers underdeter-mine the detail of the performance, the players must go beyond what is

84 stephen davies

work determinative in rendering the work. There are many ways of doingthis while remaining faithful to the piece. The differences that result markdistinctions in the interpretations the work receives.⁵ When a composerinterprets his own piece in performance, the outcome is an interpretation,not a new work version.

The production of a work version requires changes in features that areusually constitutive for pieces of the kind in question. By contrast, accurateinterpretations respect what is work-constitutive. They vary qualities,features, or levels of detail that are not identity determining for the work.For instance, it may be prescribed that the performer decorates the melodywhen it is repeated, but the detail of that decoration might be left openand the player’s choices add to her performance interpretation. Though thework includes decoration, no particular set is mandated and many choices(even if they are not all equally tasteful or effective) can be consistentwith respecting the work’s identity. And even where the interpretationdeliberately deviates from what the composer instructs, the acts involvedare ones of performance, not composition.

Where one draws the line between work versions and work interpreta-tions is likely to depend on the level of detail at which works of the relevantkind are specified. In other words, it depends on the work’s ontologicalcharacter. If the work is ‘‘thin,’’ wide variety among its instances is likely tobe a function of its interpretation, not of its re-composition into versions.If the work is very ‘‘thick,’’ significant differences between its instances,where they are not accidental, may be indicative of versions.

Interpretations within some performance traditions share important char-acteristics with work versions, however. I have in mind cases in whichinterpretations attain an autonomous standing, with the expectation thatthey are to be repeated and preserved, and in this resemble work versions.Also, where the piece being interpreted is ontologically thin, its interpre-tation involves creative decisions of the kind that composers make, andthese decisions post-date the work’s completion, which again makes themsimilar to work versions. The result can be called an interpretation version.It is not a work version, as I have characterized that notion, because it

⁵ This is not to say that interpretation comes into the picture as something added only afterfaithfulness has been achieved. Interpretation is not reserved solely for the ‘‘gap’’ between the workand the concrete detail of its performance. It reaches all the way down, as it were.


aims to establish a repeatable way of performing the source work, not torecompose that work.

Interpretation versions are common and well known, but let me intro-duce an out of the way example: in the Balinese gamelan gong kebyartradition, Teruna Jaya is a piece that has existed in its modern form formore than fifty years. Though much is common between renditions ofthe work by different ensembles, there is also considerable variety. It isexpected both that different performance groups will create their owninterpretation and that each performance group will stick closely to theinterpretation it establishes as its own.⁶ Because these interpretations areworked out, are preserved and repeated, and are associated with partic-ular groups, they are paradigms of interpretation versions. They do notinvolve post-completion changes to features that the tradition treats aswork-constitutive, but neither are they as ephemeral as interpretations thatare specific to a single performance.

A more familiar example of an interpretation version is the rock ‘‘cover.’’Because rock songs are typically rather spare as regards their work-identifying elements, re-recordings can remain faithful while differingin many respects from the original.⁷ Joe Cocker’s account of ‘‘With a LittleHelp from my Friends’’ differs considerably from the Beatles’, but is notthereby unfaithful as a performance, I maintain. As a recording, the coverattains a status in its own right as a repeatable individual—not only can itbe re-played, it can be performed again and again by Cocker or by his emu-lators—so it is a version, not a one-time-only interpretation. Because thecover does not involve the re-composition of work-constitutive elementsof the original, despite differing in many respects from the Beatles’ source,it qualifies as an interpretation version, however, not as a work version.⁸

⁶ Over the years, a group’s interpretation can evolve and alter along with wider aspects of theprevailing style, but these alterations are more gradual and less self-conscious than is the creation of theinitial interpretation.

⁷ I defend this account of rock’s ontology in Davies (2001). Gracyk (1996) takes the different viewthat the works in rock are electronic compositions that are presented via recordings and, hence, thatare not for performance (see also Fisher 1998). In his theory, the cover is a new but derivative piece,not a new performance of the same song, as I maintain.

⁸ In Davies (2001: 180–1), I suggest that Benny Goodman’s band arrangement of George Gershwin’sFascinating Rhythm and Glenn Miller’s rendering of Joe Garland’s In the Mood qualify as independentbut derivative works, because they achieved the status of autonomous, repeatable versions preserved viarecordings and notations. It is no less plausible to regard them as interpretation versions, as I imply here,however. Unlike transcriptions, which are discussed in the next section, these adaptations generateperformances that are faithful to the original, ontologically thin, sources.

86 stephen davies

For reasons that have been already acknowledged, interpretation versionscan be like work versions in many respects, and the terminologies I havespecified are not clearly observed in ordinary discourse about music. Nev-ertheless, it is worthwhile to mark their differences. We expect completedworks to receive differing interpretations and we regard interpretationalvariety and creativity as consistent with faithfulness to the work beingperformed, so long as its work-constitutive elements are preserved. Also,where works are thin, we appreciate that interpreters are creative after themanner of composers. As well, there is nothing surprising about the factthat interpretations can achieve standing in their own right as repeatable,persevering entities. By contrast with all this, it is disconcerting to have toaccept that a work’s identity can survive alteration to its work-constitutivefeatures following its authorization as a completed piece, and to allow thata single work can exist in multiple incarnations. These latter characteristicsare the hallmarks of what I have called work versions, as opposed tointerpretation versions.

5. Composers’ Versions versus Transcriptions

Just as work versions should be distinguished on the one side fromperformance interpretations and interpretation versions, so, on the other,they must be separated from derivative yet distinct new works, a prominentexample being transcriptions. Where a medium-specific piece is adapted toa new medium, as when a nineteenth-century symphony is rewritten forthe piano, the result is a distinct work, I claim (Davies 1988). Usually thetranscription postdates the original, but it can anticipate it if, for example,the composer creates a symphonic work on the piano. The first outing ofThe Rite of Spring came when Stravinsky and Debussy played the composer’stranscription for two pianos.

Work versions involve changes to work-constitutive properties. Theircreation can include alterations to pieces’ orchestrations. As mentionedabove, Stravinsky reduced the number of wind players in later versions ofFirebird and Petrushka. These modifications did not depart from the mediumin which the original was written, however. For these, the medium isthe symphony orchestra, but such an orchestra can be treated flexibly asregards its forces; in particular, with regard to the number of woodwind,


brass, and percussion lines. When he created new work versions of theseearly ballets, Stravinsky lightly reworked them in the same medium.Transcriptions also involve changes to work-constitutive instrumentations.⁹I regard transcriptions as new works, not as new work versions, however,because they involve changes in instrumentation that alter the work’smedium and its medium-specific contents. When one instrumental mediumis replaced by a significantly different one, the work’s contents must beadapted accordingly. Among other matters, this usually involves changesto the piece’s notes or their relative dispositions, and imposes differenttechnical requirements and constraints on the piece’s new performers.When a work’s features are filtered through a new medium, the impact isusually sufficiently radical that a new piece results.

A musician might begin to compose, using a finished work as her source.If she carries the process of re-composition far enough, she writes a newpiece. The new work is influenced by the original, and perhaps audibletraces of this inspiration remain detectable in quotations or allusions. In adifferent scenario, the composer does not carry the process very far and sheconceives of herself as revising the source rather than going beyond it. Theproduct is what I have called a work version. The practice of transcriptionlies between these extremes.¹⁰ The audible relation with the original ispreserved, as is the sound-structural outline and much else, yet the changein instrumental medium distances the transcription from its model, withthe result that a new work is produced.

At what point does the process of composition leave the original workbehind? When do alterations to the original cross the border between awork version and an entirely new, albeit derivative, piece? There is nosimple answer, of course. And as long as we are all clear about the degreeof arbitrariness involved, perhaps we can say what we like. Nevertheless,my intuition is that transcriptions usually achieve a degree of independencefrom their sources that versions do not. This is intended and reflectsdifferences in what typically motivates the act of reconstruction in the

⁹ If the instrumentation of a piece is not work-determinative, varying its orchestration wouldresult in an interpretation, not a transcription. Many musical pieces—folk songs, Bach’s Art of Fugue,most pop songs, much music composed prior to the standardization of the orchestra—do not havework-determinative instrumentations.

¹⁰ On the border between transcriptions and entirely new pieces is the ‘‘fantasia after’’ and ‘‘homageto.’’ Sometimes these are labeled as transcriptions, an example being Percy Grainger’s ‘‘transcription’’of Tchaikovsky’s ‘‘Waltz of the Flowers.’’

88 stephen davies

first place. Work versions re-compose their sources, but in a deliberatelyrestrained way, so that they do not threaten to count as new pieces. Alsorelevant is this: if we were to count transcriptions as work versions, notas new works, there would be the implication that the composer of thetranscription is a co-author of the work. We would have, not Liszt’stranscription of Beethoven’s Fifth, but Beethoven’s Fifth as composedby Liszt. When one composer completes another’s work, the final resultshould be credited to both. When one composer transcribes another’spiece, she deserves acknowledgment as the transcription’s author, but not asthe co-author of (a version of) the original.¹¹

6. Versions in other Artforms

Much of what I have written about work versions in music could be appliedto works produced in other artforms. The novelist can rewrite her alreadypublished book. The movie can be re-released as the director’s cut, themodified-for-TV print, and the DVD supplement, which might includedeleted scenes and an alternative ending. The artist can add to a previouslycompleted oil painting and the sculptor can cut bits from a marble statuethat was designated as finished long before.

These last cases provoke the following thoughts: it is common todistinguish artforms with works that must be singular from those in whichmultiple instances of the work are possible. Handmade sculptures, drawings,and paintings are singular; prints, cast statues, novels, plays, songs, movies,and musical works are potentially multiple. Some philosophers, Currie(1988) being one, reject this division, however. They regard all artworksas potentially multiple. A doppelganger of Mona Lisa would instance thesame work as Leonardo’s painting, provided it was accurate enough. Doesthe earlier observation—that oil paintings and marble sculptures can existin more than one version—show that ontological multiplists are right?

Handmade paintings and sculptures can be made in sets. An artist mightpaint more or less identical copies of a single scene, and sell one to the

¹¹ Sometimes transcribers deserve and get equal billing with the composer of the original, as Raveldoes for his orchestral transcription of Mussorgsky’s Pictures at an Exhibition, which was written for thepiano. This equal acknowledgment is justified by the level of creativity and originality displayed byRavel, but we do not identify him as co-author of the work.


palace, one to the church, and so on. We might regard these paintings aspresenting different versions of the same subject matter, but there is no reasonto treat the members of the set as versions of a single, generic painting.Members of the set can be identified as discrete but related works. When itwas said earlier that there can be more than one version of an oil painting,the relevant example was not this but a different one, which I now describe.The artist completes a work but returns (much later, let us assume) to thesame canvas or statue and takes to it again with brush or chisel. In doingso, he does not obliterate the original, though he does alter it in ways thatshould be relevant to its identity. He adds an angel or two in the sky, trimsfat from the statue’s thighs, or whatever. As a result, the work exists throughtime in more than one version, and to that extent it appears to be multiple.

I deny that this counts against the commonsense view that handmadepaintings and sculptures are singular, however. The multiple versions ofplays and novels are numerically distinct and coexistent. This is possiblebecause they are types or kinds indicated as such by an artist. By contrast,the painter in oils or the sculptor in marble can create a version onlyby laboring directly and concretely on the original work and its material.The new version supplants the earlier one; in other words, such worksexist in only one version at a given time. The work preserves its identitythrough the process of change, but this does not make it a universal or type.Instead, it might be compared to a person, who can be stooped and graynow yet be the same as the child who formerly was blond and upright. Oilpaintings and hewn statues can have successive versions, but unlike playsand novels, they cannot have coexistent versions (and instances), whichis what would be required if they were to be multiple in the relevantsense. Rather than deducing from the possibility of their having versionsthat paintings and sculptures are all potentially multiple, it is more sensibleto observe that the ontological differences between singular and multipleworks dictate a corresponding difference in what is involved in creatingversions of them.

7. Translations in Literature and Poetry

Here is a new issue. What is the status of translations of literary worksto a language other than that in which they are written? Are translations

90 stephen davies

equivalent to work versions or to transcriptions? Is a translation a variantthat instances the original work, or is it a distinct but derivative piece?

If we ask a person if she has read War and Peace, she is likely to answer‘‘yes’’ if she has read an English translation. It would be odd if she replied:‘‘No, only the translation.’’¹² But it would be hasty to conclude fromthis that the translation thereby counts as an English-language version ofTolstoy’s novel. It is likely that the respondent interprets the question asasking if she is familiar with Tolstoy’s work via its translation, not as askingif she accepts their co-identity.

People turn to a translation usually because they do not read the languageof the original and therefore, for the case of War and Peace, cannot accessTolstoy’s work as he wrote it. In the past, musical transcriptions werethe same. Before there were radios, recordings, or affordable subscriptionconcerts, piano transcriptions provided the main route to many orchestralworks because households frequently possessed a piano and a competentpianist. It would have been natural in those times to claim acquaintancewith Beethoven’s Fifth on the basis only of Liszt’s transcription. Nowadayswe are liable to insist that a person does not really know Beethoven’s Fifthuntil she hears the work as it was written for orchestra. A person’s claimto have read Tolstoy’s work by reading a translation might seem natural,then, only because the original remains comparatively inaccessible to mostnative speakers of English.¹³ But in that case, it could be that the translationis best regarded as a distinct piece, which is the way Liszt’s transcriptionshould be considered.

Notwithstanding these observations, there are good reasons for thinkingof the translation as a work version. It is intended and represented as avariant of the original, not as a new, though derivative, piece. And thisis plausible, to the extent that it is possible to achieve a high degree of

¹² That answer might be more likely, though, if the question had been ‘‘Have you read Voiynah eemeer?,’’ supposing the person to understand this as the Russian title.

¹³ Compare the role of literary translations with that played by mass reproductions of singular orlimited edition paintings and sculptures. Reproductions can range from photographs to duplicates donein the same medium as the original. We are liable to regard these as standing in for the original but notas work versions. Perhaps this is because they are often less than fully faithful to the original. If we hada matter replicator that could reproduce physical objects down to the molecular level, we might beinclined to accept clones of the originals as work equivalents. They are not work versions, because theypreserve rather than modify work-identifying features. And neither are they genuine instances, if theworks in question necessarily are singular or limited in number.


accuracy (as regards content and general mood) in converting one languageto another. Admittedly, it can be difficult for a foreigner to appreciate thenuances and associations of a story in translation, but that problem maybe cultural rather than linguistic. A foreigner who could read the work inits original language might miss exactly the same features through a lackof sensitivity to the conventions and values of the culture in which thestory finds its home. If the work is the story, and if the story survivesits translation, the translation is best regarded as a version, not as anautonomous work.

Though I stand by it, this conclusion should be viewed warily. The moreresistant to translation are linguistic and semantically relevant features of thework, the nearer translation approaches the case of musical transcription;that is, to being an autonomous though related work. An example might bethat of translating from a tone language (such as Mandarin) to a non-tonelanguage (such as English). The relevance of pitch to meaning may beexploited in the original to present artistically relevant possibilities thatcannot survive in translation. A different, more obvious case is that ofpoetry or rhyming verse. It is notoriously difficult to reproduce in anotherlanguage the subtle semantic content and ease of expression of somethingwritten in rhyming iambic pentameter while respecting the same formalconstraint.

The nearer language comes to being merely a vehicle for meaning, sothat the content of the story is ‘‘transparent’’ to the language in which itis written, the nearer translations approach work versions so long as thetranslator aims successfully to produce a literal rendering of the original. Butthe more the language of writing becomes a medium that draws attention toitself and, prism-like, modulates, filters, and inflects the message expressedthrough it—or alternatively, the freer and more ‘‘creative’’ the translatortries to be—the more translations are analogous to musical transcription.In this latter case, translations must function like transcriptions, eitherbecause they lose the original’s interplay between medium and content orbecause they reproduce an appropriate effect via the different resources andpossibilities of the alternate language. In other words, where the treatmentof the language of the original is such that a translation cannot convey whatis required simply by telling the same story under similar formal constraints,the translation might better be seen as a new but derivative piece than as

92 stephen davies

a version of the original. But where the audience is not multilingual, thetranslation then provides a useful substitute, if not a work version.¹⁴

References

Anderson, E. (1966) The Letters of Mozart and his Family, 2nd edn, London:Macmillan. 2 vols.

Currie, G. (1988) An Ontology of Art, London: Macmillan.Davies, S. (1988) ‘‘Transcription, Authenticity and Performance.’’ British

Journal of Aesthetics 28/3: 216–27.Davies, S. (2001) Musical Works and Performances, Oxford: Clarendon Press.Deutsch, O. E. (1966) Mozart: A Documentary Biography (2nd edn), trans.

E. Blom, P. Branscombe and J. Noble, London: Adam & Charles Black.Fisher, J. A. (1998) ‘‘Rock ’n’ Recording: The Ontological Complexity

of Rock Music,’’ in P. Alperson (ed.), Musical Worlds: New Directionsin the Philosophy of Music, University Park, PA: The Pennsylvania StateUniversity Press, 109–23.

Gracyk, T. (1996) Rhythm and Noise: An Aesthetics of Rock Music, Durham,NC: Duke University Press.

Further Reading

Davies, S. (2003) Themes in the Philosophy of Music, Oxford: OxfordUniversity Press.

Dilworth, J. (2001), ‘‘A Representational Theory of Artefacts and Art-works,’’ British Journal of Aesthetics 41/4: 353–70.

Livingston, P. (2003) ‘‘Pentimento,’’ in B. Gaut and P. Livingston (eds),The Creation of Art, Cambridge: Cambridge University Press, 89–115.

Rohrbaugh, G. (2003) ‘‘Artworks as Historical Individuals,’’ EuropeanJournal of Philosophy 11/2: 177–205.

¹⁴ I am grateful to Paisley Livingston for challenging me to clarify my views on this topic and toJonathan McKeown-Green and others too numerous to list conveniently for comments and suggestionson drafts of this paper. The versions will all be mine.

PA RT I I

Musical Expression


4

Expression in MusicD E RE K MAT RAV E RS

Progress in enquiry depends as much on getting the right questions asit does on getting the right answers. This is nowhere better illustratedin recent philosophy than in the discussion of the expression of emotionby instrumental music. What I hope to do in this paper is to argue forframing the question in one particular way, and, having done so, evaluatethe answers that have been given to that question (even if the answers werenot originally given in answer to that question).

It has become more or less traditional to approach this through attemptingto make sense of the claim ‘the music is sad’. Clearly this sentence is simplya placeholder; nobody should take it to be an example of serious criticism(Kivy 1989: 181). The narrowness of focus consequent on taking this as theplaceholder does, however, sideline several issues. In particular, it directsour attention to the expression of sadness by a brief passage of music, ratherthan the expression of sadness by an entire piece (although the latter maywell emerge from relations between the former). However, if we can firstidentify and second solve the philosophical problem this simple placeholderraises, that would itself be progress.

There have been many attempts to identify the problem, most alongthe lines of pointing out that, in the central case, we operate with a looseequivalence between ‘being sad’ and ‘feeling sad’. As music is insensate itcannot feel sad, and hence, given the equivalence, cannot be sad. This,however, does not seem the most perspicuous way of stating the problem,as we are happy to use ‘sad’ of insensate objects or events (such as ‘a sadoccasion’, ‘sad news’, ‘a sad book’). Rather, I think, we should think firstabout what we could expect philosophy to contribute to the debate andframe the question so as to draw on this expectation. It is part of philosophy

96 derek matravers

(at least) to explicate what we mean by our claims: that is, to provide aperspicuous account of their content. Hence, the problem would be toexplicate the content of ‘the music is sad’: what would someone who madethat claim mean? The answer to that question, at least as a preliminary,seems to me reasonably obvious: they mean that the experience of themusic has a certain phenomenology—when one listens appropriately, onewill have an experience of it the right description of which is ‘the music issad’. Thus, to explicate the content of the claim will be (at least in part) tothrow light on the character of this experience so as to make it clear whyit is rightly described in terms drawn from the emotions.

Compare this with the philosophical work on pictorial representation.An approach here, perhaps popular enough to be regarded as the orthodoxapproach, is to say what it is for an object to be a picture by providing acharacterization of the experience to which it gives rise. This view is mostclosely associated with Richard Wollheim, for whom a picture of an x is anobject that gives rise to a certain experience: that of seeing an x in the picture(provided also the artist intends us to see an x in the picture). A great deal ofeffort has been expended on clarifying the nature of the experience of this‘seeing in’. Whatever the virtues of this approach in the case of pictorialrepresentation, it is not obviously correct. That is, it is not obvious why theanswer to the question ‘What makes it the case that this object is a picture ofan x?’ is ‘Because of the nature of the experience to which it gives rise’. Afterall, unless one is a sort of idealist, one would not think that form of answerappropriate in attempting to define ‘car’ or ‘bed’. However, the train of rea-soning given above seems to provide a good reason for this form of answerin the case of expression. That is, what one appears to be claiming when oneclaims that a piece of music is sad is that the experience of hearing it has a cer-tain phenomenological characteristic. If the foregoing is correct, the prob-lem is to throw light on the character of the experience of hearing the music.

I said above that any solution should also make it clear why the musicis rightly described in terms drawn from the emotions. This, however,may have been too hasty. Inspired by the thought that the experience ofa passage of sad music is not the experience of the natural expression ofsadness, one might think that the content of the experience is a purelymusical property—one of a certain limited set of musical constructions,perhaps—or a sui generis aesthetic property. This approach has beencriticized by Roger Scruton:

expression in music 97

Consider the application of an emotion term—such as ‘sad’—to a work of art(or, for that matter, to an event, or a letter, or anything that cannot literally be inthe emotional state of sadness). To understand the word ‘sad’ is to know how toapply it to people in order to describe their emotional state. The criteria for theapplication of the term ‘sad’ concern the gestures, expressions and utterances ofpeople on the basis of which I describe them as sad, and to grasp the concept ofsadness is to know how to apply it on the basis of these criteria. When we applythe concept to art, however, it is arguable that these criteria are not, or need notbe, present. Does this mean that the term ‘sad’ is ambiguous?

(1982: 38)

Scruton is right to think that the entailment that the term is ambiguous isa reductio of the view. It is worth noting a small complexity here. Scrutonclaims that difference in criteria of application, rather than difference inproperty being referred to, is sufficient to ensure ambiguity. Thus it isnot clear that his point can be used, as I have attempted to use it, tocriticize the notion that the property being referred to is distinct from thatto which ‘sad’ would refer in the central case. However, we can state thepoint more generally. The content of the claim ‘the music is sad’ shouldoverlap in the central and the aesthetic cases. Realists and non-realiststypically differ as to what they take the content to be: the former take itto be attributing a certain property, the latter to be expressing a certainexperience. However, unless the realist can show that the property is notdistinct from the property that would be referred to in the central case, andunless the non-realist can show that the experience expressed is not distinctfrom an experience that would be had in the central case, the danger ofambiguity will loom.

A second thought might be to characterize the experience of expressionby looking at its causes in the music (this is arguably the focus of Kivy(1989)). We might try to justify this approach by noting that, in the centralcase, we frequently account for an emotion by citing its cause: the womanis afraid because she has seen the tiger running towards her. Clearly, thereis an interesting question as to what the properties of music are that causea listener to hear it as expressive. Two further matters, however, are notso clear: first, the question of whether this is a concern for philosophy andsecond, whether, even if we did have an answer to our question, the natureof the experience would be illuminated. I think the answer to both of thesequestions is ‘no’.

98 derek matravers

The herb Puck puts on Titania’s eyes causes her love of Bottom. Theidentity of the cause would be a matter of empirical investigation, towhich Philosophy would have nothing to contribute. Furthermore, theidentification of the cause would not be an account of Titania’s feelings,nor explain what is meant by ‘What angel wakes me from my flow’rybed?’ If this were right, however, it would appear to make mysterious theabove case, in which citing the cause of the emotion (the charging tiger)did seem to throw light on the nature of the experience. The appearance isdeceptive, however, as the two cases are relevantly different. In the case ofthe tiger, the cause of the emotion is closely related to its object. The tigeris the cause of the woman’s fear, and her fear is of the tiger. Citing theobject of the emotion—the state of affairs towards which the emotion isdirected (the charging tiger)—does something to account for the subjects’feelings, and does something to explain what is meant by ‘I am scared ofthe tiger’. Citing the cause looks to be illuminating because we assumethe cause will lead us to the object, and thus are able to grasp the object.Indeed, we are usually right to do so because the cause of an emotionusually is a belief (the tiger is charging) the content of which refers to theobject of the emotion (the charging tiger).

It is uncontroversial that some properties of music (including propertiessuch as the context in which it is usually heard) cause the experience ofexpression in a listener, which results in the listener uttering ‘The musicis sad.’ The task we have set ourselves is to throw light on the nature ofthe experience. The above discussion suggests finding the properties of themusic that cause the experience will aid us in our task only if the cause leadsus to the object. Geoffrey Madell has argued just that: dissonant propertiesof music arouse a desire for the resolution of dissonance, and a frustrationif that is not achieved or a satisfaction if it is (2002: 28). However, thefrustration at the failure to resolve, or satisfaction at the resolution of,dissonance is not yet the experience of expression (although it might be thecause of, or part of, such an experience). Madell seems to face the dilemmaof either having an emotion with a bona fide object that does not amountto expression, or a case of expression but with the wrong object (2002:128). Putting Madell’s view to one side, I cannot see that we have anyreason to think that the cause of our experience of expression should also beits object. The object of the experience of expression would, presumably,have to fall within a reasonably narrow range fixed a priori. By contrast,


there is no reason to suppose that there are a priori limits on what propertiesof music can cause the experience of expression, nor that some of thoseproperties will not turn out to be quite unexpected, even bizarre.

The right approach, then, is to characterize the experience directly.It might be thought that this is a task unsuited to philosophy. That is,the tools of philosophy, or at least, analytic philosophy, are not suited tophenomenological description. Furthermore, the task seems pointless. If Iintrospect, on hearing a piece of sad music, and report what it is like tohave that experience, who can gainsay what I report? Matters are not quitethat bad, however. In the first place, the task is not simply to describe whatI experience, but to come up with a description with which everyonewho has the experience can agree. This is not always an easy task, as wecan see if we consider the gustatory case. It sometimes takes an expert toidentify what we taste in a bottle of wine. The second constraint is thatthe elements that make up the description fit into a naturalistic account ofthe experience being caused by the music. Third, which is a special caseof the second, the elements themselves must not be philosophically dubious.The second constraint is a qualification of the above claim that the searchfor causes is irrelevant. If a description of the experience fits the knownfacts about the relevant musical properties, then so much the better for it.This is compatible with the view that philosophical theories do not owe usany account of the causation itself.

It is optimistic to think that an introspective awareness of the phe-nomenon is going to yield a unique correct description. This means thatin adjudicating between competing descriptions, much weight is going tofall on the second criterion. This does pose a problem. Let us say that anaccount goes something like this. The elements of the music that enter intothe experience are a, b and c, and the description of the experience putsthose elements (and perhaps some others) into a particular configuration.Let us say that we have two competing accounts—one puts the elementsinto configuration A and the other into configuration B (they need notagree on the list of elements—different philosophers might have differentviews as to what is plausible). How are we to know which is correct? Thereis no independent check whether it is configuration A or configurationB that we are experiencing: both accounts would claim that what it isto experience configuration A (or B) is simply to have the experience ofexpression.

100 derek matravers

This is, I think, a problem and should dampen our optimism aboutthe robustness of philosophical accounts of expression. However, anysuch account is no worse off in this respect than at least some accountsof representation. Consider, for example, the competing accounts ofrepresentation proposed by Kendall Walton, Richard Wollheim and RobertHopkins (Walton 1992; Wollheim 1987; Hopkins 1998). In none of thesecases is there any independent check on whether the analysis given of theexperience of representation is as they say it is: their claim is, rather, thatto experience the elements of their analysis in the configuration claimed issimply to have an experience of representation. The burden, in decidingbetween them, rests on the dubiousness (or not) of the proposed elementsand the plausibility (or not) of the purported manner in which they arecombined.

There are a number of attempted descriptions of the experience ofexpression in the literature. I shall divide them into two sorts. The first sortis characterized by taking us to be imagining the music, or imaginativelyconceptualizing the music, in such as way as to provide us with a reason toapply an emotion term to it. I shall examine three versions of this: those ofPeter Kivy, Stephen Davies and Jerrold Levinson. The second sort describesour experience in a way that stops short of imaginatively characterizing themusic. I shall examine two versions of this: the standard dispositional (or‘arousal’ theory) and that of Kendall Walton. There is much of merit in theliterature that I shall not discuss, including Aaron Ridley’s account, whichis a hybrid of the two sorts, and Malcolm Budd’s account, which is similarto that of Walton (Ridley 1995; Budd 1989). In addition, I shall not discussRoger Scruton’s important contribution (Scruton 1997). Finally, I woulddraw the reader’s attention to Budd’s Music and the Emotions, a work thatcleared the ground of previous theories and did much to generate recentinterest in this topic (Budd 1985).

Peter Kivy’s The Corded Shell (later published with additional essays asSound Sentiment) was an important early contribution to the recent debate(Kivy 1989). Perhaps because it was discovering (or rediscovering) theground of the debate, it elides distinctions prominent in later literature.As already intimated, it does not clearly distinguish the search for theproperties that cause the expressive experience from the character of thatexperience. Furthermore, even when focused on the latter, a number ofcharacterizations are suggested. The most prominent is that we ‘animate’ the


music, and experience it as animated (1989: 59). The notion of animation isnot, however, any more immediately perspicuous than expression. Kivy’sattempt to elucidate the notion by providing examples is not, I think,successful. His principal example is a circle in which three lines are drawnthat we will ‘inevitably’ see as a face. However, that is a simple case ofpictorial representation, and it is unclear what is supposed to generalize tothe case of expression. A further claim that he makes is that animation isthe same phenomenon as ‘seeing as’ and ‘seeing in’ (1989: 172). As thecomparison with aspect perception appears in several other places in thediscussion of expression, it is worth a brief digression.

The phenomenon of aspect perception was introduced into contem-porary philosophy by Wittgenstein (Wittgenstein 1976: 193–216). Hediscusses various cases in which the object that we see remains unchanged,although our experience of it changes. The most well-known case is thatof the ambiguous figure, ‘the duck/rabbit’, in which the same set of linescan be seen as a duck-picture and as a rabbit-picture. It is unclear, withoutfurther explanation, what in this phenomenon is supposed to be analogousto expression. It cannot be the relation between the figure and what it isseen as: in both cases, that is the relation of pictorial representation, and all(including Kivy) agree that expression is not an instance of that. It mustbe, rather, that the way in which we process the figure (either as a duck-picture or a rabbit-picture) has a phenomenological upshot. Analogously,the way in which we process the music has a phenomenological upshot:we hear it as expressive. However, we hardly needed the analogy withaspect perception to tell us that.

We might try to provide a fuller or more illuminating descriptionof the experience of expression than its being ‘animated’. Elements inKivy’s work suggest that he might conceive of expression as a matter of‘experienced resemblance’ (Kivy 1989: 53, 142). This is a notion familiarfrom Hopkins’ account of pictorial representation. Consider, for example,hearing a foghorn. One might experience this sound as resembling thatof a whale. Hearing the music as sad might be a matter of experiencinga resemblance between the music and a person expressing sadness. Thistheory has at least three things to recommend it. First, it is the right sort ofexperience to make sense of there being a connection between music andthe emotions, and to justify the utterance ‘the music is sad’. Second, it fitswith a plausible account of how expression is caused. Several philosophers

102 derek matravers

have claimed that what one could broadly call the ‘movement’ propertiesof music are isomorphic to the movement properties of natural expression.Putting the matter at its crudest, sad people tend to move slowly and thisslowness is mirrored in some music. If there is such an isomorphism, thatis, if certain music does resemble the natural expression of sadness, thenit is a plausible that, in such cases, resemblances cause the experience ofresemblance. Although the theory fits with a plausible causal story, it isnot conceptually tied to it. However, the antecedent—the claim that sadmusic is music that resembles sad people—is difficult to defend. Such aresemblance is not necessary—the major chord differs in expressive qualityfrom the minor chord, yet both resemble the natural expression of emotionas much as, or as little as, the other (this claim is discussed further below).Neither is resemblance sufficient—there is plenty of low, slow music that isstately rather than sad. Indeed, if there is a resemblance between the natural,non-audible, expression of emotion and music it is trivial. This does notundermine the claim that the experience of expression is the experienceof resemblance. Some music is experienced as resembling the naturalexpression of emotion (through whatever cause), and it is this music that isexpressive. If we find that such music also exhibits some isomorphism tothe natural experience of emotion that can be characterized independentlyof experience, that would be no more than a step along the way to part ofa causal account.

Experienced resemblance is not, however, a good account of the experi-ence of expressive music. The claim that a resemblance can be experiencedacross sense modalities makes doubtful sense. I can understand experiencinga sound (the foghorn) as resembling another sound (that made by a whale),but it is difficult to see how this generalizes to cross-modal experiencedresemblance: experiencing a sound (the music) as resembling a sight (aperson moving slowly). However, if we stay with a single modality expe-rienced resemblance will fail our first test: it is not a description of theexperience of expressive music on which competent listeners will agree.Sad music is not experienced as resembling the sounds made by sad people(except in certain special cases).

Stephen Davies has provided an account that attempts to overcome theseproblems (Davies 1994: 228–40).


Our experience of musical works and, in particular, of motion in music is likeour experience of the kinds of behaviour which, in human beings, gives rise toemotion characteristics in appearances. The analogy resides in the manner in whichthese things are experienced rather than being based on some inference attemptingto establish a symbolic relation between particular parts of the music and particularbits of human behaviour.

(1994: 239)

An ‘emotion characteristics in appearance’ is a type: a perceivable propertyof a person that is the criteria by which we would judge that person to befeeling an emotion. We can use emotion terms to refer to such appearances:for example, we might say that the St Bernard dogs have ‘sad faces’, wheresuch a comment is free from implications concerning the dogs’ psychology.Davies’s claim is that we experience expressive music in a way that is likeour other experiences of emotion characteristics in appearance.

I am not sure the introduction of ‘emotion characteristics in appearance’does solve the problem identified above. That is, I am not sure whetherthere is a type of appearance, tokens of which are exhibited in the face ofthe person expressing sadness, the face of the St Bernard, and in a piece ofsad music. That the sad person and the St Bernard present the same typeappearance is made more plausible by the fact that in each case the token issomething visible. Can, however, the same type appearance be presentedas something experienced through a different sense modality altogether(Levinson 1996: 105)? Davies’s analogue of the move made above (thatthe ‘emotion characteristic in appearance’ presented by music is a token ofan audible type—the sound characteristically expressed by a person in thethroes of an emotion) is rejected as not being true to the phenomenology.Instead, Davies affirms that

the expressiveness of music depends mainly on a resemblance we perceive betweenthe dynamic character of the music and human movement, gait, bearing andcarriage.

(Davies 1994: 229)

However, the appearance exhibited in human movement involves changeof location, whilst the appearance exhibited in the relevant sense of musicalmovement does not, which makes it difficult to see how they can each

104 derek matravers

exhibit the same type appearance. Davies’s solution to this is to point outthat ‘many widespread, mundane descriptions of non-musical phenomena’are described in terms that apply to movement yet do not involve anychange of location: time flows, stock exchange indices plunge, governmentslurch to the right (Davies 1994: 235–6). We can readily concede theappropriateness of such characterizations; however, this falls short of whatDavies requires. For the worry was not with whether emotion terms couldbe applied appropriately to ‘objects’ detected by different sense modalities,but whether there was a type of emotion characteristic in appearance thatcrossed the sense modalities. Davies’s examples support only the formerclaim.

A second worry with Davies’s account is whether emotion characteristicsin appearance that are centred on movement properties cover all cases.Sometimes, as in the Dies Ira of Mozart’s Requiem, it seems to do withtimbre. Davies could concede that, at times, we experience the music asbeing of a type with vocally expressive utterance. The principle remainsthe same: the attribution is made on the grounds of experiencing the musicas exhibiting an emotion characteristic in appearance; it needs to remain‘emotion characteristic’ even if the type of appearance might change.However, to return to a point brought up above, what of the minor andthe major chords? They seem to exhibit the same appearance, although theyhave different expressive properties. In a private communication, Davieshas argued that combinations of notes (A-C-E) are heard within a tonalcontext. That combination of notes is a minor chord, the tonic of A minor,but can also feature in a sequence as part of a piece of music in C major.When it does so, it might mark a point of relative tension, but not sadness.Hence, in isolation, it is simply a combination of notes. To hear it in theminor key is to contextualize it in a certain way, and thus make it part ofa (virtual) emotion characteristic in appearance. However, it is not clearwhy this would help. The two chord sequences (one in A minor and theother in C major) will themselves only exhibit an emotion characteristicin appearance (with respect to movement) as much (or as little) as eachother and yet they will afford different expressive experiences; it is still thedifference between major and minor (except this time a difference in key)that is bringing the experience about. The problem has been moved ratherthan solved.


Jerrold Levinson has argued for a richer description of the experience ofemotion:

A passage of music P is expressive of an emotion or other psychic condition E iffP, in context, is readily and aptly heard by an appropriately backgrounded listeneras the expression of E, in a sui generic, ‘musical’ manner, by an indefinite agent,the music’s persona.

(Levinson 1996: 107)

Imagining a mechanism for expression (such as playing the piano) is notpart of the experience: rather, we simply hear the music as the expressionof an emotion by a persona through musical output. This relieves Levinsonof any obligation to try to tie, through similarity claims, the nature of ourexperience of the music to the nature of our experience of the expressionof emotion in the central cases. There are two ways in which we couldinterpret Levinson’s claim (Levinson seems to hover between the two(1996: 118)). The first would take Levinson to be attempting to throw lighton the experience by providing us with a perspicuous description of it. Thatis, not all descriptions of the same event are equally enlightening. The sameevent can be a moving of my arm, a flicking the switch, an illuminating theroom and a scaring of the burglar. Thus, Levinson is proposing that insteadof describing our experience as hearing music or even hearing expressivemusic, we describe it as the hearing of the music as the unmediatedexpression of an emotion by a persona. The second reading is that what itis to hear music as expressive is to make the music the object of some kindof imaginative endeavour: we ‘hear the music with imagination’—that is,in however attenuated way, we intentionally imagine the music to be theunmediated expression of emotion by a persona. Each of these readingscan be seen to face the same root problem. In the first case, the problemis whether the re-description is enlightening. In the case of the movedarm, some of the descriptions are enlightening because (roughly) they referto the intention behind the action, or fit the action into a context. If welook at an example that is closer to expression, the account of pictorialrepresentation in terms of ‘seeing in’ is only enlightening to the extentthat we have an understanding of seeing something in something else thatis broader than, and prior to, its application to the pictorial case. In thesecond case, the problem is whether my grasp of the project of imagining

106 derek matravers

the music to be an unmediated expression of emotion by a persona isindependent of my grasp of expressive music. If it is not, the claim thatI engage in the project of imagining the music to be the expression ofemotion by a persona would add nothing to my understanding of theexperience of expressive music. This is related to the problem expressedabove. If we ask what the experience of music as the direct, unmediated,expression of emotion by a persona would be like, or how we wouldknow if we had succeeded in our imaginative project, Levinson would—Iassume—say that the experience is like the experience of expressive musicand we would know if we had succeeded were we to have that experience.We have no check as to whether we have succeeded in our imaginativeproject independent of our hearing the music as expressive.

Is my grasp of what it would be for an experience of instrumental musicto be the unmediated expression of emotion by a persona independent ofmy grasp of the experience of expressive music? Levinson argues that it is:

[T]he extension of ‘sounding like’ beyond the range of actual human behaviouralexpression, in the form of music sounding like, or as if it were, an alternate, specif-ically ‘musical’ mode of expression of emotion, is both natural and imaginativelyunproblematic, suitably understood.

(1996: 116)

This seems to me unduly optimistic. If the account was that expressivemusic was music that was heard as if it were within the range of actualhuman behavioural expression of emotion, it would be enlightening wereit true—although obviously it is not true. The problem of expression surelyis just that we do not hear expressive music as falling within the range ofactual human behavioural expression of emotion.

This is a version of an objection considered by Levinson, owed to HubertEiholzer. To this objection Levinson argues for two possible responses. Thefirst is to ‘maintain that listeners need not know, in any detailed fashion,what music’s being a natural mode of expression would amount to but onlybe disposed to posit, in imagination, that music was such a mode’ (1996:120). The second is that we drop the claim that the music is the form ofexpressing, and instead maintain that the experience is of expressing toutcourt on which, if we reflect upon it, we would decide that it constituteda sui generis mode of expressing (1996: 121). Both responses retreat fromthe claim that music is experienced as a sui generis mode of expression.


However, as it is the experience we are trying to characterize, it is unclearthat Levinson’s account can survive intact (to be fair, Levinson allows thatboth responses are a concession). The first seems to involve a dilemma.Either our positing, in the imagination, that music is a natural mode ofexpression results in our experiencing it as a natural mode of expression orit does not. If it does, then Eiholzer is not answered: we still need somegrasp of what it is to hear music as a natural mode of expression. If itdoes not, then Levinson will have failed to capture what it is to experiencemusic as expressive, for it is not to experience music as not expressive, andto imagine, of it, that it is the natural mode of expression. The secondresponse aims to remove the music as the vehicle of the persona’s emotionfrom the experience, and place it (if it is to appear at all) in the listener’slater reflection. All that remains of the attempted elucidation of the contentof ‘the music is sad’ is that it is ‘readily and aptly heard by an appropriatelybackgrounded listener as the expression of sadness’.

Levinson might object that this neglects the claim that further reflectionwill drive the listener to the conclusion that the mode of expression is thesui generis mode of expression by a persona. However, this raises an issuethat threatens to undermine the entire account. What are the grounds ofLevinson’s claims? How does he know that further reflection will take thisform? More generally, how does he know an appropriately backgroundedlistener will hear the music in the way he describes? He might arguethat anyone who is familiar with the experience of expressive music willaccept the description. This again seems unduly optimistic: according toStephen Davies at least, ‘it seems straightforwardly false that there are publicconventions or consensus regarding these matters’ (Davies 1997: 101). Thisdoes seem to be a problem: Levinson is giving a naturalistic account of theclaim that music is sad, in terms of how an appropriately backgroundedlistener would hear the music. If, however, the backgrounded listenerclaims not to hear the music like that, it is unclear what the resources areto which Levinson could appeal in order to convince him.

The theories I have considered so far all claim that we experience themusic as, in some way, appropriate for description in emotion terms. Itis experienced as resembling the expression of emotion, or as manifestingthe appearance of the expression of emotion, or as emotion expressedin a musical manner. There is, however, another way of answering thisquestion: that sad music is music that makes us sad. Stated in this bald

108 derek matravers

form, the (so-called) ‘arousal theory’ has been the subject of much scorn.In retrospect, however, we can see that it got two points roughly right.First, let us consider a slightly sophisticated version: a passage of musicexpresses an emotion if, among the mental states caused by the music, issome non-cognitive state (a feeling) which stands in the right relation tothe appropriate reaction to the expression of that emotion in the centralcase (Matravers 1998: 146). To give an example, music expresses sadness if,among the mental states it arouses is something like the feeling componentof the normal reaction to the expression of sadness in the central case. Thefirst point it got roughly right is that it never concerned itself with thecausation behind this. That is, it had no commitments as to the natureof the properties that caused the experience of expression. The secondpoint it got roughly right was to focus on the nature of the experienceof expression. Both of these, I have argued, are the right stance for aphilosophical theory to take.

The question, then, is whether the account of the experience is correct.On the face of it, it appears foolish to equate the experience of music assad with a conjunction of the experience of music and an experience ofsadness. I will return to that after considering two more obvious problems.An aroused feeling is not necessary in that some people appear to experienceexpression whilst claiming not to be experiencing the relevant feeling. It isnot sufficient in that plenty of things arouse our feelings, without therebybeing expressive. I cannot fully enter into the matter here, but a defenceof the arousal theory would run along the following lines. First, the claimthat one was not experiencing the relevant feeling may be compatible withexperiencing some kind of appropriate non-cognitive state. That is, thelistener might be disavowing the mental state usually referred to by theterm ‘feeling’, or terms such as ‘sad’ or ‘joyous’. Those terms are usuallyused in a context in which the mental state is dominant: one does notusually claim to have those mental states unless they are strongly presentin experience. The arousal theorist does not require such strong presence,only the presence, among the mental states caused by the music, of somenon-cognitive state that stands in the right relation to the appropriatemental state in the central case. There is a danger here of reducing thefeeling to a cipher, in which case it loses any explanatory weight. A balanceneeds to be struck between having little enough of the non-cognitive state


to make sense of the listener claiming that he is not experiencing a feeling,but enough of it to play the required role in the account.

To rebut the claim that the aroused non-cognitive state is not sufficient,the arousal theorist needs to pick out a psychological role filled only by thosearoused feelings relevant to expression. In short, the listener experiencesthe feeling as ‘tracking’ the music; it varies as the music varies. This is alsoa more plausible account of the experience of expression than the simpleconjunction proposed above. I say ‘more plausible’, although for most itwould not be plausible enough, for the analysis remains in terms of (at least)two separate mental states: the music and the aroused non-cognitive state.The question is how this gap is to be closed.

It is not clear whether this is a problem and, if it is a problem, it is notclear whether it is peculiar to the arousal theory. The analysis provided is interms of two separate mental states (the music and the non-cognitive state)which we do not experience as separate. The relation between the termsof a philosophical analysis of an experience and its phenomenology is not,however, obvious. The two separate mental states are co-instantiated inconsciousness in the particularly intimate causal relation of the one trackingthe other. Is it absolutely clear that this could not be an account of thephenomenology—especially given the fact that so many listeners attest tothere being some effect on their feelings and emotions by expressive music?In other words, these are elements we have anyway—it is obvious that thisis not the way in which they are arranged.

Kendall Walton has proposed a more nuanced account that shares withthe arousal theory the thought that the solution to the problem lies notin imagining the music to be a certain way, but in the way in which ourresponse to the music is characterized:

I propose that, although music does not in general call for imaginative hearing orimaginative perceiving, it often does call for imaginative introspecting. We men-tioned the possibility that music is expressive by virtue of imitating behaviouralexpressions of feeling. Sometimes this is so, and sometimes a passage imitates orportrays vocal expressions of feelings. When it does, listeners probably imagine (notnecessarily consciously and certainly not deliberately) themselves hearing some-one’s vocal expressions. But in other cases they may instead imagine themselvesintrospecting, being aware of, their own feelings.

(Walton 1988: 359)

110 derek matravers

In a later paper, he elucidates this a little:

Anguished or agitated or exuberant music not only induces one to imagine feelinganguished or agitated or exuberant, it also induces one to imagine of one’s auditoryexperience that it is an experience of anguish or agitation or exuberance.

(Walton 1994: 55)

Let us grant that claim that some music is expressive by virtue of imitatingbehavioural expressions of feeling and focus on the claim that what it isto experience expression is for the hearer to imagine of his awareness ofauditory sensations, that it is an introspective awareness of states of mind.I say this is in the arousal tradition, because, first, there is no commitmentto any causal account of how this state comes about, and second, becausewe are not imagining of the music that it be appropriate for description inemotion terms.

This account faces a familiar difficulty. Is our grasp of what it would beto imagine of one’s auditory experience that it is an experience of anguishor agitation or exuberance independent of our grasp of the experience ofexpressive music? Is there any independent check that it is this, rather thanany other imaginative project, that we are pursuing? One problem withWalton’s account (a general problem with a number of Walton’s accounts,and one of which he is aware) is the absence of a robust account of the imag-ination. In answer to the question of whether this is the way an appropriatelybackgrounded listener would hear the music, I would be inclined, likeStephen Davies, to scepticism: there is simply no consensus on these matters.

For Jerrold Levinson, the main problem for Walton’s account (whichlooks as if it will generalize to being a problem for the arousal theory) is thatit ‘casts the activity of perceiving musical expressiveness in too egocentrica light; it represents expression in music as in effect the expression of thelistener’s own, albeit imaginary, feelings’. By contrast, Levinson maintains,‘expressiveness in music ... is something we encounter fundamentally asresiding ‘out there,’ as existing anterior to our own minds’ (Levinson 1996:94). The argument, however, is double-edged as we can hear music asbeing either an expression of emotions from the outside, or as a vehicle forour own emotion. The distinction was marked by R. K. Elliott concerninglyric poetry, in a way that generalizes to music:

If a work is experienced as expression, experiencing it from within involvesexperiencing this expression after a certain imaginative manner as one’s own.


Experiencing it from without is experiencing it as expression, but not experiencingthis expression as if it were one’s own.

(1967: 146)

Walton might claim that hearing music from without is already accountedfor in his theory by his admission that sometimes ‘listeners imagine (notnecessarily consciously and certainly not deliberately) themselves hearingsomeone’s vocal expressions’, reserving the account discussed here forhearing from within. What of Levinson’s own account? As the experienceis one of hearing the music as the expression of a persona, can it accountfor hearing music from within?

Levinson’s definition of expression states a necessary as well as a sufficientcondition. Thus, it seems as if he is committed to expression being theexperience of an emotion being expressed by someone else (the musicalpersona). This, however, seems contradicted by claims Levinson has madein an earlier paper (‘Music and Negative Emotion’):

If one begins to regard music as the expression of one’s own current emotionalstate, it will begin to seem as if it issues from oneself, as if it pours forth from one’sinnermost being.

(1982: 328)

Is the claim that our experience can be of the music seeming to ‘issuefrom oneself ’ compatible with the claim that, necessarily, to hear musicas expressive is to hear it as being expressed by the musical persona? Theanswer to this depends on the relation between the listener and the musicalpersona. In explicating what it is to hear music from within, Levinsoncould construe the relation in at least two ways. The first would be thatthe listener experiences the music as described in the later essay, but thenempathizes with the persona and takes on, at least in imagination, thepersona’s emotion (1982: 327). That is, the relation is one of ‘identifyingwith’ the persona in the way in which one might, for instance, identifywith a friend who is experiencing some emotion. The second would bethat the relation between listener and persona is one strict identity: weexperience the music as a sui generis expression of our own emotion andthere is no further figure involved.

Which of these two options should we favour? The issue depends onwhether hearing from within and hearing from without are on a par, or

112 derek matravers

whether we take hearing from without to be primary. Levinson wouldappear to favour the latter, in which case he would take the first option.If so, the claims in the two essays are compatible. If we take hearingfrom within and hearing from without to be on a par, and thus take thesecond option, the claims in his two essays would not be compatible. If thiswere the case, Levinson would need to weaken the definition in ‘MusicalExpressiveness’ to the claim that it captured only one (albeit important andcentral) experience of expression.

Empathy and sympathy look to be a resource to help the arousal theoristmake the distinctions necessary here. Recall the version of the arousaltheory considered above: a passage of music expresses an emotion if,among the mental states caused by the music, is some non-cognitive state (afeeling) which stands in the right relation to the appropriate reaction to theexpression of that emotion in the central case. The appropriate reaction tothe expression of emotion can take two forms. Consider a situation in whichyou are faced with someone expressing sadness. One reaction would beempathize—that is, to adopt the type of emotion being expressed—that is,to feel sadness. Another would be to sympathize—that is, to feel an emotionthat responds to the emotion being expressed. One virtue of the arousaltheory is its simplicity: amongst the mental states aroused by the music issome non-cognitive state. For this to help the arousal theorist, he would firsthave to make sense of the notion that the non-cognitive state aroused couldbe either empathetic or sympathetic, and then show how this distinctioncould solve the problem of experiencing expression from within or without.

The first looks to be a problem because the difference between empathyand sympathy is not in the quality of the phenomenological state, but inthe context. A necessary condition for empathy, for example, is that theobserver experiences the same emotion as the expresser, which is difficultto accommodate in the musical case. Can we be sure, however, that thephenomenology of empathy does not differ from that of sympathy? Doesthe sadness we feel towards someone feel the same as the sadness we feelwith someone? It is not clear that this is the best way to put the question oreven, put like this, where we might look for an answer. Much work in thetheory of the emotions recently has been to explore the complexities of theintentionality of feelings. Animals and pre-linguistic infants can certainlydirect their attention on things outside them, yet we do not attribute tothem the kind of cognitive apparatus required by traditional accounts of


intentionality (Deigh 1994). It might be that there is a type of expressivemusic that arouses a feeling that feels as if it is directed on objects, and atype of expressive music that arouses a feeling that feels like it is our ownfeeling (although brought about in virtue of the expression of others).

The second issue is more difficult to deal with, if only because of thecomplexity of the phenomenon to be explained. The arousal theory beingexplored relies on there being a ‘right relation’ between the non-cognitivestate aroused, and the feeling state involved with the response to expressionin the central case. However, there is nothing internal to the arousal theoryto restrict the right relation to central-case emotions felt to other people(it is only that hearing from without is the usual case). There might becases in which the right relation is to central case emotions felt ourselves.The point, put at its most general, and covering both the issues raised, isthat it is not clear that the feeling components of emotions are simple andhomogenous. What sadness feels like might differ depending on whetherit is our sadness per se, an empathetic sadness or a sympathetic sadness.Whatever complexity that can be found can be used by the arousal theoristto add complexity to the response to the music. This is, of necessity giventhe shakiness of a science of feelings, speculative. However it has someintuitive force and is the obvious area for any dispositional theory to explore.

I turn now to a problem for all of these accounts, and that is normativity.Each of these could be an account of what it is for a listener to experiencemusic as expressive. However, what they aspire to be is what it is for musicto be expressive. A passage of music is expressive if it is appropriate that itbring about a certain experience. The most plausible way of meeting thisaspiration seems to be to use some version of that empiricist standby, thequalified observer in the right perceptual circumstances. However, evenhere there are grades of difficulty. For example, one could give up claimsto truth or falsity, and instead rely on the fact that some music is rewardingwhen heard in one way rather than another. Unless one is secure in one’sown judgement about which music is to be heard how, one will be opento guidance from ‘expert’ listeners. Terms such as ‘the appropriate way tolisten to this’ will have a natural role in that discourse. The stonier road istrod by the realist, who needs to provide determinate truth conditions tounderlie the claim that sad music is music that possesses the (real) expressiveproperty of sadness. For the realist, which would seem to include Levinson(2001), the judgement of the ‘appropriately backgrounded listener’ has a

114 derek matravers

constitutive role. The objection (here made to Levinson in particular) hasbeen forcefully put by Roger Scruton:

The invocation of a ‘reference class’ of listeners opens the way to a radicalscepticism: how is this class to be defined and by whom? The natural gloss onLevinson’s definition is to identify the reference class as the class of those who areable to discern the expressive content of a work of music. But this would be toreduce the definition to vacuousness.

(Scruton 1997: 353)

Apart from reducing the commitment to realism, and accepting the conse-quent restrictions on the aspirations of criticism, I can see no way aroundthis problem. As it is hardly a problem for expression in particular, and as itdoes nothing to help us weigh one account against another (as opposed tothe further commitments of advocates of the accounts) I shall say no moreabout it. What will help in weighing the accounts against each other is theirplausibility in capturing the phenomenon, whether we have a grasp of theelements of the accounts independently of expression, and whether theseelements can be fitted together in a perspicuous way. It is not obvious to methat the dispositional account is in a worse position than that of its rivals.¹

References

Budd, M. (1985) Music and the Emotions, London: Routledge and KeganPaul.

Budd, M. (1989) ‘Music and the Communication of Emotion’, Journal ofAesthetics and Art Criticism 47 2: 129–37.

Davies, S. (1994) Musical Meaning and Expression, Ithaca, NY: CornellUniversity Press.

Davies, S. (1997) ‘Contra the Hypothetical Persona in Music’, in M. Hjortand S. Laver (eds), Emotion and the Arts, Oxford: Oxford UniversityPress, 95–109.

¹ Stephen Davies and Jerrold Levinson were (as usual) generous and helpful in their comments onthis essay. It has also been much improved by suggestions from the editor. Jerrold Levinson addressesthe problems raised in the final section (on normativity) in an exchange with Stephen Davies in Kieran(2005) that came out too late to be dealt with in this paper.


Deigh, J. (1994) ‘Cognitivism in the Theory of Emotions’, Ethics 104 4:824–54.

Elliott, R. K. (1967) ‘Aesthetic Theory and the Experience of Art’, inH. Osborne (ed.), (1972), Aesthetics, Oxford: Oxford University Press.

Hopkins, R. (1998) Picture, Image and Experience, Cambridge: CambridgeUniversity Press.

Kieran, M. (ed.) (2005) Contemporary Debates in Aesthetics and the Philosophyof Art, Oxford: Blackwell.

Kivy, P. (1989) Sound Sentiment: An Essay on the Musical Emotions, Philadel-phia, Penn.: Temple University Press.

Levinson, J. (1982) ‘Music and Negative Emotion’, in Music, Art andMetaphysics: Essays in Philosophical Aesthetics (1990), Ithaca, NY, andLondon: Cornell University Press, 306–35.

Levinson, J. (1996) ‘Musical Expressiveness’, in The Pleasures of Aesthetics,Ithaca, NY: Cornell University Press, 90–125.

Levinson, J. (2001) ‘Aesthetic Properties, Evaluative Force and Differencesof Sensibility’, in E. Brady and J. Levinson (eds), Aesthetic Concepts: EssaysAfter Sibley, Oxford: Oxford University Press, 61–80.

Madell, G. (2002) Philosophy, Music and Emotion, Edinburgh: EdinburghUniversity Press.

Matravers, D. (1998) Art and Emotion, Oxford: Oxford University Press.Ridley, A. (1995) Music, Value and the Passions, Ithaca, NY: Cornell

University Press.Scruton, R. (1982) Art and Imagination: A Study in the Philosophy of Mind,

London: Routledge and Kegan Paul.Scruton, R. (1997) The Aesthetics of Music, Oxford: Oxford University Press.Walton, K. (1988) ‘What is Abstract about the Art of Music?’, Journal of

Aesthetics and Art Criticism 46 3: 351–64.Walton, K. (1992) ‘Seeing-In and Seeing Fictionally’, in J. Hopkins and

A. Savile (eds), Psychoanalysis, Mind and Art: Perspectives on Richard Woll-heim, Oxford: Blackwell, 281–91.

Walton, K. (1994) ‘Listening with Imagination: Is Music Representational?’,Journal of Aesthetics and Art Criticism 52 1: 47–62.

Wittgenstein, L. (1976) Philosophical Investigations, Oxford: Blackwell.Wollheim, R. (1987) Painting as an Art, London: Thames and Hudson.

116 derek matravers

Further Reading

Budd, M. (1985) Music and the Emotions, London: Routledge and KeganPaul.

Davies, S. (2006) ‘Artistic Expression and the Hard Case of Pure Music’,in M. Kieran (ed.), Contemporary Debates in Aesthetics and the Philosophyof Art, Oxford: Blackwell, 179–91.


Levinson, J. (1996) ‘Musical Expressiveness, in The Pleasures of Aesthetics,Ithaca, NY: Cornell University Press, 90–125.

Levinson, J. (2005) ‘Musical Expressiveness as Hearability-as-Expression’,in M. Kieran (ed.), Contemporary Debates in Aesthetics and the Philosophyof Art, Oxford: Blackwell, 191–204.

Scruton, R. (1997) The Aesthetics of Music, Oxford: Oxford University Press.

5

Explaining Musical ExperiencePAUL B OG HOS S IA N

Music and the Emotions

1

I start with the observation that we often respond to a musical performancewith emotion—even if it is just the performance of a piece of absolutemusic, unaccompanied by text, title or program. We can be exhilaratedafter a Rossini overture brought off with subtlety and panache; sombre andmelancholy after Furtwangler’s performance of the slow movement of the‘‘Eroica’’. And so forth. These emotions feel like the real thing to me—oranyway very close to the real thing. When one experiences them, it takestime for them to wear off, and one gets irritated with the companion who,because not similarly moved, wants immediately to start discussing whereto go for dinner. Like many others, I am drawn to the philosophy of musicby a need to understand how such emotional responses are possible. Howcan absolute music move us in the way that it does, and to the extent thatit does?

2

In seeking an answer to this question, we do not just seek any sort ofanswer. For example, we would not be satisfied with a brute physiologicalexplanation along the following lines:

In listening to music, we are exposed to sounds. Sounds are vibrationsin the air. These vibrations cause our ear drum to vibrate, whichin turn causes nerve impulses to travel up the auditory nerve to thebrain. In the brain, these nerve impulses cause certain neurons to fire,leading to the perception of sound. In certain cases, the firing of the

118 paul boghossian

neurons that constitute the perception of sound triggers the firing ofcertain other neurons, leading to the experience of specific emotions.

I take it that, even if an explanation along these lines is actually true, it isnot the sort of explanation we are looking for. Why not? What is missingfrom the brute physiological explanation just outlined?

3

What is missing, I think, is any explanation of the rationality of ouremotional response. For all that the brute explanation cares to claim, musicmight induce emotional states in us in just the way that a drug might:certain chemicals cause us to feel things, and so do certain sounds. End ofstory.

On this view, it is pointless to ask whether it makes sense for us torespond to those sounds in the way that we do, in just the way in whichit is pointless to ask whether it makes sense for us to respond to Prozac orto marijuana in the way that we do. And yet we think that it does makesense for us to be moved by music, that it isn’t just a matter of a chemicalresponse.

In fact, we not only think it’s rational to be moved in this way, weare especially admiring of those who are capable of the right emotionalresponse, and critical of those who aren’t. We take the presence of the rightemotional response to be indicative of understanding. We recommendmusic appreciation classes to those who profess not to see what the fuss isabout.

So the question becomes: how could we explain the rationality of ouremotional response to music?

4

Peter Kivy thinks that he can justify a minimal sense in which it can makesense to be moved by absolute music, and that is the sense in which onecan be moved by the sheer beauty of the music, full of appreciation for theexcellence of its craftsmanship and so forth (Kivy 2001: 92–118).

But that doesn’t come close, in my view, to explaining the diversity offeeling that is aroused by our experience. Think of how differently you feelafter listening to the witty elegance of the Mozart Piano Concerto K. 271

explaining musical experience 119

as opposed to the brooding gloom of the Tchaikovsky Sixth Symphony.They are both examples of craftsmanship and beauty.

5

I think our only hope of vindicating the range of emotional responses thatwe have to absolute music depends on our being able to see absolute musicas telling us things, in much the way in which we see characters in opera orfiction as telling us things. In other words, we need to be able to assimilatethe problem of absolute music to the problem of ‘feeling for fiction’ byfinding meaning in absolute music.

That is not to deny that there is a genuine problem of explaining therationality of our feeling for fiction. But everyone is agreed that there mustbe a solution to that problem.

Kivy says:

The question I am raising is how we are emotionally aroused by what the nineteenthcentury called absolute music ... It is important to remember this because when theresources of language are added to the musical work, the terms of the argumentare radically changed. I have no quarrel, for example, with someone who says thatwhen he attends a performance of La Traviata, he experiences real and intensesorrow over the death of Violetta, ... This is not to say that there is no philosophicalproblem in just how the emotions of sorrow and love can be aroused by the fatesof fictional characters ... But the presence of language, with all its potential forconveying concepts, and the presence of delineated characters, such as Violetta andAlfredo ... provide materials for arousal of garden variety emotions far exceedinganything that can reasonably be postulated in absolute music. And that is whyabsolute music poses a problem far beyond opera, oratorio, song and programmemusic to those who wish to claim that it arouses the garden variety emotions.

(Kivy 2001:101–2)

It makes sense to be moved by opera, Kivy tells us, because characters inopera can tell us sad things. The problem for absolute music, he says, isthat, lacking language, title or text, it can’t tell us anything; a fortiori, itcan’t tell us anything sad, happy or whatever. Hence, it cannot make senseto be moved by absolute music.

My thought is a modus tollens on Kivy’s. Since it so obviously doesmake sense to respond to music with emotion, there must be a sense inwhich music is capable of telling us things. The question is how.

120 paul boghossian

6

The fact that we can rationally respond to music with real emotions, I havebeen saying, is indirect evidence that there must be musical meaning. Notso much, perhaps, representational meaning—propositions that tell us howthings are—though some, myself included, would be prepared to allow alimited role for such meanings; but, rather, expressive meaning, the capacity,principally, to express emotions.

But an inference from the legitimacy of our emotional responses is notthe only sort of evidence that we have for the existence of musical meaning.Let me mention three other sources.

First, there is the sheer phenomenology of listening. Any reasonablyexperienced listener will hear the Tristan Prelude as suffused with theexpression of yearning, whether or not they knew anything about theopera to follow, or even whether they knew that there is an opera thatfollows.

A second important source for our conviction that music possessesaudible expressive properties derives from our evaluative thinking aboutmusic. We value some pieces over others because of what we take tobe their greater expressive power. Derivatively, we value some musiciansover others for their greater ability to unlock that expressive power. (Acriticism that is often made, rightly in my view, of a famous New Yorkopera conductor, is that he routinely sacrifices expressive meaning infavor of lush beauty, leading to performances that may be marvelouslysensuous but are otherwise superficial and unsatisfying. We understand thiscriticism.)

A final source for our conviction in the expressive capacity of musicderives from the role that music plays in opera or film. Clearly, musicenhances the expressive impact of opera. The ceremony of the grail at theend of Parsifal wouldn’t be half as powerful as it is if it were accompaniedby ‘‘Baa Baa Black Sheep’’, would it? But how could music contribute inthis way to the opera’s expressive capacity if it didn’t have an expressivecapacity all its own that it was bringing to the scene? Philosophers routinelyconcede the expressive impact of opera. But part of that expressive poweris provided by the music, something it couldn’t do unless it had someexpressive power all its own to contribute.


Explaining Musical Expression

7

In the final analysis, the real problem for musical meaning is not to justifybelief in its existence, but to explain its possibility. How is it possible formere sound, lacking speaker intention, or any of the other resources whichmake linguistic meaning possible, to express meanings? What properties ofthe sounds could constitute their expressive capacity?

Clearly, we would be looking at the musical properties of the sounds,properties like that of pitch, rhythm, harmony and melody. Sounds are heardas having the expressive properties they have because they are heard as havingcertain musical properties: it is something about the shape of the melodywhich opens the fourth movement of Tchaikovsky’s Sixth Symphony thatmakes us hear it as sombre. So what we are asking is: how could possessingcertain musical properties amount to expressing a state of mind? RogerScruton (1999) has argued that, in a sense, there is nothing to explain. Hisargument is distinctive and it will be worthwhile to linger over it.

Scruton’s starting point is the observation that mere sound is not theintentional object of musical perception. Even when sound is understoodnot merely as a physical phenomenon but as a ‘‘secondary object’’—amental object—which exists only when it is heard, it is not what we hearwhen we hear music. The intentional object of musical perception is rathertone, where tone is characterized by such musical variables as pitch, rhythm,harmony and melody. But nothing can literally have such properties andso nothing can literally be a tone. A melody must rise or fall. But thereis nothing either out there or in here that rises or falls in the way that amelody does. As applied to sound, therefore, he concludes, the concepts ofpitch, rhythm, harmony and melody are metaphorical. To describe music,says Scruton:

we must have recourse to metaphor, not because music resides in an analogy withother things, but because the metaphor describes exactly what we hear, when wehear sounds as music.

(Scruton 1999: 96)

And the use of any metaphor cannot ultimately be explained.

122 paul boghossian

It does not seem strained to suggest that Smetana’s music expresses the shining andsilken qualities that we hear in it. Smetana’s music is not literally shining or silken.But its expressive power is revealed in its ability to compel these metaphors fromus, and to persuade us that they fit exactly. Of course, it is a mystery that they fit.But the mystery is immovable. Every metaphor both demands an explanation andalso refuses it, since an explanation would change it from a metaphor to a literaltruth, and thereby destroy its meaning.

(Scruton 1999: 141)

8

Let me first get rid of the distracting claim, that Scruton ultimatelytakes back, that sounds are not the intentional objects of musical expe-rience. Surely, we hear sounds when we hear music! In an afterthoughtto his chapter on ‘‘Tone,’’ Scruton shows some awareness of this. Hesays:

Finally, we should not think of sounds and tones as distinct individuals—as thoughtones really existed apart from sounds. Perhaps the best way of understanding therelation between the two is in the way Spinoza understood the relation betweenmind and body. For Spinoza reality can be conceptualized in two ways: as mentalor as physical. But that which we conceptualize in these two ways is one.

(Scruton 1999: 79)

Invoking Spinoza’s obscure account of the relation between mind andbody is not likely to help here. If sounds and tones are the same individuals,then sounds are equally the intentional objects of musical experience. Atany rate, it is intuitively overwhelmingly clear that sounds are heard whenmusic is heard. (Imagine Dick Cheney dressed up as Santa Claus to entertain(scare?) the children at a White House Christmas party. Is Dick Cheneythe intentional object of the children’s experience? Santa Claus? Surely, theright answer is: Dick Cheney, seen as Santa Claus.)

If we set the strange claim about intentionality to one side, we are leftwith three other theses, which jointly ground Scruton’s claim that thereis no explaining musical expression: First, the thesis of musical anti-realism:sounds do not literally have the musical properties we hear them as having.Rather, and this is the second thesis, the musical and expressive descriptionsunder which we hear sounds when we hear them as music are metaphors.And, finally, although it is true that metaphors can be more or less apt,


there is no explaining any particular metaphor’s aptness. Every metaphorboth demands an explanation and also refuses it.

I think all three theses are deeply problematic, though here I will settleonly for questioning the second.¹ Suppose we agree with Scruton that themusical descriptions under which we hear sounds are literally false. Still, is itreally appropriate to call these perceptually exercised descriptions metaphors?

I would have thought that it is essential to metaphor that our use of it beintentional. Having experienced something and wishing to illuminate sometruth about it, we intentionally use a false description to bring that truthto light. ‘‘Jimi’s on fire’’ we say, for example, about a particularly excitingperformance by Jimi Hendrix.² But could that really be what’s going onwhen I hear the theme that launches the Goldberg variations? Do I havean experience of the sounds qua sounds, notice something about them andthen intentionally decide to illuminate what I heard by hearing them asconstituting that famous melody? That is surely absurd, for two reasons.

First, the tendency to hear it as that melody is as far from an intentionalmental act as anything gets. It is completely ineluctable. Scruton mentionsseveral examples of musical perception that do seem subject to the will. Forexample, the four opening drum beats of the Beethoven Violin Concertocan be heard either as part of the ensuing melody or just a preamble toit. But if his thesis were right, it would have to hold quite generally; anymusical metaphor would have to be optional from the standpoint of thewill. But they don’t seem to be.

Second, if the use of the notion of metaphor is to be justified, we wouldhave to be aware of some layer of musical experience with a perfectly literalcontent that our musical metaphors would be designed to illuminate. Butthere doesn’t seem to be such a layer of experience; for the most part, ourexperience is always already musical.

If this is right, then we can no longer ground the claim that there can beno explaining music’s expressiveness by appealing to the general claim thatthere can be no explaining any metaphor’s aptness.

At any rate, the anti-explanatory thesis is independently implausible.The point is that the expressive properties of music are clearly groundedin its purely musical properties. It is because a passage has certain musical

¹ This represents a change of heart about musical anti-realism, to which I had been inclined to bemore sympathetic. See Boghossian (2001).

² The example is Stephen Yablo’s, though not the point it is being used to illustrate.

124 paul boghossian

properties that it is heard as having certain expressive properties, as Scrutonin effect acknowledges in discussing the ‘‘Todesklage’’ from Act 2 ofWagner’s Die Walkure:

This theme contains a tragic, and yet questioning expression. It is a normal exerciseof the critical intelligence to look for the features which are responsible for sopowerful an effect: the accumulated suspensions, and the final Neapolitan cadencefinishing on a seventh chord, with its ‘‘unsaturated’’ and yearning character.Remove the suspensions and the tension goes. Alter the final cadence and we have(with a slight change of rhythm) the serene introduction to Mendelssohn’s ScottishSymphony in A minor, Op. 56.

(Scruton 1999:164–5)

Even if the musical properties are not themselves real properties of thesounds, but are only experienced as though they are real properties of thesounds, there would still be the question: how could the fact that soundsare heard as having certain musical properties explain why they are heardas having certain expressive properties? So, even if we were to concede themusical anti-realism, that would not invalidate the question with which thissection began: How is it possible for mere musical sound, lacking speakerintention, or any of the other resources which make linguistic meaningpossible, to express meaning?

9

The view I am most inclined to favor is a version of the resemblance theory,a view that has been very ably developed by Malcolm Budd (1995) andStephen Davies (1994). I would put it like this:

A passage P is expressive of E just in case P sounds the way a personwould sound who was expressing E vocally, or sounds the way aperson would look who was expressing E gesturally.

This kind of account leans, of course, on the idea that persons havecharacteristic ways of expressing their inner states. For example, there aretypical—and typically natural—ways that sad people look, typical waysthat they sound, and so forth. The account I favor, then, seeks to explaina passage’s being expressive of sadness by equating this property with theproperty of sounding like the way sad persons sound (leaving out for nowthe difficult cross-modal case).


Jerrold Levinson (2005) has an interesting objection to this idea. If yousay that sad music is music that resembles sad sound, he observes, you haveto say to what degree, since everything resembles everything else to somedegree. But the degree of resemblance required cannot be specified interms of some fixable degree of resemblance between the two. It can onlybe specified as whatever resemblance is sufficient to induce appropriatelisteners to hear the music as sad.

Levinson’s conviction that the resemblance relation cannot be spelledout in substantive terms—in terms which avoid what Crispin Wright hascalled, in a different connection, a ‘‘whatever it takes formulation’’—liesbehind his own dispositional account of musical expression:

P is expressive of E iff P, in context, can readily be heard, by a listener experiencedin the genre in question, as an expression of E.

(Levinson 2005)

He tentatively glosses a listener readily hearing P as an expression of E as:the listener readily imagining that P is an expression of E.

I think this is a challenging argument; but I would like to resist it becauseI have qualms about the dispositional account of musical expressivenessto which it seems inexorably to lead. That dispositional account is itself aversion of a no-explanation view of musical expressiveness.

10

Most of the objections that worry me are explicitly considered by Levinson;but I don’t think they have been sufficiently answered.

First, I think the gloss—hearing P as an expression of E in termsof imaginings—is ill-advised. It’s just too easy to imagine a particularpassage as being the expression of a variety of different—even contradicto-ry—emotions. Imagining is too unconstrained and so the resulting contenttoo indeterminate.

Second, Levinson’s definition is of ‘‘P’s being expressive of E.’’ This isa property that a passage can literally have. One of the motivations forthinking that music has such expressive properties is, as we have seen, thatwe seem to be able to hear them in the music. But can our hearing revealto us that a particular musical passage is expressive of sadness, for example,on Levinson’s view? To do so, it would seem, our hearing would have toreveal to us that P is a passage that has the dispositional property to induce

126 paul boghossian

listeners who are appropriately versed in the genre to hear the passage asthough it were a literal expression of sadness. I don’t see how that could bean audible property of the music.

Levinson has a reply to this objection. He says that qualified listenerstacitly assume that they are indeed qualified listeners, and so, for suchlisteners, hearing an expression of sadness in a passage becomes tantamountto hearing the passage’s being expressive of sadness.

Qualified listeners arguably tacitly assume, while listening to music that they arequalified listeners, listening appropriately. Thus on the view I favor, hearing anexpression of E in a stretch of music becomes, for such listeners, tantamount tohearing the music’s expressiveness of E. All a qualified listener need do to readilyhear the expressiveness of the music is to readily hear expression of emotion in it.

(Levinson 2005)

Whether this reply is adequate depends on what one takes the relevantdatum to be. What are we claiming when we say that an appropriatelyqualified listener can just hear the sadness in the music? On a stronginterpretation, it is to say that an appropriately qualified listener can know,on the basis of hearing alone, that the music is expressive of sadness, in justthe way in which it is natural to say that an appropriately qualified taster canknow, on the basis of taste alone, that a particular substance is sweet. Ona weaker interpretation, the claim would be not that a listener can knowon the basis of hearing alone that a particular passage expresses sadness, butonly the psychological claim that most listeners will feel no need to makean overt inference from their hearing to the attribution of expression.

Levinson’s observations can accommodate the weaker claim but not thestronger one. However, it is arguable that it is the stronger claim that needsrespecting. As a matter of psychology, there may be no inference when anexperienced observer sees some dark clouds and believes that it will rain.But as a matter of justification, the knowledge would have to be based oninference. Something is a perceptual justification of the belief that P just incase transitions from perceptions as of P (to the extent that there are any)to the belief that P are justified, without need of additional inference.

Levinson’s listener is like the experienced observer, moving swiftly fromhis hearing to the attribution of expression. But he cannot be said toknow, purely on the basis of hearing, that a particular passage has a certainexpressive property.


Arguably, though, I never need to infer that some music I hear isexpressive of sadness, let alone an inference that proceeds from somecomplicated premise about how others are likely to react. Rather, I hearthe expression in the music much as I taste the sweetness in the wine.

Third, I worry that Scruton’s objection that reference to the appropriateclass of listeners is doomed to vacuity will stick. There is a possible worldin which there are incredibly sophisticated musicians who write highlymathematically intricate music, perhaps a la Babbitt, but who profess tohear no expression in any of the nineteenth century music we play them.They have emotions, so they are not Spocks (as in Star Trek). They expressthose emotions as we do. But they simply don’t see the resemblancesbetween musical sadness and human sadness that underlie judgments ofexpressiveness. Of course, we will judge them incompetent in the genre.But will we have any basis for this judgment other than their failure to hearexpression in the music.

This is, of course, a standard problem for this sort of ideal observertheory of a given fact. You have to make sure that your conception of whocounts as an ideal observer isn’t controlled by a prior conception of the factin question, for the aim is to reduce the fact to the responses of the idealobserver.

11

What, then, should we say about Levinson’s argument? He makes it soundas though there is a perfectly general problem here: everything resembleseverything to some extent; so there is no saying in what respects musicneeds to resemble the expression of emotion if it is to express that emotion.However, there is reason for doubting this general argument.

Levinson agrees that listeners will often use judgments of resemblance tomake judgments of expression. If we ask one of Levinson’s ideal observerswhether some passage P is expressive of E, he will give an answer andthat answer won’t be a stab in the dark. He will have reasons and thosereasons will consist in judgments of resemblance. We could raise a similarpuzzle for this use of judgments of resemblance: given that everythingresembles everything else to some degree or other, how could a judgmentof resemblance help a listener to make a justified judgment of expression?To what extent should the musical passage sound the way a person would

128 paul boghossian

sound who was expressing sadness before we could be justified in sayingthat the passage expresses sadness?

Well, it’s obviously hard to say, but there had better be a solution to thatproblem if judgments of resemblance are to do what virtually everyoneagrees they must be able to do: namely, underlie judgments of expression.

But if we can solve that problem in its epistemological guise, then I don’tsee a problem of principle in solving it in its analytic guise. If we can say towhat extent a passage P must sound like the natural expression of E for itto be a good guide as to whether P is expressive of E, then we ought to beable to say to what extent P must sound like E for it to be the case that P isexpressive of E. If there are problems with this analytic claim, they won’tstem from an inability to specify the degree of resemblance that’s required.

Conclusion

In this short paper, I have tried, very schematically, to indicate how Iwould be inclined to navigate my way through this very difficult terrain.The requisite details must await another occasion.

References

Boghossian, P. (2001) ‘‘On Hearing the Music in the Sound,’’ Journal ofAesthetics and Art Criticism, November 2001.

Budd, M. (1995) Values of Art, London: Penguin Press.Davies, S. (1994) Musical Meaning and Expression, Ithaca, NY: Cornell

University Press.Kivy, P. (2001) New Essays on Musical Understanding, Oxford: Oxford

University Press.Levinson, J. (2005) ‘‘Musical Expressiveness as Hearability-as-Expression,’’

in Matthew Kieran (ed.), Contemporary Debates in Aesthetics and thePhilosophy of Art, Oxford: Blackwell.

Scruton, R. (1999) The Aesthetics of Music, Oxford: Oxford UniversityPress.

Yablo, S. (2000) ‘‘Apriority and Existence,’’ in P. Boghossian and C.Peacocke (eds.), New Essays on the A Priori, Oxford: Oxford UniversityPress.


Further Reading

Boghossian, P. (2001) ‘‘On Hearing the Music in the Sound,’’ Journal ofAesthetics and Art Criticism, November 2001.

Levinson, J. (2005) ‘‘Musical Expressiveness as Hearability-as-Expression,’’in Matthew Kieran (ed.), Contemporary Debates in Aesthetics and thePhilosophy of Art, Oxford: Blackwell.

Scruton, R. (1999) The Aesthetics of Music, Oxford: Oxford UniversityPress, especially Chapters 3 and 6.

6Persona Sometimes Grata:On the Appreciationof Expressive MusicA A RON RID L E Y

Several writers have discussed the possibility that, in hearing pieces of musicas expressive, listeners imaginatively construct for themselves a musical‘persona’ whose affective states the music is then heard as expressing. Thisthought has polarized opinion: Jerrold Levinson, for instance, claims thatthe construction of such a persona is always necessary if music is to be heardas expressive (1996); Stephen Davies, by contrast, claims that it is nevernecessary (2003). Apparently floating around in between, I have suggestedthat the construction of a persona is sometimes necessary, sometimes not(Ridley 1995). I want to defend that view here. But I want to make clear, indoing so, why that view is not the compromise position that it might seemto be. Levinson and Davies appear to disagree, first and foremost, aboutwhether the construction of a persona is necessary if music is to be heardas expressive at all. My claim, by contrast, is a critical claim: whether theconstruction of a persona is necessary or not depends upon the particularcharacter of individual pieces of music. And this isn’t a compromise becausewhat interests me is the appreciation of expressive music, rather than thebare possibility of hearing music as expressive—the issue that primarilyexercises Levinson and Davies. Indeed, or so I will argue, my issue is priorto theirs, and a correct treatment of it indicates that, while Davies is in onesense right about the question of bare possibility, Levinson’s intuitions are,in another sense, closer to the mark.

persona sometimes grata: expressive music 131

1

Let’s begin, though, with a case that has nothing directly to do with musicat all, although it is very familiar from the literature on musical expression.Consider a weeping willow. It droops; it slowly sways; it does all of theusual willowy things. And it very naturally attracts epithets such as ‘sad’,‘melancholy’ or ‘downcast’, even though no one thinks that these termscapture anything about the willow’s frame of mind, since no one thinksthat there is anything of that sort to be captured. Nor do these termsnecessarily reflect the states of mind induced by the willow in its beholders:I may perfectly reasonably characterize the tree as ‘melancholy’ whateverthe sight of it makes me feel, including nothing. And yet it doesn’t seem tobe hard to see why the relevant expressive ascriptions might be thought toapply. The willow is ‘sad’ or ‘downcast’ just because it looks as if it is.

On the face of it, this observation might be meant in either of two senses.It might be taken to mean, first, that we see the tree as if it were a sadperson: we imaginatively construct a human figure out of it, and diagnoseits posture as one of sadness. Or it might be taken to mean, simply, that wenotice—and label ‘sad’—features of the tree (droopiness etc.) that, werethey the features of a human being, would betoken sadness. In this secondcase, it is tempting to say, we do not see the tree as if it were a sad humanbeing; rather, we recognize features of it that it shares with sad humanbeings, but see those features as features of the tree. In the first case, wehave constructed a persona; in the second case we have not. In either case,however, it appears that we have seen the tree as ‘sad’, as expressive ofsadness.

This might seem to suggest that, at least as far as willows are concerned,the construction of a persona is not necessary for the experience ofexpressiveness. We can see the droopiness of the willow as ‘sad’, it appears,without imputing either the droopiness or the sadness to the human figurewhom we might see in the tree. The tree itself, as one might want to say,is droopy and hence, without further ado, appropriately to be seen as sad.

But there are reasons to wonder whether this might not be a bit quick.For one thing, the adjective ‘droopy’ seems already to be laden withemotive connotations of just the sort—of ‘sadness’, of ‘melancholy’—thatit is being enlisted to justify when those terms are applied to the willow.

132 aaron ridley

What if, instead, one saw and described the willow simply as possessedof thin and pliable parts which, given gravity, tend to hang downwards?There are some grasses like that—not to mention party-streamers—andone would strain to describe these as either droopy or sad. And this wouldseem to make the ‘droopy, hence sad’ move in the case of the willowlook rather under-motivated. Nor does the most obvious sort of attempt toremotivate it look terribly promising. The willow, it might be said, unlikegrasses and streamers, has a fundamental uprightness of build that makesthe downward tendency of its outer parts especially and strikingly droopy,and so that much more likely to be seen as sad. But this, clearly enough,is to come perilously close to conceding that we see the tree in that wayonly because we relate its basic uprightness to the uprightness of a humanfigure, and see the downwardness of its branches as ‘droopy-hence-sad’in the context of that gestalt, specifically. In which case, it is going to betricky to steer off the thought that we attribute sadness to the tree onlybecause we see it as if it were a sad person—only, in other words, becausewe construct a persona. Yet this is just what was to be denied.

So does the persona story win out here, after all? I doubt it. For itis clearly open to the opponent of that story to argue that the cues thatinvite the construction of a persona in the first place must—if we are notout-and-out animists—be a feature or a pattern of features of the tree thatis already expressively suggestive, or else nothing in the situation shouldhave prompted us to see the tree as if it were any sort of person at all,let alone a sad one. And that, surely, suggests that expressive attributionsare logically prior to the construction of the persona that was, ex hypothesi,supposed to explain them. So it would seem that, despite the considerationsoffered in the previous paragraph, no persona is necessary if the willow isto be seen as sad.

But then again ...One could go on in this vein, I think, pretty well indefinitely, alternating

between plausible enough defences of either position without, in theend, arriving at anything conclusive. Do we see willows as expressive onlybecause we anthropomorphize them? Or do we (can we) anthropomorphizethem only because we see them as expressive? I find it hard to imagine whatmight settle the question either way. And there are reasons for this—reasonsthat matter if we are to get clear about the apparently analogous debate inthe philosophy of music.


2

We can summarize the disagreement between the positions discussed in theprevious section in the following rather unattractive way. The pro-personaposition holds that:

(i) a feature, F, of a bearer, B, is aptly to be seen as expressive iff B isseen as if it were a person.

The anti-persona position holds that:

(ii) a feature, F, of a bearer, B, is aptly to be seen as expressive iff Fwould be expressive were F a feature of a person.¹

One striking thing about these positions, so summarized, is the fact thatneither concerns itself with the character of B, beyond its possession ofF: (ii) has nothing whatever to say about the matter, while (i) offers onlythe thought that B must be seeable as if it were a person, which, giventhat anything can in principle be seen as if it were anything else, is notinformative.² It might be suggested, I suppose, that their relative mutenessabout B is actually a strength of these positions, since it promises a certaingenerality that might otherwise be lacking. But any mileage that this movemight seem to have evaporates, surely, once one notices that neitherposition can deal convincingly with the case in which B is, in fact, a humanbeing. And this means that neither can pretend to offer an account ofexpressiveness in general (should such a thing be possible at all).

This is not, however, the most striking thing about the indifferenceof these positions to the character of B. The most striking thing, rather,connects to the reason why the disagreement between them appears, inthe case of weeping willows at least, to be irresolvable. For it matters,surely, that a willow is a tree—that it is a natural object whose characteris ultimately independent, as one might put it, of any fact or facts about

¹ It is worth mentioning that the openings of each of these formulations might, with some plausibility,have been reversed: i.e., both might have started ‘The bearer, B, of a feature, F ... ’, so making B ratherthan F what is aptly to be seen as expressive. I have chosen the formulations given in the main textsimply because they seem to me to work more convincingly for willow trees; but—as we will see inSection 6, below—it may be difficult to justify any general preference for one formulation over theother.

² Cf. Goodman 1976: 5.

134 aaron ridley

human beings.³ Or, to put the point another way, it matters that there isnothing in the character of the tree, qua natural object, that warrants ourtaking it as being expressive for these reasons rather than those.

The upshot of this—if it’s right—is that the disagreement sketchedin the previous section was always bound to be irresolvable, at least asit was introduced there. The disagreement was set up so as to have theappearance of a conceptual dispute about the precise application-conditionsof certain expressive predicates, an impression strengthened by the placeallotted in the exposition to normative-sounding notions such as ‘aptness’and ‘appropriateness’. But really the dispute was only ever, and at most,an empirical one, and so not usefully to be addressed in the way that Iwent through the motions of trying to. The real question comes to this:Do you, as a matter of fact (about you), find that you have to imagine thatthe tree is a human figure in order to see it as sad? Or do you, as a matterof fact (about you), find that you can see the sadness without having to dothat? And in this context the notion of ‘aptness’ is to be understood notas normative, and hence as connected to the conceptual, but as referringto the normal—that is, to whatever, given human beings or some subsetof them, should turn out to be statistically commonplace. So the dispute,such as it is, never did have anything to do with the specific grounds onwhich one would be warranted in seeing the tree as sad, however it mighthave looked at first.

In saying this, I am not of course denying that a weeping willow isaptly to be seen as sad. It looks as if it is sad, after all, and that’s enough.Nothing that I have said affects that. I am denying only that the finer-grained distinctions at issue between (i) and (ii) can get any conceptualpurchase in this context. The dispute between the two could be turnedinto a conceptual one, of course. It could be turned into one if thepro- and anti-persona positions each had something—perhaps differentthings—to say about the character of trees that made it plausible to thinkthat our expressive ascriptions were constrained in a relevantly fine-grainedway. And perhaps this is possible; I don’t know.⁴ But the dispute as I’veframed it doesn’t even try to do this. And the important point, for present

³ I set aside here any issues to do with humanly orchestrated breeding, planting, cultivation, etc.:the willow that I have in mind is, as it were, all its own work.

⁴ I can think of nothing in the recent literature on the aesthetics of nature that points very promisinglyin this direction.


purposes, is that without trying to do this nothing of any philosophicalmoment can be at issue. What we have here is only the appearance of aphilosophical disagreement, not the substance of one.

3

Music, by contrast, obviously does have something to do with us. We makeit, we play it and, if it is understood at all, it is so by us. There is, moreover,a long tradition, in the West at least, of hearing music as expressiveand—correspondingly—of composing and performing music with an earto its expressive effect. In these as in other respects, it is quite difficult toimagine anything much less like a willow tree than a piece of music is.

One might guess from this that the persona-debate, transplanted intowhat is after all its proper context, music, would come into its own asthe site of a genuinely pointful philosophical disagreement. And there is,as we shall see, a measure of truth to this. But it is still worth notingthat nothing in the basic form of the disagreement changes. In particular,and as we will also see, it is worth noting that neither position, pro-or anti-, has a lot to say about the individual character of the bearers ofthe features aptly to be heard as expressive. What both do, in effect, isto help themselves to the general cultural background gestured at in theprevious paragraph—which does at least bring with it a certain ballast ofnormativity—and then move directly to their competing claims about theapplication-conditions of expressive predicates to music tout court. And this,or so I’ll suggest, simply isn’t enough.

Let’s look at Davies’s anti-persona position first. In his book, MusicalMeaning and Expression, he advances an account of musical expression thatimputes expressiveness to pieces of music in virtue of their sharing ‘emotion-characteristics in appearance’ with human beings—that is, in virtue of theirbearing features which would be expressive were those features the featuresof a person (Davies 1994). The position, in other words, is precisely a varietyof (ii), as set out above. And Davies’s project in the essay that concerns usis to defend that account by discrediting the pro-persona position, whichhe takes to be its most serious rival (Davies 2003: 154).⁵

⁵ On the same page, Davies glosses his own account as follows: ‘musical materials can be lit-erally expressive as a result of presenting to audition sounds with emotion characteristics ... Music

136 aaron ridley

He attempts to do this, primarily, by arguing that the reasons thatpeople have had for proposing the pro-persona position are not verygood ones. He devotes most attention to two such reasons. One is thatmusic, or so it is sometimes claimed, is capable of expressing the so-called‘higher’ emotions, that is, emotions whose character goes beyond ‘whatcan be presented in appearances’ because, unlike for instance ‘sadness andhappiness’, they involve cognitively complex content. Examples of higheremotions include: ‘patriotism, shame, pride, embarrassment, envy, andhope’. If it were true, Davies suggests, that music could convey these states,this would count in favour of the pro-persona position:

To make believe someone personified in the music is to imagine that person aspossessing beliefs, desires, intentions, and attitudes, even if the contents or objectsof these are not transparently conveyed. To entertain that such a person is presentin the work is also to make believe a context where cognitively complex emotionsmight be musically expressed. If it is appropriate to hear music as expressing notsolely happiness and sadness but also more subtle, cognitively rich emotions, [thepro-persona position] provides for this possibility (Davies 2003: 156).

But Davies doubts that it is true that music can convey such cognitivelycomplex states, and he has a reasonably easy time of it showing that oneattempt to establish a claim to the contrary is not persuasive (2003: 161–6).⁶

I am more impressed than Davies is by the stronger defences of thecapacity of music to express the higher emotions.⁷ But for the sake ofargument let’s agree with him that these states lie beyond music’s compass,and that the capacity to express them cannot be enlisted on behalf of thepro-persona position. This still leaves open the possibility that to hear musicas expressive of (some of) the other emotions might (sometimes) requirethe construction of a persona. And certainly it is hard to see how theseemotions’ alleged lack of cognitively complex content might be supposedto rule this possibility out.

The other main reason in favour of the pro-persona position thatDavies considers is that a listening conducted in its terms may ‘explain

is sad-sounding in much the way that basset-hounds ... are sad-looking. Though it takes imaginationto hear music’s expressiveness, it does not take more than is needed to see ... willow trees as downcast’(2003: 154).

⁶ His target in these pages, incidentally, is Robinson 1994, plus, on occasion, Robinson and Karl1995.

⁷ In particular, I am impressed by Levinson 1990.


the structure and coherence of [a] work where a purely technical accountwill be inadequate’ (2003: 166). His grounds for rejecting this claim,however, seem to be merely stipulative. He insists, without argument, thatany persona-construction ‘comes after the recognition of musical unityand the like and, therefore, does not account for that experience’ (2003:167). But this, clearly enough, is to presuppose, rather than to give areason for accepting, the absolute priority of the technical. Indeed, hesuggests that ‘If the work strikes us as episodic and disjointed, it is notplain that we should prefer a coherent’ experience of it that depends uponthe imaginative construction of a persona; and it is clearly his view thatwe should not prefer such an experience (2003: 168). Yet it is hard tosee how this (negative) preference is to be justified: for, on the personaaccount envisaged, which does not presuppose the absolute priority ofthe technical, the work in question will precisely not strike us ‘as episodicand disjointed’; and so the issue, in this form at any rate, would notarise.⁸

But—again—let’s concede the point for the sake of argument, andaccept that pieces of music are unified or not irrespective of their expressiveproperties, and so that the construction of a persona can never trumpthe findings of technical analysis. How much would this show? Only,surely, that the possibility of structural integrity cannot be taken as asign that the pro-persona position is right. And this hardly shows thatit’s wrong.

Davies presents himself as discussing these two pro-persona reasons underthe umbrella of a larger pro-persona reason, namely, that ‘one cannotunderstand ... music (fully) without invoking the presence of [a persona]within it’ (2003: 160)—a reason that, as I hope that the foregoing will havemade clear, cannot be taken to have been defeated by the reflections offeredon the alleged impossibility of music’s expressing the higher emotions orof its having unity imparted to it by its expressive character.⁹ But Davies’sprincipal conclusion, that apparent emotion-characteristics are all that isneeded, at a general level, to explain our experience of music as expressive,still has the room, as it were, to be completely right as far as it goes. Ifhis account is persuasive, after all—as it might well be—it establishes that

⁸ See also Davies 2003: 155–6, for further arguments of the same kind. I have tried to say somethingabout the sources of this sort of question-begging elsewhere: see Ridley 2004, passim.

⁹ I try to say something more positive than this in Section 4, below.

138 aaron ridley

it is a sufficient condition of our expressive ascriptions’ being at least aptthat they are grounded in the possession by pieces of music of featuresthat would be expressive if they were the features of a person. And thisthought, clearly enough, would survive the truth of a claim to the effectthat this piece of music, or some particular detail of a piece of music, doesin fact require the construction of a persona if its expressiveness is to begrasped fully, that is, in a way that goes beyond being merely apt.¹⁰ It isalso, of course, a thought that comes more or less for free once one reflectsfor a moment on the general cultural background, as I put it a momentago, against which music is heard. We are highly attuned to hearing andtalking about music as expressive; we are alert for and proficient at spottingthe general expressive character of pieces of music; and this, in effect,is all that Davies’s account trades upon. And his account would stay inthe business proper to it even if it could be shown, which I doubt, thatdetecting emotion characteristics in appearance was only made possible bysome prior but now largely discarded habit (whether at an individual ora cultural level) of hearing expressiveness in music as the expression of aconstructed persona.

What I draw from this brief discussion of Davies’s position is threefold.First, the grounds that he gives for denying that it is (ever) true that‘one cannot understand ... music (fully) without invoking the presence of[a persona] within it’ are not sufficient to establish that conclusion, evenwhen one lets through the (surely over-generous) concessions made above.Second, Davies’s account should be regarded as evincing his commitmentto, rather than as offering a defence of, the view that expressiveness isrelatively insignificant to musical understanding (hence his rather quickway with the claims that music might express the higher emotions, or thatexpressiveness might impart musical unity). And third, and following onfrom the previous point, there is little sign that Davies’s position is drivenby close critical inspection of individual pieces of expressive music: instead,it seems to be motivated by quite general considerations about expressiveascriptions, of a sort that sometimes take us rather close to weeping willowterritory.

¹⁰ That is, accurate as far as it goes. So, for example, someone might quite aptly, in this sense,describe the ending of King Lear, or of Mahler’s Ninth Symphony, as ‘sad’. (I leave it open whetherthese examples might not work as a virtual reductio of this sense of ‘aptness’.)


4

I turn now—still more briefly—to Jerrold Levinson’s pro-persona position.Levinson, like Davies, is concerned to tell a story about all musicalexpressiveness; and, as the following should make clear, he is an advocateof a version of (i) (see Section 2, above):

A passage of music P is expressive of an emotion or other psychic condition E iffP, in context, is readily and aptly heard by an appropriately backgrounded listeneras the expression of E, in a sui generis, ‘musical’, manner, by an indefinite agent,the music’s persona.


There are of course some additional details here which are important if oneis to arrive at a rounded grasp of Levinson’s view. But for our purposeswe can set these aside, and note, simply, that he regards the presence ofa persona in the listener’s experience as a necessary condition of music’sbeing heard as expressive at all.

But what, one might reasonably ask, if—like Davies—I attend to mymusical experience (either in general, or in some particular case) and findthat it includes the apprehension of expressiveness, but not the imaginationof a persona? Can I (or Davies) just be wrong about something like that?Levinson’s answer, in effect, appears to be yes, we can. He says:

the imaginative hearing-as that the view suggests typically goes on when suchexpressiveness is registered need not be highly foregrounded, nor ... need it be verydeterminate in regard to the nature or identity of the expressing agent.


So: we may believe that we are not imagining a persona, but the explanationfor that is simply that we have not noticed that we are doing so (althoughwe are).

I complained above about the places in Davies’s account where argumentseemed to have given way to stipulation, and it is only fair to raise the samecomplaint here. Why, exactly, should I accept what Levinson is telling me?Or, to put the question another way, what would he say to me if, in a spiritof good will, I listened again and really paid attention this time, but couldstill find no persona-imaginings in my experience (although I did hear the

140 aaron ridley

music’s expressiveness)? Surely we’d have reached a stand-off.¹¹ And thesort of stand-off we’d have reached, it seems to me, is non-coincidentallysimilar to the one that—perhaps inevitably—ended the parallel ‘debate’about willow trees.

5

Such an outcome is not inevitable in the case of music, however. Thereare at least two ways of avoiding it. One would be to stick at the level ofgenerality that Davies and Levinson appear to prefer, and to say somethinglike this: that (a) Davies’s account tells us what must be minimally truewhenever music is heard as expressive (for which persona-construction isnever necessary); while (b) Levinson’s account tells us what must be truewhen the expressive character of any piece of music is understood, notmerely in a minimally apt way, but fully (for which persona-construction isalways necessary). Neither party would be likely to welcome this resolution,of course. For Davies, the cost of conceding that persona-construction isalways required for a full understanding would surely be too high; whileLevinson, I imagine, ought to baulk at having to concede that, for it to betrue that someone aptly hears pieces of music as expressive of what theyexpress, persona-construction need never go on.¹²

But, actually, it seems to me that none of us should welcome thisresolution. It is possible, as I’ve indicated on a couple of occasions, thatDavies’s account may be more successful in locating the minimally necessaryconditions for expressive hearing to be possible at all; and part (a) of theresolution does at least preserve a version of that. But the problem, in my

¹¹ But what about Levinson’s positive arguments for the necessity of a persona? Am I just neglectingthese? Yes and no. I’m not neglecting them insofar as they feature in the essay under discussion, since Ican find none there. I am, however, neglecting them as they (implicitly) feature elsewhere in Levinson’swritings. The reason for this, though, is that they seem to be very much like the pro-persona argumentscanvassed in the previous section. And these, while they may in fact be persuasive (within limits), Ihave decided—for the sake of argument—to set aside, as Davies would have us do. The larger strategicpoint of proceeding in this way should become clear in Section 5.

¹² In fact—and at odds with the main thrust of his essay—Levinson might, at one level, justabout countenance this concession. He writes: ‘it may not, indeed, be strictly necessary that thelistener ... imagine [a persona], as long as he sees what sort of imagining is prompted by the music’(Levinson 1996: 118). I’m not sure exactly what this means, but it does seem to gesture in the directionthat I have mentioned.


view, is that both parts of the resolution inherit from the original twopositions their least attractive, and most gratuitous, feature—namely, thedetermination to move directly from the background cultural conditionssketched at the beginning of Section 3 to an utterly general account ofwhat makes for apt, or as it might be ‘full’, understanding of expressivemusic under any circumstances whatever. And while the (usually implicit)inclusion of the relevant cultural background is the reason why this debatedoesn’t merely collapse into a version of the one about willows, the directmove made by both parties from background to generality nonethelessensures that the threat of some such collapse is never far away.

The other way of avoiding irresolvable impasse, at least of the sortthat has so far arisen, is therefore, it seems to me, to repudiate anypremature ambition to generality, and even to suspend judgement (maybefor a very long time indeed) about the very possibility of generality inthis context. In Section 2—perhaps cryptically—I attributed the futilityof the willow ‘debate’ to the failure of either position to have muchto say about the character of the bearers of the features at issue; and,at the beginning of Section 3, I observed that the parallel disagreementabout music carried over that indifference. It should now be clear whatI was getting at. Levinson’s and Davies’s respective versions of (i) and (ii)depend only minimally, if at all, upon any sort of critical engagement withindividual pieces of music: the premature demand for generality precludesthat. And what I am suggesting now, of course, is that any prospectof philosophical progress in this area depends precisely upon, and mustbe rooted thoroughly in, an engagement of just that kind—namely, anengagement that pays proper critical attention to the character of particularpieces of music.

In my previous efforts to say something about musical personas this ishow I tried to proceed (Ridley 1995: Chapter 8; 2004: Chapter 5). Andthis is why what I came up with—the thought that persona-constructionis sometimes necessary, sometimes not—does not represent any kind ofcompromise between the two positions that we have been discussing.¹³My concern, unlike theirs, was to tease out some of the philosophicalconsequences of the critical appreciation of music. And that, for the reasons

¹³ If anything would represent such a compromise, it would be the (unappetizing) ‘resolution’sketched at the beginning of the present section.

142 aaron ridley

I have given, is a project that must come before any grander, and perhapsmore metaphysically glamorous, undertaking is attempted.¹⁴

6

Let me conclude with two musical examples. The first, which is relativelyDavies-friendly, is the G Major Fugue from Book II of J. S. Bach’s TheWell-Tempered Clavier. There is no question that this piece has expressivefeatures; nor any question, I think, that these are of a predominantly up-beatkind. Does this work call for the construction of an expressive persona? Ithink not. Of course, there is nothing to prevent one from constructing apersona if one wants to (I’ve just tried—it’s easy). But, to my ear, nothingis gained by doing so: the music doesn’t emerge the richer for it, nor isthere anything about its expressive character, specifically, that appears toreward this way of listening. In this case, it seems to me, the piece is aptly tobe heard as possessing up-beat expressive features, and that’s it. The reasonfor this, I would suggest, is simply that the fugue’s expressiveness is not themost interesting, or even a very interesting, thing about it; so that the kindof imaginative effort that goes into the construction of a persona actuallydistracts attention from aspects of the work that deserve it more. And thisshould alert us to one of the many pitfalls of generalizing too soon. For ifnot every work that is expressive is interesting primarily in virtue of thatfact, there can be no grounds for assuming that all expressive works areto be listened to in the same way, with the same pattern and balance ofattention.¹⁵

The second example is of a work—Samuel Barber’s Adagio for Strings—which does, I think, call for the construction of a persona. I have chosenthis piece for several reasons, but two of the more prominent ones arethese: first, it is at least plausible to say that the Adagio is expressive ofsadness, say, or depression, and not of any of the higher emotions; and,

¹⁴ I should note that my approach is obviously not immune to impasses of its own. Somecritical disagreements may be simply irresolvable. But these, at least, are impasses that come withthe territory—with art, with music, with appreciation: they are not the creatures of a particularphilosophical project.

¹⁵ I noted at the end of Section 3 that Davies seems to regard expressiveness as relatively incidentalto musical understanding. And this, of course, is why the Bach example is ‘Davies-friendly’. Davies’smistake is to assume that examples of this sort have the field to themselves.


second, the work can undoubtedly be considered unified and coherent on‘technical’ grounds alone. It is, in other words, exactly the sort of workthat should, if Davies is right, least reward being listened to with a personain mind. On Davies’s account, of course, the Adagio would aptly be heardas exhibiting, merely, a succession of sadness- or depression-characteristicsin appearance;¹⁶ and his faith in this view is bolstered by the followingreflection:

A work that might be heard as laying out the developing gloom of a depressedpersona could be experienced no less convincingly as indicating the unconnectedmoods of a series of personas, each of whom is (independently) more depressedthan the last.

(Davies 2003: 167)

This seems to leave the idea of a persona (as opposed to a successionof emotion-characteristics in appearance) doing no work. Now I’ve triedlistening to the Adagio—surely an appropriate test-case—in the way thatDavies suggests, and find that any sort of experience of it, let alone aconvincing one, as having to do with the ‘unconnected moods of a seriesof personas’ is just not to be had (or at least not in a way that leaves thefaintest trace of a musical experience intact). And the sense of outlandishnessprovoked by this suggestion is only increased when Davies floats the ideathat someone might defend a preference for one persona over several, orvice versa, as ‘a form of inference to the best explanation. Where evidenceunderdetermines theory’, he says, ‘we may still prefer some theories overothers’ on the basis of, e.g., ‘predictive power, economy of elements,elegance of structure, and the like’ (2003: 167).¹⁷

This brings out, I think, the gulf that separates Davies’s project notonly from mine, but from any project that is at all rooted in the actualexperience of listening to music: no one concerned with the latter, surely,would have it cross their mind that what might be at issue was an ‘inferenceto the best explanation’.¹⁸ Rather, the issue would present itself as a criticalone.¹⁹ Is Barber’s Adagio, like Bach’s fugue, most rewardingly or revealingly

¹⁶ Davies doesn’t discuss this piece, but what he would have to say about it is clear.¹⁷ Davies goes on to reject this idea, although not on the grounds that most people would.¹⁸ Nor, I suspect, would many think that what they were after was a theory. There is a distinct whiff

of scientism here.¹⁹ As Levinson, to his credit, also recognizes. He considers a worry that is outwardly similar to

Davies’s—one persona or several?—and suggests that ‘the answer will vary with the genre and

144 aaron ridley

to be experienced as (merely) possessing expressive features? Or is it bestexperienced as the expression of a persona—or, indeed, as the expressionof a series of personas?

I have already indicated that the Adagio strikes me as impervious to amultiple-persona experience.²⁰ And, despite the absence from it of the sortsof cues—having to do with the higher emotions and structural unity—thatDavies regards as pointing to the principal, if in his view spurious, reasonsfor the attraction of the pro-persona position, I have suggested that theAdagio is, indeed, to be heard as the expression of a (single) persona. Sowhy do I say this? What does this way of hearing the work have to offerthat is missing from the experience of hearing it as a string of ‘emotion-characteristics in appearance’? The answer, I think, is depth. Experiencedin Davies’s preferred way (which, unlike the multiple-persona experience,seems to me to be perfectly possible), the work exhibits a sustained characterof sadness and a structural unity that can be accounted for in purely technicalterms—that is, without reference to the work’s expressive features. Indeed,heard thus, those features are essentially a side-effect of the aspects of thework to which a technical account would appeal; and this is to makethose features incidental, and, in one important sense, superficial. Heard asthe expression of a persona, by contrast, the work’s expressive dimensionmoves centre-stage, and affords an absorbing portrait of a very particularframe of mind, the character and unity of which are not heard as any sortof side-effect of something else, but as presented directly. And this providesthe possibility of a richness and depth of musical experience that Davies’salternative, in the case of this particular work, does not. The Adagio issimply better, more interesting, when experienced as the expression ofa persona. And since none of the relevant cultural background or musichistory counts against such a listening, we should—on ordinary critical-appreciation grounds—prefer it. (It is an idea somewhat like this, I take it,that underlies, and is then over-stated in, Levinson’s pro-persona position.)

One final thought, and one—by now predictable—caveat. The thoughtis this: if what I have said about the Bach fugue and the Barber Adagio is

style of the work involved, the particulars of its formal construction, and the way in which the localexpressiveness of its parts relate to any global expressive aim that might appear to be in view’ (Levinson1996: 107, fn.55).

²⁰ This is not of course to say that all works are similarly resistant. For example, it seems to me thatElgar’s Enigma Variations might very plausibly be heard in this way (and not just because Elgar said so).


persuasive, this has consequences for what, precisely, we hear as expressive.For, while in the former case it seems plausible to say that the upbeatnessresides in certain features of Bach’s music, in the latter case it is altogethermore tempting to say that it is the music itself that is heard as sad, and not justthis or that about it. And this, if it’s right, means that any attempt to givea general formulation (however persona-tolerant) of the conditions underwhich music is to be heard as expressive runs into a problem. For either ofthe obvious ways of framing such a formulation—namely, ‘a feature, F, ofa bearer, B, is aptly to be heard as expressive iff ... ’ and ‘the bearer, B, of afeature, F, is aptly to be heard as expressive iff ... ’—deforms the landscapein advance, as it were, and places arbitrary limits upon the ways in whichpieces of music are to be heard. This thought confirms me in the view thatgenerality is not a valuable goal in this area of philosophy.

And now the caveat, which is related. I said a moment ago that it doesn’tfollow from a work’s being expressive that it is interesting primarily invirtue of that fact—and so that it doesn’t follow that all expressive musicis to be listened to in the same way. A corollary of this is that it doesn’tfollow from the fact that a work is interesting primarily in virtue of itsexpressiveness (as Barber’s Adagio, I think, is) that it is therefore to be heardin a given way—for example, as the expression of a persona. It may be; itmay not be.²¹ One just has to listen and see. So, again, there is no placehere, at the level that matters, for generalizing in advance of the specificexperiences that individual pieces of music have to offer.²²

References


Davies, S. (2003) ‘Contra the Hypothetical Persona in Music’, in his Themesin the Philosophy of Music, Oxford: Oxford University Press, 152–68.

Goodman, N. (1976) Languages of Art, Indianapolis: Hackett.Levinson, J. (1990) ‘Hope in ‘‘The Hebrides’’ ’, in his Music, Art, and

Metaphysics, Ithaca, NY: Cornell University Press: 336–75.

²¹ Delius’s Summer Night on the River, for example, seems to me to be a piece that is primarilyinteresting in virtue of its expressive character, but which is not best heard as the expression of a persona.

²² My thanks to Maria Alvarez, Chris Janaway, Alex Neill and Kathleen Stock for comments onearlier versions of this essay.

146 aaron ridley

Levinson, J. (1996) ‘Musical Expressiveness’, in his The Pleasures of Aesthetics,Ithaca, NY: Cornell University Press: 90–125.

Ridley, A. (1995) Music, Value and the Passions, Ithaca, NY: CornellUniversity Press.

Ridley, A. (2004) The Philosophy of Music: Theme and Variations, Edinburgh:Edinburgh University Press.

Robinson, J. (1994) ‘The Expression and Arousal of Emotion in Music’,Journal of Aesthetics and Art Criticism 52: 13–22.

Robinson, J., and Karl, G. (1995) ‘Shostakovitch’s Tenth Symphony andthe Musical Expression of Cognitively Complex Emotions’, Journal ofAesthetics and Art Criticism 53: 401–15.

Further Reading


Davies, S. (2003) ‘Contra the Hypothetical Persona in Music’, in his Themesin the Philosophy of Music, Oxford: Oxford University Press, 152–68.

Levinson, J. (1990) ‘Hope in ‘‘The Hebrides’’’, in his Music, Art, andMetaphysics, Ithaca, NY: Cornell University Press, 336–75.

Levinson, J. (1996) ‘Musical Expressiveness’, in his The Pleasures of Aesthetics,Ithaca, NY: Cornell University Press, 90–125.

Ridley, A. (1995) Music, Value and the Passions, Ithaca, NY: CornellUniversity Press.

Ridley, A. (2004) The Philosophy of Music: Theme and Variations, Edinburgh:Edinburgh University Press.

PA RT I I I

Musical Meaning


7Can Music Functionas a Metaphor of Emotional Life?J E NE F E R ROB INS ON

Introduction

Over the past thirty years or so there has been a significant change inAnglo-American music theory. The idea of ‘‘pure’’ instrumental music as aformal, quasi-mathematical structure with no reference to anything beyonditself has given way to more contextual approaches, in which instrumentalmusic is treated as having meaning and expressive qualities which linkit to wider human concerns. Instead of focusing solely on the intricatetechnicalities of harmonic analysis, music theorists are turning to the studyof musical semiotics, so-called musical ‘‘topoi,’’ and, more generally, thepsychological and emotional meanings to be found in even the purest ofpure music.¹

However, as formalist critics persist in pointing out, it is not obvious inwhat sense pure music without words can mean anything at all. Meaninginvolves some kind of reference of a sign to a signified, and accordingto the formalists instrumental music has no ‘‘extra-musical’’ referenceat all. Some formalists are happy to grant that music can express lots ofthings—melancholy, joy, serenity, restlessness—but deny that expression isa form of meaning (Kivy 1989, 1990). On this view, to possess an expressivequality is not to refer to anything. Others disagree. Most notably, NelsonGoodman has suggested that expression is indeed a kind of reference, butof a metaphorical sort (Goodman 1968, Ch. 2). Some theorists go further,

¹ There has also been much written on the political meanings of music, but I will not address thisissue directly.

150 jenefer robinson

arguing that music can itself function as a metaphor of emotional life(Ferguson 1960).

In this essay I argue that Goodman’s analysis of expression as meta-phorical exemplification contains the germs of two different conceptionsabout musical meaning and expression. One of these is the familiar ideathat musical expression is just a matter of a musical piece or passagemetaphorically having or possessing certain expressive qualities, such assadness, tenderness, jollity, serenity, or nervousness. The other is the moreintriguing notion that musical expression sometimes goes beyond the merepossession of qualities and can be understood in terms of the exemplificationby an extended piece of music of some more complex psychological oremotional drama, such as a struggle to victory or a nostalgic attempt torecover a lost past. It is this second set of cases that have tempted somepeople to say that music can function as a metaphor for complex humanemotional and other psychological states.

I begin by laying out Goodman’s view. Then I discuss a critique ofthe view by Stephen Davies, which seems to hint at this distinction, butin fact makes a different distinction that I find problematic. Then I turnto Anthony Newcomb’s more sympathetic interpretation of Goodman’stheory, according to which a piece of music as a whole can sometimesbe understood as metaphorically exemplifying some large-scale story ornarrative. The trouble with this proposal is that it suggests no way todistinguish between appropriate and inappropriate candidates for what isbeing exemplified. In the final two sections I develop Newcomb’s viewby suggesting that some pieces of music are appropriately heard in termsof a particular overarching structural metaphor, in particular the metaphorof music as discourse, and by proposing criteria of appropriateness for suchmetaphors. Finally, I suggest that if a piece of music is appropriatelyunderstood as metaphorically a discourse of one sort or another, then itmay also be appropriate to hear such music as expressing something aboutthe emotional life of the characters participating in the discourse. A piece ofmusic as a whole can, then, in a sense, function as a metaphor of emotionallife, to the extent that it can appropriately be experienced in terms of alarge-scale structural metaphor—music as oration, music as drama, musicas narrative etc.—in which there are characters whose emotional lives areexemplified by the music.

can music function as metaphor of emotional life? 151

1

The basic premise of Goodman’s Languages of Art is that works of artare symbols in symbol systems. A novel is in a language, which is onekind of symbol system; pictures are in pictorial symbol systems; musicalnotation is a third kind of symbol system. Since art works are symbols insystems, our appreciation of art is largely cognitive. Traditionally, art workshave been discussed aesthetically in terms of three important features:their subject matter, their formal features and the feelings, thoughts orattitudes that they express. Much of the force of saying that works of artare symbols in symbol systems is that for Goodman these three aspectsof art can be analyzed as different kinds of symbolic relation or referencerelation between the work and its subject, its formal features and itsexpressive features. So the work represents or denotes its subject matter;² itexemplifies its formal features; and it metaphorically exemplifies or expressesits expressive features. Aesthetic appreciation is a matter of being able todiscriminate among what the work represents, exemplifies, expresses, andso on.

Exemplification, whether literal or metaphorical, is a kind of reference(of a sample to a label):

a sample a exemplifies a (label) predicate b iff b denotes a and a refersto b.

Expression is metaphorical exemplification:

a sample a expresses a (label) predicate b iff b metaphorically denotesa and a refers to b.

Goodman summarizes his view of expression as follows:

if a expresses b then: (1) a possesses or is denoted by b; (2) this possession ordenotation is metaphorical; and (3) a refers to b.

(Goodman 1968: 95)

As is well known, Goodman uses the tailor’s swatch as his exampleof exemplification: the swatch possesses many qualities (yellow, tweedy,

² This is an oversimplification. Not every subject exists and those which do not cannot be denoted,according to Goodman.


square), but it refers to only a subset of them (yellow and tweedy, but notsquare). The swatch exemplifies those qualities that it both possesses andrefers to. In the symbol system of swatches, a swatch refers to its color andtexture but not to its size and shape. Similarly, a work of art possesses manyqualities, only some of which it refers to and hence exemplifies. Expressivequalities are qualities possessed and referred to by a work of art, but theydiffer from exemplified qualities in being metaphorically rather than literallypossessed.³ According to Goodman’s theory of metaphor, when a term fromone realm, such as the word ‘‘sad’’ which literally applies to the realm ofhumans and their psychological states, is transferred from its home realm(human emotion) to a new realm (music), there is a transfer not only of theterm itself (‘‘sad’’) but also of the whole family of terms (emotion words) towhich it belongs. The new realm (music) is then organized and structuredby the application of the new set of terms. What counts as sad and as happyin the new realm is partly a function of how these terms functioned in theold realm. Terms which literally apply to human psychological states are notthe only terms which can be metaphorically applied to music. Goodmanexplicitly allows for the expression of non-psychological qualities such as‘‘weight’’ (said to be expressed by Daumier’s Laundress) (Goodman 1968:87) and ‘‘fluidity’’ (offered as an example of what a building could express)(Goodman 1968: 91).

Goodman’s theory is at once suggestive and schematic. Notoriously, hesays very little about what he means by a work’s ‘‘referring to’’ the qualities itexemplifies and expresses. I myself think that he means that certain qualitiesin a work are more aesthetically significant (although not necessarily moresalient) than others and that it is these qualities that the work exemplifiesand expresses rather than merely possesses.⁴ On this interpretation, if a pieceof music expresses sadness, this means for Goodman that it is metaphoricallysad and that its sadness is one of its more aesthetically significant properties

³ Some commentators have thought that Goodman says that expressed properties are metaphoricallypossessed and metaphorically referred to, but I think it is clear that Goodman thinks the referenceinvolved is real and literal; he never characterizes a work of art as ‘‘metaphorically referring to’’ what itexpresses.

⁴ Goodman characteristically talks as if languages and symbol systems can be understood in abstractionfrom their users, so he says that it is the swatch and the work of art that refer, not the person handlingthe swatch or experiencing the artwork. Furthermore, it is strictly speaking inaccurate to say, as Ihave done, that on Goodman’s view a work exemplifies and expresses qualities. For Goodman, whatis expressed and exemplified is always only a label, such as a predicate. For more discussion of theseissues, see Robinson (2000).


(it ‘‘refers to’’ its sadness in this sense). In short, the expressive qualitiesof music are qualities metaphorically possessed by the music which areaesthetically significant and which are describable by predicates which areat home in some realm of human experience beyond the music.

At the same time, however, his suggestion that the music refers to orsignifies those qualities that it metaphorically possesses, (or, strictly, thepredicates that denote it), also encourages a rather different idea of musicalexpression that Goodman would probably reject: the idea that expressivemusic refers to or signifies not just predicates or musical qualities, but otherphenomena—events, actions, ‘‘plots’’—in realms of human experience‘‘beyond the music.’’ As we will see later, this idea is in many ways morefruitful than Goodman’s ‘‘official’’ theory of expression.

2

Goodman’s theory of expression has never been very popular. Nevertheless,it must be admitted that he has pinpointed something very importantin critical discourse: the way that critics often resort to metaphoricaldescriptions in order to characterize a work of art. The trouble is that criticsuse metaphors in their critical discourse in order to describe many aspectsof a work, not just what it expresses. In a discussion of the language usedby art historians, Carl Hausman has pointed out that figurative languageis often used to ‘‘direct appreciation’’ (Hausman 1991: 125)⁵ of individualworks, and gives as an example Svetlana Alpers’ description of the point ofview in a particular Dutch painting:

When figures enter they are captives of the world seen, entangled Gulliver-like inthe lines of sight that situate them.⁶

Similarly, in describing the color in a Chardin painting, Michael Baxandallsays that among the striking color devices, ‘‘the most obvious is thered-lacquered table assertive, but almost unstable.’’⁷ Critics standardly usemetaphors to pick out important qualities in works of art. However, thequalities they pick out may be representational, as in the Alpers example, or

⁵ Notice that the use of metaphor is one of the methods suggested by Sibley (1959) in his classicpaper ‘‘Aesthetic Concepts’’ for directing attention to aesthetic qualities in works of art.

⁶ Cited by Hausman (1991: 124). ⁷ Cited by Hausman (1991: 119).


formal, as in the Baxandall case. They are not necessarily what are usuallycalled expressive qualities. Consequently, it would seem that there is nonecessary connection between being described metaphorically and being aquality that is expressed by a work of art.

Stephen Davies has criticized Goodman on different grounds. He thinksthat the qualities expressed by works of art are possessed not metaphoricallybut literally. Thus for Davies descriptions such as ‘‘sad music’’ and ‘‘mourn-ful music’’ are literal rather than metaphorical, and furthermore they are‘‘ineliminable’’ in the sense that the properties they refer to cannot beindicated in any other way. He distinguishes such ineliminable metaphors,which are ‘‘mandated by the nature of the music itself ’’ (Davies 1994:146), from metaphorical descriptions of features of music that could beeliminated in favor of a literal description of the same feature.

Davies suggests that Goodman is committed to the view that there is adistinction between ‘‘the case in which the music’s possession of a featureis metaphoric and that in which a description of the music is metaphoric’’(Davies 1994: 146). He thinks that the idea that expression is metaphoricalexemplification ‘‘suggests that its being metaphoric is a feature of themanner in which music possesses or presents the property in question’’(Davies 1994: 146), and that this in turn suggests that perhaps music itselfcan be metaphoric rather than the language used to describe it. Daviesthen argues that this way of making the distinction is misleading; it shouldrather be thought of as a distinction between two kinds of metaphoricaldescription of music, ineliminable descriptions (which later turn out to beliteral, not metaphorical at all) and eliminable descriptions: ‘‘for musicalmetaphor the use of the metaphor is ineliminable, which is not so for(merely) metaphoric descriptions of music’’ (Davies 1994: 147). So, forexample, the term ‘‘sad,’’ which appears to be metaphorical when appliedto music, is ‘‘ineliminable,’’ whereas other metaphorical descriptions ofmusic are ‘‘gratuitous’’ or ‘‘eliminable.’’⁸

Davies gives as an example the entry of the soloists in the first movementof Mozart’s Symphonia Concertante for Violin and Viola, K.364, whichcan be metaphorically described as ‘‘homing pigeons, first perceived asdots in the distance, which, on drawing near, flutter to their dovecote’’

⁸ Budd (1989) has argued that ‘‘sad’’ in ‘‘sad music’’ is ineliminable, and for this very reason cannotbe metaphoric.


(Davies 1994: 146). The difference between the two sorts of case mightbe ‘‘dramatized,’’ says Davies, as a difference between ‘‘the metaphoricdescription of a literally possessed property and the literal description of ametaphorically possessed property’’ (Davies 1994: 148). Davies says: ‘‘Anymusical feature might be characterized metaphorically,’’ but ‘‘some musicalfeatures seem to require metaphoric description in expressive terms, for theycannot be indicated readily except by such words as ‘sad’’’ (Davies 1994:146). The idea that the term ‘‘sad’’ is ineliminable when applied to musicsuggests that only this term can adequately capture a particular quality inthe music: ‘‘sad’’ is the unique term which can do this. By contrast, the useof the homing pigeons metaphor ‘‘might be regarded as gratuitous, sincethere are many other, equally informative accounts’’ that could be offeredof the passage (Davies 1994: 146).

There are a number of problems with this discussion. First of all, thereis no evidence that Goodman wants to distinguish between ‘‘a musicalproperty’s being metaphoric and a musical property’s being describedmetaphorically’’ (Davies 1994: 146). Since Goodman thinks that only labelscan be exemplified or expressed, there is no room in Goodman’s ontologyfor such a distinction. A ‘‘metaphoric musical property’’ for Goodman can’tbe anything other than a property named by a label used metaphorically.

Second, even if there is a distinction between ineliminable and eliminablemetaphorical descriptions of music, I do not think that Davies has givena plausible account of this distinction. It is true that if the music expressessadness, then on Goodman’s account, given his nominalism, the term ‘‘sad’’is ‘‘ineliminable,’’ since the music both is sad and refers to the label ‘‘sad.’’It does not for Goodman refer to any synonyms of ‘‘sad,’’ since Goodman issuspicious of the whole notion of synonymy. Davies, however, is not com-mitted to any such view. If we admit synonymy, then I think we have to saythat the term ‘‘sad’’ when applied to music is not strictly ‘‘ineliminable.’’ Ifthe music is sad, then its quality of sadness may be captured by several syn-onymous or near-synonymous terms such as ‘‘mournful,’’ ‘‘melancholy,’’and ‘‘miserable,’’ by similes (it is like a woman weeping and sighing) and byfurther metaphors (it is a cry of woe and despair). Each of these terms picksout—with greater or less precision and appropriateness—the sad qualityin the music. So the term ‘‘sad’’ itself is not ‘‘ineliminable,’’ although therange of suitable replacements is confined to synonyms or near synonymsthat refer to a similar kind of emotional experience.


What is arguably ‘‘ineliminable’’ is that each of these descriptions iscouched in terms of human emotional experience. Each of them invitesthe listener to hear the music in terms of human psychological states. Itdoes not matter very much whether the term is literal (‘‘sad’’ maybe) ormetaphorical (it is a cry of woe); both these descriptions invite the listenerto hear the music in terms of human emotion. We cannot ‘‘eliminate’’either description by a translation into non-emotional terms: meaning is lostwhen we say ‘‘the music is played slowly in the minor key on the cellos.’’Nor can we eliminate ‘‘sad’’ in favor of a different kind of metaphor: ‘‘darkmusic,’’ for instance, invites us to hear it in terms of a different realm ofexperience.

Davies also wants to say that descriptions such as ‘‘sad music’’ areliteral rather than metaphorical. Certainly if ‘‘sad music’’ is indeed ametaphor, it is arguably a pretty moribund one. Even if Davies wereright about ‘‘sad music,’’ however, we cannot generalize this conclusionto all descriptions of expressive qualities. Many of the terms we use todescribe the expressive qualities of music are by anyone’s standards usedmetaphorically: in describing music as weighty or fluid, fearful, anxious, orangry, metallic, watery, or ethereal, roaring or twittering, sunny or breezy,stealthy or heavy-footed, we use metaphors to bring our experience of themusic into contact with other areas of our lives. All of these predicatesare describing the musical surface in terms drawn from some other realmof human experience, and can bring us to see some new and unexpectedquality in the music. If I am brought to hear the belligerent quality in themusic, I have been brought to hear something new (to me) in the musicthrough the application to music of a term from a ‘‘foreign realm.’’⁹

Davies contrasts the term ‘‘sad’’ which he says is ineliminable, with hishoming pigeons story, which, he claims, is ‘‘eliminable’’ or ‘‘gratuitous,’’since ‘‘there are many other, equally informative accounts [he] might offerof the passage’’ (Davies 1994: 146). But the two examples are more alike thanDavies acknowledges. The mere fact that there are other ways of describingthe same passage does not entail that the homing pigeons metaphor is‘‘eliminable.’’ It is only if there were other equally effective ways of picking

⁹ Here I skirt a number of issues about aesthetic qualities that are not germane to my main purposein this essay. For example, I have not discussed whether moribund metaphors such as ‘‘sad’’ whenapplied to music have acquired secondary literal meaning rather than metaphorical meaning. Nor am Icommitted to any particular position on the question of realism vis-a-vis aesthetic qualities.


out the same quality that the metaphor would arguably be ‘‘gratuitous.’’However, saying that the soloists enter softly together, gradually get louder,and then stop at the same time is not as informative as the metaphoricaldescription. The quality picked out by the metaphor is the way in whichthe soloists gradually approach from a distance and eventually land gentlytogether. Perhaps the doves and dovecotes are eliminable (although thechoice of bird is not gratuitous, since pairs of doves have often been taken asmodels of love and fidelity, as in La Fontaine’s fable, Les Deux Pigeons), butwhat is not eliminable—or otherwise describable—is the companionablequality of the soloists’ journey and their arrival together. The descriptioninvites us to hear the music in terms of the human experience of travellingand coming home. Indeed, much music can be conceived of in this way,as either literal or psychological journeyings. Any attempt to get at thisquality in the music must involve the idea of journeying. If the two soloistsdo not arrive by motor-boat or pull up in a pick-up truck, this is becausethe music suggests a certain quality in the arrival which is captured by theimage of gentle, companionable flight.

I conclude that Davies is wrong to assume that sadness is somehow ‘‘in’’the music, whereas the gentle arrival of the soloists is just a fanciful way ofdescribing a fact about the music that could be described in many otherways. Both are descriptions of how we experience the music; both can bereplaced without loss of meaning only by synonyms or near synonyms. Theequation of musical metaphor with ineliminable and hence literal linguisticdescriptions of music fails because many clearly metaphorical descriptions(‘‘fiery,’’ ‘‘breezy,’’ and ‘‘belligerent’’ as well as the gentle arrival) are justas ineliminable as moribund metaphorical or literal descriptions such as‘‘sad.’’¹⁰

There is an interesting moral to be drawn from Davies’s homing pigeonstory, however. I suspect that the reason why it seems adventitious to him isthat he is talking about a piece of ‘‘pure’’ music by Mozart, rather than, say,a tone-painting by Vivaldi. One can easily imagine a musical representationof the flight of two fond birds by a violin and a viola playing in the wayDavies describes. What is really ‘‘gratuitous’’ about the story of the homing

¹⁰ Frank Sibley has distinguished three kinds of metaphorical description, which are often applied tomusic: (1) ‘‘‘scenarios’, more or less extended pictures, programs, or narratives: e.g. ‘A Landler dancedby ogres’’’; (2) ‘‘descriptions in terms of substances and processes’’; and (3) descriptions ‘‘in terms ofqualities’’ (Sibley 1993: 167). According to this categorization, the homing pigeon story is a ‘‘scenario.’’


pigeons in the Mozart example is that, while the metaphor does indeed getus to hear qualities present in the music—the quality of cordial companionshipon a journey, the sense of a gentle landing—there is no particular reasonfor hearing the piece as a whole in terms of the journey metaphor. There isno reason for thinking that the piece is about or refers to traveling homingpigeons. The metaphor can get us to hear qualities in the musical surface,but it does not explain the underlying structure of the piece in any way.

It is difficult to describe the way music sounds, just as it is difficult todescribe the way a painting looks. When critics say such things as ‘‘Thissounds like two homing pigeons coming from afar and landing gentlytogether,’’ or ‘‘This passage sounds silken, even though it’s played on thebassoons,’’ they are trying to help listeners identify how the music sounds,particular qualities of melody, harmony, rhythm, dynamics, orchestrationor timbre that may be hard to describe literally. Significantly, however,such remarks carry no implications that the music is about or refers to orexemplifies homing pigeons or silk, or, more generally, birds or materials.Indeed formalist critics are as inclined to describe music in such terms asanyone else.¹¹

If, however, a piece of music really is about two homing pigeons on ajourney or about materials or textures, that will affect the kinds of descriptionthat seem to be appropriate to the music. A piece entitled ‘‘Textures’’ invitesdescription as ‘‘silken’’ or ‘‘rough-hewn.’’ A piece about journeying birdsinvites description in journey-metaphors. In general, what the structure ofa piece of music is conceived of as exemplifying as a whole may well affectthe choice of metaphorical descriptions it invites of its parts. If a pieceof music is thought of as metaphorically an outpouring of feeling, then itmight seem peculiarly appropriate to call it ‘‘mournful’’ or ‘‘tender.’’ If itis thought of overall as metaphorically a struggle between opposing forces,then ‘‘belligerent’’ might be an appropriate term to use. If it is thoughtof as an actual or psychological journey, then it would seem just right todescribe it in terms of gradual approach and gentle landing.¹²

¹¹ Even arch-formalist Eduard Hanslick uses very flowery, metaphorical language in his musiccriticism, while at the same time insisting that he is merely describing the character of the music itself,not saying that the music represents or expresses any quality in the ‘‘extra-musical’’ world. E.g. incriticizing Bruckner’s Eighth Symphony, Hanslick writes of its ‘‘muddled hangover style’’ and of ‘‘thesheer weight and monotony of this interminable lamentation.’’ (Hanslick 1978: 288–9).

¹² It could be argued, however, that almost any piece of Western tonal music lends itself tointerpretation in terms of the journey metaphor. Contrast, say, Balinese gamelan music, which is not


Of course many pieces of ‘‘pure’’ music do not announce themselvesas about any specific things or events. Nevertheless there may be goodreasons why we conceive of a piece of music as a whole as metaphoricallyexemplifying some specific theme or other. How we decide which over-arching metaphors are appropriate to a particular piece is an issue I addressat length in Section 4.

3

Anthony Newcomb (1984a) sees some important advantages in Goodman’stheory of expression. He emphasizes that Goodman’s account maintainsthe ‘‘transparency’’ of the musical symbol: in calling a piece of music sad orfiery, we are focusing on properties of the music, even though the music isbeing described in terms of properties of experience ‘‘beyond’’ the music.

The two-way reference posited by exemplification seems to fit much betterthe flow of reference back and forth between musical objects and expressivemeaning—between extramusical and intramusical patterns—that is at the heart ofmusical expression.

(Newcomb 1984a: 623–4)

So while there is indeed reference by the music to the ‘‘extramusical,’’ thisdoes not mean that our attention is diverted from the music to somethingelse. The music refers to an ‘‘extramusical’’ process, but we can also hearthis ‘‘extramusical’’ process in the musical process.

Another advantage that Newcomb sees in Goodman’s theory is that it isnot only emotional qualities that can be expressed. Musicologists have oftenfound philosophers’ exclusive concern with sadness and other emotionalterms to be too narrow, and have pointed to the wealth of terms actuallyemployed by music critics to describe music. As we have just seen, manydifferent kinds of qualities can be exemplified by music. Besides being sad,happy, serene and sentimental, music can exemplify qualities of the weather(stormy, sunny, misty, breezy), natural substances (silken, metallic, fiery,ethereal, watery), animal sounds (twittering, growling, roaring), human and

readily describable as traveling towards or seeking a goal. Again, many pieces of Western tonal musiclend themselves to being heard as expressive of emotion, hence perhaps the ubiquity of emotion wordsto describe musical qualities.


animal movement (stealthy, clumsy, limping, on tiptoe) and many more.There would seem to be no theoretical limit on the kinds of qualities thatmusic can exemplify.

Newcomb’s own view of expression is that it results from:

the metaphorical resonances or analogies that a viewer-listener-reader findsbetween properties that an object possesses and properties of experience out-side the object itself. Thus expression results from intrinsic properties of an artworkbut also from the metaphorical resonances these properties may have for theperceiver.

(Newcomb 1984a: 625)

These ‘‘resonances’’ are ‘‘often broadly intersubjective’’ among a ‘‘class oflisteners,’’ those familiar with music in this style; ‘‘expressive interpreta-tion ... is concerned with how the [intrinsic] properties are connected to theresonances for a class of listeners’’ (Newcomb 1984a : 625). As Goodmaninsists, it is the critic’s task to decide which of a work’s properties itexemplifies and expresses and, further, to bring ‘‘these qualities into therelationships and patterns that make the best argument for the work athand’’ (Newcomb 1984a: 626).

Although Newcomb seems to endorse Goodman’s theory of expression, Ido not think that the theories are entirely consistent. First of all, Goodmanwould have nothing to do with the idea of expressive ‘‘resonances’’suggested to listeners who then interpret a piece of music in terms ofthese perceived ‘‘resonances.’’ For Goodman, interpreters are restricted tofinding the qualities that a work of art actually has. By contrast, Newcombseems to think that different classes of listeners will find different resonancesin a piece and thereby arrive at different interpretations, all of which canbe justified and argued about.¹³

Secondly, Newcomb emphasizes the importance of large-scale musicalmovement, or as he puts it, ‘‘process or overall form’’ (Newcomb 1984a:626) rather than musical detail, so that a whole symphony or movementof a symphony might express perseverance through adversity or struggleto victory.¹⁴ Possibly Goodman’s theory can cope with these examples,

¹³ This seems to me an advantage to Newcomb’s way of thinking about these things, but I will notpause here to discuss rival theories of critical interpretation.

¹⁴ These are examples of ‘‘plot archetypes’’ in Newcomb’s terminology. See Newcomb 1984b.


but this is not obvious. Goodman’s own examples suggest that the labelsexpressed by artworks are mostly simple terms like ‘‘sad.’’

Thirdly, although Newcomb does not develop this theme, it seems tome that Newcomb’s discussion of how an extended piece or passage ofmusic can be expressive of some large-scale process points the way to amore important distinction between two somewhat different ways in whichmusic can express or metaphorically exemplify. On the one hand, musiccan exemplify—in Goodman’s sense—aesthetically significant expressivequalities that are emergent qualities of the ‘‘musical surface,’’ and on theother hand, some music also exemplifies what Newcomb calls ‘‘expressivemeaning,’’ the overall meaning of a work taken as a whole.

If we look at Goodman’s examples, we find him talking only aboutproperties of the ‘‘musical surface.’’ Terms like ‘‘mournful’’ or ‘‘growling,’’typically describe details of the way the music sounds, or, more generally ofthe character of the music, but they carry no implication that the music inquestion is about human emotion or animal noises. Sometimes, however,a piece of ‘‘pure’’ music does have an overall ‘‘expressive meaning.’’ Forexample, Beethoven’s Fifth can be described as being about ‘‘perseverancethrough adversity’’ or ‘‘struggle leading to victory.’’ On this view, themusic is not just a surface that has expressive qualities; the music as a wholerefers to and is about a process of perseverance through adversity or struggleleading to victory.

This does not mean that the music points away from itself and towardsperseverance and that paying attention to perseverance distracts us fromthe ‘‘music itself.’’ The music is a ‘‘transparent symbol.’’ As Newcombsays, when music is truly expressive of some meaning, such as ‘‘struggleleading to victory,’’ that meaning is (metaphorically) exemplified by themusic. The music has the same structure as the ‘‘extra-musical’’ reality itrefers to; it is isomorphic to these ‘‘extra-musical’’ processes. The musicdoes not merely refer to struggle leading to victory, as the words ‘‘struggleto victory’’ do;¹⁵ the music itself metaphorically struggles and we canexperience struggle in the music itself. Indeed we experience the drawn-out process of a struggle to victory both as a musical and as a psychologicalprocess. For example, one particular heroic-sounding theme might beheard as struggling for predominance over another darker theme in a minor

¹⁵ In Goodman’s terminology, ‘‘struggle to victory’’ denotes a struggle that leads to victory.


key, and then over a period of interaction, the ‘‘hero’’ theme can beheard as triumphing over the darker minor theme. In such cases there isnot just one simple quality being exemplified such as ‘‘fiery’’ or ‘‘heroic,’’although perhaps these might be appropriate terms to describe the herotheme itself. Rather, what is (metaphorically) exemplified is an extendeddrama which gives expressive meaning to the piece as a whole. In short, themusic metaphorically exemplifies [a certain kind of] perseverance throughadversity, or struggle leading to victory.

Newcomb believes that the same piece of music has a wide ‘‘expressivepotential,’’¹⁶ which can be described—verbally—in many different waysby different listeners. Different expressive interpretations have to makesense of the music as a whole; each one has to illuminate transitions anddevelopments; each one has to fit together the many diverse elementsof the whole. One-word ‘‘labels,’’ such as ‘‘fiery’’ or ‘‘heroic,’’ althoughuseful for pointing out expressive qualities of the music that might need tobe mentioned in the context of an expressive interpretation, will never beadequate all by themselves to explain large-scale expressive meaning in music.

The distinction I am drawing may be what Leo Treitler has in mindwhen he contrasts what most people think of as one-word metaphori-cal descriptions of music (‘‘sad music’’) with the ‘‘metaphorlike effects’’(Treitler 1997: 47) possible when one considers a large-scale piece of musicas a whole.¹⁷ The second movement of Schubert’s E flat Piano Trio,D.929, opens with a ‘‘long discursive theme in C minor, accompaniedby a trudging figure’’ that ‘‘registers as the tattoo of a funeral march’’(Treitler 1997: 47). Later a second soaring theme enters, which through its‘‘vigorous, affirmative development,’’ seems ‘‘as though it might be capableof lifting the movement out of the hopelessness of the C-minor music butin the end always slumps back into it’’ (Treitler 1997: 47). Towards the endof the final movement, Allegro moderato, a rondo, in which the themeskeep repeating in ‘‘an endless chatter of mindless energy,’’ unexpectedlythe solemn tattoo theme from the second movement reappears on thecello. Treitler comments: ‘‘Suddenly it is as though this is what all thatpatter had been intended to suppress’’ (Treitler 1997: 48). Treitler doesnot give an analysis of these ‘‘metaphorlike effects’’ in music, but I would

¹⁶ This is a phrase Newcomb borrows from Cone (1974: 166).¹⁷ I say ‘‘most people think’’ of ‘‘sad music’’ as a metaphorical description, because Treitler, like

Davies, thinks ‘‘sad’’ is really literal when applied to music.


suggest that what he is talking about is an example of ‘‘expressive meaning’’in Newcomb’s sense: the Trio as a whole is experienced as having a‘‘narrative’’ dimension, in which the piece exemplifies a wider pattern ofhuman experience.

4

In his classic work on aesthetics, Aesthetics: Problems in the Philosophyof Criticism, Monroe Beardsley articulated what is still the most centraldifficulty in the theory that music exemplifies or is an iconic sign of extra-musical processes, whether psychological, physical, political, or whatever.Using Ravel’s Bolero as his example, Beardsley says that if it is:

an iconic sign of those psychological processes that have all its qualities, then ofcourse it is a sign only of itself. If it is taken to be a sign of processes that are similarto it in being characterized by increasing intensity, then we cannot tell simply fromhearing the music that it is these processes it signifies rather than others. The musicmust be supplemented by a verbal rule that prescribes which of its qualities are theiconically significant ones. And in the absence of such a rule, we cannot decideamong the innumerable possible qualities, so that if the music is a sign at all, it isambiguous.

(Beardsley 1958: 336)

Here in a nutshell is the problem with theories such as Newcomb’s: thevery same musical passage or movement may apparently exemplify hoppingup the stairs and down again, rising to heaven and falling to hell, feelinguplifted and then feeling dejected, a sunny day in the mountains followedby a gloomy evening in the motel. All these possibilities are consistent withthe ‘‘expressive potential’’ of the music. How are we to know which ofthem (if any) is the most appropriate in any given case? Without a ‘‘verbalrule,’’ as Beardsley puts it, how can this question be decided?

The short answer is that the question is decided by reference to thelarger context in which the music is embedded. If it is a whimsical piece oftone-painting describing the capering of small children, then hopping upand down the stairs may be what the piece is about; if it is a cantata aboutthe death and resurrection of Christ, then perhaps it should be heard asexemplifying rising to heaven and falling to hell; and if it is an expression of


personal feeling it might be appropriate to hear it as exemplifying feelingsof uplift followed by dejection. But then the question arises: if there is nohelpful title to guide us, how do we know whether the piece should beheard as a tone-painting, a meditation on the resurrection or a piece ofemotional expression?

My suggestion is that music as an art has historically been conceivedof in terms of certain large-scale or umbrella metaphors, what Lakoff andJohnson have called ‘‘conceptual metaphors,’’ that structure our musicalexperience (Lakoff and Johnson 1980). In their well-known example,‘‘Argument is war,’’ Lakoff and Johnson show how we structure ourexperience of argument in terms of this conceptual metaphor, thinking ofthe participants in an argument as taking sides, as defending and attacking,as winning or losing, and so forth. Similarly, music—or at least music in theWestern tradition—has always been understood in terms of one or anotherconceptual metaphor. The music Newcomb discusses, for example, comesfrom the repertory of nineteenth- and twentieth-century Romanticism.In various essays Newcomb has discussed Schumann’s Third symphony,Wagner’s music drama, Siegfried, and Mahler’s Ninth. The ruling metaphorsfor such works include music as psychological drama or story. A symphonyis not literally a play, but in the wake of the invention of opera, symphoniescould be thought of as wordless operas or dramas.

The big question, however, is when, if ever, is it appropriate toexperience ‘‘pure’’ music as metaphorically a drama or a story of somesort? My answer is that we need to find the answer in the history ofmusic and music reception. Different metaphors are current at differentperiods of music history for thinking about music. Formalism is itself atheoretical position that arose in particular historical circumstances andis more appropriate to some musical pieces than to others. We bestunderstand a piece of music if we understand how composers and theirculture conceived of their music, in terms of what conceptual metaphors.

To illustrate, I will briefly consider the period in the mid- to late-eighteenth centuries when the baroque period gave way to the classical,the polyphonic style gave way to the newly invented sonata form, and thesuite—derived from dance forms, such as the sarabande, minuet and giguefamiliar from Bach’s suites—gave way to the symphony and the stringquartet. Scholars have defended a number of different views about howthis new ‘‘absolute’’ music was conceived of by its contemporaries, each


of which relies on a different conceptual or structural metaphor, such asmusic as mathematical structure and music as discourse. My goals here aretwofold. First I want to point out how all these different views assumethat the musicians and theorists of this period thought of music in termsof certain recurrent structural metaphors.¹⁸ Second, I want to emphasizethat one of the most widely accepted of these structural metaphors at thisperiod was the metaphor of music as discourse.

For most Enlightenment thinkers purely instrumental music was despisedas trivial, a ‘‘mere twittering,’’ written ‘‘for the entertainment and idleenjoyment of the few’’ (Fubini 1991: 253). Before 1800 music was inex-tricably connected to words, and only music with words—especially therecently invented form of opera—could have any serious content or couldexpress ideas and emotions:

The older idea of music, against which the idea of absolute music had toprevail, was the concept, originating in antiquity and never doubted until theseventeenth century, that music, as Plato put it, consisted of harmonia, rhythmos,and logos. Harmonia meant regular, rationally systematized relationships amongtones; rhythmos, the system of musical time, which in ancient times included danceand organized motion; and logos, language as the expression of human reason.Music without language was therefore reduced, its nature constricted: a deficienttype or mere shadow of what music actually is.

(Dahlhaus 1989: 8)

Given such assumptions, the new ‘‘absolute’’ music without languagerequired justification.

John Neubauer has argued that the roots of this aesthetic shift lie inan old tradition going back to Pythagoras, according to which music is(literally) a brand of mathematics. In the medieval university curriculum,the quadrivium consisted of astronomy, geometry, arithmetic, and music.Neubauer argues that the revitalization of Pythagoreanism ‘‘motivated thechange from an aesthetics of affects and expression [the baroque] to an aes-thetics of structure [the classical]’’ (Neubauer 1986: 194). He cites thinkerssuch as Rameau, who emphasized ‘‘a combinatorial and permutationalprinciple [of music], which allows the generative construction of chordsand their cadential progression within the framework of the major-minor

¹⁸ Cf. Cook (1990: 4): ‘‘a musical culture is, in essence, a repertoire of means for imagining music.’’


tonal system’’ (Neubauer 1986: 83), as well as a deductive principle accord-ing to which music derived from natural (i.e. non-conventional) harmoniclaws. Music, as Pythagoras claimed, reflects the mathematical constitutionof the universe. According to Neubauer, this strand of thinking continuedthrough the eighteenth century and laid the groundwork for the idea ofmusic as an autonomous art form, distinct from poetry and the poetic goalsof expression and imitation.

I find this account unconvincing—or at least incomplete—for a numberof reasons. First, although it is true that the idea of music as fundamentallyconnected to mathematics is present throughout the history of Westernmusic, having more or less influence at different periods, yet at the timeof the new ‘‘classical’’ style, the ideas of Enlightenment figures such asRousseau were in the ascendant, and Rousseau explicitly denigrated musicwithout words because he thought that, unlike poetry, it had no content.Such proto-Romantic figures had little interest in music’s mathematicalproperties. Second, if we think of early masterpieces in the new classicalstyle, such as Haydn’s early string quartets, although they do exemplifysymmetry and proportion, their dramatic, human qualities are just as strikingas their carefully crafted form.

A theory I find more plausible has been developed by Mark Evan Bonds,who argues that the most important metaphor for understanding earlyventures in sonata form is the oration: a speech, a piece of rhetoric. Musicwas thought of as a discourse that seeks to persuade and emotionally affectits audiences. In his book, Wordless Rhetoric: Musical Form and the Metaphorof the Oration, Bonds traces the idea of music as rhetoric throughout theeighteenth and early nineteenth centuries, and shows how this temporalmetaphor—of a development of thoughts over time—eventually, in thenineteenth century, gives way to the spatial metaphor of music as anorganism: music as a living thing rather than a series of events. In theeighteenth century:

The instrumental work was seen as a wordless oration, and its form was viewed notso much as a harmonic or thematic plan but as an ordered succession of thoughts.

(Bonds 1991: 53)

Bonds argues that sonata form was understood by contemporaries onthe model of an oration with a Hauptsatz, a main idea, which was


then elaborated in various ways; the way the central idea was elaborateddetermined the structure of the movement as a whole.

Whether or not sonata form was thought of on the model of anoration, it is clear from numerous eighteenth-century sources that musicwas indeed regarded (metaphorically) as a rhetorical discourse of some sort.Enrico Fubini notes that the idea of sonata form as a reasoned discoursewas a commonplace of contemporary thinking. He suggests, however,that rather than thinking of it as an oration, it is more illuminating tothink of the rise of sonata form as a narrative form comparable to thenewly burgeoning literary art form of the day, the novel. Haydn, whomore than anyone else, was responsible for this new musical language,wrote music that can perhaps best be conceived in this way. ‘‘Haydnhad the gift of creating musical discourse and conversation’’ (Fubini 1991:255), which is heard to best advantage in his string quartets. In contrastto baroque counterpoint and polyphony, ‘‘a substantially repetitive stylein which variety was obtained through ornamentation and contrastingtimbres and dynamic levels,’’ Haydn’s sonata form ‘‘is narrative and notconfrontational; it seems to be offering in modern, Enlightenment termsthe ideal of a solid musical language which might lead to the most excitingspiritual adventures, and which is capable of allowing the expression ofthe most sophisticated emotions within a logical structure which can beunderstood by any reasoning person’’ (Fubini 1991: 257).

Similarly, in discussing the development of the classical symphony,Charles Rosen emphasizes the influence of opera, and comments that theclassical symphony should be thought of as (via the metaphor of) a drama:

The application of dramatic techniques and structure to ‘‘absolute’’ music wasmore than an intellectual experiment. It was the natural outcome of an age thatsaw the development of the symphonic concert as a public event. The symphonywas forced to become a dramatic performance, and it accordingly developed notonly something like a plot, with a climax and a denouement, but a unity of tone,character, and action it had only partially reached before. Unity of action was, ofcourse, one of the classical requirements of tragedy, and the symphony as dramagradually abandoned every trace of the looseness of the [baroque] suite.

(Rosen 1972: 155)

With Beethoven and on into the next century the dramatic possibilitiesof sonata form were fully exploited, and symphonies and string quartets


became wordless mini-operas, full of dramatic conflict in which differentcharacters are pitted against each other. Moreover, with the Romanticmovement came the notion that art in general involves the outpouring offeeling; the metaphor of pure instrumental music as drama fits nicely withthe metaphor of music as psychological drama, and the metaphor of musicas psychological exploration.

In summary, the various theories about how the new ‘‘absolute’’ musicwas conceived of each rely on an overarching conceptual or structuralmetaphor, such as music as mathematical structure or music as discourse,and among these metaphors the most widely accepted seems to have beenmusic as discourse, including music as oration, music as novel or narrative,music as conversation, and music as drama.¹⁹

I now want to suggest that unless we grasp the metaphors that structurea piece of music, we do not hear it as it was meant to be heard and weare likely to miss some of its most important expressive qualities. Thus, ifwe think of Haydn’s music as merely pretty tunes or pleasingly organizedmusical structures, we miss some of its most important qualities, whichdepend upon our hearing it in terms of the metaphor of ‘‘string quartetas civilized conversation.’’ Haydn’s string quartets are often said to be‘‘good-humored,’’ ‘‘conversational,’’ and ‘‘full of give and take.’’ If wedo not hear them as conversations, however, it is unlikely that we willhear those expressive qualities in the music. It is only because we hearthe music in terms of the metaphor of conversation (we experience itas exemplifying—referring to and sharing a structure with—a particularkind of conversation) that we are able to detect the good humor and theconversational quality of the music. We have to hear the instruments as‘‘voices,’’ as interrupting each other, as denying or confirming what eachother ‘‘says,’’ as repeating with emphasis or repeating with irony; as tossingideas back and forth; as commenting amusingly on each other’s ideas, andso on and so forth. This whole way of listening to a string quartet isgoverned by the conceptual metaphor: ‘‘this music is a conversation.’’ Andgiven the wit with which Haydn develops the idea of string quartet as

¹⁹ Romanticism and post-Kantian idealism also spawned other conceptions of music. Some post-Kantian idealist philosophers thought of music as having metaphysical meaning. Schelling held that artis the expression of the Absolute or the Infinite, and Schopenhauer claimed that pure instrumentalmusic is the representation of the Will, the force or energy that constitutes metaphysical reality andforms the basis of all of nature.


Allegro

5

10

Example 1. Haydn Quartet in E Major Op. 54 No. 3, opening bars (mm. 1–14)

conversation, it is highly unlikely that the reference to conversation in thismusic is merely fortuitous.

So, for example, we find Charles Rosen describing the opening ofHaydn’s E Major Quartet Op. 54 No. 3 (Example 1) in terms of a ‘‘sociablecomedy’’ (Rosen 1972: 141). The second violin and viola begin but areinterrupted by the first violin with a new phrase; the second violin andviola try again and are again interrupted. Then, ‘‘the second violin andviola, resigned, give up their phrase and accept the first violin’s melody;begin it—and are again comically interrupted’’ (Rosen 1972: 142). Thepoint is that there would be no comedy if we did not hear the music asdiscourse or conversation among four voices. Haydn’s wit consists in partin how he exploits the idea of the string quartet as conversation.

Now, it may be that formalists will claim that music can be heard aswitty and conversational even if they deny that it refers to or exemplifiesa conversation. Certainly, we can hear music in terms of any kind ofscenario we choose, just as Davies suggests it is possible to hear a passagein the Mozart Symphonia Concertante in terms of the companionable flight


of two homing pigeons. The homing pigeon story gets us to hear a certainquality in the music even though it is not appropriate to hear the music asexemplifying or referring to the journey of two homing pigeons. However,it is hard to imagine hearing a conversational quality in a quartet withoutexperiencing the quartet as a conversation. Moreover, it is impossible tohear Haydn’s treatment of the conversational aspect of the quartet as wittyunless you hear this conversational aspect.

More importantly, the difference between the homing pigeons in Mozartand the conversation in Haydn is that whereas it is inappropriate to hear astory about homing pigeons in the Mozart piece, hearing a Haydn StringQuartet as metaphorically a conversation is an appropriate way to hearit, authorized by the way in which it was received by contemporary andlater audiences and was probably thought of by its composer. In short, thewitty, conversational quality of the Quartet Opus 54 No. 3 is part of theexpressive meaning of the whole, not just an adventitious description ofsomething in the musical surface. The piece itself is (partly) structured bythe metaphor of conversation and (metaphorically) exemplifies a particularconversation, one in which, among other things, two of the participantsare constantly interrupted by a third.

5

What does it mean to say that music is metaphorically a discourse? Withoutdiscussing and defending in detail any particular theory of metaphor, it isperhaps fair to say that there are two main types of theory: on the onehand theories which claim that metaphors involve a transfer of meaningand, on the other hand, the view associated with Donald Davidson,²⁰that metaphor is primarily a pragmatic rather than a semantic affair, thatmetaphors keep their primary meaning but simply induce us to changeour attitudes to what is being described metaphorically. If we are told thatmusic is metaphorically a conversation, we are being invited to think ofmusic as a conversation, just as Romeo’s calling Juliet the sun invites us tothink of Juliet as the sun. We are also invited to organize our experience ofmusic in a new way: to experience music as a conversation is to hear it as an

²⁰ See especially Davidson (1978).


interchange of voices, as a means of social communication, in which eachvoice listens to the others, comments on the others, interrupts the others,echoes the others, accepts or denies what the others assert and so on.

Second, in being told that music is metaphorically a conversation, weare being encouraged to experience the same feelings and attitudes to themusic that we experience towards conversation, just as calling Juliet thesun invites us to experience the same attitudes to Juliet that we alreadyhave towards the sun. Romeo’s calling Juliet the sun reflects his new anddifferent experience of Juliet: she is the center of the universe, the brightestthing in all the heavens, the giver of light and life, and so on. Similarly,our experience of music as a conversation reflects a new and differentexperience of the music; for example, we experience it as civilized andcivilizing, as witty, as a synthesis of reason and imagination (Fubini 1991:256). Our attitude towards the music changes when we think of it as aconversation.

When we say that a is metaphorically b, the focus is on a and our changedattitude to a. At the same time, our attitude to b also shifts a little by beingbrought into juxtaposition with a (Black 1962). So the sun seems to be aless distant entity, more intimately concerned with human affairs, perhaps,when considered as an appropriate description of a young and charminggirl. Similarly, when we consider music as some sort of discourse, our mainattention is on the music, but we may also notice our attitudes to discoursechanging as well, as we think of it as potentially musical.

Third, metaphors, as Arthur Danto has pointed out, are intensionallyopaque (Danto 1981: 188–9). To say that ‘‘No man is an island’’ is notto say that no man is a body of land completely surrounded by water.Similarly, to say that music is a discourse does not mean that music is aset of sentences in a natural language uttered by a speaker or speakers. Itis quite beside the point to object, as the formalist might, that a stringquartet isn’t really a conversation, because in the quartet no-one is reallyuttering sentences or making assertions. We all know this. The music isonly metaphorically a conversation. The metaphorical description simplyencourages us to have a certain sort of experience of the music.

If I am right to accept the view that at the beginning of the classicalperiod, ‘‘pure’’ instrumental music in sonata form was widely thoughtof as a kind of discourse—as an oration, or a story, a conversation ora drama—and that these ways of thinking about music were embedded


in corresponding metaphors, the question arises: is it correct to say thatparticular pieces of music can be thought of as themselves metaphors for‘‘extra-musical reality’’? If Beethoven’s Fifth is metaphorically a drama, canit perhaps be a metaphor for struggle leading to victory? If Haydn’s Opus 54quartets are metaphorically conversations, can they also serve as metaphorsfor friendly companionship?

The short answer to this question is ‘‘No.’’ Because a piece of ‘‘pure’’ orabsolute music does not have any literal meaning, it cannot function as ametaphor. Why, then, do some people have such a powerful inclination totreat music as metaphor? One reason is probably that we think of music interms of the metaphors that apply to it, in the ways I have been discussing:on the one hand broad over-arching conceptual metaphors—music is adrama, music is a conversation—and on the other hand metaphors thatcharacterize details of the musical surface, such as music as ‘‘anxious,’’‘‘stormy,’’ ‘‘belligerent’’ and so on.

However, to describe music metaphorically is not the same thing as totreat a piece of music as itself a metaphor. We talk metaphorically about lotsof things without wanting to say that those things are themselves metaphors.Just because a friend is a jewel, we don’t feel inclined to say that the friendherself is a metaphor for jewelry. When people say that music is a metaphorfor something, or—more circumspectly, like Newcomb—that music has‘‘metaphorical resonances,’’ I think they mean simply that the piece, whenconceived of metaphorically as a certain kind of process, thereby exemplifiesa process or processes in the ‘‘extra-musical’’ world, i.e., it refers to thoseextra-musical processes and it is structurally isomorphic to them.²¹ Forexample, if we think of a Haydn string quartet as a conversation, thenwe can experience the instruments of the quartet as exemplifying voicesin conversation, agreeing, disagreeing, commenting wittily, uniting andresolving their disagreements, and so on. The quartet is not a metaphor forconversation; it is metaphorically a conversation, and it exemplifies a partic-ular (kind of) conversation: it refers to this conversation and is isomorphicwith it. Similarly, if a Beethoven symphony is metaphorically a drama, thenit can be experienced as a particular (kind of) drama in which, for example,

²¹ Even the formalistically inclined Malcolm Budd suggests a space for exemplification in musicwhen he allows that ‘‘the appreciation of music is infused with the perception of relations betweenparts as such relations, and ... these relations are not specific to music but obtain outside it’’ (Budd1995: 170).


there is a struggle to victory. The symphony exemplifies struggle to victory:it is about struggle to victory and it has the same structure as struggle to victory.

Instead of allowing, like Newcomb, that each listener may bring quitedifferent metaphorical resonances to a piece, however, I am insistingthat there are better and worse ways of metaphorically conceiving of aparticular musical work, and restrictions on what it is appropriate to hearit as exemplifying. It is only if the music is metaphorically a drama or anarrative, that it makes sense to hear it exemplifying a struggle to victory.At the same time, of course, there is room for multiple interpretations ofthe same piece, within the confines of the metaphor ‘‘music as drama.’’

So far I have suggested that Goodman’s account of expression asmetaphorical exemplification can be read both as an account of expressivequalities of parts of the musical surface, and—in Newcomb’s version—asan account of the ‘‘expressive meaning’’ of a piece as a whole, the wayin which some musical works metaphorically exemplify some story orsequence of events or process in the ‘‘extra-musical’’ world. Now, at leastsome works of art also contain expressions of emotion in a more literalway. Thus King Lear, expostulating with the elements on the heath, there-by expresses his rage and grief. The protagonist of Schubert’s Winterreiseexpresses his forlornness and despair. In concluding this essay I would liketo point out that if pure music is conceived of as metaphorically a kind ofdiscourse, it too can contain expressions of emotion of this sort.

If a string quartet is conceived of as metaphorically a conversation, thenit refers to voices in conversation. Voices are expressive and they canexpress particular thoughts, emotions, feelings and attitudes. Similarly, ifa symphony is metaphorically a drama, then it refers to characters and a‘‘plot;’’ characters too can express particular thoughts, emotions, feelings,and attitudes. Moreover, composers can express their emotions, attitudesand so on through the interplay of voices or characters that they create.Edward T. Cone and Jerrold Levinson have each suggested that in musicwe can hear a musical persona or perhaps a set of characters who behavesomewhat like the protagonists in a drama.²² If a symphony expressesnostalgia for an Arcadian past, which is threatened by a desire for aglitzy urban life (as in Newcomb’s reading of Mahler’s Ninth, second

²² Cone (1974); Levinson (1996); see also Karl and Robinson (1995). My most recent attempt toexplain musical expression is in Robinson (2005).


movement),²³ it is a persona or character in the music who is expressing thisnostalgia and sense of conflict. The character may be, as Cone says, someparticular musical element—a theme, a key, or a set of instruments—ratherthan the music as whole. So when we say that a (whole) piece of musicexpresses nostalgia for an idyllic past, we are probably saying somethingmuch more complicated than that the music (as a whole) is metaphoricallynostalgic and refers to that particular quality, as in Goodman’s theory. InNewcomb’s interpretation of the Mahler movement, he hears the music asrepresenting a character who is conflicted between the desire for a past stateof Arcadian innocence (represented by a theme from the first movement,which is recalled in the second), and the desire for a more sophisticatedand exciting urban life (represented by a seductive waltz theme).²⁴ In hisreading, this character expresses nostalgia for the past, and a desire for theglitzy present. Whatever you think of the details of his account, it showshow, if music is metaphorically a drama in which there are characters, thesecharacters can express their emotions and attitudes.

To conclude: formalists complain that we should not treat music asanything other than itself, not as language, and not as painting. If wedescribe music in metaphorical terms, it is only because it is so difficult tocapture the character of a piece of ‘‘pure’’ music. I have argued, however,that although metaphorical descriptions of the ‘‘musical surface’’ play animportant role in music criticism, there are more fundamental metaphorswhich do not just describe the surface of music but structure our wholeexperience of it. To hear music as metaphorically a drama is to hear it ascapable of referring to characters and exemplifying a plot; to hear musicas metaphorically a conversation is to hear it as capable of referring tovoices in conversation and as exemplifying the unfolding of a particularconversation. And once we hear characters and voices in the music, wecan also hear the emotions and attitudes that they express.²⁵

²³ Newcomb (1997).²⁴ Leonard Ratner has described how music exploits ‘‘topoi’’ which aid in expressive interpretation,

such as the landler, an ‘‘innocent’’ peasant dance, versus the waltz, a sophisticated ‘‘urban’’ dance. SeeRatner (1980), especially Chapter 2.

²⁵ I am grateful to Stephen Burton, Russell Dancy, Lydia Goehr, Kim Lockwood, Mary SueMorrow, Edward Nowacki, Kathleen Stock, Leo Treitler, and Nick Zangwill for useful commentsand/or discussion of these issues. This essay is a much revised and extended version of an article thatappeared in a special issue of the Revue Francaise d’Etudes Americaines 86 (2000): 77–89. I would like tothank the editors of the Revue for permission to adapt this article.


References

Beardsley, M. (1958) Aesthetics: Problems in the Philosophy of Criticism, NewYork: Harcourt, Brace and World.

Black, M. (1962) ‘‘Metaphor,’’ in J. Margolis (ed.), Philosophy Looks at theArts, New York: Scribner’s, 218–35.

Bonds, M. E. (1991) Wordless Rhetoric: Musical Form and the Metaphor of theOration, Cambridge: Cambridge University Press.

Budd, M. (1989) ‘‘Music and the Communication of Emotion,’’ Journal ofAesthetics and Art Criticism 47: 129–38.

Budd, M. (1995) Values of Art: Pictures, Poetry and Music, London: Penguin.Cone, E. T. (1974) The Composer’s Voice, Berkeley, Calif.: University of

California Press.Cook, N. (1990) Music, Imagination and Culture, Oxford: Clarendon Press.Dahlhaus, C. (1989) The Idea of Absolute Music, trans. R. Lustig, Chicago:

University of Chicago Press.Danto, A. (1981) The Transfiguration of the Commonplace, Cambridge, Mass.:

Harvard University Press.Davidson, D. (1978) ‘‘What Metaphors Mean,’’ Critical Inquiry 5: 31–47.Davies, S. (1994) Musical Meaning and Expression, Ithaca, NY: Cornell

University Press.Ferguson, D. N. (1960) Music as Metaphor, Westport, Conn.: Greenwood

Press.Fubini, E. (1991) The History of Music Aesthetics, trans. M. Hatwell, Lon-

don: Macmillan.Goehr, L. (1992) ‘‘Writing Music History,’’ History and Theory: Studies in

the Philosophy of History 31: 182–99.Goodman, N. (1968) Languages of Art, Indianapolis: Bobbs-Merrill.Hanslick, E. (1978) Hanslick’s Music Criticism, trans. and ed. H. Pleasants,

New York: Dover.Hausman, C. R. (1991) ‘‘Figurative Language in Art History,’’ in S. Kemal

and I. Gaskell (eds.), The Language of Art History, Cambridge: CambridgeUniversity Press, 101–28.

Karl, G. and Robinson, J. (1995) ‘‘Shostakovich’s Tenth Symphony andthe Musical Expression of Cognitively Complex Emotions,’’ Journal ofAesthetics and Art Criticism 53: 401–15.



Kivy, P. (1990) Music Alone: Philosophical Reflections on the Purely MusicalExperience, Ithaca, NY: Cornell University Press.

Lakoff, G. and Johnson, M. (1980) Metaphors We Live By, Chicago, Ill.:University of Chicago Press.

Levinson, J. (1996) ‘‘Musical Expressiveness,’’ in The Pleasures of Aesthetics :Philosophical Essays, Ithaca, NY: Cornell University Press, 90–125.

Neubauer, J. (1986) The Emancipation of Music from Language, New Haven,Conn.: Yale University Press.

Newcomb, A. (1984a) ‘‘Sound and Feeling,’’ Critical Inquiry 10: 614–43.Newcomb, A. (1984b) ‘‘Once More ‘Between Absolute and Program

Music’: Schumann’s Second Symphony,’’ 19th Century Music 7: 233–50.Newcomb, A. (1997) ‘‘Action and Agency in Mahler’s Ninth Symphony

Second Movement,’’ in J. Robinson (ed.), Music and Meaning, Ithaca,NY: Cornell University Press, 131–53.

Ratner, L. (1980) Classic Music : Expression, Form, and Style, New York:Schirmer Books.

Robinson, J. (2000) ‘‘Languages of Art at the Turn of the Century,’’ Journalof Aesthetics and Art Criticism 58: 213–18.

Robinson, J. (2005) Deeper than Reason: Emotion and its Role in Literature,Music, and Art, Oxford: Oxford University Press.

Rosen, C. (1972) The Classical Style, New York: Norton.Sibley, F. (1959) ‘‘Aesthetic Concepts,’’ Philosophical Review 68: 421–50.Sibley, F. (1993) ‘‘Making Music Our Own,’’ in M. Krausz (ed.), The

Interpretation of Music, Oxford: Clarendon Press, 165–76.Treitler, L. (1997) ‘‘Language and the Interpretation of Music,’’ in J.

Robinson (ed.), Music and Meaning, Ithaca, NY: Cornell UniversityPress, 23–56.

Further Reading

Bonds, M. E. (1991) Wordless Rhetoric: Musical Form and the Metaphor of theOration, Cambridge: Cambridge University Press.

Budd, M. (1995) Values of Art: Pictures, Poetry and Music, London: PenguinBooks.


Cone, E. T. (1974) The Composer’s Voice, Berkeley, Calif.: University ofCalifornia Press.


Goodman, N. (1968) Languages of Art, Indianapolis: Bobbs-Merrill.Kivy, P. (1989) Sound Sentiment: An Essay on the Musical Emotions, Philadel-

phia, Penn.: Temple University Press.Levinson, J. (1996) The Pleasures of Aesthetics: Aesthetics Philosophical Essays,

Ithaca, NY: Cornell University Press.Robinson, J. (ed.) (1997) Music and Meaning, Ithaca, NY: Cornell University

Press.Robinson, J. (2005) Deeper than Reason: Emotion and its Role in Literature,

Music, and Art, Oxford: Oxford University Press.

8

The Structure of Irony and Howit Functions in MusicE D DY Z E MACH A ND TA MA RA BA LT E R

1

Irony is an aesthetic property, that is, a feature of objects that is necessarilyrelevant to their aesthetic evaluation.¹ Musicologists hold that many worksof music have that property, and we agree. However, contemporarystudies of irony in music² tend to focus on one kind of irony³ or on asingle composer.⁴ Usually, they offer no general semantic theory of irony,showing how music can be ironic.⁵ In this paper we present a generaltheory that shows why there are many kinds of irony and how musicdisplays them, analyzing one example of each musically realizable kind ofirony. Like other analyses of irony in music we use texts, too, but onlyas supporting material, for if music can be ironic that property must liein the music itself. Our study also differs from that of most musicologistsin that we ground it in today’s philosophy (e.g., possible-world semantics)and not in philosophies of the composer’s time. We do not doubt thathistorical material is valuable in art interpretation, yet historical material,

¹ The concept of an aesthetic property is due to F. N. Sibley, mainly in Sibley ([1959] 1978); also seeBrady and Levinson (eds.) (2001). A non-aesthetic property may be relevant to aesthetic evaluation (e.g.that the author is six years old is a good reason for believing his work is not very deep. However, tenderage is only inductively linked to lack of emotional depth, and so the former fact is only contingentlyrelevant to the latter).

² Among the latter Longyear (1970), Bonds (1991), and Sheinberg (2001) are especially noteworthy.³ For example, romantic irony in Longyear (1970) and Dill (1989).⁴ For example, Schumann’s lieder to Heine’s texts in Brauner (1981), Dill (1989), and Suurpaa (1996).⁵ Sheinberg’s work (2001), though extensive and illuminating, is no exception. It presents an array

of edifying quotes on various kinds of irony and sophisticated analyses of certain specimen, but nothinglike a semantic theory is attempted.

the structure of irony/how it functions in music 179

even when its relevance is not in doubt,⁶ can only show that the piecewas thought (by its author or his contemporaries) to be ironic; it cannotwarrant an attribution of an aesthetic property. What aesthetic propertiesan item has is an objective fact about it. One would not use a vintage1800 medical theory to diagnose Schubert’s illness; similarly, one shouldnot exclusively use a vintage 1800 aesthetic theory to analyse his music.A theory of music may use historical sources, but must also be conversantwith analytic aesthetics.

2

We start, not with musical, but with verbal irony. Suppose that speaker S,to express his opinion that Joe is stupid, says: ‘‘Joe is very wise.’’ Since Sthinks that Joe is not very wise, does he lie? No; S’s interlocutors readilyunderstand that S paid Joe no compliment. S’s token sentence is ironic.What does that mean?

To deny that S lied, one may suggest that ‘‘wise’’ has an additional sense:it also means ‘‘stupid,’’ and S used that sense in his token-sentence. Somewriters do hold that view about metaphor: metaphors, they say, extend theliteral sense of words⁷. But stupid is not an extension of the sense wise; nodictionary lists it under ‘‘wise.’’ Further, since all terms may be ironicallyused, that view would imply that every term also means the opposite of itsusual sense. If that were so, contracts were worthless. We therefore agreewith Davidson (1978) and Scruton (1997: 84) that an expression retains itsliteral meaning in all contexts. What, then, about S’s sentence ‘‘Joe is verywise’’? Davidson bites the bullet: he would say that what S said is falseor truth-valueless. But if what S says is false, why do all S’s interlocutors,who share his assessment of Joe’s intellectual stature, say that what he saidis true? Do they lie, too?

No doubt, what S says implies that Joe is not wise, but this cannot beall that it means. If to speak ironically were, as Quintillian alleged, tosay one thing and mean another, that is, if irony were an operator that

⁶ Cf. Daverio (1990), among others. Most of those who wrote about irony in Beethoven citeSchlegel and Tieck. Longyear (1970) does so too, yet he is careful to note we have no evidence thatBeethoven was familiar with the writings of Schlegel and Tieck on irony.

⁷ G. Lakoff and M. Johnson (1980); S. Glucksberg and B. Keysar (1993).

180 eddy zemach and tamara balter

took a sentence ‘p’ into its negation, ‘‘∼p,’’ then it would be a mosttedious and pointless device: why risk error and confusion instead of plainlysaying ‘‘not-p’’? Yet, as we know, irony does not make conversationtedious but enlivens it, rendering it more entertaining. So, irony is notjust a lugubrious way of achieving an effect that could be achievedmore frugally by negation. We hold that ‘‘p’’ keeps its literal sense evenwhen ironically used: it makes us project a situation, p, which literallysatisfies ‘‘p.’’

Some writers, who, like us, hold that metaphorical (including ironical)phrases retain their literal meanings do so by stipulating special undefinedoperators. Searle (1978, 1979) holds that to speak metaphorically is to use anirreducible kind of illocutionary force. Walton (1990) regards the sententialoperators ‘‘Make-believe (p)’’ and ‘‘Fictional (p)’’ as basic and irreducible.We think that leaves them unintelligible. Our most basic concepts mustindeed be left undefined, but surely these are not among our most basicconcepts? We offer a definition: ‘‘p is fictional’’ means ‘‘p obtains in apossible world other than the real one’’ and ‘‘S makes believe that p’’ means‘‘S projects p in some world, when some real situation Cp is taken to be itscounterpart.’’ Metaphors, one of us suggested (Zemach 1999b; 2001) aremodal terms: the sentence ‘‘Metaphorically (Fx)’’ is true iff x is F, not inthe real world, but in x’s home world: a world where x’s nature is apparent.⁸A similar semantic analysis is developed below for irony.

3

Let us build a definition gradually, starting with a linguistic competencecondition. To understand a sentence is to project a situation⁹ that satisfiesit. For example, one who understands the sentence ‘‘Joe is wise’’ projects(envisages) the situation, that Joe is wise, in some world. Let ‘‘p’’ be true inw. Call the situation in w that satisfies ‘‘p’’, ‘‘pw.’’ So, ‘‘p’’ makes us projectpw (‘‘→’’ stands for ‘‘projects’’):

1. (∃w)(‘‘p’’→ pw).

⁸ A common objection is that a metaphor may be logically impossible; for a reply see Zemach (2001).⁹ Scruton (1997: 93) calls it an ‘‘unasserted thought’’; his discussion is illuminating, but that term is

not, for it conflates the thought with the situation that would satisfy it.


Now, pw has a counterpart in the real world (W0); call it ‘‘CpW0.’’ CpW0and pw are genidentical: they make one trans-world individual. For thesentence ‘‘p’’ to be ironic, however, these situations must be qualitativelyopposed to each other. In symbols (‘‘x|y’’ stands for ‘‘x is the inverse of y’’):

2. (∃w)(pw|CpW0).

In words: A situation p (satisfying ‘‘p’’) in some world w is qualitatively theinverse of its counterpart in the real world. Now put (1) and (2) together:

1+2. (∃w)[(‘‘p’’→ pw) & (pw|CpW0)].

In words: ‘‘p’’ makes us project a situation that satisfies it in a world, whereit sharply contrasts with its real-world counterpart.

Finally, we need to add that the projected situation is superior to its realcounterpart. Envisaging pw, it becomes evident that CpW0, the real-worldcounterpart of pw, is far inferior to it. Compared to its ideal counterpart pw,the real situation looks deformed. Thus, irony is bitter-sweet: it is amusing,but also painful, to imagine intellectually challenged Joe as wise. Theprojected situation pw, where Joe is very wise, shames and hence mocksthe real situation CpW0. Ironically to say ‘‘p’’ is therefore more rhetoricallyeffective than to assert ‘‘not-p,’’ for the contrast demonstrates the ludicrousdeficiency of the real situation. With all the conditions combined, we get(‘‘>’’ stands for ‘‘much better than’’):

3. (∃w)[(‘‘p’’→ pw) & (pw|CpW0) & (pw>CpW0)] ⊃ I‘‘p’’.

The last conjunct of the definition is important. Irony is not a mere tongue-in-cheek use of language: it bites. If Joe is very wise and his friend S fondlycalls him ‘‘silly,’’ that is no irony: it is a kind gesture, an understatementused to compliment. A similar understatement is used by the basketballplayers who call their tallest ‘‘shorty.’’ Such expressions neither bite normock, so the trope involved is not irony. An amicable understatement, oran exaggeration, used as endearment, or as an underhanded, coy, praise, isnot irony. Irony involves criticism, a dramatic exposition of depravity.

That a given token is ironic is linguistically and extra-linguisticallyindicated. In our example, the highly laudatory words ‘‘very wise’’ put uson guard. The situation that satisfies it is projected and, having had its effect(shocked amusement, due to the known facts about Joe) we see that ourinterpretation must be further elaborated. The methodological constraint


of charity is invoked: interpreting S’s words literally is uncharitable to S,making him a liar, or dead wrong. We cannot ignore the literal meaningof ‘‘very wise,’’ so, we take it to have a modal rider. Another hint thatthe sentence is ironic is given by S’s tone of voice. It is taken to have asemantic role, indicating the modal operator, irony.

Music is not a language, so verbal irony in music is extremely rare.However, we hope to have shown that use of words is not essential toirony. Its essence lies in projecting a situation that makes reality lookdeformed, and such projection need not be prompted by verbal means. Itmay be evoked by certain real-life situations, too.

4

Joe is a clumsy and reckless driver but never had an accident. Jane is anadept and scrupulous driver, yet she is killed in a traffic accident. Thatsituation, we say, is ironic. What is an ironic situation?

Like verbal irony, situational irony invokes an ideal situation that iscompared with the one at hand; the two situations are genidentical yetdissimilar, the real one being a deformed version of its projected counterpart.In verbal irony a speaker uses language to indicate an ideal situation in apossible world; in situational irony the ideal situation is indicated by ournorms. We have strong intuitions on how things should be: what is right,just, and fair in a particular case; how the world should then be. It is asif we all jointly do what a speaker does in verbal irony: we indicate acounterpart of the real situation at hand. The present situation looks like awizened, degenerate version of its projected counterpart. It looks funny yetterribly wrong. The contrast between it and its possible counterpart makesus regard it with a sad smile.

In our example, the real situation CpW0 makes us project a just andproper situation pw, where people get what they deserve. There, it is Joeand not Jane who is killed in a motor accident. The projected situationpw makes us see the real situation CpW0 as a grotesque and deformedversion of itself. Thus, the possible situation serves vividly to denigrate itsreal counterpart.

An ironic situation is modal since it requires interworld comparisons. Asentence is modal if its evaluation in reality depends on its truth in other


worlds. Similarly, a situation is modal if appreciating it requires that wecompare it to a counterpart; if it is ironic, that comparison finds it wanting.The real situation CpW0 (the good driver is killed and the bad comes tono harm) is ironic in that it projects a just counterpart, pw, which puts it toshame. Let us define:

4. (∃w){[CpW0→(pw)] & (pw|CpW0) & (pw>CpW0)} ⊃ I(CpW0).

In one of the very few articles devoted to situational irony Lee Miller (1976)requires an additional condition to hold for an ironic situation. An ironicsituation, he says, though it runs contrary to the hopes and efforts of thoseinvolved, displays a particular propriety: a kind of poetic justice. But, to useMiller’s own examples, what poetic justice is there in the wealthiest womanwinning the lottery, or in a father’s constantly yelling at his children to bequiet? Miller’s general diagnosis, we think, is correct: ironic situations areuniquely apt; they are characterized by extra significance. But the aptness,we think, is that which makes us regard the two situations as counterparts,and the extra significance is the demonstration of the wretched nature ofreality. We see the ironic situation as profoundly instructive, almost proof-like, for it shows the reality as a degenerate version of what there should be.

Theorists of music do not often discuss situational irony, either. This isstrange, for two central, universally agreed upon, characteristics of musicmake it amenable to situational irony. First, music raises expectations:competent listeners project situations they consider right for the (musical)conditions at hand. Second, listeners compare and contrast the anticipatedsituation (a musical event) with the one that does occur in the work. Thesetraits, however, are necessary, but not sufficient, conditions for musicalirony, for contrasting situations and frustrating expectations have manyother uses too: they create tension, drama, or simply add interest to thework. When does the presence of these features indicate situational irony?

A situation may challenge our assumptions on what is proper and right,expressing fundamental contradictoriness, by defying our expectationsabout the proper location of musical material of a certain kind. Thus, apiece that begins with a closing gesture flies in the face of our expectationsfrom a musical event proper to the beginning of a work (in the givenstyle). Such ‘‘closing-beginnings’’ exist in the music of Haydn and latercomposers, mainly in string quartets. Kramer (1973) and Lochhead (1979)discuss the contrast between these two states, the projected and the actual


AllegrettoViolino1

Violino II

Viola

Violoncello

pizz.

5

poco cresc.

poco cresc.

poco cresc.

Example 2. Beethoven: String Quartet in F Major, Op. 135, i, mm. 1–10.

(Kramer names them ‘‘clock-time’’ and ‘‘gestural-time’’), in Beethoven’s StringQuartet Op. 135 (Example 2). The musical event at the beginning of thatwork suggests closing or cadencing. It thus abuses its actual (clock-time)place at the head of the piece. At that hallowed location we expect (henceproject) normative tonic stability and clear periodicity. Instead, we get acompound ten-measure sentence that closes with a final cadence (in m. 10all four parts have an F in four different registers). Two surprising Phrygiancadences open the piece, each of which is followed by a sprightly three-notemotive that mocks its serious mood and questioning gesture (mm. 1–4). Anonchalant sequence follows and further develops the mocking three-notemotive. A terminal cadence occurs in m. 10, and then there starts a new,unrelated musical idea. That beginning makes a mockery of the projectednormative beginning. It undermines the belief in the good order of thingsand includes comic elements. Thus, we say that the actual beginning ofOp. 135 is ironic.

One of the few who investigate situational irony in art, Lars Ellestrom,uses Magritte’s painting In Praise of Dialectic as an example. It shows the insideof a room through an open window, but that room only contains the facadeof another building (1996: 202). Instead of a family scene we get an inferior


situation: a forbidding externality.¹⁰ The frustration of our expectations bythe exterior wall that stands where an inside should have been parallels theabove closing-beginning in Beethoven’s Op. 135. In Magritte’s work ‘‘theinside is the outside’’ (ibid.); likewise, Beethoven shows that the beginningis the end.¹¹ The competent listener’s tendency to project the norm createsa paradox. Beethoven seems to resolve it at the end of the work, but thatsolution only intensifies the irony. The ‘‘closing’’ sentence that openedOp. 135 reappears in its end, doing, so to speak, its ‘‘duty’’ to conclude thepiece. Yet this ending redoubles the irony: after we consented to forgo ourprevious conceptions and accept a new musical idiom wherein that musicalevent has the role of a commencing gesture, the new norm is forfeited. Wemust abandon our newly acquired norm and go back to the old one. Theopening sentence concludes the very work that taught us not to expect itthere. The destructive irony here is radical and most effective.

Another way to express situational irony is by frustrating a meticulouslyprepared expectation for harmonic resolution or a key area. The suddendislocations of the apparent tonal stability in the finale (sonata-rondo) ofBeethoven’s violin sonata Opus 30, No. 3 are ironic, says Longyear (1970:656). There are two tonally surprising shifts in that finale. First, after thefinale’s second episode there is a false reprise of the theme; that false reprise,in the major mediant (mm. 133–6) had been carefully prepared, yet isabruptly abandoned. The movement proceeds to a preparation for a correctrecap (m. 141). Second, in the coda there is a surprising shift to the flatsubmediant (m. 177) after a prolonged dominant; this unexpected tonalityis ‘‘stated in a ‘vamp’ accompaniment’’ and precedes another statement ofthe main theme in a wrong key (Longyear 1970: 656–7). Longyear believesthat these shifts show romantic irony, for they destroy an illusion. Leavingaside, for now, the question of romantic irony (but see Section 8 below),we ask: is it situational irony? We think not. Even if we expect neither the

¹⁰ Although the situation that contrasts with the projected one is not real, but merely painted, it isone we see in front of us, so it serves here as ‘‘real.’’ Other substitutes for reality are situations we seein front of us on the stage. (We thank Kathleen Stock for pointing it out to us.)

¹¹ Following Muecke (1969), Ellestrom writes ‘‘An ironic situation is a situation that displaysa striking incongruity between an expectation and an event’’ (p. 210). But that condition, whilenecessary, is not sufficient, for it fails to distinguish an ironic situation from a merely surprising one.Ellestrom himself goes beyond it when he claims that in music, ‘‘when we feel that two contrastingmoods are mutually exclusive, and yet in a way make sense when jumbled together, it is irony thattickles our ears’’ (p. 205). This is reminiscent of Miller’s (1976) notion of the overall appropriateness ofthe ironic situation.


major mediant nor the flat-submediant, their surprising appearance doesnot amount to contradiction, as required in situational irony. Listeners areled to project a more normative, perhaps a banal, tonic area, which doesnot materialize; but the projected situation (tonic area) does not mock therealized situation (major mediant) and does not contradict it.

There is, however, a late movement by Beethoven, the scherzo in StringQuartet Op. 131, where unexpected modulations to the mediant may beretrospectively interpreted as cases of situational irony.¹² In that movementBeethoven modulates to the mediant very early in the scherzo sections, thendirectly moves from the dominant seventh chord of the mediant (markedwith a fermata) to the tonic, E major. That shift brings to mind the deceptivemove in the coda of the violin sonata Op. 30, no. 3, from a dominant seventhchord with a fermata to the flat submediant. In Op. 131 the deceptionis double, since what may sound as the submediant in G# (an E majorchord) is, in fact, a correct (tonic) reprise of the A section of the scherzo(Example 3). A deceptive move (to bVI) turns out to be a move to the tonic.

30Molto poco adagio

Tempo1

più

A

dim. più

dim. piùun poco più adagio

dim. più

39ritard. in Tempo

Example 3. Beethoven: String Quartet in C-sharp Minor, Op. 131, v, mm. 30–48.

¹² Longyear (1970) briefly mentions this movement, calling it, too, a case of romantic irony.


These early modulations to the key of the mediant are ironic because,once it becomes clear that the E major chord is a correct beginning of thereprise, the modulation to G# minor looks wrong. Beethoven makes itlook as if this (unconventional) modulation to the key of the mediant was amistake by writing no modulation back to the home key. Instead, he makesthe players slow down, and seem utterly confused. They hesitate, as if theywonder how they arrived at this G# minor area and got stuck there, notknowing how to return to E major. So what they do is just give up and startall over again, without modulation. That makes it look as if the modulationto the G# minor area is an error they made. Now, to clinch the point,it does make sense to modulate to G# minor toward the next movement,which is in G# minor, and is immediately linked to the present movementwithout a pause; but there, no such modulation occurs. It thus appears asthough things got mixed up; the said modulation came at the wrong time.The composer has, then, intentionally made the actual situation to looklike an aberration, a deformation of another, normative, one.

5

Dramatic irony is a modification of situational irony. A situation in a dramais said to manifest dramatic irony if its protagonist does not comprehendthe situation she is in, while the audience does. For example, the scenewhere Rigoletto cheers on the gang of courtiers not realizing that the ladythey abduct is not Countess Ceprano but his own daughter, Gilda; or, inWilhelm Tell, the scene where Governor Gessler boasts how he will subduethe Swiss, not knowing that his Swiss assassin is already in place, ready tokill him. Analysing this we see that in dramatic irony, too, a counterpart tothe present situation is projected, which is the opposite of the situation athand. That projected situation, however, is good for its projector only; itis not (as in situational irony) recognized as good by us all. The situationsRigoletto and Gessler project are good for them, but sharply contrast withthe actual situations, which we now see as their negative counterparts.

Dramatic irony is modal, for it interprets the real situation via a possiblecounterpart. Now, in dramatic irony the projected situation is the onethat looks ridiculous; does this mean that dramatic irony deviates from thebasic structure of irony, where reality is mocked and criticized? No, for


the inadequacy of the situation projected makes the present protagonist,who projected it, look ludicrous. The unrealistic protagonist is, therefore,the butt of the dramatic irony. It is pathetic to be so naıve as to believe anidealized counterpart of grim reality. Let us define:

5. (∃w){[CpW0→S(pw)] & (pw>S CpW0) & [BS(w = W0)]& (w �=W0)} ⊃ I(CpW0).

In words: A situation is fraught with dramatic irony if its protagonist sprojects a counterpart situation (that is much better for s than the real one)and falsely believes it is real.

Dramatic irony is the most prevalent kind of irony used in music. Inevery B-movie the music forewarns the spectators, before the hero has anyinkling of it, that the new man is the villain, or that the heroine is about tofall in love. Music is a forecaster, interpreter, and prompter in all operas, sodramatic irony is ubiquitous in them (see Burnham 1994 and Sawyer 1999).The example we chose (the cavatina of Barbarina from Le nozze di Figaro)is, however, subtler; it is threefold. The protagonists see it as a vulgar joke;the audience perceives it as dramatic irony, but, upon reflection, may alsotake it as meta-poetic criticism.

Barbarina’s cavatina is the only aria in Le nozze di Figaro written in aminor key from start to finish (Example 4). Suffused with appoggiaturasand chromatic chords (diminished seventh and augmented sixth chords) itexpresses deep sorrow, yet it is sung by a minor character. Dent (1960: 112)thinks the cavatina has a structural role: it signals the opera’s transition froma light to a somber atmosphere; but this does not explain why Barbarina,

29

- du ta!- ah, chi sa do ve- sa rà?- E mia cu gi- na?- e ilpa dron,- co sa- di rà?- co sa- di rà?-

Example 4. Mozart: Le nozze di Figaro, Act IV, Scene I, No. 23, Cavatina, L’hoperduta, mm. 29–36.


an unlikely figure to express a deep sentiment, is chosen to introducethat somber tone. Also, as Kivy (1980) notes, the crying motive in thecavatina is appropriate to a greater loss than that of a pin. What is thepoint of that conspicuous incongruity? The mystery deepens when wenote that the entire section was added by Da Ponte: it is not found inBeaumarchais’ play. Da Ponte also gave Barbarina her name (Beaumarchaiscalls her ‘‘Fanchette’’). Since the libretto is quite faithful to the play (itomits, but does not add scenes!) and only one other very minor character(Don Curzio) is renamed, one cannot but wonder why this scene wasadded. Is it only for the sake of the joke? This cannot be the whole story.Had Mozart only wished to mock Barbarina for the weight she attaches toher tiny loss he could have set her words to a recitative; he would not havewritten for them dark, heart-rending, music that makes all jokes go sour.

A pin may be the most fungible object there is. It is (even in the eigh-teenth century) a mass-produced, cheap, common object (the Count sayswomen have them everywhere), hence most easy to replace. Indeed, Figaroimmediately replaces the pin Barbarina lost by taking one off Marcellina’sdress. To him, her moaning seems silly: pins are indistinguishable; theyare an entirely generic product. A dirge for a lost pin is absurd, and maybe funny. But the music disallows that interpretation: it tells us that thiscavatina has a serious import that is incompatible with the frivolous spirit ofthe opera-buffa. The pin must have a special significance here that neitherFigaro nor Barbarina grasp, but the audience should.

The pin Barbarina lost was the seal of Susanna’s letter, setting out (onher wedding night!) a tryst at which the Count will deflower her. Shepretends to promise him the prize he wanted to take by force, that is, theright to break her seal. Thus the seal of Susanna’s letter symbolizes hervirginity. That symbol is further developed when, upon contact with theCount, it draws blood. Ironically, the roles are here reversed: it is not thevirgin Susanna, when her seal is broken, who bleeds, but the man whotries to break her seal. In symbolic, ironic, revenge, the seal (the pin thatseals her letter) pricks his finger. Virginity is the subject of the entire opera:it starts when Almaviva wants to reinstitute his feudal droit de seigneur tothe virginity of his vassals. So, if the seal is virginity, then in mourning theloss of the seal Barbarina really mourns the loss of virginity. What is lost isnot just a pin: it is virginity. Indeed, the Count also embraced Barbarina,kissed her, and promised to give her whatever she wants if she would love


him (no. 21). So it may be that it is her virginity too, not only Susanna’s,that (unknowingly) she mourns. Barbarina, of course, never realizes the fullmeaning of her lament over the pin, but there is no question that Figarodoes. To Marcellina he explicitly says: ‘‘Ah, that ‘pin’, mother, is the sameone the Count lately renounced.’’ The right the Count lately renouncedis, of course, the infamous Jus prima noctis.

That is one level of dramatic irony; but there is a higher one. All ofMozart’s operas present a view of love according to which love is generic:the identity of the beloved is of little importance. In Cosı fan tutte thewise cynics (Don Alphonso and Despina) cause the girls to switch theirlovers—and then switch them back again. In Die Zauberflote, Papagenoadmits he is attracted to all women; he wants, as he often says, a (generic)woman (Ein Madchen oder Weibchen ... ). In Figaro, Cherubino voices thatsentiment when he candidly describes (e.g., in Voi che sapete) the love hefeels as generic: it may shift objects or pick one at random. Susanna isthe lady with the seal, while Cherubino is one whose patente has no seal(end of Act II). Most blatant in his generic attitude to erotic objects is, ofcourse, Don Giovanni. In Figaro that attitude is manifested by the lewdCount, by Figaro’s condemnation of all women (the aria Aprite un po’ quegliocchi), by the fact that every lover in this opera accuses his or her partner ofinfidelity, and by the ease with which one person can pass, even during anassignation, for another.

Such is the society depicted in Mozart’s operas. Here, in a statementthat transcends that society, Mozart criticizes it; it lost, he says, somethingvaluable. People in that society cannot see themselves as tragic, but wecan, so the music advises us of the composer’s judgment. Barbarina, then,speaks for the composer. That is why her cavatina is so profoundly sad, soout of step with the carefree levity of the musical material in Figaro. Figaro,for instance, even when given the most awful news, is assigned light andcheerful music. No one in this opera has notes as sad as those in Barbarina’slament over a lost pin.

Thus it is not only the plight of women that this plaintive music laments,but the entire notion of romantic, non-generic, individually ‘‘sealed’’ love;it is non-fungible love that Mozart yearns for. Why is Barbarina chosento express that yearning? Because she is the proverbial Ingenu, too naıveto play the games of dissimulation and deceit that all others in that society


play. Muecke’s example of Ingenu irony is the child who cries out loud (inAndersen’s The Emperor’s New Clothes) that the King is naked. In Muecke’swords, ‘‘even simple innocence or ignorance may suffice to see through thewoven complexities of hypocrisy’’ (1969: 91). Simple-minded Barbarina isthus the perfect choice for noting how sad the world of her sophisticatedbetters and elders is. Muecke says: ‘‘the ingenu is himself sometimes avictim of irony, exposed as well as exposing’’ (92). Surely, that is true ofBarbarina; moaning over a lost pin is indeed farcical, but it exposes thefar greater farce of the Almaviva court, which, essentially, is Mozart’s ownsociety.

Another contribution of the music to the sense of this cavatina is byalluding to the Countess’s big recitative and aria in Act III (no. 19).Barbarina’s cavatina ends abruptly with exactly the same half cadence—asustained Italian augmented sixth chord resolving to a dominant—thatends the Countess’s recitative, ‘‘fammi or cercar da una mia serva aita!’’ (‘‘[he]makes me seek help of my servant!’’) (Example 5). The reference is toSusanna, but the musical allusion makes us think of Barbarina, who alsoexpresses the agony of the Countess. Another link between Barbarina andthe Countess is structural: the cavatina of the countess, Porgi amor, opensAct II of the opera; Barbarina’s cavatina opens Act IV. Barbarina’s loss isthereby associated with the loss of the Countess, a loss of her ‘‘treasure’’(mio tesoro): love. Finally, note that the main aria of the Countess (Dove

24

fam mi or- cer car- da u na- mia ser va- ai ta!-

Example 5. Mozart: Le nozze di Figaro, Act III, Scene VIII, No. 19, Recitative,mm. 24–25.


sono, which mourns the loss of happy moments and the forfeiture of oathssworn by her husband) is written to the theme of Mozart’s Agnus Dei inthe ‘‘Coronation’’ Mass, K. 317 (completed in 1779). Mozart was not inthe habit of pillaging his own works for melodies. The rare hyper-textualconnection shows, then, an affinity of moods. A critical message hiddenbehind the jocular face: that is the gist of dramatic irony.

6

Another kind of irony is total rejection: treating all alternatives to realityas flawed. Muecke (1969) calls that kind of irony, made famous byKierkegaard ([1841] 1989), ‘‘general irony’’; others call it ‘‘cosmic irony,’’‘‘infinite irony,’’ etc. It is a negative attitude to all possibilities, whateverthey may be. General irony is therefore anti-modernist, for modernismbelieves that a better alternative to reality can be envisaged.

Like all other kinds of irony, general irony, we say, is modal, for itsevaluation in reality depends on its evaluation in alternative worlds. Asituation that displays general irony does so by invoking its counterpart. Inthe kinds of irony discussed above the projected situation pw is consideredbetter than the real situation. General irony, however, implies that noalternative to the real situation can make a difference; all possibilia areequally flawed. That attitude is typical to fin de siecle symbolists bored byeverything and to existentialists for whom the depravity of a situation liesnot in its quality but in its very being (as Sartre puts it, it is its facticity thatmakes a situation unbearable). Nineteenth-century works like Goethe’s DieLeiden des jungen Werthers, turn-of-the-century Decadence works such asPicasso’s Les Demoiselles d’Avignon, and mid-twentieth-century works suchas Sartre’s La Nausee and Celine’s Voyage au bout de la nuit, all flaunt generalirony. The irony of Romanticists and Modernists presented the world associally, i.e., redeemably, flawed; general irony regards every situation withequal ennui or disgust.¹³ According to Kierkegaard, that attitude is essential

¹³ Several theorists assimilate general irony to romantic irony. Esti Sheinberg is one: romantic irony,for her, has three kinds: aesthetic distance (romantic irony proper), infinite negation, and existentialirony (2001: 61–2). That classification has some merit, yet it misleads. General irony condemns allalternatives, while romantic irony is corrective: it criticizes an element in the work, or, by shatteringthe illusion, alerts us to the nature of the work as an artefact.


to achieve religiosity: only those who sense the pull of nihilism make the‘‘jump’’ to religion. Wittgenstein, in the Tractatus, holds a similar view.¹⁴We define that kind of irony as follows (Where ‘‘D’’ is ‘‘depraved’’):

6. (∀w){(CpW0→ pw) ⊃ [ ∼(pw>CpW0) & D(CpW0)]} ⊃ I(CpW0).

In words: a situation is ironic if all the counterparts it projects are as flawedas it is.

Patricia Herzog (1995) claims that Beethoven’s Diabelli variations expressan ironic-sceptic attitude to life. Her characterization of that attitude makesit a case of general irony. These variations, Herzog says, take the listenerto diverse and contrary worlds; just as one begins to get accustomed toa world, expecting it to endure, one is ejected into its opposite. It is anarrative of successive rejections that has no goal and no end; so (as we putit), it manifests general irony. William Kinderman (1987) interpreted thatwork in a very different way: according to him, these variations do havea goal; their unfolding is a teleological process through which Beethovenovercomes the waltz.¹⁵ The finale of Beethoven’s Op. 101 (The Deed)generates a similar dispute. Wilfred Mellers (1983) finds in it cosmic irony;Robert Hatten disagrees. Although Hatten, too, thinks the movementcontains a ‘‘Romantic-ironic projection of the persona,’’ he takes it toenhance the achievement of ‘‘a positive outcome’’ (1994: 186). As thesedebates show, it is hard to find works that consist of incongruities betweensegments, all of which are equally presented as wrong, and therefore maybe said to express equally inadequate views. We saw that even a workfilled with incongruities and shifts in level of discourse can be seen as ateleological narrative; alternatively, the negation may be construed as adevice for creating artistic distance. Perhaps only synchronic presence ofdiverse levels, styles and topics can clearly embody general irony. Themusic of Mahler and Stravinsky often has concurrent presentations of non-congruent layers, though not throughout entire movements. Stravinsky, inthe first miniature piece from Three Pieces for a String Quartet (1914) doesthat, too.

¹⁴ ‘‘If there is any value that does have value, it must lie outside the whole sphere of what happensand is the case. For all that happens and is the case is accidental. What makes it non-accidental cannotlie within the world, since if it did it would itself be accidental. It must lie outside the world’’ (6.41).

¹⁵ Kinderman’s interpretation is that the clashing variations in the middle section of the work areindeed ironic, as evidenced by the direct parodies of the waltz. The final variations, however, go beyondirony to uplift the original waltz; there, the waltz is used to create lyrical, richly textured, variations.


Conversation and communication between partners is, perhaps, thecommonest metaphor for the string quartet genre.¹⁶ We suggest that in thefirst of his Three Pieces for String Quartet (later named ‘‘Dance’’) Stravinskygoes all out to combat it. Like many pieces by Stravinsky, ‘‘Dance’’ hasstratified layers, but here stratification is radical: each layer includes anindependent, musical idea that repeats as an ostinato over and over again,to the very end. The first violin repeats five times a 23-beat phrase (thefifth repetition is shortened) consisting of a major tetrachord (G-A-B-C).It is reminiscent of a naigrysh, the Russian folk instrumental ‘‘dance-until-drop’’ tune (Watkins 1994: 260); yet here the phrase is much longer thanin typical naigryshi. The viola and cello repeat many times a much moremonotonous 7-beat ostinato; the second violin alternates between singleand double statements of a descending minor tetrachord (F#-E-D#-C#)that sound like impulsive interruptions. These independent layers clashwith each other. Clashes occur within layers, too; for instance, althoughthe ostinati of the viola and of the cello are of the same length, the viola’sD clashes with both the Eb and the Db of the cello. The relation betweenthe layers varies due to their different phrase lengths. A three bar dissonantdrone (a minor ninth) in the viola part starts and ends the piece; added tothe extended folk tune in the first violin and to the repeating ostinati itrenders the piece a grotesque folk dance.

A performance of that piece creates an impression of miscommunication,where each interlocutor repeats, ad nauseam, an unrelated idea. It picturesincommensurable non-coalescing world versions that cancel each otherout. It is a world that does not and cannot make sense. In brief, it carriesgeneral irony. Watkins argues, in detail, that this piece is influenced bycubism, in particular, the cubist appeal for simultaneism.¹⁷ We acceptthat, but note that cubism itself has a nihilistic streak; it criticized classicalpainting for giving a definite, unambiguous, ‘‘literary,’’ sense to the visualworld. Cubism had another streak, too, a constructive, modernist streakexemplified, e.g., by Fernand Leger. But it is the first, nihilistic, streakfound in Decadence art from Alfred Jarry (Ubu roi) to Jean Cocteau andDadaism that is relevant to Stravinsky’s work of that period.

¹⁶ For a comprehensive discussion of that metaphor, see Mara Parker (2002).¹⁷ ‘‘Dance’’ was conceived by Stravinsky after discussing a possible collaboration with Jean Cocteau

on his ballet David (Watkins, 1994: 258–65). We thank Marianne Kielian-Gilbert for this reference.


What most inclines us to interpret the Stravinsky piece as divulginggeneral irony is the original choice of the string quartet as its medium,while flagrantly defying the tradition of the string quartet as a paradigmof cooperation. Stravinsky’s piece retains enough of the early eighteenth-century tradition, for instance, a longer and more interesting melody inthe first violin part, to make the link with it evident.¹⁸ It thus seems to bea deliberate attempt to ironize and negate the notion of a joint effort tomake sense of the world. Cooperation is ruined precisely where a musicianwould most look for it: in the string quartet.

7

Instead of showing the depravity of the real situation by projecting an idealsituation that sharply contrasts with it, one may project a situation thatis much worse than the real one, but does not sharply contrast with it.This kind of irony is not immediately recognized as such: we first take theprojected situation pw to be real. So, when we realize our mistake, we seethe real situation in a new light. Since the projected bad situation pw couldso readily be mistaken for the real situation CpW0, we now see the latterin the light of the former, that is, as depraved. That is parody: a projectedworse counterpart reveals the real situation as also flawed.

Swift’s ‘‘A Modest Proposal’’ proposes to relieve the poverty in Irelandby selling Irish children as meat to be ‘‘Stewed, Roasted, Baked or Broiled.’’Since this piece is ironic, does it mean the opposite of what it seems to say,i.e., that the Irish should not be treated like cattle? No; that semantics makesirony otiose. The text projects a situation where cannibalism is advocated.Only later we realize that, by hyperbole, Swift criticizes a real situation (cf.Booth 1974: 105–20), i.e., that a tacit modal operator precedes the text. Inthe possible situation pw the proposal is seriously made. The depravity ofthe real CpW0 is revealed when one notes its proximity to the projectedpw. Let us define:

7. (∃w){(‘p’→ pw) & ∼(pw|CpW0) & (pw<CpW0)} ⊃ I‘p’.

¹⁸ Stravinsky denied any influence of Schoenberg or Webern on these pieces, claiming it ‘‘looksahead’’ to later works such as Piece Faciles (1915) and to ‘‘my so aberrant ‘neoclassicism’ ’’ (Igor Stravinskyand Robert Craft [1960], p. 89).


In words: a text that projects a situation which is far worse than the realone yet does not contrast with it, is ironic (a parody).

Elsewhere, one of us argued that pictures are like binoculars for possibilia:they enable us to inspect items from other possible worlds.¹⁹ When thepicture is a caricature, the possible item seen in (i.e., via) the cartoon isassumed genidentical with some real item. In visual parody an item isprojected via a cartoon and is grasped as a counterpart of a real item,say, a famous politician. Usually, a caricature mocks a well-known personby projecting an object that is seen in the cartoon, capitalizing on theirassumed genidentity and perceived similarity. Since the two are similar, wetend to impute to the real person properties of his projected counterpart.In parody, these are negative properties; once you see (thanks to the artist)that the projected object has them, you see them in the real object, too.Take a cartoon that projects a man with a huge nose. You say ‘‘voila,General de Gaulle!’’ De Gaulle in reality was not so deformed, he didnot have such a huge nose; but in the possible world you now descry,he does. The counterparts, the one with monstrous proboscis, pw, and hisreal counterpart, CpW0, are similar enough for you to impute to the realone the deformities that only his projected counterpart has. The funny,possible, projected De Gaulle illuminates the real one, so that you now seethe real De Gaulle in the light of his unreal, deformed counterpart.

Unlike literary and visual parodies, musical parody is not a possibleobject we project but a real piece of music which suggests that anotherreal object, being its counterpart, also has its manifest drawbacks. Call theparody ‘‘pW0,’’ for it occurs in the real world, and its alleged counterpart‘‘CpW0’’; pW0 says of itself that, essentially, it is CpW0 (an older work,style, or genre). By presenting itself as CpW0’s true self, pW0 mocks it andputs it to shame, since if CpW0 is as flawed as pW0 evidently is, then it isnot good at all. Musical parody is thus immediate, funny, and vicious.

Not every musical borrowing is a parody, but every musical parody isa borrowing—one that sheds unfavourable light on its source. It may bea mere irreverent joke that entertains by casting aspersions on a veneratedpredecessor, or a serious attempt to reform our taste by exposing the oldermodel. ‘‘Look at me,’’ the parody says, ‘‘see how flawed I am? Well, for allintents and purposes I am a doppelganger of the august CpW0; it is no better

¹⁹ For a discussion of seeing in, cf. Wollheim (1986, 1987, 1998) and Zemach (1999a, 2002).


than I am.’’ The spoof pW0 tries to persuade us that the grand old work (orstyle, or genre) is really, deep down, not better than it, its counterpart, is.

The Theme-and-Variations form is particularly suited to parody since itprovides clear reference to the item parodied: its theme. Usually, a claimthat a piece of music pW0 parodies another requires weighty evidence; inthe case of theme-and-variations that claim is obvious, for each variation isassumed genidentical with the theme that it is supposed to elaborate. Whenthe theme is by another composer, the variations, functioning as explicitquotations, have an obvious external counterpart which they paraphraseand may ridicule. Consider Beethoven’s Diabelli variations. In some ofthem Beethoven abandons many features of Diabelli’s waltz: its structure,its melodic and harmonic framework, etc. Those variations that do maintainsalient qualities of the theme such as its contour and the oft-repeated chordor single note, parody it. Indeed, some traits of the theme seem gross; nowonder Beethoven considered it banal and ludicrous (cf. Schindler [1860]1996 and Kinderman 1987). Yet, as Tovey (1972) showed, Beethovenfound in it a rich vein for a profusion of diverse variations, and so onlysome of them reflect his low opinion of the theme. We shall confine theillustration of parody to two variations, nos. 1 and 13.

The first variation in the cycle is a pompous march (Example 6).Apparently, that variation was one of the latest to be written, c.1823(Kinderman 1987: 29–34), Beethoven placed it first only later, perhaps inorder to declare his ironical intentions at the outset. The ironical treatmentof the theme is manifested here, first, by its being a radical antithesis ofa waltz; it is a ponderous, pompous march in quadruple meter. This is adeliberate defiance of convention, which demands that one begins a workin this form with a variation that closely resembles the theme. Anotherindication that the Diabelli waltz is here made fun of is Beethoven’sadherence to its most ludicrous features: eleven repeats of the tonic triad, a

Alla Marcia maestoso

Example 6. Beethoven: 33 Veranderungen uber einen Walzer von A. Diabelli, Op.120, variation 1, mm. 1–5.


monotonous rhythm, and repeats of the strange dynamics, including a greatmany sforzandi. These sforzandi may have added a mischievous quality to thewaltz, but in Beethoven’s variation they amplify the heavy pretentiousnessof a march. That is a veritable cartoon: contempt for Diabelli’s waltz isexpressed by the variation’s being cast in a contrasting form (a march) whileadhering to the waltz’s ridiculous features, magnified and exaggerated. Inreality the old CpW0 (the waltz) looks much better than the new pW0(variation no. 1) makes of it. This does not imply that the variation isof no aesthetic value: it is a good cartoon, a successful parody, brilliantlyhighlighting the depravity of Diabelli’s waltz.

Variation 13 is the most biting parody in the cycle (Example 7). It reducesthe theme more than any other variation in the cycle, and negates (bringsad absurdum) almost every element of the waltz. It shows its contemptto Diabelli’s theme by ignoring it. The many chord-repeats, which are,perhaps, the most preposterous feature of the waltz, are mocked by beingreduced to a thunderous silence in the form of surprising general pauses.²⁰Half of the variation (46 beats out of 96) consists of silences only! Whilethe first variation in the cycle ridicules Diabelli’s chords by exaggeratingthem, variation 13 hints at them briefly with a dotted rhythm that suggestsa march, raising expectations for yet another pompous march, but instead

Vivace

91. 2.

cresc.

Example 7. Beethoven: 33 Veranderungen uber einen Walzer von A. Diabelli, Op.120, variation 13, mm. 1–16.

²⁰ Replacing chord-repeats with GPs is reminiscent of Laurence Sterne’s ploy in Tristram Shandy.Sterne fills a page with asterisks or blank spaces to make fun of a prevalent literary convention thatused such devices to avoid description of sexual matters. Mark E. Bonds (1991) draws a similar analogybetween the irony of Sterne and Haydn’s irony. Cf. Imeson (1996).


we get a long rest followed by two notes in a weak small voice. Finally, thegentle turn motif from the waltz becomes a short, grotesque march gesture.

8

An interesting case of irony is when the projected pw and its real counterpartCpW0 are not only genidentical (the same item in different worlds), butfully identical: the very same situation. In that case the work mocks someor all of its own features: it may deride its own harmonies, melodies, voiceleading, and emotive content; it may poke fun at its poetics and deny its owncandour and aesthetic value. Such self-denigration and self-degradation iscalled Romantic irony. According to the argument of the present articleromantic irony is the epitome of all irony, since, as we argued, all ironyis, essentially, self-directed. Irony, we said, is always based on the claim ofone item that another item is, deep down, genidentical with it. Only acounterpart of a given item can impute its own traits to that given item.Irony would be neither funny nor clever without that dramatic assumptionof a trans-world, or (in musical parody) at least inter-temporal, genidentity.Self-ridicule, then, is the basis of irony as such. Romantic irony only takesit to the extreme by making the mocking situation not only genidentical,but plainly identical, with its target. We define:

8. (∃w)(CpW0→ pw) & D(pw) & [CpW0= pw) ] ⊃ I (CpW0)

In words: a situation that projects a depraved situation which turns out tobe none other than it, itself, is ironic.

We mentioned that some writers compared works of Haydn andBeethoven to novels of Sterne; these writers attributed romantic ironyto both composers.²¹ The coda that ends Beethoven’s string quartet Op. 95is often regarded as an archetype of romantic irony in music.²² An earlierexample, however, one that has not been noted, is Haydn’s coda in stringquartet Op. 76, No. 1, in G Major (Example 8). Though the entire quartetis in the major mode, its final movement is mostly dramatic, in the Sturmund Drang style, in the minor mode with orchestral sound effects (e.g.,

²¹ Such comparisons date back to the eighteenth century. See Bonds (1991) and Kinderman (1996).²² See Longyear (1970); Hatten (1994, chapter 7).


179

pizz.

pizz.

pizz.

184

cresc.

cresc.

arco pizz.

cresc.

arcopizz.

cresc.

arco pizz.

Example 8. Haydn: String Quartet in G Major, Op. 76, No. 1, iv, mm. 179–88.

the opening tutti in unison) and sigh gestures. Only in m. 139 does themode change from G-minor to its parallel key. In m. 180 a buffa-likecoda starts: the violin has a banal, music-box-like, melody while the otherinstruments accompany it in pizzicato. That melody repeats twice andends in a laughing gesture: a quick descending scale marked staccato playedsimultaneously by three instruments (mm. 187, 195). The coda starts onthe last beat of m. 180 in a delay, after the expectation for a tonic on thedownbeat of this measure is frustrated (in the first violin part in m. 179the leading tone is left in the air, unresolved). It thus seems that the codapokes fun at the very lofty sentiment the movement has just so patheticallyavowed.

Such humorous, twaddle codas are found in other movements by Haydn,too, but the debunking effect of this coda is similar to the effect of thewidely discussed coda (the addendum, as Hatten (1994: 187) calls it) thatcloses Beethoven’s Op. 95: deflating bathos that sharply contrasts with thepreceding noble high drama. In both works the abrupt switch to the comic


is a drop from a zenith to a nadir level of discourse. That is romantic irony.Unlike in Beethoven’s Op. 95, the coda in Haydn’s Op. 76 No. 1 doesnot directly follow the serious, dramatic minor music; a few measures inmajor precede it. Yet the precipitous drop from the high genre suggeststhat here Haydn mocks, hence defiles, his own grandiose gesture in thismovement. It represents the serious, pathetic portion in minor, indeed, allSturm und Drang music, as a vain exaggeration. It is as if Haydn confesseshere to perjury, as if he says ‘‘Do not believe me. My sighs were phoney,my tribulations trivial, and my heroic posture ludicrous. It was all a sham.’’

In his last quartet, Op. 135, Beethoven follows the general scheme ofHaydn’s Op. 76, No.1. As in Haydn’s work, all four movements are inthe major mode, but the finale begins in minor. None of Beethoven’s

240 1. 2.

Si repete la seconda parte al suo piacere

247Poco adagio Tempo I.

pizz.

pizz.

pizz.

pizz.

253

Example 9. Beethoven: String Quartet in F Major, Op. 135, iv, mm. 240–58.


earlier quartets has such a plan. The resemblance to that work of Haydnbecomes even more pronounced in the ironic coda that comes after themovement has cadenced in the tonic. That coda lightly dismisses both thequestion (‘‘Muss es sein?’’ in minor, marked Grave) and the answer (‘‘Esmuss sein!’’ in major, marked Allegro) in the finale (Example 9). It startsby presenting the answer motive, for the first time, in a mock-pathetictone; due to the slow tempo, fermatas, and endings on diminished seventhchords, it now sounds like a question. The core of the coda, we suggest,gives the real answer: the naıve melody, played pizzicato, which (as inHaydn’s Op. 76, No.1) sounds like a music-box tune, derides all thatpreceded it, including, most notably, the lofty assurance of the ‘‘Es musssein!’’ answer.

George Edwards regards such music-box endings in Haydn as ‘‘nonsenseendings,’’ claiming that Haydn’s works ‘‘search for resolution or closuremore often than they find it’’ (1991: 228). We suggest that the commonscheme of the above examples divulge romantic irony. The item that thecomposer mocks is juxtaposed to the item that mocks it. That immediate(or almost immediate) succession, where the two items are thrust againsteach other, makes them look like counterparts. In romantic irony theprojector and the projected are one; it is self-reflection that borders onself annihilation. Had the music been funny, ridiculous or gross from thestart, it would have been a parody (denigrating something else) or a mere(non-referential) joke. If the buffoonish situation is relegated to anotherpart in the work, not immediately adjacent to the lofty situation it mocks,we would still see them as counterparts, but not as identical, hence it wouldbe parody rather than romantic irony. It is therefore the amalgamation thatgenerates romantic irony in music. Romantic irony is found, then, wherethere is, first, a sharp contrast between the two items, and second, a motivicand physical proximity between them. These features make us hear thelater musical event as referring to the former, and, through it, to the workas a whole. Reference to an earlier segment of a work is usually achievedthrough motivic connection.²³ When enhanced by physical proximity,we cannot but regard it as identity. We therefore must see the work asferociously turning on itself in a grand act of self-immolation.

²³ For instance, the beginning of the addendum in Beethoven’s Op. 95, F-F#-G, hints at theprevious function of these pitches, mainly the pathos that the F-Gb projects from the very start of thequartet.


References

Bonds, M. E. (1991) ‘‘Haydn, Laurence Sterne, and the Origins of MusicalIrony,’’ Journal of the American Musicological Society 44/1: 57–91.

Booth, W. C. (1974) A Rhetoric of Irony, Chicago, Ill.: University of ChicagoPress.

Brady E. and Levinson, J. (eds.) (2001) Aesthetic Concepts, Oxford: ClarendonPress.

Brauner, C. S. (1981) ‘‘Irony in the Heine Lieder of Schubert andSchumann,’’ The Musical Quarterly 67/2: 261–81.

Burnham, S. G. (1994) ‘‘Mozart’s Felix Culpa: Cosı Fan Tutte and the Ironyof Beauty,’’ The Musical Quarterly 78/1: 77–98.

Daverio, John (1990) ‘‘Reading Schumann by Way of Jean Paul and HisContemporaries,’’ College Music Symposium 30/2: 28–45.

Davidson, D. (1978) ‘‘What Metaphors Mean,’’ Critical Inquiry 5: 31–47.Dent, E. J. (1960) Mozart’s Operas: A Critical Study, London: Oxford

University Press.Dill, H. J. (1989) ‘‘Romantic Irony in the Works of Robert Schumann,’’

The Musical Quarterly 73/2: 172–95.Edwards, G. (1991) ‘‘The Nonsense of an Ending: Closure in Haydn’s

String Quartets,’’ Musical Quarterly 75/3: 227–54.Ellestrom, L. (1996) ‘‘Some Notes on Irony in the Visual Arts and

Music: The Examples of Magritte and Shostakovich,’’ Word & Image 12:197–208.

Glucksberg, S., and Keysar, B. (1993) ‘‘How Metaphors Work,’’ inA. Ortony (ed.), Metaphor and Thought (2nd edn), Cambridge: CambridgeUniversity Press, 401–24.

Hatten, R. S. (1994) Musical Meaning in Beethoven, Bloomington, Indiana:Indiana University Press.

Herzog, P. (1995) ‘‘The Practical Wisdom of Beethoven’s Diabelli Varia-tions,’’ The Musical Quarterly 79/1: 34–54.

Imeson, S. (1996) ‘‘The time gives it proofe’’: Paradox in the Late Music ofBeethoven, New York: Peter Lang Publishing.

Kierkegaard, S. ([1841], 1989) The Concept of Irony, With a ContinualReference to Socrates, ed. and trans. H. V. Hong, and E. H. Hong,Princeton, NJ: Princeton University Press.


Kinderman, W. (1987) Beethoven’s Diabelli Variations., Oxford: ClarendonPress.

Kinderman, W. (1996) ‘‘Beethoven’s High Comic Style in Piano Sonatasof the 1790s, or, Beethoven, Uncle Toby, and the ‘Muckcart-Driver’,’’Beethoven Forum 5: 119–38.

Kivy, P. (1980) The Corded Shell: Reflection on Musical Expression, Princeton,NJ: Princeton University Press.

Kramer, J. D. (1973) ‘‘Multiple and Non-linear Time in Beethoven’s Opus135,’’ Perspectives of New Music 11/2: 122–45.

Lakoff, G., and Johnson, M. (1980) Metaphors we live by, Chicago, Ill.:University of Chicago Press.

Lochhead, J. (1979) ‘‘The Temporal in Beethoven’s Opus 135: When AreEnds Beginnings?,’’ In Theory Only 4/7: 3–30.

Longyear, R. M. (1970) ‘‘Beethoven and Romantic Irony,’’ in P. H. Lang(ed.), The Creative World of Beethoven (essays originally published in TheMusical Quarterly), New York: W. W. Norton & Co. Inc., 647–64.

Mellers, Wilfrid (1983) Beethoven and the Voice of God, New York: OxfordUniversity Press.

Miller, C. L. (1976) ‘‘Ironic or Not?,’’ American Philosophical Quarterly 13/4:309–13.

Muecke, D. C. (1969) The Compass of Irony, London: Methuen.Parker, M. (2002) The String Quartet: Four Types of Musical Conversation,

1750–1797, Aldershot: Ashgate.Sawyer, J. E. (1999) ‘‘Irony and Borrowing in Handel’s Agrippina,’’ Music

and Letters 80/4: 531–59.Schindler, A. F. ([1860], 1996) Beethoven As I Knew Him, ed. D. W.

MacArdle, trans. C. S. Jolly, Mineola, New York: Dover.Scruton, R. (1997) The Aesthetics of Music, Oxford: Oxford University Press.Searle, J. (1978) ‘‘Metaphor,’’ in A. Ortony (ed.), Metaphor and Thought

(2nd edn), Cambridge: Cambridge University Press, 83–111.Searle, J. (1979) Expression and Meaning, Cambridge: Cambridge University

Press.Sheinberg, E. (2001) Irony, Satire, Parody and the Grotesque in the Music of

Shostakovich: A Theory of Musical Incongruities, Aldershot: Ashgate.Sibley, F. N. ([1959], 1978) ‘‘Aesthetic Concepts,’’ in J. Margolis (ed.),

Philosophy Looks at the Arts, Philadelphia, Penn.: Temple UniversityPress, 64–78.


Stravinsky, I. and Craft, R. (1960) Memoirs and Commentaries, London:Faber and Faber.

Suurpaa, L. (1996) ‘‘Schumann, Heine, and Romantic Irony: Music andPoems in the First Five Songs of Dichterliebe,’’ Integral 10: 93–123.

Tovey, D. F. (1972) Essays in Musical Analysis: Chamber Music, London:Oxford University Press.

Walton, K. (1991) Mimesis as Make-Believe, Cambridge, Mass.: HarvardUniversity Press.

Watkins, G. (1994) Pyramids at the Louvre: Music, Culture, and Collage fromStravinsky to the Postmodernists, Cambridge, Mass.: Harvard UniversityPress.

Wittgenstein, L. ([1922], 1961) Tractatus Logico Philosophicus, trans.D. F. Pears and B. F. McGuinness, Oxford: Blackwell.

Wollheim, R. (1986) ‘‘Imagination and Pictorial Understanding,’’ Proceed-ings of the Aristotelian Society 60 (suppl. vol.): 45–60.

Wollheim, R. (1987) Painting as an Art, Princeton, NJ: Princeton UniversityPress.

Wollheim, R. (1998) ‘‘On Pictorial Representation,’’ Journal of Aestheticand Art Criticism 56: 217–26.

Zemach, E. M. (1999a) ‘‘Look, This is Zeus!,’’ in K. Krausz, andR. Shusterman (eds.), Interpretation, Relativism and the Metaphysics ofCulture, Amherst, NY: Humanity Books, 311–33.

Zemach, E. M. (1999b) ‘‘Metaphors and Ways of Life,’’ in J. Hintikka(ed.), Aspects of Metaphor, Dordrecht: Kluwer, 243–54.

Zemach, E. M. (2001) ‘‘A Modal Theory of Metaphor,’’ Theoria 67: 60–74.Zemach, E. M. (2002) ‘‘What Do We See On TV?,’’ in R. Lorand (ed.),

Television: Aesthetic Reflections, NY, Washington: Peter Lang Publishers,89–106.

Further Reading

Bribitzer-Stull, M. (2004) ‘‘Did You Hear Love’s Fond Farewell? SomeExamples of Thematic Irony in Wagner’s Ring,’’ Journal of MusicologicalResearch 23/2: 123–57.

Diener, B. S. (1992) ‘‘Irony in Mozart’s Operas’’ (Ph.D. dissertation), NewYork: Columbia University.


Finson, J. W. (1994) ‘‘The Intentional Tourist: Romantic Irony in theEichendorff Liederkreis of Robert Schumann,’’ in R. L. Todd (ed),Schumann and His World, Princeton, NJ: Princeton University Press,156–70.

Hoeckner, B. (1997) ‘‘Schumann and Romantic Distance,’’ Journal of theAmerican Musicological Society 50/1: 55–132.

London, J. (1996) ‘‘Musical and Linguistic Speech Acts,’’ The Journal ofAesthetics and Art Criticism 54/1: 49–64.

Lowe, M. (2002) ‘‘Falling from Grace: Irony and Expressive Enrichment inHaydn’s Symphonic Minuets,’’ The Journal of Musicology 19/1: 171–221.

Nestrovski, A. (c.2000) ‘‘Beethoven’s Ironies,’’ in D. Greer, I. Rumbold,and J. King (eds.), Musicology and Sister Disciplines: Past, Present, Future,New York: Oxford University Press, 428–38.

Spitzer, M. (1996) ‘‘The Retransition as Sign: Listener-OrientedApproaches to Tonal Closure in Haydn’s Sonata-Form Movement,Journal of the Royal Musical Association, 121/1: 11–45.

PA RT IV

New Issues


9

Music and Electro-sonic ArtG ORD ON G RA HA M

In this paper I want to discuss two questions. First, is electro-acoustic music,music? And, second, how important is the answer we give to this question?Though the term ‘electro-acoustic music’ now has a wide currency, usingit to explore this issue is clearly a recipe for confusion, and so I shall usethe term ‘electro-sonic art’ instead. The questions thus become: is sonic artthe same as music? And does it matter whether it is or not? Or, to put thesecond question in a slightly different way, is music the whole of sonic art?I shall begin with some reflections on the nature of music.

1

Music is for listening to as nothing else is. Such an assertion is neitherstartling nor novel and may even sound banal. In fact, I mean it to bea commonplace and not to be confused (at least at this stage) with anysophisticated contention in the philosophy of music. For example, PeterKivy has identified a ‘prescriptive listening code embodied by the aestheticattitude’, and argued persuasively that music which is specially meant tobe listened to in accordance with this code—the music of the concert andthe concert hall—is not music per se, but one phase of its history anddevelopment, albeit an enormously important one (Kivy 2001: 55). Myproposition, by contrast, means only to assert that (pace John Cage perhaps,to whom further reference will be made at a later stage) an essential,indispensable feature of music is that it is heard (rather than seen, touched,smelt, tasted or otherwise apprehended).

210 gordon graham

The proposition that music is for listening to as nothing else is, shouldnot be confused with the false proposition that music is only for listeningto. On the contrary, music is one of the performing arts, and most composedmusic has been written first and foremost for performance. However, thedominance of ‘art music’ (that has an audience) over folk music (thatgenerally does not) has given listening to music a certain prominence andthis is what makes the indispensability of listening worth stressing. Because,in view of what people often say about music, that music has to be heard isone of those propositions illustrative of the fact that truisms are not alwaysobvious. There is a recurrent tendency for both theoretical explanations ofmusic in general and musicological interpretations of particular pieces topass beyond the experience of listening, and to think of it as primarily ameans to an experience of some other kind.

There are two familiar candidates for what other kind of experience thismight be—the emotion the music expresses is one, and the images or ideasrepresented in it another. Of these two candidates, emotion is the moredominant, despite the early onslaught of Hanslick (Vom Musicalisch-Schonen,1854) and the much more recent arguments mounted by Kivy (1989) andothers. It is rare to find a programme note that does not try to identify boththe emotional origins and the emotional content of a composition. Theorigins are usually sought in the composer’s biography (or autobiography),and the content of the music is analysed in terms of emotional statesexpressed within it and likely to be experienced by the audience. Whenformalism about music was pre-eminent, such commentary tended tobe dismissed as mere ‘music appreciation’, but this appeal to emotion isnot an idiosyncratic feature of those who write programme notes; almosteveryone’s first attempt at explaining the nature and value of music iscouched in terms of emotional expression, and it is an explanation withwhich sophisticated philosophies of music have often persisted.

Sometimes, this kind of analysis is combined with an account of repre-sentation in sound—the use of music to paint pictures and tell stories asin (supposedly) programme music. Occasionally, the intellectual contentof music is construed, even, as going beyond mere representation to theincorporation of moral, political, or religious ‘ideas’. Generally, though,expressivism remains the dominant view because even when some elementof representation or philosophizing is thought to be present, it is usually inthe feelings associated with the pictures, stories or ideas that the meaning

music and electro-sonic art 211

of music is ultimately said to lie. In some cases such intellectual contentis almost coincidental—the peasants in Beethoven’s Pastoral symphony,for example; and in some it is highly intentional—the paintings depict-ed by Mussorgsky, for example, or the idea of resignation expressed inBeethoven’s late quartets. But the essential idea is the same: that music, likepigment, is a material medium through which something else is realized.This thought was expressly articulated by Wassily Kandinsky, who tellsus that ‘all the arts are identical. The difference [between them] manifestsitself by the means of each particular art ... Music expresses itself by sound,painting by colours etc.’ (quoted in Chipp 1968: 346–7).

This idea that sound is the means of musical expression whose real contentlies elsewhere seems to me an important mistake. Part of the purpose ofthis paper is to trace some of its implications, but to do so with an eyeto the issues surrounding tonality and the innovations of atonality andelectro-acoustic composition.

2

There is a powerful argument against any suggestion that musical com-position and performance are means to some other form of experience,emotional or intellectual. This may be termed ‘the redundancy argument’because it turns on the idea that construing music in either way makeslistening to it redundant in principle. Suppose that when we listen to (say)the second movement of Elgar’s Cello Concerto, we are hearing ‘an old man’smelancholy’ (as Elgar himself declared); or that we are ourselves experi-encing the emotion of sadness (as many commentators on this movementhave asserted), and that it is in this experience that the meaning or valueof Elgar’s composition lies. There are clearly other ways of experiencingthe same thing—having Elgar himself tell us of his melancholy in words,for example, or feeling sad in the same way by recalling something thathas happened to us in the course of life. If such possibilities are alternativemeans to the same end, it follows that, though Elgar’s music may be oneroute to this experience, it is entirely dispensable should these other meansbe available to us.

The same point may be made about other supposed musical meanings.Wilfred Mellers has claimed that Beethoven’s late quartets ‘say’ the same

212 gordon graham

as T. S. Eliot’s Four Quartets (Mellers 1988: 657–8). If so, the music isnot uniquely valuable. We can read Eliot’s poems instead of listening toBeethoven’s music. In this way (quite contrary to his intention) Mellers’sattribution of meaning makes the music redundant. So too with the strictlyrepresentational. If Mussorgsky’s music truly is a depiction of (for example)the Great Gate at Kiev, then we have a choice; listen to the music or lookat the picture. Neither is essential, in which case we can ignore the musicso long as we have the picture (and vice versa of course).

‘The redundancy argument’ is my name for a long established and familiarline of thought. Its first articulation is to be found (in a slightly differentform) in Hanslick. But in my view its force has not always been properlyrecognized. This is perhaps the result of its being construed as an unnecessarydemonstration of a self-evident proposition—that sound is indispensable tomusic. This is, rather, the incontestable proposition on which it is based.The argument itself is intended to show that any attribution to music of non-aural meaning or content—emotion or depiction for example—carries theimplication that this self-evident proposition is false. The identification ofmusical significance with emotional experience or intellectual content leadsto the absurd conclusion that in principle we can get everything music hasto offer without actually listening to it.

Anyone who wants to resist this implication has available at least twosorts of reply. The first is to deny that the proposition from which theimplication is drawn is as truistic as I have alleged. The fundamentalcontention, it might be said, is not about sound, but about music—thatthe content of music is itself music and nothing but music. Far from beinga truism, this contention is just one of a number of theories of musicalcontent. However, it is only the second half of this contention that conflictswith other theories of content. While there can be competing theoriesabout what the content of a composition can coherently be said to be, itsbeing a piece of music is an indispensable part of its having the value it does.Roger Scruton, for instance, argues at length in favour of the view that‘The expressive qualities of a work of music form the most important partof its content’ (Scruton 1997: 344). A competing theory (L. B. Meyer’sperhaps) would be that the structural properties of a work are the mostimportant part of its content. But both theories rightly take it for grantedthat it is musical expression and musical structure that we are talking about,and the important point is that this musicality is essentially, not contingently


related to its value or importance. Whatever it is we get from it, we cannotget it in non-musical form.

A second response to the redundancy argument (with respect to expres-sivism) would be that the relation between music and emotion is moresophisticated than I have allowed. Music, it might be claimed, is not merelya means to the expression and apprehension of emotion. But what elsecould the relation be? One familiar line of thought is that the redundancyargument does not rule out the expression of emotion as realized in music.What exactly does this mean, though? What it must not mean, obviously,is ‘realized in a way not susceptible to the redundancy argument’, sincethis would simply beg the question. What it could mean is that emotionalterms seem to be directly applicable to music, and not simply to thestates of mind of composers and audiences. A minor chord is naturallydescribed as ‘sad’, irrespective of the feelings of those who play or hearit, and people spontaneously describe much more extended musical piecesin a similar way. However, all this shows, I think, is that the language ofemotion, whose home is in the context of human feeling, can (perhapssurprisingly) be applied to music as well. It does not actually show that thefull implications of the originating context are preserved in the extensionof that language to music. And if there were reason to think that they are,we would still have to assert that emotion ‘realized’ in this particular wayrequires to be heard.

This contrasts with those instances in which, for example, music is usedmerely as a signal. Famously, the opening bars of Beethoven’s Fifth meantsomething special as D-Day drew near, and were intended to do so. Anon-aural signal could have performed the same role. But if there is someemotion denoted that is intimately connected with those opening bars, therealization of it in this form is uniquely valuable. In short, music is forlistening to as nothing else is—the proposition with which I began.

3

What could give music this indispensably aural character? The answer Iwant to explore is that music precisely consists in intentionally organizedsound. This is a familiar contention, but it needs both explication andrefinement. Jerrold Levinson writes as follows:

214 gordon graham

It has often been suggested that music be defined as organized sound. But this ispatently inadequate. While sonic organization has some plausibility as a necessaryfeature of music, it is hardly a sufficient one. The output of a jackhammer, theticking of a metronome, the shouts of a drill sergeant during a march ... are allorganized sounds, but not instances of music.

(Levinson 1990: 269–70)

Levinson himself is willing to accept the ‘necessary but not sufficient’interpretation. For present purposes, his intuitive rejection of alternativetypes of sonic structure as music is a matter yet to be discussed. However,the important point is this: even if we do interpret intentional sonicorganization as only a necessary condition of something’s being music, itcan still be the case that its sonic character is an essential part of its value.In the same essay Levinson invites us to ask why music is important to us.His answer is that ‘actively engaging with organized sounds, intentionallyproduced for such purpose, can enrich experience’ and he identifies thefollowing forms of enrichment: conveying emotional content, stimulatingthe ordering faculties of the mind, providing insights into the mysteriesof the psyche, making one pleasurably aware of one’s body, securingpermanent alterations in outlook and attitude (ibid.: 278). Now all of theseends can be obtained in other ways and cannot therefore secure the uniquevalue of music. What the intentional organization of sound alone secures(and what is not expressly mentioned here) is the enrichment of strictlyaural experience. It is for this reason that listening is essential, and for thisreason that the sonic properties of music are indispensable.

It is this fundamental thought that the redundancy argument rests upon.Let us agree that music may prompt in us emotional experiences orintellectual insights. Yet if and when it fails to do so, it can still be valuedas an extension and exploration of aural experience itself. Moreover, itsessentially aural nature allows us to explain important differences betweenthe creations and explorations of different composers. National anthems can(and often do) stir people deeply, and the music of ‘avant garde’ composersmay be full of ‘ideas’. But the ultimate mark of a superior composition isthat it is more worth listening to. What makes it ‘more’ worth listening tois not a simple matter of quantity of notes or instruments, but the creationof sounds that we naturally have to describe in metaphorical or analogicallyextended language. This is a point worth stressing. The fact that pure or


absolute music can be described as bright or architectonic or mournful isspecially interesting not because the music is in some way connected withlights, or buildings or even, as is so commonly supposed, with emotions(even if it is so connected), but because it shows that sound—puresound—can have properties far more sophisticated than those of volume(loud or soft), pitch (high or low), and tempo (fast or slow).

A further, similar and crucial point is to be made about performancewhose role is to reveal the sonic properties of a musical composition in a waythat no musicological analysis of the ‘meaning’ of the piece or study of theemotional life of the composer could do. A gifted performer enables us todiscover the existence of interesting and unusual sonic properties that left toourselves we might have missed. That is how a particular performance canbe specially worth listening to. Composers imagine and performers realizesonorities that it may well be natural to describe as exhilarating, melancholic,agitated and so on, but in the experience of listening it is nonetheless themusical sound itself that is the principal object of the attentive mind.

By contrast, in speaking, noise-making can be a mere means of expressionor communication. What matters is the meaning of what is said, not theaural quality of its saying. It is true that focus on phonetic propertiesmay be too narrow. Sameness of ‘what is said’ does not make twoexpressions interchangeable. For example, ‘rabbits are animals’ and ‘bunniesare animals’ could be said to say exactly the same thing, but the latter wouldbe inappropriate in a zoology textbook (an example I owe to StefanoPredelli). Plainly, there are cases in which what is communicated dependsnot only on what is said, but on how it is said. Moreover, there is poetry, inwhich aural properties do matter, and there are some natural languages inwhich pitch makes a difference to meaning. Still, for present purposes thesecomplications can be set aside. The irrelevance of accent, rhythm and pitchin ordinary English is enough to make the point. While the high-pitchedutterance of a sentence spoken quickly in a Scottish accent says exactlywhat the same sentence slowly uttered in a low-pitched Welsh accent says,it is precisely such variables that matter in music.

Some of Steve Reich’s compositions are interesting in this connection.Reich has used his celebrated technique of ‘phasing’ to separate the soundof speech from its meaning, and thus turn it into a form of music. Hecontends that, so transfigured, language offers us a ‘window on the humansoul’. But a more natural thought would be that untransfigured language is a

216 gordon graham

window on the soul (when we let it be), and that the interest in transfiguredspeech lies in the special aural rather than semantic properties it has.

Listening to music is a peculiarly human phenomenon. Animals otherthan humans can listen as well as hear, of course, but while we have everyreason to think that their being equipped with auditory powers is enoughto let them hear music, its sophisticated structural properties mean that onlyhuman beings can listen to it. This is why reflection and analysis can helpus to hear the properties of a piece of music. Indeed where the musicis of a highly developed sort we may need the vocabulary of technicalanalysis to isolate and describe the structures of sound to be found withinit. Nevertheless, if we are to know that these sounds have that property,listening is inescapable. And so it should be of course. The problem withthe familiar alternative explanations of music’s value is (as I argued) thatthey make the music redundant. Emotional ‘highs’ are to be had elsewhere;what the music ‘says’ can be said more intelligibly in words; what it is saidto depict can be painted; any values that it exemplifies, such as beauty orgracefulness, are exemplified in other arts. But once we locate the value ofmusic in its unique ability to extend and explore aural experience, we haveattributed to it a distinctive value. Since we cannot have that experiencenon-aurally, the activity of actually listening to music is ineliminable.

However, in the business of listening both composer and performerhave a crucial role. Just as the painter directs our visual perceptions, so thecomposer and performer direct our aural perceptions. Listening to musicis not just a matter of sound pouring into a receptor, but of the mindbeing directed through a series of perceptions. We are, so to speak, steeredthrough our experience. It is as though the composer were saying ‘It mustbe heard this way’ by actually making us hear it that way. An analogy mightbe this: we enter a series of underground caverns where our journey cantake alternative routes through spaces of differing shape, dimension, andatmosphere, lighted by different means. Each composer is the guide whodecides upon the lighting and the order in which we pass from cave tocave; the performer is the guide who leads us through the caverns.

This analogy touches upon an important debate in the ontology ofmusic: is composition a matter of creation or discovery? The parallel withthe underground caverns would seem to favour the sort of Platonism inmusic that Peter Kivy has defended (Kivy 1993: Ch. 2). The shapes anddimensions of the caves are there for all to see, but they can be seen only


this way or that. The way we see them and the ways that are especiallyworth seeing, are matters for which we rely upon our guides. Applyingthis to music it seems that the melodies and harmonies are ‘there for all tohear’, and only await the composer and the performer to point them out.Now, while I am sympathetic to Platonism in music (and to Kivy’s defenceof it), it does not seem to me that the centrality of listening to music requiresus to think in this way. The parallel with the caverns is only intendedheuristically, so to speak. It is consonant with thinking of composition (andimprovisation) as pure invention, the creation ex nihilo of musical structuresthrough which we are led. Either way, we can correctly be said to enjoyenriched aural experience.

But whether we think of musical composition as invention or discovery,there is a further question. Is it only music that can enrich aural experience?Is music truly unique in this regard? Hitherto the answer has perhapsseemed obvious, but recent technological innovations can be construed asraising a doubt on this point. There are now other possibilities, and thatis why we have to deploy the term ‘sonic art’. By ‘sonic art’ I mean anyauditory construction whose purpose is to explore and enrich the world ofaural experience. Does this necessarily mean music?

4

It might be thought that the answer to this question depends upon ourdefinition of music. Clearly, if music is defined as any sonic construction,then the answer is an uninteresting ‘yes’. But to include any piece ofintentionally organized sound under the category of music begs certainquestions. Levinson (as noted earlier) thinks that such a definition hascounter-intuitive implications—‘The output of a jackhammer, the tickingof a metronome, the shouts of a drill sergeant during a march ... are allorganized sounds, but not instances of music.’ It is not clear, however, thatthis differentiation is supported by his own definition of music, namely:

sounds temporally organized by a person for the purpose of enriching or intensifyingexperience through active engagement (e.g.) listening, dancing, performing withthe sounds regarded primarily, or in significant measure, as sounds.

(1990: 273)

218 gordon graham

In recognition of this fact, Levinson amplifies his definition with a distinc-tion between what is music and what can be treated or regarded as music:

One way to ignore this distinction is to claim, in the spirit of John Cage’s Zen-inspired reflections, that any and all sounds are music. This is simply false ... . WhatCage shows perhaps is that any sounds can be listened to as if they were music ... Itdoes not follow that all sounds are at present music. The whirr of my blender andthe whistling of the wind are not instances of music.

(1990: 275)

The qualification ‘at present’ may be of some significance. Levinson’scounter instances are plausible because most people perceive an evidentdifference between music properly so called and ‘the whirr of my blenderand the whistling of the wind’, even if the difference is one they findhard to articulate. On the other hand, it is precisely these other types ofsound that composition using digital technology is keenest to deploy andexploit. As digital compositions become more widely heard (as film scores,for example), it may be that the sounds classified as ‘music’ extend beyondthose so regarded ‘at present’.

The main difference between music traditionally understood, and thesounds that characterize more recent compositions (intuition tells us) istonality, a difference equally detectable by the tutored and untutoredear. Most people incline to Hanslick’s view that ‘[t]he material out ofwhich the composer creates, of which the abundance can never be exag-gerated, is the entire system of tones, with their latent possibilities formelodic, harmonic and rhythmic variety’ (Hanslick 1854: 28). Accord-ingly, they reserve the label ‘music’ for tonal sounds. Furthermore, mostpeople (I imagine) share Roger Scruton’s opinion that ‘[t]onality pro-vides a paradigm of musical organization ... [a]nd attempts to depart fromtonality, or to discard it entirely, seem only to confirm its authorityover the musical ear’ (Scruton 1997: 239). That is why the expression‘electro-acoustic music’ is confusing, and the term ‘electro-sonic art’ tobe preferred. Two further issues arise, however. First, given the historyof composition in twentieth-century music, it is highly contentious tomake tonality the paradigm of music. This is precisely what the advocatesof atonality, Schoenberg chief among them, expressly denied. Second,even if the perceptible difference between music and electro-sonic art issufficient warrant for using different terms to refer to them, this tells us


nothing about their respective aesthetic value. I shall discuss these issues inturn.

Scruton, as is well known, defends tonality against atonality and heargues that if those composers who were (are) persuaded that tonality isexhausted do succeed in extending composition and performance in newdirections:

it is either because tonality can be extended or because its effects can be preservedthrough the kind of oblique tonal thinking shown by Skriabin, Stravinsky, Brittenand Berg. The possibility remains that tonal music is the only music that will everreally mean anything to us, and that, if atonal music sometimes gains a hearing, itis because we can elicit within it a latent tonal order.

(Scruton 1997: 308)

A preference for tonality is widely regarded as indefensibly conservativeand is usually countered by the observation that popular responses to atonalcomposers simply repeat the experience of the greatest of tonal composers.If near riots greeted the first performance of Stravinsky’s Rite of Spring, thisis no more adverse a reaction than Beethoven’s late Quartets prompted. Yetwhile episodes from the history of music undoubtedly provide a salutarycounter to unthinking conservatism, there does seem to be some basis forregarding the twentieth century as innovatory to a degree unmatched byearlier centuries. As Gerald Abraham has remarked:

[e]ven such a drastic innovation as the Florentine recitativo was gradually assimilatedto the melodic tradition in ariosos and finally in Wagner’s scores. Monteverdicould employ his prima prattica and seconda prattica side by side; the Westerntradition was strong enough to absorb both. [But then] the first signs of a moreserious crisis showed early in the twentieth century. A sense that the greattradition was approaching a dead end was reflected in Debussy’s non-functionalharmony, in Skryabin’s experiments with a completely new harmonic system, andin Schoenberg’s revolt against tonality.

(Abraham 1979: 820)

However, it is not with Schoenberg or Skryabin that the most radicalinnovations are to be found. If we include in the definition of ‘tonal’ music(as Scruton does) the existence of a tonic, a dominant tone, then atonalmusic is indeed atonal. But odd though this may sound, atonal music iscontinuous with tonal. That is why it is less misleadingly called ‘serialism’.

220 gordon graham

The difference is the method of composition. Scales and chords are aban-doned in favour of a self-imposed system of composition and organization.But even if the end result sounds repellent to the conventionally educatedear, it is still tones that are being serialized. A more complete break withtonality is to be found in composers such as Varese and Lutoslawski.Varese declared himself to be engaged in a ‘fight for the liberation ofsound and for my right to make music with any sound and all sounds’(www.zakros.com/mica/soundart/s04/varese−text.html). Accordingly, hiscompositions Deserts and Poem electronique consist in amalgamated sounds ofthe sort that have become much more familiar with the advent of digitaltechnology. Not only is there nothing we could recognize as melody orharmony, and no key or mode, if the individual sounds out of which it iscomposed are isolated, they are not found to be tones either, but more likeLevinson’s jackhammer or blender.

Varese seeks ‘a whole new world of unsuspected sounds’. Similarly, inthe second movement of Lutoslawski’s Trois Poemes d’Henri Michaux singingis abandoned for speaking and shouting. Lutoslawski has devised a notationwhich instructs the ‘choir’ about pitch and volume, a notation that looksa little like ‘music’, and another special notation for the percussion thataccompanies it. The effect is highly dramatic and aurally impressive, butas the need for special notation indicates, the composition signals a moveaway from music, precisely because it seeks to go beyond the confines ofresonant tones. To repeat, serialism devises deliberate rules of compositionthat bear no relation to established harmonic structures. This makes itsoutcomes strange to the ear raised on Bach and Beethoven, which is whyit is readily described as atonal. But it still takes the twelve tones of thechromatic scale as the basic sonic material. It is with compositions such asDeserts and Trois Poemes, not atonalism, that we move to non-musical sonicart, a move that finds its full realization with the advent of recording anddigital technology.

Varese is widely regarded as a founding figure in electro-sonic art, butwith hindsight the devices at his disposal seem exceptionally primitive. Still,the gap between his compositions and those of even the most innovativeserialists seems to me radical enough to warrant replacing the term ‘music’with that of ‘sonic art’. It has been widened almost immeasurably bythe deployment of new technology. The most telling difference between(for instance) serial composition and Steve Reich’s ‘phasing’ is that the

www.zakros.com/mica/soundart/s04/varese_text.html


latter is not the application of rules of composition, but a technologicalprocess for combining recorded sounds. Digital composition widens thegap still further; in preference to using actual recordings it manufacturesand manipulates its own sounds. At the end of this development, it seemsto me, we are left with things so different that it makes for confusion tobracket music and electro-sonic art together. But if the two are not thesame, how do they compare?

Varese was a classical musician by training who saw his work standingin the long tradition and denied that he had any desire ‘to disparage andeven to discard the great music of the past’. ‘No matter how original, howdifferent a composer may seem,’ he says, ‘he has only grafted a little bitof himself on the old plant’ (www.yale.edu/oham/CD1transcript.html).But notwithstanding the sincerity of this assertion, the continuity is hardto hear, the graft and the plant hard to identify. In my view, however,this does not matter, or to be more precise, it does not matter from anevaluative standpoint. The nature and the status of music and electro-sonicart are separate questions. We can and should conclude, I think, thatthe aspirations of Varese, now that they have finally been realized withdigital technology, have indeed resulted in something wholly new. A moredifficult question is whether this wholly new thing should be accorded thekind of value that hitherto music properly so called has uniquely enjoyed.

5

The first point to be made of course is that electro-sonic art fits Levinson’sdefinition of music—sounds regarded primarily as sounds and temporallyorganized for the purpose of enriching or intensifying experience throughactive engagement (which I have modified to read ‘aural experience’).The evaluative question, accordingly, is twofold: does electro-sonic artoffer as much as music does by way of exploring and enriching our auralexperience? And does it offer the same or similar opportunities for activeengagement?

One way of answering the first of these questions is to consider howmuch of the language that we use to analyse and describe music can alsobe used in the analysis of electro-sonic art. As far as analysis is concerned,it seems to me clear that the two share a good many of the same structural

www.yale.edu/oham/CD1transcript.html

222 gordon graham

properties; theme, variation and recapitulation are obvious cases in point.Counterparts to a tune and its accompaniment are harder to discern, butinsofar as some such relationship can be characterized in terms of relativeaural importance, it seems plausible that electro-sonic art can containsimilarly related aural sequences.

On the matter of description, the position is perhaps less clear. Here wereturn to the matter of expressiveness. I have argued against the familiaridea that music is a medium of emotional experience. This seems to me apost-Romantic fallacy. But as I noted earlier, the mistake lies in a fallaciousinference from an undoubted fact. The undoubted fact is that musicalsound can be described in affective language. To repeat, a minor chord isrightly and naturally described as ‘sad’, but we cannot infer from this alonethat composing, playing or hearing a minor chord is in any wider way anexperience of sadness. It is the same mistake, though far less obvious, asone that would be made by anyone who tried to compare the noise levelsof loud and soft colours. The language is right; the reasoning is wrong.

But it is precisely because the language is right that we are able to marvelat the richness of music, whose strictly aural properties can be described insuch diverse ways. Can the sound constructions of electro-sonic art sustaina similarly rich descriptive vocabulary? I do not know what the answerto this question is, partly because electro-sonic art is in relative infancy.Electro-sonic constructions are rightly described as captivating, intriguing,exciting, even haunting. Whether they are, or can also be moving, tranquil,transcendent, humorous, and so on, are matters waiting to be resolved. Itis upon their resolution that the question of electro-sonic art’s significancewill ultimately turn. Enthusiasts will say ‘of course they can’, but morecompelling evidence will come from the ways in which, eventually, a muchwider cross-section regularly makes use of electro-sonic compositions.

In addition to the applicability of structural analysis and expressivedescription to electro-sonic art, there is a further substantial question aboutactive engagement. Music is a communal as well as a natural product. Askedto name the really ‘big’ figures in the world of music, we immediatelycome up with composers—Bach, Beethoven, Mozart. Asked to identifyoutstanding ‘musicians’, we naturally think of performers—Pavorotti orJacqueline du Pre, say. Yet even the greatest names will have had teacherswhose names are virtually unknown, and the meaning and value of thecomposer’s and performer’s work requires realization in the ears of the


listener. Music exists to be heard, and neither its composition nor itsperformance can simply spring into existence by virtue of the ingenuity ofindividuals. Even the categories of composer, performer, teacher and listenerdo not exhaust all the forms of participation. There are also those, knownand unknown, who have invented, made and perfected the huge varietyof instruments whose distinctive sounds, both alone and in combination,produce (so to speak) the matter out of which this world is made.

Is the same range of participation available in electro-sonic art? There issome reason to think that the answer might be ‘no’. I shall consider onlythe case of performance. Before the advent of digital technology moderncomposers employed a number of devices to extend the range and characterof sounds that appear in their compositions. For instance, the ‘prepared’piano has nuts and bolts and other items inserted into the instrument, andis played as much by striking and plucking inside the piano itself as bydepressing the keys, which are in any case thumped by arms and elbows aswell as being played by fingers in the normal way. Some composers haveincluded noise producers that are not instruments, for example, vacuumcleaners, bat and ball, and gun shots. Others have included passages ofscreaming and shouting as well as singing.

These innovations are sometimes derided. Whether they should be ornot is of no special concern here. Rather, the important point to note inthe present context is that such innovations are child’s play in comparisonto what can now be done by means of digital technology. But unlikeits forerunners, the composition of this kind of sound is inextricablyconnected with the technology which makes it possible. Composers likeLutoslawski had to devise systems of notation that would instruct singers inthe production of sounds; composition which deploys digital technologyhas no need to do this. More tellingly, there is no place for it, becausethe composition and its realization are one and the same. In other words,there is no gap, either temporal or conceptual, between composition andperformance, the sort of gap that there is between a musical score andits realization. What this implies, however, is that electro-sonic art hasno counterpart to what has been an essential feature of music, namelyperformance, and by eliminating performance, electro-sonic art eliminatesinterpretation.

Why does this matter? It matters not just because for most people activeengagement beyond listening means performance and not composition, but

224 gordon graham

also because interpretation in performance is a significant further avenuefor the exploration of sound. Conventional music, in other words, providesfor a collaborative exploration of sound by composer and performer (andarranger, sometimes). In digital composition, by contrast, the composerdoes everything. This can be true of musical sounds as well as non-musicalones. The music of Vangelis, for instance, cannot be performed; it canonly be listened to. But this is a non-standard case. By contrast, directlyrecorded composition, like Reich’s, is characteristic of non-musical sonicart. What this means is that electro-sonic art is more limited than music inthe opportunities it offers for active engagement.

This contention is likely to be disputed by electro-sonic composers, whowill point out that increasingly electro-sonic art is not simply listened to overheadphones, but ‘performed’ in the sense that the digital product is played toan audience through different arrangements of speakers in different acous-tics. It is true that for the most part it is composers themselves who performin this way, but that might change. As it reaches new maturity, peoplemay emerge who are distinguished performers of electro-sonic art but notthemselves composers. Perhaps so, though the nature of the medium lendssome credence to the view that performance and composition can never beseparated in the way that they are in conventional music. Many great com-posers were fine performers, but by no means all, and there is nothing toprevent someone composing a violin sonata who cannot play the violin. Bycontrast, a digital composer must have mastered the very same technologythat needs to be deployed in performing it, at least in the sense described.

Digital technology is a source of enrichment insofar as it allows com-posers to overcome the aural limits of conventional musical sounds, asit unquestionably does. Nevertheless it may bring with it a still greaterlimitation yet. Since its products are to an important extent fixed and final,they give relatively little scope for something we value—participationin active music making. If so, then, whether we call it music or not, aworld in which digital composition gradually came to dominate, thoughtechnically much more sophisticated, would in fact be a poorer one. But justfor that reason, those who subscribe to the view that the value of musiclies in the world of sound that it enables us to explore can persist with ahigh estimation of its value even in the world of digitally generated andmanipulable sounds. And for just the same reason, it is unlikely that peoplewill ever cease to value conventional composition.


References

Abraham, G. (1979) The Concise Oxford History of Music, Oxford and NewYork: Oxford University Press.

Chipp, Herschel B (ed.) (1968) Theories of Modern Art, Berkeley, LosAngeles and London: University of California Press.

Hanslick, E. (1854) Vom Musicalisch-Schonen, trans. G. Payzant (1986), Onthe Musically Beautiful, Indianapolis: Hackett.

Kivy, P. (1989) Sound Sentiment, Philadelphia, Penn.: Temple UniversityPress.

Kivy, P. (1993) The Fine Art of Repetition: Essays in the Philosophy of Music.Cambridge: Cambridge University Press.

Kivy, P. (2001) New Essays on Musical Understanding, Oxford and NewYork: Clarendon Press.

Levinson, J. (1990) ‘The Concept of Music’, in his Music, Art and Meta-physics, Ithaca, NY, and London: Cornell University Press.

Mellers, W. (1988) Man and His Music (revised edn), London: Barrie andJenkins.

Scruton, R. (1997) The Aesthetics of Music, Oxford and New York:ClarendonPress.

Further Reading

Kivy, P. (2002) Introduction to a Philosophy of Music, Oxford: OxfordUniversity Press.

Madell, G. (2002) Philosophy, Music and Emotion, Edinburgh: EdinburghUniversity Press.

Sharpe, R. A. (2000) Music and Humanism, Oxford: Oxford UniversityPress.

10

Thoughts on RhythmROG E R S CRUTON

1

Perceptual information must be assembled into comprehensible units if it isto guide us around the world. This is as true of the ear as of the eye; henceGestalt principles operate in the auditory as in the visual sphere, thoughapplied to temporal rather than spatial configurations (Bregman 1990;Gelfand 1998). Some philosophers might wish to speak of a ‘nonconceptual’or ‘preconceptual’ unity in the auditory Gestalt—even of a nonconceptualcontent (Peacocke 1983: Ch.3). For the unities and wholes that we hear are,in the first instance, presented under no description. When, for example, aregular sequence of scale steps is interrupted by noises that mask individualpitches, we hear a continuous sequence, interrupted by foreground noise,rather than discontinuous sections of a scale. This perception does notrequire us to conceptualize the sequence as a scale, as musical movement,or indeed as anything else. It requires only that we group together thesounds that precede the noise with those that follow it.

It is clear, therefore, that we could group sounds into coherent temporalfigures without hearing them as music. Hence there is a real questionas to what must be added to transform an experience of sound into anexperience of music. To say that we must hear the sounds as music iseither vacuous or false (false if it implies that we apply that very concept,a concept which many a singing, listening, dancing infant has yet toacquire). I argue elsewhere that sounds become music when organizedrhythmically, melodically or harmonically—with the implication that eachform of organization is sufficient to provide an experience of music. But Ialso suggest that these forms of organization pertain to the intentional rather

thoughts on rhythm 227

than the material object of perception. Melody is something that we hearin a sequence of sounds, and is not something that would be mentionedin a description, however complete, of the sounds themselves, judged asitems in the material world (Scruton 1997).

2

Our ways of grouping and streaming individual sound events—both ineveryday perception and in musical attention—reflect a peculiar metaphys-ical feature of sounds, which is that they are ‘pure events’. Unlike a carcrash, a sound is not a change in something else. It is a self-subsisting object.When a sound occurs we assume that there is some physical cause. But wedo not need to identify that cause in order to identify the sound, and thesound is fully intelligible as sound without reference either to the causeor to any other spatio-temporal particular. (Sounds should be compared inthis respect with tastes and smells.)

One consequence of this is that auditory streams can be organizedinternally, by reference to audible features of the sound events, and withoutinvoking any order in the objects that produce them. An illustration isprovided by the octave illusion studied by Diana Deutsch (Deutsch 1982).In this experiment headphones are placed over the subject’s ears, and thenotes of a descending scale and an ascending scale played in each ear, inthe patterns shown on lines 1 and 2 (Example 10). The two ears thereforereceive, respectively, the inputs shown on lines 3 and 4. What they hear,however, are the two continuous sequences of lines 5 and 6. It is as thoughthe sounds gravitate towards neighbours, where ‘neighbourhood’ is definednot by the physical proximity of the causative events, but by adjacent placeson the pitch spectrum. Yet the sequences as heard are played into neitherear, and represent no causally unified process in the physical world. Theauditory Gestalt is not merely incongruous with the physical events thatproduce it. It is organized according to principles that are intrinsic to theworld of sounds, and which would be operative even if there were nophysical events that could be identified as the causes of the individualsounds.

The streaming that occurs in Deutsch’s experiment is not yet musicalstreaming. This kind of grouping by pitch proximity and sequential

228 roger scruton

Example 10. Diana Deutsch: pattern hearing.

‘flow’ could occur even in the experience of the wholly unmusicalperson—perhaps even in the experience of an animal. Musical experience,I suggest, involves the importation of a spatial framework, and the organi-zation of the auditory field in terms of position, movement and distance.Those spatial concepts do not literally apply to the sounds that we hear.Rather they describe what we hear in sequential sounds, when we hearthem as music. In other words the concepts that provide the fundamentalframework for musical perception are applied metaphorically, and I haveelsewhere tried to give an account of what this means and why it is impor-tant (Scruton 1997). The relation between auditory ‘streaming’ and theperception of musical movement can be likened to the relation betweenthe perception of shapes and the perception of their figurative content.In the second case you see something in the shape that in the first case youmerely see as a shape. In like manner the musical experience arises whenyou hear movement in a stream that you might otherwise hear merely as a


stream.¹ (The word ‘stream’ here is not a metaphor, but a primitive termdenoting a continuous temporal Gestalt.)

The musical order emerges when we adopt the ‘acousmatic’ attitudeto the world of sounds,² attending to sounds without focusing on theirmaterial causes. There is a virtual causality that governs musical movement,as when one note in a melody is heard to bring its successor into being, eventhough sounded on another instrument in another place, and this virtualcausality organizes the acousmatic Gestalt. The acousmatic experience isassociated with a certain kind of listening, in the concert hall or at home.But it occurs also when musicians play together and become united in thefirst-person plural of the band, the music flowing through them as thoughwith a force of its own (Schutz 1964: 159–78). It occurs too when wedance to music, and match the movement of our body to the movementthat we hear. Dancing is both a response to musical movement and a wayof understanding it as movement. And it is partly through its connectionwith the movement of the body and its social meaning that music acquiresits moral character, a point made in other terms by Plato.

3

My subject in what follows is rhythm, and the distinction between rhythmsimposed by metre, and rhythms generated by musical movement. By metreI mean the measuring and parcelling out of the temporal sequence. Not allmusic has a metre, and not every metre is like the metres familiar in Westernmusic, which govern the divisions and subdivisions that correspond to timesignature and bar-line. There are musical traditions that derive metricalpatterns by adding note-values and not, as we do, by dividing larger unitssymmetrically. In classical Arab music, for example, rhythmic cycles arecomposed of time units added together to make often asymmetrical patterns,which do not permit whole-number division (Wright 1980: Sections 1–4).

¹ On the distinction between ‘seeing in’ and ‘seeing as’, see Richard Wollheim’s account of‘representational seeing’, in Wollheim (1987). For reasons given in Ch. 5 of Scruton (1997), I do notregard ‘hearing in’ as sufficient foundation for musical representation, and in general prefer to discussthese phenomena as varieties of ‘double intentionality’, rather than ‘representational perception’.

² The term ‘acousmatic’, introduced in this sense by Pierre Schaeffer in Schaeffer (1966), is explainedin Scruton (1997, Ch. 1).

230 roger scruton

The metres (talas) classified by Sharngadeva, the thirteenth-century Indiantheorist, and co-opted in our time by Messiaen, are likewise formed throughaddition rather than division, so that the rhythmic unit is assembled fromnote-values, rather than deduced by whole-number division within a bar.³Yet more complicated are the patterns of African drum music, in whichmultiple metres are sustained indefinitely, producing patterns which cannotbe assigned a single measure. (See the examples notated by Jones (1959),who convincingly argues that drum music is composed of repeated metricalunits, with bar lines that fail to coincide.)

Metre is—as its name implies—a form of measurement, in which time-values are ordered into repeatable segments. Christopher Longuet-Higgins(1987: 150–68) has argued that we understand metrical organization by akind of ‘generative grammar’, through which smaller units are derived in arule-governed way from larger.

Thus I hear the first note of Parsifal (Example 11) as the second of fourcrotchet values, and therefore locate it on a weak beat, after the strongbut silent beat that begins the bar. The off-beat experience is intensifiedby the tie, which pushes the second note on to the second quaver valueof a further subdivision within the bar. I do not carry out this calculationconsciously. Nevertheless by unconsciously latching on to the generativehierarchy I am able to assign a measure, a beat and a temporal value to thenotes that I hear, and thereby strain after the music as it flees my attempt tofix it to a downbeat, catching up with it, so to speak, only at the beginningof the third bar.

Whatever the plausibility of that account, we should recognize that itapplies only to the kind of metrical organization with which we are familiarfrom our own tradition, founded on the division and subdivision of the bar.

Example 11. Wagner: Prelude to Parsifal.

³ See Grosset (1913), and the discussion of modern Indian tala in Fox Strangways (1914: 191–224).


The oriental traditions referred to above introduce metres that are generatedby addition, not division, and which therefore impose quite another orderon the musical line—one in which syncopation cannot easily take root.Moreover, even if we confine ourselves to works in our own tradition, wemust acknowledge that metrical order and rhythm are distinct. In graspingthe rhythm of the opening phrase of Parsifal I am distinguishing accentedfrom unaccented beats; I am responding to differences in stress, hearingcertain notes as part of an ‘upbeat’, others as initiated by a downbeat, andgrouping the tones into separate rhythmical units, as in Example 12.⁴ Theseother aspects of rhythm belong not to number but to life: they are featuresof the virtual energy that flows through the music, and which causes me tomove with it in sympathy.

The ancient musicologist and pupil of Aristotle, Aristoxenus of Taren-tum, argued, in this connection, that rhythm is a temporal order imposedupon, though not inherent in, the sequence in which it is heard.⁵ Thesequence is ‘rhythmized’ by the perceiver, by dividing it into up-beatand down-beat, and assigning a variable duration to each (the first alwaysshorter than the second). It is clear from Aristoxenus’s account that he hasdancing primarily in mind, up-beat and down-beat being explicitly con-nected to the arsis and thesis—lifting and falling—of the foot; nevertheless,his discussion reminds us that rhythm is a phenomenal, not a mathematical,property of a sequence, and that our capacity to perceive it is dependentupon our wider ability to respond to movement.

Often the metre is an ad hoc attempt to place bar-lines across an organismthat has no such divisions, as in certain parts of the Danse sacrale from theRite of Spring (which, incidentally, is measured out somewhat differently in

Example 12. Wagner: Prelude to Parsifal.

⁴ The term ‘up-beat’ is contentious, and can mean one of at least two things: the entire preparatoryphrase that precedes some musical emphasis, or the short intake of breath that leads directly into it. Thefirst use began with J. J. de Momigny and was elaborated into a theory by H. Riemann, in Riemann(1903), to be further adapted by Edward T. Cone in Cone (1968).

⁵ Aristoxenus, Elementa Rhythmica, surviving fragments of Book 2 translated in Barker (1989), vol. 2.

232 roger scruton

Example 13. Bartok: Music for Strings, Percussion and Celesta, 1st movement.

Example 14. Boulez: Le marteau sans maıtre.

the four-hand piano score and the orchestral score). Consider the openingtheme of Bartok’s Music for Strings, Percussion and Celesta (Example 13).Here the bar-lines and time-signatures seem like a tentative analysis of stresspatterns, rather than an authoritative division into beats. This impression iseven more evident in the measures from Le marteau sans maıtre (Example 14),where time-values are established by addition and not division within themusical line. (A comparable example is given in Example 15, from Lemaıtre sans marteau by Nabil Az-Zeloub, a poignant lament over the deathof Jacques Derrida.) In such works the time-signature is like a piece of


Example 15. Az-Zeloub: Le maıtre sans marteau.

speculative commentary placed above the score, and identifies no featureof the musical process. An analysis of Boulez’s rhythms, for example,would show them to be derived by the superposition of two or moremetres, each formed by addition of time-values, rather than division ofthe bar. There is a real question as to whether we hear the result as arhythm at all: paradoxically, the attempt to escape from metrical divisionhas led, in Le marteau, to the cancellation of rhythm by metre, though ametre hidden behind the time-signatures that straddle the bars. Here themathematical order seems to arrest the forward movement, rather than toguide it. (Likewise with Az-Zeloub, who uses in this piece the himar metreof his native Algeria, transformed by successive augmentations which arethen rearranged in a permutational sequence. The result is again entirelya-rhythmical.)

234 roger scruton

In the classical tradition metrical divisions coincide with significantjunctures in the rhythm. However, as I suggest below, this coincidencecan occur in two quite distinct ways. In one kind of case the metre is laidacross the movement like a grid. In another kind of case it emerges from themovement, as though ‘precipitated out’, in the way that the hexagonal cellsof the honeycomb are precipitated out of the honey-making within them.

4

The musical phenomena that we group together under the rubric ofrhythm have their counterparts in other areas of human activity. Stress andmeasure occur in dancing and also in the movements required by certainkinds of physical work, such as those remembered in sea-chanties, in theoccupation songs of India, and in the Negro spirituals (Bucher 1909). Stress,accent, metre and grouping all occur in speech, and speech-rhythms areboth patterns and constraints when set to music. The rhythms of hymntunes are classified according to their syllabic rather than their metricalstructure since, from the liturgical point of view, it is more importantto know what texts can be sung to them than to assign a time-signatureto their musical line. And folk rhythms reflect syntactical features of theassociated languages (Bartok 1931). I suspect that there might be a usefulcontrast to be made between composers whose rhythmic organizationprimarily reflects dance patterns, and those for whom speech patterns aremore important. Tchaikovsky, Dvorak, Elgar and Stravinsky belong to thefirst kind; Mussorgsky, Wagner, Janacek, Puccini, Britten and Schoenbergto the second. (In Stravinsky’s Les Noces, for example, the composer usesnonsense syllables, precisely in order to reconstitute speech-rhythms asdance-rhythms. In Stravinsky it is not words that give the meaning ofthe dance, but dance that gives the meaning of the words.) A. H. FoxStrangways has argued that classical Indian music uses time-value ratherthan accent to emphasize a note because that is how you emphasize asyllable in Sanskrit. He explains the complex Indian talas as derived fromthe verse metres of Sanskrit liturgy and poetry, in the way that the metresdanced by the Greek chorus reflect the quantities exhibited in the verse.⁶

⁶ See the illuminating discussion in West (1992, 129–59).


Music supporting movement and music supporting words will differ inmeasure, accent and stress. In dancing and marching small-scale repetitionsare necessary, whereas in song the strophe overrides small-scale regularitiesand encourages their variation. If there were nothing to rhythm savemeasure each piece of music could be assigned a clear and unambiguousrhythmic character simply through tempo and metre. But this wouldbe to ignore the effects, not only of grouping and stress, but also ofmelodic and harmonic organization which, in our tonal tradition, exerttheir own gravitational pull over the rhythmical movement. A very obviousinstance of this is the device of suspension, in which a phrase is held backfrom its melodic or harmonic closure across a metrical closure, as in thestraightforward example of repeated suspensions from Elgar in Example 16.The result is a tie across the bar line—hence a form of syncopation—andalso a clash of movements, as metrical closure anticipates the harmonicclosure that follows. Atonal music, which (officially at least) does notadmit suspensions, cannot generate this kind of rhythmic pulse. Nor canit reinforce rhythmical closure by creating melodic and harmonic closuresthat coincide with it.

The rhythmic effect of harmony is not confined to cases where theharmony is explicitly stated. It can be witnessed in an unaccompanied line,such as that from Parsifal in Example 11. Here the first seven notes (all offthe beat) arpeggiate the triad of A flat, to which they add first the sixthand then the major seventh, creating an implied dissonance that is resolvedon to the C minor triad, as G, C and E flat are sounded successively in anemphatic down-beat. The intense up-beat experience here is dependenton the cadence implied in the melodic line and could not be reproducedmerely by reproducing the note values.

Example 16. Elgar: Violin Sonata, 1st movement.

236 roger scruton

5

The considerations raised in the previous section suggest a use (thoughnot the only possible use) for the distinction between ‘beat’ and ‘rhythm’.‘Beat’ denotes a pattern of time-values and accents, while ‘rhythm’ denotesthe movement that can be heard in that pattern, and which may beinfluenced by harmony and melody so as to reach across metrical closuresand establish contrary motions of its own. A piece of music may havea strong beat but little or no rhythm, and some of the most rhythmicalpieces in our tradition are characterized by a light beat and a refusalto emphasize the bar-line. Often a composer will present accompanyingfigures which shift the accent sideways through the music, as in the triofrom Dvorak’s New World Symphony (Example 17), where a rhythmicalcell, replete with melodic and harmonic meaning, imposes its own micro-metre on the over-arching triple time. To see the point of this it issufficient to imagine how uninteresting Ravel’s Bolero would be, wereit to consist only of the underlying beat, overlayed by a melody thatexactly ran in its groove. The rhythm of this piece is generated withinthe never-ending syncopated melody, which plays against the beat like asquash-player against a wall. The beat is what makes the rhythm possible,but in itself it is without rhythmic interest. When Jazz musicians distinguishbeat from ‘swing’ it is partly this that they have in mind. It has beenplausibly argued by Gunther Schuller (1968) that swing involves a feel forthe residue of African polyrhythm that underlies Jazz syncopation. But itshould not be thought that swing can be reduced to a metrical outline.Swing is a phenomenological feature of the music. It is not understoodby counting but felt in the musical line, and Louis Armstrong’s muchquoted remark, that if you don’t feel it you’ll never know what it is,conveys a profound truth about the nature of rhythm. Metre is the frame;rhythm the life that grows on it. Jazz idioms like ‘swing’ and ‘groove’ are

Example 17. Dvorak: New World Symphony, 3rd movement.


Example 18. Sibelius: Violin Concerto, 3rd movement.

therefore adopted in order to remind us that rhythm is not reducible tometre.

In this connection we should draw a contrast between ostinato rhythm,in which a relentless beat subjects the music to a discipline that mighthave nothing much to do with its melodic and harmonic movement, andrhythm which adapts to and takes its accents from the musical movement.Stravinsky in Oedipus Rex sustains the chorus with an ostinato 6/8 beat,on G and B flat, and the result is rhythmical but static. The beat is like anexternal force, constraining the music from outside. It is the steady marchof fate that can be deflected by nothing in the action.

This does not mean that ostinato is a substitute for, or denial of,rhythm, or that it can be understood without those other features—accent,stress and grouping—which I earlier mentioned as fundamental to therhythmical order. On the contrary, even the simplest ostinato can beheard in competing ways, if accent and grouping are left ambiguous bythe melodic line. (Witness the sustained ostinato of the last movement ofSibelius’s Violin Concerto, a fragment of which is given in Example 18.)Such examples show that measure is never sufficient in itself to determinerhythm, even when made fully explicit to the ear.

6

There is an extreme case of the ostinato phenomenon, in which rhythmseems to become detached from harmonic and melodic organization, so asto be fired at them from outside, as it were. I refer to rhythmic ‘backing’,

238 roger scruton

as exemplified by a certain style of pop. This might have a mechanical‘tic-toc’ character, as in ‘No Son of Mine’ by the group Genesis (fromthe appropriately named album, We Can’t Dance); or it may depend uponmixing, in which melody and harmony are smeared together so as tobecome indistinct, leaving only percussive ostinato to establish some kindof measure—as in ‘Be Here Now’ by the group Oasis. Here rhythm hasfallen away from the music altogether, leaving a bare shell of melody anda harmonic progression without cogent voice leading, both overwhelmedby the percussive noise from next door.⁷

We encounter in the idiom adopted by Oasis a disaggregation of musicinto beat and pitch, neither giving true support to the other. The drum-kit has the role of marshalling the music to its stride, forbidding alldeviation, while adding nothing to the melodic or harmonic development.(Contrast the Dvorak; here the lilting phrase that generates the rhythm isalso replete with melodic and harmonic implications, which subsequentlyunfold through the melodic line.)

The use made of the drum-kit in contemporary pop is to a great extentan innovation. Classical jazz introduced the drum-kit as an embellishmentto a pre-existing rhythm, often sounded on the off-beat and hidden, as itwere, behind the strumming of the banjo. The rhythm was generated bythe syncopated voices of the instruments, each of which played its part inbreathing rhythmic life into the bar-lines. Strictly speaking, New Orleansjazz has no need of percussion, which it uses—if at all—purely ornamen-tally, and in deference to the African polyrhythms which, according toSchuller’s plausible view of the matter, are remembered in the syncopations(Schuller 1968). In modern pop percussion has a constitutive, rather than anornamental use: without it, there would be no music, since the beat—onwhich everything depends—would not exist. (All that, it seems to me, isalready implied in the word ‘backing’.)

In the early days of rock you find a jazz-like use of the drum-kit—notto impose a rhythm synthesized outside the melodic line, but to emphasizeand vary a rhythm generated within it. The locus classicus of this is ElvisPresley, whose extraordinary voice, with its barely perceptible micro-rhythms and tremors, produces melody and rhythm together, so that the

⁷ The prohibitive cost attached to the copyright of pop music prevents me from giving examples. Ihave been interested to discover that the worse the music, the more expensive it is to reproduce it.


one is inseparable from the other. In ‘Heartbreak Hotel’, for example, therhythm is compellingly announced by the solo voice, and the bass andpercussion seem merely to take it up and prolong it.⁸

Equally impressive in this respect is Eric Clapton, who uses the guitarrather as Elvis uses his voice, to set the music in motion before the drum-kitenters, so establishing the rhythmical identity of the piece as an internalfeature (‘Lay Down Sally’ is an effective instance). It is interesting to notethat Clapton, in his days as leader of Cream, was criticized by Jimi Hendrixas rhythmically incompetent—a criticism taken up by David Hendersonin his biography of Hendrix (Henderson (1983) ). Cream was disbandedin 1969, and Clapton turned his back on Hard Rock, an idiom which inany case destroys the need for rhythmic competence. The guitarist that hasemerged from this is surely one who has ‘got rhythm’ in just the sense thatGershwin intended.

7

The dance-forms adopted by Bach and Handel were attached to elaboraterituals and courtesies, and required complicated steps and formations fromthe dancers. In the ancient dances to which Debussy, Ravel and Respighilooked back with such poignant emotions, partners were assigned bycourtesy and exchanged by rule, with people of all ages participatingwithout embarrassment in a dance which could at any moment placethem side by side and hand in hand with a stranger. In a very realsense the dancers were generating the rhythm that controlled them, andgenerating it together, by attentive gestures governed by a ritual politeness.The experience of dancing as a ‘dancing with’ (usually with a group,subsequently with a single partner) survived right down to the days of rock’n roll. This ‘withness’ of the dance is captured by the baroque and classicalidioms, in which rhythmic organization is not imposed but extracted bythe metre, as song-like phrases weave around and embellish one another,moving to closures of their own.⁹

⁸ Presley’s vocal style—heavily influenced by both R&B and Gospel—is a subject in itself, addressedin part by Henry Pleasants in Pleasants (1974: 270 ff ), and by Peter Guralnick, in Guralnick (1982, 1987).

⁹ On the ‘withness’ of the dance see Scruton (2000: 164–76).

240 roger scruton

The human need for this kind of dancing is still with us, and explainsthe current craze for Salsa as well as the periodic revivals of ballroomdancing and Scottish reels. The ‘withness’ of the reel was noticed andcommented upon by Schiller, who regarded what he called ‘English’dancing as confirming the connection between beauty and gentility. Hiswords are worth quoting:

The first law of gentility is: have consideration for the freedom of others, the second;show your freedom. The correct fulfilment of both is an infinitely difficult problem,but gentility always requires it relentlessly, and it alone makes the cosmopolitanperson. I know of no more fitting image for the ideal of beautiful relations than thewell danced and multiply convoluted English dance. The spectator in the gallerysees countless movements which cross each other colourfully and change theirdirection wilfully but never collide. Everything has been arranged so that the first hasalready made room for the second before he arrives, everything comes together soskilfully and yet so artlessly that both seem merely to be following their own mindwhile never impeding the other. This is the most fitting picture of a maintainedpersonal freedom, which respects the freedom of others.

(Schiller 2003: 173–4; I have slightly amended the translation)

It is undeniable that, for many if not most young people, the experienceof ‘withness’ is absent from their dancing, which typically involves neithercomplicated steps nor formations. The normal dancing of the disco floorinvolves little or no contact with or recognition of a partner, and mayoccur with no partner at all. You dance to Heavy Metal by head-banging,slam-dancing or ‘mashing’ (pushing people around in the crowd). Suchdancing is not really open to people of all ages, but confined to the youngand the sexually available. Of course, there is nothing to forbid the oldand the shrivelled from joining in: but the sight of their doing so is anembarrassment, all the greater when they themselves seem unaware of this.

The social impoverishment of disco has a rhythmical cause. The pulseof disco music sets the dance in motion and controls its beat, but it doeslittle to suggest how or with whom you should move. The dance, likethe rhythm, remains external to the music, a kind of generalized ‘settingin motion’ rather than a balletic commentary on the musical line. You seethis at its most extreme in techno music, especially when embellished withstrobe lights and similar psychedelic effects. Such dancing is like throwingoneself into a pool of collective emotion, to be swept away in its frenzy.


There is nothing you can do, either to create or to embellish the rhythm.And communication with a partner is rendered impossible by the noise,the lights and the sheer formless press of the crowd to every side. Schillersaw ‘English’ dances as emblematic of personality, freedom and the civiccommunity. If we were to view disco dancing through the same Kantianspectacles we should describe it as ‘pathological’—an event in whichfreedom is displaced by empirical causality, personality by nature and willby desire. Sarabandes, galliards and reels were social dances in the very realsense of being society-forming dances. Modern rock is crowd-forming,rather than society-forming.

8

Dancing shapes the body rhythms and trains the ears of those who engagein it, and changes in the dance-culture will lead of their own accord tochanges in the rhythmical organization of music—even the music of theconcert hall. This is what we witnessed in the nineteenth century, whengipsy rhythms affected the music of Brahms (explicitly, in works like the Gminor piano quartet, but implicitly in many of the other chamber works,songs and concerti), and when the waltz exerted an ubiquitous fascinationthat was to begin with Schubert’s drawing-room waltzes for piano, and toculminate in Ravel’s monumental La Valse.

This is one of the factors that must be borne in mind when we considerthe relation between serious and popular music today. People have becomeused to the ostinato beat of pop, which throbs in the background of lifeand shapes the expectations of all of us, like it or not.

It is hard to attract modern audiences to music in which rhythm is eithermelodically generated or measured out in Messiaen’s way, by additionrather than division of time-values; it is comparatively easy to attract themto music with an ostinato propulsion, regardless of its melodic or harmonicinvention. Hence the popularity of John Adams, whose ‘The ChairmanDances’ (from Nixon in China) and ‘Short Ride in a Fast Machine’ typifya new kind of ostinato writing, with the melodic instruments assignedessentially percussive tasks, and with continual repetition of elementalrhythmical cells. For the pop-trained ear this music is easy to listen to,since its rhythmic structure does not have to be deciphered by following

242 roger scruton

the melodic line but is imposed by a regular external emphasis. Composersused to be wary of ‘the tyranny of the bar-line’; in Adams, however, thetyranny is accepted, as a benign dictatorship which gives us all what wewant.

9

This bears on the controversy surrounding atonal music. Much musictheory conceives tonality as a melodic and harmonic system, in whichboth harmony and melody are goal-directed. Chords in this system arenot simultaneities, but nodes in a web, bound by functional relations tothe chords that precede and follow them. Musical order is achieved whenvoices move together towards closures that are harmonically driven andmelodically complete. Theories of tonality—such as that defended bySchenker—see these features as giving the essence of tonal music. Hencethey have a tendency to leave rhythm out of the picture (Schenker 1979;Salzer 1952–62). And those who propose atonality as a genuine alternativeto the tonal tradition often describe it in similar terms, as replacing thehorizontal and vertical relations of pitches with a new order based on thepermutation of pitch-class sets. Rhythm is again left out of the picture,often with counter-intuitive results, since the experience of rhythm is anexperience of grouping, and will therefore affect the segmentation of themusical surface on which the pitch-class structure depends.¹⁰

It seems to me, however, that we should see tonality as in part arhythmical system, and recognize that the difficulties of atonal music areexperienced as much at the rhythmical as at the melodic and harmoniclevel. The harmonic and melodic principles of tonal music arose out of thedesire to make satisfying sequences, in which movement begins, continuesand comes to a conclusion—not only on the large scale but also regionby region and voice by voice. Rhythm is not an addition to this, buta part of it, both sustaining and sustained by the harmonic and melodicrelations. Measure is useful not only because it corresponds to familiarmovements of the body but also because tonality is founded in repetition,

¹⁰ See Nicholas Cook’s devastating demolition of Allen Forte’s set-theoretic analysis of Stravinsky’sExcentrique, in Cook (1987: 138–51).


with the music constantly returning to identifiable places, and constantlydeparting on some new but related journey of its own. Take away theold tonal order, however, and rhythm has a tendency either to collapseentirely—since it is not driven by the forward movement of the musicalline—or else to become external to the music, a frame supporting an inertdisplay of musical fragments.

The atonal composer therefore has a serious dilemma. To suspendorganized ‘pitch class sets’ on an ostinato frame is to take a big stepbackwards into banality. On the other hand, to reconstitute rhythm outsidethe expectations fostered by the tonal order—expectations of grouping,closure, accent and so on—is to lose the connection between rhythmand the life-phenomena that give sense to it. In response to this problemMilton Babbitt has produced ‘serialized’ rhythms in which the twelve pitchintervals are set in relation with twelve equal time intervals, and bothsubjected to systematic permutation. However, the very procedure showsthat it is not rhythm, but measure, that is being reconstructed. Serializedrhythm is really serialized metre, and is as little part of the rhythmicsurface of the music as is the drum-kit in synthetic pop. It seems to methat Messiaen’s rhythmical experiments were prompted at least in part byan awareness of this problem. Messiaen sought to reinvent rhythm as aninternal feature of the musical line, even when the melodic and harmoniclanguage had abandoned all the old forms of closure.

Messiaen’s experimental approach was taken further by Stockhausen,who attempted to reconstruct the rhythmic dimension of music by local-izing movement in competing orchestral units. Stockhausen’s large-scalemusic shifts great blocks of sound through musical space, often with anindifference to human life and bodily sensations that reminds one of aslave-minder building a pyramid. The result is a meticulous, though hid-den, metrical order, which generates an entirely a rhythmical surface. Ifyou were to introduce rhythmic motion into a piece like Gruppen, forexample, it would have to be an external rhythm, like a pop-ostinato,laid on top of the musical structure but generated outside it. In itself themusic has no enduring pulse; the emphasis generated within one episodedoes not survive to the next, and therefore cannot combine with it ina dance step. There is a curious parallel here, between Stockhausen andJohn Adams: both think of rhythm in ostinato terms, the one thereforerejecting it, as an extra-musical device, the other accepting it as the sole

244 roger scruton

organizing principle of a musical surface that in all other respects is whollymonotonous.

10

Like his teacher Olivier Messiaen, Stockhausen uses metres formed bythe addition rather than division of time-values. The result may bea-rhythmical, but it is meticulously measured. Messiaen’s music, on theother hand, is not merely highly rhythmical but organized rhythmically, sothat rhythm permeates the music with the same intensity as it permeatesa symphonic movement by Beethoven. Messiaen has himself written ofthis subject, emphasizing the way in which rhythm can be altered through‘added values’ and ‘non-retrogradeable rhythms’ so as to defy metricalsymmetries, and compel the listener to cling to the musical movement,rather than the bar-line, in following the score (Messiaen 1956).¹¹

Messiaen took the idea of additive metre from the decı-talas ofSharngadeva (see above, Section 3), in which the notion of a beat isreplaced by that of the shortest note-value (matra), and metres are classifiedas repeatable sequences, of which Sharngadeva lists 120. Some of the talas areextremely long—no. 35, simhanandana, has 21 separate ‘beats’, comprising62 matras—and could not, therefore, be heard as self-contained rhythmicalunits, on the model of the bars in Western music.¹² And Messiaen’s useof them does not compel us to hear rhythmic groups in his music thatcorrespond exactly to the metrical order: for example, in the TurangalılaSymphony at its most rhythmical, we tend to hear curtailed polyrhythmsbased on division of the bar, rather than the underlying additive order.

Messiaen’s principal device is illustrated in Example 19, showing themethod of ‘added values’. The first sequence (Example 19 a, b and c)shows three bars in standard metre; the second (d, e and f) shows thetransformation of these bars either by adding a rest or lengthening oneof the notes by a fraction. Example 20 illustrates the method in practice:Messiaen’s transformation of a Peruvian folksong, Delirio (Example 20a) inthe second of the Harawi songs (Example 20b).

¹¹ Messiaen’s theories and practice are lucidly presented in Johnson (1989: Ch. 4).¹² The talas are listed in Appendix 2 to Johnson (1989).


Example 19. Rhythmic augmentation.

(a)

(b)

Example 20. (a) Delirio (Peruvian folk song). (b) Messiaen: Harawi no. 2 ‘Bonjourtoi, colombe verte’.

In such examples time-signature is clearly of little importance.Example 19f, for example, can be assigned a time-signature; indeed theperformer will be helped, up to a point, if the bar is prefaced withthe metrical sign 13/16. But this measure misrepresents the rhythmicalmovement: the bar has six beats, each subdivided and one of themstretched. By recursively adding or subtracting note-values in this way,Messiaen attempts to build character into his motives, which develop overtime by acquiring or losing note-values, while retaining the same numberof ‘syllables’ overall. (Compare the way in which a face changes over time,as age first strengthens and then weakens the flesh, always retaining a certainrecognizable outline.)

Example 21 illustrates the method of ‘non-retrogradable rhythms’. HereGroup B is the retrograde of Group A, with the central five semi-quaver note shared between them. The whole sequence forms a rhythmic

Example 21. Non-retrogradable rhythms.

246 roger scruton

palindrome which is unchanged when played in retrograde, and hence‘non-retrogradeable’. Again this metre can be assigned a time-signature,but the result (19/16) is even less informative than in the previous examples.What we hear is a rhythm which seems to flow back into itself, constantlyreturning the energy that propels it to its source. Messiaen’s rhythms arecompletely emancipated from the bar-line, and the metrical markings ofhis scores say little or nothing either about the underlying additive metre,or about the movement that can be heard in it.

The result owes its rhythmic vitality not to any regular beat that canbe mapped on to bar-lines, but to the energy that flows from note tonote and which is inseparable from accent and grouping, both enforcedby the melodic line. A work like Canteyodjaya (Example 22) offers thesupreme instance of an organization in which rhythmical order has beenemancipated completely from measure, with accent and grouping becomingthe unaided vehicles of the rhythmic movement. Certain commentatorshave described Messiaen’s music of this period as conveying a new, or atany rate unconventional, experience of time: the forward-moving temporalorder of the classical style, which leads to the relentless drive of a Beethovensymphony, has been replaced by a kind of circularity, in which everythingreturns into itself as though in a state of rapt contemplation (Davidson2001). One can see the point of such descriptions, whether or not oneendorses the claim that it is time, rather than musical sequence, that is beingexperienced differently.

Example 22. Messiaen: Canteyodjaya.


However, a word of caution is in order. Messiaen, like Babbitt andBoulez, often uses metre in ways that make no contact with rhythm,as this is perceived by the listener. The metrical schemes deployed inChronochromie, for example, involve permutations of tala-like sequences, ofa kind that cannot be followed by the ear, still less by the body.¹³ If wehear rhythms in these permutations it is not because of the mathematicalorder but in spite of it. Fred Lerdahl has argued that musical understandinglooks for elaboration rather than permutation in its object (Lerdahl 1988),and therefore that serial organization will always be irrelevant to theorganization that we actually hear in the musical surface. Even if not trueof all musical phenomena, Lerdahl’s observation is surely true of rhythm,which we understand through patterns, accents and repetitions, creating amovement in which we can join. The passage quoted from Canteyodjayain Example 22 can be supplied with a full metrical analysis, in terms offundamental sequences and their successive permutations. But the rhythmresides in the groupings of semi-quavers, the pauses, and the accentednotes, which shake us around as we try to fit them to patterns that cannotquite contain them. The experience of rhythm here cannot be capturedby metre, and certainly not by the arcane metrication that preoccupied thecomposer.

Example 23. Bulgarian Christmas carol.

¹³ See the table of permutations set out in Johnson (1989: 177).

248 roger scruton

11

Although Messiaen deserves credit for recognizing that melodic and har-monic innovation require new ways of grouping tones, and that this, inturn, requires new forms of rhythmic organization, we should recognizethat the practice of adding value to notes has been widespread in folk-music, and is by no means confined to oriental traditions. Example 23 givesa Bulgarian Christmas carol, in which the lengthening of certain syllableshas led to the stretching of a bar of four semiquavers to 9/16 and even10/16 time, while retaining the same basic rhythm of four beats to thebar.¹⁴ Fox Strangways, in his closely observed study of Indian classical musicas it was before the disease of Western harmony, persuasively argues thatmeasure is prior to rhythm in the Indian raga because melodies originated ashymns in the Sanskrit language, which emphasizes a syllable by lengtheningrather than stressing it, as in ancient Greek prosody (Fox Strangways 1914).Moreover, the metaphysically conceived tala system of Sharngadeva doesnot represent the rhythmic organization that we hear in Indian classicalmusic, as will be evident to anyone who has swayed to Ravi Shankar.Most of the melodies written down by Fox Strangways can be divided intobars of equal length, with the stretching of syllables marked by a pause.Structurally speaking, it is true, additive measure dominates rhythmicaldivision. But this can be accounted for not only by the peculiarities of theSanskrit language, but also by the early emergence in India of a listeningculture, with dancing as an art to be observed rather than an event to join.

12

The need for rhythmical organization to change with the change inharmonic language was widely recognized in the wake of Messiaen’steaching. But it was not Messiaen who first observed the connection.Wagner was already aware that genuine melodic and harmonic innovationrequire a new approach to rhythm. If Wagner is not often thought of asa rhythmic innovator, this is because the ostinato conception of rhythmhas to a great measure obscured the origins of rhythm in the grouping

¹⁴ Harmonized by Raina Katzarova, and quoted in Bartok (1976: 48).


experience, and obscured the roots of that experience in speech, song anddance.

In fact the path-breaking character of Tristan und Isolde is shown asmuch in the rhythmic organization of the musical line as in melody andharmony. This is so in spite of the fact, and also because of the fact, thatWagner hardly ever uses percussion as a rhythm-generating device. Thetimpani play an enormous role in Tristan, but it is a melodic role, consistinglargely of prolonged tremolandi which swell the melodic line and suggestreserves of unexplored emotion beneath it. And the other instruments ofthe orchestra are used melodically rather than percussively, even in themost frenetic passages such as the scene in which Tristan tears the bandagesfrom his wound, and the music enters a kind of rhythmic catastrophe as itstrives to keep pace with his delirium.

Wagner’s motives in Tristan have pronounced rhythmical contours,but, unlike the motives in Beethoven say, their emphasis is rarely onthe downbeat. Consider the motive sometimes known as the ‘look’(Example 24), which sets the Prelude in motion, after the interruptedcadence on to F major. This establishes a complex rhythm in 6/8 time,with a fractured triplet followed by a crotchet-quaver sigh. The melodicline here, and the chromatic movement in the bass which lifts the harmonyfrom F major to G major, endow the rhythmic pattern of this phrase with akind of completeness. This is a rhythm that has been ‘sung out’, and whichhenceforth bears the memory of the phrases that sang it. Wagner promptlyrepeats it, incorporating it into no less than five subsequent motives, whichhe draws together with the original ‘look’ to form a continuous sequence(Example 25 illustrates the musical process).

Example 24. Wagner: Tristan and Isolde, motive of ‘The Look’.

Example 25. Wagner: Tristan and Isolde, elaboration of ‘The Look’ motive.

250 roger scruton

In the classical style the closures imposed by tonal melody and diatonicharmony are aligned with rhythmical closures reinforced by bar-lines. InTristan all three forms of closure are minimized or avoided entirely: for theyare symbols of a law-governed order that passion has undermined. Hencethe motives that take up the rhythmic organization of the ‘look’ involveties across bar-lines, off-beat accents, and the ubiquitous fracture in thetriplet, that together create a rhythmic profile that cannot be understoodapart from the melodic pulse in which it originates. The prelude to Tristanis already moving in the direction of Messiaen, with the bar-line as theeffect, rather than the cause, of the rhythmic organization, which deployshighly energized and repeatable cells, and in which accent and groupingare often at variance with those suggested by the metre.

13

Rhythm is a property of dancing and also of speech. As Aristotle pointedout, we hear speech differently from other sounds,¹⁵ since speech is anexpression of soul, and is organized semantically. The surface order ofspeech may be highly irregular and punctuated by silences; but it is heard asflowing from a single source. Its movement is the movement of meaning,and its pulse is the pulse of life. Speech is therefore a paradigm for us of arhythmical organization generated not by measure and beat but by internalenergy and the intrinsic meaningfulness of sound. Hence there is a greatdistinction to the ear between speech and chant, in which speech soundsare suspended at a single pitch and on a single pulse. A chanting voicesounds from another region, a place of unseen powers. The ostinato chantrepresents the inexorable order that controls us and which forces our wordsto move in time to its commands.

(One of the most remarkable developments within the sphere of popis the ‘rap’ artist, who speaks toneless rhyming prose along the groove ofa relentless percussive rhythm, so replacing speech-rhythms with chant-rhythms. This is an extreme form of the ostinato experience, in which notonly harmony and melody but speech itself are annihilated by metre.)

¹⁵ De Anima, 420b.


Often Janacek is given credit for being the first composer to build speechrhythms into his melodic line. It seems to me, however, that speech rhythmshave been part of phrasing in classical music at least since Monteverdi andthe stile rappresentativo, and that the conscious attempt to give them musicalform began with Wagner, not Janacek. Wagner himself discusses the matterin Opera and Drama, referring to the accents that are natural to the spokenlanguage as underpinning the musical divisions within the bar (Wagner1887–8: 103–41). His use of Stabreim arose directly from his search for alanguage that could be sung out without losing the inflections of naturalspeech. Stabreim derives from two musically important features of Germaniclanguages, namely their preference for accent over quantity as a form ofemphasis, and their wealth of plosive consonants.

Schoenberg was heir to Wagner in this as in other things. Unlike somany of his successors on the path of atonality, Schoenberg recognized thatatonal music would succeed only if it could emerge as a rhythmical system.His early atonalism was therefore as much a rhythmical as a harmonicexperiment. Schoenberg did not wish to repudiate the connection betweenrhythm in music and rhythm in life. Instead he wished to replace dance-rhythm by speech-rhythm as the organizing principle of the musical line.The instrumentation and harmony of a work like Pierrot Lunaire are dictatedin part by the desire to generate speech-like rhythms in all the instrumentalvoices. Tonal harmony compels voices to move together, to magnetizeeach other, to work towards the same points of closure and stasis—as ina dance. Hence in tonal music, even when speech-rhythm inspires themelodic line, a reminiscence of dance-rhythm will inhabit the harmony.

In order to free the music entirely from this dance-like togetherness,Schoenberg pulls the harmony apart, so that what we hear is a simultaneityof utterances rather than a sequence of harmonizing songs. The instrumentsspeak to us, in rhythms that match the Sprachgesang of the voice. Thelater development of the serial technique likewise has a pronouncedrhythmical meaning. The serialization of the pitch sequence permits a kindof polyrhythmic structure, in which accents in the various voices seldomcoincide, and the unified chorus gives way to a conversation (examples ofinterest: the chorus of the Israelites wandering in the desert in Moses andAaron—contrast the flat and ineffective dance rhythms that accompany theworship of the golden calf).

252 roger scruton

14

Rhythmic organization is fundamental to musical meaning and intimatelyconnected with the moral character of music. The externalized rhythmof much contemporary disco music invites people to move in crowd-likerather than conversation-like ways. You do not ‘move with’ rhythm of thiskind; you surrender to it, are overwhelmed by it. You lose yourself in it, asin a drug. And that is a feature not only of the way disco music is received,but also of the moral character that we hear in it. By contrast the ostinatorhythms in the Bartok piano sonata, for example, have no reality apart fromthe sharpened harmonies and emphatic melodic line that produce them.You do not lose yourself in these rhythms: rather you discover somethingother than yourself—an idealized form of human life, in touch with the soiland with the natural world.

The Darmstadt orthodoxy (as propagated by Stockhausen, Boulez andtheir immediate followers) effectively killed off rhythmical organizationas an intrinsic feature of the musical line. Somehow the dancing musicalline that Messiaen discovered through the talas disappeared from the musicof his followers. The idea of additive metre became institutionalized andmathematized, and accents lost all connection with the body. People canlisten to Stockhausen’s Gruppen and move ‘along with’ it, even be movedby it. But the experience of rhythm is absent from Stockhausen’s block-like sounds, as it is absent from the meticulously metrical music of BrianFernyhough. In reaction, composers like John Adams have tried to addressour need for rhythm by adding rhythm to melodic lines and harmoniesthat do not have the strength to generate rhythm out of their own innermovement. The rhythm is pumped in from outside, not breathed outby the melody. It seems to me that we are placed by this music inthe vicinity of the ‘crowd’ experiences that dominate the world of pop.There are, by contrast, composers still writing who invite us to join inrhythms that are intrinsic to the melodic and harmonic life of their music:David Matthews, for example, in his string quartets and symphonies; JohnBorstlap in ‘Psyche’; Michael Berkeley in his concertos; Oliver Knussen inhis two operas on childhood themes, Higgledy-Piggledy-Pop and Where theWild Things Are. And these works make systematic use of tonal relations,illustrating the thesis that tonality is a rhythmical system, arising when voices


‘move with’ each other, as in a dance. They illustrate both the depth andthe importance of the controversy surrounding musical modernism. This isnot a controversy about harmonic systems only; it is primarily a controversyabout the nature of listening, about the role of music in the shaping of ouremotions, and about the connection between music and life.

References

Barker A. (1989) Greek Musical Writings, London, vol. 2. CambridgeReadings in the Literature of Music (Cambridge: Cambridge UniversityPress, 1984–9).

Bartok, B. (1931) Hungarian Folk Music, trans. M. D. Calvocoressi, London:Oxford University Press.

Bartok, B. (1976) ‘The So-called Bulgarian Rhythm’, in his Essays, ed.B. Suchoff, New York: St Martin’s Press.

Bregman, A. S. (1990) Auditory Scene Analysis: The Perceptual Organizationof Sound, Cambridge, Mass.: MIT Press.

Bucher, K. (1909) Arbeit und Rhythmus, Leipzig und Berlin: BG Teubner.Cone, E. T. (1968) Musical Form and Musical Performance, New York:

Norton.Cook, N. (1987) A Guide To Musical Analysis, London: Dent.Davidson, A. E. (2001) Olivier Messiaen and the Tristan Myth, Westport,

Conn., and London: Praeger.Deutsch, D. (1982) ‘Grouping Mechanisms in Music’, in D. Deutsch (ed.),

The Psychology of Music, New York: Academic Press.Fox Strangways, A. H. (1914) The Music of Hindostan, Oxford: Oxford

University Press.Gelfand, S. A. (1998) Hearing: An Introduction to Psychological and Physiological

Acoustics, New York: Marcel Dekker.Grosset, J. (1913) ‘Histoire de la musique: Inde’, in A. Lavignac (ed.),

Encyclopedie de la musique et dictionnaire du conservatoire, vol. 1, pt. 1, Paris:Delgrave, 287–324.

Guralnick, P. (1982) Lost Highways: Journeys and Arrivals of American Musi-cians, New York: Harper and Row.

Guralnick, P. (1987) Liner notes to Elvis Presley: The Sun Sessions CD(BMG/RCA 6414-2-R).

254 roger scruton

Henderson, D. (1983) ’Scuse Me While I Kiss the Sky: The Life of JimiHendrix. New York: Bantam Books.

Johnson, R. S. (1989) Messiaen, 2nd edn, London: Dent.Jones, A. M. (1959) Studies in African Music, 2 vols, London: Oxford

University Press.Lerdahl, F. (1988) ‘Cognitive Constraints on Compositional Systems’,

in J. A. Sloboda (ed.), Generative Processes in Music, Oxford: OxfordUniversity Press.

Longuet-Higgins, C. (1987) Mental Processes: Studies in Cognitive Science,Cambridge, Mass.: MIT Press.

Messiaen, O. (1956) Technique de mon langage musical, Paris: Leduc.Peacocke, C. (1983) Sense and Content: Experience, Thought and their Relations,

Oxford: Oxford University Press.Pleasants, H. (1974) The Great American Popular Singers, New York: Simon

and Schuster.Riemann, H. (1903) System der musikalischen Rythmik und Metrik, Leipzig:

Mayer und Wigand.Salzer, F. (1952–62) Structural Hearing, 2 vols, New York: Dover.Schaeffer, P. (1966) Traite des objets musicaux, Paris: Editions du Seuil.Schenker, H. (1979) Free Composition, trans. E. Oser, New York: Longman.Schiller, F. von (2003) ‘Kallias or Concerning Beauty: Letters to Gottfried

Keller’, in J. M. Bernstein (ed.), Classic and Romantic German Aesthetics,Cambridge: Cambridge University Press.

Schuller, G. (1968) Early Jazz: Its Roots and Musical Development, NewYork: Oxford University Press.

Schutz, A. (1964) ‘Making Music Together’, in Collected Papers, vol. 2, TheHague: Martinus Nijhoff: 159–78.

Scruton, R. (1997) The Aesthetics of Music, Oxford: Oxford University Press.Scruton, R. (2000) Perictione in Colophon, South Bend Indiana: St Augus-

tine’s Press.Wagner, R. (1887–8) Oper und Drama, dritter Theil: Dichtkunst und Tonkunst

im Drama der Zukunft, in Gesammelte Schriften und Dichtungen, vol. 4, 2ndedn, Leipzig: Fritzsch.

West, M. L. (1992) Ancient Greek Music, Oxford: Clarendon Press.Wollheim, R. (1987) Painting as an Art. Princeton, NJ: Princeton University

Press.


Wright, O. (1980) ‘Arab Music’, Sections 1–4, in Stanley Sadie (ed.), TheNew Grove Dictionary. London: Macmillan.

Further Reading

Bregman, A. S. (1990) Auditory Scene Analysis: The Perceptual Organizationof Sound, Cambridge, Mass.: MIT Press.

Cone, E. T. (1968) Musical Form and Musical Performance, New York:Norton.

Cook, N. (1987) A Guide To Musical Analysis, London: Dent.Deutsch, D. (1982) ‘Grouping Mechanisms in Music’, in D. Deutsch (ed.),

The Psychology of Music, New York: Academic Press.Messiaen, O. (1956) Technique de mon langage musical, Paris: Leduc.Scruton, R. (1997) The Aesthetics of Music, Oxford: Oxford University Press.Scruton, R. (2000) Perictione in Colophon, South Bend, Indiana: St

Augustine’s Press.


Index

Adams, John 241, 243, 252Alpers, Svetlana 153Aristotle 250Aristoxenus of Tarentum 231Armstrong, David 64Armstrong, Louis 236

Babbit, Milton 243, 247Bach, Johann Sebastian 80 n., 83, 87 n.,

142–5, 239Barber, Samuel 142–5Barker, Andrew 231 n.Bartok, Bela 232, 234, 248 n., 252Baxandall, Michael 153Beardsley, Monroe 163Beatles, the 85Beethoven, Ludwig van 88, 90, 172, 179

n., 184–7, 193, 197–9, 201, 211–2,213, 219, 249

Berkeley, Michael 252Black, Max 171Boghossian, Paul 123 n.Bonds, Mark Evan 166, 178 n., 198 n., 199

n.Booth, Wayne 195Borstlap, John 252Boulez, Pierre 232–3, 247, 252Brady, E. and Levinson, J. 178 n.Brahms, Johannes 241Brauner, Charles S. 178 n.Bregman, Albert S. 226Britten, Benjamin 234Bruckner, Anton 81Bucher, K. 234Budd, Malcolm 100, 124, 154 n., 172 n.Burnham, Scott G. 188

complete, indicators that a musical workis 79–80

Cage, John 209Celine, Paul 192Chopin, Frederic 81Clapton, Eric 239Cocker, Joe 85

Cocteau, Jean 194Cook, Nicholas 242 n.Cooke, Deryck 83Cone, Edward T. 162 n., 173–4, 231 n.Currie, Gregory 32 n., 50 n., 88

Dahlhaus, Carl 165Darmstadt orthodoxy, the 252dance 239–42Danto, Arthur 171Davidson, Donald 170, 179Davidson, Audrey Ekdahl 246Davies, Stephen 25 n., 28, 33, 36–8, 50 n.

56, 81, 85 n., 86, 100, 102–4, 107,124, 130, 135–9, 142–4, 154–7, 162n., 169

Daverio, John 179 n.Debussy, Claude 86, 239Delius, Frederick 145 n.De Momigny, J.J. 231 n.Dent, E.J. 188Deutsch, Diana 80, 227–9Diabelli, Anton 197–9Dill, Heinz J. 178 n.Dodd, Julian 56, 63 n., 73Dummett, Michael 59Dvorak, Antonin 234, 236, 238

Edwards, George 201Eiholzer, Hubert 106–7Elgar, Edward 144 n., 211, 234–5Eliot, T.S. 212Ellestrom, Lars 184, 185 n.Elliot, R.K. 110–11electro-sonic art 220–4emotional responses to music 108–9,

112–13, 117–20are evidence of musical meaning 119–20see also musical expression, as arousal

Fernyhough, Brian 252Fox Strangways, A.H. 234, 248Fubini, Enrico 165, 167Furtwangler, Wilhelm 117

258 index

Garland, Joe 85 n.Genesis 238Gershwin, George 85 n.Glucksberg, S. and Keysar, B. 179 n.Gracyk, Theodore 85 n.Grainger, Percy 87 n.Godlovitch, Stan 25 n., 35–6Goethe, Johann W. von 192Goodman, Benny 85 nGoodman, Nelson 53, 55, 133 n., 149, 161,

173–4, 174Guralnick, Peter 239 n.

Handel, George Frideric 239Hanslick, Eduard 158 n., 210, 212, 218Hass, Robert 81 n.Hatten, Robert 193, 199 n., 201Hausman, Carl 153Haydn, Joseph 80 n., 166, 167–70, 172,

183, 199–202Heidegger, Martin 52 n.Heine, Heinrich 178 n.Henderson, David 239Hendrix, Jimi 123, 239Herzog, Patricia 193Hopkins, Robert 100, 101Hume, David 62

Imeson, Sylvia 198 n.instrumentalism 25–6

Janacek, Leos 234, 251Jarry, Alfred 194Johnson, Robert Sherlaw 244 n.Johnson, Mark, see Lakoff, George and

Johnson, Mark

Kandinsky, Wassily 211Kant, Immanuel 61 n.Karl, Gregory 173 n.Katzarova, Raina 248 n.Keysar, B., see Glucksberg, S. and Keysar,

B.Kierkegaard, Søren 192–3Kinderman, William 193, 197, 199 n.Kivy, Peter 34 n., 63 n., 95, 97, 100–1,

118–19, 149, 189, 209–10,216–17

Knussen, Oliver 252Kramer, Jonathan D. 183

La Fontaine, Jean de 157Lakoff, George and Johnson, Mark 164,

179 n.Leger, Fernand 194Lerdahl, Fred 247Levinson, Jerrold 25 n., 26, 29, 30–2,

34–5, 39–49, 50 n., 52 n., 56, 63 n.,64 n., 72 n., 100, 103, 105–7, 110–12,114, 125–8, 130, 136 n., 139–40, 143n., 173, 178 n., 213–14, 217–18, 221

see also Brady, E. and Levinson, J.Liszt, Franz 88, 90Lochhead, Judy 183Longuet-Higgins, Christopher 230Longyear, Rey M. 178 n., 179 n., 185, 186

n. 199 n.Lutoslawski, Witold 220, 223

Madell, Geoffrey 98Magritte, Rene 184–5Mahler, Gustav 79, 83, 164, 174, 193Matthews, David 252Mellers, Wilfred 193, 211–12Messiaen, Olivier 230, 241, 243–8, 250,

252metre, varieties of 229–34Meyer, Leonard B. 212Miller, Glenn 85 n.Miller, Lee 183, 185 n.Monteverdi, Claudio 251Mozart, Wolfgang Amadeus 79–80, 81,

83, 118, 154, 157 169–70, 188–92Muecke, D.C. 185 n., 190–1, 192music

acousmatic nature of 27, 229atonal 218–20, 242–3created or discovered? 216–17electro-acoustic, see electro-sonic artindispensably aural character of 209–17

musical expression 95–146, 120–8and empathy/ sympathy 112–13and normativity 113–14as ‘arousal’ 108–9, 112as imaginative characterisation of

music 100–7, 110–13, 139–45as metaphorical 121–4, 150–74as metaphorical exemplification 150–74cannot be sui generis 96–7comparison with aspect perception 101comparison with pictorial

representation 96, 100

index 259

in terms of a musical persona, see personaaccount of musical expression

is not adequately characterised in termsof causes 97–9, 117–18

Levinson’s dispositional accountof 125–8

resemblance account of 101–5, 124–5,127–8, 135

whether it is a form of meaning 119–24,149–74

musical irony 178–203comparison with verbal irony 179–82dramatic 187–92general 192–5parodic 195–9romantic 199–203situational 182–7

musical meaning 149–206musical ontology 23–92musical performance

its relation to the work 70–2see also performance means essentialism

musical worksare they sui generis? 75–7as norm-types, see type-token theoryin what sense are they ‘repeatable’?

73–5the essential meaningfulness of 56–63the essentially created nature of 64–70

Mussorgsky, Modest 88 n., 211, 212, 234

Neubauer, John 165–6Newcomb, Anthony 159–63, 172–4Nowak, Leopold 81 n.

Oasis 238

Parker, Mara 194 n.Peacocke, Christopher 226performance interpretations 83–6performance means essentialism 39–41persona account of musical

expression 105–7, 111, 130–45,173–4

relevance of individual character ofmusic to 136–45

Picasso, Pablo 192Plato 229Pleasants, Henry 239 n.

Presley, Elvis 238–9Puccini, Giacomo 234

Rameau, Jean-Philippe 165Ratner, Leonard 174 n.Ravel, Maurice 88 n., 163, 239, 241‘redundancy argument’, the 211–16Reich, Steve 215–16, 220, 224Resphighi, Ottorino 239rhythm 226–55

and morality 252–3different from beat 236ostinato 237–9which reflects dance patterns 234–5,

239–42which reflects speech patterns 250–2

Riemann, H. 231 n.Rohrbaugh, Guy 73, 75Rosen, Charles 167, 169Ridley, Aaron 100, 130, 137 n., 141Robinson, Jenefer 136 n., 152 n., 173 n.Rousseau, Jean-Jacques 166Rossini, Gioachino 117

Salzer, Felix 242Sartre, Jean-Paul 192Sawyer, John E. 188Schaeffer, Pierre 229Schelling, Friedrich von 168 n.Schenker, Heinrich 242Schiller, Friedrich von 240Schindler, Anton F. 197Schlegel, Karl W.F. 179 n.Schoenberg, Arnold 195 n., 218–19, 234,

251Schopenhauer, Arthur 168 n.Schubert, Franz 79, 162, 173, 179, 241Schuller, Gunther 236, 238Schumann, Robert 164, 178 n.Schutz, Alfred 229Scruton, Roger 27–8, 96, 114, 121–4,

127, 179, 180 n., 212, 218–19, 229 n.,239, 227

Searle, John 180Shankar, Ravi 248Sharngaveda 230, 245, 248Sheinberg, Esti 178 n., 192Sibelius, Jean 237Sibley, Frank 157 n., 178 n.

260 index

Skryabin, Alexander 219sonicism 23–51

pure 27–31timbral 31–50arguments from scores against

timbral 31–8Stravinsky, Igor 81–2, 86–7, 193–5, 219,

231, 234, 237Sterne, Laurence 198, 199Stockhausen, Karlheinz 243–4, 252Suurpaa, Lauri 178 n.Swift, Jonathan 195

Tchaikovsky, Pyotr Ilyich 87 n., 119, 234Tieck, Ludwig 179 n.Tolstoy, Leo 90Tovey, Donald F. 197transcriptions 86–8translations in literature and poetry 89–92Treitler, Leo 162type-token theory 23–4, 52–6, 63–4, 67,

71–2, 73

universals, musical works as 52–3

Vangelis 224Varese, Edgar 220–1verbal irony, semantic analysis of 180–2Vivaldi, Antonio 157

Wagner, Richard 124, 164, 230–1, 234,235, 248–50, 251

Walton, Kendall 25 n., 100, 109–11,180

Watkins, Glenn 194Webern, Anton 195 n.West, M.L. 234Wittgenstein, Ludwig 62, 101, 193Wollheim, Richard 64 n., 96, 100, 196 n.,

229 n.Wolterstorff, Nicholas 23 n., 24, 55, 63 n.,

73, 74 n.work versions

by others 83characterisation of 82contrasted with performance

interpretations 83–6contrasted with transcriptions 86–8examples of 81–2in art forms other than music 88–9paintings and sculptures as

successive 88–9translations are a form of 90–1

Wright, Crispin 125Wright, Owen 229

Yablo, Stephen 123 n.

Zemach, Eddy 180, 196 n.

Documents

Philosophers on Music