Every Performance Is a Stage: Musical Stage Theory as a Novel Account for the Ontology of Musical Works

This article defends Musical Stage Theory as a novel account of the ontology of musical works. Its main claim is that a musical work is a performance. The significance of this argument is twofold. First, it demonstrates the availability of an alternative, and ontologically tenable, view to well-established positions in the current debate on musical metaphysics. Second, it shows how the revisionary approach of Musical Stage Theory actually provides a better account of the ontological status of musical works.


i. introduction
In our common linguistic practices, we seem to bear conflicting attitudes toward musical works. For example, when someone says "Scarlatti's Sonata in D minor came after Chopin's Sonata in C minor," she is uttering an acceptable and possibly truthful sentence if she is talking about the program of yesterday's concert but not if she is referring to the years in which these pieces were composed. In this and similarly ambiguous cases, a term apparently devoted to designating a work of music seems to pick out a sound event (a performance) in one case, but something else (traditionally, a repeatable entity) in another.
Type-token theorists take the latter as straightforward: 'Chopin's Sonata in C minor' refers to a repeatable entity, that is, the musical work, and the apparent reference to a performance, that is, one of the instances of the work, is explained derivatively, as in "[The performance of] Scarlatti's Sonata in D minor came after [the performance of] Chopin's Sonata in C minor." At bottom, this attitude is shared by views otherwise opposed to the type-token approach. For instance, musical perdurantists consider the work designated by 'Chopin's Sonata in C minor' as a perduring entity: performances are temporal parts, and their fusion constitutes the work of music. Nominalists, on the other hand, take it to be a class of performances. For both of these positions just as for type-token theories, then, a nonliteral reading of the sentence above is needed, as in the assumption of an implicit "performance of" caveat (see Caplan and Matheson 2006, 59-69;Goodman 1968).
The aim of this article is to challenge the concept of musical work that has been taken as a model by traditional ontological theories and to turn this focus on its head: the work we refer to with 'Scarlatti's Sonata in D minor' is literally the sound event that occurred yesterday after Chopin's Sonata in C minor. It is our apparent talk about alleged repeatables, then, that needs to be analyzed and explained from the viewpoint of the theory that is defended. In so doing, the view that is proposed gives a promising prominence to the sonic/performative dimension, a dimension which, in a sense, has remained as an afterthought in alternative theories. Recognizing the priority of the sonic aspect of music is one of the main motivations for my view. This article focuses on music, but the same approach could straightforwardly be extended also to other performing arts such as dance and theater, since they standardly share the same structure as music: a composer/author writes on behalf of American Society for Aesthetics the instructions that performers should follow on a script/score. 1 For reasons that will soon be apparent, the view that I propose is called "Musical Stage Theory" (henceforth MST): MST: the musical work is a stage/performance connected by a privileged relationship to other stages/performances.
According to MST, then, the work is a performance, an event we have sensory access to. From now on, I will refer to it as a work-as-performance. Every performance is thus a different work, even if, as will be explained, the act of grouping performances together according to a certain relationship also plays a role in our everyday notion of musical works. By identifying works with performances, that is, with sound events uncontroversially accessible by the senses, MST accounts for the commonsensical commitment to the idea that we should be able to have direct epistemological grasp of musical works. As argued in Section V, this is an advantage exclusive to MST. A second advantage is that of providing a systematized way of describing the relationship which stands between performances, allegedly, of the same "work" through what will be called the Repeatability-relation.
The significance of MST is twofold. First, it is intended to show the availability of an ontologically tenable alternative to well-established views in the current debate on musical metaphysics. Second, it also aims to show the independent advantages of the identification of works and performances.
In the central section of this article, MST is developed against the background of an independently motivated move proposed by Theodore Sider (1996Sider ( , 2001 in the ontology of material objects and persistence. In the final section, I more tentatively put forth considerations to show that MST not only is a tenable alternative, but in fact provides a better account of the ontological status of musical works than the alternatives currently on offer.
ii. the legacy of sider's stage view The explanation of how MST overcomes certain apparent difficulties affecting the identification of works with performances takes as its starting point an analogy with certain aspects of Sider's metaphysics. The comparison with Sider's Stage View is particularly promising for MST since it suggests an encouraging strategy for overcoming prima facie shortcomings of MST. This section describes the points of contact and contrast between Sider's view and MST as well as the benefits which result from this theoretical connection.
Sider puts forward his Stage View in dialectic confrontation with other accounts of persistence, namely, endurantism and perdurantism, with the aim of providing a more satisfying solution to the puzzle cases of identity over time (Sider 1996, 433). Endurantists describe objects as enduring or three-dimensional entities which are "'wholly present' at all times at which they exist" (Sider 2001, 3). Enduring objects do not have distinct temporal parts; instead they occupy in their entirety each of the momentary spatiotemporal regions through which they persist. On the other hand, perdurantism adopts a four-dimensional approach and claims that objects persist through time by perduring, that is, by having different temporal parts. Objects are thus described by perdurantists as four-dimensional worms made up of the sum of different spatiotemporal stages.
The Stage View, in contrast, claims that ordinary objects such as people are instantaneous spatiotemporal stages. They are not threedimensional entities as endurantists contend; yet, they are also not four-dimensional continuant objects which perdure through time, by having temporal parts at each of these times as perdurantists claim. I suggest that a move parallel to the one that Sider makes from perdurantism to the Stage View can be made within musical ontology. In doing so, I place myself in direct dialogue with musical perdurantism, which advocates the application of a rival theory of persistence, namely, perdurantism, to the ontology of musical works. MST maintains, in agreement with musical perdurantism, that musical works are concrete entities but holds that they are not, contra the perdurantist, fusions of temporal parts. Rather, MST claims that the work is a single performance/stage connected by a privileged relationship to other appropriate stages/performances. For Sider, the referent of a singular term such as 'Elizabeth II,' namely, Elizabeth, is an instantaneous spatiotemporal stage, which does not extend through time. Rather, it is related by a counterpart relation to certain other past and future stages. These aspects of Sider's view also provide a fruitful framework for the development of my main thesis. In particular, according to MST, (1) musical works are spatiotemporal stages, just as ordinary objects are in Sider's view; (2) stages are connected by a unity relation which is described as a relation between counterparts similar to Sider's counterpart relation (the I-relation); (3) although 'Chopin's Sonata in C minor' refers to a performance-stage, a systematic shift analogous to that invoked by Sider supports certain other readings of sentences involving that term; (4) as in Sider's analysis of reference to people, this shift is contextually modulated, in a sense that will become clearer later on.
One caveat before continuing with the analogy: Sider's stages are instantaneous while performances are temporally extended. This is not, however, a stumbling block for the analogy: as Katherine Hawley points out in her interpretation of the Stage View, a salient temporal interval may well be established, which allows us to set the boundaries of the stages according to our interests (see Hawley 2001, 59ff.). A performance-stage, in the sense relevant for MST, can thus be described as the sum of all the instantaneous stages that constitute a sound event, the salience of which is determined by the complete performance, from the first to the last note prescribed by the composer's instructions. 2 Sider's account reaches a prima facie impasse when dealing with temporal properties. For instance, intuitively, "I was once a baby" seems like a true description of my past; yet, for Sider, this is not a true attribution of a past property to myself, since, strictly speaking, the occurrence of 'I' in that sentence refers only to my present stage. In order to overcome this difficulty, Sider appeals to an idea which, in its general form, will also play a central role in MST: for Sider, the relation between any stages of a single person (understood as personworm, as will soon be clear) is privileged in respect to the relation between stages of distinct persons. For example, "my" baby stage or "my" 20-year-old stage bear some sort of privileged relation to myself (that is, to "my" present stage) than to other objects, such as the number two or my mother's 30-years-old stage. 3 Sider calls this relation the "Irelation": "If we accept the Stage View, we should analyze a tensed claim such as 'Ted was once a boy' as meaning roughly that there is some past person stage, x, such that x is a boy, and x bears the Irelation to Ted" (Sider 1996, 437). The I-relation, thus, provides a way to explain how certain distinct stages are linked together more closely than others.
However, despite taking the basic entities of our ontology to be instantaneous spatiotemporal stages, Sider's Stage View also appeals to aggregates of stages or worms. This is necessary for Sider to explain how, in our talk, we sometimes seem to employ singular terms so as to refer not to instantaneous stages but to certain collections of stages. This happens, for example, when we ask how many people have been sitting at this desk for the last hour. Assuming the Stage View, our reply should be "infinitely many stages." But here Sider proposes a "partial retreat" from the strict claim of the Stage View: "The stage view should be restricted to the claim that typical references to persons are to person stages. But in certain circumstances, such as when we take the timeless perspective, reference is to worms rather than stages" (448). The difference between Sider's view and perdurantism stands in the fact that, despite allowing for the existence of four-dimensional worms, Sider refuses to identify them with persons and objects in general (444).
Interestingly, especially when it comes to some aspects of MST discussed below, this shift is contextual and interest governed. In addressing the issue of counting coinciding objects, Sider claims that we can attribute different meanings to the same predicate depending on the speaker's interests. To illustrate this point, Sider partially modifies Lewis's example of coinciding roads. Suppose Jane wants to reach the farm and she asks us how many roads she must cross to get there. Jane and the farm are divided by a winding road and, in order to avoid misunderstandings, our answer should be "three" even if the "three" roads are connected miles away with each other. Instead of referring to the road, in this case we refer to road segments to facilitate our talk (440-441). For Sider: "The predicate 'road' does not always apply to 'continuant' roads-it sometimes applies to road segments" (441). The "harmless indeterminacy" (Sider 2001, 199) which is implied by this relativization involves a contextualist analysis of the truth value and of the target of the speaker's utterance. This is analogous to a contextual attribution of meaning to the predicate 'person': while often it applies to person stages, it sometimes also applies to worms, for example, when we take a timeless perspective (190).
Having summarized these aspects of Sider's ontology, I now proceed to the development of its analogue in musical ontology, MST.
iii. musical stage theory The I-relation between stages allows stage theorists to comply with our ordinary talk about objects: what I strictly speaking am, that is, a present stage, is for Sider interestingly related to "my" previous stages, thereby allowing for everyday attributions as in "I was once a baby." Similarly, MST provides an account of certain features of our talk about works by appealing to a privileged relation among performances (that is, for MST, strictly speaking among different works).
Every work-as-performance is related to its counterparts through an ontologically important relation: the Repeatability-relation (henceforth R-relation), which, as we shall see, reflects the sort of intuitions that type-token theories explain in terms of exemplifiables and their instances. So, although a work (say, Chopin's Sonata in C minor) is a performance-stage, this work is related to other performances-stages by means of a privileged R-relation, parallel to  According to MST, the R-relation as a privileged "horizontal" relation between different entities explains what is commonly understood as repeatability, that is, as a "vertical" relation between a work-type and its exemplars. 5 For this reason, much of what traditional theories hold about the work-performance relation may be adapted to the alternative ontological outlook promoted by MST. I propose the following as components of the R-relationship: 1. a causal relation which links the works-asperformances together and/or which connects the works-as-performances to the relevant act of composition, 2. the performers' intentions to play precisely that performance, and 3. a sufficient degree of similarity between the works-as-performances.
As I explain in more detail below, we appeal to these requirements differently in different contexts when speaking of musical works.
The causal relation mentioned in (1) has two aspects: it is a relation between works-asperformances (for instance, between Azhkenazy's and Lilya Zilberstein's performances "of" Chopin's Sonata in C minor) and also a relation between performances and acts of composition (say, between Zilberstein's performance and Chopin's composition). 6 So, it is by virtue of, say, Zilberstein's acquaintance with Azhkenazy's performance or with Zilberstein's and Azhkenazy's common historical relation with Chopin's act of composition that both events are described as performances "of" Chopin's Sonata in C minor. 7 According to requirement (2), the connection between performances involves the performer's intentions: the performer intends to initiate a sonic event precisely by virtue of the causal connection in (1). So, what is required is not only Zilberstein's acquaintance with Azhkenazy or her historical connection with Chopin, but also her intention that these relations guide her performance. Incidentally, this condition is also essential not only to support the causal relation described above but also to rule out cases of unintentional performances such as the scenario of the wind blowing through the canyon (see Wolterstorff 1980, 74).
Lastly, the similarity mentioned in (3) requires that all suitably R-related performances share at least certain aspects of a sonic (harmonic, rhythmic, and melodic) profile. 8 Here as before, the details remain negotiable, since their specification is not an exclusive burden of MST: for anyone, supposedly, there will be vague limits, at some point or another, where an utterly incompetent execution ceases to be a performance of anything.
It is worthy of note that, as a consequence to the approach presented above, MST's assessment of the R-relation is contextual: a sound event qualifies as (what we would commonly call) a performance of W if and only if it satisfies a contextually salient collection of R-desiderata. The R-relation between performances or stages may intelligibly depend upon contextual factors, such as the interests of the speaker or the musical background that she takes to be appropriate. Thus, a sound event may qualify as an instance of W notwithstanding its peculiar sonic profile in some settings (undemanding student recitals, or even arrangements and adaptations) but not in others.
In the previous section I mentioned how the Irelation allows Sider to account for our references to apparent continuants, such as when someone says "I was once a baby." Thanks to a similar strategy, the R-relation allows MST to explain our uses of terms such as 'Chopin's Sonata in C minor' in ways that do not seem to refer to a single work-asperformance, but rather to a collection of performances. This explanation, parallel to Sider's appeal to worms, highlights a further sense in which the R-relation is contextual.
I already discussed a first sense in which the Rrelation is contextually assessed: the links that we establish between performances depend on the context and on the speakers' interests. In other contexts, furthermore, we might not want to refer to performances at all. For example, when someone says "I enjoyed Chopin's Sonata in C minor yesterday night," she refers to a performancestage, that is, according to MST, to what strictly is the work. Yet, in other cases she adopts a 'timeless perspective' and asserts that "Chopin's Sonata in C minor was performed many times in the late nineteenth century in Paris." In this case, what is at issue is a talk directed toward a collection of R-related stages: suitable stages/performances of it took place in Paris. So, while in the first example the person refers to the work-as-performance, that is, strictly speaking, the work, in the second example she is referring to what one might call the work-as-construct.
The work-as-construct is correlated to the work-as-performance just as worms are correlated to stages in Sider's view. According to MST our terms for works primarily refer to stages, that is, performances. Yet, MST also grants that our linguistic attitudes systematically shift between the level of discourse about works-as-performances to that of discourse ostensibly directed toward the work-as-construct. The work-as-construct is indeed a collection of performances, as well as of information we have in respect to the "work," such as provenential and historical information, information on performing traditions, and so on.
It follows that, on the basis of the ontological theory proposed, sentences about works-asconstructs do not involve reference to any single entity. Thus, we need to analyze the sentence: "Chopin's Sonata in C minor was performed many times" in a different way: this sentence means that Chopin's Sonata in C minor (the relevant stage) has the tensed property "having been performed many times" by virtue of there being many previous stages S n , such that (1) S n have the property "being performed" and (2) S n are R-related to the present stage.
As a result, the shift between work-asperformance and work-as-construct allows MST to paraphrase apparently problematic statements such as: "Beethoven's Fifth Symphony is a great work but you wouldn't know it from last night's performance." The concern with how to paraphrase it may stem from the fact that, apparently, the sentence is showing a mismatch between the evaluation of the "work" and of "last night's performance," thus recognizing them as two separate entities, a distinction which is rejected by MST. MST can accommodate this by resorting to the shift of reference between work-as-performance and work-as-construct. While the first part of the sentence is referring to the value of the work-asconstruct (in relation to Beethoven's corpus, the history of music, and so on), the second part of the sentence is referring instead to the properties of the work-as-performance. The incompetence of some performers does not undermine the worth of the work-as-construct but just that of the work-asperformance.
Thus far I have been talking about the reference for terms for musical works (such as "Scarlatti's Sonata in D minor came after Chopin's Sonata in C minor") and its shifts, but I have done so mostly from a metaphysical viewpoint: what was of interest was the status of their referents, as in my discussion of work-as-performance and work-asconstruct. Still, the idea of shifts of reference may also raise semantic considerations that need to be addressed. Just like in Sider's view, where 'person' sometimes refers to a stage and sometimes to a worm, in MST 'work of music' sometimes refers to a stage, that is, the work-as-performance, and sometimes to the work-as-construct.
Note first that the claim is not one of accidental ambiguity. The relationship between the interpretations of the linguistic uses of the term is clearly systematic. It is so in my account to no lesser extent than in traditional accounts (which, as indicated, need to reanalyze 'Chopin's Sonata in C minor' as 'a performance of Chopin's Sonata in C minor' in some cases). A similar methodology applies to cases of presumed co-reference. Suppose we say: "Yesterday, Beethoven's Fifth, a masterpiece of the Classic period, occurred after the Brandenburg VI." In this sentence, the parenthetical clause identifies Beethoven's Fifth as a masterpiece of the Classical period which, over here, anachronistically always comes before the Brandenburg VI.
The problem is that, if these are cases of coreference, either Beethoven's Fifth is a work dating back to the Classical period or it was composed before the 1700s. But note first that this is as much a problem for alternative ontologies of musical works. It is, furthermore, a problem for pretty much any traditional account of other one-many scenarios: it is relatively uncontroversial that 'novel' has a work reading and a physical object reading, but that we can say, for example, "Austen's Emma, a beloved masterpiece, is on the top shelf." The problem is there, but it is a purely semantic issue. 9

iv. objections to musical stage theory
This section addresses a few lingering objections to MST. The first concern is that, if musical works are performances, then compositions which are never performed do not deserve the status of musical works. This concern is legitimate: unperformed compositions are not musical works for MST. However, this does not preclude ordinary assertions about so-called unperformed compositions as works being at least loosely speaking true.
Think of what we would commonly call an unperformed work-suppose that Bob writes a score named Blue Carpet which never gets performed. Plausibly, Bob would not be willing to follow MST in claiming that, strictly speaking, Blue Carpet is not a work because it still has not been performed. Bob's concern is understandable. It is indeed true that by looking at a score, even people who are only scarcely musically literate are able to get some information about the piece. For instance, they can know whether the piece is short or long, which instruments it calls for, whether the structure is simple or complex, and so on. Yet, Blue Carpet according to MST, is not a work of music. It is, nevertheless, a "would-be work." In particular, many among those with an access to Bob's score would understand the shape "it" (that is, a performance that follows that score) would take. Although not a work, Blue Carpet bears all the characteristic features of something that provides the conditions for the existence of a particular workthat is, for the realization of a performance. 10 Indeed, the sense in which MST deals with cases such as Blue Carpet may once again be further clarified by an analogy with Sider's theory. He addresses a similar problem when he takes into consideration sentences about stages that do not presently exist, for example "Socrates was wise." The sentence 'Socrates was wise' cannot be a de re temporal claim about the present Socrates-stage since there is no such present stage. Nor can we take it as being about one of Socrates's past stages, for lack of a distinguished stage that the sentence concerns. What we must do is interpret the sentence as a de dicto temporal claim. Syntactically, the sentence should be taken as the result of applying a sentential operator 'WAS' to the sentence 'Socrates is wise'; the resulting sentence means that at some point in the past, there is a Socrates-stage that is wise. (Sider 1996, 450) Sider thus interprets these sentences as de dicto claims to which sentential operators are applied. I will call Sider's strategy the "operator-strategy." This move is analogous to the application of the modal operator 'possibly' in modal counterpart theory: "According to counterpart theory, an object, x, has the property possibly being F iff some counterpart of x in some possible world has F. . . . The temporal operator 'was,' and 'will be' are analogous to the modal operator 'possibly'" (Sider 2001, 194).
MST can apply a similar operator-strategy to account for sentences which refer to unperformed works. Instead of using the modal operaftor 'possibly,' I use the operator 'in potentia.' So, were we to say, for instance, "Blue Carpet is an example of choral counterpoint," when Blue Carpet is a composition that has never been performed, our conversational contribution amounts to the claim that "in potentia Blue Carpet is an example of choral counterpoint," that is, Blue Carpet has the potentiality of being an example of choral counterpoint in the future. Even if Blue Carpet cannot at present be considered a musical work by MST, since there is no performance 'of' Blue Carpet available, still its composition creates the conditions for a performance to exist in the future and, thus, for Blue Carpet to be a full-fledged work.
The application of the modal operator 'in potentia' is functional for accounting for unperformed compositions as 'would be' works. In this way, MST manages to preserve the commonsensical attribution of the label 'work' to unperformed compositions and, at the same time, to explain it without detracting from its main claims. In this respect, MST fares better than musical perdurantism, which faces the challenge of explaining how an empty fusion of performances can still be called a "work." The second concern that needs to be addressed is related to the compliance of MST with the commonsensical feature of the creatability of musical works. 11 Indeed, MST's central claim, that the work is the performance, counterintuitively entails that a work is created when its first note is played and it ceases to exist when the last note fades out. Interestingly, the strategy that is repeatedly applied to defend MST against previous objections can once again yield its fruit. For instance, the claim we ordinarily express by saying "Chopin's Sonata in C minor was created in 1828" means that Chopin's Sonata in C minor (the relevant stage) has the tensed property "having been created in 1828" by virtue of a previous stage S, such that (1) S has the property "occurring in 1828," (2) S is R-related to the present stage, and (3) no other R-related stage occurs prior to S.
Admittedly, this account yields the conclusion that Chopin's Sonata in C minor can truly be described as being created at the time of its first performance, that is, of S. This account of creatability could still be considered unsatisfactory if, as common sense perhaps intimates, the creation of the work is identified with its act of composition and not with its first performance. Indeed, according to MST, barring cases in which an act of composition consists in a full-fledged performance, the composition is merely a set of instructions necessary to performers in order to transform it into sounds but which by itself is not sufficient to qualify as a stage/work. 12 Nevertheless, even if it cannot be recognized as a work, the relevance of the act of composition is unquestionable: without it, it would be impossible to originate a sequence of performances, that is, of works. In this sense, the act of composition is part of the work-as-construct, since it is part of the set of provenential information that we have in regards to the "work." Here, as in the case of unperformed compositions, I can apply the modal operator strategy and add the operator 'in potentia' to the sentence at issue. MST can thus grant that when we say "Chopin's Sonata in C minor was created in 1828," meaning that it was composed and not performed in that year, we are applying the operator 'in potentia' to our sentence: "Chopin's Sonata in C minor was in potentia created in 1828." Whatever came to light in 1828 is not a full-fledged performance, but it has the potentiality of becoming a stage as understood in the austere sense.
Indeed, MST fares better than its direct rivals, type-token theory and musical perdurantism, in addressing the creatability of musical works. The desideratum of creatability has proven to be particularly problematic given the type-token account of musical works as eternal abstracta. The type-token theorists typically reply that we do not need to think of musical works as creatable entities but instead as stemming from an act of creativity. However, this claim encounters well-known difficulties. 13 Musical perdurantism does not fare better in this respect: the identification of musical works with fusions has the counterintuitive consequence that musical works come into being when the first performance is played. 14 By resorting to the shift to the work-asconstruct, MST manages to acknowledge the relevance of the act of composition. In addition, the adoption of the modal operator strategy allows MST to explain in a systematized and consistent way the role of the act of composition as creation in potentia and the case of unperformed compositions as would-be works.

v. advantages of musical stage theory
It has already been shown that MST is an ontologically tenable alternative to other theories in musical ontology. In this section, the aim is to provide some preliminary reasons for thinking that MST is not only an alternative but a better alternative. There are two main advantages of MST: (1) it straightforwardly accounts for our epistemological grasp on musical works, and (2) it accounts for the widespread relevance that we give to performers in musical practice and in our everyday interaction with music. This advantage partially derives from the appeal to the R-relation, which provides a systematic strategy for overcoming prima facie shortcomings of MST and also allows for greater flexibility in referring to works and performances.
As mentioned in Section I, one advantage of MST is that it straightforwardly accounts for expressions such as "Yesterday I listened to Chopin's Sonata in C minor" thanks to the identification of musical works with performances. The main ontological views currently on the market at best justify this accessibility by recognizing a sort of indirect epistemic contact with the work and by a heavy-handed rephrasing of common expressions. Type-token theorists strive to account for the grasp that we can have of the work by positing a kind of mediated (and arguably not well defined) relation of hearing a work-type through its tokens (see Wolterstorff 1980, 56ff.;and Dodd 2007). The traditional type-token view claims that listening to a performance and being aware of its properties is sufficient to have a grasp on the essential nature of the work itself. However, the main objection that type-token theorists need to face is that, as abstract entities, works cannot enter causal relations and possess audible properties. 15 At best, they can enter causal relations derivatively, along the lines of Quine's "deferred ostension": the individuation of tokens suffices for the individuation of the related type, so that "it is possible to listen to a work of music by listening to one of its performances" (Dodd 2007, 12). Despite securing a way to handle the type-token relation, though, what the type-token view cannot secure is the direct access to the work which common statements such as "Tonight I listened to Beethoven's Fifth Symphony" seem to imply. 16 Indeed, even materialistically oriented views such as musical perdurantism struggle in this respect. According to musical perdurantism, musical works are fusions of temporal parts where these temporal parts are performances. When listening to a performance, then, we are listening only to a part of the work: it is impossible to hear worksqua-fusions in their entirety; only their temporal parts are accessible (Caplan and Matheson 2006, 61-63). 17 Nominalists and action theorists meet similar issues: neither a class of performances nor the composer's actions can typically be objects of acquaintance on the part of the audience (see Goodman 1968 andD. Davies 2004). Instead, the identification of works and performances posited by MST straightforwardly accounts for our access to the work: we have direct grasp of the work by listening to it.
A second advantage of MST consists in acknowledging the importance of musical practices and of performers for the nature of works. This advantage derives from its appeal to the R-relation. In previous sections, the R-relation is presented as the relationship which holds between different performances and that underwrites claims about those performances being of the same "work." MST thus adopts the same strategy for explaining our shift of reference from the work-asperformance to the other uses of the term 'work' and for grouping performances together.
Analogously to Sider's use of multiple unity relations to explain puzzles of identity over time, the multiple requirements of the R-relation show that there is no single way to classify performances. It depends on the speaker's interests whether we should identify a performance as a performance "of" Chopin's Sonata in C minor on the basis of the intentionality requirement even if it contains mistakes, idiosyncratic embellishments, and other departures from the score. Note that the aforementioned contextuality need not (and is not) a question of "tolerance," as when we adopt a certain lassaiz faire in the exemplification of a type (as, say, in the case of a badly scribbled letter 'A'). It is, rather, part and parcel of MST's account of the very nature of the metaphysics of musical works and of our ensuing ways of talking about music. In this sense, MST may perhaps stand in a better position than traditional typetoken theories: at least as long as a musical work is being identified with a single type (as traditional type-theories allege), any deviation from the constraints of the type may well be tolerable, but are nevertheless a source of defectivity. 18 The contextuality envisaged by MST is not a tolerance of departures from fixed standards, 19 since univocal standards are nowhere to be found. There are no invariantly incorrect performances: each performance should be analyzed following distinct and contextually assessed criteria. In this way MST is nearer to the actual (and admittedly fuzzy) practices of reception: each one of us has a slightly different opinion on how and when a performance is correct, and it is hard to say who is right and who is wrong.
MST gives the concept of musical work a new meaning, better suited to interpreting widespread musical practices which are neglected by other theories. The ordinary concept of work that we refer to when making claims such as "Chopin's works have been performed many times," and which is assumed by type-token theorists as the model of musical work, is arguably a legacy of the late nineteenth-century fashion of isolating what was deemed art from other expressions of human creativity. As a result, when we hear someone talking about a work of Classical music, the idea that pops into our minds is that of an entity that ought to be reproduced by respecting the composer's instructions and the performing tradition associated with it. 20 The lens of the "Classical music" tradition has been the medium through which both musicologists and theorists of the ontology of music interpreted the concept of musical work. But the immutability and the aura of untouchable authority which emanated from the works as commonly intended is distinctive only of a very limited period of time in the Western tradition. Before and after that, the work of music was interpreted more as a flexible entity: from the Early Music practice of thoroughbass and partimenti, through accompanied music in the Early Classical period, up to more modern styles such as jazz, a component of improvisation and unexpectedness has always been at the heart of music. Unsurprisingly, then, traditional theories struggle to account for improvisation in their ontological framework because of its ephemeral nature. MST as an account for the ontology of musical works is instead able to seamlessly include musical improvisations within its framework. 21 Improvisations are not lesser musical works than performances based on a set of instructions. What counts for something to be a work, in fact, is that it is a sonic product, intentionally created and performed, and which is immediately graspable by a potential audience. Indeed, in their immediateness, improvisations are musical works par excellence. The performance is not only a means to an end, where this end is the work as an abstract entity. The performance is the means and the end of music.

vi. conclusion
In this article, MST was proposed as a novel account of the ontology of musical works and as an alternative to other views. The proclaimed aims of MST are to be ontologically defensible and to explain the reasons behind our seemingly inconsistent linguistic attitudes toward musical works. Ontological tenability was reached by appealing to an independently motivated account given by Sider in the ontology of space-time: the Stage View. Section II showed how a parallel with Sider's view is beneficial for MST to explain both the relationship between performances and the shifts of reference occurring in our language. The latter takes place between the talk of the work-asperformance-what for MST is, strictly speaking, the work-and the speech aimed at the work-asconstruct. As explained above, the rationale behind this shift of reference is also central for replying to possible objections against MST.
In addition to being a plausible alternative to other theories in the ontology of musical works, MST also appears to be a better alternative. MST has the advantage of straightforwardly accounting for our epistemological access to musical works. Furthermore, the revision of the concept of musical work that it promotes has the favorable consequence of including different musical practices, and of valuing the active contribution of performers. 22

CATERINA MORUZZI
University of Nottingham Nottingham, United Kingdom internet: caterina.moruzzi@nottingham.ac.uk references 1. The same approach could less straightforwardly be applied to other multiple-instantiation artworks such as poetry, literature, and etching. Here, indeed, we have a material object which instantiates the work, not a performance as in the case of performing arts. Thanks to an anonymous referee for pointing out the necessity of drawing this parallel.
2. I refer here generally to the composer's instructions so as not to exclude cases of compositions which do not result in a fully written-out score.
3. Sider's analysis of tensed claims is analogous to David Lewis's modal counterpart theory. The temporal operators used by Sider to explain how, for example, despite being a stage, I was once a child, are analogous to the modal operator 'possibly' used by Lewis to say that, despite being five feet tall, I could have been six feet tall. (See Sider 2001, 194-195.) I return to the application of modal operators within MST in Section VI.
4. My R-relation should not be confused with Lewis's R-relation. Lewis distinguishes the R-relation from the Irelation in his Survival and Identity (Lewis 1983, 58-60), where the R-relation refers to the mental continuity between person-stages and the I-relation to the relation which holds between the stages of a continuant person. 5. In this respect, then, MST follows Christy Mag Uidhir in his proposal of replacing the standard notion of repeatability with the notion of "relevant similarity" (2013,. 6. What I mean for causal relation between score and performance is that, barring the availability of other sources, the score is necessary for the performance to take place. The relation between score and performance is then a counterfactual relation of the form "If score S were not available, the performance P would not have occurred." Regarding the causal relation between works-as-performances, instead, I refer to particular cases in which a performer is not performing from a score, but she is recreating a performance that she has listened to in the past. Also in this case the relation is of counterfactual dependence: "If the act of listening to P did not occur, the performance P1 would not have occurred." 7. It is necessary here to point out an important distinction between the causal connectedness posited by the R-relation and the one posited by the I-relation. In the case of performances, it is possible for two performances A and B to be independently connected to the act of composition without being causally dependent one upon the other (for example, the performer of A need not be familiar with the performance of B). In the case of the I-relation for continuants, instead, each stage is connected with the previous ones, and stages from different branches are not connected despite having the same source. Thanks to an anonymous referee for pointing out this disanalogy.
18. Admittedly, a contextual assessment of the correctness of performances and of their belonging to the group of "performances of work X" is also allowed by this view. Many type-token theorists, for example, grant that in order for a performance to be an instantiation of a type, it just needs to "fulfill an adequate number" of the criteria of correctness (Kivy 1993, 51). (See also Dodd 2007, 32;Levinson 1980, 26;and Wolterstorff 1980, 76.) Still, the flexibility granted by type-token theories is allowed only within the sphere of what is deemed a "correct" performance. A referee suggested that type-token theorists may adopt an alternative strategy. They could say that we refer to different sorts of types on different occasions: sometimes types are connected by one sort of relation, and sometimes by another. Despite leading to a proliferation of types, this may indeed be an option type-token theorists can take for advocating for contextuality. I thank the referee for the remark.
19. As in Peter Unger's (2002) contextual approach to 'flat': in many everyday scenarios, much counts as flat from the purposes of the conversation. Yet, 'flat' is absolute, in the sense that something is strictly flat iff it is perfectly deprived of bumps.
21. For an overview on the topic see Lewis and Piekut (2016).
22. This work was supported by the Arts and Humanities Research Council, United Kingdom (grant number 1504272). I would like to thank the editors and anonymous referees for helpful comments on earlier drafts of this article.

this is an open access article
This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited.