TYPES, CONTINUANTS, AND THE

ONTOLOGY OF MUSIC
Julian Dodd
Are works of music types of performance or are they continuants? Types are
unchanging entities that could not have been otherwise; continuants can undergo
change through time and could have been different. Picking up on this distinction,
Guy Rohrbaugh has recently argued that musical works are continuants rather than
performance-types. This paper replies to his arguments and, in the course of so doing,
elaborates and defends the conception of musical works as types of performance. I
end the article by arguing that the conception of musical works as continuants is
under-motivated and, ultimately, obscure.
I. INTRODUCTION
A WORK of music is a repeatable entity. A performance of Vaughan Williamss
Fantasia on a Theme by Thomas Tallis, for example, is not itself the work; the work
is the thing of which this performance and others are instances. So what is the
ontological nature of the work? How, in other words, is the phenomenon of
repeatability to be explained?
I believe that the traditional and most natural answer to this question is the
correct one. Vaughan Williamss Fantasia is a performance-type: an abstract entity
whose identity is determined by the condition a particular must meet, if it is to
count as one of its tokens.
1
The of relation obtaining between a performance of
the piece and the piece itself is the familiar relation that holds, for instance,
between an inscription of table and the word. However, in a recent article,
2
Guy
Rohrbaugh has argued that such a type/token view of repeatable artworks is un-
sustainable and must be replaced by a conception of these works as continuants.
3
British Society of Aesthetics 2004 342
British Journal of Aesthetics, Vol. 44, No. 4, October 2004, pp. 342360, DOI: 10.1093/aesthj/ayh042
1
For a defence of a version of this view, see my Musical Works as Eternal Types, British Journal of
Aesthetics, vol. 40 (2000), pp. 424460, and Defending Musical Platonism, British Journal of Aesthetics,
vol. 42 (2002), pp. 380402.
2
Guy Rohrbaugh, Artworks as Historical Individuals, European Journal of Philosophy, vol. 11 (2003),
pp. 177205. Hereafter AH.
3
In fact, Rohrbaugh describes his conception as one that treats artworks as historical individuals,
but this is just a terminological difference. I shall concentrate on the case of musical works
throughout but, like Rohrbaugh (AH, p. 177), whose own focus is on photographs, I take my
considerations to apply to repeatable artworks generally.
The purpose of this paper is to demonstrate that Rohrbaughs arguments fail, and
that it is his own counter-proposal that cannot be sustained.
II. TEMPORALITY, MODAL FLEXIBILITY, AND TEMPORAL FLEXIBILITY
Let us start with some definitions. An entity is temporal just in case it either comes
into or goes out of existence (or both).
4
An entity is modally flexible just in case it
is possible for it to differ with respect to its intrinsic properties. An entity is tem-
porally flexible just in case it is capable of change in its intrinsic properties over time
(AH, p. 178). It is Rohrbaughs contention that musical worksunlike typesare
temporal, modally flexible, and temporally flexible, and that their having these
features demonstrates them to be continuants (AH, p. 179). Let us take a look at
his arguments for this view.
I do not wish to dispute that types are atemporal, modally inflexible, and tem-
porally inflexible. But why should we think that works of music are any different?
Rohrbaugh argues swiftly for the thesis that repeatable artworksunlike
typesare temporal existents. Photographs, he claims, come into existence when
they are taken, and are destroyed once their embodimentstheir respective
negatives and printshave been destroyed (AH, pp. 190191). A photograph,
Rohrbaugh says (AH, p. 191), only exists if it possible to see what it is like,
and this requires there to exist either the negative or some other embodiment of
it. It comes into existence at the moment at which the film is exposedwhen its
negative is producedand goes out of existence once this negative and all of its
prints have ceased to be. Clearly, this line of argument transfers directly to the
case of music. Fantasia on a Theme by Thomas Tallis, we are supposed to accept,
came into existence once its score was produced, and will go out of existence
once it is impossible for anyone to know what it sounds like: once there exist no
copies of the score, no recordings, and no memories of it.
When it comes to the claimed flexibility of works of music, Rohrbaugh trusts
that a consideration of examples will demonstrate our intuitions to point in the
same direction as his: namely, towards the thesis that musical works could have
been otherwise and are capable of genuine change. Thus, in defence of the thesis
that musical works are modally flexible, Rohrbaugh claims that Bruckners 9th
Symphony might have been finished had he lived longer, and that we can all
imagine cases in which a piece could have differed slightly, if its composer had
been of a mind to have altered things (AH, p. 182).
Rohrbaugh admits that the case for musical works being temporally flexible is
harder to make. Such works are not physical objects, after all, so it is tempting to
think that their properties are fixed for all time. True enough, it could be argued
JULIAN DODD 343
4
This emends the definition given by Rohrbaugh, who takes a temporal entity to both come into and
go out of existence (AH, p. 178). My account of temporality leaves room for a position discussed in
section III below: the view that musical works are temporal (that is, brought into being by their
composers) but, once created, everlasting.
that if a photographs negative deteriorates, then we are prevented from seeing
how the photographthe abstract objectonce was, which, in turn, might be
taken as evidence that the photograph itself has changed. But the crucial differ-
ence between a photograph and a musical work is that the latter has no physical
object analogous to a photographs negative that may serve as the locus of change
(AH, p. 188). Nonetheless, Rohrbaugh appeals to examples in order to persuade
us that musical works can genuinely change. Composers, he says, may revise their
works after they first compose them (AH, p. 189); and it is also tempting to
describe a folk song as having changed, if it is sung differently as time passes.
How, asks Rohrbaugh, are we to describe the fact that a song or story is sung or
told differently as it is passed down from person to person if not as change in its
structure? (AH, p. 188).
Of course, someone may grant that Rohrbaugh is correct about works from the
folk tradition, and yet nonetheless view scored works as unchanging. Does not a
works score fix its properties for all time? In response to this kind of objection,
Rohrbaugh appeals to theoretical unity. While accepting that there are strong
intuitions that Austens Emma and Beethovens Tempest Sonata are not changing
(AH, p. 188), he claims that a general framework which allows for the possibility
of change in all artworks is the more powerful one (AH, p. 188). Given that
sculptures and paintings may change, and that (arguably) folk songs may change
too, any lingering intuitions that scored musical works lack temporal flexibility
should be regarded as errant.
It is easy to see where these considerations lead. If Vaughan Williamss Fantasia
could have had other intrinsic properties than the ones it actually has, and is
susceptible to change over time, it is not a type, but a continuant. That is to say,
works of music are temporal entities with a
a life story. They are all subject to change over time, and all, had their life stories gone
differently, could have been somewhat different than they in fact are. (AH, p. 199)
On this view, works of music, though not themselves physical entities, have more
in common with particular substances than with types, propositions, and sets.
A question remains, however: what, precisely, does the repeatability of the
Fantasia consist in, if not the fact that it is a type that can be repeatedly tokened?
The answer, for Rohrbaugh, is that the work is a continuant that comes to be
embodied by its performances, which themselves are a subset of the works con-
crete, physical embodiments:
5
the items upon which the work depends for its
existence. As he himself puts it:
the existence of all such items is rooted in the physical world. They ontologically
depend for their existence and qualities on what passes in this, ultimately physical,
world. (AH, pp. 199200)
344 TYPES, CONTINUANTS, AND THE ONTOLOGY OF MUSIC
5
A works embodiments will also include copies of the score and, perhaps, memory-traces of the
work (AH, p. 203, n. 16).
In this way, the repeatability of the Fantasia is coupled with what might seem to
be a more naturalistic construal of musical works than the type/token view offers.
To use Rohrbaughs own parlance (AH, p. 200), works of music, for him, count
as real. Although they are higher-level objects (AH, p. 199), the fact that they
depend for their existence upon items in the physical world ensures that they are
naturalistic entities rather than occupants of some kind of Platonic heaven or
Fregean third realm.
We are now clear, at least, about what Rohrbaugh takes to be the desiderata of an
ontology of music. But, nonetheless, his own positive proposal might still strike
one as alien or, worse, obscure.
6
As he freely admits, he denies that musical works
fit into any antecedently understood ontological category, and takes it that the
ontology of repeatable works of art is fascinating and important precisely because
it turns out to be an occasion on which aesthetics should not be beholden to the
metaphysics on offer, but rather should drive new work in metaphysics (AH,
p. 197). I have two things to say about this. First, we shall, in fact, see in section V
that our suspicion that such new metaphysical work is mystificatory is quite
justified. But before we get on to that, it is also worth keeping in mind that
Rohrbaughs claim that such revisionary metaphysical thinking is needed is only
as strong as his case against the type/token view. And, as I shall demonstrate in
sections III and IV, Rohrbaughs arguments against the type/token view are unsound.
III. ROHRBAUGHS ARGUMENTS: A CRITIQUE
Let us first of all focus on Rohrbaughs arguments for the temporality of repeatable
artworks. At once, two questions need to be distinguished: whether such works
can be said to come into existence, and whether they can be said to be capable of
being destroyed. When it comes to the latter question, Rohrbaugh just asserts that
we agree that such works can go out of existence, baldly stating that [a] photo-
graph is destroyed . . . [when] there are neither prints nor the possibility of future
prints (AH, p. 191). But it is far from clear to me, at least, that our intuitions
point in this direction. To return to the case of works of music, Jerrold Levinson
has admitted that it is hard to deny the residual pull of the idea that works of
music, once created, last forever. Such a work might, he says, just inhabit the
abstract realm of the universe . . . forever. Why should it lapse into non-existence,
just because we do?
7
This last question is a deep one. Given that works of music are not physical
JULIAN DODD 345
6
A feeling that will not be assuaged by the claimed kinship with David Kaplans account of the
metaphysics of words (AH, p. 204, n. 22). The worries we have concerning Rohrbaughs account
of the ontology of repeatable artworks transfer across directly to Kaplans proposal that we view
words as continuants ontologically dependent upon utterances, inscriptions, and thought-episodes
(Kaplan, Words, Proceedings of the Aristotelian Society, suppl. vol. LXIV [1990], p. 98).
7
Jerrold Levinson, What a Musical Work Is, Again, reprinted in his Music Art and Metaphysics
[Ithaca, NY: Cornell U.P., 1990], p. 263.
entities, just what would constitute their destruction? It seems to me to be an
unquestioned anti-realism to suggest, as does Rohrbaugh, that such a repeatable
artwork is dependent for its existence on its being possible for it to be instanced,
or that what he terms our lack of access (AH, p. 191) to a work ensures the
works destruction. All in all, the claim that the existence of a work is sustained by
a historical flow (AH, p. 198) of performances, playings, memories, and the like
is assumed rather than argued.
But what of the claim that works of music are brought into being by their
composers? Notably, while he is undecided about whether such works go out of
existence, Levinson, for one, is convinced that they do not pre-exist their com-
position.
8
Once more, however, Rohrbaughs arguments lack bite. Focusing on
the case of photographs once more, Rohrbaugh claims that a photograph comes
into existence at the moment of the films exposure. Photographs, he states,
come into existence when they are taken. At the moment the button is pressed, the
shutter opens and closes, exposing the film, and we say that we have taken a
photograph. The phrase has what is called success grammar. If I forget to load the
film and blithely snap away at your birthday party, then I should correct my claim to
have taken photographs of it; without exposed negatives, no such photographs exist.
What is so important about the moment at which the film is exposed? This event
determines certain crucial facts about what the photograph is like, in particular,
structural facts. (AH, p. 190)
This reasoning, however, is question-begging. For notice how the linguistic data
are susceptible to a wholly different explanation by a Platonist about photographs:
someone who thinks that photographsqua typesexist atemporally. Admit-
tedly, if the camera contains no film, then no photograph can be taken. But this
does not entail that the photograph itself only comes into existence once the film has
been exposed, only that the taking of a photographa certain eventrequires the
exposure of film and the production of a negative. A Platonist is free to regard
the taking of the photograph as the creation of a negative which counts as an
atemporal types first token.
Along the same lines, a Platonist will not accept that the taking of a photograph
determines certain crucial facts about what the photograph is like (AH, p. 190),
if this is supposed to mean that the event in question brings it about that the
photograph is a certain way. What is thus determined by the photographers point
of view, his equipment, and the process of exposure is the nature of the negative
and, hence, of which atemporal type (that is, photograph) the negative and future
prints will count as tokens. Rohrbaughs talk of the moment of exposure bringing
a photograph into existence can simply be reformulated as a matter of this
moments determining which atemporal photograph is tokened.
346 TYPES, CONTINUANTS, AND THE ONTOLOGY OF MUSIC
8
Levinson, What a Musical Work Is, pp. 6568.
I leave it to the reader to decide upon the merits of such a Platonism about
photographs. But the mere fact that such a position exists in metaphysical space
serves to indicate how Platonism about musicwhich is, perhaps, intrinsically
more plausible than its photographic cousincan be defended against the style of
thinking exhibited by Rohrbaugh. Vaughan Williamss production of its score did
not bring Fantasia on a Theme by Thomas Tallis into existence; rather, this act of
composition was the creation of a recipe for the production of its tokens. The
composers indication of the work did not determine the works nature; it
determined which (atemporal) work he composed (that is, creatively discovered).
Let us now consider Rohrbaughs claim that musical worksunlike typesare
both temporally and modally flexible. My response to this will be to present
Rohrbaugh with the following dilemma. Either cases in which a work appears to
be flexible are really cases in which there is more than one (inflexible) work; or
else, they are cases in which the single inflexible work has a vague structure: a
structure which both allows for the work to be variably scored, and for its occur-
rences to differ structurally. I shall explain what it is for a works structure to be
vague in the next section. In the remainder of the present section I shall argue
that Rohrbaughs examples of putative temporal flexibility see him impaled on
the dilemmas first horn.
To my mind, Rohrbaughs examples fail to convince. It is quite true, for
example, that we commonly describe works as being revised, but such talk is
quite unreflective and, in any case, it is not obvious that we should regard a
revised work as a work that has changed, as opposed to being a work whose
composition was derived from study of the original. Imagine, for example, that in
1920, some ten years after the composition of the Fantasia, Vaughan Williams
returns to the score and, as perhaps he would put it, decides to make some
revisions. To this effect, he produces a new score differing from his 1910 score in
several significant respects. Would we say that the work had been changed by
Vaughan Williamss actions in 1920? I dont think so. What we would say, I
contend, is that Vaughan Williams had composed a new version of the piece. That
is, we would regard the works as distinct although belonging to the same broader
type by virtue of sharing much of the same tonal structure.
9
Rohrbaughs claim that revisions change works is also subject to a worry of an
altogether more theoretical nature. When an object changes, it no longer exists in
its previous state. If a painting, for example, suffers discoloration at t, then it no
longer exists in its pre-discoloured condition at t. But now look again at the
imagined case in which Vaughan Williamss Fantasia is revised in 1920. It would
be a mistake to deny that the piece-as-indicated-by-the-1910-score still existed
after Vaughan Williamss revisions. Indeed, given that the original score would
still be recoverable, the original version could continue to be performed alongside
JULIAN DODD 347
9
Here I agree with Levinson. See his What a Musical Work Is, Again, p. 234.
the 1920 version. This being so, it is ontologically on a par with the piece scored in
1920, rather than being an earlier period in a single pieces life. If the piece had
changed, its pre-1920 state would no longer exist post-1920. But, since it is still
performable from this time, it plainly does exist. Consequently, it is not that
Vaughan Williamss revisions have changed the work; it is that they have resulted
in the composition of a distinct piece based upon the original.
Needless to say, these considerations apply to what is presented by Rohrbaugh
as the case in which a single song is sung with different lyrics through time.
Again, Rohrbaugh has said nothing that prevents us from describing this example
as one in which distinct versionsbased upon the originalare sung down the
years. And the worry about the very idea of a musical work changing recurs. It
makes little sense to say that a song has changed with respect to its lyrics at t, if it
could still be sung with those lyrics after t, as would evidently be possible. In such
a case, the original version unquestionably still exists after t, and so there is no
sense in which it is a mere temporal stage in something that has since changed.
These considerations, it seems to me, are sufficient to undermine the examples
Rohrbaugh takes to demonstrate that musical works are temporally flexible. And
in the wake of such objections, it would be a mistake to think that an appeal to
theoretical unity could salvage matters. In particular, Rohrbaugh is mistaken in
thinking that a general framework which allows for the possibility of change in
all artworks is the more powerful one (AH, p. 188). It is, doubtless, true that
those works of art that are concrete particulars can change: it does not come as
news to be told that a sculpture can become more weather-beaten over time. But
it is misguided to suppose that admitting that sculptures and paintings are con-
tinuants puts pressure on us to regard works of music in the same way. To put it
bluntly, no theoretical unity should be expected here because, while sculptures
and paintings are physical particulars, works of music are not. Given that works
of music fall into a distinct ontological category from sculptures and paintings,
there is no reason to expect that they too should be continuants, especially as our
original concern precisely consisted in wondering what change in an abstract
entity could consist in.
IV. VAGUE TYPES
To claim that a work can change through time thus seems unwarranted, and
perhaps even incoherent. But one might still doubt whether all cases of apparent
temporal flexibility are really cases in which a new, and unchanging, work has
been composed. Imagine that the revisions made to the score of Fantasia on a
Theme by Thomas Tallis in 1920 are minor, footling even. Perhaps just a note is
changed here and there. In a situation such as this our intuitions do not suggest
that the work scored in 1920 is distinct from the work scored in 1910. Likewise,
the examples taken to indicate works modal flexibility resist the kind of
explanation given in the previous section. Couldnt Bruckner have completed his
348 TYPES, CONTINUANTS, AND THE ONTOLOGY OF MUSIC
9th Symphony? Couldnt Beethoven have composed the Tempest Sonatathat
very same workwith just a few different notes (AH, p. 184)? The answer to
these questions is surely yes. The crucial point to grasp, however, is that recog-
nizing such truths does not require us to accept that works of music are either
temporally or modally flexible. For at this point our dilemmas second horn kicks
in. Works of music are vague types, and hence admit of being variably scored,
and, as a result, admit of differently structured tokens: tokens that are sonically
multifarious.
In order to see how this is possible, it will be helpful to examine Rohrbaughs
own conception of the type/token theory, and, in particular, his claim that no
type for which structure is essential could have had differently structured tokens
(AH, p. 184). Rohrbaugh here supposes types to be wholly determinate, so that
each of a types tokens will share their structure with the type itself. On this
conception of a type, it follows that, if two sound events differ in their struc-
tureas would an occurrence of the Fantasia and an occurrence of our imagined
1920 revisionthey cannot be tokens of the same type. Indeed, to return to the
case of putative temporal flexibility once more, it is the thought that types are
wholly determinate that prompts Rohrbaugh to say that what it is for a work to
change its structural features is for later occurrences to differ in their structural
features from earlier occurrences (AH, p. 189; my emphasis). According to
Rohrbaugh, then, if it is possible for performances of a work to differ structurally,
this demonstrates both that the work is not a type, and that the work is modally
flexible. Likewise, if earlier and later performances of a piece differ structurally,
this just goes to show that the performances in question are not tokens of the
same type, and that the piece itself has changed.
What I want to argue now is that these transitions in thought are invalid. The
key premise in Rohrbaughs thinking iswe have just notedthe following
thesis:
(RA) If sound events e and e differ structurally, they cannot be tokens of the
same sound event-type.
Assuming (RA) to be true, and rightly inferring that if (RA) is true, then the
type/token theory is sunk, Rohrbaugh comes to conclude that the fact that a
musical works actual and possible occurrences may differ structurally can only
be explained by taking such a work to be modally and temporally flexible. But, as
I shall now explain, (RA) is false and, what is more, grasping why this is so
enables us to explain away the appearance of modal and temporal flexibility in
musical works.
The identity of a type, we have agreed, is determined by the condition that
something must meet to count as one of its tokens. But this explanation of the
criteria of individuation for types does not rule out the possibility that such
JULIAN DODD 349
conditions can vary with respect to their permissiveness. To put it another way,
since a type is just a way of binding together tokens according to whether they
meet a certain condition, it is possible for this condition to admit of a certain
fuzziness. In other words, it is possible that the principle by which such tokens
are bound can allow for such tokens to vary in certain respects. Let us call such
types vague. Not all types are vague, of course. The sign a, for example, is deter-
minate: although tokens may differ in size, and may be written in different fonts,
they must all have the same structure: they must all have the same shape as the
occurrence within quotation marks in the first clause of this sentence. But some
types most definitely are vague, one such example being the type: House. A house
is a type of building, but the condition that something must meet to be a house is
such that its tokens may differ structurally: they can differ with respect to the
number of storeys they have, the number of rooms they have, and so on. The
point here is that the condition that something must meet to be a house does not
determine the number of floors, or the number of rooms, it must have. Given
that this is so, houses canliterallydiffer structurally and yet be tokens of the
same type. The condition that something must meet to be a token of the type
House allowsby its very natureits tokens to be multifariously structured.
Let us now return to the case of musical works. It is a familiar point that the
words traditionally used by composers to set the tempo of works are themselves
vague; and even if a composer specifies a pieces tempo precisely (by saying, for
example, that it should be played at a rate of 85 crotchets per minute), it is accep-
ted that a tempo approximate to this is perfectly acceptable.
10
What this suggests is
that a works tempo is vague: there is no single precise tempo at which a piece
should be performed. Noticing this feature, the type/token theorist will reply to
Rohrbaugh by taking further properties of works to be analogously vague, thus
enabling two properly formed performances to sound quite different: to differ
structurally, in other words. This being so, the type/token theorist now has the
means available to cope with the kinds of examples that Rohrbaugh takes to be
indicative of musical works modal flexibility. Fantasia on a Theme by Thomas Tallis,
it may be argued, is a modally inflexible, yet vague, type of performance. That is
to say, the work is such that its token performances may differ to some degree in
their sonic properties; and, what is more, it is such that it may be indicated by two
different scores, if these scores are nonetheless sufficiently alike. Two scores
which differ merely by a note or two may count as scores of the same work since
it is not written into the works identity that any performance should be struc-
tured exactly like that. Naturally, whether, in the present example, we decide to
count the actual and counterfactual scores as scores of the same work will depend
upon how exactly the particular example is elaborated; but the crucial point is that
350 TYPES, CONTINUANTS, AND THE ONTOLOGY OF MUSIC
10
See Stephen Davies, Musical Works and Performances: A Philosophical Exploration (Oxford: Oxford
U.P., 2001), pp. 5960.
there is nothing in principle that rules out the possibility that differing scores may
indicate the samefuzzy, but unchangingwork of music.
Somewhat surprisingly, though, Rohrbaugh doubts that the phenomenon of
vagueness can be of much help to the type/token theorist. He accepts, for
instance, that two performances that are qualitatively identical, except for a single
A-note, one of which is played at 440 Hz and the other at 441 Hz, may be said to
differ slightly in their structure and yet could still be said to share the same,
albeit vague, structure (AH, p. 202, n. 8). But he then goes on to claim that the
range of this phenomenon is narrower than that of modal flexibility where we are
considering, e.g., the substitution of a completely different note or sequence of
notes (AH, p. 202, n. 8). However, it is impossible to motivate the attempt to
hold the line at this point. To accept that two tokens of a type may differ in their
structureeven slightlyis to deny (RA). And once it is granted that types may
be vague in structure even to the slightest extent, it takes a very small step to reach
the conclusion that this phenomenon could have a wider range.
The point, then, is that once Rohrbaugh accepts the principle of a vague
structure, his case is lost. And indeed, we can appeal to the notion of a vagueyet
inflexibletype to explain the fact that Bruckner might have completed his
(actually unfinished) 9th Symphony. Given that the unfinished final movement
was, in fact, substantially composed,
11
the actual score and imagined completed
score are sufficiently alike for them to count as scores of the one, vague work. A
possible world in which a work is finished is not a world in which the work itself
is any different; it is a world in which the works score has been completed in
such a way that the identity of the (vague) work indicated is preserved. Naturally,
performances using the completed score would sound different to performances
following the actually extant score, but this is just one more case of a types
having a vague structure, and hence binding its tokens in a more permissive
fashion.
Needless to say, we can imagine cases in which the completion of a works
score is more extensive. Perhaps one or more movements were never started, and
those begun are comprehensively revised, so that the completed works score
bares very little relation to the score of the unfinished work. But in such cases the
pressure upon us to regard the completed work and the work pre-completion as
identical largely dissipates. The type/token theorist is free to insist, with some
justification, that the two works are distinct, and hence that the case provides no
suggestion that works of music are modally flexible. Consequently, cases in
which we are asked to imagine that an incomplete work has been finished look to
be either explicable by the type/token theory, or else to be cases in which we
JULIAN DODD 351
11
According to Dave Lampson, at the time of Bruckners death, approximately 17 minutes of the
unfinished final movement had been fully scored, and the remainder had been sketched out. See
his Anton Bruckner: The Completion of Symphony #9, http://www.classical.net/music/comp.lst/
articles/bruckner/symphony9.html
should deny that an existing work and the imagined one are one and the same.
Either way, the type/token theory survives.
V. AGAINST THE CONTINUANTS VIEW
Having replied to Rohrbaughs arguments against the type/token theory, I would
now like to turn the spotlight upon his own positive proposal. This proposal, let
us remind ourselves, is that works of music are continuants dependent for their
existence upon their embodiments (AH, pp. 199200). In this section we shall see
that any attempt to elaborate this idea ends in obscurity.
The conception of musical works as continuants may take one of two forms,
depending on how the ontological dependence of the supposed continuant on its
embodiments is understood. The first version views the ontological dependence
of a musical work upon its embodiments to be that of constitution, perhaps drawing
an analogy between works of music and David Kaplans conception of words.
Kaplan has famously suggested that utterances and inscriptions are stages of
words, which are the continuants made up of these interpersonal stages;
12
and one
might think of musical works similarly: namely, as historical individuals whose
performances, playings, and other embodiments are their temporal parts. On
such a view, Vaughan Williamss Fantasia will turn out to be a perduring entity: an
entity whose persistence through time is a matter of the succession of its temporal
parts, and for which change consists in the difference in those successive
temporal parts.
13
Equally clearly, such a conception will have to treat musical
works as concrete (that is, spatially located) entities, because a perduring entity
cannot have concreta as its temporal parts without itself being concrete. As David
Lewis has explained, a temporal stage of an object must be of the same ontological
kind as the object itself.
14
It is at this point that the perdurantists intended analogy
between persistence through time and extension through space has bite.
Time-slicing an object is like space-slicing a piece of salami: one slice is pretty
much like any other.
Although Rohrbaugh himself appeals to Kaplans conception of words in his
exposition (AH, p. 204, n. 22), his own version of the view of musical works as
continuants is, in fact, crucially different. For, according to Rohrbaugh, such
352 TYPES, CONTINUANTS, AND THE ONTOLOGY OF MUSIC
12
Kaplan, Words, p. 98.
13
See Theodore Siders Four-Dimensionalism (Oxford: Oxford U.P., 2001) for an excellent discussion
of perdurantism and endurantism.
14
Challenged to explain what a person-stage is, Lewis replies by pointing out that nothing could be
simpler to do: A person-stage is a physical object, just as a person is. (If persons had a ghostly part
as well, so would person-stages.) It does many of the same things that a person does: it talks and
walks and thinks, it has beliefs and desires, it has a size and shape and location (Survival and
Identity, Appendix B, in his Philosophical Papers, vol. 1 [Oxford: Oxford U.P., 1983], p. 76). As for
people, so for works of music: an entitys temporal parts cannot be concrete without the entity
itself being so.
works are higher level objects, dependent on but not constituted by physical or
spatial things (AH, pp. 198199; my emphasis). Since, on this alternative
elaboration, musical works are not taken to have their embodiments as temporal
parts, Rohrbaugh is free to regard such works as enduring (rather than perduring)
entities: items wholly present at all times at which they exist, and which change
by means of gaining and losing properties through time.
15
Equally, he is free to
acknowledge the intuitive pull of the thesis that works of music are abstract. As
he himself puts it, if works of art are in time but not in space, then they are at
least in good company (AH, p. 200).
Ultimately, though, such twists and turns matter little, for neither version of
the view of musical works as continuants is sustainable. To begin with, the idea
that the Fantasia is a concrete, perduring entity of which its performances are
temporal parts has a couple of alarmingly counter-intuitive consequences. First of
all, if a performance is a temporal part of a work, it follows that it is impossible for
an audience at such a performance to hear the work in its entirety. Any given
performance of the Fantasia is but a temporal part of the work, and to have heard
the whole piece would have required an audience member to have audited all of
its constituent temporal parts.
16
What could a defender of the perdurantist version of the continuant view say
by means of reply? Perhaps tu quoque. He might allege that, if anything, the type/
token theorist is in a worse position. Abstracta, he could claim, are imperceptible,
so is it not the case that the Fantasiaif an abstract typecannot be heard in
performance at all? Actually, no. Before we get too carried away by the thought
that we cannot be said to perceive types, we need to remember that there is a
wholly unremarkable sense in which one may demonstrate an abstract type which
stands behind a token that is present before one in space. In such a case of what
Quine has called deferred ostension,
17
one demonstrates the abstract type by
virtue of pointing at one of its concrete tokens. So, for example, when teaching
my daughter to click her fingers I might click my own fingers and then say Do
this. The demonstrative this plainly does not refer to the action I have just
performed (an unrepeatable particular): this has gone forever. Rather, what I
demonstrate is an action-type: the type of which my own action was a token. Now,
JULIAN DODD 353
15
In correspondence Rohrbaugh has pointed out that he favours a conception of musical works as
endurants.
16
This point has been well made by Eddy Zemach: If Jones tells me that he heard Beethovens Missa
Solemnis last night, he would probably be very insulted if I responded, You mean, of course, that
you heard a part of the Missayou could not have heard it all! He would rightly protest that he did
hear the whole Missa indeed (i.e. he did not leave in the middle). If I insisted that in order to hear
the whole Missa one has to hear all its occurrences, including its past and future ones, he would
probably believe that I had gone completely out of my mind (Four Ontologies, Journal of
Philosophy, vol. LXVII [1970], p. 243).
17
W. O. Quine, Ontological Relativity, in his Ontological Relativity & Other Essays (New York:
Columbia U.P., 1969), pp. 3941.)
the fact that this type is abstract does not entail that it cannot be demonstrated,
just so long as such demonstrative reference is ensured by the presence of one of
its tokens. So why should not an analogous move be made when it comes to
the question of whether a work of musican abstract typecan be listened to?
Listening to a piece of music, one could say, is a kind of indirect listening.
18
One
listens to a performance or playinga particular sequence of concrete sounds
and, in so doing, thereby listens to the type of which this concrete sound-
sequence-occurrence is a token. This is the sense in which one can listen to (the
whole of) the Fantasia by listening to a performance or playing of it. And, if this is
right, the type/token theorist is substantially better off when it comes to satisfying
our intuitions concerning the perceptibility of musical works.
The second counter-intuitive feature of the perdurantist incarnation of the
continuant view is its eventual commitment to the thesis that musical works are
scattered objects. Let us suppose that there occur two performances of Fantasia on
a Theme by Thomas Tallis, e and e, occurring in different places, but starting and
ending at the same time. The problem for the perdurantist is this: his doctrine
demands that both e and e be counted as temporal parts of the work, but common
sense says that they cannot be. Fairly obviously, an entity cannot have more than
one temporal part beginning and ending at the same time. (There was, for
example, only one fifth minute of the 1975 FA Cup Final, a part located at
Wembley Stadium; it makes no sense to say that there was an additional fifth
minute of the match taking place somewhere else.) So how could the apparent
commitment to such an absurd consequence be explained away? Only, I suggest,
by treating e and e as two spatial parts of one and the same temporal part. That is,
the obvious move to make is to claim that some of the Fantasias temporal parts
areor, at least, could bescattered objects. Ultimately, however, it is not clear
how such a move could help matters. Forby the previously introduced
principle that an entity and its temporal parts cannot differ in their ontological
natureit would thereby follow that the Fantasia itself is a scattered object, and,
once more, this has clearly counter-intuitive consequences. It is no less weird to
suppose that an audience listens only to a spatial part of a work than it is to
suppose that they can only listen to one of its temporal parts. Furthermore, we
have been given no explanation of the nature of the temporal part that is
supposedly composed of, say, a performance in London and another in Sydney.
To be sure, a jacket located in London and a pair of trousers found in Sydney may
be parts of a scattered objectnamely my suitbut it is unclear why (leaving
aside purely pragmatic motives) we should count two spatially discontinuous per-
formances as parts of the same thing in analogous way. In the end, this response
to our second objection looks like the kind of move made by a philosopher in
retreat.
354 TYPES, CONTINUANTS, AND THE ONTOLOGY OF MUSIC
18
This way of putting it is due to Jerrold Levinson.
Rohrbaugh, of course, may grant all this: after all, he is no perdurantist. But, as
it happens, he is no better off, since, as we shall now see, he side-steps the above
objections at the considerable cost of embracing metaphysical obscurantism. For
one thing, although his acceptance that musical works are abstract enables him to
do justice to our intuitions on this question, it simultaneously removes what
seemed to be a major motivation for viewing such works as continuants in the
first place: namely, the desire to treat them naturalistically. Rohrbaughs account,
in its commitment to abstracta, seems to be no more naturalistically respectable
than the type/token theory. In reply, Rohrbaugh could point out, I suppose,
thaton his view, but not according to the type/token theorythe Fantasia is
existentially dependent upon physical objects (AH, p. 199). That is to say, he
could stress that it is distinctive of his view alone that the work exists only if at
least one of its embodiments exists. But it is important to realize that the
type/token theorist can also allow that musical works are, in an important sense,
rooted in their performances. For although the type/token theory holds that
works can exist untokened, it also accepts that they are intrinsically of such
sound-events. They are not ghostly inhabitants of some mythical Platonic realm,
but things that we can only think of as kinds of performance, and can only refer
to by means of referring to such concrete items.
The analogy to draw here is one between musical works and chess moves. A
chess move is, essentially, something done with a piece. Of course, that this is so
does not entail that chess moves depend for their existence upon pieces in an
analogous sense to that in which Rohrbaugh supposes musical works to depend
upon their embodiments. It would be extremely odd to insist that if every chess
set in the world were destroyed, there would exist no more moves in chess. But,
nonetheless, chess moves are conceptually dependent upon pieces: we can only think
of a move as something done by a piece. Michael Dummett, having characterized
chess-moves in this way, describes them as intrinsically of chess pieces; and he
goes on to deny that they are, in his words, self-subsistent objects.
19
This
combination of relationsnamely conceptual dependence upon, and existential
independence ofis precisely what holds between a musical work and perform-
ances thereof. The challenge to Rohrbaugh is to explain why musical works need
be existentially dependent upon performancesas opposed to merely being
conceptually tied to themif a satisfactorily naturalistic approach to them is to be
made good. We may agree that musical works must, indeed, be rooted in the
physical world (AH, p. 199), but why must we reject the analogy with chess
moves and construe this dependence as existential, rather than merely concep-
tual, in nature?
One reason for denying that the existence of a work of music depends upon
JULIAN DODD 355
19
See Michael Dummett, Freges Myth of the Third Realm, in his Frege and Other Philosophers
(Oxford: Oxford U.P., 1986), pp. 261262.
the historical flow of its embodiments has been mentioned already in section III.
It is contrary to our intuitions to suppose that a work of musicthat is, an
abstract objectgoes out of existence once it has no physical embodiments. A
natural thought has it that an entity without a spatial location precisely does not
require anchorage in the spatial world in order to exist. True enough, the
dependence of a set upon its members is an exception to this rule, but it is an
exception for a specific reason: namely, that a sets members are essential constituents
of it. Since sets are constructions out of their constituents, they have their
actual members essentially.
20
However, such an explanation of the existential
dependence of a musical work upon its embodiments is plainly unavailable to
Rohrbaugh. Rohrbaugh, remember, specifically denies that a works embodiments
are its constituents. Consequently, the idea that an abstract object can depend for
its existence upon physical entities, and yet these physical entities not be its con-
stituents, remains a puzzling one. Given that it is hard to come by other instances
of such a relation of non-constitutive existential dependence between an abstract
entity and concreta, we are entitled to suspect that Rohrbaughs reasoning here is
ad hoc.
But let us put our initial reservations to one side, and see how the proposed
relation of existential dependence could be characterized. Rohrbaugh himself
says little on this matter, but we can, at least, try to reconstruct his thinking. With
this aim in mind, the first thing we should note is that Rohrbaugh, in supposing
the sense in which a musical work depends upon its embodiments to be existential,
is taking sides on the question of the nature of what we might call, neutrally,
ontological dependence. (Other examples of ontological dependence that stand in
need of explication would be the claimed dependence of a trope upon a sub-
stance, and the dependence of a set upon its members.) In speaking of a work of
musics embodiments as those things on which it ontologically depends for its
continued existence (AH, p. 198; my italics), Rohrbaugh is supposing that this
notion of ontological dependence should be explained existentially. I shall return
to this in a moment because, for now, another feature of the supposed ontological
dependence of a musical work upon its embodiments is worth highlighting.
In short, it is that the claimed dependence can only be generic in nature.
21
Rohrbaughs claimif it is to be remotely plausiblemust be that the work
ontologically (that is, existentially) depends upon there being an embodiment,
but not upon there being a particular embodiment. There is no embodiment such
356 TYPES, CONTINUANTS, AND THE ONTOLOGY OF MUSIC
20
This point about the identity conditions of sets is well made by, among others: Nicholas
Wolterstorff, On Universals (Chicago, IL: Chicago U.P., 1970), pp. 178180; Peter Simons, Token
Resistance, Analysis, vol. 42 (1982), p. 198; and David Wiggins, Sameness and Substance (Oxford:
Blackwell, 1980), p. 113.
21
For this notion of generic ontological dependence, see, for example, Kit Fine, Ontological
Dependence, Proceedings of the Aristotelian Society, vol. 95 (1995), p. 287, and Jonathan Lowe,
Ontological Dependency, Philosophical Papers, vol. 23 (1994), p. 34.
that the work would go out of existence, were this particular embodiment not to
exist. The existential dependence in question is not, in this sense, de re.
Putting together these two featuresthe generic nature of the claimed
dependence of a work upon its embodiments, and Rohrbaughs assumption that
ontological dependence is existentiala first attempt at capturing what
Rohrbaugh is looking for would seem to be the following:
(DG) x ontologically depends upon objects of kind K =
df
(x exists
something y exists such that y is of K).
22
However, there are two reasons why (DG) cannot provide Rohrbaugh with the
notion he requires. First, if the definienss consequent expresses a necessary truth
(as it would, if the objects of kind K were the real numbers), then the definiens
comes out as true whatever entity x is taken be. Any entity at all will turn out to
be ontologically dependent upon the real numbers, which is absurd. Second, the
existential construal of ontological dependence allows for such dependence to be
symmetrical. This is especially clear if we consider non-generic dependence,
which, on the existential account, would amount to
(DN) x ontologically depends upon y =
df
(x exists y exists).
23
Given (DN), {The Eiffel Tower} ontologically depends upon The Eiffel
Tower, which is the result we expect; but it is also the case that The Eiffel Tower
ontologically depends upon its singleton. This, needless to say, offends against
the intuition that ontological dependence is asymmetrical: that, in this case, sets
depend ontologically upon their members, but that the converse does not hold.
Consequently, the existential construal of ontological dependence fails to capture
the intuition that motivates the search for an account of ontological dependence
in the first place.
Now let us return to (DG). Given that (DN) allows for symmetrical
dependence, there is no reason to suppose that (DG) should not. And, indeed,
the case of works of music looks to be a case in point. Let us grant, for the sake of
argument, that the Fantasia is, in the non-generic sense explicated by (DG),
existentially dependent upon its embodiments. Is it not equally true that the
existence of any embodiment is dependent for its existence upon the work? Any
performance of the work is a performance of that work, and, as such, could not
exist unless the work did. So the dependence would seem to go both ways and,
JULIAN DODD 357
22
Lowe, Ontological Dependency, p. 35.
23
Amie Thomassons notion of constant ontological dependence would seem to involve a commit-
ment to (DN), as well as to something akin to (GN). See her Fiction and Metaphysics (Cambridge:
Cambridge U.P., 1999). Inasmuch as she construes ontological dependence existentially, she
would seem to fall foul of both objections to such a construal.
once more, the moral can only be that the definiens fails to catch hold of the
fugitive notion.
In the light of these concerns, it might seem tempting to follow Jonathan Lowe
in reconstruing ontological dependence, not as existential dependence, but as
identity-dependence. According to Lowes alternative account of the non-generic
notion,
24
we should replace (DN) with
(DN*) x ontologically depends upon y =
df
(the identity of x depends upon
the identity of y),
where, for the identity of x to depend upon the identity of y is for which thing of
its kind x is to be (at least, partially) determined by which thing of its kind y is.
25
Clearly, this yields the right answers when it comes to the example of The Eiffel
Tower and its singleton. Is {The Eiffel Tower} ontologically dependent upon
The Eiffel Tower? Yes, since what makes {The Eiffel Tower} that set is that it has
The Eiffel Tower as its only member: the axiom of extensionality is a criterion of
identity for sets. Is The Eiffel Tower ontologically dependent upon {The Eiffel
Tower}? No, because the identity of The Eiffel Tower is not to any degree fixed
by the identity of {The Eiffel Tower}. What makes The Eiffel Tower that object
has nothing to do with the identity of any set.
Presumably, the application of this idea to the notion of generic ontological
dependence that Rohrbaugh needs would see us replace (DG) with something
like
(DG*) x ontologically depends upon objects of kind K =
df
(the identity of x
depends upon the identity of K).
Unfortunately, though, there is a clear reason why this move will not help
Rohrbaugh. Put bluntly, it is that
(1) (the identity of x depends upon the identity of K)
does not entail
(2) (x exists something y exists such that y is of K).
(2), remember, is what Rohrbaugh wishes to defend, if x is a work of music and
K is taken to be the type whose tokens are xs embodiments. However, secondary
qualities, if construed as response-dependent properties of objects, turn out to
be a counter-example to the thesis that (1) entails (2). Let us suppose that
358 TYPES, CONTINUANTS, AND THE ONTOLOGY OF MUSIC
24
Lowe, Ontological Dependency, p. 41.
25
Lowe, Ontological Dependency, p. 41.
experience enters into the analysis
26
of redness in the following sense: an objects
being red consists in its being such as to look red (to normal observes in standard
circumstances). If this is the case, it follows that it is necessary that the identity of
the property of redness depends upon the identity of a certain kind of response:
namely, the kind whose tokens are experiences as of redness. But it does not
follow from this that the property of redness exists only if at least one such experience
as of redness exists. For it is quite compatible with the response-dependent view
of colour that things keep their colour in the dark and, indeed, that things were
coloured before there was any sentient life. Somethings being red consists in its
possessing a disposition to look red to normal observers in normal circumstances,
and it still possesses that dispositional property in the dark, and it had that
disposition even before anyone was around to look at it. If a normal observer had
looked at the object in normal conditions, it would have looked red to her. In
other words, if x in (1) is a colour-property, and K is the kind of perceptual
experience that fixes the colour-propertys identity, (2) does not follow from (1).
Given that secondary qualities provide a counter-example to the thesis that (1)
entails (2), there is no reason to suppose that the supposed identity-dependence
of a musical work upon its embodiments will entail the thesis that Rohrbaugh
wishes to defend: namely, that such works depend for their existence (AH,
p. 199) upon their embodiments. Earlier, we doubted whether this thesis
accorded with our intuitions; it is now apparent thateven according to the most
plausible account of ontological dependence available to Rohrbaughthere is no
good reason to believe it. For, on this account, such dependence does not turn
out to be existential at all.
This, it seems to me, presents Rohrbaugh with a dilemma. Either the sense in
which a work of music is supposed to depend ontologically upon its embodi-
ments is obscure; or else, if we accept Lowes account of the notion, works do not
depend for their existence upon their embodiments. The choice for Rohrbaugh is
invidious. If he is to provide a satisfying defence of his construal of musical works
as continuantsan account which must include an explanation of the precise way
in which such works depend ontologically upon their embodimentsthen he
cannot leave this notion of dependence unexplained. But, equally, if he accepts
that such dependence is not existentialthat it amounts to no more than identity-
dependence in Lowes sensethen the motivation for taking his view of musical
works starts to crumble. For, as we noted earlier, the type/token theorist can
acknowledge that musical worksqua typesare intrinsically of their perform-
ances in a strikingly similar way. A type is individuated by the condition
something would have to meet be an instance of that type. So, applying this fact
about the individuation of types to musical works, it follows that a work is
individuated by the condition a performance would have to meet to be a
JULIAN DODD 359
26
Colin McGinn, The Subjective View (Oxford: Clarendon Press, 1983), p. 8.
performance of that work. Works, on this view, are conceptually dependent upon
their performances (although, to reiterate, this is quite compatible with the fact
that works are eternal entities, and hence existed before they were performed, or
even composed).
The end result is that if Rohrbaugh seeks to avoid obscurity by retreating to
Lowes construal of ontological dependence, he nonetheless faces the following
problem. Given that the type/token theory can deliver the same kind of (non-
existential) dependence of a work upon its performances as can Rohrbaugh, what
could possibly motivate rejecting the type/token theory in favour of the con-
ception of musical works as continuants? On the one hand, we have an account
of musical works that places them in a familiar ontological category; on the other,
we have a theory that requires us to introduce new kinds of objects into our
ontology (AH, p. 197). Rohrbaugh himself admits that we should only introduce
such new objects if they can be understood to serve some widespread systematic
and philosophical need (AH, p. 197). No such need exists.
27
Julian Dodd, Centre for Philosophy, School of Social Sciences, University of Manchester,
Manchester M13 9PL. Email: Julian.Dodd@man.ac.uk
360 TYPES, CONTINUANTS, AND THE ONTOLOGY OF MUSIC
27
Many thanks to Peter Lamarque and Jerrold Levinson, who both offered invaluable comments on
an earlier draft of this paper.