Language-Created Language-Independent Entities
CUNY Graduate Center
A striking feature of properties is how swiftly and easily we appear to get committed to their existence. They exhibit what I've elsewhere called a something-from-nothing feature1: a trivial transformation takes one from a sentence in which no reference is made to a property to a sentence that evidently contains a singular term whose referent is a property. Thus, from
Fido is a dog,
whose only singular term is 'Fido', we can infer
its pleonastic equivalent
Fido has the property of being a dog
wherein the ostensible singular term 'the property
of being a dog' evidently refers to the property of being a dog. Subject
to a qualification that I'll later discuss, every suitably closed predicate
'F' has its nominalization, 'the property of being F', which is seemingly
guaranteed of reference.2
Other putative entities also exhibit a something-from-nothing feature, and, not surprisingly, propositions are among them.3 Thus, again from
Fido is a dog,
whose only singular term continues to be 'Fido',
we can infer its pleonastic equivalent
That Fido is a dog is true
wherein the ostensible singular term 'that Fido
is a dog' evidently refers to the proposition that Fido is a dog. More generally
(but subject to the same qualification to which I've already alluded), every
suitably closed indicative sentence 'S' has its nominalization, 'that S',
which is seemingly guaranteed of reference.
A small elaboration is prudent before zeroing in on my oxymoronic topic. No one will dispute that 'the property of being a dog' is ostensibly a singular term and that its referent, if it has one, is the property of being a dog. This property, assuming there to be such a thing, is abstract, or immaterial, in that it doesn't occupy space or have any other physical properties. And it's mind and language independent in at least two senses. First, it exists in possible worlds in which there are neither speakers nor thinkers, which is just to say that something might have had, or failed to have, the property of being a dog even if there had been neither thinkers nor speakers. Second, while the property of being a dog may be expressed in any language, it itself belongs to no language; it isn't French, English, Japanese, or anything else.
I use 'proposition' as a term of art for the referents of that-clauses; that is to say, that-clauses are referential singular terms and 'proposition' occurs superfluously in 'the proposition that snow is white'. Let me, then, briefly indicate why I take that-clauses to have referents and what properties I take those referents to have. If there's a good reason for taking that-clauses to be referential singular terms,4 it's because the hypothesis that they are is the best way to account for the evident validity of inferences such as:
Betty believes that snow is white.
That snow is white is true.
Therefore, Betty believes something true (to wit,
that snow is white).
Betty believes that snow is white.
That snow is white is Alfred's theory.
Therefore, Betty believes Alfred's theory.
If 'that snow is white' refers, then it refers to
what Betty believes, and what she believes is that snow is white. But what
manner of thing is this thing, that snow is white, which is the referent
of the that-clause singular term? Well, it's abstract in the way properties
are abstract, and it's also mind- and language-independent in the way properties
are mind- and language-independent.5
In addition, it has a truth condition--that snow is white is true iff snow
is white--and it has its truth condition both essentially and absolutely.
Essentially, in that it's a necessary truth that that snow is white is true
iff snow is white, in contrast to the sequence of marks 'snow is white',
which has its truth condition contingently on our actual linguistic practices.
Absolutely, in that that snow is white has its truth condition without relativization
to anything, in contrast to the sequence of marks 'snow is white', which
has its truth condition only in English or among us. So, whatever else propositions,
the referents of that-clauses, may be, they're at least abstract, mind-
and language-independent entities that have truth conditions, and have their
truth conditions both essentially and absolutely. I return now to my main
Properties and propositions have been on many philosophers' Index for a long time. Their reputation has been tarnished by their being abstract, by their not having good individuation conditions, and by their not being obviously theoretically indispensable. But philosophers have also been made wary of the ontological status of properties and propositions by the something-from-nothing language game which seems to deposit them, with so little effort, into our ontology. What are we to make of the ontological status of properties and propositions in view of the something-from-nothing feature?
One severe position that's been entertained is that there's no reason to posit such things as properties and propositions, for the singular terms that purport to refer to them aren't genuinely referential singular terms. That is, we can account for the truth of the true sentences containing 'the property of being a dog' and 'that snow is white' without supposing those expressions to have denotations. Some proponents of this position have sought support in the fact that both 'Fido has the property of being a dog' and 'That Fido is a dog is true' are pleonastically equivalent to 'Fido is a dog'. Given this, they wonder, why not say that all the ontological commitments are owned by the more parsimonious version, neither of the ostensible singular terms 'the property of being a dog' and 'that Fido is a dog' having a genuinely referential function?
But not much support can really be gotten from these equivalences. There's more than one reason for this, but the most obvious is that the something-from-nothing feature doesn't imply that every sentence containing a singular term of the kind in question can be paraphrased without that singular term, and in fact there's no getting rid of the that-clause in, say, 'Ralph believes that snow is white'. The something-from-nothing feature secures that every closed sentence and predicate has its nonlinguistic shadow--propositions for sentences, properties for predicates--but once those entities are secured we allow ourselves some ways of talking about them which aren't equivalent in meaning to sentences that contain no overt reference to them.
A more serious problem for the not-a-referential-singular-term line derives from the fact that quantifiers can bind variables that occupy the positions of the singular terms in question. From 'Fido has the property of being a dog' we can infer 'There is some property that Fido has' and from 'Ralph believes that Fido is a dog' we can infer 'There is something that Ralph believes'. If the terms on which we're existentially generalizing don't refer, then one must say that the inferred existential generalizations don't involve genuinely objectual quantifications. The trouble with this line--I'm now convinced after trying to live with it6- -is that there are serious philosophical difficulties involved in trying to spell out the kind of nonobjectual quantification involved. For reasons that I can't now go into, straightforward substitutional quantification can't give us what we need, and when one finishes spelling out the sort of quasi-substitutional quantification that is needed, one is left wondering what problem this move was supposed to solve.
Another, but equally severe position, holds that there are no such things as properties and propositions even though the singular terms in question purport to be genuinely referential. For although they purport to refer, they also fail to refer. On this view, 'Fido has the property of being a dog' and 'That Fido is a dog is true' are false or without truth-value, just as 'God is omnipotent' is false or without truth-value if 'God' doesn't refer. One might try to soften the counterintuitiveness of this position by mimicking Hartry Field's line on mathematics: '1 + 1 = 2' isn't true, since the singular terms '1' and '2' don't refer, but it's nevertheless "true in the fiction" of mathematics.7 But in the case of properties and propositions it's far from clear what theory of them will stand to them, as the abiding fiction, in the way that mathematics stands to numbers. However, the most serious problem with this reference-failure line is that it's hard to see what could enable us to override our conviction that both 'Fido has the property of being a dog' and 'That Fido is a dog is true' are pleonastically equivalent to 'Fido is a dog' and to hold, instead, that they don't even have the same truth-value. Some terms have "algorithms for elimination" built into them.8 For example, we learn that there are no witches when we learn that no women have certain causal powers, and we learn that there is no phlogiston when we learn that no substance is given off in burning. But what algorithms for elimination do properties and propositions have? Our linguistic and conceptual practices give bases for asserting sentences that ostensibly entail reference to properties and propositions, but these practices give us nothing like an algorithm for elimination. Some philosophers would deny the existence of properties and propositions because we've no need for these things in any of our serious explanatory theories. But: (i) It's unclear what the force of this would be even if it were true: serious science probably has no use for fictional entities like Sherlock Holmes, but I doubt that we should on this account withhold truth from 'The fictional character Sherlock Holmes wasn't created by Mickey Spillane'. (ii) Propositions, at least, do arguably (and consistently with their something-from-nothing feature) play an indispensable role in our commonsense psychological explanations. (iii) Issues of need are vague along several dimensions anyway. Do we need a bottle opener given that with proper training we can do the job with our teeth? This sort of question is pressing because we can reasonably assume that those of our notions that enjoy well entrenched uses are fulfilling some perceived need. Why should explanatory needs provide the criterion for existence?
It may seem that if we reject the two foregoing positions, then we're liable to be stuck with what might be called heavy-duty Platonism. Propositions and properties actually exist; the singular terms that appear to refer to them really do refer to them. Moreover, these entities are as ontologically and conceptually independent of us as rocks, dinosaurs, trees, electrons, or people. In no sense are properties and propositions products of our linguistic or conceptual practices. Consequently, in no sense can we study and learn the nature of these things simply by observing how we're inclined to talk about them. Consequently, properties and propositions are as much in the world as potential objects of discovery as islands or quarks.
Some philosophers hope for an existence-affirming alternative to heavy-duty Platonism. The idea is to allow that propositions exist but to give a minimalist, or deflationary, account of their ontological status. The deflationary view shares with heavy-duty Platonism the assumption that 'the property of being a dog' is a genuinely referential expression that succeeds in referring to the actually existing property of being a dog, and likewise, mutatis mutandis, for 'that Fido is a dog' and the proposition that Fido is a dog. How, then, might the view hope to be "deflationary"? Well, in juxtaposition to heavy-duty Platonism it would maintain that properties and propositions are not as ontologically and conceptually independent of us as rocks and electrons, that there is a sense in which they're products of our linguistic or conceptual practices, a sense in which properties and propositions are mind- or language-created entities. Consequently, we needn't hold that properties and propositions are potential objects of language-independent discovery in the way that islands and quarks are, and consequently we needn't hold that they, like islands and quarks, have "hidden and substantial nature[s] for a theory to uncover."9 David Armstrong has supplied the guiding metaphor for this deflationary picture: properties are but shadows of predicates, propositions but shadows of sentences.10 To unpack this metaphor, and the sense in which properties and propositions are language- or mind-created, would be to elaborate a sense in which properties and propositions are hypostatizations of our ways of talking about properties and propositions. These things exist all right, and not merely in a manner of speaking, but, at the same time, they exist, somehow, as a result of a manner of speaking: they're somehow products of the pleonastic transformations of the something-from-nothing features that earn them a place in our ontology. In a sense, these entities are creations of our linguistic and conceptual practices--our ways of introducing referential and quantificational talk of these things--and there is nothing more to their natures than is determined by those practices. Now it's precisely here that the perennial problem threatens to dash these vague hopes: How can any form of conceptualism about properties and propositions be true when properties and propositions, by their very nature, exist in all possible worlds, hence exist however we do or don't conceptualize anything? We know in what sense refrigerators and shoes are human artifacts: they wouldn't exist but for the human activities that literally create them. But nothing we or anyone else ever did literally created the property of being a dog or the proposition that no triangle is a circle. In what sense, then, might these things be mind or language created? How can conceptual or linguistic activity create things that exist independently of the activities that create them?
There is, I think, a sense in which linguistic practices can create things that, in another sense, exist independently of those practices, and I'll try now to say what it is. It's possible to say quite directly the sense in which properties and propositions are "language created," but I think that what I have to say about properties and propositions will gain both clarity and force by my first advancing a sense in which fictional entities are "creations" of a certain hypostatizing linguistic practice. In the end, we'll see that the issue of whether there's "really" a sense in which the entities in question are language created is perhaps a boring verbal question. The real interest of what's to be said lies in its literal consequences.
Fictional entities, I submit, are a paradigm of entities that both exist independently of certain linguistic practices and yet are, in a sense, created by those practices. They can serve as a model for the sort of claim I want to make about properties and propositions.
We use fictional names--names like 'Othello', 'James Bond', and 'Sherlock Holmes'--in at least two different ways. That is to say, we have at least two sorts of practices involving fictional names.
The pretending use. John Le Carré's novel The Night Manager begins with the following sentence:
On a snow-swept January evening of 1991, Jonathan
Pine, the English night manager of the Hotel Meister Palace in Zurich,
forsook his office behind the reception desk and, in the grip of feelings
he had not known before, took up his position in the lobby as a prelude
to extending his hotel's welcome to a distinguished late arrival.
This is an example of the pretending, or fictional,
use of language, and, in particular, of the fictional name 'Jonathan Pine'.
What's noteworthy is that the displayed token of 'Jonathan Pine' doesn't
refer to anything, and Le Carré is not trying to use it to refer
to anything. Likewise, in producing the sentence, Le Carré isn't
making any true or false assertion. He's not trying to tell us something
about some night manager named 'Jonathan Pine'. Consequently, the utterance
doesn't constitute a true or false statement, since it's not making, or
even trying to make, any kind of statement at all. Rather, Le Carré
is pretending, or making as if, he's referring to a real man named 'Jonathan
Pine', and he's pretending, or making as if, the rest of the sentence is
telling us something about this man, and hence making a true or false statement
about him. More exactly, Le Carré is counting on it's being mutual
knowledge between him and his readers that he's writing fictionally, where
this involves a deliberate simulated play on the conventional assertion-making
use of indicative sentences. In short, what's characteristic of the pretending,
or fictional, use of language is that when involved in it speakers aren't
making true or false statements, though there's a sense in which they're
pretending to, and when they use fictional names in these sentences they're
not referring to anything, though again there's a sense in which they're
The hypostatizing use. Now consider the following utterances involving the use of fictional names.
Jonathan Pine isn't nearly as famous as James Bond.
Sherlock Holmes was created by Sir Arthur Conan
Chili Palmer, the protagonist of Elmore Leonard's
Get Shorty, wasn't based on any one actual gangster.
As regards this use, it's reasonable to hold two
things. First, a speaker, in producing these sentences, might well be making,
and intend to be making, true assertions. It may well be that she intends
to be, and is, imparting genuine information, that her utterances express
complete propositions that are and purport to be literally true. Second,
a large part of what explains the fact that these utterances are making
true statements is that the fictional names in them refer, and are being
used to refer, to fictional characters, abstract entities of a certain kind.
That is to say, although there's no real person James Bond, there really
does exist the fictional character James Bond, which really existing thing
is an abstract entity of a certain kind, and is the referent of the token
of 'James Bond' in the displayed example. Let me hasten to add that while
I believe both these things I've offered as "reasonable to believe,"
I don't regard them--especially the second--as being self-evident. Still,
the position has been well argued for by Saul Kripke and others,11 and for present purposes
I'm content to take it as a working hypothesis. At the same time, it won't
hurt very briefly to mention three prima facie reasons for taking the occurrences
of the fictional names in the most recently displayed examples to refer
to really existing fictional entities. (i) It may be hard to see how the
displayed utterances can be literally true otherwise, and they do seem to
be literally true. (ii) That these occurrences of the names actually refer
to fictional characters seems supported by the fact that we can rewrite
the examples thus:
The fictional character Jonathan Pine isn't nearly
as famous as the fictional character James Bond.
The fictional character Sherlock Holmes was created
by Sir Arthur Conan Doyle.
The fictional character Chili Palmer wasn't based
on any one actual gangster.
(iii) We accept existential generalization on these
names and over fictional entities. For example:
There are more fictional female detectives now
than ever before.
Common sense--our ordinary, unreflective ways of
speaking--evidently contains a commitment to the existence of fictional
entities. It's philosophers who find this commitment problematic, but perhaps
it will seem less problematic when we see how we incur that ontological
We need next to notice an important connection between the pretending practice and the hypostatizing practice, a connection that gives rise to a something-from-nothing feature as regards the hypostatizing use of fictional names. The connection is that whenever one of us uses a name in the fictional way (in which case one's use refers to nothing), then that automatically enables any of us to use the name in the hypostatizing way, in which case we are referring to an actually existing fictional entity. A corollary of this is that it's a feature of our hypostatizing use of fictional names that the existence of a fictional entity supervenes on someone's using the entity's name in the pretending, or fictional, way. In other words, our hypostatizing use of fictional names has it that in every possible world in which 'n' is used in the pretending way, a corresponding fictional entity n exists in that world. Searle puts the point succinctly (I've altered his text to conform to our running example):
By pretending to refer to (and recount the adventures
of) a person, [Le Carré] creates a fictional character. Notice that
[he] does not really refer to a fictional character because there was no
such antecedently existing character; rather, by pretending to refer to
a person [he] creates a fictional person. Now once the fictional character
has been created, we who are standing outside the fictional story can really
refer to a fictional person.12
Now we can raise the question this section has been
building towards. Given our hypostatizing use of fictional names, fictional
entities are created in a straightforward and unproblematic way by the pretending
use of names: the fictional entity Jonathan Pine was quite literally and
straightforwardly created by John Le Carré's use of 'Jonathan Pine'
to pretend, in the way definitive of fiction, to refer to a real person.
The claim meant to be interesting, however, is that, in a sense, fictional
entities are created by the hypostatizing use of fictional names. The question
at issue, then, is: In what sense (if any) might fictional entities be creations
of that use of fictional names?
Needless to say, there's a clear sense in which fictional entities are not created by our hypostatizing use of fictional names: if in any possible world the name 'n' is used in the fictional way, then the fictional entity n exists in that world, whether or not speakers in that world have the hypostatizing use of fictional names. For that itself is a feature of our hypostatizing use; i.e., it's a consequence of our hypostatizing use that we deem the existence of fictional entities to supervene on the use of their names in the pretending way. How, then, might it be said that fictional entities are created by our hypostatizing use of fictional names? In what way can fictional entities be things that are both created by the practice of using fictional names in the hypostatizing way and things that exist independently of that practice?
It may be useful to begin with an example of things whose existence is wholly independent of linguistic practices. To this end, consider a possible world just like ours, only there are apparently no trees there (or even tree-like things). Consequently, the inhabitants of this possible world have neither the word 'tree' nor the concept of a tree. Nevertheless, we can easily imagine them first discovering trees, and then introducing the word and concept consequent on that discovery. This is precisely what happened as regards the discovery of electrons. In 1897, J.J. Thomson discovered the electron. Prior to this there had been speculation about subatomic particles, but no one really had the concept of an electron. What Thomson discovered was that so-called cathode "rays" were not waves of radiation, as they had been thought to be, but rather minute particles of matter each carrying an electric charge. He also found that they weighed far less than hydrogen atoms. But it took some years for scientists to realize that these particles could be emitted from atoms themselves, or to realize that atoms were not simple but had these new particles as constituents. Here there was the discovery of electrons, and then the gradual introduction of electron talk into the language as physicists discovered more and more about the nature of electrons.
There's a further important feature of the independence of trees (or electrons) from any practices we have of talking about them. Once the denizens of our imagined possible world have discovered trees and acquired the concept of, and a word for, them, they will want to learn about trees. What they won't do in order to learn more is observe their practice of speaking about trees; it's trees directly that they'll study. The nature of trees isn't revealed, or determined, by our ways of talking about trees; to learn about the nature of trees, you've got to study trees.
Now let's return to fictional entities, which, I shall claim, are not wholly independent of our hypostatizing use of fictional names. Letting a stand for the actual world, we shall consider a possible world b just like our world a except the inhabitants of b don't have, and never had, the hypostatizing use of fictional names. They have the pretending use; they lack the hypostatizing use. Now, we are the inhabitants of b , and we're just like we actually are, in a , except that, lacking the hypostatizing use of fictional names, we lack the concept of a fictional entity. In b , Le Carré wrote a novel called The Night Manager in the course of which he pretended, in the way definitive of fiction, to be discoursing about a real man named 'Jonathan Pine'. Given the nature of our hypostatizing practice in a and the facts about b , we can say the following. (i) The fictional character Jonathan Pine exists in b . (ii) Knowing the facts about b , we, in a , know that numerous fictional characters exist in b , including the Le Carré-created fictional character Jonathan Pine. (iii) Qua inhabitants of b , who lack the hypostatizing use of fictional names, and hence the concept of a fictional entity, we are ignorant of the existence of fictional entities. In b , we don't know that the fictional character Jonathan Pine exists, nor do we know that any fictional characters exist, although, of course, we do know that Le Carré used the name 'Jonathan Pine' without referring to anyone but pretending to refer to a real man.
OK, fictional entities exist in b but the inhabitants of b are ignorant of this. What would it take for them to discover--i.e., come to know of- -the existence of these entities? What is the crucial epistemological difference between us in the actual world, a , and us in b such that, as a result of that difference, we actually know about the existence of fictional entities but the inhabitants of b don't? The answer is simple: we have the hypostatizing use of language and they don't. It's necessary and sufficient for them to know of the existence of fictional entities that they adopt a certain linguistic practice, viz. the hypostatizing use of fictional names. Then, and only then, will they know what we know. What's required to bring us in b up to epistemological snuff with us in a is nothing more nor less than adopting a certain way of speaking.
There is another, but closely related, difference. Once the inhabitants of b adopt our hypostatizing use of fictional names, and thereby acquire the knowledge of the existence of fictional entities that we have, they'll also acquire all the same knowledge about fictional entities that we have. This is because there can be nothing more to the nature of fictional entities than is determined by the hypostatizing language game that recognizes them in our ontology. The contrast is with cats, islands, electrons, and whatever else enjoys the highest degree of ontological and conceptual independence from our linguistic or conceptual practices. They, as Mark Johnston would put it,13 have hidden and substantial natures for empirical investigation to discover; but there can be nothing more to the nature of fictional entities than is determined by our hypostatizing use of fictional names. The "science" of them may be done in an armchair by reflective participants in the hypostatizing practice.
So much for my gloss on the sense in which our hypostatizing use of fictional names "creates" fictional entities that exist independently of that use: (1) If, though we had fiction, we didn't have the hypostatizing use of fictional names, then we'd have no knowledge of the myriad fictional entities whose existence supervenes on the pretending use of fictional names; but all it would take to make us cognizant of fictional entities--and to give us all the knowledge about them that we currently enjoy--would be simply to adopt the hypostatizing use of fictional names. (2) Fictional entities can have "no hidden and substantial nature for a theory to uncover. All we know and all we need to know about [them] in general" is determined by our hypostatizing use of fictional names.14
Now, have I really shown that fictional entities are in any sense created by our hypostatizing use of fictional names? Let's hold that question for a little while until we can reapply it to properties and propositions.
What holds for fictional entities holds, mutatis mutandis, for properties and propositions. The analogue of the hypostatizing use of fictional names consists in those practices that license the something-from-nothing transformations already detailed. This is the practice that allows us to transform a predicate 'F' into the singular term 'the property of being F', and to transform a sentence 'S' into the singular term 'that S'. Constitutive of these transformations are those that make
a is F
pleonastically equivalent to
a has the property of
That a is F is true.
The first transformation yields the instantiation
schema for properties:
x has (instantiates) the property of being F iff
x is F,
while the second yields the truth schema for propositions:
(The proposition) that S is true iff S.15
A corollary of these transformations is that properties
and propositions exist regardless of whatever linguistic or conceptual practices
we indulge in. From
Necessarily, something is or is not F
Necessarily, something has or does not have the
property of being F,
which implies the existence of the property of being
F in all possible worlds, hence the existence of that property whatever
linguistic practices do or don't obtain. And from
Necessarily, S or not S
Necessarily, that S is true or that not S is true,16
and, by equivalence,
Necessarily, (the proposition) that S is or isn't
which implies the existence of the proposition that
S in all possible worlds,17
hence the existence of that proposition whatever linguistic practices do
or don't obtain.
At the same time, properties and propositions seem to be creations of our something-from-nothing language games in the same way that fictional entities are creations of our hypostatizing use of fictional names. First is the "discovery point." Imagine a possible world b that's exactly like the actual world, a , except that we in b don't have linguistic practices that license the formation of property or proposition singular terms ('the property of being F', 'that S'); in b , we don't play the something-from-nothing language games, and consequently we in b lack the concepts of properties and propositions and, therefore, are ignorant of the existence of the myriad properties and propositions that in fact exist in b (for
it's a consequence of the property- and proposition-introducing
language games we play in a , the actual world, that all the same properties and propositions exist in b as exist in a ). Consequently, we in b lack all the knowledge that we actually have about the existence of properties and propositions. What would it take to bring us in b up to epistemological snuff with us in a ? It's simple: what we'd need to do, and all that we'd need to do, is adopt the property- and proposition-introducing language games we actually play. We certainly couldn't become aware of the existence of, say, trees in that way! We'd have to first discover trees and then introduce talk about trees. Not so with properties and propositions: there's no way we could discover them and then introduce our talk about them. Knowledge of their existence comes with, and only with, the adoption of a certain manner of speaking. This is part of the basis for saying that properties and propositions, unlike trees, are (in a sense) creations of our linguistic practices of talking about those things.
Then there's the "nature-determination point." To learn about the nature of trees there's no substitute for studying trees. One wouldn't get far by studying our talk about trees, our tree-introducing language games. But there's nothing more to the nature of properties and propositions than is determined by our property- and proposition-hypostatizing linguistic practices. What we can learn about them is what our linguistic practices license us to learn about them. By contrast, what we can learn about the nature of trees--about, say, their origin and constitution and biological processes--is undetermined by any ways we have of talking about trees. Yet this talk of linguistic practices determining natures borders on the shamelessly vague. Let me try to do a little better in the following way.
There is an important difference between, on the one hand, "linguistic posits" like fictional entities, properties and propositions and, on the other hand, those entities that are not linguistic posits, entities enjoying the highest degree of independence from our linguistic and conceptual practices. The difference is that the essences of the latter can be discovered by a posteriori, scientific investigation, whereas those of the latter can't be discovered in any such way. Whatever belongs to their essence can be read off the something-from-nothing linguistic practices that posit them in our ontology.
I don't mean to put too fine an edge on the notion of essence. In the strictest, philosophical sense, a feature belongs to the essence of a thing only if it characterizes that thing in every possible world in which it exists, but weaker notions of essence are available, as de facto crucial properties, that don't strain our modal intuitions so severely. Think of the insights about natures we've absorbed from Kripke and Putnam over the past thirty years. What's essential to being a dog is, roughly speaking, having a certain genotype, but what that is, and even that it is a genotype, is entirely undetermined by any way we have of speaking about dogs. Analogous truths apply to islands, electrons and water. These entities have "hidden and substantial nature[s] for a theory to uncover."18 Not so, however, as regards linguistic posits. Whatever belongs to them essentially--even in the most relaxed sense of essence--is directly or indirectly determined by the something-from-nothing language games by which they're deposited in our ontology. For example, it belongs to the essence of the property of being a dog that a thing has that property just in case it's a dog. We get this from the pleonastic equivalence that takes us back and forth between 'So-and-so is a dog' and 'So-and-so has the property of being a dog'. Likewise, it belongs to the essence of the proposition that Fido is a dog that it's true just in case Fido is a dog, and we get this from the pleonastic equivalence that takes us back and forth between 'Fido is a dog' and 'That Fido is a dog is true' (more colloquially, 'It's true that Fido is a dog').
But might not scientific investigation reveal that the property of being a dog was identical to the property of being of such-and-such genotype? Only if our linguistic practices license that discovery, and in fact it's at best highly indeterminate whether they do. The language game we play with properties allows us19 to nominalize any predicate 'F' as 'the property of being F' and to be assured of its referring to the property of being F. In this way, we know that there's the property of being a dog and the property of being of such-and-such genotype. Then science tells us (we can imagine) that all and only dogs have the property of being of that genotype, and that being of that genotype determines the phenotypical properties by which we identify dogs as dogs. Can we then go on correctly to assert that the property of being a dog just is the property of being of such-and-such genotype? Well, we can certainly assert it, and if enough people join in, a linguistic practice will develop that licenses such identities. But that's what it would take, and our present practices constitutive of property talk leave, at best, the matter indeterminate. An analogy with fictional entities may both clarify and reinforce my point about linguistic determination. The essential properties of fictional entities are directly or indirectly determined by our hypostatizing use of fictional names. These practices determine that the fictional character James Bond is an abstract entity, and that his existence and most of what we can say about him derive in familiar ways from a certain body of fiction. These practices may be indeterminate in what they license, and if something pertaining to the nature of a fictional entity is not entirely fixed by our hypostatizing use of fictional names, then there is nothing that we might discover that would fix it. For consider the case of Superman and Clark Kent. In the Superman fiction, Superman and Clark Kent are one and the same person. But is the fictional character Superman identical to the fictional character Clark Kent? This is not determinately answered by our use of fictional names, and hence not determinately answered.
So much for the sense in which properties and propositions are "linguistic posits," creations of our linguistic (and hence conceptual) practices. Even granting the considerations on which I base this claim, have I really shown that there's a sense in which properties and propositions are creations of linguistic or conceptual practices? I've done my best to make sense of the idea that linguistic practices can both give rise to entities and, at the same time, determine us to deem those entities to exist in counterfactual circumstances in which the creating practices didn't obtain. But I doubt that my claim can have a determinate truth-value. Nor does it matter. First, I might hope to have unearthed the truth in conceptualism about properties and propositions, the view that they're somehow mind- or language-dependent, somehow shadows of predicates and sentences respectively. I've been trying to say what someone tempted by conceptualism should have said, and hoping to say something that theorist would be content to say, something that relieves the itch of conceptualism. Second, the two points that sustain my creation point--the discovery point and the nature-determination point--can hope to be literally true, and they are not without interest under any label. Let me give an example that illustrates an importance I came to give the nature-determination point while working on the problem of vagueness.
Paul Horwich begins the preface to his Truth with an advertisement for the deflationary theory of truth his book will present:
Perhaps the only points about truth on which most
people could agree are, first, that each proposition specifies its own
condition for being true (e.g. the proposition that snow is white is true
if and only if snow is white), and, second, that the underlying nature
of truth is a mystery. The general thrust of this book is to turn one of
these sentiments against the other. I shall show that truth is entirely
captured by the initial triviality, so that in fact nothing could be more
mundane and less puzzling than the concept of truth. (p. xi)
Horwich's point is that there's nothing more to
our concept of truth, as applied to propositions, than is determined by
the infinitely many instances of the schema
That S is true iff S.
I'm sympathetic to Horwich's project about truth,
but I have been emphasizing something different--that the instances of this
schema are indeed "trivialities"--and trying to account for why
that's so. My point has been this. There can be nothing more to our notion
of a proposition than can determined by the ways we have of talking about
propositions, of, if you will, introducing them into our ontology. The first
and primary way they enter in are simply as nominalizations of sentences.
Given a sentence 'S' you can form its nominalization, 'that S', and you
can be assured of a reference for that nominalization by virtue of the fact
that 'S' is pleonastically equivalent to 'That S is true' and by the fact
that each instance of the law of excluded middle of the form
That S is true or not true
presupposes the existence of (the proposition) that S. Now, it's precisely because of our ability to rewrite 'S' as 'That S is true' that we have it that each instance of the displayed truth schema for propositions is a "trivial" conceptual truth. But this tells us as much about the nature of propositions as it does about the nature of truth. It's constitutive of the very nature of a proposition that they be "shadows of sentences" in the sense revealed. (It's because of this that we can enlarge our proposition-using practices to include intentional constructions like 'A believes that S'; but this is a story for another day.) The truth schema for propositions,
(The proposition) that S is true iff S,
is a corollary of the most fundamental way we have
of admitting propositions into our ontology: as the nonlinguistic correlates
of the that-clause nominalizations indicative sentences determine, and find
employment for, in the pleonastic transformation that moves back and forth
between 'S' and '(The proposition) that S is true'.
The upshot is that we can read a conceptual truth about propositions, and perforce about truth--namely, the just displayed truth schema for propositions--right off of the hypostatizing language-game that introduces them into our ontology. And here is one reason that's important:20
given the truth schema for propositions plus the law of excluded middle plus the platitude that a proposition is false just in case its negation is true, we get the principle of bivalence for propositions: every proposition is either true or false. This principle severely constrains solutions to the problem of vagueness. It means that either there are no vague propositions or that every vague proposition is either true or else false. But given the way that propositions enter as the shadows of sentences--give me any (closed, indicative) sentence 'S' and I'll give you, trivially, the proposition, that S, it determines--it's the latter that we must say: vague sentences express vague propositions, and every vague proposition (and a fortiori every vague (closed, indicative) sentence) is either true or else false. This, needless to say, is fraught with significance for the theory of vagueness.
I have said that, subject to a certain qualification, each predicate 'F' determines a property, the property of being F, thanks to determining a nominalization, 'the property of being F', which can't fail of reference. A similar claim was made for propositions, but for simplicity I'll now confine attention to properties. I turn to the qualification.
Consider the true statement that
Doghood is a property that doesn't instantiate
If what I said about the property-hypostatizing
pleonastic transformation were taken without qualification, this would yield
Doghood has the property of being a property that
doesn't instantiate itself,
thus giving rise to the property that doghood is
being said to have. By excluded middle we then get,
The property of being a property that doesn't instantiate
itself either does or doesn't instantiate itself.
But if it does, it doesn't, and if it doesn't, it
does: contradiction! Evidently, no one should want to say that the property
of being a property that doesn't instantiate itself is a counterexample
to the law of excluded middle; evidently, one should want to say that there
can be no such property. A question and a worry arise for the pleonastic
conception of properties (and propositions, though this shall remain implicit)
this paper has been concerned to advance. The question is how to qualify
the "trivial transformation" that allows us pleonastically to
rewrite ' a is F'
as ' a has the property
of being F'. The worry is that the required qualification will undo the
sense in which properties are hypostatizations of the ostensibly pleonastic
transformation that takes predicates into their property nominalizations.
Properties, one might suspect, will turn out to have natures that really
are unrevealed by our casual, commonsense ways of talking about properties.
The question and the worry can be dealt with together in a way that supports the proposal that properties are hypostatizations of our predicate-nominalizing practice.21 Pretend that we didn't have ways of nominalizing predicates to form property singular terms and thus didn't have the concept of a property. Then there was a huge convention attended by all English speakers (soon followed by similar conventions for all, or most, other natural languages) in which it was stipulated that every predicate 'F' was to have its nominalization, 'the property of being F', and this was to be assured of having reference in that the criteria for using these nominalizations would secure that the nominalizations were grammatical argument terms in true predications, one of which was to be the pleonastic transformation that rewrites ' a is F' as ' a has the property of being F'. But then it was discovered that our practice wasn't entirely coherent: these stipulations led in certain cases, such as the one lately considered, to absurdity. Yet no new convention was convened. Since the practices served us well in all other cases we continued to indulge in them, simply discounting the troublesome applications. In effect, this meant that what was acceptable in our practice could, on reflection, be put no better than in the following inelegant way: Every predicate 'F' determines its shadow property, the property of being F, except when that leads to absurdity. Clearly, this little fiction does nothing to undermine the idea that, in the fictional world described, properties are hypostatizations of certain predicate-nominalizing practices.
My point is that the actual world is as if it had the fictional world's history. The reason why the property version of the heterological paradox is a paradox is that our implicitly accepted practice really does license the move from
Doghood is a property that doesn't instantiate
Doghood has the property of being a property that
doesn't instantiate itself.
Naturally, the reflective among us repudiate this
application when we uncover the bomb it conceals. At the same time, we're
party to the practice the rest of the time. Our practices are imperfect,
indeed not fully coherent, but they serve us well most of the time, and
properties and propositions are their offspring--in a sense.
1 "A Paradox of Meaning," Noûs, 28 (September 1994): 279-324. See also my "Meaning and Value," The Journal of Philosophy, Vol. LXXXVII, 11 (November 1990): 602-614.
2 "Suitable closure" pertains to predicates (better, predicate forms) like a token of 'kissed it' whose contained token of 'it' doesn't refer.
3 In "A Paradox of Meaning," pp. 304-8, I briefly discuss something-from-nothing features associated with events and states. Mathematical entities also exhibit the feature, but I'll wait for another occasion to explore the consequences of this.
4 For present purposes, I'm content to construe 'singular term' broadly so that it accommodates readings of that-clauses on an analogy with Russell's treatment of primary occurrences of definite descriptions.
5 To say that the proposition that snow is white exists in possible worlds in which there are neither thinkers nor speakers is merely to say that the proposition would have existed (and had a truth value) even if there had been neither thinkers nor speakers.
6 See my Remnants of Meaning (Cambridge, Mass.: MIT Press, 1987).
7 See, e.g., the Introduction to Field's Realism, Mathematics and Modality (Oxford: Blackwell, 1989).
8 See my "Physicalism," in Philosophical Perspectives, 3 (1990): 153-185.
9 Mark Johnston, "The End of the Theory of Meaning," Mind & Language, 3, 1988: 23-42; p. 38. In this paper, which influenced my own development of what I like to call the pleonastic conception of properties and propositions, Johnston advances what he calls "Minimalism about meaning." I'm not prepared to accept much of what Johnston says about propositions (i.e., sentence meanings), but I count his theory as deflationary in the present sense.
10 Armstrong, in Universals (Boulder: Westview, 1989), was characterizing, but not advocating, "Predicate or Concept Nominalism, where properties, et cetera, are as it were created by the classifying mind: shadows cast on things by our predicates or concepts" (p. 78).
11 Kripke presented his arguments in his John Locke lectures, delivered at Oxford University in 1973. The line is also explicitly represented and defended in John Searle, Expression and Meaning (Cambridge: Cambridge University Press, 1979), ch. 3.
12 Op. cit., pp 71-2.
13 Op. cit.
14 Johnston, op. cit.
15 Cf. Paul Horwich, Truth (Blackwell: Oxford, 1990).
16 This sounds better when put in terms of still another pleonastic equivalent: 'Necessarily, it's true that S or it's true that not S'. See Horwich, op. cit., p. 17, n. 1.
17 This may be a slight exaggeration. While we accept
Necessarily, Harry Truman is or isn't dead.
we seem not to have any determinate way of understanding
what this comes to in possible worlds--states of affairs that might have
obtained--in which Truman didn't exist.
18 Johnston, op. cit.
19 Subject to a forthcoming and already alluded to qualification.
20 A reason not lost on Horwich: see Truth, pp. 80-87.
21 What follows has been influenced by Charles Chihara's "The Semantic Paradoxes: A Diagnostic Investigation," The Philosophical Review, 88 (1979): 590-618.