On Not Denoting
On Not Denoting
The literature on descriptions tends to operate with a limited class of examples, mainly descriptions of people and places (“the queen of England”, “the capital of France”). This can bias us in favor of certain theories of their semantics. We can think of descriptions as referring to people and places (Frege and Strawson), or we can think of descriptions as quantifying over people and places (Russell). But there are many other types of descriptions that lend themselves far less readily to such theories: “the end of the line”, “the impossible dream”, “the shortest way home”, “the day after tomorrow”, “the big home sale”, “the height of absurdity”, “the reason why”, “the truth about Mary”, “the logic of existence”, “the problem of consciousness”, “the Greek gods”, “the mystery of life”, etc. etc. Here we are not dealing with concrete particulars unabashed reference to which can be assumed unproblematic; some ontological hesitancy might be presumed. Are we really to believe that the use of such descriptions necessarily commits the speaker to a robust ontology of corresponding entities? The speaker might turn around and say that of course she has no belief that such things really exist: if a theory of descriptions commits her to such existence, she might demur from the theory. [1] When a speaker says, “the golden mountain must be worth a lot of money” are we to assume that she believes that the golden mountain exists to be referred to and quantified over? It is noteworthy that these descriptions don’t have names or demonstratives associated with them, unlike descriptions of people and places: there isn’t an antecedent assumption of existence-committed reference of the kind suggested by names and demonstratives. We are not already referring to such things in other ways; the description is the only way we have to speak of them. So a referential or quantificational treatment of them is not intuitively natural: their meaning is not naturally taken to involve the kind of ontological commitment we associate with other referential devices. Their “logical grammar” is not that of a straightforward assertion (or presupposition) of existence, whether by singular term (name or demonstrative) or by existential quantifier.
The idea that descriptions are name-like or demonstrative-like has never seemed particularly attractive—they seem semantically sui generis—and this has fueled the ascendancy of Russell’s quantificational theory. But that theory too has some untoward consequences, notably the consequence that a freestanding description is really a whole sentence: if I utter the sentence fragment “the queen of England”, I have really uttered the whole sentence “there exists a unique queen of England”. The film title The French Connection is equivalent to There Exists a Unique Connection that is French or some such. Maybe descriptions imply or presuppose existence and uniqueness (or maybe not: see above), but it is hard to accept that they state existence and uniqueness (this was Strawson’s point). If they did, it would be possible to negate them; but we can’t say “It’s not the case that the queen of England”. So neither of the two standard theories looks very attractive, despite their hegemony: the suspicion grows that they have been accepted mainly because of a lack of any better alternative. The appearances suggest that descriptions are not referring devices like names and demonstratives (which don’t contain the particle “the”) and they are not like quantified expressions either: they are what they are and not some other thing. The meaning of “the” is unique—which is why we have the word. But then, what kind of semantics do they have? Their meaning appears not to be referential (in any clear sense of that elastic term), either in acts of singular reference or in acts of quantification; but if not, what kind of meaning do they have? Are we to say that they have non-referential meaning? That would not be unprecedented, since words like “not”, “and”, and “if” are also used non-referentially; but descriptions are at least nouns, unlike these words, so the idea of reference clings to them more tightly. This is really a problem: descriptions fit none of our referential paradigms, but they also don’t fit non-referential paradigms. They occupy a curious semantic no-man’s land. We might say that they serve to “introduce a topic” or “identify a subject”, but these phrases don’t really help to pin down how they specifically function. Indeed, they seem to challenge the whole referential framework: the concept of reference (or lack of reference) is unable to cope with them. They don’t denote, but they don’t not denote either. Russell could have written a paper called “On Not Denoting” and proceeded to discuss the semantic peculiarities of “the”. He was right that descriptions don’t denote like names and demonstratives, but he was wrong to suppose that they denote in the way quantifiers do, i.e. with variables ranging over existent entities. There is no denoting at all going on with descriptions as such, as opposed to certain instances of them, but what is going on remains obscure. [2]
A radical solution may be proposed: the whole framework of referential semantics deriving from Frege and developed by others needs to be abandoned. It is hard to remember that in the old days meaning and reference were not closely associated: meaning was held to consist in ideas of the mind; it was not a matter of word-world relations. Today we would say that meaning is a matter of concepts, which are psychological entities: they have a role in the mind but any supposed relation to things outside the mind is purely incidental. This is non-referential semantics (meaning is completely “in the head”). According to a view like that descriptions have meaning in virtue of internal psychological factors, so that the idea of reference never comes into the picture. In the extreme, this internalist semantics says the same thing about all expressions, including names and demonstratives, but we could restrict it to the case of descriptions: they express concepts but they don’t refer to anything or quantify over anything (i.e. refer by means of variables). The case of descriptions, then, provides support for such a non-referential semantics—though it is certainly a contrarian point of view (compared to orthodoxy). Short of that, we have an unsolved semantic problem on our hands, despite the enormous amount of attention paid to the word “the”: we still don’t know what this word means. We know what “that” means and we know what “there is” means, but “the” leaves us baffled. Russell was right to fret mightily over this little word, and Strawson was right to question his theory, but in fact it remains as puzzling now as it was over a hundred years ago. [3]
[1] I am not saying that the descriptions listed clearly fail to refer to existent entities; I am saying that a theory that says they definitely do is going out on a limb that no semantics should venture. We don’t want to end up saying that a speaker’s sentences commit him to the existence of things he expressly repudiates: a nominalist, say, should be able to use these descriptions with a clean conscience.
[2] The semantics of descriptions should not be dependent on their type of subject matter: it should be the same whether we are speaking of entities acknowledged to exist or things to which we are reluctant to ascribe existence. Indeed, it should be neutral with respect to whether anything exists. Building assertions of existence into the very meaning of descriptions burdens them with far too much ontological responsibility. Questions of existence are far too controversial to be presupposed by semantics.
[3] Note to experts: I am well aware of the complexities of this subject, and the myriad ways of wriggling out of objections, and the passions aroused by the meaning “the”: but I am trying to cut through all that to expose a basic weakness in our thinking about descriptions since Russell wrote “On Denoting”.