A DEFENCE OF MILL'S THEORY OF NAMES

By David Cole
Philosophy Dept.
UMD
revised March 17, 1998

J. S. Mill held that the meaning of a name is just the referent of the name:

	...proper names are not connotative: they denote the individuals 
	who are called by them; but they do not indicate or imply any 
	attributes as belonging to those individuals.  (Mill 1961, p. 20)
Michael Devitt and Kim Sterelny, in their text _Language and Reality: An Introduction to Philosophy of Language_, rehearse four well-known arguments against Mill's theory. They conclude that we should follow Frege and postulate senses; the only other alternative is to follow Meinong and Lewis and inflate ontology. I will defend Mill's theory and try to show how we can respond to each of the four objections without postulating senses or inflating ontology.

1. Identity statements:

"Everest is Everest"

is not informative, whereas

"Everest is Gaurisanker"

is informative. On Mill's account, they have the same meaning. But "...the sentences do differ in meaning. So the names must differ also; the Millian view is wrong" (D&S, p.26)

Reply: This objection to Mill's theory turns on epistemic considerations that are irrelevant to the semantics of names. D&S argue that the sentences differ in meaning by pointing to "the very different roles these sentences play in peoples' lives" and to remark that "they are epistemically and cognitively so different" (26). But they no where show that any of _that_ bears on meaning, especially on their preferred account of meaning in terms of truth conditions. Why should we suppose that the two sentences differ in truth conditions, given that "Everest" and "Gaurisanker" co-refer?

Note in particular how we know the first is true: there is a (contingent?) rule that in a single context each occurrence of a name has the same referent as every other (not true in decently encrypted messages!). But that only tells us pseudo-apriori that the first is true, not what it means. It might be about a dog, a game, a mountain. So someone could know that the first is true without knowing what it means -- the meaning of the referring term "Everest" is irrelevant to the special epistemic status that the first sentence enjoys, and it enjoys that status only because of certain knowledge we have about how our language works. Mill could agree that the two sentences differ in that we might know that the first is true but not know that the second is true, but that doesn't show that we know what either means, nor that they differ in meaning, nor in partcular do these epistemic considerations show that the meaning of "Everest" is anything more than a particular mountain.

Consider now the meta-statements:

"Everest" and "Everest" co-refer.

"Everest" and "Gaurisanker" co-refer.

Clearly, at this meta-level, it is true on Mill's theory that the first is not (especially?) informative but the second may be an important discovery. The discovery is a discovery about the meaning of "Everest" and "Gaurisanker", namely that they have the same meaning. Since Mill's theory has the consequence that the former sentence is trivial but the latter can be an important discovery, this difference can hardly be cited against Mill's theory.

Finally, note that we can construct an argument of exactly the same form with terms that are _stipulated_ as having the same meaning. Suppose for example that the FDA stipulates that "hotdog" and "furt" shall mean the same. Still, there's a difference between

Hotdogs are hotdogs.

and

Hotdogs are furts.

Folks who have no idea what hotdogs are, and so don't know what "hotdog" means, can nevertheless know that the first is true. While persons who don't know that "hotdog" and "furt" mean the same thing will not know that the second is true. But this difference in epistemic status is no reason to conlude that there is a difference in meaning, or in what makes the two sentences true.

Moral: Sometimes two sentences have the same meaning, but some people don't know this. When this happens, the sentences will play different roles in those peoples' lives, and may have different epistemic status.

2. Negative existence statements:

One way of putting this objection (not quite how D&S put it) is: if Mill's view were correct, then the true sentence

"James Bond does not exist."

would have a subject term which lacked a referent and so, on Mill's theory, would lack all meaning (on the theory, the referent is the only meaning a name has). But then the sentence would be meaningless, and so not true.

Reply: The sentence concerns the exotic realm of truth in fiction. Perhaps the sentence can best be understood as saying:

The "James Bond"-stories are fictional.

That is, as used in the works which appear to refer to one "James Bond", "James Bond" does not in fact refer. In understanding the situation along the lines suggested here, we practice what Quine called "semantic ascent" -- we move to the meta-level and produce paraphrases of the original sentence, paraphrases that talk about language. This is similar to the way we treat the "nobody" constructions in the King's confusions in Alice in Wonderland: we use paraphrases to clarify and avoid unwanted ontological committment. What appears to be in a referential position in the original sentence does not occupy that position in the paraphrases that more clearly capture the truth conditions. With myth and fiction, the paraphrase involves talking about certain myths and works of fiction. This move does _not_ involve us in supposing that "James Bond" refers to a "fictional character" nor to an "idea' -- it is simply the suggestion that, like "nobody", "James Bond" doesn't refer to anyone or anything at all. These considerations bear on the next objection.

3. Empty Names objection:

The following sentence is true:

James Bond is disgustingly successful.

But on Mill's account the sentence couldn't be true, because the subject term does not denote. So it would be meaningless, and that would make the sentence also meaningless, and not true. And D&S are especially stern: don't try to tell us it refers to a) a fictional character, which gives up all sense of reality or pretense to being scientific, or b) an idea (an idea isn't the referent in "Reagan is wrinkled", so it shouldn't be the referent here).

Reply: Again, these are peculiar constructions on which to rest a refutation of a theory. Anyone who did not know the Fleming stories who heard the displayed sentence asserted, would suppose there was a real person (or maybe a product) named "James Bond". So on the face of it, these sentencess are simply misleading.

But we sophisticates do talk this way sometimes. Now consider the truth conditions for such talk. It seems plausible to hold the truth conditions are close to those given by David Lewis in "Truth in Fiction": If what Fleming wrote were true, James Bond would be disgustingly successful. This seems to nicely capture what we are doing in making claims like the above (and with "Hamlet was neurotic", etc.). We enter into a little make-believe pretense, talking _as-if_ the stories were true. But so understood, there is no incompatibility with Mill's theory. If "James Bond" denoted a person with the biography set forth by Fleming, that person would be disgustingly successful. In the original sentence, we aren't trying to refer by uttering "James Bond", we are making a claim about the representations made in the novels.

4. Opacity

I'll set this one out as an explicit contradiction:

Falwell believes Bob Dylan corrupted America.

Falwell does not believe Robert Zimmerman corrupted America.

These are both true (just ask Falwell). But on Mill's theory, since Bob Dylan and Robert Zimmerman are the same person, the two names mean the same thing, so the sentences mean the same thing (compositionality), except that the second is negated. But then the theory has the consequence that two true sentences are contradictory.

Reply: This is well trod turf. The basic idea of the reply I prefer is that once again in the relevant de dicto readings of the displayed sentences, we are not using "Bob Dylan" or "Robert Zimmerman" to refer. We are mentioning them. We are describing a certain mental representation that Falwell has. Falwell believes "Boby Dylan corrupted American". He may even have said this to himself. He believes "Bob Dylan" is the name of a person who corrupted America. He may well have all these Dylan representation without the Zimmerman representation. But this is consistent with Mill's view. In de dicto belief attribution the attributer is not using the terms in the attribution, and is not using them to refer.

For example, an atheist does not contradict himself in saying "Falwell believes God will punish Bob Dylan". On de dicto readings of belief attribution, there is mention of terms which may or may not refer, but there is not referring use of the terms. Consider a parallel argument to this objection to Mill, which does not turn on Mill's theory but rather on ordinary synonymy: Falwell believes telling lies is immoral. But Falwell does not believe mendacity is immoral. (just ask him!). But "mendacity" and "telling lies" mean the same thing -- see dictionary. So the sentences are contradictory (again by conpositionality), yet both true -- which is impossible. Again, all is solved if we don't take the opaque contexts as use, but as mention, as describing someone's representation of reality. Two expressions that mean the same thing might appear in otherwise identical de dicto belief attributions, yet the one be true and the other false. So there is nothing surprising if this happens with names, and is consistent with Mill's theory. (Other commentators boldly take the line that has the consequence that Falwell believes that Zimmerman did not corrupt America and Falwell believes that Zimmerman (i.e. Dylan) corrupted America. (e.g. Martin 1987 pp. 185-9) But, presumably, he doesn't know he believes both these.)

It is revealing, I think, that there is no problem with the paradigm case of opacity, direct quotation:

Falwell said "Bob Dylan is corrupting America".

Falwell did not say "Robert Zimmerman is corrupting America".

These are both true, and the failure of substitutivity here counts not at all against Mill's theory. Mill would say the two names mean they same thing, namely their shared referent, but they are still distinct names. Saying the one sentence is not saying the other, assenting to the one is not assenting to the other, and de dicto believing the one is not believing the other.

Bibliography:

Devitt, Michael and Kim Sterelny 1984 _Language and Reality: an Introduction to the Philosophy of Language_ MIT Press

Frege, Gottlob 1892 "On Sense and Reference" reprinted in _The Philosophical Writings of Gottlob Frege_ Peter Geach and Max Black (eds.) 1952 Oxford: Blackwell

Lewis, David "Truth in Fiction" APQ?

Martin, Robert M. 1987 _The Meaning of Language_ MIT Press

Mill, John Stuart 1961 _A System of Logic_ Longmans