If the problem of universals was the great controversy of the Mediaeval philosophy, questions about naming may qualify as the parallel controversy in 20th century philosophy. This can be examined by way of Michael Devitt and Kim Sterelny's Language and Reality, which is subtitled "An Introduction to the Philosophy of Language." This is an accessible presentation, as it should be for an intended textbook, but it also presents fairly rigorious arguments for certain points of view.
The beginning of the modern problem of naming is given as the received wisdom of John Stuart Mill that "proper names" are not "connotative," i.e. they do not have any discursive meaning that can be explained.
...they denote the individuals who are called by them; but they do not indicate or imply any attributes as belonging to those individuals. [quoted from Mill, p.29]
Thus, a proper name refers directly to an individual and does not have any other meaning. This is the basic "direct reference" theory, and Devitt and Sterelny refer to it occasionally as the "Millian paradise," i.e. a simple, common sense theory that would be nice, if it could be maintained.
Unfortunately, or perhaps fortunately, it could not be maintained. Gottlob Frege (1848-1925) pointed out that the same individual may have different names, whose meaning is somewhat different. His own classic example was that "Hesperus" is the name of the "Evening Star," while "Phosphorus" (or "Lucifer") is the name of the "Morning Star"; but it turns out that the Evening Star and the Morning Star are the same thing, the planet Venus. The identity of the object, however, does not make it correct to call Venus in the evening "Phosphorus." Since logic demands that things produced by the free substitution of identities should be identical, the lack of identical truth for "Venus in the evening is Hesperus" and "Venus in the evening is Phosphorus" means that "Hesperus" and "Phosphorus" are not identical in meaning.
Also, if the entire meaning of a proper name is in the reference, then a sentence like "Pope John Paul II is Karol Wojtyla," should be no more meaningful or informative than "Pope John Paul II is Pope John Paul II." The second sentence, however, is a trivial tautology that tells us nothing, while the former gives us real information. A proper name, therefore, must have a meaning independent of its reference. In Frege's terminology, that is the "sense" of the name -- completing Frege's distinction between "sense" and "reference." "Phosphorus" and "Hesperus," "Pope John Paul II" and "Karol Wojtyla," refer to precisely the same objects but have different "senses."
So far so good. The difficulties come when we try and specify the content of the "sense." The classic form of this, building on Frege, was Bertrand Russell's theory of "definite description." In this, certain predicates, the description, apply to a certain individual and to no other. The "definite" of the "definite description" is that of the definite article: the President of the United States. Thus, at any one time, there is only one President of the United States, so that predicate uniquely specifies one individual. What goes along with this is that "sense determines reference" [p.38], so that the sense of every name is a definite description that picks out just the named individual.
There turn out to be many problems with this. The real problem, however, is one that Devitt and Sterelny do not consider, though many of their counterexamples depend on it: No specification of predicates can define an individual, for there can always be another possible individual for any number of predicates that, as universals, are specified. Individuals are concrete, which means that potentially an infinite number of predicates could apply to them. We can come nowhere near specifying such a number. Indeed, since an actual infinity is impossible, it only exists as a limit. Any finite number of predicates is still abstract, still a universal. If Russell really thought that a definite description applied to one individual and to no other, he was really making a very elementary mistake.
Devitt and Sterelny's ultimate, knock-down argument against description theories depends on this characteristic of individuals. They cite Hilary Putnam's "science-fiction" counterexample [p.62-63], that somewhere there could be a "Twin Earth," which has exactly the same characteristics and history as the actual earth. The name "Einstein" on the Twin Earth therefore would have exactly the same "sense" as the name "Einstein" on Earth (originator of Relativity, died in Princeton, New Jersey, etc.), but it would actually refer to the Twin Einstein (and the Twin Princeton), not to the Einstein known to us. Sense therefore cannot determine reference. The possibility of a "Twin Earth," of course, depends on the possibility of there being different individuals referred to by the same predicates. An elaborate counterexample is not necessary for that, just the principle that concrete individuals cannot be defined by abstract predicates.
Why would Russell have been led astray? One influence was certainly Leibniz, in whose metaphysics individuals are not truly concrete but are complexes of universals that are unique of their kind. From this came the principle of the "identity of indiscernables," i.e. if two individuals cannot be told apart, then they are the same individual. This renders individuality into a matter of predicates. In all this, there is also the temptation of the idea that, as a practical matter in ordinary life, definite descriptions are used to indentify individuals, as anyone knows who has ever needed to meet an unknown individual in a public place -- "I'll be the person wearing the red carnation" will usually be enough to do it. As we know, however, some descriptions will be better than others. Looking for someone with a beard, glasses, and a tweed coat will usually serve to pick them out, but not at a philosophy conference.
Ironically, another idea from Leibniz served to help undermine description theories, and that is of "possible worlds" semantics or metaphysics. If it is possible that I could be different than I am, then in some possible world I would be different than I am. The sense of what I am would then, however, be different; and so the original description would no longer determine the proper reference. A description of me would no longer specify me in some other possible world. Indeed, I could even have a different name in many possible worlds. So what makes me me in another possible world? Saul Kripke, who largely reintroduced Leibniz's "possible worlds" device into 20th century philosophy, said that proper names are "rigid designators," i.e. they refer to the same individuals across possible worlds, regardless of the differences in what they are like. As senses as descriptions change across worlds, however, they are not going to be as "rigid" as the names need to be [pp.51-52]. But, of course, if names directly refer to what is concrete, then the predicates, and the senses, are secondary.
Even if possible worlds metaphysics sounds like too much, and Devitt and Sterelny blanche at some varieties of it, it does serve to highlight the generality of descriptions. Thus, one might think that a description like "The President of the United States in 2000" would uniquely specify an individual, namely Bill Clinton. But it is actually possible that someone else might have been President in that year, so it is not just the description, but certain factual contingencies, that serve to fix the reference.
Having demolished description theories and their variations, Devitt and Sterelny then introduce their alternative, a "causal" theory in which an actual individual is "dubbed" with a name ("reference fixing"), and then a causal chain of transmission ("reference borrowing") transfers the name to new users [pp.66-67]. In fact, they say of a term, "It is meaningful because it has an underlying causal network" [p.74]. They eventually find some problems with this and modify it, in a descriptive direction, slightly. The real problem, however, is again more fundamental. In a "causal" theory of names, what is being caused? The object does not cause the name. The name does not cause the object (at least for Devitt and Sterelny). No, it is the naming speaker who causes the name to be the name. But then that is what everyone already believes anyway, and it is not a philosophical theory. What makes it a philosophical theory is the idea that the naming speaking makes the name refer to the object. This, however, begs the question. To make a name the name of something presupposes that there are names and that names refer.
"Dubbing" something with a name therefore does not create the nature of reference or the role of names, it merely "fixes" the reference of a particular name. A "causal" theory of names thus presupposes the existence of direct reference, upon which it is parasitic. "Dubbing," it turns out, is not even necessary. The quote from page 74 above was about a non-existent object, an "empty" name, which is meaningful, not because anything was ever dubbed, but just because a "causal network" brings the name to the user. But no causal network ever brought the name "Sherlock Holmes" to Sir Arthur Conan Doyle. He directly "dubbed" a non-existent object with the name, i.e. he began using the name in the absence of any real individual to be referred to by it.
The real problem with a "causal" theory of naming is how causality really makes any difference, how it adds something to direct reference. There is also a difficulty with the "network." Devitt and Sterelny don't seem to like the "Cartesian assumption" that users are aware of the meanings of the words that they use, but this "causal network" that provides meaning in their theory is something that, as a practical matter, no one will in most cases ever know. The chain of transmission of most names would be far more difficult to reconstruct than the chain of transmission of urban legends -- which are notoriously difficult, indeed impossible, to track down. So how is it that something that we really cannot ever know about, not just don't happen to know about at the moment, is going to count as the "meaning" of a word? This is beyond rejecting Cartesianism. It is proposterous -- as Devitt and Sterelny complain is the term with which some critics of causal theories reject them [p.78].
All that "causality" accomplishes in the theory is to link a name to something in reality. But this reality does not need to be a real denoted object, since fictional names can be picked up from the "network"; but the reality does not even need to be such a network, since there is no such network for someone who coins a fictional name. Indeed, the only irreducible link to reality is through the user; and the user who coins a fictional name is, unfortunately, going to have a Cartesian understanding of what he means by it. This is consequently going to be a poor kind of theory for realists like Devitt and Sterelny, who are trying to tie names to external reality.
Again, the fundamental flaw is simple enough. Causality is not a cognitive relation, and that is because, as Hume famously perceived, the cause and the effect do not need to resemble, contain, or even suggest, each other in the slightest. I may be using the name "Pope John Paul II" because I have borrowed the name, and the reference, from someone else, but the meaning, reference, and truths about that name have nothing whatsoever to do with where I got it. To offer the source as cognitively significant is to commit a kind of genetic fallacy. The source is so irrelevant that, in most cases, it is quickly forgotten -- which is why chains of transmission (the "Network") are so difficult to reconstruct.
Devitt and Sterelny try to turn such a criticism around, by damning reference without causation as "magical."
As Putnam puts its, "meanings just ain't in the head"...
Indeed, how could meaning be in the head? Meaning depends on reference and reference relates a person and his word to one particular external object and not others. An internal state of a donkey might be an excellent symptom, even a perfectly reliable symptom, of that donkey's having been kicked by its owner. But that internal state alone cannot make it true that the donkey has been thus kicked. Equally, no internal state of a man can make him the father of Wayne. And internal facts about salt cannot alone make it soluable in water, for salt's solubility derives as well from features of water. No more could internal facts be sufficient to determine reference. To suppose otherwise is to suppose that an internal state has a truly magical power. [p.63]
Meaning can be "in the head" because it is intentional and representational. The effect of a kick on a donkey is not a representation; it is the effect of a sufficient cause, which means that other causes for the same effect are possible. But this is not an example to use if one is to defend a causal theory, since any cause is sufficient to its effect, and knowledge could always therefore be explained as much by Descartes' Deceiving Demon as by any factual conditions in the world. Devitt and Sterelny want to avoid Cartesian solipsisim by considering "external facts" as part of meaning. But external facts are just that, external. If they cannot be made internal, then they can have nothing whatsoever to do with meaning or knowledge. They would be nothing to us. They are made internal through the representations of perception and consciousness. Devitt and Sterelny's causal theory of meaning thus must also represent a commitment to a causal epistemology, just the sort of thing, one might say, that was refuted by the example of Descartes himself. What they certainly would consider "magical" would be a non-naturalistic metaphysic, a Kantian phenomenology of empirical realism. This is the only way, however, to avoid the externalist paradoxes of causal epistemology.
So, if Mill was wrong, reference theories are wrong, and a causal theory is not even relevant, what do we do? For one thing, we can go back to Frege and ask about the relationship of sense and reference. Is it necessary that sense determines reference? Well, if individuals cannot be defined by any number of predicates or descriptions, then clearly the sense is not sufficient to determine the reference of any proper name. A name can only refer to a specific individual by a direct reference, as Mill and the direct reference theorists held. However, there are senses. There are definite descriptions of individuals, and these are actually used in ordinary life for the purpose of identifying real individuals.
If sense does not determine reference, but reference is an independent part of meaning, we should consider if this takes care of the objections that Devitt and Sterelny list against description theories.
This raises a general question. Even if sense does not determine reference, there still can be analytic predications of names -- "Hesperus is Venus appearing in the evening sky" looks like one them. Are there characteristics that are essential to individuals? If a name retains its reference across all possible worlds, then there can be no essential characteristics, because all such characteristics could be different in some other possible universe. But, as a matter of fact, is there anything essential in this world to "Socrates"? Well, yes. Socrates asked questions in the marketplace at Athens. If we are talking about someone who did not do that, it could not have been Socrates. Do we then have an analytic predication? Well, yes.
The problem is about whence the necessity of analytic propositions derives. In Kant's classic categorical analytic propositions, the necessity is the Law of Non-Contradiction applied to the relationship of the predicate to the meaning of the predicate already contained in the subject. But then, in equally Kantian terms, one can ask about the basis of the "synthesis" by which the various meanings of the subject term are associated with each other in the first place. There can be various grounds for that synthesis, and the necessity of an analytic proposition can be no stronger than the necessity of that ground. Analytic propositions about Socrates or Aristotle thus depend on the historical facts about Socrates and Aristotle. This is usually not thought of as a basis of necessity, but here it is. Similarly, the analytic necessity of a stipulative definition is based on the fact of the stipulation having been made. Other grounds of necessity can be stronger, as the necessity of physical law or the a priori necessity of the principle of causality.
If there is thus "unwanted necessity" in predications about names, this can be defused with a consideration of analyticity itself. Indeed, for all that Kripke is quoted by Devitt and Sterelny, his suggestion that there can be "analytic a posteriori" propositions is not mentioned in this context. The factual necessities of analytic predications about Socrates and Aristotle are indeed a posteriori.
Like analyticity, this raises a general issue. If we are going to accept anything like the "Cartesian assumption" that users are aware of the meanings of the terms that they use, why doesn't this create problems? It doesn't because "sense" can mean different things. Where we are talking about real things (Socrates, etc.) or natural kinds, to which Devitt and Sterelny devote a large part of their Chapter 5, there is going to be a difference, as Locke put it, between real senses and nominal senses. What the Cartesian user is aware of is his version of the sense of something, which is the nominal sense, while the real thing, or the real kind, will have actual or true characteristics, which will only be contained in a true or real sense. Locke, of course, was talking about essences, not senses, but the issues are parallel and related. The characteristics of a thing that make it what it is are its essence. The sense that we have of things may or may not correctly represent its essence. Hence the possibility of ignorance and error. In Frege's own theory of sense and reference, the reference of a term depended on what kind of term it was: Proper names refered to individuals, but general terms referred to concepts. Frege thus spoke of concepts, not as mental entities, but as characteristics of things, or of the world -- referents. This somewhat paradoxical terminology, however, is just a substitute for what the proper, traditional terminology would have been, i.e. "essences." What is essential to an individual is an indefinite number of factual characteristics. What is essential to a natural kind may also be somewhat indefinite, but it will also certainly be abstract, since it applies to a variety of objects. "Parts the hoof and chews the cudd" was sufficient to define animals that were clean and allowed for eating in the Bible [Leviticus 11:3]. As with analytic necessity, essences presuppose a synthesis, which is the result either of factual association or physical necessity.
When Devitt and Sterelny extend their causal theory to natural kinds, they are conspicuously avoiding essences. This is characteristic of 20th century philosophy. It is also characteristic of the naturalism that Devitt and Sterelny explicitly endorse. Devitt and Sterelny's commitments are not always unedifying. Their realism and representationalism are both commendable, and their confidence in science sets them apart from the post-modernist nihilists. The kind of naturalism to which they adhere, however, has problems. First, they say that their epistemology is naturalistic, meaning that all knowledge is empirical and a posteriori [p.9]. This means that their confidence in science becomes scientism, that no non-empirical issues ever come up in philosophy of language. This is not the case. More egregious is their statement that thier metaphysics is naturalistic, meaning physicalism [p.10]. This is a very heavy metaphysical commitment, with little motivation except that science and the cognescenti (and John Searle) seem to have decided that reality is physical. Since abstract universals are not physical objects, it is not surprising that a "causal" theory is preferable to a theory of essences.
While I will presuppose that Jerrold Katz has taken care of naturalism in philosophy of language, I also think that it is still the case the Berkeley took care of physicalism in metaphysics. How it is that physicalism can come back after Berkeley, Hume, and Kant seems to me a matter of inattention or negligence. All that Devitt and Sterelny can offer is this:
Physicalism is intrinsically plausible. It has excellent scientific suport from evolutionary theory, biology, and biochemistry. These sciences underscore the biochemical and physiological continuities between humans and the rest of nature. There are, we believe, no good arguments against this perspective. [p.10]
Unfortunately, there is not a shred of ontological evidence or argument here. All the evidence from the sciences is going to work just as well for Spinoza as for Devitt and Sterelny. If we are therefore to chose between Spinoza and whatever it is that Devitt and Sterelny believe, something relevent to metaphysics must be considered. Devitt and Sterelny, like most modern naturalists, are simply not interested in even getting into it. This is unfortunate, for a virtue of Devitt and Sterelny's book is that it does not privilege the position of language with respect to all other knowledge, such as has ususally been done by those, like the Positivists and Wittgenstein, who seek some ground to simply dismiss metaphysics as meaningless. The equivalent ground for Devitt and Sterelny may be their scientism. Flush with the project of cognitive science, they probably don't think that metaphysics belongs in their book. The book, however, is about the philosophy of language, and a proper philosophy book cannot be innocent of some metaphysics, where it is relevant, as it is here.