Previously I gave a brief account of how the content of a proper name could be said to be a description, rigidified if necessary. This resolves one of the problematic cases that motivated Kripke to propose rigid designation in the first place. The other, much less talked about, problematic case is natural kind terms. The problem with natural kind terms is not that no description seems to pick out their referents in the correct way. We know very well how to describe water: it is the molecule formed by two hydrogen atoms and one oxygen atom. In fact that is what makes a natural kind term a natural kind, it can be picked out in terms of its natural (physical) properties alone. The problem presented by natural kind terms is that they seem to designate the relevant natural kind even before we know what the definitive properties of that kind are, or if it is really a natural kind at all. People have used the term water (or words that meant the same thing) for a long time, but it was only recently that we discovered that it was in fact H2O. And it is not the case that is because the substance people designated with the terms before the description was known was H2O alone; we rarely encounter water in a pure state. So how then was it possible for those people to talk about water, meaning H2O, given that they didn’t have the knowledge that would allow them to mean by the use of that term the relevant natural description?
A Kripkean solution to this problem is to propose that natural kind terms rigidly designate the kinds themselves, as if the kind was some sort of abstract object. Just like a name refers rigidly to a person in the Kripkean theory of language so too does this natural kind term refer to its kind, even though in both cases the speaker may not have access to a description that picks out its target. Or so it is claimed.
Obviously this talk of abstract kind objects seems questionable, but let us set that aside. Since it is my position that a description does in fact come along with the use of every possible word (in addition the descriptions provided by knowledge of that person from context and from personal experience, in the case of proper names) it is best simply to spell out the description that is behind natural kind terms, even when they are used in the absence of any knowledge about what natural distinctions make the kind a kind.
I propose the description is as follows: “the natural kind of thing that constitutes the majority of what I experience as X”, X being a pre-theoretical description of the specific kind in question. For example, in the case of water X might be: “clear, drinkable, composes rivers, lakes, ocean, and rain, results from melting ice and condensation”. The idea behind this description is that even when we use a natural kind term pre-theoretically we still have the idea that it is a kind, and that it is a kind for reasons that are independent of us, based in natural facts rather than constructed facts. When you come across something that seems basically the same, not because you made it that way, but that just happens to be the same, then it is natural to suppose that there is a reason for the similarity outside of the human realm. (Meaning that the similarity doesn’t just exist because of, say, a similarity in the way we happen to see the objects, or think about the objects.)
Of course this description may fail, there are cases in which there is in fact no natural kind (or no single natural kind) that accounts for the similarities observed. But since both the rigid designation theory and the description theory face this same problem it doesn’t seem like an insurmountable hurdle to me. Both theories can simply assume that in such cases that what we mean when we use that term is simply what is designated by the pre-theoretic description X. (Just because the rigid designation theory makes some terms rigid designators doesn’t mean that is has to claim that every term is such a designator.)
Again, just like with the case of proper names, the rigid designation and descriptions approaches can explain basically the same facts. And given that I think that we should prefer the descriptions approach. Again it separates meaning from an ill-defined conception of how reference works. (How rigid designators manage to reference objects has never been clearly spelled out. Kripke puts forward a partial explanation in which reference is passed from person to person causally, but it is never explained how reference is maintained over time, how it is transferred through causal interactions, or what it is at all. On the other hand when dealing with descriptions reference is just a way of talking about which real objects best fit the description, which isn’t mysterious in the least.) And again it explains how we can know what a natural kind term designates, while the rigid designator theory does not. And so, for basically the same reasons, I prefer descriptions as an explanation in these situations as well.