Here’s an argument that the semantic content of some term is directly referential. It’s designed to mirror the regress argument for foundationalism in epistemology.
The basic idea is that descriptivism can’t be true for every term. Descriptivism is (roughly) the view that the semantic content of a proper name is some description (or cluster of descriptions that the speaker associates with the name. The version I have as my target extends this to predicates/kind terms. Let’s call this view Full Blown Descriptivism
Now consider some term A. If Full Blown Descriptivism (FBD) is true, then the meaning of A is some description say, the F. F will be a term or a phrase that contains terms. To keep things simple, let’s assume that it is a term. Assuming F is a meaningful term, then if FBD is true it’s meaning will be some other description say, the G…and so on and so on. Let’s call this fun, little chain a “meaning chain”
Either meaning chains are such that
- All of them go on infinitely
- All of them go in a circle
- Some terminate in some meaningless term
- Some terminate in some term with non-descriptive semantic content
It seems like (1) couldn’t be true. I certainly don’t have an infinite vocabulary, but I think my words are meaningful. I suppose a proponent of FBD could go with (2). I don’t have an argument against (2), but it seems odd to think that all of my terms have their meaning ultimately ground in themselves. I think we can rule out (3).
However, a proponent of FBD can’t accept (4). So it looks like we have an argument against FBD, and for the conclusion that at least some semantic content is non-descriptive.
If this argument were sound, then I suspect we would be well on are way to a decent indirect argument for Millianism.
(Posted from my Android mobile phone)
I wonder if FBD might be a straw man. Consider that your conclusion, that FBD is false, is hardly an argument in favor of a robust version of Millianism since it tells us nothing about the semantic content of names (or any other class of terms). So a descriptivist need not be worried by your conclusion, in particular, he need not worry that we must concluded that names have no descriptive semantic content. The falsity of FBD will only be a problem for someone who holds that there are no terms that lack a descriptive semantic content. But does anybody actually hold this view? It is not obvious that Searle (for example) need be committed to it.
Jason,
That’s right. We don’t have an argument for Millianism yet. I hadn’t yet posted how I thought the rest of the argument might go.
The way I intend to proceed is show that FBD is false, and then press the descriptivist to provide us with a non-descriptivist semantics for whatever terms will be non-descriptivist. Or take the time to try and flesh out (on my own) what such a theory would look like…
I suspect that whatever the result is it will look enough like a direct reference view that it will be vulnerable to some of the same objections that Millianism about proper names…in which case the hybrid semantic theory will lose it’s main virtues over Millianism. (I only have a vague hunch that things will pan out that way)
Mark Kalderon just suggested one such view to me …the descriptivist need not say that all descriptive content is a description. Descriptive content could be simply a cluster of properties…I’m going to think more about this option and update the post (or add a new one)
Thanks for the comment
Hey Andy,
I think Kalderon is right and I suspect that, unfortunately, things will not turn out for you the way you hope they will. So consider Russell era KBA/KBD descriptivism. Now drop the part where there are any logically proper names. You can’t Frege’s puzzle-ize on the non-descriptive terminal bits of meaning because they are properties presented to you in your immediate experience. That’s basically Russell’s criterion for being a meaning that isn’t composite and descriptive (again, ignoring the logically proper names). And you don’t get a problem of empty expressions since you just use complicated descriptions built out of properties present in your immediate experience to serve as the semantic content of even apparently empty expressions. Now this view is false as the day is long, but I think it nicely shows how there’s no neat route to Millianism here.
Hey Chris,
That seems right. The other thing I mentioned to Kalderon is that the argument should probably just be restricted to versions of descriptivism that accept the hypothesis that semantic content requires a description that the person associates with the term. And then I was thinking that we’d argue against hybrid-theories (or non-description descriptivism theories) on other grounds. One of those grounds was that I had the above suspicion.
But I see you’re rightthat the Russelian hybrid theory won’t be vulnerable to my complaint. Do you think it has problems that other versions of descriptivism don’t? I’m curious because it may make the above argument still useful. Also, I wanted to look into other theories of this kind. Do you know of any recent defenders of something in the spirit of this kind of view?
I don’t think anyone sane defends anything like this. I think what shows that is false is all the arguments that you know show that it’s false. My point was just that the foundationalist-style argument doesn’t do it since it does not cut ice against this sort of view.