I think Berkeley would endorse the following argument:
- The rules governing a bit of language cannot tell agents to perform or refrain from actions in certain circumstances unless the agents can recognize the obtaining or not obtaining of those circumstances prior to the introduction of that bit of language.
- A word refers to an object only if the rules governing that word tell the agent to behave differently with respect to the use of that word depending on whether that object is present. (E.g. a necessary condition of ‘rabbit’ referring to rabbits is that the rules governing ‘rabbit’ specify that different ‘rabbit’ sentence are assertable in the presence of rabbits from those that are assertable in the absence of rabbits.)
- Prior to learning ‘thing language’ (in Quine’s sense), no one is capable of recognizing the presence of mind-independent material objects.
- The words of ‘thing language’ do not refer to mind-independent material objects.
Obviously Berkeley didn’t actually give this argument. (For one thing, he didn’t know about Quine.) In saying that Berkeley would endorse it, I mean to make two interpretive claims: first, that Berkeley endorses (something like) each of the premises, at least implicitly, for reasons which are independent of and prior to his immaterialism, and, second, that considerations similar to those raised in the argument are in play when Berkeley is arguing about the meaningfulness of Locke’s talk of material substrata.
Among these premises, I think (3) is the one that is easiest to attribute to Berkeley; it follows from what Martha Brandt Bolton calls the theory of ‘idea-objects’. Prior to the introduction of any conventions regarding the use of ideas as signs, ideas only represent by (exact) resemblance. But, by the Likeness Principle, ideas could not resemble mind-independent material objects. The fact that these are premises in Berkeley’s main arguments for immaterialism shows that he takes them to be prior to immaterialism.
Pinning (1) on Berkeley is trickier, but I think it follows from his view that suggestion arises by habituation, and linguistic rules are typically followed by suggestion. That is, in order to follow a linguistic rule, I have to be conditioned to pass from one idea to another, or from an idea to an action, or something like that, by experience. But I can’t experience the correlation unless I can experience both correlata first. These views are also clearly prior to immaterialism.
I am much less confident about (2). A section in a yet-to-be-written chapter of my dissertation will address questions of reference in Berkeley’s theory of language. I still have more thinking to do about it. It sure seems like something like this is going on in Berkeley’s account of how physical object talk works, but the theory also needs to be able to deal with spirits, and that’s hard (see Cummins).
For present-day philosophers, I suppose the obvious premise to reject is (3), the same one that Berkeley most clearly endorses. Quine, however, seems to be committed to (3); that is, he seems to think that it’s only by learning the ‘thing-language’ that one can come to think about (e.g.) rabbits, or even some one particular rabbit. Quine’s inscrutability of reference thesis, however, commits him to rejecting (or at least refraining from endorsing) (2).
(cross-posted at blog.kennypearce.net)