In “Conceptual Truth” (Proc Aris Soc Supp 80: 1-41), Timothy Williamson makes the following argument. The context is that he’s attacking Jackson’s argument that there must be some common doctrine held by people who use terms with the same meaning.

bq. Putnam’s insight is relevant far beyond the class of natural kind terms, as Burge observed (1986). Even where we cannot sensibly divide the linguistic community into experts and non-experts, the picture of a natural language as a cluster of causally interrelated but constitutively independent idiolects is still wrong, because it ignores the way in which individual speakers defer to the linguistic community as a whole. They use a word as a word of a public language, allowing its reference in their mouths to be fixed by its use over the whole community. Such verbal interactions between speakers can hold a linguistic practice together even in the absence of a common creed which they are all required to endorse.

Whatever its merits as an argument against Jackson, this seems to me to be a quite bad argument against the view of “natural language as a cluster of causally interrelated but constitutively independent idiolects”. The problem is that individuals may choose to defer to anyone at all. If Williamson’s argument is to work, the ‘linguistic community’ has to be the whole world, and there has to be just one natural language. But that’s crazy, so Williamson’s argument doesn’t work.

We’re all familiar with examples of common loan words like ‘Schadenfreude’. That looks like a case where speakers of English (or other languages) defer, to the extent they defer at all, to experts who are not English speakers. That is, they defer to Germans. But Germans aren’t part of the linguistic community of English speakers.

Now it might be argued that really English speakers are only deferring to other English speakers. After all, ‘Schadenfreude’ is a loanword that has been incorporated into English. But I don’t think this response can be maintained. For one thing, the first English speakers who started using ‘Schadenfreude’ did not defer to other English speakers. For another, the kind of pattern we see here, namely borrowing words from other languages, can happen all the time, and on an ad hoc basis. An individual speaker may choose to defer to English speakers, or Bengali speakers, or Latin speakers, or speakers of any other kind of language, on a moment to moment basis. If Ishani and I find it convenient to adopt some term from Bengali into the language we use to talk to one another, we can, and we are under no obligation to use that term the way that other English speakers do.

If Williamson’s argument against idiolects, and for public languages in a more traditional sense, is going to work, there needs to be a linguistic community that goes with each language. And speakers must be required, in virtue of speaking that language, to defer to it. But this isn’t how language works. We can choose to defer to whoever we want at any time. Or to not defer if we insist on using a term idiosyncratically.

It’s true, and important, that the meanings of terms in my mouth is determined in part by the usage of experts, other language users and so on. But this isn’t inconsistent with the picture of overlapping idiolects. I could well choose to have the meaning of ‘sofa’, or ‘Schadenfreude’, in my language determined by the usage pattern of a broader group. What would be a problem for the idiolect view is if I was required, in virtue of speaking the language of some community, to defer to that very community. But I’m not. And unless Williamson wants to say there is really only one linguistic community, consisting of the whole world, and one public language, which we all speak fragments of, I don’t see how facts about deference can help sustain the traditional picture of public languages.