sotanaht wrote:Effectively, in order to have a meaningful exchange, there must be a commonly (between both parties) accepted source of meaning, which implies a third party source.
Like with scientific questions, that third-party source is not another person, but the world around you.
If Alice and Bob disagree on a fact, they don't need High Priest Charlie to settle the dispute, because they can both look at the same world and figure out what the fact is together.
Similarly (though not perfectly analogous), if Alice consistently produces sign X in some relation to a given phenomenon, Bob can figure out that when Alice says X, she means that phenomenon. He can then also say X to signal that phenomenon to Alice, and voila, the start of language.
(I do agree with your general point about shifting meaning being generally bad though, because it breaks the language that already existed and requires people do go back to this figure-out-what-people-mean-by-things stage. But we do have that stage to fall back on, which is why we can continually recover from the ongoing destruction of the existing language, because we made it up in the first place and we can make more when we have to. It just sucks that we have to again when we already had a perfectly good one until someone broke it.)
It's true that in a completely verbal exchange wholly divorced from life in the real world, there's no way to sort out real meaning (syntax maybe, but not semantics), but that's because meaning comes from reference to the world. (That, incidentally, is why Searle's Chinese Room fails to actually understand Chinese. It doesn't disprove functionalism, because the room can't do a huge class of functions that real Chinese speakers can: things like, given a photo of a duck on a lake, answer "what kind of bird is on the water?" It can tell you that a lake is a body of water and a duck is a kind of bird, but it can't
look for birds when prompted and return "I found one, and it's a duck").