the problem that languages solve

Believe it or not, the question of what language really is can be quite contentious within linguistics. Here's one statement:

"Languages represent a class of solutions to a problem that is undeniably unique in its scope and nature: the problem of mapping a hyperdimensional meaning space onto a low-dimensional channel."
(Bates et al (1998) in Bechtel & Graham (eds), A Companion to Cognitive Science, p595)

I have a lot of respect for Bates et al, but on this occasion I think I have to disagree. I don't think that providing a mapping between meaning and sounds is really the core problem that languages solve. Rather, I would suggest, the problem is communication, and the links between sound and meaning are only the means of solving that problem. To me anyway, it makes more sense to think of things that way round.