Uffff there is some event horizon beyond which the systematizing impulse collapses in on itself and pulls you in black hole style. This is a comprehensive and erudite field survey with tons of interesting observations by a brilliant borbetomagus of the language field, but language isn't math and beyond a certain point I--personally--can't help but be appalled by the scholastic rationalism of the many fine distinctions and inferences and cloud scaffolding all throughout here. I knew how I felt about that in the case of generative syntax, say, or constraint-based phonology, but I do certainly appreciate Levinson and his wordhoard for helping me clarify that it applies even for a field in which I feel such a deep investment and interest as pragmatics. Humanistic, impressionistic, hermeneutic, stochastic methods are the methods you should use when you don't have scientific evidence, as long as we're talking about the real world (and not math, and I guess also theoretical physics has a really good track record of not needing real-world evidence, but it's somehow closer to math than the rest of the sciences, I guess?). And certainly we have much wider crosslinguistic-descriptive and more extensive psycholinguistic and sociolinguistic work on all this stuff at present. From the perspective that this was a field in its empirical toddlerhood, this is an amazing work. But some of it is superseded and other of it is just not my philosophical bag, though I'm not saying it can't be yourn. ( )