TH: I still have a dog-eared copy on my shelf from undergrad days; great read, very stimulating, and probably dead wrong -- but hey, that's rational enquiry for you.
I think the Whorf-Sapir idea is interesting but wrong, personally. I was totally turned on by some of the citations from Indian languages, like "he's cleaning the rifle with a ramrod" coming out something like "he is causing a reciprocal motion of a dry spot in a long hollow thing, using his hand".
I've read a bunch of Amerind linguistics since, but I've never really gotten my head around polysynthesis. The Languages of Native North America (Cambridge Language Surveys) by Marianne Mithun gives the best description of it that I've seen. I highly recommend it if you're at all interested in such things. She dismisses Greenberg's Amerindian hypothesis, but I'm convinced it's right. See this example *t'ina *t'ana *t'una (meaning "son", "child", "daughter", found throughout the Americas) or any of Merritt Ruhlen's books.
The whole concept of 'a language' is a bit fuzzy, like the concept of 'species;' difficult to make a hard distinction between dialects (sub-species) because of the variations within a population of speakers, or to draw an absolute boundary between them.
There is an exact parallel with ring species, called "dialect chains". There are some in Eskimo regions, in Bantu Africa, and until a few hundred years ago, between Italy and France. I find it fascinating to be able to go from Italian, by small steps, always maintaining communication, and wind up with French.
Another parallel, which I pointed out on a thread a week or so ago, is that linguists can make very educated guesses as to what PIE's sounds, parts of speech, grammar and vocabulary were like, just like the biologists who figured out an ancestral hormone and receptor.
I tried reading Chomsky when I was in college, and it bored me stiff. Rather like pseudo science, there were a lot of assumptions and damn little empirical research (and most of that in English).
When I discovered Greenberg, on the other hand, I was floored. I predict that twenty years from now, Chomsky will be a footnote, and Greenberg will be considered a combination of Linnaeus and Darwin.
I agree regarding Chomsky. His death will be celebrated, but not so much his birth. And that's just by linguists. I call him the American Lysenko.
Whorf also has a neat diagram of English spelling. It's actually a finite state machine, although called such by Whorf. I've used it as a basis for a random pronounceable password generator.