It's kind of ridiculous that Chomsky is #1. You can't deny it was influential. Chomsky singlehandedly split the linguistics field into two camps, and it's taken the better half of the century to put his ideas to rest.

He basically argued that all languages are governed by the same set of rules, that such rules were innate, and that languages could thus be parsed into trees. The problem arises that a lot of primitive languages don't even seem to support the notion of recursion, let alone something that parses to a syntax tree. And the statistical NLP folks have a major bone to pick with such ideas, because real language apparently isn't so clean.

What seems more plausible is that more advanced societies developed rules of logic that they then imposed on their communication, hence the property that well-written English ought to be pars-able to a syntax tree. But even then you have ambiguities in how the tree is structured.

http://en.wikipedia.org/wiki/Syntactic_ambiguity

The flip side is that Chomsky's framework hasn't been useless to science. You can do some interesting things if you can parse a piece of text's syntax tree with halfway decent accuracy. One example would be idiom-finding. If a word seems misplaced, or used out of its usual context (e.g. "kicked the bucket"), then it is likely an idiom. One could simply do an analysis using the frequencies of the words as they appear next to each other, but applying a filter on only those phrases that are likely to be idioms improves your results by quite a bit.