one thing i think we should not discount is the transformational nature of language (see transformational grammar by andrew radford for more)
Radford's books are not too good, IMHO. I remember I once tried to read that book on my own, before I became a linguist. After the chapter on X-Bar theory, the book just became incomprehensible. The intro chapter to the book is one of the best introductions to Chomsky's ideas-- but a terribly uncritical one.
Radford has two newer books which are introductions to more recent attempts at defining a new grammatical model within the Chomskian school. The attempts are a mess, and by now quite likely to be very different from whatever Radford describes in these newer books. I used one of these two books in a course looong ago, and it sucked, too.
If you really want to learn about Chomsky's theories of grammar, I'd recommend this book by Peter Culicover-- it covers more or less the same ground as the Radford book you mention, is more concise, and better written. It's not the newest stuff from the Chomskian school, but if you are not a linguist, you probably want to take a look at an actual linguistic theory that has been influential, and not a quicksand bog of current work which will doubtlessly soon abandoned, which means that the most appropriate Chomsky work for you is that from the '80s.
and the principles of Chomsky (also described in the book).
Chomsky is a doctrinaire who has built an academic cult around him.
For example, back in the late '50s and '60s, Chomsky and his school argued that grammars needed to have transformational rules because of the existence of so-called "long distance dependencies", which they argued could not be encoded with contex-free grammars.
However, in the late '70s and early '80s, it was shown that the argument for transformational rules had been wrong since the very beginning-- a context free grammar not only could account for long distance dependencies, but could actually do it *better* than transformational rules. A whole other body of analyses which made use of transformational rules were also shown not to require them at all. And another of the major original motivations for transformations, accounting for the related meanings of active, passive and interrogative sentences, vanished with the coming of age of formal semantic analyses for natural language.
So have the proponents of transformational rules relinquish them, given that the arguments for having them have over time proven to be more and more misguided, and arguments against them more and more convincing? No. They expanded their use. And because of their influence in academia, this kind of linguistics, which is what Radford's book describes, is the mainstream in the US. Pick up Pinker, for instance, and this is what you will be shown. The approaches which put this kind of linguistics in question, which have been around for over 20 years, are a niche within US linguistics, and the onlooker has to dig deep to see them-- and if he does see them, he doesn't find a unified voice like in the mainstream, but rather a morass of dissenting viewpoints. Which, for the most part, write for themselves and each other, not for the outsiders.
The most accessible (but by no means representative) introduction to such an alternative viewpoint is Sag and Wasow, but frankly I don't think that book is that good an introduction to linguistics. It is very clear and explicit about its material, though, and if you are determined to read like something like Radford, this book would make a good contrast point.
IMHO this whole situation is now slowly changing. Chomsky's influence over USian linguistics is certainly fading-- the major new linguistic framework in the US in the 90s was not due to him, for instance, and now he's retiring. But only time will tell.
[ Parent ]