Chomsky always was a bit pompous in his self estimation (q.v. his recent estimation of himself as the Galileo of language), but at least his early work is for the most part legible. There is a strange division of labor in mainstream linguistics where Chomsky is afforded the de facto title of philosopher in chief. The rest of the standard theory/government-and-binding/principles-and-parameters/minimalist crew would appear to prefer their role as sheep, only elaborating upon or justifying the philosophical insights of their ersatz shepherd. But this early work, Cartesian Linguistics, does develop a program and attitude which has been rewarding to many, even if the underlying assumptions of the Cartesian program turn out to be neither valid in the long run, nor justified by their sources.
This book is pompous in another way, too. About a third of the book is made up of lengthy quotations from French and German. In the edition I own, these are untranslated, so I am left to recall my French as best I can and skip the German. A slightly less pompous edition is now available with the passages translated.
Nevertheless, Cartesian Linguistics presents a unified and in some ways compelling argument for the innateness of languages, an ideology of language which characterizes almost all work in modern syntax and semantics. This ideology of language has been seriously questioned in recent work of an anthropological, typological, cognitive, or historical bent, but the Cartesian emphasis still would appear to hold a place of honor in introductory linguistics classes.
The error in Chomsky's ways seems to be in his assumption that language is primarily the organ of thought, if not only an organ of thought. As absurd as it may seem, Chomsky has over the years stuck to his guns that sociocultural and communicative aspects of language are some kind of side-effect of language as thought. There is of course a great deal more to communication than generating, transmitting and decoding thought, but that would align linguistics with other fields of study, thereby calling into question the autonomy, or what Hanks called the irreducibility of language. Chomsky did not worry about the purity of linguistics, however, when he launched his sorties into the sagacious field of philosophy.
Chomsky misreads many of the quotes he would have us read. I have read Humboldt in translation, not in the German, yet I can tell that Chomsky's reading of Humboldt is violently wrong. Humboldt was not a generativist. His view of language was dialectical, seeing the expressions of the speaker as being both ergon and energeia. Ergon is the language as work, as in a work, something that has been created and now constrains linguistic creation as the sedimented, ossified, and more or less frozen product of historical tradition. Energeia is the language as work, as in working, and includes the creative effort to move against and beyond the dead mass of ideas to pull the language into new directions by means of novel forms and expressions. Chomsky is a bit dim or possibly dishonest in seeing his own ideas reflected in Humboldt. For Chomsky, ergon is innateness and energeia is generative creativity. The actual ideas in Humboldt are dialectically historicist and romantically relativist; not rationalist, but closer to Sapir and Whorf than to anything in Chomsky.
Chomsky also fails to take into consideration the historical grounding of the Port-Royal grammarians. The Port-Royal Grammar was not a theoretical attempt to justify a belief in the innateness of language. In fact, it was more akin to modern prescriptive grammar but with a vengeance. The idea was to abolish variation (condemning variation as irrational to be sure) for the sake of a nationalist ideological unity. For the Port Royall folks, there could be only one true reasoning, and so only one variety of Language (which happens to be French) would be tolerated. Chomsky's appropriation of this prescriptive French tradition (which is alive and well, have you been to Paris?) is an act of academic piracy.
I will wrap up by stating that there are numerous vibrant ideas in Cartesian Linguistics, nevertheless. Some later work in pragmatics and Montague semantics is anticipated here in spots, and the door was left open (at least in the mid-sixties) for some work of a more anthropological bent. At the end of the work, there is also an attempt to save (or apologize for) the at times violent comparisons between his work and that of his 'rationalist' predecessors:
"...a certain distortion is introduced by the organization of this survey, as a projection backwards of certain ideas of contemporary interest rather than as a systematic presentation of the framework within which these ideas rose and found their place."
Chomsky could use a bit of this humility and hermeneutical insight today, I feel.