Noam Chomsky is one of the great intellectual figures of all time. Not only is he the dominant figure in linguistics, but he has repeatedly launched new revolutions to replace his older discoveries. He is very modest about his own achievements, calling them pre-Galilean, as if he is clearing the way for someone else to discover the true nature of the human language capacity; but as far as impact on a subject goes, he ranks with Darwin, Newton, Einstein, and Descartes.

He is very cautious about the implications of his theories on broader problems of cognition and evolution, and has been criticized by practitioners in those fields. His insistence on the autonomy of language can be questioned, but in linguistics it is the approach that, at the moment, works best, and generates the best explanations across many languages, and is used by if not the majority of linguists then at least a greater number than any other. Everything he puts forward is tentative and open to criticism and disconfirmation.

Most of the detail of his work is in syntax, though he has also made a very important contribution to phonology. From his early work in syntax he has progressively generalized and abstracted, trying to come up with explanation for how humans use language. His latest theories are extremely powerful, but minimalist. The specifics of the inherent machinery are reduced to the simplest that is logically necessary for language to work.

The explanatory programme has been at the heart of his work from his very first publication, The Morphophonemics of Modern Hebrew in 1951. The work most people regard as launching the first Chomskyan revolution is Syntactic Structures (1957), and with Aspects of the Theory of Syntax (1965) we have what is known as the Standard Theory. Despite the name, this has now been completely superseded by the Minimalist Program.

Before him, linguistics was mainly descriptive. The leading school was structuralism, which exhibited structural patterns as an advance on traditional grammar, but was still in a sense a collector's field, full of glittering specimens of great interest and diversity from exotic languages. Almost all the examples Chomsky uses are from English. He briefly alludes to things being done differently in Italian or Japanese, but does not need to amass a collection, since his target is the innate human language capacity, and his assumption is that this is a common human inheritance. He is explicit about this being an assumption, and that he uses it because it seems to be right, on what we know so far. If it proves false in any way, his theories will change.

In 1968 together with the phonologist Morris Halle, he produced the book The Sound Pattern of English. As an analysis of English it has some surprising ideas, such as that the relationship between divide and division is a regular part of modern English rather than a hangover from Latin and Middle English. It contains a general theory of generative phonology, with sounds built up from distinctive features. Though Chomsky has not returned to this area, and the theory has been modified by others, its importance is emphasized by the fact that it is usually cited simply as SPE or (undated) as Chomsky & Halle. So also with his book Aspects.

I'm afraid I've never seen anything by him that is readable by non-linguists. He is difficult and compact. However, Neil Smith, 1999, Chomsky: Ideas and Ideals, CUP, is a good up-to-date and untechnical introduction to both the linguistics and the politics.

Below I give a more detailed exposition of his theories of Universal Grammar. I can not here discuss any theory in any detail: that requires individual nodes. I just want to give an idea of his range, and the development of his thought.

Overview of theories

There are at least three major aspects of language use that require explanation. One is how speakers create utterances, the next is how listeners understand utterances, and the third is how young children become fluent without being taught: production, comprehension, and acquisition. (Speakers and listeners include users of sign language: only the phonetic component is different for them, not the underlying capacity.)

In his early work, that of Syntactic Structures and Aspects, Chomsky appeared to concentrate on production -- though he would say he was neutral between speech and hearing. The speaker converts an internal representation of their thought into a phonetic form. The speaker's internal knowledge of their language is called the grammar. It is a generative grammar because it contains rules for generating sentences. He analysed various formal systems by the strength of their generative power: this is called the Chomsky hierarchy.

A grammar that can describe all and only the actual sentences of the language is said to have descriptive adequacy. Languages are very, very complicated, so generative rules to encompass them are also very complicated. This raises the problem of how the listener understands: they have to take a transformed surface structure and undo the transformational rules to reveal the deep structure that succinctly conveys the thought. But the listener at least knows the language. A baby exposed to a particular language has to extract all its grammar from a continual blurred and fragmented wash of sound.

To tackle this problem of explanatory adequacy, by 1970 Chomsky was reducing the number of rules needed, by generalizing them. Instead of different rules for noun phrases and verb phrases and other kinds, the concept of an X phrase (XP) was abstracted. Instead of rules for moving words in questions and relative clauses and other structures, a generalized Move α rule subsumed them. Move α interacted with the boundaries and constituents of the XP. This programme was called X-bar theory, after the notation for a constituent of XP higher than X, and reached its peak in Lectures on Government and Binding (1981).

From the lectures arose the name Government and Binding Theory, a name Chomsky rejects, as GB is only one module and is not specific to his theories. This module describes how words interact across the sentence; θ-role theory (θ for theme) describes what words you can choose at the deep level (D-Structure); and Case theory describes what their final form will be at the surface level (S-Structure). Continuing to abstract, Chomsky was seeking general grammatical mechanisms that could handle any particular language. As much as possible, grammar was being moved out of the individual language. The less grammar a child has to learn, the better for explanatory adequacy.

This is Principles and Parameters theory (P&P), probably the biggest revolution of them all. As much as possible, the rules of grammar are located in Universal Grammar (UG), the common (genetic?) inheritance of us all. The grammatical possibilities of individual languages are choices from a set: for example, adjectives can precede or follow nouns. The principles are the same, and the parameter is what sets one language apart from another. A child growing up amid English picks up the English setting for that parameter on a category. The E-language is the external milieu; the I-language is the individual's internal development.

The lexicon of a language is its internal dictionary of words as speakers know them. They know that persuade needs two humans and an idea: Mary persuades Bill to do the dishes. They know how to pronounce persuade, and they know the logic of what happens out in the world if Mary succeeds in her persuasion. The lexicon needs to contain grammatical information: in this case the three θ-roles this verb has. There is no need to have a lot of specific syntactic rules about subjects, objects, and to-complements, if the lexical form of the word already insists on these roles being satisfied. Therefore this part of the grammar is removed from the language, and universal rules handle the coordination of roles.

The latest form of Chomsky's theories is no longer called a theory but a programme, and his most important recent book has that as its title: The Minimalist Program (1995). He is removing everything that is unnecessary. Over the years the generative and transformational rules that used to constitute a language's grammar have been abolished, as have levels of structure. The lexicon holds grammatical information specific to each word, plus the parameter values that make the language's grammar unique. UG is the faculty of the mind/brain (a term he uses to forestall argument about details) that mediates between the rest of our experience and the actual uttered form of the language. There is LF (logical form) and PF (phonetic form), and UG is the universal grammar that connects them.

Many thanks to Professor Smith for nit-picking.