The Minimalist Program (Current Studies in Linguistics)

Chomsky's theories are complex, but so is the subject matter he deals with. He takes simple sentences of English (and, more importantly, sentences which are NOT part of English) as empirical evidence w.r.t. how the computation of human language works. It is well worth any scholar's (or amateur's) time to criticize his theorizing if he or she thinks a criticism is in order. But there is a right and a wrong way to do this, and unfortunately, people often do it the wrong way. The right way involves dealing with the subject in all of its complexity, and takes far more intellectual energy.

To take an example of the kind of work Chomsky tries to do, consider these sentences (an asterisk indicates a sentence that no native speaker of English would accept as part of his language).

1a) It seems that John is tired.
1b) *It seems John to be tired.
2a) John seems to be tired.
2b) *John seems that is tired.

As speakers of a language, we combine words into sentences in various ways. But there is no a priori reason why (1b) and (2b) shouldn't be acceptable parts of our language. Similar distinctions are pervasive throughout all human languages. The goal of Chomsky's theory is to figure out why the ill-formed sentences are not part of our language. That is, how does our brain combine words to adequately and automatically pair sound to meaning, and why are some ways acceptable and others not?

Keep in mind, any reasonably intelligent person can come up with simple or complex ad hoc rules to account for the distinctions in (1) and (2) above. I could, for example, say something along the lines of:

3) A finite verb needs a subject, and a noun phrase must be the subject of a finite verb.

Ignoring the enormous number of difficulties a statement like (3) would face, there is nothing in (3) which explains why it, and not some other rule (such as its converse) should be the appropriate way to combine words. Actually, Chomsky's earlier works focused on doing just this: figuring out what kinds of rules adequately describe language.

In more recent works, however, such as The Minimalist Program (MP), the goal is far more ambitious: to find the fewest and simplest set of operations which can derive the structures of human language. In MP, roughly, these rules are basically Select (choose some words), Merge (combine them), and Delete (remove anything not needed for understanding the sentence). What is often seen as Chomsky trying to "fix" his theories is really his trying to simplify and simplify the generalizations he (and other researchers) comes up with.

But language is complicated, and that is an empirical fact. It shouldn't be surprising that the computation the brain uses to produce and understand language is a complex bit of machinery. The Minimalist Program is staggeringly complicated, but in it, Chomsky is able to show just how far a few simple combinatory processes can go.

The right way to criticize Chomsky is to come up with viable alternatives to the problems he presents. There are scholars who do this from many fronts, roughly categorizable into two groups: the generativists and the functionalists. The generativists share Chomsky's belief that the brain is equipped with a specific mechanism for producing and and understanding language. The non-Chomsky generativists simply disagree w.r.t. how that mechanism works. Frameworks under this category include Head-Driven Phrase Structure Grammar (HPSG), Lexical-Functional Grammar (LFG), and Tree-Adjoining Grammar (TAG). There are others. The second group feels that language is part of a more general cognitive mechanism. They have developed frameworks such as Discourse-Functional Grammar, Cognitive Grammar, and Construction Grammar.

While most scholars who have to take a side on the matter feel very strongly that their position is the closest to answering the most important questions, few would claim that there is anything fundamentally wrong with inquiring into any framework. Making various assumptions and testing differing hypotheses is bound to contribute to our knowledge of how language works. More often than not, different theories end up helping each other and/or converging on similar conclusions.

If you care about the matter, though, than it is important to consider what it is that you want a theory of language to do. We understand very little about how the brain works, so, at this point, Chomksyan theory is not the place to turn if you want to find a way to, say, learn a second language more quickly. But no one (except some unfair critics) ever said that Chomsky's theories were supposed to be able to do that. They are theories of the mind and how it works, not pedagogical theories.

One critic below claims that MP only accounts for 63% of English syntax. Ignoring the problem of how he could have possibly come up with such a number, one must be wary of anyone who claims have all the answers: no one does. There have been fantastically thorough descriptions of language, but no serious scholar would claim to have figured out everything or even most things. For those who actually read the Minimalist Program, they will find it abound with Chomsky's caveats: he claims over and over again that he does not have all the answers, that the details of language are not completely-understood, and that there is still a lot of work to be done. That being said, the minimalists do often have less than satisfying solutions to even the simplest constructions. But the careful reader will notice that this is not because they are foolish or ignorant, but because they are giving themselves a lot less machinery to work with. This goes back to my earlier point: anyone can describe a problem or come up with various solutions, but it is a far more difficult thing to come up with a simple architecture which describes the complex details of language in general terms. The complication of the MP is that they start with the bare necessity of conception (we interpret language mentally, we speak language physically, and we combine a finite set of words into an infinite number of sentences), and try to strip away as many ad hoc rule systems as possible.

To address some of the specific criticisms listed below:
1. "Chomsky's theories don't hold up in computers"
To what extant? Chomsky's phrase structure has been mostly successful in computer applications, and many computational linguists have used his ideas to create parsers of all kinds. The fact that they don't hold up in any sort "direct transfer" IS due to the fact that we have yet to program a human brain. We are also unable to make machines that see the way we do or hear the way we do. The critics have to concede this point: we have not been able to simulate or program a human brain. Not the Chomskyans, not the critics. The argument that a theory of the mind must hold in artificial computers crucially assumes that that we understand how to program computers as sophisticated as our brain. But we don't. When we come up with a computer that can learn language by watching and listening to human beings (like a child), a similar argument would hold a lot more weight.

2. "Chomsky's theories have proven useless to computational linguistics and artificial intelligence, and thus have implications for nothing."
Ignoring the validity of the premise, who ever claimed that science's goal was to facilitate immediate technology. In this case, Chomsky has never claimed that all the answers have been found, so how exactly is it supposed to have created "magical machines" by now? The implications are supposed to be for learnability, Plato's problem of acquisition (see e.g. Chomsky's "Language and Problems of Knowledge," which discusses Plato's view that language was too complex for children to learn, so they must be born with linguistic knowledge; Plato concluded something along the lines of reincarnation, but the point is the same), and the biology of human beings that allows us to invariably learn the language of our environment without really trying while no other animal, no matter how hard it tries, can quite break the code (even signing apes are only doing something on par with what we (humans) do with a phrase book in a foreign country). Claiming a theory (ANY theory) to be moot because of its lack of technological implementability is on par with criticizing Newton's theory of gravity because he failed to build a spaceship.

3. "None of Chomsky's theories have a scientifically verifiable basis"
Except that they do. For example, he might claim that if X is true, no human language should permit some construction Y. This is done over and over, for example on page 288 where he claims that expletives failing to force agreement on a finite verb should co-occur with an expletive associate which DOES force agreement. In English, for example, the sentence in (4) below is the evidence:

4a) There seem to be some mice in here.
4b) *There seems to be some mice in here.
4c) There seems to be a mouse in here.

(4a) and (4c) show agreement with the associates some mice, a mouse and not with the expletive there. (4b) is ill-formed because seems is not the third-person plural form of the verb, but the associate is third-person plural. If Chomsky's claim is right, linguists with knowledge of various language would need to check to see how the analogues of the constructions of (4) work in their language. If his prediction turns out to be false (e.g. some language which normally has verbal agreement has expletives and no agreement with the associate), the theory is updated to account for the new data. This is, you'll observe, the scientific method; and no theory is ever proven more than tentatively. That is, science dictates that every theory is subject to revision if new data calls it into question. Critics who make remarks such as these are presumably referring to technological verification, but technology is a concept which should be kept distinct from science. I can create technology (e.g. a bucket of water) without ever doing science, and I can do science (e.g. dropping objects of various weights to see if weight affects the speed the object falls) without ever creating any technology. "Verification" in generative linguistics amounts to seeing how generalizations hold across different languages/dialects/constructions.

4. "His grammars are meaningless to normal people, and thus entirely useless."
This, of course, assumes that the benchmark of an adequate theory is its "meaning" to normal people (whoever those people might be). Pedagogical research is important for its own purposes and perhaps more widely applicable to more concrete problems, but beware of criticisms which amount to: "Einstein's theory of space-time is useless. He has yet to build us a time machine or slow down time or let us travel faster than light or through worm holes. Further, his physics haven't done the least bit of good for athletes or kept us from dying when we fall off buildings. His theory is as abstract as it is useless." Keep science and application distinct.

If you want to criticize Chomsky on his own level, start reading some of the above theories (or one of many others I did not name) or even make one of your own, joining the league of brilliant scholars who do more than name call. If you want to work on improving Chomsky's theories, join the league of brilliant scholars who are employed in this difficult task. If you choose to remain on the sidelines or be an amateur enthusiast, remember to keep science and technology separate, and to evaluate theories with regards to what they are actually supposed to account for. Otherwise, join the group of anti-Newtonists who said that the apple fell to the ground because that was its proper place.

This is a very interesting text. Chomsky, asks the following question in his "Minimalist Program": Suppose that we do have a basic language faculty (inherent in all people and by nature--very creative mental faculty), then what can we say about this system? How well designed is it? Can we ask deeper questions about the mind? What is the fundamental nature of human intellectual capacities? Although these of questions might be a bit premature to even pose, Chomsky gives some fascinating responses to these questions.
This book, not only summarizes some previous work, but it also breaks new ground as well. I would suggest it to anyone who is interested in the philosophical and linguistic questions of the mind and language. The only thing one should remember is that the text will be incomprehensible if he/she doesn't have a very strong background in linguistics (esp Chap 4!). If not, there are prep books that will help you understand the basic underlying concepts in the Minimalist Program. Also, Chomsky gives a lot of interviews where he explains, in lay terms, what his work is dealing with.

Lastly, regarding the other reviews, please stick to criticizing the book, and not the author. If one disagrees with what's being said, then refute the argument in an intelligble manner. Don't resort to childish name-calling, bashing and other foolish activities.

Minimalism, like GB and Perameters, is yet another figment of Chomsky's imagination - much like his idea that McDonald's, Israel, and the American government are out to poison him.

Chomsky's theories are to Linguistics what Creationism is to Real Science. You start with your conclusion ("language is binary!" or "grammar is universal!") and then you come up with arguments that will support your conclusion while at the same time ignoring the enormity of evidence that refutes it.

On a lighter note, Chomsky's theories have consistantly failed to contribute anything viable to linguistics, technology, or humankind for the last 40 years. In this, his 6th or 7th infallible theory, Chomsky has decided to downsize his program in the hope that it will all somehow miraculously become correct.

Although Chomsky calls his program "minimalist", it is anything but. Chomsky's Minimalism is a hodgepodge mixture of about 119 individual theorums which bare no thread of coherence, and yet still only manage to describe about 63% of English. In spite of this, he still claims his model of Syntax is "universal" and applies to all languages. But let's face the facts.

- Chomsky's Syntactic models don't hold up in computers. Not in 1960s computers, not in today's computers. Blame it on semantics, blame it on parole, results are inevitably the same: crash, crash, crash.

- Chomsky's syntactic structures crash because they don't exist. An arbitrary idea rooted in a concept that doesn't exist can't work. Because it doesn't exist. Chomsky can blame this on "untestability" all he wants, but the fact remains that every theory he's come up with has proved completely useless to both computational linguistics and artificial intelligence, and thus has implications for nothing.

- None of Chomsky's theories have a scientifically verifiable basis either, so he's never actually "discovered" anything. He's simply made up imaginary processes, designated them with terms he finds pleasing (e.g. "percolation", "pied-piping") and passed them off as linguistic truths. In the process, all he's done is invent a fantasy world full of abstract and 'conveniently opaque' terminology that allows Mitniks to pose as specialists in this or that field of his imaginary science.

- That being said, none of Chomsky's programs are the least bit useful in teaching or acquiring language. His grammars are meaningless to normal people, and thus entirely useless.

In conclusion, given the fact that language can't be stuffed into arbitrary mathematic formulas just because Anglocentric hippies prefer to be thought of as 'scientists', and given the fact that genuine Syntax is obviously language specific, and given the fact that genuine Syntax could be explained in as short and simple a fashion as basic Morphology were it not for this deep-seated Chomskian bureaucracy that has enveloped modern Linguistics to the point where the field has become virtually useless for any practical application, all I can say about Chomsky's Minimalist Program is: what filth. Here's to the 46th year of new ideas being rejected in maintenance of ignorance.

Oh, we're sorry. This book is the absolute last word in linguistics and its relation to the neuro-computational-cognitive sciences. Until Chomsky writes the Extended Last Word in Linguistics, a few years down the line.

Though Chomsky's writing style tends to be opaque, and takes work to understand from even the best linguists, this book should be a required text for every theoretical linguist. Minimalism has become a major force in Syntax since the publication of this book, and little can be done in the field without at least knowing about the theory. Even proponents of LFG and GPSG, for example, should read it so they are conversant with competing theories.

One caveat; beginning linguists will want to read up on transformational grammar, Government and Binding theory and Principles and Parameters before tackling this text. There is a very straightforward introductory textbook about Minimalism by Andrew Radford. Also, look for more recent developments of the program (Minimalist Inquiries, Derivation by Phase etc.)