One thing Noam Chomsky gets right is his emphasis on the sentence. With a newspaperman for a father, I've always placed emphasis on the paragraph, but I'm persuaded that Chomsky's focus is right. The sentence is what makes language more than the cooperative sharing of perceptions. As usual , it is nothing Chomsky said that brings me to his side. I've been reading Stanley Fish's new gem, How to Write a Sentence and how to read one. He makes the case that sounds right to a person kuje ne like me who loves language and its skillful use.
The first chapter is titled, "Why the Sentence?" and talks about it as the "medium" of writing and speaking. Fish sees the sentence as the brick that lets you say what you have to say.
If you can write a sentence in which actors, actions, and objects are related to one another in time, space, mood, desires, fears, causes and effects, and if your specification is delineated with a precision that communicates itself to your intended reader, you can, by extrapolation and expansion, write anything: a paragraph, an argument, a treatise, a novel. [pp. 7-8]
He rejects the notion that words are the basic brick because "just piling up words, one after the other, won't do much of anything until something else has been added" .
That something else is form. Fish is a champion of teaching writing by teaching form. His maxim is, "A discipline in form is a discipline in thought" . Also, "You shall tie yourself to forms and the forms shall set you free" . Fish's formalism, however, is quite remote from Chomsky's syntactical focus. Fish is interested in pragmatic forms that enable users to say what they want to say. Language, Fish says, organizes "the world into manageable, and in some sense artificial, units that can then be inhabited and manipulated" . He rejects syntactical formalism as "merely taxonomic. It explains nothing" . His argument is that sentences organize things by assigning them logical relationships.
If I thought Fish was using "logical" correctly, I would wail that his definition is wrong, but I believe that since he has merely chosen the wrong word his argument is garbled but not mistaken.
If I were the author of Fish's fine book, I would have said that sentences organize things by assigning them perceptual relationships. And apparently Fish agrees with me on the point that language is cooperative perception. He says something well that I have said poorly many times on this blog:
Language is not a handmaiden to perception; it is perception; it gives shape and form to what would otherwise be inert and dead. 
The importance of the logical/perceptual distinction is that logical relationships do not depend on a point of view while perceptual ones do. Fish knows better than most that with language there is no escaping point of view, so frankly I'm a little shocked to see him use the cliché about logic. It goes to show that even the most meticulous thinkers cannot fully escape the careless usages of their time. The slip makes it a bit harder for his readers to understand how verbal thought is unlike mathematical or procedural thought. I know many logical thinkers who are terrible writers and many good writers who are poor on logic. How, given Fish's sloppy definition, can that be?
And by talking about logic, Fish doesn't make clear enough just what it is that distinguishes a sentence from an equation. Sentences link a topic (the subject) and news (the predicate) by an action; equations can be exciting and beuatiful while remaining static. Fish does mention this connection-by-action point, "a verb shows up, providing a way of linking …." , but he does not sell it very hard. And it is crucial. I often edit the work of logical thinkers and the point that is hardest for them to get is that they should be linking things, not listing them.
I'm being more prickly than I feel, probably driven that way by the difference between what Fish has to say and what this blog is about. Fish is describing the writing (and reading) of sentences as a means of teaching would-be writers how to improve their style. I'm interested in how speech began. We intersect in our consideration of just what language is, and then point in different directions. Fish wants to show how to put together complex, written sentences.
One point that hangs over Fiish's whole project is a direct challenge to the evolutionary view of language. It is plain from Fish's discussion of his experience teaching college students how to write that composing sharp sentences does not come naturally. He repeatedly refers to practicing writing sentences just as a child learning to play an instrument practices hour upon hour at playing scales.
How are we to reconcile this with the sort of common dogma that appears in, say, Tecumseh Fitch's The Evolution of Language: "Each normal human being is born with a capacity to rapidly and unerringly acquire their mother tongue, with little explicit teaching or coaching" ?
We either have to say that Fish is wrong and that training in how to write sentences is the equivalent of being taught how to digest food, or that Fitch is, at a minimum, overstating his case.
I go with that minimalist argument, and I think that's why Fish can get away with not pounding more energetically on the basics of a sentence. That much we do learn as part of becoming old enough to trot off to kindergarten. Fish proposes an entertaining exercise in understanding how sentences work. He quotes the opening of Lewis Carroll's nonsense rhyme, "Jabberwocky," and says, "Now replace the nonsense words with good English words in a way that leads to a meaningful sequence," . Fish reports that his students can do this because:
If you are a speaker of English, you know—although it is the kind of knowledge you may have never have articulated—that there are only certain classes of words that follow… You know … that a "linking verb" such as "was" introduces a state or a condition or a location… Making these substitutions is the easy part. Explaining how you know how to do it is harder … [27-28]
How do you know those things without having had them articulated? As Fitch says, they've picked it up without much direct coaching or training.
A second point is that Fish does not teach rules. "Examples," he says, "not rules, are what learning to write requires" . His method is to provide an example of a sentence, such as this zinger, "Interior decorating is a rock-hard science compared to psychology practiced by amateurs,"  and encourage people to imitate the form by replacing the terms while retaining its structure. To do that even poorly—e.g., rap singing is high tone church music compared to Handel's Messiah performed by lawyers—you have to recognize the functional role of the words being replaced and imagine alternatives that match the function.
Fish's method is the Socratic one of drawing out from the students a conscious statement of what those functional roles are and how they work. This kind of teaching is very high level, requiring understanding and insight made possible by our evolutionary history.