First off, Stanley Fish (apart from the man himself) is a very unfortunate name, and this is coming from someone whose last name is, among other things, an implement for catching this man's surname. That aside, what to make of his book?
Backing up from my subjectivity inasmuch as is possible, let's look at the author's objectives. They are stated on page eight: ". . . I promise to give you both sentence pleasure and sentence craft, the ability to appreciate a good sentence and the ability to fashion one." (My emphasis.) Well, that's quite something — a promise to grant the reader certain abilities. That's a big task for any author, especially for one concerning himself with literature and (non-fundamental) writing. Some might say that it smacks a bit of hubris. It certainly struck me that way. I will say in advance that I think the thesis/aim of this book is a great one, and one very worthy of attention; I will also say that I enjoyed a great many things about Fish's methodology, strategy, and prose. But, almost everything comes down to his promise to readers: to make us both able to appreciate and create good sentences. So how does Fish go about this?
Roughly speaking, Fish's foremost precept is this, which is first found on page eleven: "Sentence craft equals sentence comprehension equals sentence appreciation." While one could take issue with the pithy word "equals", let's just take it at face value. If I can craft a sentence, I should be able to comprehend it; and if I comprehend it, I can therefore appreciate it. We have to remember that his audience is not people learning a language, but those that already know one, and who are trying to be better at it; with that in mind, I think his formulation a good one. Understanding sentences is obviously necessary for their appreciation, though one could always argue that even someone that cannot write can at least walk the borders of appreciation (even though their appreciation will be significantly less than someone who has learned proper sentence-craft). So far, so good.
After his first chapter, he then provides a few definitions, so that reader and author can come to terms. The first one is critical: what is a sentence? (Answer: it is: 1) An organization of items in the world; and 2) A structure of logical relationships.) Its basic formulation in English (and in other SVO languages) is constructed around the relationships of "doer/doing/done-to"; all the rest is additive (and sometimes lovely) fluff. Fish decries most English-learning and grammar-based programs, believing (like me) that knowing rote forms (e.g. parts of speech, what an adpositional phrase is) gives one only the appearance of knowledge, and not any actual ability to read or write. So, he gives his one rule of writing, and his cardinal error. The one rule is to make sure one's sentence-components relate properly (e.g. proper conjugation, singular/plural use), and the one error is that one's sentences don't turn out to be relationships but just random strings of words. In short, make sure your sentences have meaning, and don't just break down into strings of words. I don't think anything in this chapter is particularly controversial.
So how does one learn proper sentence-craft? Form, form, form! He's got this excellent quote on page 27: "When it comes to formulating a proposition, form comes first; forms are generative not of specific meanings, but of the very possibility of meaning." Of course you must have something meaningful to say to your audience, but without forms, you can't say anything. You'd just be mumbling to the void. To speak via metaphor: just as a pianist practices scales in order to play grander things, so too the writer must practice set forms in order to be ready to use them when the need arises spontaneously. One must be ready for the sentences one needs, and this requires both practice and conscious analysis. Ultimately, this prepares you for the purpose of writing: expressing content. People "write sentences in order to produce an effect, and the success of a sentence is measured by the degree to which . . . [it's] achieved." This only makes sense. If we have something truly worth saying, we need to know to whom we're speaking, as well as the best way to express it. That almost (almost) goes without saying. Language is an ordered reality, and Fish views writers as "selectors" who choose words in service to the expression of their content. The goal is not that the sentence contain everything that can possibly be said, but simply that it contains those things necessary to an author's purpose. And, from there, we can judge the sentence based on its efficacy.
So, we have literally infinite content in the world, though we, through our perceptions and sense-making apparatus, have a distinct set of content-directives to express; with our individuality online, we need to then decide how to say what it is we want to say. So, how many types of sentences are there that we need to learn? Well, there are many structures of sentences (e.g. "Even though . . ." // "But for the _________, . . ."), and we can and should practice these, but, in short, there are two primary types of sentences: the additive and the subordinating. Fish starts with the logical choice, being the subordinating sentence, in which things are given order in terms of causality, temporality, and precedence. Basically, these sentences give the effect of everything being in its right place; each word has been chosen carefully for its intended effect, and all is right in the world.
And then there's the additive, which is one that flummoxes me to no end. Read this gem from Gertrude Stein:
"When I first began writing I felt that writing should go on I still do feel that it should go on but when I first began writing I was completely possessed by the necessity that writing should go on and if writing should go on what had colons and semi-colons to do with it, what had commas to do with it what had periods to do with it what had small letters and capitals to do with it to do with writing going on which was at the time the most profound need I had in connection with writing."
Uf. Well, that's the extreme end of the additive style, which is clearly a style of greater freedom (one is reminded here of Frost's "playing tennis without the net"), informality, flow, stream of consciousness, and connectivity. This style has beautiful and strange effects on a reader, and anyone who has read Woolf knows what's up. Were I to continue the rest of my review which I don't intend to do but would find quite lovely and fun and jaunty I should do the rest of it that is my review in this additive style à la Stein but I keep finding myself wanting to add just a little comma here or a semicolon there because they have everything to do with it, that thing being writing and so I shall stop but only after I have come to the end of my rope with this beast let loose in my head and finally I hope you've enjoyed it.
Something interesting that Fish writes about is the meta-"content" of each style; that is, both of these sentences "say" something, even without regard to content. The subordinating style implies order, foresight, and conceptualization, and the additive style brings to mind carefree whimsicality, wind-in-the-trees nonsense, and a confusing looseness. This is true almost before the content is added. Yet, neither sentence type is better than another (this depends on purpose); nor is either more "pure". Both are, at heart, different ways of organizing experience. We all have our preferences, and while it may not behoove you to write your philosophical treatise like Stein, it may be equally unfitting to describe a dream like an argument from the Summa Theologica.
And then, to me, the book goes completely off the rails. Fish drifts into sentence-appreciation, and honestly just ends up picking apart sentences willy-nilly for the last chapters. He doesn't actually tell us how to read a sentence (nor how this would relate to appreciation explicitly); there is almost no guidance in this section of the book (though there wasn't much in the writing part, either), which calls back to mind his promise to readers. Let's have it again. "I promise to give you both sentence pleasure and sentence craft, the ability to appreciate a good sentence and the ability to fashion one." Well, he certainly gave the reader plenty to enjoy and learn in terms of sentence pleasure and craft, but about the abilities promised, I'm not quite convinced of his efficacy as a teacher. If someone completely inexperienced in writing (though functionally literate) picked up this book, for sure the book would fail; there are only three or four listed exercises for practice, and few criteria other than "make sure your sentence's relationships are solid". And, in terms of reading sentences for pleasure and appreciation, he never actually states how this is done. He just jumps from sentence to sentence, randomly picking them apart, but never giving anything larger in alignment with his objectives. One can read these last two chapters many times and not come away with an ability to appreciate even the finest sentence, or without having gained the ability to say why they appreciate something (which seems like an attendant good). (And how can one promise to increase another's appreciation, anyway? That's a monumental task, truly.)
Verdict: great book for sentence theory, general linguistic fun, and an introduction into some damn good sentences; non-existent pedagogically, and little help for the would-be connoisseur of sentences. A very mixed bag, and ultimately confusing (and confused) as to the book's proper audience.