I've been reading Stanley Fish's recent booklet "How to Write a Sentence: And How To Read One", seduced by passages like this one on page 2:
… just piling up words, one after the other, won't do much of anything until something else has been added. That something is named quite precisely by Anthony Burgess in this sentence from his novel Enderby Outside (1968):
And the words slide into the slots ordained by syntax, and glitter as with atmospheric dust with those impurities which we call meaning.
Before the words slide into their slots, they are just discrete items, pointing everywhere and nowhere. Once the words are nestled in the placed "ordained" for them — "ordained" is a wonderful word that points to the inexorable logic of syntactic structures — they are tied by ligatures of relationships to one another. They are subjects or objects or actions or descriptives or indications of manner, and as such they combine into a statement about the world, that is, into a meaning that one can contemplate, admire, reject, or refine.
But the following 158 pages left me disappointed. Prof. Fish gives us plenty of rhapsodizing about how great syntax is, and plenty of oohing and aahing about how great specific sentences are:
A sentence is, in John Donne's words, "A little world made cunningly." (Donne is speaking of the human body, but that is just another composition.) I want to bring you into the little worlds made cunningly by as many writers as I can cram into a short book.
This is a laudable ambition; but in his tour of great sentences, there's almost no syntactic analysis — and neither is there any careful analysis at any other level of linguistic structure. Nor is there any advice to the reader about where or how to learn more about the structure and function of these "little world[s] made cunningly", other than his recommendation of Graff and Birkenstein's practical compendium of rhetorical templates They say/I say.
This is probably just as well, because what little linguistic analysis Prof. Fish gives us is full of assumptions from old-fashioned grade-school grammar, about whose inadequacies he's curiously incurious. If he were writing about about the body, he'd be enthusing about the precarious balance of the four humours; but since he's writing about syntax, his central assumption — obviously false and curiously unexamined — is that sentences are all about agents, actions, and (optionally) things acted upon:
It is important to understand that the relationships that form the sinews and relays of sentences are limited. There is the person or thing performing an action, there is the action being performed, and there is the recipient or object of the action. That's the basic logical structure of many sentences: X does Y to Z. (Sentences can also come without objects, as in "Joe walks.") [...] The instances are infinite, although the form remains the same (this is a key point, and I shall return to it): doer, doing, done to.
He repeats this over and over again in different places and in slightly different ways:
Let me say again that by "forms" I do not mean parts of speech or any other bit of abstract machinery. I mean structures of logic and rhetoric within which and by means of which meanings — lots of them — can be generated. The logical structures are the ones we have already met: the structure of relationships between actors, actions, and the objects acted upon.
A sentence, I declared, is a structure of logical relationships; the relationships are finite and learnable; the contents that can find expression in the structures formed by the relationships are infinite and incapable of being catalogued. So it follows, I argued, that content will be a distraction and that the skill of writing well-formed, clear, and tightly organized sentences will be acquired by focusing on forms. I explained that what I meant by "forms" is not the list of parts of speech or kinds of clauses or grammatical errors found in many textbooks, but the logical forms that link actor, action, and the object of action in a way that make available simple and complicated predications.
And this is not just mechanical repetition of something he was taught in the fifth grade. When Prof. Fish attempts a bit of linguistic analysis that deals with individual clauses, those hypothetical actions materialize in his description, one per clause, even if the idea makes no sense in relation to the examples analyzed:
Although the forms of argument are more numerous than the forms of sentence structure, their number is limited; they can be catalogued (Graff and Birkenstein do it), they can be added to your argumentative repertoire, and you can use them not simply to arrange thoughts but also to create thoughts. [...]
There is no limit to the forms you can practice this way: "Even though," "Were I to," "Notwithstanding that," "Depending on whether," "In the event that." Each of these forms exists to make available a certain content, and apart from them, that content could not be briskly produced. An "Even though" sentence – "Even though I was exhausted, I watched another episode of Law & Order" — makes a complicated statement: it specifies two actions that follow one another but, in the usual course of things, shouldn't (if I'm exhausted I should turn off the TV), and it signals awareness of the oddness of the sequence even before we know exactly what its specifics are. (emphasis added)
It may very well be useful for students to practice writing the specific kind of "Even though" construction that "specifies two actions that follow one another but, in the usual course of things, shouldn't". However, it's easy to discover that few real-world "Even though" sentences meet this condition — all you need to do is to look at Fish's own example. In the phrase "I was exhausted", what is the action? Who is the actor, and who or what is the "object acted upon"?
This embarrassing oversight was almost unavoidable, because it's relatively rare for an "Even though" construction to involve a relationship between two actions in a pragmatically-incongruous temporal sequence. A typical real-world "Even though" sentence is this one from the current NYT:
Even though Wells Fargo is the nation's biggest lender to consumers, its losses on bad loans have been lower than the rest of the industry.
A statement about Wells Fargo's rank in consumer lending does not denote an action, and neither does one about the size of its losses. And the relationship between the two propositions, in this case, does not involve any obvious temporal sequence.
In other examples, the inter-clause incongruity may be more subtle — again from the current NYT:
Experts have said that a million people died in earthquakes in the 20th century and that this century might see 10 times as many deaths, with as many as a million killed in a single quake. That is, unless major efforts are made to fortify the world’s growing cities, which are expected to be homes to billions of added residents.
Even though the rate of earthquakes over time seems to be more or less unchanging, the world's population explosion means that more people are moving into quake zones, which are often near coasts. The result, the experts say, is the prospect of continuing trauma.
Again, there's no sequence of actions, but there's also no simple pragmatic incongruity between the unchanging rate of earthquakes and the trend for people to move into quake zones. Rather, the point is that a predicted increase in quake-related deaths is not due to increased quake frequency, but rather to larger populations at risk: here the rhetorical pattern links three propositions, not two.
We can even find cases where the temporal sequence between the clauses is reversed — thus
Even though he will not speak to reporters now, Zimmerman did speak to one before the Broncos played the Vikings in 1994.
That's from the COCA corpus, which has 5,733 sentences starting with "Even though". A spot check of the first 100 hits suggests that at most a few percent involve a temporal sequence of two actions. It's hard to get a crisp and reliable count, because Prof. Fish gives no definition of his terms, as I guess is appropriate given his anti-foundationalism; but in the 100 hits I checked, there were only a handful of examples like the following, where it seems clear that there's a temporal-sequence incongruity between two things that we could plausibly call actions:
Even though hydrodynamics research has driven the successful steps in the first years, it is economics and manufacturing that will have the final say on bringing this technology to the world.
The book's other exercises in linguistic analysis exhibit the same sort of confident overgeneralizations, implicitly refuted by the very examples chosen to illustrate them. For example:
The subordinating style orders its components in relationships of causality (one event or state is caused by another), temporality (events and states are prior or subsequent to one another), and precedence (events and states are arranged in hierarchies of importance).
At least there are now states as well as events in Fish's ontology (unless "actions" are meant to include both states and events, contrary to the ordinary language meaning of the terms?), but his list of "causality …, temporality …, and precedence" is repeated several times, and really seems to be intended to be exhaustive. Despite this, his own chosen examples of the "subordinating style" include phrasal relationships of implication, (spatial) location, description, exemplification, concession, and so on.
It's certainly encouraging to see Prof. Fish, a leading light of post-modern literary studies, continuing to beat the drum for linguistic form. I just wish he'd undertake to learn something about it.