Question pronouns are important but hard to visualize.
Yesterday I posted a notice about a review article by Heather K.J. van der Lely and Steven Pinker titled "The biological basis of language," that reports the brain supports two separate syntax networks which the authors call basic and extended syntax. I did not go into the grammatical analysis of that paper because it was irrelevant to my focus, which was on the multi-step evolutionary process implicit in their account. In this post, I do want to look at their grammatical analysis because it differs from the way I look at language. Can I accept the van der Lely-Pinker findings without accepting their thoughts about syntax?
In terms of generative grammar, the heretical idea espoused by the authors is the existence of a basic syntax, something that generates simple phrases. It is a finding more in keeping with the early Chomsky, but by now generativists have simplified the system so much that there is no room for a basic syntax. It is one grand system.
Basic syntax does fit perfectly into my view that language began as a way of sharing perceptions by drawing attention to them, either present or imaginary. In this system, words are concrete and phrases depend on side-by-side relations. The relations that are expressible are case relations, which are perceivable relations:
- man chase cat: we can perceive a man chasing, and a cat being chased so both agent and object relations predate language.
- man run home: we can also perceive running to home, so indirect objects also predate language.
- hurt man hand: we can perceive a hand as part of a man, so genitive relations also predate language.
By saying the relations predate language, I don't mean a languageless animal conceptualizes the relation but they can use the perceivable relation. I have seen dogs entertain themselves by scaring chickens. Conceptually, they are clueless; but perceptually they know they can act (agent relation) and the chickens will respond (object relation).
Similarly, first graders can use the basic case relations without having any formal knowledge of them. They can act upon these relations, and one way of teaching the meaning of the formal terms is to use concrete examples.
The Jump from Basic to Extended Syntax
Generative grammarians conceive of syntax as a language-only way of thinking. Doubters argue that syntax is part of a more general intelligence and that rules of syntax evolve, so that what is spoken becomes a rule. The paper's authors seem to take the generative position. Extended syntax is generative syntax.
Consider this sentence: In the quest for office, candidates chase donors around the clock. The nouns–office, candidates, donors, clock—are all abstract or metaphorical. Picturing an office or a clock will only cause confusion. Candidates and donors are both generic terms inspiring generic images. If you take the phrase 'around the clock' literally you will imagine a fixed place instead of a changing time. So it seems pretty different from basic language, but another look suggests some possible links. The relation between candidates and donors matches the relationship expressed in man chase cat. Quest for office also expresses a perceptible relationship (indirect object). Then the prepositions—in and around—also suggested a concrete, spatial relationship. So we speak about a set of abstract notions but organized them as though they described something concrete. I tend to doubt that sentences using extended syntax have broken completely with the perceptual system that supports basic syntax.
However, the authors of the review article do accept the generative rules.
The first example of extended syntax used by the authors is Who did Joe see? Children with G-SLI have a tough time interpreting this sentence. According to generative theory, the construction of the sentence demands an operation called Move. In normal English sentences the structure is subject-verb-object (Joe saw Pete), but in this question the object has been moved to the front of the sentence. Movement is said to be a universal property of languages. Thus, the reason people with G-SLI have trouble with the sentence is because they cannot use the Move operator in generating or parsing sentences.
I confess that I have never liked the Move operation. I have an instinctive distaste for it, but I like to think I can be persuaded that my gut reaction is wrong. It seems to me I've spent my whole life changing my mind, but my distaste for Move is constant. Here are some better reasons for doubting:
- Nobody has ever offered a solid reason explaining why natural selection should change our biology to include Move. What survival depends on it?
- I doubt that it is really universal. By chance I am familiar with a Bantu language, Swahili, that does not use Move. The way to ask who did Joe see in Swahili is Joe amemwona nani? The last word in the sentence is the Swahili equivalent of who. The Swahili way of saying Joe saw Pete is b. The structure is the same in both Swahili sentences. (I see on the Internet that, probably as a result of long contact with English, these days nani is sometimes moved to the front.)
- There could be some other reason for standardizing putting Who at the start of a sentence. Rhetorically there is an advantage to putting who first. It alerts the listener immediately that a question is coming and stresses the point. A salesman might say, "Three! Three cars I sold today." The object has been moved to the front for reasons of emphasis, not out of syntactical requirements. (So why doesn't Swahili use the same rhetorical trick? It puts a prosodic stress on nani (but not on Pete) instead of using a locational switch).
Of course, these doubts raise the question of why people with G-SLI have problems with questions. I think a more likely explanation is in the blankness of the word who. It does not point to anything, it does not evoke anything. I notice the American Heritage Dictionary gives it a particularly useless definition, "What or which person: Who left?"
The vocabulary of basic syntax appears to be purely perceptual, associated with things we can perceive. Children with G-SLI seem to understand a sentence like, "Barbie bakes the cakes," and they can say that, "Barbie drinks the cakes," is wrong. But these sentences are entirely visualizable.
Some variations on these sentences go beyond the visualizable: (a) Barbie baked the cakes; (b) who baked the cakes; (c) Barbie loves the cakes.
These simple-seeming sentences all use extended syntax. Sentence (a) sets the event in the past, but looks exactly the same as a sentence set in the present. Sentence (b) requires no Move operation, but still uses a fill-in-the-blank reference, and (c) uses a subjective verb that evokes an emotional state rather than a visible one. Thus, even without the syntactical demands, the language-user must be able to get beyond basic perception.
I consider the distinction between basic and extended language to be a major find, something that I feel sure will be part of my thinking from this point on. It should not be surprising that I disagree with the authors over what underlies the difference. We disagree on how language works. The great thing about science is that even when questions of interpretation persist, progress can continue.