When some linguists hold out that there certainly must be a universal grammar to all human language, and cultural theorists stubbornly resist any kind of schematizing whatsoever, you know there's an opportunity for synthesis. The suggestion has been made that language is simply fundamental to humyns' way of understanding the world—for example, all children learn to speak, and at around the same age, without any deliberate instruction. But the effort to notate a universal structure for culture has a) stalled and b) come under fire as a fascistic attempt to pre-scribe and normalize human behavior.

Here's an unguent proposal. What if language is fundamental to our per-/con-ception, and yet there simply is no common universal language? What if each person derives a system of language, based on all the speech s/he hears, which "connects the dots" of that speech with a systematic framework? For common cases, co-native speakers would connect the dots the same way because of overwhelming trends in usage; but for fringe cases, two people might have a different construction for grammaticality. That would explain why NY Times author Stephen S. Pickering thinks this example sentence is incorrect, and why he needs to cobble up a round-about explanation for why it nonetheless makes sense.

The sentence is: "Toni Morrison's genius enables her to create novels that arise from and express the injustices African Americans have endured." The rule by which he declares the example to be incorrect is, I would argue, entirely ad-hoc and minted for the sole purpose of finding an error—like any other rule that grammarians might invent to justify their cavil. He repeats an argument that "her" must for some reason refer to "Toni Morrison's" (or is that a mistake for "Toni Morrison's genius," the subject of the sentence?). But why can't "her" simply refer to "Toni Morrison," the only human being referred by the sentence? This rule is just as good as the other, both conjured after the fact in order to justify a position pro or con toward the sentence. My point is not that language is formless, or structureless, but rather that there is no publicly-held rulebook for it. Each speaker's mind has its own structure for its language, but there may well be variation among perfectly fluent speakers.

In retrospect, this makes more sense than the Universal Grammar idea. The formal construct of "a grammar," may well be part of the human mind, and that may be what allows us to learn to talk so quickly. But since people of wildly different cultures construct wildly different "grammars" for themselves, why isn't it possible that two nearby speakers of the "same language" arrive at slightly different grammars for that "same" language?

In the literature of Computer Science there is a body of techniques known as "grammar induction" or "grammar inference" which can be used, given a body of examples of a language and a body of non-examples, to derive a grammar for that language. The program begins life agnostic about any particular linguistic forms, but terminates with a particular set of formal rules. Perhaps it is just such an apparatus—a universal grammar-maker and not a universal grammar—which exists in humyn minds.