The One Model that Works

Massimo Piattelli-Palmarini is a firm defender of generative grammar, but reading his defense has made me even more confident of the notion that language is perception by other means.

A book review in the current issue of Biolinguistics (registration free but required) has persuaded me that only the speech-as-group-perception model (favored by this blog) can account for the evolution of speech. (See: What I’ve Learned About Language) That’s not to say the idea that “language is perception by other means” is the only view around, far from it. It is just that none of the many rival definitions of language proposed to account for its evolution have clear rebuttals to the objections proposed by orthodox generative-grammarians, and neither have the orthodox any way of accounting for the origins of their subject.

I have come to (or perhaps just been strengthened in) that opinion after reading Massimo Piattelli-Palmarini’s review of Eva Jablonka and Marion Lamb’s book, Evolution in Four Dimensions (reviewed on this blog some time ago; see Selection Without Genes). The review (available here) leaves me with two certainties:

• Generative grammarians are looking but still have no idea how to account for the natural origins of language as they describe it.

• The forceful objections generative grammarians can raise to other accounts of language do not hold for the perception theory. Indeed, they reinforce it.

The review is written with an impressive bluntness. For example, here is its the book’s claim that grammatical structures are designed to handle events and situations.

Sorry, but it’s just not so. … Grammatical structure is only sensitive to actor, patient (or theme, more generally) and in some cases the instrument or the modality of action. Period. … For everything else, we have to … use adjunctions, circumlocutions, add further separate sentences, develop a whole discourse, and so on. [p. 243; italics his]

To put it in properly intimidating terms: “The paucity of syntactic theta-roles, with respect to all the things we are interested in in our life, is one of the central observations in linguistics.” In other words, of all the interesting relations about events and situations only a very few can be expressed syntactically. Grammar let’s us say who did what (John led Mary into his house) but it is no help in saying, for example, why who did what.

We can say Because John lusted after Mary, he led her into his house, letting lusted after stand for the motivation even though the structure is the same as for an action. Motivation in this case becomes an action, and it takes a bit of subtle meditation to realize that the syntactical structure is a kind of syntactical metaphor for expressing the invisible. We could also say John lustily led Mary into his house, letting the adverb imply a motivation. But compare lustily with into. There is nothing implicit about the preposition; it states a spatial relationship. Of course it is possible to imagine a language that has a series of markers similar to prepositions that would be used to express a motivational relationship. A language could say John garnled Mary into his house, in which the garn- prefix has the same syntactic effect as saying led-motivated-by-lust and perhaps contrasts with grunled which has the syntactic effect of saying led-motivated-by-pride. But although we can imagine such a language, none are found in nature. Why not?

This kind of problem is more than merely a fantastical-nuisance, a bit of whimsy posed to be irritating. If, as a great many people say (and Piatelli-Palmarini singles out Michael Arbib, Terrence Deacon, Michael Tomasello), language is “a cultural invention that has become common knowledge,” then why do these same limitations crop up in all the cultures of the world? The constraints must be inherent to the system, not cultural inventions.

Most of those who support the cultural evolution of language argue that language is a symbolic system. Terrence Deacon is the leader of this movement, insisting that our capacity to think and communicate symbolically is the defining human characteristic, but Piatetelli-Palmarini will have none of it. He writes:

There are at least four major differences between words and all non-linguistic symbols: (A) aspectual reference, (B) headedness, (C) internal structure, and (D) edge features. [238-239]

Although these technical terms sound abstract and remote from experience, they identify qualities of perception.

Aspectual reference refers to a way of perceiving, a point of view. Perceptions are built into the expression. For example, a transaction in which money is exchanged for a bicycle can be described as buying a bike or selling a bike, depending on the speaker’s point of view. Computer languages and animal signals have no point of view, because their communication is not a form of mutual perception.

Headedness means that strings of words are about something. “The California highway commissioner report” is about a report. You can modify the phrase in many ways, deleting California, adding stupid, etcetera but even if you change report to study the phrase is about a document out there in the world. You cannot alter the phrase so that it looses its subject (its 'head') without destroying the phrase itself, just as a perception can never be so abstract that it loses its focus.

Internal structure refers to the requirement that words pilot attention somewhere. It is not enough for a word or phrase to have a syntactically correct form. Chomsky came up with the famous nonsense sentence, Colorless green ideas sleep furiously. Although it has proper structure, it is unintelligible. Piattelli-Palmarini would say the famous sentence ignores the words’ internal structure, which is true, but not as helpful as just saying the words leave a listener with nowhere to direct attention. We can refer to Scipio’s destruction of Carthage and imagine a specific event, but we are at a loss when it comes to imagining Scipio’s sleep of the bed, even though the two phrases have the same abstract structure.

Edge features refer to language’s ability to broaden a perspective. The cattle are lowing focuses on some animals. The cattle in the north meadow are lowing broadens our attention span. Only the cattle in the north meadow are lowing broadens the perspective still further, implying something about the activity of cattle in other areas.

Stated as abstractly as the generative tradition puts it, these properties may seem like arbitrary features of a diagrammatic concept, but if you look at the synonymous list of psychological properties—point of view, focus, directed attention, and perspective—it is plain that we are talking about perception.

Shared perception also explains the limitations on grammar that Piattelli-Palmarini noted and clarifies why no language conjugates verbs in the garnled manner. Perception, shared or personal, is limited “to actor, patient (or theme, more generally) and in some cases the instrument or the modality of action” [243].  We cannot perceive motivations or many other things that interest us. We can have opinions about those things, and find ways to talk about them if we care to, but those ways will be metaphorical or depend on narrative or poetry or wit. They will not be built into the syntax.

The review also makes it clear that generative-grammarians themselves have no deeper answers. It never explains anything; it only describes features and says they are inherent to the system. It offers no hint as to why language has the features it has rather than others. The perception model is different, saying the features are not arbitrary; they are the properties of any perceiving system.

A final argument for the perception model comes from the review’s own close. The review is, after all, about a book on evolution and the author’s frustration that a “classic, neo-Darwinian, functionalist explanation of the evolution of language” [241] cannot possibly account for the rise of the non-functional, arbitrary system he takes language to be. He concludes:

The prima facie appealing and almost irresistible hypothesis that the need to communicate has shaped the evolution of language is countered by the huge corpus of data collected in many languages and dialects [and summarized briefly in the review]. [244]

But the perception model has a different hypothesis: the need to cooperate has shaped the evolution of language and given us the capacity to perceive together. We are on the right track at last.

Comments

[Matthijs Westera]

I wish to congratulate you, both on this great blog and on this small triumph for the perception model, which I dare say is the most sensible, most elegant and most promising view on language I have encountered so far (not too many, I admit).

[JoseAngel]

An excellent way to turn the tables on generative grammar's abstract rationalism. Your view would seem to be much in line with Pinker's cognitive analysis of the structures of word combinations in THE STUFF OF THOUGHT. Is that the case?
--------------------
BLOGGER: As I recall, Pinker says that language starts with the abstract and moves to the concrete, so that would be the reverse of my view. But I'm away from my library just now and cannot take Pinker off the shelf for a peek.