Friday, October 21, 2011

Objects, Description, and Objectification

It has just occurred to me that Latour’s interest in description in his domain—“No scholar should find humiliating the task of description“—and my interest in description in mine stem from the same root, the need for objectification. Description is a way of providing objects for one to think about. Mathematics can do this as well, that is, whatever mathematics may be to the pure mathematician, other thinkers use is as tool for description.

The trouble with language is that it can nominalize anything, but that does not mean that conceptually useful objects are associated with every noun or noun phrase. It all depends: Just what do you want to do with the concept? Ordinary language has useful tools for dealing with macro-scale physical objects, from, say, grains of sand, hairs, and aphids to mountains, oceans, and planets. We can name them and characterize them in many ways, if crudely.

But what of the state, by which I mean a socio-political object? Surely states exist: Monaco, Norway, Sri Lanka, Senegal, Brazil, and so forth. But what are they? They are abstract objects, no? Sociologists, anthropologists, political scientists, and philosophers have devoted considerable effort to the objectifying the state.

And then we have, for example, memes, those cultural objects named by Richard Dawkins. The term has become common, and there’s been a fair amount of pop science speculation about memes in cultural evolution. But I don’t know that anyone has provided a conceptually useful and widely accepted account of memes such that one can identify and describe individual memes and the processes in which they participate. That’s the basic substance of my critique of Dawkinsian memetics: no useful descriptive tools.

And then we have my old hobby horse, salt and sodium chloride. Here we have a substance that is objectified in two conceptual domains: 1) ordinary language, where it is a white granular substance that tastes salty, and 2) chemistry, where it is a compound consisting of one sodium and one chlorine atom bound in a certain way. The sciences have given us many re-descriptions of ordinary objects. Think of the difference between the more or less flat earth or ordinary experience and the round earth of astronomy and long-distance navigation. Or of the fact that the shock of an electric eel and the flash of a lightening bolt are much the same thing—electron flow—on vastly different scales.

Then we have language. How do we objectify it? When he named the metalingual function as one of the basic functions of language, Roman Jakobson in effect asserted that language contains the seeds of its own objectification, of its own description. But only the seeds. It is one thing to be able to point out, in a general way, this or that utterance: “I can’t hear you”, meaning I can’t hear what you’ve said.

But what about syntax? How does one describe that? Pāṇini’s 4th century BCE description of Sanskrit is the oldest such description we have. Ever since then philosophers and linguists have been offering different ways of characterizing syntax. The most recent wave of such descriptions is conventionally attributed to Noam Chomsky, who characterized syntax analogy to mathematical proof. A sentence was said to be grammatical if and only if you could derive its structure from the basic rules of grammar. Those rules then constitute a description of the language’s syntax.

Chomsky’s project splintered into many. And alternative approaches of formalization have arisen. Thus there is no formal approach to syntax on which all linguists agree, and there are some who, I’m sure, would deny that formalism is the way to go.

My point here is simply that, while modern linguistics offers a variety of different approaches to the objectification of syntax, there is no reason to believe than any of them is the objective truth about syntax. However it is that one would determine the objective truth of a proposed approach to syntax, such a determination has not happened. Objective truth presupposes objectification, but objectification does not guarantee objective truth.

And then there’s meaning. How does one objectify the meaning of words, phrases, sentences, dialogues, and so forth? One cannot do that by describing the symbols alone. Yes, one must describe the symbols, but one must also characterize the system that uses the symbols, and the world in which they’re used.

As far as I can tell, that is the lesson of artificial intelligence and computational linguistics. That is a hard lesson to learn, very hard. I doubt that we’ll stop learning it in the foreseeable future, if ever.

No comments:

Post a Comment