In lieu of an abstract, here is a brief excerpt of the content:

9 The Social Ecology of Meaning Chapter 7 presented games of partial information as a general model of communication in natural language. This class of games makes possible a very fine analysis of ambiguity and disambiguation. The context can a¤ect the subjective estimate of probabilities, allowing one or another ambiguous form to ‘‘pop out.’’ Downtown in the financial district, the word bank will most likely be interpreted as a financial institution. Out in the country near a river, bank is likely to be interpreted as the side of a river. There is a lot more to be said about lexical items and lexical meanings beyond accounting for their ambiguity. In this chapter, I explore some other aspects of lexical items. There is a half-truth that I told in chapter 5 for which I now want to make amends. In that chapter, Abélard, the falsifier, and Eloı̈se, the veri- fier, played a game of picking and choosing objects from a world model in order to verify or falsify a sentence. I pretended at the time that both the verifier and falsifier had perfect access to the set of elements that denoted by a word. For example, if the sentence was (1) Some dog barks. then both players had perfect access to the things in the world model that count as dogs. In other words, if something is a dog, then the players know it is a dog, and if the players know something is a dog, then it is, in fact, a dog. In general, this can be simulated by listing the things that a word denotes. Of course, in a full semantics for natural language, words can denote all sorts of peculiar things like functions from sets to sets, and so on. For the moment, though, I want to focus on simple concrete nouns. Let’s take dog as an example. We can simulate knowledge of what dog means by simply listing the things in the world that count as dogs: (2) DOG ¼ fSami, Faye, Ginsberg, Rover, MacDu¤, Sandy, Pip, Fido, Adorno, Flip, Blue, Tucker, Apple Sauce, Two Dot, . . .g. Of course, I can’t list all the dogs in the world, even if I could in principle pick them out accurately. This points out a weakness in this kind of treatment of word meanings. The idea of listing the extension of a word the set of things that the word denotes is a mathematical trick; the idea is to simulate the traditional notion of an Aristotelian definition. An Aristotelian definition would give some set of necessary and su‰cient conditions that things must have in order to count as instances of whatever the word denotes. That is, anything that satisfies the conditions in the definition would count as an instance of the word and nothing else would count as in instance of the word. An Aristotelian definition should cover all and only the things denoted by the word. It’s clear, though, that this isn’t how words work. I, for one, know the Aristotelian definitions for almost no words. I failed to give a useful definition for tiger (concluding, in fact, that tigerness was a primitive essential property) and admitted that I know almost nothing about trees and plants. I know that dandelions are flowering plants, but I wouldn’t be able to pick one out in a garden with any degree of accuracy. Things get worse when we look farther afield. Consider a word like tra‰c, as in (3) The tra‰c was light on the Atlantic City Expressway today. What does the word tra‰c mean here? Presumably, some density of number of cars on a road. How many cars count as tra‰c? Two, a dozen, more? Well, that all depends on the road. If I’m driving around central Baja California, then two or three cars can count as tra‰c. There aren’t many cars on the road down there. But what about the New Jersey Turnpike? If there are only two or three cars around me on the New Jersey Turnpike, then that’s not any tra‰c at all. The right way to think about the meanings of words and phrases is not in terms of necessary and su‰cient conditions but rather appropriateness of usage. Does the word signal the intended meaning to the hearer in the context? Thus, working out appropriate usage involves...

Share