Polysemy and the lexicon
Concluding remarks
In my 1992 paper, I suggested that a search for the meaning, or meanings, of a word, such as open or window, may be ill-conceived. Speakers acquiring English do not acquire the lexicon word by word, nor, for any given word, do they acquire a finite number of discrete meaning packages. Neither, a fortiori, do they understand complex expressions by making choices amongst the various meanings supposedly associated with the words in question. They do not, for example, understand the expressions open a book, open your eyes, open the door, open the box, open a wound, and so on, by combining a meaning of open that they have acquired with a meaning of the direct object nominal; this is irrespective of whether the word meanings are presumed to be schematic or highly particularized. It is more plausible to assume that speakers understand these expressions because they have learned them as such. They have encountered the verb open in association with actions performed on various basic-level kinds of objects, such as doors, books, and eyes. If asked to state the meaning of open a book, speakers are readily able to describe the action, they can perform it, and mime it. But to state the meaning of open, abstracting away from the action performed on the book, is not only a difficult task, it is also an artificial one. Speakers may, no doubt, perceive some commonality between the different kinds of opening events. However, this perception of commonality is by no means necessary to guarantee mastery of the word– speakers merely need to have learned the relevant collocations. Appeal to established usage patterns and their associated practices readily resolves the disambiguation problem (or, rather, non-problem). Open the window is not perceived as being ambiguous, and no choices amongst the various senses of open and window are involved. This by no means entails that open and window have fixed invariant senses, merely that open the window has a learned sense, it is associated with an “embodied” practice, and, as Searle pointed out, it comes with easily statable truth-conditions. New uses, when encountered, will be interpreted on the basis of their similarity with already familiar uses. In the case of Searle’s example open the Sun, speakers would need to determine which of the established uses of open is to be taken as a model for interpreting the expression. They might construct some fantastic science-fiction scenario, in which opening the Sun would be analogous to, say, opening a wound. Or they might well conclude that the expression is simply nonsense.
I have focused on corpus evidence for the reality of usage patterns, showing that at least some uses of over resist a compositional analysis. This casts doubt on the idea that the word might have a finite number of discrete meanings. What was true of over, is no less true of all over. This expression also has a number of particularized uses, with their own distinctive semantic values, which, once again, resist a strict compositional analysis.
Wittgenstein’s Blue Book opens with the question: “What is the meaning of a word?” (Wittgenstein 1958: 1). Wittgenstein proposed to bring this rather abstract question– a question which, as he put it, is liable to induce in us a kind of “mental cramp”– “down to earth” by operationalizing it, that is, by replacing it with a more practical question, concerning how one might set about determining the meaning of a word. His answer, as is well-known, was that meaning was to be studied by examining how language is used: to say that a word has a meaning is to say that it has “a use in our language” (p. 69). The aphorism, to be sure, begs the question of what “use in [a] language” entails, and how it is to be characterized. This can be read as an attempt to come to grips with this issue, with respect to a (very) limited range of uses of over, and of all over. The picture that emerges is that a speaker of a language does not learn a list of words (with their meanings) and a list of rules for combining words, as proposed in the earlier citations from Jackendoff and Baker. A speaker has learned a vast number of word uses, some more frequent than others, to be sure, and some more “idiomatic” than others. This approach need not be in conflict with the argument from linguistic creativity, which is standardly put forward in support of the dictionary + grammar book model of language. On the contrary. Idiomatic expressions can themselves be creatively extended to new usage situations, and are subject to creative modification. Creativity need not be the bogey of a non-generative, usage-based linguistics.
