Our pre-analytic concept of the meaning of a word is not well defined. We have some intuition that there is a kind of objective meaning that the word somehow reaches out to grab hold of. Although this view is simple and agrees with our intuition there are several reasons to object to it, for example it doesn’t explain how people can disagree on the meaning of a word, and it introduces a metaphysical entity, the abstract meaning, which is a move many philosophers try to avoid.
The first step beyond this intuitive understanding of meaning was taken by Wittgenstein (as far as I know), who argued that the meaning of a word was given by how we use the word. If I use a word, say X to refer only to those objects that are blue then X means the same thing as the word “blue”. If tomorrow we started to use X to refer to objects that were red then X would mean the same thing as the word “red”. Thus the meaning of a word is taken out of the abstract realm and turned into something we can observe.
The next step was taken by Bruce Goldberg in his article “The Linguistic Expression of Feeling” (1970). In it he argued that the meaning of a word can not be determined simply by its association with various objects. For example consider four blue chairs. If you say X while pointing at each chair clearly you have demonstrated to me the use of X, but I still don’t know whether X means blue, or chair, or blue-chair. Additionally it seems impossible that our internal “meaning” of X could be determined by knowing a list of objects or situations which it applied to, because there are an infinite number of possible objects a word like blue might be applied to, but when we think of “blue” we do not see in our mind’s eye this endless catalogue. Under Wittgenstein’s theory we are forced to assume such an endless mental catalogue because without it there should be some object that we would be unable to apply X to, because the objects is not in the catalogue, and we have no method for extending a finite catalogue of associations to cover this new case.
For these reasons Goldberg argued that the meaning of a word is determined by the class of objects that we would assign the word to, not the objects themselves. Of course we must also assume that we have a method of assigning objects to classes, and although the method was not explicitly sated it seems safe to assume that the categorization is based on rules since that is the simplest solution. Thus I determine what is “blue” by a process in which I associate the object with a category, presumably with the help of a rule that says “objects are blue when you observe light in the following part of the visual spectrum”, and it is this category provides meaning to the word “blue”. Of course language is not learned though the communication of these rules, but after being shown enough objects in conjunction with a word we are able to abstract the rule from the examples provided.
Wittgenstein’s work however contains an objection to just this kind of improvement. Wittgenstein asks us to consider the word “game”. There are many kinds of games, so many in fact that it is impossible to give a rule that encompasses them all. Despite this we are able to apply the word “game” to various situations without hesitation, indicating that we aren’t using a rule to determine when to use “game”.
So there seems to be something right in Wittgenstein’s analysis and something right in Goldberg’s analysis. One solution is a kind of a compromise between them. Instead of having a formal rule, such as “apply word X when …” we could assume that the rule is based on exemplars, meaning that our minds hold several objects as key examples of when to use the word. When we come across an actual object our minds subconsciously decide if it is close enough to the exemplars that we have associated with the word. If it is sufficiently similar we place it in the class that is associated with the word in question, and if it isn’t we exclude it from that class and associate it with another instead. Of course different people may have different exemplars with which they classify objects, but this doesn’t imply that the word means something different to each person. As Goldberg argued in his article we should say that the meaning of a word is the same for two people when the classes they associate the word with contain identical members; the manner in which they assigned the objects to the class is irrelevant to this claim. This kind of rule resolves our objections to Wittgenstein, because the meaning of a word is no longer solely its use, and it resolves our objections to Goldberg because the meaning of a word no longer has to be decided by a formal system, although we might see this as simply a clarification of Goldberg’s work since he never explicitly stated how the classification worked.
Note: by object here I mean anything that we might label with a word, including thoughts, feelings, abstract categories, events, and of course physical objects.