The distributional analysis of meaning

In much of the statement of meanings in dictionary entries, the linguist necessarily looks beyond the realm of language, as he is dealing with the use of words in sen­tences in relation to the whole of human experience, and he may have to rely on sciences other than linguistics and on unsystematized ‘common sense’. For this reason some lin­guists have tried to redefine or reconsider meaning in so far as it is relevant to linguistics as equivalent to distri­bution. That is to say, the meaning of a word, as far as it concerns the linguist within the strict confines of his subject, is to be understood as the range of its occurrences in sentences consisting of other words. Just as there are prob­ably no words exactly alike in meaning in all contexts, so there will probably be no two words in any language sharing exactly the same lexical environment (distribution). The temptations of this treatment of meaning are strong. Re­liance on sciences and experience outside linguistic compe­tence is eliminated; and, especially with the use and devel­opment of statistical methods and computer techniques (the mechanical collection and sorting of data), considerable precision and exhaustiveness, not available in semantics otherwise, seems within reach.

The objection, however, to this method of treating meaning is that it fails to ‘save the phenomena’. Meaning is everywhere understood as involving the relation of language to the rest of the world, and such meaningfulness is an essential part of any definition of language. Distributional studies enable one to state a great deal about the total functioning and use of words in a language, but just not what is ordinarily understood to be their meanings and what is rightly expected from the entries in a dictionary, for which information such traditional terms as reference and denotation, and more modern systems of analysis such as context of situation, however imprecise they may be in many cases, have been devised.

The distributional approach to meaning, while inadequate as a complete treatment of it, or at least as the way one would ordinarily expect it to be treated, brings to notice the significant fact that part of the total meaning of many words in all languages is to be determined by their individual relations with other words, in both the basic dimensions of linguistic analysis, syntagmatic and paradigmatic. Words, as individual lexical items, are structurally related to each other in languages and the uses of languages, as they are as members of grammatical classes and as are grammatical and phonological elements generally, though in different ways. The syntagmatic relations between words as lexical items have been studied under the title of collocation, and the paradigmatic relations are considered in the theories of the linguistic field.