Fluctuating Linguistic Structures

Abstract

When asked about the object of their discipline, most linguists would agree that linguistics deals with language. This statement is surely not wrong, but it is empty, as empty as the statement that “physics deals with nature”. The everyday notion of language is very vague; without further precisions, it is almost worthless. Numerous attempts have been made toward this end, and two of them have deeply affected the course of language studies in this century: Saussures’s trichotomy of “language”, “langue”, and “parole” and Chomsky’s dichotomy of “competence” and “performance”. The Saussurian conception and particularly his notion of langue as a system of structural relations was — in its original and many modified forms — basic to the various directions of structuralism. In the last two decades, this role was largely taken over by Chomsky’s distinction and by his notion of competence. Like most attempts to clarify everyday concepts, both langue and competence in their technical sense tend to neglect many properties of the original, vague concept. They are idealizations, and the value of such idealizations crucially depends on whether the properties they neglect are essential features of the concept to be clarified, whatever might be understood by an “essential feature”. In this chapter, I will put forward some features of natural languages that I think to be essential. Moreover, I think that most linguists would agree with me here. Nevertheless, these features are neglected or widely neglected by both the concept of langue and the concept of competence. This does not necessarily mean that their empirical investigation has been neglected as well. But the tendency to do so is quite strong, of course, and even if it is resisted, a certain gap arises between the empirical work on the one hand and the theoretical background on the other. Several strategies have been worked out to handle such a gap between apparent linguistic facts and a theoretical framework that as yet has been unable to account for them: 1) The first and simplest strategy is to ignore certain features of natural languages. For example, most existing semantic theories simply disregard the vagueness of word meanings, so that they necessarily fail to give an adequate account of word semantics. 2) A second well-known strategy may be called “abstracting away”, i.e., the existence of facts inconsistent with the theory is efficiently recognized, but they are excluded by an abstraction process. The classical notion of competence, for example, involves both an ideal speaker-listener and a totally homogenous linguistic community, even though it is quite apparent that real linguistic communities are quite heterogeneous. Few linguists working in that framework would deny this fact, but they exclude it by abstracting from heterooenity. 3) A third and more sophisticated strategy is that of cover terms; the linguistic phenomena in question are “covered” by a newly coined term or terminology, they are accounted for by an additional index in a formal representation, and so on. This attitude of “treating” problems by making them in some sense invisible might be called “terminological immunization”.