It has been commonplace in linguistics to accept as 'given' that language in some way constitutes a 'system' but very little effort has been put into the discovery of the type of system involved, the majority of linguists concentrating on the specification of the elements and linkages internal to the assumed system. Typically, language has been thought of, though rarely defined as, a 'closed system' similar in structure to those of the physical sciences. Such a view has led during this century to increasing reductionism, to the acceptance of the hypothetico-deductive approach to investigation of phenomena and to the adoption of models rooted in the axioms of symbolic logic and of mathematics.

In contrast, the social sciences have viewed their objects of study as 'open systems', interacting with, rather than isolated from, their environments. The distinction between these two types of system, derived from work in cybernetics in particular, can be seen to be of crucial importance to linguistics, since what is now required, as a result of sociolinguistic and psycholinguistic research, is a radical reorientation of the discipline to its data. It is now necessary to see language as an open system consisting of a virtually closed core but with several subsystems, so to speak on its periphery, which are very open and have strong degrees of connectedness with the systems which make up its environment. The task resolves itself into the creation of an adequate theory and explanatory model which reflects both the closed 'mechanical' aspects of the code itself and the open pattern and evolutionary aspects of the individual and social use of the code, and then links these two aspects in a systematic way through sets of statements connecting the inner form and the external functions of the overall system.

The paper has the following structure: an outline of the notions 'whole', 'system' and 'model' in the philosophy of science, the impact of these concepts on scientific method, the relationship of the terms to more common linguistic concepts such as langue and parole and their modern partial synonyms, the characteristics of 'closed' and 'open' systems in the sciences and, finally, a consideration of the relevance of 'systems thinking' to linguistics.