Links to pages: 354, 355, 356, 357, 358, 360
Comments Welcome
George W. Grace
University of Hawaii
December 14, 1984
[I should point out that this will be a far-from-thoroughgoing discussion of either boundaries or open systems. What I am most immediately concerned with is laying some groundwork for re-examining the problem of genetic subclassification in linguistics.]
In Grace 1984a and 1984b I pointed out that human languages are systems of the kind which I called "open". By "open systems" I meant what I described (Grace 1984b: 19) in the following way: "systems which do not have clear boundaries--where some things belong to the system more clearly than do others and still others may not belong at all, although there is no (non-arbitrary) basis for saying for sure." I want to develop the concept of open system somewhat further here because I believe that we will never be able to talk about some of the things that we need to talk about unless we have a way of talking about open (in this sense) systems.(1)
The key characteristic of an open system is the absence of natural boundaries. However, bounded systems, i.e., systems which do have boundaries, are not all of the same kind. In some cases, the boundary has a function which is essential to the understanding of the system. Such systems may be spoken of as intrinsically closed). In other cases, the boundary is an incidental by-product of the processes through which the system evolved. Such systems may be spoken of as incidentally bounded (or incidentally closed). We may then refer to systems which are either open or incidentally bounded as intrinsically open systems.
I will first say a few words about boundaries.
In one kind of intrinsically closed system the boundary is an essential feature of the process of differentiation, and the differentiation in fact consists in the erecting of a partition or establishment of a contrast. This kind of case may be illustrated by phonemic differentiation as it appears in the various structuralist views. In these views it could be (and was) said that "the business of a phoneme" was to be different from other phonemes. A phonological system was seen as being a system of "phonological oppositions" (Prague School) or "phonemic contrasts" (American School). An example of such opposition from the natural world might be that of the male and female sexes.
At the other end of the spectrum are incidentally bounded systems--where the boundary which arises is a quite incidental by-product of the process which produces the differentiation. In fact the differentiation may itself be a by-product of the changes from which it results.
If boundaries of the second type (which we might call "gaps") separate anything, the fact that they do so is incidental. For example, the processes involved in biological speciation may eventually result in populations which are incapable of interbreeding, but the evolution which leads to this degree of differentiation does not occur with this end in view. Likewise, linguistic differentiation may ultimately lead to linguistic varieties which cannot be used by their speakers to communicate with each other at all ("mutual intelligibility" is lost), but the linguistic changes which ultimately lead to this result do not occur with this end in view (or else the very fundamentals of our theories of linguistic change are misguided).(2)
There is one consequence of what has just been said that I would particularly like to call attention to. It is that for a way of talking about linguistic differentiation (as distinguished from linguistic change as that is presently conceived) to make it a matter of talking about boundaries is to misinterpret the whole process. At least as far as linguistic differentiation is concerned we should assume openness to be natural and boundaries to be an accidental by-product. Let me now return to the problem of dealing with openness.
It may help to make the notion of "open system" more concrete if I point out that the categories represented by the meanings of ordinary language words are intrinsically open systems. In Grace 1981 I proposed the term "categorization by exemplars" for this kind of categorization. I (1981: 80) defined it as follows: "Categorization by exemplars consists in taking a sample of one or more examples as representative of the category. Other individuals are then assigned to the category on the basis of similarity to the exemplars, or perhaps of similarity to members of a chain of similar individuals which leads ultimately to one of the original exemplars. The membership of such a category is not necessarily indeterminate, but it typically is. Attention is likely to be focused on the exemplars."(3)
There are several other kinds of open (or intrinsically open) systems with which I have been concerned at one time or another. As was just pointed out, there are, of course, ordinary word meanings. Also in Grace 1984b I was particularly concerned with the systems of assumptions which constitute what I called "subject-matter views".(4) Such systems are, as was emphasized there, open systems, and I am convinced that it would be impossible to have any meaningful discussion of such views without recognizing their openness. Finally, and this is the intrinsically open system which most concerns me here, there is the individual language. Of course my immediate concern is with the languages (including various proto languages) which are the basis of genetic subgrouping in linguistics, but that is not the only context in which the assumption that languages are closed systems can cause trouble.
I am far from prepared to propose a full blown conceptual apparatus for approaching open systems.(5) I will have only a few general suggestions. However, I would like to point out that an approach which is designed for open systems would be equally appropriate for many closed systems as well. Per contra, our current practice of applying a closed-system approach to open systems often yields quite unsatisfactory results.
What is distinctive in an open-system approach is not that the systems to be investigated must lack boundaries, but rather that our attention is not focused upon whatever boundaries may happen to exist or, if none have been discovered, upon the search for boundaries. What is needed is a way to recognize and exploit systematic relations without feeling any obligation to try to define boundaries delimiting the system. It will be convenient, therefore, to use the term presumptively open system for a system which is to be approached using an open-system approach--that is, which we are to study as if it were an open system even though we have not established that it is unbounded (and probably have not been concerned to determine whether or not it is). Presumably at the very least, all intrinsically open systems (recall that this means systems which are either open or only incidentally bounded) should be treated as presumptively open.
To begin with I would like to define (presumptively) open system in a very general way. I propose that we recognize a presumptively open system as existing if there are at least two elements which are related in any of the following three ways:
1. They interact: one acts upon the other or each acts upon the other.
2. They co-occur (go together): the presence of one implies (with greater than chance frequency) the presence of the other or there is mutual implication.
3. They are similar (partially equivalent or intersubstitutable).
The presence of a relation connecting as few as two such elements establishes the existence of a system. More such elements can be added on the basis of their being related to some element already in the system. An intrinsically open system therefore has the structure of a chain (and a presumptively open system is to be presumed to have that structure until there is evidence to the contrary). (For convenience, I will talk as if the members of a system are its elements. It is not the elements in themselves which compose the system but rather the elements in the way in which they relate to particular other elements--the elements as relata rather than as elements per se).
One other matter which requires brief mention here is that of the internal homogeneity of systems. There are at least two ways in which homogeneity in a system might be conceived. First, one might say that a system is homogeneous if the fact that an element belongs to the system tells us something (else) about that element. That is, a homogeneous system would in this conception be one in which every constituent element possessed some common attribute. If there were one attribute shared by all constituent elements which was not shared by any non-constituent element, then that attribute could serve as a defining property (criterial attribute) of the system.
Clearly, open systems cannot be homogeneous in that sense, and of course, chains are not necessarily so. There is, however, another possible conception of homogeneity. If we are able to arrive at some basis for measuring the relative closeness of two elements (or two links in the chain) according to the relations on which the system is constructed (i.e., this particular relation holds between A and B in a greater degree than between A and C or B and C), then we might define a homogeneous system as one in which the closeness of relation between any member element and any other member element is greater than that between any member element and any non-member. In this sense also, of course, an open system could not be homogeneous although an incidentally bounded system could. But even an intrinsically closed system would not necessarily be homogeneous in either sense.
The main point to be made is that systematicity should not be taken to imply homogeneity (for on such confusions rest prejudices of many kinds).
All of what has been said suggests two conclusions about the world and our relations to it. First, there are very many systems in the world--in fact their number is presumably limited only by the imagination of their observers. Increased recognition of this fact seems destined to take us one step farther from the great-chain-of-being view of the natural world which (I think) has left a still-discernable mark on the ideology of science. Second, since a chain can be entered at any point, any extensive system can be viewed from any of a large number of starting points (a truly open system presumably from an infinite number) which is to say that it can be seen and described from a large number of different perspectives. The recognition of this fact will, I think, tend to loosen the hold of the constant-point-of-view ideal which is also a discernable part of the ideology of science.
If we do not undertake to study open systems, we will be obliged to ignore most of the structure in the world. However, our usual tactic in studying open systems is to treat them as if they were closed, either by arbitrarily assigning them boundaries or by assuming that such boundaries actually exist and that we will discover them in the course of events. I have no objection to these tactics as such just so long as we do not permit ourselves to be deceived by them into thinking that the particular systems (the particular prototypical patterns) that we have chosen for study have some special claim to reality. In fact, the number of systems which are there to be recognized is infinite, and there is much to be said for a strategy which envisages recognizing as many of them as possible and exploiting each just for so long as the returns are truly great and then setting it aside for another. If we take a particular one too seriously, we may be tempted to cling to it long after its maximum returns have been collected.
1. However, the term, "open system", has also been used to designate a system with clearly identifiable boundaries, but boundaries which permit some interchange with the external environment to take place. That is the kind of system of which I. Prigogine, in particular has written--structures which are "far from equilibrium" or "dissipative structures" (e.g., Glansdorff and Prigogine 1971). Such structures no doubt also are to be found in linguistic phenomena, but they are not what I have in mind here. Back up
2. However, current linguistic theory has, to my way of thinking, moved so far in the direction of mysticism that I cannot really be at all sure that theoretical linguists would not find at least a possibility of some kind of larger purpose in the processes of biological and linguistic differentiation which I have cited as examples. Back up
While prototype semantics is particularly concerned with what kinds of examples are conventionally taken to be most representative of the category by speakers of a language, my concern is somewhat different. It is with the construction of categories starting from a pattern represented in a single prototypical individual or set of similar individuals and proceeding to include new individuals on the basis of pattern recognition. More specifically, the principle for inclusion of an individual is recognition of a common pattern in that individual and a prototypical individual or some other individual who already belongs to the category. Back up
4. The principal examples of subject-matter views discussed in Grace 1984b are the traditional view of the nature of language and an alternative view which I propose there. Back up
5. In fact, I am inclined to imagine that dealing with open systems will require probably the greatest revolution in the ways of thinking of "Modern Civilization" since the latter was first concocted. But I also think that it is only by means of this revolution that language can finally assume its rightful place in the ideology of science. Back up
Glansdorff, P. and I. Prigogine. 1971. Thermodynamic theory of structure, stability and fluctuations. London, etc.: John Wiley and Sons. Back up
Grace, George W. 1981. An essay on language. Columbia SC: Hornbeam Press. Back up
Grace, George W. 1984a. More on autonomous text. Ethnolinguistic Notes, Series 3, no. 21. Printout. 9 pp. Back up
Grace, George W. 1984b. The linguistic construction of reality. Printout. 234 pp. Back up
| Home Page | The Ethnolinguistic Notes | The Ethnolinguistic Notes, Series 1 and 2 | Ethnolinguistic Notes, Series 3 | The Ethnolinguistic Notes, Series 4 | Reflections: Language Evolution |
| Reflections: Knowledge of Language | Personal Page | The Human Predicament | Why Write Unpublishable Things? | Modest Proposals | Odds and Ends | Pictures |
Last updated 4 September 1996