Introduction

Sometimes the same thing is not even synonymous with itself.

—Y. R. Chao

The first thing that is necessary for an analysis of graphic representation in linguistics is a definition of graphic representation in linguistics. In this book, the term "graphic representationM embraces most, though by no means all, of the figures or diagrams that are used in linguistics, and that are graphic. ''Graphic" rules out not only prose description (Hockett's [1947] formulation of the item-and-arrangement and item- and-process models of grammar, for example), but also mathematical notation and systems of representation that involve concatenations of symbols rather than diagrams. Thus the reader will not find here any treatment of mathematical models in linguistics, nor any application of mathematical techniques (such as graph theory or calculus) to linguistic concepts, beyond the rudimentary mathematics necessary (as in Chapter 6) to illuminate a particular graphic device. Again, the reader will not find here any treatment of systems of representation that concatenate symbols--themselves a kind of mathematics, an algebra--such as the systerns of categorial grammar or tagmemic analysis. Even within these limits, this work is not meant to be exhaustive. Every sort of diagram current in linguistics could not possibly be included. For instance, the graphic representation used for stratificational grammar, as well as that used for systemic linguistics--both greatly modified tree diagrams--is not discussed in great detail, though the tree diagram itself, of course, is. The diagrams tha are included are included because they illustrate the approach to linguistics taken in this book. Some, like the tree diagrams for immediate constituent analysis, represent major theoretical trends; others, like the matrix for Bell's Visible Speech, while comparatively minor elements in linguistics as a whole, serve to make a particular point--the provenience of a figure, for example, or its spread over various schools of linguistics. The diagrams included here, then, constitute a selection from, not a collection of, the diagrams in current use in linguistics.

״Representation," the other half of the term, requires that the figures or diagrams be used to represent something; that they make statements about the world; that they have meaning. Accordingly, we can divide graphic representation into form and meaning. We can separate the figure, whatever it is--say, a tree diagram--from the meaning it conveys in a given instance--say, immediate constituent analysis. This dichotomy between form and meaning makes possible the taxonomy of figures in linguistics that comprises the first three chapters of this book. It is a taxonomy of figures on the basis of graphic criteria into three types--tree diagrams, matrix diagrams, and box diagrams--and a discussion of the meanings for which the three types stand. It, too, is not exhaustive; it is meant, not as a comprehensive survey of the figures in use in linguistics, but as a demonstration of the feasibility of the approach taken in this book to the problem posed by the graphic representation of models in linguistic theory.

The problem lies in the existence of homonymy and synonymy: most of the figures can stand for more than one meaning, and most of the meanings can be expressed by more than one figure. When one looks at any given diagram, therefore, one cannot know, when faced with a tree diagram, for instance, whether it sets out the data in a family tree, in a taxonomy, in a componential analysis, or in an immediate constituent analysis. One cannot know until he looks at the data and at the context of the work in which the diagram appears and infers the statement that the diagram is making: if the figure appears in a work on historical linguistics and the data are the names of languages, its meaning must be a genealogy of the data; if the figure appears in a work on syntax and the data are words, its meaning is probably the immediate constituent analysis of the data; and so on. In short, the process by which diagrams in linguistics are interpreted is a process of circular reasoning. Using diagrams to make statements about the data amounts to begging the question.

So it is that graphic representation in linguistics is a thing sometimes not even synonymous with itself. As things stand, linguistics takes with respect to diagrams the position of Humpty Dumpty on words. "When I use a word," says Lewis Carrollfs character, "it means just what J choose it to mean--neither more nor less; to which Alice sensibly responds with the question, "whether you canmake words mean so many different things." It is not only that you cannot do so except at the price of circular reasoning, at the price of draining all the various figures of meaning, so that they are ready at any moment to mean what their users choose them to mean. It is also a matter of the influence of graphic representation on the development of linguistic theory--an influence, as things stand now, that is neither governed nor allowed for.

The mere possibility of a taxonomy of figures suggests the weight of graphic representation in linguistic science. It suggests as well the autonomy of graphic representation in linguistic science; for our classification, which proceeds by what we might call graphic least common denominators, cuts across other classifications of linguistic theory--by period, by type of theory (structural versus transformational), by subject phonology versus morphology). Thus we should not be surprised to find that a figure can come to have a life of its own, to influence the theory out of which it grew. In looking, in Chapter 4, at examples of graphic representation having directed the development of linguistic theory, we will take the first of two perspectives on the subject. From this vantage point, the vantage point of the philosophy of science, we view graphic representation as a component of linguistic science, as occupying in linguistic science the place of models in the biological and physical sciences. The influence of graphic representation on the development of linguistic theory is the counterpart of the influence of models on the development of biological and physical theory--the double helix in biology, for instance, or the wave model of light in physics. But what is it in graphic representation that enables it to furnish models for linguistic science?

With this question, we arrive at the second of our two perspectives on graphic representation in linguistics. In Chapter 5, we view it from the vantage point of the principles of graphic design. This explains how figures can serve as models in linguistic science: the principles of design are by definition the principles that relate a figure--a graphic design--and a meaning--the statement the design is meant to express. They are, so to speak, the rules of correspondence between form and meaning. Thus the two perspectives converge on a conception of graphic representation as an analogue for linguistic data. A model is an analogue seen from the viewpoint of the philosophy of science — the aerial view, so to speak, looking down on it as a part of the larger landscape of linguistic science; a graphic design is analogue seen from the viewpoint of the principles of design--the view from inside looking out. The relationship of analogy between figure and datum, between design and meaning, is what enables graphic representation to influence linguistic theory.

Up to this point the discussion of graphic representation in linguistics provides, not answers, but questions: it allows us to see problems like that of the influence of representation on theory, but it does not tell us what to do about them--how to undo the influence, or how to exploit it. Chapter 6 explores the possibility of new models for linguistics, models that grow out of old ones under the application of the principles of design. Language being the sort of thing it is, linguists cannot hope to vindicate the models they use by discovering actual physical structures that match them; there will be no double helix for linguistics. The double perspective taken here on the subject of graphic representation in linguistics suggests, however, that the solution to the problem of model-building lies in more model-building.

Ann Harleman Stewart

Seattle, Washington

February 1975