Defining terms or describing things?

THE CENTRAL THESIS of this paper was provided by one of the "big five" precursors of general semantics. It was Cassius J. Keyser who said:

   I can think of no greater improvement in our human discourse than
   that which would result if writers and speakers would stop the well-
   nigh universal and vicious practice of confusing definition and
   description ... In any useful sense of the term definition, a thing
   is definable if and only if it is possible to indicate at least one
   mark serving to discriminate that thing from all things else. But any
   true statement about a thing, even if true of a million other things,
   is a partial description of it. A vast majority of the so-called
   definitions encountered in literature are, even when true statements,
   nothing but partial descriptions. And when such a partial description
   is submitted as a genuine definition, one is bound to infer that the
   author either does not understand the essential nature of definition,
   and so is fooling himself, or is engaged in trying to deceive
   others. (1)

If the voice of Keyser were the only one calling for this discrimination between two levels of abstraction, it might be dismissed as a mere personal whim. But such is not the case. In Principia Mathematica, Whitehead and Russell declared that "... a definition is concerned wholly with the symbols, not with what they symbolise." (2) And one might add that a description is concerned with what the symbols symbolize, not with only the symbols themselves. And then Jacques Rueff has said: "... the statement that the definition expresses the essence of an object of the external world does not and cannot have any sense. An object is the sum-total of sensations. A definition, on the other hand, is a sum-total of non-contradictory words. The two are of distinctly separate orders." (3)

These voices, among others, afford strong support for a fundamental postulate: namely, definitions are always language directly about language, whereas all of those descriptions of interest to us are language directly about non-verbal things. According to this postulate, it is only one step from a description to the thing described; whereas there are at least two steps from a definition to the non-verbal realm. All of the traffic from definitions to the non-verbal things must take a detour through language. According to this postulate, there is no guarantee that definitions are dependable guides to the life facts unless the terms defined are adequate representatives (in the given culture) of the facts in question. If we agree to call an apple by the name banana, a new definition of the term banana would be needed. But to re-define the term banana without concern for the things now called apples, however entertaining the process, would not serve to distinguish apples and bananas. As the Columbia Associates in Philosophy have pointed out, "the definition must prove the means of identifying the thing defined and no other." (4)

Now in place of the terms apple and banana, substitute such terms as appeasement and negotiation, and you will surely realize a legitimate need for the rigorous, keyserian usage of the term definition.

But the keyserian usage is not the conventional usage in speech circles--nor many other circles, for that matter. Take the argumentation and discussion literature for example. Under the single label of definition we find both definition and description in the keyserian sense. According to the speech authors, an object is defined by explaining its purpose or function or how it works; a term is defined by the substitution of other terms; either an object or a term is defined by citing examples; such things as the Monroe Doctrine are defined by their history; a term is defined by etymology, usage, or context; and either terms or objects are defined by association, negation, analysis. And the most common pattern of these 'definitions' was presented as if, in defining terms, a non-verbal thing were assigned directly to a broad class (called genus), then to a narrower sub-class (called species), followed by a differentiation of the given thing from other things assigned to its sub-class. (5) Now is that what speakers do? Strictly speaking, it is not. Perhaps that is what biologists do, but that is not what speakers [per se] ordinarily do. Speech is not biology: biologists often deal with non-verbal things directly; speakers do not. For the most part, the specimens of speech are not things but words; it is words which speakers ordinarily classify and differentiate. The words may be representative of the actualities to which they are said to refer, but being 'accurate' is not a necessary function of words. Speakers ordinarily classify and differentiate things only indirectly and only if the verbal maps are adequate. As A.B. Johnson expressed it, "... un-verbal things are no party to our verbal disquisitions. They exhibit themselves just as our senses and our intellect discover, unaffected by our speculations, unchanged by our definitions." (6) The point is that a large sampling of the argumentation and discussion literature did not disclose an explicit distinction between defining terms and describing things. Indeed, the practice of confusing the two does seem to be well-nigh universal.

This is not to say that the disposition to establish, and maintain an agreement of meanings is socially undesirable. Far from it. But times have changed. As Korzybski said:

   In scientific literature of the old days, we had a habit of demanding
   'define your terms.' The new 1933 [1950] standard of science really
   should be 'state your undefined terms.' In other words, 'lay on the
   table your metaphysics, your assumed structure, and then only proceed
   to define your terms in terms of these undefined terms.' This has
   been done completely, or approximately so, only in mathematics. (7)

Of course the define-your-terms habit is still with us. Perhaps the demand is not demanding enough. But Keyser explained the situation this way:

   No discourse ... can define all of its important terms. The reason is
   plain: there is no way to define a term except by means of other
   terms; and so if we define certain terms by means of others, then
   those by still others, and so on, in the hope of defining all of our
   terms, we are bound to use, sooner or later, directly or indirectly,
   the terms first defined as means for defining others; and so our
   behavior will resemble that of a kitten pursuing its tail--a charming
   motion but no journey. (8)

Thus it is that anyone who faithfully defines his terms can spin himself into one of Wendell Johnson's verbal cocoons. And be it observed that he can never define his way out of it--not, at least, in the keyserian sense of definition.

So we have some advice for the faithful definer: don't spin yourself into a verbal cocoon. Learn to discriminate between definition and description. Instead of the single category, set up two categories: definition and description. Under the label of definition enter the language-directly-about-language procedures. Reserve for description only those procedures involving language directly about non-verbal actualities. This constitutes your measuring stick. Apply this measuring stick. You will discover, eventually, that proper classification depends upon the situation. When you have made this discovery, dig out one of your old speeches. Re-discover one of your so-called definitions, and find within it some undefined term. Working in the old pattern of definition-only, define this term; and then, from your new definition, select another undefined term--and so on, until you are thoroughly exhausted. Like Keyser's kitten, you will have had a "charming motion but no journey."

Now go back to the original specimen of language about language. As before, select one of its undefined terms. What actualities does this term represent--or misrepresent? Ask yourself, "What is the territory for this verbal map?" Describe that territory--not completely (for that's impossible)--but describe it adequately for your purposes. And then, if necessary, change your verbal map to fit the territory. And if you still need a definition, use the undefined term (but now its territory has been partially described) in your definition. Thus, by making description a prerequisite of definition, you may avoid verbal cocoons.

And now that you have followed this advice, you will surely appreciate an illustration by A.B. Johnson--an illustration which encourages what workers in general semantics call an extensional orientation.

   What is the moon? If an infant were to ask me this question, I might
   tell him to go into the street, and on looking towards the sky, he
   would discover something that looks like a large round piece of
   silver. That is the moon. You may say that my designation will not
   enable the child to find the moon, and you may give him some better
   description. We probably shall not altercate, because we shall
   understand that our words are intended to merely point out to the
   child something that is different from the words. But suppose I were
   to ask a philosopher to tell me what the moon is; he might say that
   the moon is an opaque globe of land and water, like our earth. He is
   not attempting to designate an existence, as I did to the child. My
   words were not supposed to be the moon itself; but the philosopher's
   definition is the moon verbally at least. You probably now understand
   what I mean by saying, that in all verbal discussions we should
   discriminate whether we are attempting to define a word, or to
   designate an existence. The discrimination is seldom made, and the
   want of it produces much contention and confusion. (9)

In these times of "much contention and confusion," it seems to me we might well be prepared to answer this question: Defining terms or describing things?

[Original] Editor's Note by M. Kending:

For the author's reference in the first sentence to the "big five precursors of general semantics," see Alfred Korzybski, "Fate and Freedom" (1923), reprinted in Irving J. Lee, ed., The Language of Wisdom and Folly (New York: Harper, 1949), pp.341-357. "All human achievements are cumulative; no one of us can claim any achievement exclusively as his own; we all must use consciously or unconsciously the achievements of others ... Much of what I will say has been said before by many others ... the names of a few stand prominent." (Whitehead and Russell, Poincare, Keyser, Einstein.) As regards the author's use of "precursors" he has written, "I applied this label from a time-binding point of view. But also, my study of Keyser convinced me that Korzybski's 'mathematical philosophy' stems from and goes beyond the Keyserian mathematical philosophy in numerous fundamental respects (e.g. the extensional interpretation of 'real' variables in Keyser's writing; the extensional devices for handling multiordinal terms as variables in Korzybski's writings)."

REFERENCES

1. Cassius J. Keyser, "Humanism and Pseudo-Humanism," The Hibbert Journal, XXIX (1931), p.229.

2. Alfred N. Whitehead and Bertrand Russell, Principia Mathematica (Cambridge: at the University Press, 1910), I, p.11.

3. Jacques Rueff, From the Physical to the Social Sciences, trans. Herman Green (Baltimore: The Johns Hopkins Press, 1929), p.30.

4. Columbia Associates in Philosophy, An Introduction to Reflective Thinking (Boston: Houghton Mifflin Co., 1923), p.33.

5. See, for example, A. Craig Baird, Discussion; Principles and Types (New York: McGraw-Hill Book Co., 1943), pp.28-37; James H. McBurney and Kenneth G. Hance, The Principles and Methods of Discussion (New York: Harper and Brothers, 1939); William T. Foster, Argumentation and Debating (2nd ed. rev.; Boston: Houghton Mifflin Co., 1945), pp.25-35.

6. Alexander B. Johnson, The Meaning of Words, etc. (Originally published, 1854; Milwaukee, Wisconsin: John Windsor Chamberlin, 1948), p.91.

7. Alfred Korzybski, Science and Sanity: An Introduction to Non-Aristotelian Systems and General Semantics (First Edition, 1933, Second Edition 1941, Third Edition, 1948, International Non-Aristotelian Library Publishing Company; Institute of General Semantics, Lakeville, Conn., Distributors), p.155.

8. Cassius J. Keyser, The Pastures of Wonder: The Realm of Mathematics and the Realm of Science (New York: Columbia University Press, 1929), pp.72f.

9. Alexander B. Johnson, A Treatise on Language, ed. by David Rynin (Berkeley and Los Angeles: University of California Press, 1947), p.178.

ELTON S. CARTER, PH.D.

From General Semantics Bulletin Nos. 6-7, Spring-Summer 1951. Dr. Carter taught at State College, Pennsylvania.