Hume and 'The Meaning of a Word'

  • 30 130 2
  • Like this paper and download? You can publish your own PDF file online for free in a few minutes! Sign Up
File loading please wait...
Citation preview

Hume and "The Meaning of a Word" James F. Zartman Philosophy and Phenomenological Research, Vol. 36, No. 2. (Dec., 1975), pp. 255-260. Stable URL: Philosophy and Phenomenological Research is currently published by International Phenomenological Society.

Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at JSTOR's Terms and Conditions of Use provides, in part, that unless you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you may use content in the JSTOR archive only for your personal, non-commercial use. Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed page of such transmission.

JSTOR is an independent not-for-profit organization dedicated to and preserving a digital archive of scholarly journals. For more information regarding JSTOR, please contact [email protected]. Tue Apr 24 02:42:17 2007

DISCUSSION HUME AND "THE MEANING OF A WORD" Expositions of Hume's view of language often contain expressions like the following, which are intended to formulate one of his basic principles: A. The meaning of a word is nothing but the idea which it expresses.l B. The meaning of a term is an image.2 Often, too, the meaning of a word is said to be the idea that it "stands for." Hume does not explicitly state A or B, or anything like them; Flew speaks of B as an assumption that is "neither argued nor fully expounded." A and B are commonly held to be false or mistaken by those who attribute them to Hume, and although we may have to concede that Hume's semantics is, in the end, inadequate, one may ask whether we have to condemn him for holding to a principle as patently inacceptable as A or B. Those who ascribe A or B to Hume have in mind the "ideational theory" of Locke; and they assume that Hume does not differ from Locke except, perhaps, on the question of abstract ideas. It is also commonly held that Hume used some form of an "empiricist criterion" of meaning in order to eliminate metaphysical terms from philosophical discourse," and that the ideational theory provided him with the foundation for such a criterion. In this discussion I shall try to show (a) that Hume does not have to be interpreted as having held to the primitive form of the ideational theory, and ( b ) that the more advanced view I shall derive from Humean principles a l s ~ generates a version of the empiricist criterion. There are a few reasons why we might question Hume's committal to A or B. They are: 1. A says that a word "expresses" an idea, while B leaves the wordidea relation unclear; neither A nor B uses the relation Hume actually describes as holding between words and ideas. H. Basson, David H u m e (Harmondsworth, Middlesex, 1958), p. 43. Flew, Hume's Philosophy of Belief (London, 1961), p. 36. Flew uses "image" as the equivalent of Hume's "idea." 3 For this theory, see W. F. Alston, Philosophy o/ Language (Englewood Cliffs, N.J., 1964), p. 22 ff. 4 Ibid., p. 63 f. 1 A. 2 A.

2. A and B are identities, and thus have a certain strong force, while it may be sufficient for the purpose to assign to Hume only the use of ideas as a test of meaningfulness. 3. Although it may be only an infelicity in their expression, A and B seem to imply an identity between the meaning of a word and a single idea, whereas, in the case of general words at least, Hume recognizes that one word can be connected with a number of different, but resembling, ideas. 4. A and B are not explicit in Hume, and there would be a presumption in favor of a principle that couId be more closeIy derived from his actual statements, and could be shown to be more in keeping with the originality of his philosophy. Let me first point out in connection with number 2, that an "empiricist criterion" is adequately stated in the form of an "if and only if" expression, providing us with a test or criterion of meaning. fulness. Compare the two principles: The meaning of a word is the idea that it stands for. A word is meaningful if and only if it stands for an idea. The first entails the second, but I do not think that the second entails the first. The first is a strong, "reductionist" principle that would involve its defender in certain difficulties avoided by the second, for the latter avoids something called "the meaning of a word" and also avoids saying what that thing is. The second would be even safer if it eschewed use of the relation of "standing for," which is unclear, and whose use, it might be argued, does make the second seem to entail the first. In any ease, I see no compelling reason why the second, or something like the second, could not be attributed to Hume in preference to the first. The second is sufficient for Hume's main purposes, and does not gratuitously commit him to a reductionism unsupported by his actual methods. Had I not an entirely different proposal in mind, I should want to defend something like the second form as preferable to A, B, and their like, in the reconstruction of Hume's "assumptions." The hypothesis I shall offer will not attempt such a reconstruction, but suggests a posi'tion that Hume could have derived from original aspects of his philosophy. A role will be assigned to a criterion similar to the second principle above, but based on Hume's own views. I take as fundamental what Hume says in the chapter "Of Abstract Ideas" in the T r e a t i ~ eHe . ~ asserts that words "raise up" ideas. On hearing a word, such as "man," an idea is called up in the mind5

P. 17 ff. of Selby-Bigge's edition (Oxford, 1888), which is hereinafter referred to.

in this case, it will be the idea of some particular man, e.g., Socrates. (Of course, for Hume all ideas are of particular objects.) Words can cause us to have ideas, then, and since a causal connection is a correlation, we have this: C. Words are correlated with ideas. It would be misleading to say that a word is correlated with an idea, if this were taken to mean that the same word must always call up the same ideaa6"Man" might perfectly well cause me to imagine Socrates on some occasions, but on others Agathon or Alcibiades. Even if I were accidentally always to imagine Socrates on hearing the word "man," other ideas, as we shall see, will be no less involved with the word. C, however, does not account for what ideas a word can call up, nor how many ideas can be connected with the same word, and for this Hume introduces a second element. Hume says that words raise up both ideas and customs of the mind. The custom associated with a word is formed as a result of our applying the word to different, but resembling objects, and enables us to pass from the idea of one such object to others. To C we must then add: D. Words are correlated with customs. Hume says the custom connected with a word can expose errors: if "All men are snub-nosed" seems true when we imagine Socrates as our sample of man, the custom involved with the word "man" enables us to go on and think of other men who are not snub-nosed, thus producing an exception that falsifies the generalization. This example of mine and Hume's example about triangles clearly show how the primitive ideational theory would be inadequate, if we take it to have involved only one idea with a word. The introduction of a one wordmany idea relation, made possible through Hume's customs, is a great step forward. Hume's account of the origin of these customs, saying that they are acquired by frequently applying the same word to resembling objects, is objected to by Kemp Smith. He insists that Hume "evades" rather than solves the problem of generalitye7Although I shall later touch on that issue, it is fairly clear, at least, how these customs work, assuming it is possible to acquire them. If I have applied the word "W" to objects O,, 02,03,etc., the custom or habit thereby formed, and called up when "W" is heard again, enables me to think of 0 , or 0, or 03,o r the like, ad libitum, in addition to whichever idea 6


One of Alston's criticisms, [bid.

The Philosophy of Dal>ld Hume (London, 1949), p. 261 f f .



of one of these first struck my mind. Hume offers four "reflexions" on the working of custom, designed to clear up lingering difficulties, , of which the fourth will be useful later. Although there is further detail in ~ u m e ' sdiscussion, we nowhere find anything called "the meaning of a word." We find only words, ideas, and customs. A and B, however, attributed to Hume an assumption about what the meaning of a word is, which saw in his views no progress beyond a simple ideational theory, presumably with its one word---one idea relation. We have already seen, in D, clear evidence of Hume's having moved a step beyond such a simple theory. If we assume that some place might be found in Hume's system for a thing called "the meaning of a word," there is little reason to think that A and B have told us what it is. We might choose to assign to Hume only some form of the empiricist criterion of meaningfulness, but I propose that we ask what would be a Humean answer to the question what the meaning of a word is. If we turn for help to the example set by Hume's analysis of causal inference, an interesting parallel may be drawn. Looking for the meaning of a word is not unlike looking for the necessary connection between cause and effect. In both instances, besides ideas (and words) we have only a custom. In our thinking about causes there is a custom connecting one idea (of the cause) and another idea (of the effect). In language, there is a custom connecting a word to a set of ideas. In the former case, Hume comes to the "extraordinary" conclusion that the necessary connection between cause and effect cannot be other than a "determination of the mindb' produced by a custom, to pass from the idea of the cause to the idea of the effect. The hypothesis I propose is that the following is an acceptably Humean view: E. The meaning of a word is the custom, connected with the word, that enables us to pass from one idea correlated with the word to other ideas that resemble it. Understanding the meaning of a word would consist in the acquirement or possession of such a custom, not in merely possessing o r forming an idea, in connection with the word. For example, to understand "man" is not to think of Socrates o r Agathon or Alcibiades, etc., but to have the custom of imagining men, such as these, in connection with the word. E in no way destroys or minimizes the importance of ideas for meaning; like A and B, it is still within the tradition of the ideational

theory. A, B, and E alike recognize the importance of C, which affirms the connection between words and ideas, but E alone also assigns a role to custom, which Hume clearly emphasizes. Some argue that ideas are not relevant to meaning at a11.8 an issue that cannot be gone into here, but I think it may be said that E, a Humean view, represents a considerable advance over other theories that do tie meaning to ideas. When the question arises whether a certain word is meaningful, the test is to look for the ideas, if any, with which it is connected. We saw something of this sort before, but we can now achieve a formulation that is distinctly Humean: F. A word is meaningful if and only if it is correlated with ideas. We have thus a form of the empiricist criterion and also, in E, a treatment of meaning itself that follow known Humean ideas. As F suggests, to describe the ideas with which a word is correlated is not to produce its meaning, but to show that it satisfies a criterion justifying its use. The actual meaning of the word has more to do with the custom, and hence with the pattern or principle, that enables us to choose and call up the appropriate ideas. The identification of the meaning of a word with a custom, rather than with an idea, as in A and B, eliminates the error, or possibility of error through misinterpretation. that was the crux of the third objection to A and B. That is, it becomes clear that meaning is not a matter of an idea. The word "man," for example, does not have as its meaning this or that particular idea, and surely Hume introduced customs in order to avoid such a mistake. A word, a general word, at least, obviously is connected with a group of ideas. If the meaning were directly identified with such a group, little advantage would be gained, however, because, for one thing, the group of ideas actually called up is not the same for all men, whereas the custom is essentially the same for a given word. Two persons might imagine different sets of ideas on hearing the word "man," and if these were its meaning (we might have to say: its meaning for them), it would then have different meanings. But the custom that enables them to imagine these ideas is the same, in that in both cases it is the custom of imagining men, not just certain men. Necessarily, one thinks of only a few ideas on any occasion, according to one's degree of genius, but the custom connected with a word places the whole world of ideas at one's disposal, to pick out any that are suitable? Thus, the 8 9

See Flew, o p . cit., p. 23 tf.

That is the sense I make of Hume's fourth comment on customs, p. 23 f.

custom is general whereas the ideas we actually call up are private, varying, and few in number. One of the principal objections to A and B is that ideas are private.10 One cannot produce one's ideas for inspection, an argument goes, hence the doubt might always arise whether the ideas one associates with a word are conformable to those held by others. A custom of the mind, one might also object, is as personal and private as the ideas it enables one to call up, but there is the reply that a custom will be universal or nonprivate in so far as it is general. A custom of imagining men cannot be entirelv private, for it will necessarily follow a pattern based on the publicly observable resemblance between objects of that kind. However, we must here recall that Kemp Smith has objected to Hume's account of generality, in particular, to flume's use of resemblance. If Kemp Smith is correct, a larger question than can be settled here, Hume's failure will at least have been not in a simplistic identification of meaning with an idea, but in his attempt to introduce custom in order to overcome the deficiency of the ideational theory. In any case, it is proper to object that E employs only a psychological element, a custom or habit, while understanding the meaning of a word is rather a matter of obeying a rule with respect to the formation of ideas. A meaning or concept is a rule, as Kaflt observed. Hume mentions "general rules" and uses them in his analysis of causal inference," but there are no grounds for replacing customs with rules in E, if we are to keep to Hume. Nevertheless, while A and B are still a world away from the Kantian view, Hume's customs are the psychological counterpart of rules. It is certainly a testimony to Hume's genius that he can be reinterpreted along the lines that lead to E, rather than A or B. We can see how he moved toward a more correct view, while keeping within the bounds of empiricism.




See Flew, op. cit., chapter 11 passim. 173 ff.; compare p. 141 ff.

11 P.