C. L. STEVENSON AND ETHICAL ANALYSIS ETHICS and Language by C L* Stevenson has been received with the interest proper for one of the few serious philosophical books published during the war, I value some of the discussions of specific points, of which the chapter on Intrinsic and Extrinsic Value is a good example. There are, however, some aspects of the main tenets of the book which to me are unsatisfactory and call for examination. The general aim of Stevenson's book has been well stated by A* Duncan-Jones. "The kind of theory Stevenson most wants to deny is . . ,what I propose to call the 'distinctive concept' theory.. .any theory to the effect that at least one concept is involved in the analysis of moral judgments which is not involved in the analysis of any other kind of judgment; and that this concept is some characteristic of the objects about which moral judgments are made, of motives, or of actions, or of results of actions."1 The most influential contemporary exponent of the distinctive concept theory is G. E. Moore, whose early work, Principia Ethica, has been much read. It is interesting, therefore, to rind that Stevenson acknowledges that he was at one time greatly under the influence of G. E. Moore's writings.2 Stevenson was much impressed by Moore's denunciation of the naturalistic fallacy. He saw the force of Moore's objection that none of the statements that have been offered as definitions of "good" were equivalent in meaning to "good." He was, however, not satisfied with Moore's explanation of why no analysis offered was satisfactory- He could not "identify the quality that is spoken of."3 He was not alone in this. It is safe to say that few philosophers would now maintain (as Moore once maintained) that we see that something is good in the same way that we see that an orange is yellow. Stevenson first expounded his explanation of the difficulty in defining "good" in a series of articles in Mind for 1937 and 1938. This book 1 Review of Ethics and Language in 2 Ethic's and Language^ p. 272. Mind, 1945, p, 372. *Ibid.f p, 10& 422 DISCUSSION adheres to the line laid down there with considerable expansion and elaboration. He maintains that the reason no attempt at definition can give a satisfactory equivalent for "good" is that "good" has an emotive meaning which is not easily duplicated. What Stevenson calls the descriptive meaning of a word or sentence he agrees can be defined, but the definition will not have the same emotive effect. Stevenson asserts that the emotive effect of a word is helpful in producing agreement in attitude on a matter. And it is agreement or disagreement in attitude that according to him is the main part of an ethical dispute. He holds that disagreement in attitude may occur even when there is complete agreement in belief. Stevenson in consequence proposes to describe ethical decisions in terms of agreement and disagreement in attitude- He further claims: "Any definition which seeks to identify the meaning of ethical terms with that of scientific ones* ♦ .without further explanation or qualification is extremely likely to be misleading."* This is because "it will suggest that the questions of normative ethics give rise to an agreement or disagreement that is exclusively in belief/*5 We shall have to consider later whether Stevenson is justified in describing ethical decisions solely in terms of agreement and disagreement in attitude. Stevenson has taken as the setting for his study the function of ethical statements in the transition from disagreement to agreement in ethical attitude. For interpreting ethical statements he first proposes what he calls the first pattern of analysis* Stevenson speaks of this as a pattern of analysis because it is "a pattern for making definitions, not a specific definition,"* "The first pattern deals not with any one ethical term, but with a great many of them. Even when it is illustrated in connection with one term, such as 'good/ there will be several alternative senses which must be recognized/*7 The first analysis in this pattern occurs in Chapter II. Stevenson introduces a definition of "this is good" as "synonymous with 'I approve of this; do as well.* *>s He does not claim that this is a completely satisfactory analysis of the sentence. It is, however, offered "as a working model or first approximation to analysis."* Later a definition of the first pattern is characterized as one in which "good" is assigned a descriptive meaning that refers to the speaker's favorable attitudes, and an emotive meaning that "may serve to evoke * Ibid., p. 20. 7 Ibid., pr &). * Ibid., p. 20. e Ibid., p. 8i. *Ibid.3 p. 89. 9lbid.t p. 81. 423 THE PHILOSOPHICAL REVIEW the favor of the hearer."10 The heart of the first pattern is the principle that "if anyone says that he approves of something, he introduces no new descriptive meaning by adding that it is good/'11 (It is possible that the belief in the powerful effect of "good" toward producing agreement in attitude stems more from the need of it to explain the difficulty of defining good than from psychological observation. It has been my experience that calling something good will do little to bring acquiescence even with children.) After a long discussion interpreting ethical disputes and other ethical theories in terms of his first patternt Stevenson provides "the second pattern of analysis," "The distinguishing features of the second pattern lie solely in the added descriptive meaning that it provides, "la The second pattern is described as one in which '* This is good' has the meaning of 'This has qualities or relations X, Yt and Z. ■ ; except that 'good' has as well a laudatory emotive meaning which permits it to express the speaker's approval, and tends to evoke the approval of the hearer,"13 II In judging the correctness of Steven son fs account we must first be clear what it is that he is trying to do. In this I follow Moore's famous remark that the difficulties and disagreements of philosophers "are mainly due...to the attempt to answer questions, without first discovering; precisely what question it is which you desire to answer,"14 I have remarked that Stevenson seems to be trying to answer questions phrased under the influence of Moore's writings* Can an analysis of "good" be given? If so, what is it? If not, why not? Now I am not at all satisfied with Stevenson's answers, I find that a large part of my dissatisfaction clusters around his use of the word analysis. In some passages of Stevenson it is possible to interpret the word analysis as having no more technical a meaning than that of the process of study (or the result of this process) employed on the situation in which ethical questions and statements are made. In this sense of the word it is the name of a process which may produce a variety of satisfactory analyses as end products for a single situation, It is a process which describes a situation in terms which lend emphasis to lüIbid., p. 90. ™ Prtiicipitt Ethica, p, vii. uIbid.r p, 105, lfl Cf. Ethics and Language, pp. 20, 13 Ibid., p. 206. 89, " Ibid., p. 207. 424 DISCUSSION certain aspects and ignore or minimize others. This usage is related to the use of the word in chemistry, e.g., Qualitative Analysis, and in mechanics, e.g,, Analysis of Forces as expressed in the Law of the Parallelogram of Forces. I shall call this analysis of situations. Now there is another sense of the word, found in the writings of G. E. Moore among others. In this sense the word is a name for a process (or the result of a process) directed toward the meaning of statements and the ideas, concepts, and propositions expressed in the statements, I shall call this analysis of meaning. What is distinctive about this sense of analysis is that complete equivalence of analysis and analys-andum is held as the goal of the analysis* In the more ordinary use (analysis of situation) equivalence for the purposes in hand is all that is required. Analysis of meaning is what classically is called definition, and in passages quoted above we find Stevenson using this word.}3 Stevenson explicitly recognizes that to be acceptable an analysis of the meaning of a sentence must be substitutable for the original, He states baldly: "A definiendum and its defmiens have the same meaning. And in urging the recognition of the importance of emotive meaning he says that his definition will not give complete equivalence, "These phrases fare no better than the working models."17 They "resemble 'This is good' in a rough way; but there is no situation in which they may replace the latter without changing emotive subtleties. The same is true of all other efforts to find an exact definition,"1* But this applies only to the emotive meaning, for if we leave emotive meaning out of the picture we find that the *'descriptive meaning may he defined, though not without complications of ambiguity and vagueness/'18 It appears then that Stevenson is not merely describing situations of ethical dispute and decision in terms of agreement and disagreement in attitude. He is also offering an analysis in these terms as the meaning or, more modestly, a correct meaning of ethical statements. Stevenson insists, however, that his interest is not "wholly in the study of language* "ao "We have the additional task of marking off and emphasizing the distinguishing aspects of ethical issues and methods."31 Yet he expects to employ the same two patterns of analysis in "the study of language" and in "marking of and emphasizing the distinguishing aspects of ethical issues and methods*" 36 Cf. Ethics and Language, pp. 20t 89. "/iirf., p. 82. 1T Ibid., p. 81. ™Ibid.f pp. Sc^a 425 ibid., p. Ibifl.t p. Ibid., pr 89. THE PHILOSOPHICAL REVIEW If the two patterns of analysis are intended to analyze the meaning of sentences, it would seem that such different sets of notions as those employed in the first pattern and those employed in the second pattern could not both produce a correct analysis of the meaning of the same token-statement* Yet he seems to think that the meaning analysis offers is in part a matter of choice by the analyst. In terms of analysis of meaning it is hard to see how Stevenson can justify such statements as: "A judgment may be analyzed as.. . . "22 ''This may be assigned a number of meanings.*'23 It is significant that he says "he assigned" rather than ;*in different contexts possess" In both of these passages it sounds as though Stevenson thought that there were several correct alternative analyses. We find him speaking of "any second-pattern analysts of an ethical judgment"34 rather than an analysis of a second^pattern judgment. He seems to believe that a given utterance may with propriety be analyzed in whichever of several ways you like. We find that he calls the "inadequacy** of his working models ^'only partial,*' because "they provide a meaning which can be assigned to the ethical terms and which is well suited to certain context s.*'SB But "they must be supplemented by a number of alternative definitions."2* And finally "definitions of all sorts must not be conceived as exhausting the possibilities of ethical language but only of revealing by example its enormous flexibility/'37 But this is not telling us what usages do prevail, but what usages might prevail Stevenson seems to alternate between two incompatible views of the relation of his patterns of analysis to each other. In some passages he speaks of them as not equivalent, as two classifications of different ways in which people use "good." "Two patterns are required because the ethical terms, as used in everyday life* are vague. Whenever a term is vague there is no sharp distinction between its strict descriptive meaning and what it suggests. If an analyst makes such a distinction, in the hope of bringing clarity to ordinary discussions, he will not do well to insist that the distinction must be made in one way, to the exclusion of all others. It is more important to understand the flexibilities of common words, and the varieties of meaning they may 'naturally* be assigned, than to msist on some one meaning that they should be given."2* And sometimes the meaning offered at the end Ibid., p. 93. Ibid., p. 56. Ibid., p. 207. Ibid., p. 36. w/foV£, p. 36. p. 36. M Ibid., p. 206 42(5 DISCUSSION of analysis is assumed to be the meaning of the speaker, "It will be assumed throughout that the speakers use ethical terms in accordance with the first patterns But far easier to find are passages where Stevenson regards the two patterns as alternative methods of correct analysis. At one point Stevenson says that "intrinsically good does not readily lend itself to a second-pattern definition."3* But on the next page he inclines the other way: "the same consideration appears in new phraseology. The difference between the patterns here as elsewhere is only of linguistic interest. It has no bearing on the nature of ethical disagreement or on the extent to which it can be resolved."31 "The patterns have been shown to be parallel in all central respects. We may conclude that the choice of one pattern in preference to another is a choice between forms of language."as "However we shall find that one pattern may often be more useful than another because of its comparative freedom from confusion and misunderstanding.'*33 Stevenson justifies his deviation from what he has accepted as the sum of analysis by the assertion that common usage is vague and ambiguous. He maintains that the job of analysis is to supply clarity where it does not exist. "We have always a choice of making its descriptive meaning rich or poor. And it is of great importance to realize that neither choice will violate the elastic requirement of * natural English usage.*ItAi "Vagueness of the ethical terms will be removed by limiting their descriptive reference to the speaker's own atti- Such is the typical situation with vague terms. When subject to definition or analysis, their meaning is seldom treated as a fait accompli (as might be the case when a scientific term is analysed for the benefit of a beginner) ; rather the meaning Is in the course of becoming what the analyst makes of it* And the analyst has a choice in deciding what it shall become, no matter how anxious he is to abide by common usage. As Wittgenstein once remarked: To remove vagueness is to outline the penumbra of a shadow. A sharp line fa there after we draw it, not before.3' Yet how does this justify Stevenson's statements? It is of course proper for the analyst or anyone else to try to change the world. But as analyst he is expected to analyze what the meaning is that the sentence does as a matter of fact have, however vague, ambiguous, or ill-suited that may be- Whatever else he may go on to do, he is expected 51 [hid.f p, 115, Ibid,, p, 242, tudes/'a& K Ibid,, pt 239. a Ibid., pt 340. B Ibid., p+ 241, *■ Ibid,; p. 36. *Ibid.t\\ So, 427 THE PHILOSOPHICAL REVIEW to do this first. And we rightly feel cheated if we are not given what we have reason to expect, but instead given something else because the analyst thinks it is better than what we asked for. Furthermore the point of Wittgenstein's remark was to warn against supposing that to remove vagueness and ambiguity in the interests of clarity was to analyze the meaning that the sentence had. Do T misinterpret Stevenson's words? I am certainly unfair to him, if I should be mistaken in supposing that he is intending to offer correct analyses of the meaning of sentences. His repeated statements that both patterns of analysis may be applied to the same sentence with satisfactory results would incline one to suppose that he was speaking of analysis of situation. Yet we have seen that he repeatedly claims to be analyzing the meaning of sentences. The contradictions and misapprehensions of the aim of analysis might be explained. Stevenson may not have been explicitly aware and attentive to the two kinds of analysis. As a result he may have said of one what is true only of the other, III But either pattern taken separately will not provide a correct analysis of meaning for many utterances. Let us examine the first pattern. It is quite true that when I say