¹ An exception is Professor Gustav Bergmann who, though confining philosophy to formal linguistic analysis, yet insists that such analysis yields features that show us something about the world. My only fundamental objection here is directed against his Wittgensteinian contention that this sort of thing cannot itself be said, that thus categorial analysis, though it actually reveals general features of the world, cannot itself include or investigate this claim.

² I think it is a too narrowly nominalistic view of experience that leads to such a denial of the empirical significance of categories as is to be found in W. R. Dennes' “The Categories of Naturalism” in Naturalism and the Human Spirit. Combined with this there is a failure to recognize that language can reveal what it does not assert.

³ By ‘value statements proper’ I mean declarative sentences with value-terms as predicates, such as, ‘Pleasure is good.‘

⁴ I have in mind Professor Charles L. Stevenson's Ethics and Language.

⁵ By Herbert G. Bohnert, in “The Semiotic Status of Commands”, Philosophy of Science, vol. 12, 1945, pp. 302–316.

^5a Or, at least his early position, as stated in Principia Ethica.

⁶ I believe that William K. Frankena's critism of Moore, in The Philosophy of G. E. Moore, amounts to a combination of these two objections. He says, “By regarding goodness as a simple intrinsic quality and defining ‘right’ as meaning ‘conducive to as much good as possible,’ Principia transforms statements of the form ‘We ought to do X’ into mere statements of fact” (p. 100). They are, that is, simply predictions that X will produce as much exemplification of a certain quality (good) as any alternative.

⁷ See for esample the bibliography given by Bohnert, Philosophy of Science, vol. 12, 1945, p. 315. One of the most promising articles in English is “The Logic of Value Imperatives,” by Thomas Storer, Philosophy of Science, vol. 13, 1946, pp. 25–40.

⁸ Op. cit., p. 41. He has more recently characterized this view as “silly.” This is a matter I shall not dispute with him. The point is that he was apparently unsatisfied with the position that good is a property.

⁹ P. 274.

¹⁰ P. 590.

¹¹ The closest approximation to either is the following statement: “Anyone who saw reason to doubt the existence of characteristics answering to our description of ‘non-natural’ might fairly use this conclusion to show that ‘good’ is not a name of a characteristic at all” (op. cit., p. 65). To clarify this another sentence must be quoted: “I propose to describe a ‘natural characteristic’ as any characteristic which either (a) we become aware of by inspecting our sense-data or introspecting our experiences, or (b) is definable wholly in terms of characteristics of the former kind together with the notions of cause and substance” (ibid., p. 62).

¹² Ibid., p. 66.

¹³ Op. cit., p. 283.

¹⁴ “A Suggested Non-naturalistic Analysis of Good,” Mind, N. S. vol. 48, 1939, pp. 1–22.

¹⁵ Loc. cit. p. 21.

¹⁶ Foundations of Ethics, p. 168. Lest we interpret ‘resultant attribute’ to mean an attribute whose possession is caused by the possession of another attribute, Ross immediately goes on to say, “No doubt there are causes which cause this patch to be yellow …; but I can perceive [it] to be yellow… without knowing anything of the causes that account for this.” It seems to me that ‘resultant’ is thus a poor word to express Ross's idea. ‘Dependent’ would perhaps be better.

¹⁷ Here is the ground, I think, for Ross's characterization of good as a “resultant property.”

¹⁸ I omit entirely an extremely interesting possibility. Can value comparatives (such as ‘Pleasure is better than pain,’ or ‘Take a walk rather than sitting there idle’) be adequately expressed in sentences of the type ‘f{a}’? If not, what form should such comparatives take, and can absolute valuatives be defined in terms of it? On the analysis given above, a value comparative cannot be simply a declarative with a value predicate of second (or higher) degree (referring to a value relation). The comparative element, since valuative, must enter the form of union of subject and predicate. I might suggest the following: ‘fg{a}’ to be read, ‘it were better than a be f than g.’ It might then be possible to define absolute valuatives in terms of comparatives, as follows: (f)(x)(f{x} = Df f̄f̄ {x}), where ‘f̄’ designates non-f.

¹⁹ On this point see Alf Ross, Philosophy of Science, vol. 11, 1944, p. 39.

²⁰ Mr. Thomas Storer, in a paper already referred to, shows by using both the matrix and the postulational techniques, how such a general logic, including both valuative and declarative connectives, could be developed.

²¹ To say that nevertheless they are really synonymous (granting that pleasure is the standard of good, and therefore that everything good is pleasant and everything pleasant is good) is to commit the “naturalistic fallacy.” See the excellent article by W. K. Frankena, “The Naturalistic Fallacy,” Mind, N. S., vol. 48, 1939, pp. 464–477. The decisive point is that ‘pleasure’ is a descriptive predicate, referring to an observable quality; ‘good’ is not, it indicates that we are in the valuative dimension.

^21a The connective, ‘if and only if,’ in this legitimacy condition is analogous to, but not identical with, the ‘if and only if of the truth condition. The latter connects two declaratives (albeit a semantical and an object-language declarative), whereas the former connects a declarative and a valuative. Simply as illustrative of the character of this analogy, I suggest the following as being true in a broad way to ordinary language (where ‘L’ is an abbreviation for ‘legitimate,’ ‘N’ for ‘non-legitimate,’ and ‘I’ for ‘illegitimate’):

²² The suggestion conies from W. E. Johnson's term, ‘characterizing tie’ (cf. Logic, vol. i, pp. 10 ff.), but there is no intention of following Johnson in holding that ties are simply forms of thought.

²³ In an article on “Value Propositions and Verifiability,” Journal of Philosophy, vol. 34, 1937, pp. 589–602, Wilbur M. Urban speaks of the “authentication” of an evaluative judgment, and says that it is quite different from the verification of a factual judgment. But beyond saying that values can be “shown forth” and thus experienced and acknowledged in their own right, Urban does not make authentication and its relation to verification clear.