¹ Logical syntax and semantics, Lg. 30.230-7 (1954).

² I want to make it clear that the remarks which follow are critical not of logical syntax and semantics as such, but of the claim that these disciplines furnish solutions to linguistic problems. I have borrowed freely from various critical accounts of the theory of meaning, including W. V. Quine, From a logical 'point of view, esp. Ch. 2, 7, 8 (Cambridge, Mass., 1953); M. G. White, The analytic and the synthetic: An untenable dualism, John Dewey: Philosopher of science and freedom 316-30 (New York, 1950), reprinted in Linsky, Semantics and the philosophy of language 272-86 (Urbana, 1952).

³ I will not discuss these proposals separately, since, as I will try to show, they fail for the same reasons.

⁴ Thus, for Bar-Hillel no doubt the proper interpretation of this relation will be such that Mary was seen by John follows from John saw Mary, but not This gas was at such-and-such a temperature from This gas was at such-and-such a pressure, or This dry twig will burn from This dry twig will be thrust into the fire. Similarly, from Socrates is a man will follow Socrates is a rational animal, but perhaps not Socrates is a biped or Socrates has a finite number of arms. The transformation rules of logic are rules of valid inference; hence in this discussion the terms ‘inference’ and ‘transformation’ will be used interchangeably.

⁵ Note that any behavioral definition of synonymy will probably yield paradoxical results with respect to such notions as logical truth, analyticity, and synonymy itself. It is not hard to imagine the result of asking ‘ordinary people’ what is the status of a sentence like There are as many even integers as there are integers altogether, or There are uncountably more real numbers than rational numbers, but between any two real numbers there is a rational number, or simply a tautology too complex to be readily grasped.

⁶ See Quine, op.cit., for a discussion of Carnap's systems and a general discussion of synonymy and related notions.

⁷ Thus Carnap points out that it is purely a matter of convention whether only ‘logical laws’ or also ‘physical laws’ are given among the transformation rules of a language. For any given language there is no general way of deciding how to differentiate between the two.

⁸ Carnap, Meaning postulates, Philosophical studies 3.65 ff. (1952). Incidentally, Carnap here (66) explicitly states his present view that the explications which he is developing ‘refer to semantical language-systems, not to natural languages ... the problems of explicating concepts of this kind for natural languages are of an entirely different nature.‘

⁹ Even if true, this would be a minor point for linguistics; for the linguist's problem is not the size of his ad-hoc lists but their arbitrariness.

¹⁰ N. Goodman, The problem of counterfactual conditionals, Journal of philosophy 44.113 ff. (1947), reprinted in Semantics and the philosophy of language.

¹¹ The last case has some specific linguistic interest. Though in the cited case the expressions are ‘syntactically identical’, in other cases they will differ. Thus tall and thin are equally stressed in a tall and thin man, but black is more heavily stressed than white in a black and white flag (as if black-and-white were here a single word).

¹² Journal of philosophy 40.113 ff. (1943), reprinted in Semantics and the philosophy of language.

¹³ Philosophical review 60.22 ff. (1951), reprinted almost without change as Ch. 2 of From a logical point of view.

¹⁴ I am not arguing that a model cannot be interesting unless it is a special case, or measurably a limiting case, of something real; but Carnap, by this analogy, implies that it is in this strong sense that artificial languages are necessary. Such a view, I have argued, is incorrect. A general argument ‘for or against’ models is pointless. A model must prove its worth: it must analyze, not assume, some interesting notions, and it must be applicable in a reasonably unambiguous way.

¹⁵ I. Scheffler, On synonymy and indirect discourse, to appear in Philosophy of science. A special case of the same point was discussed by Benson Mates, Synonymity, University of California publications in philosophy 25.201-26 (1950), reprinted in Semantics and the philosophy of language 111-36.

¹⁶ For instance, Goodman's account, in The Structure of appearance (Cambridge, 1951), of the notion ‘all alike’, or ‘at a time at a place’ as opposed to ‘at a time and at a place’; or Austin's account of ‘call’, ‘describe’, and ‘match’, in How to talk: Some simple ways, Proceedings of the Aristotelian Society 53.227 ff. (1952/53). Although these and other particular semantic studies, whether in the form of models or not, may have some interest for linguists, they do not clarify the general notion of meaning or synonymy, and they provide no grounds for demonstrating that the conclusions at which they arrive are correct descriptions of the way in which language is used. Hence they have no direct bearing on the program of developing an adequate linguistic theory.