. The analytic/synthetic distinction is first explicitly made by Kant. A proposition is analytic, on Kant’s view, if the predicate is covertly contained in the subject, as in ‘Roses are flowers’. A proposition where the predicate is attached to the subject but not contained in it is synthetic, as in ‘Roses are red’. The contradictory of a synthetic proposition is always synthetic whereas the contradictory of an analytic proposition is usually called ‘analytically false’. Kant’s distinction was partly anticipated by Leibniz, who distinguished ‘truths of reasons’ from ‘truths of fact’, and had the idea of containment, and by Hume, who distinguished ‘relations between ideas’ from ‘matters of fact’.
Kant’s distinction can easily be extended to conditional propositions, which are analytic if the consequent is contained in the antecedent, e.g. ‘If this is a rose, it is a flower’, and otherwise synthetic. Some other kinds of propositions raise difficulties, for instance, existential propositions like ‘There exist black swans’, where containment does not seem to apply, and the notion of containment is anyway hard to analyse. In general in ‘Red roses are red’ the containment is straightforwardly verbal. But in what sense precisely is the predicate ‘contained’ in the subject in ‘Roses are flowers’, or the consequent in the antecedent in ‘If all men are mortal and Socrates is a man, then Socrates is mortal’?
Because of this difficulty Kant himself proposed an alternative definition now often adopted: a proposition is analytic if its negation is, or is reducible to, a contradiction or inconsistency; otherwise the proposition is synthetic. A proposition which is true because it exemplifies a certain logical FORM, as ‘Bachelors are bachelors’ exemplifies the form ‘x’s are x’s’, can be called explicitly analytic. A proposition which is true because of certain definitions, as ‘Bachelors are male’ is true because of the definition of ‘bachelor’, is implicitly analytic or true by definition. Explicitly analytic propositions, and sometimes implicitly analytic ones too, can be called logically true or logically necessary.
A proposition like ‘Nothing is both red and green all over’ seems to be true in virtue of the meanings of the words involved, but not true by definition: ‘red’ is not defined in terms of ‘not green’, nor ‘green’, in terms of ‘not red’. This proposition therefore must be called analytic, if at all, in a sense even wider than that of ‘implicitly analytic’.
Recently the analytic/synthetic distinction has been attacked, especially by Quine, who argues that any clear account of the implicitly analytic would require notions like meaning, definition and synonymy, which themselves presuppose the implicitly analytic. He also alleges that the point of calling something analytic is to give a reason why is cannot be revised in the light of experience, and then claims that no statements are immune to such revision. Some statements are revisable with little effect on others (suppose ‘I see a cat’ is taken as true: it could be revised, i.e. rejected as false, by simply dismissing the experience as a hallucination). The rejection of other statements, such as the laws of logic, would profoundly affect our whole way of talking, but Quine thinks it is still possible. Scientific laws form an intermediate case. Thus Quine ends by saying that ‘analytic’ even in the narrow sense of ‘explicitly analytic’ cannot be applied absolutely, but at best as a matter of degree to those statements we are least willing to revise. Controversy still rages over this, especially concerning the laws of logic: is it simply that any sentence now expressing such a logical truth could one day change its meaning and fail to do so, or is there more to it than this? Cf. LOGIC (on deviant logics), PARACONSISTENCY.
The distinction has also been attacked, in a less fundamental way, by Waismann, who claims that it is not a sharp one, and that statements such as ‘I see with my eyes’ and ‘space has three dimensions’ cannot be unambiguously classified in accordance with it.
A further problem about the analytic/synthetic distinction, for those who accept it, is how it relates to the A PRIORI/empirical and necessary/contingent (see MODALITIES) distinctions. It is usually assumed that nothing can be both analytic and empirical, or both analytic and contingent, and in fact Kripke defines ‘analytic’ as what is both a priori and necessary, though he makes an important related claim (for which see A PRIORI). Kant, though he took ‘analytic’ in the wider sense, as ‘implicitly analytic’, treated analytic propositions as trivial and uninformative, like TAUTOLOGIES. He and others have claimed that the propositions of mathematics, etc., must be synthetic a priori, while logical positivists and others have vigorously denied that anything can be both synthetic and a priori. Often the synthetic a priori, which before Kripke was generally assumed to coincide with the synthetic necessary, is defended merely by interpreting ‘analytic’ in a narrow sense. Thus the issue at least partly depends on distinguishing senses of ‘analytic’ and giving reasons for preferring one to another. It is still disputed whether a substantial notion of synthetic a priori is needed for statements like ‘Nothing can be red and green all over’, or ‘If A exceeds B and B exceeds C then A exceeds C’; and also whether the laws of logic themselves can properly be called analytic. How too should we classify the statement itself that no synthetic statement is a priori? (Cf. POSITIVISM for the objection to the verification principle that it cannot account for its own status.)
Certain problems concern the relation between sentences and the statements they are used to make. Does ‘The fat cow which I see is fat’ make an analytic statement, although it apparently implies the synthetic statement that I do see a cow? And does ‘I exist’, since it cannot be uttered to make a false statement, make an analytic statement?
All the above must be distinguished from the question of the analytic and synthetic methods, deriving from Greek mathematics. See also SENTENCE.
T.Burge, ‘Philosophy of language and mind’, Philosophical Review, 1992. (See pp. 4–11 for three senses of ‘analytic’, with discussion.)
R.Descartes, Reply to Second Objections (to his Meditations), last few pages. (Analytic and synthetic methods.)
H.P.Grice and P.F.Strawson, ‘In defense of a dogma’, Philosophical Review, 1956. (Defence of analyticity against Quine. See also A.Sidelle, Necessity, Essence and Individuation: A Defense of Conventionalism, Cornell UP, 1989 (see its index).)
*J.F.Harris and R.H.Severens (eds). Analyticity, Quadrangle Books, 1970 (Readings. Includes Quine, Grice and Strawson, and bibliography.)
I.Kant, Critique of Pure Reason, Introduction, § 4.
S.Kripke, Naming and Necessity, Blackwell, 1980 (original version, 1972). (See especially pp. 39, 122, n. 63.)
H.Putnam, ‘The analytic and the synthetic’ in H.Feigl and G. Maxwell (eds), Minnesota Studies in the Philosophy of Science III, Minnesota UP, 1962, reprinted in his Mind, Language and Reality: Philosophical Papers vol. 2, Cambridge UP, 1975. (Claims that the distinction does exist but should not be overestimated.)
W.V.O.Quine, ‘Two dogmas of empiricism’, in From a Logical Point of View, Harper and Row, 1953, chapter 2.
*A.Quinton, ‘The a priori and the analytic’, Proceedings of the Aristotelian Society, 1963–4. (Distinguishes several senses of ‘analytic’ and rejects synthetic a priori for each of them.)
L.Resnick, ‘Do existent unicorns exist?’, Analysis, 23, 1963, pp. 128 ff. (‘Fat cow’ example. Cf. J.J.Katz, Linguistic Philosophy, 1972, pp. 146–73; pp. 156–7 claim analytic sentences are not always true.)
R.Robinson, ‘Analysis in Greek geometry’, Mind, 1936, reprinted in his Essays in Greek Philosophy, Clarendon, 1969. (Greek origins of analytic and synthetic methods).
F.Waismann, ‘Analytic-synthetic’ (in six parts). Analysis, 10, 11, 13, (1949–1953); reprinted in his How I See Philosophy, Macmillan, 1968.
This is the complete article, containing 1,264 words
(approx. 4 pages at 300 words per page).
