Semantics and Abstract Objects Author(s): W. V.

Quine Reviewed work(s): Source: Proceedings of the American Academy of Arts and Sciences, Vol. 80, No. 1, Contributions to the Analysis and Synthesis of Knowledge (Jul., 1951), pp. 90-96 Published by: American Academy of Arts & Sciences Stable URL: http://www.jstor.org/stable/20023638 . Accessed: 07/08/2012 12:23
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at . http://www.jstor.org/page/info/about/policies/terms.jsp

.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected].

American Academy of Arts & Sciences is collaborating with JSTOR to digitize, preserve and extend access to Proceedings of the American Academy of Arts and Sciences.

http://www.jstor.org

AND ABSTRACT OBJECTS SEMANTICS

language. the other

of in the philosophy Let me begin by deploring two tendencies to confuse meaning with reference One is a tendency ; of the category to widen is a tendency excessively
terms.

to confuse meaning with reference is the one The tendency which Frege took pains to combat, though it was, even in Frege's than the other. There is a long tradition time, less widespread or be for the distinction between and denotation, connotation and this is a distinction tween intension and extension; squarely At the level of singular between meaning and reference. terms, and reference were less clearly distinguished. however, meaning that there has been a connection between Mill's doctrine Perhaps names no have and the but proper only reference, connotation, a of mistaken but inclination to the popular meaning identify proper name with the object named. has commonly been Another type of expression whose meaning is the statement. have confused with its reference Statements of propositions, been treated as names these latter frequently as entities of a sort better describable as mean being construed case of statements. in the of The confusion proper names ings are opposite; and that in the case of statements is the name of as meaning the named thought object, for lack of connota is its meaning, the statement of as naming tion, whereas thought to name. for want of anything better in all and reference Frege kept meaning separated sharply
cases. He construed

the class of all things of which and he con the term is true ? strued a statement as naming its truth value. On the other hand, to be the proposition, he took the meaning (Sinn) of a statement this being analogous to the intension or idea which is the meaning For singular of a general term. terms, likewise, he recognized distinct from the objects named. meanings, terms are themselves and statements Actually general singu lar terms for Frege, since he treats each of these as a name of a ? a class or a truth value. So Frege is a prime single entity ? of second I the to that tendency example promised deplore

a general

term

as

nammg

its

extension

SEMANTICS

AND

ABSTRACT

OBJECTS

91

of excessively me postpone
about meaning.

the category of singular terms. But let widening this matter until I have said what I have to say

the cleavage between meaning and reference is properly the of is what problems loosely called semantics become heeded, so fundamentally into two provinces distinct as not to separated at all. I shall call these the theory of deserve a joint appellation and the theory of reference. 'Semantics' would be a meaning name were the for of it not for the fact that good theory meaning, some of the best work in so-called semantics, notably Tarski's, be The main in the longs to the theory of reference. concepts theory of meaning, apart from meaning itself, are synonymy (or sameness of meaning), of meaning), significance (or possession of meaning). Another is (or truth by virtue and/analyticity or analyticity of the conditional. The main concepts entailment, in the theory of reference are naming, truth and denotation sense in the term in which the (or 'truth-of', 'green' denotes or is true of each green thing). or the class of is extension, Another denoted also the notion of values of variables. The objects; of reference is the locale of the so-called semantical theory such as Grelling's and Berry's and the Epimenides. paradoxes, The theory of meaning is not troubled with paradox, but it has troubles of a different order. The most conspicuous ques tion is as to the nature of its objects: what sort of things are intended to be ideas, somehow ??. They are evidently meanings? mental ideas for some semanticists, Platonic ideas for others. of either sort are so elusive, not to say debatable, that Objects there seems little hope of erecting a significant science about them. It is not even clear, granted meanings, when we have two and when we have one; it is not clear when expressions should be regarded as alike in meaning and when they should not. If a standard of synonymy should be arrived at, we may as entities will not reasonably expect that the appeal to meanings have played a very useful part in the enterprise. A felt need for meant entities may derive from an earlier to appreciate failure that meaning and naming are distinct. Once the theory of meaning is sharply from the separated as the busi it is a short step to recognizing theory of reference, ness of the theory of meaning of expres simply the synonymy of expressions, or and the analyticity sions, the meaningfulness of statements; meanings entailment as obscure inter themselves, This is a step that mediary entities, may well be abandoned.1 Frege did not take. When

92

QUINE

The notions of synonymy, meaningfulness, and en analyticity, in intimate ways. are synony tailment are connected Statements in a broad sense, if their biconditional or in is analytic, mous, other words if they entail each other. terms are synon Singular are synonymous if their identity is analytic. Predicates ymous are when to their if, they applied variables, universally quanti fied biconditional if is analytic. An expression is meaningful with itself. A statement is analytic if synonymous, synonymous in our broad sense, with some arbitrarily chosen specimen ? ' = in tying this well say (x) (x x) \ But there is great difficulty knit group of concepts down to terms that we really understand. even with the elimination The theory of meaning, of the mysteri ous meant entities, strikes me as in a comparable state to theology ? are blithely but with the difference that its notions used in the supposedly most and hard-headed brands of phi scientific losophy. As an empiricist I consider of that the cognitive synonymy statements should consist in sameness of the empirical condi tions of their confirmation. A statement is analytic when its condition of verification operational is, so to speak, the null con dition. But I am using terms here which we cannot pretend to understand until we have made in the substantial progress what good hands the prob theory of confirmation. Considering lem of confirmation is in, it may seem that optimism is in order. if synonymy and analyticity and the rest are to be However, made sense of only in terms of an eventual account of confirma are tending to be insufficiently tion, I think philosophers chary of the circularity in resting of involved their eventual account as confirmation such and upon synonymy concepts analyticity. So much for meaning. Now let me return to the theory of to deplore the practice and others of exces reference, by Frege the category of terms that name. Once the sively widening of is well from there that of theory meaning separated naming, is no need to seek a named object in order to insure meaningful ness. I prefer to consider that naming of singu is the function lar terms in the original narrow terms sense, excluding general and statements. terms and statements General have referential of their own, but of a different functions ; type than meaning a general term is true of this and that object and false of this and that object, and a statement is true or false. A general term has its extension, viz. the class of all things of which it is true, and a statement has its truth value ;but there is no need to treat as the general term as a name of its extension, nor the statement

SEMANTICS

AND

ABSTRACT

OBJECTS

93

can be formed for classes and a name of its truth value. Names else ; but I prefer not to regard for truth values, as for anything as names at all. On this point, the general terms and statements no means I in agree with Carnap's Meaning general, though by and Necessity. out of to keep general terms and statements I want Why the category terms will be clearer later. Meanwhile of singular I shall in principle, consists. let us consider wherein, naming assume that we use a language is funda in which quantification as mental. recognize Clearly, then, the things which we must to be true, are just the there being, in order for our discourse the range or ranges of our vari things which must fall within we affirm in order that the statements ables of quantification come out true. This is obvious from the meaning of the quanti fiers : 'each object x is such that', 'there is something x such that'. It follows that a term need be regarded as a name of something for the variable of a substitutable only if the term is always to false from without truth universal leading quantification,2
hood or nonsense.

treats general terms as names of classes To say that Frege as names of truth values, then, is simply to say and statements of general terms for quantified for substitution that he provides for quantified of statements and substitution class variables
truth-value variables.

of this procedure is that it leads to a false disadvantage of our of most of the ontological accounting presuppositions we say that some dogs are black, (rx) discourse. When (x is a admit some black dogs into our dog . x is black), we explicitly to such abstract entities as universe but do not commit ourselves or to class of hence it is misleading black the things; dogkind construe the words 'dog' and 'black' as names of such entities. in discourse which quanti Classes are genuinely presupposed call quite There are occasions which fies over class variables. ; e.g., where we want to say that directly for this sort of discourse some zoological is where we Another species are cross-fertile. want to define ?ancestor in terms of parent, by Frege's method: x is ancestor of y if x belongs to every class which contains y There is thus serious motive and all parents of its own members. over classes ; and, insofar, there is a place for for quantification such singular terms which name classes ? terms as singular of "the class ancestors". let and But Napoleon's "dogkind" us not read such terms into positions where no appeal to classes is involved. The

94 I would distinguish, (Qx) (Rx) (2) (x

QUINE

therefore, is-a-dog

between . x

: and (1) (2)

is-black)

(x s dogkind

. x s class-of-black-things).

is a case of the form: (3x) (x y x e z), or:

x e ?). (Rx) (x e a " " The "dogkind" and in (2) are singular class-of-black-things in positions to variables terms, capable of occurring appropriate of quantification. On the other hand the general terms or predi
cates "is-a-dog" and "is-black", in (1), are not to be regarded

as meaningfully I would substitutable for variables. represent the form of (1) merely in which by a schema '(Rx) (Fx . Gx)\ the 'F' and 'G' are purely diagrammatic and do not admit of quantification. I grant that (1) and (2) as wholes are equivalent statements. to the part of language But they differ in that (1) belongs on the question which is neutral of class existence, whereas (2) is a degenerate of the It specimen part of language. higher as a whole to be equiv is degenerate merely in that it happens alent to (1). The bulk of logical reasoning takes place on a level which does not presuppose abstract entities. Such reasoning proceeds can be rep the laws of which by mostly quantification theory, over class resented through schemata involving no quantification of what Much is commonly variables. in terms of formulated the algebra of classes or the algebra of relations actually needs
no such things as classes and relations, and can be easily reformu

within schematically quantification theory plus perhaps theory. identity Thus I consider in an all-purpose formulation of it a defect the formulation the theory of reference repre if, like Frege's, to abstract entities from the very beginning sents us as referring rather than only where there is a real purpose in such refer ence. to wish terms from Hence distinct my keep general terms. abstract singular en I am not here arguing of abstract over-all repudiation In discussing I did urge the use the theory of meaning tities. as entities. lessness of meanings My thesis there was, in effect, over such alleged in quantifying that no gain is to be sought
entities as meanings. But quantification over classes or other

lated

abstract

entities

is the essence

of set theory

and

indeed

of most

SEMANTICS

AND

ABSTRACT

OBJECTS

95

as well of classical mathematics ; and mathematics is, figuratively as literally, to we reckon with. I that urge only something between that put no obstacles in the way of distinguishing clearly on an ontology which of part of discourse depends squarely does not. abstract entities and that part which In depriving terms or predicates of the virtues of general names I do not thereby deny that there are certain classes con than in the fashion of being named. nected with them otherwise On the contrary, in that part of the theory of reference which has to do with classes there is occasion to speak of the extension ? the class of all things of which of a general term or predicate is true. There is occasion in that part of the theory the predicate of reference to assume such classes and to quantify over them. One such occasion arises when in the theory of reference we treat the topic of validity of schemata of pure quantification theory; for a quantificational sch?ma is valid when it comes out true for all assignments of classes as extensions of the schematic predi cate letters. The general of quantificational theory validity thus appeals to classes, but the individual statements represented by the schemata of quantification theory need not ; the statement of itself, no appeal to involves, '(3?) (x is a dog . x is black)' the abstract extension of a predicate. there is occasion in the theory of validity to speak Similarly of truth values of statements, truth-functional e.g., in defining But there is no need to treat statements as names validity. of these values, nor as names at all. When we simply affirm a statement we do not thereby appeal to any such entity as truth to have that special subject values, unless the statement happens
matter.

It can indeed prove

convenient
as names

and
?

elegant
perhaps

in special
of 0 and

systems
1, as in

Church's This is perhaps as a matter better regarded system. of making names of 0 and 1 serve the purpose of statements, for the special system; and I have no quarrel with it. Similarly as making his singular Frege may be represented terms, plus do the work of general terms ; and with this again, membership, as a means merely of absorbing lower logic into a particular system of higher logic for the sake of elegance, there is no quar I object to is projection rel. What of these special expedients into the general theory of reference. It is preferable in the gen eral theory of reference to proceed in such a way as not to impute to portions of discourse which special ontological presuppositions are innocent of them.

96

QUINE

in the higher theory of validity Even it happens that the ap and extensions of predicates peal to truth values of statements can finally be eliminated. can be For, truth-functional validity test of computation, in redefined and validity by the tabular can redefined to be the quantification theory simply by appeal rules of proof is Here (since G?del has proved them complete). a good example of the elimination of ontological presuppositions, domain. in one particular In general it is important, I think, to show how the purposes of a certain can be met with a reduced of mathematics segment to show how an erstwhile non just as it is important ontology, can be accomplished constructive proof in mathematics by con The interest in progress structive means. of this type is no an out-and-out more intolerance of abstract upon dependent an out-and-out than it is upon of non entities intolerance our The important constructive thing is to understand proof. on to tab the diverse of diverse instrument; keep presuppositions of our theory, and reduce them where we can. It is portions to discover, thus that we shall best be prepared the eventually, that has always over-all of some assumption dispensability rankled as ad hoc and unintuitive.
iWhen Unity months ciation, peared reading '' Two in of I presented at the meeting of the Institute for this paper the I had no thought in April, of publication. Science 1950, Eight Asso of the American Philosophical later, at the Toronto meeting and of empiricism,'' it ap I presented "Two my paper dogmas in in January, I now in the Philosophical Review observe 1951. proof the above nineteen that '' in pp. 22 f, though lines the main recurred the two almost papers verbatim are unlike in

dogmas, content.

2With

appropriate

type

of

variable,

in case

there

are

distinctions.