"Synonymity" has been a major topic in philosophy since the publication of Rudolf Carnap's Meaning and Necessity in 1947, though it was discussed earlier in the writings of W. V. Quine and C. I. Lewis. After Quine and Morton White launched their attacks on the tenability of the analytic-synthetic distinction, around 1950, the two topics became closely linked.
Synonymity and the Analytic-Synthetic Distinction
Analytic statements, in Quine's account, fall into two classes. Those of the first class, exemplified by (1), are logically true.
(1) No unmarried man is married.
Quine has no objection to the notion of analytic truth as used here, for he has what he regards as an acceptable account of the notion of logical truth in terms of which the notion of analytic truth is partially explicated. "The relevant feature of this example is that it not merely is true as it stands, but remains true under any and all reinterpretations of 'man' and 'married.' If we suppose a prior inventory of logical particles, comprising 'no,' 'un-,' 'not,' 'if,' 'then,' 'and,' etc. then in general a logical truth is a statement which is true and remains true under all reinterpretations of its components other than the logical particles" (all quotations from Quine are from "Two Dogmas of Empiricism").
All logical truths are analytic. The problems that beset analyticity, however, concern those purported analytic truths which are not logical truths. These are typified by
(2) No bachelor is married.
This is not a logical truth, for it does not remain true under every reinterpretation of its nonlogical components, "bachelor" and "married." If (2) is nevertheless to be considered analytic, it is because we can turn it into the logical truth (1) by replacing synonyms with synonyms. Thus, since "bachelor" and "unmarried man" are synonyms, we may replace the former with the latter in (2) in order to arrive at (1), a truth of logic.
It might appear that a generalization of the above considerations would yield a satisfactory account of the notion of an analytic statement. The generalization would go as follows: a statement is analytic if and only if it either (1) is a logical truth or (2) is transformable into a logical truth by the substitution of synonyms for synonyms. This account is rejected by Quine and White on the ground that synonymity (or synonymy, as Quine prefers) is no clearer a notion than analyticity. In Quine's words, "We still lack a proper characterization of this second class of analytic statements, and therewith of analyticity generally, inasmuch as we have had in the above description to lean on a notion of 'synonymy' which is no less in need of clarification than analyticity itself."
Interchangeability Criterion of Synonymity
A natural response to Quine is that we can give an acceptable account of synonymity in terms of interchangeability. The suggestion is that the synonymity of two linguistic forms consists simply in their interchangeability in all contexts without change of truth-value—interchangeability, in Gottfried Wilhelm Leibniz's phrase, salva veritate. Benson Mates has offered an argument to show that if two expressions are synonymous they are interchangeable everywhere salva veritate. Following Gottlob Frege, Mates assumes that the meaning of a declarative sentence is a function of the meanings of the words which compose the sentence. Furthermore, two declarative sentences having the same meaning will necessarily have the same truth-value. It follows from these two assumptions that the replacement of a word in a sentence by another word synonymous with it cannot change the meaning of that sentence and hence cannot change its truth-value. Thus, if two words are synonymous they are interchangeable everywhere salva veritate.
In spite of the reasonableness of the above argument, the proposed interchangeability criterion soon runs into difficulty. Consider the synonymous pair "bachelor" and "unmarried man." The following statement is true:
(3) "Bachelor" has fewer than ten letters.
But the result of replacing the word bachelor by its synonym unmarried man is the false statement
(4) "Unmarried man" has fewer than ten letters.
This case can presumably be set aside on the ground that quoted expressions should themselves be understood as words functioning as names for their quoted contents. The interchangeability test is then interpreted as not applying to words such as bachelor when they appear as fragments of other words, such as "bachelor. " This makes the account of synonymity rest on the notion of wordhood, but Quine does not object on this account.
Perhaps Quine does not take seriously enough the difficulties involved here. Consider the synonymous pair "brothers" and "male siblings." Replacement of the former by the latter in
(5) The Brothers Karamazov is Dostoevsky's greatest novel
turns a true statement into one which is not true,
(6) The Male Siblings Karamazov is Dostoevsky's greatest novel.
Quine cannot object to this replacement for the same reason he objects to substitution of synonyms for synonyms within the context of quotation marks, for he cannot reasonably claim that titles are all single words.
The most serious problem connected with the interchangeability criterion is that the requirement is, apparently, too strong. Problems about wordhood aside, it is doubtful that paradigmatic synonym pairs like "bachelor" and "unmarried man" can pass the test. Consider the statement
(7) Jones wants to know whether a bachelor is an unmarried man.
Suppose it true, as it may well be, of some man named "Jones." Replacement of synonym for synonym here yields a statement that is no doubt false,
(8) Jones wants to know whether a bachelor is a bachelor.
Carnap's "Intensional Isomorphism"
Carnap intended the concepts of intensional isomorphism and intensional structure to be explications of the ordinary notion of synonymity. Intensional isomorphism is explained in terms of logical equivalence (L-equivalence) when the usual application of the latter notion is extended beyond full sentences to cover various sentence parts. For example, two names "a " and "b " are L-equivalent if and only if "a = b " is logically true (L-true). Two (one-place) predicate expressions "P " and "Q " are L-equivalent if and only if "(x )(Px ≡ Qx )" is L-true. (This means that it is L-true that whatever has the property P also has the property Q, and conversely.) An analogous definition extends the notion of L-equivalence to many-place predicates (expressions for relations). Expressions for which the relation of L-equivalence has been defined in this manner are called "designators." If two designators are L-equivalent they are said to have the same intension.
Intensional structure is explained thus: "If two sentences are built in the same way out of corresponding designators with the same intensions, then we shall say that they have the same intensional structure" (all quotations from Carnap are from Meaning and Necessity ). For example, consider the expressions "2 + 5" and "II sum V." These occur in a language S in which "2," "5," "II," and "V" are designations for numbers and "+" and "sum" signs for arithmetical operations. We suppose that according to the semantical rules of S, "2" is L-equivalent to "II" (and thus the two have the same intension), "5" is L-equivalent to "V," and "+" is L-equivalent to "sum." With regard to this example Carnap says, "…we shall say that the two expressions are intensionally isomorphic or that they have the same intensional structure, because they are not only L-equivalent as a whole, both being L-equivalent to '7,' but consist of three parts in such a way that corresponding parts are L-equivalent to one another and hence have the same intension." In our example corresponding parts correspond spatially, but this is not a necessary condition. Thus, Carnap regards "5 > 3" as intensionally isomorphic to "Gr(V,III)" because the (two-place) predicates ">" and "Gr" are L-equivalent and so are "5" and "V" and "3" and "III." The (two-place) predicates "correspond," regardless of their positions in the sentences. The sentence "(2 + 5) > 3" is intensionally isomorphic to "Gr(Sum(II,V),III)" because "2 + 5" is intensionally isomorphic to "Sum(II,V)" and the predicate expressions are L-equivalent, as are "3" and "III." On the other hand "7 > 3" is not intensionally isomorphic to "Gr(Sum(II,V),III)" even though "Gr" is L-equivalent to ">," "3" to "III," and "Sum(II,V)" to "7." They are not intensionally isomorphic because "Sum(II,V)" is not intensionally isomorphic to "7," although these expressions have the same intension (are L-equivalent). Intensional isomorphism of two expressions requires the intensional isomorphism of all corresponding subdesignators.
Consider Carnap's extension of the use of "≡" so as to hold between predicators. According to this extension, if Ai and Aj are two predicators of degree 1, the following abbreviation is allowable:
Ai ≡ Aj for (X )(AiX ≡ AjX ).
Now let us assume as L-true a sentence of the following form:
(1) Ai ≡ Aj.
This sentence will be intensionally isomorphic to
(2) Ai ≡ Ai.
But (1) is not intensionally isomorphic to
(3) (X )(AiX ≡ AiX ),
which is the definitional expansion of (2). Sentence (1) will not be intensionally isomorphic to (3), because (3) contains a designator, "(X )," which cannot be matched to a designator in (1). The point of this criticism is that an expression can be intensionally isomorphic to another expression without being isomorphic to a third expression which has the same meaning as the second according to a definition. For this reason intensional isomorphism seems not to be an adequate explication of synonymity.
In "A Reply to Leonard Linsky," Carnap says that the ordinary notion of synonymity is imprecise. He concludes that more than one explicans must be considered. He proposes a series of seven possible explicata, at least some of which would not be affected by the above criticism.
The most serious argument against Carnap's program is that of Benson Mates: Let "D " and "D ′" be abbreviations for two intensionally isomorphic sentences. Then the following are also intensionally isomorphic:
(1) Whoever believes that D believes that D.
(2) Whoever believes that D believes that D ′.
Now the following sentence is true:
(3) Nobody doubts that whoever believes that D believes that D.
But (4), which is intensionally isomorphic to (3), is very likely false:
(4) Nobody doubts that whoever believes that D believes that D ′.
If anybody even doubts that whoever believes that D believes that D ′, then (4) is false, and the consequence is that two intensionally isomorphic sentences will differ in truth-value. But since two synonymous sentences cannot differ in truth-value, it follows that intensional isomorphism is not adequate as an explication for synonymity.
According to Hilary Putnam, Carnap believes that his theory in its present form cannot refute Mates's criticism. However, other philosophers (notably Alonzo Church) disagree with Putnam and (apparently) Carnap over the soundness of Mates's argument.
One of the most widely discussed contributions to the topic of synonymity is Nelson Goodman's "On Likeness of Meaning." His view is particularly attractive to nominalistic philosophers who would avoid "abstract" entities, such as thoughts, senses, and meanings, in their semantical theories. Goodman proposes to explicate the notion of synonymity solely in terms of words and their "extensions"—the objects to which they apply. His account is confined to predicate expressions.
Suppose we say that two predicate expressions have the same meaning if and only if they have the same extensions—are true of the same things. A fatal objection to this view is that there are clear cases where two words have the same extension but do not have the same meaning. Centaur and unicorn, for example, have the same (null) extension, yet they differ in meaning.
We thus see that any simple identification of sameness of meaning of two expressions with sameness of extension must fail. But Goodman argues that we can still give an extensional account of sameness of meaning; although two words may have the same extension, certain predicates composed by making identical additions to these two words may have different extensions. Centaur and unicorn have the same (null) extension, but there are centaur pictures that are not unicorn pictures. Thus, "centaur picture" and "unicorn picture" have different extensions. Goodman concludes that "difference of meaning among extensionally identical predicates can be explained as difference in the extensions of certain other predicates. Or, if we call the extension of a predicate by itself its primary extension, and the extension of any of its compounds a secondary extension, the thesis is formulated as follows: two terms have the same meaning if and only if they have the same primary and secondary extensions." Suppose that in accordance with our nominalistic inclinations we exclude thoughts, concepts, attributes, meanings from the extensions under consideration. This means that when considering the identity of meaning of, for example, centaur and unicorn we will ignore such secondary extensions as those of "thought of a unicorn" and "thought of a centaur" or "concept of a unicorn" and "concept of a centaur." "If the thesis is tenable, we have answered our question by stating, without reference to anything other than terms and the things to which they apply, the circumstances under which two terms have the same meaning" (all quotations from Goodman are from "On Likeness of Meaning").
Let us see how Goodman's solution works. The predicates "(is the) morning star" and "(is the) evening star" have the same (primary) extension but differ in meaning. This difference is explained by Goodman as being due to a difference in the secondary extensions of these predicates. There are morning-star pictures that are not evening-star pictures and vice versa.
Now consider any predicates "P " and "Q. " Consider the actual ink marks which constitute any inscription of the phrase "a P that is not a Q. " Such an inscription will itself be part of the (secondary) extension of the predicate "P, " for it will be part of the extension of the expression "P -description." But no inscription of the phrase "a P that is not a Q " will be part of the extension of the expression "Q -description." It follows from this that "P " and "Q " have different (secondary) extensions and hence that they are not synonymous. Since "P " and "Q " are any predicate expressions, no two predicates are synonymous. For example, any inscription of the phrase "a centaur that is not a unicorn" will be part of the extension of the expression "centaur description," but it will not be part of the extension of the expression "unicorn description." Hence, "centaur" and "unicorn" have different secondary extensions (though they have the same primary extension), so they differ in meaning.
The discussions of the interchangeability criterion of synonymity and of Goodman's extensional criterion lead to the same radical conclusion. No two expressions are synonymous. Many philosophers regard this result as a reductio ad absurdum of the proposed criteria. Goodman seems to regard the result as a reductio ad absurdum of what is "commonly supposed" about synonymity. It is not clear whether he thinks that these views are commonly supposed only by the philosophers who discuss such questions or that they are held by those who in ordinary language sometimes declare two words to be synonymous. What is "commonly supposed," according to Goodman, is that (1) some predicates are synonymous with others and (2) synonymous expressions can replace each other "in all nonextensional contexts without change of truth-value."
Goodman holds that the two requirements are incompatible, and we can see why. "A P that is not a Q " is a P -description, not a Q -description; "a Q that is not a P " is a Q -description, not a P -description. On the supposition that "P " and "Q " are synonymous the following two statements have the same truth-value, if the interchangeability criterion is correct.
(1) "A P that is not a Q " is a P -description.
(2) "A P that is not a Q " is a Q -description.
However, the first statement is true and the second false. Thus, the predicates "P " and "Q " are not interchangeable everywhere, even in extensional contexts. But since "P " and "Q " are any predicates, no predicates are interchangeable everywhere. It follows from this that either no predicates are synonymous or synonymous predicates are not interchangeable everywhere.
In the face of this dilemma Goodman takes the alternative of declaring that "the relation of exact synonymy between diverse predicates is null." This is to say that no two predicates (or expressions of any kind, presumably) are "exactly synonymous." To many it has seemed more reasonable to abandon the interchangeability criterion. If no two expressions are synonymous or mean exactly the same thing, it is hard to see how the expressions "synonymous expressions" and "mean exactly the same thing" could have any currency in our language. Is it really credible that whenever we say two expressions are synonymous we are wrong? Is it not much more likely that the philosophers who discuss these issues have supposed that our concepts are governed by criteria which in fact do not apply? Consider a dictionary of synonyms. Is it credible that it is wrong in every entry because no two terms are synonymous? Surely not.
The above, or something like it, represents the response of the ordinary-language philosophers to the radical conclusions discussed in the earlier parts of this article. Such philosophers observe that a pair of terms may be regarded as synonymous "for certain purposes." This requires that they be interchangeable not everywhere but only in contexts relevant to the given discussion. It is wrong, these philosophers argue, to treat language as though it were a calculus governed by exact rules. But it is one thing to complain that the philosophers have distorted our actual use of the concept of synonymity and quite another to supply a careful and complete account of what that use is. Such an account remains to be given.
Several of the papers cited below are reprinted in Semantics and the Philosophy of Language, edited by Leonard Linsky (Urbana: University of Illinois Press, 1952).
Quine, W. V. "Two Dogmas of Empiricism." In From a Logical Point of View. Cambridge, MA: Harvard University Press, 1953. Ch. 2.
White, Morton. "The Analytic and the Synthetic." In Linsky, Semantics (see above). Ch. 14.
White, Morton. Toward Reunion in Philosophy. Cambridge, MA: Harvard University Press, 1956.
Mates, Benson. "Synonymity." In Linsky, Semantics (see above). Ch. 7.
Quine, W. V. "Notes on Existence and Necessity." In Linsky, Semantics (see above). Ch. 5.
Carnap, Rudolf. Meaning and Necessity. Chicago: University of Chicago Press, 1947.
Carnap, Rudolf. "A Reply to Leonard Linsky." Philosophy of Science 16 (4) (1949): 347–350.
Church, Alonzo. "Intensional Isomorphism and Identity of Belief." Philosophical Studies 5 (5) (1954): 65–73.
Church, Alonzo. "On Carnap's Analysis of Statements of Assertion of Belief." Analysis 10 (1950): 97–99.
Lewis, C. I. An Analysis of Knowledge and Valuation. La Salle, IL: Open Court, 1946.
Lewis, C. I. "The Modes of Meaning." In Linsky, Semantics (see above). Ch. 3.
Linsky, Leonard. "Some Notes on Carnap's Concept of Intensional Isomorphism and the Paradox of Analysis." Philosophy of Science 16 (4) (1949): 343–347.
Putnam, Hilary. "Synonymity and the Analysis of Belief Sentences." Analysis (April 1954): 114–122.
Sellars, Wilfrid. "Putnam on Synonymity and Belief." Analysis (1955): 117–120.
Goodman, Nelson. "On Likeness of Meaning." In Linsky, Semantics (see above). Ch. 4.
Goodman, Nelson. "On Some Differences about Meaning." Analysis (March 1953): 90–96.
Rudner, Richard. "A Note on Likeness of Meaning." Analysis 10 (1950): 115–118.
Thomson, James. "Some Remarks on Synonymy." Analysis 12 (1952): 73–76.
Rollins, C. D. "The Philosophical Denial of Sameness of Meaning." Analysis 11 (1950): 38–45.
Shwayder, David. "Some Remarks on 'Synonymity' and the Language of Semanticists." Philosophical Studies 5 (1) (1954): 1–5.
Leonard Linsky (1967)