The most distinctive doctrine of the logical positivists was that for any sentence to be cognitively meaningful it must express a statement that is either analytic or empirically verifiable. It was allowed that sentences may have "emotive," "imperative," and other kinds of meaning (for example, "What a lovely present!" or "Bring me a glass of water!") even when they have no cognitive meaning, that is, when they do not express anything that could be true or false, or a possible subject of knowledge. But—leaving aside sentences expressing analytic statements—for a sentence to have "cognitive," "factual," "descriptive," or "literal" meaning (for example, "The sun is 93 million miles from the earth") it was held that it must express a statement that could, at least in principle, be shown to be true or false, or to some degree probable, by reference to empirical observations. The iconoclasm of the logical positivists was based on this criterion of meaning, for according to the verifiability principle a great many of the sentences of traditional philosophy (for example, "Reality is spiritual," "The moral rightness of an action is a nonempirical property," "Beauty is significant form," "God created the world for the fulfillment of his purpose") must be cognitively meaningless. Hence, like Ludwig Wittgenstein in the Tractatus Logico-Philosophicus, they held that most of the statements to be found in traditional philosophy are not false but nonsensical. The verifiability principle, it was maintained, demonstrates the impossibility of metaphysics, and from this it was concluded that empirical science is the only method by which we can have knowledge concerning the world.
The verifiability principle stands historically in a line of direct descent from the empiricism of David Hume, J. S. Mill, and Ernst Mach. It has some affinities with pragmatism and operationalism, but it differs from them in some important respects. Pragmatism, as presented by C. S. Peirce, William James, and John Dewey, is the view that the "intellectual purport" of any symbol consists entirely in the practical effects, both on our conduct and on our experiences, that would follow from "acceptance of the symbol." This view, unlike the verifiability principle, makes the meaning of a sentence relative to certain human interests and purposes and to the behavior adopted for the realization of these purposes. Operationalism, as held by P. W. Bridgman and others, is the view that the meaning of a term is simply the set of operations that must be performed in order to apply the term in a given instance. Thus, according to this view, the meaning, or rather a meaning, of the term length is given by specifying a set of operations to be carried out with a measuring rod. Moritz Schlick and other logical positivists sometimes said that the meaning of a sentence is the method of its verification. But, unlike the advocates of operationalism, they meant by "the method of verification" not an actual procedure but the logical possibility of verification. The verifiability principle had among its immediate antecedents Schlick's Allgemeine Erkenntnislehre (Berlin, 1918) and Rudolf Carnap's Der logische Aufbau der Welt (Berlin, 1928). It was first formulated explicitly by Friedrich Waismann in his "Logische Analyse des Wahrscheinlichkeitsbegriffs" (1930) and subsequently by Schlick, Carnap, Otto Neurath, Hans Reichenbach, Carl Hempel, A. J. Ayer, and other logical positivists in numerous publications.
Problems Raised by the Principle
The controversial questions concerning the principle are: (1) What is it to be applied to—propositions, statements or sentences? (2) Is it a criterion for determining what the meaning of any particular sentence is, or is it simply a criterion of whether a sentence is meaningful? (3) What is meant by saying that a statement is verifiable, or falsifiable, even if in practice it has not been, and perhaps cannot be, verified, or falsified? (4) What type of statement directly reports an empirical observation, and how do we ascertain the truth-value of such a statement? (5) Is the principle itself either analytic or empirically verifiable, and if not, in what sense is it meaningful? (6) Is the question that the principle is intended to answer (that is, the question "By what general criterion can the meaning or the meaningfulness of a sentence be determined?") a logically legitimate question?
what is the principle to be applied to?
In some of the earlier formulations of the verifiability principle it is presented as a criterion for distinguishing between meaningful and meaningless propositions. However, in an accepted philosophical usage, every proposition is either true or false, and hence a fortiori a proposition cannot be meaningless. To meet this point some of the later exponents of the principle say that a grammatically well-formed indicative sentence, whether it is cognitively meaningful or not, expresses a "statement"; the term proposition is retained for what is expressed by a cognitively meaningful sentence—that is, propositions are treated as a subclass of statements. The verifiability principle is then presented as a criterion for distinguishing between meaningful and meaningless statements. This procedure, however, presupposes a usage for "cognitively meaningful sentence," and indeed it is sentences that are normally said to be meaningful or not. Consequently, in still other formulations the principle is presented as applying directly to sentences; the objection to this is that sentences are not normally said to be true or false, and hence they are not said to be verifiable or falsifiable.
In order to meet these difficulties, sentences, statements, and propositions may be distinguished in the following way: A sentence, as we shall understand it, belongs to a particular language, it is meaningful or not, but it is not properly said to be true or false, or to stand in logical relations to other sentences, or to be verifiable or falsifiable. A statement is what is expressed in certain circumstances by an indicative sentence, and the same statement may be expressed by different sentences in the same or in different languages; a statement is properly said to be true or false, it does stand in logical relations to other statements, and it is verifiable or falsifiable. What can or cannot be said of statements applies equally to propositions, except that a proposition cannot be meaningless, that is, it cannot be expressed by a meaningless sentence.
For convenience we shall sometimes speak of sentences as being verifiable or not, and of statements as being meaningful or not. But, more strictly, we shall understand the verifiability principle as claiming that the cognitive meaning or meaningfulness of a sentence is to be determined by reference to the verifiability (or falsifiability) of the statement expressed by the sentence.
a criterion of meaning or meaningfulness?
The earliest presentations of the verifiability principle identified the meaning of a sentence with the logical possibility of verifying the corresponding statement, and apparently, in the last analysis, with the occurrence of certain experiences. This has some initial plausibility in the case of "empirical sentences," that is, sentences containing, apart from nondescriptive expressions, only empirical predicates (for example, "red," "round," "middle C"). An empirical predicate is, by definition, one that stands for a property that can be observed or experienced. Consequently, in the case of such a sentence as "This is red," there is a natural tendency to say that the meaning of the sentence is given by the experience that would verify it. The meaning is understood by anyone who can use the sentence for the purpose of identifying red objects when he sees them and cannot be understood by anyone who cannot identify red objects. It might be argued that a congenitally blind person could be said to understand the sentence "This is red" if he were able to identify red objects in some other way, by touch, for example. But in that case, an early adherent of the verifiability principle might reply, the predicate "red" has, for the person in question, not a visual but a tactual meaning. Our ability to understand empirical predicates, he might say, is plainly restricted by our capacity for sensory discrimination. For example, a person may be able to give a verbal definition of "C ♭" as "the note midway between the notes designated by 'C' and 'C ♯'"; but there is an important sense in which he does not know what "C ♭" means if he is not able to discriminate quarter tones. It may be fairly objected, however, that this argument rests on the ambiguities of the words meaning, stands for, and designates ; for example, the sense in which a term may be said to have a "tactual meaning" if it designates something tactual is not the sense in which a sentence may have a "cognitive or factual meaning." Moreover, it cannot be correct to identify the meaning of a sentence with the experiences that would verify it, for the characteristics that can be appropriately attributed to an experience cannot be appropriately attributed to the meaning of a sentence, nor conversely—for example, the meaning of a sentence does not occur at a particular time or with a certain intensity, as does an experience. And finally, if the meaning of a sentence were identified with the experiences of a particular person, the verifiability principle would result in a radical form of solipsism.
To meet these objections some other early formulations of the principle identified the meaning of a statement with that of some finite conjunction of statements directly reporting empirical observations. As will appear in more detail later, there are two main replies to this: (1) there are many types of statement whose meaning is not equivalent to that of any finite conjunction of observation statements, and (2) to identify the meaning of one statement with that of another is simply to say that the two statements have the same meaning, and this is not to explain or to give the meaning of the original statement.
For the foregoing reasons, it cannot be held that the verifiability principle is a criterion for determining the meaning of any particular sentence. In its later formulations it is presented simply as a criterion for determining whether a sentence is cognitively or factually meaningful.
In their early formulations Waismann, Schlick, and others held that the cognitive meaning of a sentence is determined completely by the experiences that would verify it conclusively. According to Waismann, for example, in "Logische Analyse des Wahrscheinlichkeitsbegriffs," "Anyone uttering a sentence must know in which conditions he calls the statement true or false; if he is unable to state this, then he does not know what he has said. A statement which cannot be verified conclusively is not verifiable at all; it is just devoid of any meaning." This was sometimes called the requirement of "strong verifiability." It says, in effect, that for any statement S to be cognitively meaningful there must be some finite consistent set of basic observation statements O 1 · · · On, such that S entails and is entailed by the conjunction of O 1 · · · On. The principal objections to this requirement are: (1) a strictly universal statement, that is, a statement covering an unlimited number of instances (for example, any statement of scientific law), is not logically equivalent to a conjunction of any finite number of observation statements and hence is not conclusively verifiable; (2) any singular statement about a physical object can in principle be the basis of an unlimited number of predictions and hence is not conclusively verifiable; (3) statements about past and future events, and statements about the experiences of other people, are not conclusively verifiable; (4) even if an existential statement (for example, "Red things exist" or "At least one thing is red") is verifiable in the required sense, its denial cannot be verifiable in this sense, for its denial (for example, "Red things do not exist" or "Everything is nonred") is a strictly universal statement. Hence, the requirement of strong verifiability would have the strange consequence that the denial of an existential statement would never be meaningful, and this would involve the rejection of the fundamental logical principle that if a statement S is true, then not-S is false, and that if S is false, then not-S is true; (5) if a statement S is meaningful by the present requirement and N is any meaningless statement, then the molecular statement S or N must be meaningful; (6) the present requirement presupposes that observation statements are conclusively verifiable, for unless this is so, no statement at all, not even a statement that is logically equivalent to a finite conjunction of observation statements, will be conclusively verifiable—or cognitively meaningful.
It was sometimes suggested that conclusive falsifiability rather than conclusive verifiability should be the criterion of a cognitively meaningful statement. The criterion of conclusive falsifiability says, in effect, that a statement S is meaningful if and only if not-S is conclusively verifiable. Consequently, objections analogous to those already considered still apply: (1) existential statements are not conclusively falsifiable, for if S is an existential statement, not-S is a strictly universal statement; (2) even if a universal statement is conclusively falsifiable, its denial is not conclusively falsifiable, since its denial is an existential statement. Hence, the present criterion would have the consequence that the denial of a universal statement would never be meaningful, and again this would involve the rejection of the fundamental principle of logic mentioned before; (3) the present criterion is open to the special objection that a universal statement (for example, "Whatever is pure water boils at 100° C.") would be meaningful, that is, conclusively falsifiable, only if the corresponding negative existential statement (for example, "There is an instance of pure water that does not boil at 100° C.") were assertable, and a fortiori meaningful; but this negative existential statement would be meaningful, that is, conclusively falsifiable, only if the corresponding universal statement were assertable, and a fortiori meaningful. To escape from this circle it would be necessary to have a different and independent criterion of significance for either universal or existential statements; (4) if S is meaningful by the present requirement and N is any meaningless statement, then S and N must be meaningful; (5) again, the present requirement presupposes that basic observation statements are conclusively verifiable.
To meet the preceding difficulties the later formulations of the verifiability principle require of a meaningful statement that it should be related to a set of observation statements in such a way that they provide not conclusive verifiability but simply some degree of evidential support for the original statement. This was sometimes called the requirement of "weak verifiability." It says that for any statement S to be cognitively meaningful there must be some set of basic observation statements O 1 · · · On such that S entails O 1 · · · On and that O 1 · · · On confirms, or gives some degree of probability to, S. A formulation of this kind was given by Ayer in the first edition of Language, Truth and Logic (1936). He held that a statement is verifiable, and hence meaningful, if one or more observation statements can be deduced from it, perhaps in conjunction with certain additional premises, without being deducible from these other premises alone. The qualification concerning additional premises is introduced to allow, among other things, theoretical statements in science to be verifiable.
But this formulation, as Ayer recognizes in the second edition of his book, permits any meaningless statement to be verifiable. For if N is any meaningless statement and O some observation statement, then from N together with the additional premise if N then O the observation statement O can be deduced, although O cannot be deduced from the additional premise alone. To meet objections of this kind Ayer introduces a number of conditions; he says (1) "a statement is directly verifiable if it is either itself an observation-statement, or is such that in conjunction with one or more observation-statements it entails at least one observation-statement which is not deducible from these other premises alone," and (2) "a statement is indirectly verifiable if it satisfies the following conditions: First, that in conjunction with certain other premises it entails one or more directly verifiable statements that are not deducible from these other premises alone; and secondly, that these other premises do not include any statement that is not either analytic, or directly verifiable, or capable of being independently established as indirectly verifiable."
These conditions are designed inter alia to prevent obviously meaningless statements from being verifiable simply by occurring as components of verifiable molecular statements as in the objection to the requirement of strong verifiability (see above), and the objection to the requirement of conclusive falsifiability. The conditions are, however, insufficient for this purpose. As Hempel remarks, according to the present formulation if S is meaningful, then S and N will be meaningful, whatever statement N may be. And Alonzo Church has shown that given any three observation statements O 1, O 2, and O 3, no one of which entails either of the others, and any statement N, it is possible to construct a molecular statement from which it follows that either N or not-N is verifiable. Such a molecular statement is one of the form (∼O 1 · O 2) ∨ (O 3 · ∼N ). For (∼O 1 · ∼O 2) ∨ (O 3 · ∼N ) together with O 1 entails O 3, and so the molecular statement is directly verifiable; but N together with (∼O 1 · O 2) ∨ (O 3 · ∼N ) entails O 2, and therefore N is indirectly verifiable. Alternatively, (∼O 1 · O 2) ∨ (O 3 · ∼N ) may by itself entail O 2, and in that case ∼N and O 3 also entail O 2, and therefore ∼N is directly verifiable.
Difficulties of the kind raised by Hempel and Church obtain when a component of a molecular statement is superfluous as far as the verifiability of the molecular statement is concerned, that is, when the inclusion or exclusion of the component makes no difference to the verifiable entailments of the molecular statement. To eliminate components of this kind, R. Brown and J. Watling have proposed that for a molecular statement to be verifiable, either directly or indirectly, it must contain "only components whose deletion leaves a statement which entails verifiable statements not entailed by the original statement, or does not entail verifiable statements entailed by the original statement." This stipulation is designed to ensure that every component of a verifiable molecular statement either is independently verifiable (that is, "entails verifiable statements not entailed by the original statement") or else contributes to the meaning of the molecular statement in such a way that the molecular statement entails verifiable statements not entailed by any of its components (that is, any of the components alone "does not entail verifiable statements entailed by the original statement"). The intention of these stipulations is to ensure that a meaningless statement cannot occur as a component of a verifiable molecular statement and derive verifiability from the statement in which it occurs.
In two important articles titled "Testability and Meaning" (1936–1937), Carnap distinguished the testing of a sentence from its confirmation; a sentence is "testable" if we know of a particular procedure (for example, the carrying out of certain experiments) that would confirm to some degree either the sentence or its negation. A sentence is "confirmable" if we know what kind of evidence would confirm it, even though we do not know of a particular procedure for obtaining that evidence. Carnap considers four different criteria of significance—complete testability, complete confirmability, degree of testability, and degree of confirmability. All of these exclude metaphysical statements as being meaningless. The fourth criterion is the most liberal and admits into the class of meaningful statements empirical statements of the various kinds that were excluded by the requirement of conclusive verifiability or the requirement of conclusive falsifiability.
Each of Carnap's criteria determines a more or less restrictive form of empiricist language, and this, according to his view, is the same thing as a more or less restrictive form of empiricism. Carnap is largely concerned in these articles with giving a technical account of the formal features of such languages. One of the most serious difficulties he encounters is that of giving a satisfactory account of confirmability. His procedure is, in effect, to regard as cognitively meaningful all and only those statements that can be expressed in a formalized empiricist language.
Similarly, Hempel, in his article "Problems and Changes in the Empiricist Criterion of Meaning" (1950), discussed the proposal that a sentence has cognitive meaning if and only if it is translatable into an empiricist language. A formalized language is characterized by enumerating the formation and transformation rules of its syntax and the designation rules for the terms of its basic vocabulary. An empiricist language is one in which the basic vocabulary consists exclusively of empirical terms. As Hempel explains, dispositional terms may be introduced by means of "reduction sentences," and the theoretical constructs of the more advanced sciences (for example, "electrical field," "absolute temperature," "gravitational potential") can be accommodated by allowing the language to include interpreted deductive systems.
Hempel claims for his criterion that it avoids many of the difficulties of the earlier formulations of the verifiability principle. The logic of a formalized language may ensure that no universal or existential statement is excluded from significance merely on account of its universal or existential form and also that for every significant statement its denial is also significant. The vocabulary and syntax of a formalized empiricist language ensures that no meaningless statement will be admitted as significant, even by occurring as a component of a verifiable molecular statement.
Nevertheless, leaving purely formal objections aside, the main difficulty of both Carnap's and Hempel's treatment of the verifiability principle is that of giving an adequate characterization of an empiricist language. An "empirical term" or an "observation predicate" is one that designates a property that is in principle observable, even though in fact it is never observed by anyone. But if the property has never in fact been observed, how are we to know that it is observable?
It may be said that a basic observation statement "Pa," asserting that an object a has the observable property P, is meaningful only if the experiences that would verify the statement could occur. But "could" here cannot mean "factually could," since we can speak meaningfully of occurrences that are factually impossible. Apparently what is meant is that the experiences in question must be logically possible. But then it seems that the only sense that can be given to saying that the experiences are logically possible is that the statement "Pa " is contingent. However, in "Pa " the object a is simply named or referred to, and the property P ascribed to it—and it seems that every statement of this form must be contingent. Thus, unless a further explanation of the expression "observation predicate" is forthcoming, we have no way of distinguishing between those basic observation statements that are meaningful and those that are not.
Schlick, in an early article titled "A New Philosophy of Experience," claimed that to understand a proposition we must be able to indicate exactly the particular circumstances that would make it true and those that would make it false. "Circumstances" he defined as facts of experience; and thus it is experience that verifies or falsifies propositions. An obvious objection to this view is that sense experience is essentially private, and hence apparently the cognitive meaning of every statement must be essentially private. Schlick attempted to avoid this objection by distinguishing between the content and form of experience. The content, he said, is private and incommunicable—it can only be lived through. But the form of our experiences, he claimed, is expressible and communicable, and this is all that is required for scientific knowledge. However, Schlick's distinction between content and form cannot save his view from the objection of solipsism; for if the meaning of every descriptive expression is to be found, in the last analysis, in private experience, then this is so not only for qualitative words but also for the relational words that are supposed to describe the form of experience.
Thus, the first problem concerning statements reporting empirical observations is that they should be expressible in such a way that their meaning is not private to any one observer. The logical possibility of verifying a given statement can then be explained without mentioning the experiences of any particular person or indeed the experiences of anyone at all. If basic observation statements can be formulated in the required way, they express logically possible evidence, and hence any statement suitably related to a set of observation statements is verifiable in principle, even though no one is ever in a position to have the relevant experiences, that is, to verify the statement in question.
In order to achieve this result some adherents of the verifiability principle regard certain statements describing physical objects as basic (for example, "This is a black telephone"); others attempt to achieve the same result while still regarding sense-datum or phenomenal statements as basic (for example, "Here now a black patch" or "This seems to be a telephone"). In either case, there is the difficulty of explaining how these statements are related to the experiences that would verify them.
The question whether a statement reporting an empirical observation is conclusively verifiable is, as we have seen, of importance for the criterion of conclusive verifiability and for that of conclusive falsifiability. It has also been thought to be of importance for the criterion of weak verifiability or confirmability, for, it has been said, unless basic statements are certain, or in some sense incorrigible, no other statement can be even probable or confirmable. Finally, as we noted before, there is also the problem of explaining what is meant by saying that a basic observation statement is verifiable in principle, that is, that certain experiences are logically possible, if in fact the experiences in question never occur.
is the principle itself meaningful?
It is sometimes objected that the verifiability principle itself, according to the criterion it lays down, must be either analytic or empirically verifiable if it is to be cognitively meaningful. But if it is analytic, then it is tautological and uninformative; at best it only exemplifies a proposed use of the terms "cognitive meaning" and "understanding." And if it is empirically verifiable, then it is a contingent statement about the ordinary use or some technical use of these terms and at best is only confirmable to some degree by the relevant evidence. In either case, it is objected, the principle cannot be the decisive criterion of cognitive meaning that its adherents suppose it to be.
One reply to this objection is that a criterion that determines a certain class of statements cannot have the same logical status as the statements in question. For example, the statement that expresses the principle of causality in effect determines a class of statements, namely, the class of causal statements, but obviously it is not itself a causal statement. Similarly, the verifiability principle, which claims to delimit the class of cognitively meaningful statements, cannot be expected to have the same logical status as the statements it delimits.
In order to understand the status of the verifiability principle, in the form in which it was held by the logical positivists, the following considerations are relevant: (1) They claimed that an essential difference between their empiricism and the earlier empiricism of Hume, Mill, and Mach was that it was based not on any particular psychological assumptions but only on considerations of logic. They may have believed that it is factually impossible for us to have experiences radically different in kind from those that we now have, but they did not present the verifiability principle as stating or implying this. But then, if the possibility of mystical or religious experiences is allowed, it seems that at least some metaphysical statements are verifiable and therefore meaningful. This conclusion has been accepted by some later adherents of the verifiability principle, but it is evident that the logical positivists wished to present their criterion of meaning in such a way that it would exclude all metaphysical statements from the class of meaningful statements.
(2) It might be argued, as Ayer once did, that it is meaningful to say that mystics have unusual experiences, but that nevertheless we can have no grounds for supposing that their experiences are relevant to the truth or falsity of any statement of fact, since we have no grounds for thinking that the "object" of such experiences could be described in ordinary empirical terms. The statement "Mystics have experiences that they report by the sentence 'Reality is One'" is empirically verifiable in the ordinary way. But the statement "Reality is One" is not empirically verifiable in the ordinary way. To this, however, the mystic may reply that he can describe in ordinary empirical terms the kind of preparation or discipline he recommends, and if we are not willing to carry out the appropriate procedure we are simply refusing to consider the possibility of verifying mystical statements. The antimetaphysical import of the verifiability principle, he may say, is apparently based on the assumption that we cannot have experiences radically different in kind from those that we now have.
(3) Some of the logical positivists (Schlick, the early Ayer) claimed that the verifiability principle is in effect a statement of the sense of "cognitive or factual meaning" and "understanding" that is actually accepted in everyday life. Schlick, for example, said that the verifiability principle is "nothing but a simple statement of the way in which meaning is actually assigned to propositions, both in everyday life and in science. There never has been any other way, and it would be a grave error to suppose that we believe we have discovered a new conception of meaning that is contrary to common opinion and which we want to introduce into philosophy" ("Meaning and Verification"). But, as we have seen, if the verifiability principle is simply a contingent statement about a certain linguistic usage, its logical status cannot justify the degree of confidence that its adherents place in it.
(4) Finally, the principle has been regarded as a recommendation or a decision concerning the use of the expression "factually meaningful statement." It has been claimed that this decision prevents radical intellectual confusion and that it promotes clarity in the discussion of many philosophical questions. Carnap and Ayer, among others, have taken this view of the status of the verifiability principle. It should be noted that this does not imply that the principle is regarded as an analytic or necessarily true statement. A principle that expresses a linguistic recommendation is no doubt closely related to a corresponding analytic statement, but the recommendation itself is not tautological and uninformative. A recommendation or a decision has a different logical status; it is not successful by being true or unsuccessful by being false.
more recent criticisms
Following the later work of Wittgenstein it is now widely held among philosophers that to ask whether a sentence is meaningful is simply to ask whether the words that compose the sentence are used according to the rules or practice of a language. Understanding a word, it is said, does not involve "knowing what the word stands for" or "being able to recognize what the word designates"; it involves only the ability to use the word in accordance with certain linguistic rules. Furthermore, the rules governing the correct use of different kinds of words differ enormously, and hence there is not just one way of misusing the words that occur in a sentence and thereby rendering the sentence meaningless. Each of the sentences "I do not exist," "The round square feels depressed," "Nonbeing is infinitely perfect," and "The Absolute enters into but transcends all change" involves a violation of one or more linguistic rules, but of quite different rules. Consequently, it is said, it is not possible to give a general criterion of the meaningfulness of a sentence. The verifiability principle is an attempt to answer the question "Under what conditions is a sentence cognitively or factually meaningful?," but this question, according to the view now widely held, is not one to which it is possible to give an answer that is both general and informative. Two further criticisms are made of the verifiability principle: (1) the principle, it is said, is not at all a criterion of the meaningfulness of a sentence but simply a characterization of an "empirical sentence," (2) the principle confuses the question of whether a sentence is meaningful with the different question of whether the statement it expresses can be known to be true or false. These more recent objections to the verifiability principle occur in most post-Wittgensteinian discussions of the topic of meaning. A useful summary of the arguments is given by J. L. Evans in "On Meaning and Verification."
Truth theory of meaning
It is convenient to begin by examining the second of these two further criticisms. It is concerned with the fact that one component of the verifiability principle is the thesis that the meaning of a statement is given by its truth conditions. This idea, which may be called "the truth theory of meaning," had been employed and stated by philosophers before the discussions of the Vienna circle. It is assumed, for example, by Bertrand Russell in his theory of descriptions. And Wittgenstein, in the Tractatus, said explicitly, "To understand a proposition means to know what is the case if it is true."
The formal correctness of this view can be seen from the following definition of the meaning of a statement in terms of its truth conditions. "Die Sonne scheint means that the sun is shining = DfDie Sonne scheint is true if, and only if, the sun is shining"; in general, "S means that p = DfS is true if, and only if, p." Nevertheless, it has to be admitted that the truth theory provides no effective clarification of the notion of cognitive or factual meaning. For even if the truth conditions of a statement S can be enumerated exhaustively in terms of a finite conjunction of observation statements O 1 · · · On (and, as we have seen, in very many cases this cannot be done) this entitles us to assert only that S and O 1 · · · On have the same meaning. But this does not clarify what the meaning of S is, or what it is for S to be meaningful. To say simply that two statements have the same meaning is not to say what either statement means or what it is for either statement to be meaningful.
For the kind of clarification that is being sought we now need a different and independent explanation of the meaning of an observation statement. Furthermore, the definition of the meaning of a statement in terms of its truth conditions provides no clarification unless the notion of truth is further explained. The truth of a statement can be defined in terms of its meaning in the following way. "Die Sonne scheint is true =Df Die Sonne scheint means that the sun is shining, and the sun is shining"; in general "S is true =Df S means that p, and p." But obviously it would be circular to employ this definition of truth in an attempt to clarify the notion of cognitive meaning. The two preceding definitions show, however, that there is a close connection between the notion of cognitive or factual meaning and the notion of truth. And hence, in reply to the second of the two further criticisms of the verifiability principle mentioned above, it may be argued that there must be a close connection between understanding a sentence as expressing a statement of fact and its being possible for one to know whether the statement is true or false.
Meaning and experience
The first of the two further criticisms of the verifiability principle is concerned with the fact that another component of the principle is the thesis that the truth conditions of a statement can be known only by reference to experience. This is the traditional doctrine of empiricism or positivism. The logical positivists (with the exception of Neurath, Carnap, and others, who at one time adopted a "coherence theory" of truth) held this view on the grounds that there are only two ways in which the truth-value of a statement can be ascertained, either a priori or a posteriori. According to their doctrine, if a statement can be known to be true a priori, then it is analytic and tautological and hence not a statement of fact. Therefore, if a statement is a statement of fact, it cannot be known a priori—its truth-value can be ascertained only by reference to experience. The simple dichotomy (either a priori or a posteriori) on which this argument is based has been criticized in more recent philosophy. W. V. Quine, for example, maintains that for the most part the statements that compose the corpus of knowledge have their truth-values determined by linguistic and pragmatic considerations, as well as by the occurrence of certain sensory experiences. He allows, however, that statements "on the periphery" have their truth-values determined by experience. Thus, even in a more qualified version of empiricism the difficulty still remains of making clear what it is to know that a statement is true "by reference to experience."
Nevertheless, the criticism of the verifiability principle now being considered admits that for a sentence to be an "empirical sentence" it must express a statement that is in some sense verifiable, that is, the truth conditions of which can be known by reference to experience. And it may be argued that the grounds on which this is admitted are such that they compel a similar admission for every sentence that can be understood as expressing a statement of fact. It is evident that if a form of language can be used to describe the world—that is, to make statements—its rules cannot be wholly syntactical, that is, of the kind that govern simply the formation and transformation of sentences in the language. For the language to be descriptive it must also have semantic rules, for example, rules that relate the use of its basic predicates to certain states of affairs in the world. Semantic rules may be said to govern directly the use of basic predicates and to govern indirectly, via definitions and other syntactical means, the use of nonbasic predicates. The more detailed analysis of a semantic rule—that is, an account of how such rules function in a language—is a difficult matter that we need not attempt here. For our present purpose it is sufficient to note that it would be a contradiction to say that a language was descriptive but had no semantic rules; similarly, it would be a contradiction for someone to say that he could understand a sentence as expressing a statement although he had not been able to ascertain the semantic rules of the language in which the sentence was expressed.
We can now see why many present-day philosophers say that the verifiability principle is simply a characterization of an empirical sentence. If a sentence is used to describe an experienceable state of the world, then the semantic rules governing its predicates relate those predicates, directly or indirectly, to that state of the world. It follows that the sentence expresses a statement that is in principle verifiable. But consider the position of a philosopher who maintains that he uses certain sentences to make statements about the world, although these statements are not verifiable in any sense at all. This position seems to be simply incoherent. If the sentences in question express statements, the use of the predicates that occur in them must be governed by semantic rules; how can these rules be known or explained to anyone else if the states of affairs which the sentences are supposed to describe are not experienceable in any way at all? The philosopher in question may eventually admit that the relevant states of the world are, after all, experienceable—but intuitively or by some other special kind of experience. This, apparently, would be a psychological claim, to the effect that we are capable of types of experience other than those we usually associate with the normal functioning of our sense organs. The onus of proof to show that such experiences are possible plainly rests upon the philosopher in question. But even if such experiences do occur, and are of such a kind that they can be associated, via semantic rules, with the descriptive expressions of a language, this will not provide an exception to the requirement laid down by the verifiability principle—it will, in fact, be simply an extension of that requirement to types of sentences that formerly could not be understood as expressing statements of fact.
For a further examination of this question, it would seem that the correct approach would be to give a completely general analysis of "knowing the use of a predicate." Such an analysis cannot be given here, but the following outline may be suggested. In the case of a basic predicate it may be held that (1) an essential part of the use of the predicate is to identify a property, (2) an ability to use the predicate to identify the relevant property does not constitute knowing its use, unless the user also knows what the ability consists in, and (3) the user cannot be said to know this if it is impossible for him to have any kind of experience of the property in question.
Thus, to revert to the first and main criticism of the verifiability principle, it may be admitted that to ask whether a sentence is meaningful is to ask whether the constituent words are used according to the rules of a language. And it may be admitted that the rules governing the use of different kinds of words differ immensely and that there is not just one way in which a sentence can be meaningless. Nevertheless, if the foregoing remarks are correct, a sentence cannot be understood as expressing a statement unless the use of the descriptive expressions that occur in it are governed by semantic rules; and these rules cannot be known or explained to anyone else unless it is possible for the users of the language to have some kind of experience of the states of the world to which the descriptive expressions in question are related. These requirements are, perhaps, all that is essential in the claim made by the verifiability principle in its later formulations.
formulations and favorable discussions
Ayer, A. J. Language, Truth and Logic. London: Gollancz, 1936; 2nd ed., 1946.
Ayer, A. J. "Logical Positivism—A Debate" (with F. C. Copleston). In A Modern Introduction to Philosophy, edited by Paul Edwards and Arthur Pap. Glencoe, IL: Free Press, 1957; 2nd ed., rev., New York, 1965.
Ayer, A. J. "The Principle of Verifiability." Mind 45 (1936).
Ayer, A. J. "Verification and Experience." PAS 37 (1936–1937). Reprinted in Logical Positivism, edited by A. J. Ayer. Glencoe, IL: Free Press, 1959.
Brown, R., and J. Watling. "Amending the Verification Principle." Analysis 11 (1950–1951).
Carnap, Rudolf. "Testability and Meaning." Philosophy of Science 3 (1936): 419–471; 4 (1937): 1–40. Reprinted in Readings in the Philosophy of Science, edited by Herbert Feigl and May Brodbeck, 47–92. New York: Appleton-Century-Crofts, 1953.
Hempel, C. G. "The Concept of Cognitive Significance: A Reconsideration." Proceedings of the American Academy of Arts and Sciences 80 (1951): 61–77.
Hempel, C. G. "Problems and Changes in the Empiricist Criterion of Meaning." Revue internationale de philosophie 4 (1950): 41–63. Reprinted in Semantics and the Philosophy of Language, edited by Leonard Linsky. Urbana: University of Illinois Press, 1952. Also reprinted in Logical Positivism, edited by Ayer (see above).
Reichenbach, Hans. "The Verifiability Theory of Meaning." Proceedings of the American Academy of Arts and Sciences 80 (1951). Reprinted in Readings in the Philosophy of Science, edited by Herbert Feigl and May Brodbeck. New York: Appleton-Century-Crofts, 1953.
Schlick, Moritz. "Form and Content." Three lectures given in London in 1932. Reprinted in Gesammelte Aufsätze.
Schlick, Moritz. "Meaning and Verification." Philosophical Review 45 (1936): 339–368. Reprinted in Gesammelte Aufsätze and in Readings in Philosophical Analysis, edited by Herbert Feigl and Wilfrid Sellars. New York: Appleton-Century-Crofts, 1949.
Schlick, Moritz. "A New Philosophy of Experience." Publications in Philosophy of the College of the Pacific, No. 1. Stockton, CA, 1932. Reprinted in Gesammelte Aufsätze 1926–1936. Vienna: Gerold, 1938.
Schlick, Moritz. "Positivismus und Realismus." Erkenntnis 3 (1932–1933): 1–31. Reprinted in Gesammelte Aufsätze and translated in Ayer, ed., Logical Positivism (see above).
Waismann, Friedrich. "Logische Analyse der Wahrscheinlichkeitsbegriffs." Erkenntnis 1 (1930–1931).
Whiteley, C. H. "On Meaning and Verifiability." Analysis (1938–1939).
Whiteley, C. H. "On Understanding." Mind 58 (1949).
Barnes, W. H. F. "Meaning and Verifiability." Philosophy 14 (1939).
Berlin, Isaiah, "Verification." PAS 39 (1938–1939).
Church, Alonzo. Review of Ayer's Language, Truth and Logic, 2nd ed. Journal of Symbolic Logic 14 (1949).
Copleston. F. C. "Logical Positivism—A Debate" (with A. J. Ayer). In A Modern Introduction to Philosophy, edited by Edwards and Pap (see above).
Copleston, F. C. "A Note on Verification." Mind 59 (1950). Reprinted, with "A Further Note on Verification," in Contemporary Philosophy. London: Burns and Oates, 1956.
Ducasse, C. J. "Verification, Verifiability, and Meaningfulness." Journal of Philosophy 33 (1936): 230–236.
Evans, J. L. "On Meaning and Verification." Mind 62 (1953): 1–19.
Ewing, A. C. "Meaninglessness." Mind 46 (1937): 347–364.
Kneale, W. C. "Verifiability." PAS, Supp. 19 (1945).
Lazerowitz, Morris. "The Positivistic Use of 'Nonsense.'" Mind 57 (1946). Reprinted in The Structure of Metaphysics.
Lazerowitz, Morris. "The Principle of Verifiability." Mind 46 (1937): 372–378.
Lazerowitz, Morris. "Strong and Weak Verification." Mind 48 (1939) and 59 (1950). Reprinted in The Structure of Metaphysics. London: Routledge and Paul, 1955.
Lewis, C. I. "Experience and Meaning." Philosophical Review 43 (1934). Reprinted in Readings in Philosophical Analysis, edited by Herbert Feigl and May Sellars. New York: Appleton-Century-Crofts, 1953.
O'Connor, D. J. "Some Consequences of Professor Ayer's Verification Principle." Analysis 10 (1949–1950).
Russell, Bertrand. An Inquiry into Meaning and Truth. New York: Norton, 1940.
Russell, Bertrand. "On Verification." PAS 38 (1937–1938): 1–15.
Stace, W. T. "Metaphysics and Meaning." Mind 44 (1935): 417–438. Reprinted in A Modern Introduction to Philosophy, edited by Paul Edwards and Arthur Pap. Glencoe, IL: Free Press, 1957; 2nd ed., New York, 1965.
Stebbing, L. Susan. "Communication and Verification." PAS, Supp. 13 (1934).
Waismann, Friedrich. "Language Strata." In Logic and Language, edited by A. G. N. Flew. 2nd series. Oxford: Blackwell, 1953.
Waismann, Friedrich. "Verifiability." PAS, Supp. 19 (1945): 119–150. Reprinted in Logic and Language, edited by A. G. N. Flew. 1st series. Oxford, 1951.
Warnock, G. J. "Verification and the Use of Language." Revue internationale de philosophie 5 (1951): 307–322. Reprinted in A Modern Introduction to Philosophy, edited by Paul Edwards and Arthur Pap. Glencoe, IL: Free Press, 1957; 2nd ed., New York, 1965.
Wisdom, John. "Metaphysics and Verification." Mind 47 (1938): 452–498. Reprinted in Philosophy and Psycho-analysis. Oxford: Blackwell, 1953.
Wisdom, John. "Note on the New Edition of Professor Ayer's Language, Truth and Logic." Mind 57 (1948). Reprinted in Philosophy and Psycho-analysis.
Alston, W. P. "Pragmatism and the Verifiability Theory of Meaning." Philosophical Studies 6 (1955).
Bridgman, P. W. "Operational Analysis." Philosophy of Science 5 (1938).
Grice, H. P. "Meaning." Philosophical Review 46 (1957): 377–388.
Heath, A. E. "Communication and Verification." PAS, Supp. 13 (1934).
MacKinnon, D. M. "Verifiability." PAS, Supp. 19 (1945).
Morris, C. W. "The Concept of Meaning in Pragmatism and Logical Positivism," in Proceedings of the Eighth International Congress of Philosophy (1934). Prague, 1936.
Nagel, Ernest. "Verifiability, Truth, and Verification." Journal of Philosophy 31 (1934): 141–148. Reprinted in Logic without Metaphysics. Glencoe, IL: Free Press, 1956.
Nelson, E. J. "The Verification Theory of Meaning." Philosophical Review 43 (1954).
other recommended titles
Bealer, George. "The Incoherence of Empiricism." The Aristotelian Society, Supp. 66 (1992): 99–138. Reprinted in Rationality and Naturalism, edited by S. Wagner and R. Warner. Notre Dame, IN: University of Notre Dame Press, 1993.
Coffa, Alberto. The Semantic Tradition from Kant to Carnap. Cambridge, U.K.: Cambridge University Press, 1991.
Cohen, Jonathan. "Is a Criterion of Verifiability Possible?" Midwest Studies in Philosophy 5 (1980): 347–352.
Dummett, Michael. "The Metaphysics of Verificationism." In The Philosophy of A. J. Ayer, edited by Lewis Hahn. La Salle, IL: Open Court, 1992.
Edington, D. "The Paradox of Knowability." Mind 94 (1985): 557–568.
Feigl, H. "The Weiner Kreis in America." In The Intellectual Migration, edited by D. Fleming and B. Bailyn. Cambridge, MA: Harvard University Press, 1969.
Hempel, Carl. Aspects of Scientific Explanation. New York: Free Press, 1965.
Johnston, Mark. "Verificationism as Philosophical Narcissism." In Philosophical Perspectives, 7: Language and Logic, edited by James Tomberlin. Atascadero, CA: Ridgeview Press, 1993.
Okasha, Samir. "Verificationism, Realism, and Scepticism." Erkenntnis 55 (2001): 371–385.
Quine, W. V. O. "Two Dogmas of Empiricism." In From a Logical Point of View: Nine Logico-Philosophical Essays. Cambridge, MA: Harvard University Press, 1980.
Smart, J. J. C. "Verificationism." In Cause, Mind, and Reality: Essays Honoring C. B. Martin. Norwell, MA: Kluwer, 1989.
R. W. Ashby (1967)
Bibliography updated by Benjamin Fiedor (2005)