Truth (Princeton Foundations of Contemporary Philosophy) Read online

  So in place of the traditional three-cornered realist-idealist-pragmatist debate, one has today a three-cornered realist-antirealist-deflationist debate, complicated by each of the three positions coming in several variant versions and by the presence also of several less popular views on the scene. That debate will be the primary topic of this book.

  1.3 PARADOXES

  Russell's work in logic became very influential among philosophers in the decades following the First World War, especially among the logical positivists, the dominant school of the period. But many positivists' views on truth were less like Russell's than like Dewey's, in that they tended to hold that the concept of truth has no role to play in a scientifically oriented philosophy. Ironically, one of Russell's own discoveries contributed to the spread of this tendency, which was also found among logicians and mathematicians.

  In the background was a crisis in the then-novel branch of mathematics known as set theory, where antinomies or paradoxes had emerged. Some of these (Burali-Forti's, Cantor's) were quite technical, but Russell discovered one that is easily stated. Consider the set R of all sets that are not elements of themselves. Then R is an element of R, which is to say, is one of the sets that is not an element of itself, iff R is not an element of R—a contradiction! Russell's paradox reminded many of a fact long known but little cited in the realist-idealist-pragmatist debate, that the notion of truth, too, is subject to paradoxes.

  In particular it reminded many of a paradox attributed to Aristotle's contemporary Eubulides, called the pseudomenos or liar. Suppose I say, “What I am now saying is not true.” Then it seems that what I am saying is true iff what I am saying is not true—another contradiction! Medieval logicians had added similar examples, under the label insolubles: Suppose Socrates says, “What Plato is saying is true,” while Plato says, “What Socrates is saying is false.” Modern mathematicians and logicians and philosophers now added more examples, finding that truth is but one of a family of related notions, all of which involve what seem to be similar contradictions.

  One member of this family of notions—called alethic, from the Greek for “truth”—is that of a predicate or an adjective being true of something. This notion gives rise to an exact parallel to Russell's paradox of the set of all sets that are not elements of themselves, namely, Grelling's paradox of the adjective that is true of all adjectives that are not true of themselves. “English” is English, “short” is short, and “polysyllabic” is polysyllabic. Hence these three are autological or true of themselves. By contrast, “French” is not French, “long” is not long, and “monosyllabic” is not monosyllabic. So these three are heterological, or not true of themselves. What about “heterological”?

  Another member of the family of alethic notions is that of definability, where an object is definable iff it is the one and only thing of which some predicate is true. This notion likewise gives rise to several paradoxes, of which Berry's, the easiest to state, goes somewhat as follows: “The smallest natural number not definable in English in twenty-six syllables or fewer” defines a natural number in English in twenty-six syllables. Other definability paradoxes (Richard's, König's) were more technical, and became entangled with debates over the status of set theory. Many mathematicians were only too happy to join those philosophers who classed alethic notions as unacceptably “psychological” if not damnably “metaphysical” in character.

  However, one important mathematician and logician, Alfred Tarski, and following him quite a few philosophers, attempted to rehabilitate the notion of truth by “solving” or “resolving” or “dissolving” or at least blocking the paradoxes. Debate over the solution (or insolubility) of the paradoxes and debate over the nature (or lack of nature) of truth proceeded separately over most of the last century, but it has become increasingly clear in recent years that the two questions cannot really be kept apart. (For instance, some think deflationism makes it harder, while others think it makes it easier, to deal with the paradoxes.) Accordingly, the paradoxes will be a secondary topic of this book.

  1.4 PLAN

  It is with Tarski, probably still today the writer most often cited in discussions of truth by philosophers in the analytic tradition, that we begin our survey of theories of truth. To accommodate readers differing in their degree of background and interest in technical matters, we divide our account of Tarski's work in two. The first half of chapter 2 gives a nontechnical account of the most often discussed aspects of Tarski's views, which should be enough to enable the reader to follow allusions to those views later in the book. More technical material is confined to the starred sections making up the second half of the chapter, which readers who so choose may postpone or omit.

  We then turn to the deflationism-realism-antirealism debates, taking deflationism first. Tarski gave a central role in the theory of truth to the principle that saying something and saying it is true are equivalent. Common to all forms of deflationism is the claim that, roughly speaking, this equivalence principle is all there is to the theory of truth. Chapter 3 is devoted to the exposition of various deflationistic positions from Ramsey's time to the present; a half-dozen variants, some more radical, some more moderate, are described. The present authors are sympathetic to the general idea behind deflationism, but not fully satisfied with any of the existing versions. It is our dissatisfaction with existing versions that will be most evident in this chapter, while our sympathy with the general idea will become more evident when we turn in later chapters to de-flationism's “inflationist” (realist and antirealist) rivals and critics.

  One popular objection to the equivalence principle, and hence to any theory of truth, such as deflationism, that embraces it, goes as follows. If you have never practiced cannibalism, then neither “Yes” nor “No” is an appropriate answer to “Have you stopped eating people?” since the question seems to presuppose that you at least used to eat people. This suggests that “You have stopped eating people” is neither true nor false. But if it is not true (as well as not false), then to say that it is true is to say of what is not that it is, which is false, and we have a case where saying something is not equivalent to saying that it is true, because the latter is false while the former is not false (though not true, either). Similar purported counterexamples turn on the phenomena of vagueness and relativity, which we lump together with presupposition in chapter 4 under the bland label “indeterminacy.” We survey various lines of defense deflationists have taken against purported indeterminacy counterexamples, without pretending to achieve a full resolution of the issues. Paradoxical examples like the liar are often viewed as further cases of indeterminacy, and the discussion of presupposition and vagueness is in some respects a warm-up for tackling the paradoxes later in the book, though it will be seen when we come to them that the paradoxes involve an additional twist.

  Chapter 5 is devoted to views we classify as “realist,” namely, views taking truth to involve standing in some appropriate relation to some portion(s) or aspect(s) of reality. Russell and Moore were realists about truth in this sense at key stages in their careers, though each changed his mind about truth more than once—our attaching just one view to Russell's name in §1.1 and to Moore's in §1.2 was a caricature—and though their versions of realism were distinct and incompatible. Among the heirs of Russell and Moore there is even more disagreement than there was between those pioneers themselves. There is disagreement both over the nature of the “appropriate relation” involved (which some do and some don't call “correspondence”) and over the “portion(s) or aspect(s)” of reality involved (which some do and some don't call “truthmakers”). Moreover, there is a division between those satisfied with a fairly abstract and metaphysical account of these matters, and those who see a need for a more concrete and physical account. Along with the different realist views we consider also a realist objection to deflationism alleging that the latter cannot explain why true beliefs are useful, and an objection to realism and deflationism alike alleging that
the notion of truth has an evaluative role that both groups wrongly neglect.

  Chapter 6 is devoted mainly to those who call themselves “antirealists.” They reject both deflationism and what we call “realism,” though they do not much discuss either. What they do discuss at length, and most emphatically reject, is something else that they call “realism,” which amounts to what others call truth-conditional semantics, to which they oppose something called verification-conditional semantics. (Explaining the tangled usages of “realism” will be one of our tasks in this chapter.) Along with antirealism we take note in the same chapter of a more recent position, pluralism, which holds that a realist view of truth may be more appropriate for some “domains of discourse” and an antirealist for others. The discussion of pluralism concludes our survey of contemporary theories of the nature of truth, insofar as those can be discussed without bringing in issues about the paradoxes.

  We begin our examination of views on liar-style paradoxes in chapter 7, with an account of the work of Saul Kripke, whose “Outline of a Theory of Truth” (1975) has probably been the most influential work on its topic since Tarski's “The Concept of Truth in Formalized Languages” (1935). Our Kripke chapter is organized like our Tarski chapter, with a nontechnical account, containing what is needed to follow discussions in our next and final chapter, coming in the first half, followed in the second half by sections starred as optional reading, containing more technical material.

  Tarski held that it is impossible to avoid paradox unless one distinguishes the language for which one is formulating a theory of truth from the language in which one is formulating that theory. Kripke in the end seems to concede, however reluctantly, that he has been unable to avoid something like Tarski's split or the “ghost” of it. It is on this point that several subsequent writers have sought to improve on Tarski and Kripke alike. Several proposals are considered in chapter 8, especially views advocating deviation from classical logic and views emphasizing the role of context in communication. Also considered is the defeatist view that no proposed solution to the paradoxes can ever be wholly successful, because the intuitive notion of truth ultimately is simply incoherent. Finally, a connection between this issue of the solvability or unsolvability of the paradoxes and the issues between deflationism and inflationism is briefly sketched. The book then ends, not with a final verdict on the issues, but with suggestions for further reading.

  Before launching into our survey we must address a question that we have sidestepped so far, but can hardly hope to continue evading when we get down to closer consideration of the views of specific authors. The question is this: What kinds or sorts of things or items are true, or as is said, are bearers of truth? Presumably the same kinds of things are false as are true, so what we are really asking is: What kinds of things bear truth or falsehood, or as is said, bear truth values? Or if (as is surely the case) more than one kind of thing can do so, which are the fundamental truthbearers? An early confrontation with this question is unavoidable, since the writers whose work we survey often have quite strong opinions on the matter, so that the position adopted on it can affect the whole character of an author's account of truth.

  1.5 SENTENCES

  One answer quickly suggests itself. Some truths have been written down. (The preceding remark provides an example.) Other truths have only been spoken. (Readers will have to provide their own examples.) When truths are written or spoken it is sentences that are written or spoken, and so it may seem that it is sentences that are the truths, the bearers of truth. The same conclusion can also be reached in a different way. Consider the following dialogue:

  (6) X: Where there's a will, there's a won't.

  Y: That's true.

  Z: What's true? I didn't hear.

  Y may answer Z with a direct quotation, thus:

  (7) Y: X said, “Where there's a will there's a won't,” and that's true.

  It seems that the quotation of a sentence designates a sentence, the one quoted. If so, it seems to be a sentence that Y is calling true in (6).

  But what are sentences? The distinctions that have to be made in response to this question, though they may at first seem pedantic, have proved fundamental to the study of language. Sticking with speech rather than writing for the moment, suppose each of ten greeters on a reception line says successively to each of ten guests, “I'm glad to see you.” Is that a hundred sentences or one sentence a hundred times? The usual answer is that it is one sentence type and a hundred sentence tokens. Does our language have one sentence type “The post office is near the bank” with two meanings, or two with the same pronunciation? Using Greek-derived words for pronunciation and meaning, we may say there is one phonological type but two semantic types. A phonological type is a sound pattern, a semantic type a sound pattern plus a meaning. Once made, the distinction phonological vs semantic can be seen to apply to tokens as well: Producing a phonological token amounts to emitting sounds, as both parrots and people can do, but only the people and not the parrots can thereby speak meaningfully, which is what producing a semantic token amounts to.

  Writing (or recorded as opposed to live speech) complicates the story. If the greeters have laryngitis and each greeter writes on a card, “I'm glad to see you,” and successively shows it to each guest, then there is one type, ten inscriptions of the type, and one hundred presentations of the inscriptions to a potential reader. Usually the word “token” is used with writing for what there are ten of, but to maintain parallelism with speech it might better be used for what there are one hundred of. As with speech one distinguishes phonological from semantic, so with writing one must distinguish orthographic from semantic, since two meanings may share a spelling.

  Ambiguity can prevent an orthographic or phonological type from having a fixed truth value: “A bank is an especially dangerous place to be during a flood” is true if riverbanks are meant but false if moneybanks are meant. An even more common phenomenon than ambiguity is so-called indexicality, the kind of dependence on features of context (such as who is speaking to whom) that is signaled by what are called indexicals (such as “I” and “you”). It can prevent even a semantic type from having a fixed truth value: “I'm glad to see you” is true when said by a greeter sincerely glad to see the guest, and false when said by one merely being polite. In general, a sentence can be the bearer of a fixed truth value only if we understand “sentence” in the sense of semantic token.

  Where indexicality is absent, and any possible semantic token would have the same truth value, there will be no harm if we say that the semantic type has that truth value in a secondary sense, and if ambiguity is absent as well, that the orthographic or phonological type has that truth value in a tertiary sense. Thus some types may be recognized as derivative truthbearers even if semantic tokens are recognized as the primary truthbearers. There is, however, a rival proposal.

  1.6 PROPOSITIONS

  Returning to the dialogue (6), Y may answer Z with an indirect rather than a direct quotation, thus:

  (7’) Y: X said that where there's a will there's a won't, and that's true.

  What is called a that-clause, the word “that” with a sentence coming after it, is often held to designate a proposition, the proposition expressed by that sentence. If so, it is a proposition that Y is calling true in (7’), and that we are calling true when we use the form of expression “That _____ is true” or its stylistic variant “It is true that _____.”

  Many take propositions to be the primary bearers of truth, with sentences being called “true” only in a derivative sense, a sentence being true if it expresses a true proposition. According to this propositionalist view, what we took earlier to be reasons for giving first place to semantic sentence tokens as bearers of truth are better taken as reasons for giving them first place as expressers of propositions (but only second place, after propositions, as bearers of truth). Some proponents of the rival sententialist view, while still insisting on sentences as the primary truthbearers, al
low that a proposition may be called true in a derivative sense if it is or could be expressed by a true sentence. Many sententialists, however, are suspicious of the whole idea of propositions, not least on account of the frequency of disagreements about them among propositionalists themselves.

  Propositionalists generally agree that, just as different tokens of the same sentence type may express different propositions, so inversely the same proposition may be expressed by tokens of different sentence types. If Jack says to Jill, “I am younger than you are,” and Jill says back to Jack, “You are younger than I am,” they have expressed the same proposition, that he is younger than she is. Similarly if Jack says to Jill in English, “I love you,” and then in French «Je t'aime.» But here agreement ends. A sentence like “Jack fell down” has a certain grammatical structure. Does the proposition it expresses have a structure as well? One of us may say “Jack loves Jill,” calling them by name, and another of us, “He loves her,” pointing first to the one, then to the other. Have we expressed the same proposition? To each question, some say yes and some say no. And then there are embarrassing questions asked by linguists. If “that the earth moves” designates the same thing that “the proposition that the earth moves” denotes, why is it that we can say

  (8a) Copernicus hypothesized that the earth moves, while the Inquisition anathematized the proposition that the earth moves.

  but not

  (8b) Copernicus hypothesized the proposition that the earth moves, while the Inquisition anathematized that the earth moves.