Jerrold Katz had been fighting off cancer for some time, but he lost that fight on February 7, 2002. I was fortunate enough to visit with Professor Katz in Manhattan on some occasions in the 90's, most recently in 1997. The last time I heard from him was in September 2001. He had just finished his new book, Sense, Reference, and Philosophy. As it happened, the contract for its publication was signed right before his death, and the book is now out from Oxford University Press . This completes the project that began with The Metaphysics of Meaning, reviewed here, and Realistic Rationalism [MIT, 1997] -- a provocative title in modern philosophy if there ever has been one. I am very sorry that we have lost Professor Katz. He was a real philosopher. I am glad that I was able to know him in a small way and enjoy some of his conversation and insight.
The Metaphysics of Meaning, by Jerrold J. Katz, is one of the most important books in all of recent philosophy. And Katz himself is that rare person in 20th Century philosophy: Someone who talks about language and actually knows something about real (i.e. natural) languages. With a doctorate in philosophy but a sober curiosity about what everyone at the time was talking about -- language -- Katz went off to do graduate work in linguistics with Noam Chomsky himself. The result is a fully informed linguistic philosopher.
This prepared Katz to talk about something that everyone was talking about also but didn't seem to know much about either: Meaning. Trying to identify or define the nature of meaning was one of the major projects of Anglo-American philosophy from the 30's to the 60's, and the whole effort was an utter and catastrophic failure. Logic and Logical Positivism inspired the project, but Logical Positivism as a school never wanted to understand meaning in anything other than denotational and ostensive terms. Otherwise it might have to dirty its hands with talk about metaphysics, whose possibility it had rejected a priori. "Intention" and "connotation" raised too many uncomfortable metaphysical questions.
This led to two serious consequences:
Katz demolishes the next generation of reductionistic theories. Wittgenstein and Quine were no longer Logic Positivists, but they were not much better. Wittgenstein served the historically useful function of discrediting the Positivist notions of ostensive definition, but his own theory, that "meaning is usage," is non-referential and, in effect, autistic. Popper already observed that Wittgenstein's definition was inconsistent with Tarski's definition of truth, that "p" is true if and only if p; for the only difference between "p" and p will be the difference between mention and use, where the former does not refer and the latter does. Katz goes far beyond such formalism in his criticism, with vast insight into the nature of language. Indeed, Benson Mates' statement in Elementary Logic is immediately followed by an obvious paradox: Explaining why he only speaks about sentences, which are visible, and not about propositions, which are the traditional metaphysical entities, Mates proceeds to say that "It's raining," and "Es regnet," mean the same things in English and German, respectively. Since they are clearly not the same sentences, it is obvious that they must refer to something beyond themselves to have the same meaning. That must either be the propositions, which Mates has denied, or the extension, which, as we have seen, cannot explain meaning for natural languages. As Katz says, we must say that "It's raining" and "Es regnet" are "synonymous," but this turns out to be a term that really cannot be used in Mates, Wittgenstein, or Quine. They all lack the third thing, that which is neither language nor external objects; and this, it happens, is what is necessary for the concept of meaning.
Katz does not give a positive metaphysical theory of meaning. That will follow. But his critique of the "naturalistic" theories of meaning in Wittgenstein and Quine that meaning cannot consist of "natural" objects, whether of language itself ("usage") or of natural objects ("extensionality"), is decisive enough; and it is obvious that a non-naturalistic theory is necessary. Katz does give us a "proto-theory," which contains decisive unorthodox insights on the nature of language. The most important part of the "proto-theory" is the principle that the syntactically simple elements of languages cannot be the semantically simple elements. This is contrary to the assumptions of Frege's Begriffsschrift approach to logic, which identified atomic syntactic with atomic semantic elements, and upon which Russell, Wittgenstein, and most representations of symbolic logic are based. The idea even occurs in sophisticated linguistics that symbolic logic may be used as a sufficient "semantic representation" of natural languages.
As Katz says:
We postulate that the sense of the syntactic simple "woman" is complex, consisting of the sense of "human", the sense of "adult", and the sense of "female". On this postulation of decompositional sense structure for "woman", the redundancy of "a woman who is female" is immediately accounted for with the same intuitively obvious notion of redundancy that accounts for the redundancy of expressions like "a woman who is a woman".
This case is exactly parallel to that in which Chomsky postulated an underlying syntactic structure in order to extend the account of subject and direct-object relation in sentences like "John loves Mary" to sentences like "John is easy to please" and "John is eager to please". By parity of reasoning, we postulate an underlying semantic structure in order to extend the account of redundancy in expressions like "woman who is a woman" to expressions like "woman who is female". Decompositional postulations require a grammatical locus for the unobservable complex senses they postulate; so we are led to taking the step of positing that grammatical structure contains an underlying level of sense structure. [pp. 64-65]
Terms like "redundant," "synonym," "antonym," etc. created great difficulties for the old, non-decompositional theories of meaning. As Katz says in regard to Wittgenstein:
The case of antonymy is, of course, of special interest in connection with the impasse Wittgenstein reached in connection with elementary sentences. "The spot is red" and "The spot is blue" contradict each other; yet cannot do so within the semantics of the Tractatus, since, being elementary, they contain no logical operators, and their nonlogical vocabulary cannot contribute a form of negation. The problem is intractable as long as we try to solve it within a semantics derived from logic, where, as Wittgenstein said, "the application of logic decides what elementary propositions there are" -- and what propositional structure is related to necessary incompatibility. There is simply no negative element to account for the incompatibility of the sentences. [p. 72]
As Wittgenstein could not account for the antonymic senses of "red" and "blue," Quine decided, famously, that the notion of analytic truth was one of the "dogmas of empiricism." Quine simply could not account for how a predicate could be "contained" in the meaning of the subject, when the subject was obviously a distinct syntactic item. But, as Katz says:
Semantic properties and relations like analyticity and analytic entailment, which also depend on sense containment, can be accounted for on the same decompositional hypotheses used to account for redundancy and superordination. [p. 65]
Thus, even Katz's "proto-theory" directs us into a profoundly different course than that found in most of Twentieth Century philosophy. The importance of this cannot be overestimated. The century began with Frege making a serious mistake that has distorted logic and philosophy ever since; and since much of the irrationality of philosophy in the century has been based on the failures of the logicians and rationalists, the human consequences may have been of tragic magnitude: Any encouragement for dogmatism and fanaticism would have been more than was desirable for the appalling tyrannies of fascism, communism, and fundamentalism that we have witnessed.
There is little on which Katz errs. One point is the status of synthetic a priori propositions as necessary or not. He says on page 312 that, "To say that a proposition is necessary is to say that it is true in every possible world, but philosophical skepticism presents us with examples of possible worlds in which synthetic a priori propositions are false." It is true that synthetic a priori propositions are not true in every possible world, as those are defined, and this means that in a certain sense they are not necessary; but it is a mistake to think that "necessary" always means "true in every possible world." There is more than one kind of necessity, and "possible worlds" analysis only addresses one of those. There is a natural and venerable counterexample to the "possible worlds" theory of necessity: Aristotle's discussion in On Interpretation about the contingency of propositions about the future in contrast to the necessity of propositions about the past.
In logic we may be accustomed to treating things sub specie aeternitatis and so perhaps need not take seriously a form of necessity that is time dependent; but since we live in time and are always faced with the unpleasant truths of a past that cannot be altered, this form of necessity, very much of our world and moment, lays a claim on us that philosophy since Aristotle (or at least since Al-Fârâbî) neglects at its peril. Be that as it may, if there is at least a prima facie case for a form of necessity apart from possible worlds or the necessity of formal logic, then this does open the possibility that there are various forms of necessity. Kant's common sense conclusion, therefore, that synthetic a priori propositions are necessary because experience cannot contract them need not be rejected out of hand because of an understanding of modality that may be reductionistic.
Another basic difficulty with quantified "possible worlds" in modal logic is that it seems to beg the question. If necessity and possibility can easily be defined in terms of each other (the necessary is the not possibly not), then the problem is to provide a meaning for modality that doesn't use either concept. But "possible worlds" obviously uses one of the concepts -- possibility. Trying to avoid that problem by making the worlds somehow actual (as some do) sounds more like science fiction than like logic. On the other hand, there may already be in Plato a sense of relating modality to actuality in terms of time: the necessary is what is eternally true or existent; the impossible is what never is so; the contingent is what is so at one time but not at others; and the possible is what is not so at one time but not at others. "Eternity" is certainly a lot to think about, but no more so than an infinite number of possible worlds (including ones where the Wizard of Oz is real, or where Hitler wins World War II). Less grandiosely, however, the future itself may simply be taken as the possible, with rules on coming-to-be as the forms of necessity.
If nothing else, there is simply no sufficient reason why we should think that "possible worlds" define the nature of necessity, or why this definition should not be circular, if we first think in terms of logical necessary, and then reflexively define the possible world in those terms. Such difficulties, however, are trivial compared to the watershed analysis of The Metaphysics of Meaning. If Logical Positivism, Wittgenstein, and Quine all fall, then the whole project of meaning in 20th Century philosophy, which has mostly been empiricist, reductionistic, and naturalistic, is a failure; and it becomes obvious that one of the avenues of inquiry branching off from Kant has been a blind alley. Since other, equally blind alleys, like Hegelianism, have subsequently been revived, and vast new theories of nihilism and irrationalism are being actively promoted, there will be no lack of challenges for the philosophers of the 21st Century.
Sense, Reference, and Philosophy, Jerrold J. Katz