Copyright 1996 and 1999 by Russell Eliot Dale All rights reserved
1. Quine (1960), p. 78.v

2. Davidson, in the following passage, can be read as claiming that Quine doesn't aim at a theory of meaning at all: "The idea of a translation manual with appropriate empirical constraints as a device for studying problems in the philosophy of language is, of course, Quine's. This idea inspired much of my thinking on the present subject, and my proposal is in important respects very close to Quine's. Since Quine did not intend to answer the questions I have set, the claim that the method of translation is not adequate as a solution to the problem of radical interpretation is not a criticism of any doctrine of Quine's" (Davidson (1973), p. 129). Of course, providing a theory of what Davidson calls radical interpretation is not exactly the same as providing a theory of meaning in the sense of this dissertation. So Davidson's denial that Quine is trying to solve the problem of radical interpretation need not be seen as a denial that he is trying to suggest a theory of meaning.3

3. Katz and Fodor (1963). See also Katz and Postal (1964).

4. Most philosophers today seem to use the expression "semantic theory" to mean a theory that somehow says what expressions mean or denote and I will stick to this usage. But this is a verbal point, not one of substance.\

5. Bruce Vermazen in his review of Katz and Postal (1964) and Katz (1966) lists twenty-five such semantic properties and relations (Vermazen (1967), p. 364, note 3).

6. Vermazen, discussing a specific Markerese translation of an English sentence says the following: "...it hardly seems to 'provide a characterization of the meaning' in any more interesting sense; considered as a string of English words in parentheses, it is more or less unintelligible, and since Katz assures us that it is not a string of English words, but a collocation of representatives of concepts, it is only more clearly unintelligible. And the meaning of a sentence is a fairly important concept to accommodate in a semantic theory" (Vermazen (1967), p. 355). retatiA version of this criticism in to be found in Lewis (1969), pp. 171: "...I have notrse, endorsed Katz's account of the nature of...interpretations.... Katz takes them to be expressions built out of symbols called 'semantic markers'.... I find this account unsatisfactory, since it leads to a semantic theory that leaves out such central notions as truth and reference." See also Lewis (1970) where he essentially repeats this criticism. sycholSearle also gives a version of the criticism in Searle (1972), pp. 26g ba-27. the eDavidson, without mentioning Katz, Fodor, or Postal or explicitly makinge re reference to Katz/Fodor-style theories seems nonetheless to have something like that sort of theory in mind when he criticizes translation-style theories in his "Radical Interpretation": "...in the general case, we can know which sentences of the subject language translate which sentences of the object language without knowing what any of the sentences of either language mean (in any sense, anyway, that would let someone who understood the theory interpret sentences of the object language)" (Davidson (1973), p. 129). ssent Evans and McDowell in their introduction to Evans and McDowell (1976) alsother give a version of the criticism, explicitly calling attention to Katz/Fodor-style theories and citing Lewis (1970) as a source for their views. ical sMartin Davies repeats Evans and McDowell's criticism, citing them and Lewiscorr (1970); see Davies (1981), p. 4. espondFor a recent reply by Katz to this style of criticism of Katz/Fodor be -style theories, see Katz (1990), pp. 211-214. Essentially, as far as I can see, Katz's response to the sort of criticism found in the above places is that they all overlook that Markerese comes with an "intended interpretation" (Katz (1990), p. 212). But it seems like the criticism in question here is that the interpretations are not explicit and that no provision is made for how to make the interpretations explicit. It is hard to understand why Katz should have thought that a property like analyticity should have been something for which to provide an explicit definition in terms of semantic markers while a property like meaning that snow is white should be left to a grasping of an implicit convention. And this is especially so if, as was claimed in Katz and Fodor (1963), a Katz/Fodor-style theory is supposed to be a characterization of a natural-language user's linguistic competence: for surely a natural-language user doesn't understand natural language by grasping "intended interpretations" of Markerese formulas! The complaint by the above philosophers is exactly that as Katz/Fodor-style theories are sketched, they won't explicitly entail that a sentence like "snow is white" has the property of meaning that snow is white or at least the property of being true just in case snow is white. To appeal to implicit intended interpretations doesn't look like an adequate response to this complaint. The provision of rules that would entail that a sentence means what it does would.Of course, another move in defense of Katz/Fodor-style theories can be the identification of Markerese formulas with Mentalese formulas. This is essentially the move suggested by Fodor that begins the translational tradition in question in the main text. This defense doesn't provide for an explicit statement of the semantics of Markerese formulae either, but it doesn't have to directly. Standing in a cognitive relation to a translation theory could, for all we know, explain the competence of natural-language users. But standing in a cognitive relation to a Katz/Fodor-style theory without the identification of Markerese formulae with Mentalese formulae leaves big holes in the theory of natural-language competence: there is still the question how do native speakers know the meanings of sentences that such a theory doesn't answer.

There are other criticisms of Katz/Fodor

-style theories besides the above. For example, Katz/Fodor-style theories have been criticized by a number of writers on the grounds that they require an untenable conceptual atomism, as it might be called. See Putnam (1970) for this sort of criticism. I am inclined to think that this sort of criticism of Katz/Fodor-style theories is telling. I am not really discussing the conceptual-atomistic aspects of Katz/Fodor-style theories in the main text though.

7. Actually quite a bit more could probably said concerning the prehistory of the translational tradition as their are many theorists besides those who I have mentioned here who have thought that translation is the central notion for a proper understanding of speech and language. Welby is among these; see Welby (1983) and Welby (1985). See also Steiner (1975) where various views of translation are compared with various other views of speech and language. See also Parkinson (1977) and Stewart (1977), both in Vesey (1977).

8. For a good basic discussion of how all that works, see Tanenbaum (1976). 'Translators' in the present sense come in two varieties called 'interpreters' and 'compilers'. Sometimes people talk only in terms of compilers when they really mean to speak more generally of any sort of higher-level language 'translator'. Fodor does this in Fodor (1975) which I am about to discuss.6

9. See Miller and Johnson-Laird (1976), pp. 158 ff. Miller and Johnson cite sources for their views as well, so though I start this vague history with them, this is probably not quite correct.@

10. Miller and Johnson-Laird (1976), p. 159.

11. See Fodor (1975), chapter 3. Fodor cites Miller and Laird-Johnson as sources for his view, and he explicitly rejects their procedural semantics. See Fodor (1975), p. 116, note 12.7

12. Fodor (1975), p. 116.w

13. None of this, by the way, is to say that Fodor in Fodor (1975) accepted the conceptual atomism of Katz/Fodor-style theories. That is an independent issue that I am not treating here.9

14. Schiffer does seem to be on the way to something like a translational-theory of meaning in Remnants of Meaning (Schiffer (1987a)), chapter 7 with his discussion of Harvey who processes sentences by translation to and from Mentalese. But Schiffer doesn't seem to have come up with the idea of formulating a theory of meaning in terms of such translation until later.*

15. Schiffer (1993), pp. 244-245.

16. Schiffer (1993), p. 241.z

17. Schiffer (1993), p. 246.z

18. Ibid.g

19. Of course, both the employment-clause and the whole-language-clause are open sentences by themselves, so if we are really being picky, we couldn't say they state anything. When I speak of these clauses as expressing or stating things I should be taken as meaning within [T].

20. [P] can be restated as: for any language L and population P, there is a practice in P of meaning L iff for some subset S of the set { x is actually tokened at some time by a member of P in order to mean y}, the sentences of S are often used by members of P to mean what they mean in S and SL. I say "for some ... subset S ..." because the members of a population may speak more than one language. The idea here is that for any set S that is a subset of a language L, and for any population P, if often a member of P in order to mean a proposition utters an S-sentence that means in S , then often that same member of P in order to mean utters an L-sentence that means in L .

21. Note that I do not say "tokened in B's belief-box" as if B came to believe what A said. Of course, it would be true to say that but perhaps confusing - the M-expression that means what A said is, indeed, tokened in B's belief-box on present presuppositions but it is embedded in another sentence. A doesn't have to believe what B said to understand it.

22. Schiffer (1993), p. 251. See also Schiffer (1994b), p. 297 and Schiffer (1994c).

23. See Ray (1995) and Schiffer's response, Schiffer (1995). See also Voltolini (1995).

24. S1 may, of course, be either identical to or a proper subset of the set {x is actually tokened at some time by a member of P1 in order to mean y}.)

25. The term "transfer function" is usually used by engineers and in engineering literature to denote an input/output mapping for some limited range of inputs and outputs, and not a for every possible input and output of a system. But there is no harm in speaking in the way I do in the main text.

26. Often, for mechanisms that people build to compute functions, we design the physical input and output states so that one specific interpretation of input and output states is quickly available to us: a certain sequence of key strokes on a calculator represents the number 39.432, and a specific arrangement of the activated LCD's of an LCD-display represents the number 946.83. But these don't really pin down one function as the one determined by the mechanism in anything but a sort of heuristic sense. For some discussion of related issues, but in a wholly different context, see Katz (1985b), pp. 198-199.\

27. I believe that this sort of view lies behind Wittgenstein's picture theory in the Tractatus and behind Hertz's views - which Wittgenstein acknowledged in the Tractatus as sources of his own - of how and why differential equations work so nicely in modeling physical systems.

28. Two functions g and g' will be isomorphic just in case there is a mapping fI such that (a) both for any item x in the domain of g, fI(x) is in the domain of g', and for any item x in the domain of g', there is an item y in the domain of g such that fI(y)=x, (b) for any two items x and x' in the domain of g, if xx', then fI(x)fI(x'), and (c) for any two items x and x' in the domain of g, if g(x)=g(x'), then g'(fI(x))=g'(fI(x')). I suppose that for a given infinite translation function f, there are infinitely many translation functions isomorphic with f since there are infinitely many sentences to construct such functions from.X

29. Be careful not to confuse the notions of a translational theory of meaning and a translation-theory for a language L. The former is a certain kind of theory of the actual-language relation. The latter is merely a statement of a finite definition of a function from each sentence of a language L to its M-translation. On the present suggestions, Schiffer is attempting to provide a translational theory of meaning by supposing that to speak a language L is to stand in a certain cognitive relation to a translation-theory for L.

30. Schiffer has told me in conversation that he would like L-determining translators to be understood as M-sentences in somewhat the way I have just outlined things here. But there is quite a bit of tension here that will become quite vivid when I discuss my second counterexample to Schiffer's theory on the present reading below. sent sNote also that when I say that an Lto p-determining translator construed in the present way determines exactly one translation-function, I of course mean not that the function has no members which are themselves functions, but rather that there is no function that is a proper superset of the function determined by an L-determining translator. I am speaking somewhat loosely then, but I expect that this should be clear enough.

31. If one took the present point more seriously, there is the following point too. Whereas it is somehow sort of plausible that we have some sorts of unconscious knowledge of the things the compositional-semantics theorist of meaning wants to say that we have knowledge of - things like word-meanings and denotations and conditions on syntactic structures - it really doesn't seem too plausible on the face of things that we know any of this neurological stuff that on Schiffer's theory we would have to know, at least on the present reading. Compare the questions, "do I really unconsciously know anything about my own language of thought?" and "do I really unconsciously know that a sentence of the form 'some F is G' is true just in case there is something in the extension of 'F' that is also in the extension of 'G'?" I wouldn't be all that surprised to learn that the answer to the latter question is yes. But it would take an awful lot, I think, to get me to understand how the answer to the former question could be yes. So, on this score, it seems that compositional-semantic theories of meaning are somewhat more plausible than the present sort of translational theory of meaning. But I won't pursue this line any further.i

32. I always took Schiffer's character Harvey from chapter 7 of Schiffer (1987a) to work by some such hard-wired process, so I was surprised to learn from Schiffer that he would like his L-determining translators to be understood as M-sentences in the way I discussed in the main text above.

33. Harvey is a character Schiffer first introduced in chapter 7 of Schiffer (1987a) who processes utterances via possession of a non-semantic translation scheme of some sort.

34. Schiffer (1993), p. 245.z

35. Note that fL is the set-theoretic function, an abstract object, and not an L-determining translator of some sort.

36. Schiffer, in Schiffer (1993), refers to an L-determining translator as "the translator", but don't confuse my use of the term "translator" in my expression "actual-translator relation" with Schiffer's. Again, my "translators" are abstract objects, not brain states or mechanisms of some kind. They are the translation-functions which map L-sentences, for some language L, onto their M-translations.: