Abstract
Paul Horwich’s Use Theory of Meaning (UTM) depends on his rejection of Paul Grice’s distinction between natural and non-natural meaning and his Univocality of Meaning Thesis, as he wishes to deflate the meaning-relation to usage. Horwich’s programme of deflating the meaning-relation (i.e. how words, sentences, etc., acquire meaning) to some basic regularity of usage cannot be carried through if the meaning-relation depends on the minds of users. Here, I first give a somewhat detailed account of the distinction between natural and non-natural meaning in order to set the stage for Horwich’s critique of it. I then present Horwich’s critique of the distinction and show how that rejection accords with his overall view of meaning as use. Horwich’s rejection of the distinction between natural and non-natural meaning, I argue in the last section, is ill founded, and because UTM depends on this rejection, UTM is stillborn.
This is a preview of subscription content, log in via an institution to check access.
Similar content being viewed by others
Notes
‘I have had a long-standing sympathy for Wittgenstein’s view that the meaning of a word derives from its use’ (Horwich 1998, p. vii). Sellars is also identified as one of the philosophers belonging to the group of ‘philosophers whose views could reasonably be labelled use theories of meaning’ (Horwich 1998, p. 43).
Christopher Gauker also rejects Grice’s distinction between natural and non-natural meaning. See Gauker 1994 and 2003. I will not address Gauker’s reasons for rejecting the distinction here. See Borge forthcoming, for a critique of Gauker’s position.
Any use theory of meaning that is either deflationary in spirit or in some other way tries to bypass any dependency on users’ minds, like, for example, Sellars’ theory, which is more behaviouristic in spirit, will have to reject the distinction between natural and non-natural meaning and, most likely, run into the same sort of problems as UTM does.
The formulations of the test are mine, as are the names, but they are otherwise in keeping with Grice. Grice distinguishes five differences between natural and non-natural meaning (Grice 1957, pp. 213–214) and so does, for example, Martinich (Martinich 1984, pp. 113–114). My Agent Test contains two of the differences between natural and non-natural meaning pointed out by both Grice and Martinich.
Notice that ‘His spots mean measles’ is not of the syntactic form of ‘His spots mean that p’, because ‘measles’ is not a sentence, but one could say that it is shorthand for ‘His spots mean that he has measles’.
See Dretske for a similar point (Dretske 1991, pp. 55–56).
I will not dwell on or develop this fact here.
Here is a complication for the picture I am presenting. Some terms are, according to some philosophers, object dependent. They cannot have meaning unless they succeed in picking out an object. Suppose ‘Bill Clinton’ is one such term. The sentence ‘Bill Clinton exists’ means that Bill Clinton exists and also entails that Bill Clinton exists, but the sentence ought still, within the larger scheme of things, be a case of non-natural meaning. There is no space in this paper to pursue this complication any further, other than to mention that if you argue for object-dependent linguistic terms, then you need to make amendments to your overall view of non-natural meaning.
This, I believe, is how Grice originally intended his example of the recent budget to function. There is, however, a different way to interpret the sentence ‘The recent budget means that we shall have a hard year’. One could think that the budget might mean we will have a hard year by not depicting our situation correctly, but instead reflecting excessive optimism, and an observer might then notice that it will have negative consequences. Here the word ‘mean’ can be substituted with ‘has as a consequence’; ‘The recent budget has as a consequence that we shall have a hard year’. Similarly, ‘Ringo’s eating habits mean that he will become fat’ should be read as ‘Ringo’s eating habits have as a consequence that he will become fat’, and so on. Compare the sentences (i) ‘His excessive drinking means that he is an alcoholic’ and (ii) ‘His excessive drinking means that he will become an alcoholic’. The case described in (i) is a case of natural meaning, but what about (ii)? It seems to me that it should count as a case of natural meaning, as the case is similar to the example of ‘Black clouds mean that it will rain’ –’Black clouds have as a consequence that it will rain’— and the latter is clearly a case of natural meaning. If (ii) counts as a case of natural meaning, then the variant of the recent budget case under discussion, where the recent budget has as a consequence that we shall have a hard year, should also count as a case of natural meaning.
This, of course, does not mean that one cannot imagine other types of scenarios where other types of correlations hold between clothing and temperatures. For example, suppose that people are scantily clad inside an apartment building. Suppose, furthermore, that I know that the heating system works by overheating whenever it is cold outside. I might then say ‘The way people are dressed today means that it is cold outside’. If the people clothing in the building reflects the temperature in the building and I am correct about the heating system, then I see no reason why this should not count as a case of natural meaning.
Another example of a non-causal natural meaning case could be the following. The fact that it is a big natural number means that there are lots of numbers less than it.
A critic might at this point complain that it is not clear why restatement in itself supports factivity. That you can restate ‘His spots meant measles’ as ‘The fact that…’, the critic continues, only shows that the original sentence requires that it has to be a fact that he had spots, not that it has to be a fact that he has measles. This complaint only gets its bite by denying the conclusions reached in the Deceptive Child case and is, as such, misguided. Let me repeat the line of thinking in the Deceptive Child case. If the spots meant measles, then he has measles. Conversely, if he does not have measles, then the spots did not mean measles (even though this is what the child’s mother took them to mean).
Notice that there are some types of natural meaning connections that – at first glance — might look like counterexamples to the Factivity Thesis, but that are not. One can for example say: ‘X meant (naturally) the tiger was about to attack and so I killed it, and so it didn’t attack’. This, however, is no counterexample to factivity, because ‘The tiger was about to attack’ does not require that it attacked.
This is, of course, how it is in the normal case, but that is not to say that one cannot imagine a case where measles meant something non-naturally. Suppose the masters used slaves as road signs and suppose, furthermore, that measles on a road sign, i.e. a slave, meant stop. In this case it seems right to say that the measles or measles on a slave non-naturally meant stop and that the masters used the measles or the slaves with the measles to mean stop. Notice that one and the same item may have both natural and non-natural meaning and that they may coincide.
One could claim that there is a superior being, God, behind every natural meaning-relation, in which case I would have to rewrite my explanation of natural meaning and the natural meaning-relation in terms of minds that are not God’s or God-like. Hereafter I will ignore this complication.
Hereafter, I will accept this basic assumption and not challenge it. One might, however, suspect that the reverse compositionality argument made by Fodor and Lepore (Fodor and Lepore 2001) will be troublesome for Horwich.
Not all use theorists of meaning would be happy with Horwich’s conception of a use theory of meaning. A critic might object that the very point of any use theory of meaning is to avoid problematic or suspect types of entities – entities that go under names like ‘sinn’, ‘sense’, ‘notion’ ‘abstract object’, and, indeed, ‘meaning’. Here is what, for example, Willard Van Orman Quine writes: ‘John Dewey, and in later years Ludwig Wittgenstein, stressed rather that there is no more to the meaning of an expression than the overt use that we make of the expression. Language is a skill (…) Now we find me urging that there is no place in the theory of meaning for meanings, commonly so called. Meaning, or use, yes; meanings, no’ (Quine 1987, pp. 130–131). I will not pursue this complication any further here.
I am here following Horwich’s practice of marking concepts with capital letters.
This is not Horwich’s own example of synonymy, but he likes using the word ‘perro’ in his examples and he uses it throughout the book Meaning, synonymous with the English word ‘dog’.
Devitt calls Horwich’s argument against Grice’s distinction ‘The Univocality-of-Meaning Argument’, and Devitt correctly identifies it as the first argument Horwich presents in his book from 1998 (Devitt 2002, p. 115). There is much to agree with in Devitt’s critique of Horwich’s theory of meaning, particularly Devitt’s claim that Horwich unduly neglects the relation between language and thought, and that the latter needs to be given priority. But Devitt seems to accept Horwich’s critique of Grice’s distinction between natural and non-natural meaning, Devitt’s only complain being that ‘univocality gives no preference to the use theory of concepts over any other’ (Devitt 2002, p. 115).
In Reflections on Meaning (2005), Horwich attempts to distance himself from the idea that the meaning- relation is an indication-relation. Horwich writes that ‘[t]hus UTM’s claim, that to have a certain meaning is to exemplify a certain use-property, might be factored into an analysis of “mean” as “exemplify”, and an analysis of meanings as use-properties. The idea is that just as a flower has a colour and a stick has a length, so a sound has a meaning; just as colours and lengths are properties, meanings are also properties. Therefore the logical form of a simple meaning-fact is “M(ws)”, which entails “S”s w has (or exemplifies) the property of M-ness. And each such properties, M-ness, reduces to a use-property’ (Horwich 2005, p. 34). Unfortunately, Horwich says nothing about how this change would influence his rejection of the distinction between natural and non-natural meaning. If he does not reject the distinction, his deflation of the meaning-relation to his use properties fails to get off the ground. My critique of Horwich is not undermined by the changes made by Horwich in his Reflections on Meaning. Because Horwich does not provide us with a new way of rejecting the distinction between natural meaning and non-natural meaning, we might safely assume that his rejection of the distinction in Meaning still holds and that whatever holds for ‘indication’ also holds for ‘exemplify’. For this reason, I will stick to Horwich’s arguments from Meaning.
This is why I do not think it makes much difference that Horwich, in his book from 2005, wants to talk about an exemplification-relation instead of an indication-relation, as both relations are going to give the hearer reason to believe whatever is exemplified according to UTM, and both relations are going to be factive.
This is inaccurate. It is not the observation of smoke that indicates fire. It is rather the presence or occurrence of smoke that indicates fire. This though is easily fixed in the way suggested here, and I will not dwell on it any further.
The argument for this claim is found in three different examples of the acceptance of basic meaning-properties on p. 45 in Meaning. One of the examples goes as follows; ‘the explanatorily fundamental acceptance property underlying our use of “red” is (roughly) the disposition to apply “red” to an observed surface when and only when it is clearly red’ (Horwich 1998, p. 45). Horwich also writes that ‘the meaning-constituting property of a word is the characteristic which provides the explanatory basis of its overall use. And (…) that this characteristic is itself a regularity of use –something like “the fundamental law” governing the behaviour of the word’ (Horwich 1998, p. 170).
See also Horwich 2005, pp. 40–41.
Perhaps Horwich subscribes to Timothy Williamson’s theory of knowledge, where what you have evidence for is what you know, because, according to Williamson, ‘one’s total evidence is simply one’s total knowledge’ (Williamson 2000, p. 9).
Horwich’s talk of exemplification does not make any difference here.
Grice himself, it seems, would not necessarily be opposed to couching the difference between natural meaning and non-natural meaning in terms of different indication-relations. In sections from ‘Meaning Revisited’ where he tries to make good on the idea that ‘one would represent the cases of nonnatural meaning as being descendants from, in a sense of “descendant” which would suggest that they were derivative from and analogous to, cases of natural meaning’ (Grice 1982, p. 292), he writes ‘that X’s nonnatural production of an expression of pain is not to be supposed to indicate every feature which would be indicated by a natural production’ (Grice 1982, p. 295).
The reader may wonder why I have not yet raised the following concern about Grice’s distinction between natural and non-natural meaning: if ‘mean’ is ambiguous between a natural sense and a non-natural sense, then some test of ambiguity ought to apply to claims about ambiguity in relation to ‘means’. One could, for example, try to take one of Zwicky and Saduck’s tests of ambiguity and apply them to the two alleged senses of ‘mean’ (Zwicky and Sadock 1975). Take the conjunction reduction test for testing lexical ambiguity, for example. The evidence for ‘bat’ being ambiguous is that one cannot ‘reduce’ the sentences ‘Ringo has a bat in his attic’ (cricket bat) and ‘George has a bat in his attic’ (vampire bat) to ‘Ringo and George have bats in their attic’. How do our two types of meaning fare in the conjunction reduction test? Consider the following; ‘Ringo encountered something that meant fire’ (Ringo saw smoke) and ‘George encountered something that meant fire’ (George heard the sentence ‘There is a fire’). In this connection is it good or bad to say; ‘Ringo and George encountered something that meant fire’? To my ears, this does not sound too bad. It is, however, to be expected that this sentence will not sound as bad as the sentence containing the two senses of ‘bat’. If one thinks, as I do, that Grice is correct in thinking (see quotes above) that the natural sense and the non-natural sense of ‘mean’ are not only merely related (non-natural meaning being descendent from natural meaning), but also have a common root (in both cases where ‘x means y’, then y is a consequence of x), then one should not expect the sentence above to sound as bad as the example with the ambiguous word ‘bat’, as the alleged two senses of the word ‘mean’, unlike the two senses of the word ‘bat’, are related. Dennis Stampe, for example, writes about ‘the polysemy of the verb mean’ (Stampe 1974, p. 283). While lexical ambiguity means that one word has multiple meanings, polysemy means that a word has multiple meanings that are closely related. That is, if you think that polysemy is a type of ambiguity as, among others, Cruse and Crystal do (Cruse 1986, p. 80 and Crystal 1992, p. 307). To complicate matters, there are other researchers like Zhang 1996 and Tuggy 1993 who distinguish polysemy from ambiguity. If, however, we look at a classic polysemous term like ‘in’, it is singled out by Zwicky and Saduck’s tests. Take the sentence ‘She came in a flood of tears’ and ‘She came in a Mini Morris’. These two do not reduce to ‘She came in a flood of tears and a Mini Morris’. A more promising place to look for a notion of ambiguity that fits Grice’s two senses of ‘mean’ would be in Aristotle’s Metaphysics and his notion of ‘pros hen’ ambiguity (Aristotle Metaphysics Γ.2., 1003a35ff, reprinted Aristotle 1984). Aristotle’s example is the term ‘healthy’. ‘Healthy’ is ambiguous between a core meaning, that of health, the thing shared by all healthy beings, and derivative meanings, things that indicate or contribute to health (like my food being healthy, my complexion being healthy, etc.). Consider the sentences ‘Ringo’s urine is healthy’ and ‘Ringo’s food is healthy’. The combination of these two sentences brings out the ambiguity: ‘Ringo’s urine and food are healthy’. This latter sentence is much less bad than ‘Ringo and George have bats in their attic’, because the terms are so related, but it still shows, according to philosophical orthodoxy, the ambiguity of ‘healthy’. I find the sentence ‘The smoke and his remarks meant that there is a fire’ just as bad as the sentence ‘Ringo’s urine and food are healthy’. One might relate Aristotle’s idea of a primary sense of ‘healthy’ and certain closely related, but derivative, senses of ‘healthy’ (like exercise, food, complexion, urine, etc. being healthy) to Grice’s idea of ‘the cases of nonnatural meaning (…) being descendants from, in a sense of ‘descendant’ which would suggest that they were derivative from (…) cases of natural meaning’ (Grice 1982, p. 292) and argue that the ambiguity between the natural and non-natural sense of ‘means’ is of the pros hen type.
Gupta 2003, pp. 664–665.
One might also point out that this line of thought does not sit very comfortably with his other claim that ‘most of a person’s public language is intimately, automatically, and non-intentionally linked to his language of thought’ (Horwich 2005, p. 60). I am not, however, going to pursue this any further here.
If we look at how Aristotle writes about this in the often-quoted passage from Metaphysics, we find that, contrary to how Tarski interpreted him (Tarski 1944), it is not so much about what it is for something to be true or false, but rather what it is to say something true or false. ‘To say of what is that it is not, or of what is not that it is, is false, while to say of what is that it is, and of what is not that it is not, is true’ (Aristotle Metaphysics, 1011B 25–27, reprinted Aristotle 1984).
Appealing to generics will not help. Of course, it is a reasonable thing to say that, for example, generically whenever someone uses the word ‘dog’ then it is generally true that a certain concept is deployed or present in the mind of the speaker, and that the generic permits individual counterinstances. That, however, is to deny that that ‘what is meant must obtain’ (Horwich 1998, p. 20), and so it is not the theory that Horwich defends.
Horwich 1998, pp. 46–47.
Notice that not everyone accepts the idea that we generally say true things when we use language. Peter Unger has argued that most of what we say is false (Unger 1975, pp. 252–271).
Allan Gibbard has complained that ‘Horwich speaks, though, as if there is such a thing as the ideal law that governs the deployment of a word. That does not fit the way idealization works in other sciences’ (Gibbard 2008, p. 150). I will not pursue Gibbard’s point any further here.
The idea of division of linguistic labour was originally introduced in Putnam (1975).
This paper was presented at University of Tromsø and ECAP 6: Sixth European Congress of Analytic Philosophy in Krakow. I am grateful to all the participants. I also thank Jan Harald Alnes, Michael Devitt, John Hawthorne, Ernie Lepore, Tom McKay, Mark Scala, Adam Sennet and Margrethe Bruun Vaage for comments and critique. The paper is part of the research project CCCOM, Communication in Context, supported by the European Science Foundation within the EuroUnderstanding EUROCORES programme.
References
Aristotle (1984). Metaphysics. Translated by W.D. Ross in J. Barnes (Ed.), The complete works of Aristotle, vol. 2. Princeton, NJ: Princeton University Press.
Bezuidenhout, A. (1998). Is verbal communication a purely preservative process? Philosophical Review, 107, 261–288.
Block, N. (1986). Advertisement for semantics for psychology. Midwest Studies in Philosophy, 10, 615–678.
Borge, S. (2003). The word of others. Journal of Applied Logic, 1(1), 107–118.
Borge, S. (Forthcoming). ‘In defence of the received view’. Philosophical Psychology.
Burge, T. (1993). Content preservation. Philosophical Review, 102, 457–488.
Burge, T. (1997). Interlocution, perception, and memory. Philosophical Studies, 86, 21–47.
Coady, C. A. J. (1992). Testimony: A philosophical study. Oxford: Oxford University Press.
Cruse, D. A. (1986). Lexical semantics. Cambridge: Cambridge University Press.
Crystal, D. (1992). An encyclopedic dictionary of language and linguistics. Oxford: Blackwell.
Devitt, M. (2002). Meaning and use. Philosophy and Phenomenological Research, 65, 106–121.
Dretske, F. (1991). Explaining behaviour. Cambridge: MIT Press.
Field, H. (1977). Logic, meaning and conceptual role. Journal of Philosophy, 69, 379–409.
Fodor, J., & Lepore, E. (2001). ‘Why compositionality won’t go away’. In The compositionality papers (pp. 43–62). Oxford: Clarendon Press, 2002.
Gauker, C. (1994). Thinking out loud. Princeton: Princeton University Press.
Gauker, C. (2003). Words without meaning. Cambridge: MIT Press.
Gibbard, A. (2008). Horwich on meaning. Mind, 117, 141–166.
Grice, H. P. (1957). ‘Meaning’. In Studies in the way of words (pp. 213–223). Cambridge, Mass: Harvard University Press, 1989.
Grice, H. P. (1982). ‘Meaning revisited’. In Studies in the way of words (pp. 283–303). Cambridge, Mass: Harvard University Press, 1989.
Grice, H. P. (1989). ‘Retrospective epilogue’. In Studies in the way of words (pp. 339–386). Cambridge, Mass: Harvard University Press.
Gupta, A. (2003). Deflationism, the problem of representation, and Horwich’s use theory of meaning. Philosophy and Phenomenological Research, 67, 654–666.
Harman, G. (1982). Conceptual role semantics. Notre Dame Journal of Formal Logic, 28, 252–256.
Harman, G. (1987). (Non-solipsistic) conceptual role semantics. In E. Lepore (Ed.), New directions in semantics (pp. 55–81). London: Academic Press.
Horwich, P. (1998). Meaning. Oxford: Clarendon Press.
Horwich, P. (2005). Reflections on meaning. Oxford: Clarendon Press.
Martinich, A. P. (1984). Communication and reference. Berlin: Walter de Gryter.
Peacocke, C. (1992). A study of concepts. Cambridge: MIT Press.
Putnam, H. (1975). ‘The meaning of ‘meaning’. In Mind, language and reality: Philosophical papers 2 (pp. 215–271). Cambridge: Cambridge University Press.
Quine, W. V. O. (1987). Quiddities: An intermittently philosophical dictionary. Cambridge: Harvard University Press.
Sellars, W. (1969). Language as thought and as communication. Philosophy and Phenomenological Research, 29, 506–527.
Stampe, D. (1974). Towards a grammar of meaning. In G. Harman (Ed.), On Noam Chomsky. Critical essays (pp. 267–302). Garden City: Anchor Press/Doubleday.
Tarski, A. (1944). The semantic conception of truth and the foundation of semantics. Philosophy and Phenomenology, 4, 341–375.
Tuggy, D. (1993). Ambiguity, polysemy, and vagueness. Cognitive Linguistics, 4, 273–291.
Unger, P. (1975). Ignorance: A case study for skepticism. Oxford: Oxford University Press.
Williamson, T. (2000). Knowledge and its limits. Oxford: Oxford University Press.
Wittgenstein, L. (1953). Philosophische Untersuchungen. Oxford: Blackwell.
Wright, C. (2001). Rails to infinity. Cambridge: Harvard University Press.
Zhang, Q. (1996). Fuzziness – vagueness – generality – ambiguity. Journal of Pragmatics, 29, 13–32.
Zwicky, A., & Sadock, J. (1975). Ambiguity tests and how to fail them. In J. Kimball (Ed.), Syntax and semantics 4 (pp. 1–36). New York: Academic Press.
Author information
Authors and Affiliations
Corresponding author
Rights and permissions
About this article
Cite this article
Borge, S. Horwich on Natural and Non-Natural Meaning. Acta Anal 29, 229–253 (2014). https://doi.org/10.1007/s12136-013-0196-2
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s12136-013-0196-2