wilson - grice on meaning- the ultimate counter-example (1970)

Upload: ninni-andersson

Post on 07-Apr-2018

225 views

Category:

Documents


0 download

TRANSCRIPT

  • 8/3/2019 Wilson - Grice on Meaning- The Ultimate Counter-Example (1970)

    1/9

    Grice on Meaning: The Ultimate Counter-ExampleAuthor(s): N. L. WilsonReviewed work(s):Source: Nos, Vol. 4, No. 3 (Sep., 1970), pp. 295-302Published by: Blackwell PublishingStable URL: http://www.jstor.org/stable/2214430 .Accessed: 01/11/2011 07:37

    Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .http://www.jstor.org/page/info/about/policies/terms.jsp

    JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new formsof scholarship. For more information about JSTOR, please contact [email protected].

    Blackwell Publishing is collaborating with JSTOR to digitize, preserve and extend access to Nos.

    http://www.jstor.org

    http://www.jstor.org/action/showPublisher?publisherCode=blackhttp://www.jstor.org/stable/2214430?origin=JSTOR-pdfhttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/stable/2214430?origin=JSTOR-pdfhttp://www.jstor.org/action/showPublisher?publisherCode=black
  • 8/3/2019 Wilson - Grice on Meaning- The Ultimate Counter-Example (1970)

    2/9

    Grice on Meaning: The UltimateCounter-example'

    N. L. WILSONMCMASTER UNIVERSITY, CANADA

    Recently (in [3]), H. P. Grice has taken another crack atelaborating and defending the theory of meaning first offered in[2]. I shall try (I) to show that Grice is on the wrong track ofwhat he is ostensibly after, (II) to show that he is on the righttrack of something else, and (III) to sketch what I believe to bethe right track of what Grice is ostensibly after.

    I

    Grice concedes (on p. 151 of [3]) that the definiendum ofprimary interest is (A): "By uttering x U meant that p." Neverthe-less he devoted most of the 1957 paper and a considerable portionof the present one to a treatment of (B), "By uttering x U meantsomething." Then, elaborating on that treatment he proceeds to ananalysis of (A). For the moment I shall deal with (B) and I shalldiscuss the initial unrefined definiens (taken over from the 1957paper) because my objections will apply equally to the refinedversions. The initial definition s (more or less, p. 151 of [3]):

    U meant something by uttering x iff, for some audienceA, U uttered x, intending (1) A to produce a particular[emphasis added] response r, (2) A to think (recognize)that U intends (1), [and] (3) A to fulfill (1) on the basis

    of his fulfillment of (2).1 Some of the ideas in this paper stem from the period during which the

    author was a recipient of NSF grant GS572.

    s95

  • 8/3/2019 Wilson - Grice on Meaning- The Ultimate Counter-Example (1970)

    3/9

    296 NOU'S

    I suggest that we have here a confusion between what theutterer ntended to say (convey)-i.e., the alleged fact he intendedto report-and what he intended to accomplish by saying it. [AtOberlin I taxed Grice with this point. His reply, if I remember itcorrectly, was: "But that's my thesis I may be mistaken, but I'mnot confused."] I shall speak of these intentions respectively as theprimary (linguistic) intention and the secondary intention. Theremight be cases where we would to talk of tertiary ntentions. At anyrate it seems to me that the primary and secondary intentions mayvary independently, and, if so, Grice's analysis is not just faulty,but on the wrong track. If I am right it would explain why so manypeople have had so much fun playing counter-examples with Grice.-To make out my case I shall have to move back and forth betweenthe two definienda above, (A) and (B).

    Suppose someone offers 'Qx and Rx' as an analysis of 'Px'. Toattack the analysis one invents a (consistent) scenario in whicheither a is Q and R but not P (too broad!) or in which a is P butnot Q and R (too narrow!). I shall do the latter. Suppose (andhere beginneth the first scenario) I am conversing with Grice. I say'Snow is white'. By uttering 'Snow is white' I mean that snow iswhite. It follows that by uttering 'Snow is white" I mean some-thing. According to Grice it follows that I intend (1), (2) and (3)above. Now I do intend to say (report the fact) that snow iswhite, but the only secondary intention I have is to avoid havingany of the intentions Grice attributes to me in this or any sub-sequent papers of his. If Grice amends his definition, I amend myscenario. He can't win. That is why I call this the ultimate counter-example. One might cavil at the suggestion that one can intend to

    avoid intending to bring about y, but it hardly matters, since onecan certainly avoid trying to accomplish y.Now the foregoing may sound a bit like the open question

    argument. It is to the extent that I dogmatically claim that in-tuitively we can see that there is no inconsistency in a scenariowhose last act is:

    Grice: So. By uttering 'Snow is white' you meant something. Didyou intend to produce this response in me?

    Myself: (correctly-that's essential): No sir!Grice: That response?Myself: No. There is no response such that I intended to elicit

    that response from you. But you can't claim that I wasn'treally talking to you. Of course I was. If you hadn't heard

  • 8/3/2019 Wilson - Grice on Meaning- The Ultimate Counter-Example (1970)

    4/9

    GRICE ON MEANING 297

    me I would have repeated myself in a louder voice. As amatter of fact I didn't even intend that you should abandonyour views. If I had, I might, of course, have played intointo your hands. In any case-and this is more to thepoint-you can only intend to produce an effect if youdeem it possible to do so (as you have pointed out) andeverybody knows it is not possible to persuade aphilosopher to give up a cherished doctrine. To repeat, Iwasn't trying to produce any particular response n you atall. My only intention was to behave in a counter-exemplary way.

    Now let us turn for the moment to (A), the more interestingof the two definienda. It is central to Grice's view that "genericdifferences n type of response would be connected with genericdifferences within what is meant" ([3], p. 165). Now the word"generic" may, I think, be ignored; otherwise it constitutes anescape hatch that would trivialize the position. (How generic?!)We have the claim, roughly, that if the definition s of the followingform:

    By uttering x U meant that p iff, for some A, U intended Ato produce response r and ....

    then there is a specifiable covariance between p and r, betweenwhat is meant and the response that the utterer intends to elicit.Again, a counter-example s forthcoming. The previous scenariowill serve and moreover, we may relax a restriction. Now I mayindeed, without imperiling my position, utter 'Snow is white' mean-in that snow is white, with the sole intention (however naive) ofdislodging Grice from his position. Here there is no correlationbetween what I meant and what psychic change I intended toproduce in my interlocutor.

    Now, however, we must wonder how much importance s tobe attached to a successful counter-example, nd the more ingeniousand intricate the scenario, the more we wonder. Perhaps no inter-esting philosophical position is immune to attack by some kind ofcounter-example. may say, too, that I don't think a position is tobe dismissed ust because it has a certain prima facie implausibility.But where there is that implausibility hen a counter-example mayserve to nail down the case against the position beyond reasonabledoubt. And the covariance claim, the claim that what is meant andthe hoped for response co-vary, does seem implausible. If I wish

  • 8/3/2019 Wilson - Grice on Meaning- The Ultimate Counter-Example (1970)

    5/9

    298 NOtS

    my guests to leave, there are any number of different things Imight say (and mean by what I say) in order to shoo them out.On the other hand, if I say, "Ifs getting pretty late," meaning thatit's getting pretty late, there might be any one of a number ofdifferent things I expect of my audience. There just is not the kindof covariance necessary to make Grice's theory go.

    II

    At this point I shall take time out to indulge in a bit of sym-pathetic exegesis. If a philosopher mentioned the problem (let ussay) of analyzing "x causes y" and immediately shifted his attentionto "x causes something" we should think he was wasting our time.Grice's concern with "By uttering x U meant something"' s puzzlingbut I don't believe it to be in the same case. The 1957 paper canbe read as simply elaborating the distinction between those caseswhere we have "natural" meaning (symptomatizing, being a sym-tom of) and those cases where we have "non-natural" meaning(signaling). The following scenarios are reminiscent of Grice'sfrown examples in the 1957 paper, but my treatment is slightlydifferent.

    (a) I cough (spontaneously). You infer that I have a chest cold.The cough meansN (symptomatizes) a chest cold. (Cf. "Smokemeans fire", "Those spots mean measles".)

    (b) I say to you, "I have a chest cold." By my utterance I signalto you that I have a chest cold. You infer that I have a chest cold.But this inference-if that is what it should be called-is vastly

    different fromthat in

    (a). There you observe my response andengage in causal theorizing about it. Here you "read" my response,as Austin puts it. There is "uptake" r understanding. And of coursewhat is involved in having this kind of understanding s preciselywhat is of such very great interest to a language-theoretician.

    (c) My boss appears at my door. I don't want to dismiss him andat the same time I don't want to ask him in. Knowing that he has ahorror of germs I cough several times, deliberately, hus simulatingcase (a). ["That's quite a cold you've got there. I'll be on my wayand let you take care of it."] (Sub-cases: (i) I really do have acold and (ii) I really don't have a cold.) Here we have a case ofwhat could be called quasi-signaling. There is the intention of con-veying information (or, possibly, misinformation) and at the same

  • 8/3/2019 Wilson - Grice on Meaning- The Ultimate Counter-Example (1970)

    6/9

    GRICE ON MEANING 299

    time the intention of concealing that there is that intention. I couldhave signaled that I had a cold simply by saying so [but since myboss is slightly paranoid I thought it more prudent to avoid sayinganything he could misconstrue and instead, to jockey him intotaking the initiative].

    Now it is possible to signal y by signaling x. [I signal to myhostess that it has been a very pleasant evening by saying so in anappropriate one of voice. And thereby signal that I am gratefulfor having been invited.] And one can quasi-signal by signaling. Ishall not give an example. Suffice it to say that Lifemanship n its

    most advanced forms depends on this fact. But it does not seempossible to signal by quasi-signaling. And the reason is that the co-vertness of the original quasi-signaling will carry right through. TheStampe-Grice ycophantic bridge player (in [3], p. 154) is, I judge,a case of quasi-signaling by simulated quasi-signaling. He is quasi-signaling to his boss that he wants him (the boss) to know that hehas a good hand and he does so by a too transparent quasi-signal-the simulation of a simulated smile (supposing that's possible).What is covert is the deliberate transparency f the "inside" quasi-

    signal. What is clear in any case is that the variety of possiblecombinations must be just about limitless.)

    At any rate Grice's papers may be read as pointing out thatwhatever else signaling is, it is of the essence of signaling (in con-trast to its cousins) that the signaler intend that the signalee shouldrecognize that the signaler is deliberately signaling and, moreover,that the signaler ntend, or at least take for granted, hat the signaleeshould recognize that he is (primarily) signaling this and not some-thing else. (Grice has bent his efforts to breaking the above cir-

    cularity.) This observation on signaling seems to me to be trueand interesting. It is, however, a contribution to what would becalled the pure sociology of speech behavior, not to semantics ormeaning theory. To the latter we now turn.

    III

    I want to go back and consider (A), "By uttering x U meantthat p". It seems to me that one can get some leverage on theanalysis of (A) by asking: How does one attack the question 'Vhatdid U mean by uttering x?"? Suppose I am crossing the KalahariDesert. An African comes up to me and says (utters), 'Hut sutrawson ona rillaraw'. How do I find out what he means? I do not

  • 8/3/2019 Wilson - Grice on Meaning- The Ultimate Counter-Example (1970)

    7/9

    300 NOUS

    inquire into his intentions. Once I have deciphered what he means,then I may be able to hazard a guess as to what his intentions werein regard to me. What I do is follow him back to his village, whipout my trusty tape-recorder and make motions like a field-workerin linguistics. How we do field-work is a question of theoreticaldescriptive inguistics. I shall leave that to the experts and insteaddo theoretical descriptive semantics. The latter is presumably ome-thing like the former but is much more artificial. (We cheat indifferent ways. )

    An illustration. The second sentence of III.50 of Thucydidesis 'Mutilenai6n eiche katheilon'. How do I find out what it means?I do not have a Greek-English exicon and I know only that thetext is about the Peloponnesian War. What I do is rent a computer,stuff it with all we know about the Peloponnesian War and thenfeed into it as a corpus of utterances the Greek text of Thucydides.If computers are half as good as they are supposed to be then thisone ought to be able to decipher the text before you can say 'Zeus',come up with a Greek-English dictionary, a translation of the textand, in particular, ell us that the sentence in question means thatthey razed the walls of the Mytilenaeans. Of course any corpus ofutterances can be given an arbitrary nterpretation-any numberof interpretations, or that matter. But not every corpus has a single,right interpretation, ight in the sense of being that interpretationunder which the largest possible number of sentences of the corpuswill be true. (What else could possibly make an interpretation"right"?) The computer operates in accordance with what I haveelsewhere called the Principle of Charity. It seeks that interpreta-tion which, in the light of what it knows of the facts, will maximize

    truth among the sentences of the corpus. In effect, it plays a verylarge number of games of twenty questions with the corpus andplays them simultaneously.

    The point is that it is no good asking the computer o deciphera single sentence. It cannot work with anything less than a fairlylarge corpus. Ask Ventris (See [1]). And of course the corpus mustbe "integral" n the sense of stemming from a single source, amonolingual ndividual or language community.

    As a first stab, then, let us try:

    By uttering x U meant something =df there is a corpus of utter-ances (including x) associated with U whose one right inter-pretation is such that under that interpretation x means that(or, is true if and only if) p.

  • 8/3/2019 Wilson - Grice on Meaning- The Ultimate Counter-Example (1970)

    8/9

    GCICE ON MEANING 301

    There is no circularity here. We explain 'he means p' in terms of'it means p' and the latter is understood from pure formal seman-tics. And the merit of this approach, f ultimately successful, is thatthe definiens contains no intentional words like "intends" or "be-lieves".

    By uttering x U meant something =df there is a corpus of utter-ances (including x) associated with U which has a right inter-pretation and endows x with an interpretation.

    It will be noted that the second of these definitions is in effect a

    definition of "Utterance x means something (is meaningful) andis not gibberish". The Grice I presented in ?II, on the other hand,is in effect, defining "Utterance x meansNN something i.e., is a caseof non-natural meaning (signaling) and is not a case of symptoma-tizing or quasi-signaling."

    Now the class of speech acts proper and the class of (lin-guistic) signalings are distinct. There can, I should think-Gricenotwithstanding-be speech acts which are not signalings; and onecan signal by spouting gibberish. (That is the point of Ziff's example

    of the irritable academic (in [5], p. 2)). At any rate it seems clearthat there are two quite distinct analysanda floating around hereand I strongly suspect that all the participants (including myself)in the discussions prompted by Grice's 1957 paper have simplyconfused them.

    Dennis Stampe (in [4]) has argued convincingly that mean-ing is not a doing. There are some relevant things that one can do,viz., utter a token of 'Snow is white', or report the fact that snow iswhite. But meaning that snow is white is not something that you

    can do. And I think we can see why. It has already been done, soto speak, in virtue of the corpus of utterances lying behind youand giving meaning to what you currently ay. To put it differently,it is always false to say anything like this: "When I said 'Snow iswhite' two minutes ago I meant by that utterance that sulphur isyellow." I just do not have the independent control over my mean-ings. Grice would point out that I can't utter 'Snow is white' andintend or reasonably expect to induce in my audience a belief thatsulphur s yellow. But-to probe a little deeper-the reason is thatboth of us are in a very real sense in the grip of our corpora, outpast responses and experiences. But I am not simply saying thatwe are creatures of habit. I am making the systematic point thatthe question "What does he mean?" can be answered, not with

  • 8/3/2019 Wilson - Grice on Meaning- The Ultimate Counter-Example (1970)

    9/9

    302 NOS

    reference to present psychological states but only with reference topast output, or at least to some corpus which he shares in part.We can break our linguistic patterns but that requires an explicitrepudiation of the past: "Now folks, from here on in, when I say'Snow is white' I shall mean that sulphur is yellow." (But thiskind of switch depends upon established linguistic convention, andI move that we all agree not to let the possibility further muddyour already muddy waters.)

    Grice has drawn attention to a puzzling and important eatureof signaling and done some major spade work. He is, I think, mis-taken in trying to analyze the meaning of an utterance in terms ofthe immediate, local social transaction that takes place when aperson utters a token with meaning. It seems to me that meaninghas to be approached holistically, in terms of a whole corpus ofutterances. And I think there are philosophical gains to be madeby trying to go as far as possible in treating the meaning of descrip-tive utterances without any reference to the social transaction.

    REFERENCES

    [1] John Chadwick, The Decipherment of Linear B, Cambridge, 1958.[2] H. P. Grice, "Meaning," Philosophical Review, LXVI (1957): 377-388.[3] H. P. Grice, "Utterer's meaning and intentions," Philosophical Review,

    LXXVIII (1969): 147-177.[4] Dennis Stampe, "Toward a grammar of meaning," Philosophical Review,

    LXXVII (1968): 137-173.[5] Paul Ziff, "On H. P. Grice's account of meaning," Analysis, XXVIII (1967):

    1-8.