Navigazione – Mappa del sito

HomeNumeri34The circle of deference proves th...

The circle of deference proves the normativity of semantics

Christopher Gauker
p. 181-198

Note dell’autore

I thank Alberto Voltolini for helpful comments on an earlier draft of this paper.

Testo integrale

1The question whether semantics is a normative discipline can be formulated as a question about the meaning of the word “means”. If I assert, “The word ‘gatto’ in Italian means cat, what have I done? The naturalist about meaning claims that I have asserted that a certain natural relation obtains between Italian speakers’ tokens of “gatto” and cats. Or at least, I have asserted something about the way Italian speakers use the word “gatto”, which way presumably has something to do with cats. The normativist claims, on the contrary, that what I have said is that in speaking Italian one ought to use the word “gatto” in a certain way, which way has something to do with cats. What I have done is endorse a certain proposal about how to use the word, which, if accepted, will have normative force.

2Normativism, as I am defining it, does not merely say that the truth of a claim about meaning has conditional consequences for what one ought to say. To say that a knife is sharp is simply to describe it, not to endorse any kind of rule. Still, the claim that a certain knife is sharp may have conditional normative consequences, such as that if one wishes to cut something and can choose between this knife (the sharp one) and that one (a dull one), one ought to choose this one. Similarly, one might hold that in saying “gatto” means cat one is only describing a certain relation or a certain use and still acknowledge conditional normative consequences such as that if one says of some object “Questo è un gatto” and in so doing wishes to say something true, then the thing referred to with “questo” ought to be a cat, and that if one uses “gatto” in speaking Italian and wishes to be understood by other Italian speakers, then what one is talking about ought to be a cat. Even a naturalist can grant that facts about meaning have such conditional normative consequences; but for the normativist, the normative import of statements about meaning goes well beyond this.

3Moreover, normativism does not claim merely that meaning claims describe prevailing norms. Suppose we say to someone, “You have a right to vote”. This could be taken as merely a description of the rights accorded to that person by the laws of the land. It expresses the same as, “In this country, you are granted the right to vote”. To say that much is not yet to endorse any right to vote for that person. The normativist about meaning does not say merely that assertions about meaning describe prevailing norms in this way — they do not merely say that in certain communities people have certain views about how a word ought to be used. In many cases we can merely describe norms in this way, as we can merely describe the rights that a political system accords. But normativism about meaning, as I define it, holds that the semantics of a language cannot be exhaustively expressed merely by describing semantic norms.

4In this paper, I will argue that normativism about meaning is true, and in the course of this defence I will define it more precisely. One way to argue for normativism would be by elimination: Every attempted naturalistic explication of semantic properties that has been tried so far has failed. Therefore, the only position that remains is one according to which statements ascribing meanings are endorsements of norms. However, I will attempt something more decisive, a refutation from obvious truths. The key premise of my argument will be that for every group of users of a word, the members of that group regard themselves as responsible to the usage of the other members of the group. Since they regard themselves as responsible to one another’s usage, we can call this fact the circle of deference.

5In characterizing the position of the naturalist, I will frequently speak of the uses of words. My assumption is that, for the naturalist, it is something about the use of a word that determines the meaning of the word. Presumably, the meaning-determining use of a word will include not only actual tokenings but nonactualized dispositions to use the word in counterfactual situations. Our epistemological problem, in determining what a word means, will be both to extrapolate the use from the actual and counterfactual uses and then to discover the meaning on the basis of the use. According to one view, the pertinent uses are uses in expressing thoughts. According to a different view, the pertinent uses are primarily uses in describing the world (directly, not just by expressing thoughts that describe the world). A better view, though still not my own, would be that the pertinent uses are uses in coordinating group action. The question of which of these characterizations of uses is correct, for purposes of specifying meanings, will not arise for me, however, since I will not take meaning to be determined by any kind of use. There are, of course, uses of a word, and we can call the collection of all past and present uses the use (so far). But my claim is that there is no such thing as the use of a word that we might treat as determining meaning.

6I will not have anything special to say about extensions and intensions. A commonplace view is that an intension is, or can be modelled as, a function that takes a world w as input and yields as output an extension-at-w. In denying that there will be a naturalistic account of meaning, I am denying as well that there will be a naturalistic account of extensions and intensions. But I will not proceed by attacking naturalistic conceptions of extensions and intensions, because, as I said, I will not be arguing by elimination. Whether there might be any place for the concepts of extension and intension in a normativistic conception of meaning is a question that I will not take up.

1 A Sense of Responsibility to the Community

7A prima facie objection to naturalism about meaning is that every member of a linguistic community recognizes that it is at least possible that he or she is misusing a word and stands ready to be corrected. In some cases, of course, misunderstanding is very unlikely. I am as confident as I am of anything that I know what the word “chair” means in English and that for many years my uses of the word “chair” have been correct. I cannot imagine how anyone could possibly persuade me that the thing I am sitting in now is not properly called “chair”. But even in this case, what I am confident of is that my uses have conformed to the meaning that the word has for others and that others will not try to correct me. My confidence does not consist in my arrogating to myself the right to decide what the word “chair” will mean among the people I talk to. Taking my own case as representative, this seems to show that we all think of meanings as standards that we are obliged to conform to. And in that case, if someone asserts that a certain word has a certain meaning, then, it seems, he or she must be taking a stand on what that standard is to which we are obliged to conform, just as the normativist claims.

  • 1 This would seem to be the position of Donald Davidson in his papers, “Communication and Convention” (...)

8This objection to naturalism rests on the assumption that where there is the possibility of correction, there must be some standard of correctness distinct from one’s own usage. Thus, one way to respond to the objection would be to simply deny this assumption. Someone might want to correct my usage on the assumption that I want my usage to conform to that of others, but there is no necessity in the assumption that I do want my usage to conform. There need not be any community-wide meaning at all. The meaning of a word, it may be said, is always the meaning that the word has for so-and-so. The meaning that a word has for one member of a linguistic community is likely to be very similar to the meaning that it has for another member of that same linguistic community. So if we make some assumptions about one interlocutor on the basis of our experience with another, we will not usually go far wrong. But in principle each of us could be a linguistic community of one. We could be a ship-load of pirates, speaking English and Italian and Russian and Arabic, provided each of us has been around long enough to have learned to understand the languages that the others speak to us. The important thing for purposes of communication is just that each interlocutor knows the meanings of every other interlocutors’ words when he or she uses them1.

9The problem with this response is that it does not really succeed in expunging a consciousness of and respect for norms from our conception of linguistic communication. Suppose, as this reply would have it, that every member of the community could speak his or her own language and that every other member of the community would be able to understand that language. Suppose that such a situation would not exceed any innate boundary on the number of languages a person can understand. Still, there could not be a society in which no member spoke the same language from one moment to the next. Even if the English-speaking pirate understood Russian, the Russian-speaking pirate would have to go on speaking Russian from one day to the next if the English-speaking pirate were to have any chance of understanding him. So the Russian-speaking pirate would have to hold himself accountable to the norms of his own language. (It would not have to be literally Russian, of course, the language spoken by millions of people in a place we call “Russia”). He would have to speak in such a way that his words will have a more or less constant meaning across time. Such constancy could arise by accident, without any effort on the part of the speaker. But insofar as a speaker reflects on his choice of words, he will recognize a need to conform to the norms of his own personal language. Even if he did not think of his obligation as an obligation to speak the same language across time, he would recognize a need to speak in such a way that his interlocutors could understand him, which would entail the requisite constancy. So if the initial observation that every speaker considers himself or herself obliged to conform to some kind of norms of usage merited an answer from the naturalist in the first place, it continues to do so under the pirate ship scenario.

  • 2 The pertinent conception of coordination problem is that which David Lewis so vividly explicated in (...)

10Since nothing was gained by denying it, let us retain the assumption that interlocutors typically conceive of themselves as speaking the same language. For each word in the lexicon, each interlocutor wants to give that word the same meaning as the other interlocutors give to it. We can put this by saying that each wishes to speak the same language as the other. Consequently, they face a sort of coordination problem2. X wants to speak the language of Y, and Y wants to speak the language of X. So Y wants to speak the language that X thinks Y is speaking, and X wants to speak the language that Y thinks X is speaking. So X wants to speak the language that Y thinks X thinks Y is speaking, and... so on. They can escape from this regress of reflections and coordinate their choice of language only if each makes his or her choice of language on some basis other than a consideration of which language the other one wants to speak. Insofar as they do not expect to coordinate in the choice of an entirely new language, each somehow grounds his or her choice in the uses of words that he or she has encountered in the past and bases on this a conception of the language spoken by the community that guides his or her own uses. Thus we seem to affirm that, as I said at the start, we all think of meanings as standards that we are obliged to conform to.

2 Merely Conditional Obligations

11There is another way for the naturalist to respond to the objection we started with, namely, to grant that we each think of meanings as standards that we are obliged to conform to, but then to deny that we should conclude from this that statements about meaning are endorsements of norms. The alternative is to say that meanings create at most conditional obligations, as I will now explain.

12The position of the naturalist is that the actual and counterfactual uses of words by the members of a linguistic community somehow determine a correct interpretation of the words, phrases, sentences and texts in the language of the community. By “interpretation” here I mean an assignment of meaning to each atomic lexical item and a means of composing the meanings of complex phrases, sentences and texts on the basis of these lexical items, their meanings and the manner of their composition. According to the naturalist, an ascription of meaning to an expression in this sense simply ascribes to the expression a property of the sort that the expression has as a consequence of the uses of words in the community. In making such ascriptions, no endorsement of norms need be involved.

  • 3 This is the only kind of semantic normativity that Åsa Maria Wikrforss is prepared to concede in (...)

13Nonetheless, the naturalist may grant that each member of the community is indeed under a kind of obligation to conform to standards, but that the obligation is only conditional on wanting to be understood. To each member of the community, we may say: “Given that word w has meaning m in your community, if you want to speak in a way that will be understood by the members of your community, you ought to use w in a manner that is consistent with its having meaning m”. The correction to which each one considers himself or herself subject is only correction under the assumption that he or she wishes to speak in a way that will be understandable to others3. In this way, a naturalistic account of meaning is combined with an acknowledgement that meanings constitute conditional norms.

  • 4 I did not grant this much in my brief argument for normativism on pp. 21-24 of my Words without Mea (...)

14However, the naturalist ultimately has to face up to fact that the posited linguistic community is a myth. Let us grant the naturalist, for the sake of argument, that there may be groups of people, of just the right size, exhibiting sufficiently rich but not overly diverse uses of a word w, such that on the basis of those actual and counterfactual uses of w in the community we might extrapolate a use and on the basis of the use determine a meaning. If we take a community too small, the number of tokenings of w in the history of that community may be too small for us to draw any definite conclusions. Even their collective disposition with respect to w may be too impoverished for there to be a particular meaning. If the community is very large and the dispositions with respect to w are very diverse, then there may not be enough coherence in the usage of the community for there to be a particular meaning. But there may be communities of just the right size, neither too small nor too large, whose uses of w determine a particular meaning for w. Given the right group of speakers and tokenings of the words of their language, we can sort the uses that are correct (true, relevant or useful) from those that are incorrect (false, irrelevant or useless) and can somehow determinately extrapolate a meaning from the ones that are correct. So in other words, let us grant the naturalist, for the sake of argument, that there are answers to the classic doubts about indeterminacy4.

15Still, the posited linguistic community is a myth because there is not just one such community. However we count linguistic communities, there are many. If each was a closed cell whose members did not interact with people in other linguistic communities, then we might be able to identify linguistic communities with these cells; but that is not the case. In each linguistic community, there is great deal of interaction with members of other linguistic communities. There is no reason to think that our method of assigning meanings to words will be invariant across different selections of community. If we interpret “cat” with respect to city-dwellers, we might interpret it as referring to just the instances of a certain kind of domestic pet (i.e., as having a meaning that determines such an extension). But if we choose a community that includes wilderness adventurers, then we might interpret “cat” as including lions and other big cats. Supposing that our method of interpretation does yield a definite interpretation sometimes, it may yield different definite interpretations given different choices of community. If the pirate ship scenario is a possibility, then a candidate community might even be a community of just one person (but not a community consisting of just one one-minute long time-slice of a person).

16Naturalism was threatened by the thought that meanings present themselves to speakers as norms. That is, each speaker thinks of his or her usage as subject to correction, and so it is seems reasonable to think of statements attributing meanings as expressions of such obligations — as endorsements of norms. The naturalist can answer this threat by representing the obligations as merely conditional: If you wish to be understood by the members of your community, you must use your words in ways that are consistent with their having the meaning that they have in your community. In particular, given that the meaning that w has in your community is m, if you wish to be understood in your community, you ought to use w in a way that is consistent with its having the meaning m. But now if, as it seems, there is no definite community that speakers can be thinking of when they think this, it is not very plausible that this is really the content of their thought when meanings present themselves to them as norms.

  • 5 On pp. 128-129 of Diego Marconi, Lexical Competence, MIT Press, 1987, and in “The normative ingredi (...)
  • 6 See Gareth Evans’s story of Turnip in his “The Causal Theory of Names”, Proceedings of the Aristote (...)

17In this light we can understand the temptation to posit certain privileged communities — experts and baptizers. The thought is that it’s chemists whose usage of “water” determines its extension, biologists whose use of “insect” determines its extension. It’s Gödel’s mom who determines that “Kurt Gödel” will denote Kurt Gödel, when she says, “I hereby dub thee ‘Kurt Gödel’”. If we could say that it was the usage of the experts that necessarily determined the meanings of common nouns and the acts of original baptizers that determined the referents of proper names, then we could say that it was always the usage of such experts and baptizers to which other speakers regarded their usage as conditionally responsible. But as Marconi rightly insists, it just isn’t so5. The experts do not regard their own usage as necessarily correct as a consequence of their being experts. They may disagree amongst themselves. Moreover, the nonexperts know this about the experts even if they are not in a position to challenge the experts themselves. An initial baptism has some claim to being a determinant of the reference of a proper name, but a lot of times there is no clear baptism, and the reference of a proper name can change over time through a misunderstanding that settles into a conventional use6. In any case, there are many words for which there does not even seem to be a candidate class of experts, such as “delicious” and “immoral”.

18Perhaps, though, the naturalist can stand up to the threat by introducing the concept of a minimal community. To a first approximation, we might say, for any group of interlocutors G and any word w, the minimal community for G and w is the smallest group of people H such that every member of G is a member of H and the uses of w in H are not too impoverished to determine a definite meaning for w. We can improve on this definition by giving priority not to the smallness of the community but to closeness of the contact between its members and the members of G. Thus:

For any group of interlocutors G and any word w, the minimal community for G and w = (definition) the group of people H such that (a) H includes G, and (b) the actual and counterfactual uses of w in H are not too impoverished and not too diverse to determine a meaning for w, and (c) for every other group H’ that includes G such that the uses of w in H are not too impoverished and not too diverse to determine a meaning for w, the amount of interaction that regularly takes place between the members of H and the members of G is greater than the amount of interaction that regularly takes place between the members of H’ and the members of G.

19There is still some vagueness in this definition, but let it pass. As far as it goes, this appears to be a reasonable definition of the community to which a person would want his or her use a given word to conform, inasmuch as conforming his or her use of the word to the use of that word in the community so defined would maximize his or her chances of being understood in conversation with the other members of G. Inasmuch as this definition of the minimal community is relative to a group of interlocutors, it allows that in different conversations an individual might adopt different standards, depending on who the other interlocutors are.

20With this definition in hand, we can say that the content of the conditional obligation felt by each interlocutor in a group of interlocutors is the obligation to conform his or her use of a word to the meaning that it has in the rest of the minimal community (for that group and that word) if he or she wishes to communicate with other members of the minimal community. The idea is not that each of us regularly contemplates the uses of words in the minimal community as I have defined it here. Probably I am the first person who has even thought of the minimal community in just this way. Rather, the idea is that each of us senses some obligation to conform his or her use of a word to the use of that word in some larger community without having any very clear idea about the defining features of that community. Nonetheless, insofar as the minimal community, as I have defined it here, is the community of people from whom we would be disposed to take instruction (either explicit instruction or instruction in the form of example-setting), it is fair to articulate and define the conditional obligations that we consider ourselves subject to in these terms.

3 The Circle of Deference

21There is still a problem for this proposal, and I think it is a decisive objection to the conception of semantic norms that I have constructed for the naturalist. The problem, putting it crudely, is that the members of the minimal community do not know what their words mean. They consider their use of a word to be responsible to that of yet other groups. The uses of w in a community may not be too impoverished or too diverse to determine a meaning for w, and yet the uses of w may fail to determine a meaning for w just because the members of the community do not themselves consider their own use to be authoritative and are prepared to defer to the usage of others outside their community. That is, they are prepared to alter their usage in order to conform to the usage of others. These others to whom they are prepared to defer may even include some of the members of the group we started with. So there may be no group whose use of a word is given as a settled fact that we can plug into our theory of the way in which use determines meaning.

22To get a clearer grip on the problem, let us approach it by way of another, merely apparent problem. As I have defined the minimal community H for G and w, H includes G. How can the use of w in H determine a meaning for w for the interlocutors in G if H includes G and the members of G are not sure what w means? This need not be a problem, however, if the other members of H do not possess the same doubts. So suppose there are members of H who use w with confidence, without ever having to defer to others in deciding whether a given use of w is correct. Then we can suppose that the meaning that our naturalistic theory determines for w on the basis of the uses of w in H is the meaning that our theory assigns to it on the basis of the uses of w in mouths of those members of H. To put it another way, our naturalistic theory may tell us that that the meaning a word has in a community is the meaning it has in that community by virtue of the uses of the word among those members of the community whose dispositions are settled.

23However, there is a further complication to consider in that there need not be any members of the minimal community for group G and word w whose dispositions with respect to w are settled. Let the minimal community for a group of interlocutors G and word w be H. Since H is the minimal community for w and G, the uses of w in H are not too impoverished and not too diverse to determine a meaning for w. But it does not follow that the other members of H will regard themselves as authoritative over w. They will not think that what the word means is just whatever meaning the uses of w in H determine it to be. They too, on the present hypothesis, consider themselves conditionally obliged to conform their use of w to the meaning it has in the minimal community for w and the group of interlocutors that they happen to be engaged with. In defining the use of w for purposes of plugging that use into our theory of the way in which use determines meaning, surely we do not want to include uses that the members of the community would themselves disavow in light of criticisms that they would accept. So while H may indeed be the minimal community for group G and word w, it may be clear that the uses of w in H cannot all by themselves determine a meaning for w in H.

24On its own, this observation still does not yield any decisive objection, since we can find minimal communities for groups of interlocutors in H. We can say that the meaning that w has for G is the meaning it has in the minimal communities for other groups of interlocutors in the minimal community H for G and w. In other words, for each group of interlocutors J1, J2, J3, ..., in H, other than G, there will be a corresponding minimal community K1, K2, K3, ... , for w. If the uses of w in each of K1, K2, K3, ..., determine a meaning for w in each J1, J2, J3, ..., respectively, then the meaning of w in H might be determinable as a composite of the meanings of w in J1, J2, J3, ..., and the meaning of w for G could be identified with that composite meaning that w has in H. Accordingly, the conditional obligation that each member of a group of interlocutors G considers himself or herself as under will be formulated thus: If you wish to communicate with other members of your minimal community using a word w, make sure that your use of w conforms to the meaning that w is determined to have by the uses of w in the minimal communities for groups of interlocutors in your minimal community.

25Of course, the minimal communities K1, K2, K3, ..., for the groups of interlocutors J1, J2, J3, ... and w in the minimal community H for G and w may have no settled dispositions with respect to w either. Even that is not a decisive objection, because it can be answered in the same way (by appeal to the minimal communities for the groups of interlocutors in each of K1, K2, K3, ...). The really decisive objection is, rather, that the chains of deference are not grounded in any independent foundation. In a world containing infinitely many people, the problem could be that each community depends on another community ad infinitum. But since there are only 6.6 billion people in the actual world, the problem is instead that the chains of dependency are bound to circle back. (Here I am assuming that present people do not defer to future people, but maybe they do). The people whose use of w the people in the J’s and the K’s defer to may include members of G. And for any K , the people whose use of w the people in K. defer to may include members of any of the J’s, who in turn may defer to members of G. And of course these considerations iterate; so we may have even longer chains of deference that start with members of G and circle back to the members of G. Call this situation the circle of deference.

26The problem for naturalism is precisely this circle of deference. For a given word w and a given group G, there may indeed be a group of people H who satisfy the definition of a minimal community for G and w. And for each group of interlocutors J in H other than G there may be a minimal community K for J and w. And so on. But we do not find a meaning for w in G at any of these levels, because at no such level are the dispositions with respect to w independent of the dispositions of the members of G. There may simply be no subset of the users of w, however remotely connected with G, whose dispositions with respect to w are so independent of the use of w by the members of G that we can treat their use of w as determining the meaning of w for G.

27There is a way out of the circle of deference, but it is a way that leaves naturalism behind. Start with an analogy. Suppose you and I need to meet, and we can meet in your office or we can meet in mine. Both choices are acceptable, which is not to say that they offer objectively equal advantages. (Maybe the chairs are more comfortable in your office). If there were a practice on your side always to meet in your office, that would settle it. But on the contrary, your practice is always to make a choice that depends on the choices of others. Similarly for my practice. This interdependency may remain even if there is a past practice of meeting in your office that inclines us to do so again. If there is a settled practice, then in light of it we may act without hesitation. But if there is any doubt, then nothing remains but for one or the other of us to make a proposal: “Let’s meet in my office!” or “Let’s meet in your office, as usual!”

28The situation with respect to choice of meanings is similar to the situation with respect to choice of meeting places. Each interlocutor wishes his or her use of a word w to conform to the use that w has among other members community (as always, conditionally on wishing to communicate by means of w). Precisely because each wishes to conform to the use of w by the other, there is no independent source that they can go to find out what that use is (e.g., the use in the minimal community). Often they will act without hesitation. The question whether a given use is correct often does not arise. But in the face of any doubt, nothing is left to them but for someone to make a proposal. That, I suggest, is how we should view a claim about the meaning of a word. It need not be an arbitrary proposal. The proposal may be one that the proposer inclines to due to past uses of w. The proposal may be a proposal to model future uses on certain past uses. Nonetheless, a claim about the meaning of a word retains the character of a proposal, because the past uses of the word in the community are not determinative, because they do not reflect a settled use.

4 Outsider Normativity

29I have denied that statements that ascribe a meaning to a word or phrase describe a natural relation or property, and I have asserted that such meaning statements are endorsements of norms. Offhand there would seem to be a third alternative that ought to be considered, namely, that meanings statements are descriptions of norms. But I will argue that in fact there is no way for us to describe linguistic norms without endorsing some others; so this apparent alternative reduces to the position I have affirmed.

30Again, we might say to someone, “You have a right to vote”, meaning thereby only that in this country the addressee is granted a right to vote. Or an anthropologist might describe the norms of the Yoweezowee tribe saying, “Before a young man will be permitted to take a wife, he must camp out in the forest by himself until he can kill and bring back a wild boar”. In saying this, the anthropologist is not articulating a normative principle that she herself endorses. She is merely attributing to the Yoweezowee endorsement of this norm. Similarly, it might be said, if a semanticist says, “In Italian, ‘gatto’ means cat, then the semanticist is merely describing a norm that prevails in the community of Italian speakers. That one in particular might not seem very useful to the Italians themselves, since in Italian it comes out “‘Gatto’ significa gatto, but there could be others such as “The extension of ‘gatto’ is confined to animals”, or “‘Gatto’ applies to x if and only if x has such and such genetic features”. So semantics is normative, we might say, but only in the sense that it describes the norms that prevail in a community without endorsing them. In short, the normativity of semantics is what we might call outsider normativity.

31In order to conclude that the Yoweezowee govern themselves by norms described in the marriage rule, we need to know more than merely that young men in the Yoweezowee tribe rarely or never marry before camping out in the woods alone and bringing back a boar. That would be a basis only for an empirical generalization. In order to know that they are governed by some norm, we have to find some more or less explicit acknowledgement of that norm in what they say. Perhaps they actually assert the rule themselves, in their own language. In denying a young man the right to marry, they explain that he has not brought back a boar. So similarly, if we want to conclude that the Italians are in fact governed by some semantic rule, we will have to find some evidence, in what they say, that they are aware of the rule as such. In each case, then, we can recognize that a rule is acknowledged and operative in the community only insofar as we can interpret some things the members of the community say. In the case of the Yoweezowee, we have to be able to understand the words in terms of which they formulate the marriage rule. In the case of the Italians, we have to be able to understand the words in terms of which they formulate their semantic rule for “gatto”. We can recognize that a rule is operative in community only insofar as we can interpret the linguistic expression of the rule in the community’s language.

32What this shows is that our understanding of the semantics of the language of a community cannot consist exclusively in the semantic rules that we recognize as operative in the community. Rather, we have to make in addition semantic claims about the words in terms of which they formulate their semantic rules. The semantic claims by means of which we understand their semantic rules cannot be exclusively those that we thereby come to recognize as operative in the community. The semantic rules may be formulated in terms of vocabulary items the meanings of which are not in turn adequately explicated by semantic rules acknowledged and operative in the community. Or the semantic rules for a term may be formulated by means of other terms the meanings of which are adequately explicated by further semantic rules acknowledged and operative in the community. But in that case, either the series of semantic rules bottoms out in terms the meanings of which are not adequately explicated by semantic rules operative in the community, or the semantic rules lead in circles. In the case of a rule like “‘Gatto’ significa gatto" the circle is maximally small. But, as I noted, there may be other sorts of semantic rules, and then the circularity may become apparent only when we consider the totality of semantic rules acknowledged and operative in the community. In the case of circularity too our understanding cannot consist exclusively in semantic rules that are themselves acknowledged and operative in the community.

33In short, the semantic rules that are acknowledged and operative within a community will inevitably fall short of an adequate, non-question-begging specification of the meaning of each term of the language. This shortfall shows up in the fact that in many cases there may be no way to decide whether a word applies in a given case solely on the basis of semantic rules acknowledged in the community. When some new kind of furniture is invented by some Italian furniture designer, will it be correct to call it “sedia”? Or will “poltrona” be preferred? That remains to be seen, and there may be nothing that decides the matter apart from the practices that settle in among speakers of Italian when the case arises and the proposals they make to one another along the way.

34Similarly, in the case of “gatto” there is room for some surprises. Some weasel-like animal running wild in the mountains of Borneo might turn out to be, genetically speaking, a cat. Or we might find that our understanding of genetics has never been quite right, and when we correct our errors, we might find that the Borneo weasel is really a cat. It need not be the case that the Italians’ response to the Borneo weasel is determined by an acknowledged semantic rule pertaining to “gatto”. Whether they respond by applying the word “gatto” to the Borneo weasel or instead invent a new term for things having a certain genetic make-up while reserving the word “gatto” for things that look like typical cats will depend on the proposals that come to be accepted among them, and that in turn may depend on arbitrary factors, such as whether the discovery is first reported in a popular magazine or in a scientific journal.

35The lesson we should draw is that there is always more to the semantics for a language than the semantic rules that may happen to be acknowledged and operative in that community. In other words, the normativity of semantics cannot consist exclusively in what I have above called outsider normativity. A further consequence we may draw is that the community’s possession and mastery of a language does not rest on its members’ knowing the semantic rules governing their words. This does not have the absurd consequence that a language cannot be taught. It can be taught, but it is taught by example. People have the capacity to learn to speak a language that others can understand by following positive examples and avoiding negative ones. I don’t put this by saying that they “generalize from examples”; for that would suggest that they know some generalization, which I deny. (How it works, psychologically speaking, is something that no one yet understands).

5 The Nature of Meaning Proposals

  • 7 There is a similarity, but likewise a considerable difference, between this proposal and Saul Kripk (...)

36So when we say something like, “In Italian, ‘gatto’ means cat, what we are doing is not describing a natural relation between uses of a word and objects in the world. Nor are we merely reporting on the norms that are acknowledged and operative in the community. So what are we doing? My answer is, roughly, that in making a statement ascribing a meaning to a word or phrase what we are doing is making a proposal about how to use it. Alternatively, we may say that we are endorsing a norm, inasmuch as the proposal becomes normative insofar as it is accepted7.

37For example, what we are doing when we assert, “In Italian, ‘gatto’ means cat”, is proposing that when we speak Italian we shall use “gatto” as we would use “cat” when speaking English. As a matter of fact, many people will find that proposal helpful when it comes to interacting with Italians. Of course, it will have that effect only when it is said to someone who knows how to take such proposals. For instance, one would misunderstand the proposal if one took it as proposing that we simply substitute just the word “gatto” for the word “cat” and otherwise speak English words. Moreover, in most cases, no such simple proposal will do the job. For example, we might say that “andare” means to go, but of course the Italian use of “andare” is very different from the English use of “go”. There is really no way to formulate a proposal concerning “andare” other than to provide a lot of examples of the use of “andare” in sentences, provide English sentences in place of which one might propose to use these Italian sentences, and hope that the student catches on. A good Italian-English dictionary will provide just such examples.

38But actually what I want to say is something a little more subtle than this. Syntactically speaking, it is not correct to characterize a sentence ascribing a meaning as a proposal. A statement of the form “w means m” is not a statement of the form “Let’s do x” or a statement of the form “I hereby endorse the doing of x”. Moreover, the difference in syntax points to a difference in semantics, which shows up in the possibility of forming meaningful compounds. A sentence of the form “w means m” can be a meaningful component of a more complex sentence in a way that sentences of these other forms cannot be. We can meaningfully write, “If ‘gatto’ means cat, then that animal may be called ‘gatto’”, whereas it does not quite make sense to say, “If I hereby endorse the calling of cats ‘gatto’, then that animal may be called ‘gatto’”, and it is definitely ungrammatical to say, “If let’s call cats ‘gatto’, then that animal may be called ‘gatto’”.

  • 8 For a thorough development of the assertibility semantics for conditionals, see my Conditionals in (...)

39So it is not quite right to say that a statement ascribing meaning makes a proposal. Such statements about meaning statements only gesture in the direction of an account, whereas what is needed is a precise semantics. The semantics I would offer is a semantics formulated in terms of what I call assertibility in a context. Contexts are structures built up out of sentences and other linguistic objects such as demonstratives. There is a base level of contexts, which I call primitive contexts, and then for purposes of explicating the semantics of a language that permits conditionals, contexts are defined as being either primitive contexts, or sets of primitive contexts, or sets containing primitive contexts and sets of primitive contexts, and so on. For each kind of sentence (atomic, negation, disjunction, conditional, etc.), we define the conditions under which sentences of that kind are assertible relative to a context. For every class of logical terms (broadly speaking) that we add to the language (such as the class of quantifiers), there will be a corresponding development in the definition of a context. An understanding of the conditions under which an utterance of a sentence is assertible is reached by explicating the conditions under which a context in which the sentence uttered is assertible is the context that pertains to a given conversation8.

40On this approach, we avoid altogether the idea that declarative sentences express propositions modelled as sets of possible worlds, and so there is no need to construe declarative sentences as drawing a distinction between the worlds in which they are true and the worlds in which they are not. So we are free to characterize a meaning sentence as a declarative sentence and so to allow that it forms compounds in conditionals, can be quantified into, and so on, without thereby committing ourselves to treating it as a classification of possible worlds. The challenge that we face is, rather, to explain the conditions under which a primitive context containing an atomic meaning sentence of the form “w means m” is the context that pertains to a conversation. And then beyond that, we have to explain the conditions under which a more complex structure, such as a set of primitive contexts some of which contain atomic meaning sentences, is the context that pertains to a conversation.

  • 9 I have previously stated this proposal in “A New Skeptical Solution”, op. cit.

41So, what does it take for a primitive context containing a sentence of the form “w means m” to be the context that pertains to a conversation? The primary function of meaning statements, I would like to suggest, is conversational repair9. For example, if someone uses a pronoun “she,” the hearer may not be able to make the anaphoric connection between that pronoun and any other more specific noun phrase. In that case, the hearer may initiate a repair by asking “Whom do you mean?” and the speaker may complete the repair by saying, “I meant the lady who cleans the lobster tank every week”.

42In that example, meaning was ascribed to the speaker, the person who uttered the word “she”, but likewise statements about the meaning of words and phrases are typically issued in the course of conversational repair. A speaker’s choice of words may leave some uncertainty over how to they are to be taken, and a statement about meaning elicits a reformulation. X, speaking to Y, describes Z as “officious”. What X has in mind is that Z is overly concerned with the rules and proprieties of his position as department head. Thus X commits a common error in English diction. W overhears this conversation, but doubts whether X knows what “officious” means. So W says to X, “The word officious’ means overly eager to offer his or her services, not overly concerned with the rides and proprieties of his or her office” This in turn will elicit a clarification from X. In this way, W uses a claim about meaning to help X communicate to Y.

43More generally, statements of the form “w means m" function as proposals that resolve a certain sort of coordination problem in which the way words are used somehow becomes an issue. So what it takes for a primitive context containing such a sentence to pertain to a conversation is that the conversation is one in which such a resolution of a coordination problem is called for. A fundamental explication of this claim would start with a fundamental account of the nature of linguistic communication in terms of which we could specify the nature of the sorts of coordination problems that interlocutors face in choosing words. Since I am not providing any such fundamental theory of linguistic communication in this paper, I am resting content with the rather vaguer claim that meaning statements serve the purpose of conversational repair.

44Likewise, the use of meaning statements in the course of teaching someone a second language can be viewed as conversational repair, broadly speaking. The situation needing repair may be that a person is traveling to Italy and does not speak Italian and needs to learn how to speak Italian. The statement “In Italian, ‘gatto’ means cat, not cake” may belong to a larger course of instruction that fulfills that need. In this case there hardly seems to be any room for proposals, and to characterize the speaker as endorsing something wrongly suggests that the speaker is in a position to impose his or her will. That is not to say that the theory does not apply in this case, but only that in applying it one has to take care not to give the wrong impression.

45A proposal that resolves a coordination problem in the manner of a meaning statement need not take the form of a meaning statement. For example, in the case of a logical word such as “if”, it may take the form a statement of the conditions under which a sentence formed with that word is assemble in a context. A defense and exposition of such a proposal may even take the form of a long book. More generally, theoretical semantics can be cast in this mold. The theories put forward in theoretical semantics may be understood as proposals of an uncommonly elaborate sort, which the theorist will formulate in some kind of technical vocabulary that he or she uses in making a variety of such proposals.

46What is special about atomic meaning statements of the form “w means m” is that they resolve coordination problems by means of providing a model, namely, the expression m. A statement of this form is effective in resolving a coordination problem only insofar as the parties to whom the proposal is made in this way already engage in a practice of using the expression m. So if saying “In Italian ‘gatto’ means cat” is to be effective in resolving a coordination problem, each of those to whom the statement is made must already engage in a practice of using the word “cat”.

47The case of homophonic specifications of meaning deserves special mention. The question is whether a sentence like

48(*) The English word “cat” means cat.

  • 10 Marconi concurs. See section 2 of “The normative ingredient in semantic theory”, op. cit.

49is significant. It is possible to imagine a situation in which such a statement achieves some purpose, but such sentences cannot be truly informative10. In my terms, the proposal resolves no coordination problem, since there must already be coordination in the use of the word “cat” if the proposal is to be understood. To this it may be said that since the Italian sentence, “La parola inglese ‘cat’ significa gatto”, is significant and (*) is its translation, the latter must be significant as well. The answer to this is that while, yes, (*) is significant in the sense that a systematic semantic theory will assign a meaning to it, it is not significant in that, taken literally, it cannot serve the proper function of meaning statements, namely, to make a proposal that resolves a coordination problem.

50I have been careful not to formulate my thesis by saying that meaning statements endorse a practice. My thesis has been that it is a mistake to think that a meaning can be identified with a practice. If there were a definite practice that we could identify as the meaning of a word or phrase, then my objection to naturalism would fail. A statement of the form, “w means m” we could then say, simply characterizes the nature of the practice governing w as being the same as that governing m. Rather, the normativist philosophy of semantics is what we are left with once we recognize that due to the circle of deference, there cannot be, for every word and phrase, a definitive characterization of a practice governing its use.

6 Meaning and Truth

51The claims I have made about meaning statements might seem to yield an unacceptable consequence, namely, that there is no objective truth. If two sentences have the same meaning, then they have the same truth value (both true, both false, or both neither). More precisely if two sentences have the same meaning, then relative to any given context they have the same truth value. In this sense, meaning (or meaning plus context) determines truth value. But now I am saying that statements about meaning are better viewed as proposals than as descriptions. Doesn’t it follow that an ascription of truth value is likewise only a proposal and not a statement of fact?

52No, it does not. Here I will only summarize the key points that need to be made:

  1. The assertibility of a sentence in a conversation is an objective matter — dependent on what is really going on in the world — inasmuch as the content of the context pertinent to a conversation is in part a matter of that. So a speaker can be wrong, by asserting what is really unassemble in the context pertinent to his or her conversation, though the speaker and everyone else in the world thinks the speaker is right.

  2. In the framework of the semantics of assertibility in a context that I briefly described in the previous section, we might say that the meaning of a sentence is the set of contexts in which it is assemble. In that case, an analogue of the claim that meaning determines truth value will be the trivial truth that if two sentences are assemble in all of the same contexts, then for any given conversation they are both assemble in the context that pertains to it or not. Moreover, if we accept this approach to formulating semantic claims, then we will accept as well the following constraint on the making of the sort of proposals that meaning statements are supposed to be: For any two meaning sentences m1 and m2 that ascribe the same meaning to two different sentences s1 and s2 for any context Γ, if m1 and m2 are both assemble relative to Γ, then for each context Δ countenanced in Γ, s2 is assertible in Δ if and only if s2 is assertible in Δ.

    • 11 For a precise development of an assertibility semantics for the truth predicate, see my “Semantic (...)

    Truth cannot be identified with assertibility, because assertibility is a relation between a sentence and a context, and truth is not a relation to a context (though whether a sentence has the property of being true may be relative to context). However, an account of truth may take the form of an account of the assertibility conditions of sentences of the form “It is true that p.” And the objectivity that pertains to assertibility pertains as well to truth inasmuch as a sentence of the form “It is true that p” is assertible in a context if and only if the sentence p is assertible in that context11.

Torna su

Note

1 This would seem to be the position of Donald Davidson in his papers, “Communication and Convention”, in his collection Inquiries into Truth and Interpretation, Oxford University Press, 1984: 265-280 and “A Nice Derangement of Epitaphs”, originally published in 1986 and reprinted in his collection, Truth, Language and History, Oxford University Press, 2005: 89-109.

2 The pertinent conception of coordination problem is that which David Lewis so vividly explicated in his book, Convention: A Philosophical Study, Harvard University Press, 1969.

3 This is the only kind of semantic normativity that Åsa Maria Wikrforss is prepared to concede in her paper “Semantic Normativity,” Philosophical Studies 102, 2001: 203-226 (see 211).

4 I did not grant this much in my brief argument for normativism on pp. 21-24 of my Words without Meaning, MIT Press, 2003, and in fact I don’t believe it. I suspect that a good argument against naturalism could be made by arguing that any naturalistic account of meaning would leave meaning wildly indeterminate. This is simply not the strategy I am pursuing in this paper.

5 On pp. 128-129 of Diego Marconi, Lexical Competence, MIT Press, 1987, and in “The normative ingredient in semantic theory”, in W. Hinzen, H. Rott (eds.), Belief and Meaning: Essays at the Interface, Ontos Verlag, 2002: 215-228.

6 See Gareth Evans’s story of Turnip in his “The Causal Theory of Names”, Proceedings of the Aristotelian Society, Suppl. vol. 47, 1973: 187-208.

7 There is a similarity, but likewise a considerable difference, between this proposal and Saul Kripke’s skeptical solution to the problem of rule-following in his Wittgenstein on Rules and Private Language: An Elementary Exposition, Harvard University Press, 1982. For a comparison, see my paper, “A New Skeptical Solution”, Acta Analytica 14, 1995: 113-129.

8 For a thorough development of the assertibility semantics for conditionals, see my Conditionals in Context, MIT Press, 2005. That book does not, however, address specifically the case of conditionals having sentences ascribing meanings as components.

9 I have previously stated this proposal in “A New Skeptical Solution”, op. cit.

10 Marconi concurs. See section 2 of “The normative ingredient in semantic theory”, op. cit.

11 For a precise development of an assertibility semantics for the truth predicate, see my “Semantics for Deflationists”, in J.C. Beall and B.radley Armour-Garb (eds.), Deflationism and Paradox, Oxford University Press, 2005: 148-176.

Torna su

Per citare questo articolo

Notizia bibliografica

Christopher Gauker, «The circle of deference proves the normativity of semantics»Rivista di estetica, 34 | 2007, 181-198.

Notizia bibliografica digitale

Christopher Gauker, «The circle of deference proves the normativity of semantics»Rivista di estetica [Online], 34 | 2007, online dal 30 novembre 2015, consultato il 14 juin 2024. URL: http://0-journals-openedition-org.catalogue.libraries.london.ac.uk/estetica/3943; DOI: https://0-doi-org.catalogue.libraries.london.ac.uk/10.4000/estetica.3943

Torna su

Diritti d’autore

CC-BY-NC-ND-4.0

Solamente il testo è utilizzabile con licenza CC BY-NC-ND 4.0. Salvo diversa indicazione, per tutti agli altri elementi (illustrazioni, allegati importati) la copia non è autorizzata ("Tutti i diritti riservati").

Torna su
Cerca su OpenEdition Search

Sarai reindirizzato su OpenEdition Search