Navigation – Plan du site

AccueilNumeri47variaReference fixing and the stiffnes...

varia

Reference fixing and the stiffness of reference or three (would be) puzzles concerning names

Ernesto Napoli
p. 179-195

Résumé

This paper deals with three puzzles concerning reference. The first concerns the Kripkean distinction between reference fixing by description and meaning giving2. The second concerns the compatibility of two properties of names, arbitrariness and rigidity across world. The third concerns the tenability of the notion that a name means its bearer considering that a name is meaningful even when the bearer is no longer. The solution of the first consists in seeing that the distinction involved is not essentially a modal distinction. The solution of the second consists in setting things right with our modal talk. The solution of the third consists in realizing the persistence across times of the enjoyment of certain properties.

Haut de page

Texte intégral

I am grateful to Paolo Casalegno and Achille Varzi for their comments on predecessors of the paper, to Daniela Capra for a point concerning the double services of “here” and to Andrea Bianchi for the insightful dedication with which he has attended to the improvement of the paper.

1. Reference fixing versus meaning giving

1. The meaning of a name is its referent
2. There is a difference between fixing the referent and giving the meaning

 

1The combination of 1. and 2. as formulated does not constitute a puzzle. Still it becomes a puzzle, as soon as one asks himself: fixing the referent and giving the meaning of what, and is precipitous enough to answer: of a name. For then he would have that:

1’. The meaning of a name is its referent
2’. There is a difference between fixing the referent of a name and giving its meaning

2Kripke knows better than asserting 2’. He either talks of fixing the referent versus giving the meaning, or of fixing the referent of a phrase («one meter») and giving the meaning of a phrase.

3Even so the distinction is unhappily phrased. As I stress further on, to introduce an abbreviation is not to give meaning but to make allowance for the use of a shorter expression in place of a longer one. Where the longer expression has a meaning the shorter will inherit it. However the point of abbreviation is economy rather than meaningfulness. It is misplaced to see the introduction of an abbreviation as a way of giving meaning to a (short) expression rather than a way of simplifying a (long) expression. The point is clear enough. But it has been obscured by the unfortunately influential notion of semantic marks, and by Frege’s notion that names have sense. Both invite the conclusion that names are abbreviations whose meaning is given by longer expressions.

4I think it is best not to follow Kripke in talking of the «distinction between “definitions” which fix the reference and those which give a synonym».

5For the introduction of a name by description does not exemplify a definition at all. Neither a nominal definition (which is the only definition there is: only words can be defined) nor, in general, what is improperly called «real definition», i.e. a characterization of the nature of the referent of the name.

6In fact many names are introduced or could be introduced with the help of descriptions which make leverage on a contingent and temporary relation between the introducer of the name and the candidate namee and so have nothing to do with the nature of the prospective namee. For example I could say: «we do not know her name. For the sake of brevity let’s call the woman sitting right to the left of J.B. Anne».

7I think it should be best to talk of the difference between fixing the referent of a nominal form by description (rather than ostension), and providing a (shorter) synonym of a description. And this is the difference between introducing a name via a description and introducing an abbreviation of a description.

8To introduce a name is to fix the referent of, or (indifferently) to give (assign) the meaning to, a word. To introduce an abbreviation is nothing of the sort. No reference fixing, no meaning giving. Just the stipulation that one expression can be used instead of a more cumbersome one.

9Kripke tries to substantiate the difference by modal considerations. What can be evinced from what he says in Naming and Necessity is the following. When the referent of «N» is fixed by the description «the F» the statement «N is the F» is a priori and contingently true. When «N» is stipulated to be a shorter synonym of «the F», the statement «N is the F» is a priori and necessarily true.

10I will now proceed to show that being «N is the F» a priori and necessarily true is neither sufficient nor necessary for «N» to be a shorter synonym (abbreviation) of «the F» and that being «N is the F» a priori and contingently true is neither sufficient nor necessary for «N» to be a name introduced via the description «the F».

11Before doing this let me make a general point. Different modal properties are not the ground for interesting distinctions; at most they are consequences of interesting (a-modal) distinctions. The distinction between fixing the referent of a nominal form by description and providing a synonym of a description is an interesting one. But it can be clearly drawn quite independently of modal considerations. It is the distinction between introducing a name by description and introducing a notational abbreviation of a description. A name introduced by description is not a notational abbreviation of the description used for its introduction. Conversely a notational abbreviation is not a name, since it is neither the name of the expression it abbreviates, nor of what (if anything) is designated by the expression it abbreviates. A notational abbreviation of an expression means whatever the expression means, so it is synonymous with it. On the other hand a name introduced by description is not synonymous with it. The reason is simple. A description does not mean what it denotes and yet where the description is used for fixing the referent of the name, the name means the denotatum of the description.

  • 3 Ibidem: 25, fn. 3.

12Maybe the name does not even mean the denotatum of the description, but it rather means what the description, at the moment of the introduction of the name, manages to single out as the prospective namee. If the speaker and the audience share, to borrow Kripke’s words3, appropriately false beliefs, a description may serve to call attention to an individual which is not the denotatum (but it is held to be the denotatum).

13The point is simply that names are not abbreviations and abbreviations are not names. Both names and abbreviations are introduced by stipulations. But the type of stipulation is different. To introduce a name is to stipulate that a certain expression is to stand for, to be used to talk about, a certain entity. To introduce an abbreviation is to stipulate that a certain expression is substitutable for another, can be used in place of another expression, to talk about whatever the other expression is used to talk about.

1.1. That «N is the F» is a priori and contingently true is neither sufficient
Nor necessary for «N» to be a name introduced by «the F» (rather than an abbreviation of «the F»)

Not sufficient

14One could think that if «N is the F» is contingently a priori true «N» cannot be an abbreviation of «the F». The reasoning would be the following. If «N» is an abbreviation of «the F», «N is the F» is no other than «the F is the F». If «the F is the F» is true (as it is if proper, i.e. denoting exactly one thing) then it is necessarily true. Hence if «N is the F» is true and «N is the F» is «the F is the F» then «N is the F» is necessarily true.

15The reasoning has a flaw. It is not the case that if «the F is the F» is true then it is necessarily true. «The F is the F» is true iff «the F» is proper. Yet «the F» may be proper at the actual world and fail to be proper at a world different from the actual. This is to say that «the F is the F» may be true at the actual world without being true at all worlds, i.e. may be contingently true. If so the contingent (a priori) truth of «N is the F» is not sufficient for N being a name (introduced by «the F») rather than an abbreviation of «the F».

Not necessary

16One could think that where the name «N» is introduced by the description «The F» the sentence «N is the F» is a priori true. This need not be the case.

  • 4 Ibidem: 80, fn. 34.

17A name can be introduced and its reference fixed even by a description false of the object. Venus was fixed as the referent of «phosphorous» by way of «the morning star» even though Venus is not a star. Says Kripke: «In such cases, the description which fixes the referent clearly is in no sense known a priori to hold of the object, though a more cautious substitute may be. If such a more cautious substitute is available, it is really the substitute which fixes the reference…» 4. There is here an obvious question: what fixes the referent if a more cautious substitute is not available? I think that the referent is fixed by the description actually used and not by any better substitute. Venus is there to be seen and the description is no more than a way of calling attention to it. The capacity of a description to direct the attention to a specific object does not depend on its being true (even less on being uniquely true) of the object. Kripke talks of a description, which fixes the referent only to add that it does not really fixes the referent. I do not think that sense can be made of the difference between fixing and really fixing the referent. In fact if the description which fixes the referent does not really fix it, the very notion of a better substitute is void. Only if the description used is circumstantially sufficient to isolate the prospective namee it is possible to look for a substitute. But if the description used is sufficient to isolate the prospective namee in spite of its being false of it, a refined substitute true of it is quite superfluous. It arrives too late, when the job has already been done.

18Considering that a conjunction of conditions constitutes a necessary condition iff each conjunct constitutes a necessary condition, what I have said, if correct, is enough to prove the point. The point was that the contingent a priori truth of «N is the F» is not a necessary condition for «N» to be a name (rather than a notational abbreviation) and I have argued that a priori truth is not necessary.

  • 5 Ibidem: 60.

19For the sake of the unconvinced I can address contingency. The contingency of a statement of the form «N is the F» does not constitute a necessary condition for «N» to be a name (introduced by «the F»), if we are inclined, alike Kripke, to considering «p» as the name of an irrational number rather than the notational abbreviation of «the ratio of the circumference of a circle to its diameter» in spite of the fact that «p is the ratio…» is not just a priori true but also necessarily true5.

20Admittedly he has for this no more than a vague intuitive feeling. And this is hard to substantiate if we accept that a distinguishing trait of a notational abbreviation is to give rise to an apriori and necessarily true statement. In general, if this were the criterion, all terms belonging to an apriori science like mathematics and introduced by description would exemplify notational abbreviation rather than reference fixing. (Obviously «introduced by description» is to mean introduced by a description formulated in the language of the theory. This excludes the introduction of «2» via «my preferred number» or «the number of Tom’s brothers» or any other description using non mathematical expressions.)

21One could wonder how a statement of the form «N is the F», or for that matter any statement, could ever be a priori and contingently true. The wonder is not at all extravagant if, as I have argued, being «N» a name introduced by «the F» is not sufficient for the contingent a priori truth of «N is the F».

22Now, the example Kripke has in mind is «(one) meter is the length of bar S at t». This sentence is contingently true since the length of bar S at t could have different in different environmental conditions (temperature, pressure, humidity). But why is it a priori true? One should beware of thinking that the statement is a priori true since names are arbitrarily assigned. One should beware of grounding apriority on conventionality.

23«One meter is the length of bar S at t» is true iff «the length of bar S at t» is proper and its truth is a priori knowable iff it is a priori knowable that «the length of bar S at t» is proper. Is that so? Well, it is so provided it is the case that it is a priori true that any solid, or if you prefer any body at a time, has a unique length.

24One could think that if we revert to indexicals we have examples of contingently a priori true statements, «I am here now», and even examples of contingently true statement of the form «N is the F». One such would be «I am the utterer of the just uttered token of “I”», where the existence and uniqueness of the utterer of the uttered token of «I» is guaranteed by the very utterance.

25Yet «I am the utterer of the just uttered token of “I”» is not really of the form «N is the F». An individual name in use (i.e. as part of a sentence) is a determiner phrase (Dp). An indexical in use is a proDp. Still a Dp and a proDp are formally distinct. In fact if it can be known a priori that any occurrence of «I am the utterer of the just uttered token of “I”» is true, it cannot be known a priori that any occurrence of «Paul is the utterer of the just uttered token of “I”» is true.

  • 6 See Donnellan 1979: 51.

26Further, it is doubtful that «I am here now» provides an example of an apriori true statement. The sentence «I am here now» does not mean anything. Rather an utterance of «I am here now» said by x at place p and time t means that x is at place p at time t. Yet it is not known or knowable a priori that x is at place p at time t. what is known a priori is that any utterer is at the place of the utterance at the time of the utterance, i.e. that any utterance of «I am here now» cannot but be true, no matter what it means. But this is quite different from knowing of an utterance that what it means is true; as already stressed, though not in connection with indexical sentences, by Donnellan among others6.

  • 7 Kaplan 1989: 509.

27Kaplan has said that «I am here now» is in some sense universally true and that one need only to understand the meaning of it to know that it cannot be uttered falsely7.

28The claims, as phrased, are a bit sloppy. The sentence is said to be universally true only in the sense that all of its utterances, or, as Kaplan prefers to say, occurrences, are true. Yet the truth of all occurrences of a sentence does not make the sentence itself true. In fact being the sentence «I am here now» meaningless it cannot be that reckoning that it cannot be uttered falsely reposes on knowledge of its meaning. Rather it reposes on knowledge of the constraints on the interpretation (meaning assignment).

29It must be further noticed that all occurrences of «I am here now» are true provided «here» is a pure indexical rather than a demonstrative. In many places, such as museums and cross country ski tracks there is a map with a circle somewhere and an arrow indicating the circle and a comment «you are here». Admittedly the information is as a rule quite right. Still it might be wrong. During a journey one can put his finger on a spot on the map and say «we are here» even though, unfortunately, that is not the case.

1.2 That «N is the F» is a priori and necessarily true is neither sufficient
nor necessary for «N» to be a notational abbreviation of «the F»

30Not sufficient given that where «N» is the name of an abstract entity accessed a priori (and not via ostension of an exemplar or sample or instance of it, as it is the case with natural kinds) and «F» designates a necessary property necessarily uniquely satisfied (smallest prime number), «N is the F» is necessarily and a priori true (even if «N» is not an abbreviation of «the F»).

31Not necessary given that where «the F» is an empty description the statement «N is the F» is, no matter if «N» is a notational abbreviation, i.e. no matter if «N is the F» is elliptical for «the F is the F», not true and hence not a priori and necessarily true. Given the emptiness of the description we have either that the statement is neither true nor false (informal Frege) or false (Russell). We could have truth only by forcing a denotation (formal Frege: the empty class or Kaplan: exotic entity). I have qualms with forced denotation. It is technically convenient but conceptually unsatisfactory: a forced denotation is not the denotation the description has but the denotation the description is made to have, i.e. a conventional assignment. But the denotation of a description is not a matter of convention. So forced denotation is not denotation at all. (Hard to say how a would be description of a concrete individual can turn out to denote an entity which is not an individual, e.g. a set).

32From the fact that «N» is a definitional abbreviation of «the F» it does not at all follow that «N is the F» is (a priori and) necessarily true. Definitional abbreviations are substitution rules and rules are neither true nor false. Rather they are established or not, followed or not. That N =df the F means no more than any occurrence of «the F» can be replaced by an occurrence of «N». From this rule it does not follow that «N is the F» is a priori and necessarily true and not even that «N is the F» is a priori and necessarily true, if true at all.

33A description does not have a denotation or the denotation it has by convention. Where «the F» is the would be description of a concrete individual only a posteriori can be verified whether the condition «F» is (uniquely) satisfied. Hence «the F is the F» cannot be an a priori truth. And even if true cannot be necessarily true. Being the description of a concrete contingent individual there will be worlds where it is not satisfied. Only if «the F» is the would be description of an abstract entity (i.e. an entity which exists at all worlds) and «F» means a necessary (and a priori assessible) property, «the F is the F» might be a priori and necessarily true. what is crucial is that even in this case the a-priori and necessary truth of «N is the F» is not a consequence of the stipulation that N =df the F but rather of quite independent factors, namely the unique denotation of the description, the nature of the entity denoted and the nature of the property indicated by F.

34What I have been saying leads naturally to an objection of this tenor. To introduce an abbreviation of a description is to introduce a synonym of the description, and it would seem that where «N» has been stipulated to be synonymous with «the F» the ensuing statement «N is the F» is an analytic statement, and an analytic statement is a priori and necessarily true. So how could you be right in arguing that the a priori and necessary truth of «N is the F» is not necessary for «N» to be an abbreviation of «the F»? Are you suggesting a terminological revision or rather a different conception of analyticity?

  • 8 Kripke 1980: 39.

35I’ll answer these questions by reverting to and commenting on what Kripke says on analyticity in Naming and Necessity. «Let’s just make it a matter of stipulation that an analytic statement is in some sense true by virtue of its meaning and true in all possible worlds by virtue of its meaning. Then something which is analytically true will be both necessary and a priori. (That’s sort of stipulative)»8. Kripke’s suggestion is to consider «analytic» a notational abbreviation of «true and necessarily true in virtue of meaning alone». (Here it is obvious that «in virtue of meaning» is elliptical for «in virtue exclusively of meaning» considering that the truth [or falsity] of a statement in any case [but the case of logical truths] depends on the meaning of the statement).

36To give a definition of a word is not to ensure that anything satisfies the definition. After all a definition proper (i.e. the introduction of a notational abbreviation) is no more than the introduction of a substitution rule to the effect that any occurrence of the longer expression (definiens) can be replaced by an occurrence of the shorter expression (definiendum). So even if we were agreed to adopt the suggested notational abbreviation it still remains to be seen whether there are analytic statements.

37I must say that I find it strange that Kripke of all should suggest the above definition and ungrounded the consequences by Kripke derived. On the one hand the definition is redundant for anyone holding, as Kripke does, that meaning is world indifferent (i.e. that the meaning of an expression is relative to a language but not to a world). If meaning is world indifferent then it goes without saying that if a statement is true in virtue of meaning alone then it is necessarily true in virtue of meaning alone. On the other hand it does not follow, though Kripke finds it obvious, that if a statement is true in virtue of meaning alone then it is a priori true. Suppose a certain individual has two names, «A» and «b». Under the supposition, «A is b» is true in virtue of meaning alone.

38Suppose further that knowledge of the meaning of a name is a priori for a competent speaker. (Stipulations are arbitrary and so do not depend on what in the world.) Still «A is b» need not be a priori knowable on the part of a competent speaker. The reason is that the arithmetic for stipulations is ordinary arithmetic: one stipulation plus one stipulation makes two stipulations and not a new third one. It is stipulative that a certain term refers to a certain individual and that a certain (different) term refers to a certain individual, but it is not stipulative that to the two terms refer to the same individual. Hence a speaker can know that «A» refers to a certain individual and that «b» refers to a certain individual but fail to know whether they refer to the same individual. Hence «A is b» though true in virtue of meaning alone is not a priori true for a competent speaker, hence it is not a priori true. All the above may seem extravagant. But it is not. Suppose that on one occasion a man is introduced to you under the name «A» and on another occasion under the name «b». (It is not that unusual that people should use and be known under different names in different circles). If you do not recognize the man, you know all right that the referent of «A» is the man introduced to you on the first occasion and that the referent of «b» is the man introduced to you on the second occasion. Still you do not know for this that the referent of «A» is the referent of «b». Notice that where «A» and «b» stood for different individuals, «A is b» would be analytically false. (A posteriori necessarily false in virtue of meaning alone).

39The question now is: can we say that a statement of the form «N is the F» where «N» is the notational abbreviation of «the F», i.e. of the form «The F is the F», is analytic, i.e. true (or I insist, false) in virtue of meaning alone (and hence necessarily true in virtue of meaning alone)? It seems to me that the answer is negative. A statement of the form «The F is the F» is true iff exactly one thing is F the description, i.e. the condition indicated by «F», is uniquely realized. (If no thing or more than one thing is F the statement is not true, namely false or neither true nor false according to your preferred analysis of the description). A description does not mean its denotation; whether it has or not a unique denotation is not fixed by its meaning alone. Hence «the F is the F» cannot be true or (untrue) in virtue of meaning alone, i.e. analytically true or untrue.

2. The reference puzzle

40The puzzle derives from the apparent inconsistency of two apparently sound statements:

 

1. A name is a rigid designator i.e. means necessarily what it means
2. A name could have meant something different from what it means

 

41Here I should immediately pause to consider what it is to be meant by «name». I have claimed that the first (hypothetical) puzzle can be dissolved by distinguishing between the introduction of a name by description and the introduction of an abbreviation of a description, i.e. by distinguishing between names and abbreviations. A name, I have said, is not an abbreviation and an abbreviation is not a name (at best it is an abbreviated name). To introduce a name is to assign a meaning to an expression. To introduce an abbreviation is not to assign a meaning to an expression but to make an expression, which (or whose components, where the expression is complex) has already been assigned a meaning, substitutable by a shorter one. This shorter expression will inherit rather than being assigned the meaning of the longer one. Notice that a new name is introduced even when a word already in use is given a new referent. In fact, standardly, new names of individuals are introduced by giving a novel assignment to a word which has already received a number of assignments. A new name rarely is a new word. But a new abbreviation is standardly a new form or at least is an old form by accident rather than design. Coming back to the present puzzle we see that 1. and 2., as formulated, do not constitute a puzzle, i.e. a combination of two prima facie equally sound and yet conflicting statements, but simply a combination of two statements one of which (the second) is a patently false statement. In fact if a name is a meaningful form then it could not mean something different from what it means without being another name.

42So we had better to reformulate the puzzle as:

1’. A name is a rigid designator, i.e. means necessarily what it means. Or, if you prefer, if a nominal form means a, then it necessarily means a
2’. A nominal form might have been associated to something different from what it has been associated to, namely there could have been a different name of the same form

43Nominal forms have arbitrarily assigned referents. If you prefer there is no natural connection between a name and its bearer. The bearer might have been imposed a different nominal form, and the same nominal form could have been made to stand for a different bearer.

  • 9 Ibidem: 21, fn. 21.

44If anything the puzzle seems to become more puzzling when, with Kripke9, we specify 1. as:

1’’. A name rigidly designates its referent even when we speak of counterfactual situations where that referent would not have existed

45To use Kripke’s terminology a name is a de jure rigid designator, i.e. a designator whose reference is stipulated to be a single object, whether we are speaking of the actual world or of a counterfactual situation. On the other hand a definite description «the F» which happens to contain a predicate «F» that in each possible world is true of one and the same unique object is a merely de facto rigid designator.

46The point is correct but misleadingly presented as a difference between two types of rigidity instead of as the difference between two types of designators. Difference, the latter, which is quite independent of modal considerations, and which can be so put: names refer, descriptions do not. A name is stipulated to stand for something (which henceforward is the referent of the name) while definite descriptions are not stipulated to stand for something. Rather when proper they are true of or satisfied by a unique thing. Satisfaction, unlike reference is not just a matter of arbitrary association between an expression and something else.

47The seemingly contradictory 1. and 2. are based on, and seems to do no more than articulating, the notion that meaning unlike satisfaction and truth is a sheer conventional matter, that meaning is world insensitive while satisfaction and truth are world sensitive.

48If so we have to tell apart counterfactual and counterlinguistic hypotheses, i.e. hypotheses which concern satisfaction and truth and hypotheses which concern meaning. To make a counterfactual hypothesis concerning an individual is to suppose that an individual does not have the properties it has. To make a counterlinguistic hypothesis concerning a name is to suppose that a name does not refer to what it refers. A counterfactual hypothesis and a counterlinguistic hypothesis are formulated in a language. That is all they have in common. For a counterfactual hypothesis has nothing to do with language, while a counterlinguistic hypothesis has exclusively to do with language.

49The way things are called has no bearing on how the world is and conversely how things are (their nature and relations) have no bearing on how they are called. (The world has a bearing on naming only in so far as only what there is can be named). This is to say that stipulations (conventions) are world indifferent.

50Any counterfactual hypothesis is made using the accepted language, i.e. using the expressions with the meaning they have been stipulated to have. To suppose that a different stipulation had been made is not a counterfactual hypothesis but a counterlinguistic (or, more generally, a counternormative) hypothesis.

51Counterlinguistic and counterfactual hypotheses are so far apart that it hardly makes sense to combine them in one statement, as testified by:

If «cat» meant the species dog then it would be the case that cats barks

52A quite natural reaction to this conditional will be: eh, what are we talking about? The antecedent is about a word, the consequent about cats. What the hell have words to do with cats?

53For sure a hypothesis concerning what a word could have meant has no bearing on what that word means and hence on the evaluation of a hypothesis formulated in words that mean what they mean (and not what they could have meant).

54«Cat», as used in the consequent, refers to the species cat and cats do not bark and do not start barking even if it is supposed that «cat» means the species dog. The supposition is very short-lived. It is confined to the antecedent. For in the consequent the words, in specie «cat», are not used according to the supposition but for what they actually mean. If we wanted to produce a statement that clearly makes sense, we should keep talking of words all the way through and say:

If «cat» meant the species dog, «cats bark» would be true

55However, «cats bark» would be true if and only if dogs bark. Hence if «cat» meant the species dog it would still be the case that dogs, rather than cats, bark.

56The moral is this. Fix the language and then you can proceed to talk about the world, the way it is or might be or might have been. Or do not fix the language but keep toying with what your expressions might mean or might have meant.

57Then you never reach the stage of making claims about the world.

58The dissolution of the puzzle is at hand provided we take seriously the distinction between counterfactual and counterlinguistic hypotheses, languages and worlds, assignments and evaluations, by separating what is unduly subsumed under the one and the same expression, namely «necessarily»: quantification over worlds and quantification over conventions.

59In short the puzzle is dissolved as soon as we adequately reformulate 1. and 2 as:

1’’’. ∀L (if E means-L a then (∀w) (E means-L a at w))
2’’. ~ (∀L) (E means-L a)
(Where «L» is a variable for languages, «w» is a variable for worlds, «E» is a schematic letter for [quoted] expressions and «a» is an individual constant.)

60What is a language? From the present concerns a language needs not be anything more that a set of ordered pairs whose first element is a word (sound, linguistic form) and whose second element is something different (either a different linguistic form or a non linguistic entity). This set is not a function given that the same word can be associated to different entities. So characterized a language is always changing if for no other reason than the introduction of new names. Would we say that a language is substituted by another when a new born child is assigned a nominal form (no matter if a new nominal form, or an old one)? Is then Italian or English a fiction? May be not. There are important factors for the identification of a language: morphology, surface grammar (i.e. parameters fixing of universal grammar) and last but not least mutual intelligibility. Still I do not think that it is here at all important what a language is. We could forget languages and reformulate 1. and 2. in terms of stipulations (conventions). A stipulation (convention) is no other than an ordered couple (word [form], content), whose first element is a linguistic entity and whose second element is either a non- linguistic entity or a different linguistic entity. (The second disjunct is to take care of the possibility of naming expressions).

1’’’’. (∀C) (if E means-C a then (∀w) (E means-C a at w))
2’’’. ~ (∀C) (E means-C a)
(where «C» is a variable for conventions, «E» is a schematic letter for [quoted] words and «a» is an individual constant.)

61Eventually we have no longer a puzzle.

1’’’’. states that a meaningful word means necessarily what it means, i.e. it has at all worlds the meaning it has at the actual world. This is not at all surprising once it is appreciated that all words get their meaning in the actual world but the nature of the actual world has no bearing on what is named what. The association is arbitrary (intentional) hence is not dictated by the nature of the world. But if the association words-things is not dictated by the nature of the world in which it comes be established, a fortiori is not dictated by the nature of a world where it does not come to be established.

  • 10 Soames 1984: 411-429.

62Some, for example Scott Soames10, have claimed that languages are abstract objects whose semantic properties are essential. The claim even if true is hardly explanatory. The question is why semantic properties are essential. To answer that any change in the semantic properties is a change of language is just a restatement of the claim which does not provide any reason for it. The reason, I submit, is simply that the reference of a word is a sheer conventional matter and conventions are world indifferent. If they depend on what there is they do not depend on how things are. If you prefer, assignment is valuation independent. But something which is world indifferent is not going to change from world to world. Semantic properties are vacuously essential.

2’’’. states that not for all conventions a word means what it means under a certain convention. This is to say that no matter what a word is associated to it might be associated to other things, i.e. appear as first member of ordered couples with different second members.

63Being arbitrary, actual conventions might have been different from what they are. But here we must be careful with «might». Actual conventions might have been different from what they are in the very actual world. If a word were conventionally correlated to b instead of a not for this the actual world would be any different. To suppose that a word meant something different from what it means, it is not to suppose that the world were different from what it is.

64In fact certain words are ambiguous, i.e. they are actually correlated to different things. It is crucially relevant in this respect that certain words do have incompatible meanings. For example «apparent» means both only apparent, i.e. fake, and evident, i.e. not fake. If conventions were state of affairs, «apparent» could not be ambiguous the way it is. If meaning were a natural relation then no word could be in this relation with two properties that exclude each other.

65Having offered my solution of the reference puzzle I must consider possible sources of bafflement. The puzzle was that a word means necessarily what it means and might have meant (or come to mean) something different from what it means. The suggestion for the solution of the puzzle has been to account for the second statement in terms of quantification over conventions rather than worlds. A word means necessarily, i.e. at all worlds, what it means, a word is not bound to mean what it means, i.e. not under all conventions it means what it means under a certain convention.

66One may feel that the distinction between worlds and conventions is not as clear cut as it would be required to make the solution convincing. The point is that even if it were the case, as I have argued, that conventions are not bits of the worlds, still they can be actual or possible exactly as worlds. One could then urge: what is an actual convention if not a convention that is established or operative in the actual world? But if so a convention is in the world and cannot be really told apart from the world. This line of reasoning, though prima facie plausible, is not to be followed.

67«Actual» is a modifier of both «world» and «convention». If meaning is not a natural relation, conventions are not natural phenomena or state of affairs. Hence an actual convention is not a convention which belongs to the actual world. An actual convention is simply an actual convention. If anyone is unsatisfied by this he should explain why he is satisfied to hold that the actual world is the actual world. Actuality is a primitive elusive quality which is not explained by being attributed to worlds rather than conventions.

68Having dealt with one possible source of bafflement I should now dwell on a second one having to do with the essentiality of semantic properties. The essentiality of semantic properties pertains not to words or expressions but to words or expressions that have semantic properties. If an expression «N» is correlated to a thing a, then is essentially correlated to the thing a. If «N» were correlated to something different from a, say b, it would a different name. A different name, not a different expression. The expression is not bound to mean a rather than b or even, if there are functional expressions with a purely syntactical role, anything at all. In the end to say that semantic properties are essential is not a very substantial thesis. It reduces to the platitude that any thing is what it is. If a name is an expression with a referent then it is an expression with the referent it has.

69If the thesis of the essentiality of semantic properties has seemed quite substantial it is because the distinction worlds/conventions is not taken very seriously. Usually it is said that a meaningful word (a name) means necessarily i.e. at all words what it means at the actual world. But if the meaning of a word is a matter of convention and conventions are world insensitive we should rather say that a word actually means that to which is correlated by an actual convention and possibly means that to which is correlated by a possible convention.

70So far so good (let’s suppose). Yet there is a final bafflement. Conventions are arbitrary i.e. are not dictated by the world. This has led me to qualify them as world insensitive or indifferent. Still any convention no matter if actual or only supposed is possible iff the terms to be correlated are actual. A possible convention is not a correlation between possible expressions and possible things. It is a supposed rather than actual correlation between actual expressions and actual things. If so it would seem that after all conventions are actual world dependent rather than world indifferent. I think that this conclusion, though very natural, is not the one we should draw. We should rather say that a world is actual things dependent and a convention is actual things and actual expressions dependent.

71Admittedly, it is perfectly natural to suppose that there might have been individuals that there are not. It is also natural to suppose that these individuals, where human (canine, feline, …), would have had a name. For example Tom and Mary might have had a son and have called him «marc». If so, it will be said, the naming convention would hold between an actual expression and a possible object, hence a possible convention would not be in general actual things dependent.

72I do not like the conclusion since I feel uncomfortable with the notion of possible object. It is not here the place to deal with the almost intractable problem of how to make sense of «Tom and Mary might have a son and have called him “marc”» without recourse to (quantification over) possible objects.

73I will just notice that the hypothetical naming in question is dependent on a counterfactual hypothesis concerning the existence of the prospective namee. The prospective namee has no identity but the one provided by the non identifying condition «being a son of Tom and Mary». Even if the condition were «being the son of Tom and Mary» it would still be non identifying since Tom and Mary could have had more than one son or no son. If we think, as we should, that a naming relation is possible only between determinate entities it would seem that a naming dependent on a counterfactual condition is not after all possible.

74One could object that identifying conditions are not available even for actual objects. The point is, though, that actual objects are given (and not given via identifying conditions) while possible objects either are given by conditions or are not given.

75A final consideration.

  • 11 Kaplan 1989: 497.

76I have suggested that the way out from the second puzzle, resulting from the combination of arbitrariness and rigidity of names, is to distinguish between counterfactual and counterlinguistic hypotheses, between quantification over languages and quantification over worlds. I am not satisfied with the way out from the puzzle since I am not satisfied with the puzzle. For there is a puzzle only in so far as we are inclined to see names as rigid designators. The very notion of a rigid designator applied to names does not make much sense. Semantically, words are names in virtue of having been assigned a referent, of having been arbitrarily associated to an entity. Assignments are evaluation free, while evaluations are assignments dependent. While the notion of evaluation at a world makes perfect sense, the notion of assignment at a world does not make sense. So neither does the notion that an assignment is the same at all worlds, i.e. that a name is a rigid designator. I am just repeating here Kaplan’s succinct point in Demonstratives (p.497): «The referent, in a circumstance, […] is simply independent of the circumstance, and is no… function (constant or otherwise) of circumstance» but for the fact that Kaplan was unable to avoid talking of the referent in a circumstance11.

3. Wittgenstein’s puzzle

77This is a puzzle for anyone holding that the meaning of an individual name is the individual named by the name. The problem, to make it short, is this: how can the bearer of the name be the meaning of a name?

  • 12 Wittgenstein 1953.

78At §40 of Philosophische Untersuchungen12, Wittgenstein remarks: If N.N. dies we say that the bearer of the name has died, not that the meaning of the name has died. If the name ceased to have meaning, it would be meaningless to say: N.N. has died. Hence, Wittgenstein concludes, the meaning of a name is not its bearer.

79It should be immediately noticed that the problem raised by Wittgenstein is how can «N» refer to a when a does not any longer exist and not, mind you, how can «N» refer to a when a has never existed. No expression could be a name of, i.e. refer to, what has never existed. Concrete individuals come and go out of existence in this very world. So the problem is how the name of a can still refer to a when a has passed away. I have insisted above that if «N» means a then necessarily «N» means a. The reason is that nothing short of a different stipulation could make it the case that N does not refer to a. but in such a case necessarily «N» does not refer to a. Stipulations once enforced are untouched by the vagaries of the world(s) and the destiny and perishable nature of the referent. It is the stipulation, i.e. the reference and not the referent that is resilient.

80Pace Wittgenstein, strange as it may sound, if the bearer of the name dies then the meaning, i.e. the individual meant by the name, dies. So instead of saying, as admittedly we are inclined to say, that N.N. has died (is dead) we could as well say that the meaning of «N.N.» has died. Wittgenstein is no doubt right to hold that if the name ceased to have meaning it would be meaningless to say that N.N is dead. The crucial step, in Wittgenstein swift argument, is the passage from saying that N.N is dead (or, if you prefer, that the meaning of «N.N.» is dead) to saying that «N.N.» is therefore without meaning.

81The step is incorrect. To convince yourself of the incorrectness think of the following. Joyce is the author of Ulysses. Joyce is dead, hence the author of Ulysses is dead. I hope everybody will agree that not for this Ulysses is (now) without an author and that the author is still Joyce. As long as there is the book the book has an author. The author is the individual who wrote the book. The death of the author has no bearing on the existence of the book (this is maybe one of the reasons people write books). But there is no book without an author and there are plenty of books with dead authors. In short books come into existence by being written and as long the book exists (there are some copies around) it has the property of having an author. If no copies of the book were around we would say that there was a certain book which had a certain author.

82Likewise as long there is «N.N.», as long the name is used, the name has a reference. The death of the referent has no bearing on the existence of the name (this is why people are worried that their name could be used to say nasty things about them after their death). The death of the referent has no bearing on the reference of the name. Names are often used in gossip, that is in absence of the referent. It is not going to make a significant difference whether, excuse the cynicism, the referent is absent temporarily or permanently. The interesting thing about books and names is that we say that a book has an author even when the author is dead and that a name has a reference even when the referent is dead. A book has the author it had, and the name has the referent it had. Authors pass away and referents too. But books do not lose the property of having an author, nor do names lose the property of having a referent. Both authors and referents are required for no longer than the time required to do the job. Introducing a name is much quicker than writing a book. But when the job is done it is done. And nothing is going to disrupt it, short of the extinction of the book or of the name.

Haut de page

Bibliographie

Donnellan, K.

1979, The Contingent a Priori and Rigid Designators, in p. French, T. Uehling, H. Wettstein (eds.), Contemporary Perspectives in the Philosophy of Language, Minneapolis, University of Minnesota Press

Kaplan, D.

1989, Demonstratives, in J. Almog, J. Perry, H. Wettstein (eds,), Themes from Kaplan, New York, Oxford University Press

Kripke, S.

1980, Naming and Necessity, Oxford, Blackwell

Soames, S.

1984, What is a theory of truth?, “The Journal of philosophy”, 81: 411-429

Wittgenstein, L.

1953, Philosophische Untersuchungen, Oxford, Blackwell

Haut de page

Notes

2 Kripke 1980: 57.

3 Ibidem: 25, fn. 3.

4 Ibidem: 80, fn. 34.

5 Ibidem: 60.

6 See Donnellan 1979: 51.

7 Kaplan 1989: 509.

8 Kripke 1980: 39.

9 Ibidem: 21, fn. 21.

10 Soames 1984: 411-429.

11 Kaplan 1989: 497.

12 Wittgenstein 1953.

Haut de page

Pour citer cet article

Référence papier

Ernesto Napoli, « Reference fixing and the stiffness of reference or three (would be) puzzles concerning names »Rivista di estetica, 47 | 2011, 179-195.

Référence électronique

Ernesto Napoli, « Reference fixing and the stiffness of reference or three (would be) puzzles concerning names »Rivista di estetica [En ligne], 47 | 2011, mis en ligne le 30 novembre 2015, consulté le 18 avril 2024. URL : http://journals.openedition.org/estetica/1946 ; DOI : https://doi.org/10.4000/estetica.1946

Haut de page

Droits d’auteur

CC-BY-NC-ND-4.0

Le texte seul est utilisable sous licence CC BY-NC-ND 4.0. Les autres éléments (illustrations, fichiers annexes importés) sont « Tous droits réservés », sauf mention contraire.

Haut de page
Rechercher dans OpenEdition Search

Vous allez être redirigé vers OpenEdition Search