Sie sind auf Seite 1von 24

Journal of Philosophy, Inc.

Singular Terms Author(s): Michael Devitt Source: The Journal of Philosophy, Vol. 71, No. 7 (Apr. 18, 1974), pp. 183-205 Published by: Journal of Philosophy, Inc. Stable URL: http://www.jstor.org/stable/2025347 . Accessed: 14/02/2014 20:39
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at . http://www.jstor.org/page/info/about/policies/terms.jsp

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

Journal of Philosophy, Inc. is collaborating with JSTOR to digitize, preserve and extend access to The Journal of Philosophy.

http://www.jstor.org

This content downloaded from 75.97.192.97 on Fri, 14 Feb 2014 20:39:06 PM All use subject to JSTOR Terms and Conditions

THE JOURNALOF PHILOSOPHY


VOLUME LXXI, NO. 7, APRIL I8, 1974

~~~~~~

-+

is that of explaining the nature of the link betweenname and object in virtue of which the formerdesignatesthe latter.' From Frege and Russell through to Strawsonand Searle, of the object that the solution has been sought in the descriptions usersof the name associatewith the name. Saul Kripke has shown of names are mistaken.They are misthatall such "sense-theories" He has also inditaken not merelyin details but in fundamentals.

SINGULAR TERMS

names ofproper in giving thesemantics HE mainproblem

* Earlier draftsof this paper were given at the Australian National University and the Universityof Sydneyin October, 1972. It has benefitedfrom comments made on those occasions. Also, it has benefited from many other comments, especiallythose made by HartryField. 1 I use the term 'designate' to express the relationship between any definite token and its object, and also that between the person who prosingular-term duicedthat token and the object (but see the qualificationsat the end of secs. 5 and 7). I have no special fondnessfor the term,but it seems more apt thall th-le other available ordinary semantic terms,'refer' and 'deniote'. Ordinary language provides us with very few words to express the relationpresumably,the lack of interestin ships betweenwords and the world (reflecting, semantics in ordinary life). All of them have very wide uses. So, in startingon scientific semantics,we have available to us only a meager vocabulary for marking the many distinctionsthat we may find appropriate. For example, the links between proper name and object, definitedescription and object, and demonterms strativeand object are all very different, and we might well use different for them. Also, we might use different terms for the relationship between the user of these singular terms and the objects. However, we can manage well enough here with the one term,'designate'. Nothing hinges on this terminological question. In doing semantics we are not "analyzing ordinary usage." We are concerned with the nature of the relationships themselves,whatever they are called. Thus with a proper name, we are concerned with the nature of a certain relationship it has to just one object (which we might ordinarily say is "its bearer," "the object it refersto," "the object it designates," etc.). The relationship in question is picked out by its crucial bearing on the truth value of sentencescontainingthe name. of usage in sec. 6. discuission There is a further

I83

This content downloaded from 75.97.192.97 on Fri, 14 Feb 2014 20:39:06 PM All use subject to JSTOR Terms and Conditions

I84

HTHE

JOURNAL OF PHILOSOPHY

cated wherethe truth of thematterlies, namelyin a "causal theory" of propernames.2This will be the centralconcernof the paper. In part I I shall develop the causal theory, confining my attention to nonemptynames in "purely referential position."3 In. part II I shall first considerdefinitedescriptions, arguing that a distinction recently drawn by Donnellan4 is to be explained in terms of causal linksto objectssimilarto thoserevealedin part I fornames. Second, I shall draw and explain a similar distinctionfor demonstratives I shall reand personalpronouns.Making use of thesedistinctions, turnto thediscussionof propernames in part iI.
I

1. The centralidea of the causal theory of propernames is thatour presentuses of a name, say 'Aristotle', designatethe famous Greek philosopher not in virtueof the variousthings we (rightly) Aristotle, believe true of him, but in virtue of a causal networkstretching back from our uses to the firstuses of the name to designate from Aristotle.Our presentuses of a name borrowtheirreference earlieruses. It is thissocial mechanismthat enables us all to designate the same thingby a name. This central idea makes our present uses of a name causally dependenton earlieruses of it. These causal links do not,however, take us to the object. In virtueof what do the first uses of a name designate a certain object? We can see, perhaps, how we are dependent on our ancestors,but how did they manage? Other questions occur to us. What is the nature of this causal How did it begin and how did it grow?What has mycausal network? connectionto Aristotle got to do withmypresentact of designating
heard Kripke'sviewsat Harvardin Fall 1967.As a resultI 253-555.I first The Semantics of namesin myPh.D. thesis, of proper a causal theory developed paper has 1972.Kripke's Proper Names: A Causal Theory, HarvardUniversity, respects. in certain them but goesbeyond withhis 1967lectures, muchin common lines to thosein my thesis, are along similar One or twoof thesedevelopments paper is beforeKripke'spaper was available.The present whichwas written ii of the thesis. chapter drawnfrom largely are to be foundin KeithS. Donnellan, of sensetheories criticisms Somesimilar op. cit., in Davidsonand Harman, Descriptions," "ProperNamesand Identifying pp. 356-379. 8 I use 'name(s)'as short name(s)'. for'proper bothrequire names, and empty specialtreatan opaque context, Nameswithin ment. 4 "Reference Review,Lxxv,3 (July Philosophical Descriptions," and Definite "Donnellan 1966"; and "PuttingHumptyDumpty 1966): 281-304,hereafter, "Donnellan TogetherAgain,"ibid., LxXVII, 2 (April 1968): 203-215,hereafter 1968." in D. Davidson and G. Harman,eds., Semantics 2"Naming and Necessity," 1972), pp. Reidel; New York: Humanities, of Natural Language (Dordrecht:

This content downloaded from 75.97.192.97 on Fri, 14 Feb 2014 20:39:06 PM All use subject to JSTOR Terms and Conditions

SINGULAR TERMS

I85

him? Could a use of a name be causally linked in the appropriate way to morethanone object? 2. First Uses of a Proper Namne.A paradigmsituationfornamingis one in which a name is given to a previously unnamed object in a face-to-face confrontation at a "naming ceremony."The sort of ceremony that leaps to mind here is a christening or the ceremony launching of a ship. Mostly,however,such formal and elaborate proceduresmerely give religiousand public expressionto what has already been establishedinformally and more privately. The object in theparadigmis likelyto be a humbleone, and so we shall take such a case to illustratea naming ceremony. (Thinking about names has not been helped by limiting attention to the famous and the grand.) Consider the case of our late cat. We acquired her as a kitten.My wife said, "Let us call her 'Nana' after Zola's courtesan."I agreed. Thus Nana was named. This is the typicalway for a name to be bestowed,but thereare others.We shall discuss these in section 9. What happened to those presentat the naming of Nana? They perceived the ceremony, using at least theireyes and ears. To perceive somethingis to be causally affected by it. As a result of the effect it had on them,theywere in a positionto use the name 'Nana' later to designate the cat. What theygained at the ceremony, it seems appropriate to say, was "an ability to designate Nana by 'Nana'." 5 Let us expand this storya little,considering my situationat the naming.I gained the abilityfromperceiving the complex eventthat constitutedthe naming ceremony.I saw Nana. I saw my Nvife. I heard my wife'ssuggestion. I was aware of agreeing.I knew which a name for.6 As a resultof the causal interobject she was suggesting action at that ceremonyamong my wife, Nana, and myself,an interactionin which Nana occupied a certain place (that of an object being named), I gained my ability.
5 In order to gain this ability they must already have several other abilities. To gain the ability to use this name, they must already have the ability to use names in general. And they must realize that a name can be bestowed on an object by a ceremonyof the sort witnessed.This requires, inter alia, that they have the ability to use 'her' to designate objects (see fn. 6). 6Clearly the role of 'her' in my wife's remark is important here. It was because it designated Nana that the name was bestowed on her at this ceremony. We discuss the role of personal pronouns in sec. 7. Foreshadowingthat discussion, we can say now that 'her' designated the cat because of the place she had in the causal explanation of my wife's utterance. My wife might have used a definitedescription (e.g., 'our cat') instead of a pronoun to pick out Nana. See sec. 5 for discussion of this.

This content downloaded from 75.97.192.97 on Fri, 14 Feb 2014 20:39:06 PM All use subject to JSTOR Terms and Conditions

i86

THE JOURNAL OF PHILOSOPHY

A few minutes later I exercised my ability: I said "Nana is hungry." That first use of the name designatedNana. How? It designated her, because it was in fact produced by an ability that arose out of the above ceremonyin which she had a certain place. In other words, it was because Nana had that special place in the causal explanation of my utterancethat the name designatedher. The centralidea of the causal theory was that presentuses of a name are causallylinked to first uses. We see now that first uses are causallylinked to the object.7 Our account rests on talk of "abilities to designate objects by name." What is such an ability? We cannotoffer but we a reduction, can point toward one. As materialistswe expect the advance of to show us that it is a certain science,particularly neurophysiology, sort of state of the central nervous system.It is a state which is broughtabout in a language user by perceptionof a naming ceremony (and in other ways to be described) and which is apt to produce (in part) certain sortsof utterances, viz., utterancesusing the name in question. It is such states,whatever they are, that the links betweennames and theirobjects.8 largelyconstitute We shall say that underlying the use of a name is a causal chain groundedin the object the name designates. The chain underlying myfirst use of 'Nana' beginswithNana at her naming ceremony; it runsthrough of thatceremony; myperception fromthenon it is my ability thus gained to use 'Nana' to designateher. 3. Later Uses of a Proper Name. Two of us gained our abilities to designateNana by her name at the naming ceremony. All others, borrowtheirreference or indirectly, directly fromthese two. fromone of the two. I mightintroMany gain the abilitydirectly duce them to the cat: (i) "She is called 'Nana'," or (ii) "This is Nana." This ceremony plays the role forthemthat the earliernaming ceremonyplayed for me. Their perception of Nana in that introductionwill mean that underlyingtheir later uses of 'Nana' will be causal chains grounded in her. I mightpass on the ability in Nana's absence: (iii) "Our cat is called 'Nana'," or (iv) "Nana is our cat." An ability thus gained would also be causally grounded in Nana, although this is not so apparent(forNana is absent).In the first place, in using or mentioning the name 'Nana', I have exercisedmy abilitywhich is causally
them to pointing, but that is another story which we shall not attempt here.
7 In sec. 6 we shall consider some apparent exceptions. 8 Can we relate these states to nonlinguisticbehavior? I think we can relate

This content downloaded from 75.97.192.97 on Fri, 14 Feb 2014 20:39:06 PM All use subject to JSTOR Terms and Conditions

SINGULAR TERMS

I87

grounded in Nana. The person addressed hears my remark.This causallylinkshim throughmy abilityto Nana. In the second place, we mustconsiderthe role of 'our cat'. So farwe have not considered (save in note 6) the role of the other singular termsused to designate Nana at the timeabilities are gained. There was, forexample, my wife'suse of 'her' in the naming ceremony, my use of 'she' in (i), 'this' in (ii), and now 'our cat' in (iii) and (iv). Such singular termshave an importantrole in causally groundingthe ability to designateby name in the object. We must set aside showing this, however, until we have discussedthese termsin part II.9 There is one other importantway I mightpass on the ability: I might use thename in an ordinary predication. For example,I might say, (v) "Nana is hiding." Someone who hears this is in a position to borrowhis reference fromme.10If he does, he gains an ability thatis causallygroundedin Nana via the abilityI exercisedin making theremark. We have seen how those presentat a naming ceremony can pass on the ability to designatean object by its name. There are many usersof a name who neitherwere at the namingceremony nor have come to their use fromanyone who was at it. We are all in this situationwith'Cicero'. Consider again the case of Nana. Those who gained the name from the two of us present at the naming ceremonywere then in as good a positionto pass it on as we were.And theypass it on in similar ways. People are told, "The Devitts' cat is called 'Nana'," or "Nana is an unusually patternedcat," and thereby gain the appropriateability.Their later uses of the name designateNana because she is in factthe object at the base of the causal chains underlying those uses, chains that run throughseveralpeople's abilities. And so the chains continue: people acquire and use the name long afterNana is withus.
9 Had Nana another name known to the auditor, I could of course pass on 'Nana' by using it. 10Sense-theories require much more extensiveknowledge for someone to use a name properly.What we primarilyreject in rejecting these theories is this insistence on knowledge: the link between name and object is not mediated by descriptionsassociated with the name which are true of just that one object. However, we need to go further:we do not require that a user of a name have any substantial set of beliefs involving the name (whether true or false). And now we seem to be going even further:we are requiring scarcely any beliefs. Perhaps this goes a little beyond our ordinary intuitions. Precisely where we ordinarilydraw the line is unclear, although it is clear that we do not require many beliefs.In the light of the causal theorythere seems no reason to reject a case like the one discussed(see also fn.36, below).

This content downloaded from 75.97.192.97 on Fri, 14 Feb 2014 20:39:06 PM All use subject to JSTOR Terms and Conditions

i88

THE JOURNAL OF PHILOSOPY-IY

Under each of our uses of a name lies a causal chain. These causal for the chains are linked togetherand form the cansal networ-k name. So far we have consideredonly those uses of a name which pass it on. In fact,most uses of a name are to an audience that already in a memberof the audience the has it. Each of theseuses reinforces causal linkages abilityhe has with the name. It establishesfurther betweenhim and the object. Underlyinga person'suse of a name be many causal chains all grounded in the object: may therefore his use. The over-allnetunderlying theremay be a causal network workfor the name is the union of all such individual networks. It is a commonplacethata personin a positionto pick up a name may fail to do so: he may fail to pay attentionat an introduction; the requiredabilityto designatethe object is not acquired. Further, a person who had the use of a name may lose it; the abilityfades throughlack of exercise.11 on 4. Ambiguous Proper Names. So far I have ignored,as writers proper names are prone to, the fact that most proper names have more than one bearer; they are "ambiguous." The ambiguityof 'Nana' was clear fromthe start,for it was Zola's use of it that led to its being bestowedon our cat. And thereare many names much more ambiguous than 'Nana': consider 'John', for example. We need to extend our discussion to take account of this fact of
ambiguity.12

Which object does a name designate?It is natural to say that it designatesthe object the speaker had in mind or meant. This was an insightof sense-theorists.'3 Clearlywhat we need then is a satisanalysisof this vague talk. With such an analysis in hand, factory the solution to our problem of ambiguitywould be in sight: a speaker designatedone object and not another by 'John', because he had it in mind. In general,one has an object in mind in virtueof a causal connectionbetweenone's state of mind and the object. With the help
11 This loss of ability is a failure of memory.For a sense theory,what is forgotten is the required associated descriptions.For the causal theorythe inability to produce descriptionsusually associated with a name is evidence for loss of ability with the name, but does not constituteit. 12 Many philosophers have felt that ambiguities in names are removed by the context of use, by which is meant the context external to the speaker's mind. I have criticized this view, claiming that the context is only the guide to an underlyingreality,not the reality itself; see my "Semantics and the Ambiguity of ProperNames," forthcoming. 13 See particularly, P. F. Strawson, Individuals (London: Methuen, 1959), p. 182.

This content downloaded from 75.97.192.97 on Fri, 14 Feb 2014 20:39:06 PM All use subject to JSTOR Terms and Conditions

SINGULAR TERMS

I89

of this,we are in the position to incorporatethe insightof sensetheoristsinto our causal theoryby giving the following rough analysisof having an object in mind in using a name (meaning an object bya name):14 For anyx, y,and z, x had y in mindin uttering a token of thename z (x meant type a token y in uttering of thenametype z) ifand onlyif x had an ability to designate y byz and thatability was exercised in theproduction ofthattoken ofz. What bearing does thishave on the problemof the ambiguity of names? Take the name type 'John'. It is probably the case that most of us can designate about thirtydifferent people with this name. Those who can each have the same numberof distinctabilities involvingthe name, causally based on that number of objects. When theyutterthe name having a certainperson in mind, there is (normally)one and only one of these abilities exercised in the productionof the token. Which object a person has in mind depends on whichabilityhe in factexercises. We can say roughlyalso that a name tokendesignatesan object if and only if thespeakerhad the object in mind (meant the object) in uttering the token. Why do we qualifyour claims here ("roughly,""normally")?Because there are a number of cases that throwdoubt on them as theystand. These are cases where "thingsgo wrong"for a speaker. We shall discussthemin section10. More than one abilityto use a certainname may be involved in the causal explanation of an utteranceeven though only one is exercised.Consi-der, for example, my earlier-mentioned utterance, "Nana is hungry." AlthoughI exercisedmyabilityto designateour cat in saying this,my ability to designateZola's courtesanby the name (settingaside here any problems of such talk about empty had somie causal role in myutterance:it was partly names) certainly because of that ability that our cat was given her name. The utterance,however,was about the cat, not the courtesan, because the abilityused to produce it arose froma naming ceremony involving the cat; we are not concernedwith the causal explanation of that ceremony. Each timewe hear a name used, we must,in understanding it, associate it with an ability (unless we form a new ability on the of it). It is possible to do this wronglyand hence to misstrength
14Such locutions so thatone can mean Tully but can be construed opaquely, not mean Cicero.Our concernis with the transparent conistrual.

This content downloaded from 75.97.192.97 on Fri, 14 Feb 2014 20:39:06 PM All use subject to JSTOR Terms and Conditions

I90

THE JOURNAL OF PHILOSOPHY

understandthe remark.Misunderstandings are common with very ambiguous names like 'John'. We shall consider the consequences of themlater (in section 10). As a resultof manyremarks using the name type 'John', we acquire many beliefs concerning various different people of thatname. The beliefsconcerning people are, in some sense,"stored"separately with theirrespective abilities. We shall return to the discussion of proper names in part iii. 5. Donnellan's Distinction.We have claimed that names designate theirobjects because theyare causally linked to them. Can we say anythingsimilar about definite descriptions? At first sightit seems not: a definite description designatesthe object it describes.15 Howof a recent distinctionmade by Keith Donnelever, investigation lan 16 suggests thatthisis too hasty. Donnellan distinguishestwo uses of definite descriptions,an use: "attributive" use and a "referential" in an assertion whousesa definite A speaker description attributively or whatever is theso-and-so. aboutwhoever A speaker states something in an assertion, on the who uses a definite description referentially to enable his audienceto pick out otherhand,uses the description about and states whomor whathe is talking about that something person or thing (1966, p. 285). Donnellan brings out his distinctionby giving a number of exof situations wherehe claimsa personis "speakamples,particularly to" (295), saying somethingof ing about" (ibid., 286), "referring even though the description (301), someone in using a description, describe that person. These are referential does not correctly uses Attributive differ in uses of the description. thisrespect.We would naturallymark the distinctionby saying that in a referential use the speaker has a certain object in mind in using the description, use he does not.17 whereas in an attributive Consider two of Donnellan's examples. Suppose said ... in 1960before he had anyidea thatMr. Gold. .. someone in 1964,"The Republican waterwouldbe the Republicannominee in 1964 will be a conservative" candidateforpresident (293).
15 M,oreprecisely, the definite description 'the F' designates the one and onlyobjectthat'F' is trueof. 16Donnellan1966 and 1968.My own awareness of thisdistinction is due to for manyyears. who has been urgingit in lectures C. B. Martin, 'I Donnellanimplicitly marksit thisway himself in manyplaces (e.g.,ibid., 287).

1U

This content downloaded from 75.97.192.97 on Fri, 14 Feb 2014 20:39:06 PM All use subject to JSTOR Terms and Conditions

SINGULAR TERMS

I9I

Suppose thejudgmentwas based on an assessment of over-alltrends within the party.The descriptionin that utteranceis used attributively:the speakerdoes not have any particularobject in mind; he is speakingabout whoeverhappens to become the candidate in 1964.In contrast, . .. suppose thatJones has been charged Smith's witlh murder and he has been placedon trial.Imaginethatthere of Jones's is a discussion odd behavior at his trial.We might of his sum up our impressions behavior by saying, murderer is insane"(286). "Smith's The description here is used referentially. The speakerhas a certain object in mind,namelyJones. Leave aside fora momentthe questionof thesenmantic significance of this. The distinctionseems to be a good one. However, it cries out for explanation. How can a speaker manage to "refer"to an object using a descriptionthat does not describe it? In virtue of what is there"a rightthingto be picked out by the audience" (304)? We have earlier given a causal analysis of having an object in mind in using a name. This points to what we need to say here. It was because of our experiencesof Jones during his trial, and our beliefs about him, that we used 'Smith's murderer'in that utterance. Similarly, it was becautse of my experiencesof Nana, and my beliefs about her, that I used 'our cat' in those earlier remarks aimed at passing on her name, (iii) and (iv). In a sense, the object itself leads us to use the particulardefinite description in such cases. On the otherhand, Goldwaterhad (near enough) no role at all in bringingabout the use of 'The Republican candidate for president in 1964'. There was no causal link between the speaker and Goldwaterin virtueof which the speakerutteredwhat he did. There can be a causal link of the required kind even thoughthe speakerhas had no directexperienceof theobject: it will be a causal connectionrunning throughothersback to speakerswho did experience the object. Thus, someone who has heard about our cat from me, but has never met her, can have her in mind by 'the Devitts' cat'. And we can all have Aristotle in mind by 'the philosopherwho taught Alexander the Great'. One can "borrow" the abilityto have something in mind. We have said next to nothingabout the natureof the causal connectionrequiredforhavingan object in mind in usinga description. Rather,we have relied on the veryobvious difference betweenthe causal explanationsof referential and attributive uses; in particular, on the very obvious difference in the causal role of the object

This content downloaded from 75.97.192.97 on Fri, 14 Feb 2014 20:39:06 PM All use subject to JSTOR Terms and Conditions

192

THE JOURNAL OF PHILOSOPHY

described.We shall now consideranother case which will lead us to say more. However, to say much more requires an analysis of here. belief-contexts, whichcannot be undertaken Donnellan emphasizes that the descriptionin a sentence can on one occasion be used attributively and on anotherreferentially. We have illustrated is insane'. a referential use in 'Smith'smurderer Now considerthefollowing situation: we come upon poor Smithfoully From the brutal murdered. manner of thekillingand the factthatSmith was the mostlovable person in theworld, we might exclaim, "Smith's murderer is insane." . . . assume ... that... we do notknowwhomurdered Smith (285). We do not have anyone particularin mind; the use of the description is attributive. Yet clearly there is a causal link between the murdererand us (via the corpse) in virtue of which we used the description 'Smith'smurderer'. 'What distinguishes the use of this descriptionfrom the earlier useswe have classified as referential is thathereit is not experienceof the object that leads us to use the description to referto it. Before, we had actuallyseen Jonesat his trial; and it was myacquaintance with Nana that led to my use of 'our cat'. Now, however,we may have never seen the murderer. Or, if we have, seeing him has not led us to use the description:we do not associatethe personwe saw withthemurder. It would seem that,fora speakerto have the object in mind,his use of the descriptionmust be based on perceptionof it. (He need not have perceivedtheobject himself, of course.Those who have perceived it can pass on the abilityto others.)Further, it is preferable that thisperceptionbe of the face-to-face variety. Considerthe other extreme.Suppose that, at the time we come upon Smith foully murdered,we see a man fleeingin the distance whom we take to be the murderer. Many would doubt that thisis sufficient forus to have a person in mind in using 'Smith'smurderer'. We would not "fix"on theobject. The indubitablecases of having have a sufficient an object in mind are based on face-to-face perceptionof it. It is, indeed, appropriateenough that having an object in mind should be based on face-to-face ofit. perception We have discovere-d thathavingan object in mind is not a notion withsharpborders.But thisis no surprise, nor is it important. Donnellan has detected,at the level of intuitivesemantics, a difin "meaning"or "function" ference whichis markedby theordinary expression'havingan object in mind'. My claim is that thisdistinc-

This content downloaded from 75.97.192.97 on Fri, 14 Feb 2014 20:39:06 PM All use subject to JSTOR Terms and Conditions

SINGULAR TERMS

I93

It is best tion is best drawn by looking to the cause of utterances. drawn this way, because the causal theory goes a long way toward explainingit. Donnellan's term'referential' is not appropriateforme. I follow Quine's use of 'refer',according to which even predicatesrefer.I therefore replace 'referential' with the somewhatclumsy'designational'. I prefertalk of tokens to talk of uses of types.So, in my hands, Donnellan's distinctionbecomes that between attributive definite-description-tokens (briefly"a-descriptions") and designational definite-description-tokens (briefly "d-descriptions"). According to my usage, then,both a-descriptions and d-descriptions may refer, but onlyd-descriptions maydesignate. 6. The SemanticSignificance ofDonnellan's Distinction.The semantic significance of the causal link betweena name-token and an object is clear, but what is the significance of that between a description-token and an object?18 Afterall, even an a-description enables us to say somethingabout the object described.We have already granted to Donnellan's distinction,and hence to this link, some at the level of intuitivesemantics. significance But Donnellan makes more substantialclaims for the distinction at that level. These concern the role of a d-description and, related to this (and more importantly), thetruth values ofsentences containing one. (1) Accordingto Donnellan, a d-description "refers"to the object the speakerhad in mind even when it does not correctly describe thatobject.Further, thesentencecontaining the d-description is true or falseaccordingas the predicatein it is trueor falseof that object which the speaker had in mind.19 In all Donnellan's examples the d-description does not correctly describeanything;so the choice is betweenreference to what the speaker had in mind and reference failure.Donnellan plumps for the former. He does not discuss any where the speaker has one object in mind but the d-deexample he uses correctly scription describesanother.However, the implication of his discussionis clear: in such a case the d-description refers to thefirst object and the truth value of the sentencedepends on its

characteristics.20 Donnellan's claims are too strong. First, many of his remarks using

18 Kripke, op. cit., p. 343n, doubts that Donnellan's distinctionhas any semantic significance. 19See, e.g., Donnellan 1966, p. 295. Donnellan allows that in some "extreme this may not be the case. circumstances," 20 See ibid., p. 301, particularlythe sentence: "It does not matterhere whether or not the woman has a husband or whether,if she does, Jones is her husband" (my emphasis).

This content downloaded from 75.97.192.97 on Fri, 14 Feb 2014 20:39:06 PM All use subject to JSTOR Terms and Conditions

194

THE JOURNAL OF PHILOSOPHY

the term'refer'seem to presuppose that,pre-theoretically, there is a clear-cutsemantic notion picked out by this term which it is our task to investigate.21 In fact,this termin philosophyis largely a termof art gaining its meanings fromthe semantic theoriesin which it is embodied. This is true of Donnellan's use, as it is also of mine. The pre-theoretical ("ordinary")use of the termis so loose thatit can encompassa variety of sulch meanings.Donnellan's claims about "reference"become substantial,rather than merelyverbal, therefore, only when we see theirbearingon his claims about truth. Suppose that I was under the misapprehensionthat Nana was our neighbor'scat which we were looking after for a while; my wife had told me this storyin order to get Nana into the house. The day afterher arrival,she disappears. Talking about this later in the day, I say, "Our neighbor'scat has disappeared." Now, in fact,our neighborhas a cat, Jemima,whom I have never seen or heard of and who is safelyat home. Did 'our neighbor'scat' refer to Nana or Jemima?My claim is that,taken on its own, this is a purely verbal question. Clearly, my description is semantically linked to both cats,though the links are of a different kind: I had Nana in mind, but the object my descriptioncorrectly described was Jemima. Whetherwe say thatthedescription "refers to" ("designates,"or "denotes") the one or the otheris of no interest until we see what followswithinthe theory fromso saying. This brings us to the second point. Donnellan wiould say (it to Nana and, hence, that what I said was seems) that I referred true.We are here faced with a substantialquestion to which Donnellan gives a simpleanswer.It seemsto me that the correct answer is far fromsimple. When we attend to the semanticlink to Nana we are indeed inclined to say the sentenceis true,but when we attend to the different link to Jemima we are inclined to say it is false. Consideringthe whole picture,we don't know what to say. There is one thing we might say which, despite its paradoxical sound, seems to me to be right: that it was partlytrue and partly false.'We shall discussthenotionof Partial truth in section10. In cases of the sort Donnellan discusses,where the description does not correctly describe anythingand so there is only one object involved,we are morelikelyto accept that the statement is true or falseas the case maybe withthatone object. But wre mightprefer
21 See e.g., ibid., p. 293. This is a common presupposition,as some of Donnellan's referencesto the views of others show. At one point (1968, p. 210) Donnellan himself claims that dispute here is merelyverbal. Yet many of his claims about "reference"have the ring of something far more substantial than this would allowv.

This content downloaded from 75.97.192.97 on Fri, 14 Feb 2014 20:39:06 PM All use subject to JSTOR Terms and Conditions

SINGULAR TERMS

195

to say, and I thinkwe should say, that it is partlytrue (false) and partlytruth-valueless. Initiallywe grantedto Donnellan's distinction, and hence to the causal link,some intuitivesemanticsignificance. We have now gone a little further:we have seen that it bears on the truthvalues of statements: because I had Nana in mind, "Our neighbor'scat has disappeared" is not simply false; though it is not simply true, as Donnellan claims, either.And the distinctionhas more significance yet. (2) It has oftenbeen noted that manyof the definite descriptions that we ordinarilyuse correctly describe lots of objects; Russell's uniqueness condition is not satisfied. Consider the sentence,"Put the book on the table"; the world is full of books and tables. Following our earlier usage, we shall call these,"ambiguous" definite descriptions. Justas the solution to the problemof ambiguousnames has been sought in the context,so also has that of ambiguous descriptions. One is inclined, as Donnellan points out, to save Russell's view by relyingon the context"to supply further qualificationson the description to make it unique" (Donnellan 1968,p. 204n). Donnellan himselfseems to suggestthat the contextsettleswhich object an ambiguous d-description refersto.22We, on the other hand, make use of our explanation of Donnellan's distinction to offer an analysishereanalogous to our earlieranalysisforambiguousnames. An ambiguous d-descriptiondesignates the object the speaker had in mind;23i.e., it designates the object that causally results in the use of the description.Our earlier speaker designated this book and that table because of their special place in the causal explanation of his utterance. The external context is merely 24 a guide to thisreality. This discoveryis certainlyof semantic significance. The causal link has a role in determining the designatumof a d-description and, hence, the truthvalue of the sentencecontainingit. (3) We lhave pointed out earlier (in section 2) that a definite descriptionmay be used at a naming ceremonyto pick out the object to be named. So the connectionbetween a name token and its object may be mediatedby a description. Clearly if that connecHis remarks here are only suggestive: he is not attempting an analysis. We are now ignoringcases where the descriptiondoes not correctly describe the object in mind. 240ur argument against the contextual view here would be similar to that against it for names; see my "Semaniticsand the Ambiguityof Proper Names," Op. cit.
22
23

This content downloaded from 75.97.192.97 on Fri, 14 Feb 2014 20:39:06 PM All use subject to JSTOR Terms and Conditions

I96

THE JOURNAL OF PHILOSOPHY

tion is to be an unbroken causal chain, as we have claimed, that mustbe a d-description; description is causally only a d-description based on the object. And what we findin all normal naming ceremonies,indeed, is that the mediatingdescription is a d-description. We can, however, inventsome veryabnormalnaming ceremonies whereit is not. "Let us call theheaviestfishin the sea 'Oscar'." 'The heaviest fish in the sea' is an a-description. If we go on to use 'Oscar', the name would not be causallybased on an object. Would this count as a naming ceremony, and would 'Oscar' count as a name? We need not legislateon this ratheruninteresting question. 'Ve shall merelynote that the "naming-ceremony" and the "name" are abnormal and that theydiffer fromnormal ones in the respect noted. We shall mark the difference by calling the likes of 'Oscar' "attributivenames" (briefly, a-names),reservingthe term 'name' forthenormalones.25 This bearing of our discussion of definitedescriptionson the semanticsof names may not seem verysignificant. For, even if what we have claimed to be abnormal were normal, what difference would it make to the semanticsof names?The causal chain would not run rightto an object, but it would still be linked to the object describedby the description. This is a good point,and it brings us to our finalremark in thissection. (4) Perhaps the mostsignificant aspect of the distinction we have made betweend-descriptions and a-descriptions, and hence between names and a-names,is its importantbearing on the semanticsof propositional-attitude and modal contexts.But this we must leave to anothertime. We have mentionedearlier (in section5) that to say more on the natureof Donnellan's distinction we need an analysisof belief contexts. 'We see now that we need this also to appreciate its full significance. 7. Demonstratives. We next consider demonstratives and personal pronouns(briefly, "demonstratives"). When a demonstrative is used "out of the blue" to designatean object, it is clear that thereis some causal link betweenthe speaker
25 A more difficult case is the following.Suppose we detect unexplained irregularities in the movementof the planets. This leads us to conclude that there is a planet of a certain mass (M) in a certain orbit (0) outside the range of our telescopes. One of us says: "I name the planet of mass M in orbit 0 'Vulcan'." On our account the descriptionis an a-description, and 'Vulcan' an a-name. Yet 'Vulcan' would function much like a normal name. The justificationl for our procedure is alluded to in par. (4) below.

This content downloaded from 75.97.192.97 on Fri, 14 Feb 2014 20:39:06 PM All use subject to JSTOR Terms and Conditions

SINGULAR TERMS

197

and the object in virtueof which he uses the demonstrative. He is perceivingthe object,or has just perceivedit. It is the causal action of the object on him that led him (in part) to do what he did. Because of thiswe can trulysay thathe had thatobject in mind in using the demonstrative. Thus, at our earlier naming ceremony, my wifehad Nana in mind in using 'her' in that it was the causal action of Nana on mywifethat led her to use the pronoun.26 Commonly,the "out of the blue" demonstrative will be accompanied by a pointinggestureof hand or eye towardthe object. On its own this gesture would very often be insufficient to identify the designatum.What determines that one aspect and not another of the vaguely indicated environmentis designated is that the speaker had that aspect in mind. We look to what caused the behavior in order to remove ambiguities.27 Sometimes, no gesture is called for-it is not with 'I' 28-and, other times,none is given. Again we look to tlle cause of the utteranceto determine reference. Suppose that the object pointed to and the object in mind are Which is designated? different. This is similarto the question raised 'our neighbor'scat' in section6 and to the question that concerning will come up in section 10 for various names. I urge a similar
answer.29

is not used "out of the blue," the speaker When a demonstrative may not have an object in mind. And a demonstrative is oftennot so used: it "may depend for its referenceupon determinants in antecedent In such a case the verbiage";30 It is a way to cross-refer.31 from the singular term on demonstrative borrows characteristics which it depends. If that singular termis causally linked to an object so that the speaker had that object in mind, then so also is If not, not. If it is dependenton an earlier "out the demonstrative. of the blue" demonstrative, or a name, or a d-description, then it will be so linked. If, on the other hand, it is dependent on an a26 It can be shown (after a discussion of belief-contexts) that this causal link between demonstrativeand object is the basic link on which the links between name and object, and description and object, ultimately depend. 27 Note that, at this point, designation is related to nonlinguistic behavior. 28 The brief general account we are giving here degenerates somewhat with 'I', but nevertheless is still applicable. A full account of "out of the blue" demonstratives would have to take note of the differences among demonstratives,and of the role that an accompanying general term can play. The account here is much simplified. 29 Discussions with Ross Poole have helped me in writing these last two paragraphs. 30,W. V. Quine, Word and Object (Cambridge, Mass.: MIT Press, 1960), p. 113. 31 Definite descriptions (e.g., 'the man') can also be used to cross-refer. The remarksin this paragraph apply to such descriptionsalso.

This content downloaded from 75.97.192.97 on Fri, 14 Feb 2014 20:39:06 PM All use subject to JSTOR Terms and Conditions

i98

THE JOURNAL OF PHILOSOPHY

singularterm,then it will description, an a-name,or an indefinite description and the demonstranot. For example, both the definite of the following tive are causally linked to an object in the first sentences, but neitherare in the second: hours. becauseshe hasn'teatenforseveral Our cat is hungry noris it a whale. in thesea is nota shark The heaviest fish I call a demonstrative where the speakerhas an object in mind a "designational demonstrativetoken" (briefly,"d-demonstrative") and one where the speaker does not have an object in mind an Ac"a-demonstrative"). token" (briefly, "attributivedemonstrative and d-demoncording to my usage, then, both a-demonstratives but only d-demonstratives may designate. stratives may refer, forname-tokens, and d-descriptions, I shall use the term'd-term' 'a-term'. Analogously, d-demonstratives. to propernames. We mustnow return
III

8. Further Involvements ofan Object. Nana is involvedin the causal forher name at morepoints than its beginningat her namnetwork groundedin her. is multiply ing ceremony;the network place, because of the role of other singuThis arises,in the first abilities with names. lar terms in passing on, and reinforcing, Suppose I pass on 'Nana' by means of (ii), "This is Nana," together with a pointinggesture.Nana will be both mediatelyand immediately causally involved in this passing on. She will be mediately involved via the ability I exercise in using her name, an ability She will be immediately grounded in her at the naming ceremony. she is present at the involved in that 'this' is a d-demonstrative: utterance,and her presenceleads to my use of the demonstrative. Thus, someone who gains an ability fromthis utterancewill gain one that is doubly grounded in Nana. And the situation would have been similarhad Nana been absentand had I used a d-description to pass on her name, say,by means of (iii) or (iv). Nana is always mediately involved when her name is used to designateher. However,in thesecondplace, she maybe immediately involvedin that use (even thoughno othersingulartermfeatures). Suppose Nana is presentand her presenceleads someone to desighis abilityto do this.He is in the nate her by name, thusexercising of his ability positionwherehe has her in mind quite independently led could have him to This use a demonstrative to use her name. to designateher, but instead it led him to use her name. She is

This content downloaded from 75.97.192.97 on Fri, 14 Feb 2014 20:39:06 PM All use subject to JSTOR Terms and Conditions

SINGULAR TERAIS

199

again cauisallyinvolvcd bollt mediatelyand immediatelyin that utterance. at its naming,then If an object is picked out by an a-description any tokenbased on thisnamingwill be an a-name(section6); the object is not involved in the causal networkfor that termat the netIt maybecomeinvolvedlater,however, in waysjust work'sbeginning. indicated. If it does, then the networkbecomes grounded in the object: a-namesare replacedbynames. 9. Other Waysof Naming. Many names are acquired not at a naming ceremony but throughuse. Nicknames,in particular,are combut ratherare used, seem apt, monly not bestowed ceremonially, and hence catch on. Other names may be similarlyacquired. A previouslyunnamed animal or place may be called by a certain name on some occasion and the name catch on. In criminal and underground political circles, people often adopt new names. Authorsoftenadopt pseudonyms. A name of this kind may have to be used several times for an object beforewe would accord it the statusof being itsname (or one of its names). But thisis not important. Eaclhof theseuses,even the first, designates theobject. How can thisbe? In virtueof what does such a first use designate the object? Our answeris along familiarlines. The speakerhad the object in mind. He had it in mind in virtueof a causal connection. This connectionmight have led him to use a certain description had he been searchingfora description to designateit, or a certain had he been searchingfor a demonstrative, demonstrative but did lead him to use a certainname when he was sear-ching for an apt name for it. Part of what he intended was to bestow the name (provisionally, perhaps) on the object. are Some names or use. When a acquired withouteitherceremony it automaticallytakes its parents' last child is born in our society, name; those presenthave the name already available to label it. The causal networkstartsat the birth,with these face-to-face conbetween the child and the first frontations users of its name. Many have claimed that we cannot referto "futureobjects" by Our theory accordswell with such claims: causes must prename.32 cede effects; so the naming ceremonyinvolving the object must precede the causal networkthat it gives rise to. These claims,however,seem unduly rigid. There seem to be occasions where we do
32 See, e.g., Richard M. Gale and Irving Thalberg. "The Generalityof Predictions," this JOURNAL, LXII, 8 (Apr. 15, 1965): 195-210; references are to be found there to earlier writingsby Peirce, Ryle, Mayo, and Ayer.

This content downloaded from 75.97.192.97 on Fri, 14 Feb 2014 20:39:06 PM All use subject to JSTOR Terms and Conditions

200

THE JOURNAL OF PHILOSOPHY

use the names of a "futureobject," particularly where thereexists a plan or blueprintfor the object. We need not go along with the claims. We can allow such names although, in our terminology, theyare a-names. 10. Mistakes and Other Failures. The picture presentedso far has been briefly as follows.When a name occurs in a statement, there is underlyingthat occurrencea causal networkgrounded in an object. In virtueof this the name token designatesthe object. This is an idealized picture. Many thingscan go wrong,and typically some of them will have gone wrong. More than one object will usuallybe appropriately linked to a name token. First, twoobjectscan be involvedat thegainingor reinforcing ofan "ability," one mediatelyand one immediately. This is apparent in light of our discussionof how an ability can be doubly grounded in an object (section 8). Suppose for example my statement(ii), "This is Nana," is false: it is actuallyJemima.(I am mistaken, or perhaps I am lying.) Any "ability" gained as a result will be groundedin Nana via my use of her name, and in Jemimavia the demonstrative. Would later uses of 'Nana', arisingfromthis,designate Nana or Jemima,neither,or both? Or suppose that the object immediately involved,in the use of 'Nana', as a result of her presence,is not Nana (as we supposed in section 8), but Jemima. In a sense, the speaker has both cats in mind in using the name. What is designatedby that use and hence will be designated by someonewho uses thename as a resultof it?33 The examples become harder when we consider the role of definite descriptions. Suppose thatJemimais the object correctly described as "our cat" and yet I state, "Nana is our cat." Here we must distinguishmistakesand lies.34First,suppose I am lying: I have Jemimain mind in using 'our cat', knowingquite well that Nana is not our cat. Once again, any "ability" with 'Nana' gained fromthiswill be causallygroundedin both cats. Secondly,suppose I am mistaken:I have Nana in mind in using 'our cat', but wrongly thinkthis description is trueof her. We have alreadydiscussedthe difficulties of such a description(in section 6). We need add here
objectsinvolved:theremay be one or even none. (b) The speakermightbe withthe resultthathis lie "cancelsout" his mistake both lyingand mistaken, von Thun). (a possibility suggested by Manfred 84 I sparethereader descriptions of plausiblesettings forthemistakes and lies we are herecontemplating.
(iv) that would normally pass on a name may be false and yet there not be two
83 I

mention some further complications in passing.(a) A sentence like (i) to

This content downloaded from 75.97.192.97 on Fri, 14 Feb 2014 20:39:06 PM All use subject to JSTOR Terms and Conditions

SINGULAR TERMS

201

only that, although Jemima will be involved in some way in an ability arisingout of this mistake,only Nana will be causally involvedin theappropriateway. We have seen so far how two objects can be involved in the one causal network at thegainingor reinforcing of an "ability."Second, more than one ability,and hence more than one network, mayhave an immediaterole in the productionof a name token. When this happens,we have slips of mind or tongue,cases of "crossedwires."A classical example of this sortof occurrencewas supplied by Canon William Archibald Spooner.35Canon Spooner once delivered a sermonthat included many uses of 'Aristotle'.He was leaving the pulpit at the end when suddenly he stopped, returned,and announced to the congregation, "When in mysermonI said 'Aristotle' I meant St. Paul." We are inclined to say that Spooner had St. Paul in mind but designatedAristotle.Two abilities had a role in the production of the tokensof 'Aristotle': the St.-Paul-ability set the mechanismsin motion, but the Aristotle-ability intervenedin theprocess, substituting its token. Third, and finally,we note that misunderstandings can lead to the involvementof more than one object in a causal network. In section4 we pointed out that,on hearinga name, we must associate it with an ability(or forma new one) to understandit. We can do thiswrongly and hence misunderstand. Consider the followingsituation.Joe has a numberof politically well-informed friendswho frequently discuss the history of socialism. They oftenuse the name 'Liebknecht',sometimesto referto Wilhelm, the father,and sometimesto Karl, the son. Joe, who knowslittleof politics,finds himself on the edge of thesediscussions and takes all these uses of the name to be about the one person. Later he uses it in a statement. Does he designateWilhelmor Karl? Joe is not only confused himself, he spreadsconfusion. He spreads it mostobviouslyto thosehe passes the name on to. But he spreads it also among thosewho alreadyhave both uses of the name. Whichever way theyinterpret Joe's remarks(unless theyare aware of his sorrystate) theywill be tainted by his confusion.Instead of reinan abilityby establishing forcing new causal linkagesto one of the objects,each such remarkdamages the abilityby bringingboth objectsinto one network. In section 4 we qualified our remarkson having an object in mind in using a name and on designating an object by name. The
35 Graham Nerlich reminded meofthis.

This content downloaded from 75.97.192.97 on Fri, 14 Feb 2014 20:39:06 PM All use subject to JSTOR Terms and Conditions

202

THE JOURNAL OF PHILOSOPHY

threetypesof failurewe have just discussedamply demonstrate the need forthesequalifications.36 What more can we say now? We shall say no more of having an object in mind. This notion was but a steppingstone to our account of designation(by name), which is forus that relation betweenname and object which has a crucial bearingon truth. What mustwe conclude about designation and truthas a resultof thisdiscussion? I shall make use here of an idea of HartryField's.37 He has argued thatmanyscientific terms are refererntially His main indeterminate. example is the Newtonian term 'mass'. He claims (in the light of the special theory of relativity) that thereis no matterof factabout whetherthis termdenoted "relativistic mass" (= total energy/c2) or "proper mass" (= nonkineticenergy/c2). The term,however,was not denotationless.Rather, it "partiallydenoted" both relativistic mass and proper mass. With the aid of this new semantic notion Field is able to give a truthdefinition yeldingdesired truthvalues for Newton's utterances.For example, the following Newtonian assertion comesout true: To acceleratea body uniformly betweenany pair of different velocities, moreforce is required if the massof the bodyis greater. Otherscome out neithertruenor false. A fairlystraightforward development of this approach enables us to assign appropriate"degreesof truth"to these truth-valueless sentences.38 For example, instead of saying merely that 'a' partiallydenotes b and partiallydenotes c, we say that it denotes b to degreep and c to degreeq. We can thenexplain the degreeof truth of the sentence containing 'a' in terms (partly) of these degrees of denotation.39 What the failureswe have discussedin this section show is that a name tokenmay "partiallydesignate"more than one object: to a certaindegreeit designatesb, to anotherdegree,c. With the help of this we can give a truth definitionyielding intuitivelydesirable
86 Further reasons for qualification,at least if we are tryingto capture all our ordinary intuitions on the matter, are to be found in Kripke, op. cit., particularly p. 301. Kripke gives some examples that suggest elements of truth in sense-theories.And we could give others. These examples strike me as peripheral. It is appropriate that we should revise our intuitions about them in the light of the causal theory. 37 "Theory Change and the Indeterminacyof Reference," this JOURNAL, LXX, 14 (Aug. 16, 1973): 462-481. 38The development is also due to Field, but does not appear in the article cited. 39 Cf. the standard referential semantics, which explains truthin terms(partly) of denotation; see Field, "Tarski's Theory of Truth," this JOURNAL, LXIX, 13 (July13, 1972): 347-375.

This content downloaded from 75.97.192.97 on Fri, 14 Feb 2014 20:39:06 PM All use subject to JSTOR Terms and Conditions

SINGULAR TERMS

203

truthvalues. Thus, suppose Joe half-designates Wilhelm and halfdesignatesKarl by 'Liebknecht'. The followingsentencesof Joe's:
Liebknechtwas a socialist. Liebknechtwas a Swiss. Liebknechtwas the proto-martyr of GermanCommunism.

come out true,false,and half-true, In other cases, the respectively. assigning of degreesof designation will not be so simple.In most,of course,therewill be arbitrariness in the decision.What is important to our theory is thatwe look to the causal explanation of the token in question in makingour decision. Many sentencesthat we would before have deemed true may come out not fully true, but, say, 90 per cent true. But this need not disturb us. If Field is right, partial truth is all we can hope forfrommuch of our mostcherished science. 11. Earlier InsightsCaptured. We shall conclude by notingvarious insightsof earlier theories of proper names which are fully or partlycapturedby our theory. First,we must agree with Mill: a proper name designatesbut it does not "implyany attributes." 40 Both Russell and the earlyWittgenstein held an almost mystical view of the relationship betweena name and its bearer.It is a relationshipof theutmostintimacy:the natureof a name is such thatit immediately and directly focusesattention on the object (and that's all it does). We have seen thatnames are, in a quite clear sense,immediate pointers to their objects. But the relationshipis not the least bit ineffable; it is a matterforscientific investigation. Sense-theories make use of a certainfact: a man's abilityto use a name is accompanied by various beliefs about its bearer which have arisen largely from the historyof his experienceswith the name. For us, each of theseexperiences will forma link in the causal networkunderlying his presentuse; the networkembodies the history.And, of course,the name user's beliefsabout the object will in part reflectthat history.We differ from sense-theorists in not making the connectionsbetween name and object depend on the truth of thosebeliefs. Some philosophershave indicated the importanceof reference borrowingto the use of names.41 We give it a centralrole. But we do not requirethatname userskeep trackof theirborrowings. Identitystatements pose a difficulty for a theoryof names. The
40 A Systemof Logic (London: Longmans, eighth edition revised, new impression 1961), p. 20. 41 E.g., Strawson, op. cit., p. 182n.

This content downloaded from 75.97.192.97 on Fri, 14 Feb 2014 20:39:06 PM All use subject to JSTOR Terms and Conditions

204

THE JOURNAL OF PHILOSOPHY

difficulty arises from what Frege called "the differing cognitive values" of 'a = a' and 'a = b'.42 Awareness of this difficulty has pushed some philosophersaway fromthe natural view that identity is what it seems to be, a relation betweenobjects, toward the view thatit is a relationbetweensigns(e.g.,ibid.,p. 10). Objectionsto the latterview led Frege to his sensetheory of names and led others 48 to even moredesperateexpedients. betweenthe two statements that has One aspect of the difference been a source of worryis as follows: whereas 'a = b' seems to be and "empirical,"'a = a' seemsto be "ana"synthetic," "contingent," lytic,""necessary," and "a priori." This aspect need not worryus. We must sharply distinguishthe "metaphysical"term 'necessary' fromthe "epistemological"term 'a priori', as Kripke has pointed out (260-263). If we do, it is not hard to see that 'a = b', if true, is necessarilytrue (for names), even though it may be known empirically (305-311). values" of However,we muststill explain the "differing cognitive the two statements. This we can easily do. Frege rightly saw that the solution to the difficulty lay in the different "modes of presentation" of the object associatedwith 'a' and 'b' (op. cit.,p. 57). Frege's mistake was to embody these modes within "senses." For us the modes are the causal networksunderlyingthe names. There is nothing more to the "meaning" of names than these networks. 'a' will be a verydifferent fromthatunderlying network Underlying 'b'.44 Thus the "cognitivevalue" of 'a = a' will be very different fromthatof 'a = b'. We began this paper with the claim that the main problem in of propernamesis thatof explainingthenature givingthe semantics of the link betweenname and object in virtueof which the former designatesthe latter.In consideringthis problem,many have seen an object, the that the user of a name must,in some way, identify amount to? object he "has in mind." What does this identification The receivedanswerhas been that it is the speaker'sabilityto pro42 Philosophical Writingsof Gottlob Frege, Peter Geach and Max Black, eds. (Oxford: Blackwell, second edition corrected,1960), p. 56. 43 E.g., D. S. Shwayder,"'= '," Mind, LXV, 257 (January 1956): 16-37; David in R. J. Butler, ed., Analytical Philosophy: Wiggins, "Identity-Statements," Second Series (Oxford: Blackwell, 1965), pp. 40-71; Leonard Linsky, "Reference and Referents," in Charles E. Caton, ed., Philosophy and Ordinary Language (Urbana: Illinois, 1963), pp. 74-89. 44 This will be so even if the speaker associates the same descriptionswith the names. Hence the absurdity of John Searle's claim that 'Tully = Cicero' is twvo "analytic" for most of us; see "Proper Names," Mind, LXVII, 266 (April 1958): 166-173, reprintedin Caton, op. cit., pp. 154-161.

This content downloaded from 75.97.192.97 on Fri, 14 Feb 2014 20:39:06 PM All use subject to JSTOR Terms and Conditions

DEMONSTRATIVE

CONSTRUCTIONS,

REFERENCE,

AND TRUTH

205

descriptionof the object. This is mistaken.41 duce an identifying depends not but identification The speaker is indeed important, he could or would do but on what he did: for underon anything lying what he did was a causal networkgrounded in an object. Only in this way does a speaker identifyan object. Causal networkslink names to theworld.
MICHAEL DEVIIT

of Sydney University DEMONSTRATIVE CONSTRUCTIONS, AND TRUTH * REFERENCE,

and acquisition of ment play a lead role in the employment LJ natural languages. Such sentences appear more than any other kind in day-to-day communication.They occur repeatedly in ordinaryempirical thinking.And they are the firstsentences to be taught to a first-language learner (or radical translator).Inand knowledge, deed, given the limitsof our intelligence, memory, it is doubtfulthatwithoutthemwe could learn language or utilize it to describeparticularobjects,events,or experiences.The reason for this virtual omnipresenceof sentencesinvolving a demonstrative elementis thattheyare peculiarlydependentfortheirinterpretation on the context of their use. Their truth value typically or thinkingthem depends on someone'sactuallyspeaking,writing, dividend forlanguage learning in a relevantcontext.The resulting is obviousness: since the correctinterpretation of these sentences occurs,as it were, on the spot, many of them can be taught with relativeease to the novice-one has merelyto correlatethemwith the appropriatespots.The dividend for communication and thinking is economy:a gesturesaves a thousandwords. My aim in this paper is to motivateand sketcha unifiedformal theoryof some of the most ordinarykinds of sentencesinvolvinga will emphasizea point that has demonstrative element.The theory seriousnessin recent formal acnot been treated with sufficient counts.The point is that ordinarysentencescontainingdemonstra45 The move from a correct view to a mistaken one is nicely illustrated by Searle's move from his "axiom of identification"to his "principle of identification"; see his Speech Acts: An Essay in the Philosophy of Language (New York: Cambridge, 1969), pp. 77-88. * I am indebted to Gilbert Harman, David Kaplan, Dana Scott, and John Wallace for criticismof earlier versionsand to Harry Deutsch for many helpful and stimulatingconversations.

ENTENCES whosenormal a demonstrative eleuse involves

This content downloaded from 75.97.192.97 on Fri, 14 Feb 2014 20:39:06 PM All use subject to JSTOR Terms and Conditions