Kaplan, D. 1989, Demonstratives

8/12/2019 Kaplan, D. 1989, Demonstratives

1/42

17DemonstrativesAn Essay on the Semantics, Logic,Metaphysics, and Epistemology o fDemonstratives and OtherIndexicalsDavid Kap lan l


2/42

T a b l e o f C o n t e n t sPreface

I. I n t r o d u c t i o nH. D e m o n s t r a t i v e s , Indexicals, and Pure IndexicalsTwo Obvious PrinciplesI V. R e m a r k s on R ig id Desi gnatorsV. A r g u m e n t fo r P ri nc ip le 2 : P ur e Indexi cal sVI. T e r m i n o l o g i c a l R emarksVI. ( i ) C o n t e n t and Ci rcumstanceV I. ( i i ) Char ac terV II . E a r l i e r A tt em pt s: I ndex Theor yV II I. M o n s t e r s Begat by EleganceIX. A r g u m e n t f or Pr inciple 2: True Demonstrat ivesIX. ( i ) T h e A rgumentsIX. (i i) T h e Fregean Theory of DemonstrationsIX. (i i i) T h e Fregean Theory of DemonstrativesIX. (iv) Argument Against the Fregean Theory ofDemonstrativesX. F i x i n g the Reference vs. Suppl ying a SynonymX. ( i ) R ei c he n ba c h on Token ReflexivesXI. T h e Mean ing o f IndexicalsXII. D t h a tX II I. C o n t e x t s , Tru th, and Log ical Tru thXIV. S u m m a r y of Findings (so far ); Pure IndexicalsXV. F u r t h e r Detai ls : Demonst ra ti ves andDemonstrationsXVI . A l t e r n a t i v e Treatments of Demonstrat ionsXVI I, Ep i s te mol og i ca l RemarksXVII I . T h e Formal SystemXIX . R e m a r k s on the Formal SystemXX. A d d i n g 'Says'XXI. R u s s e l l on Egocentr ic Part iculars and TheirDispensabilityXXI I, O n P roper Names

482

483489489492492498500500505507510512513514516516518519520521522523524527529541546553557558

Demonstratives 48 3

P r e f ac eIn about 1966 I wrote a paper about quantification into epistemologicalcontexts. Th er e are very difficu lt metaphysical, logical, a nd epistemo-logical problems involved in providing a treatment of such idioms whichdoes not distort our intuitions about their proper use and which is upto contemporary logical standards. I did n ot then, and d o not now,regard the tre atment I provided as fu lly adequate. An d I became moreand more intrigued with problems centering on what I would like to callthe semantics o f direct reference. B y this I mean theories o f meaningaccording to which certain singular terms refer directly without the me-diation of a Fregean Sinn as meaning. I f there are such terms, t hen theproposition expressed by a sentence containing such a term would involveindividuals directly rather than by way of the individ ual conepts ormanners of presentation I had been taught to expect. Le t us call suchputative singular terms ( if there are any) directly referential terms andsuch putat ive propositions (if there are any) singula r propositions. Eveni f English contained no singular terms whose proper semantics was oneof direct reference, could we determine to introduce such terms? A n deven if we had no dir ectly referential terms and introduced none, is therea need or use for singular propositions?

The feverish development o f quantified modal logics, more generally,of quantified intensional logics, of the 1960s gave rise t o a metaphysicaland epistemological malaise regarding the problem of i dentifying indiv id-uals across worldswhat, in 1967, I called the problem of Trans -Wor ldHeir Lines. Thi s problem was really just the problem of singular propo-sitions: those which involve individuals directly, rearing its irrepressiblehead in t he possible-world semantics tha t were then (a nd are now) sopopular.

It was no t t hat according to those semantical theories any sentencesof the languages being studied were themselves taken to express sin -gular propositions, i t was jus t t ha t singular propositions seemed to beneeded in the analysis of the nonsingular propositions expressed by thesesentences. For example, consider(0) 9 s ( F x A - O F ) .This sentence would not be taken by anyone to express a singular propo-sition., But in order t o evaluate the truth-valu e o f the component

DF


3/42

484 Da v id Kaplan

(under some assignment of an i ndivid ual to the variable 'x') , we mustfirst determine whether the proposition expressed by i ts componentF r

while Frege's solution: t ha t a, a n d [ 3 ,t h o u g hr e f e r r i ngt ot h es a met h i ng ,

do so by way o f different senses, would be blocked. Also : because o f thefact that the component of the proposition is being determined by theindividual rather than vice versa, we have something like a violation ofthe famous Fregean dictum that there i s no road back from denotationto sense [propositional component]. (Recently, I have come to think thatif we countenance singular propositions, a collapse of Frege's intensionalontology into Russell's takes place.)I can draw some little pictures to give you an idea of the two kindsof semantical theories I want t o contrast.

(under an assignment of an indi vidu al to the variable 'x' ) is a necessaryproposition. So in the course of analyzing (0), we are required to deter-mine the proposition associated w ith a formul a containing a free vari-able. N ow free variables under an assignment o f values are paradigmsof what I have been calling directly referential terms. I n determininga semantical value for a formula containing a free variable we may begiven a value for the variablethat is, an individual drawn from theuniverse over which the variable is taken to rangebut nothing more.A variable's first and only meaning is its value. Therefore, if we are toassociate a proposition (not merely a truth-value) with a formula con-taining a free variable (with respect to an assignment of a value to thevariable), t hat proposition seems bound to be singular (even i f valiantattempts are made to disguise this fact by using constant functions toimitate i ndividual concepts). Th e point is, that if the component of theproposition (or the step i n the construction o f the proposition) whichcorresvonds t o the singular te rm is determined by the in divid ual andthe indi vidual is directly determined by the singular termrather thanthe indiv idua l being determined by the component of the proposition,which is directl y determined by the singular te rmthen we have what Icall a singular proposition . [Russell's semantics was like the semanticaltheories for quantified intensional logics that I have described in tha talthough no (closed) sentence of Principia Mathernatica was taken tostand for a singular proposition, singular propositions are the essentialbuilding blocks of all propositions.]The most important hold-out against semantical theories that re-quired singular propositions is Alonzo Church, the great modern cham-pion o f Frege's semantical theories. Chur ch also advocates a version o fquantified intensional logic, but with a subtle difference that finesses theneed for singular propositions. (I n Church's logic, given a sentential for-mula containing free variables and given an assignment of values to thevariables, no proposition is yet determined. A n additional assignmentof senses to the free variables must be made before a proposition canbe associated wi th the formula.) I t is no accident th at Church rejectsdirect reference semantical theories. F o r if there were singular termswhich referred directl y, i t seems likely tha t Frege's problem: h ow can

= , if t rue, differ in meaning from ra = c e ' , c o u l db e

c.I-c,,,,.... ,e, ,.cv -,.. 'q...',,,,Q-c,1,:.N1

Is,e'dcZS'.-'4-

c, e.0 4zze' . t , t ' 'a c e c,(S' 4 0,- .

C -c to'C

C 0 4 ( 3 _ p$ C,0.0reP

N:.00'er

,,,,, 4. +,eS N>o ' ,',.5.

Fregean Picture

LANGUAGE(singular term) d e n o t e s(This relation is definedas the product o f the othertwo relations)

Demonstratives 4 8 5

PROPOSITIONAL COMPONENTSense ( a concept, something

like a description in purelyop q u a l i t a t i v e language)2 (This relation is, in general,empirical; the individual who fallsunder the concept, i.e., who, uniquely,' ' has the qualities)

INDIVIDUAL


4/42

486 David Kaplan

Di rec t Reference P ic tur e

PROPOSITIONAL COMPONENT

LANGUAGE(singular term)(This relation is determinedby the conventions or rulesof the language)

refers INDIVIDUAL

(These pictures are not entire ly accurate fo r several reasons, amongthem; that the contrasting pictures are meant to account for more thanjust singular terms and t hat the relation marked 'refers'involve a kind of Fregean sense used t o fix the referent.)I wo n't go into the pros and cons of these two views at th is time,Suffice it to say that I had been raised on Pregean semantics and wassufficiently devout t o wonder whether th e ki nd of quantification in tomodal and epistemic contexts that seemed to require singular proposi-tions really made sense. ( M y paper "Quan tifyi ng I n" can be regardedas an attempt to explain away such idioms for epistemic contexts.)2 But there were pressures fr om quarters other than quantified in-tensional logic i n favor o f a semantics of direct reference. F i r s t ofall there was Donnellan's fascinating paper "Reference a nd DefiniteDescriptions."3 T h e nt h e rew e red i s cu s s io n sI

in 1958 in which he argued with respect to certain natural kind termslike 'tiger' and 'gold', that if their Fregean senses were the kind of thingthat one grasped when one understood the terms, then such senses could

may already

2David

Philosophy of Language, ed. A, P. Mart inic h (Oxford; Oxfo rd University Press,1985).3Kei

(1966): 281-304; reprinted in Martini ch, op. cit.


not determine the extension of the terms. An d finally Kripke's Prince-ton lectures of spring 1970, later published as Naming and Neeessity,4were just beginning to leak out along with their strong attack on theFregean theory of proper names and their support of a theory of directreference.

As I said earlier, I was intrigued by the semantics of direct refer-ence, so when I had a sabbatical leave for the year 1970-71, I decided towork i n the area in which such a theory seemed most plausible: demon-stratives. I n fall 1970, I wrote, f or a conference at Stan ford, a paper"M in t ." ' Using Donnellan's ideas as a starting point, I tried t o de-velop the contrast between Fregean semantics and the semantics o f di-rect reference, and to argue that demonstrativesalthough they couldbe treated on a Fregean modelwere more interestingly treated on adirect reference model. Ultimately I came to the conclusion that some-thing analogous to Donnellan's referential use of a definite descriptioncould be developed using my new demonstrative, "d th at ." In the courseof this paper I groped my way to a formal semantics for demonstrativesrather different in conception f rom those that had been offered before.In spring 1971,1 gave a series of lectures at Princeton on the seman-tics of direct reference. By this time I had seen a transcrip t o f Namingand Necessity and I t ried to relate some of my ideas to K ri pke' s.6 I a l s oad writte n ou t the f ormal semantics for my Logic o f Demonstratives.That summer at the Irvine Philosophy of Language Institu te I lecturedagain on t he semantics of direct reference and repeated some o f theselectures at various institutions in fall 1971. And there the matter hasstood except for a bit of updating of the 1971 Logic of Demonstrativesnotes in 1973.

I now thi nk t hat demonstratives can be treated correctly only on adirect reference model, but that my earlier lectures at Princeton andIrvine on direct reference semantics were too broad in scope, and thatthe most important and certainly the most convincing part of my theoryis just the logic of demonstratives itself. I t is based on ju st a few quite

4Saul Kripke, "N ami ng a nd Necessity," i n Semantics o f N atur al Language,ed. G. Harman and D . Davidson (Dordrecht; Raids , 1972); revised edition pub-lished as a separate monograph, Naming and Necessity (Oxford : B asi l Blackwell, 1980). References are to the revised edition.5David

Academic Press, 1978); reprinted in Martinich, op. cit.6Altfamili ar wit h Naming and Necessity, I have enthusiastically adopted the 'analyticalapparatus' and some of the terminology of that, brilliant work.


5/42

488 David Kaplan Demonstratives 48 9simple ideas, but the conceptual apparatus turns out to be surprisinglyrich and interesting. A t least I hope that you wil l find it so.In this work I have concentrated on pedagogy. Philosophically, thereis little here that goes beyond the Summer Institute Lectures, but I havetried, by limiting the scope, to present the ideas in a more compellingway. Some new material appears in the two speculative sections: X VII(Epistemological Remarks) and XX (Adding 'Says'). I t is my hope thata theory of demonstratives will give us the tools to go on in a moresure-footed way to explore the de. re propositional attitudes as well asother semantical issues.

I . In t roduct ionI believe my theory of demonstratives to be uncontrovertable and largelyuncontroversial. Th is is not a tribute to the power of my theory bu t aconcession of its obviousness. In the past, no one seems to have followedthese obvious facts out to their obvious consequences. I do that. W ha tis original with me is some terminology to help fix ideas when things getcomplicated. I t has been fascinating to see how interesting the obviousconsequences of obvious principles can be.'

Demonstrat ives, Indexica ls, and Pur e Indexica lsI tend to describe my theory as 'a theory of demonstratives', but thatis poor usage. I t stems from the fac t that I began my investigationsby asking what is said when a speaker points at someone and says,He is suspicious. ' The word 'he ', so used, is a demonstrative, andthe accompanying pointing is the requisite associated demonstration. Ihypothesized a certain semantical theory for such demonstratives, andthen I invented a new demonstrative, 'dthat', and stipulated that itssemantics be in accord with my theory. I was so delighted wi th thismethodological sleight of hand for my demonstrative rdthat ', tha t whenI generalized the theory to apply to words like ' I' , 'now' , 'here', etc.words which do not require an associated demonstrationI continuedto call my theory a 'theory of demonstratives' and I referred to thesewords as 'demonstratives'.

That terminological practice conflicts with what I preach, and I willtry to correct it. (But I tend to backslide.)The group of words for which I propose a semantical theory includesthe pronouns 'I' , 'my', 'you', 'he ', 'his', 'she', 'i t' , t he demonstrativepronouns 'that', 'this' , the adverbs 'here', 'now', 'tomorrow', 'yesterday',the adjectives 'actual', 'present', and others. These words have usesother than those in which I am interested (or, perhaps, depending onhow you individuate words, we should say that they have homonyms inwhich I am not interested). For example, the pronouns 'he' and 'his' areused not as demonstratives but as bound variables in

7Note

the correct use of certain words and I propose detailed analyses of certain notions.I reeognice that these matters may be controversial. I do not regard them as partof the basic, obvious, theory.


6/42

490 David Kaplan

For what is a man profited, if he shall gainthe whole world, and lose his own soul?What is common to the words or usages in which I am interestedis that the referent is dependent on the context of use and that themeaning of the word provides a rule which determines the referent interms of certain aspects o f the context. T he term I no w favor for these

words is 'indexical' . Oth er authors have used other terms; Russell used'egocentric parti cula r' and Reichenbach used 'toke n reflexive'. I prefer'indexical' (which, I believe, is due to Pierce) because it seems less theoryladen than the others, and because I regard Russell's and Reichenbach'stheories as defective. ,

Some of the indexicals require, in order to determine their referents,an associated demonstration: typically, though not invariably, a (visual)presentation of a local object discriminated by a point ing.9 T h e s e i n -exicals are the tr ue demonstratives, and 'th at' is their paradigm. T hedemonstrative (an expression) refers to that which the demonsirationdemonstrates. I call tha t which is demonstrated the 'demonstr atum'.A demonstrative without an associated demonstration is incomplete.The linguistic rules which govern th e use of the t rue demonstratives'tha t', 'he', etc., are not sufficient t o determine their referent in all con-texts of use. Something elsean associated demonstrationmust beprovided, T h e linguistic rules assume that such a demonstration ac-companies each (demonstrati ve) use of a demonstrative. A n incompletedemonstrative is no t vacuous like an improper defin ite description. Ademonstrative can be vacuous in various cases. For example, when itsassociated demonstration has no derrionstratum (a hallucination)orthe wrong kind of demonstratum (pointing to a flower and saying 'he'in the belief that one is pointing to a man disguised as a f l o w e r1 ) o rtoo many demonstrata (pointing to t wo intertwined vines and saying

9However,ade

the speaker's part, as when someone shouts Stop t hat man while on ly one manis rushing toward the door. M y noti on of a demonstration is a theoretical concept.I do not, in t he present work, undertake a detailed 'operational' analysis of thisnotion although there are scattered remarks relevant to the issue. I do consider,in XVI below, some alternative theoretical treatments of demonstrations.101amawa

be appropriate for flowers, but it is not so in English; (2) t ha t a background storycan be provided that will make pointing at the flower a contextually appropriate,though deviant, way of referring to a man; for example, if we are talking of greathybridizers; and (3) th at it is possible to t reat t he example as a referential useof the demonstrative 'he' on the model o f DonneIlan's referential use of a definitedescription (See Reference and Definite Descriptions ). Unde r the referential use


'that vine'). B u t it is clear that one can distinguish a demonstrativewith a vacuous demonstration: no referents from a demonstrativ e w ithno associated demonstration : incomplete.All this is by way of contrasting true demonstratives with pure index-

icals. Fo r the latt er, no associated demonstration is required, and anydemonstration supplied is either for emphasis or is i rr et ev an t.11 A m o n ghe pure indexicals are '1', 'now ', 'here' (in one sense), 'to morr ow', andothers. T h e linguistic rules which govern Their use full y determine th ereferent for each con te xt. No supplementary actions or intentions areneeded. Th e speaker refers to himself when he uses '1', and n o point ingto another or believing that he is another or intending to refer to anothercan defeat this reference.

Michael Bennett has noted that some indexicals have both a pureand a demonstrative use, 'Here' is a pure indexical inI am in here

and is a demonstrative inIn two weeks, I will be here [point ing at a cit y on a map].

treatment we would assign as referent for ' he' whatever the speaker intended t odemonstrate, I intended the example to exemplify a failed demonstrat ion, thus,a case in which the speaker, falsely believing the flower to be some man or otherin disguise, but having no particular man in mind, and certainly not intending torefer to anything other than tha t man, says, pointing at the flower, He has beenfollowing me around all day.ll I have in mind such cases as pointin g at oneself while saying ' I' (emphasis) o r

pointing at someone else while saying 'I' (irrelevance or madness or what?).'The re are certain uses of pure indexicals that might be called 'messages recordedfor later broadcast', which exhib it a special uncertainty as to the referent of 'here'and 'now'. I f the message: '' I am no t here now is recorded on a telephoneanswering device, i t is t o be assumed tha t t he tim e referred to b y 'no w' is t hetime of playback rather than the time of recording. Donnellan has suggested tha tif there were typically a significant lag between our production of speech and itsaudition (for example, i f sound traveled very very slowly), our language mightcontain two forms of 'now': one for the time of production, another for the time ofaudition. The indexicals 'here' and 'now' also suffer from vagueness regarding thesize of the spatial and temporal neighborhoods to which they refer. These facts donot seem to me to slur the difference between demonstratives and pure indexicals.130r course it is certain intentions on the p art of the speaker that make a partic-ular vocable the first person singular pronoun rather a nickname for Irving. Mysemantical theory is a theory of word meaning, not speaker's meaning. It is basedon linguistic rules known, explicit ly or implicitly, by all competent users of thelanguage.


7/42

492 David Kaplan

Two Obvious Pr inc ip lesSo much for preliminaries. M y theory is based on two obvious principles.The first has been noted in every discussion of the subject.Principle 3. The referent of a pure indexical depends on the context,and the refe rent of a demonstrati ve depends on the associated dem-

onstration.If you and I both say 'I' we refer to different persons. The demonstratives't ha t' and 'he' can be correctly used to refer to any one of a wide varietyof objects simply by adjusting the accompanying demonstration.

The second obvious princ iple has less often been formul ated exp lic-itly.Principle 2 Indexicals, pure and demonstrative alike, are directly ref-erential.

IV. Remarks on Rigid Des ignatorsIn an-e a r li e rd ra ftIadoptedtheterm

'rigid designators', and tried to explain that my usage differed from his,I am now shying away from t hat terminology, Bu t because it is so wellknown, I will make some comments on the notion or notions involved.The term 'rigid designator' was coined by Saul Kripke to character-

ize those expressions wh ich designate the same th in g in every possibleworld in which th at thin g exists and which designate nothing elsewhere.He uses it in connection wi th his controversial, though, I believe, cor-rect cl aim that proper names, as well as many common nouns, are ri giddesignators. There is an unfort unate confusion in the idea tha t a propername would designate nothing if the bearer of the name were not to

Kripke him self adopts positions which seem inconsistent w iththis feature of rigid designators. I n arguing that the object designatedby a rigid designator need not exist in every possible world, he seemsto assert that under certain circumstances what is expressed by 'Hitlerdoes not exist' would have been true, and not because 'H it le r' would havedesignated nothing (in that case we might have given the sentence notruth-value) b ut because wh at 'Hi tle r' would have designatednamely141ha

Approaches to N atur al Language, ed. J. H inti kka et al. (Dordrech t: Reidel, 1973),especially appendix X.

Demonstratives 49 3

Hitlerwould not have existed.15 F u r t h e r m o r e ,i ti sas t r i k i nga n di m -

port ant feature of the possible world semantics for quantified intensionallogics, which Kripke did so much to create and popularize, that variables,those paradigms of rigid designation, designate the same individual inall possible worlds whether the individual exists or n o t .1 6 1 / 1

1h a te ve rKr ipke'sintentions(didhe,asIsus

own concept?) a n d whatever associations or even meaning the phrase'rigid designator' may have, I intend t o use 'directly refe rential' for anexpression whose referent, once determined, is taken as fixed for all pos-sible circumstances, i.e., is taken as being the proposit ional component.For me, the intuitive idea is not that of an expression which turnsout to designate the same object in all possible circumstances, but anexpression whose semantical rules provide directly that the referent inall possible circumstances is fixed to be the actual referent. I n typicalcases the semantical rules wi ll do this only impl icitl y, by prov idin g a wayof determining the actual referent and no way of determining any otherpropositional comp on ent .17 We should beware of a certain confusion in interpreting the phrase'designates the same object i n al l circumstances'. We do n ot mean t hatthe expression could no h ave been used t o designate a different obje ct.1 . 5K r ip ke ,NamingandNecessit

' T h e matter is even more complicated, There are two 'definitions' o f 'rig id desig-nator' in Naming and Necessity, pp. 48-49. The first conforms to what seems tome to have been the in tende d conceptsame designation in all possible worldsthe second, scarcely a page later, conforms to the more widely held view that arigid designator need not designate the object, or any object, at worlds in whichthe object does not exist. Accordi ng to t his conception a designator cannot, ata given world, designate something which does not exist in th at world. T he in-troduction of the notion of a strongly rig id designatora rigid designator whosedesignaturn exists in all possible worldssuggests tha t the lat ter idea was upper-most in Kripke's mind. (T he second definition is given, unequivocally, on page146 of Ident ity and Necessity, in Ide ntity and Individuation, ed. M, K. Mu nitz(New York: New Yor k Universi ty Press, 1971).) I n spite o f the t ext ual evidence,systematic considerations, including the fact that variables cannot be accountedfor .o t he rw i se ,leave

Here, and in the preceding paragraph, in attempting to convey my notion of adirect ly referential singular term, I slide back and forth between two metaphysicalpictures: th at of possible worlds and tha t o f structured propositions. r t seemsto me that a truly semantical idea should presuppose neither picture, and be ex-pressible in terms of either, Kripke's discussion of rigid designators is, I believe,distorted by an excessive dependence on the possible worlds picture and the asso-ciatedsemantical style. Fo r more on the re lationship between the two pictures, seepages 724-25 of my How to Russell a Frege-Church, The J ournal of Philosophy72 (1975): 716-29.


8/42

494 David Kaplan Demonstratives 49 5We mean rather th at given a use o f the expression, we may ask of whathas been said whether it would have been true or false in various counter-factual circumstances, an d in such counterfactual circumstances, wh ichare the individuals relevant to determining truth-value. T hu s we mustdistinguish possible occasions of usewhich I call contdxtsfrom possi-ble circumstances of evaluation o f what was said on a given occasion o fuse. Possible circumstances of evaluation I call circumstances or, some-

, t i m e s , just counterfactual si tuations. A directly referential term maydesignate different objects when used i n different contexts. B u t whenevaluating what was said in a given context, onl y a single object wi ll berelevant to the evaluation in all circumstances. T h i s sharp distinctionbetween contexts o f use and circumstances o f evaluation must be keptin mind if we are to avoid a seeming conflict between Principles 1 and2 .18Tolookatthematterfromanot

the obviousness of both principles (I have not yet argued for Princip le 2)the disti nctio n between contexts o f use and circumstances of evaluationis forced upon us.

If I may wax metaphysical in order to f ix an image, let us think of thevehicles of evaluationthe what-is-said in a given contextas proposi-tions: Don't think of propositions as sets of possible worlds, but ratheras structured entities looking something like the sentences which expressthem. For each occurrence of a singular term in a sentence there will be acorresponding constituent in the pro position expressed. T he constituentof the proposition determines, for each circumstance of evaluation, theobject relevant to evaluating the proposition in tha t circumstance. I ngeneral, the constit uent o f the proposition wil l be some sort of complex,constructed fro m various attributes by logical composition. B ut in thecase of a singular ter m which is directly referential, the constituent of

.1 t h e prOposition is just the object itself. Thu s it is tha t i t does not justturn out th at the constituent determines the same object in every cir-cumstance, the constituent (corresponding to a rigid designator) just isthe object. There is no determining to do at all. On this pictureandthis is really a picture and not a theorythe definite description(1) T h e n[(Snow is slight A n2= 9 ) V ( S n o wi ss l i g h tA

2' = n + 1) ]1 9

' I think i t likely that i t was just the failure to notice this distinction that led toa failure t o recognize Principle 2. Some of the histo ry and consequences of theconflation of Context and Circumstance is discussed in section VII .' I would have used 'snow is white ', b ut I wanted a contingent clause, and so many

would yield a constituent which is complex although it would determinethe same object i n a ll circumstances, Th us , (1), though a rigid desig-nator, is not directly referential from this (metaphysical) poi nt of view.Note, however, t ha t every proposition which contains the complex ex-pressed by (1) is equivalent to some singular proposition which containsjust the number three itself as consti tuen t.2 The semantical feature that I wish to hig hlight in call ing an expres-sion directly referential is not the fact tha t it designates the same objectin every circumstance, but the way in which it designates an object inany circumstance, Su ch an expression is a device o f direct reference,This does not i mpl y t hat it has no conventionally fixed semantical ruleswhich determine its referent in each context of use; quite the opposite.There are semantical rules which determine th e referent in each con-text o f usebut tha t is all. Th e rules do not provide a complex whichtogether with a circumstance of evaluation yields an object. They justprovide an object,

If we keep in mind our sharp dist inctio n between contexts of use andcircumstances o f evaluation, we wi ll not be tempted to confuse a rulewhich assigns an object to each context wi th a ' comple x' wh ich assignsan object to each circumstance. For example, each context has an agent(loosely, a speaker). T hu s an appropriate designation rule for a directlyreferential term would be;(2) I n each possible context of use the given term refers to theagent of the context.But this rule could not be used to assign a relevant object to each cir-cumstance of evaluation. Circumstances of evaluation do not, in general,have agents. Suppose I say,(3) I do not exist.Under what circumstances would what I said be true? I t would be truein circumstances in which I did not exist. Am on g such circumstancesare those in whic h no one, and thus, n o speakers, no agents exist. Tosearch a circumstance of evaluation for a speaker in order to (mis)applyrule (2 ) would be t o go off on an irrelevant chase.

people (possibly including me) nowadays seem to have views wh ich allow th at'snow is white' may be necessary, I am ignoring propositions expressed by sentences containing episternic Operators

or others for which equivalence is no t a sufficient condition f or interchange ofperand,


9/42

496 David Kaplan

Three paragraphs ago I sketched a metaphysical picture of the struc-ture of a proposition. The picture is taken from the semantical partsof Russell's Principles of Mathemalics.21 T w o y e a r sl a t e r ,i n

O nD e -

no t ing ,22e v enR u ss e llr ej ec te dthatpicture,ButIstilllike

not a part of my theory, but it well conveys my conception of a directlyreferential expression and of the semantics of direct reference. (The pic-ture needs some modifica tion in order to avoid difficulties which Russelllater notedthough he attributed them to Frege's theory rather thanhis own earlier theory.)23 If we adopt a possible worlds semantics, all direct ly referential terms2 1Be rt ra ndRussell,ThePrinciplesofMa

2 2Be rt ra ndRussell,OnDenoting,Mind

2 3He reisadifficultyinRussell's19

Consider the proposition expressed by the sentence, 'The centre of mass of theSolar System is a point'. Ca ll the proposition, 'P'. P has in its subject place acertain complex, expressed by the definite description. Call the complex, 'Plexy'.We can describe P/exy as th e complex expressed by 't he center o f mass of thesolar system'. Can we produce a directl y referential term which designates Plexy?Leaving aside for the moment the controversial question of whether 'Plexy' is sucha term, le t us imagine, as Russell believed, th at we can di rectly refer t o Plexyby affixing a kind of meaninp marks (on the analogy of quotation marks) to thedescription itself. N ow consider the sentence ' th e center of mass of the solarsystem ' is a po int'. Because the subject of this sentence is directly referentialand refers to F lexy, the proposition the sentence expresses will have as its subjectconstituent Plexy itself. A moment's reflection will reveal tha t this proposition issimply P again. But this is absurd since the two sentences speak about radicallydifferent objects.(I believe the foregoing argument lies behind some of the large ly incomprehensi-ble arguments mounted by Russell against Frege in On Deno ting, though thereare certainly other difficulties in t hat argument. I t is n ot surprising that Russellthere confused Frege's theory w it h his own o f Principle o f Malheanatics. T h efirst footnote of On Denoting asserts that the two theories are very nearly thesame. )The solution to the difficulty is simple, Regard the 'object' places of a singularproposition as marked by some operation which cannot mark a complex. (Therealways will b e some such operation.) F or example, suppose th at no complex is(represented by) a set containing a single member. Then we need only add t omark the places in a singular proposition which correspond to directly referentialterms. We no longer need worry about confusing a complex with a propositionalconstituent corresponding to a directl y referring term because no complex wil lhave the form -(x). I n particular, Flexy { F l e x y l . Thi s technique can also beused to resolve another confusion in Russell. He argued that a sentence containinga nondenoting directly referential term (he would have called it a nondenoting'logically proper name') wou ld be meaningless, presumably because the purpo rtedsingular proposition would be incomplete. But the braces themselves can fill outthe singular proposition, and if they contain nothing, no more anomalies needresult than what, the development of Free Logic has already inured us to.

will be regarded as rigid designators in the modified sense of an expres-sion which designates the same thing in all possible worlds (irrespectiveof whether the thing exists in the possible world or n ot ). However, asalready noted, I do not regard all rigid designatorsnot even al l stronglyrigid designators (those t hat designate something th at exist s in a ll pos-sible worlds) or all rigid designators in the modified senseas directlyreferential. I believe that p roper names, like variables, are dire ctl y ref-erential. They are not, in general, strongly rigid designators nor arethey rigi d designators in the original sense.25 W h a t i sc h a r a c t e r i s t i co fdirectly referential terms is t hat the designatum (referent) determinesthe propositional component rather than the proposition al component,along wit h a circumstance, determining the designatum. I t is for thisreason that a directl y referential term t hat designates a contin gently ex-isting object w ill still be a rigid designator in the m odified sense. Th epropositional component need no t choose its designatum f rom those of-fered by a passing circumstance; i t has already secured i ts designatumbefore the encounter w it h the circumstance.

When we think in terms of possible world semantics this fundamentaldistinction becomes subliminal. Th is is because the style of the seman-tical rules obscures the distinction and makes it appear that directlyreferential terms differ fr om ordina ry definite descriptions only in thatthe propositional component in the former case must be a constant func-tion of circumstances. In actual fact, the referent, in a circumstance, ofa directly referential t erm is s imply independent o f the circumstance andis no more a function (constant or otherwise) of circumstance, than myaction is a function of your desires when I decide to do it whether youlike it or not. T he distinction t hat is obscured by t he style of possibleworld semantics is dramatized by the structured pr opositions picture.That is part o f the reason why I like it.Some directly referential terms, like proper names, m ay have no se- k-mantically relevant descriptive meaning, or at least none that is specific:1t h-a t'disti

icals, may have a limited kind of specific descriptive meaning relevantto the features of a context of use. S ti ll others, like ' clthat ' terms (seebelow), may be associated with full-blown Fregean senses used to fix thereferent. B u t in any case, the descriptive meaning of a direct ly referen-tial term is no part of the propositional content.24 This is the first sense of footnote 16. Th is is the second sense of footnote 16.

Demonstratives 49 7


10/42

498 D a v i d Kaplan D e m o n s t r a t i v e s 4 9 9

V. Argu ment fo r P r inc ip le 2 : Pure Index ica lsAs stated earlier, I believe this principle is uncontroversial, B u t I hadbest distinguish it from similar principles which are false. I am not claim-ing, as has been claimed for proper names, that indexicals lack anythingthat might be called 'descriptive meaning'. Indexicals, in general, havea rather easily statable descriptive meaning. B u t i t i s clear tha t thi smeaning is relevant only to determining a referent i n a context of useand not to determining a relevant indiv idua l in a circumstance of evalu-ation. Let us return to the example in connection wit h the sentence (3)and the indexical 'P. The bizarre result of taking the descriptive mean-ing of the indexical to be the propositional constituent is that what Isaid in utt ering (3) would be true in a circumstance of evaluation if andonly if the speaker (assuming there is one) of the circumstance does notexist in the circumstance. Nonsense I f that were the correct analysis,what I said could not be true. Fr om which i t follows that

It is impossible that I do not exist.Here is another example to show that the descriptive meaning of anindexical may be entirely inapplicable in the circumstance of evaluation.When I say,

I wish I were not speaking now.The circumstances desired do not involve contexts o f use and agentswho are not speaking. The actual context of use is used to determinethe relevant individual: mea nd time: nowand then we query thevarious circumstances of evaluation with respect to that individual andthat time.Here is another example, not of the inapp licabili ty of the descriptivemeaning to circumstances but of its irrelevance. Suppose I say at to, "Itwill soon be the case that all tha t is now beautiful is faded." Considerwhat was said in the subsentence,

Al l that is now beautiful is faded.I wish to evaluate that content at some near future t ime tl. W h a t i s t h erelevant time associated with the indexical 'now'? Is it the future tim etl? No, it is to

See how rigidly the indexicals cling to the referent determined in thecontext of use:(4) I t is possible that in Pakistan, in five years, only those whoare actually here now are envied.The poin t of (4) is t hat the circumstance, place, and time referred t o bythe indexicals 'actually', 'here', and 'now' are the circumstance, place,and time of the context, not a circumstance, place, and time determinedby the modal, locational, and temporal operators wi thi n whose scopethe indexicals lie.

It may be objected that this only shows that indexicals always takeprimary scope (in the sense of Russell's scope of a definite description).This objection attempts to relegate al l direct reference to im pli ci t useof the paradigm of the semantics o f direct reference, the variable, Thus(4) is transformed into,

The actual circumstances, here, and now are such tha t it ispossible that in Pakistan in five years only those who, in thefirst, are located at the second, dur ing the thi rd, are envied.

Although this may not be the most felicitous form of expression, itsmeaning and, in particular, its symbolization should be clear t o thosefamiliar with quantified intensional logics. The pronouns, 'the first', 'thesecond', and 'the third' are to be represented by distinct variables boundto existential quantifiers at the beginning and identified wit h ' the actualcircumstance', 'here', and 'now' respectively.(5)

(3w)(3p)(3t) [w=the actual circumstance A p=here A t= nowA 0 In Pakistan In five years Vx( x is envied --, s is locatedat p during i n w)]

But, such transformations, when thought of as representing the claimthat indexicals take primary scope, do not provide an alternative toPrinciple 2, since we may sti ll ask of an utterance of (5) i n a context c,when evaluating it with respect to an arbit rary circumstance, to what dothe indexicals 'actual', 'here', and 'now' refer, The answer, as always, is:the relevant features of the context e. [In fact, although (4) is equivalentto (5), neither indexicals nor quantificat ion across intensional operatorsis dispensable in favor of the other.]


11/42

500 David Kaplan

Perhaps enough has been said t o establish the following .( T i ) The descriptive meaning o f a pure indexical determines the

referent of the indexical with respect to a context of use butis either inapplicable or irrelevant to determining a referentwith respect to a circumstance o f evaluation.

I hope that your intuition will agree with mine that it is for this reasonthat:(T2 ) When what was said in using a pure indexical in a context a

is to be evaluated with respect to an arb itr ary circumstance,the relevant object is always the referent of the indexical withrespect to the context a.This is just a slightly elaborated version of Principle 2.

Before turning to true demonstratives, we will adopt some terminol-ogy.

VI. Termino log ica l Remarks'Principle 1 and Principle 2 taken together imply that sentences contain-ing pure indexicals have two kinds of meaning.V I . ( i ) Con t en t and C i r c ums tanceWhat i s said in using a given indexical in different contexts may bedifferent. Th us if I say, today,

I was insulted yesterdayand you utte r t he same words tomorrow, wha t is said is different. I fwhat we say differs i n trut h-value, t hat is enough to show th at we saydifferent things. B u t even if the truth-values were the same, i t is clearthat there are possible circumstances in which what I said would be truebut what you said would be false. Thus we say different things.Let us call this first kind of meaningwhat is saidcontent. Thecontent of a sentence i n a given context is what has tradi tiona lly beencalled a proposition. Strawson, in noting that the sentence

The present king o f France is baldcould be used on different occasions t o make differen t statements, used'statement' in a way similar t o our use of content of a sentence. I f we

Demonstratives 50 1

wish to express the same content in different contexts, we may have tochange indexicals. Frege, here using 't houg ht' for content of a sentence,expresses the point well.

If someone wants to say the same today as he expressedyesterday using the word 'today', he must replace this wordwith 'yesterday'. Although the thought is the same its verbalexpression must be diff erent so t hat the sense, whic h wouldotherwise be affected by the differing times of utterance, isreadjusted.26I take content as a notion applying not only to sentences taken in a

context but to any meaningful part of speech taken in a context. Th uswe can speak of the content of a definite description, an indexical, apredicate, etc. I t is contents th at are evaluated in circumstances o fevaluation. I f the content is a proposition (i.e., the content of a sentencetaken in some context), the result of the evaluation will be a truth -value.The result of evaluating the content of a singular term at a circumstancewill be an object (what I earlier called 'the relevant object'). In general,the result o f evaluating the content o f a well-formed expression a at acircumstance wi ll be an appropria te extension for a (i .e., for a sentence,a truth-values for a term, an indivi dual; for an n-place predicate, a setof n-tuples of individuals, etc.). This suggests that we can represent a26 From The Thought: A Logical Inq uiry, Mind 65 (1956): 289-311. I f Frege hadonly supplemented these comments wi th the observation th at indexicals are de-vices of direct reference, the whole theory o f indexicals would have been his. B ut

his theory of meaning blinded him to t his obvious point, Frege, I believe, mixedtogether the two kinds of meaning in what he called Sinn. A tholigh t is, for him,the Sin n o f a sentence, or perhaps we should say a complete sentence. Sinn is tocontain both the manner and context of presentation [of the denotation], accord-ing to Clber Sinn und Bedeutung (Zei tschr ift f i r Philosophic end philosophischeEr it ik 100 (1892); trans. as On Sense and Nominatum, in Contemporary Read-ings in Logical Theory, ed. Copi and Gould (Macmillan, 1967); mistrans. as OnSense and Meaning, in Martini ch, op. cit.). Sinn is fi rst i ntroduced to representthe cognitive significance of a sign, and thus t o solve Frege's problem: how canreviftrue

taken to represent the truth-conditions or content (in our sense). Frege felt thepull o f the two notions, which he reflects in some torture d passages about 'I ' inThe Thought (quoted below in XVI I). If one says Today is beautiful on Tues-day and Yesterday was beauti ful on Wednesday, one expresses the same thoughtaccording to the passage quoted. Yet one can clearly lose track of the days andnot realize one is expressing the same thought . I t seems then tha t t houghts arenot appropriate bearers of cognitive significance. I return to th is to pic in ) 0/1I.A detailed examination of Frege on demonstratives is contained in John Perry'sFrege on Demonstratives, Philosophical Review 86 (1977): 474-97.


12/42

502 Da vi d Kaplan

content by a function from circumstances of evaluation to an appropriateextension. Ca rna p called such functio ns intension&The representation is a handy one and I will often speak of con-tents in terms of it, bu t one should note th at contents which are disti nct

but equivalent (i.e., share a value in all circumstances) are representedby the same intension, Am on g oth er things, this results in the loss ofmy di stinc tion between terms which are devices of direct reference anddescriptions which tur n ou t t o be r igid designators. (Re call the meta-physical paragraph of section IV.) I wanted the content of an indexicalto be just the referent itself, but the intension of such a content will bea constant function. Ils e o f representing intentions does not mean I amabandoning that ideajust ignoring it temporarily.A fixed content is one represented by a constant function. A l l di-rectly referential expressions (a s w ell as all rigi d designators) have afixed content. [ Wh at I elsewhere call a stable content.1Let us sett le on circumstances fo r possible circumstances o f evalu-ation. B y this I mean both actual and counterfactual situations withrespect t o which it is appropriate to ask for the extensions o f a givenwell-formed expression. A circumstance w il l usually include a possiblestate or histo ry o f the world, a time, and perhaps other features as well.The amount of information we require from a circumstance is linked tothe degree of specificity of contents, and thus t o the kinds of operatorsin the language,

Operators of the familiar kind treated in intensional logic (modal,temporal, etc.) ope rat e o n contents. (Si nce we represent contents byintensions, i t is n ot surprising th at intensional operators operate oncontents.) T h u s an appropriate extension for an intensional operatoris a function from intensions to extensions.27 A m o d a lo p e r a t o rw h e napplied to an intension will look a t th e behavior of the intension wit hrespect to the possible state of the world feature of the circumstancesof evaluation. A tempor al operator wil l, similarly, be concerned with27 As we shall see, indexical operators such as I t is now the case that ," " I t is actuallythe case tha t," and "dt hat " ( the last takes a term rathe r than a sentence asargument) are also intensional operators. They differ from the familiar operatorsin only tw o ways; first, thei r extension (the function from intensions to extensions)depends on context, and second, they are directly referential (thus they have a fixedcontent). I shall argue below (in section VII: Monsters) that all operators that canbe given an English reading are a t most' intensional. Not e th at when discussingissues in terms of the formal representations of the model-theoretic semantics, Itend to speak in terms of intensions and intensional operators rather than contentsand content operators,

Demonstratives 50 3

the time o f the circumstance, I f we bui lt the ti me o f evaluation intothe contents (thus removing time from the circumstances leaving only,say, a possible worl d histo ry, and makin g contents specific as t o t ime),it w ould make no sense to have temporal operators, T o put the pointanother way, if what is said is thought of as incorporating reference toa specific time, o r state o f the world, o r whatever, i t is otiose to askwhether what is said would have been true at another time, in anotherstate of the world, or whatever. Tempo ral operators applied to eternalsentences (those whose contents incorpora te a specific ti me of evalua-tion) are redundant. A n y intensional operators applied t o perfect sen-tences (those whose contents incorporate specific values for all featuresof circumstances) are redundant.282 6Th enotionofredundancyinvolvedc

building the ti me of evaluation into contents, or making contents specific as totime, or taki ng what i s said to incorporate reference to a specific time, what Ihave in mind is this. G iven a sentence S: ' I am writing ', in the present contextc, which of the following should we take as the content: (i ) the propositio n thatDavid Kaplan is writing at IO A.M. on 3/26/77, or (ii) the 'proposit ion' that DavidKaplan is writing? The proposition (i) is specific as to time, the 'proposition' (ii)ithe scare quotes reflect my feeling that this is not t he tr aditional notion of aproposition] is neutral with respect to time. If we take the content of S in c to be(ii), we can ask whether it would be true at times other than the time of c. Thuswe thin k of the temporall y neutral 'prop osition' as changing its truth-va lue overtime. Note that it is not just the noneternal sentence S that changes its truth-value over time, but the 'proposition' itself. Since the sentence S contains anindexical 'I', it will express different 'propositions' in different contexts, But sinceS contains no temporal indexical, the ti me o f the context wil l n ot influence the'proposition' expressed. An alternative land more traditional] view is to say thatthe verb tense in S involves an impl icit temporal indexical, so tha t S is understoodas synonymous with S': ' I am writin g now'. I f we take this point of view we willtake the content of S in c to be (i ). In this case what is said is eternal ; i t does notchange its truth-value over time, although S will express different propositions atdifferent times.

There are bot h technical and philosophical issues involved in choosing between(i) and (ii). Philosophically, we may ask why the temporal indexical should betaken to be implicit (making the proposition eternal) when no modal indexicalis taken to be implicit. A ft er all, we coithi understand S as synonymous withS": ' I am actually writing now', T he content of S" i n c is not only eternal, itis perfect, It s tru th changes neither through time nor possibility. I s there somegood philosophical reason for preferring contents which are neutral with respectto possibility but draw fixed values frota the context for all other features of apossible circumstance whether o r no t t he sentence contains an expl ici t indexical?(It may be that the t radit ional view was abetted by one of the deli ghtfu l anomaliesof the logic of indexicals, namely that S, S', and S" are al l logically equivalentSee Remark 3, p. 547.) Technically, we must note that intensional operators must,if they are not to be vacuous, operate on contents which are neutral with respect


13/42

504 Da v i d Kaplan Demonstratives 5 0 5What sorts of intensional operators to admit seems to me largelya matter of language engineering. I t is a question of which features o fwhat we intuitively think of as possible circumstances can be sufficientlywell defined and isolated. I f we wish to isolate location and regard i t as afeature of possible circumstances we can introduce locational operators:'Two miles north it is the case that' , etc. Such operators can be iteratedand can be mixed with modal and temporal operators. However, to make

such operators interesting we must have contents which are locationallyneutral. T ha t is, i t must be appropriate to ask if what is said would betrue in Pakistan. ( Fo r example, 'I t is raining' seems to be locationallyas well as temporally and modally neutral.)This functional notion of the content of a sentence in a context maynot, because of the neutralit y o f content wi th respect to time and place,say, exactly correspond to the classical conception of a proposition. Butthe classical conception can be introduced by adding the demonstratives'now' and 'here' t o t he sentence and tak ing the content of the result.I will continue to refer to the content of a sentence as a proposition,ignoring the classical use.Before leaving the subject of circumstances of evaluation I should,perhaps, note that the mere attempt to show that an expression is di-rectly referential requires that it be meaningful to ask o f an individualin one circumstance whether and wi th what properties it exists in an-other circumstance. I f such questions cannot be raised because they areregarded as metaphysically meaningless, the question of whether a par-ticular expression is directly referential (or even, a rigid designator) can-not be raised. I have elsewhere referred to the vie w tha t such questionsare meaningful as haecceitism, and I have described other metaphysicalmanifestations of this view,2 9 Ia d v o c a t et h i sp o s i t i o n,a l t h ou g hIam

to the feature of circumstance the operator is interested in. Thus, for example, i fwe take the content of S to be (i), the application of a temporal operator to sucha content would have no effect; the operator would be vacuous. Furthermore, ifwe do not wish the iteration of such operators to be vacuous, the content of thecompound sentence containing the Operator must again be neutral with respectto the relevant feature of circumstance. This is not to say that no such operatorcart have the effect of fixing th e relevant feature and thus, in effect, renderingsubsequent operations vacuous; indexical operators do ju st th is. I t is just tha tthis must not be the general situation. A content must be the kind o f entity thatis subject to modification in the feature relevant to the operator. [The textualmaterial to which this note is appended is too cryptic and should be rewritten.)29 How to Russell a Prege-Church. Th e pronunciation is: Hefo-ee-i-tis-m. Theepithet was suggested by Robert Adams. I t is not an accident that it is derivedfrom a demonstrative.

uncomfortab le w ith some of its seeming consequences ( fo r example, thatthe world might be in a state qualitatively exactly as it is, bu t w ith apermutation of individuals).It is hard to see how one could thi nk about the semantics o f indexicalsand modality without adopting such a view.

VI . ( i i ) Charac te rThe second kin d of meaning, most prominent in the case of indexicals,is that which determines the content in varying contexts. The rule,

'I' refers to the speaker or writeris a meaning rule of the second kind. Th e phrase 't he speaker or writ er'is not supposed to be a complete description, nor it is supposed to referto the speaker or writ er of the word T (Th ere are many such.) I t refersto the speaker or writer of the relevant occurrence of the word 'I', thatis, the agent of the context.

Unfortunately, as usually stated, these meaning rules are incompletein that they do not explicitly specify that the indexical is directly refer-ential, and thus do not completely determine the content in each context.I wil l return to this later.Let us call the second kind o f meaning, character. T he character of

an expression is set by li nguis tic conventions and, in turn, determines thecontent of the expression in every con te xt .30 B e c a u s ec h a r a c t e ri sw h a t

is set by linguistic conventions, it is natural t o th ink of i t as meaning inthe sense of what is known by the competent language user.Just as it was convenient to represent contents by functions fr ompossible circumstances to extensions (Carnap's intentions), so it is con-venient to represent characters by functions from possible contexts tocontents. (A s before we have the drawback tha t equiv alent charactersare identi fied.31) T h i sg i v esu st hef o l lo w in g

3 0Th isdoesno

another context, you can decide whethe r the contents are the same. I may twiceuse 'here' on separate occasions and not recognize that the place is the same, ortwice hear 'I' and not know i f the content is the same. W ha t I do know is this:if it was the same person speaking, then the content was the same. [More on thi sepistemological stuff later.]311ain

to see whether the revisions in Fregean semantical theory, which seem plainlyrequired to accommodate indexicals (this is t he 'obviousness' of my theory), canthrow any light on it. Here we assume that aside from indexicals, Frege's theory


14/42

506 D a v i d Kaplan D e m o n s t r a t i v e s 5 0 7

Character: Contexts C o n t e n t sContent : Ci rcumstances Ex te ns ion s

or, i n more famil iar language,Meaning + Context IntensionIntension + Possible World E xt en s i o n

Indexicals have a context-sensztive character. I t is characteristic o fan indexical that its content varies with context. Nonindexicals havea fixed character. T h e same content is invoked in al l contexts. T h i scontent will typically be sensitive to circumstances, that is, the non-indexicals are typically not rigid designators but will vary in extensionfrom circumstance to circumstance. Etern al sentences are generally goodexamples of expressions with a fixed character.

Al l persons alive in 1977 w il l have died by 2077expresses the same proposition no matter when said, by whom, or underwhat circumstances. T he truth- value of that proposition may, of course,vary w ith possible circumstances, bu t the character is fixed. Sentenceswith fixed character are very useful to those wishing to leave historicalrecords.

Now that we have two kinds of meaning in addition to extension,Frege's principle of intensional int erc han ge' becomes two principles;is correct, roughly, tha t words and phrases have a kind of descriptive meaning orsense which a t one and the same ti me constitutes their cognitive significance andtheir conditions of applicability.

Kripke says repeatedly in Naming and Necessity th at he is on ly provi ding apicture o f how proper names refer and th at he does not have an exact theory.His picture yields some star tli ng results. I n the case of indexicals we do have arather precise theory, which avoids the difficulty of specifying a chain of corrununi-cation and which yields many analogous results. I n facing the vastly more difficul tproblems associated with a theory of reference for proper names, the theory ofindexicals may prove useful; i f only t o showas I believethat proper names arenot indexiceds and have no meaning in the sense in w hich indexicals have mean-ing (namely a 'cognitive content' which fixes the references in all contexts). [Theissues that arise, involving token reflexives, homonymous words with distinct char-acter, and homonymous token reflexives wi th the same character are best savedfor latermuch later.]3 2Se e

Chicago Press, 1947).

(F1) Th e character of the whole is a function of the character o fthe parts. T ha t is, i f two compound well-formed expressionsdiffer only wit h respect t o components which have the sameCharacter, then the Character of the compounds is the same.

(F2) The Content of the whole is a function of the Content of theparts. T h a t is, i f two compound well-formed expressions,each set in (possibly different) contexts differ only with re-spect to components which wizen taken in their respectivecontexts have the same content, then the content of the twocompounds each taken in its own context is the same.

It is the second principle that accounts for the often noted fact thatspeakers in different contexts can say the same thing by switching in-dexical& (And indeed they often must switch indexicals to do so.) Fregeillustrated this point w ith respect to 'today' and 'yesterday' in "Th eThought." (But note that his treatment of 'I' suggests that he does notbelieve that utterances of a n d 'you' could be simila rly related )Earlier, in my metaphysical phase, I suggested t ha t we should thi nkof the content o f an indexical as being just the referent itsel f, and I re-sented the fact tha t the representation of contents as intensions forced usto regard such contents as constant functions. A simil ar remark applieshere. I f we are not overly concerned wi th standardized representations(which certainly have their value for model-theoretic investigations) wemight be inclined to say tha t t he character of an indexical-free word orphrase just is its (constant) content.

VI I , Ea r l ie r A t t em p ts : I nd ex T heo r yThe following picture seems to emerge. Th e meaning (character) o f anindexical is a function from contexts to extensions (substituting for fixedcontents). The meaning (content, substituting for fixed characters) o f anonindexical is a f unction fro m circumstances to extensions. Fr om thispoint of view it may appear that the addition of indexicals requires nonew logic, no sharp distinctio n between contexts and circumstances, jus tthe addition of some special new features ('conte xtual ' features) t o thecircumstances of evaluation. (For example, an agent to provide an in-terpretation for T. ) Thus an enlarged view of intension is derived. Theintension of an expression is a function from certain factors to the ex-tension of the expression (with respect to those factors). Originally suchfactors were simply possible states of the world, but as it was noticed


15/42

508 David Kaplan

that the so-called tense operators exhibited a structure highly analo-gous to that of the modal operators the factors with respect to whichan extension was to be determined were enlarged to include momentsof time. W he n it was noticed tha t contextual factors were required todetermine the extension o f sentences containing indexicals, a s till moregeneral noti on was developed and called an index. The extension of anexpression was to be determined wit h respect to an index. T he intensionof an expression was tha t f unctio n whic h assigned to every index, theextension at that index.

The above example supplies us with a statement whose truth-value is not constant bu t varies as a funct ion of i EThis situation is easily appreciated in the context of time-dependent statements; t ha t is, in the case where I representsthe instant of time. Obviously the same statement can betrue at one moment and false at another. For more generalsituations one must not thi nk o f the i E I as anything assimple as instants of time or even possible worlds. In generalwe will have

i = (up,t,p, a,...)where the index i has many coordinates: fo r example, w is aworld, t is a time, p = (x, y,z) is a (3-dimensional) positionin the world, a is an agent, etc. A l l these coordinates canbe varied, possibly independently, and thus affect the truth-values of statements which have indirect references to thesecoordinates. [From the Advice of a prominent logician.]

A sentence rt, w a st a k ent obel o gi c al l ytrue

(in every 'structure'), and 0 0 was taken to be true at a given index(in a given structure) ju st in case 0 was true a t every index (in thatstructure). Thu s the familiar principle of modal generalization: i f k 0,then 0 0 , is val idated.This view, in its treatment of indexicals, was technically wrong and,more importantly, conceptually misguided.Consider the sentence

(6) I am here now.It is obvious that for many choices of indexi.e., for many quadruples(w, x, p, t) where w is a possible world h istory, x is a person, p is a place,

Demonstratives 50 9

and t is a time(6) w ill be false. I n fact, (6) is true only wi th respectto those indices (w, x, p,t) which are such tha t i n the w orld history w,x is located at p at the time t. T hus (6) fares about on a par with(7) D a v i d Kaplan is in Portland on 26 March 1977.(7) is empirical, and so is (6).

But here we have missed something essential to our understandingof indexical& Intuitively, (6) is deeply, and in some sense, which wewill shortly make precise, universally, true. O ne need o nl y understandthe meaning of (6) to know tha t it cannot be uttered falsely. No suchguarantees apply to (7). A Logic of Indexicals which does not reflect thisint uit ive difference between (6) and (7) has bypassed something essentialto the logic of indexical&

What has gone wrong? W e have ignored the special relationshipbetween T , 'here', and 'no w'. Her e is a proposed correction. Le t t heclass of indices be narrowed to include only the proper onesnamely,those (w, x, p, t) such tha t in the world w,x is located at p at the ti me t.Such a move may have been intended ori gina lly since improper indicesare like impossible worlds; no such contexts could exist and thus thereis no interest in evaluating the extensions of expressions with respect tothem. Ou r reform has the consequence tha t (6) comes out, correc tly, tobe logically true. Now consider(8) 0 I am here now.Since the contained sentence (namely (6 )) is true at every proper i ndex,(8) also is true a t every proper index and thus also is log ical ly true. (A swould be expected by the aforementioned principle of modal generaliza-tion.)

But (8) should not be logically true, since it is false, I t is certainlynot necessary that I be here now. But for several contingencies, I wouldbe working in my garden now, or even delivering this paper in a loca tionoutside of Portland.The difficulty, here, is the attem pt to assimilate the role of a context

to tha t of a circumstance. The indices (w, x ,p, t) tha t represent contextsmust be proper in order that (6) be a tru th o f the logic of indexicals, butthe indices tha t represent circumstances must include improper ones inorder tha t (8) not be a logical truth .

If one wishes to stay with this sort o f index theory and blur theconceptual difference between context and circumstance, the minimalrequirement is a system of double indexing, one index for context and


16/42

510 David Kaplan Demonstratives 6 11

another for circumstance. I t is surprising, looking back, tha t we (for Iwas among the early index theorists) did not immediately see that doubleindexing was required, for in 1967, at UCLA, Hans Kamp had reportedhis work on ' no w' " in which he had shown tha t double indexing wasrequired to p roperly accommodate temporal indexicals along wit h th eusual temporal operators. B u t it was four years before i t was realizedthat this was a general requirement for (and, i n a sense, the key to) alogic of indexicals.However, mere double indexing, wit hou t a clear conceptual under-standing o f what each index stands for, is stil l no t enough to avoid allpitfalls.

VII I . Monsters Begat by E leganceMy l ibera lity wit h respect to operators on content, i.e., intensional op-erators (any feature of the circumstances of evaluation t hat can be wel ldefined and isolated) does not extend to operators which attempt tooperate on character. A re there such operators as 'In some contexts it istrue that ', which when prefixed t o a sentence yields a truth if and onlyif in some context the contained sentence (not the content expressed byit) expresses a content that is true i n the circumstances of that context?Let us try it:(9) J o some contexts it is true that I am not tired now.For (9) to be true in the present context i t suffices th at some agent ofsome context not be tired at the time of tha t context. (0), so interpreted,has nothing t o do with me or the present moment. B u t this violatesPrinciple 2 Pri nci ple 2 can also be expressed in more theory laden wayby saying that indexicals always take primary scope. I f this is trueandit isthen no operator can control the character of the indexicals withinits scope, because they wil l s imply leap o ut of its scope to the fr ont ofthe operator. I am n ot saying we could not construct a language wit hsuch operators, just that English is not o ne .' An d such operators could\ not be added to it.

There is a way to control an indexical, to keep it from taki ng pri maryscope, and even to refer it to another context (th is amounts to changingits character). Use quotation marks. I f we mention the indexical rather3 3Pu bl is h

'Thomason alleges a counterinstance: 'Never put off until tomorrow what you cando today'. What should one say about this?

than use it, we can, of course, operate directly on it. Carnap oncepointed out to me how i mportan t the difference between direct andindirect quotation is inOtto said " I am a fool."Otto said that I am a fool.

Operators like 'In some contexts it is true that', which atte mpt to meddlewith character, I call monsters. I claim t hat none can be expressed inEnglish (without sneaking in a quotation device). I f they stay in themetalanguage and confine thei r atten tion to sentences as i n

In some contexts " I am not tire d now" is truethey are rendered harmless and ca n even do socially useful work (asdoes, 'is valid' {see below)).

I have gone on at perhaps excessive length about monsters becausethey have recently been begat b y elegance. I n a specific appli cationof the theory of indexicals there w il l be ju st certain salient features ofa circumstance o f evaluation. S o we m ay represent circumstances b yindexed sets of features. This is typical of the model-theoretic way. Asalready indicated, al l the features o f a circumstance wil l generally berequired as aspects of a context, and the aspects of a context may all befeatures of a circumstance. I f not, a lit tle ingenuity may make it so .353 3Re ca llthatinaparticul

include all elements with respect to which there are content operators, and theaspects of a context must include all elements with respect to which there areindexicals. Thus, a language with bot h the usual modal operators '0 ', 97, andan indexical modal operator 'I t is actually the case that' wil l contain a possibleworld history feature in its circumstances as well as an analogous aspect in itscontexts. I f a circumstance is an aspect of a context , as seems necessary for thedefinition of truth, then we only need worry about aspects of contexts that arenot features of circumstances. The most prominent of these is the agent of thecontext, required to interpret the indexical I n order to supply a correspondingnonvacuous feature to circumstances we must treat contents i n such a way tha t wecan ask whether they are true for various agents. (N ot characters mi nd you, bu tcontents.) Th is can be done by representing the agent by a nesstrala term whichplays the syntactical role of 'I' but gets an interpretation only with respect to acircumstance. Let a be a special variable that is not subjec t t o quanti fication andlet b be a variable not in the language. Ou r variable a is the neutra l. We wish t ointroduce content operators which affect the agent place and which can be iterated.Let R be a relation between individuals, for example 'aRb' for 'b is an uncle of a'.Then we may interpret the operator ORO as (3b)(aR6 A ( I s ) (b = a A O)J I f i s'a walks', ORO comes to 'an uncle of a walks'. The indexical 'I' can be representedby an operator 0

1 f o rw h i c h'

equivalent to replacing the neutral a by t he indexical T.


17/42

512 Da v i d Kaplan

We could then represent contexts by the same indexed sets we use torepresent circumstances, an d instead of having a l ogic of contexts andcircumstances we have simp ly a two-d imensional logic o f indexed sets.This is algebraically very neat and it permits a very simple and elegantdescription of certain important classes of characters (for example, thosewhich are true at every pair (i, i), though the special significance ofthe set is somehow diminished in the abstract formulati on) ,as B u t i talso permits a simple and elegant intr oducti on of many operators whichare monsters. I n abstracting from the dis tinct conceptual roles playedby contexts of use and circumstances of evaluation the special logic ofindexicals has been obscured. O f course restrictions can be put on thetwo-dimensional logic to exorcise the monsters, but to do so would beto give up the mathematical advantages of that fo rmu la ti on .37IX . A rgumen t fo r P r inc ip le 2: T ru e Demonstra t ivesI return now to the argument that all indexicals are directly referential.Suppose I point at Paul and say,

He now lives in Princeton, New Jersey.Call what I saidi.e., the content of my utterance, the propositionexpressed'Pat'. Is Pat true or false? True Suppose that unbeknownstto me, Paul had moved to Santa Monica last week. Would Pat havethen been true or false? False No w, the tr icky case: Suppose tha tPaul and Charles had each disguised themselves as the other and hadswitched places. I f that had happened, and I had uttered as I di d, thenthe proposit ion I woul d have expressed woul d have been false. B u t inthat possible context the proposi tion I wou ld have expressed is not Pat.That is easy to see because the proposition I would have expressed, hadI pointed to Charles instead of Paulcall this proposition 'Mike' notonly would have been false but actually is false. Pa t, I would claim,would still be true in the circumstances of the envisaged possible con-3 6Se e,forexa

Philosophical Logic 2 (1973): 77-96 . Segerberg does metamathematical work i nhis article and makes no special philosophical claims about its significance. Th athas been clone by others."T he re is one other difficulty in identi fying the class of contexts with the class ofcircumstances. T he special relationship between the indexicals T , 'here', 'n ow'seems to require that the agent of a context be at the location o f the contextduring the time of the context. Bu t this restriction is not plausible for arbit rarycircumstances. I t appears tha t thi s approach will have difficu lty in av oiding theproblems of (6) and (8) (section VII).


text provided that Paulin whatever costume he appearedwere stillresiding in Princeton.I X . ( i ) Th e A r gum en t sI am arguing that in order to determine what the truth -value of a propo-sition expressed by a sentence contain ing a demonstrative would be underother possible circumstances, the relevant individual is not the indivi dualthat would have been demonstrated had those circumstances obtainedand the demonstration been set in a context of those circumstances, butrather the in dividu al demonstrated i n th e context which d id generatethe proposition being evaluated. A s I have already noted, i t i s char-acteristic of sentences containing demonstrativesor, for that matter,any indexicalthat they may express different propositions in differentcontexts. We must be wary of confusing the proposition that would havebeen expressed by a similar utterance in a slightly different contextsay, one in which the demonstratum is changedwith the propositionthat was actually expressed. I f we keep this distincti on i n mi ndi.e., wedistinguish Pat and Mikewe are less likely to confuse what the t rut h-value of the proposition actually expressed would have been under somepossible circumstances with what the truth-val ue of the propositi on th atwould have been expressed would have been under those circumstances.When we consider the vast array of possible circumstances wi th re-spect to which we might inquire i nto the tr uth of a proposition expressedin some context c by an utterance u, it quickly becomes apparent thatonly a small fraction of these circumstances wi ll involve an utterance ofthe same sentence in a simil ar context, and that there roust be a way o fevaluating the truth-value of propositions expressed using demonstra-tives in counterfactual circumstances i n which no demonstrations aretaking place and no indivi dual has the exact characteristics exploited i nthe demonstration. Surely, it is irrelevant to determining whether what Isaid would be true or not in some counterfactual circumstance, whetherPaul, or anyone for that matter, looked as he does now. All that wouldbe relevant is where he lives. Therefore,(T3 ) the relevant features of the demonstratum qua demonstra-

turn (compare, the relevant features of the x Px qua the xFa)namely, that the speaker is pointing at i t, t ha t i t hasa certain appearance, is presented in a certain waycannotbe the essential characteristics used t o identi fy t he relevantindividual in counterfactual situations.


18/42

514 Da v i d Kaplan

These two arguments: th e distinction between Pat and Mike, a nd con-sideration o f counterfactual situations in which no demonstration occurs,are offered to support the view tha t demonstratives are devices o f directreference (rig id designators, i f you wi ll ) and, b y contrast, t o reject aFregean theory of demonstratives.IX , ( i i ) Th e Fregean Theory o f Demonst ra t i onsIn order to develop the latter theory, in contrast to my own, we turn firstto a p ortio n o f the Fregean theory which I accept: th e Fregean theoryof demonstrations.

As you kno w, for a Fregean the paradi gm o f a meaningful expres-sion is the definite description, whi ch picks out or denotes an ind ividual,a unique individual, satisfying a condition s. T he individual is calledthe denotation o f the definite description and the cond ition s we mayidentify with the sense of the definite description. Since a given individ-ual may uniquely satisfy several di stinc t conditions, definite descriptionswith distinct senses may have the same denotation. And since some con-ditions may be uniquely satisfied by no individual, a definite descriptionmay have a sense but no denotation. The condition by means of whicha definite description picks out its denotation is the manner of presen-tation of the denotation by the definite description.

The Fregean theory of demonstratives claims, correctly I believe,that the analogy between descriptions (short for 'definite descriptions')and demonstrations is close enough t o p rovide a sense and denota tionanalysis of the 'meaning' o f a demonstration. T h e denotation is t hedemonstratum (that which is demonstrated), and it seems quite nat-ural t o regard each demonstration as presenting its demonstratum i na particular manner, which we may regard as the sense of the demon-stration. T h e same indivi dual could be demonstrated by demonstra-tions so different in manner of presentation tha t i t would be informat iveto a competent auditor-observer t o be tol d t hat the demonstrata wereone. Fo r example, it mig ht be informative to you for me to tell youthat

Tha t [po inting to Venus in the morning sky] is identical withthat [pointing to Venus in the evening sky].

(I would, of course, have to speak very slowly.) T h e two demonstra-tionscall the first one 'Phos' and the second one 'Hes'which accom-panied the two occurrences of the demonstrative expression 't ha t' have

Demonstratives 51 5

the same demonstratum but dis tinc t manners of presentation. I t i s thisdifference between the sense o f Hes and the sense o f Phos that accounts,the Fregean claims, for the informativeness o f the assertion.

It is possible, to pursue the analogy, for a demonstrat ion to have nodemonstratum. Th is can arise in several ways: thr oug h hall ucination,through carelessness (not not icing , i n the darkened room , t ha t the sub-ject had jumped off the demonstration platform a few moments beforethe lecture began), through a sortal conflict (using t he demonstrativephrase th a t , where F is a common noun phrase, while demonstrat-ing something which is not an F), and in other ways.

Even Donnellans's important distinction between referential and at-tributive uses of definite descriptions seems to fit, equally comfortably,the case of demonstrations.'

The Fregean hypostatizes demonstrations in such a way th at it isappropriate to ask of a given demonstration, say Phos, wh at would ithave demonstrated under various counterfactual circumstances. Phosand lies mig ht have demonstrated distinct ind ivi dual s."

We should not allow our enthusiasm for analogy to overwhelm judg-ment in this case. There are some relevant respects in which descrip-tions and demonstrations are disanalogous. F ir st, as Davi d Lewis haspointed out, demonstrations do not have a syntax, a fixed formal struc-ture in terms of whose elements we might try to define, either directlyOr recursively, the notion of sense." Second, to different audiences (forexample, the speaker, those sittin g in front of the demonstration p lat-form, and those sitting behind the demonstration platform) the samedemonstration may have different senses. O r perhaps we shoul d saythat a single performance may involve distinct demonstrations from theperspective of distinct audiences. (" Exa ct ly like proper names " says theFregean, "as long as the demonstrat um remains the same, these fluctu-ations in sense are tolerable. B u t they should be avoided in the system3 81 have wri tte n elsewhere, in appendices YU and \ Pil l of "Bob and Carol and Tedand Alice," o f these matters and won't pursue the topic now.

it could then be proposed that demonstrations be individuated by the principle:(/2 = d2 i f and Only if, for all appropri ate circumstances c, the demonstratum ofsitinc=th

is that the same demonstration is being performed in two diffe rent contexts if thestandard audience can't determine, from the demonstration alone, whether thecontexts are distinct or identical. Thi s makes the indi viduation of demonstrationsmore epistemological than the metaphysical proposal above.4 0A lt ho u

of pictures. See P. Suppes and Rottrnayer, "Automata," in Hanettook of P ercep-lion , vol. 1 (New York: Academic Press, 1974).


19/42

516 Da v id Kaplan

of a demonstrative science and should not appear in a perfect vehicle ofcommunication. )IX . ( i i i ) T he Fregean Theo ry o f Demonst ra ti vesLet us accept, tenta tive ly and cau tiously, the Fregean theory of demon-strations, and turn now to the Fregean theory of demon stratives.41 According to the Fregean theory of demonstratives, an occurrence ofa demonstrative expression functions rather like a place-holder for theassociated demonstrati on. T h e sense of a sentence contain ing demon-stratives is to be th e result of replacing each demonstrative by a con-stan t whose sense is given as the sense of the associated demonstrati on.An impo rtan t aim of the Fregean theory is, of course, to solve Frege'sproblem. An d it does that qui te neatly. Yo u recall that the Fregeanaccounted for the informativeness o f

That [l ies] = that [Phos]in terms of the disti nct senses Of Hes and Phos, No w we see th at thesenses of the tw o occurrences o f ' tha t' are identified w it h these twodistinct senses so that the ultimate solution is exactly like that given byFrege originally. T he sense of the le ft 'th at' differs from the sense of theright 'that'.I X . ( i v ) A r g um en t A ga i ns t t he F regean Theo r y o fDemonst rat ivesLet us return now to our original example:

He [Delta] now lives in Princeton, N ew Jerseywhere 'Delta' is the name of the relevant demonstration. I assume thatin the possible circumstances described earlier, Paul and Charles hav-ing disguised themselves as each other, Del ta would have demonstratedCharles. Therefore, according to the Fregean theory, the propositionlus t expressed, Pat, would have been false under the counterfactual cir-cumstances of the switch. But this, as argued earlier, is wrong. There-fore, the Fregean theory of demonstratives though it nicely solves Frege'sproblem, is simply incorrect in associating propositions with utterances.

Let me recapitulate. We compared two theories as to the propositionexpressed by a sentence containing a demonstrative along with an asso-4 1Th e

sial theory of indexicats. On the contrary, it has the fascination of the speculative.

Demonstratives 51 7

ciated demonstration. Bo th theories allow that the demonstration canbe regarded as having both a sense and a demonstratum. M y theory, thedirect reference theory, claims tha t in assessing the propos ition in coun-terfactual circumstances it is the actual demonstratumi n the example,Paulthat is the relevant individual . T he Fregean theory claims th atthe proposition is to be construed as if the sense of the demonstrationwere the sense of the demonstrative. Thus, i n coun terfactual situationsit is the individ ual that would have been demonstrated th at is the rele-vant indi vidual. Ac cordi ng to the direct reference theory, demonstrativesare rigid designators. Accordi ng to the Fregean theory, t he ir denotationvaries in different counterfactual circumstances as the demonstrata ofthe associated demonstration would vary in those circumstances.

The earlier distinction between Pat and Mike, and the discussionof counterfactual circumstances i n which, as we would now put it, thedemonstration would have demonstrated nothing, argue t ha t wit h re-spect to the problem of associating propositions with utterances thedirect reference theory is correct an d t he Pregean theo ry is wrong.

1have carefully avoided arguing for t he direct reference theory byusing modal or subjunctive sentences for fear the Fregean would claimthat the peculiarity of demonstratives is not that they are rigid designa-tors but that they always take primary scope. I f I had argued only onthe basis of our intuitions as to the truth-value of

If Charles and Paul had changed chairs, then he (Delta )would not now be living in Princetonsuch a scope interpretation could be claimed. But I didn't.

The perceptive Fregeans among you wi ll have noted t ha t I have saidnothing about how Frege's problem fares under a direct reference theoryof demonstratives. And indeed, if 'that' accompanied by

Documents

Kaplan, D. 1989, Demonstratives