Anda di halaman 1dari 11

19 Logicand Conversation

H. P. CRICE

Il is a commonplace ol' philosophical logic that there are. or appear to be. divergenccsin nrcaning between. on the onc hand. at lcast somc ol'what I shall call the lbrmal deviccs. - . n . v . l . ( V r ' ) .( 3 t ) . ( r . r )( w h e nt h e s ca r e qrven a stan(lardtw<l-valuc<l interpretationlrnd. on thc othcr. what are taken to bc their analogsor counterpartsin natural language* rrrclr cx;rrcssions as n{)/, und. rtr. il. ull. sontt' Ittr ut lca.r! rnt'\, tht. Somc logicians may at srrmctimc havc wantcd to claim that lhere arc rn l'actno such divcrgences: but such claims. if' nradeat all, have been somewhat rashly made. lnd thosc suspcctcd of making them havc hrcn subjectedto sonrepretty rough handling. Thosc who concede that such divcrgcnces crist adhcrc. in thc main. to onc or the other ,rf lwo rival groups. which I shall call the lbrrnalist and the inlbrmalist groups. An outlinc ,rl a not uncharacteristic formalist position rnay be liiven as follows: Insolar as logicians rrc concernd with the fbrmulation of very ucneralpatternsof valid inlbrence,thc formal Jr'vices possess a decisive advantagc over their natural counterparts.For it will be possibleto !onstruct in terms of the formal dcviccsa svs-

tem of vcry general formulas. a considcrahle number clf'which can be regardecl as. or are closelyrclatcd to. patternsol infbrences the expressionof which involves somc or all of the dcvices:Such a systemmay consislola cerlain set ol'simplc formulas that must bc acceptable il'the dcviccs have the mcaning that has bcen assigncdto them. and an indefinite number ol' lurthcr lbrmulas, many of which arc lcss ohviously acceptablcancl each of which can be shown to bc' acceptableif'thc membcrs ol'thc original set are acceptable.Wc havc. thus, a way of handling dubiously acceptablcpalterns of infcrencc. and it. as is sometimes possihlc. we can apply a dccision procedurc,wc have an cven bettcr way. Fuflhcrmorc, liom a frhil<r sophical point ol' view. thc posscssion by lhe natural counlcrpartsof'thoscclements in thcir meaning, which they do nol sharc wilh the corresponding lbrmal dcvices, is to be regardcd as an imperlection of natural lanSuagcs:the elements in question are undcsirablc excrescenccs.For the prcsence of thcsc elcments has the result both that the concepts within which they appear cannot be preciscly or clearly deftncd.and that at leastsomc stale-

Rcprintcd by permission of Harvard (Jniversiry Pressfronr Stutlicj in rlu' ll'ut ol l4.ortl.t, bv paul Gricc. iar l96l{. . v 1 5 h y H . P . G r i c e , ' c 1 9 8 9h y T h e P r e s r d e na rn d F e l l o w s o l ' H a r v a r d( o l l e s e .

305

306

AND RELEVANCE IMPLICATURE CONVERSATIONAT of the natural counterparts of these devicesi this logic may be aided and guidedby the simplified logic of the formal devices but cannot be supplantedby it. lndeed, not only do the two logics differ, but sometimesthey come into conflict: rules that hold for a formal device may not hold for its naturalcounterpart. questionof the placein phiOn the general losophy of the reformation of natural language,I shall, in this essay,have nothing to say.I shallconfinemyselfto the disputein its relation to the allegeddivergences. I have, moreover. no intentionof enteringthe fray on behalf of either contestant,I wish. rather. to maintainthat the commonassumption of the contestants that the divergences do in fact existis (broadly speaking) a commonmistake, and that the mistakearisesfrom inadequatc attention to the natureand importance of the conditionsgoverningconversation. I shall. therelbre, inquire into the general conditions that, in one way or another,apply to conversationas such,irrespective of its subjectmatter. I beginwith thecharacterization ofthe notion ol'implicature'.

ments involving them cannot. in some cira definitetruth valuei be assigned cumstances, is not oftheseconcepts and the indefiniteness only objectionablein itself but also leaves open the way to metaphysics-we cannot be excertainthat noneofthesenaturallanguage 'loaded'.For these pressions is metaphysically as used in natural the expressions, reasons, cannot tre regardedas linally acceptaspeech. ble, and may turn out to be, finally, not fully intelligible.The proper courseis to conceive inand bcgin to constructan ideal language. the sentences corporatingthe formal devices, of which will be clear. determinatein truth value,and ccrtifiablyfree from metaphysical will implications; the foundations o[ science now be philosophically secure, sincethe stalements of the scientist will tr expressible (though not necessrrilyactually expressed) (l do not wish to within this ideallanguage. suggest that all formalistswould accept the wholeof thisoutline. but I think thatall would acceptat leastsomepart of it.) To this. an informalistmight reply in the -fhe philosophical followingvein. demandfor rcsts on cenainassumptions an idcallanguagc theseare. that that should not be conceded; the primary yardstickby which to judge thc adequacy of a language is its ability to serve thc nccds of science, that an cxpression cannol be guaranteed as fully intelligibleunless an explication or analysisof its meaninghas been provided.and that evcry explicationor analysismust take thc form of a precise dcfinition that is the expression or assertion of a logical equivalencc. Language serves many important purposes besides thoseof scicntihcinquiry; we can know perfectlywell what an expression (and so a fortiori that it is intelligible) means without knowing its analysis. and the provision of an analysis may (and usuallydoes) consistin the specification, as generalized as possible, of thc conditions that count fbr or against the applicability of the expression beinganalyzed. Moreover.while it is no doubt true that the formal devices are cspecially amenable to syslematic trealmentby the logician. it remainsthe casethat there are very many inferences and arguments, expressed in naturallanguage and not in termsofthesedevices, which are nevertheless recognizably valid. So there must be a placefor an unsimplitied.and so more or lessunsystematic, logic

IMPLICATURI Suppose that A and B aretalkingabouta mutual friend,C. who is now workingin a bank. A asksB how C is getting on in hisjob. and B rcplies,Oh quitt' n'ell.I think: he likr"rlti.t tttlleogut,.r, and he ha.sn'tbcen lo pri.xtn vu. Ar this point. A might well inquire what B was implying. whal he was suggcsting. or evcn what he meant by sayingthat ('had not yet becnto prison.l-he answermight be any one of such things as that C is the sort of Jrrson likely to yield to the temptationprovidedby his occupation,that C's colleagues are really very unpleasant people,and and trcacherous so forth. lt might.ol'course, be quite unnecessary for A to make suchan inquiry of B, the answerto it treing,in the context,clearin advance. It is clearthat whatever B implied.suggested, meant in this example.is distinct from what B said.which wassimply that C had not been to prison yet. I wish to introduce. as termsof art. the verb implicateandthe related nouns inrplicature(cf. implving) and implicatum (cf. what is implied\. The point of this

TOCIC AND CONVERSATION maneuveris to avoid having, on each occasion. to choosebetweenthis or that member of the family of verbsfor which implit'ateisto do generalduty. I shall,for the time beingat extent to a considerable least.haveto assume of the meaningof an intuitive understanding .rn.r, in such contexls.and an ability to recogof the family nizc particularverbsas members I can, howwith which implicatcis associated. ever.make one or two remarksthat may help to clarily the more problematicof theseaswith the namely.that connected sumptions. .ra.t'. word thc meaningof in which I am usingthe word ln the sense .ra.l',t intend what someonehas said to be of ntionalmeaning to the conve related closcly he has ultered.Supthe words(thc sentence) l/c posesomeone to haveutteredthe sentence of is in thc grip ol u vitr'. (iivcn a knowledge ofthe but no knowledge language. thc English of the utterance'onc would circumstances had about what thc speakcr know somcthing that he wasspeaking said,on the assumption Onc literally. and spcaking English. standard would know that he had said.about someparticular malc personor animal .r' that at the (whatcver that was)'citimc of the utterancc of a certain to rid himself thcr ( l).r wasunable trail or (2) somc part of kind of bad character .r's personwascaughtin a certainkind of tool accounl' of or instrumcnl (approximate of what But lbr a full identification course). had said,one would nccdto know the spcaker (a) the identityof .x,(b) the time of utlcrance. ancl (c) the meaning.orr the particularoccasion ofutterance.ofthe phraseirr thagrip d o (l) and (2)1.]-his vicc la decisionbetween s rt open Ieave my of use .ta1' brief indicationof whethera man who says(today) Ilarold W'ilson is a greal man and anotherwho says(also today) Ilrr' Briti.shPrinr( lllinisttr is u ,qrcal man would,if eachknew that the two singular have said the terms had the same reference, is made decision whatever But same thing. that I am about this question.the apparatus ofaccounting about to providewill be capable on the depend might that implicatures lbr any presence of one rather than anotherof these uttered.Such singularterms in the sentence would merelybe relatcdto differimplicatures enl maxlms. the conventionalmeaningof In somecases the words usedwill determinewhat is impli-

307 cated. besideshelping to determine what is said. If I say (smugly).He i.t an Englishman; he i.g.therelitre,brave.I have certainly committed myself.by virtue of the meaningof my that his beingbrave words,to its beingthe case of (followsfrom) hisbcingan is a consequence But while I havesaid that he is Englishman. and said that he is brave.I do an Englishman. the lavored not want to saythat I have,raful(in that it follows from his being an En' sense) glishman that he is brave,thoughI havecerthat this is and so implicatcd. tainly indicated, of so. I do not want to say that my utterance false .speaking. would be, .stric!1.1, this scntence in questionlait to should the consequence arc conventional, hold. So .rolrcimplicalures thisdisunlikethe onewith which I introduce cussion of implicature. of nona ccrtainsutrclass I wish to represcnt which I shall call implicatures, conventional (rrnt,u.\ulionulimplicatures,as being essentially connectcdwith certain gcnerall-eatures ol'discoursc: so my ncxt stcp is to trv to say what these featuresare. The lollowing may provide a hrst approximation to a gencral do not normally principle. Our talk exchanges reof disconnected consistol'a succession marks.and would not bc rationalif thcy did. at to somedegree They are characteristically. efforts:and eachpaflicipant least, cooperative in them, to somc cxtent.a comrecognizes or at leasta mon purposeor set of purposes, 'I'his purpose or direction. mutuallyaccepted (c.9.'by start from the fixed may be direction an lnttral proposalot a question lbr discuslt sion).or it may cvolveduringthe exchange: may be lairly dcfinite,or it may be so indcfilatitude to nite as lo leavevery considerable conversalron)' (as in a casual the participants But at each stage,.rrrne possibleconversational moveswould be excludedas conversaWe might then formulate tionatlyunsuitable. a rough generalprinciplc which participants will be expecled (celcris paribus) to observc' contribunamely:Make your conversational at which tion such as is required.at the stage purpose or direction by the accepted it occurs, in which you areengagcd' o[the talk exchange Principle One might labelthis the Cooperative (CP). that somesuch general On the assumption one may pcrprinciple as this is acceptable' under one or categories four hapsdistinguish

308

AND RETEVANCT IMPTICATURE CONVERSATIONAT

anotherotwhich will fall certainmore specific maxims and submaxims' the lollowing of which will, in general,yield resultsin accorPrinciple'Echoing dancewith the Cooperative categories Kant, I call these Quantity,Quality, of QuanRelation.and Manner.The category to the quantity of informationto be tity relates provided.and under it fbll the followingmaxlms:

supermaxim-'Be perspicuous'-and various maximssuchas: of expression. | . Avoid obscurity ambiguity. 2. Avoid prolixityl. unnecessary 3. Bebrief(avoid 4. Beorderly. And one might needothers. of someof It is obviousthat the observance than urgency maximsis a matterof less these asis reasinformative yourcontribution l. Make a man who hasexis the observance ofothers: ofthc exchangc). quired (for the currcnt purposes pressed with undueprolixitywould.in himself more informayour contribution not makc Do 2. general.be open to milder comment than tive than is rcquired. he bcwould a man who hassaidsomething it might be lelt that false. Indeed. lieves to be (The second maxim is disputable:it might bc of least lrrst maxim the importance the of at transsaid that to be overinformative is not a not be includcd is it should such that Quality gressionof the CP but merely a wasteof time. in a scheme of the kind I am constructing; However. it might be answered that such overother maxims come into operatron on onl-v' it is in that may be confusing inlormativeness t h e t h a t t h i s m a x i m o l ' Q u a l i t yi s a s s u m p t i o n and there may also liablc to raise side issues: While lar satistred. may as this be correct. so be an indirect cffcct. in that thc hcarcrs may generation thc implicaturcs is it of conccrncd i s t h a l l h e r c b e m i s l e da s a r e s u l lo l ' t h i n k i n g play seems role not to a totally fionr ditlerc-nt some particulat ytint in the provision of the t h c o t h e rm a x i m s a . nd it will bc convenrcnl. cxccssof intirrmaticln. Flowever this may be. lbr thc present at least, to trealit asa nrenrbcr there is perhaps a difl'erent reason lor doubt of list the of maxims. about thc admission of this sccond maxim. Thereare.of coursc. all sortsof othcr maxnamely. that its cffcct will bc sccurcd by a latcr (acsthetic. ims social. or moral in charactcr). maxim, which conccrns relevancc.) suchas 'Bc polite'.that are alsonormallyob[Jnder the categoryof Quality hlls a superserved by participants in talk cxchanges, and maxim-'Try to make your contribution onc thcsc may alsogencrate nonconventional imthat is truc'-and two more specificmaxims: plicatures-'l.he convcrsalirtnal maxims.however. :rndthectlnversational inrplicaturcs conto bc lalsc. l. Do not saywhat you belicve nectcd with thcm. arc spccially (l connccted l. Do not say that lilr which you lack adequatc hope) with the particularpurposes that talk cvidcnct(andso,talk e'xchangc) is adaptcd to servc and tinder the category()l'Rclation I placea single primarily employcdto servc.I havestated is 'fhough 'Be relevant.' maxim, namely. the m y m a x i m sa s i l ' t h i s p u r p o s c w c r cI m a x i maxim itsellis tcrsc, its formulation conccals mally clli'ctivecxchangc of information: this a number of problems that exerciseme a good specification is, ofcourse,too narrow.and the dcal: questionsabout what diflbrcnt kinds and schemencedsto be generalizcd to allow lirr lbcuses of relevance there may bc. how these purposes suchgcneral asinllucncing or directshift in the course of a talk exchangc.how to i n gt h ea c t i o n s ofothcrs. allow for thc fact thal subjects ofcclnversation As one of my avowcdaims is to seetalking are legitimalely changed.and so on. I lind the as a special caseor varictyol'purposive. intrealment of such qucstions exceedinglydifltdeed rational.behavior.it may bc worth notcult. and I hope to revert to them in a latcr ing that the specilic expectations or prcsumpwork. tions connectedwith at least some of the Finally. undcr the category ol Manner. lbregoing maximshavetheir analogues in the which I understand as relating not (like the sphereof transactions that are not talk exprevious categories)to what is said but, rather, changes. I list brieflyone suchanaloglbr each to /rou'what is said is to be said. I include the conversational category.

TOCICAND CONVERSATION |. Quontit"v.lf you areassisting me to mend a car. I expectyour contribution to be neither more nor less than is required.Il. for example. al a particularstageI need four screws, I expecl you to hand me four. ratherthan two or six. 2. Qualitv. I expectyour contributions to be genuine and not spurious. Ifl needsugar asan ingredient in the cakcyou are assisting me to make.I do not expect you to hand me saltlif I needa spoon. I do not expecta trick spoon madeof rubber. 3. Rclatitm I expecta partner'scontribution to be appropriate to the immediateneeds at eachstage of thetransaction. lf I am mixing ingredients for a cakc. I do not expectto be handcda good book, or evenan oven cloth (thoughthis might be an appropriatc contribution at a laterstage). 4. lllanner. I expecta partner to make it clcar what contributionhc is making and to exccutc his perfclrmancc with rcasonablc dispatch. Theseanalogies are relevantto what I rcgardas a fundamcntal qucstion aboutthe Cl, and its attendantmaxims,namely.what the is lbr the assumption hasis which wc seemto make,and on which (l hopc) it will appcar that a grcat range of implicaturesdcpends. that tafkerswill in general (rctcrispurilrrr.r and in the absencc of indications to the contrary) procecdin the manncr that thcseprinciplcs prescribe. A dull but. no doubt at a cenain level.adequate answer is that it is just a wellrccognized empirical lact that peopledo behavein these ways:they havelearned to do so in childhoodand have not losr thc habit of doingso:and.indeed. it would involvea good dcal o[cffort lo makea radicaldeparture from thc habit.lt is mucheasier. for example, to tcll the truth than to inventlics. I am, howcver.enough of a rationalistto want to {ind a basisthat underlies thesefacts. undeniable thoughthey may be; I would like to be ableto think oflhe standard typeofconversational practicenot merelyas something that afl or most do in lact follow but as something that it is rcasonahle for us to follow.that we shouldzr.rlabandon.For a time. I was attractedby the idea that observance of the Cp and the maxims,in a talk exchange. could be thoughtof asa quasi-contractual matter. with

309 parallels outsidethe realm of discourse. If you pass by when I am struggling with my stranded car, I no doubt havesomedegree ol expectation that you will offbr help. but once you join me in tinkeringunderthe hood,my expectations becomestrongerand take more specilicforms (in the absence of indications that you aremerely an incompetent meddlerh and talk exchanges seemed to me to exhibit. characteristical ly, certai n l'eatures that joi ntly distinguish cooperative transactions: l. The panicipants havesomecommonimmcdiate aim, like getting a car mended: their ultimateaimsmay,of course, be independent and evcn in conflict-each may want to gct the car mendedin order to drive olf. leaving the other stranded. In charactcristic talk cxchanges. thereis a commonaim evcnif. as in an over-lhe-wall chat. it is a second-order one. namcly,that eachparty should.for the tinrc bcing.identifyhimself with the transirory conversational interesls of'the other. 2. l'he contributionsof the participanrs shouldbe dovctailed, mutuallydcpcndent. 3. Therc is some sort of undcrstanding (which may be explicit but which is olrcn tacit)that,otherrhingsbeingequal.thc transaclion should continue in appropriate stylc unless both particsare agrceable that it should terminate. You do not just shovcolr or start doingsomelhing elsc. But whilesomesuchquasi-contractual basis as this may apply to somecases, thcre are too many typesof exchange. like quarreling and letterwriting.that it failslo fit comfortably. In any case, one feelsthat the talker who is irrelcvant or obscure has primarilv let down not hisaudiencc but himsclf. So I would like to be able to show that obscrvance of the CP and maximsis reasonable (rational) alongthe followinglines: that any one who cares aboutlhe goals that are cenlral to conversation/communication(suchas givingand receiving information,influcncing and being influenced by <-rthers) must be expected to havean inlerest, given suitablecircumstances. in participation in talk exchanges that will be profitable only on the assumption that they are conductedin gencralaccordance wirh the CP and the maxims. Whether any such conclusion can be reached, I am uncertain:in any case,I am fairlysurethat I cannotreach it until I am

310

CONVERSATIONAT IMPLICATURE AND RELEVANCE

the competence a good deal clearerabout the nature of releof the hearerto work out, or grasp intuitively, that the suppositionmenvanceand of the circumstances in which it is required. tionedin (2) is required. Apply this to my iniIt is now time to show the connectionbetial example. to B's remark that C hasnot yet tweenthe CP and maxims,on the one hand, been to prison. In a suitablesettingA might implicatureon the other. and conversational rcasonas follows:"(t) B has apparentlyviomay fail to latedthe maxim 'Be relevant'and A participantin a talk exchange so may be ways, whichinclude fulfill a maxim in various regarded as having floutedone of the maxims perspicuity, yet I haveno reason the following: to 4^\oftJoining L He may quietly and unostentatiously suppose that he is opting out from the operaviolalea maxim; if so,in somecases he will be tion of the CP: (2) given the circumstances, I liableto mislead. can regardhis irrelevance as only apparentif, 2. He may opt otil from the operationboth and only if. I suppose him to think that C is of the maxim and of the CPI he may say.inpotentially (3) B knows that I am dishonest: dicate,or allow it to becomeplain that he is capable of working out step(2). So B impliin the way the maxim unwillingto cooperate cates that C is potentially dishonest." requires.lle may say. for example,I canntil Thc presence of a conversational implicasavmorc: mv lips arc sealed. lure must be capableof beingworkcd out; for 3. He may be lacedby a clash:He may be evenifit can in factbe intuitively graspcd. ununable. for example, to fulfill the lirst maxim lessthe intuition is replaceable hy an arguof Quantity(Be as inlormative as is required) ment, the implicature(if prescntat all) will without violating the second maxim of Qualnot count as a conversational implicaturc; it ity (llave adequate evidcncefor what you will bc a conventional implicaturc. To work ev). out that a particularconversalional implica". 4. He may llout a maxim; that is. he may r/ ture is prescnt. thc hearer will rey'ly on thc lblblatanrlyfail to fullill it. On thc assumprion l o w i n gd a t a :( l ) t h e c o n v e n t i o n h m l eaning olthat the speaker is able to lutlill the maxim thc words uscd.togetherwith thc identity of and to do so withoutviolating another maxim any refcrences that may be involved:(2) the (becausc of a clash), is not opling out, and is CP and its maxims:(3) the contcxt,linguistic not. in view of the blatancyof his pcrfor(4) other items or otherwisc. ol'the utterancc: mance.trying to mislead.the heareris faccd ofbackground knowledge; and (5) the fact(or with a minor problem:How can his saying supposed fact) that all rclevantitems lirlling what hc did say be reconciledwith the sup under thc prcvioushcadings are availahle to positionthat hc is observing thc ovcrallCP? both participants and both participants know This situation is onc that characleristicallv or assumc this to be the case. pattcrn A general givcsriseto a conversalional implicature: lor the workingout of a conversational anrl impliwhen a conversational implicatureis gener- caturemightbe givcnas lblklws:"lle hassaid ated in this way, I shallsay that a maxim is that /; thereis no reason to suppose that hc is tring c.rploit<,d. not observing the maxims. or at least thc (-lr. I am now in a position to characterize he could not be doing lhis unless the he thought notion of conversalional implicaturc. that 17; A man he knows(and knowsthat I know thar who. by (in. when)saying(or rnakingas if to hc knows)that I can seethat the supposition say)that 2 has implicatedthat q. may be said that he thinksthat 4 rs rcquircd; hc hasdone to have conversationally implicated that 4, nolhingto slop me thinkingthat 4; he intends provided that ( l) he is to be presumed me to think. or is at leastwilling to allow to be observing the conversational maxims.or al mc to think. that r7: and so he hasimplicated leastthc Cooperative Principlc;(2) the supthar 4." position that hc is awarcthat,or thinksthat,4 is rcquired in orderlo makehissaying or mak[xAMPt-ts or coNvrRsATtoNAt ing as if to sayZ (or doing so in /lro.rc terms) IMPTICATURT consistent with this presumption: and (3) the speaker thinks (andwould expectthe hearer I shall now offcr a number of examples,which to think that the speaker thinks)that it is within I shall divide into three groups.

TOGICAND CONVERSATION no ma-rimis violated, oRoupA: Examples in v,hich or at leail in w,hit'h it is nol dear thatanvmaximis t,ioluted A is standingby an obviouslyimmobilized by B: the followingexcar and is approached changetakesplace: ( l) A: / am ou!til pctrol. roundtht'utrner. B'. I-hcrt'is a garagc (Gloss:B would be infringing the maxim 'Be relevant'unless he thinks.or thinks it possible, is open,and haspetrol to sell; that the garage is, or at least so hc implicatesthat the garage may be open,etc.) In this example, unlike the caseof the rethe unmark Hc hasn't hcen lo prixtn ].'(,1, B's remarkand A's slatedconncctiontretween remark is so obviousthat. even if one inter'Bc perspicpretsthe supermaxim of Manner. uous.'asapplyingnot only to the expression of whal is said but alsoto the connection of what is said with adjacent rcmarks. there seemsbe no case for regardingthat superin this example. maxim as infringed Thc ncxt cxampleis perhaps a little lessclearin this respect:
.stt,rtr ro hovr'r, rirl, cnd thc:;t' 12\ A: Snith dtx,sn't dars B: Ile hu:; lven puinq u ht ol vi.sit.s to Ncv, l'rrk lutaly. B implicates that Smith has. or may have, a girlfricnd in New York. (A gloss is unneccssary in view of'that given for the previous example.) In both examples. the speakcr implicates that which he must be assumed to believe in order to preservethe assumption that he is observingthe maxim of Relation. in x'hith u mu.rim is violated. cRoup B: F).rumple.r b!'tht \uppo.\ttton but it,; t,iolutionis trt ba cxplaine'd ,,1u clu:;htith un(tlh.'r,nu.tim A is planning with B an itinerary lor a holiday in France.Both know that A wants to see his friend C, if to do so would not involve too great a prolongation of his journey: (3\ A: l|'hcrc dttcsC lival' B'. Srnnax'ht:rc in tha Stntthof'['-rance. -l'here (Gloss: is no reason to supposethat B is opting out; his answeris, as he well knows, less

311 informative than is required to meet A's needs. This infringementof the first maxim of Quantity can be explainedonly by the sup positionthat B is awarethat to be more informative would be to say somethingthat infringed the secondmaxim of Quality. 'Don't saywhat you lack adequate for'. so B e.,idence implicatesthat he does not know in which town C lives.) cRot.rp that c: E.rumplas that involva e.rpktilutitm, is,u pnrtdurc hr u'hich is llrrutad.litr the a ma-rim purpo;;t,ol in u convarsational impliruIur(b)' 11ctting
mtunl tl .something ol th<' naturc ol a hgure ol sltau h

In theseexamples. though some maxim is violatedat the levelof what is said.the hearer is entitledto assume that that maxim. or at leastthe overallCooperative Principle,is observed at the levelof what is implicated. (la\ ,l llouting ol the hrst maxim ol Quunlrl.r' A is writinga testimonial abouta pupil who job. and hislctis a candidate lor a philosophy ler reads as lbllows:'Dear Sir. Mr. X's command of Englishis excellent, and his attendance at tutorials has been rcgular. Yours. etc.' (Gloss: n cannot be opting out. sinceif why writc at he wishedto be uncooperative, all? tle cannotbe unable, throughignorance, to saymorc,since the man is his pupil; moreover. he knowsthat more informationthan He must.therelbre. be wishing this is wanted. to impart information that he is reluctantt<r writedown.This supposition is tenable only if he thinks Mr. X is no good at philosophy. '[his. then.is what he is implicating.) Extreme examples of a floutingof the trrst maximof Quantityareprovided by utterances llke lltnnen ere *\,men of patent tautologies and ,lar is v'ar. I would wish to maintainthat at the level of what is said. in my favored sense. such remarksare totally noninformative and so, at that level,cannotbut infringc the firsl maxim of Quantityin any conversational context.They are, of cours,informative at the levelof what is implicated,and the hearer's identilication of their informative on hisability content at thislevelis dependent of this parto cxplain the speaker's selection ticular patenttautology. (lb) An inliingementoJ'thesecondmarim 'Do ol' Quantitt,, not give more information

312

AND RELEVANCE IMPTICATURE CONVERSATIONAT joy' and then the irony interpretant'You are my bane.' Meiosis.Of a man known to have broken a little up all the furniture, one saysHe w'as into-ricatt'd. Hyperbole.Everv nic'egirl lovesa sailor. (2b) Examples maxim in which the second of Quality.'Do not saythal for whichyou lack not is floutedare perhaps evidence'. adequate easyto find. but the following seemsto be a specimen.I say of X's wife. She is probahlv conhim thiso,cning.In a suitable dcc<'iving gesture or toneof voice. text.or with a suitable reason it may be clearthat I haveno adequate for supposing this to be the casc.My partner, the assumption that the conversato preserve that assumes tionalgameis still beingplayed. I am gctting at some relatedpropositionfor theacceptance ofwhich I do havea reasonable 'l'he might well be rclatedproposition basis. her husband. or that sheis givento deceiving possiblythat shc is the sort of pcrson who would not stopshortofsuch conducl. (31 l',vmplc.r in :r'hit'hun itnplttuturctt hv rul. u.rtli.rtinttlrrtnt ultpurorl, t'iu<'hievcd ()luli()n ttl llrc tnu.tirrtttl Relutirtnare perhaps rarc.but the lbllowingsccnls1obc a goodcandidatc.At a gcntecl tca party.A says ,t/rs..\'i r 'Iherc un rild bug is a momcnt of appalledsilcnce.and then B says7h{, A't,uthuhus heen tlutte ddightlirl this .urmmer,husn't rll B has lllatantlyrefused to make what he saysrelevanl to A's preceding remark.He thereby implicates that A's remark should not be discusscd and. pcrhaps morc spccifically. that A hascommitteda sociaigafl'e. (4) F,.uumplc: ntu.rims.lullin v'hicltuarittu.t ittg ttntlcr lltt, .rltpermu-ritnflc pcr.rpittutu.t urt lltnrtcd We mustrcmember .lurbigttit.t'. thal we are conccrncd only with ambiguitythat rs delibintends erate. and that the spcaker or expects 'l'he problcm to be recognized by his hcarcr. thc hcarcr has lo solvc is why a spcakcr should.whcn slill playingthc convcrsalional game.go out ol'his way to choosc an ambiguousutterance. Thercarc lw() typcsof cases: (a) Examplcs in whichthereis no diflbrence. rlr no strikingdiflbrence. between two interpretations ol' an uttcrancc with respect to straightlorwardne'ss: neither interprctation is notably more sophisticated. less slandard.

lhat the on the assumption than is reqLtired', af sucha maxim should be admitted existence A wants to know whetherp, and B volunteersnot only the information that p, but information to the effectthat it is certainthat p, that for its beingthe case and that the evidence p is so-and-so and such-and-such. and if it B's volubility may be undesigned. by A it may raisein A's mind a is so regarded doubt as to whetherB is as certainas he says he is ('Methinks the lady doth protest too it much'). But if it is thoughtof as designed. would be an obliqueway of conveyingthat it whetheror not controvcrsial is to somedegree that suchan impliarguable lr. It is. however. lo the caturecould be explainedby reference maxim of Rclationwithout invoking an alleged maxim of Quantity. second (2a) l:..ramples in u'hit'hthc /ir,tt tnaxim ttl Quulitv i.slknucd Inn.l'. X. with whom A has bcen on close a sccrctof A's termsuntil now. hasbetraycd both rival A and his audience to a business ((iloss:lt know lhis. A says.t' i.su.lrnc.lricnrl. is perl'ectly ohvious to A and hisaudience'that what A has said or has made as if to say is hc doesnot believe. and the audisomclhinB, cncc knowsthal A knowsthat this is obvious is enSo.unlcss A's utterance to the audiencc. A musl be trying, to gel across tirclv pointless. than the one he pursomeother proposition 'Ihis must be lx)rts to be putting forward. somc obviouslyrclatedproposition:thc rnost proposition is thc contradicrclatcd obviously tory ol'thc onc hc puqrorlsto bc puttinglirrward.) l'ldulh()r lixamplcslike l'orrur( th( uurm involvecategoi tr trt.1' utllt'acharacteristically rial falsity.so the contradictory of what the spcakcrhas made as if to say will. strictly speaking, bc a truism;so it cannolbc llal that -l-he is trying to gct across. sucha spcakcr most is that the speaker likcly supposition is attributing to his audiencc somelbature or l'eaturcs in respectof which the audienceresemblcs (morc or lcsslancil'ully) thc menlionedsubstancc. It is possiblcto combine metaphorand irony by imposing two stagcs of on the hearer intcrpretation.I say lrrrr ur., !hc crcant in m.t' rrrli,c. intendingthe hcarerto reach first the 'You are my prideand interpretant mctaphor

TOCICAND CONVERSATION than the more reconditeor more far-fetched 'Never Blake's lines: other.We mightconsider seekto tell th], love, Love that nevertold can by introduced be.'To avoidthc complications of the imperativemood. I shall the presence sentence. Lunryht to tell the related consider mr lova.lovethat nc'tcrlold can Dt'.Theremay refer bs a doubleambiguityhere.Mv fuuc'may to either a state of emotion or an object of cmotion. and /ovc lhat ncvar!old run be may meaneither'Love thatcannotbe told' or'love that if told cannot continueto exist.' Partly of the poet and of thc sophistication hccausc (that the parlly because of internalevidence is kept up).thereseems to bc no alambiguity are that the ambiguities lernative to supposing both and thal the poetis conveying deliberate what hc would be saying if ont: interpretation wcrc intcndcdrathcrthan thc othcr.and vice vcrsa; thouBhno doubt thc pocl is not explicitly sayingany onc of thesethingsbut only them (cl. 'Sinccshe or suggesting conveying Inaturc] prickedthec out lirr women'spleasurc.minc be'thylovc.anclthy lovc'susetheir trcasurc'). is in n s w h i c ho n e i n t e r p r e t a t i o th) F.xample notably lcss straightforwardthan anothcr. 'I'ake of the British Gcnthecomplcxcxamplc cral who captured the provinceot'Sind and I'ct'L'uvi. Thc ambiguity nl backthe message se ('l havcSind'/'l havesinned') is phoinvolved acand thecxprcssion ncmic.n()tmorphcmic: it is in a but since tuallyuscdis unambiguous. translanguage lbrcign to speaker:rnd hearer, lation is calledlbr. and the ambiguityresides into native English. translation in thc standard Whcther or not the straighlforwardinterit prctant('l havesinned')is bcingconveyed, interpresecmsthat the nonstraightforward 'l'here might be stylistic reasons tant musl be. lbr conveying merelyits nonby a sentence intcrpretant. but it would be straightforward pointless, and perhaps also slylisticallvobiectionable,to go to the trouble of frndingan conveys thal nonstraightforwardly expresion theeffort on an audience thatp, thusimposing if this inthis interpretant, involvcdin frnding tcrprelantwcrc otioscso lar as communicaWhetherthe straightfortion was concerned. ward interpretant is also being conveyed to dcpendon whelhersucha supposiseems tion would conflictwith other conversational

313 requirements. for example,would it be relecould vant. would it be something the speaker be supposed to accept,and so on. If such requirementsare not satisfied. then the straighrlf forward interpretantis not beingconveyed. could they are, it is. If the author of Pct't'avi to think that he had naturally be supposed lor excommittcd somekind of transgression. his ordersin capturing ample. had disobeyed to such a lransgresslon Sind. and if rel'erence interests o[ would be relevantto thc presumed then he would havebeenconthe audience. he would vcyingboth interpretants: otherwise he convcving only the nonstraighttbrward one. How do I exploit, lirr thc purOh.rutritv. posesof communication.a deliberatcand overt violation of the rcquirementthat I il'thc CoObviously. shouldavoidobscurity'l I mustintend is to operatc, opcrativc Principlc what I am saying my partncrto understand thc ohscurityI import rnl(r my ultcrdespite that A and B arc havinga conarrcc. Suppose vcrsation ol'a third party,tbr in thc prescncc a child,thenA mightbc deliberatcly cxamplc, obscure. though nol too obscurc.in the hope and the third party that B would undersland B to sccthat A if A expects not. l"urthcrmore, reasonit sccms obseurc. is bcingdelibcratcly that, in nrakinghis convcrsaableto suppose tionalconlributionin this way,A is implicating that the contcntsol' his communication shouldnot bc impanedto the third party. . C'ompare thc f'uilurc Io ltc hricl rtr nu't'inct remarks:
(x) ,l1l.r.r\' .sung'llttnr Su'tcl Ilttnu' (b) tlr.rr .\' pnulttttrl o s(rt(.\ ttf .stntnd: lltttl utrn' 'llrnnr '\ttt't'l .tptuulul t ltt:dy lith lht' .sutn' ttl ' I lrnttt

that a reviewerhaschoscnto ultcr Suppose (b) rathcrthan(a).(Gloss: Why hashe selcctcd that rigmarolein place of the conci*- and to tnPresumably. .ran,g,) nearlysynonymous betweenMiss dicate some striking difference to which the wclrd and those X's performance The most obvious is usuallyapplied. sing,ing suf: is that Miss X's performance supposition from somehideousdefect.The reviewer f'ered is what is liketyto knowsthat this supposition springto mind. so that is what he is implicating.)

314

CONVERSATIONAI. IMPLICATURE AND RETEVANCE of expression arrX, implicates that the X does not belongto or is not otherwise closelyconnectedwith some identifiablepenon, the implicature is presentbecause the speakerhas failedto be specific in a way in which he might have been expectedto be specific.with the consquence that it is likely to be assumed that he is not in a positionto be specific. This is a familiarimplicature situation and is classifiableasa failure.for one reason or another. to fulfillthe firstmaximof euantity.The only diltcult questionis why it should,in certain cases.be presumed.independentlyof information about particularcontexts of utterance. that specilication of the closeness or remote_ nessof the connectionbetweena particular personor object and a l'urtherperconwho is mcntioned or indicated by the utterancc shouldbe likely to be of interest. The answer musl lie in the following region; Transactions betwecna personand other persons or things closelyconncctcdwith him are liable to be very different as regardstheir concomitants and results from the sames()r1 ol'transactions involving only rcmotelv connecterJ Dersons or things:the concomitants and rcsuits. lirr in_ stancc. of my finding a hole in my rool'arc likcly to be verv diftL'rcnt liom the concomitantsand rcsults of my findinga holcin somc_ one else'sroof. Information.like moncv. is oflengivcnwithoutthegivcr's knrlwing tojust whal usethe recipicntwill wanr to puf ii. tt' someone lo whom a transaction is ntcntionecl givesit lurthcr consideration. he is likelv to l i n d h i m s e l fw a n t i n gt h e a n s w c r s ro lunhcr questtons that the speaker may not be ablc to rdenlifyin advance: if the appropriatc spccificationwill be likelyto enable the hearcrioan_ swera considerable variclyol'suchquestions ftrr himscll.thcn therc is a presumption that the speaker shouldincludeit in his rcmark:if no1,thenthereis no suchpresumption. Finally,we can now show thal. convcrsationalimplicature beingwhatit is.it nrustpos_ sess certainfeatures: L Since,to assume the presence of a con_ versational implicature. we have to assunle principle is being that at leastthe Cooperative observed, and sinceit is possible to opt out ol theobservation of thisprinciple, it lollows that a generalized conversational implicaturecan be canceled in a particularcase.lt may be explicitly canceled, by the additionof i clause

CENERATIZED CONVERSATTONAT IMPIICATURE


I have so far considered only cases of what I might call'particularized conversational implicature'-that is to say. casesin which an implicature is carried by sayingthat p on a particularoccasion in virtue o[specialleatures of the context.cases in which thereis no room for the idea that an implicatureof this sort is normally carried by sayingthat p. But there are cases o[ generalized conversational implicature.Sometimes one can saythat the useof a cerlainform of wordsin an utterance would normally (in the absence of special circumstances) carry such-and-such an implicature or type of implicature.Noncontroversial examples are perhaps hard to find, sinceit is alt too easyto treat a generalized conversational implicature as if it wcrea conventional implicalure.I offer an examplethat I hope may be lairly nonconlroversial. Anyone who uses a sentence of the lorm .{ ts tttt'ctittgu troman this cvenin.g would nor_ mally implicate that the person to be met was someone other than X's wife. mother,sister. or perhapscven close platonic friend. Similarfy. if I wereto sayX u.ctr! into a hunt vcstcrdu.r utul limnd a !ttrttti.st'in:;itlt tht lrrnt dtxr. my hearerwould normally be surpriscd if somctime latcrI reveated that the house was X's own. I could produccsimilar linguistic phen<rmena involvingthe cxpressions u Rardcn. a tur. a utllcgc. and so on. Sometimes. however.therc would normally bc no such implicature ('l have beensitting in a car all morning').and sometimcs a rcverse implica_ ture ('l brokc a fingcr ycsterday'). I am inclinedto think that onewould not lenda svm_ patheticcar l() a philosopher who suggeired that there are threc senses of the form of ex_ prcssion an i';one in which it means roughly 'something that satisfies the conditions delin_ rngthe word.Y,'another in whichit means ap 'an proxlmately X (in the lrrst sense) that is only remotcly relaled in a cerlainway to some personindicated by the context.'andvel an_ other in which it mans.an X (in fhe first sense) that is closelyrelatedin a certainway to someperson indicated by the context.'Would we not much preferan accounton the following lines (which, of course.may be incorrect in detail);When someone. by usingthe form

TOCICAND CONVERSATION that statesor implies that the speakerhas canceled. out. or it may be contextually opteci that usuallycarriesit if the form of utterance is usedin a contextthat makesit clearthat the is oPtingout. speaker that a particular 2. Insofaras the calculation requires. implicatureis present conversational contextualand backgroundinformabesides of what hasbeensaid tion. only a knowledge of the utcommitment (or of theconventional as the manncro{'expresand insofar tcrance). it will not no rolein thecalculation. sionplays to hnd anotherway ol'sayingthe be possiblc samething. which simply lacksthe implicaexceptwhcre some special ture in question. is itsell-relcversion of thc substituted f'eature of an implicature vant to the cletermination l[ (in virtueof one of the maximsof Manner)onc may nondetachability' we call this feature implicaconversational expccl a gencralized loture that is carriedby a lamiliar. nonspecial cution to have a high dcgreeof nondetachability, sinccthc calcuapproximately. J. To speak tmof a convcrsational lation of the prcsence ol' prcsupp()scs an initiat knowlcdge plicaturc the the conventionallbrce of the exprcssion a thc implicature. of which carries uttcrance a condition will be implicatum conversational

315 that is not included in the original specificaconventional lbrce' tion of the expression's Though it may not be impossiblefor what imas a conversational startslife. so to speak. to supplicatureto becomeconventionalized. posethat this is so in a given casewould rejustification. So' initiallyat least' quirespecial implicata are not part of the conversational to the employmeaningof the expressions attach. which theY ment of impli4. Sincethe truth of a conversational by the lruth of what rs catum is not required said(whatis saidmay be true-what is impliis not carthe implicaturc catcdmay hc lalse). of ried by what is said.but only by the saying 'puttingit that way.' what is said,or by ima conversalional 5. Since.1o calculate bc supto whal has plicaturc is to calculale that the supposilion posc<l in ordcr to prescrve the CooperativePrinciple is being observed' sp'cand sincc there may be variouspossible a listof whichmay beopen' cilicexplanations. implicatumin such cascs thc conversational explanawill bc disjunctionof such specilic tions:and if the list of theseis open.thc inrplicatumwill havejust the kind of indetermtnacy that rnany actual implicatado in lact seemto possess.

Anda mungkin juga menyukai