Clark and Carlson 1982 Hearer Ans SA

7/28/2019 Clark and Carlson 1982 Hearer Ans SA

1/43

Linguistic Society of America

Hearers and Speech ActsAuthor(s): Herbert H. Clark and Thomas B. CarlsonReviewed work(s):Source: Language, Vol. 58, No. 2 (Jun., 1982), pp. 332-373Published by: Linguistic Society of AmericaStable URL: http://www.jstor.org/stable/414102 .

Accessed: 10/03/2013 08:23

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .http://www.jstor.org/page/info/about/policies/terms.jsp

.JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of

content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms

of scholarship. For more information about JSTOR, please contact [email protected].

.

Linguistic Society of America is collaborating with JSTOR to digitize, preserve and extend access toLanguage.

http://www.jstor.org

This content downloaded on Sun, 10 Mar 2013 08:23:39 AMAll use subject to JSTOR Terms and Conditions
http://www.jstor.org/action/showPublisher?publisherCode=lsahttp://www.jstor.org/stable/414102?origin=JSTOR-pdfhttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/stable/414102?origin=JSTOR-pdfhttp://www.jstor.org/action/showPublisher?publisherCode=lsa


2/43

HEARERS AND SPEECHACTSHERBERTH. CLARKand THOMASB. CARLSON

Stanford UniversityIn conversations involving more than two people, most utterances are intended to beunderstood not only by the people being addressed, but also by the others. Theseutterances cannot be accounted for in current theories of speech acts unless severalbasic changes are made. In our proposal, the speaker performs two types of illocutionaryact with each utterance. One is the traditional kind, such as an assertion, promise, or

apology; this is directed at the addressees. The other, called an informative, is directedat all the participants in the conversation-the addressees and third parties alike. It isintended to inform all of them jointly of the assertion, promise, or apology being directedat the addressees. We present evidence that every traditional illocutionary act is per-formed by means of an informative.*Althoughhearersplay an essential role in speech acts, that role has neverbeen fully examined. Considerrequests, such as this one from 'Othello':

(1) Othello, to Desdemona, in front of lago and Roderigo: Come,Desdemona.In Searle's 1969 theory and its descendants-the STANDARDHEORIESs wewill call them-Othello's request 'counts as an attemptto get H to do A'. Itis an attempt by Othello to get the 'hearer'H to go with him. This, of course,is incorrect:by 'hearer', Searle really means 'addressee'.' AlthoughOthellohas an audience of three 'hearers'-Desdemona, Iago, andRoderigo-he isn'ttryingto get all three of them to go with him. His request is for Desdemonaalone. She is anaddressee,notjust a hearer.The standardheoriesare theoriesabout illocutionaryacts directedat addressees.Are there illocutionaryacts directedat hearerssuch as lago and Roderigo?The standard heories, by their silence on the question, appearto assume theanswer is no.2 Thistoo seems incorrect.AlthoughOthello sn'taddressing agoand Roderigo, he intends them to understandwhat he is saying. Indeed, heintends them to understand n the same way that he intends Desdemonatounderstand-by means of their recognitionof his intentions,just as theories

* The work reported here was supported in part by Grant MH-20021 from the National Instituteof Mental Health. For advice and suggestions, we are indebted to many colleagues, especiallyKent Bach, Keith Brown, Eve V. Clark, Helen B. Clark, David Evans, Charles Fillmore, JamesFox, Gerald Gazdar, Jerry Hobbs, Paul Kay, George Lakoff, A. P. Martinich, Jerry Morgan,Geoffrey Nunberg, Mary Louise Pratt, Ivan Sag, and Robert Stalnaker. A version of this paperwas presented at the Stanford Pragmatics Workshop, Asilomar, California, June 1980.

Searle is in good company. Austin 1962, Bach & Harnish 1979, Bennett 1973, Chomsky 1975,Davison 1975, Fraser 1975, Garner 1975, Gordon & Lakoff 1971, Kempson 1975, 1977, Lewis1969, and Morgan 1977 all use 'hearer' for 'addressee'. Donnellan 1968 and Grice 1968 refer to anundifferentiated 'audience'. Others, including Fillmore 1972, Green 1975, and Katz 1977, haveused 'addressee'-though still others, such as R. Lakoff 1972 and Ervin-Tripp 1976, have usedthis interchangeably with 'hearer'.2 Allusions have been made to the effect of a third party on the significance of a speaker'sutterance to an addressee (Bird 1975, Rubin 1978, Verschueren 1978), but no discussion of illo-cutionary acts toward these third parties has taken place (see also fns. 9 and 10, below).

332

This content downloaded on Sun, 10 Mar 2013 08:23:39 AMAll use subject toJSTOR Terms and Conditions
http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp


3/43

HEARERS AND SPEECH ACTSof illocutionary acts require. The difference is that what lago and Roderigo areto understand is not that they are to go with Othello, but that he is requestingDesdemona to go with him. As a first conjecture, then, Othello is performingillocutionary acts directed at all three hearers. However, the ones he is directingat lago and Roderigo aren't the same as the ones he is directing at Desdemona.In this paper, we argue that this conjecture is correct: Speakers performillocutionary acts not only toward addressees, but also toward certain otherhearers. We define a type of hearer we call a PARTICIPANT,hose role as heareris distinct from the roles of both addressee and overhearer. In ex. 1, Iago,Roderigo, and Desdemona are participants; Desdemona is also the addressee.Then we take up three hypotheses:

(2) THEPARTICIPANTYPOTHESIS.ertain illocutionary acts are directedat hearers in their roles as addressees, and others are directed athearers in their roles as participants.The first class, called ADDRESSEE-DIRECTEDLLOCUTIONARYCTS, includes allthe familiar illocutionary acts such as assertions, requests, promises, and apol-ogies. It is the second class, called PARTICIPANT-DIRECTED LLOCUTIONARYACTS,that is new.

(3) THEINFORMATIVEYPOTHESIS.he fundamental kind of participant-directed illocutionary act is one by which the speaker jointly in-forms all the participants fully of the illocutionary act that he issimultaneously performing toward the addressee or addressees.These illocutionary acts will be called INFORMATIVES. This leads to the thirdhypothesis:

(4) THE INFORMATIVE-FIRST HYPOTHESIS.All addressee-directed illocu-tionary acts are performed by means of informatives.

By this hypothesis, Othello makes his request of Desdemona by means of aninformative directed at lago, Roderigo, ANDDesdemona.3This proposal has far-reaching consequences for speech-act theories. Eversince Austin 1962, the act of speaking has been divided by levels into componentacts that are causally related: phonetic acts, locutionary acts, illocutionaryacts, and perlocutionary acts, among others.4 Our proposal is to add a newlevel-a new component act-to this causal chain. Consider an analogy (cf.Austin, 107): in shooting a gun, a person tenses certain muscles, by means ofwhich he crooks his right index finger, by means of which he pulls the trigger,by means of which he shoots the gun. If no one had described the act ofcrooking the right index finger, which is a necessary component of this causalchain, then the theory of gun-shooting would be incomplete. For speech acts,since no one has described the act of informing participants, which is just asnecessary a component of speaking, speech-act theories are likewise incom-

3 Note that the third hypothesis presupposes the second, which in turn presupposes the first.So it is possible to accept the first hypothesis alone, the first two alone, or all three.4 By a 'causal' relation, we mean a 'by-means-of relation. For discussion, see Austin 1962,Grice 1968, Goldman 1970, and our ?5.

333



4/43

LANGUAGE, VOLUME 58, NUMBER 2 (1982)

plete. They need to be amended by a new level of component acts, theinformatives.We take up the hypotheses roughly in order. In ?1, we describe situationsthat appear to require participant-directed illocutionary acts. In ?2, we describethese situations more completely. In ?3, we argue that informatives are a typeof illocutionary act. In ?4, we lay out the evidence for the informative-firsthypothesis. In ?5, we show how this analysis accounts for two kinds of indirectillocutionary acts. In ?6, we take up a family of participant-directed illocu-tionary acts called partial informatives. In ?7, we analyse how informativesare used in deception. In ?8, we summarize.1. FIVEPROBLEMS.re hearers like Iago and Roderigo truly the targets ofillocutionary acts? We will describe five problems for the standard theoriesthat appear to be resolved only if the answer is yes. For each problem, we willconsider examples in which the speaker performs a traditional illocutionary acttoward one hearer, and must be assumed at the same time and with the sameutterance to be informing other hearers of that act. These other hearers willbe called SIDE-PARTICIPANTS,nd the acts of informing them INFORMATIVES.1.1. CONVERSATIONS. n ordinary conversations, the information that theparties acquire accumulates in a principled way (see Gazdar 1979, Stalnaker1978). Imagine Ann, Barbara, and Charles in a conversation. When Ann asksBarbara a question, Charles is expected to keep track of that question, eventhough he is not being addressed. And when Ann asks HIMa question, he isexpected to keep track of the fact that Barbara-and Ann-are keeping trackof this question too. That is, the parties to a conversation generally adhere toa PRINCIPLEFRESPONSIBILITY:ach is responsible at all times for keepingtrack of what is being said, and for enabling everyone else to keep track ofwhat is being said. Each party keeps a cumulative record that becomes partof everyone's COMMONROUND,n the technical sense of Karttunen & Peters1975, Stalnaker 1978, and Clark & Carlson 1981. With each contribution to theconversation, the current speaker presupposes the common ground already

established; and all the parties, the speaker included, add what is new in thatcontribution to their common ground.The problem this poses for the standard theories is that speakers cannotfulfill this responsibility without the use of informatives. Consider this example:(5) Charles, to Ann and Barbara: What did the two of you do today?Ann, to Charles, in front of Barbara: We went to the museum.Barbara, to Charles, in front of Ann: Before that, we went to thetheater.

Here Ann and Barbara take turns telling Charles what they did together. WhenAnn asserts that they went to the museum, she is addressing Charles. Shecan't be addressing Barbara, since she would be telling her something it wasobvious to the two of them that Barbara already knew (cf. Searle 1969, Stal-naker 1978). Yet if the conversation is to accumulate, Ann must let Barbaraknow what she is telling Charles. Otherwise, Barbara cannot keep track ofwhat is being said; she may repeat things Ann has already said. So Ann must

334



5/43

HEARERSAND SPEECHACTSinform Barbarathat she is telling Charles that they went to the museum.Barbarapresupposesjust that when she says Before that, we went to thetheater.If BarbaraunderstandsEnglish, how could Ann fail to let her know whatshe was telling Charles?There are manyways. Ann could have told CharlesWe went to the place you recommended to me this morning, knowing thatBarbara had no idea he had recommendedthe museum. In that case, Annwould have good reason for believingthat Barbaracould NOTdeterminewhatshe was telling Charles and therefore would not be able to buildon what shehadjust said: Barbara might mistakenly go on And then we went to the museum,duplicatingAnn's contribution.So, to fulfill her responsibility,Ann must domore than tell Charlesthey went to the museum. She must tell him in such away that she simultaneously nforms Barbaraof what she is tellinghim.The need for informatives s especially clear for certaintypes of ellipsis inconversation;e.g.,

(6) Charles, to Ann, in front of Barbara: Did you like the museum?Ann, to Charles, in front of Barbara: Yes, I did.Charles, to Barbara, in front of Ann: What about you?Barbara, to Charles, in front of Ann: I liked it too.WhenCharles asks Ann his question, he must also be informingBarbarawhathe is askingAnn. Otherwise,he cannot be certainthat Barbarawillunderstandhis highly elliptical question, What about you? Imagineinstead that Charleshad asked Ann Did you like the place I recommended this morning?, knowingthat Barbarawas not privy to that conversation:he couldn't then expect herto understandWhataboutyou? So Charlesmust do two thingsin utteringDidyou like the museum?:(a) ask Ann whether she liked the museum, and (b)informBarbara that he is asking Ann this. Only then will his next ellipticalquestionbe felicitous.The necessity for informatives s also obvious wheneverthere is a changeof addressee inthe middleof anutterance.Thisoftenoccurs withtagquestions,as in this example from GrahamGreene's Thehuman actor:

(7) Hargreaves, to Percival, infront of Daintry: Daintry began his checkwith those, [turning o Daintry]didn'tyou?In the first half, Hargreavesis makingan assertionto Percival. Yet he mustalso be informingDaintry of that assertion; otherwise, he couldn't turn toDaintryand ask himto confirm t. Daintry,for example,mustunderstandwhathis check and those referto. Withthe tag question, thereis a similarproblem.As Millar&Brown(1979:44)pointout forsimilarScots examples:'Thespeakeris not seeking confirmation of his propositionfor his own sake-he is notuncertainof the truthof his proposition-but for the sake of the thirdparty.'Withhis tagquestion, Hargreaves'main aimis not to askDaintry he question,but to inform PERCIVALhat he is asking Daintry that question. Informativesare essential in both halves of his utterance.

1.2. INDIRECT LLOCUTIONARYCTS.In the standard theories, the speakercan performone illocutionaryact toward an addressee and thereby perform

335



6/43


another toward the same addressee. The first is called a direct illocutionaryact; the second, an indirectillocutionaryact (Bach & Harnish1979,Morgan1978, Searle 1975).Take this example:(8) Ann, to Barbara, in front of Charles: Barbara, I insist that you tellCharles who we met at the museumtoday.Ann is directly assertingthat she insists that Barbara ell Charlessomething,and she is indirectly requestingBarbara o do this. Both the assertionand therequestare addressed to Barbara.Now considerthis:(9) Ann, to Charles, in front of Barbara: Charles, I insist that Barbaratell you who we met at the museumtoday.

AgainAnn is directly assertingthat she insists that Barbara ell Charles some-thing,and she is indirectlyrequestingBarbara o do this. In thiscase, however,the assertion is addressed to Charles,and the requestto Barbara.Thus, thereare two distinct types of indirectillocutionaryacts: those like 8, in which thedirect andindirect addresseesare the same;andthose like9, in whichthey aredifferent. These two types will be called LINEAR and LATERALndirect illocu-tionary acts, respectively. The indirect addressees in the two types will becalled LINEARnd LATERALddressees, respectively.Lateralindirectillocutionaryacts pose a problemfor the standard heories.In 9, the standardtheories say that Ann's only direct illocutionaryact is theassertion to Charles. But if she isn't performingany illocutionaryact towardBarbara,how is it possible for her to performan INDIRECTllocutionaryacttoward Barbara-the indirect request? In the standardtheories, all indirectillocutionaryacts are performedby means of direct ones aimed at the samehearer.These theories predictthat the lateralrequestin 9 is not possible.Onesolutionis to denythat Ann's request n 9 is indirect.Instead,one wouldsay that she is performing wo DIRECTllocutionaryacts-an assertion towardCharles, and a request of Barbara.There are two drawbacksto this. One isthat, in9, Annis clearlyaddressingCharles,notBarbara.She uses the vocativeCharles;and she speaks of Charlesas you, but of Barbaran the thirdperson.This solutionwouldrequirea drasticrevision of the notion of addressee. Thissolution also obliterates the parallelsbetween 8 and 9: in 8, Ann's request isclearly indirect;in 9, it shouldbe indirect for the same reasons.A second solution is to dropthe 'indirectperformance riterion' hatindirectillocutionaryacts must be performed by means of direct illocutionaryactsaddressedto the same hearer or hearers. This move, however, would be self-defeating.In Searle's 1975 heory, it is thiscriterion hatmakes anillocutionaryact indirect.5 If it were dropped, indirect acts would become formallyindis-tinguishable rom directacts, andmuchof the motivation or the theorywoulddisappear.

A thirdsolution is to bringin informatives. In uttering9, Ann is informingBarbaraof her assertion to Charles, and she is thereby makingher indirectrequestof Barbara.This solutionretainsthe idea thatAnn'srequestof Barbara5 Cf. the inclusion principle discussed in ?4, below.

336



7/43

HEARERS AND SPEECH ACTSis indirect. It also retains the indirectperformancecriterion,since Ann's in-directrequestis performedby means of a direct nformativeo the samehearer.The analysis which we will offer later differsfromthis solutiononly in somedetails.Lateralindirectillocutionaryacts must be contended with because they areso common:e.g.,

(10) Charles, to Ann, in front of Barbara: Ann, what's playing at thetheaternext week?Ann, to Charles, infront of Barbara: Sorry, I don't know. But Barbaradoes.Barbara, to Charles, in front of Ann: 'Much ado about nothing'.In telling Charles that Barbara knows what's playing, Ann is indirectlyandlaterallyaskingBarbara o answerCharles'question. Again:

(11) Father, to son, infront of daughter Julia: Ned, go do your homework.Ned, to father, in front of Julia: I can't. Julia stole all my pencils.Julia, to father, in front of Ned: No, Papa, I did not.WhenNed addresses his father,he is laterallycharginghis sisterwith stealinghis pencils, a chargeshe then laterallydenies. Lateralindirectness ike this isquite an ordinarypartof conversations.6In talkinglaterally,the speakerdoesn't appear o be speaking o the indirectaddressee, but to someone else, and this appearance s often useful. One ex-ampleis what Greenburg1964has called the 'thirdpersoninvisible'. Imaginethat a grownson has broughthome a femalefriendfor his mother'sapproval:(12) Mother, to son, in front of his friend: Does she want another cup ofcoffee?In asking her son this, the motheris really tellingthe woman indirectlythatshe is refusingto speakto her. If the motherhadsaiddirectlyI refuseto speakto you, she would be belying her own words. She can accomplishwhat shewants only indirectly AND laterally, e.g. by telling her son I refuse to speak tothat woman, or as in 12.

In other examples, the speakermakes a pretenseof speaking inearlywhenthe primary llocutionaryact is lateral and indirect;e.g.,(13) Mother, to three-month-old, in front offather: Don't you think yourfathershouldchange your diapers?

6 Evidently the use of lateral indirectness is acquired quite early. Sully (1896:474-5) gave severalillustrations from the study of C., a somewhat precocious five-year-old:'One day (the end of the seventh month) he was playing on the Heath under the eye of hismother. He had put on one of the seats a lot of grass and sand as fodder for his wooden horse.While he went away for a minute a strange nurse and children arrived, making a perfectlylegitimate use of the bench by seating themselves on it, and in order to get room brushingaway the precious result of his foraging expedition. On coming back and seeing what hadhappened he turned to his mother and swelling with indignation exclaimed loudly: "What doyou mean by it, letting these children move away my things?" Of course this was intendedto intimidate the real culprits, the children.'

337



8/43


(14) Father, to dog, in front of son: Lassie, Daniel is about to take youfor a walk.(15) Mechanic, to automobile, in front of co-worker: Damn it, car, I needtwo more hands to help me bolt your fender back on.In 13, the mother is laterally and indirectly asking father to change the baby'sdiapers; in 14, the father is ordering the son to take the dog for a walk; andin 15 the mechanic is asking his co-worker for help. The pretenses are clear,because the direct illocutionary acts are aimed at things that couldn't be realtargets of what the speakers are saying.An extreme form of lateral indirectness is required when direct illocutionaryacts are precluded by taboo. Certain Australian groups have 'mother-in-lawlanguages' for talking in the presence of relatives with whom it is forbidden

to speak (Dixon 1972, Thomson 1935). However, when a woman wants tocommunicate with her son-in-law, whom she cannot address directly, she canaddress a nearby dog or child. Thomson (485) describes how a Wik Monkanwoman can ask her son-in-law for tobacco:'She may not use direct talk (wik koi'um) which is: raindir rjaiya mai ken ya'a, "Son-in-lawI tobacco nothing," but in gonk wonk tonn [the avoidance language], speaking to her daughter'sdog: kemiari mampi uaiya kon katume, "Daughter's son (i.e., the dog is the child of her tuwa[child]), I tobacco nothing." If he has none, instead of replying directly raiya ya'a, he againaddresses the dog: gaindari raiya katum, "Son (to his dog), I (have) nothing," or if he has asmall piece only he may say inwe rainda wetta, "Here (is) son no good." Freely rendered"Son, here is a little no good piece."'

Interestingly, the Wik Monkan term for the avoidance language is yonkwonk tonn, literally 'speech side another', which Thomson glosses as 'one sidetalk'. The woman is speaking 'one side' to her son-in-law. This is contrastedwith wik koi'um or 'straight talk', i.e. speech to linear addressees.So lateral indirect illocutionary acts, though ubiquitous, cannot be accountedfor by the standard theories. They appear to require the recognition ofinformatives.1.3. DESIGNATINGADDRESSEES.In conversations with three or more parties,

one person can 'speak to' the others without knowing which of them he isaddressing; e.g.,(16) Charles, to Ann and Barbara: Please return my map, whichever ofyou has it.

Imagine that Ann and Barbara know that Ann has the map, and that Charleshas no idea which one of them has it. So Charles is making a request of Ann,but he doesn't know this. As far as he is concerned, he is aiming what he issaying as much at Barbara as at Ann.The standard theories aren't equipped to handle 16; they presuppose thatthe speaker knows to whom he is addressing each illocutionary act. A request,says Searle (1969:57), is 'an attempt to get H to do A', where H is 'the hearer''in the presence of' whom the sentence is uttered. Charles, however, isn'ttrying to get both Ann and Barbara to return his map. The person he is tryingto get to return the map is the one who has it, whichever one that may be.

338



9/43

HEARERS AND SPEECH ACTSCharles makes this explicit with his vocative, whichever of you has it. Theproblemis that Charles'sintentions towardAnn and Barbaraare identical.Soif he is performingan illocutionaryact toward, say, Ann-which he must be,if he is makinga request-then he must also be performingan illocutionaryact towardBarbara.One solution is to assume that Charles is performing wo illocutionaryactssimultaneously.He is makinga requestof the personwith the map, and he isinforming he otherpersonof thatrequest.He lets the situationsort out whichperson is which. In this way, Charlesis treatingAnn and Barbara n virtuallythe same way. Later, we will returnto such examplesand offer a solutioninwhich Charles treats Ann and Barbaraentirely symmetrically.The point fornow is that examples like 16 also appearto require nformatives.

1.4. PUBLIC IDE-PARTICIPANTS.n many occasions, government officials,television newsmen, and others are ostensibly addressingcertainhearers,buttheir primaryaim is to informthe on-lookingpublic of what they are sayingto these hearers.Consider,first, a television interviewbetween, say, Crothersand SenatorSmyth. In private, theirtalk might go like this:(17) Crothers,to SenatorSmyth:Well,Joe, what do you think of the NewHampshirestink?Smyth,to Crothers: t's a goddammess. If Billdoesn't watch his ass,Bert may take away all his marbles.Before the television camera, however, Crothersand Smythtreat the unseenviewers as side-participants,and theirutteranceschange radically:(18) Crothers, to Smyth: Senator Smyth, what do you think of Jones'scontroversialremarksin the New Hampshireelection campaignlast week?Smyth, to Crothers:They were unfortunate. f SenatorJones doesn'twatchhis step, BertApplemanmay get impatientwithhimandcutoff all his campaign unds.Crothers, to Smyth:You're speakingof Bert Appleman,the Demo-cratic PartyNationalChairman,aren'tyou?Smyth, to Crothers: Yes, I am.

The features that Crothersand Smythadd to accommodatethe side-partic-ipants are of three types. First, the private references to Joe, to 'the NewHampshirestink', to Bill, to Bert, and to 'marbles'are filled out so that thetelevision audience can understand hemtoo. Second, whenSmythdoesn't fillout his references enough, as with the mentionof Bert Appleman,Crothersrequests clarification,even though he himself knows perfectly well who isbeing referredto. Third,both men move into a registerappropriate or theirpublicpersonae:they avoid informalityand offensive expletives.Consider, next, open letters addressedto 'the President'or 'the oil com-panies' or 'Membersof Congress', which are publishedas politicaladvertise-ments in newspapersand magazines,or 'letters to the editor', which are ex-presslywrittenfor possible publication n the editor'snewspaperor magazine.

339



10/43


Althoughthese letters are addressedto the President,the oil companiesetc.,their main targetsare the newspaperand magazinereaders who are the side-participants n these public acts. Without informativesto these readers, thepurposeof these publicacts would be lost.On other occasions, side-participants re desirablebut not necessary. Con-siderpublic expressionsof thanks,congratulations,pologies,andcondolences:

(19) Mayor, tofireman:Onbehalf of the citizens of SanFrancisco,I thankyou for savingthree lives in lastThursday's ire,and inappreciationI give you this medal.(20) Duchess of Kent, to Billy Jean King on winning the women's finalsat Wimbledon:Congratulations n your fine performance.Here isthe winner's cup.(21) The President, to Queen Elizabeth, as she steps out of an airplanein Washington:Welcometo the United States.In all three cases, the mainpoint is to express a politicallyimportant eelingin public. The public is intended to bear witness to these feelings. Althoughthese speakerscouldhave expressedthese feelingswithoutanyguests present,it is ordinarilythe public participation n the expression of the feelings thatcounts.With certain public acts, therefore, it is the informatives-the acts of in-forming he side-participants f what is beingsaid-that fulfillthe mainpurpose

of what is being said.1.5. INSTITUTIONALITNESSES.n Austin's classic work, the first three il-locutionaryacts mentioned are the marriagevow I do, the christening namethis ship the Queen Elizabeth, and the bequest I give and bequeath my watchto mybrother.These attractedhis attentionbecausethey arepartsof 'acceptedconventionalprocedures'in the churchand the law; so it is easy to see whenthey are mis-invoked, mis-applied,or mis-executed-i.e., when they are in-felicitous. As it happens,manysuchproceduresalso appear o require nform-atives for felicitous performance.Consider he questioningof the defendantat a murder rialby theprosecutingattorney:(22) Attorney, to defendant, in front of judge, jury, and court officials:Whendid you arrive at the bank?The attorneyis askingthe defendant(the addressee)a question. But if this isto count as an officialquestion, as a partof the officialtrial,then he must alsobe informingthe judge, jury, and other court officials that he is asking thedefendant that question. The informativesdirectedat these institutionalwit-nesses are mandatory.To see this, consider several ways in which the attorneycould succeed inasking the defendant the right question, but without properly informingtheinstitutionalwitnesses. He could speak in too low a voice; in that case, thejudge wouldask himto repeatthe questionso all could hear.Or he could leavethe questionambiguous or thejudge andjury between two interpretations fbank; nthatcase, thejudgewould askhimto clarifythe question.Moresubtly,

340



11/43

HEARERS AND SPEECH ACTShe could phrase the question When did you arrive at the place where Maryworks?,knowingthat the suspect would understand hat he was referring othe bank, but that thejudge andjury couldn't(since they don't know of Maryor where she works). In that case, the defense attorneycould object, and thejudge would ask for a rewording.As in ordinaryconversations,it isn't enoughfor the attorneyto askthewitness a questionthat thewitnessfullyunderstands.Withhis utterance,he mustalso properly nformthe institutionalwitnesses ofthat question.In many legal settings, the institutionalwitnesses sign documentsaffirmingthat they witnessed the appropriate llocutionaryacts towardthe addresseesand accept these as felicitous-as not beingfalse, or fraudulent,or insincere.Thisincludeswills, contracts,andpassportapplications,as wellas mostactionsincourt. With a will, for example, it isn't legallysufficient or a personto makea bequest sincerelyand in sound mind:he mustproperly nform wo witnessesthat he is doing so, and they must attest to this by signingthe will. There iseven a personspeciallydesignatedas a legallycertifiedside-participanto suchacts: the notarypublic.Consider, finally, the marriagevow made by the groom to the bride, asprescribed by the Book of common prayer:

(23) Groom, to bride, in front of minister and wedding company: In theName of God, I, John, take you, Mary,to be my wife, to have andto hold fromthis day forward[etc.]Johncould make this vow to Marysincerelywithoutanyoneelse around.Butfor it to count as a MARRIAGE OW,he must also inform the minister and theweddingcompanythat he is makingthis vow. As the Book of commonprayerprescribes, 'it is required ... that the ceremony be attested by at least twowitnesses.' Whatthese witnesses must apparentlyattest to is not merelythatJohnsaid therightwords,butthathe meantwhathe said.Beforethe ceremony,the witnesses areasked, 'If any of you can showjust cause why they [thebrideand groom] may not lawfully be married,speak now; or forever hold yourpeace.' If the witnesses believed thatJohn's intentionswere in any way insin-cere-e.g. that he planned to abandon Mary for another woman after theceremony-they could not in good conscience attest to his marriage vow.7The problemraisedby these examples is the same as before: the standardtheories say nothingaboutillocutionaryacts directedat hearersotherthan theaddressees. Yet for an attorney'squestionto be official, for a will to be legal,orfor a marriagevow to be proper,the speakermust act toward he institutional

7 Marriage appears to be the only Christian sacrament that requires official witnesses (aside fromGod-who, when not the addressee, is a side-participant in most sacramental speech acts). In aRoman Catholic marriage, witnesses may not have to be full side-participants in the sense we meanit, even though they ordinarily will. According to one Roman Catholic handbook (Jone 1959:523),'Witnesses need not have the explicit intention of acting as witnesses to a marriage. It is sufficientif they do so accidentally, even though the contractants are not aware of this ... Marriage is alsovalid if the witnesses are forced to assist by violence, fear or deception.' We are indebted to A.P. Martinich for bringing these points to our attention. For a discussion of sacramental speechacts but without mention of witnesses, see Martinich 1975.

341



12/43

LANGUAGE, VOLUME 58, NUMBER 2 (1982)witnesses as well as toward his addressees. He must fully inform the witnessesof what he is doing to his addressees. Without informatives, these institutionalprocedures would collapse.

2. AUDIENCEDESIGN. f speakers relied solely on conventional linguisticdevices to convey what they meant, everyone who knew the language shouldhave equal ability to understand them. But the examples we have offeredsuggest quite the opposite: when the speakers design their utterances, theyassign different hearers to different roles; and then they decide how to saywhat they say on the basis of what they know, believe, and suppose that thesehearers, in their assigned roles, know, believe, and suppose. That is, a fun-damental property of utterances is one that we will call AUDIENCEESIGN.Tocharacterize informatives properly, we must first characterize the roles towhich these hearers are assigned, and the ways in which speakers design theirutterances with these hearers in mind.

2.1. HEARERROLES.Conversations consist, very roughly, of sequences ofutterances among two or more people; with each utterance, the speaker per-forms one or more illocutionary acts directed at addressees. Consider one ofthese utterances:(24) Ann, to Barbara, infront of Charles, with David eavesdropping: Bar-bara, when did the two of you arrive last night?

For every addressee-directed illocutionary act such as Ann's question to Bar-bara, we can identify four basic roles:(a) SPEAKERagent of the illocutionary act). This is the person who performsthe illocutionary act: in 24, the role is filled by Ann.8(b) PARTICIPANTSn the addressee-directed illocutionary act. These are thehearers who the speaker intends to 'take part in' the illocutionary act that isdirected at the addressees. In 24, the participants include Barbara and Charles,but not David. What it means to 'take part in' an illocutionary act will bespelled out below.(c) ADDRESSEESf the addressee-directed illocutionary act. Certain of the

participants are intended to take on additional roles as addressees of the illo-cutionary act. In 24, Barbara is the addressee. The addressees are the partic-8 Just as the notion of hearer can be differentiated into various roles, so can that of speaker.Goffman 1979 distinguishes between the 'animator' of an utterance (the person uttering the words),the 'author' of the utterance (the person 'who has selected the sentiments that are being expressedand the words in which they are encoded'), and the 'principal' (the person who is 'committed towhat the words say'). Suppose that George (standing with Jane) says to Julia, in front of Margaret,Julia, Jane and I congratulate you on your new discovery. Although George is the animator andthe author of the utterance, he is merely the spokesman for the congratulations, for which he and

Jane together are the principals. Informatives may aid in these distinctions. In George's utterance,George himself is informing Julia and Margaret of something; but what he is informing them of isthe congratulations that he and Jane are jointly performing. George is the sole agent of the in-formative, but George and Jane jointly are the agents of the congratulations. Informatives wouldenable us to distinguish between the agents of the two illocutionary acts. In other ways too, thisanalysis would fit nicely into the informative analysis proposed in ?4, below.

342



13/43

HEARERSAND SPEECHACTSipantswho are, or could be, designatedvocatively in the utterance,as Barbarais named in 24. They need not coincide with the set of hearersreferred o byyou, but need only be a sub-set of those hearers: in 24, you refers to bothBarbaraandCharles,whereasthe only addresseeis Barbara.Theparticipants,then, divide into two sub-sets: those who are also addressees, like Barbara,and those who are not, like Charles. It is convenient to call the latter partic-ipants SIDE-PARTICIPANTS, the term we have already been using for thesehearers.

(d) OVERHEARERSf the addressee-directed illocutionary act. These are thehearerswho are NOTntendedby the speakerto 'takepartin' the illocutionaryact, in the favored sense of 'take part in', but who are neverthelesslisteningin. In 24, David is an overhearer.9These four roles are defined by the speaker. He defines his own role asspeaker;he defines who is to 'takepartin' his illocutionaryact, separating heparticipants rom the overhearers;and he defines who amongthe participantsare to be addressees. He makes these role assignmentsby the way he designshis utterance,andby the way he positionshimself withrespectto the audience.We will take up these methods in our discussion of audiencedesign and roleassignment. On any occasion, the speaker may not succeed in getting hishearers to recognize the roles to which they are being assigned, despite aflawless performanceon his part;hearersdo makemistakes.For ourpurposes,however, these cases are irrelevant.Whatis relevant are the speaker'sinten-tions about who is to assume which roles. It is these intended roles that arebeing designatedas speaker, addressee, participant,and overhearer.2.2. TYPESOF AUDIENCEESIGN.The speaker designs his utterance withthese roles in mind. In this way, audiencedesign can be dividedroughlyintoparticipantdesign, addressee design, and overhearerdesign."?The basic de-sign, as we will present it, is for participants.Addressees andoverhearersaretaken care of in modulationson the basic design.

9 These distinctionsaren'taltogethernew. Virtually he same ones have been made in studiesof thesequentialorganization f conversation seefn. 10),althoughnot withrespect o thespeaker'sillocutionary ntentions. Goffman'sdistinctions 1975:260; ee also 1978, 1979)are very close toours:

'Observe now that, broadly speaking,there are three kinds of listeners to talk;those whoovERhear,whetheror not theirunratifiedparticipations inadvertentand whetheror not ithas been encouraged; hose who are ratifiedparticipants ut (in the case of more thantwo-persontalk)are not specificallyaddressedby the speaker;and thoseratifiedparticipantswhoARE ddressed,that is, oriented to by the speaker n a manner o suggestthat his words areparticularlyor them,and thatsome answer s thereforeanticipatedromthem,moreso thanfrom the otherratifiedparticipants.'10Ourconcept of audiencedesignhas obvious roots in the notionof 'recipientdesign'used instudiesof the sequentialorganizationof conversation cf. Sudnow1972,Psathas1979).Sacks etal. 1974,who attributeheconceptto Garfinkel1967,describe t as follows:'By "recipientdesign"we referto a multitudeof respects in whichthe talkby a party n a conversation s constructedor designed n ways whichdisplayan orientationandsensitivityto the particular ther(s)who aretheco-participants.'Ournotion of audiencedesignencompassesoverhearers s well as addresseesand 'co-participants'.

343



14/43

LANGUAGE, VOLUME 58, NUMBER 2 (1982)The basic designis characterizedby the principleof responsibilityproposedearlier.By this principle,the speaker s responsiblefor designinghis utteranceso that all the partiesto the conversationcan keep trackof what he is saying.This defines what might be called a CANONICALCONVERSATION.With eachcontribution,the speakerassigns to every otherpartyin the conversationtherole of participant.The conversationcan thereforeaccumulateas awhole, andthe commongroundthat accrues is easy for everyone to keep trackof.In non-canonicalconversations, this simplicity is lost. ImagineA talkingwith B, C, and D. If the conversation isn't canonical, A must keep trackseparatelyof his commongroundwith B, with C, and with D. If he knows Bwell, C moderately, and D very little, then there will be large disparitiesincommongroundto begin with. If he informsB, C, and D of differentthings

during he conversation,and if he is informedby them of different hings,thesedisparitiescan only growand become even more difficult o keeptrack of. Thespeaker's task is therefore greatly simplifiedin that most conversations, ormost parts of conversations, are assumed to be canonicalby all parties. Theparties accomplish this by always assigning all other parties to the role ofparticipant.Although most conversations, or parts of conversations, are canonical,speakerscan and do deal with disparities n commongroundwhen they must.They exploit these disparitiesto say one thing to one group of participantswhile saying somethingelse to another,or to carryout elaboratedeceptions.We will returnto these possibilitieslater.The addressees are the ostensible targetsof what is being said. Ordinarily,they are the participants or whom the speakerhas the mostdirect and obviousgoals in designinghis utterances.In 5, when Anntells Charles We went to themuseum, she is makingsure both Barbaraand Charlesunderstandwhat sheis doing, but she has designedherutterancewithCharles n particularn mind:she wants to get himto believe thatthey wentto the museum. As a preparatorycondition, then, she must assume, very roughly,that Charlesdoesn't alreadybelieve what she is asserting. She doesn't have to assume this for Barbara;indeed, Ann thinks that Barbaraalreadydoes believe the proposition hat sheis asserting.Most of Searle's 'felicity conditions'for speech acts are satisfiedas partof addresseedesign, not merelyparticipantdesign.Speakers also design their utteranceswith overhearersin mind. Althoughthey don't intend the overhearersto 'take part in' what they are saying-inthe favored sense of 'take part in'-they realize that the overhearers cannevertheless form conjecturesor hypotheses aboutwhat they mean.The pur-pose of overhearerdesign is to deal with these hypotheses. By designingtheirutterances ust right,speakerscan leadoverhearers o formcorrecthypotheses,incorrecthypotheses, or even no coherenthypothesesat all. If they know theiroverhearers, hey can even designwhatthey say to fit theminparticular.Thesewe will call KNOWNOVERHEARERS. et speakers also recognize the possibilityof UNKNOWNOVERHEARERS, nd they can design what they say with them inmind too. That is, speakers can harborintentions toward both known andunknownoverhearers,and can design theirutterancesaccordingly.These in-

344



15/43

HEARERSAND SPEECHACTStentions, however, are differentfrom those that speakers have toward theparticipants,as we will argue, in that they are not intended to be recognizedas intendedto be recognized."In the most obvious examplesof overhearerdesign, speakerstry to preventoverhearers romcorrectlyguessingwhat they are saying. For example,thereis the secrets-in-a-crowd cenario:

(25) Ann, to Barbara, on crowded bus: Do you remember that thing aboutyou-know-who that we were talking about last week? Well, ithappened.Ann's references are designedto be opaqueto everyone but Barbara,who isthe only one for whom the last week's conversationis commonground.Sim-ilarly, there is the spelling ploy used by parentsin front of children:(26) Father, to mother, in front of Johnnie: What did you think of theb-i-c-y-c-l-ewe saw in the store?With spies, this sort of preventionis a sine qua non of communication.Onestrategy s to switch to a languagenot knownto potentialoverhearers,namelya spy code.12Another is to switch to a code that masqueradesas a genuinelanguage,and thus leads overhearersto the wronghypotheses.'3Overhearersare generally not meant to realize how utterances have beendesignedfor them. The deceptionmay' benign, as when a workersings thepraises of his boss while knowingthatti. 'oss may be overhearing.It is less

benignwhen the same workerwinksbroaulyat his co-workersto indicatetheirony of his remarks.Othertimes, the designmay have moreseriousends, aswhen Hamlet, in Act III, Scene 1, plays the madman n talkingto Ophelia-apparently o draw two knownoverhearers, he KingandPolonius, into think-ing that he is going mad. In these examples, the conjectures to which theoverhearersare led are sometimescorrect, and sometimesnot.1ConsiderwhatGoffman1978has called 'responsecries', as whena man,walkingalonedownthe street and slippingon some ice, says Oops loud enoughfor the people who happento bewatchingto hear. As Goffmanargues, the man intendsthese self-imprecationso be heardby

peoplenearby,in orderto let them knowhe is awareof what befell himand is in full control.Yethe isn't addressing hose people. He intendshis Oopsto appear o be a self-imprecationhattheyjust happento overhear. In our terms, he intends themto recognizewhat he meant,but not bymeansof theirrecognitionof his intention hatthey do so.12 Kahn 1967gives the followingexample of a telegramfrom PresidentLincoln to ColonelLudlowduring heCivilWar: Guard dam heyatwaylandbrown orkissingvenuscorrespondentsat neptune are off nelly turning up can get shy detainedtribune and times richardson he areascertainandyou fillsbellythis if detainedpleaseodor of ludlowcommissioner.'Translation:ForColonelLudlow.Richardsonand Brown,correspondents f the Tribune, apturedat Vicksburg,aredetainedat Richmond.Pleaseascertainwhy theyaredetainedandget themoff if you can. ThePresident.'13 As in this WorldWar II conversationbetweenKurusu, he JapaneseAmbassador o the US,and Yamamoto,the ForeignOfficeAmericanDivisionChief, shortlybeforePearlHarbor fromKahn):'But withoutanything,they want to keep carryingon the matrimonial uestion[i.e. thenegotiations].They do. In the meantimewe're faced withthe excitementof havinga childborn[i.e., we're faced with a crisis].Ontop of thatTokugawa i.e. the army] s reallychampingat thebit, isn't he?'

345



16/43

LANGUAGE, VOLUME 58, NUMBER 2 (1982)Anotherelement in overhearerdesignis politenessandregister(see Comrie1976, Levinson 1979). In most societies, certain words are taboo in certaincircumstances,in which they may not even be overheard. In urbanAmerica,obscenitiesfreely used in lockerroomsareoftenavoidedinpublicplaceswherethey may be overheard.14The goal is to projectthe speaker'spublic personaand to avert overhearerdiscomfort. Many aspects of speech registers, fromtheoratorical o the intimate,have relatedorigins,as do theAustralianmother-in-law' languagesmentioned above. Because it is forbidden to use ordinaryspeech in the presence of one's mother-in-law r otherspecifiedkin, whathasdeveloped is a special register, with its own vocabulary,that is used near atabooed kin who mightpossibly overhear.2.3. ROLEASSIGNMENT.s a critical part of audience design, the speakermust designate which hearersare to take which roles. It is essential that thespeakerand participants,and the speakerandaddressees,mutuallyrecognizewhich hearers are being designatedas participantsand which as addressees.How speakersaccomplishthis is a complicatedtopic (see Goffman1979).Wewill mentiononly the majordevices by which they do this: physical arrange-ment,conversationalhistory, gestures,mannerof speaking,andlinguisticcon-tent. Most utterancesrely on some combinationof these five factors.Participantsare often distinguished rom overhearersby physical arrange-ment. Hearers in the same groupas the speakercan ordinarilyassume they

areintendedto be participants,whereasother hearerscannot.Whatconstitutesa group (a 'with', in Goffman's 1971 terminology) s highly constrainedbyphysicalarrangement.Thepeoplemustbe neareach other relativeto the spaceavailable-not separated by obvious physical or psychologicalbarriers,andaccessible to each other auditorilyand visually (see Goffman1963, 1971).Participants realso distinguished romoverhearersby the historyof the on-going conversation. If certainhearers were participantsduringthe last utter-ance, and if the speaker gives no indication to the contrary, then they canassumethatthey are also participantsorthe currentutterance.It mayhappenthat two partiesin a groupexchangeso manyremarks hatthey come to definetheir own conversation, with the others splittingoff into a separateconver-sation. It can also happenthat one partyin a groupcomes to be ignored,andis no longertakento be a participantn what is being said.Addressees aregenerallydesignated n partby gestures.They can be pickedout by eye contact, sometimesaccompaniedby a handgestureor nod of thehead. Furthermore,certainpeople in a groupcan be excluded as addresseesor participantsby the speaker's turninghis back on them. The advantageofgestures is that they are public acts, easily recognizedsimultaneouslyby allthe partiesinvolved.Addressees and participantscan also be designatedby mannerof speaking.14 Strikingly, it isn't the meaning of a particular expression that is eschewed, but rather its formor sound. As Randolph 1928 has noted, people will use circumlocutions like She's ready to go orThe hammer's back just to avoid saying The gun's cocked. The idea is, apparently, that anoverhearer might hear only the critical word, and mistake it for an obscenity.

346



17/43

HEARERS AND SPEECH ACTS

By whispering, a speaker can select a small group of people as participants oraddressees, letting everyone else know they are not participants. By speakingin a markedly loud voice, a speaker can do just the opposite. At a restaurant,Ann might say to her companion This could do with a little salt, asking himto pass the salt. By saying this loudly within earshot of the waiter, she coulddesignate the waiter as participant and indicate that he instead was to fulfillher request. By using a high pitch, a speaker can designate children as opposedto adults as participants; by over-articulating, a speaker can likewise designateforeigners.Finally, of course, addressees, participants, and overhearers are often des-ignated through the content of what is said. Addressees can be defined byvocatives and other devices. Participants can be brought in by prefatory ut-terances, such as George and Julia, I want you to hear what I have to say toEdward, and in other ways. Overhearers can be excluded as participants bysimilar devices. Later we will take up devices for designating addressees insome detail.In summary, speakers intend certain hearers to fill certain roles, and theydesign their utterances accordingly. In particular, they have devices for gettinghearers to recognize mutually who is to take which roles. Most of these devicesderive not from the content of what is said, but from such factors as the physicalarrangement of the hearers, the history of the conversation, gestures, and themanner of speaking. All these factors, whether 'linguistic' or not, must beconsidered part of the means by which the speaker performs illocutionary acts.They are crucial for designating to whom these acts are directed.

3. INFORMATIVES.o far our characterization of informatives has been in-formal. We now consider them more closely, taking up three issues. First, areinformatives truly illocutionary acts? Second, to whom are informatives di-rected? And third, how would informatives be characterized in the standardtheories?3.1. INFORMATIVES S ILLOCUTIONARYCTS. Informatives are speech acts-but which type are they? Since they are performed with whole utterances, theycannot be propositional acts such as reference or predication (Searle 1969).The obvious candidates are Austin's locutionary, illocutionary, and perlocu-tionary acts.Informatives are clearly not locutionary acts, since these are merely acts of'saying something' (Austin, 94). In ex. 1, Othello's locutionary act consists ofsaying the imperative sentence Come, Desdemona with come and Desdemonahaving a certain meaning and reference. Since locutionary acts aren't directedby speakers to specific hearers, Othello's act of informing Iago and Roderigoof his reques. to Desdemona cannot be a locutionary act. In Austin's scheme,

locutionary acts would be used in performing informatives, and so they wouldbe kept conceptually distinct.Nor are informatives perlocutionary acts. In uttering ex. 1, Othello is tryingto get Desdemona to go with him. If she then goes with him, she does so onlyas a consequence of her understanding what he meant. Ever since Austin,

347



18/43


consequences such as this have been called perlocutionary ffects. These aredistinguished romillocutionaryeffects, whichconsist of hearers'understand-ings of what speakersmean. Imagine,then, that (in utteringex. 1) Othello isalso tryingto get Iago and Roderigoto stay behind. If they do so, that wouldbe another perlocutionaryeffect of Othello's utterance. But it would be aconsequence of theirunderstandinghat he was askingDesdemonato go withhim. Ourinterest, however, is not in the consequences of theirunderstandingwhat he meant, but in the understandingtself. Our interest, therefore, is inillocutionaryeffects and in the illocutionaryacts thatproducedthem.In the standardtheories, illocutionaryacts are distinguishedfrom otherspeech acts at the utterance level by the fact thatthey requirereflexiveinten-tions-Grice's 'm-intentions'(Bach & Harnish1979,Grice 1957, 1968, 1969,Schiffer 1972, Searle 1969, Strawson 1964).What is required,as Searle (47)put it, is that 'the speaker S intends to producean illocutionaryeffect IE inthe hearer H by means of gettingH to recognizeS's intentionto produceIE.'So Othello intends to get Desdemonato understand hat he wants her to gowith him. But accordingto this criterion,he gets her to understand his bygettingher to recognize his intentionto get her to understand his. If inform-atives are illocutionaryacts, Othello must intendto get lago and Roderigotounderstand hat he is requestingDesdemonato go with himby means of gettingthem to recognizehis intentionto get themto understand hathe is doingthat.Note thatOthellocouldintend some overhearingguard o guess whathe meanstoo. But if he doesn't intendto do so by meansof gettingtheguard o recognizethat intention,then he isn't performingan illocutionaryact towardthatguard.Let us call a communicative ntentionthatis less than an m-intention like thisone) a PARTIAL INTENTION.There is good evidence that informativesrequirem-intentions,not partialintentions. Let us returnto our exampleof a lateralindirectrequest:

(9) Ann, to Charles, in front of Barbara: Charles, I insist that Barbara tell you who we metat the museum today.Ann is indirectlyaskingBarbara o tell Charleswho they met. By the standardtheories, requests like this, whether direct or indirect, requirem-intentions.How does Annmakethis request?She can do it, we argued,only by informingBarbaraof her assertion to Charles.Supposethatthis informativewere basedon a partial ntention; .e., Ann didn'tintendBarbara o understandby meansof her recognitionof any of Ann's intentions. If so, therewouldbe no way forAnn to get Barbara to recognize her INDIRECTntention to get Barbara to tellCharleswho they met at the museum;and this recognitionis necessary forsuch a request. Put simply, m-intentionscannot be indirectlyconveyed bypartial ntentions. Ann's informativemust involve m-intentions.Consideranotherexamplegiven above:

(16) Charles, to Ann and Barbara: Please return my map, whichever of you has it.Charles is requestingone of the women-he doesn't know which-to returnhis map;and he is informing he other womanof that request. By definition,he has m-intentions oward the first woman.But since he doesn't know whichwoman that is, he must have m-intentions owardthe second woman as well.

348



19/43


Since he is only informing the second woman of his request, that informativemust involve m-intentions.'Letters to the editor', intended for publication, force the same conclusion.When people write such letters, they write, in all respects but one, as if theywere talking to the newspaper readers and not just the editor. They intend thereaders to understand them just as if they were writing them directly-as ifaddressing them as 'Dear reader'. So if m-intentions are required for 'Dearreader', then they are also required for 'Dear editor'. 'Dear editor' is differentonly in that the writers are simply informing readers of what they are sayingto the editor, whom they address as you. Their informatives, therefore, mustinvolve m-intentions.Finally, consider three-way conversations, as in these two examples:

(27) Charles, to Barbara, in front of Ann: Did the two of you go to themuseum?Barbara, to Charles, in front of Ann: Yes, we did.Charles, to Barbara, in front of Ann: And then what?(28) Charles, to Ann, in front of Barbara: Did the two of you go to themuseum?Ann, to Charles, in front of Barbara: Yes, we did.Charles, to Barbara, in front of Ann: And then what?For Charles to expect Barbara to grasp the ellipsis of And then what? in either27 or 28, he must make sure she understood the prior question-regardless ofwhether it was addressed to Barbara herself, as in 27, or to Ann, as in 28. Howcan he make sure? In 27, it is easy, for he has asked Barbara the prior questiondirectly; he therefore has all the proper m-intentions toward her. In 28, hemakes sure only by informing Barbara of that prior question. Intuitively, how-ever, the informative in 28 serves as just as good an antecedent as the directquestion in 27. If it does, then it too must involve m-intentions.The case, however, can be made even stronger. Suppose that Ann andBarbara went to both the Tate and the British Museum, and that Barbara hadtalked with Charles only about going to the Tate. Barbara didn't know, how-

ever, what Ann had discussed with Charles-they might have talked abouteither museum-and she knew that Charles realized this. In 27, Charles couldexpect Barbara to recognize that the museum to which he was referring wasthe Tate, since it was the only museum they had discussed together; i.e., hecould expect her to pick out the Tate by means of her recognition of hisintention. In 28, however, he could NOT xpect her to know to which museumhe was referring unless he again intended her to understand by means of herrecognition of his intention; otherwise, she could equally well assume he wasreferring to the British Museum, which she could assume he had discussedpreviously with Ann. So, for the informative in 28 to serve as a proper ante-cedent for the elliptical question, it requires an m-intention. A partial intentionwon't do.If informatives require m-intentions, they must in this sense be a kind ofillocutionary act. What distinguishes them is that they are directed not ataddressees, but at participants, which may also include hearers other than

349



20/43

LANGUAGE, VOLUME 58, NUMBER 2 (1982)addressees. So good evidence exists for the participanthypothesisthatcertainillocutionary acts are directed at addressees, and others are directed atparticipants.Our scheme involves a basic contrastbetweenparticipantsandoverhearers.Participantsare intended to take partin addressee-directedllocutionaryacts,andoverhearersaren't. Now we can say what 'takingpartin' means:Hearersare intended to take part in an addressee-directed llocutionaryact wheneverthey are m-intendedto understand it. What distinguishesoverhearersfromparticipants, herefore, is that overhearersaren'tm-intended o understand.

3.2. SPEECHACTSTO PARTICIPANTS.n each of our examples, the speakerinforms the side-participantsof the illocutionaryact which he is directingatthe addressee. In ex. 5, when Ann tells Charles Wewent to the museum,sheis informingBarbara hat she is tellingCharlesthat they went to the museum.That,however,can't be all thatAnnis doing.If she is to enable theconversationto accumulate,she must also inform Charles that she is informingBarbaraofherassertion. If she doesn't, Charlescannot assume that Barbara an buildonwhat Ann has told him. The same goes for our other examples.Informatives, then, are directed at all the participants,notjust at the side-participants.This is capturedin the participanthypothesis, which states thatone class of illocutionaryacts is directedat addressees,andanotherat partic-ipants. As a first approximation,an informative s an act by the speakertomake it known to the participantswhat illocutionaryact he is performing orthe addressees.This characterization,however, isn't quite right, since it omits an essentialaim of informatives-to make it public knowledge among the speaker andparticipantswhat the speaker s doingwithhis utterance.In24, when AnnasksBarbara in front of Charles, Barbara, when did the two of you arrive last night?,she isn't informingBarbaraand Charles separately. She intends Barbaratoknow that Charlesknows, to know that he knows she knows, to know that heknows that she knows that he knows, and so on ad infinitum.This is what istechnically called MUTUAL NOWLEDGEetween Barbara and Charles (cf. Schif-fer). Ann also intends Barbara o know that her question is mutuallyknownby her andCharles,andshe intendsCharlesto know that it is mutuallyknownby her and Barbara,and so on. With these additions, we require what istechnically known as COMMONNOWLEDGEn a group (Lewis 1969). 5 To makeit clear that we are using these technicaldefinitions,we will hyphenatethemas 'common-knowledge','commonly-known'etc. Thus, in uttering24, Annintendsto make it commonly-knownamongAnn, Barbara,and Charlesthat,in uttering24, she is asking Barbarawhen Barbaraand Charles arrivedthenightbefore.

15 Schiffer (131) also considers mutual knowledge within a group, but his formulation is notequivalent to Lewis' common knowledge. For one thing, his definition allows for a proposition tobe mutually known within a group without its being mutually known between two members of thegroup. Bach & Harnish give a similar definition, but without the infinite iterations of either Lewisor Schiffer.

350



21/43

HEARERSAND SPEECHACTSIn the standard heories, each type of illocutionaryact, suchas informatives,has associated with it a set of felicity conditions-the necessaryandsufficientconditions 'for the successful andnon-defectiveperformance f the act' (Searle1969).16 ince informativesare the meansby whichspeakers'tell' participantswhat they are doingto addressees, they resemble the illocutionaryacts whichSearle has called 'representatives',andwhich Austin(as well as Bach & Har-nish) has called 'constatives'. This class includes forms of 'telling' such asassertions, reports, claims, and allegations. On that model, the felicity con-ditions for informatives should look something ike this (where 'x' is the sen-tence uttered, 'S' is the speaker, 'A' is the addresseeor addressees, 'P' is theparticipantor participants,and 'I' is the illocutionaryact which S is directingat A):17

(29) Preparatorycondition: In utteringx, S is performing addressed toA.Sinceritycondition: S wants it to be commonly-knownamongS andP that, in utteringx, S is performing addressed to A.Propositionalcontent condition:S predicatesthat, in utteringx, S isperforming addressed to A.Essentialcondition: S's utteringx counts as an attemptby S to makeit commonly-knownamong S and P that, in utteringx, S is per-formingI addressed to A.

3.3. A NOTATION.To express the parallels between addressee- and partici-pant-directed llocutionaryacts, we need a notation.Let us return o Othello'sutterance:Come, Desdemona. His request has three arguments: he speaker(Othello),the addressee (Desdemona),and the requestedact that she go withhim.Thismightbe representedas a three-place unctionas follows:Request(O,D, 'D go with 0'). Likewise, Othello's informativehas three arguments: hespeaker(Othello),the participants Desdemona,lago, andRoderigo),and therequestof which he is informing hem. This mightbe representedas follows:(30) Inform(O,D & I & R, I,)Here II stands for Request(O, D, 'D go with 0'), where I represents 'illocu-

16 Searle (54) notes that the failure to satisfy only some of these conditions is sufficient 'to vitiatethe act in its entirety'. Thus Bach & Harnish (55) separate 'success' conditions-those conditions'that are singly necessary and jointly sufficient for the performance of an act'-from felicityconditions 'that are not success conditions but are required for non-defectiveness.' We agree withBach & Harnish that the role of such felicity conditions in a theory of speech acts is unclear. Fromour point of view, it is the speaker's illocutionary intentions, as captured in Searle's essentialcondition (which Bach & Harnish consider a success condition), that are definitive for the act inquestion. As Searle himself notes (69), 'In general the essential condition determines the others.'17The informative defined here is a JOINT NFORMATIVE,n which the participants are collectivelyinformed. We take this to be the unmarked case. One can also define an ELEMENTARYNFORMATIVE,in which the speaker informs each participant separately, and doesn't intend his addressee-directedillocutionary acts to become common-knowledge among the participants. The situations that requireelementary informatives appear to be rare (see Clark & Carlson 1982).

351



22/43


tionary act'.18In ex. 1, therefore, Othello can be said to be performing woillocutionaryacts with the same utterance:(31) a. Inform(O,D & I & R, I1)b. Request(O,D, 'D go with O') = I1This notation makes it easy to express several constraints on the relationbetween the addressee-directedand participant-directedcts that can be per-formedby a single utterance.First, the agents of the two acts must be iden-tical:19n 31, Othello is the firstargumentof both the Informandthe Request.Second, the targetsof the one act-the addressees-must be includedamongthe targets of the other-the participants: n 31, the second argumentof theRequest (Desdemona)is includedin the second argumentof the Inform(Des-demona,Iago, andRoderigo).20Finally,the thirdargumentof the Informmustrefer to the addressee-directed llocutionaryact: in 31, the thirdargumentofthe Inform s the Request. In this analysis, informatives ike 31aalways comepairedwith addressee-directed llocutionaryacts like 31b.

4. INFORMATIVE-FIRSTYPOTHESIS.or each addressee-directed illocution-ary act, there is an informative o let the participantsknow of that act. In ex.1, along with Othello's request of Desdemona, there is an informativeforDesdemona, lago, and Roderigo:

(32) a. Informative:Inform(O,D & I & R, 11)b. Addressee-directed llocutionaryact: Ii = Request(O,D, 'D gowith O')What is the connectionbetween 32a and 32b?Onepossibilityis thatthey areperformed ndependentlyand inparallel.Thiscan be diagrammedas follows, with the arrows indicatingthe direction ofcausation: Inform(O,D & I & R, I1)(33) Othello's utterance

Request(O,D, 'D go withO')Othello would be utteringCome, Desdemona in order to perform he inform-ative and the request independently.This wouldbe comparable o a person'sfanninghimself in orderto cool himself and to exercise his wrist. Since he cancool himself without exercising his wrist, and can exercise his wrist withoutcooling himself, these two acts are independent.But can Othello inform Des-demona of his request without makingthat request, or make his request of

18 This notation doesn't try to capture all the niceties of the logical form of illocutionary acts.It is intended to represent only roughly the content and force the speaker intends and certainrelations among them.19 However, see fn. 8 above.20 Note that these targets can be designated either referentially, as here, or attributively, as wehave discussed in ?4.

352



23/43

HEARERS AND SPEECH ACTSDesdemona without informingher of it? It appears not. The independencemodel is inappropriate t the outset.The second andthirdpossibilitiesare morepromising.In the addressee-firstmodel, 32a is performedby meansof 32b, as in this diagram:(34) O's utterance=>Request(O,D, 'D go with O') =>Inform(O,D & I& R, 11)Othellomakes a requestof Desdemona;andby doingso underthe appropriateconditions, he informseveryone of his request. In the informative-firstmodel,in contrast, 32b is performedby means of 32a, as in this diagram:(35) O's utterance > Inform(O,D & I & R, I) =>Request(O,D, 'D gowith O')Othello informsDesdemona, Iago, and Roderigothat he is makinga requestof Desdemona, and by doing this, he also makes the request of Desdemona.Inbothmodels, one act is performedby meansof another-just as, in shootinga gun, the act of pulling he trigger s performedby means of the act of crookingthe rightindex finger.The naturalplace to comparethese two models is in speech situations withmore than one hearer, where the speaker can have differenttargets for hisinformativesand his addressee-directedllocutionaryacts. There he mustoftendistinguishaddressees from participants.As we noted earlier, he has manyways of getting the participants o recognizewho the addresseesare; but theprincipalway is with vocatives. Most vocatives studied so far in linguisticshave been the canonical kind:propernamesof single individuals,as in Come,Desdemona. But they also occur in many other forms, and these provide asurprising ine of evidence for distinguishing he addressee-firstand the in-formative-firstmodels.

4.1. THEINCLUSIONRINCIPLE. he basic function of vocatives is to desig-nate addressees. They generallydo this by distinguishinghe addresseesfromthe other participants.In ex. 1, there are two 'audiences': the participants(Desdemona,lago, andRoderigo);and the addressee(Desdemona).Let us callthese two sets of hearers 'Set 1' and 'Set 2'. In usingthe vocative, Othelloisindicatingto Set 1 that he is assigningthe role of addresseefor this particularrequest to Desdemona (Set 2) and not to Iago and Roderigo(Set 2'). He isdirectinghis vocative at Set 1 as a means of gettingthem to recognize Set 2as distinct from Set 2'. Since performing he vocative is an integralpart ofperformingboth the informativeand the request, Othellomust be performingthe informativeto Set 1 as a means of makingthe requestto Set 2.21What we will here call the INCLUSION RINCIPLEs one we have met before:when one illocutionaryact is used as a means for performinganother, thetargetsof Set 1 must contain the targetsof Set 2. This is a basic assumption

21 Consider the request One of you guys, help yourself to the last beer. In the vocative, theplural you guys requires the targets of the vocative to be more than one guy; but in the requestproper, the singular yourself requires the target of the request to be exactly one person. Thecontrast between the two sets of targets is directly indicated in the plural you vs. the singular you.

353



24/43

LANGUAGE, VOLUME 58. NUMBER 2 (1982)of Searle's analysis of indirect speech acts. When someone asserts This soupneeds salt as a means of requesting someone to pass the salt, the addresseeof the request must also be an addressee of the assertion. Searle's idea is thatthe speaker gets the addressee to recognize his request by getting him torecognize the assertion and to see that it isn't sufficient in the circumstances.Without this assumption, Searle's analysis would fall apart.The inclusion principle rules out the addressee-first hypothesis, while leavingthe informative-first hypothesis intact. Recall that, in the informative-first hy-pothesis, Othello performs his informative (one illocutionary act) as a meansof performing his request (another illocutionary act). In the addressee-firsthypothesis, he does the reverse. So in the informative-first hypothesis, thetargets of the first act (Set 1) include the targets of the second (Set 2); but inthe addressee-first hypothesis, the targets of the first (Set 2) do not include thetargets of the second (Set 1). The inclusion principle, therefore, is in directconflict with the addressee-first model. In plainer words, Othello cannot makea request of Desdemona as a means of informing Iago and Roderigo-since toinform them, he would have to be directing the request at them too, which heisn't.This conclusion is a general one. If we accept the inclusion principle, thenthe addressee-first model is ruled out in favor of the informative-first model.

4.2. THE EQUIPOTENTIALITY PRINCIPLE.An additional principle applies whenvocatives are either attributive (Donnellan 1966, 1968) or indefinite. Considerthe attributive vocative here:

(36) George, to Alistair and Fergus: The last one of you to leave, turn outthe lights.Suppose that George has no way of knowing who will leave first, Alistair orFergus. He cannot direct his vocative at the actual addressee (the person whois to turn out the lights) since he doesn't know which one that is. All he cando is direct it at both Alistair and Fergus-and intend each of them, by rec-ognizing who is picked out, to recognize which one is the addressee. So, bythe inclusion principle, George can only make his request via his informatives,as in this diagram (with the first causal link from George's utterance omitted):

(37) Inform(G, A & F, 11) =: Request(G, the last one to leave = X, 'Xturn out the lights')Only in this way is the target of the request (Set 2) included in the targets ofthe informative (Set 1).The second principle at work here is one we will call the EQUIPOTENTIALITYPRINCIPLE. When a speaker directs what he says at several hearers at once,not knowing which of them he is actually addressing, each hearer has an equalpotential of being an addressee. So the speaker must have the same intentionstoward all the hearers; he cannot have special intentions toward any individualhearer. The principle is this: Whenever the speaker cannot indicate to eachof two or more participants whether or not that participant is an addressee,the speaker must have the same m-intentions toward each of them, regardlessof who the addressees actually are.

354



25/43


In 36, the only m-intentions that George has equally toward Alistair andFergus are in his informatives. All George can do directly to either Alistair orFergus is inform him of the request he is making of the person who leaves last.That works out just right, however-since, by informing each of them of thatrequest, he will automatically make the request of the right person as well.Suppose that George utters 36 at noon, and the first person to leave does notdo so until midnight. Although, in one sense, he will have performed both theinformatives and the request at noon, the addressee of the request won't be-come specified, and hence the request itself won't become operative, untilmidnight. By the equipotentiality principle, then, George can make his requestonly by means of his informatives.The inclusion and equipotentiality principles also apply to indefinite vocatives:

(38) Schwartz, to history students: Any of you who needs a syllabus, raiseyour hand.Schwartz is directing the vocative any of you who needs a syllabus at each ofthe history students; he is thereby indicating to each of them who are theaddressees of his request and who aren't. Since these addressees are a sub-setof all the history students that he is informing, by the inclusion principle, hemust be using the informative as a means of making that request, as in thisdiagram:

(39) Inform(S, students, II) > I = Request(S, X = any student whoneeds a syllabus, 'X raise his hand')

The equipotentiality principle leads to the same conclusion. Take the pointof view of a student called Margaret. Since Schwartz cannot know whether ornot she is an addressee of his request, he must treat her precisely as he istreating all other students. All he can do is inform her that he is making arequest of any student who needs a syllabus; but that will do very well. If shehappens to be such a student, he will thereby have made a request of her, eventhough he needn't know that he has done so. If she happens not to be such astudent, he will only have informed her that he was making a request of theneedy students. To treat Margaret and her fellow students as equally potentialaddressees of his request, Schwartz must inform each of them of that request,and only by that means make the request of the particular ones who need asyllabus.It might be objected that vocatives shouldn't be considered a genuine partof the utterances in which they are found: they would be considered separateutterances whose function is to constrain the associated illocutionary acts.Thus Othello would be performing two utterances: with Desdemona, he wouldbe 'addressing' all three participants in order to designate which one is to bethe addressee of his other utterance; with come, he would be making a requestof the designated addressee. We could then separate the 'addressees' of thetwo utterances-the utterance of Desdemona is 'addressed' to three people,and the utterance of come only to one.Even if vocatives WERE assumed to be separate utterances, other addressee-specifying devices work the same way. Consider these examples:

355



26/43

LANGUAGE,VOLUME58, NUMBER2 (1982)(40) Othello, to Desdemona, in front of Iago and Roderigo: I want Des-demona to come.(41) George, to Alistair and Fergus: Would the last one to leave pleaseturnout the lights?(42) Schwartz, to history students: Any of you who needs a syllabus shouldraise your hand.Inthese parallels o exx. 1, 36, and 38respectively,the addresseesarespecifiedonly throughthe content of what is requested. The expressions Desdemonaand the last one to leave and any of you who needs a syllabus cannot beextirpatedand treated as separateutterances.Even 'hidden'vocatives, as in 40-42, areunnecessary.SupposethatGeorgeis on the roof, out of sight, and is requestingAlistair and Fergusto handhim

up certain tools one by one. Further,George doesn't know that Alistairhasthe saw. George could then say:(43) George, in the direction of Alistair and Fergus: Hand me the saw.Alistair and Fergus are both intendedto recognize that Georgeis addressinghis request to the person with the saw. How? From the fact that George isrequesting he saw, and the personwhohe intendsshouldcarryout thatrequestis the one with the saw. We mightcall this process 'addressingby attribution',as if George had said The one of you with the saw, hand it to me. He is treatingAlistair and Fergus as equally potential addressees, even though he hasn't

made that explicit with a vocative or other linguisticdevice. Here too, there-fore, Georgecan make his request only by meansof his informatives.Designatingaddressees by attribution,as in 43, is commonplacein adver-tisements. Whena televisionpitchmansuggeststo his viewers Treatyourchildto One-a-Dayvitamins,he is addressingonlythoseviewerswithminorchildren.He intends each viewer to understandwhat he is suggesting,in orderto rec-ognize whetheror not he or she is an addressee. The informativemust comefirst.4.3. THEPRINCIPLEF NDIVIDUALECOGNITION.ery often, what the speaker

is requesting s acoordinatedorjoint act on thepartof two or moreaddressees:(44) Noah, to Shem and Ham: Begin the trick now.Suppose that what Noah is requestingof Shem and Ham is a particularlycomplicatedknife-jugglingrick that the two of them can only dojointly. It isone of manysuchtricksthey coulddo; and if eitherone beginsthewrongtrick,it could be dangerous.Now Noah isn't askingShemand Hamto do theirpartsof the trickseparately.He is askingthe pairof themto do an act thatthey cando only as a pair, at least in the way he intends. He is makinga collectiverequest. If he had said Closeyour eyes, he wouldhave been makinga distrib-utive request, since he would be asking each of them to close his eyesindependently.Collective requests like 44 force us to make explicit a basic assumptionofall theories of illocutionaryacts. When a speaker pe

Date post:	03-Apr-2018
Category:	Documents
Upload:	nashwa-nashaat
View:	232 times
Download:	0 times

Clark and Carlson 1982 Hearer Ans SA

Documents