How to Be Sarcastic

7/27/2019 How to Be Sarcastic

1/13

Journal of Experimental Psychology: General1989, Vol. 118, No. 4,374-386 Copyright 1989 by the American Psychological Association, Inc.809


2/13

SARCASTIC IRONY 375the listener must recognize that intent. Furthermore, thelistener must infer what the speaker wants to communicateby saying something that is obviously not true. How dolisteners (a) recognize such intentions (i.e., to be ironic) and(b) infer specifically what a speaker intends to convey?According to Grice (1975), participants in a conversationobserve the cooperative principle. Listeners assume thatspeakers will be truthful and informative. When a speakersays something that ispatently untrue, and when both speakerand listener know this and know that each other knows this,and so on (see Clark & Carlson, 1981, for a description ofcommon ground), then a listener can make one of two inter-pretations: Either the speaker is violating the cooperativeprinciple or he or she is deliberately trying to communicatesomething byappearing^ violate that principle.When speak-ers appear to violate the cooperative principle, they implicitlyinvite their listeners to make an inference. In Grice's terms,apparent violations of the cooperative principle are taken as"conversational implicatures" (p. 45): They suggest to a lis-tener that there is a communicative intent behind the appar-ent violation. Being untruthful is one way to appear to violatethe cooperative principle. Being irrelevant, as in Example 5,is another:

MARY: Is Tom a good dancer?S H E I L A : He's got terrific taste in ties. (5 )Mary has asked a question and has received an apparentlyirrelevant answer. Rather than conclude that Sheila is beingunresponsive, Mary might infer that Sheila intends to com-municate that Tom is not a good dancer at all but that sheprefers not to say so directly, perhaps out of politeness con-siderations (Brown &Levinson, 1987).This type of analysis can account for why listeners gobeyond the meaning of what is said in cases of irony orsarcasm. A listener should generally assume that speakersintend to be truthful and relevant, even when a speaker sayssomething that is counter to the facts. One way in which aspeaker may be considered truthful in such cases is to under-stand him or her as intending the opposite of what is actuallysaid. This is precisely what the standard theory of ironyassumes: When someone says something that is patentlyopposite to the facts, then listeners interpret the utterance asmeaning the opposite ofwhat it appears to mean (Grice, 1975,1978; Searle, 1969).This form of the standard theory, however, does not ac-count for the most interesting and important aspects of irony.Tobegin with, this theory does not account for why a speakerwould choose to say the opposite of what is meant, nor doesit consider the potential relevance of saying the opposite ofwhat is meant (Sperber & Wilson, 1986). Surely if a speakerintends to communicate only proposition X, why try tocommunicate X by saying not-Jf? Why, fo r example, sa y"What a beautiful day" if all you want to communicate isthat it is a terrible day?If a speaker does follow Gricean maxims in order to berelevant, informative, and truthful, then he or she should notsa y "It's a beautiful day" merely to communicate the opposite(that it is a terrible day). After all, the listener presumablyknows that already, and so speakers must intend something

other (or more) than a straightforward description of an eventor a state of affairs. An ironic or sarcastic remark is, instead,a comment that isused to communicate the speaker's attitudetoward an event or a state of affairs (Grice, 1978), such asdisappointment with the weather itself or ridicule of a weatherforecast that had gone sadly awry. In either case, the expres-sion "It's a beautiful day" can remind a listener of what mighthave been expected and hoped for or of that inaccurateprediction. The reminding function of ironic expressions wasimplied bySperber and Wilson (1981,1986) in their treatmentof irony as a case of echoic interpretation. According toSperber and Wilson, listeners understand an ironically in-tended utterance byrecognizing that it isechoic, that it alludesto the thoughts, opinions, utterances, or behavior of someperson other than the speaker. Once theutterance is identifiedas echoic, then the source of the echoed opinion, and so on,can be identified. The speaker's attitude toward that source isalso recognized, usually as disapproval or, as in the case ofsarcastic irony, as ridicule or contempt.Such remindingcan be either implicit or explicit (Jorgeri-sen, Miller, & Sperber, 1984). If there is no immediatelyrelevant antecedent remark or event, then the reminding isimplicit, usually of an implicit positive norm or expectation

( see Boucher &Osgood, 1969, and Matlin & Stang, 1978, fora description of cultural norms of success and excellence).Ifthere is a relevant antecedent remark or event, such as anincorrect weather forecast, then the reminding is explicit, andthe object ofdisapproval or ridicule is that specific antecedentevent. In either case, the ironical utterance serves to expressan attitude, usually negative.According to Sperber and Wilson (1981, 1986), then, theliteral meaning of an ironic or sarcastic remark is preciselywhat the speaker intends, not its opposite. The utterance isinterpreted as an "echo," either of an implicit norm or expec-tation or of an explicit antecedent event. The communicativefunction of such echoic utterances is to remind a listener ofimplicit or explicit norms or expectations that are known toboth speaker and listener (Tannen, 1984). The speaker accom-plishes the reminding by saying the opposite of what is truebecause the listener can thus be reminded not only of a sharedexpectation or cultural norm but also of a discrepancy be-tween what is and what should be. By reminding listeners ofsuch a discrepancy, a speaker expresses disapproval of it.Sperber and Wilson (1981, 1986) referred to this account ofverbal irony as echoic interpretation. We propose to call thisaccount echoic reminder theory for two reasons. The first isthat this term highlights the reminder function of echoicutterances. The second reason is perhaps more important:Although all ironic utterances accomplish their communica-tive intent by reminding listeners of some antecedent event,not all such reminders are "echoic"that is, not all anteced-ent events are actual or even implied utterances. The remarkillustrated in Example 3, "Another gorgeous day," need notecho anyone's utterance, thought, or opinion. It simply al-ludes to a generalized expectation or desire for good weatherand, in so doing, expresses the speaker's disappointment atthe actual weather. Echoic interpretation, then, may well bea special case of reminders in general: allusions to prioroccurrences or states of affairs.


3/13

376 ROGER J. KREUZ AND SA M GLUCKSBERGFormulated in this way, the echoic reminder theory ofverbal irony not only provides a motivation for saying theopposite of what is true, but it also provides a principledexplanation for one of the most salient and important char-acteristics of verbal irony: the marked asymmetry of ironic

statements. In general, positive statements can readily be usedironically, as in-You're a fine friend!" (6)

in the context of a close acquaintance's failing to be helpfulwhen he or she should have been. In contrast, the negativestatement

"You're a terrible friendr1 (7)cannot be used ironically in the context of someone's beingextremely helpful, unless there is some reason to have ex-pected otherwise. For example, if a friend had earlier said thathe or she was terribly unreliable and cowardly and thencourageously foils an attempted mugging, then a negativestatement about this positive event could be used ironically.The statement is a reminder of the friend's prior predictionof cowardice and calls attention to the discrepancy betweenthat prediction and what actually happened. In so doing, thespeaker conveysan attitude of gentle mockery of that incor-rect prediction, along with admiration for the courageousbehavior.The principled reason for this asymmetry is the implicitexpectation of success or adherence to positive norms. AsJorgensen et al. (1984) put it, "Expectations of success areintrinsic to any action; culturally defined criteria ofexcellenceand rules of behavior are invoked in most value judgments.Thus it is always possible to mention those expectationsironically when they are frustrated,orto mention those normsironically when they are violated" (p. 1 IS). Implicit remind-ing, therefore, is sufficient for positive statements intendedironically because there are positive norms to be remindedof. In contrast, implicit reminding should not be sufficient fornegative statements intended ironically because normallythere are no implicit negative norms to be reminded of.Examples 8 and 9 illustrate this asymmetry:

A ca r ahead turns sharply without signaling. Y o u r companionsays, "Now there'sa terrific driver!" (8)*A ca r ahead signals a left turn, then turnsleft. Your companionsays, "Now there'sa terrible driver!" (9)

Example 8 isreadily understood as a sarcastic remark directedat the nonsignaling driver. Example 9 is uninterpretable,unless there is a relevant antecedent, such as an explicit priorexpectation that the driver in the car ahead would in fact failto signal properly (e.g., if one had expressed an opinion thatteenage drivers drove recklessly, and then a teenaged driverclearly and carefully signaled). In this latter case, the ironicremark reminds the listener of an explicit incorrect expecta-tion and, in so doing, ridicules that expectation.Jorgensen et al. (1984) tested Sperber and Wilson's (1981)echoic hypothesis by seeing whether people would be morelikely to perceive counterfactual statements as ironic when anexplicit antecedent was available. The basic assumption forthis prediction was that explicit antecedents should be easier

to identify than implicit antecedents. The results were equiv-ocal. Explicit antecedents did lead to more frequent judg-ments of irony than did implicit antecedents, but in only fourof the six test items that were used. Of more importance,Jorgensen et al. did not examine the critical contrast betweenpositive and negative counterfactual assertions. According toreminder theory, an explicit antecedent should not be neces-sary for positive statements about negative events becausepositive statements can remind people of implicit positiveexpectations or culturally shared norms. In contrast, negativestatements about positive events should require explicit an-tecedents because there are no implicit negative expectationsor cultural norms to be reminded of.Gibbs (1986a), in a similarly motivated study, asked collegestudents to understand and to remember stories that couldend with sarcastic remarks. Sarcastic remarks that explicitlyalluded to a prior statement or event in a story were under-stood more quickly than those that relied entirely on implicitantecedents; such remarks were also better remembered.Gibbs also found that positive sarcastic remarks were under-stood more readily than negative ones. However, the criticaldifference in the relative importance of explicit antecedentsfor positive and negative forms of sarcastic utterances wasagain not considered. This difference is critical because anytheory of comprehension, including the standard theory ofverbal irony, would predict that explicit antecedents shouldaid comprehension and subsequent memory.The purpose of our studies was to discriminate betweengeneral models of comprehension (including the standardtheory of verbal irony) and echoic reminder theory in thecontext of one form of verbal irony: sarcastic irony. Thestandard theory makes no differential predictions about therelative importance of explicit antecedents for positive, incomparison with negative, sarcastic utterances. For either typeof utterance, explicit antecedents merely provide additionalcontextual information. Echoic reminder theory, in contrast,predicts that positive sarcastic utterances do not require ex-plicit antecedents because such utterances can remind peopleof implicit positive expectations or cultural norms. Explicitantecedents may well be helpful because they can provideuseful contextual information, but they should not be neces-sary. Negative sarcastic utterances, in contrast, should requireexplicit antecedents because negative utterances would belesseffective remindersof implicit positive expectationsor norms.There should, therefore, be an interaction betweenthepolarityof a sarcastic utterance and the availability of an explicitantecedent. Specifically, explicit antecedents should be muchmore important forunderstanding negative sarcastic remarksthan for understanding positive sarcastic remarks.To test this prediction, we examined the role of one com-mon form of antecedent in sarcastic contexts: the victim of asarcastic remark.The availabilityofsomeone to bevictimizedprovides an explicit antecedent. When a remark is positiveabout a negative event, then victimless irony should be pos-

1 Following standard linguistic convention, we have denoted anom -alous utterances with an asterisk. In linguistics, these markers typicallyindicate syntactically anomalous utterances; in this article, they de -note pragmatically anomalousutterances.


4/13

SARCASTIC IRONY 377sible, as when someone says, as in Example 3, "Anothergorgeous day!" when the weather is terrible and when noexplicit victim, such as a mistaken weather forecaster, isinvolved. According to standard theory, the remark expressesthe opposite of what is factually the case and may also expressdislike of the weather. According to reminder theory, theremark may also express disappointment with the weatherand perhaps disdain for some imagined optimist. Providingan explicit victim as a target for such positive remarks mightaid and influence comprehension, but it should not be nec-essary, according to either theory. Now consider the oppositecase, in which someone says

""It's a terrible day, isn't it?" (10)when, in fact, the sun is shining brightly. In this case, victim-less irony is difficult (if not impossible) to interpret. Thestandard theory is silent on this issue. Reminder theory,however, predicts this outcome: Unless a listener can infersome plausible antecedent, such as a meteorologist's mistak-enly predicting terrible weather, then the statement cannot besensibly interpreted. Reminder theory, then, requires an ex-plicit antecedent for negative remarks about positive events.An explicit victim of such negative remarks can provide suchan antecedent.The logic of our three studies rests on this argument. Forstandard theory, victimless irony is acceptable and compre-hensible, irrespective of the polarity of a sarcastically intendedremark. For reminder theory, victimless irony is acceptableand comprehensible only for positive sarcastic remarks. Vic-timless irony should not be acceptable for negative remarks:These should require an explicit antecedent, such as somerelevant behavior of a victim.

General MethodIn all three experiments, weused the same set of very short stories(anecdotes, really) that had the following characteristics: (a) Eachconsisted of four sentences that described a brief interaction betweentwo people: a potential ironist and a potential victim. The twopeopleand a situation were introduced in the first sentence, as in thefollowing example:

Nancy and her friend Jane were planning a trip to the beach.(lla)(b) In the second sentence, the potential victim (in this case, Jane)said something about a future event: She uttered a positive or negativeprediction, or made a neutral statement, such as

"It's probably going to rain tomorrow,'' said Jane, whoworkedfor a local TV station as meteorologist. (lib)In this example, Jane was made out to be an expert, and she made anegative prediction. In half of the story types, the predictor was anexpert; in the remainder, the predictor was a layperson. W e felt thatan expert making a mistake would be a more salient victim than alayperson and so might provide a more effective target for a sarcasticremark, (c) An event, either positive or negative,wasdescribed in thethird sentence. If a prediction had been made, then the event eitherconfirmed or discontinued that prediction, as in

The next day waswarm and sunny. (lie)In this example, the event was positive and so discontinued Jane's

prediction, (d) The potential ironist either accurately described theevent or said the opposite of what had happened so as to make apotentially sarcastic remark:

As she looked out the window, Nancy said, "This certainly isawful weather." (11 d)In this example, the negative remark about the positive event shouldbe understood as sarcastic, ridiculing Jane's incorrect prediction. IfJane had not previously made aprediction (as in 1 Ib), but had simplysaid

"I hope everyone doesn't have the same idea", said Jane, whohated crowds, (lie)then Nancy's remark (Example 1 Id) would be difficult to interpretas sarcastic.In Table 1 we provide all the types from the a forementioned storyframe that we used; in the Appendix we provide examples from the20 story frames that we used. According to this scheme, there were20possible types ofstories. There were5initial statements that couldbe made by the potential victim: a positive prediction or a negativeprediction by either an expert or a layperson, and a neutral statementby someone whose expertise was not specified. The event that actuallyoccurred waseither positive or negative and thus either confirmed ordisconfirmed a prediction. Positive events disconfirmed negativepredictions, and vice versa. Last, the potential ironist either accuratelydescribed the event or contradicted the event. This format yielded 10types of stories in which the final remark was true and 10 in whichthe final remark was counterfactual and thus potentially sarcastic.In Experiment 1 , we asked college students to read the stories, on eat a time, and then to indicate their interpretation of the final remark

Table 1Variants of "The Beach Trip" Story

Prediction sentence: Five possibilitiesExpertise, positive prediction"The weather should be nice tomorrow," said Jane, who workedfor a local TV station as a meteorologist.Novice, positive prediction"The weather should be nice tomorrow," said Jane, who wasalways trying, with little success, to predict the weather.Expertise, negative prediction"It's probably going to rain tomorrow," said Jane, whoworkedfor a local TV station as a meteorologist.Novice, negative prediction"It's probably going to rain tomorrow," said Jane, who was al-ways trying, with little success, to predict the weather.Neutral statement"I hope everyonedoesn't have the same idea," said Jane, whodisliked crowds.

Event sentence: Two possibilitiesPositive outcomeThe next day was a warm and sunny one.Negative outcomeThe next day was a cold and stormy one.

Reaction sentence: Two possibilitiesPositive remarkAs she looked out the window, Nancy said, "This certainly isbeautiful weather."Negative remarkAs she looked out the window, Nancy said, "This certainly isawful weather."Note. The introductory sentence was the same for all stories: Nancyand her friend Jane were planning a trip to the beach.


5/13

378 ROGER J. KREUZ AND SAM GLUCKSBERGwith respect to three measures: (a) to report what the charactersmeant by their final remark in each story; (b) to rate how positive ornegative the final remark was; and (c) to rate how sensible that remarkwas. The 10 true-statement types were essentially fillers, althoughthey also provided useful baseline data. We expected that the truestatements would not elicit judgments ofsarcasm or irony, would berated in accord with their surface polarity (positive statements wouldbe judged as positive, and negative as negative), and would all makesense. Of more interest, of course, were the responses to the 10counterfactual story types.

Infive of these types, a positive remarkwasmade about a negativeevent (see the Appendix for examples).This is the canonical form forsarcastic irony, and so we expected all five of these types to beidentified as ironic or sarcastic, to be rated as negative despite theirpositive surface form, and to be judged as sensible. Presence orabsence of an explicit victim would have little, if any, effect on anyof these three measures of comprehension. In the other five counter-factual types, a negative remark was made about a positive event(e.g., a sunny day is referred to as "awful weather"). In these types,presence or absence of an explicit potential victim would make adifference. In two of these five counterfactual types, a potential victimmade an incorrect prediction. This victim was either an expert or anonexpert in the domain of the prediction (see the example in Table1). We expected these two types to be indistinguishable from the fivecanonical types that contained positive remarks about negative eventsbecause the availability of an explicit antecedent should have madeechoic reminding possible. Greater expertise on the part of a victimshould have increased the salience of the victim's gaffe and shouldtherefore have affected the degree ofnegative evaluation and perhapsthe ratings ofsarcasm or irony aswell.The three counterfactual types that contained negative remarksabout positive events without an explicit victim (i.e., either no onemade a prediction or the prediction was accurate; an example appearsin Table 2) were the critical conditions. They were critical in thatthey provided no antecedent of which to be reminded. Therefore,they should have made less sense than the stories that did provide anantecedent: implicit antecedents when positive counterfactual re-marks were made, and explicit antecedents when negative counter-factual remarks were made with an available victim. People mightalso be less willing to identify negative remarks as sarcastic when novictim is available. The results reported by Jorgensen et al. (1984)and by Gibbs (1986a) would be consistent with such a finding.Alternatively, people might judge any blatantly counterfactual state-ment aspotentially ironic or sarcastic, but they would be puzzled bynegative statements in the absence of any explicit antecedent, such asa victim.

Experiment 1: Effect of Victim Availabilityon Interpretation of CounterfactualsIn this experiment we tested the reminder theory prediction

that explicit antecedents are more important for negativesarcastic statements than for positive sarcastic statements.

Table 2Example of a Story in Which a Negative Remark fs MadeAbout a Positive Event Without an Explicit VictimNancy and her friend Jane were planninga trip to the beach."The weather should be nice tomorrow," said Jane, wh o worked fora local TV station as a meteorologist.The next day was a warm and sunny one.As she looked out the window, Nancy said, "This certainly is awfulweather."

Standard theory, as we noted earlier, makes no differentialpredictions for positive versus negative statements. Weaskedpeople to interpret and then to rate the sensibility and affectivevalence of both negativeand positive sarcastic remarks whena victim was either available or not. Availabilityof a victimin these conditions provided an explicit antecedent for thepotentially sarcastic remark.

MethodSubjects. One hundred nine introductory psychology students atPrinceton University served as volunteer subjects in groups of 12-18students each.Materials and design. Each subject received a booklet with in-

structions and 20 stories, as described earlier. Each subject saw onlyon e story of each type an d only on e frame from each story. Th e storytypes were presented in 60 orders, an d story type an d story framewere systematically varied so that each story frame appeared approx-imately equally often in each type. For example, the story frame of"The BeachTrip" (see Table 1) appeared approximately equally oftenin each of the 20 types, and any given subject saw this story in onlyone type. The distribution of story types across story frames was onlyapproximately equal because we had 109 subjects instead of a numberevenly divisible by 20.

This produced a 2 X 5 design, wherein the 10 true-statement typesserved essentially as fillers. The factors in this design were polarity ofstatement (positive or negative) and nature of antecedent. In twoconditions a victim provided an explicit antecedent in the form of aninaccurate prediction, and this victim either was an expert or wasnot. In two other conditions an accurate prediction was made. Thisprovided only a general topic as a potential antecedent, and we didno t expect this to suffice fo r interpreting a counterfactual remark assarcastic. In the fifth condition, nopredictionsconcerning the criticalevent were made, and sothere was no antecedent at all for a potentialsarcastic remark. Therefore, we expected that counterfactual remarkswould be difficult to interpret as sarcastic.This design completely confounded subjects and items, and soseparate analyses with subjects and items as random factors wereneither possible nor required.Procedure. Subjects began the experiment by reading the follow-ing instruction printed on the first page of their booklets:

We're interested in how people use language to communicateideas to each other. As you are probably aware, people canaccomplish this in a variety of ways. You can help us betterunderstand this process by participating in this experiment.In the following pages, yo u will read a number of short stories.In each story, two characters will have a brief interaction, and inthe final sentence, one of the characters will say something tothe other character.We'd like you to evaluate this final comment You will beasked what the character means by making this comment, andyou will also be asked to rate how positive or negative this finalcomment was. Finally, we would like you to indicate how muchsense the comment made in the situation.The best way to do this is to quickly read through each story,and then go backand concentrate on the important parts. Finally,when you are sure that you fully understand the interaction thatis described, you can answer the questions.Before yo u begin making your ratings, look over a few of thestories to get a feel for the amount of variety they have. If youfind yourself unsure about what kind of rating to make, just us eyour best judgment. Remember, there are no right or wronganswers; we are interested in your intuitions. Please be sure toanswer all the questions. Thank you very much for your partic-ipation.


6/13

SARCASTIC IRONY 379The subjects then read the stories and answered the three questionsfor each story. The meaning question was"What did [character name]

mean by his/her final comment?" The polarity question was "Howpositive or negative was [character's] final remark?"; subjects an-swered by indicating their response on a 7-point scale on which 1 =very negative and 7 = very positive. The sensibleness question was"How much sense did [character's] final comment make?"; again,subjects used a 7-point scale on which 1 = very little sense and 7 = alo t of sense.After subjects had completed their booklets, they were given awritten description of the purpose and design of the experiment.

Results and DiscussionThe first question that subjects answered concerned theinterpretation of the final, potentially sarcastic remark. Weinitially studied the entire set of answers to the open-endedquestion "What did [character name] mean by her/his final

comment?"; then we grouped responses into 14 categories.The critical category for our purposes was "Sarcastic." Aresponse wa s judged to belong to this category if the responseactually contained one of the words sarcastic, sarcasm, orsome variant of these specific terms. A response was judgedto belongto the "Ironic" category only if the response actuallycontained one of the words ironic, irony, ironical, or a variantof these terms. Th e other categories were "Character W asAngry"; "Compliment"; "Fact"; "Insult"; "Joking"; "Oppo-site Meant"; "Surprise"; "Character Was Right"; "CharacterWas Wrong"; and "Don't Know". In each of these categories,subjects could use a range of terms (e.g., responses such as"kidding," "joking," and "being funny" were all categorizedas "Joking"; "don't know" and "can't say" were subsumedunder "Don't Know"). Responses that paraphrased the targetremarkwere labeled as"Paraphrase,"and responses that couldnot be placed into any of the categories just mentioned werelabeled "Other."Interpretations of sarcasm or irony were rare for the truestory types: Only 2.4%of the responses indicated an interpre-tation of these stories as sarcastic or ironic. In contrast,sarcasm was the modal interpretation for 9 o f the 10 counter-factual story types. A preliminary analysis revealed that ex-pertise of victim or potential victim had no effect on propor-tion of sarcasm interpretations, and so the data were pooledacross this variable. Similarly, there was no reliable differencebetween types with a correct prediction and types with noprediction at all, and so these data were pooled into a "no-victim" condition. These data poolings yielded a 2 x 2 designfor analysis. In Table 3 we present the proportion of sarcasticinterpretations as a function of polarity of final remark andavailability of victim.These data were subjected to a 2 x 2 (Polarity X Victim)analysis of variance( A N O V A ) . As expected, the polarity of thefinal remark had a reliable effect; positive statements yieldedmore judgments ofsarcasm than did negative statements (.39vs. .32), F(l, 108) = 5.42, MS, = 0.08, p < .03. This reflectsthe well-known asymmetry of sarcastic/ironic remarks. Pres-ence or absence of an explicit antecedent in the form of avictim also had a reliable effect. Remarks in the context of anidentifiable victim were more likely to be interpreted as

Table 3Proportion of Sarcastic Interpretations as a Function ofStatement Polarity an d Victim AvailabilityAvailability of victim

Statement polarityPositiveNegative

Victim.46.41

No victim.32.23

sarcastic than were identical remarks in the absence of avictim (.44vs. .28), F(l, 108) = 39.95, MSe = 0.06, p < .001.The interaction between statement polarity and victim wasin the predicted direction but was not reliable (F< 1). In thestories that we used, of course, both participants in the con-versation knew what happened, and so counterfactual state-ments could not be interpreted as lies or as attempts todeceive. Under such circumstances, a reasonable interpreta-tion of such obviously counterfactual statements is that theywere intended sarcastically. This would account for the tend-ency to interpret negative counterfactual remarks as sarcasticeven when there was no explicit victim and, presumably, nonegative expectation or cultural norm of which to be re-minded.This finding would be consistent with reminder theory ifsuch counterfactual remarks are interpreted as attempts atsarcasm that do not quite work. To determine whether this isthe case, we turn to the ratings of sensibleness. In Table 4 wepresent the mean sense ratings for the 10 counterfactual storytypes, collapsed across expertise of potential victim. Foursubjects did not complete their questionnaires, and so theanalysis of the sense rating data was based on the data of the105 remaining subjects. A 2 x 2 A N O V A ofthese data revealedthe predicted main effects of statement polarity and availabil-ity of victim and, of more importance, the interaction of thesetw o factors.Positive statements were rated as more sensible than werenegative statements (Ms = 4.67and 3.99), F(l, 104)= 36.00,M Se = 1.57, p < .001. As expected, presence of victim hadreliable effects on the sense ratings. Th e mean sense ratingfo r remarks with an explicit victim (4.83) was higher thanthat for remarks without a victim (3.86), F(l, 104) = 90.76,MSt = 1.05, p < .001. The interaction of statement polaritywith victim was consistent with reminder theory: Presence ofa victim was more important for negative statements than forpositive ones. Positive statements with and without victims

Table 4"Sense" Ratings as a Function of Statement Polarity andVictim AvailabilityAvailability of victim

Statement polarity Victim No victimPositiveMSDNegativeMSD

5.051.554.611.65

4.351.733.371.67


7/13

380 ROGER J. KREUZ AND SAM GLUCKSBERGwere rated 5.05 and 4.61, respectively, and negative state-ments with and without victims were rated 4.35 and 3.37,respectively. This interaction was reliable, F(l, 104) = 6.50,MS.= 1.23,p


8/13

SARCASTIC IRONY 381casm ratings were 1.9 for the true statements and 6.3 for thecounterfactual statements, and the twogroups differed signif-icantly, f(18) - 21.19,p< .01.W e subjected the counterfactual statements to a separateanalysis to seewhether the presence of a victim had a greatereffect for negative statements than for positive statements.Asin Experiment 1, the expertise of the victims had no reliableeffect, and so the data were pooled. In Table 6 wepresent themean sarcasm ratings as a function of victim and statementpolarity.These data were subjected to a 2 x 2 A N O V A . In keepingwith the hypothesis that social norms provide implicit ante-cedents for positive statements, positive statements were ratedas more sarcastic than negative statements, F(l, 39) - 8.24,MS, = 1.11, p < .01. Presence of victim also had a reliableeffect, which was consistent with the notion that availabilityofan explicit antecedent is important for sarcasm, F(l, 39) =22.00, MSt = 0.81, p


9/13

382 ROGER J. KREUZ AND SAM GLUCKSBERG

included these ratings of appropriateness to see whether sub-jects' understanding of counterfactual remarks would be in-fluenced by statement polarity and availability of an explicitantecedent. Standard theory would predict no interaction ofstatement polarity and antecedent; echoic reminder theorypredicts that antecedence availability is more important fornegatively phrased remarks than for positively phrased sar-castic remarks.

Experiment 3: Degree and Appropriatenessof Sarcasm as a Functionof Statement Polarityand Victim AvailabilityMethod

Subjects. Forty undergraduate students at Princeton Universityserved as paid subjects. All were native English speakers, and nonehad served in Experiments 1 or 2.Materials and procedure. The same questionnaire as in Experi-ment 2 was used here, but the 10 true-remark story types were notincluded. In addition, subjects were asked to rate both the degree ofsarcasm and the degree to which the target statements were appro-priate in each story.Results and Discussion

As in Experiments 1 and 2, the expertise of victim had noreliable main effects or interactions, and so the data werepooled. In Table 8 we present the mean ratings for degree ofsarcasm and appropriateness. Each of these sets of ratingswere subjected to a 2 X 2 A N O V A .Rated sarcasm again varied as a function of statementpolarity; positive statements were rated asmore sarcastic thanwere negative statements, F(l, 38) = 12.54, MSC = 0.60, p


10/13

SARCASTIC IRONY 383victim, positive statements were rated as more sensible thannegative. The data from Experiments 2 and 3 essentiallyreplicated this pattern of results. When asked to rate degreeof sarcasm, people were not differentially sensitive to victimpresence as a function of the positive or negative form of thestatements. However, when asked to rate how appropriatesuch sarcastic remarks were, victim presence became moreimportant fo r negative statements than fo r positive state-ments. These data, taken as a whole, are consistent with thenotion of implicit reminding through positive social normsand expectations. These data are also generally consistentwith the findings of Jorgensen et al. (1984), who reported thatpeople had difficulty interpreting potentially ironic remarkswhen there was no explicit antecedent.In that study, Jorgensen et al. were interested in testingSperber and Wilson's (1981) echoic mention theory ofverbalirony. If we consider sarcasm as a particular type of verbalirony, and if we confine consideration of the theories to thetype of sarcasm that weused in these experiments, then echoicmention and echoic reminder theories are closely related toone another. According to mention theory, a person whoutters a counterfactual statement, intended ironically, is saidto mentionthe idea that is literally expressed by that statementin order to communicate a derogatory attitude toward thatidea (Sperber, 1984). Presumably, when sarcasm is specificallyintended, a person who may express or accept such an idea isalso derogated, albeit indirectly. The critical claim here is thatthe intended target is mentioned, not used, in the technicallinguistic sense of these terms. The difference between us eand mention is illustrated in these two expressions that con-tain the word page:

"M y typewriter gets about 250 words on a standard page." (12)"The wordpage is contained in the sentence above." (13)

In sentence 12, the word page is used to refer to the concept"page"; in sentence 13 (and in this sentence as well) the wordpage is mentioned to refer to itself as a word.In expressions such as Example 3 ("Another gorgeousday"), the sentence can be used to refer to an actual state ofthe weather. This same sentence can be mentioned as avehicleof ridicule or disapproval (Jorgensen et al., 1984). This men-tion in turn can serve as a remainder ofwhat isbeing ridiculedor disapproved of.Mention theory, then, is primarilyaddressed to the issueofhow the language itself is used. Echoic reminder theory isperfectly consistent with this account, but it is addressed to adifferent aspect of the problem, namely, how the communi-cative goal of expressing an attitude is accomplished. Inmention theory terms, an ironic utterance (and byextension,sarcastic ironic utterances) "echoes" a thought that corre-sponds to the literal meaning of the utterance. The irony issuccessful if the listener recognizes the "echo" (i.e., recognizesthe meaning of the utterance and can attribute it to "specificpeople, specific types of people, or people in general"; Sperber,1984, p. 132). In other words, irony succeeds if it drawsattention to an idea and to a source of that ideathat is, theintended target. Drawing attention to the targetbe it an ideaor a specific personcan be done only if the target is known

to the listener. As Clark and Gerrig (1984) pointed out, ironycan be accomplished only when speaker and listener sharerelevant knowledge. Clark and Gerrig used the example oftw o people, Harry and Tom, who have just attended a lecturethat ha d been unexpectedly fascinating. As they leave thelecture hall, they run into Anne, and Harry says"Tedious lecture, wasn't it?" (14)

Tom, if he now remembers the earlier expectation of a tediouslecture, will understand the statement as ironic. Anne, whoha s no relevant antecedent in memory, can understand thestatement only to be true or else nonsensical.M ention theory, because it is not explicitly addressed tosuch issues as shared relevant knowledge, is incomplete as apsychological account of how ironic or sarcastic communi-cations are accomplished. In a more recent treatment of therole of relevance in language comprehension, Sperber andWilson (1986) noted that the concept ofmention fails to covera range of cases: "Mention is a self-referential ... use oflanguage: it requires full linguistic or logical identity betweenrepresentation and original.... Wehave therefore abandonedthe term 'mention' in favor of the more general term 'inter-pretation'" (p. 264).One example of the limitations of mention is the case inwhich the literal meaning of an expression is used but sarcasticirony is intended as well. One common example of this is aspeaker's being excessively polite in order to be sarcastic.Brown and Levinson (1987) pointed out that within a givenculture, the degree of politeness of , say, requests is determinedby such factors as the amount of imposition on the addresseeand the social status differential between requester and ad-dressee, among other things. Thus a mother being excessivelypolite to her daughter, as in

"Would yo u very much mind if I asked you, please, to perhapsconsider cleaning up your room sometime this month?" (15)is simultaneously expressing several communicative inten-tions. She is directly asking a question, indirectly making arequest, and also being sarcastic. This is clearly a case of use,not mention. The statement simultaneously communicates arequest and remindsin the sense of calling attention to the speaker's attitude towards the listener's own customaryattitudes and behaviors. That this functions as a reminder ofthe attitudes and expectations of both parties is indicated bythe fact that such sarcasm would be totally misplaced if suchan excessively polite request were to be made of a stranger.This kind of sarcasm ca n succeed only when speaker andaddressee can share expectations and when the sarcastic re-mark can remind the listener (and perhaps those who over-hear) of those expectations. In Sperber and Wilson's (1986)terms, such remarks can be interpreted echoicallythat is, asalluding to something that both speaker and listener know.This example suggests that the concept of mention versusus e ca n provide an adequate account of verbal irony fo r thosecases in which literal meanings are not intendedthat is, forcases in which meanings are mentioned instead of being used.Echoic reminder theory accounts fo r these cases and also fo rcases in which literal meanings are intended and are used, as


11/13

384 ROGER J. KREUZ AND SAM GLUCKSBERGin Example 15. Mention, then, can be considered a specialcase of echoic reminding, in which implicit antecedents areechoically mentioned in order to remind a listener of thoseantecedents and, in sarcastic irony, to derogate an idea andthe source of that idea.

One other proposal merits consideration here. Clark andGerrig (1984) argued that verbal irony involves pretense ratherthan echoic mention. Citing Grice (1978), "To be ironical is,among other things, to pretend" (p. 125), Clark and Gerrigused an example that corresponds closely to the kinds ofremarks that weused in our experimental materials. A speakersays, in the context of a rain storm:"Trust the weather bureau! See what lovely weather it is: rain,rain, rain." (Cited from Jorgensen et al., 1984,p. 114) (16)

Clark and Gerrig argued that "the speaker is pretending to bean unseeing person, perhaps a weather forecaster, exclaimingto an unknown audience how beautiful the weather is" (p.122). If the pretense involved only the appearance of assertingthe opposite of what is clearly the case, then pretense theorywould be a notational variant of echoic mention theory. Clarkand Gerrig asserted that the pretense goes beyond this: Thespeaker is pretending to be someone else (the "unseeingperson") and is also pretending to be talking to some personother than the listener. If the listener recognizes this pretense,then the listener understands that the speaker is expressing aderogatory attitude toward three entities: the idea expressed,the imaginary speaker, and the imaginary listener. Theseimaginary speakers and listeners "may be recognizable indi-viduals (like the TV weather forecaster) or people of recogniz-able types [such as inaccurate weather forecasters in general]"(p. 122).Clarkand Gerrig (1984) preferred pretense theory to echoicmention theory on grounds similar to those discussed earlier.There are clear cases of irony and of sarcasm that seem notto involve mention at all, such as Jonathan Swift's (1729/1971) essay, "A Modest Proposal." This essay proposed thatIrish children be used as food to serve to the rich and soprovide income for their poor parents while simultaneouslyexpanding the culinary repertoire. Clark and Gerrig correctlypointed out that this entire essay could not be a case ofmention, let alone echoic mention. It can, however,be treatedas a pretense, in which Swift is pretending to be someone else,speaking to an imaginary audience who would accept theideas that he is proposing in all seriousness. The idea and theimaginary speaker and audience are all simultaneously dero-gated by this pretense.Clark and Gerrig (1984) were correct in asserting that thepretense theory ismore powerful than echoic mention theory.Pretense theory may, however, be too powerful and toolimited simultaneously. The pretense notion can, with littlestretching, be applied to all indirect speech acts. When some-one says, "Can you pass the salt?" pretense can be invoked:The speaker is only pretending that some obstacle preventsthe addressee from complying with the request (see Gibbs,1986b). When someone says, "It sure is hot in here!" thespeaker is only pretending to comment on the temperaturewhen in fact she wants someone to open a window.

When applied to sarcastic irony, the pretense theory in thisvery general sense can hold. However, the special sense of aspeaker's pretending three thingsbelief in the assertion,pretending to be someone else, and talking to an imaginaryaudienceseemsboth unnecessary and implausible. Who arethe imaginary speaker and audience in cases such asExample15, in which excessive politeness is used to convey sarcasm?Even when sarcasm is accomplished by verbal irony, as inExample 16 ("What beautiful weather"), the elaborated pre-tense notion seems unnecessary. As our results show, sarcasmis suspected whenever a remark is blatantly counterfactual.Sarcasm is further judged to be sensible and appropriate whenthe listener can be reminded of a positive implicit expectationor when there is an identifiable, explicit victim. The notionof pretense to accomplish such reminding seems unnecessary.Therefore, on grounds ofgenerality and of parsimony, echoicreminder theory seems preferable to pretense theory.Last, pretense theory and mention theory may not differsignificantly from one another. As Williams (1984) pointedout, there may be no linguistic difference between utterancesthat involve mention and utterances that involve pretense:"Most participant utterances seem to be explained plausiblyin terms of either mention or pretense [A] more elabo-rated version of [mention theory] would probably look verymuch like ... pretense theory" (p. 129). Both mention andpretense theory are theories of linguistic use. In both theories,the communicative function of mention or pretense is tocommunicate an attitude by calling attention to an idea thatboth speaker and listener can derogate. Both pretense andmention, then, serve to remind a listener of an idea or thesource of that idea (i.e., a victim). Echoic reminder theory isthe more general of the three theories because it applies tothose cases in which neither mention nor pretense is involved.W e agree with Williams (1984) that "the distinction be-tween the mention and the pretense theories is ultimately anissue for linguistic analysis and is not a matter for psycholog-ical test at all" (p. 129). This position is consistent with bothSperber and Wilson's (1986) and our view that mention is aconcept of linguistic use. Echoic reminding (or, as Sperber &Wilson put it, echoic interpretation) is, in contrast, apsycho-logical concept and so may hold promise as the basis for apsychologically testable theory of verbal irony in general andof sarcasm in particular. One task for linguistic theory wouldthen be to describe the linguistic devices that people use toremind one another of the intended objects of irony andsarcasm.

ReferencesBoucher, J., & Osgood, C. E. (1969). The Pollyanna hypothesis.Journal of Verbal Learning an d Verbal Behavior, 8, 1-8.Brown, P., & Levinson, S. C. (1987). Politeness: Some universals inlanguage use.Cambridge, England: Cambridge University Press.Clark, H. H., & Carlson, T. B. (1981). Context for comprehension.In J. Long & A. Baddeley (Eds.), Attention and performance IX(pp. 313-330). Hillsdale, NJ: Erlbaum.Clark, H. H., & Gerrig, R. J. (1984). On the pretense theory of irony.Journal of Experimental Psychology: General, 113, 121-126.Gibbs, R. W., Jr. (1986a). On the psycholinguistics of sarcasm.Journal of Experimental Psychology: General, 115, 3-15.


12/13

SARCASTIC IRONY 385Gibbs, R. W., Jr. (1986b). What makes some indirect speech actsconventional? Journal of Memory an d Language, 25,181-196.Grice, H. P. (1975). Logic and conversation. In P. Cole & J. L.M organ (Eds.), Syntax an d semantics, Volume 3: Speech acts (pp.41-58). NewYork: Academic Press.Grice, H. P. (1978). Further notes on logic and conversation. In P.

Cole (Ed.), Syntax and semantics. Volume 9: Pragmatics (pp. 113-127). NewYork: Academic Press.Jorgensen, J., Miller, G. A., & Sperber, D. (1984). Test of the mentiontheory of irorty. Journal of Experimental Psychology: General, 113,112-120.Matlin, M . W . , & Stang, D. J. (1978). Th e Pollyanna principle:Selectivity in language, memory and thought. Cambridge, M A:Schenkman.Miller, G. A. , & Glucksberg, S. (1988). Psycholinguistic aspects ofsemantics and pragmatics. In D. Luce, R. A. Atkinson, & R.Herrnstein (Eds.), Stevens' handbook of experimental psychology,second edition (pp. 417-472). New York: Wiley.M uecke, D. C. (1970). Irony: Th e critical idiom.London: Methuen.M uecke, D. C. (1980). Th e compass of irony (Rev. ed.). London:M e th u en .

Searle, J. R. (1969). Speech acts. Cambridge, England: CambridgeUniversity Press.Sperber, D. (1984). Verbal irony: Pretense orechoic mention? Journalof Experimental Psychology: General, 113, 130-136.Sperber, D., & Wilson, D. (1981). Irony and the use-mention dis-tinction. In P. Cole (Ed.), Radical pragmatics (pp. 296-318).NewYork: Academic Press.Sperber, D., & Wilson, D. (1986). Relevance: Communication andcognition. Cambridge, MA: Harvard University Press.Swift, J. S. (1971). A modest proposal for preventing the children ofthe poor people from being a burden to their parents or country,and for making them beneficial to the public. In T. Scott (Ed.),Th e prose works of Jonathan Swift (Vol. 7, pp. 205-216). Ne wYork: AM S Press. (Original work published 1729)Tannen, D . (1984). Conversational style: Analyzing talk amongfriends. Norwood, NJ: Ablex.Webster's ninth new collegiate dictionary (1988). Springfield, M A:M erriam-W ebster.Williams, J. P. (1984). Does mention (or pretense) exhaust theconcept of irony? Journal of Experimental Psychology: General,113, 127-129.

AppendixThe 20 story frames used in the experiments are reproduced asfollows. All are in the form of expert victim, positive prediction,negative outcome, and positive remark. For an example of all thestory types from one story frame, see Table 1.

Th e Beach TripNancy and her friend Jane were planning a trip to the beach."The weather should be nice tomorrow," said Jane, who workedfor a local TV station as a meteorologist.The next day was a cold and stormy one.As she looked out the window, Nancy said, "This certainly isbeautiful weather."

Th e LectureJohn and Steve were walking together to their morning class.As they entered the lecture hall, Steve said, "I read over theassignment pretty carefully, and I'll bet this is going to be a greatlecture."The professor proceeded to give a dry and boring presentation ofthe material.As they left the lecture hall, John said to Steve, "Great lecture,wasn't it?"

Th e ElectionTom and Dave were discussing the big election over drinks at abar."I'll bet the mayor's race will be no contest this year; Fve beenfollowing the polls pretty closely," said Dave.A couple of hours later, they learned that the mayor had beenreelected by a very slim margin.Tom commented, "Gee, that guy running against the mayor gotblown away,didn't he?"

Th e Fishing TripM ik e was preparing to go fishing, and Jim was watching him getready."I was talking to some of the guys just getting back from the lake,and they say the fish are really biting this year," Mike remarked.In the evening, Mike returned without any fish at all.Jim remarked, "Well, it looks like they're really biting this year."

Th e Stock MarketFrankhad just invested in the stock market for the first time."You did the right thing; I've been investing for years, and I thinkthe market's about to take off," his friend Larry told him.Over the next few weeks, the stock market fell sharply, an d Franklost a lot of money.Th e next time Frank saw Larry, he said "Well, I really made akilling on the market."

Th e BabyLinda stopped by to visit Mary, who was expecting her first childan d really hoping for a son."I'm sure it'll be a boy; most first births are," said Linda, who

worked in the maternity ward of the local hospital.A week later, Linda drove Mary to the hospital, where she gavebirth to a daughter.A few days later, Mary phoned Linda, and said "You must comevisit my beautiful baby boy."Th e Exam

Carol and Sam had just taken their calculus midterm."I think I did pretty well," said Sam, who had studied a lot beforethe test.A few days later, Sam found out that he had gotten one of thelowest grades in the class.


13/13

386 ROGER J. KREUZ AND SAM GLUCKSBERG

When Carol heard the news, she said to Sam, "M aybe you canhelp me study for the next test."Th e Plane

Cathy and Bob were waiting at the airport for their plane to arrive."I fly on this airline all the time; I'm sure the plane will be ontime," Cathy told Bob.The plane landed over an hour behind schedule.As they were boarding, Bo b said "Thesepunctual airlines are reallyimpressive."Th e Chess Game

Karen and Ed were playing a game ofchess.Karen knew that Ed was an expert player, and she sighed w hen Edsaid, "Y ou played well, but I'll finish you off soon."A few minu tes later, Ed lost the game.Karen said, "Gee, you sure finished M Eoff pretty fast."Th e Piano Recital

Susan waswaiting to go onstage for her piano recital.She told Paul, "I'm really going to play well tonight," since shehad practiced for many days.When it was her turn, Susan played her piece v ery poorly.After the recital, Paul remarked to a friend, "What a wonderfulperformance!"Th e Restaurant

Joan a nd M ark were waiting for service at Joan's favorite restau-rant ."This shouldn't take any time at all," said M ark, who had heardmany positive stories a bout the restaurant's service.It took over half an hour for a waiter to appear.After they ordered, Joan remarked, "The service is really goodhere, isn't it?"Th e Auction

Su e and Jerry were at an art auction."I know quite a bit about art, and the next painting looks like avery valuable one," said Sue.Few people bid for the painting, and it sold for only a few hundreddollars.Jerry said, "Wow, the bidding wa s really high for that picture."

Th e Fuel GaugeBetty and Sally were on a trip in Betty's old car."The fuel gauge in this car doesn't work, but we have enough gasto get where we're going," said Betty, and Sally knew that Bettyusually kn ew what she was talking about.A few minutes later, the engine sputtered and died.Sally said, "Well, it looks like we had enough gas."

Th e SweatersGeorge was doing his laundry, and his friend Fred was helpinghim."Y our sweaters won't shrink if you wash them with your shirts; Iwash them together all the time, so I know," said Fred.

George went ahead and washed all his clothes together, and hissweaters shrank several sizes.Holding a sweater against himself, George said, "They didn't shrinkat all, did they?"Th e Opera

Judy and Joyce had decided to go to the opera together, and theywere waiting for the orchestra to begin."Y ou should really like thisit's a great production," said Judy,who had seen the opera the night before.The performance was uninspired, and the soloists were uniformlypoor.As they emerged from the music hall, Joyce said, "I just love greatperformances!"Th e Bow ling Incident

Bill and Ann had decided to go bowling."I'm feeling pretty luck y tonight," said B ill, who was a very goodbowler.A few minutes later, they began their game, an d Bill ha d severalgutter balls in a row.Ann turned to Bill, and said "Well, I bet you're really glad wecame."

Th e CakeDianne was taking a home economics course, and she decided topractice baking a cake."I'd like you to try this cakeI'm a good baker, you know,"Dianne remarked to Jack and his brother.Diann e pulled the cake out of the oven, and it was far too dry.Jack and his brother tried a slice, and they said to each other, "Thiscake is R E A L L Y delicious."

Th e ClockHarry wa s helping Pa t move into her new apartment."Don't worry, I can move this grandfather's clock by myself," saidHarry, who was very muscular.Harry only managed to tip the clock over, and it crashed to thefloor.Pat looked up from some boxes she wasmoving, and said "You'rereally helping me out."Th e Dinner Party

Betty an d Jean were on their way to a formal dinner party."I'll bet I really ma ke a good impression," said Betty, who was verysophisticated an d graceful.During the dinner, Betty managed to spill her soup all over thehostess.As they were leaving the party, Jean remarked, "You sure were thehit of the party!"Th e Commuters

Pete and Rick were in a hurry to get to work."We can avoid all the traffic if we stay on this road; I know thisarea pretty well," said Pete.A few minutes later, their car was bogged dow n in a huge trafficjam , and Pete was slowed to a crawl.Rick observed, "This route sure saved us a lot of time!"Received July 29, 1988Revision received March 14, 1989Accepted March 21, 1988 "

How to Be Sarcastic

Documents

Transcript of How to Be Sarcastic