Irony and Language Expectancy Theory: Evaluations of Expectancy Violation Outcomes

Communication Studies Vol. 61, No. 3, July August 2010, pp. 356 372 Irony and Language Expectancy Theory: Evaluations of Expectancy Violation Outcomes Joshua M. Averbeck Language expectancy theory (LET) provides the basis for a study of the relationship of ironic and literal messages in terms of persuasiveness and expectedness. The experimental design was a 2 (literal vs. ironic) 2 (compliment vs. criticism) factor design. The expectation was that literal messages would be more expected and criticisms less preferred to compliments. The results indicated that the ironic messages tended to be negative, unexpected, and attributed to the situation as opposed to the sender. Compliments were associated with fewer behavioral intentions than were criticisms. The interaction also revealed that the ironic messages were a safe middle ground between literal criticisms and compliments in terms of fostering behavioral changes. Keywords: Indirect Language; Irony; Message Production; Persuasion There are a variety of goals one may seek to achieve during an interpersonal interaction; e.g., determine who picks the kids up from school to whether or not to purchase a new flat-screen television. Given the wide range of interpersonal goals people have, there are also a variety of strategies for satisfying them. Some researchers have attempted to study these measures by examining the acceptability of directness in close relationships (Dillard, Kinney, & Cruz, 1996), going behind someone s back to gain compliance (Beatty, Valencic, Rudd, & Dobos, 1999), and the use of indirect messages in face-to-face interactions (Anolli, Ciceri, & Infantino, 2002; Dews & Winner, 1995). Joshua M. Averbeck is a PhD Candidate at the University of Oklahoma. Correspondence to: Joshua M. Averbeck, Department of Communication, University of Oklahoma, 660 Parrington Oval, Norman, OK 73019. E-mail: jmaverbeck@gmail.com. ISSN 1051-0974 (print)/issn 1745-1035 (online) # 2010 Central States Communication Association DOI: 10.1080/10510971003776147

Irony and Language Expectancy Theory 357 It is the indirect message that may be the least appreciated. A specific instance of the indirect message is irony. An ironic message is a blatantly false message containing some counterattitudinal information with the intent of being actively detected by the receiver as being false (Averbeck & Hample, 2008; Kreuz & Link, 2002). To put it simply, it is a message that contains some falsehood that is evident to both sender and receiver. In an interpersonal context, irony serves as a comment on an undesirable or unexpected form of behavior. When an individual acts outside some expected societal or relational norm, an ironic comment may reel the deviant behaviors back in line with the violated norm (Averbeck & Hample, 2008; Katz & Lee, 1993) among other potential functions. The study of irony is of value because it is commonplace in our interactions (Gibbs, 2000), yet we don t have a firm grasp on why we use it (Dews, Kaplan, & Winner, 1995). Given that communication becomes both strategic and automatic (Kellermann, 1992), once a desirable effect is in evidence, repeated, and eventually habitualized, the presence of stylistic choices indicates some strategic value for influencing others. However, there remains some undiscovered reason for why we use irony when influencing others. There is something about making a statement that one does not actively believe to be true that is very useful in our interactions. A richer understanding of this phenomenon would inform both sender and receiver. A closer examination of irony is important because we are relatively uninformed as to why irony is used (Averbeck & Hample, 2008). Beyond an attempt to understand the unknown, a deeper understanding of irony is informative as it serves several social functions. Dews et al. (1995) explain irony functions as humor, status elevation, aggression, and emotional control. These varied functions permeate most if not all interpersonal interactions, and they extend to a variety of other contexts. Indeed, a comprehensive understanding of irony would be informative to interpersonal, emotion, humor, power, and message design scholars. To uncover the intention behind ironic messages, an expectancy violation approach will be utilized. Clearly, stating something false is not expected behavior (J. K. Burgoon, Jensen, Kruse, Meservy, & Nunamaker, 2007; Grice, 1975). Since an ironic message is, at least to some degree, counterattitudinal, an expectancy violation approach makes sense. After all, it is generally expected that one is direct with requests, the intent of the message is fairly transparent, and it does not require the receiver to untangle a counterattitudinal message in order to understand the sender s intent (Brown & Levinson, 1978). Ironic messages deviate from this expectation by being indirect and opaque rather than transparent. Although a great deal of the work with language expectancy theory has focused on language intensity (for review, see M. Burgoon, Denning, & Roberts, 2002), ironic messages also constitute violations of expectations on several levels. Ironic messages not only violate societal norms of veracity for senders but they also violate interpersonal expectations of behaving consistently with one s own attitudes. Therefore, the relationship between irony and counterattitudinal messages will be developed, language expectancy theory will be explained, and the expansions to language expectancy theory in relation to irony will be discussed.

358 J. M. Averbeck Irony and Counterattitudinal Intent An ironic message is counterattitudinal because it embodies a facetious display of an attitude (Clark & Gerrig, 2007). For example, most people believe that littering is a socially undesirable behavior. When we observe littering, we might react in a very direct and literal; You should throw your trash away. We could also be indirect and ironic; Putting our trash next to the trash can is demonstrating the true function of the receptacle. For the ironic message, it should be obvious to the target of the message that littering is not being condoned. Rather, it is a subtle attempt to call out the behavior as being unacceptable and to emphasize a social norm. Despite the prevalence of irony in our interactions (Gibbs, 2000), a question looms for researchers. Why not make the message direct? Dews et al. (1995) feel irony is a face-saving technique. Averbeck and Hample (2008) echo this sentiment in pointing out that irony may, in some capacity, serve as a relational-maintenance mechanism. Specifically, an ironic criticism softens the blow of a negative reaction by highlighting the normatively appropriate attitude instead of the actor and, thereby, alleviating any direct face threats to the target. By focusing the criticism on the actor s apparent attitude, one is no longer the focus of attack. This, in turn, benefits the addressee by minimizing the norm-violating performance (Kreuz & Link, 2002). When a wider array of attitudes is known, there will be more opportunities to soften criticisms by focusing on the attitude one should have instead of the undesirable or unexpected behavior. However, the attitude targeted by the sender should be shared by the receiver as part of some larger common ground the two share on a variety of topics. Irony stands in contrast to sarcasm. Sarcastic messages tend to be more hurtful and the focus of the message is the person, not the behavior (Averbeck, 2007). Ironic messages are also considered more appropriate, in part, because the criticism they offer is more diffuse than that in sarcasm. The common ground shared between the speaker and the hearer is a necessary component of irony. For one to understand the message as ironic, the expected behavioral pattern should be known (Glucksberg, 1995; Waldron, 1990). If the expectation is unknown, there is little chance of it being violated. If it is not violated, there is little chance of a message being generated. For example, consider a couple with a mortgage. Both partners are expected to pay their share. When one partner purchases a new big screen television and is now unable to pay his=her share of the mortgage, a relational norm has been violated. When there is considerable common ground two people share there is a greater opportunity for an ironic message (Averbeck & Hample, 2008). Essentially, the overlap in attitudes, or at least the knowledge of another s attitudes, allows one to anticipate behavior and, therefore, be able to gauge when behavior is undesirable or unexpected. When faced with norm-violating behavior, one may be direct or indirect in attempting to prevent future violations. There is also a common argument presented concerning the comprehension of irony and delivery of the message. Bryant and Fox Tree (2005) raised the question as to whether there is a sarcastic tone of voice. From the analysis, they concluded

Irony and Language Expectancy Theory 359 individuals may associate a tone of voice with a message that is negative but also the tone of voice is not critical for accurately interpreting the meaning of a message. Although many may consider the delivery of the irony message important, research relating to perceptions based on audio (Bryant & Fox Tree, 2005), computermediated ironic messages (Hancock, 2004), and text-based messages (Averbeck & Hample, 2008) all provide support for the detection of an ironic message as such without considering paralinguistic qualities. Language Expectancy Theory Language expectancy theory (LET) addresses expectations in language patterns. Specifically, it focuses on how message features (e.g., intensity, length, word choice, etc.) positively or negatively violate expectations concerning appropriate communication. Intentionally or unintentionally violating an expectation has ramifications for the effectiveness of a message (M. Burgoon, 1995). LET relates to the message and whether it is consonant or discrepant with a stereotyped response (Brooks, 1970). For instance, when one expects a relational partner to be supportive, an unsupportive message would violate that expectation. By incorporating a stylistic feature in a message, one is making a choice to observe or violate intentionally an expectation for the message. This choice is a strategic one, and LET helps to illuminate how message features positively or negatively violate (or conform to) macro-level expectations about what constitutes appropriate suasory communication attempts (M. Burgoon et al., 2002, p. 119). Using language that negatively violates the expectation, being aggressive where deemed inappropriate for instance, results in a negative appraisal of the message and=or the source. One consequence of this is that an influence attempt is likely to result in little to no attitude change (M. Burgoon, 1995). As appropriateness is tied to cultural expectations, individuals who belong to a specific group should not act outside of expectations others place on them. For instance, if a woman was being aggressive, she might be negatively violating expectations that women are not supposed to be dominant or aggressive. Persuasive effectiveness, on the other hand, is facilitated by positive violations of expectations. For example, whereas the desirable message strategy is one of civility and an individual is usually aggressive, violating the expectation of aggression by being polite would result in a greater likelihood of attitude change in the desired direction of the speaker (Hamilton, Hunter, & Burgoon, 1990). Credibility plays a crucial role in determining whether an individual is positively or negatively violating expectations. Demonstrating competence when not expected would be a positive violation, whereas failing to exhibit competence where expected would constitute a negative violation. Overall, highly credible sources are expected to continue to satisfy expectations. Sources with low credibility are not expected to meet expectations. In general, meeting or exceeding expectations, be they social or relational, is desirable for the source. Positive violations result in greater attitude change, ratings of persuasiveness, and source credibility (M. Burgoon, 1995). Negative violations

360 J. M. Averbeck elicit resistance. One is less likely to change one s attitude when faced with a message that negatively violates expectations. The negative violation would also result in lower ratings of persuasiveness and source credibility (Hamilton et al., 1990). LET and Irony The relationship between irony and LET can be understood on two levels. Initially, irony is an intentionally counterattitudinal message. Second, the ironic evaluation of an expectancy violation is an expectancy violation in and of itself. Each of these connections will be discussed below. Ironic messages, while intentionally counterattitudinal, are influence attempts designed to maintain an attitude and its corresponding behavioral intentions rather than change attitudes. According to LET, when an individual presents a counterattitudinal message to a receiver, the receiver falls in line with the desired result of the message. However, a sender does not necessarily want the receiver to accept the ironic message at face value (Sperber & Wilson, 1981). Although LET identifies the mechanism, avoiding unnecessary arousal, for changing attitudes, the public pronouncement of counterattitudinal messages does not account for bringing one s behavior back into line with a more global attitude. By global attitude, I mean an overall attitude toward a range of related behaviors (Ajzen, 2001). A local attitude is a more specific attitude-object relationship. While traditional persuasion has dealt with specific attitude objects (Rucker & Petty, 2004), LET examines the global level (M. Burgoon et al., 2002). Restoring behavior to be in line with a previously held attitude is not accounted for in the traditional attitude change model. Because the basis for an ironic message can be a social norm, as well as an individual s global attitude, the traditional influence attempt to shift an individual s interaction with a specific attitude-object may fail because of the presence of a more global attitude (Katz & Lee, 1993; Kaufer & Neuwirth, 1982). As an illustrator, consider an individual who would normally help others but actively avoids a situation in which he or she could provide assistance to an elderly woman, and as a result becomes the target of an ironic message exaggerating the benefit of putting the self above others. The attitude at play may be a religious one driving the individual to help others. Therefore, the counterattitudinal message produced is actually counterattitudinal to the receiver and may or may not be counterattitudinal to the sender. Given this situation, the influence attempt via an ironic message is counterattitudinal to the receiver with the intent to bring deviant behavior back in line with the more global attitude. The counterattitudinal message can take one of two forms. It can be a positive evaluation of a negative outcome or a negative evaluation of a positive outcome (Ivanko, Pexman, & Olineck, 2004). A positive evaluation of a negative outcome might be saying, What lovely weather we are having, when it is in fact raining. The current weather is not preferred, yet the comment is a positively valenced. Concurrently, a negative evaluation of a positive situation could be saying, It s about time I got my food, while at a restaurant notorious for slow service when

Irony and Language Expectancy Theory 361 the food has arrived very quickly. The expectation was for the service to be slow. Although the expectation itself was violated, the response itself was negatively valenced. Considering that an ironic message is an attempt to bring behavior back into line with expectations, it is an attempt to condemn negative expectancy violations. Therefore, an intentionally counterattitudinal message is not only an attempt to highlight the importance of a particular deviant behavior (Kreuz & Link, 2002) but also is an attempt to maintain the expected attitude (Schwoebel, Dews, Winner, & Srinivas, 2007). Second, the ironic evaluation of an expectancy violation is itself an expectancy violation. The difference between a positive evaluation of a negative outcome and a negative evaluation of a positive outcome has already been mentioned. However, LET suggests that, when faced with a counterattitudinal message, individuals will fall in line with the public attitude and produce fewer aggressive messages (M. Burgoon & Miller, 1971). Yet, individuals do mismatch evaluations to outcomes. Matching a positive evaluation to a positive outcome is an expected turn of events. Using a negative evaluation of a positive outcome requires the receiver to determine why the negative evaluation was used and whether this is a good or a bad thing. Thus, it should be the case that a negative evaluation of a positive outcome is a positive violation of expectations (Dews et al., 1995; Kreuz & Link, 2002). When someone is consistently incompetent at a task, a competent performance is a positive violation of expectations. A compliment on this achievement would be warranted. In fact, messages are considered to be more persuasive when they match the underpinning attitude they target (Shavitt & Nelson, 2002). However, a negative evaluation of this outcome is counter to the behavior and could be considered an ironic criticism. This again speaks more to a social expectation that one should be competent at one s job. Using irony in this instance is an attempt to bring the consistent expectancy violation (incompetence, rudeness, etc.) back in line with a larger expectation (competence, politeness, etc.). A positive evaluation of a negative outcome should be a negative expectancy violation. The matched response to a negative outcome should be a negative evaluation. For instance, when someone is expected to be friendly and is suddenly very rude, it is expected that the person will be called out on one s rude behavior and is considered an ironic compliment. However, when the unexpected rude behavior is met with an ironic compliment, the expectation is violated. In general, positive messages are expected to accompany positive behavior and negative messages negative behavior. A counterattitudinal message in such instances is ironic. The valencing of the expectancy violation depends on the matching of the ironic message to the behavior. The ironic criticism (criticism of positive behavior) is a positive language expectancy violation. One is expected, in general, to compliment desirable behavior. The ironic compliment (praise of negative behavior) is a negative language expectancy violation, as one would not normally praise negative behavior. Given the relationship between the violation of expectations and the persuasiveness of messages, ironic messages should be more effective facilitating attitude change than literal messages. A positive violation results in greater attitude change than a

362 J. M. Averbeck negative violation (Kelley, 1999). However, this is only taking into account variations in messages such as aggressiveness and language intensity (M. Burgoon, 1995). Irony, on the other hand, can successfully influence behavior without being too aggressive or intense. Therefore, in terms of violation valencing, ironic messages are more negative than literal messages. H1: Compared to literal messages, one will view ironic messages as being more negative, unexpected, and important, but more likely to produce behavior change and be attributed to the situation rather than the source. In general, compliments constitute positive expectancy violations whereas criticisms are negative ones. Because criticism is direct, it is likely to be attributed to the sender and be considered more important than a compliment (Floyd, 1997). Compliments are less likely to induce behavior change since they are expected. However, when one encounters a criticism, one is more likely to alter one s behavior to avoid future criticisms. H2: Compared to criticisms, one will view compliments as more positive, attributed to the sender, and judged to be less important, more expected, and less likely to produce behavior change. The effectiveness of an ironic message does not depend on its being a positive violation. As a result of the counterattitudinal component of irony, the speaker is not the one causing the receiver to change behavior (Giora, 1997; Giora, Fein, Ganzi, Levi, & Sadah, 2005; Giora, Fein, & Schwartz, 1998). The receiver has acted counterattitudinally or is exhibiting a change in attitude. In either case, the ironic message requires the receiver to identify the behavior as being in conflict with currently held attitudes or with a more global attitude to which they currently do not yet, but normatively should, subscribe (Smith, Atkin, Martell, Allen, & Hembroff, 2006). In that case, depending on the valence of the expectancy violation, there should be a difference in persuasiveness for an ironic message. Specifically, literal criticism should produce the greatest behavioral intentions, as it is the harshest. However, one would be apt to see it as the most negative. Ironic criticism should follow as it is also critical of behavior, but it will be viewed as less negative. Finally, compliments will follow with the ironic one being more effective than the literal ones. H3: There is an inverse relationship between message effectiveness and valencing, with literal criticisms being the most effective and negative followed by ironic criticisms, ironic compliments, and literal compliments. Method Participants Two-hundred and one undergraduate students enrolled in communication courses at a large Midwestern university participated in this study and received course credit for

Irony and Language Expectancy Theory 363 doing so. Each participant provided limited demographic information including age, sex, and grade; three individuals did not provide this information. The mean age was 20.7 and ranged from 18 to 27. Eighty-four males and 114 females participated in the study. The sample consisted of 28 freshmen (14.5%), 67 sophomores (33.6%), 40 juniors (20.4%), and 63 seniors (31.5%). Procedures The experimental design was a 2 (language: ironic vs. literal) 2 (message: compliment vs. criticism) factor design. The study was conducted via an online data collection site. The participants were randomly assigned to one of four conditions: ironic compliment, ironic criticism, literal criticism, and literal compliment. They initially completed attitude and involvement measures concerning alcohol use. Because the scenarios were from the perspective of a close friend, the participants were instructed to think of a close friend and complete the relational closeness scale concerning that friend. After reading the message, they responded to measures of expectancy violations and behavioral intentions, with the close friend as the source of the message. Message Construction The messages differed as little as possible except for the manipulations of interest. The scenario portrayed the participant and a close friend out drinking, a common occurrence on a college campus. One then read one of the four messages containing an ironic criticism, ironic compliment, literal criticism, or literal compliment. The lead-in to the ironic criticism described a situation in which an individual engaged in limited alcohol consumption. The person in a close relationship to the respondent, for which the relational closeness items were answered, was used as the sender of the message. The ironic criticism was a condemnation of the participant s excessive alcohol consumption following a situation in which one had exercised restraint despite the description of restraint by the subject. This was counterattitudinal in the situation. However, the receiver would be able to identify the message as ironic as it was intentionally false given the situation described. For the ironic compliment, the situation involved an individual engaged in excessive alcohol consumption. Again the person described as being in a close relationship to the participant was the source. In this case, the ironic compliment appeared to condone the restraint demonstrated for the alcohol consumption by the participant when, in fact, the participant had not exercised restraint. The literal, or nonironic, messages utilized the same situations as the ironic messages. However, the ironic message was replaced with a literal message matched to the situation. The situation that described excessive alcohol consumption was followed by a literal condemnation of excessive alcohol consumption. Restraint was accompanied by a literal compliment on responsible alcohol consumption. These messages were direct and factually correct for the situations. Full texts of the messages appear in Appendix A.

364 J. M. Averbeck Dependent Variables Expectancy violations and behavioral intentions were the dependent variables. Expectancy violations were assessed via four scales from work by Afifi and Metts (1998) and Wong (2005). These scales were: violation expectedness, importance, valence, and attribution. Responses to these measures indicated how much the message violated expectations, how significant a violation it was, whether it was a negative or positive violation, and to what or whom the violation could be credited. The likelihood of adopting the message s recommendation was assessed by the behavioral intention items. All measures were completed on 7-point scales anchored by strongly disagree and strongly agree where higher scores indicate agreement. Violation expectedness referred to the extent to which the content of the message was not what one would predict. The scale included three items: This message was completely unexpected, This message was not expected at all, and This message surprised me a great deal. Expectedness had a ¼.73 (M ¼ 4.06, SD ¼ 1.38). Violation importance indicated how significant one perceived the violation was to the relationship. The scale included two items: This message was a very important relationship event, and This message was a minor relational event. This scale had a ¼.80 (M ¼ 3.79, SD ¼ 1.44). To capture whether the message was considered positive or negative, violation valence was the index. The measure consisted of five items such as: This message made me feel a lot better=a lot worse about the state of our relationship. Reliability was a ¼.85 (M ¼ 4.32, SD ¼ 1.33). Finally, attribution for the violation reflected one s perception of whether the message was attributable to the situation or to the sender, and it consisted of three items, including: This message was completely=not at all due to the situation, and This message was not at all typical of how he=she acts in our relationship. Higher scores indicated the message was attributed to the person. While the sender and situation are identified, attribution was included to assess whether the message was attributed to the situation or the source. Attribution had a ¼.81 (M ¼ 4.19, SD ¼ 1.25). The measure of behavioral intentions was assessed using two items adapted from Sheppard, Hartwick, and Warshaw (1988). The items were I intend to reduce my alcohol consumption, and I plan to decrease my alcohol use. Behavioral intention had a ¼.93 (M ¼ 4.25, SD ¼ 1.66). Covariates Attitude was assessed in order to ensure that the message received would be counterattitudinal in the irony conditions. For the conditions involving a literal message, participants were randomly distributed. General attitude was assessed using 7-item semantic differential scales developed by M. Burgoon, Cohen, Miller, and Montgomery (1978). Participants rated whether Alcohol use is: acceptable= unacceptable, wise=foolish, favorable=unfavorable, positive=negative, good=bad, and right=wrong. Attitude had a ¼.94 (M ¼ 4.01, SD ¼ 1.31).

Irony and Language Expectancy Theory 365 Involvement was assessed to ensure the interest in the topic was relatively balanced across conditions. Involvement was measured using six semantic differential scales from Zaichkowsky (1985). Scale items included: important=unimportant, means a lot to me=means nothing to me, significant=insignificant, appealing= unappealing, essential=nonessential, and wanted=unwanted. Reliability was a ¼.89 (M ¼ 2.40, SD ¼ 1.19). Relational closeness was assessed through the Miller social intimacy scale (Corcoran & Fisher, 1987; Miller & Lefcourt, 1982). The participants were asked to identify the kind of relationship held with one s closest friend. Then, participants answered 19 items about that relationship on a 7-point scale ranging from very rarely to almost always. Relational closeness had a ¼.83 (M ¼ 5.32, SD ¼.69). Results Induction Check In order to confirm the attitude and involvement levels were equally distributed across conditions and that relational closeness was held constant, one-way ANOVAs were computed. For attitude, there was no main effect for language F(1, 178) ¼.33, ns, g 2 ¼.00; no main effect for message F(1, 178) ¼.22, ns, g 2 ¼.00; and no interaction effect F(1, 178) ¼.01, ns, g 2 ¼.00. For involvement, there was no main effect for language F(1, 169) ¼.69, ns, g 2 ¼.00; no main effect for message F(1, 169) ¼ 1.52, ns, g 2 ¼.01; and no interaction effect F(1, 169) ¼ 2.78, ns, g 2 ¼.02. And for relational closeness, there was no main effect for language F(1, 173) ¼.37, ns, g 2 ¼.00; no main effect for message F(1, 173) ¼.87, ns, g 2 ¼.01; and no interaction effect F(1, 173) ¼ 2.35, ns, g 2 ¼.01. These data suggested balance across conditions in terms of attitude, involvement, and relational closeness of the source of the message. Hypothesis Testing In order to test the hypotheses, a MANOVA was computed with language and message as independent variables with expectedness, behavioral intentions, importance, valence, and attribution as dependent variables. Following presentation of the omnibus results, we reported those for univariate analyses as they apply to each hypothesis below. The omnibus test revealed a main effect for language F(5, 161) ¼ 30.49, p <.001, g 2 ¼.49. Univariate tests revealed significant effects for expectedness F(1, 165) ¼ 16.13, p <.001, g 2 ¼.09, valence F(1, 165) ¼ 116.61, p <.001, g 2 ¼.41, and attribution F(1, 165) ¼ 5.55, p <.05, g 2 ¼.03. There was also a main effect for the message F(5, 161) ¼ 4.33, p <.001, g 2 ¼.12. The univariate tests revealed significant effects for behavioral intentions F(1, 165) ¼ 15.65, p <.001, g 2 ¼.09. The interaction between language and message was also significant F(5, 161) ¼ 3.85, p <.01, g 2 ¼.11. The univariate tests revealed significant effects for behavioral intentions F(1, 165) ¼ 10.68, p <.001, g 2 ¼.06. See Table 1 for all the means and standard deviations.

366 J. M. Averbeck Table 1 Means and (Standard Deviations) Expectedness Attribution Behavioral Intentions Importance Valence Compliment Literal 3.68 4.18 3.33 3.57 5.35 (1.23) (1.18) (1.55) (1.50) (0.75) Ironic 4.51 3.98 4.21 3.56 3.45 (1.21) (1.26) (1.67) (1.42) (1.06) Criticism Literal 3.72 4.57 5.00 4.08 4.96 (1.58) (1.35) (1.28) (1.40) (1.06) Ironic 4.47 3.93 4.36 3.79 3.47 (1.25) (1.19) (1.74) (1.50) (1.14) Hypothesis one predicted that participants would perceive ironic messages as more negative than literal messages. Ironic messages, compared to literal ones, would come across as more unexpected, important, and negative, as well as more likely to influence behavior intentions to reduce alcohol consumption, and would be attributed to the situation rather than the source. As noted above, there were significant univariate effects for expectedness F(1, 165) ¼ 16.13, p <.001, g 2 ¼.09, valence F(1, 165) ¼ 116.61, p <.001, g 2 ¼.41, and attribution F(1, 165) ¼ 5.55, p <.05, g 2 ¼.03. Specifically, the participants rated ironic messages as more unexpected and negative compared to literal ones. They also attributed them more to the situation rather than the source of the message. However, there were no significant effects for behavior intentions or importance. Hypothesis two predicted that compliments, compared to criticisms, would more likely be attributed to the source, viewed as less important, seen as more expected and be less likely to contribute to behavioral intentions. However, while the predicted trend was observed, the only significant difference observed was for behavioral intentions. Univariate tests revealed significant effects for behavioral intentions F(1, 165) ¼ 15.65, p <.001, g 2 ¼.09, such that the criticism showed greater impact on behavioral intentions than did compliments. Finally, hypothesis three predicted an inverse relationship between effectiveness and valencing, with literal criticisms being the most effective and negative followed by ironic criticisms, ironic compliments, and literal compliments. Univariate tests revealed significant effects for behavioral intentions F(1, 165) ¼ 10.68, p <.001, g 2 ¼.06. Again the predicted trends were generally observed, but the only significant difference was for behavior intentions. Literal criticisms had the most apparent behavior intentions followed by the ironic criticism. Both were more effective than the ironic compliment. The literal compliment was the least effective. Discussion The present study examined the relationship between expectancy violations and a linguistic practice. The participants received and react to messages that either praised

Irony and Language Expectancy Theory 367 or criticized a behavior. Ironic messages were compared to the literal ones to understand how these messages violate expectations. Overall, the results suggest ironic messages are more negative, unexpected, and, in some instances, more effective than literal ones, in that they foster greater intentions to alter one s behavior. The notion of common ground between sender and receiver has been considered to be a necessary component to understand irony (Glucksberg, 1995). Therefore, the messages were examined in a situation in which common ground should have been established. The participants rated ironic messages as being more negative and unexpected in general. Typically, this would be a negative violation of expectations and produce derogation (M. Burgoon, 1995). However, receivers were more likely to report that the ironic message was attributable to the situation, whereas the literal message was attributable to the source. The ironic message negatively violates expectations while not carrying the baggage of a negative violation of expectations. This is a unique phenomenon that seems to fly in the face of LET. However, ironic messages are in general negative and designed to obscure their negativity (Giora, 1997). The fact that they were not considered more important than literal messages also indicates that the presence of an ironic message, though unexpected, does not have lasting relational effects. The use of irony during conflict could be particularly effective to harshly criticize while avoiding potential pitfalls associated with negative expectancy violations. The ironic message either meets or exceeds linguistic expectations. The differences between compliments and criticism were minimal. They differed only in terms of behavioral intentions. The differences between compliments and criticisms may be akin to the extensive research language intensity variable (M. Burgoon et al., 2002). Perhaps, when a close friend is being irresponsible, the appropriate behavior is to criticize. In this case, one is producing a message that is not intense and, therefore, effective. However, when a close friend is behaving appropriately, a compliment is rarely necessary. When a compliment is produced, it may appear to be intense. For instance, declaring that a friend has been very responsible when responsibility is expected does provide intensifiers. Therefore, it is not surprising that the criticism is more effective in influencing intentions to change one s behavior. The differences between ironic compliments and ironic criticisms were also restricted to behavioral intentions. Whereas the ironic messages were considered violations of expectations, they were effective when the goal of message was taken into account. In both ironic compliments and criticisms there is an attempt to realign current behavior with expected behavior. Although ironic messages did violate expectations, they were not outside the bandwidth of acceptable speech for the speaker. This echoes Dews and colleagues (1995), who suggest irony is a face-saving technique. This finding was later confirmed by Averbeck and Hample (2008). A direct message that is intended to realign behavior and expectations could be a threat to positive or negative face. If one remembers that ironic messages are, in general, more apt to be attributed to the situation, then one may more freely use one without concerns of derogation. An ironic message focuses on the behavior, not the person and, thus, reduces face threats (Grice, 1975). In that respect, ironic messages provide a safe middle ground to compliment or criticize while avoiding potential pitfalls of

368 J. M. Averbeck direct, literal messages. Irony can be a vehicle for maximally effective, minimally destructive messages that seek to maintain current behaviors in others. By highlighting the expected behavior and a corresponding violation, the message functions as a behavioral maintenance tool. One of the more interesting findings of the present study was the simultaneous ratings of ironic messages as being destructive but relatively unimportant to the relationship. The ironic message is considered much more destructive than the literal message. A particular nasty ironic message, however, has little to no consequences on the relationship. The presence of a message that is so out of the ordinary and negative at face value seems to grab hold of the receiver s attention without necessarily causing the receiver to reject the message. This provides further evidence for the face-saving argument, in that messages with negative intent can be presented in an ironic way that simultaneously is recognized as being negative but effective in achieving its goal. Although the expectation was that the counterattitudinal (ironic) message would be more persuasive, the face-saving component is perhaps more informative. Both the literal and ironic criticisms were persuasive. The ironic compliment showed greater impact on behavioral intentions than the literal compliment. However, the ironic message exhibited fewer qualities likely to result in message rejection. This is keeping in line with the prediction of LET. Counterattitudinal messages did arouse fewer negative thoughts and is likely to minimize counterarguing (Rucker & Petty, 2004). When attempting to influence someone with shared beliefs, attitudes, or experiences, one must be cognizant that a message that is too aggressive or destructive can have relational consequences. However, an ironic message enables one to be aggressive without suffering relational backlash. Additionally, LET predicts that messages are more persuasive when they utilize positive expectancy violations (M. Burgoon & Miller, 1985). The findings here suggest that ironic messages, in general, appear to be negative expectancy violations because they are so unexpected and attributed to the sender. Previous studies have revealed that ironic messages focused on the behavior and not the actor (Averbeck & Hample, 2008; Dews et al., 1995). However, in ironic message reception, the message is attributed to the person and to the situation. When one could choose to be direct and confrontational, the choice to be less direct, even if it is more aggressive, is a negative violation of expectations. However, the negative violation seems to highlight the desired behavior. It appears to be a gentle nudge in the direction of a desired outcome rather than an all out shove. The present study also offers a few practical implications. Potentially, ad campaigns could use irony to disarm audience resistance to a persuasive message. The results here indicated the ironic messages were effective, and this message strategy could be employed to tackle traditionally difficult subject matters. Additionally, some of the findings here reinforce the notion of irony as a relational maintenance mechanism (Averbeck & Hample, 2008). When one wishes to encourage or discourage a behavior in relationally close partners, irony can display behavioral preferences without placing demands on others. This could be applied to interpersonal conflict resolution in cases where an undesirable behavior is the cause of the conflict. Finally,

ironic messages could be tested in a variety of theoretical models. For instance, would an ironic message produce less reactance than a literal one? Could one process an ironic message peripherally rather than centrally? Future research could examine these possibilities. Summary and Limitations Within the framework of Language Expectancy Theory, in this study, ironic messages were persuasive, but not more so than the literal criticism. The participants saw ironic messages as unexpected and destructive, attributed to the situation but not considered an important relational event. The direction of these findings indicates that ironic messages are useful for interpersonal influence. Speakers who use irony are allotted a greater bandwidth of acceptable behavior. These messages are effective, regardless of the type of criticisms they embody. There were limitations to the current study. First, the claims presented here have limited generalizability, as only one situation was used. Replication could provide stronger evidence for the claims. Additionally, it is still difficult to ascertain whether the messages were truly ironic. Absent a clever manipulation check, the claims concerning irony depend on the researcher s argument. A greater variety of ironic messages, as used in Averbeck and Hample (2008), would alleviate uncertainty concerning the effects observed here. Finally, there was only one medium text. Additional support concerning the comprehension of ironic messages for different channels could be helpful and possibly lead to a richer understanding of the processing of ironic messages. Future research should seek to compliment the language expectancy approach taken here. For instance, language intensity could be added to the ironic messages to uncover the limits irony has on masking aggressive messages. An ironic message on a relatively unimportant topic may be rated differently than an ironic message on a more important topic. In either case, the surface has been scratched on the utility of ironic messages for interpersonal influence. Further study could keep digging to uncover this linguistic phenomenon. References Irony and Language Expectancy Theory 369 Afifi, W. A., & Metts, S. (1998). Characteristics and consequences of expectancy violations in close relationships. Journal of Social and Personal Relationships, 15, 365 392. Ajzen, I. (2001). Nature and operation of attitudes. Annual Review of Psychology, 52, 27 58. Anolli, L., Ciceri, R., & Infantino, M. G. (2002). From blame by praise to praise by blame : Analysis of vocal patterns in ironic communication. International Journal of Psychology, 37, 266 276. Averbeck, J. M. (2007, November). Gauging the acceptability of an argument: An examination of effectiveness and appropriateness through irony, humor, and verbal aggressiveness. Interpersonal Division. Paper presented to the annual meeting of the National Communication Association, Chicago, IL. Averbeck, J. M., & Hample, D. (2008). Ironic message production: How and why we produce ironic messages. Communication Monographs, 75, 396 410.

370 J. M. Averbeck Beatty, M. J., Valencic, K. M., Rudd, J. E., & Dobos, J. A. (1999). A dark side of communication avoidance: Indirect interpersonal aggressiveness. Communication Research Reports, 16, 103 109. Brooks, R. D. (1970). The generalizability of early reversals of attitudes toward communication sources. Speech Monographs, 37, 152 155. Brown, P., & Levinson, S. C. (1978). Politeness: Some universals in language usage. Cambridge, England: Cambridge University Press. Bryant, G. A., & Fox Tree, J. E. (2005). Is there an ironic tone of voice? Language and Speech, 48, 257 277. Burgoon, J. K., Jensen, M. L., Kruse, J., Meservy, T. O., & Nunamaker, J. (2007). Deception and intention detection. In H. Chen, T. S. Raghu, R. Ramesh, A. Vinze, & D. Zeng (Eds.), Handbooks in information systems (Vol. 2, pp. 193 214). Amsterdam: Elsevier B. V. Burgoon, M. (1995). Language expectancy theory: Elaboration, explication, and extension. In C. R. Berger & M. Burgoon (Eds.), Communication and social influence processes (pp. 29 52). East Lansing, MI: Michigan State University Press. Burgoon, M., Cohen, M., Miller, M. D., & Montgomery, C. L. (1978). An empirical test of a model of resistance to persuasion. Human Communication Research, 5, 27 39. Burgoon, M., Denning, P. V., & Roberts, L. (2002). Language expectancy theory. In J. P. Dillard & M. Pfau (Eds.), The persuasion handbook: Developments in theory and practice (pp. 117 137). Thousand Oaks, CA: Sage. Burgoon, M., & Miller, G. R. (1971). Prior attitude and language intensity as predictors of message style and attitude change following counterattitudinal advocacy. Journal of Personality and Social Psychology, 20, 246 253. Burgoon, M., & Miller, G. R. (1985). An expectancy interpretation of language and persuasion. In H. Giles & R. N. St. Clair (Eds.), Recent advances in language communication and social psychology (pp. 199 229). London: Erlbaum. Clark, H. H., & Gerrig, R. J. (2007). On the pretense theory of irony. In R. W. Gibbs & H. L. Colston (Eds.), Irony and language and thought: A cognitive science reader (pp. 25 34). New York: Erlbaum. Corcoran, K., & Fischer, J. (1987). Measures for clinical practice: A sourcebook. New York: The Free Press. Dews, S., Kaplan, J., & Winner, E. (1995). Why not say it directly?: The social functions of irony. Discourse Processes, 19, 347 367. Dews, S., & Winner, E. (1995). Muting the meaning: A social function of irony. Metaphor and Symbolic Activity, 10, 3 19. Dillard, J. P., Kinney, T. A., & Cruz, M. G. (1996). Influence, appraisals, and emotions in close relationships. Communication Monographs, 63, 105 130. Floyd, K. (1997). Communication affecting in dyadic relationships: An assessment of behavior and expectancies. Communication Quarterly, 45, 68 80. Gibbs, R. W., Jr. (2000). Irony in talk among friends. Metaphor and Symbol, 15, 5 27. Giora, R. (1997). Understanding figurative and literal language: The graded salience hypothesis. Cognitive Linguistics, 8, 183 206. Giora, R., Fein, O., Ganzi, J., Levi, N. A., & Sadah, H. (2005). On negation as mitigation: The case of negative irony. Discourse Processes, 39, 81 100. Giora, R., Fein, O., & Schwartz, T. (1998). Irony: Graded salience and indirect negation. Metaphor and Symbol, 13, 83 101. Glucksberg, S. (1995). Commentary on nonliteral language: Processing and use. Metaphor and Symbolic Activity, 10, 47 57. Grice, P. H. (1975). Logic and conversation. In P. Cole & J. Morgan (Eds.), Speech acts, Vol. III of syntax and semantics (pp. 41 58). New York: Academic Press.

Irony and Language Expectancy Theory 371 Hamilton, M. A., Hunter, J. E., & Burgoon, M. (1990). An empirical test of an axiomatic model of the relationship between language intensity and persuasion. Journal of Language and Social Psychology, 9, 235 255. Hancock, J. T. (2004). Verbal irony use in face-to-face and computer-mediated communication. Journal of Language and Social Psychology, 23, 447 463. Ivanko, S. L., Pexman, P. M., & Olineck, K. M. (2004). How sarcastic are you?: Individual differences and verbal irony. Journal of Language and Social Psychology, 23, 244 271. Katz, A. N., & Lee, C. J. (1993). The role of authorial intent in determining verbal irony and metaphor. Metaphor and Symbolic Activity, 8, 257 279. Kaufer, D. S., & Neuwirth, C. M. (1982). Foregrounding norms and ironic communication. Quarterly Journal of Speech, 68, 28 36. Kellermann, K. (1992). Communication: Inherently strategic and primarily automatic. Communication Monographs, 59, 288 300. Kelley, D. L. (1999). Relational expectancy fulfillment as an explanatory variable for distinguishing couple types. Human Communication Research, 25, 420 442. Kreuz, R. J., & Link, K. E. (2002). Asymmetries in the use of verbal irony. Journal of Language and Social Psychology, 21, 127 143. Miller, R. S., & Lefcourt, H. M. (1982). The assessment of social intimacy. Journal of Personality Assessment, 46, 514 518. Rucker, D. D., & Petty, R. E. (2004). When resistance is futile: Consequences of failed counterarguing for attitude certainty. Journal of Personality and Social Psychology, 86, 219 235. Schwoebel, J., Dews, S., Winner, E., & Srinivas, K. (2007). Obligatory processing of the literal meaning of ironic utterances: Further evidence. In R. W. Gibbs & H. L. Colston (Eds.), Irony and language and thought: A cognitive science reader (pp. 253 267). New York: Erlbaum. Shavitt, S., & Nelson, M. R. (2002). The role of attitude functions in persuasion and social judgment. In J. P. Dillard & M. Pfau (Eds.), Persuasion handbook: Developments in theory and practice (pp. 137 154). Thousand Oaks, CA: Sage. Sheppard, B. H., Hartwick, J., & Warshaw, P. R. (1988). The theory of reasoned action: A meta-analysis of past research with recommendations for modification and future research. Journal of Consumer Research, 15, 325 343. Smith, S. W., Atkin, C. K., Martell, D., Allen, R., & Hembroff, L. (2006). A social judgment theory approach to conducting formative research in a social norms campaign. Communication Theory, 16, 141 152. Sperber, D., & Wilson, D. (1981). Irony and the use-mention distinction. In P. Cole (Ed.), Radical pragmatics (pp. 295 317). New York: Academic Press. Waldron, V. R. (1990). Constrained rationality: Situational influences on information acquisition plans and tactics. Communication Monographs, 57, 184 201. Wong, N. (2005, May). Coping with the unexpected: Perceptions of responses to a relational expectancy violation as a function of intimacy and commitment. Paper presented at the annual convention of the International Communication Association, New York, NY. Zaichkowsky, J. L. (1985). Measuring the involvement construct. Journal of Consumer Research, 12, 341 352. Appendix A: Ironic Compliment, Criticism, and Literal Messages Compliment Recently, you and your closest friend went out for a night on the town. You decide to let loose and have a few alcoholic beverages. However, you decide to limit how much