JOURNAL OF DIPLOMATIC
LANGUAGE
JDL I:4 (2004)
Differences in Participants' Estimates and Identification of Their Own and Their Partners' Sarcastic Utterances
Dr. Patricia Rockwell, Ph.D.
Department of Communication,
University of Louisiana at Lafayette
INTRODUCTION
Sarcasm is a complex communicative behavior in which listeners are expected to recognize statements that represent the opposite of speakers' true intents. This is not easy to do. Indeed, researchers have consistently found that people have difficulty recognizing sarcasm (Ackerman, 1982; 1983; Creusere, 1999;Giora, Zaidel, Soroker, Batori, & Kasher, 2000; McDonald & Pearce, 1996; Rockwell, 2000). Although extensive research has been conducted on determining how individuals recognize written irony (Clark & Gerrig, 1984; Jorgensen, Miller & Sperber, 1984; Sperber, 1984), little research has been done on how they recognize sarcasm (a type of irony) in its oral form. Also, most studies of irony and sarcasm have been conducted on American, English-speaking subjects. Researchers (e. g., Haiman, 1998) have noted that individuals from other cultures do not express sarcasm in the same manner or with the same frequency as English speakers. Indeed, Haiman reports that the Hua, a group of New Guinea Highlanders, do not use sarcasm at all. These cultural differences in sarcasm production and recognition have the potential to seriously influence negotiations that might occur between individuals from different countries. Although sarcasm represents only a small percentage of most conversations, it is of interest to diplomats because it represents an aspect of communication that can be easily misunderstood and thus may lead to relational conflict. Why is sarcasm so poorly recognized? There are a number of reasons. First, researchers have found that conversational participants are less aware of their partners' behaviors than they are of their own behaviors (Aune, 1997; Rockwell, 2000), although Neff and Karney (2002) found that people in satisfying relationships perceived each other's traits more accurately than people in less satisfying relationships. People expect that their own behavior will be noticed by others. Gilovich, Kruger, and Medvec (2002) call this the "spotlight" effect because people are more focused on their own behavior than on that of others. Thus, one would expect that people will be more aware of sarcasm they themselves produce than they are of sarcasm produced by others. People are not only more aware of their own behavior, research suggests that people see themselves in a more positive light than they see others ("self-bias") (Brown, 1991). Thus, people may see their own use of sarcasm as a sign of their own sophistication or interpersonal control; whereas, sarcasm produced by others may be perceived as thoughtless or insulting. Along with these self-oriented reasons, there are other reasons why people may not recognize sarcasm well. Researchers suggest that people typically overestimate the frequency of uncommon events, such as sarcasm. These low-frequency behaviors tend to be enumerated; whereas, high-frequency events tend to be estimated using rate information (Blair & Burton, 1987; Burton & Blair, 1991). Means and Loftus (1991) argue that low-frequency behaviors are recalled from episodic or time-based memory. Applied to sarcasm, this would suggest that because sarcasm occurs infrequently, people may remember making a specific sarcastic comment and extrapolate from that memory the belief that they produced more sarcastic remarks than they really did. Conrad, Brown, and Cashman (1993) argue that recall for event frequency may come more from a "sense of frequency" rather than from exact counts. Thus, if people consider themselves to be sarcastic they may be more likely to report that they produce sarcastic comments. Indeed, Brenner, Koehler, Liberman, and Tversky (1996) found that people tend to be overconfident in their estimates of difficult behaviors. So, with the production of sarcasm (arguably a difficult behavior), people may be more likely to overestimate the amount of sarcasm they produce. Due to these findings that people tend to overestimate certain behaviors, one might conclude that when pressed to justify their estimates, people will not be able to identify all of the behaviors that they estimated. That is, if a person estimates that he/she made ten sarcastic comments during a conversation and is then asked to identify those ten comments, it is likely that, due to the overestimation tendency, the person will not be able to identify ten comments, but will identify a smaller number. Sarcasm research has been hampered by methodological limitations. Early studies utilized hypothetical written scenarios (Clark & Gerrig, 1984; Jorgensen, Miller, & Sperber, 1984; Kaufer, 1981; Kreuz & Glucksberg, 1989; Sperber, 1984) that did not consider the reciprocal nature of sarcasm found in conversation. More recently, experimental research has investigated sarcasm in more interactive contexts (Rockwell, 2003; Rockwell & Theriot, 2001). These studies utilized procedures that required participants to rate their partners' answers to questions regarding the amount of sarcasm they perceived in each of their partners' answers. However, this method did not ask participants to identify specific comments as sarcastic, and responses to questions often included numerous utterances, some of which may have been sarcastic and some of which may have been sincere. Some field research in which natural conversations have been recorded on audiotape and later analyzed (Gibbs, 2000; Roy, 1981; Tannen, 1984) is promising. In these studies, researchers examined actual dialogue and investigated individual comments for specific features. Unfortunately, field studies do not allow for probing of participants' intents and perceptions. In field studies, judgments regarding whether an utterance represents a particular phenomenon (such as sarcasm) are typically made by researchers not by participants. In order to investigate communicators' recognition of sarcasm, the present study utilized a videotape review procedure that focused on specific sarcastic utterances produced and recognized by participants. Given the preceding concerns, the following hypotheses and research question are posed: H1: Participants will estimate more self-produced sarcastic utterances than they will identify. H2: Participants will estimate more self-produced sarcastic utterances than partner-produced sarcastic utterances. H3: Participants will estimate more partner-produced sarcastic utterances than they will identify partner-produced sarcastic utterances. H4: Participants will identify more self-produced sarcastic utterances than partner-produced sarcastic utterances. RQ1: How accurate are participants in identifying their partners' sarcastic utterances? Method Participants Participants (n =136) were tested in pairs. Participants were drawn from introductory communication classes at a large southeastern university. All participants received extra credit for their involvement. They were asked to bring a close friend or relative with them to the experiment, because previous research (Rockwell, 2003) has indicated that friends and close relations are more likely to produce sarcasm than are strangers. Participants reported that they had known their partners for an average of 5.9 years (SD = 6.21) with the range of years known extending from one to 26 years. The 136 participants included 38 men and 98 women. The average age of the participants was 22.8 years (SD = 7.36) with participants ranging in age from 18-60 years. The average educational level was 14.4 years (college sophomore) (SD = 1.38) with number of years of schooling ranging from 9-17 years. Of the 136 participants, 12 were relatives, 106 were friends, and 18 were romantic partners. Procedure Six communication majors (five seniors and one graduate student) served as research assistants. All assistants received independent study credit for their work. Assistants completed a weeklong training period before beginning work, which consisted of extensive discussion of sarcasm, and rehearsal of the post-videotape interviews Participants were asked to come to the television studio on the university campus at appointed times. When they arrived, the participants were seated on the studio set in two chairs arranged in a face-to-face position. Before the interview began, participants were asked to complete a consent form. Participants were told that they would take part in a conversation that would be videotaped for research purposes. When the participants were ready, the research assistants gave each participant a list of questions to ask each other during the conversation. The questions were designed to prompt sarcasm. The first few questions were designed to put participants at ease and to get them talking about each other. The remaining questions asked participants to make evaluations of their partners in hopes that they would make comments using sarcasm. Participants were told to continue conversing until asked to stop, even if they had completed all of the questions. Research assistants stopped all conversations at ten minutes. The research assistants then started the recording and told the participants to begin. Following the taping of the conversation, the participants were separated and each taken to separate small rooms by the research assistants. Then participants were asked to complete a form that gave the operational definition of sarcasm used in the experiment ("a negative comment or insult said with positive words; an example would be when someone says 'That's just great!' when someone spills food all over them."), and asked to estimate the number of their own and their partners' sarcastic comments using the operational definition. After participants completed the questionnaire, the research assistants played the videotape of the conversation on a videocassette recorder for each participant and asked them to identify each sarcastic comment they and their partners made. Each time participants identified a sarcastic comment the research assistant recorded the text of the comment along with the starting and ending time, which was determined from a timer graphic superimposed on the videotape. The use of the timer graphic allowed research assistants to accurately record the exact location of each sarcastic comment on the videotape rather than be forced to use the built-in timers on the videocassette recorders. This procedure allowed the research assistants to compare the sarcastic comments that were identified by both participants more easily and accurately. Finally, participants were debriefed regarding the true nature of the study and were dismissed. The entire procedure lasted approximately one hour for each dyad. After completion of each experimental session, the assistants scored the questionnaires and tabulated the sarcastic comments that participants had estimated and identified. Results Sarcasm Production and Recognition. To answer Hypotheses 1-4 regarding communicators' ability to estimate and identify their own and their partners' sarcastic remarks, several calculations were made. The 68 dyads (n =136 participants) produced 680 minutes (4,080 seconds) of taped conversation. From an examination of all 136 questionnaires, it was found that participants estimated that they produced a total of 93 sarcastic comments. When the research assistants reviewed the actual videotapes with the participants, the participants identified a total of 66 sarcastic comments that they produced. A paired t-test was conducted to compare participants' estimates of their own sarcasm production with their identification of their own sarcastic comments. This test found a significant difference between the two scores, providing an answer for Hypothesis 1 (Table 1). Next, it was determined that participants estimated that their partners produced a total of 91 sarcastic comments, but during the videotape review they identified a total of 50 comments of their partners that they believed were sarcastic. A paired t-test comparing these scores also indicated a significant difference between the estimates and the identified sarcastic comments of partners, providing an answer for Hypothesis 2 (Table 1). Table 1 Differences in Means between Participants' Estimates and Identification of Sarcasm Participant Estimates SD Identification SD t p effect size Own 2.98 3.81 1.43 2.68 5.07 .00 .23 Partner 2.54 3.25 .94 1.85 6.63 .00 .29 ________________________________________________________________________Next, a paired t-test was conducted comparing the self-estimates of sarcasm production to the participants' estimates of their partner's sarcasm production. This test produced significant results providing an answer for Hypothesis 3 (Table 2). Then, a paired t-test was conducted comparing the 66 self-identifications of sarcasm to the participants' 50 identifications of their partner's sarcasm. This test also produced significant results, providing an answer for Hypothesis 4 (Table 2). _______________________________________________________________________ Table 2 Differences in Means between Participants' Own and Partners' Sarcasm Own SD Partner SD t p effect size ________________________________________________________________________ Estimates 2.98 3.80 2.54 3.24 2.05 .04 .06 Identifications 1.43 2.68 0.94 1.85 2.60 .009 .11 Of the 66 self-identified sarcastic comments and 50 identified partner sarcastic comments, only 29 (m per dyad = .58, SD = 1.77) were in agreement. Only 14 of the 68 dyads were responsible for these 29 agreed upon sarcastic utterances. The 29 sarcastic comments (m length per comment = 2.3 seconds) that partners agreed upon as being sarcastic represented .0016 of the total amount of recorded conversation. These findings provide answers for Research Question 1. Discussion The results of this study provide new insights into the puzzle of sarcasm. The findings are derived from the questions asked of participants following a videotaped conversation. Participants were asked to estimate their own production of sarcasm and their partners' production of sarcasm. Then, while viewing a videotape of their conversation, they were asked to identify the sarcastic statements that they made and those made by their partners. This procedure allowed for the investigation of sarcasm production and recognition in an interactive context. Comparing estimates to actual identified statements revealed that participants overestimated both their own production of sarcasm and their partners' production of sarcasm. This finding suggests that people perceive themselves and others as more sarcastic than they actually are. It also supports previous research that has found that people tend to overestimate low-frequency, difficult behaviors, which certainly describes that of sarcasm. Why did participants overestimate sarcasm? There are several possible reasons. Possibly, the effect of producing a sarcastic comment colors speakers' perceptions of themselves as "sarcastic" and forces them to view their total conversation as more sarcastic than it actually is. Possibly, it is a matter of definition; speakers may perceive their own or their partners' sarcastic intents but cannot recognize it when pressed to identify individual sarcastic utterances following specific definitional guidelines. Possibly, the process of the videotape review may have taken the participants out of the role of actor/partner and placed them in the role of outside observer. This perspective may have influenced participants to respond more like observers who (Storms, 1973) typically do not recognize as many behaviors as do participants. In addition to the general overestimation of sarcasm found in this study, participants both estimated and identified more self than partner sarcasm. Why did this occur? Research on self-bias (Brown, 1991) suggests that people view themselves more positively than they view others. Participants in this study who produced sarcasm may have perceived themselves as intellectually superior to their partner and may have seen their use of sarcasm as evidence of this superiority. Also, research (Aune, 1997; Gilovich et al., 2003; Rockwell, 2000) confirms that people are more aware of their own behavior than that of their partners. This phenomenon may carry over into how people estimate their own sarcasm production as compared to their partners. Most telling in this study, however, was the poor accuracy of partners in recognizing each other's specific sarcastic comments. Participants were able to identify less than half of the sarcastic comments produced by their partners. Why did this happen? In a previous sarcasm study (Rockwell & Theriot, 2003), participants merely rated their partners' responses for amount of sarcasm; they were not asked to identify specific sarcastic utterances. In the present study, when participants were required to identify their partners' specific sarcastic utterances, they were less accurate and thus, far less sarcasm was "produced" in the conversation. This finding that participants recognized only about half of the sarcasm produced in the conversation is also particularly revealing when it is recalled that participants in this study were friends and that researchers have found that friendship typically improves production and recognition of sarcasm by participants (Rockwell & Theriot, 2001). Perhaps partners sensed their partners' sarcastic intent, but faltered when asked to identify the exact statement. The finding also suggests that participants do not "recognize" sarcastic utterances as well as once thought. It is evident from this study that communicators tend to produce more sarcasm than their partners typically perceive. If communicators intend for their sarcastic comments to be recognized, they are often unsuccessful in accomplishing this goal. The assumption, of course, has been that speakers intend for partners to recognize their sarcastic comments. This may not always be the case and communicators may not always intend for their sarcasm to be recognized. Perhaps speakers use sarcasm to express their feelings and are less concerned about whether or not these feelings are recognized. That is, perhaps sarcasm is utilized as a propositional act (as defined by Searle, 1969) in which communicators express emotion without attempting to communicate this emotion to another person. The use of sarcasm may allow speakers emotional relief without offending their partners. Another possibility may be that when receivers do recognize sarcasm, it is because the speaker's true intent "leaks" out, a phenomenon that frequently happens during deception, (Ekman & Friesen, 1969; Zuckerman, DePaulo, & Rosenthal, 1981). Indeed, sarcasm shares characteristics with deception in that both behaviors involve presenting false information. In deception, speakers wish receivers to believe the false information. In sarcasm, speakers expect receivers to not believe the false information. Researchers have shown that people are poor detectors of deceit (Burgoon, Buller, Ebesu, & Rockwell, 1995). It appears that people are also poor detectors of sarcasm. This study represents a new approach to determining how communicators use sarcasm. In this project, participants were asked to identify specific sarcastic remarks as they reviewed a videotape of their conversation. This procedure required participants to identify their own and their partners' actual utterances rather than to provide general ratings of the sarcastic content of their responses. This more exacting procedure allows researchers to determine participants' perceptual accuracy more precisely. There are limitations to this study. Certainly, the fact that the experiment took place in a television studio may have intimidated many participants. Possibly some participants who might have produced or recognized sarcasm in a less threatening venue did not do so out of fear. Even so, the quality of the video and audio produced made the review of the videotapes easier and will facilitate future research that will utilize the videotapes of the sarcastic recorded utterances. Another limitation concerns the technique used to elicit the sarcastic comments. Participants were provided with a list of questions that were designed to encourage the production of sarcasm. Thus, the results may have been attenuated in favor of more rather than less sarcasm produced. This, however, did not seem to be the case; the total amount of sarcasm produced was well under that reported in previous research findings. Indeed, without the prompting questions, it is possible that even less sarcasm might have been produced. A third limitation of the study concerns time. There may have been a fatigue element at play because the experiment took approximately an hour, even for participants who produced no sarcasm. When participants realized that if they identified more sarcastic utterances on the videotape, the experiment would last longer, it is possible that some participants may have neglected to identify some utterances. Other limitations concern the directions given to participants to "estimate" sarcastic utterances produced. The directions may have suggested guessing rather than recalling specific utterances, although participants were not pressured to make quick decisions and were allowed to spend as much time as they needed to complete this process. Confusion also may have resulted from the fact that participants were asked to make their evaluations in writing whereas their experience had been oral. Also, participating in a videotaped conversation may lead to cognitive overload. Participants must manage their impressions, follow their partners' dialog, decide on appropriate comments, and be prepared for unexpected events. Perception of sarcasm is an intriguing area that appears, from this study, to be frequently inaccurate. Given that previous research has demonstrated deficits in sarcasm recognition ability, investigation of sarcasm recognition holds promise for increasing our understanding of how the mind works and how communicators make sense of conversation. This study has made it clear how poorly communicators understand their own use of sarcasm, let alone others' use of sarcasm. Sarcasm represents a sophisticated communication tool. Understanding how communicators use this tool is a challenge that still lies ahead.