J Psycholinguist Res DOI 10.1007/s10936-014-9323-5

Context, Contrast, and Tone of Voice in Auditory Sarcasm Perception Daniel Voyer · Sophie-Hélène Thibodeau · Breanna J. Delong

© Springer Science+Business Media New York 2014

Abstract Four experiments were conducted to investigate the interplay between context and tone of voice in the perception of sarcasm. These experiments emphasized the role of contrast effects in sarcasm perception exclusively by means of auditory stimuli whereas most past research has relied on written material. In all experiments, a positive or negative computer-generated context spoken in a flat emotional tone was followed by a literally positive statement spoken in a sincere or sarcastic tone of voice. Participants indicated for each statement whether the intonation was sincere or sarcastic. In Experiment 1, a congruent context/tone of voice pairing (negative/sarcastic, positive/sincere) produced fast response times and proportions of sarcastic responses in the direction predicted by the tone of voice. Incongruent pairings produced mid-range proportions and slower response times. Experiment 2 introduced ambiguous contexts to determine whether a lower context/statements contrast would affect the proportion of sarcastic responses and response time. Results showed the expected findings for proportions (values between those obtained for congruent and incongruent pairings in the direction predicted by the tone of voice). However, response time failed to produce the predicted pattern, suggesting potential issues with the choice of stimuli. Experiments 3 and 4 extended the results of Experiments 1 and 2, respectively, to auditory stimuli based on written vignettes used in neuropsychological assessment. Results were exactly as predicted by contrast effects in both experiments. Taken together, the findings suggest that both context and tone influence how sarcasm is perceived while supporting the importance of contrast effects in sarcasm perception. Keywords

Auditory perception · Sarcasm · Communicative intent · Contrast effects

The study reported here was supported by a grant from the Natural Sciences and Engineering Research Council of Canada to D. Voyer. The authors thank Kyle J. Brymer and Jennifer A. Harding for their assistance with data collection and scoring. In addition the authors are grateful to Cheryl L. Techentin and Susan D. Voyer for their contribution to stimulus production. D. Voyer (B) · S.-H. Thibodeau · B. J. Delong Department of Psychology, University of New Brunswick, PO Box 4400, Fredericton, NB E3B 5A3, Canada e-mail: [email protected]

123

J Psycholinguist Res

Introduction A friend asks you to help her move into her new apartment. While your friend is packing up her things into boxes and lugging the boxes outside and into the moving van, you just hang around and watch, neglecting to offer any assistance or carry any items. Your friend then says to you, “You’re a big help.” In most cases, this statement would be perceived as sarcastic, but why? If instead you had only carried lighter boxes, would this same statement be perceived as sarcastic? What if you had carried many boxes and helped your friend move in half the time than it would have taken her otherwise? Now consider how the statement was uttered; did the speaker use a sarcastic tone of voice or a sincere tone? The current study aimed to explore questions similar to these while also determining what influence these two factors (context and tone of voice) have on the auditory perception of sarcasm. Before going further, it is important to explicitly state what the label “sarcasm” reflects in the present article. Somehow, this is not as straightforward as one might presume as sarcasm has proven to be a broad term, especially when considering that it stems from the broader concept of verbal irony (Gibbs 2000). However, Burgers et al. (2011a) offered a definition that represents quite well the concept of interest here. Specifically, these authors defined verbal irony as “an utterance with a literal evaluation that is implicitly contrary to its intended evaluation” (p. 190). Thus for example, stating “What a great day” in reference to a cold, rainy day uses a literally positive evaluation that implies a negative evaluation of the weather. However, the term “sarcasm” is used throughout this paper, as opposed to “sarcastic irony” or “sarcastic verbal irony” as a writing shortcut. In addition, the research presented here is particularly concerned with the situation where the contrary meaning is conveyed in part by a contradiction between what is said and the prosody used to utter it (see Cheang and Pell 2008 for more on this aspect). Therefore, our focus is on the use of sarcasm in speech. Considering this focus, when speech is viewed as a means to communicate intent, the use of sarcasm in conversation might come with a cost in terms of clarity. In particular, speakers who produces sarcasm put themselves at risk by increasing the amount of effort the listener must implement in order to comprehend the uttered statement (Gibbs and Colston 2002). Specifically, a crucial goal in spoken communication is the need for the listener to understand the intentions of the speaker, that is, communicative intentions (Sabbagh 1999). However, achievement of this goal is not necessarily obvious considering that a number of factors contribute to language processing. In fact, some researchers view the study of communicative intentions as relevant to research in social cognition (Channon et al. 2005; Walter et al. 2007), whereas others focus on the acoustic components of speech sounds (Nittrouer et al. 1989), and yet others explore the acoustic (Cheang and Pell 2008) and perceptual (Rockwell 2000) factors underlying the prosody used to convey intentions. Therefore, understanding communicative intentions clearly is a complex process that draws on multimodal information from several sources as well as on social cognitive skills. This complexity is also reflected in the large amount of research investigating markers of sarcasm both in written (e.g., Burgers et al. 2011b) and spoken (e.g., Rockwell 2000, 2007; Voyer and Techentin 2010) language. The list of markers of irony proposed by Attardo (2000) further emphasizes the multimodal and social cognitive quality of sarcasm. For example, Attardo (2000) covers cues relevant to acoustic (e.g., nasalization, intonation, etc.), typographical (e.g., “sic”, “!”, etc.) and kinesic (e.g., winks, nudges, etc.) means to convey verbal irony. However, Attardo (2000) also implies that verbal irony is poorly conveyed in written form when compared to the spoken form. Considering this statement, it is interesting that much of the research on sarcasm comprehension has relied on written vignettes, although researchers have also considered

123

J Psycholinguist Res

acoustic cues to sarcasm in spoken discourse (Bryant and Fox Tree 2005; Cheang and Pell 2008; Rockwell 2000, 2007; Voyer and Techentin 2010). One common approach to study sarcasm comprehension has been to present a brief context, followed by a statement requiring a judgement of communicative intent (e.g., Colston and O’Brien 2000a, b; Gerrig and Goldvarg 2000; Ivanko and Pexman 2003). This research clearly makes use of context as a marker of sarcasm. After all, the statement “what a great day” takes on a totally different meaning depending on whether it is uttered on a sunny or rainy day and this is true regardless of the tone of voice used that make the statement. Or, is it? Most theories of sarcasm comprehension emphasize the role of a shared context in this process [e.g., implicit display theory: Utsumi (2000); echoic reminder theory: Kreuz and Glucksberg (1989); allusional pretense theory: Kumon-Nakamura et al. (1995); direct access model: Gibbs (1994)]. However, Woodland and Voyer (2011) considered how identification of the tone of voice used to utter a statement might be affected by the context that precedes that statement. Specifically, these researchers presented a context spoken in a flat tone of voice followed by statements presented either in a sincere or sarcastic tone of voice. The context could be positive (e.g., you helped your friend move from her apartment and carried many heavy objects) or negative (you did not do anything to help while your friend moved). For these contexts, the statement “you’re a big help” was pronounced sincerely or sarcastically. Results from this experiment showed extreme and fast ratings toward “sarcastic” or “sincere” when a context and tone of voice matched (i.e., negative context and sarcastic tone; positive context and sincere tone). However, incongruent contexts and tones resulted in midrange (i. e., neutral) and slower ratings. Woodland and Voyer interpreted their findings as a twist on models of sarcasm comprehension that rely on the notion that sarcasm is perceived when a statement leads to a failed expectation (but see Kreuz and Link 2002). Specifically, such models would only apply to the verbal content and might fit better when considering exclusively written material. In contrast, for spoken material, using an inappropriate tone of voice for a given context would result in a failed expectation and produces slower, mid-range ratings. From a theoretical perspective, the manipulation of context and tone of voice by Woodland and Voyer (2011) only covered a very limited set among the many possible markers of sarcasm. Nevertheless, the interactive contribution of these two sources of information in their experiment would tend to support theories of sarcasm perception in which cues are processed by direct access (Gibbs 1994) or in parallel (Katz 2005) to reach an accurate interpretation of a situation involving sarcasm. From this perspective, the inclusion of congruent markers in the Woodland and Voyer experiment would have contributed to a fast classification of the tone of voice as participants would respond as soon as they had enough information. On the surface, these findings do not necessarily fit as well with models that posit a two stage model in which literal meaning is processed first and then compared to context (e.g., Giora 1997). Such models would likely predict minimal effects of context on response time. Specifically, as tone of voice identification was the only task required from participants, this factor would likely trump both the context and the literal meaning to produce a fast response regardless of context. However, teasing apart these theories is not the central purpose of the present study and this theoretical pursuit lays in future uses of the paradigm proposed here. In addition to these considerations, the approach developed by Woodland and Voyer (2011) raises further questions. Specifically, the allusional pretense theory (Kumon-Nakamura et al. 1995) would predict that a listener perceives a remark as sarcastic more easily when it differs substantially from their expectations; the expectation has been violated, signalling a contrast effect. The violation then causes a shift in perception and, subsequently, an interpretation of sarcasm. From our perspective then, the allusional pretense theory seems to rely heavily

123

J Psycholinguist Res

on situational disparity. For instance, if Peter shows up 5 min late for an appointment, Betty might say something like, “Thanks for being on time.” This would most likely be construed as a literal statement. However, if Peter were to show up an hour late for his appointment, the same statement would more likely be perceived as sarcastic because there is a greater disparity between what was expected and what was actually said. Such anecdotal evidence fits well with the notion that situational disparity or contrast between the context and the sarcastic statement contributes to the perception of sarcasm (Colston 2002). For example, Gerrig and Goldvarg (2000) used various written vignettes to manipulate the magnitude of situational disparity (e.g., extent of failed expectation) and demonstrated that high-disparity situations led to higher sarcasm ratings than low-disparity situations. From this perspective, the results obtained by Woodland and Voyer (2011) could be interpreted as reflecting the fact that the inclusion of a sarcastic tone of voice in addition to a literal statement that contradicts the situation (e.g., it is pouring rain outside but my friend calls it a beautiful day) further emphasized the situational disparity and this likely increased the likelihood that the utterance would be rated as highly sarcastic. In contrast, when the tone of voice and the literal content send a mixed message, situational contrast is diluted and it results in a mid-range rating. Therefore, the tone of voice used to utter a statement might contribute to its contrast with the referent context. Pexman et al. (2010) suggested that the influence of contrast on perceived sarcasm might derive from the fact that a strong contrast clearly emphasizes the speaker’s true attitude. Therefore, a congruent tone of voice in Woodland and Voyer’s experiment likely ensured that the speaker’s intentions were clearly communicated. The present study modified the methodology used by Woodland and Voyer (2011) by introducing ambiguous contexts to contribute to the literature relevant to the importance of contrast. In doing so, it focuses on context, intonation, and contrast as markers of sarcasm. Of course, the investigation of intonation as a marker of sarcasm requires the use of auditory targets. In addition, in the present study, contexts were presented as auditory stimuli, not only the targets. Keeping all stimuli in the same modality is a novel aspect of the present approach. In fact, most of the research on contrasts effects has been conducted with written material (e.g., Colston and O’Brien 2000a, b; Gerrig and Goldvarg 2000; Ivanko and Pexman 2003). Therefore, the present study introduces a novel aspect by investigating context and contrasts effects exclusively through material presented in the auditory modality. This allowed an examination of the potential contribution of tone of voice in the contrast effect. To our knowledge, the interaction between context and tone of voice when both components are presented as auditory stimuli has received little if any attention in the literature, whereas contrasts effect have not been investigated yet in this setting. Although a number of studies have considered various combinations of context and auditory statements, the Woodland and Voyer (2011) stands out as the only case where context valence and tone of voice were fully manipulated in the auditory modality. An careful search of relevant studies showed that they either did not have purely auditory conditions where both the context and target were presented in the auditory modality (Bryant and Fox Tree 2002); they did not empirically manipulate context valence (Milosky and Ford 1997); they did not examine the differential effect of the context and tone of voice used to utter a statement (Capelli et al. 1990); they focused on production, not perception (Anoli et al. 2002; Cheang and Pell 2008); or they were only interested in the affective component of sarcasm, as opposed to the sarcastic tone of voice itself (Nakassis and Snedeker 2002). In addition to these issues, a crucial point to note is that none of the auditory studies discussed above implemented a manipulation of contrast effects. Therefore, whether contrasts effects will emerge with the exclusive use of spoken material still remains an empirical question.

123

J Psycholinguist Res

Aside from empirical and theoretical issues, the identification of markers of sarcasm in auditory stimuli also has important implications from a practical perspective. For example, difficulties identifying sarcasm is often presented as a symptom of autism spectrum disorder (Happé 1994). Therefore, identifying sources of information relevant to the correct identification of sarcastic intent might contribute to programs that aim to teach individuals with autism spectrum disorders the components they need to focus on to achieve better sarcasm identification performance (Persicke et al. 2013). Similarly, Vannier and Voyer (2012) reported that the ability to identify a sarcastic tone of voice was associated with higher quality peer relationships and less negative romantic relationships. This suggests that identifying cues that contribute to the identification of sarcastic intent would also likely provide useful information for therapists aiming to improve relationship quality in their patients. Examination of markers of sarcasm in the present study therefore has both theoretical and practical implications. Essentially, the present study followed a typical approach in the study of the role of context in sarcasm perception. For example, Gerrig and Goldvarg (2000) studied contrasts effects by presenting a written context that concluded with a potentially sarcastic statement made by one of the protagonists. This approach was essentially followed here, but with all material presented through the auditory modality. In addition, the statement that followed the context was always literally positive (e.g., “what a great day”). As a starting point, Experiment 1 attempted to replicate the findings obtained by Woodland and Voyer (2011) with a more intuitive dependent variable (proportion of sarcastic responses based on a forced choice decision instead of a rating, as used by Woodland and Voyer). Experiment 2 provided the crucial examination of the role of contrast by presenting a positive, negative, or ambiguous context followed by sincere and sarcastic utterances. These context-tone pairings were either strongly or mildly congruent or incongruent. For example, contexts that refer to a “cold rainy day” or a “warm sunny day” would reflect strong congruence or incongruence, respectively, when paired with a sarcastic tone of voice. Alternatively, an ambiguous context describing a warm but rainy day would offer mild congruence for a sarcastic or sincere statement. Therefore, a sarcastic tone was paired with each of the three contexts mentioned above (positive, negative, and ambiguous) while a sincere tone was also presented and paired with the same three contexts. A positive situation was considered one that was expected to produce a sincere response and should result in a high percentage of responses in that direction. A negative situation was one that would likely produce a sarcastic response. Therefore, the positive statement uttered by the protagonist would produce large situational disparity and, when combined with a sarcastic tone, result in a high percentage of sarcastic responses. An ambiguous situation would be one through which either a sincere response or a sarcastic response could be produced. This should dilute the contrast effect and result in a percentage of sarcastic responses between 50 % and the extreme values observed in congruent pairings. Experiment 3 and 4 were conducted to replicate the results of Experiment 1 and 2 with different stimulus material to confirm the robustness of the findings. Specifically, Experiments 3 and 4 adapted for auditory presentation written vignettes drawn from a validated neuropsychological test of sarcasm perception. This allowed us to eliminate potentially stereotypical sarcastic statements in Experiment 1 and 2 as well as to generalize the results to new stimuli and from written to auditory material.

Experiment 1 The first experiment aimed to replicate the findings obtained by Woodland and Voyer (2011) with a new dependent variable. Specifically, these authors had participants rate the tone of

123

J Psycholinguist Res

voice of the statement that followed the context on a scale from 1 (very sincere) to 7 (very sarcastic). We proposed to force participants to make a simple decision by choosing either “sincere” or “sarcastic” as a response. By coding a sarcastic response as “1” and a sincere response as “0”, the resulting measure across all trials reflects a proportion of sarcastic responses that is directly interpretable without reference to the underlying endpoints of the scale. Based on this dependent variable and following the results obtained by Woodland and Voyer (2011), it is expected that the proportion of sarcastic responses will be around .50 and long response times will be obtained when context and tone are incongruent. In contrast, when context and tone are congruent, the proportion of sarcastic response will be at the expected extreme (near 0 for the positive-sincere tone of voice pairing; near 1 for the negative-sarcastic pairing) and response times will be relatively fast. Method Participants Sixty-four undergraduate students (34 women, 30 men), ranging between the ages of 17 and 42 (M = 20.03, S D = 4.15) participated in Experiment 1, although one female declined to report her age. These participants were all recruited in various sections of introductory psychology and they received bonus credit for the course. All reported normal hearing. The experiment followed proper ethical guidelines and was approved by the institutional research ethics board. Materials The materials were exactly the same as Woodland and Voyer (2011). Specifically, the twelve pre-recorded statements previously used and validated by Voyer et al. (2008) were used. The validation and production of these statements are presented in Voyer et al. (2008). Woodland and Voyer (2011) created two relevant situational contexts (positive, negative) to precede each statement. In this process they attempted to keep the contexts plausible and as similar as possible to each other. Considering that using a real human voice to narrate the context statements might introduce biasing prosodic cues, each context was produced by a computer generated voice with the text-to-speech program created by AT&T research labs (http://www.2.research.att.com/~ttsweb/tts/demo.php). This program allowed the production of contexts presented in a completely monotone human-sounding voice. A female voice (Lauren, US English) with a 16 bit quality and a sampling rate of 25.6 kHz was used to produce the contexts. Each context was paired with its relevant statement spoken in a sarcastic or sincere tone of voice, resulting in a total of four pairings (2 contexts × 2 tones of voice) for each of the 12 phrases, resulting in 48 combinations. The context situations and statements are presented in Appendix 1 (only negative and positive contexts apply to Experiment 1). Following the approach used by Woodland and Voyer (2011), the 48 combinations were presented twice for a total of 96 trials completed by each participant. Procedure Participants were tested in groups of at most four. They were seated in front of a computer and each participant was separated from its neighbors by screens. Stimuli were delivered through Sony MDR-150 headphones with a mean peak intensity adjusted to 70 dB (as measured with

123

J Psycholinguist Res Table 1 Mean and standard deviation (in parentheses) for the proportion of sarcastic responses and response time (in ms) as a function of context and tone of voice in Experiment 1 Context

Tone of voice Sarcastic

Overall for context Sincere

Proportion of sarcastic responses (P) Negative

.92 (.10)

.52 (.22)

.72 (.12)

Positive

.65 (.26)

.11 (.10)

.32 (.14)

Overall P for tone

.79 (.16)

.32 (.13)

Response time (RT) Negative

1,230 (509)

1,652 (852)

1,441 (604)

Positive

1,808 (843)

1,274 (478)

1,542 (596)

Overall RT for tone

1,519 (629)

1,463 (615)

a General Radio sound level meter, Model 1565-A). Participants were first presented with either a positive or negative context and, following a 400 ms interval, the appropriate probe statement in a sincere or sarcastic tone was presented. At this point, participants were asked to indicate whether they thought the statement was sincere or sarcastic by clicking one of two alternative options presented on the computer screen. The next trial started immediately after production of a response. The 96 trials were presented in random order until all were completed. A program written using E-prime Version 1.1 (Schneider et al. 2002a, b) recorded response and response time (in ms). Results Proportion of Sarcastic Responses The proportion of sarcastic responses was analyzed by means of a repeated measures analysis of variance (ANOVA) with context (negative, positive) and tone of voice (sarcastic and sincere) as within-subject factors1 . This analysis revealed a significant main effect of context, F(1, 63) = 173.01, p < .01, η p = .73, reflecting the fact that, when a negative context was presented, it was typically associated with a relatively high percentage of sarcastic responses whereas a positive context produced a relatively small percentage of sarcastic responses (see Table 1 for relevant means). Tone of voice also produced a significant main effect, F(1, 63) = 272.11, p < .01, η p = .81. In this case, the main effect indicates that a sarcastic tone of voice produced a high proportion of sarcastic response whereas this proportion was relatively low for a sincere tone of voice, as seen in Table 1. More critical to the expected findings, context and tone of voice produced a significant interaction, F(1, 63) = 29.47, p < .01, η p = .32. The means and standard deviations associated with this interaction are also presented in Table 1. Considering that the ordering of the means derived from this interaction is crucial to the predicted findings, follow-up 1 Considering that the dependent variable was a proportion, these data were also analyzed with logit mixed

modeling, as recommended by Jaeger (2008). When analyzed in this manner, the results remained exactly the same in all experiments. Accordingly, only analysis of variance results are reported as this approach is likely more familiar to all readers. In addition, analysis of variance greatly facilitates the examination of the pattern of differences among means crucial to the hypotheses.

123

J Psycholinguist Res

analysis used multiple comparisons among the four context/tone of voice pairings at the .05 level of significance with a Bonferroni correction. This analysis indicated that all differences among the means presented in Table 1 achieved significance, except for the difference between the two incongruent pairings (positive context/sarcastic tone compared to negative context/sincere tone). This finding indicated that the proportion of sarcastic responses was highest for the negative/sarcastic pairing and lowest for the positive/sincere pairing, but that incongruent pairings produced mid-range scores that did not differ from each other. Response Time In Experiment 1, as well as in all subsequent experiments, analyses with response time as dependent variable required us to take into account the fact that, if we considered participants’ response to the sounds and computed response time separately for their “sarcastic” and “sincere” responses as a function of context/tone pairings, in some cases, it would produce response times based on very few (if any) responses. For example, response times for “sincere” responses to a negative context/sarcastic tone pairings would be based on very few trials. Therefore, intended tone of voice rather than actual response was used to label response time. Although this produces response times that reflect the intended tone rather than the perceived tone, we felt that this would capture the effect of congruence and eliminate empty cells in response time estimation. In addition, the presence of outliers is well documented as a potential source of extraneous variance in response time (Ratcliff 1993). Accordingly, an inverse transformation was applied to response times as Ratcliff (1993) showed that such an approach minimizes the influence of outliers on observed results while maximizing power when compared to other available approaches (e.g., individual data trimming, median response time, etc.). Nevertheless, response time data are presented in their original untransformed values in what follows for clarity of exposition. Response time data were analyzed using the same approach as with the proportion of sarcastic responses. Both the main effects of context and tone of voice failed to achieve significance, F(1, 63) = 2.67, p > .10, η p = .04, and F(1, 63) = 0.013, p > .91, η p < .001, respectively. More importantly, the interaction of context and tone of voice was significant for response time as well, F(1, 63) = 106.43, p < .01, η p = .63. Relevant means are also presented in Table 1. Follow-up analyses using the same approach as for the proportion of correct responses showed that incongruent pairings produced slower responses than the congruent pairings but no significant differences were obtained on response time within congruent or incongruent pairings. Discussion The purpose of Experiment 1 was to replicate the results obtained by Woodland and Voyer (2011) but with a more readily interpretable dependent variable. This goal was clearly achieved as results showed an appropriate proportion of sarcastic responses for congruent pairings, that is, when a negative context was paired with a sarcastic tone of voice, whereas this proportion was considerably reduced when a positive context preceded a sincere tone. Incongruent pairings (positive context/sarcastic tone, negative context/sincere tone) produced proportions of sarcastic responses near .5 that did not differ significantly from each other. To bolster these findings, response time for congruent pairings were relatively fast and similar, whereas incongruent pairings produced slower response times that were statistically similar. The results therefore parallel those obtained by Woodland and Voyer (2011) despite the change in dependent variable.

123

J Psycholinguist Res

These findings therefore can be interpreted in the context of the mechanisms used by Woodland and Voyer (2011) to explain their results. However, these will not be discussed here as Experiment 1 only aimed to set the stage for an examination of contrast effects. Using the same set of stimuli from Experiment 1, Experiment 2 also included an ambiguous context to provide a reduced amount of contrast in the design. If the results demonstrate a contrast effect, the proportion of sarcastic responses for an ambiguous context followed by a sincere tone should be between that obtained for the positive context and .5, whereas the sarcastic tone should result in a proportion between .5 and that obtained with the negative context. Experiment 2 Method Participants Ninety-six undergraduate students, 70 women and 26 men, ages 18–40 years (M = 19.77, S D = 3.38) recruited from the same source as in Experiment 1 participated in this experiment. None of them completed Experiment 1 and they all reported normal hearing. The experiment followed proper ethical guidelines and was approved by the institutional research ethics board. Materials and Procedure Materials and procedure were the same as in Experiment 1 except for the following changes. An ambiguous context relevant to each of the twelve statements was produced. These additional contexts are also presented in Appendix 1. They were prepared in the same way as in Experiment 1 and involved mostly a change in wording that weakened the contrast when compared to the negative contexts. They were constrained by the need to make them as similar as possible to the contexts used in Experiment 1. The addition of the ambiguous context resulted in three situational contexts paired with a statement spoken either sincerely or sarcastically. Thus, there were 72 combinations derived from 12 statements × 3 contexts (ambiguous, negative, positive) × 2 tones of voice (sincere, sarcastic). As in Experiment 1, each combination was presented twice, resulting in a total of 144 trials presented in random order. Results Proportion of Sarcastic Responses The proportion of sarcastic responses was examined with a repeated measures ANOVA with context (ambiguous, negative, and positive) and tone of voice (sarcastic and sincere) as within-subject factors. Results of this analysis showed a significant main effect of context, F(2, 190) = 178.07, p < .01, η p = .65. Means relevant to this effect are presented in Table 2. Pairwise comparison among these means at the .05 level with a Bonferroni correction showed that they were all significantly different from each other, with the following pattern of differences: negative > ambiguous > positive. Tone of voice also produced a significant main effect, F(1, 95) = 263.99, p < .01, η p = .74. The means relevant to this finding, also presented in Table 2, reflect a greater proportion of sarcastic response to the sarcastic tone of voice compared to the sincere tone.

123

J Psycholinguist Res Table 2 Mean and standard deviation (in parentheses) for the proportion of sarcastic responses and raw response time (in ms) as a function of context and tone of voice in Experiment 2 Context

Tone of voice Sarcastic

Overall for context Sincere

Proportion of sarcastic responses (P) Negative

.95 (.07)

.57 (.28)

.76 (.14)

Positive

.63 (.31)

.11 (.11)

.37 (.17)

Ambiguous

.91 (.10)

.39 (.23)

.65 (11)

Overall P for tone

.83 (.14)

.35 (.18)

Response time (RT) Negative

1,050 (618)

1,388 (584)

1,219 (552)

Positive

1,401 (738)

1,121 (480)

1,261 (569)

Ambiguous

1,255 (653)

1,511 (677)

1,383 (623)

Overall RT for tone

1,236 (616)

1,340 (525)

Finally, context and tone produced a significant interaction, F(2, 190) = 43.69, p < .01, η p = .32, with relevant means presented in Table 2. As in Experiment 1, differences among means were examined with pairwise comparisons at the .05 level with a Bonferroni correction. This analysis revealed that they were all significantly different from each other except for the difference between the two strongly incongruent pairings (positive context/sarcastic tone compared to negative context/sincere tone). In brief, the pattern of difference reflected the following order of means: negative/sarcastic > ambiguous/sarcastic > (positive/sarcastic = negative/sincere) > ambiguous/sincere > positive/sincere (with combinations within parentheses not differing from each other but differing from means outside the parentheses). Response Time Response time data were analyzed with the same approach as above. As before, results of this analysis showed a significant main effect of context, F(2, 190) = 14.37, p < .01, η p = .13. Means relevant to this effect are presented in Table 2. Pairwise comparison among these means at the .05 level with a Bonferroni correction showed that means for the negative and positive contexts did not differ from each other, but they both showed significantly faster response times than the ambiguous context. Tone of voice also produced a significant main effect, F(1, 95) = 22.01, p < .01, η p = .19. The means relevant to this finding, also presented in Table 2, reflects faster response times for the sarcastic tone over the sincere tone. Finally, the context by tone interaction was also significant on response time, F(2, 190) = 59.72, p < .01, η p = .39, with relevant means presented in Table 2. Again, differences among means were examined with pairwise comparisons. This analysis produced a complex pattern of differences as adjacent means were found not to differ significantly from each other. Nevertheless, the order of the means (from fastest to slowest response time) was as follows: 1-negative sarcastic; 2-positive/sincere; 3-ambiguous/sarcastic; 4-negative/sincere; 5positive/sarcastic; 6-ambiguous/sincere. Essentially, congruent pairings produced the fastest response times, whereas incongruent pairings (either strong, or weak due to an ambiguous context) slowed response time.

123

J Psycholinguist Res

Discussion Experiment 2 can be seen as forming the core of the present study. Essentially, it was an incursion into examining the importance of contrast to promote sarcasm perception but with auditory stimuli. This use of auditory information was the novel aspect of this experiment as past research has neglected the role tone of voice or prosody can play in further promoting a contrast between literal and figurative content. Essentially, a sarcastic tone of voice in the presence of a high literal contrast (i.e., it is pouring rain but I describe the weather as great) presumably would act to confirm the sarcastic intent for a listener. From that perspective, it was expected that the introduction of an ambiguous context followed by a sincere tone would result in a proportion of sarcastic responses between that obtained for the positive context and .5, whereas the sarcastic tone should results in a proportion between .5 and that obtained with the negative context. The results confirmed these predictions. Although the results with response time were not as clear-cut as for the proportion of sarcastic responses, they still supported the notion that high contrast makes it easier to perceive sarcasm, especially when the tone of voice confirms this intent for the listener. Specifically, as can be seen in Table 2, the fastest response time was obtained when the literal contrast was highest (i.e., in the negative context) and when the sarcastic intent was confirmed by the tone of voice. The incongruent positive/sarcastic pairing produced the slowest response time whereas the ambiguous/sarcastic pair produced a response time of intermediate value. This mirrors what was obtained with the proportion of sarcastic responses and provides strong support for the contribution of tone of voice to emphasize the contrast effect in sarcasm perception. In contrast, for the sincere tone of voice, it is the ambiguous situation that produced the slowest response time. One possibility to explain this finding would be that the ambiguous situation might not have been as ambiguous as hoped. After all, the contexts were not pilot tested to assess their clarity. However, considering that the results correspond to our predictions for the proportion of sarcastic responses regardless of tone of voice, and for the sarcastic tone of voice on response time, this essentially provides a manipulation check on the contexts, despite the absence of pilot testing. A more likely explanation for the apparent discrepancy of the response time results is simply that the ambiguous situation did not lead participants to expect a clearly sincere tone of voice. Plausibly, most people would not wholeheartedly describe as great a day when it is grey even though it is warm. Therefore, response times were slowed down as participants tried to assimilate the information provided by the context, literal meaning, and tone of voice. Perhaps a statement in a neutral tone of voice would have resulted in a pattern of means as a function of context similar to that observed for the sarcastic tone of voice. However, this remains an empirical question. In contrast, for the sarcastic tone, the ambiguous situation likely produced some expectations of sarcasm, but the contrast between situation and literal meaning was not strong enough to promote a very quick response time as was observed for the negative/sarcastic pairing. In any case, these results require replication before further efforts are expanded to account for them. In fact, the question of replication might be crucial here. Specifically, to our knowledge, only the material originally developed by Voyer et al. (2008) has been used as a means to demonstrate the interaction between context and tone of voice in sarcasm perception in a purely auditory task. However, some researchers might feel that many of the statements that arose from this research would lead by themselves to a sarcastic interpretation. For example, the statement “tell me about it” is often used rhetorically and most listeners would likely not

123

J Psycholinguist Res

expect that the speaker really wants to be told about an annoying phone call in detail. In fact, it could be argued that many of the statements used in Experiments 1 and 2 are essentially stereotypical sarcastic statements. Of course, the fact that participants could also perceive them as sincere argues against this possibility. Nevertheless, this discussion emphasizes the need to extend the present findings to a new stimulus set that might not be viewed as loaded with sarcasm from the start. Accordingly, a new set of statements and contexts was developed for Experiments 3 and 4. A strength of these statements is that they come from written vignettes used in the neuropsychological assessment of sarcasm perception when right hemisphere damage is suspected. Experiment 3 validated these statements in a replication of Experiment 1, whereas Experiment 4 examined whether the results of Experiment 2 could be generalized to a new set of stimuli.

Experiment 3 Method Participants A total of 84 undergraduate students (16 male and 68 female) from the same source as the earlier experiments participated in Experiment 3. They all reported normal hearing and none were involved in either Experiment 1 or 2. The age of participants ranged from 17 to 59 years (M = 21.43, S D = 7.32). The experiment followed proper ethical guidelines and was approved by the institutional research ethics board. Materials and Procedure Although the procedure used in Experiment 3 was the same as in the previous two experiments, it also required the production of a new set of stimuli. Specifically, the new selection of stimuli was based on six written vignette (contexts) taken from the Sarcasm subtest of the Right Hemisphere Communication Battery (Gardner and Brownell 1986), following the approach used by Giora et al. (2000). These phrases were each modified to provide both a positive and a negative context (the original vignettes all reflect a negative context). As before, to control biases that could occur with real human speech, these contexts were presented by the same computerized voice used in Experiments 1 and 2. The original contexts were minimally modified to keep within the word limit prescribed by the text to speech program and to ensure intelligibility from the electronic voice. The probe statements that followed the contexts were pre-recorded to convey a sarcastic and a sincere tone of voice by using a real female voice. This native English speaker produced four tokens of each of the six sentences by two tones of voice combinations, resulting in 48 different sounds. These 48 stimuli were then presented to 10 undergraduate raters not involved in later testing. In this pilot study, the raters were asked to indicate for each stimulus whether the tone of voice reflected sincere, sarcastic, or neither by clicking on the appropriate choice on the computer screen. The stimuli selected for experimentation all showed an agreement rate above 75 % among the judges, except for “Fred is a good soloist” in a sarcastic tone that only reached a peak agreement rate of 56 % despite the production of additional “corrective” tokens. However, this rate of agreement is well above chance (z = 4.82, p < .001, based on a binomial test with chance = 33.3 %). Accordingly, this stimulus was included in the final selection although its presence should be kept in mind when interpreting the results.

123

J Psycholinguist Res Table 3 Mean and standard deviation (in parentheses) for the proportion of sarcastic responses and raw response time (in ms) as a function of context and tone of voice in Experiment 3 Context

Tone of voice Sarcastic

Overall for context Sincere

Proportion of sarcastic responses (P) Negative

.94 (.10)

.55 (.31)

.76 (.17)

Positive

.58 (.32)

.09 (.12)

.32 (.18)

Overall P for tone

.75 (.16)

.33 (.16)

Response time (RT) Negative

1,291 (752)

2,143 (2,111)

1,717 (1274)

Positive

2,027 (1,099)

1,535 (1,063)

1,781 (883)

Overall RT for tone

1,659 (796)

1,839 (1,411)

The use of six situations under two contexts paired with two tones of voice resulted in 24 possible combinations. As in the earlier experiments, each combination was presented twice for a total of 48 trials. The context situations and statements are presented in Appendix 2 (only negative and positive contexts apply to Experiment 3). Results Proportion of Sarcastic Responses A repeated measures ANOVA was conducted with context (positive, negative) and tone of voice (sarcastic, sincere) as independent within-subject variables and with the proportion of sarcastic responses as the dependent variable. A significant main effect of context was found, F(1, 83) = 181.06, p < .01, η p = .69; as well as a significant main effect of tone of voice, F(1, 83) = 160.97, p < .01, η p = .66. Means corresponding to these results can be found in Table 3. They show that a sarcastic tone or a negative context produced a significantly greater proportion of sarcastic responses than the sincere tone or positive context, respectively, as should be expected from the earlier experiments. These main effects were qualified by a context by tone of voice interaction, F(1, 83) = 12.56, p < .01, η p = .13. The means and standard deviations associated with this interaction are also presented in Table 3. Multiple comparisons among the four context/tone of voice pairings at the .05 level of significance with a Bonferroni correction indicated that all differences among the means presented in Table 3 achieved significance, except for the difference between the two incongruent pairings (positive context/sarcastic tone compared to negative context/sincere tone). Therefore, as in Experiment 1, the proportion of sarcastic responses was highest for the negative/sarcastic pairing and lowest for the positive/sincere pairing, whereas the incongruent pairings producing mid-range values that did not differ from each other. Response Time The analysis of response time data used the same approach as with the proportion of sarcastic responses. This analysis revealed a significant main effect both for context, F(1, 83) = 9.30, p < .01, η p = .10, and tone of voice, F(1, 83) = 5.43, p = .022, η p = .06.

123

J Psycholinguist Res

However, these main effects were qualified by a significant context by tone interaction, F(1, 83) = 32.57, p < .01, η p = .28. The means and standard deviations associated with this interaction are presented in Table 3. Multiple comparisons were performed as was done for the proportion of sarcastic responses and, here as well, they indicated that all differences among the means presented in Table 3 achieved significance, except for the difference between the two incongruent pairings (positive context/sarcastic tone compared to negative context/sincere tone). Therefore, response time was slowest for the incongruent pairings, followed by positive/sincere, and lastly by negative/sarcastic. Discussion Experiment 3 aimed to replicate the findings obtained by Woodland and Voyer (2011) as well as those of Experiment 1 but with a new set of stimuli that were not perceived as potentially reflecting stereotypically sarcastic statements to determine how well these results generalize. Consistent with these prior findings, results from the current study showed that participants presented with a congruent pairing of context and tone of voice had faster response times than when presented with incongruent pairings. In addition, incongruent pairings yielded a mid-range proportion of sarcastic responses ranging close to .5. These findings indicate that when participants were presented with incongruent pairings, they may have had a harder time identifying the tone of voice, therefore explaining the moderate percentage of sarcastic responses and slower responses they produced. However, for the congruent pairing of a negative context paired with a sarcastic tone of voice, there was a large proportion of sarcastic responses, whereas the proportion was very small for the positive/sincere pairing, as expected. From this perspective, Experiment 3 did provide a successful generalization of findings by Woodland and Voyer (2011) and Experiment 1. The results also validate the materials used in Experiment 3, clearly showing that written vignettes can easily be converted to auditory stimuli for the study of sarcasm perception. Considering that the overarching goal of the present study is to examine contrast effects, Experiment 4 aimed to further examine these effects in a context similar to Experiment 2, but with the novel material validated in Experiment 3.

Experiment 4 Method Participants Eighty-three undergraduate students, 60 women and 23 men, aged 18 to 41 years (M = 19.57, S D = 3.16) were recruited from the same source as in earlier experiments. None of them completed any of the earlier experiments and they all reported normal hearing. The experiment followed proper ethical guidelines and was approved by the institutional research ethics board. Materials and Procedure The procedure was the same as in Experiment 2. The contexts and tones of voice were the same as used in Experiment 3. However, additional ambiguous contexts were used for Experiment 4 as shown in Appendix 2. They followed from Experiment 3, so they were

123

J Psycholinguist Res

taken from the Sarcasm subtest of the Right Hemisphere Communication Battery (Gardner and Brownell 1986) and modified to fit word limits and intelligibility. These modifications were also guided by the need to make them similar to the contexts used in Experiment 3. The addition of the ambiguous context resulted in three situational contexts paired with a statement spoken either sincerely or sarcastically for the six different situations. Thus, there were 36 combinations derived from 6 statements × 3 contexts (ambiguous, negative, positive) × 2 tones of voice (sincere, sarcastic). As in the previous experiments, each combination was presented twice, resulting in a total of 72 trials presented in random order. Results Proportion of Sarcastic Responses The proportion of sarcastic responses was again examined with a repeated-measures ANOVA with context (ambiguous, negative, and positive) and tone of voice (sarcastic and sincere) as within-subject factors. Results of this analysis showed a significant main effect of context, F(2, 164) = 148.22, p < .01, η p = .64. Means relevant to this effect are presented in Table 4. Pairwise comparison among these means at the .05 level with a Bonferroni correction showed that they were all significantly different from each other, with the following pattern of differences: negative > ambiguous > positive. Tone of voice also produced a significant main effect, F(1, 82) = 25.37, p < .01, η p = .66. The means relevant to this finding, also presented in Table 4, reflect a greater proportion of sarcastic response for the corresponding tone of voice compared to the sincere tone. Finally, context and tone produced a significant interaction, F(2, 164) = 29.19, p < .01, η p = .26, with relevant means presented in Table 4. As before, differences among means were examined with pairwise comparisons at the .05 level with a Bonferroni correction. As in Experiment 2, this analysis revealed that the means were all significantly different from each other except for the difference between the two strongly incongruent pairings (positive context/sarcastic tone compared to negative context/sincere tone). Here as well, the pattern of difference reflected the following order of means: negative/sarcastic > ambiguous/sarcastic > (negative/sincere = positive/sarcastic) > ambiguous/sincere > positive/sincere. Table 4 Mean and standard deviation (in parentheses) for the proportion of sarcastic responses and raw response time (in ms) as a function of context and tone of voice in Experiment 4 Context

Tone of voice Sarcastic

Overall for context Sincere

Proportion of sarcastic responses (P) Negative

.93 (.17)

.58 (.32)

.76 (.19)

Positive

.53 (.33)

.06 (.14)

.30 (.17)

Ambiguous

.76 (.19)

.22 (.19)

.49 (09)

Overall P for tone

.74 (.19)

.29 (.17)

Response time (RT) Negative

1,091 (603)

1,757 (1,197)

1,424 (623)

Positive

1,968 (1,706)

1,143 (537)

1,555 (978)

Ambiguous

1,600 (978)

1,529 (886)

1,564 (836)

Overall RT for tone

1,553 (903)

1,476 (699)

123

J Psycholinguist Res

Response Time Response time data were analyzed with the same approach as above. The main effects of context was significant, F(2, 164) = 6.29, p < .01, η p = .07, but it was qualified by a significant context by tone interaction, F(1, 82) = 24.59, p < .01, η p = .23, with the corresponding means and standard deviations presented in Table 4. Multiple comparisons were performed as for the proportion of sarcastic responses and, as was found in Experiment 2, the analysis produced a complex pattern of differences with adjacent means generally not significantly different from each other. The order of the means (from fastest to slowest response time) was as follows: 1-negative/sarcastic; 2-positive/sincere; 3-ambiguous/sincere; 4-ambiguous/sarcastic; 5-negative/sincere; 6-positive/sarcastic. Therefore, as in Experiment 2, congruent pairing produced the fastest response times, whereas incongruent pairings slowed response time. However, here strong contrast pairings produced slower response times than weak contrast (i.e., ambiguous context) pairings. Discussion The aim of Experiment 4 was to extend the findings of Experiment 3 to emphasize contrasts effects in sarcasm perception. As such, it also aimed to replicate the results of Experiment 2 with new material. The results clearly replicated those of Experiment 2. In fact, the proportion of sarcastic responses produced the exact same pattern of means under both Experiments 2 and 4. For response time, the general finding of slower response time for incongruent context/tone of voice pairings was also replicated, although the findings of Experiment 4 provided even stronger evidence for the importance of contrast in sarcasm perception. Specifically, the results of Experiment 4 showed slower responses for the strongly incongruent (high contrast) pairings such as positive/sarcastic and negative/sincere, whereas the presentation of an ambiguous context resulted in somewhat faster response times than for these incongruent pairings and the fastest responses were observed for congruent pairings. The results therefore still emphasize the importance of contrast effects in the interpretation of sarcasm. Further elaborations on this will now follow as we will consider the overall results of the present series of experiments.

General Discussion The purpose of the current study was to examine the interactive influence of three markers of sarcasm in conveying the communicative intents of speakers, that is, context, intonation, and contrast. The interactive contribution of these three sources of information across four experiments provides further support for theories of sarcasm perception in which cues are processed by direct access (Gibbs 1994) or in parallel (Katz 2005). Specifically, the results support the notion that context, intonation, and contrast were processed in parallel to produce the observation pattern of responses. In contrast, the present findings do not fit as well with models hypothesizing a two stage model (e.g., Giora 1997). Most likely, such models would simply predict a main effect of tone of voice as this component was the only one that participants were asked to identify. At the empirical level, research examining context and contrast effects has relied exclusively on written vignette (Colston and O’Brien 2000a, b; Gerrig and Goldvarg 2000; Ivanko and Pexman 2003). Therefore, it seemed particularly relevant to determine whether findings obtained with written vignettes would generalize to purely auditory material in light of claims

123

J Psycholinguist Res

that the communicative intent of sarcasm is conveyed more easily in speech (Attardo 2000). In addition, intonation could only be investigated by means of auditory stimuli. The use of auditory presentations of stimuli where the situational context and tone of voice were varied to produce different levels of contrast constituted the central novel aspect of the experiments presented here. An additional novel aspect was the conversion of written vignettes from a neuropsychological tool into auditory material to examine sarcasm perception in speech stimuli. Experiment 1 aimed to replicate past results obtained by Woodland and Voyer (2011) with a new, more readily interpretable dependent variable. Results were clearly replicated in showing the expected proportion of “sarcastic” responses for congruent context and tone of voice pairings. Specifically, a negative context paired with a sarcastic tone of voice produced a high percentage of such responses, whereas a positive context/sincere tone of voice produce a low proportion of sarcastic responses. In contrast, incongruent context/tone pairing produced mid-range proportions of responses. Response times were also much faster for congruent than for incongruent pairings. Taken together these results showed that context can bias the interpretation of tone of voice. Furthermore, the response time data support the notion put forward by Woodland and Voyer (2011) that the incongruence between tone of voice and situation creates a failed expectation that increases difficulty when making a decision concerning the tone of voice and slows responses. For example, a negative context would lead one to expect a sarcastic tone of voice given the literal statement that is uttered and use of an unexpected sincere tone requires additional processing time from the listener to reach a decision. This replication is important in demonstrating the robustness of the earlier findings. However, Experiment 1 only involved negative situations that contrasted highly with the literal statement. Accordingly, an intermediate amount of situational contrast was introduced in Experiment 2 and we predicted that this should affect the proportion of sarcastic responses and response time so that they would be mid-way between what was obtained for the clearly congruent and clearly incongruent pairings. This is essentially what happened for the proportion of sarcastic responses as the ambiguous context followed by a sincere tone resulted in a proportion of sarcastic responses between that obtained for the positive context and .5, whereas the sarcastic tone resulted in a proportion between .5 and that obtained with the negative context. However, these findings were not mirrored for response time as the slowest response was actually obtained for the ambiguous/sincere situation (see Table 2). This raises the possibility that there might have been something wrong with the ambiguous situations in Experiment 2. However, considering that the results were as expected for response time to the sarcastic tone (order of contexts, from fastest to slowest: negative, ambiguous, positive), a more likely account of this finding derives from the argument that was used to justify Experiments 3 and 4. Specifically, it is possible that the literal statements used in Experiment 1 and 2 lend themselves too well to sarcastic usage, to the point that they might be “stereotypically” sarcastic. From this perspective, the sincere tone might only have been believable when paired with a positive context whereas an ambiguous context still established too much of a contrast with the literal statement to cancel the expectation of a sarcastic tone of voice. This would account for the slow response time for the ambiguous/sincere pairing, although the fact that responses to this pairing were even slower than the presumably higher contrast negative/sincere cannot be readily explained with the theoretical perspective used here. Inconsistencies in the results of Experiment 2 therefore lend credence to the possibility that statements used in that experiment might be “stereotypically” sarcastic and this provided a strong justification for an extension of these findings to a new set of stimuli that might be clearly plausible when stated both sincerely and sarcastically. Experiment 3 showed that the

123

J Psycholinguist Res

results of Experiment 1 could be replicated with a new set of situations and statements based on written vignettes drawn from a validated test, paving the way for Experiment 4. In Experiment 4, the results were exactly as predicted as the ambiguous context produced a proportion of sarcastic responses and response time that were mid-way between what was obtained for the clearly congruent and clearly incongruent pairings. This suggests that the use of stereotypically sarcastic stimuli had some influence on the results of Experiment 2, where it showed clear support for contrast effects only on the proportion of sarcastic responses, not on response time. This support generalized to both dependent variables in Experiment 4, suggesting that the use of non-stereotypical stimuli in that experiment contributed to produce clear-cut results. More importantly, the results of Experiment 4 add weight to the claim that the contrast between context and literal meaning is an important marker of sarcasm and that tone of voice is an additional contributor to clarify communicative intent. Specifically, the data also supported the notion that the tone of voice exaggerated the contrast as it also affected how participants responded to the context as a function of its amount of contrast with the literal statement. Considering that all stimuli were presented aurally in the present study, it allows the generalization of past conclusions on the role of contrast in sarcasm perception derived from written material to spoken material. When considered in the broader context of language as means of communicating intent, the present experiments suggest that the proper combination of context, contrast, and tone of voice seems to leave little ambiguity when one attempts to convey sarcasm. Specifically, when a clearly negative context offers a strong contrast with a positive statement uttered in a sarcastic tone of voice, the data presented in Tables 1, 2, 3 and 4 showed that sarcasm was quickly identified as such on at least 92 % of trials. In contrast, the combination of a clearly positive context offering no contrast with a positive statement uttered sincerely produced at most 11 % of sarcastic responses and fast response time. It is thus clear that the combination of context, contrast, and tone of voice offers powerful clues to a speaker’s intent. However, these are only three of the available markers of sarcasm. For example, Attardo (2000) mentioned nine such markers that he deemed most commonly used or clearest, but he also indicated that these only reflect a partial list among a much larger set. For example, authors such a Kreuz and Roberts (1995) would likely include the use of hyperbole as a marker of sarcasm. However, the present discussion emphasizes the fact that communicating intent through language is a complex process that relies on multiple cues and social skills. From this perspective, the present study emphasizes only three such cues. At the practical level, identification of three cues to sarcasm as contributors to the accurate identification of communicative intent provides relevant information for therapists involved with problematic groups. In particular, one approach that has been used to improve the ability to perceive sarcasm for individuals with autism spectrum disorder has been to emphasize the cues that they should notice in conversation. In fact, Persicke et al. (2013) focused on the opposition between context and literal meaning. The present results suggest that ambiguous situational contrast might make communicative intent more difficult to convey, even in participants without a diagnosed deficit. Therefore the amount of situational contrast should not be ignored when training individuals with autism spectrum disorder, especially considering that they might have difficulty attending to contextual cues in the first place (Vermeulen 2014). The present results also suggest that tone of voice should be noted as an additional marker of sarcasm requiring attention when attempting to improve an individual’s ability to perceive sarcasm. In general, it is likely that emphasizing the cues to sarcasm intent identified here offers strong potential to improve sarcasm perception for individuals with autism spectrum disorder (Persicke et al. 2013) or for improving the quality of interpersonal relationships in the general population (Vannier and Voyer 2012).

123

J Psycholinguist Res

One potential drawback of the present study may be that only positive statements such as “Fred is a good soloist” or “what a great day” were used following each type of context. Although the exclusive use of positive statements may have promoted further a high contrast when paired with a negative context, it might be worthwhile to consider an examination of negative statements in the paradigm proposed here. For example, following a positive or negative context, a negative statement such as “Fred is a bad soloist” could be added and would be presented in both tones of voice. Using a negative statement would provide a higher contrast when following a positive context and might further emphasize the sarcastic intent. In fact, this approach would potentially turn an insult into a compliment or into a diluted insult (Boylan and Katz 2013). Regardless of its potential outcome, such research would further contribute to our understanding of the interplay between context and tone of voice in sarcasm perception. In summary, the present study presented four experiments examining the interaction of context and tone of voice on sarcasm perception in spoken material while exploring the role of contrast effect in this perceptual phenomenon. The results supported the need to integrate tone of voice as a contributor in theories using a failed expectation to explain sarcasm perception. The results also support models proposing direct or parallel processing of cues to sarcasm. The results indicated that there is a strong interplay between context and tone so that the tone of voice can serve to emphasize the contrast between context and literal meaning to promote the perception of sarcasm. In fact, it is legitimate to conclude that the combination of context, situational contrast, and tone of voice provided powerful cues to the sarcastic intent of speakers. These factors therefore deserve consideration in applied settings aiming to improve interpersonal communication. Appendix 1 See Table 5. Table 5 Contexts used in Experiments 1 and 2 [based on Woodland and Voyer (2011)] Statement

Context

Situation

Aren’t you smart

Positive

You’re working on a difficult problem for an assignment with a friend and you solve the problem. Your friend says...

Ambiguous You’re working on a difficult problem for an assignment with a friend and you get the solution from the T.A. Your friend says... Negative Aren’t you special Positive

You’re working on an easy problem for an assignment with a friend and you solve the problem. Your friend says... You apply for a financial award and get it. You go home and tell your room-mate the good news. Your roommate says...

Ambiguous You apply for a financial award but do not get the full amount. You go home and tell your room-mate. Your roommate says...

Isn’t she friendly

Negative

You and your room-mate apply for the same financial award and you get it but she doesn’t. Your roommate says...

Positive

You and a friend are at the mall. You see a classmate and she smiles and says hello to you. Your friend says...

Ambiguous You and a friend are at the mall. You see a classmate and she frowns but says hello to you. Your friend says... Negative

You and a friend are at the mall. You see a classmate and she frowns at you and walks away. Your friend says...

123

J Psycholinguist Res Table 5 continued Statement

Context

Isn’t this exciting Positive

Situation You’re in your favorite class and the professor tells you to do a fun exercise. Your classmate says...

Ambiguous You’re in your favorite class and you are given a boring exercise. Your classmate says...

Nice outfit

Negative

You’re in your least favorite class and you are given a very tedious and boring exercise. Your classmate says...

Positive

You walk into a room with a beautiful outfit on. Your friend says...

Ambiguous You walk into a room with a beautiful but stained outfit on. Your friend says... Tell me about it

Negative

You walk into a room with a very ugly outfit on. Your friend says...

Positive

You get off the phone and tell your roommate that you just finished planning your vacation with your friend. Your roommate says...

Ambiguous You get off the phone and tell your roommate that you just had a long conversation with a friend. Your roommate says...

Thanks a lot

Negative

You get off the phone and tell your roommate that you just had an annoying conversation with a telemarketer. Your roommate says...

Positive

Your professor gives you and your friend an easy and interesting assignment to complete. Your friend says...

Ambiguous Your professor gives you and your friend a difficult but interesting task to complete. Your friend says... Negative That worked well Positive

Your professor gives you and your friend a difficult and boring task to complete. Your friend says... Your car breaks down and you flag down a car to help. They stop and get your car running again. Your friend says...

Ambiguous Your car breaks down and you flag down a car to help. They stop but cannot help you. Your friend says...

That’s just great

Negative

Your car breaks down and you flag down a car to help. They stop but they made the problem worse. Your friend says...

Positive

You go home and tell your friend that you got a pay raise. Your friend says...

Ambiguous You go home and tell your friend that you did not get laid off from work but got a pay cut. Your friend says...

Wasn’t that fun

Negative

You go home and tell your friend that you got laid off from work. Your friend says...

Positive

You go to a party with a friend. The music was great and the people were interesting. Your friend says...

Ambiguous You go to a party with a friend. The music was great but the people were boring. Your friend says...

What a great day

Negative

You go to a party with a friend. The music was bad and the people were boring. Your friend says...

Positive

You walk outside and it is a warm, beautiful, and sunny day. Your friend says...

Ambiguous You walk outside and it is a warm, but grey day. Your friend says... Negative

123

You walk outside and it is a cold, dark, and rainy day. Your friend says...

J Psycholinguist Res Table 5 continued Statement

Context

You’re a big help Positive

Situation You’re helping a friend move into a new apartment. You carried many items and you helped her finish in half the time. Your friend says...

Ambiguous You’re helping a friend move into a new apartment. You carried many light objects but nothing heavy. Your friend says... Negative

You’re helping a friend move into her apartment but you just hang around and you don’t carry any items. Your friend says...

Appendix 2 See Table 6. Table 6 Contexts used in Experiments 3 and 4 [adapted from Gardner and Brownell (1986)] Statement

Context

Situation

Roger handled that case well

Positive

Jane and Roger were lawyers in a law firm. Jane hated Roger because he teased her for defending clients who could not afford to pay. One day Jane was at the courthouse while Roger was defending a wealthy man. He performed great; completely handling the very complex case. Jane said to another lawyer...

Ambiguous

Jane and Roger were lawyers in a law firm. Jane hated Roger because he teased her for defending clients who could not afford to pay. One day Jane was at the courthouse while Roger was defending a wealthy man. He performed fine, as hoped, handling what was a simple case. Jane said to another lawyer...

Negative

Jane and Roger were lawyers in a law firm. Jane hated Roger because he teased her for defending clients who could not afford to pay. One day Jane was at the courthouse while Roger was defending a wealthy man. He performed terribly; completely mishandling the simple case. Jane said to another lawyer...

Positive

Betsy and Don liked to play cards and were members of a bridge club. One night when Don and Betsy were partners, they won a difficult game because Don had played very skillfully. Betsy said to another player...

Ambiguous

Betsy and Don liked to play cards and were members of a bridge club. One night when Don and Betsy were partners, they won a game because they both had played well. Betsy said to another player...

Negative

Betsy and Don liked to play cards and were members of a bridge club. One night when Don and Betsy were partners, they lost an easy game because Don had played very poorly. Betsy said to another player...

Positive

Jenny and Max were shoemakers. Jenny was losing a lot of business to Max because Max sold his shoes for less than Jenny. Max was able to do this because he used low-quality materials, but the shoes Max made lasted a long time because he was a great shoemaker. Jenny said to another shoemaker...

Don plays bridge well

Max makes shoes well

123

J Psycholinguist Res Table 6 continued Statement

Doreen is a quick reader

Olivia is a good cook

Fred is a good soloist

123

Context

Situation

Ambiguous

Jenny and Max were shoemakers. Jenny was losing a lot of business to Max because Max sold his shoes for less than Jenny. Max was able to do this because he used low-quality materials, but the shoes Max made fell apart quickly because of the materials he used. Jenny said to another shoemaker...

Negative

Jenny and Max were shoemakers. Jenny was losing a lot of business to Max because Max sold his shoes for less than Jenny. Max was able to do this because he used low-quality materials. The shoes Max made fell apart quickly because he made bad shoes. Jenny said to another shoemaker...

Positive

Alice and Doreen were in the same English class. At the end of the semester the teacher awarded a prize to the student who read the most books. Both had wanted the prize, but Doreen read very fast and finished more books than anyone else. Alice said to another student in the class...

Ambiguous

Alice and Doreen were in the same English class. At the end of the semester the teacher awarded a prize to the student who read the most books. Both had wanted the prize, but someone else received it instead. Alice said to another student in the class...

Negative

Alice and Doreen were in the same English class. At the end of the semester the teacher awarded a prize to the student who read the most books. Both had wanted the prize, but Alice won. Doreen was slow and finished few books. Alice said to another student in the class...

Positive

Olivia and Penny shared a house with three other people. They all took turns cooking dinner. It was Olivia’s night to cook and she made lasagna. It came out great and was delicious. Penny said to another roommate...

Ambiguous

Olivia and Penny shared a house with three other people. They all took turns cooking dinner. It was Olivia’s night to cook and she made lasagna. It came out of the oven and turned out okay but not great. Penny said to another roommate...

Negative

Olivia and Penny shared a house with three other people. They all took turns cooking dinner. It was Olivia’s night to cook and she made lasagna. It came out burnt and almost impossible to eat. Penny said to another roommate...

Positive

Clara and Fred were violinists in the orchestra. Clara was the lead and Fred was second. Clara was jealous of Fred because the conductor chose Fred to perform a brief solo. During the rehearsal Fred played very well without making any mistakes. Clara said to another violinist...

J Psycholinguist Res Table 6 continued Statement

Context

Situation

Ambiguous

Clara and Fred were violinists in the orchestra. Clara was the lead violinist and Fred was second. The conductor chose Fred to perform a brief solo. During the rehearsal Fred played his solo and Clara said to another violinist...

Negative

Clara and Fred were violinists in the orchestra. Clara was the lead and Fred was second. Clara was jealous of Fred because the conductor chose Fred to perform a brief solo. During the rehearsal Fred played poorly making a lot of mistakes. Clara said to another violinist...

References Anoli, L., Ciceri, R., & Infantino, M. G. (2002). From “blame to praise” to “praise to blame”: Analysis of vocal patterns in ironic communication. International Journal of Psychology, 37, 266–276. doi:10.1080/ 00207590244000106. Attardo, S. (2000). Irony markers and functions: Towards a goal-oriented theory of irony and its processing. Rask, 12, 3–20. Boylan, J., & Katz, A. N. (2013). Ironic expression can simultaneously enhance and dilute perception of criticism. Discourse Processes, 50, 187–209. doi:10.1080/0163853X.2012.747041. Bryant, G. A., & Fox Tree, J. E. (2005). Is there an ironic tone of voice? Language and Speech, 48, 257–277. doi:10.1177/00238309050480030101. Bryant, G. A., & Fox Tree, J. E. (2002). Recognizing verbal irony in spontaneous speech. Metaphor and Symbol, 17, 99–117. doi:10.1207/S15327868MS1702_2. Burgers, C., van Mulken, M., & Schellens, P. J. (2011). Finding irony: An introduction of the verbal irony procedure (VIP). Metaphor and Symbol, 26, 186–205. doi:10.1080/10926488.2011.583194. Burgers, C., van Mulken, M., & Schellens, P. J. (2011). Type of evaluation and marking of irony: The role of perceived complexity and comprehension. Journal of Pragmatics, 44, 231–242. doi:10.1016/j.pragma. 2011.11.003. Channon, S., Pellijeff, A., & Rule, A. (2005). Social cognition after head injury: Sarcasm and theory of mind. Brain and Language, 93, 123–134. doi:10.1016/j.bandl.2004.09.002. Cheang, H. S., & Pell, M. D. (2008). The sound of sarcasm. Speech Communication, 50, 366–381. doi:10. 1016/j.specom.2007.11.003. Colston, H. L. (2002). Contrast and assimilation in verbal irony. Journal of Pragmatics, 34, 111–142. doi:10. 1016/S0378-2166(02)80008-X. Colston, H. L., & O’Brien, J. (2000a). Contrast and pragmatics in figurative language: Anything understatement can do, irony can do better. Journal of Pragmatics, 32, 1557–1583. doi:10.1016/S0378-2166(99)00110-1. Colston, H. L., & O’Brien, J. (2000b). Contrast of kind vs. contrast of magnitude: The pragmatic accomplishments of irony and hyperbole. Discourse Processes, 30, 179–199. doi:10.1207/S15326950DP3002_05. Capelli, C. A., Nakagawa, N., & Madden, C. M. (1990). How children understand sarcasm: The role of context and intonation. Child Development, 61, 1824–1841. doi:10.1111/j.1467-8624.1990.tb03568.x. Gardner, H., & Brownell, H. H. (1986). The right hemispheric communication battery. Boston: Psychology Service. Gerrig, R. J., & Goldvarg, Y. (2000). Additive effects in the perception of sarcasm: Situational disparity and echoic mention. Metaphor and Symbol, 15, 197–208. doi:10.1207/S15327868MS1504_1. Gibbs, R. W, Jr. (2000). Irony in talk among friends. Metaphor and Symbol, 15, 5–27. doi:10.1080/10926488. 2000.9678862. Gibbs, R. W. (1994). Figurative thought and figurative language. In M. A. Gernsbacher (Ed.), Handbook of psycholinguistics (pp. 411–446). New York, NY: Academic. Gibbs, R. W., & Colston, H. L. (2002). The risks and rewards of ironic communication. In L. Anolli, R. Ciceri, & G. Riva (Eds.), Say not to say: New perspectives on miscommunication (pp. 181–194). Washington, DC: IOS Press.

123

J Psycholinguist Res Giora, R. (1997). Understanding figurative and literal language: The graded salience hypothesis. Cognitive Linguistics, 7, 183–206. doi:10.1515/cogl.1997.8.3.183. Giora, R., Zaidel, E., Soroker, N., Batori, G., & Kasher, A. (2000). Differential effects of right- and lefthemisphere damage on understanding sarcasm and metaphor. Metaphor and Symbol, 15, 63–83. doi:10. 1080/10926488.2000.9678865. Happé, F. G. E. (1994). An advanced test of theory of mind: Understanding of story characters’ thoughts and feelings by able autistic, mentally handicapped, and normal children and adults. Journal of Autism and Developmental Disorders, 24, 129–154. doi:10.1007/BF02172093. Ivanko, S. L., & Pexman, P. M. (2003). Context incongruity and irony processing. Discourse Processes, 35, 241–279. doi:10.1207/S15326950DP3503_2. Jaeger, T. F. (2008). Categorical data analysis: Away from ANOVAs (transformation or not) and towards logit mixed models. Journal of Memory and Language, 59, 434–446. doi:10.1016/j.jml.2007.11.007. Katz, A. N. (2005). Discourse and sociocultural factors in understanding nonliteral language. In H. Colston & A. N. Katz (Eds.), Figurative language comprehension: Social and cultural influences (pp. 183–207). Mahwah, NJ: Erlbaum. Kreuz, R., & Glucksberg, S. (1989). How to be sarcastic: The echoic reminder theory of verbal irony. Journal of Experimental Psychology: General, 118, 374–386. doi:10.1037/0096-3445.118.4.374. Kreuz, R. J., & Link, K. E. (2002). Asymmetries in the use of verbal irony. Journal of Language & Social Psychology, 21, 127–143. doi:10.1177/02627X02021002002. Kreuz, R. J., & Roberts, R. M. (1995). Two cues for verbal irony: Hyperbole and the ironic tone of voice. Metaphor and Symbolic Activity, 10, 21–31. doi:10.1207/s15327868ms1001_3. Kumon-Nakamura, S., Glucksberg, S., & Brown, M. (1995). How about another piece of pie: The allusional pretense theory of discourse irony. Journal of Experimental Psychology: General, 124, 3–21. doi:10. 1037/0096-3445.124.1.3. Milosky, L. M., & Ford, J. A. (1997). The role of prosody in children’s inferences of ironic intent. Discourse Processes, 23, 47–61. doi:10.1080/01638539709544981. Nakassis, C., & Snedeker, J. (2002). Beyond sarcasm: Intonation and context as relational cues in children’s recognition of irony. Retrieved from https://software.rc.fas.harvard.edu/lds/wp-content/uploads/2012/ 04/Nakassis_Snedeker_2002.pdf, on July 26, 2013. Nittrouer, S., Studdert-Kennedy, M., & McGowan, R. S. (1989). The emergence of phonetic segments: Evidence from the spectral structure of fricative-vowel syllables spoken by children and adults. Journal of Speech and Hearing Research, 32, 120–132. Persicke, A., Tarbox, J., Ranick, J., & St. Clair, M. (2013). Teaching children with autism to detect and respond to sarcasm. Research in Autism Spectrum Disorders, 7, 193–198. doi:10.1016/j.rasd.2012.08.005. Pexman, P. M., Whalen, J. M., & Green, J. (2010). Understanding verbal irony: Clues from interpretation of direct and indirect remarks. Discourse Processes, 47, 237–261. doi:10.1080/01638530902959901. Ratcliff, R. (1993). Methods for dealing with reaction time outliers. Psychological Bulletin, 114, 510–532. doi:10.1037/0033-2909.114.3.510. Rockwell, P. M. (2007). Vocal features of conversational sarcasm: A comparison of methods. Journal of Psycholinguistic Research, 36, 361–369. doi:10.1007/s10936-006-9049-0. Rockwell, P. (2000). Lower, slower, louder: Vocal cues of sarcasm. Journal of Psycholinguistic Research, 29, 483–495. doi:10.1023/A:1005120109296. Sabbagh, M. A. (1999). Communicative intentions and language: Evidence from right-hemisphere damage and autism. Brain and Language, 70, 29–69. doi:10.1006/brln.1999.2139. Schneider, w, Eschman, A., & Zuccolotto, A. (2002). E-prime user’s guide: Version 1.0. Pittsburgh, PA: Psychology Software Tools. Schneider, W., Eschman, A., & Zuccolotto, A. (2002). E-prime user’s guide: Version 1.0. Pittsburgh, PA: Psychology Software Tools Inc. Utsumi, A. (2000). Verbal irony as implicit display of ironic environment: Distinguishing ironic utterances from nonirony. Journal of Pragmatics, 32, 1777–1806. doi:10.1016/S0378-2166(99)00116-2. Vannier, S. A., & Voyer, D. (2012). Sarcasm perception and quality of per and romantic relationships. Canadian Psychology, 53(2a), 176. Vermeulen, P. (2014). Context blindness in autism spectrum disorder: Not using the forest to see the trees as trees. Focus on autism and other developmental disabilities. 11. doi:10.1177/1088357614528799. Voyer, D., Bowes, A., & Techentin, C. (2008). On the perception of sarcasm in dichotic listening. Neuropsychology, 22, 390–399. doi:10.1037/0894-4105.22.3.390. Voyer, D., & Techentin, C. (2010). Subjective acoustic features of sarcasm. Metaphor and Symbol, 25, 1–16. doi:10.1080/10926488.2010.510927.

123

J Psycholinguist Res Walter, H., Adenzato, M., Ciaramidaro, A., Enrici, I., Pia, L., & Bara, B. G. (2007). Understanding intentions in social interaction: The role of the anterior paracingulate cortex. Journal of Cognitive Neuroscience, 16, 1854–1863. doi:10.1162/0898929042947838. Woodland, J., & Voyer, D. (2011). Context and intonation in the perception of sarcasm. Metaphor and Symbol, 26, 227–239. doi:10.1080/10926488.2011.583197.

123

Context, Contrast, and Tone of Voice in Auditory Sarcasm Perception.

Four experiments were conducted to investigate the interplay between context and tone of voice in the perception of sarcasm. These experiments emphasi...
272KB Sizes 5 Downloads 7 Views