Using Sarcasm to Compliment: Context, Intonation, and the Perception of Statements with a Negative Literal Meaning

Abstract

The present study extended findings of contrast effects in an auditory sarcasm perception task manipulating context and tone of voice. In contrast to previous research that had used sarcastic and sincere statements with a positive literal meaning, the present experiment examined how statements with a negative literal meaning would affect the results. Eighty-four undergraduate students completed a task in which an ambiguous, positive, or negative computer-generated context spoken in a flat emotional tone was followed by a statement with a negative literal meaning spoken in a sincere or sarcastic tone of voice. Results for both the proportion of sarcastic responses and response time showed a significant context by tone interaction, reflecting relatively fast sarcastic responses for the situation in which sarcasm would turn the statement into a compliment (positive context, sarcastic intonation) and fast sincere responses when the literal insult was emphasized (negative context, sincere intonation). However, the ambiguous context produced a pattern of results modulated by the tone of voice that was similar to that observed when the context/intonation pairing could not be interpreted as a compliment or an insult (negative context/sarcastic intonation or positive context/sincere intonation). These findings add to the body of literature suggesting that situational contrast, context, and intonation influence how sarcasm is perceived while demonstrating the importance of the literal meaning in sarcasm perception. They can be interpreted in the context of models of sarcasm comprehension that postulate two stages of processing.

A novel task to evaluate irony comprehension and its essential elements in Spanish speakers

“…In terms of the acoustic characteristics of ironic prosody, we found that classified as ironic stimuli had a slower speech rate than literal and unrelated statements, in agreement with previous studies ( Rockwell, 2001 ; Cheang and Pell, 2009 ; Bryant, 2010 ; Li et al, 2013 ; Peters et al, 2016 ; Voyer and Vu, 2016 ; Deliens et al, 2018 ). Ironic stimuli had higher median intensity and higher fundamental frequency (F0) (median and range) than unrelated statements only; others have found that ironic stimuli have greater intensity ( Rockwell, 2001 ; Li et al, 2013 ; Peters et al, 2016 ; Deliens et al, 2018 ).…”

“…Although the prosody of ironic statements had a lower fundamental frequency (F0) than literal statements, this difference was not statistically significant as expected based on the literature ( Rockwell, 2001 ; Li et al, 2013 ; Peters et al, 2016 ; Deliens et al, 2018 ). One possibility is that the specific acoustic conventions to express irony may differ between languages, as has been suggested ( Rockwell, 2001 ; Cheang and Pell, 2009 ; Bryant, 2010 ; Li et al, 2013 ; Peters et al, 2016 ; Voyer and Vu, 2016 ; Deliens et al, 2018 ). Another is that the acoustic correlates of ironic prosody are not intrinsic but relative to the enfolding discourse ( Bryant, 2010 ).…”

“…Other cues that can facilitate the identification of irony are prosody ( Wang et al, 2006 ) and facial expression ( Akimoto et al, 2014 ). The acoustic parameters associated with prosody in irony are lower fundamental frequency (F0; Rockwell, 2001 ; Peters et al, 2016 ), changes in F0 ( Milosky and Ford, 1997 ; Cheang and Pell, 2009 ; Bryant, 2010 ; Li et al, 2013 ; Deliens et al, 2017 ; Rivière et al, 2018 ), greater intensity ( Rockwell, 2001 ; Li et al, 2013 ; Peters et al, 2016 ; Deliens et al, 2017 ), and slower speech rate ( Rockwell, 2001 ; Cheang and Pell, 2009 ; Bryant, 2010 ; Li et al, 2013 ; Peters et al, 2016 ; Voyer and Vu, 2016 ; Deliens et al, 2017 ). The facial information that supports ironic comprehension includes smiling, raised eyebrows, eye-rolling, winking, and squinting eyes ( Rockwell, 2001 ; Attardo et al, 2003 ; Caucci and Kreuz, 2012 ).…”

Multi-Rule Based Ensemble Feature Selection Model for Sarcasm Type Detection in Twitter

“…A small dataset of only 2000 tweets was used; hence, Naïve Bayes worked well. Voyer and Vu [24] developed a model to experiment how negative literal statements would affect sarcasm detection in auditory perception. e approach achieved 80% accuracy in sarcasm identification.…”

“…e polarity and sentiment score for these tweets are analyzed using TextBlob and NLTK. e change in mood 0 10 20 30 40 50 60 70 80 90 100 Accuracy (%) State-of-the-art approaches Mukherjee et al [25] Mazen et al [24] Jihen et al [21] Jayasanka et al [56] Rajadesingan et al [34] Proposed work Figure 9: Comparison of the proposed system with existing approaches. Figure 10: Sarcastic type classification based on proposed multi-rule approach.…”