Beyond mock politeness: sarcastic interrogatives as tools for face-threatening acts

1 Introduction

A key factor in the success of human communication is our ability to relate to others – addressing, responding, and anticipating their actions (Linell 2007). In everyday interactions, participants engage in communicative projects that promote joint actions and the development of co-representations (van der Wel et al. 2021). Politeness, understood as efforts to mitigate potential threats to individuals’ self-esteem, plays a crucial role in maintaining cooperative discourse (Brown and Levinson 1987).

Politeness has been extensively studied as a linguistic, psychological, and cultural phenomenon (e.g., Brown 2022; Spencer-Oatey and Kádár 2021). With the development of politeness theory, increasing attention has also been directed toward impoliteness strategies. As Culpeper (2005) notes, impoliteness is pervasive and can manifests in various forms. During interactions, participants may attack the addressee’s image and self-esteem by ignoring them, condescending, belittling, using sarcasm, and other and other similar strategies strategies.

Within impoliteness theory, sarcasm is viewed as a strategy in which an attack on the addressee’s face is achieved through insincere politeness (e.g., “ DO help yourselves, won’t you?” said to people greedily helping themselves). Consequently, sarcasm is often labeled as “mock politeness”. However, studies on the relationship between sarcasm and (im)politeness have predominantly focused on interactions characterized by social closeness between participants (e.g., Ghezzi and Molinelli 2019; Gibbs 2000; Jorgensen 1996; Taylor 2015).

In this paper, we aim to diverge from these valuable previous approaches while offering new insights into the relationship between sarcasm and impoliteness. Rather than contradicting them, we seek to demonstrate how sarcasm can also be used to attack the addressee’s face directly, in a way that aligns more with outright rudeness than just mock-politeness. While previous research recognizes sarcasm as a face-threatening strategy, it has predominantly positioned it as a form of mitigated impoliteness, i.e., mock politeness (Culpeper 1996; Taylor 2015). Our analysis reveals that sarcastic interrogatives often function closer to conventionalized and overt impoliteness, similar to direct insults, rather than as subtle implicational impoliteness that relies on incongruity. This perspective on sarcasm as an unmitigated face-aggravating strategy has been relatively underappreciated heretofore.

To investigate this phenomenon, we compare discursive genres that differ in medium, setting (physical vs. online discourse), and communicative purposes (Hymes 1974). To illustrate this point, consider example (1), drawn from an online debate forum.

Example (1) illustrates the use of an alternative question, i.e., a closed question that limits the answer to the explicitly mentioned options (Biezma and Rawlins 2015). In this case, the alternatives target the addressee’s positive face, specifically the desire to be valued by others (Brown and Levinson 1987). By presenting two options that frame the addressee as either a liar or a self-deluded individual, the writer performs a face-threatening act (FTA), delivering critique. This is an example of a sarcastic interrogative, where the expected situation, seeking an honest response, is inverted to convey sarcasm (Camp 2011).^[2]

This paper aims to explore in depth how the interplay between the surface forms of interrogatives and other-images can be used to sarcastically attack the addressee’s face. Specifically, two key questions are posed:

1. How are sarcastic interrogatives used as a face-aggravating strategy?

2. What contextual factors influence their use?

To answer these questions, we conduct both qualitative and quantitative analyses of tokens collected from various genres. As discussed in further detail later, we demonstrate that sarcastic interrogatives function as a face-aggravating strategy through other-positioning achieved via negative connotations, ad hoc categorization, and maxim-flouting. Additionally, we show that the use of sarcastic questions as an FTA varies across genres, and is influenced by several contextual features of the communicative event, such as social distance and the likelihood of joint accomplishment.

These findings enhance our understanding of the relationship between sarcasm and impoliteness in two key ways. First, the qualitative analysis shows that sarcastic utterances do not necessarily rely on insincere politeness strategies, contrary to previous research (e.g., Taylor 2015). Second, the quantitative analysis highlights the influence of discursive environments on the frequency of sarcasm, addressing a gap in the literature (see Hancock 2004).

While prior studies have examined sarcasm in institutional contexts, research specifically comparing sarcasm usage (distinct from irony) quantitatively across different discursive environments (e.g., spoken vs. Computer Mediated Communication [henceforth CMC]) remains scarce. To our knowledge, only Hancock (2004) and Aguert et al. (2016) directly compared CMC and face-to-face (henceforth FTF) interactions, yet both focused primarily on irony and sarcasm as one of its subsets, rather than as a separate phenomenon. Additionally, these studies are limited in terms of recency and participant demographics.^[3] These limitations further highlight the contribution of our approach: a comparative, cross-genre analysis using openly available datasets. This design enhances both the reproducibility of our findings and our understanding of how sarcasm functions across different communicative contexts.

The paper is structured as follows: Sections 2 and 3 present key information on sarcasm and impoliteness, respectively. Section 4 reviews the corpora and methodology used in the study. Section 5 examines how sarcastic questions function as a face-aggravating strategy, and analyzes the contextual factors that influence their use. Finally, Section 6 concludes the paper.

2 Sarcasm and (sarcastic) interrogatives

Sarcasm is a complex linguistic phenomenon that conveys contempt implicitly and humorously, while signaling the speaker’s dissociation from others’ opinions (Lozano-Palacio and Ruiz de Mendoza 2022; Savini and Caragea 2022). Within the framework of (im)politeness theory, it typically functions as a face-threatening act that operates through the contrast between apparent politeness and intended criticism, generally equated with “mock politeness” (Culpeper 1996; Taylor 2015).

Discussions of sarcasm frequently merge it with irony, as both serve as important communicative devices to express attitudes or evaluations toward the content of a message (Abrams 1999). As a result, the two terms are often used interchangeably, and are frequently assumed to be sufficiently similar to omit explicit differentiation (Clift 1999; Ling and Klinger 2016). Furthermore, sarcasm is commonly viewed as a specific case, or sub-set, of irony rather than as a distinct linguistic phenomenon (Filatova 2012), and has even been referred to as irony’s “evil twin” (Kreuz 2020: 9).

This tendency to conflate sarcasm with irony has not only blurred important distinctions between them, but also contributed to its diminished status in academic research. Historically, sarcasm has been treated as a mere subset of irony, which has contributed to its marginalization in both philosophical and linguistic research. While irony, famously linked to Socratic dialogue, has been valued for its philosophical and epistemic functions, sarcasm has retained a predominantly negative connotation (Lee and Katz 1998). Although recent decades have seen the emergence of sarcasm as a distinct object of inquiry, certain gaps persist in our understanding, especially regarding its use in naturally occurring dialogue (Oraby et al. 2016).

A particularly underexplored area within sarcasm research concerns sarcastic interrogatives. While sarcasm has been studied in assertions and, to a lesser extent, in rhetorical questions, interrogative forms have received minimal and inconsistent attention. They are often conflated with ironic and rhetorical questions, and described using varying terminology, which obscures their distinctive function. This oversight is especially striking given the frequency of sarcastic interrogatives in oral traditions, journalism, and popular culture (Doyle 2008). However, to explore sarcastic interrogatives meaningfully, we must first establish a clear working definition of sarcasm.

In a pioneering approach, Camp (2011) provided one of the first comprehensive definitions of sarcasm as an independent phenomenon. According to her, sarcasm is a scalar meaning inversion that hinges on the transposition of pretended commitment. Specifically, sarcasm involves evoking a normative scale while feigning commitment to it, only to invert this commitment in interpretation. While the speaker appears to make an assertion, they are actually setting an evaluative standard and implying that it has been negatively violated. Let us illustrate this with the following example, taken from our own corpus research:

This sarcastic alternative interrogative exemplifies scalar meaning inversion by establishing a normative scale in which an informed and truthful response is expected. A genuine alternative question would present a balanced choice, allowing for the possibility that the addressee is correct or acting in good faith. However, in this case, the speaker feigns neutrality while framing both options negatively, implying that the only possible explanations for the addressee’s statement are ignorance (lack of knowledge) or deception (intentional misrepresentation).

The sarcasm arises from the forced commitment to this normative scale, where informed discourse is the expected standard, while simultaneously inverting this commitment by presupposing its violation. This demonstrates how sarcasm manipulates meaning not by simple opposition, like verbal irony, but by distorting the expected neutrality of discourse through a forced evaluative contrast.

Although Camp’s (2011) meaning-inversion model does not explicitly address impoliteness, it clearly predicts this phenomenon, which is central to definitions of sarcasm as a form of mock politeness. Camp’s framework presents four components of a normative scale – presupposition, pretense, inversion, and evaluative attitude. These elements provide a theoretical apparatus that naturally accounts for mock politeness dynamics.

This is particularly evident in her analysis of illocutionary sarcasm, where she demonstrates how utterances like “Thanks for holding the door” operate on politeness scales: the speaker “pretends to undertake an utterance which would be appropriate if the addressee had held the door, where door-holding ranks high on a scale of politeness” and thereby “communicates the speaker’s evaluation of the addressee’s actual behavior as rude” (Camp 2011: 618).

The strategic value of sarcasm for impoliteness is further reinforced by Camp’s observation that sarcasm enables speakers to “communicate rhetorically volatile negative attitudes while preserving deniability” (Camp 2011: 609), a characteristic that makes it particularly useful in antagonistic contexts. Indeed, her characterization of sarcasm as a “communicative bluff” that creates vulnerability to retaliation explains why sarcasm is frequently deployed in contexts of “intense social anxiety” (Camp 2011: 605) and interpersonal conflict, where face management and social positioning are paramount concerns.

Building on Camp’s (2011) definition, Michaelis and Feng (2015) offer an elaboration using a similar approach. According to them, sarcastic utterances rely on the widely understood assumption that the speaker’s intended meaning diverges substantially from their stated words, rather than merely representing the direct opposite. They describe sarcasm as an indirect rebuke, prompting the listener to reconstruct both a scale and a presupposition. This can be illustrated by the following adapted example from Michaelis and Feng (2015: 150):

The scale reconstructed by the interpreter is based on an ordering of injuries by severity. The reconstructed presupposition concerns the circumstances in which one should worry about an injured person’s death (likely not when they have a paper cut). In this example, the sarcastic utterance establishes the necessary evaluative scale, and signals that the situation is the opposite of one where a sincere utterance would be appropriate. Sarcasm, therefore, highlights the disparity between the current situation and the circumstances that would justify a sincere speech act (Michaelis and Feng 2015). The speaker appears to make an assertion, but implicitly sets an evaluative standard, suggesting that it has been negatively violated (Camp 2011).

Some constructions are particularly sensitive to sarcasm due to elements that readily lend themselves to meaning inversion, the core mechanism of sarcasm. One such construction is interrogatives, which form the focus of our current research. In the following subsection, we elaborate on the different types of interrogatives examined in this study.

3 Impoliteness and facework

Research on language in use demonstrates that everyday discourse primarily reveals speakers’ attitudes, stances, and evaluations (Thompson and Hopper 2001). In this context, interlocutors are not merely exchanging factual information, but are also expressing assessments and (dis)aligning themselves with others (Du Bois 2007). This suggests that interactions frequently involve stance-taking (Du Bois and Kärkkäinen 2012), where stance is viewed as a three-way process. In this process, speakers evaluate an object, position themselves in relation to it, and (dis)align with co-present others (Du Bois 2007).

When multiple viewpoints are introduced in discourse, participants often communicate their evaluations of a situation, positioning themselves within a particular social or moral framework (Graumann and Kallmeyer 2002). In doing so, interlocutors frequently adopt specific roles, presenting their “personhood” in relation to others through characters or stereotypes, such as mother–child or nice–crude dynamics (Wells 2015).

Positioning, defined as “the discursive production of selves” (Davies and Harré 1990), is closely tied to the concepts of image and face. In the literature, face refers to the public self-image that community members seek to project (Brown and Levinson 1987). Since this image can be maintained, enhanced, or damaged during interactions, it is understood as existing within the dynamic flow of discourse rather than being an inherent attribute of individuals (Holtgraves 2002).

For instance, Brown and Levinson (1987) emphasize that certain communicative actions, such as criticism or commands, may threaten the addressee’s positive or negative face. Positive face refers to the desire to be appreciated by others, while negative face concerns the need to act without imposition. In such cases, speakers may mitigate the threat by using strategies like hedging devices or indirect requests. Consequently, interlocutors may soften face threats, enhance their own positive face through (indirect) self-praise, or attack the face of their addressees to achieve specific communicative goals (Culpeper 2016; Speer 2012). These processes are called facework.

While politeness is seen as a strategy to reduce threats to the addressee’s face, impoliteness may be understood as a face attack. For impoliteness to occur, the speaker must deliberately convey a face attack, or the hearer must perceive the communication as a face attack (Culpeper 2005). Culpeper (2005) outlines several types of impoliteness strategies, such as direct and explicit face attacks, attacks aimed at damaging the addressee’s positive face (e.g., showing a lack of sympathy), and attacks targeting the addressee’s negative face (e.g., expressing scorn).

Recently, Culpeper (2016) has suggested that impolite utterances are those that go challenged. This challenge can be expressed in discourse through verbal responses indicating the hearer’s perception of rudeness, as well as non-verbal reactions, such as facial expressions of hurt, or counter-impolite utterances. Culpeper (2005) also argues that directness is often associated with impoliteness, but this connection can be problematic, as it tends to rely on the researcher’s intuition. To address this, Culpeper (2011) proposed a model of impolite utterances grounded in empirical research.

In this model, impolite utterances can take two forms: conventionalized formulae, such as insults, and implicational expressions. Implicational expressions are utterances that exhibit a mismatch between the context implied by one part of the turn (e.g., a compliment) and the context implied by another part (e.g., a bored tone of voice). Sarcasm, in this sense, is seen as an incongruity between context, content, and prosody (Nakamura et al. 2022).

This distinction between conventionalized formulae and implicational impoliteness is critical for our understanding of sarcasm’s role in face-threatening acts. Following Dynel (2015), impoliteness can be categorized along a spectrum from mitigated to unmitigated forms. While existing literature has predominantly positioned sarcasm within the domain of mitigated impoliteness, specifically as mock politeness that relies on the incongruity (Nakamura et al. 2022), our analysis reveals a different pattern.

As will be shown in upcoming sections, the sarcastic interrogatives in our data, particularly in online political debate, frequently function as unmitigated impoliteness, bypassing the pretense of politeness entirely. Rather than operating through implicit contrast between surface politeness and intended criticism, they directly position the addressee negatively through explicit linguistic choices. This positioning of sarcasm closer to conventionalized impoliteness formulae rather than implicational strategies represents a departure from how sarcasm has traditionally been conceptualized in impoliteness theory.

Research on the relationship between sarcasm and (im)politeness has often emphasized sarcasm’s connection to politeness strategies, particularly highlighting the incongruity between context and content. For instance, Jorgensen (1996) suggests that sarcasm can be used to critique or complain about minor issues among close acquaintances as a face-saving mechanism. She argues that sarcasm allows the speaker to soften face threats, enabling them to appear less aggressive or unfair. This mitigation helps maintain the speaker’s positive self-image and decreases the likelihood of negative reactions from others. In these contexts, sarcasm is not intended to amuse, but to temper the criticism, thereby avoiding the perception of the speaker as thoughtless or insulting.

However, other studies have suggested that sarcasm functions as an impoliteness strategy, achieved through the use of insincere, or mock, politeness (Bousfield 2008; Culpeper 2005). In these cases, sarcastic utterances may seem to support or enhance the recipient’s face, but in reality, they significantly undermine and damage it.

In the following sections, we demonstrate how sarcastic interrogatives perform FTAs through other-positioning by analyzing features of syntax, semantic transparency, and pragmatic inference. This shifts the focus of previous studies, which have emphasized the relationship between sarcasm and politeness, to explore the interplay between positioning, impoliteness, and offense.

4 Corpora and method

The analysis of sarcastic interrogatives in this study is based on data from three American-English corpora: the COCA corpus, the NLDS corpus, and the Switchboard corpus. These corpora vary in terms of contextual features, including the medium, setting, and the participants’ communicative purposes (Hymes 1974).

The COCA (Corpus of Contemporary American English) corpus contains over one billion words across eight genres: spoken, fiction, popular magazines, newspapers, academic texts, TV and movie subtitles, blogs, and other web pages. For this study, we analyzed the sub-corpora of spoken, non-scripted language from the two most recent years available – 2011 and 2012, comprising a total of 7,430,781 tokens.

The NLDS (Natural Language and Dialogue Systems) corpus includes several main categories, such as online debate forums, human-generated summaries, blogs, and casual conversations. For this study, we focused on the Internet Argument Corpus (IAC) version 2, which is a collection of corpora designed for research on political debate in internet forums (Abbott et al. 2016). The IAC consists of three datasets: four forums (414,000 posts), Convince Me (65,000 posts), and a sample from Create Debate (3,000 posts).

The Switchboard corpus is a large, multi-speaker dataset of conversational speech, collected between 1990 and 1991 (Godfrey et al. 1992). It contains approximately 2,400 two-sided telephone conversations between 543 speakers of American English (302 male, 241 female) from across the United States. All speakers were complete strangers, connected by a robotic operator and prompted to select discussion topics from a predetermined list of about 70 options, such as cars, recycling, savings accounts, and newspapers (Calhoun et al. 2010). To ensure variation, no two speakers conversed more than once, and no one discussed the same topic more than once. The conversations range from one and a half to 10 minutes, with an average duration of six and a half minutes. In total, the corpus contains roughly three million words, amounting to approximately 260 h of recorded speech (Calhoun et al. 2010).

Based on these corpora, we collected interrogative-tokens and manually annotated them as either literal or non-literal, focusing on meaning inversion. An interrogative was classified as sarcastic if it aligned with our primary definition, which treats sarcasm as a form of scalar meaning inversion (Camp 2011). In such cases, the speaker appears to commit to a proposition while implicitly signaling a negative evaluation of that commitment, thereby implying that an expected standard has been violated.

In such situations, the speaker exploited a shared assumption that the literal interpretation was implausible, thereby implying indirect criticism (Michaelis and Feng 2015). Sarcasm was further identified by features such as implicit contempt, ridicule, and the speaker’s disassociation from others’ views (Lozano-Palacio and Ruiz de Mendoza 2022; Savini and Caragea 2022).

Here are three examples illustrating the application of this annotation criterion:

This is a sincere question, in which the speaker is seeking information in an honest way. There is no meaning inversion, no criticism, and no ridicule. Therefore, this question was annotated as a literal, non-sarcastic question.

This is not a sincere question. The speaker is not seeking for information or waiting for an answer (which is evident by the fact that they answer themselves immediately). Instead, they are trying to prove their point by asking a question to which the answer is obvious, thus making a compelling and persuasive interrogative statement (Frank 1990). Therefore, it is not a literal question. However, there is no meaning inversion, and no implicit and humorous conveyance of contempt and ridicule. As a result, it cannot be considered a sarcastic interrogative either. Consequently, it was annotated as a rhetorical,^[4] non-sarcastic question.

This is also not a sincere question. It is not posed with the intention of eliciting a genuine response. Instead, it serves to deliver a pointed critique through meaning inversion and the pretense of seeking for information. In reality, the asker employs an evaluative standard, implying it has been breached negatively.

As noted by Michaelis and Feng (2015), sarcasm functions as an indirect rebuke, prompting the interpreter to establish a scale and presuppose certain truths. In this example, the rebuke targets the addressee’s reliance on misconceptions rather than facts. The scale that the interpreter reconstructs relates to proper argumentation practices – specifically, how arguments should be based on facts to be persuasive. The presupposition challenges the conditions under which one should find an argument convincing, namely when it is grounded in facts rather than fallacies. This interrogative sets the parameters for evaluating the discourse while highlighting the disparity between the current situation and one that would call for a literal utterance (Michaelis and Feng 2015), while conveying ridicule, humor, and criticism. This analysis led to this token being annotated as a sarcastic interrogative.

To ensure the accuracy of our sarcasm annotation process, we conducted an external judgment test. In this test, we presented a sample of utterances to external judges, asking them to determine whether they perceived them as sarcastic or literal. Our test comprised a total of 72 questions. This number was selected for several reasons.

First, this subset constituted approximately 6.5 % of the total utterances collected for the study, providing a representative and manageable sample size. Second, this quantity allowed external judges, particularly laypersons who were not academics, to assess the data without experiencing excessive fatigue or confusion.

Among these utterances, 36 were annotated by us as sarcastic interrogatives, while the remaining 36 were annotated as literal questions. We provided the judges with a concise document outlining our definition of sarcasm in layperson’s terms, along with several examples. The full explanation and the table of utterances used in the test are available in the appendix.

Our judges were divided into two distinct groups: an expert group and a non-expert group, comprising of native speakers. The expert group consisted of two female linguists who had previously conducted research on the topic of sarcasm and published several articles in the field. Conversely, the non-expert group comprised three native speakers of American English, including two males and one female, ages ranging from 32 to 42 years old. All non-expert participants were monolingual speakers of American English, right-hand dominant individuals, and had no diagnosed cognitive disorders or learning disabilities.

All participants received an identical short explanation demonstrating the distinction between a literal, information-seeking question, and non-literal, sarcastic interrogatives. They were instructed to review the 72 utterances and simply indicate whether they perceived each question as literal or sarcastic, without providing any further explanation for their choice.

To illustrate, consider the following two examples – one to which all five judges (both experts and native speakers) agreed with our classification as sarcastic, and one to which all agreed to our classification as literal.

After collecting all responses, we conducted a Fleiss’ kappa test to measure the agreement between our categorization of sarcasm and that of the external judges. Fleiss’ kappa assesses the extent of agreement among multiple raters beyond what would be expected by chance alone. Specifically, observed agreement was calculated as the proportion of consistent judgments averaged across all items and raters, while expected agreement represents the level of consensus predicted by chance.

For each judgment, the proportion of times each category (agreed or disagreed) was used by the raters was calculated. These proportions were then averaged across all items to obtain the average proportion of each category. The expected agreement was computed as the square of this average proportion.

In our study, the expected agreement was 0.25, reflecting the level of agreement expected by random chance. Fleiss’ kappa was calculated to be over 0.72, indicating a substantial level of agreement among the participants beyond what would be expected by chance alone.

In addition to inter-rater agreement, we examined discrepancies between our judgments and the external raters. 29.17 % of utterances were classified differently by at least one external judge compared to our original classification. To ensure the accuracy of the final analysis, we excluded these utterances from further consideration. Despite these differences, the overall high agreement supports the robustness of our annotation process.

To conclude, we believe that the sarcasm annotations in our study, validated by external judges (comprising both experts in the field of sarcasm and lay native speakers) demonstrated significant concordance as confirmed by Fleiss’ kappa. This substantial agreement among the raters reinforces the validity of our analytical approach and the accuracy of our results in identifying sarcastic content.

Following this, we performed a text-based analysis of non-literal questions to illustrate how certain interrogatives can be sarcastically used to attack the addressee’s face. We examined selected tokens that contained stretches of co-text from several perspectives: facework, other-positioning, presuppositions, and their co-occurrence with linguistic and extralinguistic cues. To better understand the contextual features influencing the use of sarcastic interrogatives as a face-aggravating strategy, we conducted a sarcasm distribution analysis using a z-test.

5 The interplay between sarcastic interrogatives and face aggravation

In this section, we present both quantitative and qualitative analyses of our data. First, we examine the distribution of sarcastic interrogatives across different corpora, highlighting how contextual features of the communicative event influence their frequency (Section 5.1). Then, we demonstrate how speakers and writers use these sarcastic interrogatives to perform FTAs, providing detailed analysis of specific patterns and their relative prevalence in our data (Section 5.2).

Before presenting the study’s findings, it is important to discuss the category of non-literal questions. As mentioned earlier, they can exhibit meaning inversion, allowing speakers to fulfill various communicative functions, such as expressing surprise, conveying disagreement, or persuasion. This study focuses specifically on a subset of non-literal questions – sarcastic interrogatives.

Sarcastic interrogatives are particularly likely to occur in contexts of disagreement, as they are used to express the speaker’s dissociation from others’ opinions (Lozano-Palacio and Ruiz de Mendoza 2022). However, not all questions that appear in disagreement contexts are necessarily sarcastic interrogatives. To illustrate this distinction, consider the following examples:

Both interrogatives in example (9) are non-literal within a conflictual context. However, the interrogative in (9a) allows the speaker to express an assertion based on information shared by the participants (Biezma and Rawlins 2017), specifically, “I, the mother, have cared for you your whole life”. In contrast, the interrogative in (9b) introduces new information into the discourse, which is not necessarily shared by the interlocutors. This distinction suggests that non-literal interrogatives can draw on common ground to reaffirm accessible information for various communicative purposes (Rohde 2006). In the following subsection, we examine how sarcastic interrogatives leverage discourse-new information to convey dissociation and criticism.

5.1 Quantitative analysis

Numerous studies on language-use highlight how contextual factors, such as the medium and setting of interactions, shape speakers’ linguistic choices (Gregory and Carroll 2018). Building on these insights, this subsection explores whether the use of sarcastic interrogatives as a face-aggravating strategy varies across different discursive environments.

As mentioned in Section 4, this examination was conducted across three corpora: the non-scripted, spoken TV corpus (COCA); the spoken telephone corpus (Switchboard); and the written web-discourse corpus (NLDS). Each corpus differs in the medium (spoken/written), the setting (telephone/FTF/CMC), and the communicative purposes of the participants (deliver information/entertain/debate).

In our analysis, a total of 787 questions from the COCA corpus were examined, among which 37 were identified as sarcastic (4.7 %). The NLDS corpus, on the other hand, exhibited a significantly higher incidence of sarcasm: 121 out of 301 questions analyzed (40.19 %) were sarcastic. Notably, the Switchboard corpus contained no instances of sarcasm (0 %). These results are illustrated in Figure 1.

Figure 1:

Sarcasm distribution across corpora.

Statistical analysis reveals that the proportion of sarcastic interrogatives is significantly higher in the CMC corpus compared to the FTF TV corpus (z = −14.87, p < 0.001). This result highlights substantial differences in the prevalence of sarcasm between spoken, FTF, interactions and written, web-based discourse.

This finding may be attributed to the differing expectations that each type of interaction elicits from participants. Locher (2013) emphasizes that interlocutors enter communicative events with expectations shaped by their understanding of the event’s social frame. Similarly, Locher and Watts (2008) argue that social frames originate from cognitive conceptualizations of behavior, formed through individuals’ experiences in social practice. These frames establish norms and expectations that evolve over time and are subject to change. According to Locher and Watts (2008), these social frames are essential for analyzing interactions, as any judgment of (im)politeness is always assessed against the specific frame of expectation for the given communicative context.

From this perspective, each corpus represents a distinct communicative frame, differing in two key aspects: social distance and the likelihood of joint accomplishment. Regarding social distance, the three corpora vary in the familiarity between participants. The Switchboard corpus, for instance, represents an artificial discursive environment where strangers, connected through an automated system, are required to select discussion topics from a predetermined list. As a result, the participants have no prior familiarity with one another.

Previous empirical research has shown that politeness tends to increase as familiarity decreases (Holtgraves and Yang 1992; Stephan et al. 2010). Additionally, sarcasm was shown to be more frequently used in close relationships compared to distant ones (Zhu and Wang 2020). This pattern may account for the complete absence of sarcastic interrogatives in the Switchboard corpus, where interactions involve socially distant participants.

Unlike the Switchboard corpus, the NLDS corpus represents a social frame characterized by ongoing engagement between participants in an online forum. Here, interlocutors are familiar with one another, regularly interacting within the same forum. In contrast, the COCA sub-corpus studied represents a discourse type that includes both high and low familiarity.

On one hand, some participants, such as panel members of talk shows and morning programs, interact regularly and develop familiarity over time. On the other hand, interactions between hosts and one-time interviewees involve participants who have never met before and are unlikely to meet again. While this discourse type includes elements of social distance, such as between interviewees and hosts, it also evokes expectations of familiarity in certain communicative settings. This familiarity can facilitate the use of sarcasm, as shared common ground makes the underlying counterfactual attitude more transparent to addressees (Kreuz 1996).

However, social distance alone cannot fully account for our findings, particularly the significantly higher proportion of sarcastic interrogatives in the NLDS corpus compared to the COCA corpus. To address this discrepancy, it is important to consider a key difference in communicative expectations between the two corpora: the likelihood of joint accomplishment.

Linell (1998) highlights that a communicative project, which aims to achieve social goals, inherently involves the participation of others. In such interactions, participants may share communicative purposes, working together toward the joint accomplishment of actions. Alternatively, they may have competing goals, each pursuing their own objectives, as is often the case in arguments.

In the NLDS corpus, interlocutors engage in debates over political and religious issues, adopting different stances and seeking to justify their positions against those of others. As a result, their likelihood of joint accomplishment is diminished. Additionally, these debates take place anonymously, which influences the online disinhibition effect, characterized by a reduced sense of restraint when communicating online (Suler 2004). This anonymity fosters an environment where participants are more inclined to use sarcastic interrogatives as a face-aggravating strategy, particularly within a discourse that features competitive, unidentified interactions.

In contrast, in the COCA TV-corpus, cooperation among participants is essential for the success of talk shows and morning programs. In this context, interlocutors share a communicative goal of entertaining the audience through emotional and relatable stories that promote a sense of solidarity. As a result, the use of sarcasm to undermine or attack the addressees may be less prevalent, as it could hinder the joint accomplishment of these shared objectives.

In this subsection, we have argued that contextual factors, such as social distance and the likelihood of joint accomplishment, may influence the use of sarcastic interrogatives as a face-aggravating strategy. Therefore, our findings shed light on how communication modalities and discursive environments shape the frequency of sarcasm.

6 Conclusions

This paper aimed to explore how speakers and writers use sarcastic interrogatives as a face-aggravating strategy and the contextual factors that may influence this usage. Through quantitative and qualitative analyses of data from three open-source corpora, which vary in medium, setting, and participants’ communicative purposes, several key conclusions have been drawn.

First, the data indicates that sarcastic interrogatives can serve as a tool for directly attacking the addressee’s face, which differs from previous conceptualizations of sarcasm as primarily a form of mock politeness. While traditional views position sarcasm as a mitigated impoliteness strategy that operates through the contrast between apparent politeness and intended criticism, our findings reveal that sarcastic interrogatives often function as unmitigated, conventionalized face-threatening acts closer to direct insults, entirely bypassing the pretense of politeness.

This is achieved through an interplay of semantic transparency, direct reference to targets, pragmatic inferencing, and the syntactic structure of the interrogatives themselves. Specifically, speakers and writers may incorporate affect-marked lexemes with negative connotations, scalar values, or presuppositions that flout the Maxim of Quality into the open slots of interrogatives. This strategy presents a negative image of the addressee, thereby conveying ridicule and criticism. Our findings highlight how sarcastic interrogatives can perform FTAs through other-positioning, rather than relying on insincere politeness strategies. In doing so, this study contributes to the growing body of literature on sarcasm and (im)politeness by uncovering an understudied aspect of sarcasm’s use.

Second, the data reveals that sarcastic interrogatives as a face-aggravating strategy are more prevalent in written, web-based, argumentative discourse than in spoken, unscripted TV interactions. We attribute this finding to the differing contextual factors that characterize each type of discourse: social distance and the likelihood of joint accomplishment. While previous research has emphasized sarcasm as a reflection of content-context incongruity, our study sheds light on another dimension of the relationship between sarcasm and context – specifically, how sarcasm usage depends on the discursive environment. Future research could expand on this by examining other types of discourse that display varying degrees of additional contextual features, such as agency and authority.