Face-work in North American Presidential Debates: A Corpus-based Multi-theoretical Approach

image restoration manifestations. The selected corpus comprises the presidential debates of seven North American elections —one per decade—, providing an overview of the evolution of both phenomena. This methodology also allows insight into the frequency of specific strategies. The analysis adopts a multi-theoretical approach to demonstrate the effectiveness of integrating Impoliteness theory (Culpeper, 1996) and Image restoration theory (Benoit and Wells, 1996) to create a comprehensive new approach to the study of face-work in presidential debates. Supported by results from Pearson’s Chi-square test, the combination of these theories was shown to be possible. The recurrence of attack-defense pairings integrating strategies from both theories also pointed to their necessary combination. While the recent incorporation of a female presidential candidate hinted at the potential gender variation of face-work in the debates, further data is needed to obtain conclusive results.


Introduction
Presidential campaigns have long received significant media attention and they maintain their status as 'newsworthy' (Benoit 2015, 63) to this day.This results in the widespread recognition of the spectacle-like nature inherent to some campaign events.Assuming their role as candidates, politicians then become performers to some extent (Atkinson 1984, 178), and they seek media attention to foster a connection with their potential voters (Hinck et al. 2021, 1).This dynamic sets expectations regarding how they must behave to prove worthy.Given the polarized nature of political discourse, particularly evident in contexts such as presidential debates, candidates should endeavor to seem 'preferable to their opponents' (Benoit and Glantz 2020, 9) by exploiting their differences through face-threatening behavior that undermines 'the target's face, image, or reputation' (Benoit 2007, 321).
Although the analysis of political discourse from an impoliteness perspective has attracted considerable interest over time, the mechanisms used by politicians in debates to salvage the face loss that they have incurred are still underresearched.This study strives to offer an approach for the combined analysis of attacks on face and subsequent restorative efforts through the combination of Impoliteness theory and Image restoration theory.
This paper is structured as follows: Section 2 seeks to characterize the object of the study by defining the face of politicians and the notion of presidential character, as well as the communicative setting at the core of this paper.Section 3 presents a literature review on impoliteness and image restoration in presidential debates.Section 4 states the research objective and the five hypotheses to be tested through the methodology also outlined in that section.Section 5 contains the findings and discussion, including the evolution of face-work in the corpus, the most common strategies, and some remarks on gender variation.Lastly, Section 6 serves as the conclusion.

Some Remarks on the Face of Politicians and Presidential Debates as a Platform for Face-work
Originating in Chinese culture, the concept of face has an inherent social nature that offers a new perspective to analyze interaction and, more specifically, how some aspects of identity are constructed-and co-constructed-through 'facework' (Goffman 1955, 216).In this paper, I utilize the traditional differentiation between positive and negative face, initially adapted to Western cultures by Goffman (1955) and adopted by Brown and Levinson (1987) in their Politeness theory.This dichotomy explains positive face as the part concerned with societal Face-work in North American Presidential Debates: A Corpus-based Multi-theoretical... Issue 41, 2024, pages 53-73 perception, while negative face pertains to our desire to be free to act without undue constraints imposed by others (Brown and Levinson 1987, 61-62).

Alicante Journal of English Studies,
When politicians engage in face-work, they tend to prioritize positive face (Harris 2001, 469;Bull and Fetzer 2010, 161) given its ties with other key aspects for those in politics such as reputation (Culpeper 2011, 24), desirability (Brown and Levinson 1987, 62) or honor (Spencer-Oatey 2002, 540).The significance of positive face in politics is underscored by Gruber's introduction of 'public positive face (PPF)' (Gruber 1993, 3), a specific kind of positive face shared by politicians which represents their need to be judged as 'a rational, trustworthy person whose political ideas and actions are better fitted to the wants and demands of the general public than his opponents' (Gruber 1993, 3).However, we must note that negative face does not go unnoticed in political discourse-especially as it is closely intertwined with positive face (Bargiela-Chiappini 2003, 461) given the fluctuating delimitations of both aspects.Culpeper et al. (2003Culpeper et al. ( , 1567) pointed out it is not rare for threats deployed against one to affect the other.
Regarding the type of face-work politicians engage in, scholars often highlight impoliteness as a defining trait of their discourse (Fuentes Rodríguez 2016, 172).For instance, Hinck et al. (2021) noted that the ability to respond to criticism with criticism was an exhibit of 'presidential character' (Hinck et al. 2021, 2).While presidential character is a significantly complex concept, charisma (Kraus 1999, 206;Chilton 2004, 46) and persuasiveness (Fetzer 2002, 181) appear to be fixed features.Candidates are also expected to demonstrate their presidential character by 'denying the legitimacy of others' (Bentivegna and Rega 2022, 2).When done via impoliteness, Alcaide Lara (2014, 250) mentioned the necessity to pair it with argumentation so as to align with the voters' expectations.
Since their inception in 1960-with Senator J.F. Kennedy and Vice President Richard Nixon as the presidential candidates-and especially since they became standard practice in 1976, televised presidential debates in the U.S. have become a major platform for political interaction.Candidates engage in an earnest fight for their face, seeking to disparage the opponent's (Blas Arroyo 2003, 398).Even before 2008, when direct interaction between candidates was first permitted (Marín 2020, 177), there was still confrontation between candidates, mainly during the 2-minute rebuttal turns from the original format.Since 2012, debates have adopted an issue-centered structure divided into 15-minute segments, but the preference for monological interventions (Albalat 2020, 75) still prevails.This influences the deployment of attacks against the opponents, for they will mainly be carried out indirectly, albeit not implicitly.Moreover, it is also a testament to the importance of the audience, to whom all messages are ultimately addressed in the debates (Alcaide Lara 2014, 251;Benoit and Glantz 2020, 9).

Impoliteness and Face-threatening Acts
Impoliteness theory challenges the premise established by Brown and Levinson (1987) by suggesting that face-threatening acts (FTAs) can be performed deliberately to cause face loss to the recipient (Culpeper 1996, 350).Given the polarized discourse of presidential debates-characterized by the Us vs. Them dichotomy (Atkinson 1984, 39;van Dijk 1997, 28)-they have been categorized as events where 'impoliteness is licensed and rewarded' (Culpeper, 2011, 216), and FTAs are 'a staple' (Bortoluzzi and Semino 2016, 16) of communication in this setting.Despite the seemingly inherent negative connotations of impoliteness, it has been argued that conflict can sometimes be necessary for democratic progress, especially in the form of 'emotionally marked criticism of the past and future actions of public persons' (Tracy 2008, 170).
Therefore, presidential debates are inherently confrontational and unravel through a succession of FTAs and defensive strategies through which candidates vie to enhance their face 'at the expense of their opponents' face' (Beck, 1996, 167), as clearly exemplified by unfavorable comparisons between 'Us' and 'Them'.This perpetuates a constant struggle for face, driven by the need to emerge victorious from this 'zero-sum game' (García-Pastor 2008, 104) where only one candidate can win, and during which their performance will be judged by the electorate who ultimately determines the outcome of the election (Blas Arroyo 2001, 33).FTAs, however, serve purposes beyond mere threat in presidential debates, as they also wield a persuasive influence over the audience-the voters-to whom candidates aim to appear as the preferable option (Benoit and Glantz 2020, 9;Benoit 2007b, 33).They also hold entertainment value, due to the 'voyeuristic pleasure' (Culpeper 2005, 45) some derive from watching conflict unfold.This aspect has contributed to the debates' second nature as 'mass media spectacle[s]' (Blas Arroyo 2003, 417) where there is little pretense of cooperation between candidates, and instead focus on the audience's judgment (Vinther 2011, 573).
The imperative to hold both the audience and the camera's attention compels politicians to display higher levels of impoliteness (Hinck et al. 2021, 5, 15) in an attempt to hold the floor for longer.Interruptions are often used (Vinther 2011, 575) with a double aim, as a means to assert dominance over the opponent and simultaneously steal the floor.Although a moderator usually regulates floor time and segment structure, the extent to which they hold actual power depends on the candidates.
While it would be challenging to make generalizations about the content and focus of FTAs in presidential debates, a distinction can be made between those aimed at positive face and those targeted at negative face (table 1).The former are typically more prevalent and, as explained by Chilton, 'can be thought of in terms of threat to personal political credibility, political consensus and national identity in the domestic arena' (Chilton 1990, 221).

Against positive face
Against The frequency of FTAs in the debates-fueled by the diverse purposes of impoliteness-has prompted numerous adaptations of Culpeper's (1996) original taxonomy of impoliteness strategies, seeking to adapt it to this communicative context.Table 1 presents the revised taxonomy adopted in this paper, which addresses critiques of the original model-particularly concerning the classification of strategies 'against positive face' or 'against negative face'and incorporates new strategies proposed by García-Pastor (2008, 108), Blas Arroyo (2001, 30-38), Bousfield (2008, 126-135), Bull andFetzer (2010, 161) andFernández García (2015, 37).Some strategies were merged due to their overlapping definitions.

Image Restoration and Defenses
When targeted by an FTA, individuals may feel compelled to defend themselves and attempt to recover the face they lost (Brown and Levinson 1987, 61;Benoit 2007, 324).The responses to such threats on face can vary based on factors including idiolect and the specific communicative event.In the case of presidential debates-and political discourse in general-politicians often choose to counter an attack with another, which translates into a preference for attack-attack pairings over attack-defense pairings as explained by Bousfield (2008, 193).This choice can lead to impoliteness or aggressiveness spirals (Culpeper et al. 2003(Culpeper et al. , 1564;;Fernández García, 2015, 53).Favoring this path also underscores the level of impoliteness inherent to presidential debates, for candidates adhere to the notion that 'the best defense is a good offense' rather than engaging in image restoration.An example of this behavior could involve accusing the initial offender of lying, which would mean a defense against the previous claim and an attack on the opponent's credibility.
This paper is concerned with attack-defense pairings, which entail image restoration efforts through restorative strategies as conceived by Benoit and Wells (1996;also Benoit andHarthcock 1999, Benoit 2015; see table 2).While less common than attack-attack pairings, image restoration efforts or defenses-'remarks that respond to or refute attacks made against candidates' (Benoit and Glantz 2020, 10)-have been observed to occur more often in presidential debates than in other genres of political discourse (Benoit 2007, 321).Their frequency in this context may be attributed to the need to recover face in front of the voters, or to the immediacy of communication in debates, which, on the one hand, increases the impulse to respond 'in the heat of the debate' (Benoit and Brazeal 2002, 228), and on the other hand, reduces 'the risk of informing or reminding voters of a possible weakness' (Benoit and Glantz 2020, 50) because they have just witnessed the attack.
Despite being the least common function1 observed in political discourse, image restoration efforts have been analyzed in the U.S. presidential debates of 1960 (Benoit and Harthcock, 1999), 1988(Benoit and Brazeal, 2002), 1992 (Benoit and Wells, 1996), 1996(Benoit et al., 1998) and 2016 (Benoit, 2018), to identify the most prevalent defensive strategies.table 2. Image restoration strategies as per Benoit (2015, 22-26) Denial: (a) denies that the offensive act actually occurred or (b) denies that he or  2003) incorporated a taxonomy of response strategies (table 3) in a later revision.However, unlike Benoit's framework, this taxonomy was not specifically developed for public discourse, thus initially less suitable for analyzing presidential debates.As depicted in table 3, only four of the proposed strategies-those in italics-do not directly correspond with those proposed previously by Benoit.They are also yet to be applied to a corpus of presidential debates.table 3. Defensive strategies as per Culpeper et al. (2003Culpeper et al. ( , 1564Culpeper et al. ( -1567) ) Direct contradiction Abrogation Dismiss or make light of the FTA Ignore the FTA Offer an explanation

Opt out on record
Treat the situation as a different activity type Two of the italicized strategies also warrant further scrutiny.The strategy Ignore H is exceedingly challenging to apply, as it is difficult to determine whether an FTA has been ignored strategically or overlooked accidentally because of other factors such as interruptions or lengthy interventions.Opt-out would also have to be ruled out in the context of North American presidential debates, given no candidate has ever abandoned an ongoing debate.This further highlights the need for a more exhaustive taxonomy of strategies to tackle the analysis of defensive behavior in presidential debates.
The collaboration of Benoit's Image restoration theory with Culpeper's Impoliteness theory is made possible by the similarities between an FTA performed in public and Benoit's concept of a persuasive attack: 'messages that attempt to create unfavorable attitudes about a target' (Benoit 2015, 2).In presidential debates, FTAs hold particular significance because they unfold in front of an audience whose opinions can be swayed by the content of the attacks and their responses (Blas Arroyo 2001, 19).Furthermore, one of the core principles of Image restoration theory proves its suitability for analyzing defenses in presidential debates, for 'maintaining a positive reputation is one of the central goals of communication' (Benoit 2015, 14) within these events.The emphasis of this theory on reputation as a primary concern further supports the viability of its combination with Impoliteness theory and its application from a face perspective.

Objectives and Methodology
The main objective of this paper is to explore the possibility of combining Impoliteness theory and Image restoration theory to develop a comprehensive framework for the analysis of face-work in presidential debates.This investigation is motivated by the key role of defenses in the context of presidential debates, characterized by a constant struggle for face.Additionally, the selected corpus will shed light on the evolution of impolite and defensive behavior in North American presidential debates spanning from their early stages (1960) to the present day ( 2020), as well as potential evidence of gender variation.The hypotheses formulated for the analysis are as follows: H1 The increase in media presence in presidential campaigns has led to the rise in impoliteness in the debates.
H2 Impoliteness theory and Image restoration theory can be combined to facilitate a comprehensive analysis of face-work in presidential debates.
H3 There is sufficient statistical evidence of association between specific FTA and defensive strategies to identify a series of recurring attack-defense pairings in North American debates.
H4 There is evidence of gender-based variation of face-work in presidential debates.
The corpus selected for the analysis (Table 4) comprises one electoral race per decade to provide an overview of the evolution of face-work.Face-work in North American Presidential Debates: A Corpus-based Multi-theoretical...The methodology employed in the analysis encompassed qualitative, quantitative and statistical approaches.The qualitative analysis started by categorizing each debate intervention as either 'containing salient face-work' or 'no salient face-work'.Those belonging to the first category were then identified as 'threatening', 'defensive' or 'both'.This part of the analysis yielded insight into the evolution of impoliteness and defenses through the years comprised in the corpus.Once done, each impoliteness and restorative strategy found in the interventions was tagged, and so was their relationship with each other when relevant.This part of the analysis sought the most common attack and defensive strategies in the corpus separately and also as a pair, i.e. the attack-defense pairings with the highest absolute frequencies.Due to the difference in the size of the debates, I worked with percentages for the comparative part of the quantitative analysis.Absolute frequencies were used to determine the most common strategies overall.

Alicante Journal of
The findings of the latter part of the analysis were then statistically validated through testing on IBM SPSS.I conducted a Pearson's Chi-square test to determine if there was a significant correlation between both sets of strategies beyond random coincidence.The test took into account the absolute frequency of each pairing in the whole corpus separately as well as in comparison with the frequencies of the rest of the pairings.The specific FTA strategies and response strategies within an Attack-defense pairing that obtained a p-value under 0.05 are considered to have a significant correlation.

Evolution of Face-work
The highest percentage of impoliteness observed in the corpus was in the 1976 presidential debates (figure 1).Despite the lack of direct face-to-face interaction between participants, the percentage of interventions 2 devoted to impoliteness (76%) surpassed those from the newer debates, even after the new format was implemented in 2008 (63%).This finding rejects Hypothesis 1.However, since the adoption of the new format-which allows direct dialogue between candidates-there has been a discernible upward trend which shows how it favors impoliteness occurrences.The grey lines, which represent interventions containing image restoration in some form, illustrate the prevailing inclination towards simultaneous threat and defense until the 2016 and 2020 debates, when there was a higher percentage of interventions devoted solely to image restoration efforts (21% and 25% respectively).As the frequency of interventions exclusively devoted to FTAs declined, those containing simultaneous restoration and threatening efforts increased in all years save for 2016 and 2020, the two instances when Donald Trump was a candidate.The instances of negative correlation between these two trends highlight how even when the attention focused on defensive behavior grows, impoliteness remains salient.The higher relative frequency of simultaneous attack-and-defense utterances compared to solely defensive utterances is a testament to why impoliteness has come to be expected from presidential candidates.
The higher frequency of interventions containing FTAs supports findings from other investigations under Benoit's (2007b) Functional Theory of Political Campaign Discourse, which often found defenses to be the least common separately and also as a pair, i.e. the attack-defense pairings with the highest absolute frequencies.Due I worked with percentages for the comparative part of the quantitative analysis.Absolute frequencies strategies overall.
The findings of the latter part of the analysis were then statistically validated through testing on IBM test to determine if there was a significant correlation between both sets of strategies beyond random absolute frequency of each pairing in the whole corpus separately as well as in comparison with the specific FTA strategies and response strategies within an Attack-defense pairing that obtained a psignificant correlation.2007) study of defenses in presidential debates concluded that the increase in attacks against a candidate prompted an increase in their defenses.This could be the reason behind the relationship between the trends observed in interventions containing FTAs, and those devoted exclusively to image restoration.

Impoliteness and Defensive Strategies
Idiolect is a major factor influencing the choice of strategy, yet participants in presidential debates are constrained by their nature as presidential candidates and conditioned by the confrontational nature of the communicative event.This resulted in similarities regarding the most prevalent attack strategies across the debates.Due to space limitations, figure 2 includes the absolute frequency of the most common strategies that appeared over twenty times in at least two of the sets of presidential debates.Refuse H (1), Associate with a negative aspect (2) and Unfavorable comparison (3) were consistently the three most common strategies throughout the debates, save for 2008, when the second most common strategy was Question Credibility (4).Note that the significant differences in the total frequency of the strategies are due to the difference in the size of the debate sets.Likewise, Trump's participation in the debates notably contributed to the spike in FTAs.His debate style could be influenced by his previous career as an entertainer, which may make him more likely to lean into the entertainment value of impoliteness.
attack strategies across the debates.Due to space limitations, figure 2 includes the absolute frequency of the most common strategies that appeared over twenty times in at least two of the sets of presidential debates.FIGURE 2. Absolute frequencies of the most prevalent impoliteness strategies.
Refuse H (1), Associate with a negative aspect (2) and Unfavorable comparison (3) were consistently the three most common strategies throughout the debates, save for 2008, when the second most common strategy was Question Credibility (4).Note that the significant differences in the total frequency of the strategies are due to the difference in the size of the debate sets.Likewise, Trump's participation in the debates notably contributed to the spike in FTAs.His debate style could be influenced by his previous career as an entertainer, which may make him more likely to lean into the entertainment value of impoliteness.The evolution of the Hinder/block strategy also bears mentioning, for it experienced a surge from 2008 onwards coinciding with the introduction of face-toface interaction and the newfound possibility to steal the floor from the opponent through interruptions.Similarly, Challenge H became more popular as candidates gained more opportunities to pose their own probing questions.Example (5) depicts a combination of both strategies in a segment of the second 2016 debate between Donald Trump and Hillary Clinton.In this instance, Trump impeded the progress of communication by repeatedly interrupting Clinton and challenging her about her deleted e-mails, a highly contentious topic in the 2016 elections.
Face-threatening strategies against positive face (80.48%) were more common in the corpus than those against negative face (27.26%), which supports the observation that positive face is especially significant for those running for office.Politicians' awareness of being watched by the audience likely led to a higher number of attacks on positive face.Although consistently fewer than those against positive face, the percentage of attacks against negative face increased from 2016 onwards, likely due to Trump's frequent interruptions-12.42% of his total FTAs were interruptions, while the second highest percentage of interruptions was Obama's 7.69%.
Regarding the taxonomy proposed in Section 3, I ultimately considered Refer to rights, duties and rules not respected, fulfilled or complied with to be threatening against both positive and negative face in the corpus.Reprimanding the opponent in front of the audience by bringing to attention unfulfilled duties-failing to implement an economic plan or remaining still in the face of war-did not target their freedom in most cases, but rather pointed out their lack of compromise and reliability.
Comparing these results with previous investigations is challenging due to the unique taxonomy used for the analysis.However, findings do align with the higher importance traditionally attributed to positive face attacks over negative face attacks in politics.
Figure 3 showcases the most common defensive and restorative strategies, considering only those with an absolute frequency exceeding five in at least two of the sets of debates.Among the strategies depicted in figure 3, two are only considered in Culpeper et al.'s (2003) taxonomy of defensive strategies (Dismiss and Offer explanation).The predominant strategies throughout the debates were Deny (6) and Reduce offensiveness (7).The latter is unique to Benoit's image restoration taxonomy and has no equivalent in Culpeper et al.'s (2003) framework.In the corpus, it was most often deployed in the Transcendence (7) and Bolstering (8) variations.
Transcendence attempts to reduce the offensiveness of the act by making clarifications or painting it in a different light; Bolstering entails reminding others of one's positive achievements.
(6) The so-called negotiation that you said I walked out on was the so-called walk in the woods between one of our representatives and one of the Soviet Union, and it wasn't me that turned it down, the Soviet Union disavowed it.[Reagan, second 1984 debate] The predominant strategies throughout the debates were Deny (6) and Reduce offensiveness (7).The latter is uniqu taxonomy and has no equivalent in Culpeper et al.'s (2003) framework.In the corpus, it was most often deployed Bolstering (8) variations.Transcendence attempts to reduce the offensiveness of the act by making clarifications or Bolstering entails reminding others of one's positive achievements.
(6) The so-called negotiation that you said I walked out on was the so-called walk in the woods between one of our repre Union, and it wasn't me that turned it down, the Soviet Union disavowed it.[Reagan, second 1984 debate] (7) As a matter of fact -I don't advocate amnesty; I advocate pardon.There's a difference in my opinion and in accordanc Court and accordance with the definition in the dictionary.[Carter, first 1976 debate] (8) Well, as soon as he travels to 112 countries and negotiates a peace deal, a cease-fire, a release of dissidents, an openin around the world, or even spends 11 hours testifying in front of a congressional committee, he can talk to me about stamina (7) As a matter of fact -I don't advocate amnesty; I advocate pardon.There's a difference in my opinion and in accordance with the ruling of the Supreme Court and accordance with the definition in the dictionary.
[Carter, first 1976 debate] (8) Well, as soon as he travels to 112 countries and negotiates a peace deal, a cease-fire, a release of dissidents, an opening of new opportunities in nations around the world, or even spends 11 hours testifying in front of a congressional committee, he can talk to me about stamina.[Clinton, first 2016 debate] The results of the analysis support the findings from previous studies conducted under Image restoration theory.Deny and Reduce offensiveness are consistently the most common defensive strategies across presidential debates, as seen in the studies carried out by Benoit and Harthcock (1999, 352), Benoit and Brazeal (2002, 227), Benoit and Wells (1996, 87), Benoit et al. (1998, 177) and Benoit (2018, 30).While the exact number of instances of the Reduce Offensiveness variants changes from this analysis to others, it is not an alarming deviation.
As predicted earlier in this paper, Opt-out and Ignore H were not observed in the analysis.Despite appearing similar, responding to an FTA with Dismiss entails acknowledging of the attack in some way, while Ignore H would imply no explicit manifestations.In contrast, all of Benoit's strategies yielded at least one occurrence in the corpus.
Additionally, although Offer explanation was considered a distinct strategy before the analysis, it could also be understood as a variation of Reduce offensiveness akin to Minimize offensiveness or even Transcendence.This observation reinforces the idea that Benoit's taxonomy is more thorough.However, the inclusion of Dismiss among the most used defensive strategies emphasizes the need for cooperation between theories to analyze responses, as it lacks an equivalent in Benoit's taxonomy.Consequently, Hypothesis 2 is confirmed.

Attack-defense Pairings
The attack-defense pairings with the highest absolute frequency in the corpus were Refuse H-Deny (9) and Negative Aspect-Deny (10), closely followed by Refuse H-Reduce offensiveness (9; in bold).These pairings were identified between sixtythree to fifty-three times in the corpus, but none were present in all the debates.Refuse H-Reduce offensiveness was the most consistent pairing, appearing in all the debates but the third in 1976 and the second in 1984.
(9) NIXON: And so I would say that in all of these proposals Senator Kennedy has made, they will result in one of two things: either he has to raise taxes or he has to unbalance the budget.If he unbalances the budget, that means you have inflation, and that will be, of course, a very cruel blow to the very people-the older people-that we've been talking about.Despite not being the most frequent pairings, the results of the Chi-Square test demonstrate the statistical significance and, in turn, the recurrence of these twelve pairings in the political discourse of North American presidential debates.This means that whenever these FTA strategies are employed, they are often met with the corresponding response strategies in the pairings.The lowest the p-value, the strongest the correlation.These findings confirm Hypothesis 3.
Seven of the statistically significant pairings consist of an FTA strategy from Impoliteness theory and a restorative strategy exclusive to Image restoration theory, further supporting Hypothesis 2. While five of the twelve pairings include a defensive strategy exclusive to Impoliteness theory (Dismiss and Offer explanation), future research could combine Offer explanation and Reduce offensiveness based on their previously unnoticed similarities and lower the number to two.
The analysis also revealed that moderators tended to intervene when faced with quick successions of interruptions.The Chi-square test results showed a significant correlation between the FTA strategy Hinder/block and the intervention of the moderator (p=0.001).
Lack of research conducted under these theories combined prevents the comparison of results.

Gender Variation
Hypothesis 4 proved challenging to confirm or reject given the limited data available.The 2016 presidential debates were the only set with a female candidate, in the form of Hillary Clinton.This makes it impossible to extrapolate any observations and attribute them to the gender variable instead of, for example, the candidate's idiolect.That said, there were noteworthy differences in the percentage of interventions Clinton allocated for image restoration (19%) and those allocated by his opponent, Donald Trump (7%) (figure 4).
In an attempt to obtain more data, the same methodology was employed to analyze the 2023 Spanish presidential debate, where there was also a female candidate: Yolanda Díaz.The findings of this analysis were similar to those of the 2016 presidential debates in the U.S. The female candidate allocated a significantly higher percentage of her interventions solely to image restoration than her two male opponents did (figure 5).Hillary Clinton's interventions containing solely image restoration doubled Donald Trump's, and the same can be observed of Yolanda Díaz's when compared to Pedro Sánchez's and almost Santiago Abascal's.These findings suggest a greater concern with recovering face from the female candidates, together with a preference for engaging in exclusively restorative efforts over engaging simultaneously in defense and attack on the part of Clinton.Still, the limited sample of female candidates prevents a definitive conclusion.

Conclusion
This paper has explored the possibility of combining Impoliteness theory and Image restoration theory to create a comprehensive framework for the analysis of face-work in presidential debates.In implementing this hypothetical new model with a representative corpus, I could observe the evolution of face-work in North American presidential debates from their first broadcast in 1960 until their latest in 2020.While impoliteness did not show the hypothesized upward trend, it seems to have been slightly growing since the 2008 debates, when candidates were first allowed to debate face-to-face.Donald Trump's presence in the last two U.S. elections may have influenced the findings, given his markedly impolite debate style.Upcoming elections could show a different trend.
While the analysis showed impoliteness dominated the debates, image restoration by itself and in combination with FTAs has also proved to be deserving of attention and research.Akin to impoliteness being considered a key aspect of presidential character and political discourse, research conducted on other genres of political discourse under Image restoration theory and more in-depth analyses like the one presented in this paper could lead to the incorporation of 'defensive abilities' to the discursive characterization of politicians.It could also prove the mediatization of politics results in a stronger need for defenses, resulting in the incorporation of some considerations for defensive behavior to the conceptualization of presidential character.The proposed model of analysis poses an interesting alternative for the analysis of face-work given the nature of defenses in presidential debates.Results showed statistically relevant attack-defense pairings comprising strategies exclusive to Image restoration theory, proving its taxonomy suitable for the analysis of restorative efforts in this context.The existence of pairings strategies exclusive to Impoliteness theory makes it impossible to disregard its approach to face restoration despite the lack of occurrences of some of its strategies in the corpus, namely Ignore H and Opt-out.
Results of the analysis partially lined up with results from previous similar investigations conducted by other researchers in terms of the similarity in the trends of the different types of interventions labeled and the most common restorative strategies..
Unfortunately, the size of the corpus of female candidates was too small to obtain conclusive results about the gender variation of face-work in the debates.Although the analysis of a functionally equivalent Spanish corpus showed similarities in the image restoration efforts of the female candidates, more data is necessary to obtain conclusive results.On another note, analyzing the full corpus of North American presidential debates could lead to the discovery of more statistically significant attack-defense pairings and even allow the labeling as such of the most frequent pairings in this corpus.Likewise, it could yield further proof of the suitability of this method of analysis.Analyzing a corpus of Spanish presidential debates could identify any shortcomings in the model when dealing with a different culture.

Figure 1 .
Figure 1.Evolution of relative frequency of each kind of intervention in the corpus.

2
Percentages were obtained by considering only interventions containing salient face-work.

Figure 2 .
Figure 2. Absolute frequencies of the most prevalent impoliteness strategies.

Figure 3 .
Figure 3. Absolute frequencies of the most prevalent defensive strategies.

Figure 4 .
Figure 4. Percentages of each kind of intervention per candidate at the 2016 presidential debates.

Figure 5 .
Figure 5. Percentages of each kind of intervention per candidate at the 2023 Spanish presidential debate. 2

FIGURE 4 .FIGURE 4 .
FIGURE 4. Percentages of each kind of intervention per candidate at the 2016 presidential debates.

it is a fact that in the last eight years the Democratic Congress has reduced the appropri-the requests for the appropriations by over ten billion dollars
KENNEDY: […] That is wholly wrong, wholly in error, and .[First1960 debate, Bolstering] (10) MCCAIN: As you notice, he starts talking about government.He starts saying, government will do this and government will do that, and then government will, and he'll impose mandates.OBAMA:All I'm going to do is help you to lower the premiums on it.You'll still have choice of doctor.There's no mandate involved.Small businesses are not going to have a mandate.[Second2008 debate]Although these pairings have the highest absolute frequency, to identify statistically relevant and potentially recurring attack-defense pairings, a Pearson's Chi-square test was conducted.The pairings listed in table 5 were the ones found to be significantly correlated with p-values under 0.05.

table 5 .
Pearson's Chi-Square test results of pairings with a significant p-value.