Abstract
While counterfactual thinking has been widely investigated, we know much less about how counterfactual (“If . . . then”) statements are employed in political communication. We analysed statements made by politicians during pre-electoral televised broadcasts, to assess whether politicians employ counterfactuals in facework. Counterfactuals were coded according to their direction, controllability, and structure. Log-linear analysis revealed that upward, controllable, and additive counterfactuals were more frequent than downward, uncontrollable, and subtractive counterfactuals, respectively. A significant three-way interaction between target, direction, and controllability also emerged. While politicians more often employed upward controllable counterfactuals when speaking about targets other than themselves, they more often used downward controllable and upward uncontrollable counterfactuals when referring to themselves. These findings advance our knowledge of how counterfactuals are employed by politicians to promote their positive face and aggravate the face of adversaries.
Keywords
“If the previous government were still in power, the yield spread between Italian and German government bonds would now be at 1200.” This statement, made by Italian Prime Minister Mario Monti on August 7, 2012, is an example of a counterfactual statement: an antecedent of a past event is postulated to have changed to alter, hypothetically, the outcome of the event (Kahneman & Miller, 1986; Roese, 1997).
Events that are perceived as negative are more likely to trigger counterfactuals, and counterfactuals are often used spontaneously as an explanatory heuristic, offering a simplified way to attribute responsibility for the negative outcome (Mandel, Hilton, & Catellani, 2005). One can therefore assume that counterfactuals are employed in communication to convey a simplified interpretation of an event and to suggest an attribution of responsibility that is consistent with the communicative goals of the speaker. However, with the exception of a small number of studies (Catellani, Alberici, & Milesi, 2004; Catellani & Milesi, 2001; Wong, 2010), there is little research on counterfactuals that are embedded in a communicative context, including the political context.
Peter Bull has convincingly pointed out that politicians devote much of their communication time to promoting and defending their face (positive face) and attacking the face of adversaries (face aggravation; Bull, 2008; Bull & Fetzer, 2010; Bull & Wells, 2012). This often involves, for example, evasive or equivocal responses to questions that are face-threatening to the politician themselves, their group, or their allies. When politicians do reply to such questions, they often attempt to talk up their positive face by, for example, outlining the positive aspects of their actions. Politicians also use strategic face aggravation to attack their adversaries. In doing so, they often recur to mitigation to reduce the full force of the attack and make it more acceptable. For example, they use humorous discourse or quote from newspaper interviews.
In the present research, we assumed that counterfactuals are among the indirect communicative strategies that politicians employ to defend their own face and to attack the face of adversaries. The creation of hypothetical alternative scenarios rather than reference to reality may be an indirect yet efficacious way of saving positive face and performing face aggravation.
To investigate this type of facework, during the months preceding the 2006 Italian General Election we video-recorded and analysed appearances on television by the two candidates for the premiership, Silvio Berlusconi and Romano Prodi. We identified all the counterfactuals embedded in their statements and analysed them according to the four main dimensions along which counterfactuals can be classified: target, direction, controllability, and structure (see Mandel et al., 2005). A counterfactual target is the actor whose actions are mutated in the hypothetical scenario (e.g., “If the government had implemented better fiscal policies, the economic condition of the country would be better today”). Counterfactual direction relates to whether the outcome envisaged in the hypothetical scenario is better (upward counterfactual) or worse (downward counterfactual) than the real outcome. Controllability concerns whether the circumstances referred to in the counterfactual antecedent were or were not under the control of the actors involved in the real event. Finally, counterfactual structure involves a distinction between subtractive counterfactuals, where an element that was present in the actual scenario is removed (e.g., “If the government hadn’t approved that economic law . . .”), and additive counterfactuals, where an element that was not present in the actual scenario is introduced (e.g., “If the government had taken special measures to reduce the inflation rate . . .”).
Some types of counterfactuals are more commonly generated than others, and we expected to find these same patterns in counterfactuals embedded in political discourse. In particular, we expected upward counterfactuals to be more frequent than downward counterfactuals (Markman, Gawanski, Sherman, & McMullen, 1995), counterfactuals focused on controllable behaviours to be more frequent than counterfactuals focused on uncontrollable behaviours (Girotto, Legrenzi, & Rizzo, 1991), and additive counterfactuals to be more frequent than subtractive counterfactuals (Markman, Lindberg, Kray, & Galinsky, 2007).
More important, we also expected that the main patterns of counterfactual use would change in accordance with the targets on which the counterfactuals were focused. Specifically, we anticipated that politicians would employ different types of counterfactuals when speaking about themselves and when speaking about others and that this would be consistent with the goals of either defending their positive face or aggravating the face of adversaries. We developed three research hypotheses in this regard.
Upward controllable counterfactuals increase the perceived negativity of the actual event and the probability that the counterfactual target is perceived as primarily responsible for that event (Nario-Redmond & Branscombe, 1996). As one of the communicative goals of politicians is face aggravation, we hypothesised that politicians would focus upward controllable counterfactuals mainly on adversaries or on other actors (i.e., make statements of the type, “If they had acted in a different way, things would have been better”).
While upward counterfactuals enhance the perceived negativity of the event, downward counterfactuals diminish that negativity (McMullen & Markman, 2000) and therefore may reduce the level of responsibility attributed to the target. As a main communicative goal of politicians is to defend their positive face, we hypothesised that politicians would focus downward controllable counterfactuals mainly on themselves (i.e., make statements of the type, “If I had acted in a different way, things would have been worse”).
Although counterfactuals that focus on controllable behaviours are generally more frequent than those that focus on uncontrollable behaviours, the generation of upward uncontrollable counterfactuals increases when individuals aim to make excuses for negative events in which they were involved (Markman & Tetlock, 2000). Hence, we expected that politicians would occasionally defend their positive face by resorting to upward uncontrollable counterfactuals focused on themselves (i.e., making statements of the type, “If I had been able to act in a different way, things would have been better”).
Method
Materials
Six televised pre-electoral broadcasts from the 2006 Italian General Election formed the basis of this study. The broadcasts were transmitted on the main channel of the Italian state television station (RAI 1) and received very high viewing figures. They provided the most prominent opportunity for the two premier candidates, Silvio Berlusconi and Romano Prodi, to expound their political programme. The broadcasts were (a) two episodes of Porta a Porta (Door to Door, a popular talk show) under the title “Berlusconi’s Italy,” featuring the incumbent leader Berlusconi as the sole guest (January 31 and March 8, 2006); (b) two episodes of the same programme, Porta a Porta under the title “Prodi’s Italy,” featuring the opposition leader Prodi as the sole guest (February 7 and March 7, 2006); (c) two Faccia a Faccia (Face to Face) programmes, in which both leaders were present (March 14 and April 3, 2006).
Coding Transcripts
The broadcasts were video-recorded and transcribed. Then two independent coders identified and coded all the counterfactuals embedded in the politicians’ statements. Counterfactuals were either explicit (if . . . then) or implicit and alluded to hypothetical scenarios (e.g., at least, next time, otherwise, if not, it isn’t that, without, though) or to expectations that had not been met (e.g., only, even, still, instead). Interrater reliability was calculated using Cohen’s (1960) Kappas and was high for the identification of both explicit (.82) and implicit (.75) counterfactuals. After coding all the counterfactuals, the implicit counterfactuals were changed into their explicit form (see Catellani & Milesi, 2001). For example, the statement—“They should have invested in the fight against tax evasion. Instead, they did not do that and they have not regained control over the state budget”—was changed to—“If they had invested in the fight against tax evasion, they would have regained control over the state budget.” All counterfactuals were then independently classified by the same two coders according to the programme, the speaker, and the four main counterfactual dimensions as specified below.
Programme
Either Porta a Porta (i.e., talk show) or Faccia a Faccia (i.e., face-to-face programme).
Speaker
Either incumbent leader (Silvio Berlusconi) or challenging leader (Romano Prodi).
Target
Either self-focused counterfactuals focused on the speaking politician, his party, or his political coalition, or opponent-focused counterfactuals focused on the opposing politician, his party, or his coalition, or other-focused counterfactuals focused on generic or largely inclusive collective actors, such as “the nation,” “international institutions,” or other individual or collective figures acting in social, political, and economic domains.
Direction
Either upward counterfactuals in which it is imagined that “If . . . , things would have been better” or downward counterfactuals in which it is imagined that “If . . . , things would have been worse.”
Controllability
Either controllable counterfactuals in which it is imagined that the target had control over the behaviour (e.g., “If the opposition had voted in favour of this law . . .”) or uncontrollable counterfactuals in which it is imagined that the target had no control (e.g., “If I could have counted on more financial resources . . .”).
Structure
Either additive counterfactuals, where an antecedent is hypothetically added to the counterfactual scenario (e.g., “If the government had given the required resources to the local authority . . .”), or subtractive counterfactuals, where an antecedent is hypothetically deleted from the counterfactual scenario (e.g., “If they had not introduced the euro . . .”).
To verify the effectiveness of the coding scheme, interrater reliability was calculated, and all Cohen’s Kappa values had high concordance between coders: .82 for the programme, .88 for the speaker, .77 for target, .87 for direction, .74 for controllability, and .81 for structure. Disagreements were resolved through discussion.
Results
Preliminary Analyses
The length of the interventions of the two politicians in the six programmes did not differ significantly: χ2(1, N = 80,638) = 0.51, p = .47. In the four talk-show episodes, Berlusconi’s utterances totalled 28,650 words and Prodi’s totalled 28,177. In the two face-to-face programmes, Berlusconi’s utterances amounted to 11,939 words and Prodi’s came to 11,872 words. The analysed texts therefore amounted to a total of 80,638 words.
The analysis of the politicians’ utterances revealed the presence of 194 counterfactuals, 73.2% of which were in implicit form. A comparison between implicit and explicit counterfactuals revealed that they did not differ with respect to any of the classification criteria. The same held true for the programme and speaker variables. We therefore dropped these variables from our subsequent analyses.
A further preliminary analysis revealed that 45.4% (88) of the counterfactuals were focused on the politicians themselves, 36.6% (71) on their opponent, and 18.0% (35) on other targets. Opponent- and other-focused counterfactuals did not differ with respect to their interactions with the other classification criteria and were therefore merged into a single target category. Hence, only two target categories, namely, “self” and “other,” were employed in the main analysis.
Hierarchical Log-Linear Analysis
A hierarchical log-linear analysis was applied to all the counterfactuals and included the following four variables: target (two levels: self, other), direction (two levels: upward, downward), controllability (two levels: controllable, uncontrollable), and structure (two levels: additive, subtractive).
The results of the tests of significance revealed a number of significant main and interaction effects. Upward counterfactuals (74.7%) prevailed over downward counterfactuals (25.3%), χ2(1, N = 194) = 47.50, p < .001; counterfactuals focused on controllable behaviours (76.8%) prevailed over uncontrollable ones (23.2%), χ2(1, N = 194) = 55.75, p < .001; and additive counterfactuals (74.7%) prevailed over subtractive counterfactuals (25.3%), χ2(1, N = 194) = 47.50, p < .001. Only counterfactual target did not have a main effect, χ2(1, N = 194) = 1.67, p = .19, because a comparable percentage of counterfactuals focused on the self (45.4%) and on other targets (54.6%).
With respect to interactions among counterfactual dimensions, backward elimination without the four-way interaction demonstrated that a four-way interaction was not required to explain the data, L.R. χ2(1, N = 194) = 0.45, p = .51. However, the combination of three-way effects was significant, indicating that at least one of the three-way effects was required to explain the data, L.R. χ2(1, N = 194) = 13.02, p < .001. Tests of partial associations showed that the target × direction × controllability interaction was the only significant three-way interaction, χ2(1, N = 194) = 8.66, p < .01.
Further analyses on the significant three-way interaction tested the plausibility of our hypotheses. First, two separate analyses on controllable and uncontrollable counterfactuals revealed a significant target × direction interaction for controllable counterfactuals, χ2(1, N = 149) = 49.58, p < .001, but not for uncontrollable counterfactuals, χ2(1, N = 45) = 0.35, p = .55. As shown in Figure 1, the percentage of upward controllable counterfactuals focused on other targets was higher than the percentage of upward controllable counterfactuals focused on the self (Hypothesis 1; 79.6% vs. 20.4%), χ2(1, N = 108) = 37.92, p < .001. The opposite was the case for downward controllable counterfactuals, which were more often focused on the self than on other targets (Hypothesis 2; 82.9% vs. 17.1%), χ2(1, N = 41) = 17.78, p < .001.
Percentage values of controllable counterfactuals as a function of target and direction.
Next, we separately analysed upward and downward counterfactuals and found a significant target × controllability interaction for upward counterfactuals, χ2(1, N = 145) = 34.08, p < .001, but not for downward counterfactuals, χ2(1, N = 49) = 1.72, p = .19. As shown in Figure 2, while the politicians mainly focused upward controllable counterfactuals on other targets instead of on themselves (79.6% vs. 20.4%), χ2(1, N = 108) = 37.92, p < .001, the pattern was reversed for upward uncontrollable counterfactuals. In this case, the percentage of self-focused counterfactuals was much higher than the percentage of other-focused counterfactuals (Hypothesis 3; 73% vs. 27%), χ2(1, N = 37) = 7.81, p < .01.
Percentage values of upward counterfactuals as a function of target and controllability.
Discussion
We investigated whether counterfactuals may be among the indirect communicative strategies that politicians employ in facework (Bull & Fetzer, 2010). Our results confirmed that this is the case and fully supported our three hypotheses. Overall, the politicians employed more upward than downward counterfactuals, more controllable than uncontrollable counterfactuals, and more additive than subtractive counterfactuals. These patterns are consistent with the findings of previous research on counterfactual thinking (Girotto et al., 1991; Markman et al., 1995; Markman et al., 2007). However, we also found that these patterns changed in accordance with the target counterfactuals were focused on. Specifically, we identified three counterfactual strategies employed for facework.
First, politicians employ more upward controllable counterfactuals when focusing on targets other than themselves, making statements of the type: “If they had acted in a different way, things would have been better.” For example, in reply to a journalist who asked whether Europe was currently helping Italy more or less than in the past, Berlusconi said, “If Prodi [the other candidate] had defended the interests of Italy when he was at the European Parliament, things would be better now” (Door to Door, March 8, 2006). The targets of upward controllable counterfactuals are more likely to be perceived as responsible for negative events (Nario-Redmond & Branscombe, 1996). The use of other-focused upward controllable counterfactuals may therefore serve the goal of face aggravation, and in our research this was indeed the most frequent counterfactual strategy employed by the politicians.
Second, politicians employ downward controllable counterfactuals mainly when referring to themselves, making statements of the type: “If I had acted in a different way, things would have been worse.” For example, in reply to a journalist who suggested that a fight against inefficiency would not solve the problem of too high public expenditure, Berlusconi said, “If I hadn’t intervened on tax evasion in the budget, the economic condition of the country would have been worse” (Face to Face, April 3, 2006). Comparing the real outcome of an event with a hypothetical worse alternative reduces the negative perception of the actual event (McMullen & Markman, 2000). Politicians therefore adopt this counterfactual strategy to try to save positive face.
Third, politicians employ upward uncontrollable counterfactuals more often in relation to themselves than others, making statements of the type, “If I had been able to act in a different way, things would have been better.” For example, in response to a journalist asking how Prodi would deal with divisions within his party regarding the withdrawal of Italian troops from Iraq, Prodi said, “If it had been up to us, we wouldn’t have entered the Iraq War” (Face to Face, April 3, 2006). Hence, it would appear that upward uncontrollable counterfactuals are employed by politicians to reduce their responsibility for negative outcomes (see Markman & Tetlock, 2000). We found that this third type of face-saving counterfactual strategy was less frequent than the other two strategies. One possible explanation for this is that admitting that certain events/actions are uncontrollable is too overt a violation of the expectations of the audience: people generally expect individuals to have control over their actions (Fiske & Taylor, 1984).
In the present study, we assessed the types of counterfactuals that are more frequently employed by politicians as face-aggravating and face-saving strategies but did not study the broader interactive context, in this case their interaction with journalists. Further studies could investigate whether politicians recur to counterfactuals more frequently in response to questions from journalists that also include counterfactuals or that are face-threatening. The assessment of how effective counterfactual strategies are in the context of audience perception is also a promising line of investigation. For example, one could experimentally manipulate the statements of politicians to investigate which of the three counterfactual strategies described here is rated as more convincing and leads to a better evaluation of the politician using it. Such studies could help to build on the foundations laid by the present research and elaborate how counterfactual communication is used by politicians seeking to convey their interpretations of reality to their audience.
Footnotes
Acknowledgements
We wish to thank Howard Giles and the anonymous reviewers for their helpful comments. We would also like to thank Patrizia Milesi, Marilena Gagliardi, Sara Galli, Federica Marelli, and Silvia Meregalli for their help in the transcription and coding of the broadcasts.
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by the Italian Council of University Research (Grant Number Cofin-PRIN 2005141050) and the Catholic University of Milan (Grant Number D1-2832C13).