Running Head: SWEARING IN POLITICAL DISCOURSE
Swearing in Political Discourse: Why Vulgarity Works
Nicoletta Cavazza1, Margherita Guidetti1
An experimental study investigated the effect of politicians’ profanity and gender on their perceived
and actual persuasiveness. Results showed that a candidate’s use of swear words increased the
perception of language informality and improved the general impression about the source. The latter
effect was particularly strong for male candidate, as female candidate was already evaluated
positively, irrespective of her cursing. In addition, though the manipulation of the politician’s
vulgarity did not directly affect participants’ self-reported likelihood of voting for him/her, an
indirect effect through language informality and impression about the candidate emerged. On the
contrary, profanity use reduced perceived persuasiveness of the message, suggesting that the
influence of swearing could be automatic and unaware. Theoretical implications are discussed.
language vulgarity, persuasion, political communication, social judgment, language informality.
1Università di Modena e Reggio Emilia
Nicoletta Cavazza, Università di Modena-Reggio Emilia, via Allegri 9, 42121 Reggio Emilia, Italy.
In the last Italian political election (2013) the Five Stars Movement became unexpectedly the
second most voted party at its first appearance on the electoral scene. Its leader, the former
comedian Beppe Grillo based his campaign exclusively on blog posts and monologue performances
in city squares and, for the first time in the Italian political arena, delivered messages replete with
swear words (Bordignon & Ceccarini, 2013).
The effects of language on impression formation and decision making in the political field
are widely studied. In particular, much attention has been devoted to the effects of attacks, outrages
and incivilities (i.e., negative campaign). However, little is known about the effect produced by
simply filling the discourse with swear words without necessarily attacking the adversary. The
present investigation explored the consequences of including swear words in political messages, in
terms of impression of the source and persuasiveness.
How do people perceive somebody swearing? Research findings are somewhat
controversial. Some studies have shown that swearers are perceived as untrustworthy, incompetent
and lowly sociable (Bostrom, Baseheart, & Rossiter, 1973; Hamilton, 1989; Paradise, Cohl, &
Zweig, 1980). However, Scherer and Sagarin (2006) failed to find any influence of profanity on
source credibility. Drawing on expectancy violation theory (Burgoon, 1993), Johnson and
colleagues (Johnson & Lewis, 2010; Johnson, 2012) hypothesized that swearing can be perceived
positively, neutrally or negatively. Indeed, they found that only negative expectancy violations had
detrimental consequences over perception of the speaker and of message effectiveness, whereas
when swearing was considered as a positive surprise the opposite was observed.
In order to study the effects of message vulgarity we have to individuate which functions
swearing fulfill (Jay, 2000). People curse mainly because profanity helps express mood, aggressive
urge, dominance, etc. (e.g., Fine & Johnson, 1984; Jay, 2009). In addition, nasty words function as
rhetoric tools used by the source in order to intensify the discourse, and to define an informal and
friendly relationship with the receiver, thus reinforcing social connections (Wajnryb, 2005).
However, the empirical evidence of vulgarity rhetorical and social functions is rare. Bostrom
et al. (1973) exposed participants to a videotaped interview including either a religious, excretory,
or sexual swear word, or to the same interview devoid of profanity. The vulgar message, regardless
of the nature of the swear word, induced greater attitude change toward the topic when the source
was a woman rather than a man, even if in both cases profanity had a detrimental effect on the
perceived credibility of the communicator. Other studies carried out in the judicial domain suggest
that the persuasive effect of vulgar language occurs without recipients’ awareness: Although
respondents perceived swear words as a sign of deceit, their presence actually increased the
believability of fictitious statements (Rassin & van der Heijden, 2005). Finally, Scherer and Sagarin
(2006) manipulated the presence/absence of a swear word in a 5-minute video addressed to
university students by a male speaker supporting a pro-attitudinal topic (i.e., lowering university
tuition). Based on reinforcement expectancy theory (Bradac, Bowers, & Courtright, 1980), the
authors hypothesized that language intensity would affect source evaluation that in turn would
affect attitude change. In this study, the simple presence of a swear word actually improved the
receivers’ attitude about the topic. The effect was partially mediated by perception of language
intensity, whereas swearing did not affect source credibility. Thus, although the effect of profanity
in terms of perception of the speaker is controversial, there is evidence of its positive effect in terms
of persuasiveness, mediated by language intensity.
Therefore, we can speculate that, beyond the expressive function, people curse also because
they have learned from experience that swear words may reinforce message effectiveness. This is
not to say that people are aware of this persuasive effect, but they may have automatized the
association between swearing and positive outcomes.
The results of the recent Italian election might be an example of the ability of vulgar
messages to convey consensus. However, to the best of our knowledge, no research study has
focused attention on the specificity of the persuasive communication in the political domain.
This would be an important test because political communication, compared to commercial
or other communicative domains, is characterized by the centrality of the source. Actually, the
candidate is the “product to sell”, and the crucial persuasive effect is not about influencing the
audience attitude towards some issues discussed in a message, but convincing them to support the
source of the message in case of election. Thus, previous findings cannot be simply generalized to
political communication in which we can expect that source evaluation would play a stronger role
in respect to that observed by Scherer and Sagarin (2006).
Furthermore, as we have seen, a bad impression is only realized when swear words induce a
negative surprise, and can be favorable instead if the surprise is positive (Johnson, 2012). Since
research in the field of political communication stresses the efficacy for a candidate of being
perceived as an “everyday man” (Barisione, 2009), a positive surprise might occur through
perception of informality. As most swearing occurs in friendly conversation (Jay & Janschewitz,
2008), a proportion of mediation between message vulgarity and its persuasiveness should be due to
the perception of language informality, which in turn should induce a positive impression about the
Finally, since men are generally observed to curse more than women (Foote & Woodward,
1973; Ginsburg, Ogletree, & Silakowski, 2003; but see Bayand & Krishnayya, 2001), and gender
stereotypes are also reflected in the appropriateness of language to be used by men and women, we
will control whether source’s gender moderates the effect of vulgarity on impression about the
candidate and voting intention.
The Present Experiment
The purpose of this study was to investigate the impact of swearing on the influence process in the
political domain. To this end, we examined the combined effect of politicians’ use of swear words
and gender on perception of language intensity and informality, impression about the source and
self-reported likelihood of voting for that candidate (Hp1). Given the scarcity of previous evidence
about gender differences in vulgarity effects, we cannot formulate a hypothesis about the
moderating role of gender: we included this variable to explore whether vulgarity works better for
either male or female source, or for both.
In addition, drawing on Bradac et al. (1980), we expected that vulgarity would increase
perception of language informality, as well as language intensity, that in turn would positively
affect the evaluation of the source and finally the behavioral intention (i.e., two step model; Hp2).
Then, in order to investigate whether people are aware of the expected effect of profanity,
we also checked if receivers perceived a vulgar political message as more/less persuasive in respect
to a neutral one.
One hundred and ten Italian adults (58.4% females) aged 20-68 years (M = 29.46, SD = 9.72) were
recruited through snowball sampling. They were highly educated (40%), predominantly
progressives (50.5% placed themselves on the left side of the 10-step left-right dimension), and
mostly workers (47.5%) and students (41.6%). Participants were asked to complete an online
questionnaire about political communication.
Design and Procedure
Participants were randomly assigned to conditions in a 2 (vulgar vs. neutral message) × 2
(source’s gender) between-subjects factorial design. After having answered three questions about
involvement in politics (Cronbach’s α = .85), participants read the following instruction: “Now,
imagine that in a few months a local election will be held in your town. While exploring the web, in
order to collect information about the candidates, you find the following post on the blog of one of
them.” They were then invited to read carefully a fictitious 15-lines blog post allegedly written by
the candidate her/himself dealing with the issue of unemployment. A pre-test confirmed that
receivers did not systematically associate the post content to left- or right-wing rhetoric or issue
ownership. The name of the blogger was manipulated as to refer either to a man or to a woman
(Mario vs. Maria Gambettini). The post was the same in the four conditions except for two swear
expressions (“a situation that pissed off everyone” and “is up shit creek”) appearing only in the
vulgarity conditions, as opposed to the neutral expressions (“a situation that worries everyone” and
“is in a tragic situation” included in the control condition.
Then participants answered a series of questions about the perceived persuasiveness of the
piece, their impression about the candidate, their likelihood of voting for him/her, and their
evaluation of his/her language. Finally, they were asked to write down what they remember of the
post, reported their position in the left-right political spectrum and completed socio-demographic
Language Perception. A scale for the evaluation of language informality and intensity was
developed for this investigation. Participants had to say if each of 6 adjectives describes the post
language from 1 = not at all to 7 = very much. An exploratory factor analysis elicited two factors,
explaining 64.99% of the variance. Thus, two indexes were calculated: informality (as a mean of
three items: direct, explicit and formal, reversed score, Cronbach’s α = .79), and intensity (as a
mean of the other three items: enthusiastic, encouraging, and enjoyable, Cronbach’s α = .63). As a
manipulation check, participants were also asked to what extent the candidate’s language was
vulgar (same response scale).
Impression of the Source. Participants evaluated, on a 7-point response scale (see above),
how much they think each of six adjectives (sincere, reliable, dishonest, skilled, qualified, and
uninformed) describes the author of the blog post. The exploratory factor analysis elicited a one
factor solution accounting for 49.08% of the variance (negative adjectives reversed), thus we built a
source general impression index (Cronbach’s α = .75).
Likelihood of Voting for that Candidate. Two questions asked participants the likelihood
with which they would vote for the candidate if they were members of this town constituency: “If
you had to vote in this town election, how likely were you to vote for Mario/a Gambettini?” “What
is the probability that in the next election, Mario/a Gambettini would be your voted candidate?”
(range 1-7, r = .90).
Perceived Persuasiveness. Participants rated the message persuasiveness through four
adjectives (persuasive, efficacious, credible, convincing; Cronbach’s α = .90) on the same 7-point
Recollection of the Blog Post. Participants’ recollection of the post they read was coded as a
score ranging from 0 (no answer or very bad memory) to 4 (very good memory) by a judge blind of
the experimental conditions.
The effect of participants’ gender, education, involvement in politics and self-reported position on
the left-right political spectrum have been tested in all analyses; they did not yield any signiﬁcant
main or interaction effects and therefore will not be presented.
The independent sample t-test performed on the perceived language vulgarity confirmed that
participants considered the message with the two swear words more vulgar (M = 4.24, SD = 2.11)
than those in the control condition (M = 1.37, SD = .71), t(106) = 9.48, p < .001.
Effects of Vulgarity and Gender of the Source
Table 1 reports descriptive statistics for our measures.
We first carried out a series of ANOVAs including the two independent variables
(candidate’s gender and vulgarity manipulation) on each dependent variable (Table 2).
The candidate’s gender, F(1,105) = 23.02 p < .001, η2p = .18, and the vulgarity
manipulation, F(1,105) = 87.52 p <.001, η2p = .45, main effects emerged on the perception of
language informality. Globally, the language was perceived as more informal when the message
was attributed to the female candidate than to the male one, and the vulgar message was perceived
as more informal than the control message. The interaction effect also emerged, F(1,105) = 6.02, p
= .016, η2p = .05, indicating that, though for both male and female candidates the vulgar message
was perceived more informal than the control message, the impact of swearing was greater when
the source was a man, F(1,54) = 69.48, p < .001, η2p = .56, rather than a woman, F(1,51) = 23.99, p
< .001, η2p = .32.
Regarding impression of the source, the main effects of candidate’s gender and vulgarity
manipulation as well as their interaction emerged. The female candidate inspired a better impression
as compared to the male candidate, F(1,106) = 46.51, p < .001, η2p = .30; furthermore, the vulgar
message globally induced a more positive impression about its author as compared to the control
one, F(1,106) = 12.62, p = .001, η2p = .11. However, the interaction effect, F(1,106) = 14.16, p <
.001, η2p = .12, revealed that the vulgar message induced a significantly better impression only in
the case of male candidate, t(55) = 5.03, p < .001, whereas when the politician was a woman the use
of swear words did not influence the judgment, t(51) = .154, p = .88.
Our independent variables did not influence participants’ perception of language intensity
and likelihood of voting for the candidate.
The same ANOVA performed on the message perceived persuasiveness yielded an
interaction effect, F(1,106)= 6.80, p = .01 ,η2p = .06: The vulgar message authored by the man was
considered less persuasive than the neutral one, t(55)=2.03, p = .047, whereas in the case of a
female source this difference did not reach statistical significance, t(51) = 1.66 p = .10. In addition,
the woman’s message was perceived as more persuasive than the man’s only in the vulgar
condition, t(53) = 2.29, p = 0.3, but not in the control condition, t(53) = 1.38, p = .174.
Indirect Effect of Vulgarity on Voting Intention
Based on the results illustrated above and in order to test our hypothesis 2, we verified a two-step
model in which vulgar message induces perception of language informality1 that in turn favors a
good impression about the author that finally increases the likelihood of voting for the candidate. To
this end, we ran Model 6 of PROCESS, the SPSS Macro by Hayes (2013), setting 5000
bootstrapped samples. This analysis allows testing three indirect paths: the first includes only
informality of language as the intermediate factor, the second includes only impression about the
candidate as the intermediate factor, and the third include the complete sequence depicted in Figure
1. Vulgar message was coded 1 and compared to control condition (coded 0).
The only significant model is the one including the complete sequence of factors we
considered (indirect effect = .27, SE = .13; LLCI = .07; ULCI = .59). All the path coefficients in the
model are positive.
Finally, the ANOVA including our independent variables on participants’ recollection of the
post confirmed that the independent factors did not induce significant variations, neither alone nor
in interaction, in the recall of the message content. This rules out the possibility that the observed
path was due to the ability of swear words to focus receivers’ attention on message content.
The results of our study shed light on the process through which vulgarity embedded in a
public political speech, even though socially sanctioned, is able to positively affect receivers’
behavioral intention. In line with hypothesis 1, we showed that a candidate’s use of swear words in
a blog post, without heightening receivers’ attention, increased language perceived informality and
improved the impression about the source. The latter effect was particularly strong when the
candidate was a man, as the woman candidate was already evaluated more positively, irrespective
of her cursing. In addition, though the manipulation of the politician’s vulgarity did not directly
affect the likelihood of voting for him/her, we found that not only the impression about language,
but also the impression about the source played the role of intermediate factor. This is apparently
incongruent with the study by Scherer and Sagarin (2006), but it is worth noting that the present
study differs from it in two main aspects. First, in the Scherer and Sagarin paper, the persuasive
message was given in speech form, while in our experiment we used a (written) blog post.
Participants in the two situations may have formulated different inferences about the source
intentionality, as saying a swear word can come across as more impassioned than to write it in a
blog post. In the latter case, the author has time to think about the wording used, put the swear word
in the text and kept it in future edits. Thus, it may seem as a planned choice and that may have
induced the reader to make an internal attribution to the author.
Second, Scherer and Sagarin operationalized the persuasive effect as receivers’ final attitude
toward the message issue. However, in the political domain the crucial output is the receivers’
global propensity to support the candidate in case of election. This is why we chose to include a
variable closer to the behavior, such as the likelihood of voting for the candidate. This specificity of
the political domain makes the present contribution particularly valuable, as it provides a first
evidence that when (rephrasing Marshall McLuhan) the source is the message and language
informality is perceived as a positive surprise, swearing can be effective in gaining consensus.
Even though we are inclined to interpret our results in terms of strategic use of a rhetorical
device (i.e. people swear not only for expressing emotions but also because, in the course of
experience, they have associated this behavior with its communicative efficacy), we cannot claim
that this is a conscious and deliberate strategy. Indeed, in line with Rassin and van der Heijden
(2005), we found friction between what participants generally think about the persuasiveness of a
vulgar message and their judgments when actually confronted with such a message. Indeed, in our
data, the vulgar message delivered by the male politician was at the same time the most influencing
and the one considered the least persuasive. Therefore, the association between swearing and its
positive consequences may be automatic and unaware, thus contrasting with self-reported beliefs.
Still, there are some reasons to be cautious when interpreting the ﬁndings from this study.
First, the results are based on a single experiment in a single context. In line with expectancy
violation theory, we cannot generalize the persuasiveness of a vulgar message: We believe that our
results emerged from a context (a blog post) that allowed such language to be perceived as
appropriate. For example, such language is explicitly sanctioned when used by a politician in the
Parliament. Second, our vulgarity manipulation was relatively minor and the manipulation check
revealed that, even though a significant difference in terms of vulgarity was perceived between the
control and the experimental condition, our participants considered the message with the two swear
words a moderate vulgar communication. This could be due to the current high familiarity with this
kind of language. We followed previous experimental studies that included one or two swear words
as well (Bostrom et al. 1973, Johnson, 2012; Scherer and Sagarin, 2006), but we cannot exclude a
non-linear effect. Further research is needed to test if an even more vulgar communication would
instead decrease message effectiveness. Third, respondents were given limited information about a
ﬁctitious politician and, participants did not know the party affiliation of the blog author. This may
have exaggerated the salience of the two swear words included in his/her message. In sum, future
research should study long-term effects of multiple vulgar messages in different contexts, beyond
our minimalistic experimental setting.
Notwithstanding its limitations, our study suggests that the appropriate use of swear words
by politicians actually works and highlights that it works through perceived language informality
that gets the politicians closer to people. In addition, the positive effect of cursing seems to apply to
the whole electorate, as this effect did not vary as a function of participants’ gender, education,
involvement in politics and self-reported position on the left-right political spectrum. These findings
outline the importance of studying the influence effect of vulgar language in order to acknowledge
that cursing should not to be seen only as an expressive behavior, since it also fulfills a strategic
communicative function. As such, our study extends the few previous results to the political field
where this has become an increasingly widespread practice.
We wish to express our gratitude to Chiara Gussoni for collecting the data, to the editor Howard
Gilles and the four anonymous reviewers for their constructive and detailed comments.
1As language intensity was not affected by message vulgarity, we did not include it in the
Barisione, M. (2009). Valence image and the standardisation of democratic political leadership.
Leadership, 5, 41-60.
Bayard, D., & Krishnayya, S. (2001). Gender, expletive use, and context: Male and female
expletive use in structured and unstructured conversation among New Zealand university
students. Women and Language, 24, 1-15.
Bordignon, F., & Ceccarini, L. (2013). Five stars and a cricket. Beppe Grillo shakes Italian politics.
South European Society and Politics, 4, 427-449.
Bostrom, R. N., Baseheart, J. R., & Rossiter, C. M. (1973), The effects of three types of profane
language in persuasive messages. Journal of Communication, 23, 461–475.
Bradac, J. J., Bowers, J. W., & Courtright, J. A. (1980). Lexical variations in intensity, immediacy,
and diversity: An axiomatic theory and causal model. In R. N. St. Clair & H. Giles (Eds.),
The social and psychological contexts of language (pp. 193-223). Hillsdale, NJ: Erlbaum
Burgoon, J. K. (1993). Interpersonal expectations, expectancy violations, and emotional
communication. Journal of Language and Social Psychology, 12, 30-48.
Fine, M. G., & Johnson, F. L. (1984). Female and male motives for using obscenity. Journal of
Language and Social Psychology, 3, 59-74.
Foote, R., & Woodward, J. (1973). A preliminary investigation of obscene language. Journal of
Psychology: Interdisciplinary and Applied, 83, 263-275.
Ginsburg, H. J., Ogletree, S. M., & Silakowski, T. D. (2003). Vulgar language: Review of sex
differences in usage, attributions, and pathologies. North American Journal of Psychology,
Hamilton, M. A. (1989). Reactions to obscene language. Communication Research Reports, 6, 67-
Hayes, A. F. (2013). Introduction to mediation, moderation, and conditional process analysis. New
York: The Guilford Press.
Jay, T. (2000). Why we curse: A neuro-psycho-social theory of speech. Amsterdam: John
Jay, T. (2009). The utility and ubiquity of taboo words. Perspectives on Psychological Science, 4,
Jay, T., & Janschewitz, K. (2008). The pragmatics of swearing. Journal of Politeness Research, 4,
Johnson, D. I. (2012). Swearing by peers in the work setting: Expectancy violation valence,
perceptions of message, and perceptions of speaker. Communication Studies, 63, 136-151.
Johnson, D. I., & Lewis, N. (2010). Perceptions of swearing in the work setting: An expectancy
violations theory perspective. Communication Reports, 23, 106-118.
Paradise, L. V., Cohl, B., & Zweig, J. (1980). Effects of profane language and physical
attractiveness on perceptions of counselor behavior. Journal of Counseling Psychology, 27,
Rassin, E., & van der Heijden, S. (2005). Appearing credible?: Swearing helps! Psychology, Crime
and Law, 11, 177-182.
Scherer, C. R., & Sagarin, B. J. (2006). Indecent influence: The positive effects of obscenity on
persuasion. Social Influence, 1, 138-146.
Wajnryb, R. (2005). Expletive deleted: A good look at bad language. New York: Free Press.
Table 1. Descriptive Statistics for Measures and Intercorrelations.
1. Language informality
2. Language intensity
3. Impression about the
4. Likelihood of voting
for that candidate
Note: *p < .05 level; ***p < .001.
Table 2. Effects of Inclusion of Swear Words in a Political Message on Language Evaluation,
Impressions about the Author, Behavioral Intention, and Perceived Persuasiveness (means, standard
deviations in parenthesis).
voting for the
Notes: Scales range is 1–7 with higher values indicating greater informality, greater intensity, better
impression, higher likelihood of voting for the candidate and higher perceived persuasiveness.
Means in a same row that do not share subscripts differ at p < .05.
R2 = .11, F(3, 104) = 4.37, p = .006.
Figure 1. The final model.
Notes: Path coefficients are b. ***p < .001; **p < .01.
Nicoletta Cavazza is associate professor in Social Psychology at the Communication and
Economics Department of the University of Modena and Reggio Emilia (Italy). Her current main
research interests include attitude change, political psychology, persuasive communication, social
aspects of eating. She has published books and scholarly articles in European Journal of Social
Psychology, Personality and Social Psychology Bulletin, Social Influence, Journal of Social
Psychology among others.
Margherita Guidetti (PhD, University of Modena-Reggio Emilia) is an adjunct lecturer of Social
Psychology at the University of Modena-Reggio Emilia. Her research interests mainly focus on
health behaviour, explicit and implicit attitudes, social influence, social judgement and reputation
management. She has published peer reviewed articles in Appetite, Psicologia Sociale, British
Journal of Health Psychology, Social Influence and Journal of Social and Clinical Psychology.