Emotional Politics on facebook . An Exploratory Study of Podemos ’ Discourse during the European Election Campaign 2014

The results of the European elections 2014 in Spain were characterized by the outstanding rise of a new party, Podemos, which obtained five seats in the European Parliament, despite being founded few months before the elections. The present study analyzes both the content and the presence of emotions in Podemos’ discourse on Facebook during the European electoral campaign. In particular, the affective content of both the party’s discourse and the comments of its followers will be analyzed through a pragmatic linguistic approach applied to a corpus of 163 posts and 215 followers’ comments. Results show an insistence on positive emotions in the party’s discourse and a prevalence of negative emotions in the comments of the citizens. keywords: political discourse, European elections, Facebook, emotions, Podemos

rise in popularity translated into 8% of the total votes and five seats in the European Parliament.
This study aims to analyze the party discourse on Facebook during the European campaign 2014, focusing on the posts published on the party's Facebook profile and paying special attention to the emotional content of both the posts and the comments published by Podemos followers.2

Internet and electoral campaigns
Political communication research has shown that electoral campaigns in Western countries are undergoing a process of modernization, whose main objective is to achieve the maximum level of electoral effectiveness through a series of techniques imported from corporate marketing (Maarek, 2009) and different refined ways of attracting media attention (Swanson and Mancini, 1996;Gibson and Römmele, 2007).
Since the emergence of the Internet in the 90s, political parties have been adapting themselves to the new tools provided by the Web 2.0 to the rhythm of the technological innovations and the process of citizen appropriation, in such a way that each election has been accompanied by numerous technopolitical innovations: blogs, video channels on YouTube, social networks, etc. (Dader, 2009).Thus, while the Internet seemed to give parties a platform to recover the political message (Bimber and Davis, 2003), i.e., to bypass the media machinery and reach militants and voters through websites (Sey and Castells, 2006;Chadwick, 2006), the Internet has been gradually incorporated as another element of the overall campaign strategy.
Despite the technological optimism that accompanied the expansion of the Internet, several studies about the partisan use of Internet generally suggest that parties have used these tools as a forum to deploy their persuasivestrategic discourse and attract media attention, rather than to recover a di-rect link with voters (Norris, 2003), 3 especially in the Spanish context (Dader, 2009;Dader et al., 2011).In this way, the accumulated experience seems to confirm the normalization hypothesis, namely, the idea that the technopolitical practices of the Internet reproduce the main features of the offline campaigns (Druckman, kifer and Parkin, 2010;Schweitzer, 2009).In other words, political actors are using the Internet mainly in order to foster marketing purposes, not to stimulate political pedagogy or to promote debate among citizens.But the political use of the Internet is not limited to electoral processes.There is, in fact, an intense academic debate on the impact of digital communication on the form and practice of politics in Western democracies.Scholars generally agree that the Internet has notably opened bottomup communication channels in political processes (Gibson and Römmele, 2007;Castells, 2009).More specifically, icTs have been conceived as a tool to face the growing citizen dissatisfaction towards democracy (Bentivegna, 2006), since social media allow for the direct contact between political actors and citizens and provide for new ways of political engagement.In that sense, cyber-optimist scholars claim that the Internet has positive effects on political participation (Lévy, 2004;Jenkins, 2008), especially intensifying the participation of people already interested in politics (Dader, 2001(Dader, , 2003) ) or even reaching new groups (Rheingold, 2004).For their part, cyber-pessimists highlight the resistance to change and democratization of political elites and the lack of interest of the majority of the citizenry in politics (Davis, 2001;Anduiza et al., 2010).
Over the last decades, several authors have emphasized that the study of emotions may help understand social behavior and collective action processes.
According to Collins (1999), emotions bind society and motivate initiative in social interaction.
Different scholars point out the crucial role of emotions in constructing the identity of social movements (Melucci, 1995;Goodwin, Jasper and Polletta, 2001), since they play a key role in gathering people together, providing a sense of group identity and promoting collective action to achieve specific purposes.This crucial role of emotions does not imply that social movements' members are simply irrationally driven or have no long-term commitment to it.Even though the emotional energy is the primary attractor, there are people who internalize these emotions, and who are strongly motivated and committed to the cause in rational terms.
According to Collins (2001), the collective rituals that take place within social movements generate two different emotional transformations.At the time of the movement formation, emotions tend to be mostly negative, since their members experience certain aspect of social life as deeply unfair and therefore mobilize to design a plan for action.In fact, one of the means for social movements to defy their enemies is the recreation of this initial vigor.Then, however, the group generally transforms the negative feelings into positive emotions, which strengthen the sense of group identity and solidarity.
This study constitutes an exploratory research into the presence of emotions in electoral discourse through social media.We assume that emotions were particularly important in Podemos' discourse during the European election campaign for different reasons.First, the party emerged a few months before the elections with a strong heritage of different organizations, such as the platform pah (Plataforma de Afectados por la Hipoteca),5 or the 15M Movement, etc.Second, researchers have pointed out the prevalence of persuasion and emotional appeals in the use of icT by political parties during electoral campaigns.
The research questions that guided this study are the following: What kind of content prevails in Podemos' political discourse on Face- Is there a relationship between the content of the posts and the emo- tions conveyed by them?

mEthoDology
This study analyzes the discourse of Podemos and its followers on the party's Facebook profile during the European campaign 2014 through two methodological approaches.First, content analysis is applied to all the posts published during the campaign in order to assess what types of issues were specially emphasized by Podemos.Second, the presence of emotions in both the posts and the comments is analyzed through the means of a pragmatic linguistic approach.

Sample and time frame
The corpus consists of all the posts published by Podemos in its Facebook profile during the official time frame of the electoral campaign in Spain (9 th to 23 rd May 2014).These posts with all their respective comments were retrieved and saved by means of the research software NodeXLGraph. 6In total, 163 posts and 7578 comments were collected.The analysis focuses on all the posts published by the party (n=163) and a sample of the comments.In order to select the sample, five posts were randomly chosen (post number 7, 56, 80, 93 and 161) and all the corresponding comments (n=215) were included in the analysis.

Content analysis
The content of the posts was categorized according to the classification scheme of electoral discourse proposed by López and adapted by Valera and López (2014).It includes four possible categories: Ideological and programmatic issues: principles, values, ideological po-1.
Government management (any information evaluating the perfor-4. mance of a concrete government or official).
Two raters classified the content of a sample of 50 posts (30% of the sample).Inter-rater reliability was calculated using Cohen's kappa (k) statistics to correct for chance agreement.The overall agreement was adequate enough (k=0.62).As a consequence, the content rating was continued only by one rater.
In addition to this methodological design, we also analyzed the deliberative nature of the posts published by Podemos, in order to observe if its Facebook profile was actually being used for citizen deliberation.We did so through a concept elaborated by Valera (2014b) named "political density" or "deliberability", which refers to discourse that contains any statement that can be rationally debated.Generally, any post including an ideological or programmatic statement can be rationally discussed and argued among users (see example 1), while other posts including information about electoral events cannot (see example 2).
To assess the agreement among coders, inter-rater reliability was calculated using kappa (k).The results showed that raters were in moderate agreement (k=0.52).Afterwards, the 42 posts in which raters disagreed were jointly recoded and three major criteria for the subsequent coding of "deliberability" and "non-deliberability" were established.

linguistic analysis
Regarding the study of emotions in political discourse, we propose a linguistic approach.Generally, two different methods have been applied to the linguistic study of emotions: sentiment analysis and pragmatic analysis.While the first tries to extract emotions through the retrieval of affective words (e.g.happy, sad, afraid, and so on) through different means (keyword spotting, lexical affinity, statistical methods, etc.), pragmatic analysis studies emotions in the context of a specific discourse.
Indeed, pragmatic research considers that «emotional words are only one way of grammatically codifying emotions in language» (Bazzanella, 2004).As a consequence, emotional language is always analyzed in the overall context of the discourse.Caffi and Janney (1994) identified six ranges of emotional devices in language, which have been successfully applied to the study of computer-mediated discourse (Laflen and Fiorenza, 2012;Vandergriff, 2013).
We chose a pragmatic approach for this exploratory study of emotions in political discourse for multiple reasons.On the one hand, Facebook posts are usually short, so the shortness may reduce the presence of explicit emotional words.On the other hand, emotional content may be masked under the metaphors commonly used in political discourse.As a consequence, in the current study the presence of emotions in discourse is analyzed using Caffi and Janney's (1994) methodology, especially focusing on two types of emotion markers: evaluation devices and proximity devices. 9he first category is based on the central distinction between positive and negative evaluations.It includes «all types of verbal and nonverbal choices that suggest an inferable positive or negative evaluative stance on the part of the speaker with respect to a topic, part of a topic, a partner, or partners in discourse» (Caffi et al., 1994, p. 354).For example, a negative evaluation device frequently repeated by Podemos is the term casta (caste), addressed to the major Spanish political parties.The word casta (caste) is an emotionalladen word negatively marked.
Proximity devices are based on the central distinction between near and far and include all types of linguistic choices that create a "distance" between the speaker and the partners or topics (Caffi et al., 1994, p. 356).Among the most frequent devices in this category the use of deixis can be mentioned (personal pronouns, temporal markers, social proximity markers, etc.).Podemos made an attentive use of the first person plural pronouns (we, our).The inclusive 'we' creates proximity with the interlocutor and involve him in the political choices, a stratagem frequently used in political discourse (cfr.for example Wilson, 1990).Indeed, the name of the party itself, Podemos,10 is the first person plural of the verb poder (to can), which clearly echoes Barack Obama's 2008 presidential campaign slogan «yes, we can».
Therefore, the emotional content of the posts and the comments was analyzed paying attention to both evaluation devices (emotional-laden lexical choices) and proximity devices (use of pronouns and deixis to create proximity or distance with followers).The posts were analyzed indicating whether they conveyed emotions (yes or no).Posts containing emotions were subsequently classified according to the polarity of the emotions (positive, negative or both).To assess the agreement of the coders, inter-rater reliability was calculated using kappa (k).The results showed that raters in this study were in substantial agreement about the presence of emotions in the posts (k=0.68) and in moderate agreement regarding the polarity of the posts (k=0.48);due to the qualitative nature of the study, these results were considered adequate and posts in which raters disagreed were jointly recoded.

RESultS
In general quantitative terms, Podemos' Facebook use during the electoral campaign showed a continued activity on the profile.The party posted 163 posts in a period of 14 days (average 11,64 posts per day) and received 7578 comments in total (average 46,49 comments per post).leaders.This shows that despite the strong leadership and great visibility of its main leader, Pablo Iglesias Turrión, Podemos avoided to run a campaign based on this leadership in social media.It rather deployed a grass roots electoral strategy trying to appeal to and mobilize the ordinary citizen.
Finally, just 1% of the posts consisted on a critique to the management of the incumbent government.This result seems particularly striking, since most of Podemos' public discourse has consisted on a fierce and well-justified critique to the traditional political parties and their management on different levels of government.However, the results suggest that this rather negative strategy was substituted by a more constructive discourse during the campaign, as the analysis of emotions will point out later on.
Moreover, the analysis of the political density shows that the posts presented slight more "non-deliberability".Table 1 reveals that, in fact, 52% of the posts (84 posts out of 163) did not contain any statement that could be rationally discussed among the followers, but were centered on specific campaign information.

Absolute frequency
Relative frequency

Deliberable posts 79 48%
Non-deliberable posts 84 52% In other words, more than half of the posts published by Podemos during the campaign did not contain any statement that could translate into a substantial deliberation among the followers.Instead of promoting citizen discursive interaction, most of the content was designed to publish campaign information, such as concrete campaign events taking place all around the Spanish territory, or different ways to contribute to the party's funding.Still, 48% of the posts did imply taking stances that could be debated among citizens.
That is, the party's discourse during the campaign was characterized by a significant positive tone, as suggested previously by the content analysis.Moreover, 21% of Podemos' publications included both positive and negative emotions (Figure 2).

Relationship between topic and emotions
In addition to the analysis of the presence of emotions, we asked ourselves about the possible relationship between the type of content published by the party and the presence of emotions.That is, were certain types of content more likely to contain emotions?showed in Table 3, ideological and programmatic posts were far more likely to include emotional content in comparison to posts related to campaign issues.Indeed, ideological and programmatic posts conveyed emotions more frequently: 44 out 53 posts (83%) included affective content. 11The association of ideological and programmatic issues with the presence of emotions constitutes an interesting result, showing that the party exposed its essential ideological stances and programmatic proposals with the help of emotional devices in order to mobilize the citizenry.

the presence of emotions in the comments
Before analyzing the emotional content of the comments, it is worth considering some basic information about the posts that were randomly selected for analyzing their comments, as their content and emotional value may influence the subsequent comments (Table 4).The results of the analysis of the 215 comments clearly show a very different pattern in comparison with the posts published by the party (Table 5).Most of the comments contained no emotional content at all (154 out of 215), even though programmatic issues were debated, the hope for a political change repeated in the posts related to campaign issues and posts showed different emotional polarity.Most of users only commented once, and no real debates took place among several users, although two of the posts randomly chosen showed political density, thus theoretically opening the way to citizens' debates.Only less than one third (61 out of 215) expressed some emotion (Table 5).That is, the party deployed a much more emotional discourse in this social network than the citizens that got involved in the online political discussions.

table 4 Posts randomly selected for the analysis of their comments
In general, the expression of negative emotions in the comments by Podemos' followers refers to feelings of mistrust, cynicism and disaffection towards political elites.These negative comments, as previously exposed, are the most common, according to our results.For example, the 6 th comment of post number 7 (example 3), shows a negative orientation towards politicians: (3) You know, what bothers me is that we are people who have very clear ideas and when they try to deceive us, they think we're stupid and we believe everything they say without consequences, that then we will not remember anything.But that is over, we are not sheep following the herd [...]12 .
In contrast, positive emotional comments refer to feelings like enthusiasm and hope and they tend to be addressed to Podemos as a party and its potential to promote a political and social change in Spain.For example, the 32 nd comment of post number 80 clearly shows a positive orientation towards the party and its campaign, as shown in example 4: (4) Today the act in Alicante was fantastic.The thrill of feeling the people get together.
than numbers, I would talk about authenticity, and in Podemos' acts you can breathe genuine enthusiasm and excitement; it's moving to see so many people together believing in the same project 13 .

ConCluSIonS
The last two decades have witnessed an increasing interest of scholars in the use of the Internet in different political participation and mobilization processes.The growth and success of social networks have recently moved academic attention to these new communication platforms.In this article, we have analyzed Podemos' use of Facebook during the European election campaign, as well as the relationship between emotions and political discourse.
According to the content analysis, campaign issues clearly prevailed in the party's discourse on Facebook (such as information about multiple cam-paign events, or about different ways to financially support the party), while ideological and programmatic-laden contents were less frequent.This confirms that Podemos' use of the Internet was electorally driven, since they mainly used it as a tool to deploy their persuasive-strategic discourse, instead of trying to recover a link with voters or promote citizen deliberation (Dader et al., 2011).In that sense, our results with regards to the topics and the limited "deliberability" (Valera, 2014b) of the posts contribute to ratify the normalization hypothesis (Schweitzer, 2009;Druckman et al., 2010).
Despite the formal bias of political communication research when dealing with online political discussions, this study presents an exploratory research of the presence of emotions on the party's discourse.Emotions are considered important in the emergence and strengthening of political and social movements (Collins, 2001), such as the 15M Movement (Belli and Díez, 2013), the Obama's campaign 2008 (Castells, 2009), or the MAS movement in Bolivia (Errejón, 2012).
Following a pragmatic linguistic approach, this research has shown a significant presence of emotions in Podemos' discourse on Facebook.These emotions were mainly positive during the campaign, since the party repeatedly appealed to feelings like enthusiasm and hope for a political change to promote mobilization.These results are in big contrast with Podemos' public discourse before the campaign, which consisted on a well-justified and systematic critique of the traditional Spanish political parties.
Moreover, our results reveal that ideological and programmatic posts were more likely to contain emotions, thus confirming the idea that emotions play an important role in processes of political and social mobilization, especially in the early stages of the building of a new political and social movement (Collins, 2001).But there were also some negative emotions found in Podemos' discourse, mainly associated to Spanish traditional political parties.That is, Podemos' discourse during the campaign 2014 was consistently presented with an emotional tone, mainly positive (hope for a change and confidence on the party) but also negative (critiques and attacks against the "caste").
According to the analysis of emotions in the development of social movements elaborated by Collins (2001), movements start from negative feelings against people, institutions or organizations that threaten their main values, and then they move to more positive feelings, such as pride or hope.The negative emotions of the 15M Movement were addressed to traditional political actors and institutions, which were considered corrupted, degraded and inadequate to solve social problems.In that sense, our analysis of emotions in both the party's discourse and the comments shows a very interesting result and a marked discrepancy.
On the one hand, the results suggest that Podemos tried to convert these first negative emotions against traditional parties positive feelings during the campaign, placing itself in the political scenario as a source of hope and change, as argued by Collins (2001) in his conception of the two emotional stages of social movements.
On the other hand, just a minority of followers' comments had some emotional content and most of them showed a lack of political density.In their majority posts were simple repetitions of campaign slogans, or questions about practical campaign issues, and citizens rarely involved in real political debates.Moreover, emotionally laden comments mainly conveyed expressions of negative emotions, such as mistrust or cynicism, in stark contrast with the party's predominant positive emotional tone.It is therefore reasonable to assume that the party's support and its surprising electoral mobilization were achieved through the negative motivations of its followers.In other words, while the party was already trying to move to a more positive discourse that could transform the citizen disenchantment into social change, the electoral base of the party remained mainly at an earlier stage, expressing negative feelings against traditional political actors and politics in general.
In conclusion, this study constitutes a first attempt to analyze the presence of emotions in political discourse through social networks.Due to the exploratory nature of the study, several limitations should be noted.First, the qualitative nature of the method used to identify emotions hinders a generalization of the results.Even though inter-rater reliability was statistically calculated and considered adequate, a subjective bias in the analysis of the data remains.Another limitation refers to the size of the sample, especially in the case of the comments, which impose a prudent discussion of the results.Future research with bigger samples of comments will have to establish if these results are generalizable.Finally, future research will also have to study the presence of emotions in political discourse in non-electoral periods.Comparative studies could also analyze the emotional dimension of multiple parties' discourse.Moreover, the deliberative dimension of online political discussions could also be considered in more detail, through a more extensive analysis of followers' interactions and participation.