Annexe A
Tableau 1. Caractéristiques des partis
Annexe B. Livre de codes
Ce que nous codons
Nos données sont constituées de 3 564 publications Facebook que 24 partis politiques ont publiées dans leur timeline dans les deux semaines précédant les jours d’élection en 2014 (parlement de l’UE) et 2017 (élections du parlement national), ainsi que deux semaines aléatoires en 2018 (période non électorale) en France, en Allemagne et au Royaume-Uni.
Comment nous codons : quelques règles générales
- Explicitement : nous codons uniquement les expressions explicites – tout ce que disent les messages, nous le prenons au pied de la lettre. Nous ne faisons pas d’hypothèses sur les significations implicites, le contexte caché, ce qui est signifié « entre les lignes ».
- Individuellement : nous codons chaque message individuellement. S’il n’y a pas de référence explicite à un message précédent, chaque message se suffit à lui-même. Nous codons chaque message comme s’il s’agissait du seul message de la chronologie.
- Unité d’analyse : nous traitons chaque message comme une unité. Si un message contient plusieurs phrases qui s’opposent les unes aux autres (par exemple, une phrase au ton positif et une autre au ton négatif), nous codons « émotions, mais difficile de distinguer si elles sont négatives ou positives ». Nous ne codons PAS positif et négatif dans ce cas – car notre unité d’analyse est l’ensemble du message, et non chaque phrase.
- De manière approfondie : nous lisons chaque message avant de le coder. Puis nous le lisons à nouveau et nous codons. Puis nous le lisons une troisième fois et vérifions si nous avons codé correctement. Si nous ne comprenons pas le message, s’il contient du texte illisible (par exemple, uniquement des émoticônes ou des lettres sans sens), nous marquons la ligne d’une couleur et passons aux messages suivants.
- Binaire : nous codons 0 si une variable ne s’applique pas et 1 si une variable s’applique. Dans les cas où nous ne pouvons pas décider, nous codons 99. Nous ne codons 99 que dans des cas exceptionnels – 99 n’est PAS un code régulier. Tous les 99 devront être recodés à la fin, donc nous l’utilisons avec parcimonie. Le codage binaire signifie également que nous décidons uniquement si un élément est présent ou non – nous ne codons PAS progressivement, en mesurant le degré d’application ou non d’un élément. Il est possible que pour certaines catégories, tous les codes soient à 0.
Liste de variables
Variable 1_TONALITÉ
Nous mesurons la tonalité générale du message. La tonalité fait référence au contenu, c’est-à-dire aux éléments de contenu positifs (succès, réussite) ou négatifs (crise, échec). S’il n’y a pas de contenu spécifiquement positif ou négatif, nous codons neutre. Si la tonalité d’un message est ambivalente, contient du contenu positif et négatif ou est difficile à évaluer pour d’autres raisons, nous codons « émotions, mais difficile à distinguer si pos ou neg ».
Veuillez noter que la majorité des messages sont généralement rédigés sur un ton neutre.
- V1_1TONEneutre.
- V1_2TONEpositif (ex : « Scotland’s hospital A&E performance has been the best in the UK for 24 months » ; « SNP is largest party in Clackmannanshire Council »).
- V1_3TONEnégatif (ex : « Only the SNP will stand up for Scotland, including our fishing industry. The Tories will sell them out »).
- V1_4TONEémotions, mais difficile de distinguer si pos ou neg.
Variable 2_ÉMOTION
Nous mesurons l’émotivité globale du message. L’émotivité ne concerne pas le contenu, mais la manière dont un message est présenté : il présente et/ou évoque des émotions positives (c’est-à-dire le bonheur, la fierté, la satisfaction) ou négatives (c’est-à-dire la colère, la peur, l’anxiété, le stress, la haine), des sentiments envers un parti, un candidat ou un événement.
- V2_1EMOneutre.
- V2_2EMOpositif (ex : « Thank you to all our volunteers who have been campaigning to secure strong and stable leadership »).
- V2_3EMOnegative (ex : « On June 9th, this man could be Prime Minister. We can’t let that happen » ; « Tories think they can do what they like to Scotland and get away with it. Don’t let them »).
- V2_4EMOmotions, mais difficile de distinguer si pos ou neg (ex : « We are the victims of our own success and now we pick ourselves up and go on to further success in the future »).
Variable 3_STYLE
Cette variable se rapporte à la langue du message. Les populistes utilisent généralement un registre de langue direct, non formel et non institutionnel, le langage de « l’homme ordinaire », de « l’homme de la rue ». Le langage informel étant également un élément typique des publications sur les médias sociaux, le codage de cette variable est un peu délicat. Les partis et les hommes politiques atténuent généralement leur langage élitiste, en essayant de faire des déclarations courtes et accrocheuses sur les médias sociaux. Nous codons le style informel uniquement si la publication est exceptionnellement informelle, de style familier, utilisant l’argot de la rue ou des jeunes, des expressions vulgaires, des émoticônes ou des abréviations (ROLF, LOL).
Nous codons également si le message contient un langage xénophobe. La xénophobie se distingue généralement par une opposition à la culture étrangère, elle exprime une « peur profondément enracinée envers les étrangers » (Oxford English Dictionary), et une « peur de l’inconnu » (Webster’s).
- V3_1STYLE : est rédigé dans un style direct, non formel et non institutionnel.
- V3_2STYLEposting : contient un langage imprécis, des allusions, des jeux de mots et une rhétorique vide, des proverbes, des stéréotypes, des clichés et autres expressions de la « sagesse populaire ».
- V3_3STYLEposting : enfreint les règles du politiquement correct, fait preuve d’impolitesse pour distinguer un parti des autres partis (ex : « Why should our taxpayers’ money be used to molly coddle the whims of the EU? »).
- V3_4STYLEposting : contient un langage vulgaire pour atteindre les « gens ordinaires » (ex : « Theresa May arrogantly risked Brexit and blew it »).
- V3_5STYLE : contient un langage agressif et provocateur (ex : « I simply cannot understand the mentality of the pathetic geeks behind this crime »).
- V3_6STYLEposting : contient des propos xénophobes.
- V3_7STYLEposting : contient un langage ironique ou cynique.
- V3_8STYLEposting : contient un langage humoristique, drôle (du moins, il essaie d’être drôle).
Variable 5_INTERACTIVITÉ
Cette variable permet de mesurer si une publication invite à des commentaires, des activités et/ou des réactions.
- V5_1INTERencourager : encourage la discussion (avec n’importe qui ; ex : « Partagez pour que tout le monde soit au courant »).
- V5_2INTERencourager : pousse à l’action (avec n’importe qui ; ex : « Allez voter ! »).
- V5_3INTERencourager : appelle à un engagement plus poussé avec le parti (interaction avec le parti, retour d’information, participation ; ex : « Si vous voulez vous impliquer dans sa campagne, contactez-nous ! »).
- V5_4INTERcontient : contient des questions qui invitent à formuler des opinions (ex : « Que pensez-vous de nos idées ? »).
Variable 6_PEUPLE
Les populistes se réfèrent généralement au « peuple » en tant qu’acteur unitaire, prétendent être le seul et véritable représentant du « peuple » et demandent plus de pouvoir pour « le peuple ». D’autres termes synonymes incluent l’homme du peuple, le citoyen ordinaire, les vrais gens, l’homme de la rue, etc.
- V6_1PEOPLE : l’orateur fait référence au peuple comme étant l’origine théorique du pouvoir en démocratie.
- V6_2PEOPLE : l’orateur fait l’éloge des réalisations du peuple, souligne les vertus du peuple.
- V6_3PEOPLE : l’orateur demande plus de pouvoir pour le peuple et/ou promeut explicitement la mise en œuvre d’éléments de démocratie directe (ex : « “Die da oben” bestimmen über “uns hier unten”! Dieses Gefühl der Ohnmacht vieler Bürger will die AfD aufheben. Einer der wichtigsten Punkte in unserem Wahlprogramm ist deshalb dieser: “Wir wollen dem Volk das Recht geben, den Abgeordneten auf die Finger zu schauen und vom Parlament beschlossene Gesetze zu ändern oder abzulehnen” »).
- V6_4PEOPLE : l’orateur présente le parti comme le véritable et seul représentant des personnes réelles (ex : « There is only one leader who can provide strong and stable leadership for the UK through Brexit and beyond » ; « Only UKIP MPs can be trusted to speak out firmly and courageously on behalf of hard-working British families » ; « We want to build a Britain for the many, not the few »).
- V6_5PEOPLE : l’orateur présente les autres partis comme des non-représentants du peuple.
- V6_6PEOPLE : l’orateur présente le parti comme le défenseur du bon sens, de la sagesse des gens ordinaires dans un monde fou, et promet un retour au bon sens (ex: « Deutschland ist vollkommen verrückt geworden »).
Variable 7_ANTIELITISM
La communication populiste se range toujours du côté du « peuple », refuse le pouvoir aux élites (les élites ont trop de pouvoir) et blâme les élites (pour tout ce qui va mal). Attention : il est important de distinguer l’anti-élitisme de la simple critique oppositionnelle des acteurs au pouvoir, comme le fait de pointer du doigt les erreurs du gouvernement (ex : « Le gouvernement n’a pas fait assez pour de meilleures écoles »). L’anti-élitisme est global : tous les autres partis politiques et leurs politiciens sont mauvais, corrompus, à blâmer pour la ruine et l’effondrement du pays et de la société en général. L’élite est perçue comme un acteur unitaire, une clique qui se serre les coudes et s’oppose au peuple. Une attaque contre un autre parti n’est pas de l’antiélitisme, mais une partie normale des campagnes électorales.
- V7_1ATTACposting : attaquant les élites politiques comme les politiciens, le gouvernement, les autres partis, « Altparteien », la classe politique (ex : « Tim Farron and the Lib Dems will say anything to get elected. Their coalition of chaos would put the UK’s Brexit deal at risk » ; « Right now, the coddled elite are laughing at us, but that will change »).
- V7_2ATTACposting : attaquant les élites économiques telles que les banques, les riches, les courtiers en bourse (ex : « Theresa May will protect workers’ pensions from irresponsible bosses who put them at risk » ; « We don’t rely on the money of a handful of super-rich donors »).
- V7_3ATTACposting : attaquant les élites médiatiques, par exemple les journalistes, les éditeurs, les médias de service public.
- V7_4ATTAC : attaquant les élites supranationales comme l’UE, l’ONU, la Banque mondiale, d’autres puissances.
- V7_5ATTAC : attaquant les élites juridiques comme les tribunaux, les cours suprêmes, les avocats, les juges, les professeurs de droit.
- V7_6ATTAC : attaquant d’autres institutions, telles que les scientifiques/experts, l’armée, les écoles, l’église, les ONG.
- V7_7ATTACposting : attaquant des élites non spécifiées, par exemple l’establishment, l’État, l’officialité, le régime, la haute société.
Variable 8_OSTRACISM
L’ostracisme était une procédure sous la démocratie athénienne dans laquelle tout citoyen pouvait être expulsé de la cité-état d’Athènes. Le populisme tend généralement à identifier des groupes comme n’appartenant pas à la société, les marquant comme des outsiders. Ils affirment que ces groupes ne font pas partie du « peuple », traçant ainsi des lignes rouges, excluant et ségréguant certains groupes, tels que les migrants, les groupes religieux, les électeurs d’autres partis. Ils peuvent qualifier certains groupes de terroristes ou d’« ennemis de l’État ».
Attention : l’ostracisme concerne les personnes ou les groupes qui existent et vivent au sein de la société, et les exclut – pas les groupes extérieurs. Si une affiche, par exemple, affirme que les réfugiés ne font pas partie de la société, il ne s’agit d’ostracisme que si elle fait référence aux réfugiés qui sont déjà dans le pays ; si elle fait référence aux réfugiés dans les camps africains et demande de ne pas laisser entrer davantage de réfugiés, il ne s’agit pas d’ostracisme.
- V8_1OSTRposting : déclare un peuple monolithique (« le peuple », nous vs eux).
- V8_2OSTRposting : nomme des groupes que le parti ne considère pas comme faisant partie de la société.
- V8_3OSTR : accuse d’autres partis d’ostracisme (déclare par exemple qu’ils sont fascistes, populistes, etc.).
Variable 9_SYMBOLS
Le populisme fait souvent référence à des symboles, c’est-à-dire à une forme ou un signe utilisé pour représenter quelque chose comme une organisation, par exemple une croix rouge ou une étoile de David. Les symboles peuvent être des événements historiques, des drapeaux, des documents nationaux importants, des lieux, etc. Les symboles sont instrumentalisés pour conférer une légitimité au parti, ou pour critiquer un autre parti. Les symboles sont des signifiants, ils sont utilisés pour homogénéiser, pour que les partisans d’un parti se rassemblent derrière un drapeau, pour donner une identité et une légitimité à une cause politique.
- V9_1SYMBOLS : l’affichage contient un symbolisme (historique).
Variable 10_STORYtelling
La communication populiste contient souvent des récits alternatifs, racontant des versions « non officielles » de la réalité, semblant révéler des secrets non connus du public (par exemple parce que les « médias du système » les cachent). Il peut s’agir de « faits alternatifs », de la vraie vérité, de liens avec des théories du complot, etc. Le populiste a également tendance à se présenter dans le rôle du « perdant », de l’outsider, du David contre Goliath. In fine, les populistes se présentent comme des victimes, réprimées par les élites politiques et/ou médiatiques.
- V10_1STORYposting : raconte les « histoires non officielles », non connues du grand public, présente des faits ou des données pour contester ou réfuter les positions de tous les autres partis ou élites (ex : « Damit stellt sich die Umweltbilanz ganz anders dar, als von sämtlichen anderen Parteien propagiert » ; « Und teilen Sie diesen Beitrag, denn über diese Fakten werden die Leitmedien wohl kaum berichten » ; « Teilen : Das muss Deutschland wissen ! »).
- V10_2STORYposting : présente le parti comme un outsider (ex : « Im Gegensatz zu anderen Parteien haben wir keine Großspender »).
- V10_3STORYposting : présente le parti ou les politiciens du parti comme des victimes (ex : « Hexenjagd der Presse nimmt immer absurdere Züge an ! »).
Variable 11_FEAR
La communication populiste évoque parfois la peur et l’anxiété, en affirmant qu’une menace imminente pèse sur la société. Ces messages cherchent à alarmer et à mobiliser les gens, en suscitant des émotions ou des actions négatives. Les messages cherchent à mettre en garde, à alerter le public et peuvent faire référence à des dangers venant de l’extérieur (guerre, migrants), de loin (apocalypse du changement climatique), de certaines politiques (effondrement du système de protection sociale ou du système fiscal) ou à des dangers plus généraux sur la fin de la civilisation occidentale, la culture locale ou nationale, la patrie, la cohésion de la société, la tradition, la décence, etc.
- V11_1FEAR : message faisant appel à l’émotion de la peur pour alarmer ou mobiliser les gens sur un sentiment ou une action négative (ex : « WATCH and SHARE: Whoever wins this election will have to keep our country safe. Our security and Jeremy Corbyn – TOO BIG A RISK » ; « Jeremy Corbyn’s nonsensical and irresponsible ideas are a grave risk to Britain’s economic security and the finances of every UK family » ; « You can’t take tea with terrorists who want to kill us »).
Variable 12_SIMPLICITÉ
La communication populiste a tendance à simplifier à l’extrême des questions politiques complexes. Le codage de cette variable est délicat, car la simplification est également un élément typique de la communication sur les médias sociaux – et des campagnes électorales en général. Les messages postés peuvent suggérer des solutions simples aux principaux problèmes de la société. Le gouvernement ou d’autres partis ne font rien, alors que la vraie solution est si simple et de bon sens – comme le sait chaque homme de la rue. Les affiches promettent de réformer, de rendre la vie plus facile et plus simple, par exemple en permettant à l’avenir de faire une déclaration fiscale sur un sous-verre à bière.
Attention : si une solution simple est proposée, mais que l’affichage contient un lien promettant plus d’informations et de détails, nous ne la codons pas comme étant trop simplifiée (ex : « We’ll stand up for the many, not the few. How? Read our 2017 manifesto to find out more. Agree with our pledges? Share it now »).
- V12_SIMPL : simplifie les problèmes ou les solutions (ex : « Theresa May will support working families by capping expensive and unfair energy prices » ; « Whether you rent or buy, everyone needs the security of a place to call home – that’s why we will fix the broken housing market ».
Variable 13_TOPIC
Cette variable mesure le sujet de l’affichage : questions politiques, vie personnelle d’un candidat, événements d’actualité ou campagne électorale elle-même. Il est possible de coder « Autre », si le sujet d’un message ne correspond pas à ces éléments. Si c’est le cas, écrivez en un mot quel est le sujet. Nous ne codons « Autre » que très rarement, car tous ces mots doivent être recodés à la fin.
- V13_1TOPICmessage : sur les questions politiques qui doivent être résolues (souvent locales ou épisodiques, individuelles) et dont la résolution doit être rapide, par exemple les impôts, les transports publics, les soins de santé, la crise des migrants, le chômage. >> Continuer avec V14.
- V13_2TOPICmessage : sur la vie personnelle d’un candidat ou d’un politicien de l’autre parti.
- V13_3TOPICmessage : commentant les événements actuels, qui ne sont pas principalement politiques, mais d’actualité (sport, événements, nouvelles).
- V13_4TOPICmessage : sur la campagne (comment elle est gérée) et sur les performances pendant la campagne.
- V13_5TOPICmessage : n’est pas lié à la politique ou aux politiques (ex : « Happy birthday, Barack Obama »).
- V13_6TOPICOther.
Variable 14_DOMAINE DE POLITIQUE
Uniquement si V13_1 a été codé, c’est-à-dire que si l’affichage concerne une question de politique qui doit être résolue, nous codons le domaine de politique concerné.
- V14_1POLICY : politique économique (prix, monnaie, chômage, investissements, impôts, réglementation financière, banques, dette des consommateurs, inflation, questions de travail, prix du carburant).
- V14_2POLITIQUE : politique sociale (aide sociale, logement, sans-abri, pauvreté, pensions).
- V14_3POLITIQUE : politique de santé (soins de santé, hôpitaux, promotion d’un comportement sain, coût de l’assurance santé, santé).
- V14_4POLITIQUE : politique d’immigration (migrants, réfugiés, intégration, contrôle des frontières, immigration).
- V14_5POLITIQUE : politique d’éducation (jardin d’enfants, écoles, université, sciences).
- V14_6POLITIQUE : politique étrangère (diplomatie, guerre, relations avec les dirigeants étrangers, institutions internationales [UE]).
- V14_7POLICYBREXIT.
- V14_8POLITIQUE : politique environnementale (climat, scandale du diesel, énergie nucléaire, pollution atmosphérique, recyclage, énergie).
- V14_9POLITIQUE : politique de sécurité (criminalité, terrorisme).
- V14_10POLITIQUE : politique des transports (trains, transports publics, péages de rue).
- V14_11POLITIQUE : politique technologique (numérisation, fibre optique, régulation de l’internet).
- V14_12POLITIQUE : politique des médias (médias de service public).
- V14_13OTHER : si les champs 1 à 12 ne s’appliquent pas, tapez le champ de la politique en un seul mot.
Annexe C
Tableau 1. Populisme de contenu et populisme de style par parti
Tableau 2. Nombre de posts par parti pendant les deux périodes hebdomadaires d’analyse
Annexe D. Variables dépendantes et indépendantes
Variables dépendantes
Le populisme dans la communication est opérationnalisé en utilisant trois indices de populisme distincts pour le populisme dans le contenu (PiC : PiC vide, PiC exclusif, PiC anti-élitiste), suivant Clease de Vreese et ses co-auteurs (2018), et un indice pour le populisme dans le style (PiS), suivant Roberta Bracciale et Antonio Martella (2017). Il y a moins de littérature sur le populisme dans le style que sur le populisme dans le contenu, et notre opérationnalisation couvre également de nombreuses variables que Werner With et ses co-auteurs (2019) suggèrent dans leur livre de code du PRN sur le populisme et dans l’analyse des styles populistes de Martin Wettstein et ses co-auteurs (2019), comme l’émotionnalisation, le bon sens comme source et le langage familier.
Le populisme dans le style contient six caractéristiques :
- l’appel à l’émotion de la peur ;
- la simplification excessive des problèmes ou des solutions ;
- l’utilisation d’un langage non précis, avec des allusions, des jeux de mots et une rhétorique vide, des proverbes, des stéréotypes, des clichés et d’autres expressions de la « sagesse populaire » ;
- la rupture des règles du politiquement correct ;
- l’utilisation d’un langage vulgaire ;
- l’utilisation d’un langage agressif ou provocateur (2014 M=.142 SD=.43, 2017 M=.216 SD=.52, 2018 M=.185 SD.53; ALL M=.289 SD=.60, R-U M=.190 SD=.49, FR M=.021 SD=.15).
En ce qui concerne les domaines politiques, nous avons codé chaque message selon qu’il couvre ou non un sujet spécifique (en cas de sujets multiples, tous les sujets ont été codés comme présents). Les messages ont été codés selon douze thèmes : économie, questions sociales, santé, migration, éducation, politique étrangère, Brexit (Royaume-Uni uniquement), environnement, sécurité, transport, technologie et médias. Dans notre échantillon, dans 46 % (1661) des messages, au moins un des sujets ci-dessus était mentionné. Chaque message a été codé de façon binaire pour la présence ou l’absence des éléments de l’index. Ce codage n’est pas exclusif car les messages peuvent contenir plus d’un élément dans chaque index, par exemple un message anti-national, anti-gouvernemental et anti-UE.
Variable indépendante
Pour calculer les variables indiquant les caractéristiques des partis, nous utilisons le Chapel Hill Expert Survey (CHES). Les données de CHES 2014 et 2017 (et 2010 pour BNP) sont combinées (scores moyens ou score d’une année si un parti n’est étudié que dans une seule vague [par exemple LREM seulement dans CHES 2017]).
La famille de partis est une variable muette, codée séparément pour la gauche radicale (3 partis), la droite radicale (5 partis), le centre (6 partis), la gauche (3 partis), la droite (3 partis) et autres (6 partis, par exemple les Verts ou les partis régionaux).
Références
Bracciale Roberta et Martella Antonio, 2017, « Define the populist political communication style: The case of Italian political leaders on Twitter », Information, Communication & Society, vol. XX, no 9, p. 1310-1329.
De Vreese Claes H., Esser Frank, Aalberg Toril, Reinemann Carsten et Stanyer James, 2018, « Populism as an expression of political communication content and style: A new perspective », The International Journal of Press/Politics, vol. XXIII, no 4, p. 423-438.
Wirth Werner, Wettstein Martin, Wirz Dominique S., Ernst Nicole, Büchel Florin, Schulz Anne, Esser Frank, Weber Edward, Dalmus Caroline, Engesser Sven et Manucci Luca, 2019, « Codebook. NCCR Democracy Module II: The appeal of populist ideas and messages », https://doi.org/10.17605/OSF.IO/RYX42 (consulté le 1er juillet 2021).
Wettstein Martin, Esser Frank, Büchel Florin, Schemer Christian, Wirz Dominique S., Schulz Anne, Ernst Nicole, Engesser Sven, Müller Philipp et Wirth Werner, 2019, « What drives populist styles? Analyzing immigration and labor market news in 11 countries », Journalism & Mass Communication Quarterly, vol. XCVI, no 2, p. 516-536.
Populism as a communication phenomenon: A cross-sectional and longitudinal comparison of political campaigning on Facebook
Le populisme comme phénomène de communication : une comparaison transversale et longitudinale des campagnes politiques sur Facebook
El populismo como fenómeno de comunicación: una comparación transversal y longitudinal de las campañas políticas en Facebook
klinger@europa-uni.de
Professor for Digital Democracy
Stiftung Europa-Universität Viadrina
kkocmichalska@audencia.com
Professor at Audencia Business School
The paper provides an empirical verification of the concept of populism as communication phenomenon. It moves away from the actor-centered approach, where parties are perceived as populistic or non-populistic according to their ideological positioning. The populism as communication approach assumes that any party may present populist content when communicating with the voters. In the analysis, we employ the data from France, Germany, and the United Kingdom during three different periods: EU elections 2014 (second-order), national parliamentary elections 2017 (first-order), and a non-electoral period from January 2018. The paper is based on the 3,564 posts produced by political parties on Facebook during six weeks.
The article first contains a comprehensive discussion of the theories and proposes the hypotheses, it is followed by the method description of the data and results are presented. The last part is dedicated to the discussion of the results.
Populism as communication
Populism is everywhere: it is evident in the news, in public discourses, and in political communication research. Over the past 30 years, the support for populist radical right-wing parties rose to a historic high (Tartar, 2017). Populism has become a ubiquitous research topic – not only due to the upsurge of right-wing and left-wing populist leaders, parties, and movements but also because populism is a mesmerizing phenomenon. It is notoriously vague (Canovan, 1999), “immaîtrisable, incontrôlable et inutilisable” (Pranchère, 2020), promiscuous (Oliver and Rahn, 2016), chameleonic (Taggart, 2000), and an opaque zeitgeist (Mudde, 2004) at the conceptual level, and therefore poses an irresistible challenge to social scientists.
Recently, Claes H. de Vreese and his co-authors (2018, p. 3) evoked the notion of “a new generation of populism researchers”. In this section, we argue that an understanding of populism as communication (Mazzoleni and Bracciale, 2018) indeed enables researchers to bridge the current debates on whether or not populism is an ideology, a discourse, a mobilization strategy, an organizational type, or a political logic (Laclau, 2005; Moffit and Tormey, 2014; Zulianello et al., 2018).
In line with C. H. de Vreese and his co-authors (2018), we view populism as a communication phenomenon that includes typical elements of content (i.e. people, anti-elitism, and the exclusion of out-groups) and of style (i.e. evoking fear, oversimplification, and style of language). Accordingly, “the emphasis is on populist messages as independent ‘phenomenon-as-such’ and no longer on a particular party family or type of politician” (de Vreese et al., 2018, p. 3). We see populism not only as actors’ attributes, but mostly as actors’ actions. We do not view populism in terms of an either/or, populist/non-populist dichotomy, but as a phenomenon that varies in degree. As a consequence, the proposed paper does not concentrate on populism as an ideology or actor-centred approach (Mudde, 2004), on its ideational character (Mazzoleni, 2008; Pauwels, 2011; Hawkins et al., 2012), its critique as “thin ideology can […] become so thin as to lose its conceptual validity and utility” (Moffit and Tormey, 2014, p. 383) or on the populism “d’en-bas” (Ivaldi, 2018).
Populism, within the context of this paper, refers to three core elements of the content of communication: the people, anti-elitism, and the exclusion of out-groups.
- People: essentially, populism appeals to the people – not for the sake of the people as such, but in opposition to the values and ideas of elites and certain minority groups who do not belong to “the people”. As an element of populism, the people is a rhetorical construct that is instrumental and that can be exploited, and its meaning is not the same as the citizens or the population of a territory. As a united, homogenous group of ordinary people, the silent majority (Oliver and Rahn, 2016), the notion of the people is a fiction. In this view, the people are united, solidary, guided by common sense, endowed with the same interests, values and opinions, and can be invoked into many forms such as the nation, peasants, voters or the proletariat (Rooduijn, 2014). In populism, the people are in a default state of crisis, threatened by others from the outside, which is why populism has been associated with a Manichaean perspective and a dualist world view of Us vs. Them.
- Anti-elitism: similar to the people, the elite refers to a homogenous group with its identity based on antagonism directed at the people (Engesser et al., 2017a). Elites are found in the political, economic and legal systems, in the media, and in supranational institutions. The core idea behind anti-elitism is the attribution of blame. When conjuring up a people in crisis, elites are blamed as either unable or unwilling to represent the people’s will and to respect the people’s sovereignty. Populism’s opposition to complexity is connected to anti-elitism: “complexity is a self-serving racket perpetuated by professional politicians” (Canovan, 1999, p. 6) suggesting that policy should be guided by the people’s common sense that is actually pure and simple. The degree of antagonism also varies: populism always presents itself as distinct from elites but, rather than blaming or shaming elites, it may oppose other groups in society as well (Moffit and Tormey, 2014).
- Out-groups: some scholars contend that the exclusion of out-groups is not a key feature and element of populism as such, but only one of radical right-wing populism. We argue, along with C. H. de Vreese and his co-authors (2018), that the exclusion of out-groups is not a defining element of all variants of populism, but is one that features both on the right and the left of the populist spectrum. The exclusion of others constitutes the horizontal dimension of populism’s inherently antagonistic character. Elites are those in power and, therefore, the enemy from above; out-groups – being groups that populists stigmatize as a threat or a burden to the people (Jagers and Walgrave, 2007) – are the enemy from within society. Once more, the out-group is a construction of “a blameless in-group opposed to a culprit out-group” (Hameleers et al., 2018, p. 872). In this sense, Cristóbal Rovira Kaltwasser (2018) differentiates between exclusionary and inclusionary populism. While the former is often rooted in notions of nativism and right-wing ideologies that oppose immigrants, ethnic and religious minorities, homosexuals or welfare recipients, the latter focuses on radical left-wing positions and constructs a homogenous group out of those affected by alleged unjust socio-economic policies and austerity measures. Similarly, Michael Hameleers and Rens Vliegenthart (2020) differentiate left- and right-wing exclusion. In our operationalization, we do not make this differentiation of who is excluded but focus on the question of whether any groups are excluded (see in the codebook in Appendix B: V8_OSTRACISM).
Based on the contributions of Jan Jagers and Stefaan Walgrave (2007) and C. H. de Vreese and his co-authors (2018), we apply a typology of populism that differentiates between four levels of populism as a communication phenomenon and that to varying degrees can be empirically identified in political messages.
- Full populism: people + elite + out-group
- Anti-elitist populism: people + elite
- Exclusionary populism: people + out-group
- Empty populism: people
In this typology, the invocation of the people is at the core of populism and produces different types of populism depending on whether or not it is linked to the elite and out-group elements. Full populism combines all three elements. Anti-elitism or the exclusion of out-groups individually do not constitute populism; both only do so in combination with the people element. Similarly, the mere invocation of the people without reference to anti-elitism or out-groups is deemed to be empty populism, the thin, “empty-shell, initial definition” of populism (Jagers and Walgrave, 2007, p. 323).
Although, regarding content, people, anti-elitism, and the exclusion of out-groups form three “pillars of populist discourse” (Bobba, 2019, p. 12), we must also consider how populism manifests in the style of messages, that is how actors present ideas and information. As previously argued, populism is not about the attributes of a political actor, but about the actions of an actor. Populism is performed; it is what is said and how it is said (Bracciale and Martella, 2017). Also, performing populism is not limited to the political right or left, since “politicians can slip in and out of the populist style” (Moffit and Tormey, 2014, p. 393). The performative style of populism has been described as direct, emotional (Canovan, 1999), simple (avoiding complexity) and blunt. The populist transgresses the limits of presumed political correctness: “like a ‘drunken guest’ [...] with ‘bad manners’ [...], the populist disrupts the normal dinner table, much to the discomfort, even alarm, of the usual patrons” (Oliver and Rahn, 2016, p. 191). Populism is emotional and evokes emotions by purposefully breaching the taboos of mainstream politics and political culture, by employing calculated provocations (Pauwels, 2011), and by taking on a narrative of underdogs (Mazzoleni, 2008), of self-victimization, or other symbolic themes. These emotions emphasize fear and anger (Hameleers et al., 2017).
To address these issues and to conceptualize populism as a communication phenomenon, we study both populism in content and populism in style. Both aspects vary in degree, as actors from across the political spectrum can employ populist content and style elements in their public communication. As James Stanyer and his co-authors point out:
It is important to reiterate that most studies are actor-centered and that we lack systematic empirical studies. Consequently, our knowledge is poor about how frequently both populist and non-populist mainstream political actors refer to the people, express anti-elitism, and exclude various out-groups in their communication. (2016, p. 361)
Populism in election campaigns
Recent empirical studies have begun to address populism as a communication phenomenon, also considering that the proliferation of social media platforms allows access to populist communication with an additional type of data. Whereas previous research was limited to highly formalized genres like public speeches and party manifestoes, social media enable scholars to monitor the use and effects of political communication in a more granular fashion. Social media have become standard platforms for election campaigns, where parties and politicians broadcast their messages unfiltered by journalists or other gatekeepers. Social media posts are less formalized than party platforms, the way content is produced, distributed and used is very different from traditional mass media (Klinger and Svensson, 2015). In the fourth era of campaigning (Roemmele and Gibson, 2020), social media posts provide numerous and fertile material to study the use of populist messages and populist elements in party communication.
Not surprisingly, populism has become a popular research topic of political communication scholars interested in the impact of social media platforms. Although the data is proprietary and only partially accessible through APIs, it has nonetheless brought about an increase in quantitative empirical research designs (Ernst et al., 2017; Stier et al., 2017; Zulianello et al., 2018; Bobba, 2019). This is all the more relevant as experiments show that populist elements have mobilizing and de-mobilizing effects on voters (Hameleers et al., 2018).
Social media offer a communication environment that significantly differs from traditional mass media outlets. The affordances of social media platforms provide a particularly fertile ground for populist communication, being mostly un-edited, enabling public communication outside of and circumventing journalistic outlets. On social media, political parties remain in control of their messages, they can tailor messages according to target groups or use micro-targeting tools for political advertisements (Kreiss, 2016). It is popularity and not news values or other professional criteria that determine the relevance and reach of a message. Indeed, there is empirical evidence that purported populist parties have profited from social media (Bobba, 2019), that they find new supporters among adolescents (Heiss and Matthes, 2017), that populist messages receive more comments online and “prompt citizens to use populist messages themselves in their comments” (Blassnig et al., 2019, p. 629).
However, previous studies have shown that social media platforms are by no means a “populist paradise” since so-called populist parties often have a centralized structure and avoid internal dissent (Jacobs and Spierings, 2019). A study on Twitter use by populist presidents in Latin America found that the prevalent communication mode remains top-down (Waisbord and Amado, 2017), confirming that populist communication on social media tends towards normalization/politics as usual (Koc-Michalska et al., 2021). In their social media practices, populist actors are not always and necessarily more apt to respond and engage with the people (Spierings et al., 2019).
While social media data call for comparative analyses, only few empirical studies have made comparisons across countries and even less so across time. Matthijs Rooduijn (2014) set an example by using data obtained from newspaper articles, comparing five countries where, over two decades (1988-2008), purported populist parties (at the time) had varying success. Nicole Ernst and her co-authors (2017) analyzed 1,400 Twitter and Facebook posts from 6 countries, published by 88 politicians from 29 political parties. They found that parties at the fringes of the political spectrum and opposition parties were more prone to populist communication. In another study on politician statements from six Western democracies, N. Ernst and her co-authors (2019) found that populist communication is associated with typical issues (such as immigration, crime and economic hardship), and that politicians from parties considered as populist indeed are more populist in their communication. They particularly stress the importance of studying the content and style of populist messages, claiming that “the ideology of populism cannot be communicated without stylistic elements” (p. 167). S. Engesser and his co-authors’ (2017a) comparative qualitative text analysis showed that “populism manifests itself in a fragmented form on social media” (p. 1109), justifying the need for a differentiated and comparative look at party messages on Facebook or Twitter. S. Engesser and his co-authors (2017b) provided a useful operationalization, based on emotionalization and negativity (p. 1282), to identify variations in content and style. Even though the thresholds they apply are debatable, Mattia Zulianello and his co-authors (2018) compared the Facebook communication of 83 political leaders from 6 Western and Latin American countries, showing that populism as communication can be empirically identified and that it varies extensively across parties and actors.
Based on the research presented above, our paper seeks to test the following hypotheses.
N. Ernst and her co-authors (2017) concluded that populist communication is mostly used by parties at the extremes of the political spectrum, both left and right. S. Engesser and his co-authors (2017a) showed that left parties tended towards attacking economic elites, whereas right-wingers attacked media elites and ostracized out-groups. Based on this, we expect to find more populism in communication among right-wing parties.
- Hypothesis 1: political ideology impacts on populism: right-wing parties have a higher populism in content and populism in style than left-wing parties.
By using a comparison across countries, election types and over time – comparing EU elections 2014, national parliamentary elections 2017 and a non-electoral period 2018 in France, Germany and the United Kingdom – we also want to determine if and how populism in content and populism in style changed over time. Previously, researchers have shown that the presence of populism and populist elements have been increasing over time in party manifestoes and traditional media (Manucci and Weber, 2017; Hameleers and Vliegenthart, 2020), and that populists have been increasingly successful in elections (Tartar, 2017). Based on this we assume that:
- Hypothesis 2: populism in content and populism in style increase similarly over time on Facebook.
While populism does not increase in a linear and simultaneous way, Luca Manucci and Edward Weber (2017) show increases of populism in party manifestoes and newspapers across four countries after 2010, including Germany and the UK. We seek to test whether
- Hypothesis 3: populism in content and populism in style increase in time at the similar level in all three countries.
Second-order elections are characterized by lower turn-outs, better prospects for small and new parties, worse prospects for government parties, a tendency towards protest voting, and perceptions that less is at stake (Reif and Schmitt, 1980). Owing to these differences, campaigns and campaign strategies vary, and we expect to detect different levels of populism in communication. In the 2014 EU election campaigns, public interest in candidates and election as well as voter turnout were at record low levels (Treib, 2014). Based on this we expect more populist communication during national elections.
- Hypothesis 4: populism in content and populism in style are more prevalent during first-order elections than during second-order elections and non-electoral periods.
To our knowledge, the only study so far connecting populism to specific topics or policy fields is N. Ernst and her co-authors (2019), finding that there are indeed typical issues connected with populist communication. A possible reason for this research gap is that populism research either focused on right-wing or on left-wing parties, thus on ideologies rather than on policy fields. If we follow the notion of populism as communication, we can determine whether and to what extent populism is linked to and more prevalent in specific policy fields (e.g. migration policy). Thus our research question is: Is there a link between populist communication and the topic of the post?
Methods and cases
In comparing party communication in election campaigns in France (FR), Germany (GER) and the United Kingdom (UK), we opted for cases based on different political and media systems (within Western democracies), in order to focus on only one feature (populism in communication) that political parties may or may not have in common. The three countries differ considerably with regard to their political systems, having centralized (FR) or more federal (GER, UK) forms of governance, presidential (FR) or parliamentary (GER, UK) systems, and majoritiarian (UK, FR) or more consociational (GER) decision-making. Their media systems cover all three types identified by Daniel C. Hallin and Paolo Mancini (2004): polarized-pluralist (FR), democratic-corporatist (GER), and liberal (UK). In their updated and empirically richer categorization of media systems, Michael Brüggemann and his co-authors (2014) classify both the UK and Germany as central types, “characterized by strong public broadcasting, strict ownership regulation, and low press subsidies” (p. 1056) and France as the southern type, combining “the highest degree of political parallelism with the least professional journalism and the least inclusive press market” (p. 1056-1057). None of the three countries had so-called populist parties in a government position.
Facebook has become a major source of information and news. In all three countries, approximately one-third of internet users received their news through social media (France 36%, Germany 31%, and the UK 39%), a large portion of which came from Facebook (Germany 24%, the UK 27%, no data for France) (Newman et al., 2018, p. 10). Unlike Twitter, with its very short informational messages, posts on Facebook can be longer, more detailed, and do not force authors to be explicit and pointed. Since Facebook has become a standard platform for the campaign strategies of political actors (Klinger and Russmann, 2017), it is a prime source of official party communication. In the campaigns we analyze, all relevant political parties had Facebook profiles that they used for party communication.
A comparison of the three countries also produces longitudinal insight into different types of elections and election campaigns, and also provides a comparison with a randomly chosen non-electoral period. All the countries participated in the (second-order) 2014 European Parliament elections and held (first-order) national parliamentary elections in 2017. The data we use originate from the political parties’ official Facebook profiles and contain all posts published during the last two weeks of the electoral campaigns prior to the 2014 European Parliament and the 2017 national parliamentary elections, as well as during two weeks in January 2018 as a non-electoral period for comparison (data accessed via Sotrender). The data contain a total of 3,564 posts from 24 political parties in Germany, the UK and France (Table 1). The data represent all major parties present during all three periods, except La République En Marche, La France Insoumise and Debout La France which did not exist in 2014 and no data are available for them in respect of the 2014 EP election. (Please see Appendix A for party in the study, and Appendix C for detailed results.)
Table 1. Number of postings for each country in time
|
2014 |
2017 |
2018 |
Germany |
421 |
546 |
215 |
United Kingdom |
586 |
686 |
273 |
France |
253 |
337 |
247 |
We do recognize that often party leaders can be more relevant politically, reach a larger audience than their parties and can impact the likeability of messages. However, we focus only on the party profiles, for reasons of comparability and to avoid the intervention of personality traits, political status, celebrity status or biographical aspects. S. Engesser and his co-authors’ study (2017a) on political leaders found that populism manifests in a fragmented form on Facebook, and our study seeks to focus on the party side of this finding.
All posts were manually coded by three intensively trained coders (German and French native speakers, fluent in English). Inter-coder reliability of Krippendorff was α=.762b. A self-reliability test was run for each coder within a few weeks after the first coding, with the mean for the coders being Krippendorff α=.904. (The dependent and independent variables and the detailed codebook are available as Appendix D.)
Results
Of the 3,564 party posts on Facebook, 376 contained only populism in content, 317 contained only populism in style, and 213 simultaneously contained both forms of populism. In total, we identified populism in 906 posts, implying that populism is not a marginal phenomenon but that it is present in about one fourth of all posts created by parties during the six weeks under consideration.
Table 2. Regression analysis for Populism in Content (PiC) and Populism in Style (PiS)
|
PiC Empty |
PiC Excluding |
PiC Anti-elitism |
PiS |
|
IRR |
|
IRR |
|
IRR |
|
IRR |
|
Time (year 2014 ref.) |
2017 |
1.170 |
|
1.251 |
|
1.079 |
|
1.702 |
** |
2018 |
.806 |
|
1.015 |
|
.874 |
|
1.434 |
|
Country (France ref.) |
Germany |
5.956 |
*** |
5.260 |
*** |
5.834 |
*** |
12.013 |
*** |
United Kingdom |
15.844 |
*** |
7.753 |
*** |
11.981 |
*** |
10.003 |
*** |
Party Size (Major Parliamentary ref.) |
Minor Parliamentary |
.676 |
|
1.084 |
|
.823 |
|
.753 |
|
Fringe |
.322 |
|
1.024 |
|
.509 |
|
.813 |
|
Governmental |
.585 |
|
.551 |
** |
.510 |
* |
1.144 |
|
Party Ideological Family (Center, Regional and Green ref.) |
Radical Left |
1.411 |
|
.770 |
|
2.715 |
* |
1.447 |
|
Radical Right |
3.094 |
** |
2.706 |
* |
3.604 |
*** |
3.250 |
** |
Right |
1.910 |
|
2.938 |
** |
2.307 |
|
1.432 |
|
Left |
1.190 |
|
1.824 |
|
1.331 |
|
.888 |
|
Constant |
.016 |
*** |
.020 |
*** |
.020 |
*** |
.011 |
*** |
(1/df) Deviance |
.4184 |
|
.5177 |
|
.4962 |
|
.4700 |
|
(1/df) Pearson |
1.1809 |
|
1.2098 |
|
1.1216 |
|
.9584 |
|
Log pseudolikelihood |
-1355.6 |
|
-1713.9 |
|
-1647.2 |
|
-1628.1 |
|
Note: GLM Negative binomial, IRR (Incidence Rate Ratios). Data was clustered by party*year. ***p<.001 **p<.01 *p<.05.
Table 2 presents the regression analysisc on each form of populism discussed above. Notably, the analysis suggests that there is little variance over time in populism as communication. The exception is populism in style that was used almost twice as much during the 2017 national elections as in the 2014 EU elections. This suggests that populism in style may be used as a campaign strategy to mobilize voters in first-order elections. The variance between countries is also significant as political parties in both the UK and in Germany are more likely to employ populist communication in content and in style than parties in France. During all three periods there was a much lower level of populism in communication within all parties in analysis prevalent in France than in the UK and Germany.
As expected and repeatedly shown in the literature, radical right-wing parties extensively employ populist communication in all its dimensions. They strongly advocate anti-elite messages and spearhead the use of populism in style. However, while radical right-wing parties are prone to use any kind of populism, they are joined by some traditional, moderate right-wing parties (the UMP in France and the Conservatives in the UK), who also use exclusionary populism. Both the radical left and radical right show a preference for strong anti-elite communication. (Appendix C provides a table comparing the results and number of posts on the party level.)
ANOVAd analysis comparing populist communication according to party political ideology suggests a variation across the party spectrum regarding both populism in style [F(4, 3559)=41.45 p=.000] and full populism (in content) [F(4, 3559)=37.05 p=.000]. The post-hoc tests indicate strong dissimilarity not only for radical right-wing parties but also for all other parties. This confirms the previous results. We found a longitudinal variation for populism in style [F(2, 3561)=7.96 p=.000] (the post-hoc test indicates a stronger difference between 2014 and 2017), but not for full populism. The results (see Figure 1) suggest that across the political spectrum, no party type is free from populist statements. The temporal development of populism in style and full populism by the radical parties is particularly notable. Since 2014, radical parties – especially radical right-wing parties – have become more populist in their communication style, whether there was an election campaign or not. In fact, their style was most populist during the non-electoral period of 2018.
Figure 1. Relative mean for full populism and populism in style by party’s ideology
Moreover, it is clear that radical right-wing parties pursue full populism: content that does not merely refer to a vague “people”, but content that at the same time is anti-elitist and exclusionary of out-groups. Take this post by the British National Party as an example for full populism:
LIKE & SHARE The terminally liberal political class were slammed last night for their Immigration “disaster” by a member of the BBC Question Time audience. Except for the fact that an individual with different opinions to the far-left BBC was even allowed to take a seat in the carefully staged audience, there’s nothing unusual about this event, oh… except that this man was ASIAN!!! […] “Now you realize that immigration has destroyed the country”, he told a disapproving panel. “Immigration has messed up the country […] the whole of the North of England is full of my community!” he lamented. “This is not England anymore!” Britain must start expelling people from the country and turning people away he said, echoing BNP policy almost word for word. […] When it takes an immigrant to tell the political class what’s what, you know they’re beyond the pale. Count yourself among the growing ranks of British people saying “enough is enough” and taking a stand. […]e
This pattern remains stable over time and is independent of election campaigns. This differs from the moderate left-wing and right-wing parties who also mobilize support by using populist messages, but more so in national, first-order elections, only to relax their emphasis on populism outside the election periods. Based on these results, we confirm Hypothesis 1. Indirectly, this finding also validates Ronald F. Inglehart and Pippa Norris’ notion that cultural values matter (2016).
Regarding longitudinal and cross-sectional variance, Figure 1 represents the prevalence of populism in the perspectives of both content and style. As in the regression analysis, the country differences are clearly visible. French parties are reluctant to use populism in content and populism in style, a trait most likely due to the difference between parties seen as populist (i.e. the Front National and France Insoumise) and all other parties. The descriptive data suggests that the former extensively uses populism in communication, whereas the latter seldom uses it. In contrast, all parties in the UK and Germany extensively use populist communication in their posts. Populism in style strongly increases over time in Germany, while empty populism or the mere invocation of the people dominates in the UK election campaigns, but clearly decreases during the non-electoral period. In France all forms of populism are at a much lower level than in the UK and Germany, but increased during the non-electoral period.
Figure 2. Relative mean for populism in content and populism in style by year
Note: to facilitate inter-index comparison, the mean score was averaged (by number of elements in each index).
Over time, exclusionary populism strongly increases in Germany and slightly increases in France, but not in the UK where it decreases during non-electoral periods. This confirms Hypothesis 2 for Germany and France, but not for the UK where the levels of both populism in content and populism in style clearly peaked during first-order election campaigns (with empty populism dominating in 2014 and 2017 and populism in style in 2018). We reject Hypothesis 3, because it only holds true for France and Germany. Hypothesis 4 is confirmed for the UK, but not for Germany and France where all forms of populism increase over time, albeit on a higher level in Germany than in France.
With regard to our research question, the most prominent policy fields that parties posted about were economic and social policy, but they were not particularly populist. Rather, we found that the perennial (“usual suspect”) topics were the most populist, with posts on migration and security policy containing more than half of all the populist elements.
Table 3. Policy fields and populism
|
Populism |
Full populism |
PiS |
N |
Migration policy |
.60 |
.48 |
.47 |
172 |
Security policy |
.50 |
.32 |
.37 |
119 |
Social policy |
.40 |
.28 |
.20 |
259 |
Economic policy |
.36 |
.23 |
.20 |
394 |
Health policy |
.32 |
.22 |
.14 |
152 |
Foreign policy |
.31 |
.18 |
.19 |
156 |
Education policy |
.20 |
.08 |
.14 |
122 |
Environmental policy |
.12 |
.07 |
.07 |
82 |
Brexit (for UK) |
.44 |
.35 |
.19 |
121 |
Populism indicates if any kind of populism (populism in content PiC or populism in style PiS) was evident in the posting. Numbers indicate the ratio of the posting containing populist element to those which do not contain such element.
While not surprising, it is striking that about half the posts by political parties on migration policy contain full populism. These posts invoked the “people”, and they contained anti-elitist notions as well as messages about the exclusion of out-groups, with the same number using a populist style. About one third of the posts on respectively security and social policy exhibited full populism. Education policy and environmental policy were the fields least related to full populism, although not completely free from populism either. Other categories that we controlled (e.g. technology, transportation, and media) did not contain any populist communication. Although only valid for the UK, posts about Brexit confirmed intuition and contained high levels of populism. Based on these findings, we conclude that populism in content and populism in style are more prevalent in specific policy fields. While populism is prevalent in messages of all the political parties across the political spectrum, it is evident that typical right-wing policy fields dominate the populist posts.
Discussion and conclusion
Our results underline that it makes sense to study populism as a communication phenomenon (rather than only as an actor type approach) that includes typical elements of content and elements of style. Populist elements were evident in the messages of all parties across the ideological spectrum, and confirm that over time a populist style is increasing particularly in the messages of radical parties on the left and right. But moderate and center parties are not always “non-populists”, as they tend to use a populist style during first-order election campaigns, but less so during second-order campaigns and in non-electoral periods. As was expected, radical right-wing parties employ all types of populism (full, anti-elite, and excluding out-groups) on a constant level over time. Regarding the three countries included in our study, all forms of populism are increasing in Germany. In the UK, populist communication peaked during the 2017 national election, particularly empty populism, but not as much in 2014 and 2018. In France, all forms of populism are also increasing, but at a lower rate. Considering that the radical-right Front National (currently RN) under Marine Le Pen, the very low level of populist messages in France is somewhat counter-intuitive. Closer scrutiny of the data reveals that En Marche, while occasionally referring to the people (empty populism), is the only case where no evidence of ostracism, the exclusion of out-groups, and the evoking of fear was found in their posts. This is remarkable in comparison to all the other parties. However, other French parties are also very reluctant to use populist communication. A possible explanation lies in the long tradition of legal constraints that prevent negative campaigning in France. Also, Front National is very centered on Marine Le Pen, a lot of communication in general and populistic in particular is emitted by the party leader, not by the party and thus not captured here.
Studying populism as a communication phenomenon allows for a differentiated view: Yes, populism is increasing and has been scoring at the ballot boxes. But in many cases, it consists only of empty references of “the people” by moderate parties in specific settings. This is hardly threatening the integrity of democratic elections. The more problematic elements, such as the exclusion of specific groups from society, xenophobia, homophobia, attacks on elites, institutions and intermediary actors of various kind, can be obscured by a global focus on populism. In many cases, referring to “the people” is a cloak to mantle illiberal, anti-democratic or authoritarian notions as a presumable “popular” agenda.
An interesting finding is that populist style increased much stronger than populist content. This underlines the importance to see populism not only as an ideology or a political strategy, but as a particular form to communicate – employed by parties across all political spectrum. We can also interpret this as a sign for the brutalization of political language and public discourse.
Populism is also not limited to the perennial topics of migration and security, as we noted populist style elements in posts on environmental issues and on education, foreign and health policies. Environmental policy, in particular, could play an increasing role in the future since radical right-wing parties have recently turned to opposing the notion of a climate crisis and attack climate activists.
To better understand how populism forms part of campaign strategies and how it mobilizes or demobilizes support, researchers need to look beyond parties nominated as “populist”. Improved comparative research designs (beyond the particularities of individual parties) and quantitative, data-based studies on more campaigns and over more extended periods, can be attained by redirecting populism research from actor-centered approaches towards content-centered approaches. The notable difference between the findings in respect of first-order and second-order elections and non-electoral periods suggests that, depending on the political situation, the prevalence and intensity of populism as communication varies.
Our study faces several limitations. We only included one case of a specific time-period. Undoubtedly it would be valuable to reproduce this study with the 2019 EU election and the next round of national parliamentary elections. Also, including party leaders, which may differ in their communication style from parties. We only used data from three countries, which are similar (Western democracies) but also distinct (electoral system). For example, in the case of France, parliamentary elections take place right after presidential elections, which impact campaign content and dynamics, and could produce different results as in the separate campaigning in the UK or Germany. We only coded Facebook posts made by the parties, excluding comments of users and parties in response (although the parties were not very responsive and rarely contributed to the comment section). The inclusion of comments would show whether or not users similarly populist as parties. Another way forward is a comparison across platforms, including Twitter and Instagram. Unfortunately, due to all restrictions on data from social media platforms, independent research is only “peeking through the keyhole” at the moment and only grasping in part what is happening on social media platforms during political campaigns. Also, little is known on how affordances of the platform can influence the popularity of certain posts (e.g. are post with populist content somehow differently promoted by the algorithms). This situation poses a threat to democracy, especially since social media platforms have become a major source of news and information, also considering that populist messages across the political spectrum, time and location, undeniably form part of this environment.