Pop populism: ethno-traditionalism beyond national borders and the populist radical right

Abstract

We test the somewhat counter-intuitive expectation that support for populist radical right (PRR) parties at the national level is associated with public support for ethno-traditional cues—a frequent feature in PRR imagery and nativist discourse—abroad. We do so by leveraging a large-scale comparative dataset that covers voting patterns during the Eurovision Song Contest (ESC) between 1999 and 2019. Looking at voting patterns for more than 30,000 country dyads (i.e., ESC points given by country A to country B in a given contest), we show that countries with stronger national support for PRR parties tend to vote more clearly for songs showcasing ethno-traditional cues (ethnic imagery and, in particular, the use of non-English national languages), even when accounting for cultural and regional ties and song characteristics. These results provide novel insights into the relationship between nativism, radical right populism, and perceptions of foreign ethno-traditionalism.

Introduction and rationale

Populist radical right (PRR) movements are usually strongly entrenched in national narratives, expressing domestic political grievances of the “pure people” against the misdeeds of a “corrupt elite” and the disrupting cultural and political interference from non-native elements (Mudde 2004). This inward focus has been reflected by anti-internationalization attitudes and suspicion towards supra-national institutions (Verbeek and Zaslove 2017), and a marked support for narratives framed in terms of nostalgia and attachment to traditions (Elçi 2022; Elgenius and Rydgren 2019; Buzalka 2018). Naturally, PRR support is thus particularly associated with support for ethno-traditionalism (e.g., Bar-On 2018). Ethno-traditionalism and nativism are strongly embedded in (cultural) ethnocratic ideals, such as returning the nation-state to the ethnic group and preserving and cultivating the national lead culture and its traditions (Mudde 2007). As such, ethno-traditionalism can be seen as the most intense expression of national (vs. international) sentiments within the radical right populist worldview.

This suggests a strong, inward-looking perspective associated with PRR as a political phenomenon, and a marked preference for one’s own customs and traditions. Does this imply, given the distrust for non-native actors often associated with populist worldview, that support for PRR is negatively associated with expressions of ethno-nationalism in other countries? Intuitively, the answer to this question could go either way. On the one hand, such expressions could be shunned as manifestations of otherness and non-nativism, and thus rejected in the realm of the ostracized out-group. On the other hand, expression of ethno-nationalism in other countries could be seen as being part of a similar struggle against national and international elites, impinging on traditions and customs via the promotion of a globalized cultural homogenization.

Surprisingly, whether domestic nativism finds an echo in popular support for expressions of domestic nativism in other countries remains strongly under-investigated. Focusing specifically on the ethno-traditionalist core of right-wing populism, in this article we investigate the extent to which support for Populist Radical Right (PRR) at the national level is associated with public support for foreign ethno-traditional cultural elements – language, music, visual imagery, and costumes. We do so by leveraging voting patterns in the Eurovision Song Contest (ESC; Gauja 2019; Kalman et al. 2019). The ESC is an annual international music competition where artists perform live an original song to represent their country, and receive votes from the public in all other participating countries. Importantly, the public cannot vote for the artist of their own country; they must vote for a foreign artist. Within the framework of the ESC contest, we test for the presence of support for foreign ethno-traditionalism by assessing whether performances that included ethno-traditional cues received stronger support from countries with higher incidences of PRR support. We consider all votes that have taken part in the ESC between 1999 and 2019, representing 21 competitions, 767 songs, and 30,822 voting dyads (i.e., points given from county A to country B during a single competition).

The ESC, started in 1956 by the European Broadcasting Union (EBU), is an ideal setting to test for the presence of support for foreign ethno-traditionalism. First, it is broadcast simultaneously to an audience of millions worldwide, with the public voting for their preferred artist. Official estimates put the overall audience of the 2022 contest at more than 183 million viewers,¹ a magnitude in line with the American Super Bowl. A 2020 IPSOS survey² found that among respondents from a subsample of thirteen participating countries, 97% had heard of Eurovision and 86% had watched a broadcast. 10% of all respondents and 23% of Swedish respondents, e.g., had voted at least once. As such, the ESC represents one of the most salient and noteworthy events in the European cultural calendar. Second, nations and not artists are in the spotlight; the competition is clearly framed as a contest between countries – both as contestants and judges (e.g., “Greece gives its 12 points to Iceland”). Countries must vote for other countries, eliminating a propensity to rate themselves as superior. There is furthermore no monetary prize at stake, the benefits are purely in terms of visibility and status.³ Third, and relatedly, because the competition is between nations, the ESC is both “a unique opportunity for expressing national identity and gaining membership in the European community” (Cassiday 2014, p. 1; see also Wellings and Kalman 2019). Our intuition is that these ethno-traditional cues are particularly picked up and valued by viewers in countries with higher incidences of PRR support.

Fourth, the viewing public votes for the winner as opposed to a competition based on objective measures (e.g., most sports) or selected solely by a group of judges (e.g., most literary prizes). Testing for public support for cultural or political elements abroad can be challenging, as citizens usually cannot take part in elections concerning countries other than theirs and are seldom surveyed about specific political preferences concerning foreign countries. The ESC votes, however, can be seen as a direct measure of aggregated national public attitudes for foreign cultural dynamics. Fifth, the competition occurs in a space that is not explicitly politicized, at least on paper; EBU rules state that “ESC is a non-political event […] No lyrics, speeches, gestures of a political, commercial or similar nature shall be permitted”). To be sure, contestants frequently push these rules to their limits and political and social relations between countries can certainly affect voting behaviors (see discussion below about “voting blocs”); for instance, the result of the 2022 contest, won by Ukraine in a landslide, was arguably in part due to the Russian invasion of that country just a few months beforehand. Furthermore, the ESC origins are deeply rooted in the desire to consolidate a pan-European public sphere, certainly stemming from the geopolitical situation of the 1950s and an eminently political endeavor in itself. Yet, the ESC remains a setting that is not directly connected to national partisan debates, such as electoral politics.

Finally, the ESC represents a conservative setting in which to test for the presence of support for foreign ethno-traditionalism. It is well-documented that the ESC has historically been a liberal and progressive competition (for example, it has been particularly friendly to queer performances, e.g., Baker 2017, 2019) and it is, therefore, unlikely that the ESC amplifies patterns of support foreign for ethno-traditionalism—if anything, the opposite is more probable.

PRR and support for foreign ethno-traditionalism

In this article, we advance the argument that domestic PRR support is positively associated with stronger support for foreign expressions of ethno-traditionalism—that is, cultural elements in other countries that are perceived to be authentic, native, or representative of the geographic region, regardless of their actual histories or provenance.⁴ In a nutshell, our intuition can be summarized as follows: radical right populism (much more so than populism in general), due to its nativist core, tends to be associated with a nostalgic defense of cultural traditions against the homogenizing and modernizing push coming from the (international) elite. Foreign expressions of ethno-traditionalism can be seen, within the PRR worldview, as allies within this broad normative struggle. We unpack the different components of this argument below.

Nativism, a key ideological feature of the Populist Radical Right, “holds that states should be inhabited exclusively by members of the native group (“the nation”) and that nonnative elements (persons and ideas) are fundamentally threatening to the homogenous nation-state” (Mudde 2007: 19). The PRR playbook therefore unfolds by pitting the native people against the elite and arguing that politics should be an expression of their native general will. As opposed to left-wing populists, speaking for an egalitarian and inclusive people, the PRR define the ingroup of the people by delineating who is not a part of it. Normally, the expression of this native will tends to be inward-looking, yielding to more negative evaluations of all elements associated with foreignness (that is, outside of the national, cultural, ethnic, religious ingroup). Along the same lines, nostalgia for the past and affective attachment to traditions are strongly linked to populism (Elçi 2022), particularly in PRR movements (Elgenius and Rydgren 2019; Buzalka 2018). Similarly, populist discourse is often characterized by a marked preference for authenticity (e.g., Montgomery 2017). PRR nativism has a “tendency to look backward in a nostalgic attempt to revive and reconstruct” a past that has been lost (Betz 2017, p. 337; see also Taggart 2000; Mudde 2007). By definition ethno-traditional cues serve these preferences for tradition, nostalgia, and authenticity. In this sense, nativism and its nostalgic components should be associated with a stronger preference for domestic ethno-traditionalism—but also with a generalized distrust of anything foreign. Yet nativism is not the only force that shapes the (radical right) populist worldview. To the horizontal conflict line between the (domestic, culturally homogeneous) ingroup and the foreign outgroup a vertical conflict line is superimposed, pitting the ordinary people against the self-serving and corrupt elite (Rooduijn 2019). Usually expressed as a strong distrust towards national elites, this vertical conflict dimension often translates also into a fierce normative opposition towards international elites and forces—including cosmopolitanism, globalization, and supranational governance—especially evident in the Eurosceptic discourse of many radical right populist parties in Europe (Kneuer 2019; FitzGibbon et al. 2017). Importantly for our argument, this opposition to international elites, cosmopolitanism, and globalization is deeply intertwined with ethno-traditionalism. Culture and cultural preferences are part of social hierarchy (Bourdieu 1984). For example, contrast American presidential candidates being expected to sample corn dogs and pork-chops-on-a-stick at state agricultural fairs, and President Barack Obama’s campaign misstep (“Arugulagate”) asking Iowa farmers whether they had “gone into Whole Foods [an American grocery store known for its high prices and organic products] lately and see[n] what they charge for arugula?”⁵ It does not take much effort to identify who is more likely to be considered the “elitist” in this example, even for non-Americans. Furthermore, tastes are increasingly becoming globalized and homogenized as evidenced by critiques of the global dominance of Hollywood, McDonaldization, and the English language. Expressions of ethno-traditionalism can be seen as the antithesis of cosmopolitanism or cultural homogenization—the retention of the nationally specific vs. the diffusion of the globally shared—and as such find an echo in (PRR) populist support. Our argument is that this is also likely across national borders, indicating a general struggle of ethno-traditionalism as a social and political phenomenon—regardless of its content—against the globalized modern cultural homogenization. A similar trend was observed in the early phases of the Brexit, where European conservative media signaled the existence of a “common ground” with British Eurosceptics (Bijsmans et al. 2018).

Why should a supporter of, say, the Italian populist party Lega Nord care about ethno-traditionalism in Romania? It is the conceptual “thinness” of (right wing) populism that makes it highly modular and readily adaptable to different national contexts, structures, and grievances. The defense of national traditions finds an echo in the identification of a common enemy across different countries—for instance, the European Union or the economic globalization more in general—so that demonstrations of ethno-traditionalism are not necessarily seen as antagonistic or competitive in other countries. On the contrary, they can be seen as coherent expressions of nativist people-as-underdog, against globalized cultural homogenization. All in all, we expect that foreign instances of ethno-traditionalism might be seen as participating to the same normative struggle against the homogenizing push from international elites. As such, popular support for foreign ethno-nationalism could be expected to be a positive function of support for the PRR domestically. The ethnic culture of kin people-as-underdog should thus be reflected into a generalized support for expressions of ethno-traditionalism.

An example from our object of study, the ESC, helps highlight how all these aspects may work in an international and transnational context. In 2017, Portuguese singer Salvador Sobral won the ESC with his Portuguese ballad “Amar Pelos Dois” (See Online Appendix D for links to videos of all the songs referred to in the text), the first winner in a decade to sing exclusively in a language other than English. Featuring only Sobral singing to accompanying piano and strings, the performance stood out from many other Eurovision entries featuring backup singers, dancers, pyrotechnics, digital enhancements, and elaborate costumes. International commentators praised the song for its authenticity and for being representative of Portuguese tradition: “Fado singers like Salvador Sobral reflect Portugal’s soul” and “Amar Pelos Dois pays tribute to Portugal’s folk traditions, borrowing from the Fado style.”⁶ Relevant to our definition of ethno-traditionalism as being about perceived tradition or authenticity, “Amar Pelos Dois” was not actually a Fado song, but it was interpreted by foreign viewers to be Fado or Fado-esque. Importantly, in his victory speech Sobral set his song in opposition to the increasing globalization of music: “we live in a world of disposable music, fast food music without any content.” Our intuition is that this type of messages resonates well with normative considerations about the nefarious role of international elites and the need to defend customs and traditions, itself more likely in societies with a higher incidence of PRR support. Summing up, in this article we test the general intuition that the magnitude of PRR support within a given country is positively associated with support for expressions of ethno-nationalism in other countries.

Data and methods

Data

We test our expectations with a large dataset that covers all 21 ESC competitions between 1999 and 2019, representing 767 unique songs and performances. Our examination starts in 1999 when countries could freely choose the language of their entry (prior to 1999, countries were required to submit entries sung in a national language) and ends in 2019, the last ESC contest before the 2020 shutdown due to the COVID-19 pandemic.⁷ Figure 1 illustrates all participating countries in the dataset; increasing color intensity reflects more frequent participation. Table C1 (Online Appendix C) presents the detailed patterns of participation for each country, and Table C2 lists all the winners between 1999 and 2019.

Fig. 1

Participating countries, 1999–2019

Under the current ESC format, six countries automatically qualify for the final: the five biggest financial contributors to the EBU (France, Germany, Italy,⁸ Spain, and UK; the so-called “Big 5”) and the winner of the previous year (which is the current year’s host). The remaining contestants compete in one of two semi-finals (random assignment) and the top 10 songs from each of the two semi-finals qualify for the final. A total of 26 countries thus participate in the final. However, all countries can vote in the final, regardless of whether their entry advances.⁹

Since 1998, citizens in all participating countries can vote for their preferred songs (“televoting”); these votes are computed by each national broadcaster, combined with the votes of a professional jury, and transformed into a normalized scale; each country assigns 12 points to their most preferred song (hence, the famous “12 points go to…”), 10 to the second most preferred song, then 8, 7, 6, 5, 4, 3, 2, and 1 points in a descending order; all other countries receive 0 points. Viewers and countries cannot vote for their own country’s entry. It is important to note that it is only from 2016 onwards that separate scores from the public vote (televotes) and the professional jury vote made public; prior to 2016 the votes assigned by each country come from a mix between the two voting groups (public and professional), aggregated at the country level. Because we can account for only part of the popular preferences, we should globally expect more conservative effects prior to 2016. To account for this, we control all models by a simple dummy variable that sorts all songs between the two periods (pre- and post-2016). Furthermore, we run a series of additional models only for the four contests in which the popular vote can be fully accounted for (2016–2019); as we will see, results are particularly strong in these additional models.

Our dataset is structured so that each observation corresponds with the vote of one country for another (country dyads), for all participating countries between 1999 and 2019, and for both semifinals (when existing) and finals. The final dataset includes information for 30,822 such dyads, computed by taking into account all voting combinations based on the participating countries (N * (N—1)); for example, a competition with 26 participants the dataset includes 26 * 25 = 650 voting dyads (A to B, B to A, A to C, C to A, and so forth). For each dyad, the dependent variable used in our models is the points given by a country to another country’s song (from 0 to 12).¹⁰ Each dyad also includes all information about the voting country for that specific year (especially in terms of the importance of populist support, see below), as well as characteristics of the song to which the points were given (e.g., language, stage performance, etc.; see below).

PRR and nationalism

We measure the presence of PRR support in the country, in any given year, by looking at the share of votes that PRR parties have received in the most recent national elections in that country and for that year. To do so, we have first identified (i) who the populist parties are, in each country and for each year (and more specifically the PRR amongst them), and (ii) their electoral fortunes in elections over time.

Unfortunately, there is no large-scale comparative inventory of populist parties covering all the countries in our database over the past 20 years. To date, the most comprehensive inventory that also includes a temporal component is the “PopuList” project (Rooduijn et al. 2019). While it is probably the most systematic and comprehensive classification of populist movements available, the “PopuList” does not cover all the countries included in our dataset. Therefore, we compiled the list of populist parties by triangulating evidence within the “PopuList” with the most systematic large-scale repository of information about parties available: the Wikipedia pages for each party, which provides a classification of the party ideological position”, including whether the party can be classified as “populist”. If certainly intuitively unusual as a systematic source, Wikipedia has been shown to provide reliable factual information on party ideological classification (Herrmann and Döring 2021). Our compiled list of populist parties relevant for our study is Table B1 (Online Appendix B); PRR are indicated with an asterisk. This classification is used to compute the share of “PRR support” in each country, per year, by simply adding up their electoral results in the national election closest to the relevant competition year. Electoral results were compiled again via information present in the Wikipedia page of each election. The procedure creates a continuous measure of PRR support, with varying scores for each country and year, varying between 0% (Albania 2005) and 72.1% (Hungary 2011).

To ensure that the effect shown comes from the radical right component (and not populism in general), we control all models also for the presence of overall “populist support” in the country. This measure is obtained by simply adding the electoral results in the national election closest to the relevant competition year of all populist parties, not only PRR—that is, all parties in Table B1. This more general measure of populist support varies between 0% (Moldova 2010) and 96.55% (Georgia 2012).

Furthermore, to ensure that the effects of PRR in the country on support for ethno-traditionalism abroad are not simply driven by the intrinsic nationalism of (some) PRR parties (e.g., Bonikowski et al. 2019), we control all models by the support for “nationalist” parties in the country in each year. To measure such support, we have replicated the procedure described above for the measure of populist support (see “PRR and nationalism” section) but applied to parties that are classified on their Wikipedia page as “nationalist.” Table B2 (Online Appendix B) lists all nationalist parties from the participating countries. We were able to check the external validity of our classification of parties as “nationalists” by triangulating our data with the 1999–2019 trends in the Chapel Hill Expert Survey (CHES; Jolly et al. 2022). The CHES dataset includes a measure of whether parties advocate cosmopolitanism or nationalism (0–10 scale). For 27 countries over time, looking at 660 parties that have a score in both datasets—that is, that have a CHES score and are classified or not as nationalist in our data—shows a clear-cut result: parties that we classified as nationalist score significantly and substantially higher in the CHES nationalism scale, t(658) = -19.08, p < 0.001. While parties that we did not classified as nationalist score just above 4.3 points on the 0–10 CHES scale, parties that we did classify as nationalist score, on average, 7.7 points.

Language and ethno-traditional elements in song and performance

We hypothesize that the number of points assigned within each dyad (that is, from country A to country B’s song) is a function of the political context of the voting country (A) and the characteristics of the song (from country B)—more specifically, whether the song or live performance contain ethno-traditional cues. For example Russia’s 2012 Eurovision entry portrayed was sung by the group Buranovskiye Babushki (Buranovo Grannies), composed of six older women, incorporated embroidered “ethnic” dresses and headscarves, shoes made of bark, “traditional” music, baking bread in an oven, and singing primarily in the regional language of Udmurt. These cues may come from a majority (e.g., songs in Estonian, etc.) or minority group (e.g., a minority language such as Udmurt in the above example, or music and dress from the Sámi people, Norway 2019). The 2022 French entry, sung entirely in Breton, is also an excellent example in this sense.

We identify such ethno-traditional cues at three levels: the language(s) of the song, the type of music, and the presence on stage of visual elements with overt ethno-traditional connotations. Collectively we refer to these as ethno-traditional “cues.” When only referring to the latter two (i.e. non-language cues) we refer to them as ethno-traditional “elements.” We argue that the language of the song, its “ethnic” genre, or the presence of “traditional” instruments, costumes, or performative elements on stage, can be interpreted by viewers in other countries as ethno-traditional cues—and supported (or not) according to the dynamics described earlier. Because the point-giving audience of Eurovision is outside the competing country, these external viewers may not be aware of what is or is not authentic, regionally specific, or shared. What matters for the purpose of this paper is that a viewer might reasonably read certain cues as “ethnic” or “traditional” for the country.

Of the 767 songs in our database, 587 (76.5%) were sung totally or partly in English and 23.5% were completely sung in a language (or languages) other than English. We also created a variable for songs that were entirely or partly sung in a national language of the country. These include songs in an official national language (e.g., songs from Finland performed in Finnish or Swedish, ESC 2012), as well as dialects, regional, or minority languages (e.g. Võro for Estonia 2004 and Samogitian for Lithuania 1999). Of the 767 songs, 320 (41.7%) were sung partly or fully in one national languages, and 232 (30.2%) were fully sung in a national language(s) (e.g., Malta 2000, sung in English and Maltese).

In addition to the song’s language, we manually coded all 767 songs, equivalent of 38.3 hours of live performances,¹¹ for the presence of ethno-traditional elements in terms of i) music that could be interpreted as ethno-traditional¹² (28% of all songs contain such cues); visual elements with overt ethno-traditional connotations, such as instruments, costumes, makeup, or dance¹³ (11% feature an ethno-traditional instrument on stage, 19% another visual cue, such as costumes). Figure 2 presents some examples of such visual elements, identified during the coding phase.

Fig. 2

Examples of visual ethno-traditional elements found

Figure 3 shows that these four ethno-traditional cues are more or less consistently present in songs and performances over the 21 years covered in this study. We combined the ethno-traditional music and performance elements into an additive index of ethno-traditional elements (α = 0.69), which we will use along with the song language in all models. The presence of these ethno-traditional cues for each of the songs is listed in Table C3 (Online Appendix C).

Fig. 3

Evolution of ethno-traditional song elements over time. N = 767

Voting blocs

Because of historical trajectories of cultural, social, ethnic, linguistic, or geographical proximity, countries participating in the ESC tend, in many cases, to vote along “friendship networks” patterns (e.g., Charron 2013)—more provocatively also referred to as “collusive voting patterns” (Gatherer 2006). For example, between 1999 and 2019, Cyprus gave Greece 12 points (the maximum value) 21 out of 22 times and Greece did the same for Cyprus 19 of 20 times. Other such observed patterns include countries from the former Yugoslavia, Turkey-Azerbaijan, Germany-Switzerland-Austria, and the Baltic states. We control our models for membership in voting blocs to account for potentially shared ethno-traditional and cultural ties between countries. To do so, we use the results of the clustering classification of countries in thirteen different “friend groups” described in Charron (2013, p. 490). The procedure inductively creates groups of countries that tend to vote for each other more frequently, looking at the Euclidian distancing between all countries when looking at the whole voting patterns over the years. Table 1 presents the members of these thirteen “voting blocs.” Membership to a bloc implies that those countries are substantially more likely to vote for other members of the same bloc when compared to other countries—that is, the relationship between all countries in each bloc is reciprocal.

Table 1 Voting blocs

The classification in Table 1 largely reflects trends discussed in the literature (e.g., Fenn et al. 2006; Gatherer 2006; Yair and Maman 1996). Adopting this classification, for the 1999–2019 period 1,808 votes (5.9% of the total) were cast within a “bloc.” Unsurprisingly, the average number of points given to a country within a bloc is substantially higher; a vote outside of a bloc averages 2.38 points (SD = 3.45), whereas a vote within a bloc averages 7.12 points (SD = 4.21); the difference is substantial, t(30,820) = -55.90, p < 0.001, d = 1.35.

Additional covariates

Song and performance

Looking at the level of the song and performance, first, “good” songs are more likely to receive points across the board, regardless of all strategic or cultural considerations (e.g., Ginsburgh and Noury 2008). Because it is methodologically questionable to establish objective measures of each song “intrinsic quality,” we simply control all models by the song’s final rank, a variable that ranges from 1 (the winner) to 28 (the last ranked country).

Second, because of primacy and/or recency cognitive biases affecting viewers’ perceptions (e.g., Crano 1977; Steiner and Rain 1989), the order in which the songs are performed on stage could potentially affect their success, which is why the EBU uses randomized draw. We indeed observe a slight recency effect: songs performed later in a competition are more likely to receive a higher score. Because all voting takes place only after all songs have been performed on stage, this is not surprising. This is the case both during semi-finals, r(23,288) = 0.04, p < 0.001, and finals, r(25,044) = 0.10, p < 0.001. Because of this, our models will control for the order of passage (“draw”) of the song during the show.

Third, the ESC features a wide variety of styles, genres, outfits, instruments, and performances, often quite elaborate, which are likely to explain the success of some songs and performances. During the 2016 final in Malmö, the hosts performed a tongue-in-cheek song describing all elements that a successful entry must have,¹⁴ such as “a man in a hamster wheel, […] a burning fake piano, […] a Russian man on skates, [… or] four women baking bread.”¹⁵ The skit suggests that “stylistic” elements are likely to drive a song’s success, above and beyond its musical quality. With this in mind, we have manually coded all 767 songs to identify the following stylistic elements: (i) its musical genre: ballads, rock, hard rock, dance, funk and soul, opera, and “ethnic” music; all these genres are coded independently and are not mutually exclusive; (ii) characteristics of the singer(s): whether the song is performed by a solo singer, a duet, or a band; (iii) the gender of the lead singer (female, male, both)¹⁶; (iv) the presence of dancers on stage (including when the singer(s) themselves dance); (v) the presence of instruments on stage; (vi) the presence of backup singers/vocalists on stage; (vii) the presence of a “key change” (modulation) at any moment during the song; (viii) the presence of any “ambiance” elements (fire or pyrotechnic props, dry ice, smoke, wind, stroboscopic lights); (ix) the presence of any “comedic”¹⁷ or satirical¹⁸ elements; and (x) the presence of any overtly sexually provocative elements, such as suggestive outfits, overt sexual innuendos, or provocative behaviors.¹⁹ Finally, (xi) the fact that the ESC is a queer-friendly competition is well-documented (e.g., Cassiday 2014; Baker 2017); as some forms of populism and nationalism are likely to be associated with homophobia and more conservative worldviews when it comes to gender dynamics (e.g., Mole et al. 2021), we have coded for the presence of overt LGBTIQ + cues (e.g., drag queen performances²⁰ or displays of gender non-conformity²¹) or classical queer tropes.²² The distribution of song and performance characteristics is illustrated in Figure C1 (Online Appendix C).

Competition and country

We run separate models for semi-finals and finals as robustness checks, and to also account for the fact that it is especially the “better” songs that reach the final. Additionally, our models include a dummy variable identifying the participating countries that are part of the biggest contributors to the EBU (“Big Five”): France, Germany, Spain, UK and Italy (the latter since 2010). Songs from these countries automatically qualify for the final and do not compete in a semi-final, and, as such, the public might be less familiar with them. Finally, at the country level, we control all models for the geographical region of the voting country (Western Europe, Southern Europe, Northern Europe, Eastern Europe, Balkans), and for their country level of ethnic fragmentation (or “fractionalization”; Alesina et al. 2003). The latter is an indicator that broadly reflects “the probability that two individuals selected at random from a country will be from different ethnic, linguistic or religious groups” (Martinez i Coma and Nai 2017, p. 78) and varies theoretically between 0.0 and 1.0. For all countries except four (San Marino, Serbia, Montenegro, Bosnia and Herzegovina²³) we simply use the latest score reported in Alesina et al. (2003).

All datasets (songs and voting dyads) and scripts are available for replication in the following OSF repository: https://osf.io/jrwzm/

Results

What drives the success of a song? Fig. 4 presents the results of linear regressions where the song’s final “success” is estimated via the presence of these ethno-traditional cues in the song and performance, controlling for all other covariates. The coefficient plot presents results based on standardized variables; full results with non-standardized variables are in Table A1 (Online Appendix A). The plot presents the results of two regressions, the first estimating the final rank of the song (reversed; higher scores reflect greater success), whereas the second estimates the total cumulative number of points received by the song (adjusted by the total number of voting countries).

Fig. 4

The success of ethno-traditional cues; coefficient plot. Coefficient plot with 95% confidence intervals. All variables standardized (M = 0, SD = 1). Full results with original non-standardized variables in Table A1 (Online Appendix A). The variable “rank” has been reversed; high scores indicate a more positive performance

More successful songs are more likely to feature a male singer and to occur later during the show. Songs that score lower are more likely to include backup singers, “ambiance” elements, provocative performances, and comedic elements. Additionally, songs score more poorly if they are performed by the ESC host country, member of the “Big Five”, or from Northern Europe, Eastern Europe, and the Balkans (when compared by songs from Western Europe). Importantly, songs that are fully in a national language are penalized; they rank worse than all other songs and receive fewer points. Inversely, the presence of ethno-traditional musical elements is rewarded, and in a greater order of magnitude. The presence of such elements is, even one of the most important factors explaining the success of the song, for both measures of success (rank and points received).

Is the success of these ethno-traditional cues a function of PRR support in the voting country? To answer this question, we use models at the voting dyads level (voting patterns between countries). The models in Table 2 are hierarchical linear regressions (HLM), where observations (voting dyads) are nested within individual songs. In all models, the dependent variable is the number of points given by country A to the song performed by country B (see methods section above). Models in Table 2 regress the number of points given within the dyad as a function of the song and performance characteristics, the percentage of PRR and nationalist support in the country that year, and a series of interaction terms between those percentages and the presence of ethno-traditional cues in the song and performance they voted for. More specifically, M1 includes the interaction between PRR support in the voting country and the language of the song (national language of the country), whereas M2 includes the interaction between PRR support in the voting country and the presence of ethno-traditionalist elements in the song or performance. Models M3 and M4 present interactions with nationalist support. All models exclude songs of English-speaking countries, which could potentially confound the dynamics at play (see Table A3 for models that include these songs).

Table 2 Populism, ethno-traditional cues, and points given

The table broadly shows countries with higher PRR support tend to support songs showcasing ethno-traditional cues, supporting our expectation. Figure 5 substantiates the effects of the four interaction terms. As shown in the two top panels, in presence of ethno-traditional cues (national language, ethno-traditional elements, plain black line) the percentage of support for PRR in the voting country is positively associated with the number of points given to the song. In other terms, greater support for songs showcasing ethno-traditional cues comes from countries with a stronger PRR presence. Inversely, in absence of ethno-traditional cues (song in other language, no ethno-traditional elements; dotted line), there is no positive relationship between support for PRR and points given to the song. Importantly, as the two bottom panels show, increasing levels of nationalist support in the voting country are not significantly or substantially associated with points given to a country, regardless of the presence or absence of ethno-traditional cues—suggesting that the dynamics at play are driven by intrinsic specificities of PRR support beyond nationalism.

Fig. 5

Populism, nationalism, ethno-traditional cues, and points given. Marginal effects with 95% Confidence intervals, calculated from coefficients in Table A3. All other variables fixed at their mean. Please note the reduced range of the y-axis (original variable ranges 0–12)

Robustness checks

A series of robustness checks show consistent trends with those discussed above (Online Appendix A). Table A2 replicates the main models but checks for the interaction between ethno-traditional cues and general populist support (that is, regardless of the ideological direction of the support, models M1 and M2) and non-PRR populist support (models M3 and M4). Results are considerably weaker than the main ones discussed in the text. Furthermore, the presence of non-PRR populist support (that is, support in the country for all populist parties excluding PRRs) is negatively, albeit weakly, associated with points given for songs that contain those elements. All in all, what matters is PRR support—much more so than general populist, non-PRR, or nationalist support.

Table A3 replicates the main models but also includes songs for English-speaking countries (excluded from the main analyses). Table A4 replicates the models but only for votes that have taken place outside of a voting bloc, thus excluding all voting dyads within any of the 13 blocs (N = 1,808 dyads excluded), whereas Table A5 replicates the models but only for ESC finals (semifinals excluded). Table A6 then replicates the main models but only for ESC contests where we can fully account for the popular vote (televote), between 2016 and 2019; results are again robust broadly speaking, even if only for songs in national language. Because of the zero-inflated nature of the dependent variable—in any given ESC, countries give points to only ten songs among all competing ones (respectively, 1, 2, 3, 4, 5, 6, 7, 8, 10 and 12 points), and give thus de facto zero points to all other songs—we have run alternative models that use instead multilevel negative binomial estimations as robustness checks (Table A7). Finally, Table A8 presents results for models that exclude the two arguably “non-European” countries (Israel and Australia), to investigate whether their inclusion interferes with more pan-European logics. Results of all these robustness checks are in line with the main results discussed above.

Conclusion

Recent years have seen the emergence of a new societal and political cleavage, between the proponents of cosmopolitanism and liberal supranationalism, on the one hand, and those more in favor of an inward-looking repli on traditions along ethno-nationalistic lines. While these dynamics are well understood from a national standpoint, in particular through the lens of voting patterns for the populist radical right, little is known as to whether they can “spill out” beyond national borders. Our investigation focused on the ethno-traditionalist core of right-wing populism and, more specifically, on whether PRR support in a given country is associated with support for ethno-traditionalist expressions in other countries. To test for such an intuition we have investigated the “hard case scenario” of popular support for songs that showcased ethno-traditional elements competing during all Eurovision Song Contests (ESC) between 1999 and 2019. Our investigation showed that support for songs presenting ethno-traditional cues—that is, the use of non-English national languages and the presence of ethno-traditional elements in the song or performance, such as traditional music, dances, costumes, or instruments—significantly increases as a function of the strength of PRR in the country assigning the points, even when accounting for cultural and regional ties.

These results could indicate the presence of a reservoir of untapped “transnational” PRR sentiments—or, at the very least, the existence of sentiments of affinity for ethno-traditional expressions that resonate beyond national borders. This is in line with the general idea that traditionalism and populist nostalgia, usually inward-looking, can find an outward-looking echo if expressed against the homogenizing force of cosmopolitanism and globalization.

In recent years, several initiatives promoting international or transnational (Moffitt 2017) cooperation among populist and PRR movements have made the headlines, from Steve Bannon’s (aborted) European initiative to establish “the infrastructure, globally, for the global populist movement” (Miller-Idriss 2019, p. 17), to groups that have formed in the European Parliament among elected MEPs of national Eurosceptic parties (McDonnell and Werner 2020). While our results do not have anything to say about the extension of populist sentiments per se across the national borders, central for these initiatives, they nonetheless suggest that inward-looking nativism is not necessarily at odds with positive feelings towards ethno-traditional expressions in other countries. More research is needed to understand whether the embeddedness of traditionalism and cultural nostalgia associated with domestic populism (e.g., Elgenius and Rydgren 2019) also translate outside of the national borders (see also De Cleen et al., 2020).

These results should also be considered cautiously, especially in terms of the temptation to conclude that they reflect a direct association between individual support for PRR parties and support for ethno-traditional expressions in other countries. We do not have individual data able to assess whether those citizens that supported populist parties in the most recent national elections in their countries are those that voted for songs with ethno-traditional cues. It might be even likely that an inverse relationship exists between voting for PRR and watching the ESC in the first place—especially knowing that the ESC is a particularly progressive show (e.g., Cassiday 2014). Our claim is, thus, simply that the level of populist support generalized in the country is associated with popular cultural choices in the aggregate. As such, the existence of a link between the two, as shown in our article, could be considered as a particularly conservative test of the idea that populist support can find an echo outside of the national borders. Further research able to disentangle the individual dynamics of populist support (or attitudes; Akkerman et al. 2014), cultural traditionalism, and support for international expressions of ethno-traditionalism, for instance by triangulating observational electoral data and experimental evidence, is necessary.