Far‐Right Populism Online: Did VOX’s Community Reproduce the Party’s Discourse During the April 2019 Campaign?

In April 2019, VOX, a far‐right populist party, won seats for the first time in the Spanish parliament. VOX successfully used social media to participate in the electoral debate and to establish a more direct link with its followers. We investigated how the VOX online community was structured during the election campaign and to what extent the most influential profiles spread the party’s messages. We accordingly analysed two samples, one composed of tweets and retweets that used the hashtags #28A, #28Abril, and #28AbrilElecciones, and the other composed of metaphorical expressions identified in tweets by influencers. Applying social network analysis to the first sample, we studied the form and structure of the network and identified key profiles in the VOX community, i.e., influencers, builders, and bridges. Using critical metaphor analysis with the second sample, we identified the main frames used by VOX influencers to explore whether they reproduced the party’s populist discourse. We found that the VOX online community in 2019 did not only include party supporters or members but was composed of varied profiles. For this reason, the populist metaphorical framing used by the VOX leadership was only partially disseminated.


Introduction
Populism is gaining momentum (Belim, 2020) in the digital age and, as pointed out by Hameleers andSchmuck (2017, p. 1425), social media "contribute to [its] success by providing an attractive environment for both politicians and ordinary citizens to disseminate their political ideas".Populism, in its turn, is transforming social media, generating specific dynamics in the articulation of communities and the dissemination of messages.
Although right-wing populist parties began to emerge in Europe in the 1990s, Spain remained outside this trend until 2019, when the far-right populist party VOX emerged to quickly obtain representation in the Spanish parliament.Parallel to its growth, VOX has progressively become more radical and has aligned itself with other parties of the European far-right (Ferreira, 2019).
VOX was created in 2013 as a far-right splinter from the Popular Party (PP) in response to discontent among militants regarding a perceived PP shift to the centre.According to Ferreira (2019), the VOX ideology has four basic pillars: Spanish nationalism based on a centralized single-nation state, ethnocentrism combined with xenophobia, authoritarianism, and the defence of anti-feminist, traditional family, and rural values.Beyond these ideological features, Capdevila et al. (2022), Eatwell and Goodwin (2019), Olivas-Osuna (2021), and Vampa (2020) state that VOX is a far-right nationalist-populist party and identify in its communication strategy the main components of the populist discourse.Since its inception, VOX has been active in the main social platform for political debate, namely Twitter.According to Guerrero-Solé et al. (2022), VOX's incursion on Twitter has greatly contributed to political polarization in two main blocks to the left and right.
In this article, we focus on an issue that, as far as we are aware, has been unexplored to date: precisely how VOX's populist message is disseminated in the party's online community.We accordingly analysed how this community was articulated and composed and how VOX's populist message was reproduced by key profiles, focusing on the period around the Spanish general elections of 28 April 2019, when VOX's proactivity in its communications ensured that VOX achieved parliamentary representation at the national level for the first time.

Theoretical Framework
In this article, in line with authors such as Charaudeau (2009), De Vreese et al. (2018), Engesser, Ernst, et al. (2017), Engesser, Fawzi, and Larsson (2017), Ernst et al. (2019), andHameleers (2019), we consider populism to be political content expressed using a specific communication style.Those authors coincide in considering populism from the perspective of a communicative logic based on two related issues: the discursive manifestation of a thin-centred ideology that can be combined with other ideologies (Mudde, 2004) and the expression of ideas using linguistic and stylistic devices that are as crucial as the ideas themselves (De Vreese et al., 2018).Engesser, Fawzi, andLarsson (2017, p. 1280) refer to a "populist communication logic," specified in a set of norms, routines, and processes that shape populist communications.
From an ideological point of view, populism is transmitted through what De Vreese et al. (2018, p. 426) denominate "core components of populism ideology."These core components are as follows: (a) sovereignty of the people, appealed to and located at the communicative centre; (b) condemnation of the elite; and (c) dangerous others who "deprive the sovereign people of their rights, values, prosperity, identity, and voice" (Albertazzi & McDonnell, 2008, p. 3).According to Aslanidis (2015, p. 12), populist discourse can be understood as a frame that "diagnoses reality as problematic because 'corrupt elites' have unjustly usurped the sovereign authority of the 'noble People' and maintains that the solution to the problem resides in the righteous political mobilization of the latter in order to regain power."Based on this populist frame, Engesser, Fawzi, and Larsson (2017, pp. 1281-1282) have elaborated a narrative that combines these components to situate them at the centre of populist political communications.Antagonisms are established between the people, whose sovereignty is at risk, and the elite and the dangerous others.In this struggle, populists define themselves above all as defenders of the people and as its ally in facing the threats posed by the elite and dangerous others.Nevertheless, who belongs to the out-group depends on the ideology of the left/right axis (Alonso, 2018).As for right-wing populism, the financial system, experts, and intellectuals form the corrupt elite.On the contrary, left-wing populism points at supranational institutions, media, and a range of traditional political institutions (De Cleen et al., 2020).The same happens with the attack on "the dangerous others," which mostly applies to right-wing populism and, thus, cannot be considered an inseparable characteristic of populism as a wider movement (Mudde, 2004), which is not necessarily exclusionary (Mudde & Rovira Kaltwasser, 2013).
Considering populism's ideological core and taking into account the presence or absence of populist's frame components, Jagers and Walgrave (2007) establish four types of populism: (a) complete populism, where political actors appeal to the people, blame the corrupt elite, and exclude the others; (b) anti-elitist populism, where they refer to the people, and they criticize the elite; (c) excluding populism, which includes references to the people and ostracizes the others; and (d) empty populism, where only the people is mentioned.
As for how populist ideology is communicated, Ernst et al. (2019) have identified three main stylistic devices, namely, negativity, emotionality, and sociability, while Engesser, Fawzi, and Larsson (2017) and Ernst et al. (2019) have proposed simplicity as a fourth device.Although not populist in themselves, these stylistic devices help package content in an attractive way.Presenting antagonism in a simple or elaborate way, with a rational/positive or emotional/negative slant, is not a mere matter of communication style but an ideological decision.

Metaphor as a Populist Stylistic Device
Because ideas are transmitted through language, linguistic choices let us access discourses' ideological content (Charteris-Black, 2004), and so we suggest that populist ideological content (the populist frame) can be understood through stylistic devices characteristic of language.We use metaphor as a framing device (Gamson & Modigliani, 1989) that not only condenses the stylistic elements of communication but also how issues are perceived.As Keating (2021, p. 499) has pointed out, metaphorical expressions contribute to the communication of ideologies; furthermore, as in the case of populists, they "can be useful for politicians who wish to counter the dominant conventional frames."Metaphor use, which favours the economy of attention, thus facilitates the use of simple and understandable language in populist communications (Charaudeau, 2009).
The metaphor, as well as a frame detection mechanism, plays a dual cognitive and persuasive role and sometimes itself gives rise to a frame (Burgers et al., 2016).In its cognitive role, metaphor is a figurative way of understanding the world (Lakoff & Johnson, 1980).Analogy, in particular, is fundamental to the linguistic construction of social and political reality because it shapes a new reality, the target domain (TD), from a known reality, the source domain (SD; Mio, 1997;Musolff, 2004).In its persuasive role, metaphor shifts attention to specific aspects of reality (Eco, 1993;Semino, 2008) to the detriment of other aspects.
We start from the premise that populist movements, like VOX, use metaphorical language to convey their particular and possibly novel perspective of the core components of the populist frame.As demonstrated previously (Capdevila et al., 2022), during the 2019 election campaign, VOX used metaphors to build a specific perspective on the three core components of populist ideology: the people personifying Spain as a living being (España viva [living Spain]); the left-leaning Spanish Socialist Workers Party (PSOE) governmentled by Pedro Sánchez as president-reflecting the elite (la dictadura progre [the politically correct dictatorship]); and dangerous others, personified as immigrants (salvajes [savages]), Catalan independentists (golpistas [insurrectionists]), and traditional media, especially when left-leaning (una secta [a sect]).

Social Media: Morphology and Echo Chamber Limits
As widely demonstrated (Barlett, 2014;Boulianne et al., 2020;Casell, 2021;Engesser, Ernst, et al., 2017;Engesser, Fawzi, & Larsson, 2017;Ernst et al., 2019;Esser et al., 2016;Gerbaudo, 2018;Hameleers, 2019), the social media are an ideal arena for populist political communications.Engesser, Fawzi, and Larsson (2017), in their analysis of how populist logic is linked to online structures of opportunity, refer to three aspects.First, populists bypass gatekeeper media and sidestep journalistic values in social media with personal and sensationalist communications (Engesser, Ernst, et al., 2017) that are deliberately people-centred (Gerbaudo, 2018).Second, non-elite participation in communications fosters anti-elitism in social media (especially Twitter), which has structurally transformed the public sphere into multiple peripheral spheres that interact with the central sphere (Sampedro, 2000(Sampedro, , 2021)).Finally, in relation to the dangerous others, since social media tend to foster homophily-i.e., the creation of echo chambers in which people hold similar opinions and share messages that confirm those shared views (Barlett, 2014)-populists become isolated in online communities, and this results in the formation of an in-group that excludes those considered as belonging to the out-group.
Because social media facilitate the transmission of messages and the creation and mobilization of homogeneous communities, populists make extensive use of different media depending on their goals.According to Hameleers et al. (2021, p. 12), if the objective is "to spread novel information on their issue positions, Twitter may have a wider reach among new segments of the audience".Van Kessel and Castelein (2016) affirm that, by enabling unmediated communications by politicians with their followers, Twitter plays a key role in social media, most especially due to inbuilt interaction mecha-nisms such as the retweet, mention, response, and hashtag functions.
The structure of social media communities is a widely studied research field (Guerrero-Solé et al., 2022).Regarding online populist communities, research has mainly focused on how social media are used for communications and have focused less on the inclusionary and exclusionary messages typical of populist discourse (Hameleers, 2019).One exception is Åkerlund (2020), who reports on how influencers contribute to reproducing the language of the far-right on Twitter.However, we still know little about how and to what extent messages are replicated by online communities.Boulianne et al. (2020) question the notion of the echo chamber in relation to the articulation of online populist communities, given that the concept has not been well defined.Any proper definition needs to focus on the level of isolation of individuals and on discourse homogeneity within echo chambers.We take into account the two dimensions of content and of relationships, i.e., we identify both the morphology (form and structure) of communities and their main articulating profiles.
Beyond the conceptual debate, at a formal level on Twitter, retweeting can often function as an echo chamber because it tends to signal ideological affinity (Alonso, 2018).In the context of a retweeting network generated around an electoral campaign, analysing profiles can indicate how influence is articulated online (Bruns & Highfield, 2015;González-Bailón & Wang, 2016).
Graph theory offers a number of metrics to calculate network node centrality, i.e., relevance (Boutet et al., 2013;Carnia et al., 2021).Using graph theory, we can map the most retweeted profiles that construct the public arena as follows: builder profiles (they retweet the most), bridge profiles (they interconnect remoter network regions; Cherven, 2015), and influencer profiles (they are the most retweeted profiles).

This research aims to answer the following research questions in relation to the Spanish 2019 general election campaign:
RQ1: To what extent was the VOX online community homophilic?RQ2: Did the VOX online community reproduce the populist framing metaphors used by the party and its leader?
To answer the two questions, we used different samples and different methodological approaches.Data were collected for the 15 official election campaign days (12-26 April 2019).
To respond to RQ1, we analysed a sample (n1) composed of all tweets and retweets (n = 917,010) posted under the hashtags #28A, #28Abril, and #28AbrilElecciones.R statistical language was used to develop the code to collect and clean the data.
Social network analysis, a tool that maps and measures social relations, was used to analyse network form and structure (i.e., to determine which communities were created and how they were related) and to profile the VOX online community and its constituent influencers, builders, and bridges from centrality metrics (Cherven, 2015;Scott, 2013;Scott & Carrington, 2012).The centrality metrics identified the most relevant nodes in the network and enabled communicative functions to be associated with profiles identified in the network of influence.Metrics were as follows: Influencers were determined from the weighted in-degree (WI) metric, which calculates the number of incoming weighted links to nodes receiving the most retweets; builders were determined from the weighted out-degree metric, which calculates the number of outgoing weighted links to other nodes (reflecting the most prolific retweeters); and bridges were calculated as the nodes that obtained the highest betweenness centrality scores, a metric that identifies the nodes that most interconnect different network regions or clusters (Cherven, 2105;Newman, 2010).
Influencers were grouped into five categories according to the description provided in their profile: (a) citizens (self-identified citizens independent of any association or political party); (b) media (including journalists); (c) non-VOX political actors (supporters and leaders of other political parties); (d) VOX political actors (VOX supporters and leaders, including regional VOX sections); and (e) miscellaneous, including unidentified Twitter users (with opaque profile descriptions), citizen associations, and security forces.
To respond to RQ2, we analysed a sample (n2) composed of metaphorical expressions (n = 984) used by the leading 50 influencers in the online VOX community.We stopped including new profiles once saturation point was reached, i.e., once detected metaphors began to be similar to those already collected.
A qualitative methodology based on critical metaphor analysis (CMA; Charteris-Black, 2004) was used to determine whether influencers reproduced populist framing metaphors used by VOX.The CMA was based on the three steps of identifying, interpreting, and explaining metaphorical expressions.We thus identified expressions referring to a semantic field that diverged from the remaining content-generating what Charteris-Black (2004) calls "semantic tension"-and grouped those expressions in terms of broader conceptual units (SDs).Once a word or a set of words have been identified as potentially metaphorical, researchers checked their original meaning in the Real Academia Española's dictionary of the Spanish language in order to confirm that these words were being used in a different way from the one that was intended in their most basic meaning.Interpretation and explanation were based on relating salient items from the identified SDs with TDs taking into account the context, which throws light on which narrative or definition of a situation arises from the use of the metaphor in question.
Two researchers coded the sample manually according to the variables listed in Table 1.To minimize subjectivity in metaphor classification, reliability was calculated by running Krippendorff's alpha inter-rater agreement test on 10% of the dataset for SD, TD, populist frame, and populist component.The high score (over  = 0.862) obtained indicated strong inter-rater reliability.

Research Question 1
To answer RQ1, we needed to determine how the retweeting network was configured during the election campaign period.The network consisted of 201,665 nodes and 527,372 links, with nodes reduced to 192,364 once zero-degree nodes were removed.The network clusters were revealed by applying a modularity algorithm with a resolution of one.Figure 1 shows the nodes with the highest WI scores, reflecting the influencers that determined the political affinity of each community.At the network morphology level, left-right polarization is very evident, reflected in both the proximity of libertarian and right/far-right communities (almost 25% of all the network nodes) and their clear isolation from the rest.
In the figures below referring to influencers, builders, and bridges, node and text sizes are proportional to the metric scores, and the colours represent the different communities (see the legend of Figure 1).

Influencers
Figure 2 shows the 527 nodes with the highest WI, receivers of between 213 and 30,405 retweets, and representing 0.27% of the total.A total of 107 nodes (representing 20.3% of the top influencers) corresponded to the right/far-right community.
Table 2 lists the 18 influencers within the VOX community who received more than 1,000 retweets during the campaign.To respond to RQ2, the content of these profiles was analysed after including the next 40 profiles   down the list and after excluding profiles without results and official accounts of parties and presidential candidates (@pablocasado_/@santi_ABASCAL).

Builders
Figure 3 shows the 430 nodes with the highest weighted out-degree scores, receivers of between 100 and 1,475 retweets and representing 0.22% of the total.Of the 430 builder nodes, 28 (6.5% of the top builders) corresponded to the VOX community.Notable is the @el_partal profile, representing the most prolific retweeter in the entire network.The WI score of many builder profiles in the VOX community was zero, indicating that their main function was disseminating messages from other profiles; although outside the scope of this study, it would be of interest to determine whether these were bot-type profiles.

Bridges
Figure 4 shows the 103 nodes with the highest betweenness centrality scores (0.05% of the total).Corresponding to the VOX community were 24 of these 103 nodes (23% of the top bridges).It can be seen that, except within the same cluster or community, the connection between profiles was very weak.This is a further indication of isolation and polarization in the studied network.The data suggest that the bridges mostly acted at the local rather than the global network level.

Research Question 2
In relation to RQ2, 984 metaphors were detected, 554 (56.3%) reflecting populist framing, i.e., the TD referred to the core components of people, elite, and/or dangerous others.These metaphors were labelled as populist metaphors (Table 3).The remaining 430 (43.7%)  metaphors referred to other frames, such as the elections, electoral results, sectoral policies (e.g., regarding the economy), VOX, Santi(ago) Abascal (VOX leader), the security forces, and other political parties.The most metaphorized component was the elite, followed closely by the people (41.88% and 41.15%, respectively) and, at a distance, by the dangerous others (16.97%).
Table 3 shows populist framing by Twitter profile.Citizens used the greatest number of metaphors overall (30.9%), followed closely by non-VOX political actors (29.88%); between them, they accounted for over twothirds of metaphors.The remaining metaphors were accounted for almost entirely by the media and VOX political actors (18.59% and 16.87%, respectively).Citizen profiles made by far the greatest use of populist metaphors (73.77%), while populist and non-populist metaphor use was broadly more balanced for the remaining profiles.The media and VOX political actors used smaller proportions of populist metaphors than metaphors referring to other aspects of the campaign.
Table 4 shows the intensities with which the different profiles metaphorized populist frames, with all the profiles prioritizing at least one of the three components.The elite was metaphorized most by non-VOX politi-cal actors (53.85%), citizens (43.56%), and the media (42.7%), and the people were widely metaphorized by VOX political actors (68.06%) and, to a lesser extent, by non-VOX political actors (40.56%).Metaphors referring to dangerous others were used least (citizens 17.33%, VOX political actors 13.39%, and non-VOX political actors 5.59%), except by the miscellaneous profiles (44%).
Through CMA, we verified whether the online community reproduced metaphors reflected in official communications by VOX and its leadership, which were analysed in a previous work (Capdevila et al., 2022).As mentioned in Section 3, metaphorical expressions do not work in an isolated manner, but they configure wider narratives under particular SDs.Tables 5, 6, and 7 show the most used SDs by Twitter profile for conceptualizing each of the core components of the populist frame in the analysed sample.

The Elite
All the analysed profiles used mainly the conflict/war/ crime SD for framing the elite, with numbers that go from 37.75% (citizens) to 76.92% (VOX political actors).The main metaphor referring to the elite by VOX and its leadership, la dictadura progre (the politically correct dictatorship; Capdevila et al., 2022), was not included in any of the analysed tweets by citizens, although the word progre (progressive) was used, seemingly to refer to the left-leaning government.expressions that conceptualized the government as a traitor, as Spain's enemy or as totalitarian were used instead.Under this SD, the metaphor el gobierno/Pedro Sánchez okupa (squatter government/Pedro Sánchez) did also stand out: Se estará llevando usted las manos a la cabeza si está viendo al presidente okupa @ivanedlm [You'll be putting your hands to your head when you see the squatter president @ivanedlm].(Lauri, 2019) Plagiador, mentiroso, okupa, traidor, corrupto, cínico, payaso, oportunista, inepto, ególatra, psicópata...¿Sigo? Patético [Plagiarist, liar, squatter, traitor, corrupt, cynical, clown, opportunist, inept, egomaniac, psychopath... Shall I continue? Pathetic].(Comando España, 2019) As far as media profiles were concerned, the metaphorical construction of the elite was more diverse, with no particular metaphor predominating.Hence, despite metaphors of war and conflict being the most common, the same metaphorical expression was rarely repeated.Even so, the metaphor of the squatter appeared frequently: Esta es la carta de un ciudadano que el Okupa presentó ante 10 millones de españoles.Dijo que la autoría pertenecía a la Junta de Andalucía, acusándoles de elaborar "listas negras" de trabajadores de violencia de género.Nadie va a emprender acciones judiciales?[This is the letter from a citizen that the Squatter presented to 10 million Spaniards.He says that it was authored by the Junta de Andalucía and accuses it of preparing "blacklists" of gender-violent workers.Is no one going to take legal action?].(Isabel Rábago, 2019) In the case of non-VOX political actors, metaphors referring to the elite were very varied.Most of the metaphors defining the government referred to it as a cowardly enemigo de España (Spain's enemy) that attacks the people and is held hostage by the parties that support it.In all cases, the metaphors reflect the SD of war and conflict (  Capdevila et al., 2022).Only the latter was present in VOX's online community.Table 6 shows how the analysed profiles used the SD of Personification to conceptualize the people.Under this category, the metaphor of "living Spain" was predominant.
The España viva metaphor (and its diametric opposite, España muerta [dead Spain]) stood out between citizens, the media, and VOX political actors, who reproduced this metaphor, mainly using hashtags, in many tweets referring to the people: El próximo #28A tenemos que elegir entre #LaEspañaViva y esta otra España: #LaEspañaMuerta Yo elijo #VOX [On #28A next, we have to choose between #LivingSpain and that other Spain: #DeadSpain.I choose #VOX].(Felipe G. Aguirre, 2019) The España viva metaphor was not mentioned even once by non-VOX political profiles, despite them using the Personification SD when giving Spain the ability to be "strong," "to wake up every morning," or "to grow up."Rather they tended to use Mechanics/Physics metaphors regarding the people, referring to aspects such as their unity and the risk of breaking it if there was no change in government.

The Others
As mentioned, VOX and its leadership focused on immigrants, the media, and Catalan separatists as dangerous others.Inside VOX's online community, Catalan separatists were the most metaphorized group, whereas the analysed profiles did not conceptualize media and immigrants that much (Table 7).
Immigrants were conceptualized as savages by VOX and its leadership (Capdevila et al., 2022), but this metaphor was not used by any profile in their online community.The groups that did metaphorize immigrants used metaphors such as madness and control (citizen profiles and media profiles, respectively): Proyecto sharia, activo en Londres, atacan a quien bebe alcohol, mujeres con minifalda y a homosexuales.También activo en Francia o Países Bajos.No podemos permitir q llegue a España, sólo @vox_es puede parar esta locura.El #28a #PorEspañaVotaVOX [The Sharia project, active in London, attacks those who drink alcohol, women in miniskirts, and homosexuals.Also active in France and the Netherlands.We cannot allow this to happen in Spain, only @vox_es can stop this madness.On #28a #ForSpainVoteVOX].(David Lorenzo, 2019) Un menor cántabro de 17 años ha muerto en San Sebastián tras recibir una brutal paliza a manos de siete jóvenes de "nacionalidades diferentes".Ya nos podemos hacer una idea de sus nacionalidades.Es hora de controlar la inmigración [A 17-year-old minor from Cantabria has died in San Sebastián after receiving a brutal beating at the hands of seven young people of "different nationalities."We can have a good idea of their nationalities.It's time to control immigration].(Caso Aislado, 2019)

Discussion and Conclusions
As indicated by Boulianne et al. (2020), the literature on the impact of social media on the spread of populist ideas is based, first, on an imprecise definition of echo chambers and second, on a lack of determination of echo chamber limits at the levels of content transmission and of relationships between profiles.In this article, we tried to throw light on those issues.First of all, in relation to RQ1, we have shown that the community that formed around VOX during the April 2019 election campaign could not be considered fully homophilic, as it did not reflect an echo chamber in VOX-supporting profiles were isolated from the rest of the Twittersphere.As indicated by Moragas-Fernández et al. (2019), the Twittersphere is divided into thematic clusters that are part of broader spheres articulated in different ways depending on the issue.During the 2019 electoral campaign, the Twittersphere adopted a specific morphology.Regarding findings for RQ1, the network morphology was clearly polarized along an ideological left/right axis, but a more precise perspective of the right revealed the coexistence of different communities interacting with VOX.There was, therefore, no absolute homogeneity in relationships, as the populist profiles of the far-right interacted with other more moderate political and media profiles.
Second of all, in relation to RQ2, the echo chamber at the node level was also found to be imperfect with regard to the transmission of populist messages.While the official VOX and Santi(ago) Abascal accounts transmitted populist messages-and so were truly populist in reproducing all the components of the populist ideology (Capdevila et al., 2022)-this was not the case with the community more broadly.Considering Jagers and Walgrave's (2007) four types of populism, i.e., complete, anti-elitist, excluding, and empty (reflecting the presence and/or absence of the various populist ideological components), we found that the analysed Twitter profiles prioritized those components differently in their communications (see Tables 5, 6, and 7).Thus, non-VOX political communications reflected a mix of anti-elitist populism and empty (people-centred) populism (see Table 4); a similar pattern was observed for citizen profiles and, to a certain extent, for media profiles (although in this case, there was also a certain degree of metaphorization of dangerous others); VOX political communications reflected mainly empty populism (around 70% of their metaphors referred to the people); and finally, unidentified profiles mainly reflected excluding populism that marginalized dangerous others, with the anonymity behind which some of those profiles hid clearly reflecting more aggressive communications.
The results of the analysis also show that the reproduction of metaphors used by VOX and referring to the different populist components was only partial.Essentially, VOX was only successful in reflecting Catalan pro-independence politicians as insurrectionists.This metaphor predominated (practically exclusively) in all the profiles analysed (see Table 7).VOX managed to partially convey the España viva metaphor to refer to the in-group (see Table 6).This metaphor, mainly transmit-ted as a hashtag, was broadly reproduced by citizens and the media, but far less so by non-VOX political actors.The metaphor of la dictadura progre used to refer to the elite had less impact, as it went largely unreproduced by VOX political actors.
The analysed profiles used metaphors mostly for the elite (see Table 4), and the metaphors used to conceptualize the elite were very varied.The metaphor most used by citizens and the media, and widely reproduced by VOX political actors, was that of the gobierno okupa.Non-VOX political actors constructed a metaphorical framework based on conflict regarding the elite, considered a cowardly enemy of Spain that attacks citizens and is held hostage by political supporters.In the echo chamber that was formed in the VOX online community during the April 2019 Spanish elections, therefore, metaphors resonated that offered different visions of the three populist components of people, the elite, and dangerous others.The diversity in disseminated content may have been marked by the fact that the community was not completely ideologically homogeneous, as evidenced by its influencers interacting with other profiles.
Beyond answering the research questions, in this article, we have also provided a methodological proposal based on triangulation for analysing the two levels of echo chambers.Social network analysis has proven to be a reliable method for community detection and for identifying the relations established between profiles at a morphological level, while CMA has let us approach small data, which, despite not being statistically representative, is necessary for determining how echo chambers operate at a content level.Likewise, we have brought forth an analysis that shows the way metaphors construct the populist frame, and so we have contributed to filling the gap mentioned in the literature section.Considering this, the aim of the article was not so much to reflect on what the results imply for VOX, but to argue that the way in which the populist frame is disseminated online is not perfect, because its public is formed by a heterogeneous community (or at least not as homogeneous as thought).
Our findings may be affected by certain limitations.In the first place, the election campaign may have conditioned the content transmitted by VOX and its community, as legislation governs what can/cannot be broadcast by political parties during election campaigns.Furthermore, the fact that parties compete with each other for a common electorate may have meant that non-VOX political actors were more reluctant to convey metaphors generated by that party.The question remains as to whether, outside the election campaign period, the analysed profiles would have behaved in the same way or whether metaphors would have been transmitted differently within the community.A final limitation is that only one populist party from the far-right was analysed, so it remains to be seen whether the communities of other populist parties, whether of a similar or different ideology, would behave in a similar way.All these limitations can be addressed in future research.

Figure 1 .
Figure 1.The main communities in the retweet network.Note: Numbers to the left of the labels are codes.

Table 1 .
Coding variables with #LivingSpain as an example.
Comment"Living Spain" reflects those who do not feel represented by the left-wing government (the elite), but now have a chance to rule thanks to VOX.

Table 2 .
Influencers in the VOX community with more than 1,000 retweets.

Table 3 .
Metaphor use by Twitter profile.

Table 4 .
Populist frames used by Twitter profiles.

Table 5 ):
(Teodoro García Egea, 2019)ro Sánchez y el adversario del PP es el PSOE.Advertir del peligro que es Sánchez no es un eslogan sino una realidad.Es una emergencia nacional que salgan de La Moncloa el #28A [Spain's enemy is Pedro Sánchez and the PP's adversary is the PSOE.To warn of the danger implied by Sánchez is not a slogan, but a reality.It is a national emergency that they exit La Moncloa on #28A].(TeodoroGarcía Egea, 2019)

Table 5 .
Source domain used by Twitter profile for referring to the elite.

Table 6 .
Source domains used by Twitter profile to refer to the people.

Table 7 .
Source domains used by Twitter profile to refer to the other.