Racial Disinformation, Populism and Associated Stereotypes across Three European Countries during the COVID-19 Pandemic

Abstract

Within the realm of disinformation, across all media platforms, a crucial subject of interest seems to be immigration, which produces the so-called ‘racial hoaxes’. Racial hoaxes are closely linked to the spread of populist ideologies and ethnic stereotypes, both of which are psychosocial processes that, during health crises, could acquire particular features based on cultural differences. This study analysed the main contents of 239 racial hoaxes in relation to three main features: the country of origin (i.e., France, Italy and Spain), the stereotypical contents, and the COVID-19 pandemic period. The results highlight some similarities across these three countries but also peculiarities in terms of topics and stereotypical contents that were magnified during the pandemic period. The peculiarities of emerging racial hoaxes are discussed in relation to the stereotype content model and the literature on populist discourses, providing valuable information for use in psycho-educational intervention, policymaking and social integration efforts.

1. Introduction

The most advanced innovations in communication brought about by the digital era have sparked an exponential increase in the critical phenomenon of information disorders, which pose a significant threat to modern society due to their harmful and far-reaching impacts.

Researchers agree on a categorisation of information disorders based on three types of warped information: misinformation, disinformation and misinformation (Frau-Meigs 2019; Lewis and Marwick 2017; Wardle and Derakhshan 2017). These processes, although slightly different from each other, involve the spreading of incorrect or untrue information to erode trust and hinder effective information sharing among people on a larger scale (Fallis 2015). In particular, disinformation relies on adversarial narratives and thrives on identity-based controversies: through multiple rhetorical strategies and forms of knowing, including truths, half-truths and value-laden judgments, it exploits and amplifies identity-driven controversies (Diaz Ruiz and Nilsson 2023). As such, the concept of disinformation goes far beyond what is false or not, as it can also use the truth and parts of the truth to disinform (Brisola and Doyle 2019).

The act of deliberately creating false or misleading information has given rise to the phenomenon of fake news (Lazer et al. 2018), meant as entire fabrications (Imhoff and Lamberty 2020) intentionally and verifiably false (Allcott and Gentzkow 2017) albeit having a realistic ground which makes them believable. Fake news is difficult to define clearly, but researchers seem to agree that it is a product of intentional deception of a mass audience by non-media actors via sensational communication that appears credible but is actually crafted to manipulate and be concealed as false (Finneman and Thomas 2018). What makes fake news particularly harmful is its parasitic nature, as it feeds off traditional news outlets, simultaneously benefiting from and undermining their credibility (Lazer et al. 2018). The emotional framing of fake news is what differentiates it from mainstream news sources, as it uses an increase in the surprise factor to influence how the information is engaged with, shared among and, later, better recalled by the public (Bessi et al. 2015; Chen et al. 2015; Horne and Adali 2017; Paschen 2019; Scardigno et al. 2023; Taute et al. 2011).

This study focused on a particular type of fake news, racial hoaxes, that appear to be multifaceted in origin and to have deep sociocultural roots. Especially they work following specific psychosocial processes, some of which fit in with those involved in stereotyping dynamics and populist movements. Based on the theoretical background and models discussed in the first part of this paper, we analysed the contents of racial hoaxes—identifiable as hoaxes since their source and nature—in terms of their stereotypical differences before and after the COVID-19 pandemic and among three countries in Mediterranean Europe that are strongly interested in immigration (i.e., France, Italy and Spain). We also analysed whether these contents, given their association with misleading news, reflected the characteristics of the populist discourses of the three countries.

1.1. Racial Hoaxes and Their Associated Stereotypes

Racial hoaxes, a particular type of informative disorder (Wardle and Derakhshan 2017), can be defined as communicative acts with false, distorted and misleading content in the form of threats to people’s health and safety, in which the culprit is a person or a group of people described in terms of their ethnicity or nationality (Cerase and Santoro 2018; D’Errico et al. 2023).

Specifically, this well-established definition of ‘racial hoaxes’ refers to ‘race’ as an umbrella term, including different types of social belongings that can be the object of prejudicial attitudes. Thus, it does not at all refer to the strict and original concept of “race”, having historical and socio-cultural connotations when thinking, as an example, of the American vs. Mediterranean areas.

Racial hoaxes can include a bias towards the person responsible for the action, which can be conveyed using marked or implicit negative evaluation. In marked negative evaluation, the news contains emotive and sensationalistic language (language bias), exaggeration or overinterpretation of the fact (factual bias) or description of the fact that considers only one side instead of the two or more sides of the issue (Litovsky 2021). Lewandowsky and Yesilada (2021) showed how radical-Islamist disinformation is characterised by fallacies such as polarisation, hasty generalisations and invoking of emotions. More in detail, although no causal relations between disinformation and hate speech were found (Cinelli et al. 2021; Vasist et al. 2024), these two phenomena are connected: they seem to have the same causes and polarising effects on societies (Bader and Bender 2022), as it was found when people justify and defend the use of hate speech through misinformation (Kim and Kesari 2021).

Misleading racial news generally contains typical linguistic forms of stereotypes, mainly focused on the absence of morality and the presence of prejudices aimed at dehumanising and attributing various types of threats to the people considered as ‘outgroup’ members (D’Errico et al. 2022). A powerful way to propagate biased information against immigrants is by reinforcing people’s anti-immigrant attitudes, as Wright et al. (2020) demonstrated in an adult sample and as Wright and Duong (2021) demonstrated in young people about the fake news against Asian Americans during the pandemic.

The psychosocial literature has highlighted how, from the theoretical model of content stereotypes, the contents of the stereotypes can be associated with different types of prejudice (Bye 2020; Fiske et al. 2007). Considering the evaluation process underlying stereotypes or prejudices, an individual can negatively evaluate and, thus, discredit the image of the other in various ways that are useful in identifying the different forms of discrediting (D’Errico and Poggi 2014).

According to the stereotype content model of Fiske et al. (2007), people can be stereotyped in two main dimensions: communion and agency.

In communion, individuals are associated with a lack of benevolence, dominance up and a lack of warmth. A recent study based on stereotypical social media contents regarding immigrants (Bosco et al. 2023) showed that immigrants tend to be described as thieves, rapists, murderers, criminals or exploiters (lack of benevolence); or are generally described as dangerous, overbearing and arrogant (dominance up) or as lacking empathic emotional competence (lack of warmth).

As for the dimension of agency, stereotypical contents can refer to a lack of cognitive competence, a lack of physical competence and dominance down (Abele et al. 2016; Poggi et al. 2013). In this case, immigrants are described as incompetent (lack of cognitive competence); dirty, sick or ugly (lack of physical competence); or parasites, do-nothings, complaining slackers or people who passively take advantage of the system (dominance down).

We used these theoretical models as the basis of our analysis in this paper, which will be defined in the next section.

1.2. Fake News and Racial Disinformation during the COVID-19 Pandemic

Within the realm of disinformation and fake news, across all media platforms, a crucial subject of interest seems to be immigration. Due to ongoing migratory flows, immigration has gained much attention as a major societal challenge. The use of fake news in this field is especially important because exposure to fake news that portrays minority groups in a negative light can have detrimental effects: fake news not only reinforces negative sentiments and prejudices (Dijk 2000) but can also invite the reader to seek more information that is consistent with and confirms pre-existing negative views (Atwell Seate and Mastro 2016; Dijk 2000; Wright and Duong 2021).

Immigrants are perceived as threats—both in truth and symbolically—because they jeopardise job availability or stability, as well as the national and religious identities of the natives in the countries where they reside (D’Errico et al. 2022). These negative and stereotypical perceptions can be further reinforced through a highly targeted type of fake news: racial hoaxes, which are stories that are carefully fabricated to legitimise and promote discriminatory and racist attitudes and beliefs against minority groups (Papapicco et al. 2022).

These attitudes can have a greater impact in uncertain times and in periods of severe collective economic, social and health challenges, such as in the COVID-19 pandemic. The reference to the COVID-19 virus as ‘the foreign virus’, ‘the Chinese flu’ or the ‘Wuhan virus’ portrays the virus as a disease of the ‘Other’ (Ivic and Petrovic 2023). Othering is the construction of positions for a group of people as ‘us’ by designating a certain category of people as ‘them’ (Liu and Self 2020). These dynamics, which are the basis of xenophobia and discrimination, already exist in various societies but become more penetrating in times of crisis when the existing patterns of division become more focused and more critical. In those times, the binary oppositions, such as we/they, self/other, civilised/barbaric and culture/nature, are emphasised. This contraposition was primarily experienced during the pandemic in reference to China and Chinese people: public discourse, media narratives and social media ascribed the COVID-19 pandemic to unhealthy culinary habits (Moreno Barreneche 2020), thus blaming China as a barbaric threat to the Western way of life (Yee 2020). Chinese people were perceived according to the phenomenon of the triple conflation (Gao 2022)—the intermingling of the health, racial and political or national domains.

1.3. Racial Hoaxes and Populism in France, Italy and Spain

Populist phenomena are paradoxically widely developed and consolidated in the sociopolitical context of Western European democracies and in the sociocultural context of digital technologies. Corbu and Negrea-Busuioc emphasised the following similarities and interconnections between populism and fake news in the new media ecosystem (Krämer and Holtz-Bacha 2020).

First, the new media landscape emphasises a new media logic that shifts from accuracy, facticity and objectivity to immediacy, rapid circulation and emotionality, thus catalysing the amplification of attractive populist and fake content. Second, in both cases, media stereotypes—mass-mediated repeated pairings of social groups with specific attributes (Arendt et al. 2015)—can act as primers since they form memory traces that become part of common knowledge and can be quickly reactivated even by short exposures to them. Thus, fake news is perceived as credible when it focuses on cognitive resonance with prior stereotypes and when it blames the (political) elite and offers highly polarised content.

Third, emotions, such as enthusiasm, fear and anger, commonly contribute to making populist messages more appealing. Anger retrieved the Manichean split into antagonistic groups (Rico et al. 2017)—which offended the elite versus good people. The chains of reactions involving emotional activation make both populist content and fake news viral and the subjects of vivid debates, and echo-chamber and segregation dynamics expand the potential of these contents to become viral (Törnberg 2018).

The results of the national elections in European countries and of the European Parliamentary elections testify to the increasing role of populism across European countries. However, populism is a multifaceted phenomenon with deep sociocultural roots. Recently, France, Italy and Spain have experienced some common political trajectories related to public mistrust, economic and responsiveness crises, erosion of traditional parties and the emergence of new political actors due to discontent with established parties (Boscán et al. 2018). However, more specific pathways can be outlined.

In France, in the wake of the overall state of instability and turmoil that characterises many democracies in Europe due to external and internal destabilising factors (Betz 1994) and in line with the populist tradition started by Boulangism in the 19th century, the Rassemblement National party rose as an alternative to the mainstream parties, which had failed to address heavy issues such as terrorism, immigration and unemployment. This movement presented itself as an opportunity to save the French national identity and to free the country from the decline caused by the threat of foreigners, who were identified as “daily viruses” (Abdeslam 2021). This party is considered the first to stress the question of immigration in the triangulation between the French and immigrants by pointing to nationalism and radical conceptions of otherness (Benveniste and Pingaud 2016), thus embodying the main features of exclusionary populism (Mudde and Kaltwasser 2017) through the slogan ‘the French First’.

Italy, in the last decades, has been characterised by the durable success of its populist parties: Lega Nord (LN) in the late 1980s, Forza Italia in the early 1990s and M5S since 2013. More recently, Italy has also experienced the rise of Brothers of Italy (FdI), whose great achievement in the 2022 general election allowed its leader, Giorgia Meloni, to become the first Italian female Prime Minister (Baldini et al. 2022). All the cited parties contributed in different ways to the creation of a type of populism that included parties that were mostly right-oriented (Bertero and Seddone 2021). This scenario was specifically amplified in 2013 when a major change occurred in LN: Matteo Salvini proposed an ideological redefinition of the party as a far-right populist party. Thus, patriotism replaced regionalism, and the European Union and immigrants replaced Rome and Southern Italians as the people’s enemies (D’Alimonte 2019). As Italy rode the wave of the European migration crisis and because the country is a focal point for maritime arrivals due to its geographical position, populist parties capitalised on the fears and threats related to immigration. Anti-immigrant political arguments became the focus of the campaign of LN (Cervi et al. 2020), which had been described as the first political entrepreneur of xenophobia (Avanza 2010). In addition, Meloni’s rhetoric frames opposed the ethnically defined national “us” to globalising forces such as transnational market integration and migratory flows (Trillò and Starita 2023).

Enríquez and del Carmen (2017) referred to Spain as an exception to other European countries in its immunity to the appeal of right-wing populism, as seen in the results of its national elections (Enríquez and del Carmen 2017). This attitude was linked to the historical weakness of the Spanish national identity, the Spanish people’s pro-Europeanism, the failure of parties that had tried to appeal to right-wing populist sentiments in Spain and the effects of the Spanish electoral system. However, in the April 2019 general elections, one of the winning parties was Vox, a new right-wing populist party. As left-wing populism was already represented by Podemos since the 2015 elections, right-wing and left-wing populism co-existed (Vampa 2020). These two opposing movements have competed on several topics, such as territorial mobilisation (the so-called ‘centre-periphery cleavage’) and several dimensions of globalisation, even involving immigration issues.

1.4. Current Research

Aim and Hypothesis

This study was conducted to analyse the main contents of a specific type of fake news, racial hoaxes, in relation to three of their main features: (1) their country of origin (France, Italy and Spain); (2) their stereotypical contents (lack of benevolence, cognitive competence, physical competence and warmth, and active and passive dominance); and (3) whether they occurred before or during the COVID-19 pandemic (i.e., before or during, starting from March 2020).

Furthermore, in line with the literature reviewed in the previous section, we determined whether these variables can be associated with populist discourses, considering potential differences among the three national contents considered.

2. Materials and Methods

2.1. Corpus and Coding Process

The development of the Racial Hoaxes Corpus emerged from a collaborative European project called ‘Stereotypes’ in which social psychologists and computational linguists in a research group collected 239 racial hoaxes—70 of them in French, 97 in Italian and 72 in Spanish. These hoaxes were sourced from fact-checking websites or newspaper articles in French, Italian and Spanish that aimed to verify or refute claims circulated on social media or other news outlets related to immigration. Specifically, AFP Factuel and Les Décodeurs in Le Monde were used for the French hoaxes, Bufale.net and Butac for the Italian hoaxes, and Maldita.es and Newtral for the Spanish hoaxes. The debunked racial hoaxes in the corpus were from 2019 to 2022. After the racial hoaxes were collected and their country of origin was determined, they were further categorised into pre- and post-pandemic based on the cutoff date of 10 March 2020.

Subsequently, stereotypes related to immigrants that were found in the title or the text of the hoax were identified in a previous study (Bourgeade et al. 2023) by encoding them according to the classical approach to stereotype contents (Fiske et al. 2002), with lack of benevolence, cognitive and physical competence, or warmth; and active or passive dominance (Bosco et al. 2023). Finally, the stereotypical contents were coded by two judges for each corpus, and the French and Spanish corpora achieved very good levels of agreement, while the Italian corpus achieved moderate agreement (Bourgeade et al. 2023).

Table 1. Descriptive Percentages related to the content of the stereotype and period differentiated by country of origin.

	Stereotypes Content						COVID-19 Pandemic
	B	P	DD	DU	W	C	Pre	Post
Italy	35	15	24	23	0	0	48	49
	36.1%	15.5%	24.7%	23.7%	0%	0%	49.5%	50.5%
France	10	1	28	31	0	0	54	16
	14.3%	1.4%	40%	44.3%	0%	0%	77.1%	22.9%
Spain	21	6	16	29	0	0	32	40
	30.9	8.3%	22.2%	40.3%	0%	0%	44.4%	55.6%

shows the relative frequencies and percentages of the stereotype contents and periods differentiated by country of origin. Our corpus did not have racial hoaxes with stereotypical content related to a lack of warmth and competence.

2.2. Data Analysis Procedure

Before the analysis, the entire corpus was translated into Italian. The psycholexicometric analysis was performed using T-LAB software (version Plus 2017) for content analysis and text mining (Lancia 2004). The corpus—a collection of comparable texts (Bolasco 1998)—had a total of 19,839 words, 3252 hapax (words with only one occurrence), and 5247 forms (different words used in the corpus). The texts were preprocessed automatically using T-LAB, which included the following processes:

-

Normalisation: file transformation to ensure proper recognition of graphic forms by resolving different ambiguities (e.g., removing excessive spaces or reducing capital letters);

-

Stopword removal: elimination of terms that did not convey specific and/or relevant contents (indefinite adjectives, articles, adverbs, exclamations, interjections, prepositions, pronouns, auxiliary verbs and modal verbs);

-

Multiword verification: identification of sequences of two or more words that constitute a lexical unit, including subsets such as compound nouns and idiomatic expressions;

-

Text segmentation into elementary contexts, with each context corresponding to a transcribed transmission;

-

Lemmatisation: restoration of inflected forms to their original form (masculine and singular; Lancia (2004)). The lemmatisation was later reviewed manually by two independent judges to remove lemmas deemed uninteresting for subsequent analyses.

Keywords—lexical units included in the analyses—were selected by two independent judges, following the criteria of frequency and semantic relevance (Lancia 2007; Melotti et al. 2023). The next section presents the results of the following two types of analyses conducted:

-

Specificity analysis differentiated in ‘typical lexical units’: comparison of the relative frequencies of the lemmas in a portion of the corpus with the expected theoretical frequencies (Giuliano and La Rocca 2012) and ‘exclusive lexical units’ (i.e., the lemmas present only in a specific portion of the corpus), performed and reported for each country; and

-

Correspondence analysis: qualitative–quantitative technique of factorial analysis applied to categorical data based on the co-occurrence matrix. It allows for qualitative interpretation of the factorial plane based on the proximity or distance and/or similarity or dissimilarity between lexical units. The closer the lexical units are, the more frequently they co-occur in the same lexical context, and the more distant they are, the less frequently they co-occur (Benzécri and Bellier 1980). This analysis was carried out and reported in the first case, considering only the country-of-origin variable, and in the second case, considering the stereotype content and pandemic period variables.

Finally, the results of the qualitative analyses of the concordances of specific words were integrated into the discussion to allow for the interpretation of not only the usage frequencies but also the specific ways in which words were employed in the context of a sentence, enabling a ‘situated’ interpretation.

3. Results

3.1. Specificity Analysis

The identified exclusive and typical lexical units by country are presented in this section. The terms that belong to a common sphere are included in the shared categories.

Considering the French subcorpus, issues emerged in the following domains in capitalised relation to the significant words (

Table 2. Specificity analysis for French RH.

	Typical Lexical Units
	LEMMA	SUB	TOT	CHI2	(p)
National cues	FRANCE	21	29	107.59	0.000
	FRENCH	12	13	84.58	0.000
	NATIONAL	5	11	12.68	0.000
	Exclusive lexical units: MARINE (7), PEN (7), CALAIS (3), ROUBAIX (3)
Economic domain	EURO	15	36	32.92	0.000
	RECEIVE	5	8	20.75	0.000
	PENSIONER	3	4	16.06	0.000
	TAXPAYERS	3	4	16.06	0.000
	JOB	4	8	11.84	0.000
	ACCOMMODATION	3	5	11.73	0.000
	ALLOWANCE	2	3	9.10	0.002
	PUBLIC HOUSING	2	4	5.92	0.014
	PENSION	3	8	5.42	0.019
	Exclusive lexical units: CREDIT CARD (6), CREDITED (3)
Religious lexicon	RAMADAN	5	6	30.83	0.000
	WEAR	3	4	16.06	0.000
	MUSLIM	7	21	10.07	0.001
	PRAYER	3	8	5.42	0.019
	CROSS	2	5	4.06	0.043
	Exclusive lexical units: SERMONS (3), HEADSCARF (3)
Migratory domain	IMMIGRATION	11	21	35.11	0.000
	IMMIGRANT	57	272	25.81	0.000
	REFUGEE	4	6	18.20	0.000
	REUNIFICATION	3	4	16.06	0.000
	ASYLUM	7	19	12.25	0.000
	APPLICANT	4	8	11.84	0.000
	ARAB	3	5	11.73	0.000
Threat/danger	CONTAINMENT	2	3	9.10	0.002
	INVASION	2	3	9.10	0.002
	STEAL	3	6	8.88	0.002
	EXPOSURE	2	4	5.92	0.014
	Exclusive lexical units: INSECURITY (3), INVADE (3)

):

(1)

National cues, where words related to the French people, politicians and locations were found. In our corpus, these cues were intended to emphasise the economic threat posed by immigrants, the individuation of France as a landing country for migrants without any control, the national identity as downgraded and contaminated (as in the example ‘A third of marriages are with foreigners who become French’) and the contextualisation of specific reported events.

(2)

The economic domain, where the included lemmas and their sentences emphasise the favourable conditions for migrants, the complementary feelings of unfairness when thinking of the French’s conditions and the denouncement of the welfare spending for migrants (e.g., ‘Pierre, a retired farmer, lives on 284 euros per month. Unfortunately for him, Pierre is not a migrant. Asylum seekers can receive a monthly allowance of 330 euros from the state’.).

(3)

The religious lexicon, the main objective of which was to stress the perceived threat in both temporary situations (e.g., risks related to the COVID-19 pandemic) and more general sociocultural traditions (e.g., ‘In Clichy, a number of rather shocking things are happening: Muslims praying in the streets every day and elected officials who can not take it anymore…’) An additional function of the religious domain is to mock migrants’ habits (e.g., ‘On this first day of Ramadan, the photo of a cow-shaped box laughing with an Arabic design is being shared on social networks as a special Ramadan promotion’).

(4)

The migratory domain, whose lexicon is related to accentuating the criminal rate (e.g., ‘Chloë, 9 years old, rapt, killed and abused in Calais by a migrant’).

(5)

Both specific and wider threats or dangers concerning the national identity, the extent of the migratory flows, and more general socio-political and economic problems, as in the following example: ‘We still have statistical evidence of the link between mass immigration and a staggering increase in insecurity’.

As for the Italian hoaxes (

Table 3. Specificity analysis for Italian RH.

	Typical Lexical Units
	LEMMA	SUB	TOT	CHI2	(p)
COVID-19 pandemic emergency	COVID	74	77	11.19	0.000
	CENTRES	65	68	9.19	0.002
	POSITIVE	27	28	4.22	0.040
	Exclusive lexical units: QUARANTINE (34); SANITARY (21); CONTAGION (17); CASE (12); COVID-19 swab test (16); OUTBREAK (12); VIRUS (10); GATHERINGS (8).
National cues	ITALY	38	39	6.70	0.009
	Exclusive lexical units: ITALIAN (55), SALVINI (23), CONTE (21), MATTEO (16), ALPINE (13), MILAN (12), ROME (12), SICILY (12), MINISTER (8), NAPLES (8), PALERMO (8), ALASSIO (7), LAMORGESE (7), VENTIMIGLIA (7)
Landing and hosting	Exclusive lexical units: SITUATION (20), BOARD (19), GUEST (17), CASE (12), EPISODE (11), REPATRIATE (8), CLIMB (8), LAND (8), UMPTEENTH (7), LANDED (7), TRANSFER (6), SEA (6), HOTSPOT (5), INTEGRATION (5)
Current affairs and crime	Exclusive lexical units: POLICE (10), ESCAPE (10), CARABINIERE (9), BARRACKS (9), KILLED (9), KILL (8), WEAPONS (7), REJECT (6), KNIFE (5), DENOUNCED (5), INSULT (5)
Socio-economic domain	Exclusive lexical units: BET (10), TWITTER (6), ZUCKERBERG (6), ECONOMIC (5), FACEBOOK (5)

), the most widely represented issues concerned:

(1)

The COVID-19 pandemic emergency. This topic that first emerged had a wide range of subthemes, where migrants are presented both as potential threats (e.g., ‘If the right attention is not paid to ghettos and buildings occupied by foreigners, there is a risk of witnessing a new explosion of COVID-19.’) and as a real danger (e.g., ‘Migrant with COVID repatriated. And now 100 agents are in quarantine.’). In addition, migrants are pointed out as benefiting from unfair aid and as objects of more general public and social debates (e.g., ‘The problem with COVID-19 is not the infection itself: that is surmountable. We have the tools, we have safety measures, and we know that the mortality rate from COVID-19 is comparable to that of any other influenza’.).

(2)

National cues, where a special focus on political personalisation was found. As a matter of fact, wide opposition between the two central national leaders is proposed. On the one hand, Salvini (representing closed-port measures concerning migrants) is depicted as a purposeful leader who takes on important positions and is invoked as a supervisory authority and a victim of partiality and bias; on the other hand, Conte (a leader trying to mitigate too restrictive measures against migrants) is defined as a weak-willed leader and a target of complaints and critics. Beyond these political issues, the national referred lexicon also includes Italian people as victims of extremely violent acts (e.g., ‘Italian boy beaten up by 3 immigrants’) and symbolic victims of unfair measures taking advantage of migrants, who have a dismissive attitude towards Italy.

(3)

Landing and hosting of migrants, where the lexicon includes the several steps and demands concerning these problems, which are connected to several risks (crime and health problems) and to favouritism from political management.

(4)

Current affairs and crime, where objects, actions and culprits give life to a wide range of criminal acts.

(5)

The socioeconomic domain. For example, the world of gambling is cited to emphasise the contradiction between immigrants’ (presumed) hardship and (real) careless behaviours. This domain is specifically marked when it relates to (1), thus connecting the socioeconomic lexicon to the COVID-19 emergency and blaming migrants’ COVID-19 infractions (e.g., ‘Migrants, they don’t care about COVID and DPCM: they crowd Snai betting halls. No rules for them’.).

Some similarities and other peculiarities emerged from the Spanish exclusive and typical lexical units (

Table 4. Specificity analysis for Spanish RH.

	Typical Lexical Units
	LEMMA	SUB	TOT	CHI2	(p)
National and geographic cues	SPAIN	9	10	101.56	0.000
	SPANISH	9	11	90.73	0.000
	Exclusive lexical units: CANARIES (6), MASPALOMAS (5), PODEMOS (4)
Ethnic and religious domain	MUSLIM	13	21	93.17	0.000
	MOROCCAN	7	20	22.85	0.000
	GYPSY	3	5	20.61	0.000
	ISLAM	3	6	16.24	0.000
	CROSS	2	5	7.94	0.004
	ISLAMIC	3	10	7.66	0.005
	MAN	4	20	4.81	0.028
	WOMAN	4	22	3.89	0.048
	Exclusive lexical units: ROMANIAN (5), NATIVE (3)
threat, crimes	ATTACKED	4	6	31.39	0.000
	OFFEND	2	3	15.69	0.000
	RAPE	2	3	15.69	0.000
	ILLEGALLY	2	3	15.69	0.000
	DEAD	2	5	7.94	0.004
	CLANDESTINE	9	52	7.81	0.005
	ARRESTED	2	7	4.71	0.030
economic side of the problem	AIDS	4	5	39.22	0.000
	PAY	5	12	21.07	0.000
	EURO	9	36	16.85	0.000
	EAT	2	3	15.69	0.000
	FOOD	2	3	15.69	0.000
	TOURISM	2	3	15.69	0.000
	TOURISTIC	2	3	15.69	0.000

). In fact, the lemmas that most commonly emerged from the Spanish corpus of fake news focused on the following domains:

(1)

National and geographic cues related to general references and contextualised events.

(2)

The ethnic and religious domains highlighted (a) the contradictions between the native population and migrants, either beneficiary or pretentious, arrogant and dismissive (e.g., ‘If you are a Spanish worker, you pay 1500 euros for [a] truck licence. And if you are an immigrant, the Andalusian government pays for it’). In addition, a too high rate of the migratory phenomenon is denounced.

(3)

Feelings of threat mostly related to crime rates as well as to the paradoxical condition of the Spanish population, which is both economically deprived and at risk of losing traditions. News events are accurately told, and details are usually proposed as aggravating conditions (e.g., ‘14 North Africans affected by scabies arrested for assaulting a girl who was looking for her boyfriend; and when the boyfriend appeared, they stabbed him in the face’). In addition, the condition of clandestinity as a beneficiary of governmental benefits and costs is proposed.

(4)

The economic side of the problem mainly reflects the migrants’ parasitic condition (e.g., ‘The Municipality of Adra once again launches aid to support the unemployed gypsy and Muslim population’).

3.2. Lexical Correspondence Analysis

Lexical Correspondence Analysis (Benzécri and Bellier 1980) is a quali-quantitative technique working on data through factor analysis. This method enables scholars to obtain a synthetic and overall view of the most significant information by extracting factors contained in the co-occurrence matrix.

Figure 1 and Figure 2 show the categories with V-Test ≥ |1.96| projected on the factorial plane, obtained through the factorial analysis of correspondences. For the sake of clarity, the terms inserted in the graphs were selected based on the criteria of frequency and semantic relevance. This method is helpful for qualitatively interpreting the factorial plan, emphasising the proximity/distance and similarity/dissimilarity among categories. Categories that are set closer co-occur more frequently in the same lexical context, whereas those ones that are farther away co-occur less frequently (Melotti et al. 2023).

Figure 1 shows the categories projected on the factorial plane, taking into consideration the variable country of origin. Through lexical correspondence analysis, 2 factors were extracted, which explain, respectively, 54.28% and 45.72% of the variance. The first factor (ordinate axis) opposed mainly the French (quadrant number 1) stereotypical lexicon associated with the economic and religious domains, with words such as free, credit card, receive, receiver, pension, allowance, Ramadan, sermons and veil, as well as a benevolent area that emerged from the Spanish (quadrant number 4) stereotypical lexicon where words associated with aggressive behaviours (i.e., lack of benevolence) and threats attributed to migrants prevailed, such as assault, offend, rape, illegally, dead and clandestine. Thus, in this first factor, we found in Spanish racial hoaxes lemmas such as applicant, immigrant and refugee, which were close to terms such as invasion, stealing and receiving, as opposed to French terms such as requesting, asylum, pension and credit card, from which emerged stereotypical contents based on a lack of benevolence or, on the other hand, a lack of dominance, from the religious and economic points of view. We further observed how, from this factor, peculiar lemmas of the national context emerged, such as Le Pen, France, French, national, marine and Calais in the French case, and Lanzarote, Canaries and Fuerteventura in Spain. This nationalistic lexicon also co-existed, in the case of the Spanish racial hoaxes, with a religious area evoked by generic terms, as in the case of Muslim, Moroccan, Islam, cross and Islamic.

The second factor (abscissa axis), on the other hand, explains how these stereotypical features of the Spanish (quadrant number 4) and French (quadrant number 1) racial hoaxes are opposed to and, thus, differ from, the Italian (quadrant number 2/3) context, since they are strongly associated with the immigrants’ lack of physical competence in pandemics. In fact, in this factor, we could not find lemmas such as COVID-19, quarantine, and positive situations, as well as national political situations such as Salvini, Italia, and Conte.

In summary, correspondence analysis reveals that racial hoaxes have quite different contents based on their socio-cultural context: the Italian ones appear as clearly distinct (far away in the graph) from the Spanish and French ones which, at their turn, are distinct from each other, having few themes in common.

Figure 2 shows lemmas and categories projected on the factorial plane obtained from the analysis of multiple correspondences, allowing scholars to consider several subparts of the corpus in relation to nominal variables. In this case, the two variables are the content of stereotypes (physical competence, dominance down, dominance up and benevolence) and the pandemic period (pre- and post-COVID-19). Two factors were extracted which explain, respectively, F1 = 36% and F2 = 25%.

From the first factor (ordinate axis) emerged stereotype contents that seemed to oppose, on the one hand, the stereotypical dimension of agency, particularly the lack of physical competence (quadrant number 1) and dominance down (quadrant number 2), with terms such as infected, positivity, infect, infection and epidemic, or aid, income, free, pension, credit card and poor, respectively; and, on the other hand, the dimension more linked to the social relations of lack of benevolence (quadrant number 4) and dominance up (quadrant number 3), associated with words such as security, arrest, stop and escape or Islam, Ramadan, church and Christian, respectively. In the first case, immigrants are stereotypically seen as carriers of disease or portrayed as individuals lacking opportunities, benefiting at the expense of others, as parasites do, and in the second case, immigrants are described as individuals who challenge national safety or the societal religion or culture.

In the second factor (abscissa axis), the contribution of the variable pandemic period emerged, since in the pre-pandemic period (quadrant number 2/3), the economic and cultural threats posed by migrants were prevalent, characterised by lemmas associated with the stereotypical contents of dominance up (quadrant number 3) and down (quadrant number 2); whereas in the period after the pandemic (quadrant number 1/4), the stereotypical contents were mainly associated with the lack of physical competence (quadrant number 1), positivity, infected and risk) that, in this period, was semantically close to lemmas of lack of benevolence (quadrant number 4) and security, such as ordinance, security, illegal immigrant, disembark, trafficking, escaping, escape and testified, as seen in several examples, such as ‘Immigration: over a thousand COVID-positive refugees devastate repatriation centres to escape’. In summary, the analysis of multiple correspondences stresses a clear time-based distinction: during the pre-COVID time, stereotypes in racial hoaxes were mostly addressed to Dominance Down and Dominance Up; with the onset of the pandemic crisis, they turned out to be mostly oriented to Affective Competence and Physical Competence.

4. Discussion

This study explored, through a psycholexicometric analysis, some typical features of disinformation, particularly of racial hoaxes, meant as entire fabrications intentionally and verifiably false albeit having a realistic ground. Recent literature has shown that these types of fake news reinforce ethnic prejudice (Atwell Seate and Mastro 2016) and thus can have a sensitive impact on digital literacy interventions and prejudice prevention among both adolescents and adults.

We hypothesised that different characteristics of racial hoaxes might emerge from (1) the national context in which the racial hoaxes spread, (2) the stereotypical contents of the hoaxes, and (3) whether the racial hoax spread before or after the pandemic emergency period.

For the first hypothesis, common themes and notable differences were found in the hoaxes from France, Italy, and Spain, three countries in Mediterranean Europe that were most strongly interested in ‘first aid immigration’.

The shared themes of all three countries’ hoaxes focus on the possible socioeconomic impact of migratory phenomena also concerning the COVID-19 pandemic, the weakening of the national identity due to the forced integration of new cultures and customs, the many contextualised portrayals of local political figures and their stances on migratory phenomena, strong criticisms of the perceived unfairness of welfare management and, finally, the recurring and persistent criminalisation of migrants. If taking into account the stereotypical content model, this overall sight showed a common reference to the features of dominance up and down, as well as to the lack of benevolence, implying especially the ‘communion’ side.

However, in line with other studies that emphasised peculiarities that emerged from national corpora of hoaxes (Bourgeade et al. 2023), France, Italy and Spain showed different levels of action on the following topics:

(1)

Religious and ethnic references: In France and Spain, there is a stronger emphasis on religious and ethnic dimensions, which include concerns about specific religious practices (such as Muslim prayers in public spaces) and the portrayal of ethnic or religious minorities as potential threats to the local culture and traditions. France’s hoaxes are specifically crafted to mock migrants’ religious habits.

(2)

Specific geographical references: While the high rate of migratory phenomena is discussed in all three hoaxes, Italy focuses on specific locations. Its concern is for ports and other specific areas to become landings for migrants without any control. This reflects the Italian political landscape, where issues related to migrants and asylum seekers have been highly visible and criticised. France also seems bothered by the high rate of uncontrolled migratory flows into its ports.

(3)

Unique political figures: Each country has its unique political figures at the centre of its hoax narratives. For example, French hoaxes depict LePen as a strong leader against migratory phenomena, fighting to protect and restore the national identity (his slogan is ‘The French first’).

(4)

Socioeconomic concerns: In Italy, socioeconomic issues related to gambling are mentioned as irrefutable proofs of migrants’ careless attitude towards Italy’s safety measures against COVID-19 and as demonstrations that they are not truly innocent victims facing hardships. This theme is not as prominent in the hoaxes from France and Spain, which tend instead to criticise the management of resources and welfare between the ‘parasitic’ migrants and the local population.

As for the contents of the stereotypes, the dimension of lack of benevolence is quite widespread in all three nations, as demonstrated by both the coding of hoaxes in stereotypical terms (Bourgeade et al. 2023) and by lemmas such as assault, hurt, rape and dead, which imply danger and a sense of threat to the safety of citizens. The dimension of active dominance is very present, especially in French and Spanish hoaxes, as demonstrated by lemmas such as threat’ veiled and Muslim, highlighting how this stereotyping process is mainly focused on religious lemmas and on the threats to cultural identity. In addition, the prevalence of the stereotypical dimension of passive dominance in the French hoaxes clearly emerged from the view of immigrants as economic ‘parasites’ who disrespectfully take advantage of those offering them help.

Italian hoaxes, while sharing the dimensions emerging in France and Spain, are further characterised by the dimensions of lack of physical competence (as attested to by lemmas such as infection, pandemic, contagion and outbreak). Migrants are definite threats to public health because they are potential carriers of the COVID-19 virus due to their lack of hygiene.

Globally, these analyses enable us to confirm the orientation of fake news to catch realistic anxieties, thus taking readers the bait, in line with overall assumptions and preconstructed cultural anchorages. In some cases, even more contingent events, such as the COVID-19 pandemic, are ridden, thus emphasising the emotional activation and spreading new risks of ethnic prejudice (Atwell Seate and Mastro 2016).

The frequency analysis was further supported by some interesting patterns highlighted by the correspondence analysis, which compared the lexicons by country of origin.

The lexicon of the French hoaxes focuses mainly on the dimension of passive dominance in describing immigrants, leveraging issues ranging from economic aid to asylum requests to create a narrative of the immigrant as someone privileged by the State and who receives ‘special treatment at the expense’ of the French population and of religious matters that threaten to undermine local culture and traditions.

On the other hand, the lexicon of Spanish hoaxes focuses mainly on the dimension of lack of benevolence, depicting immigrants as dangerous individuals, consistently involved in horrific crimes against the vulnerable and innocent.

Finally, Italian hoaxes differ from Spanish and French hoaxes as they shift towards the lack of physical competence, aiming to promote a perception of the immigrant as incapable of maintaining proper hygiene and, therefore, is a likely carrier of the COVID-19 virus.

Nevertheless, even though the contents of the stereotypes seem to overlap with the prevalent issues in each state, there is a common thread in the construction of all the hoaxes: they portray the immigrant as a ‘clandestine’ outlaw and an outsider.

This common element becomes even more prominent when looking at the data on the stereotypical content before the COVID-19 pandemic. All of the hoaxes show the significant presence of the active and passive dominance dimensions. In these cases, the immigrant is described as a threat either to the country’s economy, the citizen’s security, or even the cultural and religious identities of the nation. However, during and after COVID-19, especially in Italy, the hoaxes shifted towards the dimension of lack of physical competence.

Nonetheless, the dimension of lack of benevolence in the hoaxes remained stable before and after the pandemic, making the hoaxes consistent in their narrative of portraying the ‘other’ as dangerous and violent.

The complex configuration of the aforementioned results enabled us to relate the analysed racial hoaxes—as specified through their content stereotypes, their national configuration and their temporal dimension—to the different types of populism that are widespread across the included nations (Boscán et al. 2018). In the French context, Rassemblement Nationale has achieved a solid position in the overall political scenario. Among its core issues is its opposition to migratory phenomena in order to protect and restore the national identity (‘The French First’). Therefore, it is not so surprising that racial hoaxes stress the active dominance stereotype, thus emphasising the realistic and symbolic threats posed by migrants and strengthening the arguments of ‘exclusionary populism’ (Mudde and Kaltwasser 2017), which claims the exclusion of these social groups from social, economic or political profits, which are granted to the natives or the original people (Abdeslam 2021). In addition, the identification between migrants and Muslims supports the heavy cultural difference between ‘us’ and ‘them’, discouraging any effort to integrate them into French culture.

In the Italian scenario, right-wing populism has been referred to as ‘emergency’ populism (Cervi et al. 2020) since one of its core issues concerns landing management and all related duties or tasks, including denounces concerning general confusion and contingency. The hardest part of the Italian racial hoaxes concerns the migrants’ lack of physical competence, which refers to the relationship between migratory phenomena and the pandemic emergency, thus improving readers’ emotional activation related to the several ways in which migrants can worsen the already difficult management of the sanitation emergency. Therefore, especially in the Italian context, migrants are indiscriminately the targets of the extension of the triple conflation (Gao 2022) of the health, racial, political or national domains, which originally involved Chinese people.

As for Spain, the presence of a ‘bipartisan’ populism (Vampa 2020), represented by Podemos (for left-wing populism) and Vox (for right-wing populism), led to the great salience of immigration issues in the political debate, which was based on both different and complementary points. Thus, on the one hand, the emphasis on the cultural threat and the lack of benevolence of migrants fuels right-wing populism, which is traditionally ‘anti-immigration’ (Dennison and Geddes 2019), as it conveys a very hard line on immigration based on deportations and severe restrictions of social policies (Turnbull-Dugarte 2019). On the other hand, emphasis on the socioeconomic threat could be considered as more in line with the trope of left-wing populism orientation (Kriesi et al. 2006; Rodrik 2017), even if keeping in mind that Podemos is set in the frame of ‘inclusionary’ populism (Font et al. 2021), combining anti-elitist claims with a left-wing discourse (De Cleen et al. 2018; Kioupkiolis 2016).

Generally, we found a sensitive relationship between the content of racial hoaxes and the main points of the populist movements spread in these European countries and having specific socio-cultural pinnacles. This is an interesting starting point for further investigations aimed at both confirming these assumptions and extending the included countries.

5. Conclusions

Starting from framing and defining racial hoaxes, this work aimed to empirically investigate this phenomenon as (1) presenting stereotypical content, (2) related to contextual dimensions (nationality and COVID-19), and (3) connected with other social phenomena, such as populism. The multifaceted results enabled us to offer insights set at the intersection of disinformation studies, political communication and hate speech. As for the first domain, this study shed new light on the relations between racial hoaxes, stereotypes and prejudices, reinforcing the socio-cultural roots of fake news and their capability to adhere to the agenda-setting news value. As for the second domain, the connections between racial hoaxes and populist movements can have interesting implications in the political efforts to contrast any kind of disinformation, also keeping in mind the favourite trope of those movements and the more general political scenarios. As for the third dimension, even if hate speech was not a central topic in this study, a specific focus on this issue could further strengthen our results in improving critical information literacy (Brisola and Doyle 2019), even considering that this phenomenon is widespread in Italy (Lingiardi et al. 2020), France (Paci 2022) and Spain (Calderón et al. 2020).

This study had several limitations, mainly concerning the corpus extraction procedure. First, we gathered data mostly from fact-checking websites in Mediterranean European countries, and thus, other hoaxes were not included in our sample, such as those from other European regions. Second, the fake news was translated into Italian to create a monolingual linguistic text, which the software for lexical analysis required. This means that some linguistic nuances could have been neglected. Finally, some of the hoaxes were very brief, thus implying a different quantity of words in the analysis.

This study also had some strengths, mainly its explanation of the possible connections among stereotypes, cultural origins and racial hoaxes, and also between racial hoaxes and populism, in three nations that play an essential role in the European socioeconomic and political scenarios. Thus, the findings of this study can have essential applications in several fields, mainly related to digital literacy and to the deconstruction of the stereotypical schema of immigrants as clandestine, violent and parasitic individuals. In turn, these orientations could promote more functional social integration and policymakers’ efforts, as well as proactive and respectful citizenship, thus reducing the ‘us-them’ distance.