Hate and Perceived Threats on the Resettlement of Afghan Refugees in Portugal

Abstract

The withdrawal of US troops from Afghanistan in August 2021 precipitated a humanitarian crisis, prompting the displacement of Afghan refugees seeking sanctuary, including in Portugal. This study aims to rigorously assess public perceptions of Afghan refugee resettlement in Portugal by analyzing national news and subsequent Facebook reactions, seeking to discern sentiment, approval/disapproval, perceived realistic and symbolic threats, and potential hate speech toward migrant resettlement. Employing a mixed-methods approach with a qualitative emphasis, this cross-sectional study involved a volumetric analysis of 40 Facebook posts from mainstream news outlets, followed by a qualitative content analysis of 1000 comments from 10 selected posts based on popularity and controversy. Findings reveal a predominance of negative sentiments and low levels of approval for migrant resettlement. Emotional complexity suggests controversy and polarization, with negativity surpassing positivity. Concerns regarding economic impact ranked highest, followed by security, cultural, and religious considerations. Instances of hate speech, predominantly political in nature, with lesser occurrences of religious and nationality-based content, were evident. This research contributes to the scientific understanding of public perceptions concerning Afghan refugee integration in Portugal, providing valuable insights into societal attitudes towards humanitarian crises and refugee resettlement efforts.

1. Background

The cessation of military involvement by United States forces in Afghanistan during August 2021, concluding a twenty-year period of engagement, signified a pivotal geopolitical transition, culminating in the swift resurgence of the Taliban regime. This momentous event precipitated a significant alteration in the geopolitical landscape of the area and ushered in a severe humanitarian crisis. The immediate aftermath was characterized by an escalated displacement phenomenon, affecting an estimated 18 million individuals [1]. This scenario highlighted the acute adversity encountered by the Afghan populace, thereby triggering an extensive exodus of refugees, delineating the stark realities spawned by the geopolitical upheaval.

The status and rights of refugees [2] offers a structured lens through which to view this displacement. Refugees, distinct from migrants, are individuals compelled to flee their homelands due to imminent threats of persecution, conflict, and violence. This definition underscores the critical nature of asylum as a cornerstone of human rights protection, embodying the international community’s commitment to safeguarding individuals from severe human rights abuses [3].

The escalation in the number of Afghan refugees, notably rising to millions in the subsequent year as per UNHCR reports, exemplifies the scale and urgency of the global response required, with countries around the globe participating in the humanitarian effort to provide asylum and support [4].

The reception and integration of refugees into host countries are deeply influenced by public perception, an amalgam of cultural values, beliefs, and stereotypes, shaped significantly by media representation and societal discourse. This aspect is critical as public opinion can significantly sway political decisions and shape the societal environment encountered by refugees. Public perceptions of refugees often stem from deeply ingrained cultural values, beliefs, and stereotypes, which, when rooted in personal values, emotions, and experiences, can magnify concerns among citizens of host nations [5]. Nevertheless, the acceptance of refugees is commonly viewed as a collective responsibility about the entire nation or community, rather than a solitary concern centered on individual circumstances [6].

The Tent Foundation’s Global Report (2017) [7] presents a comprehensive examination of the global perception of refugees. When respondents were queried about their countries’ obligations in assisting refugees, the economic ramifications, security implications, and provision of financial aid, the prevailing sentiment among participants leaned towards overt negativity (47%), followed by a mixed perspective (41%) and a distinctly positive outlook (12%) [7].

A factor that also undermines the success in the reception of refugees is the surge in Islamophobia witnessed in Europe and the United States of America in recent years. Islamophobia, as defined by the United Nations, is “a fear, prejudice, and hatred against Muslims or non-Muslim individuals that leads to provocation, hostility, and intolerance through threats, harassment, abuse, incitement, and intimidation, driven by institutional, ideological, political, and religious hostility that transcends structural and cultural racism targeting symbols and markers of being Muslim” [8].

The September 11 attacks and the series of terrorist attacks in the early and mid-2010s in Paris, Brussels, Madrid, and London fueled the growth of Islamophobia in the United States of America and Europe [9], exacerbating anti-Muslim sentiment [10].

Given that anti-immigration rhetoric increasingly aligns with Islamophobia [11], this prejudice and negative attitude towards Islam result in Muslim asylum seekers having lower chances of acceptance compared to Christian asylum seekers [12]. A 2016 Chatham House survey of 10,000 people in 10 European states concluded that 55% agreed with the statement that “all further migration from mostly Muslim countries should be stopped”, regardless of whether they are migrants or refugees [13].

The beginning of the war in Ukraine has drawn considerable attention, potentially overshadowing other pressing domestic issues, such as Islamophobia, albeit to varying degrees. Amidst the chaos, the persecution of members belonging to various Islamic groups persists [14].

The Russian invasion of Ukraine has not only ignited geopolitical tensions but has also shed light on the stereotypical perspectives harbored by many Europeans regarding refugees. There has been a notable contrast in the reception of refugees, with a warm welcome extended to white, Christian (female) Ukrainians, while Muslim refugees, as those fleeing the conflicts in Afghanistan or Gaza, often face violent resistance and rejection. This disparity underscores deep-seated prejudices and challenges the notion of equal treatment and humanitarianism.

The European Islamophobia Report 2022 [14] reports on negative attitudes, discrimination, and hate crimes towards Muslims in twenty-three countries, which does not include Portugal. The European Union Agency for Fundamental Rights (FRA), in the Fundamental Rights Report 2022, provided data from Amnesty International Italy’s online monitoring, revealing that the most common targets of hate in online posts and comments are directed at Muslims (46% and 21%, respectively) [15]. According to the European Commission against Racism and Intolerance (ECRI), governments should also place particular emphasis on prevention. Political actors, opinion leaders, and other public figures must take a strong public stance against anti-Muslim prejudice [16].

In the era of digital globalization, social networks have emerged as pivotal arenas for public discourse, significantly influencing and reflecting societal attitudes toward complex issues like the refugee crisis, as observed in recent conflicts in Afghanistan, Ukraine, or Gaza. In fact, media can play an important role by investigating and uncovering problematic structures. Platforms such as Facebook have evolved into critical spaces for news dissemination, opinion formation, and community engagement. Nearly half of the global population utilizes these platforms, underlining their role in shaping public sentiment and societal norms. The dual-edged nature of social media, capable of fostering both solidarity and divisiveness, underscores the necessity of scrutinizing the online discourse surrounding refugees, particularly in understanding the nuances of hate speech, misinformation, and community support within digital ecosystems.

While numerous studies delve into the attitudes of host country inhabitants towards immigrants, the same depth of research is lacking when it comes to refugees, making information on this subject notably elusive. The most notable progress in the academic exploration of analogous work has been devoted to online racism and hate speech, particularly in Europe [17], Spain [18,19], North America [20], Brazil [21], and China [22]. In Portugal, although not extensively studied by academia, Islamophobia has been appropriately acknowledged. Almeida et al. [23], conducted a critical discourse analysis over Facebook and Twitter (now known as X) for two newspapers and a political page and uncovered trending topics such as the idea of belonging, reverse racism, denial of racism, and freedom of expression.

Araújo (2019) [24], in a more comprehensive approach, provides a performative approach to Islamophobia in Portugal, discussing how Islam and the figure of the Muslim are mobilized in the national imagination, between 2000 and 2019, emphasizing the necessity of systematically and thoroughly investigating such aspects. Six categories of Islamophobic expression were identified: attacks on individuals perceived as Muslims, attacks on property believed to be associated with Muslims, acts of intimidation, incidents that may occur in institutional contexts, public domain comments that defame Muslims and Islam, and state surveillance activities. Moreover, the author discovered ten Islamophobic narratives in Portugal: Islam and Muslims advocate violence, Islam and Muslims are too misogynistic and sexist, Islam is based on religious supremacy and is prone to autocracy, Muslims are intolerant, Muslims are not assimilable, Islam wants to “invade Europe” and control “our way of life”, Islam undermines freedom of expression, Islam promotes homophobia, Muslims use public funds to promote Islamic fundamentalism, and Islam is against modern science and rationality. In fact, another study in the Portuguese context reveals that the debate on whether or not to reduce the high numbers of involuntary migrants witnessed in 2015 has exposed the growth of Islamophobia, and beyond the discourses and practices regarding refugees and migration, a humanitarian crisis prevails [25].

In this local and global context, and considering the recent 2021 crisis, the present study seeks to explore the intricate landscape of the public perception toward Afghan refugees in Portugal, leveraging social media analytics as a lens to examine the perceived threats expressed by Portuguese citizens and the myriad of sentiments, beliefs, and discourses permeating the digital sphere, which can amplify hateful content. This investigation aims to delve into the nuanced interplay between global crises and local responses, examining the contours of empathy, fear, and societal engagement as reflected in online interactions, as well as hypermediated online Islamophobia, where the internet facilitates the quick spread and amplification of negative messages [26].

By adopting a mixed-method research design, with deductive and inductive content analysis, we explore how Portuguese citizens reacted to the news about the reception of Afghan refugees in Portugal, attending to the following specific research goals: identify the types of threats perceived by the audience and reveal the beliefs that translate the public perception of the resettlement of Afghan refugees in national territory, characterize the expressed sentiment, emotion, and approval, and evaluate and translate presence and type of hate speech directed at Afghans.

2. Materials and Methods

A cross-sectional mixed-methods research strategy with an embedded design was adopted, focusing primarily on the qualitative component over the quantitative component, as it enables a more nuanced analysis [27,28].

Considering the timeframe of the 2021 Afghan crisis (US troops withdrew from Afghanistan in August 2021, triggering the humanitarian crisis, which was felt more intensely during the year that followed), we collected all media news posted by the major TV channels in Portugal (SIC Notícias, CNN Portugal, CM TV, and RTP Notícias), between the 1 August 2021 and the 31 August 2022 (1.539 news posted). These media sources were chosen based on their higher number of followers, compared to other sources, and because they specialize in general news. We focused the analysis on the social network Facebook, since it is the most used in Portugal.

The dataset was compiled and extracted using Crowdtangle, an interface provided by the Meta Journalism Project, available for journalists and researchers. For the news selection we used the keywords “refugees”, “Afghanistan”, and “Afghan”, in singular, plural, male, and female, to cover the linguistic variation in the Portuguese language (“refugiado”, “refugiada”, “refugiados”, “Afeganistão”, “afegão”, “afegãos”, and “afegã”), which led to a selection of 40 news posts on social media. We intended to capture all news related to the conflict and especially referring to refugees, which, after an initial data scanning, led us to conclude that nationality (Afghan) and status (refugee) were the keywords mostly used in the subset of news we aimed at analyzing.

Since our core goal was to investigate the public reaction to the news, namely by identifying the public perception of the reception of Afghan refugees, and detect approval/disapproval, sentiment, and expressions of hate, we selected the most commented news with the highest emotional entropy levels. In doing so, we expected to capture the most critical news for the audience (most commented) and the most controversial (with the highest emotional entropy, according to [29,30]).

Controversial heated discussions are a prolific field for hate speech on social media, and according to Dori-Hacohen, et al. [31]; one of the main current challenges of hate speech recognition is the automatic detection of irony [32] because people verbalize an idea while implying the opposite meaning; thus, textual features alone fail in recognizing the implicit meanings of the discourse. Irony serves the additional social and emotional functions of projecting emotions like humor or anger, and ironic comments may provoke stronger emotional responses than literal comments [33]. In their research about irony, the authors introduce paralinguistic features (emoticons) to improve the detection of praise and criticism in written messages. Such methods had already been employed by other studies such as Carvalho, et al. [34] and Derks, et al. [35]. Several other authors have incorporated discrete or categorical emotion analysis (such as Facebook reactions) in the detection of hate speech, such as Markov [36], Martins [37], Alorainy et al. [38], Rodriguez [39], Plaza-del-Arco [40], Rana et al. [41], Oliveira [30], and Hessel and Lee [29], to detect sarcasm and hate.

Following Hessel and Lee’s [29] procedure, we determined the degree of controversy of the news by using Basile, et al. [42] model to compute the entropy (a quantitative measure of emotional disorder) among Facebook’s reaction set (Love, Wow, Haha, Sad, Angry, and Care) per post, as a function for determining controversy. We computed the average entropy for the set of 40 news posts related to the phenomena (1689) and selected the subset of the most commented news with above-average entropy values (>1689), which led to a sample of 10 news posts, for which we collected the first 100 comments to be submitted to content analysis, for a total of 1000 comments. The subset of the most commented and most controversial news posts is detailed in

Table 1. News subset sample.

Code	News Date	Channel	Comments	Entropy	News Title
P01	26 August 2021	SIC Notícias	1112	1.955	More than 800 Portuguese families available to take in Afghan refugees. 350 people are available to provide accommodation.
P02	15 August 2021	SIC Notícias	1023	1.890	Portugal joins EU operation and is available to receive Afghans. Information provided by the Minister of Defense, João Gomes Cravinho.
P03	28 August 2021	SIC Notícias	881	1.905	First group of Afghan refugees in two reception centers in Portugal. Three dozen more refugees are expected today.
P04	1 September 2021	SIC Notícias 1	819	1.768	Where are women’s rights when an Afghan is told he can only bring one of his wives? “It’s murder to leave them behind”.
P05	18 August 2021	SIC Notícias	775	1.920	Portugal hopes to receive Afghan refugees “as soon as possible”. Information provided by the Minister of Internal Affairs.
P06	26 August 2021	SIC Notícias 1	562	1.893	840 Portuguese families are willing to take in Afghans.
P07	5 October 2021	SIC Notícias	331	2.196	Afghanistan: More than 100 music students and teachers leave Kabul and expected to come Portugal.
P08	10 September 2021	SIC Notícias	283	1.898	A Portuguese intelligence company wants to rescue a thousand people from Afghanistan. Among them the family of the Afghan refugee living in Porto.
P09	10 September 2021	CM TV	263	1.719	SEF says that all refugees hosted in Portugal have received 10,000 euros. Since 2014, “21 million euros have been received and transferred” for the reception of refugees, the service explains.
P10	16 November 2021	SIC Notícias	189	1.821	210 more refugees from Afghanistan have arrived in Portugal. They join the 266 who have been in Portugal since the beginning of the process of withdrawal from Afghanistan.

¹ Repost of a society article published by the Portuguese newspaper Expresso.

. In doing so, we used entropy as an entry point to the most extensive discussions with the most potential to contain hate references.

We adopted both deductive and inductive content analysis. The deductive content analysis of the comments on the news was performed according to a theoretical and operational framework that was developed and supported by the existing literature, aimed at revealing the sentiment and degree of approval, the perceived threats by the local population, and the types of expressed hate speech, as depicted in

Table 2. Content analysis categorization model.

Dimension	Category	Subcategory
Sentiment	Positive
	Negative
	Neutral
Approval	Approval
	Disapproval
	Not expressed
Perceived Threat [43,44]	Realistic threat	Safety
		Economy
	Symbolic threat	Culture
		Religion
Hate speech [45,46]	Cultural
	Politic
	Race
	Gender
	Sexual orientation
	Religious
	Ethnic
	Descent
	Color
	Nationality

.

Sentiment analysis is a field of Natural Language Processing (NLP) that seeks to identify and categorize emotional states in textual samples. This method classifies sentiments as “Positive”, “Negative” or “Neutral” and is widely used in the investigation of public reactions in response to social crises [47]. In this research, this analysis is carried out using Large Language Models and Generative Artificial Intelligence, by employing ChatGPT-4 for the sentiment analysis of the comments, which were later manually reviewed by the researchers.

The classification of approval is performed manually and is intended to assess whether the interaction conveys an idea of approval or disapproval of the reception of Afghan refugees in Portugal. The classification “Approval” is assigned whenever indications favorable to the reception are expressed, and the classification “Disapproval” is assigned when the opposite is evident. The classification of “Not expressed” is used for content that does not express a particular inclination towards approval or disapproval.

In the context of perceived threats, the categorization of content is based on the Theory of Prejudice [43], which divides the perceived threats to refugees into realistic and symbolic, subsequently characterizing realistic threats as relating to safety and the economy and symbolic threats concerning culture or religion [44]. Realistic threats pose challenges to the common welfare and may manifest at the level of economic and political stability, as well as the security and well-being of the populace [43]. Conversely, symbolic threats are apparent at the level of moral values and identity within a host community. Cultural and religious differences are categorized as symbolic, as they threaten the ideals and customs of host countries [44].

The Theory of Prejudice offers a valuable framework for dissecting these perceptions and is particularly relevant in understanding the complex dynamics at play in host countries, where the resettlement of refugees often intersects with national identity, security, and economic welfare debates [5]. It is essential to assess the extent to which these threats are perceived by the Portuguese given their relevance in shaping possible beliefs and concerns regarding this social group.

Concerning the hate speech dimension included in our research, the exact definition of the term remains contentious, as it is a subjective and highly interpretable concept [36,48,49]. Nockleby [50] describes “hate speech” as communication that disparages an individual or group based on attributes such as race, color, ethnicity, gender, sexual orientation, nationality, religion, or other characteristics. More systematically, the United Nations (UN) Strategy and Plan of Action on Hate Speech defines hate speech with three main components: (1) any form of communication in speech, writing, or behavior that (2) uses pejorative or discriminatory language to attack a person or group based on (3) their religion, ethnicity, nationality, race, color, descent, gender, or other identity factors [40]. According to the Strategy, hate speech is communication that is prejudicial, bigoted, intolerant, discriminatory, or contemptuous or demeaning (“pejorative”) towards an individual or group based on their identity. However, the UN specifically excludes the State, its offices and symbols, public affairs, religious leaders, and doctrines and tenets of faiths from being considered targets of hate speech, stating that only individuals or groups can be targets. From the broader sense of the definition, it becomes evident that hate speech and offensive language often coexist.

Based on Guterres’ definition of hate speech [46], the categories of race, gender, sexual orientation, religion, ethnicity, descent, color, and nationality, as well as the cultural and political dimensions [45], were included, in an attempt for a comprehensive coverage of its domains, according to the following:

–

Cultural hate targets cultural groups and involves derogatory statements, stereotypes, or incitements against a group based on their cultural practices, traditions, or lifestyles, for example, comments about traditional clothing, festivals, or cultural practices that promote contempt or hatred;

–

Political hate is directed at individuals or groups based on their political beliefs or affiliations. This includes promoting hostility or violence against political opponents or their supporters, for example, violence against members of a political party or demeaning political figures with inflammatory and dehumanizing language;

–

Racial hate targets individuals or groups based on their race or perceived racial characteristics. This can include racial slurs, stereotypes, and calls for racial segregation or violence, which can include derogatory racial epithets or promoting the superiority of one race over another;

–

Gender hate targets individuals based on their gender, often manifesting as sexism or misogyny, which translates as derogatory remarks, stereotypes, and incitements to discrimination or violence against a specific gender, such as statements that demean women or men based on gender roles or abilities;

–

Sexual orientation hate is directed at individuals or groups based on their sexual orientation. This includes homophobic slurs, stereotypes, and incitements to discrimination or violence against LGBTQ+ individuals, such as derogatory terms for LGBTQ+ individuals or promoting the idea that non-heterosexual orientations are abnormal or harmful;

–

Religious hate targets individuals or groups based on their religious beliefs or affiliations. This includes derogatory statements, incitements to violence, or promoting stereotypes against a particular religion, such as statements that vilify a religion or its followers, or incite violence against people based on their religious practices;

–

Ethnic hate is directed at individuals or groups based on their ethnicity or ethnic origin. This includes derogatory remarks, stereotypes, and calls for discrimination or violence against members of an ethnic group, which includes ethnic slurs or promoting negative stereotypes about an ethnic group’s behavior or characteristics. It can also include blaming an ethnic group for historical events, fostering resentment and hatred, mocking cultural practices, and ridicule traditions;

–

Descent hate targets individuals based on their lineage, ancestry, or family background. This can involve derogatory remarks, stereotypes, and incitements to discrimination based on descent, such as statements that demean individuals because of their familial background or heritage;

–

Color hate targets individuals based on their skin color. This includes derogatory terms, stereotypes, and calls for discrimination or violence based on skin color, such as promoting the idea that people of certain skin colors are inferior;

–

Nationality hate is directed at individuals or groups based on their national origin or citizenship. This includes derogatory remarks, stereotypes, and incitements to discrimination or violence against people from specific countries, which includes vilifying people based on their nationality, such as calling all immigrants from a particular country criminals.

For the inductive content analysis, we identified the beliefs associated with the comments expressing realistic and symbolic threats as they emerged in the comments, so as to translate the specific contours of the major threats verbalized by the audience. We rely both on manifest coding, capturing the explicit content within the text, and on latent coding, interpreting the deeper implicit message [51]. The latter is particularly relevant for the cases in which irony is present in the text, and the literal meaning is contrary to the intended implicit meaning, which, as stated, is critical in the detection of hate speech.

Annotating hateful content remains subjective and culturally dependent, frequently resulting in low inter-annotator agreement and a paucity of high-quality training data for creating supervised hate speech detection algorithms [52]. As a result, the authors engaged in a collaborative coding approach [51], working together in synchronous meetings to develop, refine, and apply the coding scheme while analyzing the textual evidence collected. This allowed us to build consensus and consistency in intra- and inter-coding decisions, which was particularly helpful in latent coding.

3. Results

3.1. Volumetric Analysis

The overarching volumetric analysis of the selection of 40 news posts yields crucial insights into the interaction dynamics observed throughout the analysis period. An interaction is defined as any engagement form—comment, reaction, or share—associated with the publications.

The distribution of the selected publications across the delineated timeframe revealed a pronounced concentration in August and September 2021 (Figure 1). This temporal clustering is notably synchronous with the geopolitical shifts and ensuing humanitarian crisis in Afghanistan, underscoring the immediacy and relevance of these publications. The observed variation in media attention to this issue can also be attributed to the salience of the issue in the national context, which highlights the role of media in shaping public priorities. As noted by McCombs and Shaw [53], the media plays a pivotal role in shaping the public agenda by determining which issues are given prominence, thereby influencing the public perception of the issue’s importance.

Across all publications, a cumulative total of 25,945 interactions was recorded, averaging 648.6 interactions per publication.

Following the publications’ distribution over time, the highest concentration of interactions was also registered during August and September of 2021 (Figure 2).

The interactions’ peak was observed in the publication “Portugal welcomes athletes from the Afghan women’s football team”, with 5092 interactions on 23 September 2021. The lowest number of interactions was recorded for the publication “Portugal received seven refugees from Turkey this Thursday” (15) on 2 September 2021. Both publications were excluded from the content analysis sample due to their low emotional controversy level.

The partitioning of observed interactions into categories—reactions, comments, and shares—revealed a distribution of 15,858 reactions (61%), 9033 comments (35%), and 1054 shares (4%), painting a comprehensive picture of the engagement landscape.

The peak of interactions through comments was recorded for the publication “More than 800 Portuguese families willing to welcome Afghan refugees” (p01) on 26 August 2021 (1112 interactions). In stark contrast, the publications “Portugal received seven refugees from Turkey this Thursday” and “Reborn from Silence—What happens to the Afghan musicians that Portugal welcomed? See the full Special Report here”. recorded only one comment each on 2 September 2021 and 4 April 2022, respectively, both being excluded from the content analysis sample due to their low comment count. The average number of comments per publication was 225.9.

The publication with the highest count of reactions was “Portugal welcomes athletes from the Afghan women’s football team”, on 23 September 2021 (4424 reactions), which also corresponds to the peak of interactions observed (Figure 3). Conversely, 23 August 2021 recorded the lowest number of reactions to publications (“Tiago Barbosa Ribeiro writes to Rui Moreira asking the municipality to welcome Afghan refugees. Porto City Council is unaware of the letter”, with 13 reactions, excluded due to low comment count and low emotional controversy level). For the sample under analysis, the average number of reactions per publication was 396.5.

Reactions were divided into Like, Love, Wow, Haha, Sad, Angry, and Care. Figure 4 illustrates the distribution of observed reactions by type.

The most observed reactions in the analyzed publications were “Love” (1536, 27%), “Angry” (1533, 27%), “Haha” (1455, 25%), and “Sad” with 614 occurrences (11%). The least used reactions were “Care” (301, 5%) and “Wow”, with only 295 records (5%).

As emotion categorization was framed within the theoretical construct posited by [54], reactions posed as indicators of universal emotions.

Analyzing the emotions expressed in the form of reactions and considering the “Like” indicator as devoid of emotion due to its standard use as a sign of publication acknowledgment [55], thus absent from the analysis, a higher frequency of the “Love” emotion is observed, closely followed by “Angry”, “Haha”, and “Sad”, with “Surprise” and “Care” being the least observed emotions. When comparing opposite emotions and considering Haha a factor of sarcasm or mockery and Surprise as a neutral emotion, negative emotions are predominant (Angry, Haha, and Sad at 22%) over positive (Love and Care at 12%) and neutral (Surprise 2%) emotions.

The comparison between the types of emotions observed reinforces the idea of the prevalence of negative emotion (63%) over positive (34%) and neutral (5%). Reactions characteristic of negative emotions occurred 3602 times, while reactions conveying positive emotions were detected 1837 times and neutral ones only 295 times (Figure 5).

Figure 6 demonstrates that the maximum occurrence of reactions linked to positive emotions (589) overlapped with the peak of reactions with a negative emotional connotation (416). On the other hand, negative emotional reactions presented an initial greater consistency, extending longer in the temporal space and reaching a state of dilution later.

As we conclude the volumetric analysis, it became evident that the issue exhibits a decline in interest over time, transitioning from its initial critical and controversial status to a diminished level of interest. This may be explained by the concept of the issue-attention cycle, which refers to the ups and downs of attention an issue receives either from the public or mass media [56], taking into consideration that, for most issues, media attention and audiences’ attention do not hold for a long period.

3.2. Content Analysis

This methodological approach seamlessly blended deductive and inductive techniques, allowing for the incorporation of emergent themes into the analytical framework. Through this rigorous methodology, the content analysis unveils the multifaceted tapestry of public discourse surrounding the Afghan refugee crisis, enriching our understanding beyond mere quantitative metrics.

Initially, sentiment analysis shed light on the emotional undercurrents within the comments. Subsequently, an evaluation of the approval expressed in the comments was conducted, with further analysis aiming at identifying perceived threats. Lastly, the occurrence and nature of hate speech were scrutinized.

3.2.1. Sentiment Analysis

In this subsection, the distribution of sentiments—positive, negative, or neutral—present in the analyzed comments is explored (

Table 3. Sentiment distribution.

Sentiment	n	%
Positive	109	11
Negative	588	59
Neutral	309	30

).

Overall, negative sentiment was dominant, present in 588 comments (59%), followed by neutral sentiment with 303 occurrences (30%). Positive sentiment was only present in 109 comments (11%).

Table 4. Perceived sentiment verbatim examples.

Perceived Sentiment	Verbatim (Examples)
Positive	“What a beautiful gesture Portuguese people. ”(p01)
Negative	“Sad, retarded, worry about those here first” (p08)
Neutral	“Jose Carlos Vilarinho, are you a doctor?” (p03)

contains examples of the different perceived sentiments.

3.2.2. Degree of Approval

Within this subsection, an examination is conducted on the classification of comments concerning their demonstration of approval, disapproval, or lack of expressed sentiment, providing a clear view of the degree of agreement or disagreement of the commentators regarding the integration of refugees.

Table 5. Approval degree.

Approval Degree	n	%
Approval	93	9
Disapproval	231	23
Not expressed	676	68

summarizes the degree of approval observed in the sample.

Of the analyzed comments, 93 (9%) showed approval regarding migrant resettlement in Portugal. Disapproval was present in 231 comments (23%), and the remaining 676 (68%) did not expressly show signs of approval or disapproval.

Table 6. Approval degree verbatim examples.

Approval Degree	Verbatim (Examples)
Approval	“I agree, I’m Chega 1, these are real refugees and deserve to be welcomed” (p01)
Disapproval	“We don’t want any more, that’s enough” (p02)
Not expressed	“Time will tell.” (p10)

¹ Chega is a far-right Portuguese political party.

contains evidence (verbatim) of statements referring to approval, disapproval or no approval expressed.

Upon evaluating the extent of approval documented within the sample, a notable prevalence of disapproval (23%) in contrast to approval (9%) is evident, suggesting that, in samples demonstrating a tendency towards one or the other, there were indeed more expressions against migrant resettlement than in favor. The categorization of comments as “Not Expressed” (676) did not necessarily imply neutrality but rather reflected the nature of the discussions and opinions shared in response to the original publications, which were not always aligned directly with the central theme of the investigation.

3.2.3. Perceived Threats

According to the categorization model, perceived threats were organized into realistic (safety and economy) and symbolic (culture and religion). Based on the analyzed data, synthesis was also developed concerning beliefs grounded on the expressed opinions. We elucidate the fears and reservations that may influence positions and discourses related to the theme of hosting Afghan refugees in Portugal.

First, the perception of realistic threats, related to safety and economic aspects, was analyzed.

Table 7. Realistic threats.

Realistic Threats	n	%
Safety	52	22%
Economy	186	78%

contains the total of realistic threats observed.

From the total sample of 1000 comments, 238 (23.8%) contained indications of realistic threats, from the commentators’ perspective. These were divided into threats related to safety (22%) and economic concerns (78%), which were supported by claims such as the ones depicted in

Table 8. Safety and economy realistic threats.

Perceived Threat	Verbatim (Examples)
Safety	“In a few years we will be beheaded by them” (p02)
	“Carmo Abreu in less than 20 years we will have an Islamized Europe. Then we will see who finds the needed peace”. (p03)
	“One of these days we’re going to pay dearly for letting Islamists in… Look what happened to Lebanon… It’s destroyed… The same is going to happen to Portugal…” (p07)
Economy	“What matters is that we are always ready to welcome others, even if our own are hungry……” (p2)
	“Portuguese should express their dissatisfaction, as pensions are a misery and those who have never done anything for this country will get something” (p3)
	“Andre Blackie Almeida is absolutely right in what he says, Portugal has all races, and the state pays them housing, food, and employs them, while sends the Portuguese to immigrate” (p6)

.

From the total set of claims regarding safety and economic concerns, we were able to identify the following set of beliefs regarding the resettlement of Afghan refugees (

Table 9. Beliefs regarding safety and economy.

Domain	Belief	n	%
Safety	Refugees are a terrorist threat	7	13
	There will be an increase in crimes due to refugees	5	10
	Europe and Portugal are on the path to Islamization	3	6
	Refugees do not integrate and have conflicting values with Western ones	3	6
	Aid to refugees is the cause of future problems	2	4
Economy	We must help the Portuguese/ours first	6	3
	Refugees will live on social benefits/subsidies	3	2
	Refugees do not work and live at the expense of those who do	2	1
	The government does not help the most disadvantaged Portuguese	2	1
	Portugal presents itself as a rich country, but it is poor	2	1
	Richer countries should provide foreign aid	2	1
	It is inconsistent to help refugees and neglect nationals	2	1

):

In the realm of realistic threats linked to safety, there is an observed tendency in the exemplary comments to refer to beliefs related to terrorism, death threats, crime, lack of peace, and destruction.

In terms of economic threats, the main concerns expressed are related to the lack of resources for Portuguese people and the assistance provided to Afghan refugees at the expense of national citizens, also highlighting references to the already weakened economic condition of the country.

As for perceived symbolic threats,

Table 10. Symbolic threats.

Symbolic Threats	n	%
Culture	52	72%
Religion	20	28%

translates the total number of occurrences.

A total of 72 symbolic threats were counted (7.2% of the total sample), divided between 52 cultural threats (72%) and 20 religious threats (28%).

Table 11. Cultural and religious symbolic threats.

Perceived Threat	Verbatim (Examples)
Culture	“Solidarity yes, but with caution and measure. May our laws, our beliefs and our customs be respected by those we welcome. May they integrate into our society according to our customs, not requiring adaptations to the customs of the places they come from”. (p03)
	“No one has the right to come and impose rules. If you don’t agree, don’t come. Look for refuge where it is permitted. Being supportive is one thing, being tolerant to the point of accepting what we repudiate is another. Let no one come and impose anything on us! THIS NEVER”, (p04)
	“In other words… we have to integrate ourselves into the culture of the refugees we welcome and not the other way around!?!? Stop the bullshit! We have to welcome human beings who need help, but we don’t have to change our society for that reason. By the way… if this is a bigger problem why aren’t they being welcomed by countries with similar culture and traditions?” (p4)
Religion	“When Sharia law is imposed, we will see. Islam is the persecution of freedom of expression.” (p03)
	“…you might as well prepare to receive them all and in 30 years’ time install an Islamic State” (p02)
	“This isn’t the first time I’ve said, with this situation where Europe has to receive refugees and they don’t accept our religion, in about 30 years the Arabs will rule Europe, it would be better to stay in countries that have the same religion and be helped by the richest countries” (p04)

shows some examples of perceived cultural and religious threats.

From the total set of claims regarding cultural and religious concerns, we were able to identify the following set of beliefs regarding the resettlement of Afghan refugees (

Table 12. Beliefs regarding culture.

Domain	Belief	n	%
Cultural	Refugees must respect local laws and customs	5	10
	Refugees have difficulty adapting to local culture	3	6
	There are significant cultural differences	2	4
	Europe is being Islamized	2	4
	Welcoming women is more acceptable than welcoming men	1	2
	Some refugees are more adaptable than others	1	2
	European laws are superior and must be followed	1	2
Religious	Islam is a threat to European culture and identity	7	35
	Islam is incompatible with European values	4	20
	Europe will be Islamized in the future	3	15
	Muslim immigrants are religious fanatics	2	10
	Sharia law will be imposed in Europe	2	10
	Islam views women as inferior	2	10
	Islam is violent and promotes terrorism	2	10

):

Symbolic threats are present at the level of moral values and the identity of a host community, as mentioned by [44]. In this symbolic aspect, the collected comments revealed concerns related to local customs and the difficulties of adaptation by Afghan refugees. Symbolic religious threats related to Islam were also identified, characterizing it as a violent religion that restricts women’s rights. Concerns about the threat of Islam’s imposition within Western society were also present.

The analysis of perceived threats in comments on publications related to the hosting of Afghan refugees was crucial to understanding how this social group and their situation are viewed by the Portuguese and what beliefs are developed about their hosting and about the Afghan refugees themselves.

3.2.4. Hate Speech Analysis

This analysis focuses specifically on hate speech present in the comments, based on the categorization model, starting from the definition by [46] complemented by [45].

Table 13. Hate speech typologies.

Hate Speech	n	%
Cultural	0	0
Political	76	57
Race	0	0
Gender	1	1
Sexual Orientation	0	0
Religion	22	16
Ethnicity	0	0
Descent	0	0
Color	2	1
Nationality	31	23

encompasses a comprehensive compilation of flagged comments exhibiting instances of hate speech. A total of 134 comments featuring such discourse were detected, constituting 13.4% of the entire sample. The political dimension accounted for the highest prevalence, comprising 76 comments (57% of occurrences), followed by nationality with 31 instances (23% presence) and religion with 22 occurrences (16% representation). Additionally, there were isolated instances of hate speech associated with color (two instances, 1%) and gender (one instance, 1%).

Table 14. Hate speech verbatim examples.

Hate Speech Type	Verbatim (Examples)
Political	“Look at this idiot, you incompetent liars, you can’t satisfy the hunger of the Portuguese and there aren’t so few of them, shame on your ugly face” (p02)
Nationality	“Visconde Alvalade I agree, these people have no conscience, it’s like the Chinese, let’s expel them all, this is too much” (p04)
Religion	“We are witnessing the Islamization of Europe and of course my Portugal… Only those who don’t know the suburbs of Paris, London, Amsterdam… Only those who don’t see news about the problems of harassment and rape in Germany and Sweden… disturbances in Greece and Italy, is that they are not worried about what is happening… Soon we run the risk of living under their laws if nothing is done… Yes, help but with planning and very clear laws about their obligations… The Western world is doomed” (p03)
Color	“Lucy Silva, I knew it, second class nigger. Because if you lived in Portugal, you wouldn’t have the same opinion” (p03)
Gender	“Graciete Guedes, then your profile is fake. You are from Neilly-Plaicense in Franca. Nobody forces them to emigrate to work in Portugal, they just need to want to work. Help those who are socially dependent on social security. About my neurons, don’t be nervous, be careful with your menstruation”. (p02)

contains illustrative examples of hate speech encountered within the interactions pertaining to the analyzed publications.

While instances related to gender and race were indeed present, their significance for the present study on the perception of Afghan refugees is deemed minimal. These instances primarily emerged within the context of responses to secondary comments within the publications, entailing discussions among the commentators themselves rather than directly addressing the research theme.

3.2.5. Cross-Sectional Analysis

In this section, a cross-sectional analysis of the 10 publications from Sample 2 is carried out.

Table 15. Observed occurrences by news.

	P1	P2	P3	P4	P5	P6	P7	P8	P9	P10	Total
Sentiment
Positive	25	3	27	12	2	14	7	4	2	13	109
Negative	34	59	49	67	70	62	47	73	59	68	588
Neutral	41	38	24	21	28	24	46	23	39	19	303
Approval
Approval	16	4	25	10	6	12	5	8	-	7	93
Disapproval	21	24	26	22	18	28	22	27	19	24	231
Not expressed	63	72	49	68	76	60	73	65	81	69	676
Threat
Realistic Security	3	10	8	2	3	7	3	6	2	8	52
Realistic Economy	19	10	15	17	21	22	10	20	34	18	186
Symbolic Culture	-	2	6	32	-	4	2	5	-	1	52
Symbolic Religion	-	10	5	3	-	-	1	1	-		20
Hate
Cultural	-	-	-	-	-	-	-	-	-	-	-
Political	2	8	4	2	13	6	9	2	10	20	76
Race	-	-	-	-	-	-	-	-	-	-	-
Gender	-	1	-	-	-	-	-	-	-	-	1
Sexual orientation	-	-	-	-	-	-	-	-	-	-	-
Religion	1	11	4	2	1	1	2	-	-	-	22
Ethnicity	-	-	-	-	-	-	-	-	-	-	-
Descent	-	-	-	-	-	-	-	-	-	-	-
Color	-	-	2	-	-	-	-	-	-	-	2
Nationality	4	1	1	3	2	4	-	4	2	10	31

encompasses the incidence of observed occurrences, by category and by publication.

Upon analyzing Table 15, a consistent trend emerges across all 10 publications, wherein negative sentiment substantially prevails over positive sentiment. Likewise, the prevalence of disapproval outweighs approval in each publication. The occurrence of comments where approval or disapproval are “Not Expressed” holds significance in this analysis, with percentages consistently equal to or surpassing 60% in all publications except for publication p04 (49%).

In terms of perceived threats, the perception of realistic economic threats remains uniformly evident across the publications, occurring at least 10 times in each sample, thus emerging as the most conspicuous threat category. Realistic security threats, though present in all publications, exhibit relatively fewer occurrences. Remarkably, p04 demonstrates a notable variation in recorded threats about the symbolic cultural threat, with 32 occurrences, a figure significantly higher than the remaining publications where such occurrences do not exceed 6.

Regarding hate speech, the political type pervades all publications, while both religion and nationality types are distributed across most samples. Hate speech types related to gender and color are observed in only one publication each.

4. Discussion

During the delimited period for sample collection (posts), there was a particular incidence of viable posts in August and September 2021. This incidence is temporally close to the humanitarian crisis triggered by the withdrawal of US troops from Afghanistan, hence there is a direct link to the event that facilitated the rise to power of the Taliban regime and resulted in the refugee exodus. Being a global event of great magnitude, it is natural that there was increased coverage by the Portuguese media, which produced a larger volume of news relevant to the current investigation, as verified in the sample selection. It is also noteworthy that, although the analysis covered a relevant period of 2022, the majority of the posts of special relevance for analysis were recorded in 2021. It is believed that the onset of the war in Ukraine on 24 February 2022, may have contributed to the reduction in the flow of posts regarding the hosting of Afghan refugees in Portugal.

The analyzed posts registered a total of 25,954 interactions, with an average of 648.6 per post. The post that gathered the most interactions occurred on 23 September 2021. The overall analysis of the interactions (reactions, comments, and shares) revealed a predominance of reactions (61%) relative to comments (35%) and shares (4%).

A more detailed analysis of the volumetric distribution of different types of interaction over time indicates that comments reached their peak occurrence during August, while reactions and shares only reached their maximum occurrences in September. These results suggest a greater initial engagement with a new topic, resulting in higher emotional investment. As time progressed, the number of comments decreased, indicating a waning interest, leading to less direct and deep involvement and an effort economy that enhanced the increase in the number of interactions through reactions and shares.

Regarding the distribution of interactions, it is important to mention that of the five posts with the most comments, three belonged to the political domain (p2 “Information advanced by the Minister of Defense, João Gomes Cravinho.”, p4 “The Minister of Defense made it clear that Afghans coming to Portugal only have the right to bring one wife, in case there is more than one.”, and p5 “Information advanced by the Minister of Internal Administration.”). This indicator aligns with the trend observed in Facebook posts of Swedish newspapers [57], where news centered on politics/government was linked to a high volume of comments.

In the domain of emotions, and establishing a correspondence between these and Facebook reactions, it is considered that the high frequency of the “Love” emotion may indicate strong agreement or appreciation of the content, while “Anger” suggests that some of the posts may have been controversial and provoked strong emotional reactions. The prevalence of “Haha” as a possible form of sarcasm or mockery suggests that not all commentators may have taken the topic seriously. Moreover, the fact that “Care” was less observed may indicate that few felt deep or genuine empathy. The occurrence of the neutral emotion “Wow” could point to the initial shock, which has prolonged over time. The analysis revealed that negative emotions (Anger, Haha, and Sadness) were more predominant than positive ones (Love and Care), suggesting that the topic may have been polarizing or provoked strong and complex feelings. Nonetheless, the maximum occurrence of positive emotions was higher than the peak of negative emotions.

After a detailed volumetric analysis of the interactions and emotions, it became imperative to delve into the underlying content of these interactions to better understand the emotions and opinions expressed.

The content analysis that followed was applied to Sample 2, a portion (25%) of the publications were analyzed volumetrically, and the aim was to decipher the reason behind the observed patterns, providing a deeper insight into the perspectives and feelings of the Portuguese in their reactions to the publications about the hosting of Afghan refugees. In exploring the comments, the main drivers of the reactions, whether supportive, critical, or indifferent, were sought to be highlighted. The content analysis included sentiment analysis, the assessment of the degree of approval, and the identification of perceived threats (realistic or symbolic), and it also sought to assess the presence and typology of hate speech. This phase was crucial for understanding not just the magnitude but also the depth and complexity of the matter under investigation.

The content analysis sample consisted of the first 100 comments from each of the 10 selected publications. Subsequently, the analysis was conducted using a pre-defined characterization model. Considering the qualitative component of the research strategy and given that interactions not fitting the initial categorization model were identified, the Approval Degree category was added later, and to the existing typologies of hate speech in the model (race, color, gender, sexual orientation, nationality, ethnicity, and religion), political and cultural typologies were added. These changes had a significant impact on the analysis results, as will be mentioned later.

The sentiment analysis revealed that the dominant sentiment in the sample was negative, present in 59% of the comments (558), followed by the neutral sentiment in 303 (30%). In turn, positive sentiment was only manifested in 11% of the comments (109). One of the factors preventing the association of the overall analyzed negative sentiment with the perception of refugees manifested is the fact that the sample includes responses to the original comments of the publication. It was observed that engagement in discussions and the exchange of opinions among commentators resulted in a considerable number of comments with associated neutral or negative sentiments, less observable in comments of positive sentimental nature.

The approval rating assessment showed that only 9% of the comments expressed approval regarding the hosting of Afghan refugees. The observed disapproval was 23%, revealing that there were more expressions against the hosting of Afghan refugees than in favor. The majority of the comments (676) did not express a defined position on the topic under study. Similar to the sentiment analysis, it is believed that this fact was due to the dispersion of comments, often motivated by discussions and opinion sharing in responses to the original publication’s comments, and was not related to the central theme of the investigation.

Perceived threats were divided into realistic and symbolic. Realistic threats were composed of security and economic typologies. Fifty-two comments containing realistic security threats were identified. The analysis of the comments concluded that these threats were perceived due to the association of refugees with beliefs related to terrorism, death threats, crime, lack of peace, and destruction.

Regarding realistic economic threats, 186 occurrences were observed, making this typology the most expressed. Considering the current economic state of Portugal, it can be stated that this indicator aligns with the ideology defended by [58], when they mention that countries considering their own economy as Bad/Normal have a greater tendency to view the hosting of refugees as a realistic threat to the economy than countries that consider having a Good/Excellent economy. Also, within this type of threat, a tendency was observed in comments that associated the hosting of Afghan refugees with a lack of resources in Portugal, making evident a fear of allocating necessary resources for national citizens to refugees.

After this analysis, the perception of symbolic threats in the comments of the sample was assessed. These are divided in relation to culture and religion. Regarding culture, 52 occurrences were identified, with the collected comments highlighting concerns related to local customs and the difficulties of adaptation by the refugees. Symbolic threats related to religion were present in 20 comments, where beliefs were emphasized that Islam is a violent religion that restricts women’s rights. Concern was also expressed about its imposition in Western societies.

Regarding the presence and typologies of hate speech, the analysis revealed that this type of discourse was present in 134 comments, of which 76 were political in nature, 31 addressed nationality, and 22 referred to religion. Residual cases of hate speech related to color and gender were also detected, although their relevance for the current investigation is limited, as they were recorded in scattered comments that did not address the theme of hosting Afghan refugees in Portugal.

Both the symbolic threats of a religious nature and the observed hate speech related to religion are indicators pointing to another phenomenon mentioned in the literature review: the growth of Islamophobia recorded in Western societies, fueled by terrorist attacks in the United States of America and Europe, which exacerbated anti-Muslim sentiment [10].

Our study also aimed to uncover prevailing beliefs about Muslims among the Portuguese population, providing valuable insights into societal attitudes and perceptions. Comparing our findings with Araújo’s (2019) [24] identification of Islamophobic narratives in Portugal reveals both converging themes and notable distinctions, shedding light on the evolving landscape of public discourse surrounding Islam.

Araújo’s identification of 10 Islamophobic narratives reflects deeply ingrained stereotypes and prejudices prevalent in Portuguese society. These narratives portray Muslims as inherently violent, misogynistic, intolerant, and incompatible with Western values. Moreover, they propagate fears of cultural invasion, religious supremacy, and the imposition of what are designated as archaic practices such as Sharia law.

Our study corroborates several of these narratives, notably the perception of Islam as a threat to European culture and identity. This belief aligns closely with Araújo’s narrative of Islam undermining Western values and seeking to “invade Europe.” Similarly, the perception of Islam as inherently violent and promoting terrorism resonates with Araújo’s narrative of Muslims advocating violence and promoting fundamentalism.

However, our study also reveals distinct beliefs that were not explicitly captured in Araújo’s narratives. For instance, the belief that Europe will be Islamized in the future reflects anxieties about demographic shifts and cultural change, which may not have been explicitly articulated in Araújo’s findings. Likewise, the perception of Muslim immigrants as religious fanatics and the fear of Sharia law being imposed in Europe suggest unique concerns about social cohesion and religious pluralism.

Furthermore, our study identified beliefs regarding the inferior status of women in Islam, echoing Araújo’s narrative of Islam being misogynistic and sexist. However, while Araújo’s narratives focus primarily on ideological aspects of Islam, our findings also highlight concerns about practical implications, such as the imposition of Sharia law and the perceived threat of terrorism.

In light of these comparisons, it is evident that while some beliefs about Muslims in Portugal align with longstanding Islamophobic narratives, others reflect evolving anxieties and perceptions shaped by contemporary socio-political contexts. Understanding these nuances is crucial for addressing misconceptions and fostering inclusive dialogue within society.

5. Conclusions

This study employed an analytical framework encompassing sentiment analysis, approval degree assessment, perceived threat analysis, and hate speech examination to investigate public discourse on the Afghan refugee crisis within Portugal. By scrutinizing comments extracted from online publications, we aimed to elucidate the spectrum of opinions, emotions, and beliefs prevalent in Portuguese society regarding refugee resettlement.

The analysis revealed a predominant prevalence of negative sentiment and disapproval toward refugee resettlement, indicating substantial societal apprehension and resistance. Economic concerns emerged as focal points, reflecting anxieties regarding re-source allocation and perceived strain on national welfare systems. Symbolic threats, particularly those related to culture and religion, underscored deep-seated fears of cultural erosion and religious imposition. Furthermore, instances of hate speech, particularly with political, nationalist, and religious dimensions, were observed. These findings highlight the presence of divisive rhetoric within public discourse and emphasize the importance of fostering tolerance and understanding.

Comparison with existing literature on Islamophobia in Portugal revealed both persistent stereotypes and evolving anxieties about demographic shifts and cultural change. Our study provides valuable insights into the attitudes, perceptions, and beliefs shaping public discourse on the Afghan refugee crisis, informing evidence-based interventions aimed at promoting inclusivity, combating prejudice, and fostering compassionate refugee resettlement strategies.

To address the economic and safety concerns driving negative sentiments towards Afghan refugees, targeted interventions are crucial. Implementing job training and placement programs, along with small business support initiatives, can facilitate refugees’ economic integration and self-sufficiency. Enhancing community policing efforts and launching public safety campaigns could also build trust and alleviate the extensive security concerns verbalized. Additionally, fostering intercultural dialogue and school integration programs could promote mutual understanding and reduce cultural tensions from an early age.

Public awareness campaigns that highlight positive stories of refugee integration and success could be used to counterbalance negative sentiments and showcase the benefits of welcoming refugees. Moreover, partnering with NGOs for legal aid, mental health support, and social services, as well as establishing volunteer programs and community support networks, can further aid in the integration process, increase the connection and empathy among locals and promote social cohesion and inclusivity.

As Portugal and other nations navigate the challenges of forced migration, understanding the complexities of public opinion is essential for developing effective strategies that uphold human dignity, protect fundamental rights, and promote social cohesion.

This work is not without limitations. While the findings offer valuable insights into the public discourse surrounding the Afghan refugee crisis in Portugal, they may not be generalizable to other contexts or populations. Cultural, social, and political factors specific to Portugal may influence attitudes and perceptions differently than in other countries. Moreover, we have only captured a snapshot of social users from Facebook, which might not represent the general population’s view. We believe, however, that, on top of the local insights into the Afghan refugee crisis within Portugal, we have provided a relevant and comprehensive methodology to investigate this and other similar phenomena in different cultural settings, which can support a much-needed call for cross-cultural research on the aforementioned crisis and on other current crises, such as the ones deriving from Gaza and Ukraine.