The Prevalence of Prejudice-Denoting Terms in Spanish Newspapers

Abstract

Previous scholarly literature has documented a pronounced increase in the prevalence of prejudice-denoting terms in American news media content. Some have referred to this shift in journalistic discourse and related public opinion trends signaling increasing perceptions of prejudice severity in U.S. society as The Great Awokening. This work analyzes whether the increasing prevalence of prejudice themes in American news media outlets has been replicated in the news media ecosystem of a Spanish-speaking country. Thus, we computationally analyzed the prevalence of words denoting prejudice in five million news and opinion articles written between 1976 and 2019 and published in three of the most widely read newspapers in Spain: El País, El Mundo and ABC. We report that within the studied time period, the frequency of terms that denote specific prejudice types related to gender, ethnicity, sexuality and religious orientation has also substantially increased across the analyzed Spanish news media outlets. There are, however, some notable distinctions in the long-term usage dynamics of prejudice-denoting terms between the leading Spanish newspaper of record, El País, and its U.S. counterpart, The New York Times.

1. Introduction

A marked increase in the prevalence of prejudice-denoting terms in popular American news media outlets has previously been reported (Rozado 2020; Rozado et al. 2021a). Some authors have referred to this dynamic in journalistic discourse and related social attitudes shifts like rising public perceptions of prejudice severity in U.S. society as The Great Awokening (Yglesias 2019; Ungar-Sargon 2021; The Economist 2021). The journalistic trend was first identified in The New York Times (Rozado 2020), and it was later confirmed across 47 news media organizations popular in the United States (Rozado et al. 2021a). Specifically, terms denoting prejudice related to gender, ethnicity, gender identity, religious affiliation and sexual orientation have skyrocketed in prevalence in U.S. news media content within the span of just 10 years (2010–2019). The trend preceded the political rise of Donald Trump in 2015, but it could have been reinforced by it. The trend also appears to have continued unabated after Trump left office (Rozado et al. 2021b).

In this study, we seek to examine whether the rising trend in the usage of prejudice-denoting terms across American news media outlets replicates in the context of Spanish newspapers. Specifically, we examined news and opinion articles from the three most widely read generalist newspapers in Spain: El País, El Mundo and ABC (OJD: El País cierra 2020 con 70.000 Ejemplares de Difusión y El Mundo con 42.000 2021). These are considered the three newspapers of record in Spain (Newspaper of Record 2021; El País 2021). Conveniently, the three newspapers span the ideological spectrum of Spanish mainstream politics (LÍNEA EDITORIAL de Algunos Periódicos Españoles n.d.).

El País was created in 1976, six months after the death of dictator Francisco Franco. Its emergence in the Spanish journalistic landscape was considered revolutionary because it was the first newspaper with a clear pro-democracy and pro-European editorial line in the context of a news media environment that had largely been supportive of the Franco regime. Its ideological leanings since its inception have been center-left, and it has historically been aligned with the main social democratic political party in the country (PSOE) (El País 2021).

El Mundo was created in 1989. Its editorial line is often critical of the ideological left, the conservative Popular Party (PP) and Spanish regional nationalisms, which suggests a considerable amount of ideological heterogeneity among its journalists. This newspaper played a critical role in the 1996 electoral loss of the left-leaning PSOE presidency of Felipe Gonzalez, who had been in power since 1982. El Mundo was also critical of the presidency of conservative Jose Maria Aznar (1996–2004) and his support for the U.S.-led invasion of Iraq in 2003. Consequently, El Mundo also played an important role in the electoral loss of the Popular Party in the national presidential election of 2004. Overall, El Mundo is widely considered a centrist newspaper that self-defines its ideological line as secular and liberal (El Mundo (Spain) 2021; LÍNEA EDITORIAL de Algunos Periódicos Españoles n.d.).

ABC was founded in 1903. Throughout its history, it has been a conservative and pro-monarchist newspaper. Since the late 1970s, the ABC newspaper has had a center-right editorial line and close ties with the conservative Popular Party of Spain and its predecessor, Alianza Popular (ABC 2021).

In the present study, beyond reporting the prevalence of prejudice-denoting terms in the most important Spanish newspapers, we also attempt to compare the overall long-term prevalence dynamics of prejudice-denoting terms in the Spanish newspaper of record, the left-leaning El País, and its U.S. counterpart, The New York Times. The reason for choosing these two outlets is due to the long-term availability of news articles in both El País and The New York Times online domains (spanning all the way back to the 1970s). In contrast, El Mundo and ABC only have articles available in their online domains from the early 2000s onwards. The New York Times usage dynamics of prejudice-denoting terms has been studied extensively before (Rozado 2020; Rozado et al. 2021a). Such studies showed that the usage of prejudice-denoting terms in The New York Times was highly correlated with the usage of those terms in other prestige news media outlets such as The Washington Post. The present analysis will attempt to illustrate to what extent the trends around prejudice themes embedded in prestigious U.S. news media are also replicated in Spanish newspapers.

Thematic prevalence in news media content can be used for ‘agenda setting’ purposes of public opinion (McCombs and Valenzuela 2021). Increasing thematic news media coverage of terrorism or crime has often preceded growing public concern about those issues, even if the actual baseline rates of terror incidents or crime do not justify increasing media coverage of those topics (Callanan 2012; Smith et al. 2019). Previous studies on news media ‘agenda-setting’ have also found that prestigious outlets tend to precede other outlets in pioneering the increasing prevalence of topics. This is perhaps modulated by journalists striving to imitate coverage trends in prestige news media (McCombs 2005). Political biases and financial incentives within newsrooms could also impact the ‘agenda setting’ dynamics of media organizations (al-Gharbi 2020, n.d.).

Methodologically, we used a Big Data measure of discursive thematic prevalence in Spanish news media written content by computationally analyzing more than five million news and opinion articles published by the three studied newspapers over the last 44 years. Computational content analysis of news outlets content has previously been shown to be helpful in characterizing the dynamics of patterns embedded in journalistic discourse (Rozado 2021). Plotting word frequencies from a chronological corpus of written news articles can be a useful method to accurately track social dynamics within the temporal and spatial contexts where the texts were written (Rozado 2020, 2021; Rozado et al. 2021a).

The observational methodology used in this work is limiting because it does not allow us to elucidate the causal roots of the uncovered trends in journalistic discourse. Thus, our aim is descriptive. However, our contribution is original. Consequently, this study pioneers an attempt to chronologically document the prevalence of prejudice-denoting terms across a large and representative set of written articles from three of the most widely read Spanish newspapers.

2. Materials and Methods

This study examines news and opinion articles from three influential Spanish newspapers. The textual content of news articles is available in the outlets’ online domains. We only analyzed articles’ headlines and the main body of text. We did not include other article elements such as figure captions in our analysis. Targeted n-grams were located in articles’ raw data using outlet-specific XPath expressions. Tokens were made lowercase prior to estimating frequency counts.

To prevent deriving noisy frequency counts from newspapers with sparse text content for a given year, we only used outlet frequencies from years for which the outlet had more than 1 million tokens of text available in its online domain.

Yearly relative frequencies of target unigrams or n-grams in an outlet were estimated by dividing the number of occurrences of the target word/n-gram in all articles within a year by the total number of words in the year. This normalization of absolute frequency counts controlled for the variable volume of published articles in different years. Thus, enabling comparisons of relative frequencies across time spans irrespective of the variable number of published articles per year for each outlet.

The temporal availability of article content in different online newspapers is variable. El País newspaper, founded in 1976, has comprehensive online availability of news articles since its inception. For ABC and El Mundo, the online availability of news articles is more limited, and only starts in 2001 and 2002, respectively. Figure 1 illustrates the number of articles analyzed per outlet and year based on news outlets’ articles availability in their online domains. Overall, the total number of articles analyzed in the three newspapers combined was above 5 million.

In a data analysis of millions of news articles, it is not feasible to manually check the correctness of frequency counts for all articles. We have noted that, occasionally, outlet-specific XPath expressions might fail to correctly detect specific articles’ textual content due to edge cases in the underlying HTML or CSS source code in which the textual payloads of online news articles are embedded. This can result in incorrect frequency counts for a small subset of articles. However, overall, the method provides reliable results, as illustrated in Figure 2, which shows yearly min–max scaled frequencies of sample n-grams in El País content during the past 44 years.

Min–max scaling rescales the range of relative frequencies to a scale between 0 and 1 (see Equation (1)). This approach allows for the comparison of minimum and maximum frequency usage of the analyzed terms, irrespective of the ranges of their absolute or relative frequencies. The final row in Figure 2 lists the names of the last 5 prime ministers of Spain, arranged in columns chronologically, to illustrate that our methodology precisely captures the temporal dynamics of their presidencies.

$$y^{\prime }=\frac{y-\min \left(y\right)}{\max \left(y\right)-\min \left(y\right)}$$

(1)

3. Results

The prevalence of terms denoting prejudice against gender, sexual orientation, gender identity, ethnicity and religious sentiment has markedly increased in the Spanish newspaper of record, El País, since 1976 (see Figure 3). Some of the target terms, such as sexist or anti-semite, appear to show a gradual increase in prevalence since the 1980s, whereas others show a more abrupt increase in prevalence, only starting in the early 2010s (i.e., white supremacist or misogyny).

We next compared the usage dynamics of aggregate sets of related prejudice-denoting terms across three widely read Spanish newspapers: El País, El Mundo and ABC, see Figure 4. The online availability of news articles in the online domains of El Mundo and ABC only starts in 2002 and 2001, respectively. Hence, to maintain temporal consistency across newspapers, we circumscribed the next analysis to the temporal window 2002–2019.

For the sexism theme, all three newspapers in Figure 4 displayed a considerable increase in the prevalence of this topic since the year 2002. The increase was most acute in El Mundo (1306%), although was non-negligible in El País (553%). Historically, the theme has been more prevalent in the left-leaning newspaper, El País. However, in recent years, El Mundo, has caught up, and as of 2019, the overall prevalence in both newspapers is similar. The right-leaning newspaper ABC displays an overall lesser prevalence of this topic in its written content. However, even the ABC newspaper has displayed a substantial increase in the prevalence of this topic (608%) since 2002.

For the sexual orientation and gender identity prejudice theme, there has also been a marked increase over time in the prevalence of this topic for all newspapers analyzed. The increase has been noticeably higher for El País (746%). However, it is also substantial in El Mundo (325%) and ABC (252%). In recent years, the overall prevalence of this topic has been largest in El País. Again, the right-leaning newspaper ABC displays the least prevalence of this theme in the three newspapers analyzed.

The ethnic prejudice theme has only displayed mild increases in prevalence for both El País and El Mundo since 2002. However, the year 2002 exhibited an unusually high prevalence of this topic in all newspapers analyzed, and the prevalence dropped substantially in the following year and stayed relatively constant for almost a decade. The theme has, however, increased moderately since the early 2010s in both El País and El Mundo. The ABC newspaper has displayed a mild decrease in the prevalence of this topic since 2002. Overall, the prevalence of this topic is again the largest in the left-leaning newspaper El País, with El Mundo displaying a similar overall prevalence in recent years, and the right-leaning ABC lagging behind.

Finally, for the religious orientation prejudice theme, there have been increases in prevalence of different intensities across the three newspapers analyzed, with the largest increase being observed in the centrist newspaper El Mundo (321%). Again, the right-leaning newspaper ABC has displayed the lowest overall prevalence of this topic in recent years.

Overall, the time series for the different prejudice themes shows a high degree of correlation between El País and El Mundo newspapers, with ABC often displaying the least correlation with the other two outlets (see the correlation matrices in each subplot in Figure 4).

In terms of comparative overall prevalence between the different prejudice themes, gender and ethnic prejudice themes have been more prevalent in Spanish journalistic discourse than themes related to sexual or religious orientation prejudice (see y-axes scales in Figure 4).

We next analyzed the overall prevalence dynamics of prejudice-denoting terms in the Spanish newspaper of record, the left-leaning El País, and its U.S. counterpart, The New York Times, see Figure 5. Overall, the gender prejudice theme has historically been more prevalent in El País than in The New York Times. The rate of increase in this topic since the 1970s has also been larger for El País (1127% vs. 555%). As of 2019, the overall prevalence of this topic is about two times larger in El País than in The New York Times. The Pearson’s correlation coefficient between El País and The New York Times time series is large ($r=$ 0.92).

For the sexual orientation and gender identity prejudice topic, the overall prevalence and rate of increase has roughly been similar between both outlets in recent years. The percentage increase in the prevalence of this topic since the 1970s has been extremely high in both outlets (25349% for El País and 39255% for The New York Times), probably suggesting a tectonic shift in attitudes towards sexual orientation identity in the United States and Spain within the studied time span. During the 1990s, the prevalence of this topic was higher in The New York Times in comparison with El País, but overall prevalence has converged in recent years. The Pearson’s correlation coefficient between El País and The New York Times for this topic time series is also high ($r=$ 0.89).

The ethnic/racial prejudice theme displays the least degree of coupling in usage frequency between El País and The New York Times (although still shows a moderate correlation, Pearson’s $r=$ 0.56). This topic has also shown, in comparison to other themes, a milder increase in overall prevalence, especially for El País (258%). From 1990 until the early 2010s, the overall prevalence of this topic was larger in El País than in The New York Times. This coincided with a peak in the prevalence of this theme in El País content in the 1990s. The increase in this theme in The New York Times has been very dramatic within the last decade, and seemed to begin just prior to the irruption of Donald Trump onto the political landscape of the United States in 2015, appearing to accelerate dramatically after this pivotal year.

Finally, the prevalence of terms denoting religious sentiment prejudice has also increased substantially across both outlets (3404% in El País and 3150% in The New York Times). In recent years, the increase has been particularly dramatic for The New York Times. Historically, the overall prevalence of this topic has been relatively similar between both outlets. The Pearson’s correlation between both time series is moderate ($r=$ 0.72).

4. Discussion

This study has documented a marked increase in the prevalence of prejudice-denoting terms across a large and representative set of written news articles from Spain’s most popular newspapers of record. Our analyses reveal that the dramatic increasing prevalence of prejudice-denoting terms previously noted in American news media is also happening in leading Spanish newspapers. The increase in frequency of words that denote prejudice against minorities and protected groups is ubiquitous across a set of newspapers spanning the mainstream ideological spectrum of Spanish politics. However, the noted trends appear less prevalent in the right-leaning newspaper ABC. Indeed, both El País and El Mundo show a high degree of synchronicity in the temporal dynamics with which they use prejudice-denoting terms, perhaps suggesting substantial journalistic alignment in the coverage of these topics between both outlets.

For the gender, ethnic and sexual orientation prejudice themes, the rising trends in prejudice-denoting term prevalence in Spanish newspapers often preceded the political rise of Donald Trump for most of the analyzed n-grams, but appeared to hasten after 2015.

We have identified an important distinction between the Spanish newspaper of record, El País, and its U.S. counterpart The New York Times. Although for most prejudice themes, the time series of prejudice-denoting terms in both outlets are highly correlated, this is less apparent for words denoting ethnic prejudice. Since 2015, thematic prevalence of ethnic prejudice is much milder in the Spanish newspaper. In recent years, the rate of increase in this topic in El País is also weaker than that of its counterpart in the United States. This perhaps reflects the heightened racial tensions that have engulfed the United States in recent years.

In contrast, the overall prevalence of terms denoting gender prejudice is substantially larger in El País than in its New York Times counterpart. The rate of increase in this topic is also larger in the Spanish newspaper. This could suggest that concerns around gender prejudice are more acute in Spanish society or within Spanish newsrooms. Unfortunately, our methodology is limiting to allow discerning between these two alternative hypotheses. Finally, the overall prevalence and rates of increase for sexual orientation and religious sentiment prejudice appear more similar between both newspapers. The substantial correlation between the frequencies time series of both newspapers for all prejudice types suggests the existence of synchronized intellectual currents around the topic of prejudice which span across borders and language barriers.

Previous research has noted a strong correlation between the growing occurrence of prejudiced themes in American news media discourse and increasing societal concerns regarding the prevalence of prejudice in society (Rozado et al. 2021a). Such findings are highly reminiscent of previous scholarly literature modeling the influence of news media content on the formation of public opinion about crime or terrorism. Thus, increasing thematic prevalence of prejudice in Spanish news media, irrespective of underlying base rates, could be shaping Spanish public opinion regarding the perceived severity of prejudice in Spanish society. Future survey-based research trying to determine the temporal dynamics of subjective perceptions of prejudice severity in Spain could clarify the relationship between news media content and public opinion in Spanish society.

In conclusion, the presented analysis has documented a marked increase in the prevalence of prejudice-denoting terms in three widely read Spanish newspapers. The dramatic increases in such words’ frequencies, particularly post-2010, and the existence of similar trends in news outlets from the United States, suggest the existence of powerful international journalistic or social dynamics at play. We hope that our detailed characterization of the phenomena can inspire future studies exploring the causal roots for the trends described herein.