{ "paper_id": "2021", "header": { "generated_with": "S2ORC 1.0.0", "date_generated": "2023-01-19T02:10:27.751459Z" }, "title": "Using Word Embeddings to Quantify Ethnic Stereotypes in 12 years of Spanish News", "authors": [ { "first": "Danielly", "middle": [], "last": "Sorato", "suffix": "", "affiliation": { "laboratory": "", "institution": "Universitat Pompeu Fabra Barcelona", "location": { "country": "Spain" } }, "email": "danielly.sorato@upf.edu" }, { "first": "Diana", "middle": [], "last": "Zavala-Rojas", "suffix": "", "affiliation": { "laboratory": "European Social Survey ERIC Universitat Pompeu Fabra", "institution": "", "location": { "settlement": "Barcelona", "country": "Spain" } }, "email": "" }, { "first": "Carme", "middle": [], "last": "Colominas", "suffix": "", "affiliation": { "laboratory": "", "institution": "Universitat", "location": { "addrLine": "Pompeu Fabra", "settlement": "Barcelona", "country": "Spain" } }, "email": "carme.colominas@upf.edu" } ], "year": "", "venue": null, "identifiers": {}, "abstract": "The current study provides a diachronic analysis of the stereotypical portrayals concerning seven of the most prominent foreign nationalities living in Spain in a Spanish news outlet. We use 12 years (2007-2018) of news articles to train word embedding models to quantify the association of such outgroups with drug use, prostitution, crimes, and poverty concepts. Then, we investigate the effects of sociopolitical variables on the computed bias series, such as the outgroup size in the host country and the rate of the population receiving unemployment benefits. Our findings indicate that the texts exhibit bias against foreign-born people, especially in the case of outgroups for which the country of origin has a lower Gross Domestic Product per capita (PPP) than Spain.", "pdf_parse": { "paper_id": "2021", "_pdf_hash": "", "abstract": [ { "text": "The current study provides a diachronic analysis of the stereotypical portrayals concerning seven of the most prominent foreign nationalities living in Spain in a Spanish news outlet. We use 12 years (2007-2018) of news articles to train word embedding models to quantify the association of such outgroups with drug use, prostitution, crimes, and poverty concepts. Then, we investigate the effects of sociopolitical variables on the computed bias series, such as the outgroup size in the host country and the rate of the population receiving unemployment benefits. Our findings indicate that the texts exhibit bias against foreign-born people, especially in the case of outgroups for which the country of origin has a lower Gross Domestic Product per capita (PPP) than Spain.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "Abstract", "sec_num": null } ], "body_text": [ { "text": "Languages are complex and systematic instruments of communication that reflect the culture of a given population. By studying language, it is possible to observe stereotypes, a type of social bias that is present when discourse about a given group overlooks the diversity of its members and focuses only on a small set of features (S\u00e1nchez-Junquera et al., 2021; Tajfel et al., 1964) . As such, language analysis is a good way to depict, understand, and demonstrate stereotypes (Garg et al., 2018; Basow, 1992; Wetherell and Potter, 1993; Bonilla-Silva and Forman, 2000) . Nonetheless, like society, languages are not static. Variations in lexical systems can be observed over time due to a myriad of intraand extra-linguistic factors. By analyzing extralinguistic aspects, it is possible to gain insights into the dynamics of social, cultural, and political phenomena reflected in texts (Marakasova and Neidhardt, 2020) .", "cite_spans": [ { "start": 331, "end": 362, "text": "(S\u00e1nchez-Junquera et al., 2021;", "ref_id": "BIBREF39" }, { "start": 363, "end": 383, "text": "Tajfel et al., 1964)", "ref_id": "BIBREF42" }, { "start": 478, "end": 497, "text": "(Garg et al., 2018;", "ref_id": "BIBREF13" }, { "start": 498, "end": 510, "text": "Basow, 1992;", "ref_id": "BIBREF2" }, { "start": 511, "end": 538, "text": "Wetherell and Potter, 1993;", "ref_id": "BIBREF46" }, { "start": 539, "end": 570, "text": "Bonilla-Silva and Forman, 2000)", "ref_id": "BIBREF8" }, { "start": 888, "end": 920, "text": "(Marakasova and Neidhardt, 2020)", "ref_id": "BIBREF26" } ], "ref_spans": [], "eq_spans": [], "section": "Introduction", "sec_num": "1" }, { "text": "Efficient methods for performing diachronic analysis are crucial, as manually evaluating several years of text collections is unfeasible due to the large amount of data involved. As such, computational methods for diachronic linguistic analysis are of utmost importance, and ongoing research shows that word embeddings models are helpful tools to this end (Garg et al., 2018; Kroon et al., 2020; Hamilton et al., 2016; Kutuzov et al., 2018; Lauscher et al., 2020) .", "cite_spans": [ { "start": 356, "end": 375, "text": "(Garg et al., 2018;", "ref_id": "BIBREF13" }, { "start": 376, "end": 395, "text": "Kroon et al., 2020;", "ref_id": "BIBREF21" }, { "start": 396, "end": 418, "text": "Hamilton et al., 2016;", "ref_id": "BIBREF17" }, { "start": 419, "end": 440, "text": "Kutuzov et al., 2018;", "ref_id": "BIBREF22" }, { "start": 441, "end": 463, "text": "Lauscher et al., 2020)", "ref_id": "BIBREF23" } ], "ref_spans": [], "eq_spans": [], "section": "Introduction", "sec_num": "1" }, { "text": "Word embeddings are powerful representations of language, that allow for the quantification of relationships between words through efficient numerical operations inside the vector space. In this context, previous works demonstrated that such models contain machine-learned biases in their geometry that closely depict societal stereotypes (Bolukbasi et al., 2016b; Gonen and Goldberg, 2019; Garg et al., 2018; Kroon et al., 2020) , which is not surprising since stereotypes are massively present in texts used to train computational models (S\u00e1nchez-Junquera et al., 2021; Nadeem et al., 2020) . Although such language models should be carefully tested for biases and not blindly applied to widely computational applications due to ethically concerning outcomes (Papakyriakopoulos et al., 2020; Brandon, 2021; Bender et al., 2021) , they can be a valuable tool for enabling sociolinguistic analysis on large volumes of textual data. This topic establishes a collaboration between computer science, social sciences, and linguistics, as hypotheses about social phenomena can be tested on language using computational methods.", "cite_spans": [ { "start": 339, "end": 364, "text": "(Bolukbasi et al., 2016b;", "ref_id": "BIBREF7" }, { "start": 365, "end": 390, "text": "Gonen and Goldberg, 2019;", "ref_id": "BIBREF15" }, { "start": 391, "end": 409, "text": "Garg et al., 2018;", "ref_id": "BIBREF13" }, { "start": 410, "end": 429, "text": "Kroon et al., 2020)", "ref_id": "BIBREF21" }, { "start": 540, "end": 571, "text": "(S\u00e1nchez-Junquera et al., 2021;", "ref_id": "BIBREF39" }, { "start": 572, "end": 592, "text": "Nadeem et al., 2020)", "ref_id": "BIBREF31" }, { "start": 761, "end": 793, "text": "(Papakyriakopoulos et al., 2020;", "ref_id": "BIBREF34" }, { "start": 794, "end": 808, "text": "Brandon, 2021;", "ref_id": "BIBREF9" }, { "start": 809, "end": 829, "text": "Bender et al., 2021)", "ref_id": "BIBREF4" } ], "ref_spans": [], "eq_spans": [], "section": "Introduction", "sec_num": "1" }, { "text": "In this study, we analyze the dynamics of stereotypical associations with seven nationalities, in the period of 2007 to 2018. We train our word embedding models using 1,757,331 news articles published in the Spanish newspaper 20 Minutos, for the aforementioned time span. We adopt a culturally diverse perspective by taking into account some of the most representative foreign nationalities that lived in Spain in the aforementioned period according to the Instituto Nacional de Estad\u00edstica (INE) 1 .", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "Introduction", "sec_num": "1" }, { "text": "Namely, British, Colombian, Ecuadorian, German, Italian, Moroccan, and Romanian are included in this study.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "Introduction", "sec_num": "1" }, { "text": "We conduct a fine-grained analysis, studying the association of such nationalities with drug use, prostitution, crimes, and poverty concepts. Then, we compare our findings with sociopolitical variables, such as survey items from the European Social Survey (ESS), number of residents by nationality living in Spain, the rate of the population receiving unemployment benefits from the Spanish government, and the number of offenses committed in Spain by outgroup background. Additionally, we investigate the effect of the outgroups' countries of origin having a lower Gross Domestic Product per capita (PPP) than the host country (Spain) 2 . To account for both group effects and error correlation, we use multilevel Random Effects (RE) models in our analysis.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "Introduction", "sec_num": "1" }, { "text": "This paper is organized as follows. In Section we discuss related works. Subsequently, in Section 3 we state our research questions, present metrics, data, model training, and evaluation. Section 4 comprises the findings and discussion about results derived from this study. Finally, in Section 5 we present our conclusions, limitations, and future work.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "Introduction", "sec_num": "1" }, { "text": "Word embeddings showed as a valuable tool, by means of enabling efficient methods for analyzing and quantifying linguistic and social phenomena in natural language. In the context of model stereotypical bias analysis, which is the focus of this paper, the first disseminated studies concern gender bias (Bolukbasi et al., 2016a,b; Zhao et al., 2018; Gonen and Goldberg, 2019; Park et al., 2018; Zhou et al., 2019) . Nonetheless, biases can exist in many shapes and forms, which can lead to unfairness in subsequent downstream tasks (Mehrabi et al., 2019) .", "cite_spans": [ { "start": 303, "end": 330, "text": "(Bolukbasi et al., 2016a,b;", "ref_id": null }, { "start": 331, "end": 349, "text": "Zhao et al., 2018;", "ref_id": "BIBREF48" }, { "start": 350, "end": 375, "text": "Gonen and Goldberg, 2019;", "ref_id": "BIBREF15" }, { "start": 376, "end": 394, "text": "Park et al., 2018;", "ref_id": "BIBREF35" }, { "start": 395, "end": 413, "text": "Zhou et al., 2019)", "ref_id": "BIBREF49" }, { "start": 532, "end": 554, "text": "(Mehrabi et al., 2019)", "ref_id": "BIBREF28" } ], "ref_spans": [], "eq_spans": [], "section": "Related Work", "sec_num": "2" }, { "text": "Garg et al., used both pre-trained models and models trained with the New York Times Annotated Corpus to quantify gender and ethnic stereotypes in 100 years of data for the English language. The reported bias series showed strong correlations with census data and demographic changes in the United States for gender and ethnic stereotypes. Similarly, According to the Data World Bank https://databank.worldbank.org bedding models, but focusing on social class biases.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "Related Work", "sec_num": "2" }, { "text": "Most works concerning the study of machine learned biases have English as target language, since there is more availability of linguistic resources that favors such analysis. Here we cite four relevant works conducted on non-English target languages. Wevers quantified gender biases in 40 years of Dutch newspapers categorized ideologically as liberal, social-democratic, neutral/conservative, Protestant, and Catholic. The results depict differences in gender bias and changes within and between newspapers over time. Tripodi et al. investigated the antisemitism in public discourse in France, by using diachronic word embeddings trained on a large corpus of French books and periodicals containing keywords related to Jews. Using the changes over time and embedding projections, they tracked the dynamics of antisemitic bias in the religious, economic, sociopolitical, racial, ethnic and conspiratorial domains. S\u00e1nchez-Junquera et al. detected stereotypes towards immigrants in political discourse by focusing in the narrative scenarios, i.e. the frames, used by political actors. They propose a taxonomy to capture immigrant stereotype dimensions and produced an annotated dataset with sentences that Spanish politicians have stated in the Congress of Deputies. Such dataset was used to train classifiers that detect and distinguish between stereotype categories.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "Related Work", "sec_num": "2" }, { "text": "More similar to ours, is the work of Kroon et al. In their study, the authors quantify the dynamics of stereotypical associations with different outgroups concerning low-status and high-threat concepts in 11 years of Dutch news data. The authors investigate both time invariant and time variant hypotheses, focusing on the difference of associations regarding the group membership (ingroup vs outgroups).", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "Related Work", "sec_num": "2" }, { "text": "Our study distinguishes itself from the aforementioned studies by (i) the interdisciplinarity with social survey research, as the selected survey questions measure attitudes of Spanish people (the ingroup) towards immigrants (the outgroups) and can be interpreted as a proxy for cultural/economic threat perception; (ii) our choice of multilevel modeling (RE model), to combine types of phenomena (linguistic and social) and account for group effects; and (iii) the use of fine-grained lists representing crimes, drugs, poverty and prostitution concepts to investigate stereotypical portrayals. Additionally, we contribute to the scarce literature on stereotypical bias analysis with non-English data sources by using Spanish from Spain as a target language.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "Related Work", "sec_num": "2" }, { "text": "In this work, we aim to study the dynamics of the stereotypical portrayals of British, Colombian, Ecuadorian, German, Italian, Moroccan and Romanian nationalities with drugs, prostitution, crimes, and poverty concepts, which are some of the stereotypical frames associated to immigrants in the literature (Neyland, 2019; Kroon et al., 2020; Warner, 2005; Igartua et al., 2005; Light and Young, 2009) . We investigate the effect that the Gross Domestic Product per capita (PPP) of the outgroup's country of origin has in the strength of stereotypical association. Namely, our hypothesis is that outgroups coming from countries with lower PPP than the host country (Spain), are more strongly associated with such concepts, due to posing a greater economic threat to the ingroup (Meuleman, 2011; Manevska and Achterberg, 2013) 3 .", "cite_spans": [ { "start": 305, "end": 320, "text": "(Neyland, 2019;", "ref_id": "BIBREF32" }, { "start": 321, "end": 340, "text": "Kroon et al., 2020;", "ref_id": "BIBREF21" }, { "start": 341, "end": 354, "text": "Warner, 2005;", "ref_id": "BIBREF45" }, { "start": 355, "end": 376, "text": "Igartua et al., 2005;", "ref_id": "BIBREF19" }, { "start": 377, "end": 399, "text": "Light and Young, 2009)", "ref_id": "BIBREF24" }, { "start": 776, "end": 792, "text": "(Meuleman, 2011;", "ref_id": "BIBREF29" }, { "start": 793, "end": 823, "text": "Manevska and Achterberg, 2013)", "ref_id": "BIBREF25" } ], "ref_spans": [], "eq_spans": [], "section": "Method", "sec_num": "3" }, { "text": "Then, we evaluate to what extent our findings can be explained by (i) the number of residents per nationality in Spain (i.e, the size of outgroup); (ii) rates of population receiving unemployment benefits; (iii) the number of offenses committed in the Spanish territory by outgroup background and; (iii) public opinion. In order to investigate such hypothesis, we adopt the following metrics, procedures and data.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "Method", "sec_num": "3" }, { "text": "Distributional semantic models maintain the properties of vector spaces and adopt the hypothesis that meaning of a word is conveyed in its cooccurrences. Therefore, in order to measure the similarity between two given words represented by the vectors v 1 and v 2 we can apply the L 2 normalized cosine similarity, although as shown by Garg et al., one could apply the Euclidean distance interchangeably.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "Metrics", "sec_num": "3.1" }, { "text": "To quantify social stereotypes in the trained word embedding models, we used a metric referred throughout this paper as bias score, which is the same metric used in Garg et al.. Such metric has been specifically chosen because it has been externally validated by the authors through correlations with census data. The bias score captures the strength of the association of a given set of words S with respect to two groups v 1 and v 2 . Hence, when we state that a word is biased toward a group, it is in the context of the bias score metric. The bias score equation is computed as in Equation 1, where S is a set of word vectors that represent a concept of interest (e.g., crimes), v 1 and v 2 are the averaged group vectors for word vectors in group one and two, respectively. An averaged group vector is computed by simply averaging the word vectors that compose a given group. The more negative that the bias score is, the more associated S is toward group two whereas the more positive, the more associated S is towards group one.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "Metrics", "sec_num": "3.1" }, { "text": "bias score = vs\u2208S cos(v s , v 1 ) \u2212 cos(v s , v 2 ) (1)", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "Metrics", "sec_num": "3.1" }, { "text": "To refer to the representation of the outgroups inside of the context of the embedding model and the bias score metric throughout this paper, we will use the name of the nationality in italics (e.g., Spanish, Moroccan).", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "Metrics", "sec_num": "3.1" }, { "text": "We compare the similarity of concepts (i.e., word lists) related to drugs, prostitution, crimes and poverty to the concepts that represent the ingroup and the outgroups. ", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "Metrics", "sec_num": "3.1" }, { "text": "We compiled the Corpus of Spanish news 20 Minutos (Razgovorov et al., 2019) . The corpus contains 14 years of articles written in Spanish from Spain, comprising 711.840.945 distinct words, that were web-scraped from the newspaper 20 Minutos 4 website in JSON format. Due to the limited availability of data measuring the sociopolitical indicators of interest (stated in the next subsection), we consider the years 2007 up to 2018 in our analysis. According to a survey made in 2017 by Cardenal et al., about 40% of the consulted experts in the areas of political science and information science in Spain consider 20 Minutos is a neutral paper. The Figure 1 shows the number of articles and sentences per year in the corpus. Noticeably, for the years 2007 up to and including 2009 there is less data than for the subsequent years. We preprocessed the corpus, lower casing words, removing punctuation and numbers. Then, we filtered the data to create a dataset for each year of the corpus.", "cite_spans": [ { "start": 50, "end": 75, "text": "(Razgovorov et al., 2019)", "ref_id": "BIBREF37" } ], "ref_spans": [ { "start": 648, "end": 656, "text": "Figure 1", "ref_id": "FIGREF1" } ], "eq_spans": [], "section": "Corpus", "sec_num": "3.2" }, { "text": "To build a sociopolitical indicator of ethnic threat perception, we use the mean score of three survey items from the European Social Survey (ESS) (NSD, 2020) studies (2006, 2008, 2010, 2012, 2014, 2016 and 2018) . We used the Spanish respondent's answers (applying sample weights provided by ESS) of 11-point scales to the following questions: (i) \"Is [country] made a worse or a better place to live by people coming to live here from other countries?\"; (ii) \"Would you say that [country]'s cultural life is generally undermined or enriched by people coming to live here from other countries?\" and; (iii) \"Would you say it is generally bad or good for [country]'s economy that people come to live here from other countries?\". Missing data points for these time series were imputed using last observation carried forward (LOCF) strategy, which can be applied since the attitudes towards immigration tends to be stable from one year to another. Each survey was responded by at least 1500 people. The indicator of ethnic threat perception has the role of representing attitudinal data in the analysis, or in other words, identifying if the reported bias is somehow a reflection of the ingroup perceptions of these outgroups.", "cite_spans": [ { "start": 167, "end": 212, "text": "(2006, 2008, 2010, 2012, 2014, 2016 and 2018)", "ref_id": null } ], "ref_spans": [], "eq_spans": [], "section": "Sociopolitical variables", "sec_num": "3.3" }, { "text": "In addition, we use as indicators the number of foreign population by nationality residing in Spain 5 , the rate of the population receiving unemployment social benefits (foreigners from the EU excluding Spain and foreigners from outside the EU) 6 and committed offenses by background, which can be countries from the EU excluding Spain (British, Germans, Italians and Romanians), America (Colombians and Ecuadorian), and Africa (Moroccans) 7 . Such datasets are publicly available and can be found in the INE database.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "Sociopolitical variables", "sec_num": "3.3" }, { "text": "Using the datasets filtered by year, we trained skipgram embedding models using the Fasttext implementation (Bojanowski et al., 2017) . Since Spanish is a morphologically rich language, this model is a suitable choice as it takes into account the words' morphological structure. Due to the difference in the number of documents in the corpus across the years, we adopt a grid search strategy to define the optimal hyper-parameters of the models and favor embedding quality (see yearly hyper-parameters in Appendix). Only words that appeared at least 15 times in each yearly dataset were taken into account in the training phase. The resulting word vectors were L 2 normalized. We evaluate our models using two Spanish word similarity benchmarks, namely RG-65 (Camacho-Collados et al., 2015) and MC-30 (Hassan and Mihalcea, 2009) . The yearly models achieved an average of 0.72 and 0.70 Pearson correlation coefficient values in the RG-65 and MC-30 benchmarks for evaluating word similarity, respectively (variance RG \u2212 65 = 0.0003 and variance M C \u2212 30 = 0.0011). The evaluation results by year are shown in Appendix. In addition, we compute the average group vector for the ingroup and each of the outgroup nationalities and observe that, although some fluctuations can be observed for the German and Spanish, the variance is not significant. Therefore, our findings cannot be explained by the group vector variance.", "cite_spans": [ { "start": 108, "end": 133, "text": "(Bojanowski et al., 2017)", "ref_id": "BIBREF5" }, { "start": 801, "end": 828, "text": "(Hassan and Mihalcea, 2009)", "ref_id": "BIBREF18" } ], "ref_spans": [], "eq_spans": [], "section": "Word Embeddings Training and Evaluation", "sec_num": "3.4" }, { "text": "Here, we describe the process for selecting words that represent the crimes, drugs, poverty and prostitution concepts, as well as the ingroup and out- groups. The word lists used for creating the vector representations of the ingroup and the outgroups were defined according to a simple rule: the nationality in masculine singular and plural form (e.g., Espa\u00f1ol, Espa\u00f1oles). The total frequencies per year for words that compose such lists are shown in the Appendix.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "Word lists", "sec_num": "3.5" }, { "text": "In order to identify words that represent crimes, drugs, poverty and prostitution categories, we start by fitting the high-treat and low-status words used by Kroon et al. in the aforementioned concepts 8 . Then, using an embedding model trained with the whole content of the corpus instead of the yearly slices, for each of the words in the initial list we retrieve the 20 most similar words in the vector space. Afterwards, the lists increased in the step described above were revised and updated again by the authors, excluding words that fall out of the desired concept category. We exclude feminine word inflections to favor lower group vector variances since the analyzed dataset is not very large. The lists of words for used each category of concepts are shown in the Appendix.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "Word lists", "sec_num": "3.5" }, { "text": "Due to the pooled structure of the data, i.e., yearly bias score measurements for each of the outgroups, we build a panel with N = 84 observations (12 years x 7 outgroups). The stationary behaviour of the panel was verified by applying the Levin-Lin-Chu test, which is equivalent to a pooled unit root test. The non-stationary hypothesis was rejected, meaning that the panel data series altogether is unaffected by changes in time. This same test was applied to test the panel data stationary behaviour in Kroon et al.. Additionally, we performed a careful analysis of the model residuals to ensure that there were no correlation patterns.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "Panel Data", "sec_num": "3.6" }, { "text": "To investigate the dependent series, we impose a Random Effects (RE) multilevel model for panel data. A multilevel model is an extension of a regression, in which data is structured in groups and coefficients can vary by group (Gelman and Hill, 2006) . We consider the RE model an appropriate choice for this analysis, as we have pooled structured data and allows accounting for both group effects and error correlation. The following variables were used as predictors: Year trend: the years from 2007 to 2018, treated as a categorical variable. N Residents: size of outgroup residing in Spain, described in subsection 3.3. Unemployment benefits: rate of population receiving unemployment benefits, described in subsection 3.3. Perception: ingroup's perception of the outgroups, described in subsection 3.3. Offenses number of offenses committed in the Spanish territory, described in subsection 3.3. Lower PPP : dummy variable that indicates if the outgroups' country of origin has a Lower PPP than Spain. According to the Data World Bank 9 , the countries with PPP lower than Spain for the period of analysis are Colombia, Ecuador, Morocco, and Romania (LowerP P P = 1). The countries with higher PPP are Germany, Italy and United Kingdom (LowerP P P = 0).", "cite_spans": [ { "start": 227, "end": 250, "text": "(Gelman and Hill, 2006)", "ref_id": "BIBREF14" } ], "ref_spans": [], "eq_spans": [], "section": "Random Effects model", "sec_num": "3.7" }, { "text": "Analytical models should also be parsimonious, as fitting models with many random effects quickly multiplies the number of parameters to be estimated, particularly since random slopes are generally given covariances as well as variances (Bell et al., 2019; Matuschek et al., 2017) . Hence, the chosen aforementioned indicators are the ones that, to the best of our knowledge, are most appropriated (both regarding data availability and purpose) to test our hypothesis.", "cite_spans": [ { "start": 237, "end": 256, "text": "(Bell et al., 2019;", "ref_id": "BIBREF3" }, { "start": 257, "end": 280, "text": "Matuschek et al., 2017)", "ref_id": "BIBREF27" } ], "ref_spans": [], "eq_spans": [], "section": "Random Effects model", "sec_num": "3.7" }, { "text": "In this section we discuss the findings and limitations of the present research. We analyse the dynamics of stereotypical associations comprised in 12 years (2007) (2008) (2009) (2010) (2011) (2012) (2013) (2014) (2015) (2016) (2017) (2018) of Spanish local news published in the newspaper 20 Minutos, comprising 1,757,331 news items, by training and analyzing yearly word embedding language models. Our objective is to quantify stereotypes in such items towards the aforementioned outgroups, taking into account a cultural dimension by studying seven of the most prominent foreign outgroups living in Spain considering the aforementioned period of analysis. We explore the hypothesis that outgroups coming from countries which have a Lower PPP than the host country (Spain), have stronger stereotypical associations with concepts related to crimes, drugs, poverty and prostitution, as a consequence of representing a greater social threat to the ingroup.", "cite_spans": [ { "start": 157, "end": 163, "text": "(2007)", "ref_id": null }, { "start": 164, "end": 170, "text": "(2008)", "ref_id": null }, { "start": 171, "end": 177, "text": "(2009)", "ref_id": null }, { "start": 178, "end": 184, "text": "(2010)", "ref_id": null }, { "start": 185, "end": 191, "text": "(2011)", "ref_id": null }, { "start": 192, "end": 198, "text": "(2012)", "ref_id": null }, { "start": 199, "end": 205, "text": "(2013)", "ref_id": null }, { "start": 206, "end": 212, "text": "(2014)", "ref_id": null }, { "start": 213, "end": 219, "text": "(2015)", "ref_id": null }, { "start": 220, "end": 226, "text": "(2016)", "ref_id": null }, { "start": 227, "end": 233, "text": "(2017)", "ref_id": null }, { "start": 234, "end": 240, "text": "(2018)", "ref_id": null } ], "ref_spans": [], "eq_spans": [], "section": "Results and Discussion", "sec_num": "4" }, { "text": "The yearly average bias scores concerning concepts related to crimes and drugs are depicted in Figures 3 and 4 . The trends in Figure 3 show that, most of the outgroups are more strongly associated with the crimes concepts than the Spanish ingroup. The Colombian and the Romanian are the outgroups more strongly associated with crimes concepts, while the German and the British are the two outgroups less associated. In fact, for most years, the bias score values are negative for the German and the British outgroups. In contrast, for the Colombian, Ecuadorian, Morrocan, and Romanian outgroups, bias score values are always positive. A similar pattern can be observed in Figure 4 , in the case of stereotypes concerning drugs.", "cite_spans": [], "ref_spans": [ { "start": 95, "end": 110, "text": "Figures 3 and 4", "ref_id": "FIGREF3" }, { "start": 127, "end": 135, "text": "Figure 3", "ref_id": "FIGREF3" }, { "start": 673, "end": 681, "text": "Figure 4", "ref_id": null } ], "eq_spans": [], "section": "Results and Discussion", "sec_num": "4" }, { "text": "The results of the Random effects model for the aforementioned series are presented in Table 1 , and the main effects of the predictors are shown in the Model 1. In accordance to our expectations, the Lower PPP variable affects the bias significantly in both series. The positive coefficients indicate that the Colombian, the Ecuadorian, the Moroccan and the Romanian outgroups have higher stereotypical association with crimes and drugs concepts than the German, the British and the Italian outgroups. The year trend does have a significant effect, except for years 2009 and 2011 for crimes series, and years 2010 and 2011 for the drugs series. The positive coefficients indicate that the bias score for such years was higher than for the basis year, 2007.", "cite_spans": [], "ref_spans": [ { "start": 87, "end": 94, "text": "Table 1", "ref_id": "TABREF1" } ], "eq_spans": [], "section": "Results and Discussion", "sec_num": "4" }, { "text": "To further inspect the effects of the Lower PPP variable, we add interaction terms in Model 2. For both series, there is a strongly significant relationship between Lower PPP and Unemployment benefits, such that when the rate of population receiving unemployment benefits increases, the stereotype association for Colombian, Ecuadorian, Moroccan and Romanian (LowerP P P = 1) also increases, but decreases for German, British and Italian outgroups. Similarly, the interaction with the number of committed offenses in the drugs series reveals that an increase in the offenses lead to stronger stereotypical associations for the first outgroups, but not for the latter. For the series concerning crimes concepts, it is also possible to observe that the public opinion threat perception decreases as stereotypical associations increases.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "Results and Discussion", "sec_num": "4" }, { "text": "The yearly average bias scores for concepts related to poverty and prostitution are depicted in Figures 5 and 6 . For poverty related concepts, German, Italian, and British bias score values are negative for most years, meaning that poverty concepts are actually more associated with the Spanish ingroup when compared to such outgroups. The same is not true for Colombian, Ecuadorian, Moroccan, and Romanian outgroups. Again, in Figure 6 it is possible to observe that same division between outgroups. The descriptive analysis show that, overall, outgroups in the Lower PPP classification exhibit stronger association with concepts related to prostitution and poverty.", "cite_spans": [], "ref_spans": [ { "start": 96, "end": 111, "text": "Figures 5 and 6", "ref_id": "FIGREF4" }, { "start": 429, "end": 437, "text": "Figure 6", "ref_id": null } ], "eq_spans": [], "section": "Results and Discussion", "sec_num": "4" }, { "text": "The Table 2 shows the results of the Random Effects model for the aforementioned bias series. Consistently, for the two dependent series a strong effect regarding the Lower PPP variable can be observed meaning that again the British, the German, and the Italian are appreciably less associated with poverty and prostitution concepts than the Colombian, the Ecuadorian, the Moroccan, and the Romanian outgroups.", "cite_spans": [], "ref_spans": [ { "start": 4, "end": 11, "text": "Table 2", "ref_id": "TABREF3" } ], "eq_spans": [], "section": "Results and Discussion", "sec_num": "4" }, { "text": "Concerning time effects, only the years 2009 and 2011 affect significantly the poverty series, while the year trend is not significant for the prostitution stereotypical associations. Comparably to the findings described for the crimes and drugs concepts, the Unemployment benefits predictor has a significant involvement with the dependent series, indicating discrepancy between lower and higher PPP groups. Aside from the interaction with the unemployment benefits predictor, which has the same pattern described above for the crimes and drugs series, no other predictor interacts significantly with the Lower PPP group.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "Results and Discussion", "sec_num": "4" }, { "text": "The strong effect of the Lower PPP predictor on our analysis that news discourse emphasises the ethnicity of certain outgroups more than others. Furthermore, the interpretation of main effects and interactions with sociopolitical variables indicates that stereotypical portrayals seem to be dissociated from real demographic trends. Discourse is one of the everyday social practices that may be used for discriminatory purposes, for instance in intra-group discourse about resident minorities or immigrants frame these \"others\" negatively, thus leading to the reproduction of ethnic prejudices or ideologies (Van Dijk, 2000) . Our findings go in line with frames described in other studies made with European newspapers, which indicate the semantic link between foreigners, prostitution, criminality and degeneracy (Neyland, 2019; Stenvoll, 2002; Light and Young, 2009; Igartua et al., 2005; Rancu, 2011) , especially for Eastern European and Latin American backgrounds. We join previous studies pointing that media coverage can be stereotypical, associating ethnic outgroups with stigmatized attributes, and therefore having serious negative effects both on individuals and society, as news are powerful sources of the discursive demoralization of marginalised groups (Hamborg et al., 2018; Zilber and Niven, 2000; Angermeyer and Schulze, 2001; Sui and Paul, 2017; Kroon et al., 2020; Farris and Silber Mohamed, 2018; Milioni et al., 2015; Abrajano et al., 2017; Saiz de Lobado Garc\u00eda et al., 2018; Neyland, 2019) .", "cite_spans": [ { "start": 608, "end": 624, "text": "(Van Dijk, 2000)", "ref_id": "BIBREF44" }, { "start": 815, "end": 830, "text": "(Neyland, 2019;", "ref_id": "BIBREF32" }, { "start": 831, "end": 846, "text": "Stenvoll, 2002;", "ref_id": "BIBREF40" }, { "start": 847, "end": 869, "text": "Light and Young, 2009;", "ref_id": "BIBREF24" }, { "start": 870, "end": 891, "text": "Igartua et al., 2005;", "ref_id": "BIBREF19" }, { "start": 892, "end": 904, "text": "Rancu, 2011)", "ref_id": "BIBREF36" }, { "start": 1269, "end": 1291, "text": "(Hamborg et al., 2018;", "ref_id": "BIBREF16" }, { "start": 1292, "end": 1315, "text": "Zilber and Niven, 2000;", "ref_id": "BIBREF50" }, { "start": 1316, "end": 1345, "text": "Angermeyer and Schulze, 2001;", "ref_id": "BIBREF1" }, { "start": 1346, "end": 1365, "text": "Sui and Paul, 2017;", "ref_id": "BIBREF41" }, { "start": 1366, "end": 1385, "text": "Kroon et al., 2020;", "ref_id": "BIBREF21" }, { "start": 1386, "end": 1418, "text": "Farris and Silber Mohamed, 2018;", "ref_id": "BIBREF12" }, { "start": 1419, "end": 1440, "text": "Milioni et al., 2015;", "ref_id": "BIBREF30" }, { "start": 1441, "end": 1463, "text": "Abrajano et al., 2017;", "ref_id": "BIBREF0" }, { "start": 1464, "end": 1499, "text": "Saiz de Lobado Garc\u00eda et al., 2018;", "ref_id": "BIBREF38" }, { "start": 1500, "end": 1514, "text": "Neyland, 2019)", "ref_id": "BIBREF32" } ], "ref_spans": [], "eq_spans": [], "section": "Results and Discussion", "sec_num": "4" }, { "text": "We cite the following limitations of our findings. The present analysis considers only one data source, therefore our conclusions cannot be generalized to other Spanish media outlets. Although the unavailability of other diachronic corpora for Spanish from Spain limits our conclusion to a single news outlet, we argue that this study is a valuable contribution to stereotype analysis in media discourse using a non-English target language.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "Results and Discussion", "sec_num": "4" }, { "text": "Further, we acknowledge that by excluding gender inflected words, stereotypes about women that could be informative were left out. We do wish to explore gender inflected words in future work with a more suitable dataset. Lastly, we would like to point that all these nationalities have intricate and deep political relationships with Spain which certainly go beyond having a higher or lower GPD per capita.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "Results and Discussion", "sec_num": "4" }, { "text": "In this work we analyzed the dynamics of stereotypical associations concerning seven of the most prominent ethnic outgroups living in Spain using language models trained with 12 years of news items from the Spanish newspaper 20 Minutos. We investigated biases concerning concepts related to crimes, drugs poverty and prostitution, exploring the relation between the stereotypical associations and the GPD per capita (PPP) of the outgroups' countries of origin, public opinion, outgroup size, unemployment subsidy, and number of committed offenses in the Spanish territory.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "Conclusion", "sec_num": "5" }, { "text": "Our results show that the texts exhibit stereotypical associations, especially for the Colombian, Ecuadorian, Moroccan and Romanian outgroups. We conclude that the examined news articles emphasize the nationality of certain ethnicities, which hinder the integration process of already marginalized outgroups. Moreover, these associations can be further propagated and amplified through computational algorithms if available data indiscriminately (Bolukbasi et al., 2016b; Nadeem et al., 2020) , leading to concerning outcomes.", "cite_spans": [ { "start": 446, "end": 471, "text": "(Bolukbasi et al., 2016b;", "ref_id": "BIBREF7" }, { "start": 472, "end": 492, "text": "Nadeem et al., 2020)", "ref_id": "BIBREF31" } ], "ref_spans": [], "eq_spans": [], "section": "Conclusion", "sec_num": "5" }, { "text": "As future work, we aim to move to a multilingual perspective and compare outgroup stereotypes across different languages. Furthermore, we wish to examine stereotypes in political discourse, to inspect if patterns similar to the ones found in this work can be observed.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "Conclusion", "sec_num": "5" }, { "text": "\"National institute of Statistics\" https://www.ine.es/", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "", "sec_num": null }, { "text": "The PPP of the Italian outgroup for the 2007-2018 period is only slightly higher while it is considerably higher for the British and German nationalities", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "", "sec_num": null }, { "text": "https://www.20minutos.es/", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "", "sec_num": null }, { "text": "\"Estad\u00edstica del Padr\u00f3n continuo. Poblaci\u00f3n extranjera por Nacionalidad, provincias, Sexo y A\u00f1o\"6 \"Tasa de paro por nacionalidad y periodo\" 7 \"Estad\u00edstica de condenados: Adultos. Condenados seg\u00fan n\u00famero de delitos, nacionalidad y sexo\"", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "", "sec_num": null }, { "text": "Excluding the words related to the police, terrorism and lack of intelligence, which do not suit the purposes of this work.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "", "sec_num": null }, { "text": "Series named \"GDP per capita, PPP (current international $)\" available in the World Development Indicators series.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "", "sec_num": null }, { "text": "https://fasttext.cc/docs/en/options.html", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "", "sec_num": null } ], "back_matter": [ { "text": "This work was partially supported by the Universitat Pompeu Fabra (UPF), through a PhD studentship grant and the use of Marvin Cluster 10 to train and analyse the language models.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "Acknowledgments", "sec_num": null }, { "text": "In the next subsections we specify the word lists that were used to represent crimes, drugs, poverty and prostitution concepts, as well as the ingroup and outgroups. Please notice that some of the words in the lists are plural inflections that have no corresponding translation in English. We identify such words by adding '(plural)' next to the singular translation.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "A Word lists", "sec_num": null }, { "text": "Ingroup in Spanish: Espa\u00f1ol, Espa\u00f1oles'. Ingroup translation: \"Spanish\", \"Spanish (plural)\". British outgroup in Spanish: Brit\u00e1nico, Brit\u00e1nicos. British outgroup translation: \"British\", \"British (plural)\". Colombian outgroup in Spanish: Colombiano, Colombianos. Colombian outgroup translation: \"Colombian\", \"Colombians\". Ecuadorian outgroup in Spanish: Ecuatoriano, Ecuatorianos. Ecuadorian outgroup translation: \"Ecuadorian\", \"Ecuadorians\". German outgroup in Spanish: Alem\u00e1n, Alemanes. German outgroup translation: \"German\", \"Germans\". Italian outgroup in Spanish: Italiano, Italianos. Italian outgroup translation: \"Italian\", \"Italians\". Moroccan outgroup in Spanish: Marroqu\u00ed, Marroqu\u00edes. Moroccan outgroup translation: \"Moroccan\", \"Moroccans\". Romanian outgroup in Spanish: Rumano, Rumanos. Romanian outgroup translation: \"Romanian\", \"Romanians\".", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "A.1 Ingroup and outgroups", "sec_num": null }, { "text": "The table 3 shows the frequencies by year of the words that were used to create the ingroup and outgroup vector representations in our study.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "A.2 Frequency of Ingroup and outgroup words", "sec_num": null }, { "text": "In the following subsections we show the hyperparameters used to train the word embedding models and the yearly scores of the RG \u2212 65 and M C \u2212 30 semantic similarity benchmarks.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "B Word Embeddings", "sec_num": null }, { "text": "All Fasttext skipgram models were trained with 250 dimensions, five epochs and minimum word frequency of 15 occurrences. The hyper-parameters selected by the grid-search are shown below in the ", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "B.1 Hyper-parameters", "sec_num": null }, { "text": "The Table 5 shows the Pearson coefficients and pvalues for the RG\u221265 and M C \u221230 Spanish word similarity scores, for each of the yearly trained embedding models.", "cite_spans": [], "ref_spans": [ { "start": 4, "end": 11, "text": "Table 5", "ref_id": null } ], "eq_spans": [], "section": "B.2 Semantic similarity evaluation", "sec_num": null } ], "bib_entries": { "BIBREF0": { "ref_id": "b0", "title": "Media framing and partisan identity: The case of immigration coverage and white macropartisanship", "authors": [ { "first": "Zoltan", "middle": [], "last": "Marisa A Abrajano", "suffix": "" }, { "first": "", "middle": [], "last": "Hajnal", "suffix": "" }, { "first": "", "middle": [], "last": "Hans", "suffix": "" }, { "first": "", "middle": [], "last": "Has", "suffix": "" } ], "year": 2017, "venue": "Ethnicity and Politics", "volume": "2", "issue": "1", "pages": "", "other_ids": {}, "num": null, "urls": [], "raw_text": "Marisa A Abrajano, Zoltan Hajnal, and Hans JG Has- sell. 2017. Media framing and partisan identity: The case of immigration coverage and white macropar- tisanship. Journal of Race, Ethnicity and Politics, 2(1):5.", "links": null }, "BIBREF1": { "ref_id": "b1", "title": "Reinforcing stereotypes: how the focus on forensic cases in news reporting may influence public attitudes towards the mentally ill", "authors": [ { "first": "C", "middle": [], "last": "Matthias", "suffix": "" }, { "first": "Beate", "middle": [], "last": "Angermeyer", "suffix": "" }, { "first": "", "middle": [], "last": "Schulze", "suffix": "" } ], "year": 2001, "venue": "International Journal of Law and Psychiatry", "volume": "", "issue": "", "pages": "", "other_ids": {}, "num": null, "urls": [], "raw_text": "Matthias C Angermeyer and Beate Schulze. 2001. Re- inforcing stereotypes: how the focus on forensic cases in news reporting may influence public atti- tudes towards the mentally ill. International Journal of Law and Psychiatry.", "links": null }, "BIBREF2": { "ref_id": "b2", "title": "Gender: Stereotypes and roles", "authors": [ { "first": "A", "middle": [], "last": "Susan", "suffix": "" }, { "first": "", "middle": [], "last": "Basow", "suffix": "" } ], "year": 1992, "venue": "", "volume": "", "issue": "", "pages": "", "other_ids": {}, "num": null, "urls": [], "raw_text": "Susan A Basow. 1992. Gender: Stereotypes and roles. Thomson Brooks/Cole Publishing Co.", "links": null }, "BIBREF3": { "ref_id": "b3", "title": "Fixed and random effects models: making an informed choice", "authors": [ { "first": "Andrew", "middle": [], "last": "Bell", "suffix": "" }, { "first": "Malcolm", "middle": [], "last": "Fairbrother", "suffix": "" }, { "first": "Kelvyn", "middle": [], "last": "Jones", "suffix": "" } ], "year": 2019, "venue": "Quality & Quantity", "volume": "53", "issue": "2", "pages": "1051--1074", "other_ids": {}, "num": null, "urls": [], "raw_text": "Andrew Bell, Malcolm Fairbrother, and Kelvyn Jones. 2019. Fixed and random effects models: making an informed choice. Quality & Quantity, 53(2):1051- 1074.", "links": null }, "BIBREF4": { "ref_id": "b4", "title": "On the dangers of stochastic parrots: Can language models be too big?", "authors": [ { "first": "M", "middle": [], "last": "Emily", "suffix": "" }, { "first": "Timnit", "middle": [], "last": "Bender", "suffix": "" }, { "first": "Angelina", "middle": [], "last": "Gebru", "suffix": "" }, { "first": "Shmargaret", "middle": [], "last": "Mcmillan-Major", "suffix": "" }, { "first": "", "middle": [], "last": "Shmitchell", "suffix": "" } ], "year": 2021, "venue": "Proceedings of the 2021 ACM Conference on Fairness, Accountability, and Transparency", "volume": "", "issue": "", "pages": "610--623", "other_ids": {}, "num": null, "urls": [], "raw_text": "Emily M Bender, Timnit Gebru, Angelina McMillan- Major, and Shmargaret Shmitchell. 2021. On the dangers of stochastic parrots: Can language models be too big? In Proceedings of the 2021 ACM Confer- ence on Fairness, Accountability, and Transparency, pages 610-623.", "links": null }, "BIBREF5": { "ref_id": "b5", "title": "Enriching word vectors with subword information", "authors": [ { "first": "Piotr", "middle": [], "last": "Bojanowski", "suffix": "" }, { "first": "Edouard", "middle": [], "last": "Grave", "suffix": "" }, { "first": "Armand", "middle": [], "last": "Joulin", "suffix": "" }, { "first": "Tomas", "middle": [], "last": "Mikolov", "suffix": "" } ], "year": 2017, "venue": "Transactions of the Association for Computational Linguistics", "volume": "5", "issue": "", "pages": "135--146", "other_ids": {}, "num": null, "urls": [], "raw_text": "Piotr Bojanowski, Edouard Grave, Armand Joulin, and Tomas Mikolov. 2017. Enriching word vectors with subword information. Transactions of the Associa- tion for Computational Linguistics, 5:135-146.", "links": null }, "BIBREF6": { "ref_id": "b6", "title": "Quantifying and reducing stereotypes in word embeddings", "authors": [ { "first": "Tolga", "middle": [], "last": "Bolukbasi", "suffix": "" }, { "first": "Kai-Wei", "middle": [], "last": "Chang", "suffix": "" }, { "first": "James", "middle": [], "last": "Zou", "suffix": "" }, { "first": "Venkatesh", "middle": [], "last": "Saligrama", "suffix": "" }, { "first": "Adam", "middle": [], "last": "Kalai", "suffix": "" } ], "year": 2016, "venue": "", "volume": "", "issue": "", "pages": "", "other_ids": { "arXiv": [ "arXiv:1606.06121" ] }, "num": null, "urls": [], "raw_text": "Tolga Bolukbasi, Kai-Wei Chang, James Zou, Venkatesh Saligrama, and Adam Kalai. 2016a. Quantifying and reducing stereotypes in word embeddings. arXiv preprint arXiv:1606.06121.", "links": null }, "BIBREF7": { "ref_id": "b7", "title": "Man is to computer programmer as woman is to homemaker? debiasing word embeddings", "authors": [ { "first": "Tolga", "middle": [], "last": "Bolukbasi", "suffix": "" }, { "first": "Kai-Wei", "middle": [], "last": "Chang", "suffix": "" }, { "first": "Y", "middle": [], "last": "James", "suffix": "" }, { "first": "Venkatesh", "middle": [], "last": "Zou", "suffix": "" }, { "first": "Adam", "middle": [ "T" ], "last": "Saligrama", "suffix": "" }, { "first": "", "middle": [], "last": "Kalai", "suffix": "" } ], "year": 2016, "venue": "Advances in neural information processing systems", "volume": "", "issue": "", "pages": "4349--4357", "other_ids": {}, "num": null, "urls": [], "raw_text": "Tolga Bolukbasi, Kai-Wei Chang, James Y Zou, Venkatesh Saligrama, and Adam T Kalai. 2016b. Man is to computer programmer as woman is to homemaker? debiasing word embeddings. In Ad- vances in neural information processing systems, pages 4349-4357.", "links": null }, "BIBREF8": { "ref_id": "b8", "title": "i am not a racist but", "authors": [ { "first": "Eduardo", "middle": [], "last": "Bonilla", "suffix": "" }, { "first": "-", "middle": [], "last": "Silva", "suffix": "" }, { "first": "Tyrone", "middle": [ "A" ], "last": "Forman", "suffix": "" } ], "year": 2000, "venue": "Mapping white college students' racial ideology in the usa. Discourse & society", "volume": "11", "issue": "", "pages": "50--85", "other_ids": {}, "num": null, "urls": [], "raw_text": "Eduardo Bonilla-Silva and Tyrone A Forman. 2000. \"i am not a racist but...\": Mapping white college stu- dents' racial ideology in the usa. Discourse & soci- ety, 11(1):50-85.", "links": null }, "BIBREF9": { "ref_id": "b9", "title": "Using unethical data to build a more ethical world", "authors": [ { "first": "Jamie", "middle": [], "last": "Brandon", "suffix": "" } ], "year": 2021, "venue": "AI and Ethics", "volume": "1", "issue": "2", "pages": "101--108", "other_ids": {}, "num": null, "urls": [], "raw_text": "Jamie Brandon. 2021. Using unethical data to build a more ethical world. AI and Ethics, 1(2):101-108.", "links": null }, "BIBREF10": { "ref_id": "b10", "title": "A framework for the construction of monolingual and cross-lingual word similarity datasets", "authors": [ { "first": "Jos\u00e9", "middle": [], "last": "Camacho-Collados", "suffix": "" }, { "first": "Mohammad", "middle": [], "last": "Taher Pilehvar", "suffix": "" }, { "first": "Roberto", "middle": [], "last": "Navigli", "suffix": "" } ], "year": 2015, "venue": "Proceedings of ACL", "volume": "", "issue": "", "pages": "1--7", "other_ids": {}, "num": null, "urls": [], "raw_text": "Jos\u00e9 Camacho-Collados, Mohammad Taher Pilehvar, and Roberto Navigli. 2015. A framework for the construction of monolingual and cross-lingual word similarity datasets. In Proceedings of ACL (2), pages 1-7.", "links": null }, "BIBREF11": { "ref_id": "b11", "title": "El reto de medir el sesgo ideol\u00f3gico en los medios escritos digitales", "authors": [ { "first": "Anas", "middle": [], "last": "Cardenal", "suffix": "" }, { "first": "Carol", "middle": [], "last": "Galais", "suffix": "" }, { "first": "Joaquim", "middle": [], "last": "Mor\u00e9", "suffix": "" } ], "year": 2018, "venue": "", "volume": "", "issue": "", "pages": "", "other_ids": {}, "num": null, "urls": [], "raw_text": "Anas Cardenal, Carol Galais, and Joaquim Mor\u00e9. 2018. El reto de medir el sesgo ideol\u00f3gico en los medios escritos digitales. quaderns del cac.", "links": null }, "BIBREF12": { "ref_id": "b12", "title": "Picturing immigration: How the media criminalizes immigrants", "authors": [ { "first": "M", "middle": [], "last": "Emily", "suffix": "" }, { "first": "Heather", "middle": [], "last": "Farris", "suffix": "" }, { "first": "Mohamed", "middle": [], "last": "Silber", "suffix": "" } ], "year": 2018, "venue": "Politics, Groups, and Identities", "volume": "6", "issue": "4", "pages": "814--824", "other_ids": {}, "num": null, "urls": [], "raw_text": "Emily M Farris and Heather Silber Mohamed. 2018. Picturing immigration: How the media criminal- izes immigrants. Politics, Groups, and Identities, 6(4):814-824.", "links": null }, "BIBREF13": { "ref_id": "b13", "title": "Word embeddings quantify 100 years of gender and ethnic stereotypes. Proceedings of the National Academy of", "authors": [ { "first": "Nikhil", "middle": [], "last": "Garg", "suffix": "" }, { "first": "Londa", "middle": [], "last": "Schiebinger", "suffix": "" }, { "first": "Dan", "middle": [], "last": "Jurafsky", "suffix": "" }, { "first": "James", "middle": [], "last": "Zou", "suffix": "" } ], "year": 2018, "venue": "Sciences", "volume": "115", "issue": "16", "pages": "3635--3644", "other_ids": {}, "num": null, "urls": [], "raw_text": "Nikhil Garg, Londa Schiebinger, Dan Jurafsky, and James Zou. 2018. Word embeddings quantify 100 years of gender and ethnic stereotypes. Pro- ceedings of the National Academy of Sciences, 115(16):E3635-E3644.", "links": null }, "BIBREF14": { "ref_id": "b14", "title": "Data analysis using regression and multilevel/hierarchical models", "authors": [ { "first": "Andrew", "middle": [], "last": "Gelman", "suffix": "" }, { "first": "Jennifer", "middle": [], "last": "Hill", "suffix": "" } ], "year": 2006, "venue": "", "volume": "", "issue": "", "pages": "", "other_ids": {}, "num": null, "urls": [], "raw_text": "Andrew Gelman and Jennifer Hill. 2006. Data analysis using regression and multilevel/hierarchical models. Cambridge university press.", "links": null }, "BIBREF15": { "ref_id": "b15", "title": "Lipstick on a pig: Debiasing methods cover up systematic gender biases in word embeddings but do not remove them", "authors": [ { "first": "Hila", "middle": [], "last": "Gonen", "suffix": "" }, { "first": "Yoav", "middle": [], "last": "Goldberg", "suffix": "" } ], "year": 2019, "venue": "", "volume": "", "issue": "", "pages": "", "other_ids": { "arXiv": [ "arXiv:1903.03862" ] }, "num": null, "urls": [], "raw_text": "Hila Gonen and Yoav Goldberg. 2019. Lipstick on a pig: Debiasing methods cover up systematic gender biases in word embeddings but do not remove them. arXiv preprint arXiv:1903.03862.", "links": null }, "BIBREF16": { "ref_id": "b16", "title": "Bias-aware news analysis using matrix-based news aggregation", "authors": [ { "first": "Felix", "middle": [], "last": "Hamborg", "suffix": "" }, { "first": "Norman", "middle": [], "last": "Meuschke", "suffix": "" }, { "first": "Bela", "middle": [], "last": "Gipp", "suffix": "" } ], "year": 2018, "venue": "International Journal on Digital Libraries", "volume": "", "issue": "", "pages": "1--19", "other_ids": {}, "num": null, "urls": [], "raw_text": "Felix Hamborg, Norman Meuschke, and Bela Gipp. 2018. Bias-aware news analysis using matrix-based news aggregation. International Journal on Digital Libraries, pages 1-19.", "links": null }, "BIBREF17": { "ref_id": "b17", "title": "Diachronic word embeddings reveal statistical laws of semantic change", "authors": [ { "first": "Jure", "middle": [], "last": "William L Hamilton", "suffix": "" }, { "first": "Dan", "middle": [], "last": "Leskovec", "suffix": "" }, { "first": "", "middle": [], "last": "Jurafsky", "suffix": "" } ], "year": 2016, "venue": "", "volume": "", "issue": "", "pages": "", "other_ids": { "arXiv": [ "arXiv:1605.09096" ] }, "num": null, "urls": [], "raw_text": "William L Hamilton, Jure Leskovec, and Dan Juraf- sky. 2016. Diachronic word embeddings reveal sta- tistical laws of semantic change. arXiv preprint arXiv:1605.09096.", "links": null }, "BIBREF18": { "ref_id": "b18", "title": "Cross-lingual semantic relatedness using encyclopedic knowledge", "authors": [ { "first": "Samer", "middle": [], "last": "Hassan", "suffix": "" }, { "first": "Rada", "middle": [], "last": "Mihalcea", "suffix": "" } ], "year": 2009, "venue": "Proceedings of the 2009 Conference on Empirical Methods in Natural Language Processing", "volume": "", "issue": "", "pages": "1192--1201", "other_ids": {}, "num": null, "urls": [], "raw_text": "Samer Hassan and Rada Mihalcea. 2009. Cross-lingual semantic relatedness using encyclopedic knowledge. In Proceedings of the 2009 Conference on Empiri- cal Methods in Natural Language Processing, pages 1192-1201.", "links": null }, "BIBREF19": { "ref_id": "b19", "title": "Framing latin america in the spanish press: A cooled down friendship between two fraternal lands", "authors": [ { "first": "Juan Jos\u00e9", "middle": [], "last": "Igartua", "suffix": "" }, { "first": "Lifen", "middle": [], "last": "Cheng", "suffix": "" }, { "first": "Carlos", "middle": [], "last": "Mu\u0148iz", "suffix": "" } ], "year": 2005, "venue": "", "volume": "", "issue": "", "pages": "", "other_ids": {}, "num": null, "urls": [], "raw_text": "Juan Jos\u00e9 Igartua, Lifen Cheng, and Carlos Mu\u0148iz. 2005. Framing latin america in the spanish press: A cooled down friendship between two fraternal lands.", "links": null }, "BIBREF20": { "ref_id": "b20", "title": "The geometry of culture: Analyzing the meanings of class through word embeddings", "authors": [ { "first": "C", "middle": [], "last": "Austin", "suffix": "" }, { "first": "Matt", "middle": [], "last": "Kozlowski", "suffix": "" }, { "first": "James A", "middle": [], "last": "Taddy", "suffix": "" }, { "first": "", "middle": [], "last": "Evans", "suffix": "" } ], "year": 2019, "venue": "American Sociological Review", "volume": "84", "issue": "5", "pages": "905--949", "other_ids": {}, "num": null, "urls": [], "raw_text": "Austin C Kozlowski, Matt Taddy, and James A Evans. 2019. The geometry of culture: Analyzing the mean- ings of class through word embeddings. American Sociological Review, 84(5):905-949.", "links": null }, "BIBREF21": { "ref_id": "b21", "title": "Guilty by association: Using word embeddings to measure ethnic stereotypes in news coverage", "authors": [ { "first": "Damian", "middle": [], "last": "Anne C Kroon", "suffix": "" }, { "first": "Tamara", "middle": [], "last": "Trilling", "suffix": "" }, { "first": "", "middle": [], "last": "Raats", "suffix": "" } ], "year": 2020, "venue": "Journalism & Mass Communication Quarterly", "volume": "", "issue": "", "pages": "", "other_ids": {}, "num": null, "urls": [], "raw_text": "Anne C Kroon, Damian Trilling, and Tamara Raats. 2020. Guilty by association: Using word embed- dings to measure ethnic stereotypes in news cover- age. Journalism & Mass Communication Quarterly, page 1077699020932304.", "links": null }, "BIBREF22": { "ref_id": "b22", "title": "Diachronic word embeddings and semantic shifts: a survey", "authors": [ { "first": "Andrey", "middle": [], "last": "Kutuzov", "suffix": "" }, { "first": "Lilja", "middle": [], "last": "\u00d8vrelid", "suffix": "" }, { "first": "Terrence", "middle": [], "last": "Szymanski", "suffix": "" }, { "first": "Erik", "middle": [], "last": "Velldal", "suffix": "" } ], "year": 2018, "venue": "", "volume": "", "issue": "", "pages": "", "other_ids": { "arXiv": [ "arXiv:1806.03537" ] }, "num": null, "urls": [], "raw_text": "Andrey Kutuzov, Lilja \u00d8vrelid, Terrence Szymanski, and Erik Velldal. 2018. Diachronic word embed- dings and semantic shifts: a survey. arXiv preprint arXiv:1806.03537.", "links": null }, "BIBREF23": { "ref_id": "b23", "title": "Araweat: Multidimensional analysis of biases in arabic word embeddings", "authors": [ { "first": "Anne", "middle": [], "last": "Lauscher", "suffix": "" }, { "first": "Rafik", "middle": [], "last": "Takieddin", "suffix": "" }, { "first": "Simone", "middle": [ "Paolo" ], "last": "Ponzetto", "suffix": "" }, { "first": "Goran", "middle": [], "last": "Glava\u0161", "suffix": "" } ], "year": 2020, "venue": "", "volume": "", "issue": "", "pages": "", "other_ids": { "arXiv": [ "arXiv:2011.01575" ] }, "num": null, "urls": [], "raw_text": "Anne Lauscher, Rafik Takieddin, Simone Paolo Ponzetto, and Goran Glava\u0161. 2020. Araweat: Mul- tidimensional analysis of biases in arabic word em- beddings. arXiv preprint arXiv:2011.01575.", "links": null }, "BIBREF24": { "ref_id": "b24", "title": "European union enlargement, post-accession migration and imaginative geographies of the 'new europe': Media discourses in romania and the united kingdom", "authors": [ { "first": "Duncan", "middle": [], "last": "Light", "suffix": "" }, { "first": "Craig", "middle": [], "last": "Young", "suffix": "" } ], "year": 2009, "venue": "Journal of Cultural Geography", "volume": "26", "issue": "3", "pages": "281--303", "other_ids": {}, "num": null, "urls": [], "raw_text": "Duncan Light and Craig Young. 2009. European union enlargement, post-accession migration and imagina- tive geographies of the 'new europe': Media dis- courses in romania and the united kingdom. Journal of Cultural Geography, 26(3):281-303.", "links": null }, "BIBREF25": { "ref_id": "b25", "title": "Immigration and perceived ethnic threat: Cultural capital and economic explanations", "authors": [ { "first": "Katerina", "middle": [], "last": "Manevska", "suffix": "" }, { "first": "Peter", "middle": [], "last": "Achterberg", "suffix": "" } ], "year": 2013, "venue": "European Sociological Review", "volume": "29", "issue": "3", "pages": "437--449", "other_ids": {}, "num": null, "urls": [], "raw_text": "Katerina Manevska and Peter Achterberg. 2013. Immi- gration and perceived ethnic threat: Cultural capital and economic explanations. European Sociological Review, 29(3):437-449.", "links": null }, "BIBREF26": { "ref_id": "b26", "title": "Shortterm semantic shifts and their relation to frequency change", "authors": [ { "first": "Anna", "middle": [], "last": "Marakasova", "suffix": "" }, { "first": "Julia", "middle": [], "last": "Neidhardt", "suffix": "" } ], "year": 2020, "venue": "Proceedings of the Probability and Meaning Conference", "volume": "", "issue": "", "pages": "146--153", "other_ids": {}, "num": null, "urls": [], "raw_text": "Anna Marakasova and Julia Neidhardt. 2020. Short- term semantic shifts and their relation to frequency change. In Proceedings of the Probability and Meaning Conference (PaM 2020), pages 146-153.", "links": null }, "BIBREF27": { "ref_id": "b27", "title": "Balancing type i error and power in linear mixed models", "authors": [ { "first": "Hannes", "middle": [], "last": "Matuschek", "suffix": "" }, { "first": "Reinhold", "middle": [], "last": "Kliegl", "suffix": "" }, { "first": "Shravan", "middle": [], "last": "Vasishth", "suffix": "" }, { "first": "Harald", "middle": [], "last": "Baayen", "suffix": "" }, { "first": "Douglas", "middle": [], "last": "Bates", "suffix": "" } ], "year": 2017, "venue": "Journal of memory and language", "volume": "94", "issue": "", "pages": "305--315", "other_ids": {}, "num": null, "urls": [], "raw_text": "Hannes Matuschek, Reinhold Kliegl, Shravan Vasishth, Harald Baayen, and Douglas Bates. 2017. Balanc- ing type i error and power in linear mixed models. Journal of memory and language, 94:305-315.", "links": null }, "BIBREF28": { "ref_id": "b28", "title": "A survey on bias and fairness in machine learning", "authors": [ { "first": "Ninareh", "middle": [], "last": "Mehrabi", "suffix": "" }, { "first": "Fred", "middle": [], "last": "Morstatter", "suffix": "" }, { "first": "Nripsuta", "middle": [], "last": "Saxena", "suffix": "" }, { "first": "Kristina", "middle": [], "last": "Lerman", "suffix": "" }, { "first": "Aram", "middle": [], "last": "Galstyan", "suffix": "" } ], "year": 2019, "venue": "", "volume": "", "issue": "", "pages": "", "other_ids": { "arXiv": [ "arXiv:1908.09635" ] }, "num": null, "urls": [], "raw_text": "Ninareh Mehrabi, Fred Morstatter, Nripsuta Saxena, Kristina Lerman, and Aram Galstyan. 2019. A sur- vey on bias and fairness in machine learning. arXiv preprint arXiv:1908.09635.", "links": null }, "BIBREF29": { "ref_id": "b29", "title": "Perceived economic threat and anti-immigration attitudes: Effects of immigrant group size and economic conditions revisited. Cross-cultural analysis: Methods and applications", "authors": [ { "first": "Bart", "middle": [], "last": "Meuleman", "suffix": "" } ], "year": 2011, "venue": "", "volume": "", "issue": "", "pages": "281--310", "other_ids": {}, "num": null, "urls": [], "raw_text": "Bart Meuleman. 2011. Perceived economic threat and anti-immigration attitudes: Effects of immi- grant group size and economic conditions revisited. Cross-cultural analysis: Methods and applications, pages 281-310.", "links": null }, "BIBREF30": { "ref_id": "b30", "title": "Framing immigration in online media and television news in crisis-stricken cyprus", "authors": [ { "first": "L", "middle": [], "last": "Dimitra", "suffix": "" }, { "first": "Lia-Paschalia", "middle": [], "last": "Milioni", "suffix": "" }, { "first": "Konstantinos", "middle": [], "last": "Spyridou", "suffix": "" }, { "first": "", "middle": [], "last": "Vadratsikas", "suffix": "" } ], "year": 2015, "venue": "The Cyprus Review", "volume": "27", "issue": "1", "pages": "155--185", "other_ids": {}, "num": null, "urls": [], "raw_text": "Dimitra L Milioni, Lia-Paschalia Spyridou, and Kon- stantinos Vadratsikas. 2015. Framing immigration in online media and television news in crisis-stricken cyprus. The Cyprus Review, 27(1):155-185.", "links": null }, "BIBREF31": { "ref_id": "b31", "title": "Stereoset: Measuring stereotypical bias in pretrained language models", "authors": [ { "first": "Moin", "middle": [], "last": "Nadeem", "suffix": "" }, { "first": "Anna", "middle": [], "last": "Bethke", "suffix": "" }, { "first": "Siva", "middle": [], "last": "Reddy", "suffix": "" } ], "year": 2020, "venue": "", "volume": "", "issue": "", "pages": "", "other_ids": { "arXiv": [ "arXiv:2004.09456" ] }, "num": null, "urls": [], "raw_text": "Moin Nadeem, Anna Bethke, and Siva Reddy. 2020. Stereoset: Measuring stereotypical bias in pretrained language models. arXiv preprint arXiv:2004.09456.", "links": null }, "BIBREF32": { "ref_id": "b32", "title": "The Sexual Other: Discursive constructions of migrant sex workers in New Zealand media", "authors": [ { "first": "F", "middle": [ "F" ], "last": "Matilda", "suffix": "" }, { "first": "", "middle": [], "last": "Neyland", "suffix": "" } ], "year": 2019, "venue": "", "volume": "", "issue": "", "pages": "", "other_ids": {}, "num": null, "urls": [], "raw_text": "Matilda FF Neyland. 2019. The Sexual Other: Dis- cursive constructions of migrant sex workers in New Zealand media. Ph.D. thesis, Victoria University of Wellington.", "links": null }, "BIBREF33": { "ref_id": "b33", "title": "Data file edition 1.0. NSD -Norwegian Centre for Research Data, Norway -Data Archive and distributor of ESS data for ESS ERIC", "authors": [], "year": 2020, "venue": "", "volume": "", "issue": "", "pages": "1--9", "other_ids": { "DOI": [ "10.21338/NSD-ESS-CUMULATIVE" ] }, "num": null, "urls": [], "raw_text": "NSD. 2020. European social survey cumulative file, ess 1-9 (2020). Data file edition 1.0. NSD -Nor- wegian Centre for Research Data, Norway -Data Archive and distributor of ESS data for ESS ERIC.", "links": null }, "BIBREF34": { "ref_id": "b34", "title": "Bias in word embeddings", "authors": [ { "first": "Orestis", "middle": [], "last": "Papakyriakopoulos", "suffix": "" }, { "first": "Simon", "middle": [], "last": "Hegelich", "suffix": "" }, { "first": "Juan Carlos Medina", "middle": [], "last": "Serrano", "suffix": "" }, { "first": "Fabienne", "middle": [], "last": "Marco", "suffix": "" } ], "year": 2020, "venue": "Proceedings of the 2020 Conference on Fairness, Accountability, and Transparency", "volume": "", "issue": "", "pages": "446--457", "other_ids": {}, "num": null, "urls": [], "raw_text": "Orestis Papakyriakopoulos, Simon Hegelich, Juan Car- los Medina Serrano, and Fabienne Marco. 2020. Bias in word embeddings. In Proceedings of the 2020 Conference on Fairness, Accountability, and Transparency, pages 446-457.", "links": null }, "BIBREF35": { "ref_id": "b35", "title": "Reducing gender bias in abusive language detection", "authors": [ { "first": "Ji", "middle": [], "last": "Ho Park", "suffix": "" }, { "first": "Jamin", "middle": [], "last": "Shin", "suffix": "" }, { "first": "Pascale", "middle": [], "last": "Fung", "suffix": "" } ], "year": 2018, "venue": "", "volume": "", "issue": "", "pages": "", "other_ids": { "arXiv": [ "arXiv:1808.07231" ] }, "num": null, "urls": [], "raw_text": "Ji Ho Park, Jamin Shin, and Pascale Fung. 2018. Re- ducing gender bias in abusive language detection. arXiv preprint arXiv:1808.07231.", "links": null }, "BIBREF36": { "ref_id": "b36", "title": "Exclusion, marginalization and prejudice: The image of the romanian woman in spanish society", "authors": [ { "first": "Raluca", "middle": [], "last": "Oana", "suffix": "" }, { "first": "Matu", "middle": [], "last": "Rancu", "suffix": "" } ], "year": 2011, "venue": "International Journal of Diversity in Organisations", "volume": "", "issue": "5", "pages": "", "other_ids": {}, "num": null, "urls": [], "raw_text": "Raluca Oana Matu Rancu. 2011. Exclusion, marginal- ization and prejudice: The image of the romanian woman in spanish society. International Journal of Diversity in Organisations, Communities & Nations, 10(5).", "links": null }, "BIBREF37": { "ref_id": "b37", "title": "Creaci\u00f3n de un corpus de noticias de gran tamano en espanol para el an\u00e1lisis diacr\u00f3nico y diat\u00f3pico del uso del lenguaje", "authors": [ { "first": "Pavel", "middle": [], "last": "Razgovorov", "suffix": "" }, { "first": "David", "middle": [], "last": "Tom\u00e1s", "suffix": "" } ], "year": 2019, "venue": "", "volume": "", "issue": "", "pages": "", "other_ids": {}, "num": null, "urls": [], "raw_text": "Pavel Razgovorov, David Tom\u00e1s, et al. 2019. Creaci\u00f3n de un corpus de noticias de gran tamano en espanol para el an\u00e1lisis diacr\u00f3nico y diat\u00f3pico del uso del lenguaje.", "links": null }, "BIBREF38": { "ref_id": "b38", "title": "Met\u00e1fora y percepci\u00f3n: an\u00e1lisis de la ideolog\u00eda subyacente en el discurso jur\u00eddico sobre inmigraci\u00f3n", "authors": [ { "first": "Mar\u00eda", "middle": [], "last": "Ester Saiz De", "suffix": "" }, { "first": "Lobado", "middle": [], "last": "Garc\u00eda", "suffix": "" } ], "year": 2018, "venue": "", "volume": "", "issue": "", "pages": "", "other_ids": {}, "num": null, "urls": [], "raw_text": "Mar\u00eda Ester Saiz de Lobado Garc\u00eda et al. 2018. Met\u00e1- fora y percepci\u00f3n: an\u00e1lisis de la ideolog\u00eda subya- cente en el discurso jur\u00eddico sobre inmigraci\u00f3n.", "links": null }, "BIBREF39": { "ref_id": "b39", "title": "How do you speak about immigrants? taxonomy and stereoimmigrants dataset for identifying stereotypes about immigrants", "authors": [ { "first": "Javier", "middle": [], "last": "S\u00e1nchez-Junquera", "suffix": "" }, { "first": "Berta", "middle": [], "last": "Chulvi", "suffix": "" }, { "first": "Paolo", "middle": [], "last": "Rosso", "suffix": "" }, { "first": "Simone", "middle": [ "Paolo" ], "last": "Ponzetto", "suffix": "" } ], "year": 2021, "venue": "Applied Sciences", "volume": "11", "issue": "8", "pages": "", "other_ids": {}, "num": null, "urls": [], "raw_text": "Javier S\u00e1nchez-Junquera, Berta Chulvi, Paolo Rosso, and Simone Paolo Ponzetto. 2021. How do you speak about immigrants? taxonomy and stereoimmi- grants dataset for identifying stereotypes about im- migrants. Applied Sciences, 11(8):3610.", "links": null }, "BIBREF40": { "ref_id": "b40", "title": "From russia with love? newspaper coverage of cross-border prostitution in northern norway", "authors": [ { "first": "Dag", "middle": [], "last": "Stenvoll", "suffix": "" } ], "year": 1990, "venue": "European Journal of Women's Studies", "volume": "9", "issue": "2", "pages": "143--162", "other_ids": {}, "num": null, "urls": [], "raw_text": "Dag Stenvoll. 2002. From russia with love? newspaper coverage of cross-border prostitution in northern nor- way, 1990-2001. European Journal of Women's Studies, 9(2):143-162.", "links": null }, "BIBREF41": { "ref_id": "b41", "title": "Latino portrayals in local news media: Underrepresentation, negative stereotypes, and institutional predictors of coverage", "authors": [ { "first": "Mingxiao", "middle": [], "last": "Sui", "suffix": "" }, { "first": "Newly", "middle": [], "last": "Paul", "suffix": "" } ], "year": 2017, "venue": "Journal of Intercultural Communication Research", "volume": "46", "issue": "3", "pages": "273--294", "other_ids": {}, "num": null, "urls": [], "raw_text": "Mingxiao Sui and Newly Paul. 2017. Latino portray- als in local news media: Underrepresentation, nega- tive stereotypes, and institutional predictors of cov- erage. Journal of Intercultural Communication Re- search, 46(3):273-294.", "links": null }, "BIBREF42": { "ref_id": "b42", "title": "Content of stereotypes and the inference of similarity between members of stereotyped groups", "authors": [ { "first": "Henri", "middle": [], "last": "Tajfel", "suffix": "" }, { "first": "A", "middle": [], "last": "Anees", "suffix": "" }, { "first": "Robert", "middle": [ "Charles" ], "last": "Sheikh", "suffix": "" }, { "first": "", "middle": [], "last": "Gardner", "suffix": "" } ], "year": 1964, "venue": "Acta Psychologica", "volume": "", "issue": "", "pages": "", "other_ids": {}, "num": null, "urls": [], "raw_text": "Henri Tajfel, Anees A Sheikh, and Robert Charles Gardner. 1964. Content of stereotypes and the infer- ence of similarity between members of stereotyped groups. Acta Psychologica.", "links": null }, "BIBREF43": { "ref_id": "b43", "title": "Tracing antisemitic language through diachronic embedding projections", "authors": [ { "first": "Rocco", "middle": [], "last": "Tripodi", "suffix": "" }, { "first": "Massimo", "middle": [], "last": "Warglien", "suffix": "" } ], "year": 2019, "venue": "", "volume": "", "issue": "", "pages": "1789--1914", "other_ids": { "arXiv": [ "arXiv:1906.01440" ] }, "num": null, "urls": [], "raw_text": "Rocco Tripodi, Massimo Warglien, Simon Levis Sul- lam, and Deborah Paci. 2019. Tracing anti- semitic language through diachronic embedding projections: France 1789-1914. arXiv preprint arXiv:1906.01440.", "links": null }, "BIBREF44": { "ref_id": "b44", "title": "On the analysis of parliamentary debates on immigration", "authors": [ { "first": "", "middle": [], "last": "Teun A Van Dijk", "suffix": "" } ], "year": 2000, "venue": "", "volume": "", "issue": "", "pages": "", "other_ids": {}, "num": null, "urls": [], "raw_text": "Teun A Van Dijk. 2000. On the analysis of parliamen- tary debates on immigration. Citeseer.", "links": null }, "BIBREF45": { "ref_id": "b45", "title": "The social construction of the criminal alien in immigration law, enforcement practice and statistical enumeration: Consequences for immigrant stereotyping", "authors": [ { "first": "Judith", "middle": [ "Ann" ], "last": "Warner", "suffix": "" } ], "year": 2005, "venue": "Journal of Social and Ecological Boundaries", "volume": "1", "issue": "2", "pages": "56--80", "other_ids": {}, "num": null, "urls": [], "raw_text": "Judith Ann Warner. 2005. The social construction of the criminal alien in immigration law, enforcement practice and statistical enumeration: Consequences for immigrant stereotyping. Journal of Social and Ecological Boundaries, 1(2):56-80.", "links": null }, "BIBREF46": { "ref_id": "b46", "title": "Mapping the language of racism: Discourse and the legitimation of exploitation", "authors": [ { "first": "Margaret", "middle": [], "last": "Wetherell", "suffix": "" }, { "first": "Jonathan", "middle": [], "last": "Potter", "suffix": "" } ], "year": 1993, "venue": "", "volume": "", "issue": "", "pages": "", "other_ids": {}, "num": null, "urls": [], "raw_text": "Margaret Wetherell and Jonathan Potter. 1993. Map- ping the language of racism: Discourse and the legit- imation of exploitation. Columbia University Press.", "links": null }, "BIBREF47": { "ref_id": "b47", "title": "Using word embeddings to examine gender bias in dutch newspapers", "authors": [ { "first": "Melvin", "middle": [], "last": "Wevers", "suffix": "" } ], "year": 2019, "venue": "", "volume": "", "issue": "", "pages": "1950--1990", "other_ids": { "arXiv": [ "arXiv:1907.08922" ] }, "num": null, "urls": [], "raw_text": "Melvin Wevers. 2019. Using word embeddings to ex- amine gender bias in dutch newspapers, 1950-1990. arXiv preprint arXiv:1907.08922.", "links": null }, "BIBREF48": { "ref_id": "b48", "title": "Learning gender-neutral word embeddings", "authors": [ { "first": "Jieyu", "middle": [], "last": "Zhao", "suffix": "" }, { "first": "Yichao", "middle": [], "last": "Zhou", "suffix": "" }, { "first": "Zeyu", "middle": [], "last": "Li", "suffix": "" }, { "first": "Wei", "middle": [], "last": "Wang", "suffix": "" }, { "first": "Kai-Wei", "middle": [], "last": "Chang", "suffix": "" } ], "year": 2018, "venue": "", "volume": "", "issue": "", "pages": "", "other_ids": { "arXiv": [ "arXiv:1809.01496" ] }, "num": null, "urls": [], "raw_text": "Jieyu Zhao, Yichao Zhou, Zeyu Li, Wei Wang, and Kai- Wei Chang. 2018. Learning gender-neutral word embeddings. arXiv preprint arXiv:1809.01496.", "links": null }, "BIBREF49": { "ref_id": "b49", "title": "Examining gender bias in languages with grammatical gender", "authors": [ { "first": "Pei", "middle": [], "last": "Zhou", "suffix": "" }, { "first": "Weijia", "middle": [], "last": "Shi", "suffix": "" }, { "first": "Jieyu", "middle": [], "last": "Zhao", "suffix": "" }, { "first": "Kuan-Hao", "middle": [], "last": "Huang", "suffix": "" }, { "first": "Muhao", "middle": [], "last": "Chen", "suffix": "" }, { "first": "Ryan", "middle": [], "last": "Cotterell", "suffix": "" }, { "first": "Kai-Wei", "middle": [], "last": "Chang", "suffix": "" } ], "year": 2019, "venue": "", "volume": "", "issue": "", "pages": "", "other_ids": { "arXiv": [ "arXiv:1909.02224" ] }, "num": null, "urls": [], "raw_text": "Pei Zhou, Weijia Shi, Jieyu Zhao, Kuan-Hao Huang, Muhao Chen, Ryan Cotterell, and Kai- Wei Chang. 2019. Examining gender bias in lan- guages with grammatical gender. arXiv preprint arXiv:1909.02224.", "links": null }, "BIBREF50": { "ref_id": "b50", "title": "Stereotypes in the news: Media coverage of african-americans in congress", "authors": [ { "first": "Jeremy", "middle": [], "last": "Zilber", "suffix": "" }, { "first": "David", "middle": [], "last": "Niven", "suffix": "" } ], "year": 2000, "venue": "Harvard International Journal of Press/Politics", "volume": "5", "issue": "1", "pages": "32--49", "other_ids": {}, "num": null, "urls": [], "raw_text": "Jeremy Zilber and David Niven. 2000. Stereotypes in the news: Media coverage of african-americans in congress. Harvard International Journal of Press/Politics, 5(1):32-49.", "links": null }, "BIBREF51": { "ref_id": "b51", "title": "criminales, delito, delitos, agresi\u00f3n, agresiones, delincuencia, delincuente, delincuentes, malhechor, malhechores, robo, robos, hurto,hurtos, sustracci\u00f3n, sustracciones, mafia, mafias, mafioso, mafiosos, violaci\u00f3n, violaciones, violador, violadores, ped\u00f3filo, ped\u00f3filos, asesino, asesinos, asesinato,asesinatos, homicidio, homicidios, homicida, homicidas, violencia, violento, violentos,maltrato, maltratos, maltratador, maltratadores. Translations: \"faction leader", "authors": [], "year": null, "venue": "A.3 Crimes Words in Spanish: Cabecilla, cabecillas", "volume": "", "issue": "", "pages": "", "other_ids": {}, "num": null, "urls": [], "raw_text": "A.3 Crimes Words in Spanish: Cabecilla, cabecillas, ar- restado, arrestados, detenido, detenidos, sospecho, sospechos, sospechoso, sospechosos, ilegal, ile- gales, ilegalidad, clandestino, clandestinos, clan- destinidad, narcotr\u00e1fico, narcotraficante, narcotraf- icantes, traficante, traficantes, contrabando, con- trabandista, contrabandistas, aprehensi\u00f3n, aprehen- siones, incautaci\u00f3n, incautaciones, atraco, atracos, atracador, atracadores, asalto, asaltos, asaltante, asaltantes, crimen, criminalidad, criminal, crimi- nales, delito, delitos, agresi\u00f3n, agresiones, delin- cuencia, delincuente, delincuentes, malhechor, mal- hechores, robo, robos, hurto,hurtos, sustracci\u00f3n, sustracciones, mafia, mafias, mafioso, mafiosos, vi- olaci\u00f3n, violaciones, violador, violadores, ped\u00f3filo, ped\u00f3filos, asesino, asesinos, asesinato,asesinatos, homicidio, homicidios, homicida, homicidas, vio- lencia, violento, violentos,maltrato, maltratos, mal- tratador, maltratadores. Translations: \"faction leader\", \"faction lead- ers\", \"arrested\", \"arrested (plural)\",\"detained\", \"de- tained (plural)\", \"suspect\", \"suspects\", \"shady\", \"shady (plural)\", \"illegal\", \"illegal (plural)\", \"il- legality\", \"clandestine\", \"clandestine (plural)\", \"underground\", \"drug trafficking\", \"drug dealer\", \"drug traffickers\", \"trafficker\", \"traffickers\", \"smug- gling\", \"smuggler\", \"smugglers\", \"apprehension\", \"apprehensions\", \"seizure\", \"seizures\", \"robbery\", \"robberies\", \"robber\", \"robbers\", \"assault\", \"as- saults\", \"burglar\", \"burglars\", \"crime\", \"criminal- ity\", \"criminal\", \"criminals\", \"felony\", \"felonies\", \"aggression\", \"aggressions\", \"delinquency\", \"delin- quent\", \"delinquents\", \"malefactor\", \"malefac- tors\", \"stealing\", \"stealing (plural)\", \"theft\", \"theft (plural)\", \"thievery\", \"thievery (plural)\", \"mafia\", \"mafias\", \"gangster\", \"gangsters\", \"rape\", \"rapes\", \"rapist\", \"rapists\", \"pedophile\", \"pedophiles\", \"mur- derer\", \"murderers\", \"murder\", \"murders\", \"homi- cide\", \"homicides\", \"killer\", \"killers\", \"violence\", \"violent\", \"violent (plural)\", \"maltreatment\", \"mal- treatments\", \"batterer\", \"batterers\".", "links": null }, "BIBREF52": { "ref_id": "b52", "title": "Year British Colombian Ecuadorian German Italian Moroccan Romanian Spanish", "authors": [], "year": 2007, "venue": "", "volume": "", "issue": "", "pages": "", "other_ids": {}, "num": null, "urls": [], "raw_text": "Year British Colombian Ecuadorian German Italian Moroccan Romanian Spanish 2007", "links": null }, "BIBREF53": { "ref_id": "b53", "title": "drogodependiente, drogodependientes,alcohol, alcoholismo, borracho, borrachos, hero\u00edna, coca\u00edna, papelina, papelinas, bolsita, bolsitas, hach\u00eds, marihuana, sustancia, sustancias, cannabis, metanfetamina, anfetamina, speed, \u00e9xtasis, mdma. Translations: \"drug", "authors": [ { "first": "", "middle": [], "last": "Droga", "suffix": "" } ], "year": null, "venue": "", "volume": "", "issue": "", "pages": "", "other_ids": {}, "num": null, "urls": [], "raw_text": "A.4 Drugs Words in Spanish: Droga, drogas, adicci\u00f3n, adic- ciones, adicto, adictos, drogadicci\u00f3n, drogadicto, drogadictos, estupefaciente, estupefacientes, dro- godependencia, drogodependencias, drogodependi- ente, drogodependientes,alcohol, alcoholismo, bor- racho, borrachos, hero\u00edna, coca\u00edna, papelina, pa- pelinas, bolsita, bolsitas, hach\u00eds, marihuana, sus- tancia, sustancias, cannabis, metanfetamina, anfe- tamina, speed, \u00e9xtasis, mdma. Translations: \"drug\", \"drugs\", \"addiction\", \"ad- dictions\", \"addict\", \"addicts\",\"drug addiction\", \"drug addict\", \"drug addicts\", \"narcotic\", \"nar- cotics\", drug addiction, drug addiction, \"junkie\", \"junkies\", \"alcohol\", \"alcoholism\", \"drunk\", \"drunk (plural)\", \"heroin\", \"cocaine\", \" \"drug pa- per\" 11 , \"drug papers\", \"drug bag\" 12 , \"drug bags\" \"hashish\", \"marijuana\", \"substance\", \"substances\", \"cannabis\", \"methamphetamine\", \"amphetamine\", \"speed\", \"ecstasy\", \"mdma\".", "links": null }, "BIBREF54": { "ref_id": "b54", "title": "vulnerable, chabola, chabolas, chabolista, chabolistas, infravivienda, infraviviendas, barriada, barriadas, vagabundo, vagabundos, marginalidad, marginal, marginales, marginaci\u00f3n, marginado, marginados. 11 Papelina is a piece of paper to hold small amounts of drugs. 12 Bolsita is a small plastic bag to hold small amounts of drugs. Translations: \"misery", "authors": [], "year": null, "venue": "A.5 Poverty Words in Spanish: miseria, miserable, miserables", "volume": "", "issue": "", "pages": "", "other_ids": {}, "num": null, "urls": [], "raw_text": "A.5 Poverty Words in Spanish: miseria, miserable, miserables, pobreza, pobre, pobres, empobrecimiento, em- pobrecido, empobrecidos, mendicidad, mendigo, mendigos, desfavorecido, desfavorecidos, nece- sitado, necesitados, desesperaci\u00f3n, desesperados, desesperado, vulnerabilidad, vulnerables, vulner- able, chabola, chabolas, chabolista, chabolistas, infravivienda, infraviviendas, barriada, barriadas, vagabundo, vagabundos, marginalidad, marginal, marginales, marginaci\u00f3n, marginado, marginados. 11 Papelina is a piece of paper to hold small amounts of drugs. 12 Bolsita is a small plastic bag to hold small amounts of drugs. Translations: \"misery\", \"miserable\", \"miser- able (plural)\", \"poverty\", \"poor\", \"poor (plural)\", \"impoverishment\", \"impoverished\", \"impoverished (plural)\", \"begging\", \"beggar\", \"beggars\", \"dis- advantaged\", \"disadvantaged (plural)\", \"people in need\", \"people in need (plural)\", \"despera- tion\", \"desperate\", \"desperate (plural)\", \"vulnera- bility\", \"vulnerable\", \"vulnerable (plural)\", \"shanty town\", \"shanty town (plural)\", \"person that lives in shanty town\", \"person that lives in shanty town (plural)\", \"slum\", \"slums\", \"poor neigh- borhood\", \"poor neighborhoods\", \"vagabond\", \"vagabonds\", \"marginality\", \"marginal\", \"marginal (plural)\", \"marginalization\", \"marginalized (plu- ral)\",\" marginalized (plural)\".", "links": null }, "BIBREF55": { "ref_id": "b55", "title": "Prostitution Words in Spanish: Prostituci\u00f3n, prost\u00edbulo, prost\u00edbulos, prostituta, prostitutas, proxenetismo, proxeneta, proxenetas. Translations", "authors": [ { "first": "A", "middle": [], "last": "", "suffix": "" } ], "year": null, "venue": "", "volume": "", "issue": "", "pages": "", "other_ids": {}, "num": null, "urls": [], "raw_text": "A.6 Prostitution Words in Spanish: Prostituci\u00f3n, prost\u00edbulo, prost\u00edbulos, prostituta, prostitutas, proxenetismo, proxeneta, proxenetas. Translations: \"Prostitution\",\" brothel\", \"broth- els\", \"prostitute\", \"prostitutes\", \"pimping\", \"pimp\", \"pimps\".", "links": null } }, "ref_entries": { "FIGREF0": { "num": null, "uris": null, "type_str": "figure", "text": "For instance, if the word vector that represents the adjective \u2212 \u2212\u2212\u2212\u2212\u2212\u2212\u2212 \u2192 delincuente (\"delinquent\") is more strongly associated with the word vector \u2212 \u2212\u2212\u2212\u2212 \u2192 rumano (\"Romanian\") than with the word vector \u2212 \u2212\u2212\u2212\u2212 \u2192 espa\u00f1ol (\"Spanish\"), that suggests there is bias in the model. It is not the similarity between \u2212 \u2212\u2212\u2212\u2212\u2212\u2212\u2212 \u2192 delincuente and \u2212 \u2212\u2212\u2212\u2212 \u2192 rumano that determines the presence of bias, but the fact that the distances between \u2212 \u2212\u2212\u2212\u2212 \u2192 rumano and \u2212 \u2212\u2212\u2212\u2212 \u2192 espa\u00f1ol are not equal regarding the adjective \u2212 \u2212\u2212\u2212\u2212\u2212\u2212\u2212 \u2192 delincuente." }, "FIGREF1": { "num": null, "uris": null, "type_str": "figure", "text": "Number of documents and sentences per year in the 20 Minutos data included in the analysis." }, "FIGREF2": { "num": null, "uris": null, "type_str": "figure", "text": "Average group vector variance." }, "FIGREF3": { "num": null, "uris": null, "type_str": "figure", "text": "Average bias score for crimes concepts.Figure 4: Average bias score for drugs concepts." }, "FIGREF4": { "num": null, "uris": null, "type_str": "figure", "text": "Average bias score for poverty concepts.Figure 6: Average bias score for prostitution concepts." }, "TABREF1": { "html": null, "content": "", "num": null, "type_str": "table", "text": "Random Effects model predictions of bias scores for concepts related to crimes and drugs. *p < .05, **p < .01, ***p < .001. Standard errors for each coefficient shown in parenthesis." }, "TABREF3": { "html": null, "content": "
", "num": null, "type_str": "table", "text": "Random Effects model predictions of bias scores for concepts related to poverty and prostitution. *p < .05.**p < .01, ***p < .001. Standard errors for each coefficient shown in parenthesis." } } } }