Hate Speech Dynamics Against African descent, Roma and LGBTQI Communities in Portugal

Proceedings of the 13th Conference on Language Resources and Evaluation (LREC 2022), pages 2362–2370 Marseille, 20-25 June 2022 © European Language Resources Association (ELRA), licensed under CC-BY-NC-4.0 2362 Hate Speech Dynamics Against African descent, Roma and LGBTQ+ Communities in Portugal Paula Carvalho1, Bernardo Matos1,2, Raquel Santos1,2, Fernando Batista1,3, Ricardo Ribeiro1,3 1 INESC-ID Lisboa, Rua Alves Redol, 9, 1000-029 Lisboa, Portugal 2 Instituto Superior T´ecnico, Universidade de Lisboa, Lisboa, Portugal 3 Instituto Universit´ario de Lisboa (ISCTE-IUL), Lisboa, Portugal pcc@inesc-id.pt, {bernando.cunha.matos, raquel.bento.santos}@tecnico.ulisboa.pt {fernando.batista, ricardo.ribeiro}@inesc-id.pt Abstract This paper introduces FIGHT, a dataset containing 63,450 tweets, posted before and after the official declaration of Covid-19 as a pandemic by online users in Portugal. This resource aims at contributing to the analysis of online hate speech targeting the most representative minorities in Portugal, namely the African descent and the Roma communities, and the LGBTQ+ community, the most commonly reported target of hate speech in social media at the European context. We present the methods for collecting the data, and provide insightful statistics on the distribution of tweets included in FIGHT, considering both the temporal and spatial dimensions. We also analyze the availability over time of tweets targeting the aforementioned communities, distinguishing public, private, and deleted tweets. We believe this study will contribute to better understand the dynamics of online hate speech in Portugal, particularly in adverse contexts, such as a pandemic outbreak, allowing the development of more informed and accurate hate speech resources for Portuguese. Keywords: hate speech, offensive speech, Twitter 1. Introduction Most research in hate speech detection focuses on the creation of language resources, and on the develop- ment of methods and tools for automatically detecting offensive and abusive language (Fortuna and Nunes, 2018). However, the lack of consensus on the defini- tion and characterization of hate speech has led to the creation of heterogeneous resources, particularly anno- tated corpora, making it difficult to compare them (Po- letto et al., 2021). In addition, hate speech is intrin- sically dependent on the sociocultural context, which means that existing resources cannot be directly trans- ferable or easily adaptable to other linguistic and prag- matic contexts (Nozza, 2021). Although there are few corpora specifically designed for detecting hate speech in Portuguese (Fortuna et al., 2019), their usefulness is quite limited to study spatiotemporally delimited phe- nomena, such as the dynamics of online hate speech in Portugal, before and during the Covid-19 pandemic. Moreover, the data included in the existing corpora is often selected based on generic lexical-based ap- proaches, using closed lists of keywords with negative polarity, typically involving epithets and slurs that may be used to incite hatred or violence against an individ- ual or a group. Used in isolation, this selection method leaves out an immense set of potentially relevant ha- tred content, including indirect or covert hate speech (Baider and Constantinou, 2020; Kumar et al., 2018), often resorting to rhetorical figures, such as irony, sar- casm, humor, analogy, metaphor, and rhetorical ques- tions, and then preventing an in-depth understanding of the nature and …

Content truncated to 500 words