The language of cyberbullying: forms and mechanisms of online prejudice and discrimination in annotated comparable corpora of Portuguese and English

With the rapid rise of Internet use, socially troublesome online behavior quickly became widespread. Attitudes of intolerance and prejudice soon appeared, together with frequent signs of discrimination and expressions of insult. Bullying, trolling and hate speech thus became common, making it more difficult, if not risky, to interact in the virtual world. Awareness of the dangers of computer-mediated communication was initially linked to the participation of children and youth in social networks, where they were found to be easy victims of sexual predators and other abusers. However, victimization in online forums cuts across all sociolinguistic variables, reaching all age groups, all nationalities and geographical origins, school levels and ethnic groups, and also affecting individuals of both sexes and of various sexual, religious and other identities. It is, therefore, a crosscutting phenomenon that deserves a multidisciplinary analysis. The project starts by establishing a significant sample of online interactions organized according to the aforementioned sociolinguistic variables. The construction of two digital, stratified, comparable corpora, in Portuguese and English, of texts published in different Internet forums (comment boards from online newspapers, blogs, and social networks) duly annotated with computer tools (taggers), will allow the extraction of information relevant to several studies of a comparative nature. Thus, the project intends to create a solid database that will work as a research interface, open to the scientific community and applicable to different academic domains. This is an innovative contribution in light of the instruments currently available in Portugal for linguistic analysis of digital content, which will lend itself to sociological, psychological, educational and legal approaches. These are, in fact, some of the synergies that the project aims to establish by integrating specialists from complementary academic areas. The work methodology will combine a quantitative approach, from the area of Corpus Linguistics, with a qualitative approach. The former will detect textual types and genres as well as recurrent words and constructions. The latter will cross the sociolinguistic variables with morpho-syntactic, lexical-semantic, and discursive-pragmatic analyses, with a view to identifying emerging ideological communities. The aim is to know what actors - be they active or passive, victimizing or victimized - populate the virtual world of marginalization and conflict, and what their recurrent identities and attitudes are in terms of linguistic expression. The results of this extended study will allow us to refine tools of control, detection and awareness-raising in society at large, including legal regulations for Internet use, automatic identification of situations of risk and abuse, and a broader debate about the freedoms and limits of cyber-navigation.