The words of hate speech: A lexical study of homotransphobia in an Italian Twitter corpus

The everyday discursive production of dehumanising representations and stereotypical beliefs regarding the LGBTQIA+ community undermines the self-respect of both individuals and the target group by damaging their social agency and entitlement dimensions.This chapter proposes a quantitative and qualitative analysis of the TWEER corpus, which consists of 5660 Italian tweets on queer topics. The aims of the work are quantifying the presence of hate speech online and describing the main linguistic features that characterise such language in Italian. Quantitative analysis consists in a manual annotation of the corpus based on a fine-grained scheme comprising six labels (Sanguinetti et al. 2018): hate speech, intensity of hate, aggressiveness, offensiveness, irony, and stereotype. We found that hate speech covers 13 per cent of the entire corpus, but only 6 per cent of those tweets contained an explicit inciting of hatred, while most hateful tweets were superficially polite, containing dangerous prejudices against LGBTQIA+ individuals. We then propose a lexical-semantic study on a subcorpus which focuses on isolating the most representative meaning clusters in explicitly hateful texts. By analysing each lexical word, we found three main clusters, namely references to politics, health, and ethics, while mentions of sexual identity issues were far rarer, confirming that even explicit hate relies on a heteronormative matrix rather than an impulsive intolerance of certain kinds of sexual orientations or gender identities.