Polarized intersectionality in Italy: changing political alliances and the weaponization of women's public representation

This paper proposes a joint application of online network analysis and Natural Language Processing techniques to contribute to ongoing discussions about the ways in which women, their bodies and choices are included in but also constituted by contemporary public discourse for how it develops within large-scale social media discussions. It does so by exploring and comparing two cases that occurred in Italy between 2019 and 2020. The first case refers to Carola Rackete, the Captain of Sea Watch 3, who forced her way into the port of Lampedusa on 26 June 2019 to rescue migrants' lives. The second case concerns the release on 9 May 2020 of Silvia Aisha Romano, an NGO aid-worker kidnapped in Kenya and returned home converted to Islam. Both events triggered two extremely heated discussions that pivoted around opposing collective (mis)representations of the two women that were shaped at the intersection of stereotyped visions particularly of gender, age, religion, and political activism. Yet, these two discussions occurred in conjunction with diverse government alliances in the country. While Rackete disembarked in Italy during the “yellow-green” government, formed by Five Stars Movement and the League, Romano was rescued by the “yellow-red” alliance mainly constituted by the Five Stars Movement and the Democratic Party. In this work, we claim that public (mis)representations of both Rackete and Romano were tailored and circulated instrumentally as weapons in the context of highly polarized political dynamics fed by legacy media, political leaders, and a myriad of citizens. We argue that political antagonism drove the choice of stereotypes that were used by adversarial groups to recount the two women and defeat political adversaries’ projects and visions. We thus explore how switching political alliances engendered different systems of “polarized intersectionality” - that is, (mis)representations of women in the public discourse that develop at the intersection of different axes of discrimination and in conjunction with political polarization dynamics. Starting from two dataset of 227.001 tweets about Carola Rackete and 55.897 tweets about Silvia Aisha Romano, we design two semantic networks where tweets are linked whenever retweeted by the same account(s). Through the application of community-finding algorithms, clusters of tweets are isolated that approximate the conversational areas structuring the two discussions. We derive the overall political leaning of these conversational areas starting from the presence of tweets authored by political parties, leaders, and partisan media accounts. We then analyze these conversational areas through Top2Vec technique, a topic modeling based on sentence embeddings and density clustering, to identify the different topics feeding the two discussions. After comparing tweets’ communities to infer general levels of polarization, we associate political and semantic positions with specific (mis)representations of the two women. Finally, we examine our results to understand if (mis)representations of Rackete and Romano changed over time and depending on transformed political alliances. Our preliminary results seem to confirm that intersectional (mis)representations of Carola Rackete and Silvia Aisha Romano permeate both discussions and all political groups albeit taking different forms and, in fact, depending on the varying geometries of partisan oppositions.