Gloria Gennaro

Migration is a defining phenomenon of our time, around which heightened political rhetoric creates the impression of unprecedented political conflict. In the UK, this perception coincides with major political developments, such as the Brexit referendum and the rise of anti-immigration politics. However, systematic understanding of elite discourse on this subject remains limited. We introduce a novel machine learning pipeline that combines LLMs and human inputs to analyze sentiment towards immigration in political speech. Applying our model to spoken contributions in the UK House of Commons from 1923 to 2025, we report high classifier performance and stability across modeling choices. Our descriptive findings validate and extend existing qualitative research: Labour Members of Parliament (MPs) have expressed more positive sentiment than Conservatives since the early 1960s, with party divergence widening in the early 1980s. Furthermore, cross-party rhetoric has grown increasingly negative over the past fifteen years, with Conservative and Reform UK MPs now using a more negative tone than any party has in the period studied. We discuss how this pipeline and dataset can support future research, highlighting use cases on the linkages between elite discourse and public opinion, discourse moralization, and positive and negative frames.

Migration is a defining phenomenon of our time, around which heightened political rhetoric creates the impression of unprecedented political conflict. In the UK, this perception coincides with major political developments, such as the Brexit referendum and the rise of anti-immigration politics in recent decades. We apply quantitative text analysis techniques to the Hansard collection of spoken contributions in the House of Commons to examine how political elites have discussed immigration over the past century. We find that the Labour Party has consistently expressed more positive sentiment towards immigration compared to the Conservative party since the early 1960s, and that these positions further diverged in the early 1980s. The last 15 years have seen more negative rhetoric across all major parties, with immigration discussed among Conservative and Reform UK MPs in the most negative terms since 1923. We discuss possible future research avenues, supported by suggestive findings on the increasing moralization of the immigration debate and growing negativity towards asylum.

Hate speech is widely seen as a significant obstacle to constructive online discourse, but the most effective strategies to mitigate its effects remain unclear. We claim that understanding its distribution across users is key to developing and evaluating effective content moderation strategies. We address this missing link by first examining the distribution of hate speech in five original datasets that collect user-generated posts across multiple platforms (social media and online newspapers) and countries (Switzerland and the United States). Across these diverse samples, the vast majority of hate speech is produced by a small fraction of users. Second, results from a pre-registered field experiment on Twitter indicate that counterspeech strategies obtain only small reductions of future hate speech, mainly because this approach proves ineffective against the most prolific contributors of hate. These findings suggest that complementary content moderation strategies may be necessary to effectively address the problem.

The politicized nature of violence in armed conflicts can favor the persistence of group divisions into peacetime. Yet, many local violence dynamics may not correspond to the conflict cleavages, leaving the experiences of affected communities unrepresented and the politicization of memory ambiguous. We show that such "marginalized'' memories can be mobilized over time along new cleavages, following changes in the political environment. We study a case of indiscriminate violence against civilians by French-North African soldiers during the Italian campaign of World War II. Through a geo-targeted survey experiment, we document that the subjective understandings of the past perpetuated within families and communities correlate with political affiliation, and that their salience causally affects attitudes toward immigration. An analysis of post-war elections further shows that increases in a party's nationalist stance result in larger electoral gains in formerly victimized communities.

Under what conditions can legacies of past violence shape political behaviour? We propose a theory of how war victimization defines attitudes over the long run, and how these can be activated by changes in the political environment. We argue that exposure to violence by members of a different ethnic group generates hostility that spills over other outgroups; this latent hostility resonates with nationalist appeals to ingroup (national) identity against non-nationals. As a result, nationalist appeals transform latent attitudes into political support, leading to different behavioural patterns among victimized communities. To test our theory, we use a historical case of indiscriminate violence against civilians by French colonial troops in central Italy during the last phase of World War II. Using data on electoral outcomes and party platforms over the full post-war period from 1948 to 2018, we show that parties which increase the nationalist content of their electoral manifesto gain more votes in formerly victimized communities, compared to otherwise similar communities. We also discuss these dynamics in qualitative case studies of party repositioning, such as the transformation of the regionalist party Northern League into a radical right party with strong nationalist appeals. We show that the results are not driven by pre-war differences in political preferences, or in socio-economic characteristics.

Despite heightened awareness of the detrimental impact of hate speech on social media platforms on affected communities and public discourse, there is little consensus on approaches to mitigate it. While content moderation—either by governments or social media companies—can curb online hostility, such policies may suppress valuable as well as illicit speech and might disperse rather than reduce hate speech. As an alternative strategy, an increasing number of international and nongovernmental organizations (I/NGOs) are employing counterspeech to confront and reduce online hate speech. Despite their growing popularity, there is scant experimental evidence on the effectiveness and design of counterspeech strategies (in the public domain). Modeling our interventions on current I/NGO practice, we randomly assign English-speaking Twitter users who have sent messages containing xenophobic (or racist) hate speech to one of three counterspeech strategies—empathy, warning of consequences, and humor—or a control group. Our intention-to-treat analysis of 1,350 Twitter users shows that empathy-based counterspeech messages can increase the retrospective deletion of xenophobic hate speech by 0.2 SD and reduce the prospective creation of xenophobic hate speech over a 4-wk follow-up period by 0.1 SD. We find, however, no consistent effects for strategies using humor or warning of consequences. Together, these results advance our understanding of the central role of empathy in reducing exclusionary behavior and inform the design of future counterspeech interventions.