APSA Pre-Conference | Narrative and Text Analysis in the Study of Migration and Citizenship

Instructor: Tom Einhorn, Sociology, with training material from the UBC Centre for Computational Social Science

Have you previously done qualitative analysis of texts or discourse and are curious about how AI and Machine Learning can help with analysis? Does your research rely on human coding and you wonder about the potential – and limits – of computational approaches? Come to this workshop to learn more. The session will be an introduction to Large Language Models and prompt-based text classification. We will touch on ethics and IRB, with time for questions.

(Note: the session is designed for beginners and will not cover advanced techniques. No programming experience required.)

• Graphic Detentions - Efrat Arbel, Allard School of Law, UBC

This paper reflects on the process of creating a graphic ethnography focused on the lived-experiences of immigration detention in Canada. This initiative serves to humanize experience, and to shift these experiences out of their relegated invisibility and into public discourse.

• The Ethics of Being an Immigrant - Ashwini Vasanthakumar, Faculty of Law, Queen's University

• Quiet Inclusion - Shelby Carvalho, King Center on Global Development, Stanford University

Why do governments sometimes provide more services than what is specified in formal policies? Focusing on the case of refugee integration in education systems in Africa, I use narrative analysis from observation of policymaking processes, audio recordings of responses from a vignette survey experiment with elites in Kenya, and traditional interview methods to introduce the concept of ‘quiet inclusion’.

• Trump and Immigration: Discourse and Implications in the North American Space - Abdullah Alzabaidi, Huong Le, Yasmeen Abu-Laban and Geoffrey Rockwell, University of Alberta

This paper investigates significant immigration policy actions, discourse, and public narratives articulated during the first six months of Donald Trump’s second term (January to July 2025) as President of the United States. Our analysis identifies three core discursive strategies: (1) the securitization and criminalization of migration using emergency and wartime logics; (2) the portrayal of immigration as a threat to national identity and sovereignty; and (3) the characterization of executive action as a necessary, urgent response to an “invasion.”

• Manufacturing Crisis, Making Europe: How Media Discourses on Migration Produce Negative Europeanisation - Hélène Thiollet, Sciences Politiques

This project investigates how migration crisis discourses in the French press since 2015 have contributed to the Europeanisation of the French public sphere. Drawing on a longitudinal content analysis of six major French newspapers (2009–2022), we examine how migration-related crises were framed, politicised, and linked to broader challenges of governance and values at both national and EU levels.

• Issue Attention, Media Narratives, and Immigration Shocks: An LLM Classification Approach - Alexander Tripp, Political Science, Vanderbilt University

How do media narratives shift in response to immigration shocks? I develop a dynamic model of political communications that conceptualizes a common media response across immigration shocks. I find strong evidence for a humanitarian phase, when an immigration shocks prompt sharp increases in humanitarian narratives in Colombia, Germany and Poland, but no direct evidence for a threat phase, as the proportion of threat narratives does not significantly increase over time. Analysis draws on 200,000 immigration-focused articles from prominent national newspapers, classification of media coverage into themes using a few-shot ChatGPT 4o model (validated with undergraduate coders), and an interrupted time series model.

• Stereotypes of High- and Low-Skilled Immigrants to Canada: Evidence from an Online Pilot Study - Vincent Hopkins, UBC; Andrea Lawlor, McMaster University and Mireille Paquet, Concordia University

Survey research shows, across diverse contexts, that citizens tend to prefer skilled migrants. But What comes to mind when voters think about immigrant skill levels? Which stereotypes predominate? We report on an online pilot survey (N=2,422) that asks respondents to think about differential skilled immigrant groups and collects open-ended text responses. We use a Large Language Model (LLM) to measure the sentiment of text responses (positive-negative) and use LLM to measure the prevalence of stereotypes in the open-ended survey responses.

• Liking and Disliking the Major US Political Parties: Understanding the Viewpoints of Asian American and Latino Voters – Jongwoo Jeong, Political Science, Georgia State University

Analyzing open-ended survey responses documenting Asian Americans’ and Latinos’ “likes” and “dislikes” about major U.S. political parties, I find that religious and social conservatism play a pivotal role in shaping immigrants’ “dislike” of the major parties, independent from any positive affinities, such that these negative sentiments overwhelm their likes and more strongly predict turnout. I use machine learning analysis together with transformer models of natural language processing.

This final discussion will sum up lessons from the day and reflect on whether and how different methods serve distinct purposes within academic circles well as beyond them.

More information about speakers coming soon.