Authors: Molly Miranker*, Texas State University - San Marcos
Topics: Human Rights, Migration, Political Geography
Keywords: Migrant deaths, corpus linguistics, text mining, Twitter, U.S. Border Patrol
Session Type: Paper
Presentation File: No File Uploaded
In 1994, U.S. Customs and Border Patrol (CBP) implemented their “prevention through deterrence” strategy to secure the Southwest border, limit trafficking of contraband (e.g., drugs), and deter undocumented migration. Since then the U.S.-Mexico border has become an increasingly lethal migration corridor, with over 7000 unidentified border crosser (UBC) deaths recorded by CBP (Spradley et al. 2019). Currently, southern Arizona and South Texas have the highest rates of migrant death. In response, law enforcement, medico-legal, academics, and humanitarian groups have mobilized to bridge institutional and informational gaps and improve UBC recovery and identification. While forensic investigatory practices have mostly been centralized in southern Arizona, in South Texas UBC case information (Spradley et al. 2019) is frequently incomplete, inconsistent, or dispersed among multiple institutions. In this presentation, I will explore the use of corpus linguistics and natural language processing methods to assess the informational potential of CBP Twitter accounts as a potential source to link UBC case information. A year of tweets (May 2018 – July 2019) from CBP South Texas and Arizona accounts were analyzed for themes (i.e., term frequencies), content commonality/comparison, and overall sentiment. Results suggest that CBP South Texas may not report enough information on UBCs to be useful for directing search and recovery efforts. It is also unclear whether analyzing multiple accounts would add useful information to aid UBC cases. Finally, sentiment analysis may not be meaningful in this context as people may perceive CBP tweets differently based on their political beliefs.