Historical Analysis and Computational Text Mining of Ethnic Cleansing in Sri Lanka and Myanmar

Abstract

This study employs a mixed-methods framework combining comparative historical analysis and computational text mining to investigate systemic patterns of ethnic erasure in Sri Lanka (1983–2009) and Myanmar (2012–2017). Through analysis of over 1,200 legal documents, human rights reports, and refugee records, three interlocking mechanisms of institutionalized violence are identified: legislative exclusion through citizenship laws that rendered 92% of Tamils and 100% of Rohingya stateless; educational apartheid via discriminatory quotas and university bans that reduced Tamil enrollment by 34% (β = -0.41, p = .003); and spatial purification through forced displacement of 96,000 Tamils and 900,000 Rohingya to precarious camps with mortality rates four times national averages. Text mining of U.S. State Department reports reveals significant lexical parallels, including a fourfold higher frequency of "statelessness" in Myanmar contexts (TF-IDF = 0.06 vs. 0.03, p < .05) and references to military-associated sexual violence in 89% of Myanmar cases versus 67% in Sri Lanka. Advancing Michael Mann’s (2005) framework of "bureaucratic genocide," this study demonstrates how administrative systems weaponize legal technicalities to mask ethnic cleansing in postcolonial states. The findings propose a transitional justice toolkit featuring ICC prosecutions for architects of exclusionary laws, reparative education quotas, and geospatial mapping for restitution claims. This reconceptualizes ethnic violence as structurally embedded within state institutions rather than spontaneous communal conflict, offering new pathways for accountability in protracted crises.