Content Moderation in the Global South: Four Low-Resource Languages

This report synthesizes an 18-month CDT research effort examining content moderation systems across South Asia, North and East Africa, and South America, built around four case studies of low-resource languages: Maghrebi Arabic dialects, Kiswahili, Tamil, and Quechua. All four were selected because scarce digitized training data makes it difficult to build accurate, equitable automated moderation tools for them.

Methodologically, the underlying research combined interviews with social media users, digital rights advocates, language activists, tech-company representatives, content moderators, and creators, alongside an online survey of more than 560 frequent social media users spanning the studied regions.

Across the case studies, the comparative findings show that content policies are defined, enforced, and experienced unevenly depending on language and region, with platforms' natural-language-processing and large-language-model systems performing poorly wherever digitized training data is scarce. The report frames this as a structural pattern rather than isolated language-specific failures, and calls for greater investment and local linguistic expertise to close the gap with better-resourced languages like English.