A Critical Study of Algorithmic Amplification, Language Bias, and Digital Linguistic Representation on Social Media Platforms

Social media platforms function as primary sites of global communication, yet the algorithmic systems that govern content distribution on these platforms are not linguistically neutral. This study investigates how algorithmic amplification produces and sustains language bias against speakers of Yoruba and Nigerian Pidgin English in South-west Nigeria, with particular attention to the mechanisms through which platform architectures, differential amplification patterns, and training data assumptions reproduce dominant Anglophone linguistic ideologies. Adopting a qualitative research design that combines critical discourse analysis and digital ethnography, the study draws on semi-structured interviews, digital observation, and document analysis conducted over three months across six states in South-west Nigeria, with data managed and analysed using NVivo 14 and MAXQDA 2024. Findings reveal that Yoruba content achieves only 31.25% and Nigerian Pidgin content 41.25% of the engagement generated by equivalent Standard English content across X, Facebook, TikTok, and YouTube. Content moderation systems disproportionately flag Nigerian Pidgin, and neither language receives full natural language processing support on any of the four platforms. The study concludes that algorithmic bias constitutes a structural mechanism of digital linguistic exclusion with significant implications for epistemic justice, platform regulation, and the equitable representation of postcolonial language communities in digital public spheres