The Algorithmic Native Speaker: Linguistic Racism, Generative AI and English Language Education From the Global South

Generative artificial intelligence (GenAI) is rapidly becoming a language‐learning partner, writing adviser, assessor, curriculum designer and source of pedagogical knowledge. Yet its expansion into English language education has largely been discussed through the languages of efficiency, innovation, personalisation and academic integrity. This conceptual article argues that such framings are inadequate for the Global South, where English education is already organised by colonial histories, racialised hierarchies of speakerhood, marketised promises of mobility and standardised proficiency regimes. Bringing raciolinguistic theory, coloniality, native‐speakerism and critical work on language datafication into conversation, the article conceptualises GenAI as an algorithmic native speaker; that is, not a literal speaker, but a sociotechnical authority that models, judges and circulates idealised English while obscuring the racial, geopolitical and commercial conditions of that authority. Four interlocking operations are identified: (i) extractive datafication, (ii) probabilistic standardisation, (iii) automated evaluation and (iv) aspirational subject formation. Chile is used as a situated example of how these operations may enter a stratified English education system in which English is promoted as a route to global participation while Indigenous languages and local English practices remain comparatively marginal. The article argues that GenAI can intensify a ‘bilingual illusion’ in which linguistic modernity is imagined chiefly as movement between standard Spanish and standard English. However, these effects are not inevitable. Educators, policymakers and learners retain agency in deciding how GenAI is adopted, questioned and governed, so these operations are better understood as tendencies that can be amplified or interrupted than as fixed outcomes. The article proposes a raciolinguistically informed critical AI literacy for English language education organised around five pedagogical commitments: historicising, auditing, counter‐prompting, re‐voicing and collective governance. The central implication is that equitable GenAI integration requires more than detecting biased outputs. It requires transforming who is authorised to define legitimate language, knowledge and educational futures.