‘Stewardship’, not ‘ownership’: from communities of proficient non-natives and historical lingua francas to AI and LLM-generated written English

Abstract This article examines stewardship as a mechanism of supra-regional language management under conditions of extensive non-native use, drawing on historical cases of Sanskrit, Latin and Classical Arabic. Unlike models grounded in native-speaker “ownership” and state-centred language policy, stewardship emerges when proficient non-native users collectively sustain a lingua franca across long spans of time and wide—often discontinuous—spaces. In these historical cases, learned communities managed stability and controlled innovation in predominantly written high registers, preserving the languages as cosmopolitan media long after any substantial native-speaker base had receded or disappeared. The article argues that supra-regional written registers of English are now undergoing a conceptually comparable—but technologically distinct—shift in its own supra-regional management. Central is an erosion of control by traditional norm-enforcing institutions and a reconfiguration of norm provision through text-generative AI: widely used LLM-based tools diffuse standard-like formulations at scale and, crucially, do so from training inputs that include large volumes of both native and non-native English. The article concludes by framing AI-mediated norm provision as an emerging stewardship regime for global English and by outlining the empirical questions this hypothesis raises: whose usage is amplified, how institutional actors respond and what this implies for the future politics of English standards.