Uploaded October 2025 | Updated September 2026, 2 hours ago
In 1492, the Castilian grammarian Antonio Nebrija observed that "la lengua fue siempre compañera del imperio".¹ Discourses on language are indeed central to discourses of power and our concepts of language play a crucial role in legitimising political structures that legislate human difference (Deumert & Storch 2020, Errington 2008). While Western colonialist/nationalist epistemes framed global sociolinguistic order along territorial and national-ethnic lines and transformed speech into standardized writing in printed text, this talk discusses how languages are discursively and materially constructed in the context of machine-learning culture. I argue that large language models built by monopolistic digital tech industries can entail new forms of imperial exploitation (Couldry & Mejias 2024), indicative of post-national social orders.
Approaching large language models from a posthumanist and assemblage perspective (Pennycook 2024), I understand them as a specific form of mediated human interaction that is distributed in time and space. How languages are constructed in computational data infrastructures has its roots in colonial and national traditions of the age of print literacy. At the same time, there are discourses, practices and affordances specific to digitization and machine-learning. Based on the study of computer science research publications and interviews with designers of language models, I discuss which language ideological discourses guide the materialization and conceptualization of language in this cultural context. We find Anglophone hegemonies and referential, logocentric, but also non-purist ideologies of language, as consumer satisfaction, data extraction and user monitoring – rather than the formation of culturally homogeneous communities – are among the aims of building language technologies. The configuration of languages as data and as dataalgorithmic models is thus influenced by capitalist histories and techno-solutionist and surveillance-based motives (Pasquinelli 2023, Zuboff 2019). Secondly, drawing on observation and interviews with users, I examine the sensory, phenomenological experiences users have with chatbots or voice assistants. As technology design constructs the tools as agentive but opaque, users (but also designers) tend to discursively ascribe agency and authority to them, leading to their fetishization (Keane 2024). From this perspective, LLMs in their current form are authoritative linguistic infrastructures intertwined with surveillance capitalism (Tacheva & Ramasubramanian 2023). In general, the interplay of technologies of mediation and language plays a decisive role in establishing sociolinguistic and socio-political orders – where machine-learning language technologies appear as companion of yet-to-be-understood types of imperialism.
References
Deumert, Ana, and Anne Storch. 2020. "Introduction: Colonial linguistics—then and now " In Colonial and decolonial linguistics: knowledges and epistemes, edited by Ana Deumert, Anne Storch and Nick Shepherd, 1-21, Oxford: Oxford Academic.
Errington, Joseph. 2008. Linguistics in a colonial world. A story of language, meaning and power. Malden, Mass.: Blackwell.
Keane, E. Webb. 2024. Animals, robots, gods. Adventures in the moral imagination. Milton Keynes: Allen Lane.
Mejias, Ulises A., and Nick Couldry. 2024. Data grab. The new colonialism of Big Tech (and how to fight back). London: Penguin.
Pasquinelli, Matteo. 2023. The eye of the master. A social history of artificial intelligence. London: Verso.
Pennycook, Alastair. 2024. Language assemblages. Cambridge: Cambridge University Press.
Tacheva, Jasmina, and Srividya Ramasubramanian. 2023. "AI Empire: Unraveling the interlocking systems of oppression in generative AI’s global order." Big Data & Society:1-13. DOI: 10.1177/20539517231219241.
Zuboff, Shoshana. 2019. The age of surveillance capitalism: the fight for a human future at the new frontier of power. New York: Public Affairs.
-----
¹ “Language was always the companion of empire.”
In 1492, the Castilian grammarian Antonio Nebrija observed that "la lengua fue siempre compañera del imperio".¹ Discourses on language are indeed central to discourses of power and our concepts of language play a crucial role in legitimising political structures that legislate human difference (Deumert & Storch 2020, Errington 2008). While Western colonialist/nationalist epistemes framed global sociolinguistic order along territorial and national-ethnic lines and transformed speech into standardized writing in printed text, this talk discusses how languages are discursively and materially constructed in the context of machine-learning culture. I argue that large language models built by monopolistic digital tech industries can entail new forms of imperial exploitation (Couldry & Mejias 2024), indicative of post-national social orders.
Approaching large language models from a posthumanist and assemblage perspective (Pennycook 2024), I understand them as a specific form of mediated human interaction that is distributed in time and space. How languages are constructed in computational data infrastructures has its roots in colonial and national traditions of the age of print literacy. At the same time, there are discourses, practices and affordances specific to digitization and machine-learning. Based on the study of computer science research publications and interviews with designers of language models, I discuss which language ideological discourses guide the materialization and conceptualization of language in this cultural context. We find Anglophone hegemonies and referential, logocentric, but also non-purist ideologies of language, as consumer satisfaction, data extraction and user monitoring – rather than the formation of culturally homogeneous communities – are among the aims of building language technologies. The configuration of languages as data and as dataalgorithmic models is thus influenced by capitalist histories and techno-solutionist and surveillance-based motives (Pasquinelli 2023, Zuboff 2019). Secondly, drawing on observation and interviews with users, I examine the sensory, phenomenological experiences users have with chatbots or voice assistants. As technology design constructs the tools as agentive but opaque, users (but also designers) tend to discursively ascribe agency and authority to them, leading to their fetishization (Keane 2024). From this perspective, LLMs in their current form are authoritative linguistic infrastructures intertwined with surveillance capitalism (Tacheva & Ramasubramanian 2023). In general, the interplay of technologies of mediation and language plays a decisive role in establishing sociolinguistic and socio-political orders – where machine-learning language technologies appear as companion of yet-to-be-understood types of imperialism.
References
Deumert, Ana, and Anne Storch. 2020. "Introduction: Colonial linguistics—then and now " In Colonial and decolonial linguistics: knowledges and epistemes, edited by Ana Deumert, Anne Storch and Nick Shepherd, 1-21, Oxford: Oxford Academic.
Errington, Joseph. 2008. Linguistics in a colonial world. A story of language, meaning and power. Malden, Mass.: Blackwell.
Keane, E. Webb. 2024. Animals, robots, gods. Adventures in the moral imagination. Milton Keynes: Allen Lane.
Mejias, Ulises A., and Nick Couldry. 2024. Data grab. The new colonialism of Big Tech (and how to fight back). London: Penguin.
Pasquinelli, Matteo. 2023. The eye of the master. A social history of artificial intelligence. London: Verso.
Pennycook, Alastair. 2024. Language assemblages. Cambridge: Cambridge University Press.
Tacheva, Jasmina, and Srividya Ramasubramanian. 2023. "AI Empire: Unraveling the interlocking systems of oppression in generative AI’s global order." Big Data & Society:1-13. DOI: 10.1177/20539517231219241.
Zuboff, Shoshana. 2019. The age of surveillance capitalism: the fight for a human future at the new frontier of power. New York: Public Affairs.
-----
¹ “Language was always the companion of empire.”
