Machine translation and large language models have become part of daily routines in work, study, and play. Linguistic anthropologists have begun to argue that these systems are transducers before they are translators, which is to say, rather than only carrying denotational content from one language to another, they instead rebuild the tone, register, and social positions of a text in some approximate equivalent through efforts of human labour that go unnoticed (Keane and Nakassis 2026).
Instead of focusing on AI developers, this interest group focuses on everyday users. Japanese is at the centre of the group — a language in which honourifics, sentence-final particles, gendered speech and role language carry much of what speakers convey about who they are and where they stand in relation to others.
The group will meet every two weeks in the Ethnography Lab; working together, group members will compare their own hand translations against those of AI tools, material derived from a variety of sources, including (but not limited to) passages from anime, manga, short stories, and historical texts — each week housing its own theme. Over time, members may be asked to keep a short diary and discuss anonymized entries about moments when they chose AI over a friend for translation. As well, we will hear from translators, localizers, and researchers, and over the year work towards a public glossary of Japanese words and phrases that resist translation — placing human and machine renderings into comparison.
The group is most interested in how well machines transduce — not how well they translate — and in what people’s judgments of tone, politeness, and voice reveal about what they take translation, and themselves, to be. By human and machine renderings, the group hopes to give the Lab’s wider conversation about AI a view from the ordinary, multilingual settings of Toronto where these tools are used.