The world’s many endangered languages are often characterized as “unwritten” and “oral” languages. Both of these terms reveal the language ideologies still implicit in many academic approaches to language: “unwritten” defines by negation, revealing a bias towards stable, standardized abstractions of communicative behaviour (away from a dynamic conception of situated talk-in-interaction); and “oral” defines by exclusion, revealing a bias towards the vocal-auditory channel (away from the multi-modal, fully embodied nature of face to face interaction). How much of our research today is unwittingly shaped by these implicit biases?