Shaping an enduring conversation on AI in applied linguistics

Extract

The whole issue of credibility (to humans) of machine output demands investigation. Important decisions increasingly tend to be made in response to computer output.
—Computational linguist Joseph Weizenbaum 1966 (p. 42)

AI in applied linguistics is nothing new. Long before, large language models became newsworthy curiosities, applied linguists were already using and theorizing intelligent systems in language education, linguistic research, and the early architecture of intelligent dialogic systems (e.g. Weizenbaum 1966; Higgins and Johns 1984; Underwood 1984; Anderson et al. 1985). What is new is the scale, the speed, and the stakes—and with them, renewed urgency around questions the field has long called its own: language as social practice, the mediation of meaning, the nature of communicative competence, and the relationship between linguistic resources and epistemic access.

These are not questions applied linguistics should answer reactively. AI is becoming a central object of inquiry in the field at a speed our scholarly infrastructure was not built to match, and the pertinent question is not how applied linguistics will respond to the phenomenon, but how it will extend its theoretical and methodological commitments to meet it.