This article examines how the concept of a knowledge-base in language teacher education has changed since the 1998 proposal. Arguing that a knowledge-base evolves in two ways: through changes in the field of knowledge, and through changes driven by the work that knowledge supports, I describe two problems: 'translating' theory into practice and the 'positionality' of those defining what counts as knowledge. The 1998 proposal outlined a work-driven framework in response to the former without fully acknowledging the latter: who is doing English language teaching, with whom, and to what ends. Revising the knowledge-base now depends on taking that positionality into account. With this in mind, I suggest three concepts -- of teacher language use (English-for-Teaching), participation and agency, and professional confidence as a measure of outcome -- as work-driven alternatives to our present thinking.