The use of Generative AI to formulate text has thrown up a number of thorny questions concerning the position of authors and publishers in scholarship. To whom/what should a scholarly work be credited? Who has copyright? Who is responsible for the claims made therein? Who can be approached for clarification or (further) evidence of these? In short, where does accountability lie? This talk addresses these issues by applying and developing the concept of production format, a framework first devised half a century ago by Erving Goffman. In one part of a 1979/1981 paper, this sociologist deconstructs the timeworn notion of the 'speaker' of a piece of language into three distinct roles (which he terms 'animator', 'author' and 'principal'). With only a little tweaking, the framework can be readily applied to written language, and with a little modification can be brought to bear on scholarly publishing, offering solutions to the thorny questions raised above.