Metadiscourse studies, inspired by Halliday's interpersonal metafunction of language, have dominated academic discourse analysis. They tended to exclude the ideational metafunction in discourse and to reduce the reader/analyst to a concordance programme that scans texts for lists of pre-established, but not always clearly defined, linguistic features. This paper does not purport to present an alternative to metadiscourse studies. It capitalises on their findings, rehabilitates propositional content and restores the role of the reader and the analyst. It takes the research paper as the central core genre in academic discourse (Schmied, 2014, p.11) and approaches it as a Quadric Writer-(Author-Author)-Reader Encounter. The paper proposes a model for the analysis of academic discourse. It unpacks Sinclair's (1986) Attribution, Averral dichotomy into an Attribution categories cline and an Averral categories cline and relates the categories to levels of Writer-Author Presence and Responsibility. It operationalises the Attribution categories and Averral categories it defines by delimiting the Discourse Unit constituting them. To achieve this delimitation, it upgrades Hunt's (1965) T-Unit and draws on Genette's (1997) Paratext. Because the paper targets academics, irrespective of their specialism, it chose to accompany the model it proposes by illustrative analyses of discourse samples from research papers dealing with issues in academic writing and publication.