Harris (1952) wrote in his starting point of discourse studies , that the two aims of this subdiscipline are continuing descriptive linguistics beyond the limits of a single sentence and correlating culture and language (i.e. non-linguistic and linguistic behaviour) . This project deals with the first aim the links between sentences by exploring in more depth what makes a sequence of sentences a discourse. Within SFL, RST and the cognitive approach many attempts have been made to account for what is called connectivity (the cover term for cohesion and coherence). Starting within this framework, I want to propose two principles on which a connectivity theory can be based: the discursive and dialogic principle. Then I will propose ways to fill in this framework with an account of connectivity from three points of view: 1. Conjunction, in which the focus is on combining clauses, the building blocks of discourse; 2. Adjunction, which deals with the content of these building blocks, the information they contain, and how information is connected; 3. Interjunction, or how the combined pieces of information function between writer and reader, or speaker and listener. This approach will be summarized in a connectivity model. This model is the base for solving problems in discourse relation research as discussed in Renkema (2004, 2006) and Taboada & Mann (2006). The most important problems are: the architecture of proposed taxonomies; the subjectivity of labelling relations and structure representations in trees (multilevel and multiparent phenomena), the accounts for the effects of signalling and the expectations prompted by formal varieties of discourse relations. The phenomena under consideration will be discussed using a reference corpus to be composed in collaboration between Tilburg University and Simon Fraser University.