Abstract
Topic of the programme is the hierarchical organisation of textual units into structured entities by the joint contributions of relational coherence (the way smaller textual units combine into larger ones) and lexical cohesion (relations between elements within textual units, e.g., synonyms, anaphors, or collocations). Corpora from different genres (journalistic, encyclopaedic, and persuasive texts) will be investigated to take into account genre-dependent differences in textual organisation. While most of the work on relational coherence centres round local coherence relations and their signalling, the first project focusses on the configuration of discourse that relational coherence imposes on complex text segments. The hypothesis to be investigated in this project is that binary trees suffice to model the configuration of the chosen corpora. Seeming counterexamples are to be explained in terms of interacting factors, in particular, lexical cohesive devices. The second project investigates the interaction between relational coherence and lexical cohesion by looking at their predictions concerning the paragraph structure and the centrality of discourse units. While we expect a fair amount of correlation between the two predictions in all genres investigated in the programme, we hypothesize that lexical cohesion will be particularly closely aligned with discourse structure in expository texts, which are mainly thematically organised.