Annotation Guidelines for Questions under Discussion and Information Structure

We present a pragmatic, i.e. meaning-based, method for the information-structural analysis of corpus data, which is built on the idea that for any assertion contained in a text (or transcript of spoken discourse) there is an implicit Question under Discussion (QUD) that determines which parts of the assertion are focused or backgrounded (and which ones are not-at-issue, i.e. not part of the assertion at all). We formulate a number of constraints which allow the analyst/annotator to derive QUDs from the previous or upcoming discourse context and demonstrate the method using corpus examples (of French, German and English). Since we avoid making reference to language-specific morphosyntactic or prosodic properties, we claim that our method is also cross-linguistically applicable beyond our example languages.