Annotating information status in spontaneous speech

Abstract

We propose a fine-grained annotation scheme for the analysis of information status in spoken language, subdivided into a referential and a lexical level. The taxonomy is easy to handle and allows for a precise and comprehensive way of capturing the informational differences between nominal expressions in texts. First results on the prosodic marking of an item's information status in a small corpus of spontaneous monologues in German confirm the relevance of the two levels of description. There are significant effects of both levels of information status on (de)accentuation in isolation and in combination. However, results show more accents than postulated by our hypotheses.