Are We All Disfluent in Our Own Special Way and Should Dialogue Systems Also Be?

This study explores interand intra-speaker variation in use of timemanagement strategies. How do speakers differ in their use of pauses, fillers and other resources aimed at managing time while they plan their next contribution? Taking a rather qualitative approach, we describe individual speakers’ production, using a small limited-domain corpus of task-oriented German speech. It is assumed that spoken dialogue systems can benefit from mimicking individual speaking styles. This study is a first step in that direction, aiming to describe in detail speaker-specific productions of selected time-buying disfluencies for later use in synthesis.