Adaptivity to the user can enhance ease and comfort in communication with a spoken language dialogue system. In this context adaptivity may mean being adaptive to the user's expertise with the system but also to his current emotional state and his situation. In an ideal case adaptivity also means switching the modality to support the user in the best way for communicating with the system. 

This thesis adresses especially the question how adaptive behaviour can be modeled in a dialogue system and which algorithms should guide the choice of modality and the way to render (predominantly acoustic) information to the user.