Anticipation in speech-based human-machine interfaces

This work is devoted to the phenomenon of the anticipation, which is made by human interlocuters of dialog during listening. People are often able to anticipate next constituents of the unfold utterance just uttered by the opposite interlocutor of the dialog. It plays an important role e.g. in the task of simultaneous interpretation. The proposed paper is an initial study focused on this phenomenon from human-machine communication point of view. It tries to discuss aspects, applications and benefits of considering anticipation on the side of human as well as machine interlocutor of dialog. Possibility of extending machines with anticipation capability is discussed for each element of human-machine communication chain. Anticipation closely relates to sentence comprehension, mutual understanding and turn-taking. The paper proposes ideas how the rapid turn-taking can be increased by considering human anticipation capability and by involving machine anticipation techniques. Moreover it brings ideas how anticipation capability can be integrated for particular modules of the human-machine speech communication chain.