This paper argues that high segmentation in Japanese conversation adds to interactional turn co-construction so as to emphasize specifically a 'move' within the larger action involved by a turn constructional unit. Thus local contingency problems can be solved without altering the global level of the turn's pragmatic work.