Action is the basis of theater(Langer 1953) and, as such, needs to be fully incorporated in what ever model a computer is running during a computer based theatrical performance. We believe the lack of good models for action is one fundamental reason for the relative absence of experiments involving theater and computers. The attempts to wire up stages or performers have been in general concerned with dance(Lovell~Mitchell 1995), only using information about the position and attitude of the actors/dancers on the stage. The main argument of this study is that computer theater not only requires action representation and recognition but it is also an interesting domain for action research.
To support our argument we begin by examining the multiple possibilities of using computers in theatrical performances, concerning both explored and unexplored developments. Recent theatrical experiences are prefered for citation rather than old ones in order to draw a picture of the current research. Some attempts to represent and recognize actions are examined in the second part of the study, and, particularly, the work we are doing in using and recognizing action information from scripts (Pinhanez & Bobiek 1996).