Just as humans show consciousness of their body, social robots, in the way to be truly autonomous need to be aware of their body posture. Feasible gestures, moves and actions depend on the current body posture. The work developed in this paper aims to empirically show how self configuration recognition augments the degree of autonomy of a robot in the context of entertainment robotics. The integration of a classification tree for body posture identification based on data acquired from proprioceptive sensors of a NAO robot allows to interact with the robot in a more flexible and persistent manner. As a result, the robot shows a more sound behavior and greater degree of autonomy. Moreover, even if the body-awareness has been developed for minstrel robots, its application can be generalized to other contexts.