In face-to-face communication, interlocutors make use of the space between them in various respects. They may look into each other's eyes to initiate or reestablish communication, direct attention by gaze and pointing gestures to objects in the situation (e.g., on the table), or use iconic gestures to describe shapes and orientations of objects or spatial layouts while giving spoken explanations. A particular observation is that interlocutors are likely to mutually align by sharing their gesture space with that of the other (i.e. the space in front of a person’s body in which hand and arm movements are carried out while speaking), while they intimately converse about individual spatial features of joint current interest. In a scene where interlocutors are proximate to each other and their gesture spaces form a shared "interaction space", this can result in an alignment of the individual ways of referring to physical or even imaginary objects. These insights about the alignment in humans are adopted to enhance human machine interaction. One especially interesting process of alignment can be observed in gesture imitation. Alignment by imitation will be comparatively investigated on two levels of gestural interaction between a human and an artificial interlocutor: On the mimicking level only essential form properties of the human interlocutor’s gesture will be imitated, whereas on the meaning level imitation is based on an understanding of the human interlocutor's gesture. Gesture imitation will be studied in two settings with different kinds of artificial interlocutors, the virtual embodied agent Max and the robotic humanoid torso Barthoc.