Footing in human-robot conversations: how robots might shape participant roles using gaze cues