Hey, Robot, can you move forward into that room? There are multiple dimensions of meaning to a single utterance like this that must be interpreted in order for a robot to respond appropriately. One dimension is the content of the instruction: here, this involves movement to a particular destination. Second, the robot must understand what the speaker is trying to do with this utterance in the conversational context and what that obligates the robot to do in response.  Is the speaker posing a question of ability that should be answered by supplying this information? Or is the speaker giving a politely phrased command that should just be executed by the robot? The appropriate response in this case is tied directly to a third dimension of meaning. The third dimension involves interpretation of this instruction with respect to the robot’s physical surroundings, including grounding the linguistic representation to the things around it—What room is being referred to? Is it possible for the robot to enter the room, or is it blocked in some way? This talk explores the use of symbolic representations to make each of these three dimensions of meaning accessible to autonomous systems participating in collaborative, situated natural language dialogue.