The present study examined the influence of a human or computer "partner" on the production of fillers ("um" and "uh") during a question and answer task. Experiment 1 investigated whether or not responding to a human partner as opposed to a computer partner results in a higher rate of filler production. Participants produced many more fillers when responding to a human. Experiment 2 tested the possibility that this large effect was driven by the mere presence of another person. It was not. There was, however, a small effect of human presence on fillers, a novel result. That individuals modulate their filler use in response to the nature of their speech partner is a critical piece of evidence in favor of the filler-for-partner hypothesis. Collectively, our data provide convergent support for the theoretical position that fillers are not solely produced as the result of difficulties in speech planning or production, but they also play a functional role in the communicative interaction between speakers and listeners.