Biomedical applications offer an exciting growth opportunity for wireless sensor networks. However, radio frequency communication is problematic in hospital environments that are susceptible to interference in the industrial, scientific, and medical (ISM) bands. Also, RF is inherently insecure as eavesdroppers can easily pick up signals. The Intra-Body Network (IBNet) proposes a novel communication model for biomedical sensor networks. It seeks the convenience of wireless communication while avoiding interference and privacy concerns associated with RF. IBNet's solution is to utilize a subject's own body tissue as a transmission medium. Assuming that transmissions are contained within the body, IBNet solves otherwise complex problems of privacy and interference. Unfortunately, transmitting through the same medium in which we sense creates a new type of conflict; it is possible that one sensor's network transmission might corrupt an adjacent sensor's sample data. We present Body Language, a set of protocols that arbitrate IBNet's sampling/communication conflict while providing basic services such as dynamic node discovery, network configuration, quality of service, and sensor sample collection. Body Language seeks to provide these services and solve IBNet's unique communication challenges while minimizing hardware resource requirements and hence sensor node cost. In order to prove Body Language feasibility, we created an IBNet prototype environment where the protocols were demonstrated on real hardware and in real time. The prototype also offers important insight into the Body Language's computational resource requirements. Our results show that Body Language provides all services required by IBNet and it does so with a very modest footprint.