Currently no standardised gesture annotation systems are available. As a contribution towards solving this problem, CoGesT, a machine processable and human usable computational model for the annotation of a subset of conversational gestures is presented, its empirical and formal properties are detailed, and application areas are discussed.