Two female bottlenose dolphins (Tursiops truncatus) and their 2 male offspring were presented with an underwater keyboard to observe how the dolphins would use such a system to obtain specific objects and activities. When a dolphin pressed visual forms on the keyboard, whistles were generated underwater, and the dolphin was given a specific object or activity. Both vocal and nonvocal behaviors were recorded. Only the males used the keyboard. In the 1st year spontaneous vocal mimicry and productive use of facsimiles of the computer-generated whistles were recorded. In the 2nd year productive use increased significantly over mimicry, and apparent combinations of discreet whistle facsimiles in behaviorally appropriate contexts were observed. The patterns of vocal mimicry and production suggest a new model for analyzing dolphin vocalizations and vocal development with respect to signal structure and organization.