In laboratory studies of vocal behavior in animals, subjects are normally isolated in a sound-insulated chamber for recording, but such socially isolated conditions may reduce the chances that they will vocalize. Indeed, past studies using such methods have faced the challenge that subjects remained silent. Knowledge of conditions under which subjects are more likely to vocalize could thus improve experimental design. This study investigated (a) whether kea (Nestor notabilis) could be trained to increase vocal production using operant conditioning and (b) the conditions under which such training was feasible. We found that visual contact with other kea increased the chances that a subject would vocalize spontaneously, therefore making training through positive reinforcement possible. In the conditions where kea could only hear but not see the rest of the group, they were much less likely to vocalize. Subjects were quickly trained to increase vocal production while in visual contact with other kea, and the training remained effective even when visual access later was removed. The procedure described here could be used as a first step in future laboratory studies of vocal behavior, carried out before subjects are isolated, to overcome the challenge of inducing isolated subjects to vocalize.