How an unfamiliar thing should be called

An empirical method is described to derive good names for unfamiliar objects. It is based on three principles: (1) The names should be within users' linguistic capacities; (2) names should be informationally efficient; (3) names should form a classification system. The principles lead to a three-step method: (1) Subjects generate names for the objects; (2) a subset of the names, which fulfills the principles, is selected; (3) how good the names are is tested by matching and recall tasks. Steps 2 and 3 are iterated to improve the names. The names that result are natural, short, easily matched with their physical referents, and well recalled. The method is generalizable and ought to be useful in a large variety of situations where names for unfamiliar objects are needed.