Aphasic disorders in matching tasks involving conceptual analysis and covert naming.

Aphasic, non-aphasic brain-damaged, schizophrenic and normal Ss were required to select which of two pictures in each item was indicated by a third picture. The latter was related to the referent not through direct associations but only through a mediator: a homonym -- the name for clue and referent --, a situational association, or a perceptual feature common to both. The availability of these mediators was to be tested. Aphasics were as good as normals when the task could be solved via situational associations (e.g. guitar/violin -- bullfight; mediator: "Spain"). They showed poorer performance than all the control-groups not only when the solution depended on the availability of a word but also when it was necessary to identify analytically a perceptual feature characteristic for clue and referent (e.g. swan/turkey -- snowman; mediator: "white"). This result is seen as possibly corresponding to the specific deficiency of both non-fluent and fluent aphasics in the Token Test which might be characteristic of left hemisphere dysfunctions in the analytical decomposition into separate features of what presents itself perceptually and experientally as a unit or Gestalt.