The systematic distortion hypothesis, which is based on a general cognitive bias contention for memory-based ratings of persons, does not account for functional differences between different types of linguistic form. A distinction between immediate terms (e.g., verbs, behaviors) and mediate terms (e.g., adjectives, traits) is offered. We argue that these linguistic forms fulfill different functions in the description of behaviors and persons. The hypothesis that the former terms differentiate features of persons in situations and are not affected by conceptual similarity, whereas the latter are primarily used to interpret persons and are organized by their conceptual relations, is supported in an experimental design in which we varied the type of target person orthogonally to type of situation. We show that whereas immediate terms differentiated for persons, situations, and persons in situations, their co-occurrence matrices are virtually unaffected by conceptual similarity. A strong conceptual similarity effect was found for mediate terms. Both sets of findings provide support for the functional distinction made between the two linguistic forms, and they suggest that the systematic effects produced for the systematic distortion hypothesis are a function of the types of terms rather than the product of a pervasive cognitive bias. The implications of the distinction between different linguistic forms for the person-situation debate and rating scale construction are discussed.

No abstract