A quantitative method for separation of semantic subspaces

Abstract

A new method for separating affective and denotative meaning subsystems in semantic differential ratings of any homogeneous concept domain is developed and illustrated using personality ratings data. Results indicated that Osgood’s Evaluation, Potency and Activity dimensions were dominant in the underlying semantic structure of personality conceptions and that four dimensions in the "other" subspace, orthogonal to Affect, were clearly interpretable, "affect-free" descriptive features of personalities. Possible applications of this model to other social and psychological research are discussed.