Abstract
Scaled a set of adjectives, a set of nouns, and a set of verbs applying J. Kruskal's (see pa, vol. 39:145 and 3167) multidimensional scaling method to rated dissimilarities in meaning. The same sets of concepts were unidimensionally scaled on qualities closely related to the 3 most salient semantic differential (sd) factors. The clearest finding was that semantic distance associated with judgments of dissimilarity is not euclidian. The evidence strongly suggests that in forming a dissimilarity judgment, differences are suppressed on all dimensions except the 1 that maximally discriminates the members of a concept pair. The term "maximum component distance" was coined to characterize the apparently best fitting distance model. All 3 sets of dissimilarities allowed the extraction of at least 4 dimensions. The noun and adjective dimensions predicted the sd scales rather well, but no sd scale appeared to be collinear with any dissimilarity dimension. In the case of the verbs, only the good-bad sd scale showed significant correlation with the dissimilarity dimensions. (27 ref.) (PsycINFO Database Record (c) 2016 APA, all rights reserved)