multidimensional scaling

A statistical technique which has been applied in PSYCHOLINGUISTICS to quantify the MEANINGS of related LEXICAL ITEMS and define the SEMANTIC space within which these items work. INFORMANTS rate numerically a set of items (e.g. kinship terms, colors) in terms of their mutual similarity; the more similar the average ratings are, the closer these items are placed in the hypothetical space. In this inductive manner, it is hoped to establish classificatory criteria for lexical sets which might otherwise not emerge, and to develop more illuminating models of semantic STRUCTURE than are available using conventional analytic techniques. The technique is also used in PHONETICS, along with other statistical methods.