This research is an in-depth investigation into typicality, the construct which has best characterized semantic distance. After establishing the validity of a ratio level scale for rated typicality, the interrelationship among various surrogate measures of typicality were examined.; Examination of the familarity based accounts of typicality indicated that frequency of encounter only has minimal influence on rated typicality. Amount known about an item, indexed by total number of properties, demonstrated a variable relationship.; Examination of the three general property classes, individual, distinctive and common, indicated that distinctive and common properties, weighted for salience, together account for substantial variance in rated typicality.; Investigations into the versions of typicality based upon exemplar-exemplar overlap and exemplar-category overlap indicated that the two measures were highly intercorrelated. Further, it appeared that the differences between correlations across categories for these measures reflects the type of category, basic or superordinate, and its possible representation.; Derived measures of dissimilarity and distinctive features, based upon additive scaling solutions of property based similarity and overlap tables, indicated that rated typicality did not substantially correlate with the derived measures.; The results taken together provide substantial support for the interrelationship of the different measures of typicality. Limitations arise when rated typicality is interpreted as reflecting detailed property structure. It is suggested that semantic distance is best represented by the tables of similarity and overlap. The methodological implications and new research areas made possible by the ratio scale values of typicality are also discussed.
展开▼
