Semantic differential for the twenty-first century: scale relevance and uncertainty entering the semantic space

We propose an interval-valued version of the semantic differentiation method originally proposed by Osgood et al. (The measurement of meaning, University of Illinois Press, Chicago, 1957). The semantic differential is a tool for the extraction of attitudes of respondents towards given objects or of the connotative meaning of concepts. Semantic-differential-type scales are also frequently used in social-science research. The proposed generalisation of the original method is better suited for the reflection of perceived scale relevance and provides a possible solution to specific aspects of the concept–scale interaction issue and some other issues recently identified in the literature in connection with the use of semantic differential or semantic-differential-type scales. Lower appropriateness of scales as perceived by the respondents is translated into uncertainty regions and neutral answers can be distinguished from answers where the scale is perceived to be irrelevant. We suggest a modified data collection procedure and describe the calculation of the representation of the attitude towards an object as a point in the semantic space surrounded by an “uncertainty box”. The new method introduces uncertainty to the semantic space and allows for a more appropriate reflection of the meaning of concepts, words, etc. in formal models. No restrictions are introduced in terms of the availability of results—standard semantic-differential outputs including the position of objects in the semantic space and their semantic distance are available. The new method, however, reflects the uncertainty stemming from linguistic labels of the scale endpoints and from lower perceived appropriateness of the scales in the process.