Semantic Differential

Development of the Semantic Differential Method

The origins of the semantic differential method can be traced to the pioneering work of Charles Osgood and his colleagues. Osgood was interested in understanding the meaning that people attached to words. He noted that words have both a denotative meaning and a connotative meaning. Denotative meaning refers to the literal dictionary definition of a word. Connotative meaning, on the other hand, refers to the associations of meaning that are attached to a word that are not, strictly speaking, part of its formal definition.

Osgood was interested in identifying the basic dimensions of connotative meaning underlying words. To explore this issue, he constructed pairs of opposing adjectives and then had respondents rate objects on these adjective pairs. Using factor analyses, he found that adjectives seemed to reflect one or more of the three basic underlying dimensions of connotative meaning mentioned above: evaluation, potency, and activity. Evaluation refers to the good-bad continuum of meaning underlying words. Potency reflects the strong-weak continuum. Activity represents the active-passive dimension. These dimensions of meaning have consistently been replicated, and analyses by Osgood and subsequent researchers have identified certain adjective pairs as highly representative of the evaluation (e.g., good-bad, valuable-worthless), potency (e.g., large-small, strong-weak), and activity (e.g., fast-slow, hot-cold) dimensions.

Constructing Semantic Differential Scales

Drawing on this research on connotative meaning, Osgood and others have proposed that semantic [Page 879]differential items (i.e., adjective pairs) can be used to construct scales assessing people's attitudes (i.e., evaluations) toward objects, as well as the extent to which people associate activity and potency with an object. A semantic differential scale typically consists of 4 to 10 items selected to assess the dimension of interest. Figure 1 presents an example of a semantic differential scale designed to assess attitudes toward vegetarianism.

When creating such scales, a researcher must keep several issues in mind. For instance, the researcher must ensure that the adjective pairs are strongly reflective of the dimension of interest and do not tap into unintended dimensions. Thus researchers often use published results of factor analyses of adjective pairs as a guide to item selection. Of course, if selections are to be based on past analyses, it is important to carefully consider whether the meaning of word pairings will hold constant across groups and be relevant to the population being sampled.

Item selection also requires the researcher to keep in mind that adjective pairs must be easily related to the object being rated. Some adjective pairs are quite generic and are likely applicable to nearly any object. For example, good-bad could probably be sensibly applied to almost any physical object, person, social group, or concept. Other adjective pairs may reflect somewhat more specific meaning and thus be less easily applied to some objects than others. For instance, beautiful-ugly is highly evaluative in content and could quite sensibly be applied to a person (e.g., one's spouse) and many physical objects (e.g., a car, a work of art). This pair might be less meaningfully applied to other objects or concepts (e.g., magnetic resonance imaging, social security).

Figure 1 Sample of Semantic Differential Scale Measuring Attitudes Toward

...