A DEVELOPMENTAL STUDY OF THE SEMANTIC DIFFERENTIAL

Abstract

ABSTRACTThe large literature on Osgood's Semantic Differential (SD) suggests that for adults there are three major dimensions in the domain of affective meaning–evaluation, potency, and activity. Since there have been no systematic, developmental studies of the SD, the main purpose of the present study was to explore the changes in the dimensions of affective meaning as a function of age. A second purpose was to obtain test‐retest reliabilities of the SD for childrenIn determining the dimensions of affective meaning for adults, some form of factor analysis was used. An SD study, however, has three sources of variance concepts, scales, and subjects–and until the recent development of techniques for three‐mode factor analysis, a complete analysis of SD data was not possible. In the three‐mode analysis, the factors characterizing the concept, scale, and subject modes are examined simultaneously along with the relationships among the factors from the three modes. Previously, investigators were limited to the dimensions of the scale mode and ignored the factor structures in the other two modes.The typical SD study involves having subjects rate a number of concepts (e.g. Mother) on a series of bipolar adjective scales (e.g., good‐bad). In the present study, the same 20 concepts (and two repeats) were rated on the same 28, 11‐point adjective scales by 96, 110, 107, and 100 parochial school children in grades three, four, six, and in high school. Both the adjectives and concepts were high frequency words.For the concept mode, the results showed that the same five factors appeared at each age level. These factors were tentatively labelled human (My Mother, Doctor), strength (Elephant, Army), agitation (Anger, Fear), softness (Clouds, Snow), and commonness (Street versus Ghost). The first four concept factors were very similar to those found in another SD study, which also used three‐mode factor analysis.For the scale mode, five, six, six, and seven factors appeared to be significant at the four respective age levels. Four of the scale factors were common to all age levels. These four were interpreted as evaluation (good‐bad), potency (heavy‐light), activity (active‐not active), and novelty (unusual‐usual). A fifth factor, stability (steady‐changing), appeared for all groups but the sixth‐grade. Scales which defined the stability factor in other grades had substantial loadings on other factors for grade six. The scale factors common across ages were all similar to ones that had previously appeared in studies with college students.For the subject mode, there appeared to be four subject factors for grade three, while there were three factors for the three older age groups. There was a strong tendency for the first factor to be dominant in the subject mode, and this tendency increased with age. This finding was interpreted as indicating that the subjects shared a common point of view toward the SD ratings and that, as a result of having a more common set of experiences as age increased, individual differences became even less noticeable for older subjects.With respect to test‐retest reliabilities, the reliabilities of the mean ratings were greater than .95 for all ages for both repeated concepts. Intra‐subject and scale reliabilities increased for older subjects.It was concluded that the primary dimensions of affective meaning for children as young as grade three children were very similar to those for adults. It was suggested that earlier studies, which found fewer dimensions for children than adults, were restricted by the small number of concepts and scales used rather than by any limitations on the judgmental capacities of the children. The findings were also related to relevant theories of cognitive development.