ABSTRACT
Allport and Odbert’s (1936) compilation and categorization of nearly 18,000 person descriptive terms still provides the basis for the lexical approach to personality assessment. Only about 25% of that list, the “real,” stable personality traits, were used by Cattell (1943) to define the “trait sphere.” Terms referring to temporary states, social roles as well as highly evaluative terms, were excluded from further analyses. Norman (1967) grouped 18,125 descriptive terms into his domains of personality containing 15 categories. The first three categories, “Stable ‘Biophysical’ Traits,” accounted for 15% of the terms and were further reduced to 1,431 (8%) to construct a hierarchical classification system. The German lexical study (Angleitner, Ostendorf, & John, 1990; Ostendorf, 1990) was based on an initial set of 5,092 terms, but only 430 terms finally qualified as Dispositionen. A similar reduction took place for the construction of the Dutch Abridged Big Five Circumplex Model (AB5C) taxonomy, beginning with 8,690 terms culled from the dictionary by Brokken (1978) and ending with 551 AB5C adjectives (Hofstee & De Raad, 1991). The major reason for significantly reducing the number of terms was to guarantee that they reflected stable traits and to avoid ambiguous and
evaluative words. Further, this emphasis on stable traits has led most lexical researchers to limit the person-descriptive terms only to trait adjectives and to exclude nouns and verbs. One effect of this strategy has been that less stable indicators of personality such as interests, values, and various attitudes have been underrepresented in measures of the Big Five. This reduction is unlikely to affect the content of the other Big Five factors that explain most of the variance, but it is perhaps not so surprising that the Openness-Intellect factor’s content has been less stable because of differences in the way the universe of trait-descriptive terms has been typically defined and sampled.
