A New Approach to the Study of Verbal Concept Learning1

Abstract

A total of 71 adults were given a concept learning problem containing verbal statements as stimuli. A conjunctive‐category procedure was used with statements differing on 5 “dimensions,” defined as Noun‐Pronoun Subject, Gender of Subject, Tense of Verb, Sense of Verb and Locus of Activity. Performance was assessed as a function of (a) the combination of two dimensions chosen to be relevant (3 combinations) and (b) the number of irrelevant dimensions (1 versus 3). An overall analysis of variance indicated the interaction between these variables to be significant at the .05 level. Analyses of simple effects showed that, when 3 dimensions were irrelevant, variation in sentence meaning (Sense of Verb, Locus of Activity) was more easily detected as relevant than grammatical variation in a single word (Gender of Subject, Noun‐Pronoun Subject). Performance under 3 irrelevant dimensions was found to be superior to that under 1 irrelevant dimension for the relevant‐dimension combination Gender of Subject‐Sense of Verb, suggesting that the effects of irrelevant variation in the present task may be more complex than in a figural concept learning problem.