The stimulus familiarity problem in semantic memory research

Abstract

Much of the research in the area of semantic memory is concerned with determining what characteristics of words (e.g., category size) or relationships between words (e.g., instance—category relatedness) influence reaction time (RT) in sentence verification tasks. Research of this sort is essentially correlational, so that it is impossible to establish with any degree of certainty that a particular result is due to a particular independent variable, rather than to some extraneous variable confounded with it. The present study suggests that in several recent semantic memory experiments, the independent variables of interest have been confounded with the familiarity of the stimulus words, and that the RT results obtained in these experiments may reflect the effects of stimulus familiarity rather than, or in addition to, the effects of the explicitly manipulated variables. The first part of the paper discusses familiarity problems in connection with an experiment by E. E. Smith, E. J. Shoben, and L. J. Rips ( Psychological Review , 1974, 81, 214–241) which suggests that verification RT is determined not by category size, but by instance—category relatedness. It is argued that (1) relatedness was assessed inappropriately in the Smith et al. experiment; (2) category size and relatedness were confounded with the familiarity of the stimulus words; and (3) the familiarity confound played an important role in producing the pattern of results obtained by Smith et al. The second part of the paper considers the question of whether familiarity confounds are prevalent in semantic memory research. Recent experiments by R. F. Lorch, Jr. ( Journal of Verbal Learning and Verbal Behavior , 1978, 17 , 523–537) and M. McCloskey and S. Glucksberg ( Cognitive Psychology , 1979, 11, 1–37) are shown to have familiarity confounds, suggesting that such confounds may be rather widespread. Finally, the implications of the present findings for current issues and future research are discussed.