List composition effects in lexical decision and recognition memory

Abstract

List composition effects are changes in subjects' performance on words of a given frequency as a function of the frequency of other words presented in an experiment. They are of considerable theoretical interest since they modify the effect of normative word frequency, generally one of the most powerful and reliable psycholinguistic variables. In particular, they furnish a critical test of theories of the word frequency effect in lexical decision since all current models predict that any list composition effects will be speed/accuracy trade-offs. In Experiment 1 it is shown that list composition effects in lexical decision are an independent phenomenon and cannot be interpreted as a result of speed/accurary trade-offs, nor, as some investigators have suggested, shifts in response bias. These effects are incompatible with all current models. A two-process model is developed which can account for these effects. Experiment 2 develops further support for these conclusions and also shows that list composition effects are also a phenomenon of recognition memory. Theoretical issues concerning the implications of list composition effects for models of lexical access are discussed. Methodological problems related to the use of lexical decision as a tool for the study of lexical access are also considered.