The priming of word associations: A levels-of-processing approach

Four experiments are reported in which the magnitude of the associative-priming effect is manipulated by varying both the experimental task and the conditions within a given task. The between-task manipulation involved a comparison of priming effects in length categorization on the one hand and lexical decision on the other. The within-task manipulation involved varying the type of letter strings presented as the nonword stimuli in lexical decisions. Pseudowords and two types of orthographically illegal letter strings—namely, consonant strings and random letter strings—served as nonwords. Finally, the effect of varying the proportion of pseudowords to illegal nonwords across four sets of experimental materials was investigated. On the whole, these within-and between-task manipulations turned out to be very effective ways of affecting the size of the associative-priming effect. Also, the size of the lexicality effect (the difference between the RTs to word and nonword stimuli) varied systematically with these within-and between-task manipulations. The results are discussed in relation to both Shulman and Davison's (1977) parallel-processing model and Balota and Chumbley's (1984) account of lexical decision.