Subjective frequency and the lexical decision latency function: Implications for mechanisms of lexical access

Subjective word frequencies were obtained in order to determine the relationship between frequency and lexical decision times across a wide range of frequencies. The subjective ratings proved to be reliable across subject groups, even though they showed somewhat paradoxical range effects. While there were good correlations between the subjective and standard objective frequency counts across the entire frequency range, there was relatively little correlation between them for low-frequency words. Nevertheless, the subjective ratings accounted for more of the decision time variance for these low-frequency words than did the objective counts. Subjective frequency was therefore felt to be more appropriate for determining the frequency-latency relationship in lexical decision, particularly for low-frequency words. Using these ratings to supplement frequency counts, together with data from several lexical decision experiments, shows that the relationship between decision time and frequency is nonlinear, asymptoting for the highest frequencies, but with rapidly increasing reaction times for lower ones. B. Gordon's (1981, Lexical Access and Lexical Decision: Mechanisms of Frequency Sensitivity. Unpublished doctoral dissertation, Johns Hopkins University; 1983, Journal of Verbal Learning and Verbal Behavior, 22, 24-44) resonance model of lexical access can directly account for this relationship, but logogen and sequential search models require further elaboration to do so.