Beyond mean RTs in visual word recognition: Extensions of a remarkably stable three-way interaction amongst word...

Theoretically important insights in the visual word recognition literature have emerged from studying how different factors jointly affect lexical processing. Almost all joint effects are reflected in over-additive interactions, which are predicted by most current models. In contrast, additive effects pose profound challenges for these models. The present study explores the joint effects of stimulus quality and word frequency, which previously have been shown to robustly produce additive effects in lexical decision on mean response time (RT). Here, we assessed how the effects of these factors on RT distributions are moderated by nonword difficulty (pronounceable nonwords, e.g., FLIRP, vs. transposed letter nonwords, e.g., JUGDE from JUDGE). In the context of difficult transposed letter nonwords, there was an over-additive interaction between stimulus quality and word frequency in modal RTs that was opposed by an under-additive interaction between the same two factors for slower responses. In contrast, with standard pronounceable nonwords, the foregoing three-way interaction was eliminated, leaving additive effects of word frequency and stimulus quality across the RT distribution. These findings reinforce the importance of considering the interplay between task-specific and task-general processing in the context of RT distributions for current models, while demonstrating that complex empirical patterns are predictable and highly replicable in this domain.