Examining word processing via a megastudy of conditional reading aloud

Abstract

Results from a megastudy on conditional reading aloud for 2,145 monosyllabic words are reported. In stepwise regression analyses, the predictor variables accounted for over 66% of the reaction time (RT) variance. Linear mixed effect modelling on log RT indicated that every variable that related to RT in either reading aloud or lexical decision also related to RT in conditional reading aloud. Notably, differences among tasks were observed. Specifically, lexical decision showed stronger reliance on semantic information than the other two tasks, but conditional reading aloud also showed strong reliance on semantic information. Interestingly, feedback consistency affected reading aloud and conditional reading but not lexical decision. Pairwise correlations revealed that conditional reading aloud performance showed moderately strong relationships to lexical decision and reading aloud performance, whereas reading aloud and lexical decision performance were weakly related to each other. Conditional reading aloud produces reliable data that can be used to examine word processing. Theoretical challenges moving forward include how to best conceptualise and model processes involved in this task.