Reexamining Transposed-Word Effects in the Grammatical Decision Task

Transposed-word (TW) effects have been repeatedly observed when comparing responses to two types of ungrammatical sequence of words in a grammatical decision task: 1) transposed-word sequences formed by changing the order of two words in a grammatically correct sentence (e.g., The white was cat big); and 2) ungrammatical control sequences that cannot be transformed into a correct sentence by transposing any two words (e.g., The white was cat slowly). When the task is speeded grammatical decision, and when all words are presented simultaneously in a format that is typical of regular reading, TW effects are observed in both response times (RTs) and error rates. Specifically, responses to the TW sequences are slower and less accurate than to the matched control sequences. However, when written words are presented sequentially, one after the other, then TW effects typically only emerge in error rates. Finally, when the word sequences are presented auditorily, TW effects are significant in both RTs and accuracy. Building on the meta-analysis of TW effects described in Wen and Grainger (2025a), here we present a more detailed analysis of this general pattern and provide a tentative explanation for the observed differences across presentation procedure and modality.