Jonathan Mirault, Bernard Lété, Jonathan Grainger, Johannes C Ziegler
We examined the developmental trajectory of transposed-word (TW) effects in a large sample of children, adolescents, and adults using a speeded grammatical decision task (is this sequence of words grammatically correct or not?). Child and adolescent participants were recruited from grades 3 to 9 in primary and early secondary education. TW effects were measured by comparing accuracy in the classification of two types of ungrammatical word sequences: (1) TW sequences, in which a grammatical sentence could be formed by transposing two words (e.g., "You that read wrong again"), and (2) matched control sequences (e.g., "You that read wrong sleep"), in which no grammatical sentence could be formed through transposition. We found that TW effects in error rates were already significant in grade 3. Importantly, the size of the TW effect increased significantly between grades 3 and 5 before remaining stable thereafter. We also observed a strong correlation, in a large cohort of children (N = 668), between reading fluency measured with a standardized reading test and response times (RTs) to grammatically correct sentences in the grammatical decision task. We interpret the observed developmental trajectory of TW effects as evidence for a gradual increase in the flexibility of word-order encoding as reading fluency increases. Furthermore, the present findings support the validity of the grammatical decision task as a tool for investigating sentence processing in both beginning and skilled readers.