The Onset of the onset effect in reading aloud

Dutch native speakers read aloud target words (e.g., BANK) preceded by masked primes that were either whole words or letters followed by percent signs. In Experiment 1, the primes either matched or did not match the onset segment of the target (e.g., beer, heer, b%%% or h%%%); in Experiment 2, they had either the same or different onset structure as the target (e.g., beer, brug, be%% or br%%). The prime exposure duration of 33 ms and 66 ms were used. Effects of mismatching segments present in the whole-word but not in the letter primes led to slower overall naming latencies only at the longer prime exposure (66 ms), suggesting that interference from mismatching segments beyond the onset needs time to build up. Both effects of match/mismatch in onset segments and onset structure emerged surprisingly early, at the 33 ms prime duration. We suggest that the apparent early locus of these effects may be due to the early availability of the grapho-syllabic and graphemic segment information that facilitates a later phonological encoding process.