You can't Stroop a lexical decision: is semantic processing fundamentally facilitative?

It is well documented that related prime words facilitate target processing in lexical decision (e.g., doctor facilitates nurse), but interfere with target processing in the Stroop task (e.g., the word blue slows the time to name the colour red). Five experiments explored several potential explanations for these differences. In Experiments 1 and 2, all stimuli were novel (as in a typical lexical decision design). Participants were faster both to make lexical decisions and to read colour words aloud that were primed by incongruent associates (e.g., banana) relative to a neutral prime (e.g., knot). Experiments 3 and 4 used a small set of repeatedly presented stimuli (as in a typical Stroop design). Incongruent colour words facilitated lexical decisions to target colour words, but interfered with identification (reading aloud). Experiment 5 further showed that interference is still observed in identification when the distractor set size is large but the target/response set size is small. These findings suggest that semantic connections are solely facilitative and that response competition only occurs when there is a small set of repeated responses and identification (rather than lexical decision) is required. The more general problem of research fragmentation is briefly discussed.