Dissociating morphological and form priming with novel complex word primes: Evidence from masked priming, overt priming, and event-related potentials

Recent research suggests that visually-presented words are initially morphologically segmented whenever the letter-string can be exhaustively assigned to existing morphological representations, but not when an exhaustive parse is unavailable; e.g., priming is observed for both hunter → HUNT and brother → BROTH, but not for brothel → BROTH. Few studies have investigated whether this pattern extends to novel complex words, and the results to date (all from novel suffixed words) are mixed. In the current study, we examine whether novel compounds (drugrack → RACK) yield morphological priming which is dissociable from that in novel pseudoembedded words (slegrack → RACK). Using masked priming, we find significant and comparable priming in reaction times for word-final elements of both novel compounds and novel pseudoembedded words. Using overt priming, however, we find greater priming effects (in both reaction times and N400 amplitudes) for novel compounds compared to novel pseudoembedded words. These results are consistent with models assuming across-the-board activation of putative constituents, while also suggesting that morpheme activation may persevere despite the lack of an exhaustive morpheme-based parse when an exhaustive monomorphemic analysis is also unavailable. These findings highlight the critical role of the lexical status of the pseudoembedded prime in dissociating morphological and orthographic priming.