Silent letters are activated in spoken word recognition

Four experiments are reported that investigate processing of mispronounced words for which the phonological form is inconsistent with the graphemic form (words spelled with silent letters). Words produced as mispronunciations that are consistent with their spelling were more confusable with their citation form counterpart than mispronunciations that are inconsistent with their spelling in a same/different task. Cross-modal repetition priming for orthographically supported productions and their citation form counterparts was equivalent; in contrast, orthographically unsupported productions showed reduced priming relative to the citation form. The findings are discussed in light of models of cross-modal interaction between spoken and written lexical representations. We argue that the results support a restructuring model where reading promotes development of a phonological representation used during spoken word recognition.