Obligatory decomposition in reading prefixed words

Abstract

The reported experiment examines the impact of stem frequency on lexical decision responses to prefixed words. Both when the nonword distractors had nonsense stems (e.g., recodge) and real-word stems (e.g., relaugh), words with high frequency stems (e.g., unreal) were recognized more quickly than words with low frequency stems (e.g., refuel) when matched on surface frequency. This was taken as evidence that a whole-word representation exists for prefixed words, but that activation of this representation is always mediated by a representation of the stem, unlike the claims of a Dual Pathways model.