ABSTRACT
A multitude of other conclusions that have nothing whatsoever to do with morphology could, of course, be advanced regarding these findings-and this is perhaps part of the reason that consideration of polymorphemic words in explicit
(computational) theories of reading has been slow in coming. For example, Forster and Azuma (2000) have argued that nonlinguistic factors could play an important role in studies examining stem and surface frequency. Specifically, knowledge about morphological relationships-not those relationships themselves-may be implicated in a lexical decision mechanism. Similarly, it could be argued that priming studies may be contaminated by episodic and/or strategic factors; or indeed, may reflect types of relatedness having nothing to do with morphology, but that are typical of morphological relatives (e.g., semantic relatedness). For all of these reasons, the introduction of the masked priming technique (Forster & Davis, 1984; Forster, Davis, Schoknecht, & Carter, 1987) to the problem of morphological processing in visual word recognition has been particularly important. Because conscious appreciation of the prime can be eliminated through masking, this technique may offer a glimpse onto the word recognition system that is free of both the episodic and strategic factors that can contaminate longer-lag priming techniques (but see Bodner & Masson, 1997; Masson & Bodner, this volume) and the nonlinguistic factors that may play a role in studies of stem and surface frequency effects in the unprimed lexical decision paradigm.
