This study examines the role of phonological information in the visual recognition of written words. The term visual word recognition is used here to refer to the process of word-form identification as opposed to meaning retrieval (Monsell, Doyle & Haggard, 1989). Meaning retrieval implies word-form identification, however, not vice versa.

Phonological information could be generated from the visually presented word in two ways. First, the word is recognized as a particular orthographic pattern and then its pronunciation retrieved from memory (addressed phonology). Second, phonological information concerning the word can be generated via spelling–sound correspondences during the recognition process before identification is complete, which represents an instance of assembled phonology (Patterson & Coltheart, 1987). The assembled phonology route necessarily assumes that component letters or graphemes be identified to an extent that allows spelling–sound correspondence rules to be applicable. Evidence for phonological mediation naturally consists of demonstrating that the word's graphemic units activate phonemic units earlier than the word is recognized. Several methods have been used to claim evidence for phonological mediation.

Paradigms used for studying phonological mediation

Two paradigms have generally been used to provide evidence for, or against, phonological mediation in English and Chinese. They are the priming paradigm and semantic categorization paradigm.

Priming paradigm

There are English and Chinese studies using the priming paradigm. Although the priming paradigm used in Chinese studies differs in some details from the priming paradigm used in English studies, both studies aimed at showing that phonological activation is very early.