We present a protocol to explore the relative activation sequence of phonology and semantics in visual word recognition. The results show that consistent with interactive accounts, semantic and phonological representations may be processed interactively, and higher-level linguistic representations may affect early processing.