Abstract: A picture-word interference paradigm was used to investigate the relative time course of phonological, orthographic, and semantic activation in the speech production of Chinese. Subjects were asked to name pictures onto which a Chinese character was superimposed either at the same time as picture presentation or 150 ms later. The superimposed character could be either homophonic to, or semantically related to, or orthographically similar to, or unrelated to the name of the picture. It was found that while homophonic characters and orthographically similar characters facilitated picture naming at both the two SOA conditions (0 ms and 150 ms), semantically related showed inhibitory effects on picture naming in both reaction time and error rate at the short SOA and only in error rate at the longer SOA. These findings were discussed in relation to the debates between modular and interactive theories of spoken word production

Key words: speech production, lemma selection, phonological encoding, picture naming