Abstract: A homophone judgment task (Experiment 1) and a syllable monitoring task were used to investigate the sequentiality of phonological activation in the speech production of Chinese disyllabic compound words. In these tasks, a picture with a disyllabic compound name was presented, followed (Experiment 1) or preceded (Experiment 2) by a Chinese character homophonic to one of the constituent morphemes. The SOA between presentation of the picture and of the character was 50 ms for Experiment 1 and 1300 ms for Experiment 2. Subjects were asked to judge whether a morpheme in the picture name was homophonic to the character. If phonological activation of compound names is conducted sequentially, from left to right, the reaction time for the first constituent morphemes should be shorter than the time for the second morphemes. In both experiments, however, the reaction time for the second constituents was shorter than for the first constituent, and the response error rate was also lower for the first than for the second, in contradiction to predictions of most existing theories of speech production. These findings were interpreted as reflecting the effect of semantic activation of constituent morphemes on phonological activation. Critical compound words used here had a structure in which the first constituents modify the second constituents, with the second constituents as lexical head and determining the semantic and syntactic properties of the whole words. We propose that phonological activation of constituent morphemes in the speech production of Chinese compound words depends on the efficiency of semantic activation of constituent morphemes, but not on the fixed order between morphemes.

Key words: speech production, phonological encoding, sequential processing, syllable monitoring, homophone judgment