Effects of Semantic Transparency on the Recognition of Chinese Two-character Words: Evidence for a Dual-process Model

This study investigated the representations and recognition processes of Chinese two-character words. In Experiment 1, subjects made lexical decisions to two-character compounds. Semantic transparency effects were found that the reaction times (RTs) for transparent compounds were shorter than those for opaque compounds. Word frequency effects were found that the RTs for high-frequency compounds were shorter than those for low-frequency compounds. The effect of semantic transparency was larger for low-frequency compounds than for high-frequency compounds. Experiment 2 was similar to Experiment 1, except for a 100 ms stimulus onset asynchrony (SOA) between two characters. There were still significant effects of semantic transparency and word frequency. The RTs for compounds composed of low-frequency characters were shorter than the RTs for those composed of high-frequency characters. There were no significant interactions. In Experiment 3 the SOA was reduced to 50 ms. The results were similar to those of Experiment 2. In Experiment 4, subjects made lexical decisions to two-character phonetic translations and loan words, and compounds with three levels of semantic transparency. The RTs for phonetic translations and loan words were shorter than the RTs for the compounds. The RTs were longest for the initial-opaque condition in which the dominant meaning of the first character was unrelated to the meaning of a compound. In Experiment 5 with 100 ms SOA between characters, the superiority for phonetic translations and loan words disappeared. These results support a dual-processing model for the recognition of multi-character Chinese words.