In two experiments, semantic facilitation and translation priming effects in Chinese-English bilingual speakers were demonstrated with a lexical decision task. A 300-msec stimulus-onset asynchrony (SOA)was used between display ofthe prime and the target item. Experiment 1 showed that subjects' lexical decision responses were facilitated to a greater extent when primed by a translation equivalent than a semantically related between-language word. In Experiment 2, we found that pictorial, between-language, and within-language primes produced comparable effects of semantic facilitation. These results are in line with the hypothesis that lexical items in different languages and pictures are processed by means of an amodal conceptual system. In recent years, there has been an increasing interest in bilingual or multilingual processing (see, e.g., Grosjean, 1982; Hornby, 1977;Snodgrass, 1984;Vaid, 1986), not only because learning to use more than one language is very common in real life, but also because it involves very complex cognitive activities. A fundamental research topic in this area concerns lexical processing by bilingual subjects. This topic is related to important theoretical issues of the nature of the lexicon(s) of the bilingual speaker and the relationship between the internal processing of a concept and its surface form.Recently, Kirsner, Smith, Lockhart, King, and Jain (1984) conducted a series of experiments to test different models of lexical representation and processing in bilingual speakers. The three major models tested were the word-association model, the word-interconnection model, and the concept-mediation model. lllustrations of these three models are shown in Figure 1. The word-association model proposes that translation equivalents in the two languages of a bilingual speaker are directly connected to each other. Semantically related words in a given language are also directly linked. However, semantically related words in the two languages are not directly connected, but rather indirectly connected through corresponding translation equivalents. The word-interconnection model postulates that all words in the two languages, including both translation equivalents and semantically related words, are directly connected to each other. The conceptmediation model, in contrast to the other two models, assumes that neither within-language nor between-language words are directly connected; rather, they are all linked