Research on spatial information processing through the intermediate semantic representation is still a controversial problem. The present studies discussed whether language’s proficiency could influence bilinguals’ representation of spatial information. In both studies, we adopt the Proctor’s paradigm, which include inducer task and diagnostic task. For the inducer task, participants were instructed to make the keypress opposite to the meaning of the location words LEFT and RIGHT. For the diagnostic task, participants were instructed to make the keypress response assigned to the color of the circles, while ignoring the stimulus location. In experiment 1, 25 monolingual Chinese speakers and 25 blingual Mongolian/Chinese speakers participated the experiment. All participants completed working memory test and self-reporting questionaire of language proficiency. There is no significant difference in measure of working memory between bilinguals and monolinguals, as well language proficiency between Chinese and Mongolian is no difference in bilinguals. All programs were run on the standard computers. The location words were presented at the center of screen. The green and red circles were presented equally in random order on the right or left side. For inducer task, participants were asked to press“Z” or “/” key opposite to the location words’ meaning. For diagnostic task, participants were asked to press“Z” or “/” key according to the circles’ color while ignoring the location. Each participant performed two parts including Chinese and Mongolian language. The reaction time and accurate rates were analyzed by 2(language: words in Chinese or Mongolian)×2 (group: bilinguals or monolinguals)×2 (correspondence: corresponding or non-corresponding)×4 (block:1-4) ANOVAs. The results showed that there was a three-way interaction of language×correspondence×group for RT（F（1,48）=9.01, p＜ .01,η2 = .16）. For monolinguals, the Simon effect reversed when location words were Chinese, but not when the words were Mongolian. For bilinguals, the Simon effect influenced by Chinese and Mongolian didn’t differ significantly（t(24)= 0.37, p > .05）. In experiment 2, the method and procedure were similar to experiment 1, except that the participants were only trilinguals and participants were asked to perform unsymmetrical inducer task in Chinese, Mongolian and English. The reaction time and accurate rates were analyzed by 3(language: words in Chinese ,Mongolian or English)×2 (correspondence: corresponding or non-corresponding)×4 (block:1-4) ANOVAs. The results showed that there was a two-way interaction of language×correspondence for RT（F（2,38）=5.01，p＜ .05,η2 = .20）. The Simon effect disappeared when they were mixed with Chinese words. The Simon effect reversed when they were mixed with Mongolian words. The Simon effect weakened when they were mixed with English words. We found that the influence of Mongolian was similar to Chinese（t(19)= 0.41，p > .05）but different from English. In conclusion，these findings indicate that the proficiency of language could influence the connection strength with the shared semantic representation. The results support the notion of The Revised Hierarchical Model of Bilingual Memory Representation. Future research should be focused on the influence of second language’s using experience on the semantic representation .