中文摘要：

采用跨语言长时竞争启动范式探讨汉语为二语学习者言语产生中的跨语言干扰。实验1的被试选择18名俄语为母语者，实验2选择18名日语为母语者。自变量是命名语言（一语和二语）和学习条件（学过一致、学过不一致和未学过），因变量是测验阶段被试图片命名的反应时和正确率。实验分为学习阶段和测验阶段，被试分别用一语和二语命名图片，观察被试在测验阶段学过一致与学过不一致两种条件下是否都能产生促进效应。实验结果发现，无论俄语为母语者还是日语为母语者，被试只有在学过一致条件下的反应时显著快于未学过条件，而学过不一致条件下的反应时跟未学过没有显著差异。本研究结果表明，汉语为二语学习者言语产生中存在跨语言干扰，且语言间书写特征的差异对跨语言干扰没有影响。

英文摘要：

Several studies reported that there are parallel speak in only one of their two languages. So does alternatives？ Two types of selection mechanisms activations of both languages even when bilinguals intend to the nontarget language compete for the selection of lexical have been contrasted. According to the language-specific selection model, bilinguals can intentionally select the right alternative, thereby avoiding potential competition of both languages. In contrast, the non-specific language model allows competition for selection such that candidates across languages actively compete with alternatives in the unintended language, which will eventually inhibit them and allow accurate production to proceed. Recently there are two alternatives for the selection-by-competition model. One is the weak link hypothesis and the other is the Ll-repeat-benefit hypothesis. One of the claims for cross-language competition in bilingual word production comes from findings using the picture-word interference （PWI） paradigm. The second approach involves switching the languages of production to examine the consequence of having to prepare alternatives in both languages. Whether there is cross-language competition in the language production of Chinese-as-a-second-language learners？ To examine the issue above, this paper completed two experiments using the new paradigm of competitor priming paradigm, which is based on implicit memory. There are 18 Chinese-as-a-second-language learners who are Russian natives in Experiment 1 and 18 learners who are Japanese natives in Experiment 2. Participants were constructed to complete overt picture naming task during the study phase and during the test phase. Naming latencies and accuracy were recorded. 60 black-and-white line drawings were sampled from Zhang and Yang （2003）. Both experiments were carried out on computers using E-Prime software. Participants were seated in front of a computer monitor, a button box, a microphone, and a digital recorder. SPSS 16 software was used to a