Dependency of number processing on native languages : Why numbers are difficult to process in the second language?

当人们使用第二语言（L2），即他们的非母语时，即使他们在L2环境中生活了很长时间，或者即使他们是完美的双语者，数字也很难处理。我们进行了三个系列的认知心理学实验：1)在实验范式中研究现象；2)确定现象的条件；3)证明我们关于现象原因的假设的有效性；也就是说，数字处理依赖于母语。第一批关于命名和朗读任务的实验表明，只有当日语母语人士必须将单个数字作为有意义的实体来处理时，他们才需要更长的时间来回应英语数字（他们的第二语言）。第二组实验采用了准stroop任务，实验对象被要求尽快用英语或日语说出一系列图片和数字的名字。对错误的分析表明，数字会导致更多的替换错误；也就是说，用L1而不是L2来回应。这些结果支持了数字处理依赖于母语的假设。第三组实验排除了语言因素的影响，使用了听觉刺激和按“是/否”按钮的任务。听觉-视觉匹配任务和连续决策任务（决定与先前呈现的刺激相比，在一系列刺激中是更大还是更前）都表明，数字在英语中处理起来更困难、更慢，尽管它们比名词和字符容易得多。在这些实验结果的基础上，讨论了人类认知系统中数字现象的特殊性，并正在进行这种现象的建模。

Numbers are difficult to process when people are using a second language (L2), i.e., their non-native language, even after they have lived in the L2 environment for a long time, or even if they are perfect bilinguals. Three series of cognitive psychological experiments have been conducted in order i) to investigate the phenomena in experimental paradigms, ii) to identify conditions for the phenomena, and iii) to demonstrate the validity of our hypothesis about the cause of the phenomena ; namely, the dependency of number processing upon native languages. The first series of experiments with naming and reading-aloud tasks showed that Japanese native speakers needed a longer time to respond to numbers in English (their L2) only when they had to process individual number as meaningful entities. The second series of experiments employed quasi-Stroop tasks, in which subjects were to name a series of pictures and numbers as fast as possible in English or in Japanese. Analysis of errors demonstrated that numbers caused many more replacement errors ; i.e., responding in L1 instead of L2. These results support the hypothesis that number processing depends on the native language. The third series of experiments were designed to exclude the factor of language in the response, and used tasks in which stimuli were presented auditorily and subjects responded by pushing a Yes/No button. Both auditory - visual matching tasks and continuous decision tasks (to decide whether the stimulus is bigger or former in a series compared with the previously presented one) showed that numbers were more difficult and slower to process in English, even though they were much easier compared to nouns and characters. The specificity of number phenomena in human cognitive systems was discussed based on these experimental results, and modeling of this phenomena are now under way.