随着双语和多语环境的越来越普及化，语言辨别已经引起了越来越广泛的重视。语言辨别可以依赖不同的语言特征。成人的双语者可以自动地利用词汇/语义知识辨别两种学习过的语言。但是对于两种没有学习过的语言的辨别，就无法再借助于词汇/语义知识，而只能利用前词汇的语音特征或者更低级的声学/信号特征，大量行为研究已经证实节奏和语调是两个最重要的可以利用于进行语言辨别的声学/信号特征。但是目前人们对于利用不同的语言特征进行语言辨别的脑机制还知之甚少。本研究的目的即通过脑功能成像技术，探讨人的大脑是如何利用不同层次的语言特征进行语言辨别的，进而深入理解语言辨别的认知加工过程。实验分为四个条件，分别采用不同的听觉刺激作为实验材料。节奏条件的实验刺激是通过语音合成技术将20个正常的汉语和英语句子的所有元音都替换为a，所有辅音都替换为s，并把所有句子的语调都变成平调，因此整个句子只保留了语言的节奏特征；节奏+语调条件的实验刺激做法与节奏条件相同，但保留了句子中的语调特征；语音条件使用了20个正常的意大利语和日语的句子；词汇语义条件使用了20个正常汉语和英语句子。18名以汉语为母语，以英语为第二语言的被试参与了脑功能成像的扫描，在四种条件下，他们都被要求判断前后听到的两个句子是不是属于同一种语言。结果发现，节奏条件主要激活了颞叶皮层。节奏+语调条件的主要激活区域也在颞叶，但比节奏条件多激活了颞极、颞上回中后部和右脑的额下回等区域。语音条件比语调条件多激活了额下回等区域。词汇语义条件则在颞中回等脑区表现出更多的激活。对于颞极、左脑额下和左脑的颞中回这几个感兴趣区的信号变化强度的进一步分析还发现，节奏+语调条件和语音条件在左脑颞极和左脑额下回鳃部表现出交互作用，语音条件和语义条件在左脑的额下回腮部和颞中回表现出交互作用。脑成像结果不仅揭示了语言辨别的脑机制，同时验证了语言辨别过程的认知假设——即大脑可以自动分配认知资源利用最为简单有效的语言特征进行辨别加工——当刺激中只包含一种低级的声学/信号特征时，大脑只能利用该特征进行加工；当刺激中包含很多的特征时，大脑会权衡选择更为确定可靠的特征进行加工；并且当人们能够自动依赖更高级的特征后，对较低级的特征依赖会相应减少。

Bilingual environments are more than a remote possibility. Bilingualism is, in fact, more widespread than is usually acknowledged. Thus, the process of language discrimination has been more and more focused recently.There are several cues that can help persons distinguish two different kinds of language. When adult bilinguists distinguish two different languages, lexical/semantic knowledge can automatically activate to help them to discriminate. When lexical/semantic knowledge information is unavailable, persons could use phonological features or low level acoustic/phonetic cues to distinguish two different languages. For example, a Chinese speaker should have no trouble to discriminate between Italian and Japanese, because these two languages both have their own specific phonemes and sound very different with each other, such as Italian makes use of very characteristic consonants, which don’t exist in Japanese. In addition, rhythm and intonation have also been indicated to be two important acoustic/phonetic cues that enable language discrimination by strong behavioral evidence. However, the neural mechanism for language discrimination is also not well-known. The purpose of this study is to detect the neural networks when persons distinguish two different kinds of language using different cues by neuroimaging technique. We especially focus on these four cues: rhythm, intonation, phonology and lexical/semantic knowledge. Thus, we set four conditions in this experiment corresponding with these four cues separately. Our main findings are the followings. The main brain regions involved in rhythm perception in language discrimination were primarily STG, bilaterally, and there is a little right-sided trend of the activation voxels of STG. More activation for rhythm+intonation processing compared to rhythm processing was primarily found in right Inferior Frontal Gyrus, Temporal Pole and a little posterior STG. Left Inferior Frontal Gyrus were more activated when using phonological information to discriminate two unknown languages. Left Middle Temporal Gyrus were more activated when discriminate two familiar languages. More importantly, we also found when the activation in left Inferior Frontal Gyrus increase in phonology condition, the activation of left Temporal Pole in this condition decreased; when the activation in Left Middle Temporal Gyrus increased in semantic condition, left Inferior Frontal Gyrus in this condition decreased, indicating interaction between these brain areas in language discrimination.To conclude, the brain activation not only revealed that neural mechanism of language discrimination, but more importantly refined our understanding of the cognitive process of language discrimination——the brain can automatically modulate cognitive resource to select the simplest and reliable character to help discrimination; and when brain can use higher character, the activation of neural corps corresponding to lower character will be deactivated. It reflects the wisdom of cognitive ability of human brain.