双语者语言加工的双机制模式

A Dual Mechanism Model of Language Processing in the Bilingual Brain

语言加工的单机制模型认为,加工第二语言必须依靠第一语言脑网络,然而单机制模型无法解释双语者如何加工第二语言的语法特异性信息,即为什么二语熟练程度的提高并没有缩小语法加工的激活差异程度,以及为什么早期双语者也存在激活差异等。本研究提出的双语者语言加工的双机制模式,以语言加工的模块化特点为理论基础,强调专属脑网络参与语言特异性信息加工的必要性,为合理解释当前研究的新发现和争议问题提供新的思路。

Languages are uniquely human. Mastering more than one language is becoming common and important in globalized modem society. In the neurolinguistic field of bilingualism, researchers wonder how bilinguals modulate brain resources to incorporate two languages in the same brain. Debates arise in two areas： 1） is there a unique neural network, or are there rather spatially segregated networks subserving two （ or multiple） languages in the human brain？ 2） what are the determinants of activation differences between languages in bilinguals？ The single network hypothesis proposes that the second language （L2） processing relies on neural mechanism of the first language （L1）, and the activation difference comes from cognitive control. However, it fails to explain some research disputes. For example, how bilinguals process grammatical features of L2 which are lacked in L1; why intensive practices cannot narrow activation differences of grammar processing between the two languages; why early bilinguals also show neural activation differences. The dual mechanism model which the current study proposes is based on the theoretical rationale that language features are processed by different modules, as a great many brain imaging studies have suggested a correlation between increased activity in specific cortical areas and specific language functions. Inter-hemispheric differences in topographic specialization for L1 and L2 has been supported by clinical reports of unbalanced language deficits in stroke patients. Functional imaging researches also conclude that L2 is largely processed in the same brain regions as the first language, but to some degree in different brain regions from L1 either. These findings suggest that brain activation differences are processed by specific brain regions subserving linguistic features, even in bilinguals who access the second language environment early. On account of the independence of two languages in both early and late bilingual brains, it is arbitrary to declare that there exists only one brain mechanism underlying languages. We suggest that activation disparities should not be attribated to different acquisition time and inequivalent language proficiency that may lead to cognitive control only. Instead, language experience may be a dominant factor. Function-specific cortical representatives are designed by evolution and language features to determine the development of professionalized neural circuitries, which is strengthened, reorganized, or replaced with the development of language proficiency. Simply put, language proficiency and AoA can determine whether the neural changes take place and at what speed, but they do not dominate the direction the brain changes. With time going, the decline tendency of the plastic resilience is equally applied to all cognitive functions, including both syntax and semantics. Early and late bilingual learners would differ in the way how fast and easy the brain settled down inputs of the two languages, but could not change the segregation of designated cortical regions modulated by language features. More exposure to and frequent daily use of the second language may have the developing brain systems quantitatively changed, but not qualitatively altered. If the processing of two languages engaged different neural substrates in nature, the brain will consistently present non-overlapped brain activation for L1 and L2 contrast, irrespective of an early or late bilingual brain. They finally might reach a neural equilibrium of two independent language systems with neural specificity for individual language features. As for significant activation differences between L1 and L2 grammars but not semantics, the dual mechanism model would predict that the brain assimilates common properties, and accommodates features of the new language. Lexico-semantic knowledge is shared and no more specific brain regions are involved in processing conceptual features. The expansion of L2 semantics is similar to the way monolingual learners enlarge their meaning and concept pool. Therefore, both early and late acquisition of the second language concept activate a series of shared cortical regions. In contrast, as languages differentiate from each other mainly in the grammatical aspect, language dissimilarities in orthography, phonology, and syntax determine the transfer of appropriate reading skills, during which the obligatory accommodation changes involve additional cerebral snbstrates specialized for certain cognitive functions. With the increase of language practice, neural representatives of different grammatical features are most likely to be illustrated by separate cortical regions. The ever-partitioned brain regions and the segregation process of neural adaptation together result in enlarged gaps between brain regions implementing distinct LI and L2 grammatical rules. It is the enhanced integration of the assimilation process and the specific segregation of the accommodation process that cause different brain activation pattern of syntax and semantics. The model also admits that at the beginning of learning a second language, activation differences of grammars between the two languages are mainly the difference between the declarative memory and procedural memory, as late bilinguals always adopt rote memory to process the second language grammar at first.