Modelling bottom-up and top-down linguistic knowledge across different contexts of bilingual development

虽然世界上能够使用多种语言的人远远多于单语者，但理想化的单语者仍然是我们理解语言知识、使用、加工和习得的主要证据来源。本项目假定掌握、保持和使用一种以上语言的能力是人类语言能力的一个关键组成部分。我们将双语纳入语言研究的中心，并提出这是唯一能够充分理解人类语言学习能力架构的方法。通过评估输入频率和习得年龄在不同双语人群词汇和语法发展中的作用，我们试图对语言学习与其他抽象技能学习的相似之处和不同之处有一个综合的理解。我们以双语儿童和成人的语法性别习得和维持为视角，比较了输入频率和第二语言习得年龄对词汇和语法发展的特定项目和基于规则的方面的影响。我们研究了频率和年龄对规则应用的相互作用，深入研究了(i)同步双语者（ii）早期（儿童）L2学习者（iii）晚期（成人）L2学习者以及（iv）儿童移民（早期属性）和(v)成年移民（晚期属性）之间的词汇性别分配和句法一致性。我们关注以德语为目标语言的法语或英语使用者，从总共480名参与者中收集数据。在一系列的生产和理解实验中，我们测试了这些双语群体在识别单词的词汇性别时是如何依赖语音规则而不是项目频率的差异，以及他们在构建句法性别一致时是如何利用语法规则而不是非语法线索的差异。根据目前的研究方法，我们假设早期的母语和二语学习者表现出规则敏感性和输入频率的加性效应，而晚期的二语习得则表现为规则敏感性和词汇输入效应，但在句法一致性方面对语法信息的依赖不足。分析使用混合效应回归模型以及结构方程模型来模拟频率，年龄和背景变量在词汇和语法习得和消耗中的贡献和相互作用。该项目将更好地理解语言发展如何与人类一生中不同阶段的暴露频率和习得年龄等因素相互作用，并为语言是通过一般领域技能还是特定领域技能来学习的问题提供新的见解。

Although there are far more people in the world capable of using more than one language than there are monolinguals, idealized monolingual speakers remain the main source of evidence for our understanding of language knowledge, use, processing and acquisition. This project assumes that the ability to acquire, maintain and use more than one language is a key component of the human language faculty. We move bilingualism into the centre of linguistic inquiry and propose that this is the only approach capable of fully understanding the architecture of the human language learning capacity. By assessing the roles of frequency of the input and age of acquisition in lexical and grammatical development across different populations of bilinguals, we try to come to an integrated understanding of how language learning is similar to and how it is different from the learning of any other abstract skill.Using the acquisition and maintenance of grammatical gender in bilingual children and adults as a lens into the lexicon and the grammar, we compare how input frequency and age of second-language (L2) acquisition affect item-specific and rule-based aspects of lexical and grammatical development. We study interactions of frequency and age effects on rule application in an in-depth examination of lexical gender assignment and syntactic agreement among (i) simultaneous bilinguals (ii) early (child) L2 learners (iii) late (adult) L2 learners as well as (iv) child immigrants (early attriters) and (v) adult immigrants (late attriters). We focus on German as the target language among speakers with French or English as the other language, collecting data from a total of 480 participants. In a range of production and comprehension experiments, we test for differences in how these groups of bilinguals rely on phonological rules versus item frequency in identifying the lexical gender of words and how they recruit grammatical rules versus non-grammatical cues in the construction of syntactic gender agreement. In line with current approaches, we hypothesize that early L1 and L2 learners demonstrate rule-sensitivity with additive effects of input frequency, while late L2 acquisition is characterised by rule-sensitivity and input effects in the lexicon, yet underreliance on grammatical information in syntactic agreement. The analyses use mixed-effects regression models as well as structural equation modelling to model the contributions and interactions of frequency, age and background variables in lexical and grammatical acquisition and attrition. This project will yield a better understanding of how the language development interacts with factors such as the frequency of exposure and age of acquisition at different stages throughout the human lifespan, and provides novel insights into the question of whether language is learned by means of domain-general or domain-specific skills.