联结主义理论下的二语词汇知识深度习得模式的构建

发布时间：2018-04-19 14:05

本文选题：联结主义 + 模型　；参考：《天津大学》2013年硕士论文

【摘要】：在二语词汇习得的研究领域中，研究者们早就注意到仅仅记住单词的拼写和意义并不能确保学习者准确、灵活地运用词汇。近年来，二语词汇习得的关注焦点集中在对词汇知识维度的研究上，研究者们认为词汇知识是由多维度而非单一维度构成，并据此构建了各种形态不一的词汇知识框架体系。这些框架对词汇知识维度的具体分类不尽相同，但广度和深度这两个维度包含在所有框架中，是词汇知识最基本和最核心的内容。第二语言词汇知识不仅包含静态知识（广度)，即读音、词形、意义三个部分，还包括动态知识（深度）即词语频度、家族关系、句法功能和搭配关系等知识。联结主义认为词汇知识不是孤立地储存于大脑中的不同位置，，也不是以独立概念的形式被提取。词汇知识是以单元的形式并存于大脑神经网络中，不同单元之间由权重相互连接，通过激活单元并加强各单元之间的联结强度即权重从而被习得。联结主义网络的概念符合生物学家对人脑中枢神经系统的认识知识。联结主义理论是建立在神经生理学、神经心理学、认知心理学等学科的研究基础之上，试图通过构建网络模型来模拟人脑的神经活动以探索人脑的认知过程的研究范式。传统符号学信息加工理论认为人脑对信息的处理是离散的、序列的、非自组织的和局部的，而联结主义神经网络则认为人脑对信息的处理是以平行结构进行的，具有亚符号性、连续性、平行性和分布式特征（李平，2003）。此外，联结主义神经网络还具有自学习、自适应、自组织等功能。如果网络的输出发生错误，其自身就可以通过采取一定的训练策略来调整神经单元之间的权重，直到输出与期望相符。联结主义神经网络所具有的这些特征，使其对大脑的模拟更接近生物脑的实际情况，对心理实质的揭示也更符合人的真实心理活动。语言学家也因此试图通过联结主义网络对语言的产生、理解和使用过程进行模拟并做出合理的解释。语言研究者们通过构建联结主义网络模型已经成功地模拟了学习者习得词汇、语法、阅读和理解段落的过程。然而，国内外联结主义研究学者对词汇的研究以母语居多，对二词汇习得模式的模型构建仍然停留在静态描述的层面：即只研究了词汇在大脑中的存储方式和提取过程，而对产出性词汇习得的研究并不多见且未构建出覆盖词汇知识深度层面的模型。词汇研究的中心问题之一是对词汇深度知识习得的研究(Henriksen,2008)，而词汇习得是二语习得过程中的基本和核心内容。本文依据联结主义理论构建二语词汇深度知识习得的模型，探索二语习得者词汇深度知识习得的认知过程及母语对这一过程的影响，以期为二语习得者习得二语词汇提供一种心理认知模式，为二语词汇的教学提供某种理论启示。
[Abstract]:In the research field of two language vocabulary acquisition, researchers have long noticed that only memorizing the spelling and meaning of words does not guarantee the learners' accurate and flexible use of vocabulary. In recent years, the focus of attention on vocabulary acquisition in two language is focused on the study of the dimension of vocabulary knowledge, and researchers believe that vocabulary knowledge is multidimensional rather than single. The two dimensions of the breadth and depth are the most basic and core content of the vocabulary knowledge. The second language vocabulary knowledge contains not only static knowledge (breadth), but also the basic and core content of vocabulary knowledge. That is, pronunciation, word form, meaning three parts, including dynamic knowledge (depth), namely word frequency, family relationship, syntactic function and collocation knowledge.
Connectionism holds that lexical knowledge is not isolated in different positions in the brain, nor is it extracted in the form of independent concepts. Lexical knowledge coexists in the form of units in the neural network of the brain. The weights are connected to each other by activating the unit and strengthening the bond strength between the units. The concept of connectionist network is consistent with biologists' knowledge of the central nervous system of the human brain.
The theory of connectionism is based on the research of neurophysiology, neuropsychology and cognitive psychology, trying to simulate the neural activity of the human brain by constructing the network model to explore the cognitive process of the human brain. Non self organized and local, and the connectionist neural network believes that the human brain is dealing with information in parallel structures with subsymbolic, continuous, parallel and distributed features (Li Ping, 2003). Furthermore, the connectionist neural network has the functions of self-learning, self-adaptive, self-organizing and so on. If the output of the network is wrong, the network is wrong. It can adjust the weights between the nerve cells by taking a certain training strategy until the output is in conformity with the expectation. These characteristics of the connectionist neural network make the simulation of the brain closer to the actual situation of the biological brain, and the psychological essence is more conformed to the real psychological activity of the human mind. Therefore, we try to simulate and explain the generation, comprehension and use of language through connectionist network.
By building a connectionist network model, language researchers have successfully simulated the process of learning vocabulary, grammar, reading and understanding of paragraphs. However, the domestic and foreign researchers are mostly native to the study of vocabulary, and the model construction of the two vocabulary acquisition model remains at a static level: that is, only research. The storage mode and extraction process of vocabulary in the brain are studied, and the study of productive vocabulary acquisition is not much and does not build a model to cover the depth of vocabulary knowledge.
One of the central problems in the study of vocabulary is the study of vocabulary acquisition (Henriksen, 2008), and vocabulary acquisition is the basic and core content in the process of two language acquisition. Based on the theory of connectionism, this paper constructs a model for the acquisition of deep knowledge in two language, and explores the cognitive process and mother tongue of the acquisition of deep knowledge in the two language learners. The purpose of this study is to provide a psychological cognitive model for two language learners to acquire two vocabulary and provide some theoretical enlightenment for the teaching of two vocabulary.

【学位授予单位】：天津大学
【学位级别】：硕士
【学位授予年份】：2013
【分类号】：H09

【参考文献】