汉语声调远距离规则的内隐学习

The project aims to examine the implicit acquisition,transfer and brain mechanism of Chinese tonal nonadjacent regularities. Implicit learning is the term coined by Reber (1967) to describe the process by which people come to be sensitive to structural regularities without necessarily intending to do so and without being conscious of the content of the knowledge. A fundamental issue is how to characterize the structures that can be implicitly learnt. That is whether people could learn the abstract underlying regularity or only some chunks and exemplars. Because finite state grammar has high family resemblance, nonadjacent regularity provides a new perspective to explore the abstractness of implicit learning. Although there are some evidence suggesting that, under certain conditions,people can learn nonlocal dependencies.However, in these studies, the rule was not abstract enough and the role of awareness was never directly assessed, therefore leaving open the questions as to whether abstract nonadjacent rule could be learnt and whether the knowledge was explicit or implicit. As a typical tonal language, Chinese use tones to signal different meanings.Tone as an irreplaceable yet rather neglected factor of this delicate combination, is one of the initial speech features that young children acquired long before any other area of phonology during language discrimination and thus we can speculate tonal pattern might easily figure in implicit learning. Using two kins of tonal nonadjacent rule (inversion and retrograde) to control both n-gram structure and repetition patterns, in study 1, we will explore the acquisition of retrograde rule on the basis of previous sduty of inversion rule, and compare the acqusition differences between different nonadjacent rule. The purpose of study 2 is whether the knowledge people acquired could be able to generalize to different untrained lengths of lines, and the results will present a good evidence to support the view of Chomsky that natural languages are more powerful than finite state and a challenge to existing models of implicit learning. Study 3 aims to investigate the neural activity of abstract implicit learning, which will resolve a lack of consensus regarding the brain regions that support abstractness of implicit learning.

本项目以独特的汉语声调的水平映射和垂直映射这两种远距离规则为材料，通过对表面特征与底层规则之间的分离操作，探讨声调远距离规则的习得、迁移和神经机制，为内隐学习的抽象性和规则性提供强有力的新证据。具体来说，本项目将：（1）在前期水平映射规则的内隐习得研究基础上，探讨垂直映射规则的内隐习得，提升先前研究结果的普遍性；（2）考察远距离规则的内隐学习在不同长度材料上的迁移，为内隐学习的规则抽象性提供确定性证据，同时对于解答乔姆斯基关于人类语言无限生成能力（一旦人们掌握了某种语法，就能够基于有限的语言经验生成无限多的句子）提供新思路；（3）借助事件相关fMRI技术考察抽象性规则所对应的神经活动，以廓清人类远距离规则内隐学习的大脑皮层机制。

1967 年，Reber 首次采用限定状态人工语法范式证明了内隐学习的存在。此后，限定状态人工语法规则被用来进一步探讨内隐学习的抽象性问题，即内隐学习获得的是抽象的底层规则还是字母串的表面特征——组块。然而，限定状态人工语法限定了相邻要素之间的顺序，具有较高的语族相似性，无法排除表面特征的混淆。因此一些研究者开始采用相邻要素间没有顺序限制的远距离规则，在平衡组块的条件下来为内隐学习的抽象性提供强有力的全新证据。. 本项目以独特的汉语声调为材料，试图控制组块和重复结构，并采用更为抽象的远距离规则考察内隐学习的抽象性及其神经机制。研究一通过考察被试对垂直映射规则的内隐学习、比较水平映射和垂直映射这两种远距离规则学习的特点，证实被试内隐获得的是抽象规则。研究结果发现，被试不仅可以突破组块等表面特征的限制，内隐习得抽象的汉语声调对称规则，而且汉语声调水平映射规则比垂直映射规则更加容易被内隐学习，这不仅为解读汉语声调的内隐习得机制提供了强有力的支持，也为内隐学习机制在功能上可能优先使用先进先出记忆缓冲器，而非后进先出记忆缓冲器提供了强有力的证据。研究二采用迁移范式，进一步考察被试获得的知识能否灵活迁移到新情境。研究发现声调的水平映射规则和垂直映射规则都能够迁移到不同长度的材料上，为内隐学习的规则抽象性提供了确定性证据，同时对于解答乔姆斯基关于人类语言无限生成能力（一旦人们掌握了某种语法，就能够基于有限的语言经验生成无限多的句子）提供了新思路。研究三采用功能磁共振成像技术，考察远距离规则内隐学习的神经机制，探测被试运用规则进行分类判断时的激活的脑区，进一步考察远距离抽象规则的脑机制。研究结果从神经机制的角度证实了汉语声调远距离规则能够被内隐地习得，这种抽象性规则的加工主要依靠布洛卡区。