On Phonological Representations as Equilibra among Phonological Computational Device

本文的总体研究方案是在博弈论的基础上重构音系学一般理论的框架。在这个框架内，一组语音小工具被假定在这些小工具之间产生最佳平衡。本研究阐明了自然语言的音韵学过程是发音和听觉计算装置之间的平衡，并进一步构建了一个真正可验证的一般音韵学理论。“均衡”的概念来源于社会科学领域的纳什均衡概念。本研究试图实现一种语言模型，在该模型中，假设语音计算工具之间的关系是相对的，从而达到在语音计算假设的支付矩阵中评估最高的最佳语音表示。目前的框架与优选性理论的原则有很大的不同，因为它认为语法系统的输出是有限的：优选性理论假设语音输出本质上是无限的，并且通过对输出的评估过程来选择最佳的表示。如果我们把注意力集中在大脑/思想的实时处理上，我们的模型在语言计算的心理现实方面是有良好动机的。当前方法的一个重要推论是，我们可以在Jackendoff的工作中提出的三方并行架构中取消对应规则的设备。在我们的模型中，语音、句法和概念模块之间的对应关系是通过计算支付矩阵来确定的。第二个推论是，语义和语境语用计算的理论基础被统一到博弈论方法的一般原则中。问题的关键在于确定代理要素的单位层次。在我们的方法中，一个人可以是某个框架中的代理元素，而在语音描述中，我们假设一个人内化了多个计算小工具或子代理。相反，一组人也可以作为一个整体的代理人。这三个层次的理论同质性有助于我们理解理论语言学与交际理论的统一。少

The overall research scheme has been to reconstruct the framework of the general theory of phonology based on Game Theory. Within the framework, a set of phonological gadget are postulated to produce an optimal equilibrium among those gadgets. The present research elucidated the fact that the phonological processes in natural languages are equilibria among those articulatory and auditory computational device, and it goes on to construct a general theory of phonology that is truly verifiable. The notion of "equilibrium" comes from the notion of Nash Equilibrium within the realm of social sciences.The present research tried to realize a linguistic model in which the relationships among phonological computational gadgets are assumed to be relative so as to arrive at an optimal phonological representation that is evaluated highest in the payoff matrix assumed in the phonological computation assumed there. The present framework crucially differs from the tenet of Optimality Theory in that t … More he outputs of the grammatical system are finite : Optimality Theory assumes that the phonological outputs are essentially infinite and that the optimal representation is selected by the procedure of the evaluation of the outputs by the ranked universal constraints.If we focus our attention on the real-time processing in the brain / mind, our model is well-motivated with respect to the psychological reality of the linguistic computation.A significant corollary of the present approach is that we may do away with the devices of correspondence rules in the Tripartite Parallel Architecture proposed in Jackendoff's work. In our model the correspondence among phonological, syntactic and conceptual modules is worked out by the calculation done in the payoff matrices.The second corollary is that the theoretical foundation of the semantics and contextual-pragmatic computation are unified into the general tenet of our game-theoretic approach. The point has been the identification of the level of the unit of agentive elements. In our approach, a person can be a agentive element in a certain framework, while in the phonological description we assume that one person internalizes more than one computational gadget, or sub-agents. Conversely, a group of persons can also be an agent as a whole. The theoretical homogeneity of the three level may contribute to our understand of the unification of the theoretical linguistics and the theory of communication. Less