间接言语行为使用条件与强重复最优应答推理

The Using Conditions of Indirect Speech Acts and Strong Iterated Best Response Model

关于间接言语行为使用条件的博弈论解释一直存在均衡选择的问题。范·罗伊和赛文斯特的超规约信号博弈模型的简单期望收益推理所依据的"同情指数"概念不具有可操作性,不可能对间接言语行为使用条件进行明确的限定。IBR推理的假设符合直觉,可以适用于一般情况下间接言语行为的博弈解释。通过在IBR框架中引入"高阶信念"的概念,可以建立一个关于话语表达信号博弈的强IBR推理模型,这能够把对话双方的同情指数量化,并明确限定间接言语的使用条件。与弗兰克的基本IBR推理模型相比,强IBR推理放弃了等先验信念假设,通过引入概率变量考察交际双方的相互信念对其期望效用的影响,因而对实际交际情况具有更强的解释力。

The game-theoretical explanation for the using conditions of indirect speech acts has the problem of equilibrium selection. Van Rooij and Sevenster＇s Simple Expected Payoff Reasoning within their Super Conventional Signaling Game model is inadequate to quantify the concept of sympa- thy, and thus cannot offer a proper explanation for the using conditions of indirect speech acts. In contrast, the IBR model with its intuitive assump- tions is adequate to explain the conditions in general. By introducing the concept of higher-order belief, we develop a strong IBR model within game- theoretical pragmatics. Comparing to Franke＇s vanilla IBR model, our strong IBR model gives up the flat-priors assumption and studies the effect of interlocutors＇ mutual belief on expected payoff, and thus is more powerful in its explanation for actual communication.