Abstract: Language evolution is one of the important problems in social science field. Before the social simulation methods are introduced in, this problem is a typical social science problem which can’t be applied to computing experiments. It is feasible to apply social simulation methods to this problem. This paper proposes a new dynamic social network model to study the competition between two languages. This model is based on agent modeling method and social circles theory. In order to reflect the vocabulary of the language and its proficiency, a language structure with two F-length integer-strings is introduced into the agent which is a node of the network. It proves that the new model has the same dynamic characteristics as the previous model and analyzes the effect of language policy by computational experiments. The dynamic characteristics are compared between bit-strings vocabulary structure and integer-strings vocabulary structure system. The experimental results show that the range of parameter values of language coexistence is very small and it is difficult to achieve the state of language coexistence without intervention, but the range is significantly broadened to achieve the coexistence state when the intervention is undertaken. Furthermore, it is obvious that the integer-strings vocabulary structure system is more expressive.

Key words: social circle, complex network, language competition, social simulation, computational experiment

摘要： 语言竞争传播演化现象是典型的不能假设、无法进行“真实性实验”的社会科学问题，而建立在社会仿真模型基础上的计算实验是可行的方案。基于agent社会圈子网络理论构建语言竞争网络，采用两个长度为F的整数字符串来表达个体掌握语言的词汇量及熟练程度，在此基础上给出了一种新的动态微观语言竞争社会仿真模型。利用计算实验的方法验证了给出的新模型具有与前面研究者模型相同的动力学特性，分析了新模型的语言政策调控效果，比较了二进制词汇结构和整数字串词汇结构系统的动力学差异。计算实验表明：无调控情况下，语言共存参数空间很小，难以达到稳定共存状态，动态调控能让语言共存参数空间显著扩大，并稳定在共存状态，整数字串词汇结构模型具有更丰富的表达能力。

关键词: 社会圈子, 复杂网络, 语言竞争, 社会仿真, 计算实验