這個研究的目的是針對語言分化背後的驅動力這個問題,提出一個模擬模型,可以在很弱的假設狀況下,成功的演化出語言的分化現象。語言有很多面向,包含字彙,文法,還有語音,而語言的變化在字彙上是最明顯快速的,跟以前的研究一樣,這個研究以字彙作為主要的研究標的。以前的研究,所提出的演化模型,都會設定一些假設,來增加群體當中的隔離性,導致方言分化的發生,例如用來增強空間隔絕的村落,或是用來增強社會隔絕的社會標誌。這個研究設計了一個相當簡單的模擬模型,採用了讓代理人自行在環境當中演化並且尋找最佳策略的機制,代理人必須在環境當中跟與他相鄰的代理人進行囚犯困境問題的遊戲,並決定各自的繁殖率。這個模擬模型的設計,設定的假設少且弱,導致模擬模型的結果自由度很大。但是實驗的結果顯示,在這樣的模擬環境下,即使給予很大範圍的參數設定,種族優越意識都會被演化出來,並且導致方言的分化。
The purpose of this dissertation focuses on the study on the driving forces of dialect diversity. We proposed a model with very weak assumptions, and can derive dialect diversity. Language has many components: vocabulary, grammar, and vocal. This work focuses on the vocabulary since it is the most significant property during the language evolution process. Previous works often proposed some assumptions to increase the isolating force to derive dialect diversity. For example, endow the concept of a village to increase the spatial isolation, and use social marking to increase the social isolation. This work proposed a very simple model with only a few weak assumptions. We let agents who search the best strategy by themselves to evolve in a virtual environment. Agents actually play prisoners' dilemma game and effect their reproduce probabilities. The assumptions in the model are weak, so the degree of freedom is very large. The result of this work shows that, even under a great variance of parameters. Ethnocentric behavior will emerge and derive dialect diversity.