군집분석 기법을 이용한 텍스트의 계통 분석 : 수궁가 고고천변 대목을 대상으로

Abstract

In this study, a computational method was used to measure the distance values between texts, especially to measure the variations which took place during the transmission of texts. Gogochenbyeon (a scene of rabbit marvelling at the beautiful scenary of land world) is called nundaemog of Sugungga. Nundaemog is one of the most essential and important parts of a Pansori, like arias of an opera. So many Pansori singers recorded those nundaemog, and therefore, gogochenbyeon has been recorded by several famous singers. We selected 16 albums of 16 singers, and transcribed the narratives from those albums to the texts, and also put the musical expressions (rhythm, rhythm pattern, key, beat) on the Korean traditional musical score Jungganbo. From these raw materials, the text-distances of the transcribed narratives were measured using Levenshtein distance, and the distances were normalized by the length of the narrative strings. And then the texts were clustered using hierarchical clustering methods and turned into the phylogenetic trees using neighbor-joining methods. After the result, we found some differences between the lineage of the Pansori transmission and the real narratives that some singers like Park Tongjin and Song Sunseob are somewhat distant from their pupils or masters.