<p>This work presents a comparative analysis of the characteristics of social networks derived from different versions and translations of the Irish epic <i>Táin Bó Cúailnge</i>. The characteristics of the social networks are compared using graph-matching techniques and comparing structural properties, as well as truncating the larger networks to match the size of the smaller networks for further comparison. We find that the original Irish edition and the direct English translation of the <i>Táin Bó Cúailnge</i> Recension I by Cecile O’Rahilly are most similar to one another, and that the modern English versions have the furthest distance from the sources. Individually we show that each social network, regardless of the author or network size, is disassortative and display "small-world" network characteristics. They are also shown to have largely similar degree distributions, with Maximum Likelihood Estimate fitting showing that most networks can be fitted using power law distributions. The results show that the more similar a version stays to the source material, the more similar the generated social networks will be.</p>

错误:搜索内容不能为空,请输入英文关键词
错误:关键词超出字数限制,请精简
高级检索

Comparing social networks of translations: a case study on the medieval irish epic Táin Bó Cúailnge

  • Stephen McKermitt,
  • Gaëlle Clion,
  • Pádraig MacCarron

摘要

This work presents a comparative analysis of the characteristics of social networks derived from different versions and translations of the Irish epic Táin Bó Cúailnge. The characteristics of the social networks are compared using graph-matching techniques and comparing structural properties, as well as truncating the larger networks to match the size of the smaller networks for further comparison. We find that the original Irish edition and the direct English translation of the Táin Bó Cúailnge Recension I by Cecile O’Rahilly are most similar to one another, and that the modern English versions have the furthest distance from the sources. Individually we show that each social network, regardless of the author or network size, is disassortative and display "small-world" network characteristics. They are also shown to have largely similar degree distributions, with Maximum Likelihood Estimate fitting showing that most networks can be fitted using power law distributions. The results show that the more similar a version stays to the source material, the more similar the generated social networks will be.