Stanza-based networks for poetic texts: a pilot study

Abstract

Abstract We use approaches from the theory of complex networks to analyze the Prologue to Moses, a poem by the Ukrainian writer and scholar Ivan Franko. The Ukrainian text (in the original orthography from 1905 and the modern one) and several translations into Polish, English, French, German, Italian, Portuguese, Slovak, and Spanish are studied. Networks are built in a manner that links are drawn between words (considered network vertices) within a stanza. Seven network parameters are calculated: all-degree centralization, assortativity, average path length, betweenness, clustering coefficient, density, and transitivity. Their dynamics is analyzed as the vertices with the highest numbers of links are removed from the networks one by one. The data obtained for the human-made translations are compared to those based on machine translations generated by the Google Translate service. Peculiarities of certain translations are underlined. Tentative correlation between the values of parameters and levels of language analyticity/syntheticity are established. Prospects toward automatic lemmatization and related problems to be solved are briefly outlined.