The study of collocation structures in Dede Korkut Stories according to computational linguistics methods

One of the most important eposes of is undoubtedly The Dede Korkut Epos. Dede Korkut, also known as Korkut Ata is a value in itself in the deep memory of Turkishness. The legends, sayings, proverbs, phrase expressions and collocational structures about Dede Korkut are all a result of this deep memory. Dede Korkut and the Book of Dede Korkut are among the common cultural values of the Turkish world. Book of Dede Korkut is undoubtedly Oğuznâme, or part of Oğuznâme. Dede Korkut is the person who edits and narrates parts of Oğuznâme. For this reason, some parts of Oğuznâme are named Dede Korkut Stories, or Dede Korkut Epos. The Dresden and Vatican manuscripts were the first that came to mind when mentioning the Book of Dede Korkut. After the known first (Dresden) and second (Vatican) manuscripts of Dede Korkut Stories, which has a very important place in Turkish language and literature, the third (Türkmen Sahra) manuscript was recently discovered and contributed to the scientific world. The discovery of the new manuscript has created an unparalleled resource to researchers of the "Dede Korkut Tradition". Advances in information technology has made it easier to process and examine languages with computer applications and software; and it became possible to analyse all language units, from phonetic to structural qualities, from vocabulary to syntax, with statistical or rule based computational linguistics methods. In this context, the subject of "Collocation", which has been studied for 25 years in Turkish linguistics, is now studied with computational methods. However there are few studies on "Collocation" in the context of Turkish historical texts. The scarcity of collocation studies for the historical periods of Turkish, and the possibility of examining linguistic phenomena using linguistic methods has led us to study the subject of collocation with a computational perspective. Our study is based on the study of stories Written in Old Anatolian Turkish with computer modeling methods regarding collocational objects (collocational sequences and collocational sets). The stories are transferred to computer environment as plain text, the collocational vocabulary was processed in Python programming language and in AntConc 3.5.9 (2020) software, then the collocations are examined using different statistical analyses such as Log-likehood, MI value, T-score, and Dice coefficient. In the method that uses statistical significance criterion, algorithms that take into account what ratio of a given word's appearance with other words to the total usage of this word were used. In this way, both the findings of collocational vocabulary detected by the researcher with linguistic methods and those found by natural language processing methods were compared and the subject of collocation was evaluated in the context of correspondence.