LINGVO-STATISTICAL ANALYSIS OF THE OLD RUSSIAN ANTHOLOGY (RNL, F.p.I. 39)

A linguistic analysis of 29 texts of the 13th century manuscript (the Tolstoy miscellany, RNL, F.p.I. 39), containing the original works of the ancient Russian author Kirill Turovsky and translations of the texts of Cyril of Jerusalem, John Chrysostom and unknown authors is presented. The objectives of the work are a) verification of the conclusions about the authorship of three anonymous texts of the manuscript attributed to Kirill Turovsky, made in the course of their previous analysis using traditional linguistic methods and correlation methods of statistics, and also b) establishing the possibility of using statistical methods to compare small texts. To measure the distance between the texts, the Delta method and its modification the Cosine-Delta were used, a different number of the most frequent words (from 50 to 250) of each of the texts were used. The data were extracted from the electronic edition of the Tolstoy miscellany. Calculations and construction of dendrograms were carried out using the Stylo package. Dendrogram analysis confirmed the findings of other studies: the three anonymous texts - Parable of Wisdom, Sermon on Pentecost and Sermon on the 5th Sunday after Easter - do not belong to Kirill Turovsky. The dependence of the grouping of texts on the size of the texts and applied methods was found. It was found that the cosine-Delta method, in contrast to the classical Delta, turns out to be insensitive to differences in the size of texts. In addition, an increase in the number of the most frequent words when using cosine-delta allows one to find a correlation between groupings of texts and their discursive and lexico-syntactic affinity.