Analysis of Literary Works from Turkish and World Literature with Natural Language Processing

There are various reading books published by many publishers for the students' utilization. Publishers prepare these literary works separately for the stages of primary, secondary and high schools by considering the characteristics of the groups. Literary works prepared for the secondary school students were used in this study. 5 literary works from Turkish and 5 from world literature having the same title out of 5 publishers were analysed by means of NLP (Natural Language Processing) techniques within the setting of Turkish grammar rules. With this study, it is aimed to determine the differences and similarities among publishers in terms of composing literary works and Turkish use in them, at the same time, it is also aimed to reveal the distinctive characteristics of the Turkish use. Developed software and Zemberek, Turkish NLP Library, were used in the analysis processes. Results which were obtained via analyses of literary works were found out as similar in terms of proportion and the differences were at the low level in general among 5 publishers. Differences were only found out in terms of unanalysable word number, average sentence number of literary works, average word number in literary works, average word number in a sentence and the ratio of the use of punctuation marks (dot and comma). Despite the fact that the titles of the literary works are the same among publishers, these distinctions occurred as a result of the difference in terms of the contents.