Experiments on the Automated Determination of Difficulty Levels in Russian Texts

Abstract

Experiments on the Automated Determination of Difficulty Levels in Russian Texts. Reading plays a very important role in language learning; however, selecting the appropriate text is often not an easy task. Although there are already books adapted for learners, they are mostly available only in English and in limited numbers. Therefore, I attempted to develop an algorithm capable of quantitatively expressing the difficulty level of large volumes of Russian-language texts. I have made the algorithm and its associated toolkit freely available on the in¬ternet. It is important to note, however, that due to differing characteristics of texts, a single formula cannot be used to evaluate all types of texts. Thus, the formula I propose is applicable only to literary texts.
Keywords: Russian language, text complexity, automated text processing, large-scale text processing, linear regression