The study is aimed at investigating the potential for original automated Russian text analyzer RuLingva to assess linguistic metrics of written recalls of students of Russian as a Foreign language (RFL). Initially developed by the authors to estimate Russian texts readability indices, the public version of RuLingva reports on 33 metrics related to text length, readability indices, parts of speech classification, noun case, verb tenses, vocabulary frequency, lexical diversity, abstractness rating, etc. We hypothesize that the abovementioned metrics can be used to discriminate and score written expository discourse of RFL students. The corpus compiled for the study comprises written recalls of 407-word expository texts produced by 71 B2 students of Russian. Each subject’s recall was scaled against the original reading text on the following metrics: text length, average sentence length, average word length, type token ratio, word frequency, abstractness rating, local and global noun overlap, local and global argument overlap. Prior to reading the expository text, we also assessed the subjects’ general knowledge with WISC test and Russian language proficiency with Quick Russian placement Test. T tests of significance indicated a strong positive correlation (>0.05) of both general knowledge and Russian proficiency tests with the abovementioned metrics automatically assessed with RuLingva. The findings enable to narrow the range of text features predicting RFL writing quality and ways of estimating language proficiency. RuLingva as the first Automated Writing Assessment tool for the Russian Language has a potential to be successfully used in formative assessment motivating students to review their writing and contributing to both literal and inferential comprehension.

Russian text analyzer RuLingva, written recalls, text metrics, Automated Writing Assessment