Complexity Mechanism in Leveled Texts

The paper presents the research conducted to assess and contrast linguistic and cognitive complexity of primary (fiction) and secondary (leveled to teach Russian as a foreign language) texts. The research dataset comprises dataset of texts with the total amount of 22,232 tokens. Measurement of linguistic complexity performed with Rulingva (rulingva.kpfu.ru) revealed 14 statistically significant parameters discriminating the primary and secondary (leveled) texts. The parametersdiscriminators fall into three groups: morphological (the number of two-syllabic words, verbs, nouns, numerals, the average number of nouns, verbs and adverbs per sentence, ‘descriptivity’, ‘narrativity’), lexical (hapax legomena, Abstractness) and syntactic (readability, sentence length, word length in syllables or in characters). Assessment of cognitive complexity of the contrasted pairs of texts included the two-step algorithm: propositional analysis and information density measurement. The propositional analysis (by W. Kintsch) was aimed at identifying and calculating the number propositions, while the information density measurement implied normalization of the number of propositions by the length of the text (In words) followed by the contrastive analysis. The study revealed increase in the normalized information density in 7 out of 9 leveled texts and confirmed the hypothesis that though leveled texts have a lower degree of linguistic complexity they retain a higher degree of normalized information density inherent in the primary text.