Resumen de Lithuanian Pedagogic Corpus: Correlations Between Linguistic Features and Text Complexity

Loïc Boizou, Jolanta Kovalevskaitė, Erika Rimkutė

• This paper discusses the problem of automatic CEFR2 level assignment to texts. We address the correlations between the lexical, morphological and syntactic features and the different CEFR levels of the texts in the Lithuanian Pedagogic Corpus. Only the texts from coursebooks showed the correlation of investigated linguistic features with text complexity. In the coursebook sub-part of the corpus, we observed that higher language proficiency levels are associated with more complex linguistic features: their number increases in texts of higher CEFR levels from A1 to B2 (e.g., non-finite verb forms, participles, adverbial participles and half participles, dative and instrumental noun cases or longer sentences).