Corpus study of bilingual errors

The article is devoted to the corpus study of bilingual errors and to the typology of speech errors. Within the framework of the research we used the web corpus manager that is the system for annotating the corpus of texts. The system provides a three-level error classification. The corpus includes texts in various languages (Russian, French, English, Spanish, Chinese, etc.) written by native speakers as well as by intentional bilinguals. We propose a method of identification of the author's language acquisition competence and authenticity of the text (written by a native speaker or a bilingual) based on machine learning.