Sociolinguistic Variability of Russian Everyday Speech: A Corpus-Based Study

The paper presents recent results of a multilevel analysis of representative corpus data, conducted in order to identify key speech parameters (lexical, morphological and syntactic) that can diagnose some social/biological characteristics of a speaker or, more broadly, a modern Russian urban sociolect. The study is based on the everyday Russian speech corpus One Speakers Day. Specific data were obtained on the analysis of the annotated subcorpus of 289,205 tokens, which includes recorded speech days of 57 men and 48 women, which were the research participants, as well as speech fragments of 87 men and 139 women, which were their interlocutors. Thus, the total number of speakers in the subsample amounts to 144 men and 187 women. The article also begs the question of Data Mining approach usability to the subcorpus and possibilities of further research using machine learning. The results obtained are important for the optimization of speech technologies systems, for theoretical understanding of linguistic processes, as well as for monitoring various social processes taking place in modern Russian metropolis.