On means of sociolingiustic balancing of a spoken corpus (drawn on “one speaker's day”)

The article considers different means of sociolinguistic balancing involved in forming spoken corpora. It gives a review of sociolinguistic principles which underlie some of the foreign and Russian corpora (for instance, the British National Corpus and the Russian National Corpus) and states the problem of unbalanced material, which linguists often encounter. The main object of the study is the sound corpus “One Speaker’s Day”, being recorded in Saint Petersburg, and the methods of its balancing and forming used by its developers. It is the second part of the corpus, the first being an annotated collection of texts which were analyzed by the means of binary oppositions. It has not always proved suitable for the corpus. In order to obtain sociolinguistically balanced material for research, the authors designed a special questionnaire based on the data of Federal Agency of Urban Statistics and general trends of linguistic description of a city. This questionnaire was used to select the candidates to participate in recording for the project. It includes such important sociolinguistic characteristics as gender, age, place of birth and places of living of the informants, their former and current occupations, their profession and education, their native language and other languages they speak, as well as their parents’ place of birth and social origins. The importance of each factor is described and justified in this study. Using certain factors, for example gender and native language, it is possible to divide the informants of a corpus research in two well-balanced groups, but others, such as age and occupation, demand a more precise classification. Therefore, the research aims to categorize the material of “One Speaker’s Day” by possible balanced groups and subgroups and states the necessity for a detailed analysis. It suggests that certain problems of balancing a corpus mentioned in the review can be avoided using this method.