ON THE FEATURES OF STATISTICAL PROCESSING OF PHONETIC EXPERIMENTAL DATA IN LINGUISTIC RESEARCH

This article discusses the problem of using statistical methods for processing experimental data in the study of quantitative and qualitative properties of phonemes based on literary texts. The article studies the duration of vowels in the Russian language and its analogues in English in statistical terms. The duration of speech sounds is measured in thousandths of second - milliseconds (ms). In English, stressed vowels are longer than unstressed. The authors of the article note that it is not the statistical apparatus itself that presents the greatest difficulties. It turns out that it is most difficult for a phonetic researcher to see a statistical problem in his linguistic hypothesis, choose the correct statistical procedure, and then give an adequate interpretation of statistical calculations in terms of a linguistic problem. The authors regret that despite the rapid development of applied, statistical linguistics, there are no methodological developments yet, as in other applied fields. So far, the only way for a researcher is equal mastery of both linguistic problems and the apparatus of mathematical statistics.In theoretical linguistics, linguistic research is usually divided into observation and experiment. The presented statistical methods apply mainly to observations. However, modern linguistics is becoming more and more experimental, especially in its field, which is called psycholinguistics.