SLANG UNITS SEMANTIC AND DISTRIBUTIVE ANALYSIS WITH GRAK CORPUS DATA

Abstract

The article describes a corpus-based approach to semantic and distributive analysis of Ukrainian slang unit “deriban”. The study is based on the idea of distributive semantics which deals with semantic changes due to frequency of distributions in a corpus of texts. General Regionally Annotated Corpus of Ukrainian (GRAC)  was taken as data base because it is the biggest Ukrainian corpus. It contains more than 728 million tokens (v. 11), also it has a linguistic annotation: morphological and semantic. The analysis of dictionary definitions revealed two homonymous words of different origins: the first one is from drug jargon, the second one is from criminal argot. But analysis of collocations with the word “deriban” in the corpus showed the prevalence of argot origin word in various functional styles. There are two spelling variants which are revealed in the sociolinguistic dictionaries but the corpus showed a high frequency of the one variant only. The corpus allows to identify the date of the first fixation of the word, as well as the frequency distribution by year. The slang unit semantic and distributive analysis with GRAK corpus data allows revealing both preserved archiseme of the argot unit and actualized semes and also latent semes. Discussion based on word occurrence frequencies, statistical parameters of slang unit collocation with MI and logDice scores. The result shows a stable collocation with the concepts of "land", "budget", which are obviously the most subject to division. Also it revealed some latent semes. T-score using shows the regularity of slang word spelling in quotation marks, which indicates the awareness of it as a substandard unit.