Speakers
Description
The paper presents a count-based semantic vector space model for Ukrainian, which has been applied for the semantic change detection task. The approach assumes creation of multidimensional vector representations of occurrences for a particular lexeme or a group of related lexemes with further visual and quantitative analysis of the obtained semantic vector space. The multidimensional space has been reduced to 2D for visual data analysis with the Multidimensional Scaling technique. The paper described two case studies to show how the proposed R & 251D workflow helps revealing potential semantic change events and discuss benefits and limitations of the approach. One case study traces the disappearance of a regional sense, and another identifies the appearance of a new metaphoric sense that is widespread in the Ukrainian media discourse.
