Problems in the study of derivational potential: automated construction of hypothetical derivatives

The aim of the present study is to justify the use of automated methods for investigating the derivational potential of Russian lexicon and to demonstrate the principles of developing an algorithm capable of modeling derived nouns with the meaning of “a bearer of a mental attribute”. The scientific novelty of the research lies in the development of a new approach to the selection of derivates: the identification of uncodified derivatives is carried out not randomly but systematically – through automated construction of hypothetical derivatives followed by verification of their presence or absence in the internet space. During the study, a list of adjectives denoting mental characteristics of a person and suffixes capable of participating in the formation of personal nouns is determined. Subsequently, a script in Python is developed that automatically models derivatives based on this material. Additionally, the filtering of the generated derivatives is performed to exclude incorrect morpheme combinations. The result of the research is a list of hypothetical derivatives and a ready algorithm that can be adapted for constructing derived units formed by other models with different sets of stems and affixes.