Identification of Metaphors With the Help of Machine Learning

Data Mining Methods to Compare Englishes

“…The volume of the research corpus, formed based on the results of semi-automatic processing of corpus queries, amounted to 65,000 tokens. Previously, to visualize the results of a cryptotype study, Chernov's faces were used [10]. But in our research, given the rather large volume of the resulting research corpus and the need to bring together parameters of different quality (namely, 20 variants of the English language under consideration, 23 names of emotions, and 6 nominal cryptotypes), the use of this method turned out to be impossible.…”

“…This metaphor is meant as the discourse evidence of the noun question belonging to the periphery of cryptotype 'Sharp objects'. At the moment, 6 cryptotypes of the English language have been identified and described (they correspond with explicit lexical and grammatical categories of some other world languages according to a typological research): the cryptotype Res Liquidae (a class of liquids, the prototype is 'water'), Res Acutae (a class of sharp objects, the prototype is 'thorn'), Res Filiformes (a class of thin objects of unstable form, the prototype is 'thread'), Res Rotundae (a class of round objects, the prototype is 'ball'), Res Parvae (a class of hand-fitting objects, the prototype is 'apple'), Res Longae Penetrantes (a class of solid, long, pointed objects, the prototype is 'stick') [8,11,12]. For example, the nominal cryptotype of the English language Res Liquidae includes such nouns as water, blood, milk, and other nominations of objects of reality that exist in a liquid state.…”