Corpus-Based Profiles of Russian Nouns: From Grammatical Number to Lexical Semantics

Abstract

A grammatical profile indicating the relative frequency distribution of the inflected forms of a word in a corpus is a tool for exploring lexical semantics. However, previous attempts to infer semantically relevant hierarchies of nouns from frequency biases within their grammatical forms seem to have failed. In this paper we explore the distinctive power of grammatical profiles of Russian nouns using the ratio of plural forms as observed in the Russian National Corpus (cf. roditel? ?parent? having 95% plural forms and mama ?mom? having just 2% plural forms). We claim that since frequent nouns for the most part are semantically ambiguous, their profiles cannot reveal any straightforward effects for large lexical classes. Instead of working on the macro-lexical level we focus on micro-effects within specific taxonomic groups, studying grammatical profiles of body part names, kinship terms, names of vehicles and emotions. The analysis involves the notion of functional frames which represent how objectsevents are typically used and typically observed. Our case studies show that grammatical profiles help to structure each group and correlate with certain properties of functional frames associated with nouns