Prototype-Based Categorization of Words and Senses: A Review of Cognitive, Lexical-Semantic, and Computational Approaches

Abstract

This review article examines prototype-based categorization of words and senses as an interdisciplinary framework at the intersection of cognitive psychology, lexical semantics, psycholinguistics, lexicography, and computational linguistics. The paper outlines the transition from the classical view of categories based on necessary and sufficient features to prototype theory, which emphasizes graded membership, typicality effects, and family resemblance. It then analyzes how these ideas were adapted in linguistic research to describe lexical categories, polysemy, and the internal organization of word meaning through central and peripheral senses, radial semantic networks, and context-sensitive meaning activation. Special attention is given to theoretical and methodological debates, including the distinction between polysemy and homonymy, the psychological reality of discrete word senses, and the limitations of purely introspective semantic analysis. The review also discusses corpus-based and lexicographic critiques, as well as modern computational reinterpretations of prototype structure in WordNet-based modeling, word sense disambiguation, and multi-prototype vector representations. The article argues that prototype-based categorization remains highly productive when used not as a universal replacement for all semantic models, but as an explanatory framework for probabilistic structure, semantic centrality, and context-driven variation in lexical meaning.