Typologies and Automatic Detection of Regular Polysemy

Ахмадуллина, Алина

Lingvospektr · 2025-yil

Annotatsiya

The paper surveys theoretical and computational approaches to regular (systematic) polysemy – recurring, rule-governed sense alternations such as organization/building, content/container, and artifact/information. First, it outlines typologies developed in formal and lexical semantics, including Pustejovsky’s dot-type analysis and lexical-rule accounts that capture regular alternations across lemmas. Next, it distills practical patterns for annotation and modeling (selectional preferences, dependency contexts, co-predication diagnostics). The computational part synthesizes rule-based, distributional, and neural pipelines for automatic detection: (1) seed pattern discovery from lexical resources, (2) context clustering with contextualized embeddings, (3) sense-type recognition via weak supervision, and (4) linkage to multilingual inventories (e.g., BabelNet). We propose an evaluation protocol combining precision/recall on curated pattern sets with human judgments of co-predication acceptability. Short case studies for Russian and Uzbek illustrate resource-lean heuristics. The contribution is a consolidated, implementation-oriented framework for typologizing and automatically detecting regular polysemy suitable for low-resource settings as well as large-scale multilingual lexicons.

Maqola ma’lumotlari
MualliflarАхмадуллина, Алина
JurnalLingvospektr
Nashr sanasi2025-10-12
Jild9
Son1
Betlar108-112
TilRus

Kalit so‘zlar

Регулярная полисемия, систематическая полисемия, типология, генеративный лексикон, лексические правила, дизамбигуация значений, дистрибутивная семантика, BabelNet, Muntazam polisemiya, sistematik polisemiya, tipologiya, generativ leksikon, lisoniy qoidalar, soʻz ma’nosini ajratish, taqsimot semantikasi, BabelNet

Ilmiy soha

Lingvospektr jurnalidan boshqa maqolalar

Lingvospektr — barcha maqolalar