The counting principle makes number words unique.

Area of Science:

• Linguistics
• Cognitive Science
• Semantics

Background:

• Open-class lexical items are modified by selectors, narrowing their meaning (e.g., 'kinda table').
• Number words possess precise denotations and cannot take selectors, but can be broadened (e.g., 'approximately seven').

Purpose of the Study:

• To investigate the distinct distributional patterns of number words versus open-class lexical items.
• To propose a framework distinguishing sparse (open-class) and dense (number) lexical domains based on distributional properties.

Main Methods:

• Empirical analysis of number word modification in the Corpus of Contemporary American English (COCA) and Hebrew (HeTenTen).
• Computational linguistic analysis using synchronic word2vec models to compare sparse and dense lexemes.
• Examination of numeral usage in non-counting communities and lexical renewal rates.

Main Results:

• Number words exhibit different modification patterns than open-class words, supporting their distinct lexical domain classification.
• Sparse and dense lexical domains show differential distributions in word2vec models.
• Numeral use and renewal rates further differentiate sparse and dense lexical types.

Conclusions:

• Number words and open-class words belong to distinct lexical domains (dense vs. sparse) due to differences in prototypicality and modification patterns.
• The proposed sparse/dense distinction provides a robust framework for understanding lexical semantics and distribution.
• Findings contribute to theories of lexical categorization and semantic change.