Published 2026 | Version 1
Preprint Open

Lexical Generativity in English: An Empirical Study of Verb Classes, Noun Polysemy, and Prepositions

  • 1. G6 LLC

Description

Lexical Generativity in English

Lexical Generativity in English: An Empirical Study of Verb Classes, Noun Polysemy, and Prepositions

Author: Pablo Nogueira Grossi · G6 LLC · Newark NJ ORCID: 0009-0000-6496-2186 Series root: https://doi.org/10.5281/zenodo.19117399 Submitted to: International Journal of Lexicography (Oxford University Press) License: CC BY 4.0

Abstract

This article examines lexical generativity in English: the capacity of a finite lexical inventory to support a theoretically unbounded range of context-sensitive meanings in use. Drawing on three converging empirical resources — the verb classification system of Levin (1991, 1993), the frame-semantic architecture of FrameNet (Fillmore, Johnson & Petruck, 2003), and the class-membership and alternation structure of VerbNet (Kipper, Korhonen, Ryant & Palmer, 2008) — the article proposes a layered model of lexical generativity that distinguishes among:

  • (a) stored semantic primitives and qualia structure (Level I)
  • (b) argument-structure templates licensed by class membership (Level II)
  • (c) event-type composition rules governing productive meaning extension (Level III)

The empirical core consists of detailed analysis of twelve English verb classes: manner-of-motion, change-of-state, causative-inchoative alternation, communication verbs, psychological verbs, creation-and-transformation verbs, aspectual verbs, verbs of putting, spray-load verbs, contact-by-impact verbs, perception verbs, emission verbs, and verbs of appearance and disappearance. The analysis is extended to systematic noun polysemy — dot objects, type coercion, and metonymic transfer — and to the generative semantics of English spatial prepositions, with a case study of over. Throughout, the article argues that apparent lexicographic irregularities are systematic consequences of a small set of generative principles that can be stated precisely, incorporated into lexicographic description, and exploited in computational lexicography. Implications for dictionary design, large-scale lexical database annotation, and natural language processing are discussed.

Keywords: lexical generativity · verb classes · noun polysemy · prepositions · FrameNet · VerbNet · Levin classes · generative lexicon · computational lexicography · argument structure · type coercion · metonymy · qualia structure

Deposit Contents

File Description
lexical_generativity_en.pdf Main article, ~14,000 words

Article Structure

Section Content
1 Introduction: three forms of lexical generativity
2 Background: Levin verb classes, FrameNet, VerbNet
3 Theoretical framework: the three-level model
4 Verb class analyses (twelve classes)
5 Noun polysemy: dot objects, type coercion, metonymic transfer
6 Prepositions and spatial semantics (over case study)
7 Computational implications: dictionary design, database annotation, NLP
8 Discussion
9 Conclusion

Submission Status

Submitted to the International Journal of Lexicography (Oxford University Press). This preprint is posted in accordance with OUP's preprint policy.

Related Deposits

Work DOI
Principia Orthogona series root https://doi.org/10.5281/zenodo.19117399
GCM Institutional Edition (context for this article) https://doi.org/10.5281/zenodo.19513913

Coherence Bridge v8.4

coherence_bridge_v8_4.yaml is the machine-readable synchronisation file between the TOGT five-operator grammar, GCM contact geometry, and all three formal pillars. Key changes from v8.3:

  • Anantharaman–Monk source expanded to full arXiv series (arXiv:2304.02678, arXiv:2403.12576, arXiv:2502.12268); Hide–Macera–Thomas polynomial-rate follow-up (arXiv:2508.14874) noted; all five TOGT entries sharpened to precise mathematical statements.
  • Wang–Zahl source expanded to arXiv:2502.17655 + precursor arXiv:2210.09581 + Guth surveys arXiv:2505.07695 / arXiv:2508.05475; conjecture-proved status corrected throughout (was mislabelled open); claim_level_note field added to prevent misreading of analogical tag; all five TOGT entries fully populated.

Three formal pillars — sorry inventory

Pillar Lean file Proved Sorry
Discrete (Collatz) DiscreteDm3.lean v1.6 operatorDecomposition, contactForm meanContraction, lyapunovDescent, hasStructuredCycle
Continuous (Navier–Stokes) Dm3Cont.lean v1.0 operatorDecomposition, contactForm meanContraction_cont, lyapunovDescent_cont, hasStructuredAttractor
Arithmetic-analytic (BSD) BSD_dm3.lean v1.0 operatorDecomposition, contactForm meanContraction_BSD, lyapunovDescent_BSD, hasStructuredCycle_BSD

Closing the three admits on any pillar turns the corresponding conjecture into a categorical corollary of the dm³ framework.

Python Simulation — Reproduce Figures

pip install numpy matplotlib
python3 autophagy_dm3.py --out figures/

Generates all four paper figures. The nbonacci_criticality.py and nbonacci_critical_lambda.py scripts in the AXLE repository reproduce the DNLS / n-bonacci criticality figures from the companion paper (DOI: 10.5281/zenodo.20026942).

Related Deposits

Paper DOI
Principia Orthogona series root 10.5281/zenodo.19117400
This deposit (Autophagy / Triple-alpha) 10.5281/zenodo.20168812
DNLS / n-bonacci companion paper 10.5281/zenodo.20026942
Fruit-fly / MultiOrbitBioSwarm 10.5281/zenodo.19210136
GCM Institutional Edition (manifesto) 10.5281/zenodo.19513913

Keywords

dm³ operator · contact geometry · Whitney fold · autophagy · triple-alpha process · Lean 4 · Mathlib4 · formal verification · TOGT · operator grammar · coherence bridge · Collatz · Navier–Stokes · BSD conjecture · stability radius · ε₀ = 1/3 · Principia Orthogona · G6 LLC

Files

Files (139.8 kB)

Name Size Download all
md5:889be177994a4c2c009a0c6722669447
9.9 kB Download
md5:2bbe1521f49657733d4652bb507230b2
15.7 kB Download
md5:60bdf2a882749606c9a26214e1146b00
42.8 kB Download
md5:3dbe8cb584c39c5a2b03eb85ac79e162
71.3 kB Download

Additional details

Software

Repository URL
https://totogt.github.io/AXLE/index.html
Programming language
Python , Lean
Development Status
Active