Skip to content
AI News HubLIVE
Original source2 min read

Disentangling Statistical Preemption from Entrenchment in Language Models' Avoidance of Overgeneralization

Summary

An EMNLP 2026 main-conference paper runs controlled rearing experiments on language models trained on child-caregiver conversations, systematically removing preemptive versus non-preemptive evidence to test how models avoid overgeneralizations such as "Tom laughed me." The models avoid overgeneralizations but show no verb-specific preemption, only weak but non-zero abstract preemption; training-dynamics analysis indicates that in the verb-specific condition models treat competing structures as indirect positive rather than negative evidence.

SourcearXiv Computational LinguisticsAuthor: Yixuan Wang, Freda Shi, Kanishka Misra
Disentangling Statistical Preemption from Entrenchment in Language Models' Avoidance of Overgeneralization
Report an error

The correction channel is not available yet. You can copy the article reference below for later.

Correction instructions
Read article

[Submitted on 1 Sep 2026]

Title:Disentangling Statistical Preemption from Entrenchment in Language Models' Avoidance of Overgeneralization

View a PDF of the paper titled Disentangling Statistical Preemption from Entrenchment in Language Models' Avoidance of Overgeneralization, by Yixuan Wang and 2 other authors

View PDF HTML (experimental)

Abstract:How do learners avoid overgeneralizations such as Tom laughed me without explicit negative evidence? Constructionists have posited two proposals that describe indirect negative evidence against overgeneralizations: preemption (which privileges exposure to near-synonymous construction---e.g., she made him laugh) vs. entrenchment (all exposures to a verb's grammatical usages, including cases like He laughed). We disentangle these hypotheses by running controlled rearing experiments on LMs trained on child-caregiver conversations, where we systematically remove preemptive vs. non-preemptive evidence. We find that while LMs avoid overgeneralizations, they do not show preemption at a verb-specific level, instead showing weak but non-zero evidence of abstract preemption. Combined with results from analyzing the LMs' training dynamics, we find that LMs treat competing structures as indirect positive---as opposed to negative---evidence in the verb-specific condition. Insofar as preemption is the more plausible route to avoiding overgeneralizations in humans, our results point the need for there to be sensitivities to indirect negative evidence in neural network learners, and suggest new human experiments to test abstract preemption.

Comments: Accepted to EMNLP 2026 Main

Subjects:

Computation and Language (cs.CL)

Cite as: arXiv:2609.01794 [cs.CL]

(or arXiv:2609.01794v1 [cs.CL] for this version)

https://doi.org/10.48550/arXiv.2609.01794

arXiv-issued DOI via DataCite (pending registration)

Submission history

From: Yixuan Wang [view email] [v1] Tue, 1 Sep 2026 19:07:20 UTC (1,873 KB)

Full-text links:

Access Paper:

View a PDF of the paper titled Disentangling Statistical Preemption from Entrenchment in Language Models' Avoidance of Overgeneralization, by Yixuan Wang and 2 other authors

View PDF

HTML (experimental)

TeX Source

view license

Current browse context:

cs.CL

new | recent | 2026-09

Change to browse by:

cs

References & Citations

NASA ADS

Google Scholar

Semantic Scholar

Loading...

Data provided by:

Bibliographic Tools

Bibliographic and Citation Tools

Bibliographic Explorer Toggle

Bibliographic Explorer (What is the Explorer?)

Connected Papers Toggle

Connected Papers (What is Connected Papers?)

Litmaps Toggle

Litmaps (What is Litmaps?)

scite.ai Toggle

scite Smart Citations (What are Smart Citations?)

Code, Data, Media

Code, Data and Media Associated with this Article

alphaXiv Toggle

alphaXiv (What is alphaXiv?)

Links to Code Toggle

CatalyzeX Code Finder for Papers (What is CatalyzeX?)

DagsHub Toggle

DagsHub (What is DagsHub?)

GotitPub Toggle

Gotit.pub (What is GotitPub?)

Huggingface Toggle

Hugging Face (What is Huggingface?)

ScienceCast Toggle

ScienceCast (What is ScienceCast?)

Demos

Demos

Replicate Toggle

Replicate (What is Replicate?)

Spaces Toggle

Hugging Face Spaces (What is Spaces?)

Spaces Toggle

TXYZ.AI (What is TXYZ.AI?)

Related Papers

Recommenders and Search Tools

Link to Influence Flower

Influence Flower (What are Influence Flowers?)

Core recommender toggle

CORE Recommender (What is CORE?)

Author

Venue

Institution

Topic

About arXivLabs

arXivLabs: experimental projects with community collaborators

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.

Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.

Which authors of this paper are endorsers? | Disable MathJax (What is MathJax?)

Key points and analysis

Article intelligence

EngineersAdvanced

Key points

  • The study separates preemption from entrenchment by removing each type of evidence in controlled rearing experiments on language models.
  • Language models avoid overgeneralizations but lack verb-specific preemption, showing only weak abstract preemption.
  • In the verb-specific condition, models treat competing structures as indirect positive evidence rather than indirect negative evidence.
  • The authors call for sensitivity to indirect negative evidence in neural learners and propose new human experiments on abstract preemption.

Highlights and analysis are generated automatically and may contain errors. Check the original source.