arXiv — NLP / Computation & Language · · 3 min read

A Mechanistic Understanding of Pronoun Fidelity in LLMs

Mirrored from arXiv — NLP / Computation & Language for archival readability. Support the source by reading on the original site.

Computer Science > Computation and Language

arXiv:2606.16407 (cs)
[Submitted on 15 Jun 2026]

Title:A Mechanistic Understanding of Pronoun Fidelity in LLMs

View a PDF of the paper titled A Mechanistic Understanding of Pronoun Fidelity in LLMs, by Katharina Trinley and Jesujoba O. Alabi and Dietrich Klakow and Vagrant Gautam
View PDF HTML (experimental)
Abstract:Faithful and robust pronoun use is important for fair and coherent generations, yet large language models largely fail when multiple referents use different pronouns. To study the interplay of reasoning, repetition, and bias in this task, prior work relies exclusively on behavioural approaches, which may not reflect a model's internal workings. Therefore, we provide a mechanistic, model-internal perspective on pronoun fidelity, testing whether three mechanisms -- group entity binding (G), recency bias (R), and stereotypical bias (S) -- are causally implemented across several SOTA language models. Using Boundless Distributed Alignment Search, we find all three coexist as causal subspaces distributed across network depth. No single mechanism fully explains model behaviour, but a combination of the three consistently accounts for 91-99.5%. An attention head analysis further reveals two competing copying routes; group binding and stereotype share a localized concept-level route that retrieves a bound occupation-pronoun unit, while recency uses a distributed token-level route that repeats surface forms. In sum, pronoun fidelity arises from competition between simultaneously active causal subspaces.
Subjects: Computation and Language (cs.CL); Machine Learning (cs.LG)
Cite as: arXiv:2606.16407 [cs.CL]
  (or arXiv:2606.16407v1 [cs.CL] for this version)
  https://doi.org/10.48550/arXiv.2606.16407
arXiv-issued DOI via DataCite (pending registration)

Submission history

From: Vagrant Gautam [view email]
[v1] Mon, 15 Jun 2026 08:45:53 UTC (286 KB)
Full-text links:

Access Paper:

    View a PDF of the paper titled A Mechanistic Understanding of Pronoun Fidelity in LLMs, by Katharina Trinley and Jesujoba O. Alabi and Dietrich Klakow and Vagrant Gautam
  • View PDF
  • HTML (experimental)
  • TeX Source

Current browse context:

cs.CL
< prev   |   next >
Change to browse by:

References & Citations

Loading...

BibTeX formatted citation

loading...
Data provided by:

Bookmark

BibSonomy Reddit
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?)
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.

Discussion (0)

Sign in to join the discussion. Free account, 30 seconds — email code or GitHub.

Sign in →

No comments yet. Sign in and be the first to say something.

More from arXiv — NLP / Computation & Language