The Emergence and Sparsification of a Syntactic Circuit

Qing Yao; Sasha Boguraev; Tiago Pimentel; Kyle Mahowald

The Emergence and Sparsification of a Syntactic Circuit

Qing Yao, Sasha Boguraev, Tiago Pimentel, Kyle Mahowald

Correct Metadata for

Use this form to create a GitHub issue with structured data describing the correction. You will need a GitHub account. Once you create that issue, the correction will be reviewed by a staff member.

⚠️ Mobile Users: Submitting this form to create a new issue will only work with github.com, not the GitHub Mobile app.

Important: The Anthology treat PDFs as authoritative. Please use this form only to correct data that is out of line with the PDF. See our corrections guidelines if you need to change the PDF.

Title Adjust the title. Retain tags such as <fixed-case>.

Authors Adjust author names and order to match the PDF.

Abstract Correct abstract if needed. Retain XML formatting tags such as <tex-math>. You may use ... for bold, ... for italic, ... for underline, <sc>...</sc> for small-caps, <tt>...<tt> for typewriter text, <url>...</url> for URLs, <a href=...> for hyperlinks, and <par/> for paragraph breaks.

Verification against PDF Ensure that the new title/authors match the snapshot below. (If there is no snapshot or it is too small, consult the PDF.)

Authors concatenated from the text boxes above:

ALL author names match the snapshot above—including middle initials, hyphens, and accents.

Abstract

To understand the emergence of grammatical behavior in language models, we track circuits for subject–verb number agreement in pythia-1b and ten seeds of pythia-410m across training. We find that subject–verb agreement is learned early, and is initially handled by a diffuse collection of neurons. However, after agreement *behavior* has stabilized, the *mechanisms* responsible for it continue to reorganize. The attention heads which mediate agreement often shift during training: in the case of pythia-1b, behavior is controlled by two heads early in training but by a third (different) head later. Furthermore, the circuit shrinks over training, with pythia-1b’s final circuit being strikingly sparse. We take these results to be informative about how linguistic knowledge emerges and is represented in a language model.

Anthology ID:: 2026.emnlp-main.732
Volume:: Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing
Month:: October
Year:: 2026
Address:: Budapest, Hungary
Editors:: Hung-Yi Lee, Sunipa Dev, Jessy Li, Desmond Elliott
Venue:: EMNLP
SIG:: SIGDAT
Publisher:: Association for Computational Linguistics
Note:
Pages:: 16987–17012
Language:
URL:: https://aclanthology.org/2026.emnlp-main.732/
DOI:
Bibkey:
Cite (ACL):: Qing Yao, Sasha Boguraev, Tiago Pimentel, and Kyle Mahowald. 2026. The Emergence and Sparsification of a Syntactic Circuit. In Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing, pages 16987–17012, Budapest, Hungary. Association for Computational Linguistics.
Cite (Informal):: The Emergence and Sparsification of a Syntactic Circuit (Yao et al., EMNLP 2026)
Copy Citation:
PDF:: https://aclanthology.org/2026.emnlp-main.732.pdf

PDF Cite Search Fix data

Export citation

BibTeX
MODS XML
Endnote
Preformatted

@inproceedings{yao-etal-2026-emergence,
    title = "The Emergence and Sparsification of a Syntactic Circuit",
    author = "Yao, Qing  and
      Boguraev, Sasha  and
      Pimentel, Tiago  and
      Mahowald, Kyle",
    editor = "Lee, Hung-Yi  and
      Dev, Sunipa  and
      Li, Jessy  and
      Elliott, Desmond",
    booktitle = "Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing",
    month = oct,
    year = "2026",
    address = "Budapest, Hungary",
    publisher = "Association for Computational Linguistics",
    url = "https://aclanthology.org/2026.emnlp-main.732/",
    pages = "16987--17012",
    ISBN = "979-8-89176-459-0",
    abstract = "To understand the emergence of grammatical behavior in language models, we track circuits for subject{--}verb number agreement in pythia-1b and ten seeds of pythia-410m across training. We find that subject{--}verb agreement is learned early, and is initially handled by a diffuse collection of neurons. However, after agreement *behavior* has stabilized, the *mechanisms* responsible for it continue to reorganize. The attention heads which mediate agreement often shift during training: in the case of pythia-1b, behavior is controlled by two heads early in training but by a third (different) head later. Furthermore, the circuit shrinks over training, with pythia-1b{'}s final circuit being strikingly sparse. We take these results to be informative about how linguistic knowledge emerges and is represented in a language model."
}

Download as File

<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="yao-etal-2026-emergence">
    <titleInfo>
        <title>The Emergence and Sparsification of a Syntactic Circuit</title>
    </titleInfo>
    <name type="personal">
        <namePart type="given">Qing</namePart>
        <namePart type="family">Yao</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Sasha</namePart>
        <namePart type="family">Boguraev</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Tiago</namePart>
        <namePart type="family">Pimentel</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Kyle</namePart>
        <namePart type="family">Mahowald</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <originInfo>
        <dateIssued>2026-10</dateIssued>
    </originInfo>
    <typeOfResource>text</typeOfResource>
    <relatedItem type="host">
        <titleInfo>
            <title>Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing</title>
        </titleInfo>
        <name type="personal">
            <namePart type="given">Hung-Yi</namePart>
            <namePart type="family">Lee</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <name type="personal">
            <namePart type="given">Sunipa</namePart>
            <namePart type="family">Dev</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <name type="personal">
            <namePart type="given">Jessy</namePart>
            <namePart type="family">Li</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <name type="personal">
            <namePart type="given">Desmond</namePart>
            <namePart type="family">Elliott</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <originInfo>
            <publisher>Association for Computational Linguistics</publisher>
            <place>
                <placeTerm type="text">Budapest, Hungary</placeTerm>
            </place>
        </originInfo>
        <genre authority="marcgt">conference publication</genre>
        <identifier type="isbn">979-8-89176-459-0</identifier>
    </relatedItem>
    <abstract>To understand the emergence of grammatical behavior in language models, we track circuits for subject–verb number agreement in pythia-1b and ten seeds of pythia-410m across training. We find that subject–verb agreement is learned early, and is initially handled by a diffuse collection of neurons. However, after agreement *behavior* has stabilized, the *mechanisms* responsible for it continue to reorganize. The attention heads which mediate agreement often shift during training: in the case of pythia-1b, behavior is controlled by two heads early in training but by a third (different) head later. Furthermore, the circuit shrinks over training, with pythia-1b’s final circuit being strikingly sparse. We take these results to be informative about how linguistic knowledge emerges and is represented in a language model.</abstract>
    <identifier type="citekey">yao-etal-2026-emergence</identifier>
    <location>
        <url>https://aclanthology.org/2026.emnlp-main.732/</url>
    </location>
    <part>
        <date>2026-10</date>
        <extent unit="page">
            <start>16987</start>
            <end>17012</end>
        </extent>
    </part>
</mods>
</modsCollection>

Download as File

%0 Conference Proceedings
%T The Emergence and Sparsification of a Syntactic Circuit
%A Yao, Qing
%A Boguraev, Sasha
%A Pimentel, Tiago
%A Mahowald, Kyle
%Y Lee, Hung-Yi
%Y Dev, Sunipa
%Y Li, Jessy
%Y Elliott, Desmond
%S Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing
%D 2026
%8 October
%I Association for Computational Linguistics
%C Budapest, Hungary
%@ 979-8-89176-459-0
%F yao-etal-2026-emergence
%X To understand the emergence of grammatical behavior in language models, we track circuits for subject–verb number agreement in pythia-1b and ten seeds of pythia-410m across training. We find that subject–verb agreement is learned early, and is initially handled by a diffuse collection of neurons. However, after agreement *behavior* has stabilized, the *mechanisms* responsible for it continue to reorganize. The attention heads which mediate agreement often shift during training: in the case of pythia-1b, behavior is controlled by two heads early in training but by a third (different) head later. Furthermore, the circuit shrinks over training, with pythia-1b’s final circuit being strikingly sparse. We take these results to be informative about how linguistic knowledge emerges and is represented in a language model.
%U https://aclanthology.org/2026.emnlp-main.732/
%P 16987-17012

Download as File

Markdown (Informal)

[The Emergence and Sparsification of a Syntactic Circuit](https://aclanthology.org/2026.emnlp-main.732/) (Yao et al., EMNLP 2026)

The Emergence and Sparsification of a Syntactic Circuit (Yao et al., EMNLP 2026)

ACL

Qing Yao, Sasha Boguraev, Tiago Pimentel, and Kyle Mahowald. 2026. The Emergence and Sparsification of a Syntactic Circuit. In Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing, pages 16987–17012, Budapest, Hungary. Association for Computational Linguistics.