@inproceedings{staruch-etal-2026-amu,
title = "{AMU}-{GEC}: A {P}olish Native Speaker Grammatical Error Correction Benchmark",
author = "Staruch, Ryszard and
Liberek, Jaros{\l}aw and
Jassem, Krzysztof and
Grali{\'n}ski, Filip and
Klimczak, Michalina and
O{\l}ubczy{\'n}ska, Sonia",
editor = "Lee, Hung-Yi and
Dev, Sunipa and
Li, Jessy and
Elliott, Desmond",
booktitle = "Findings of the {A}ssociation for {C}omputational {L}inguistics: {EMNLP} 2026",
month = oct,
year = "2026",
address = "Budapest, Hungary",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2026.findings-emnlp.895/",
pages = "18961--18971",
ISBN = "979-8-89176-462-0",
abstract = "For many languages, grammatical error correction (GEC) remains underexplored due to the lack of dedicated datasets and evaluation benchmarks. Moreover, many existing GEC datasets are based on texts written by second-language learners, which narrows the research focus to specific error types. To address these gaps, we introduce a new Polish GEC benchmark, AMU-GEC, built from texts written by native Polish speakers and manually annotated by expert linguists. The benchmark is released as a machine learning challenge, without revealing the expected labels, to prevent contamination. To set baselines, we evaluate recent multilingual and Polish LLMs, finding that smaller models may perform better when trained on monolingual (Polish) data. We also conduct a human analysis of the best performing model to gain more insights about the model behavior when it is different from human annotations. The analysis shows that most of the changes suggested by the model do not introduce errors into the texts; instead, they typically either slightly improve text fluency or represent valid alternative corrections."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="staruch-etal-2026-amu">
<titleInfo>
<title>AMU-GEC: A Polish Native Speaker Grammatical Error Correction Benchmark</title>
</titleInfo>
<name type="personal">
<namePart type="given">Ryszard</namePart>
<namePart type="family">Staruch</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jarosław</namePart>
<namePart type="family">Liberek</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Krzysztof</namePart>
<namePart type="family">Jassem</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Filip</namePart>
<namePart type="family">Graliński</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Michalina</namePart>
<namePart type="family">Klimczak</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Sonia</namePart>
<namePart type="family">Ołubczyńska</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>Findings of the Association for Computational Linguistics: EMNLP 2026</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">979-8-89176-462-0</identifier>
</relatedItem>
<abstract>For many languages, grammatical error correction (GEC) remains underexplored due to the lack of dedicated datasets and evaluation benchmarks. Moreover, many existing GEC datasets are based on texts written by second-language learners, which narrows the research focus to specific error types. To address these gaps, we introduce a new Polish GEC benchmark, AMU-GEC, built from texts written by native Polish speakers and manually annotated by expert linguists. The benchmark is released as a machine learning challenge, without revealing the expected labels, to prevent contamination. To set baselines, we evaluate recent multilingual and Polish LLMs, finding that smaller models may perform better when trained on monolingual (Polish) data. We also conduct a human analysis of the best performing model to gain more insights about the model behavior when it is different from human annotations. The analysis shows that most of the changes suggested by the model do not introduce errors into the texts; instead, they typically either slightly improve text fluency or represent valid alternative corrections.</abstract>
<identifier type="citekey">staruch-etal-2026-amu</identifier>
<location>
<url>https://aclanthology.org/2026.findings-emnlp.895/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>18961</start>
<end>18971</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T AMU-GEC: A Polish Native Speaker Grammatical Error Correction Benchmark
%A Staruch, Ryszard
%A Liberek, Jarosław
%A Jassem, Krzysztof
%A Graliński, Filip
%A Klimczak, Michalina
%A Ołubczyńska, Sonia
%Y Lee, Hung-Yi
%Y Dev, Sunipa
%Y Li, Jessy
%Y Elliott, Desmond
%S Findings of the Association for Computational Linguistics: EMNLP 2026
%D 2026
%8 October
%I Association for Computational Linguistics
%C Budapest, Hungary
%@ 979-8-89176-462-0
%F staruch-etal-2026-amu
%X For many languages, grammatical error correction (GEC) remains underexplored due to the lack of dedicated datasets and evaluation benchmarks. Moreover, many existing GEC datasets are based on texts written by second-language learners, which narrows the research focus to specific error types. To address these gaps, we introduce a new Polish GEC benchmark, AMU-GEC, built from texts written by native Polish speakers and manually annotated by expert linguists. The benchmark is released as a machine learning challenge, without revealing the expected labels, to prevent contamination. To set baselines, we evaluate recent multilingual and Polish LLMs, finding that smaller models may perform better when trained on monolingual (Polish) data. We also conduct a human analysis of the best performing model to gain more insights about the model behavior when it is different from human annotations. The analysis shows that most of the changes suggested by the model do not introduce errors into the texts; instead, they typically either slightly improve text fluency or represent valid alternative corrections.
%U https://aclanthology.org/2026.findings-emnlp.895/
%P 18961-18971
Markdown (Informal)
[AMU-GEC: A Polish Native Speaker Grammatical Error Correction Benchmark](https://aclanthology.org/2026.findings-emnlp.895/) (Staruch et al., Findings 2026)
ACL