@inproceedings{wang-etal-2026-benchmarking-failure,
title = "From Benchmarking to Failure Diagnosis: Diagnosing Classical {C}hinese Understanding in {LLM}s via {IRT} and Error Attribution",
author = "Wang, Jiong and
Li, Tengju and
Yu, Shengquan",
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.618/",
doi = "10.18653/2026.findings-emnlp.618",
pages = "12991--13005",
ISBN = "979-8-89176-462-0",
abstract = "Literary Chinese reading comprehension is a challenging task for large language models (LLMs), requiring understanding of implicit meanings, contextual reasoning, and discourse-level integration in a low-resource setting. However, existing evaluations rely primarily on aggregate accuracy, which provides limited insight into model capabilities, error sources, and reliability under uncertainty. We propose DELit, a diagnostic evaluation framework that moves beyond aggregate metrics by introducing a hierarchical capability{--}error taxonomy, a Deep-MIRT-based ability estimation model, and a capability-decoupled residual analysis for behavioral diagnosis. Together, these components enable fine-grained interpretation of model performance at both capability and error-type levels. Experiments on mainstream LLMs reveal that (i) discourse-level understanding remains a consistent bottleneck despite strong performance on local semantic tasks; (ii) similar overall accuracy can correspond to substantially different capability profiles; and (iii) overconfident errors on difficult items concentrate on logically constrained reasoning, highlighting reliability risks that are not captured by standard evaluation metrics."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="wang-etal-2026-benchmarking-failure">
<titleInfo>
<title>From Benchmarking to Failure Diagnosis: Diagnosing Classical Chinese Understanding in LLMs via IRT and Error Attribution</title>
</titleInfo>
<name type="personal">
<namePart type="given">Jiong</namePart>
<namePart type="family">Wang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Tengju</namePart>
<namePart type="family">Li</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Shengquan</namePart>
<namePart type="family">Yu</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>Literary Chinese reading comprehension is a challenging task for large language models (LLMs), requiring understanding of implicit meanings, contextual reasoning, and discourse-level integration in a low-resource setting. However, existing evaluations rely primarily on aggregate accuracy, which provides limited insight into model capabilities, error sources, and reliability under uncertainty. We propose DELit, a diagnostic evaluation framework that moves beyond aggregate metrics by introducing a hierarchical capability–error taxonomy, a Deep-MIRT-based ability estimation model, and a capability-decoupled residual analysis for behavioral diagnosis. Together, these components enable fine-grained interpretation of model performance at both capability and error-type levels. Experiments on mainstream LLMs reveal that (i) discourse-level understanding remains a consistent bottleneck despite strong performance on local semantic tasks; (ii) similar overall accuracy can correspond to substantially different capability profiles; and (iii) overconfident errors on difficult items concentrate on logically constrained reasoning, highlighting reliability risks that are not captured by standard evaluation metrics.</abstract>
<identifier type="citekey">wang-etal-2026-benchmarking-failure</identifier>
<identifier type="doi">10.18653/2026.findings-emnlp.618</identifier>
<location>
<url>https://aclanthology.org/2026.findings-emnlp.618/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>12991</start>
<end>13005</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T From Benchmarking to Failure Diagnosis: Diagnosing Classical Chinese Understanding in LLMs via IRT and Error Attribution
%A Wang, Jiong
%A Li, Tengju
%A Yu, Shengquan
%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 wang-etal-2026-benchmarking-failure
%X Literary Chinese reading comprehension is a challenging task for large language models (LLMs), requiring understanding of implicit meanings, contextual reasoning, and discourse-level integration in a low-resource setting. However, existing evaluations rely primarily on aggregate accuracy, which provides limited insight into model capabilities, error sources, and reliability under uncertainty. We propose DELit, a diagnostic evaluation framework that moves beyond aggregate metrics by introducing a hierarchical capability–error taxonomy, a Deep-MIRT-based ability estimation model, and a capability-decoupled residual analysis for behavioral diagnosis. Together, these components enable fine-grained interpretation of model performance at both capability and error-type levels. Experiments on mainstream LLMs reveal that (i) discourse-level understanding remains a consistent bottleneck despite strong performance on local semantic tasks; (ii) similar overall accuracy can correspond to substantially different capability profiles; and (iii) overconfident errors on difficult items concentrate on logically constrained reasoning, highlighting reliability risks that are not captured by standard evaluation metrics.
%R 10.18653/2026.findings-emnlp.618
%U https://aclanthology.org/2026.findings-emnlp.618/
%U https://doi.org/10.18653/2026.findings-emnlp.618
%P 12991-13005
Markdown (Informal)
[From Benchmarking to Failure Diagnosis: Diagnosing Classical Chinese Understanding in LLMs via IRT and Error Attribution](https://aclanthology.org/2026.findings-emnlp.618/) (Wang et al., Findings 2026)
ACL