@inproceedings{hong-etal-2026-knowing,
title = "Knowing but Not Doing: Discourse-level Constraints Adherence of Large Language Models",
author = "Hong, Zhonghua and
Jiang, Sihang and
Wang, Keyi and
Meng, Xiaojun and
Wei, Jiansheng and
Liang, Jiaqing and
Xiao, Yanghua",
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.1083/",
doi = "10.18653/2026.findings-emnlp.1083",
pages = "22934--22949",
ISBN = "979-8-89176-462-0",
abstract = "Instruction following in large language models (LLMs) is commonly evaluated at the sentence or surface level, leaving discourse-level constraints underexplored. We study discourse-level instruction following by formalizing it as a structured constraint adherence problem and constructing both explicit and implicit benchmarks for evaluation. Experiments are conducted across models of different scales, revealing a systematic recognition{--}adherence gap: models can identify discourse constraints but frequently fail to adhere to them during generation with specific failure modes. This failure stems from a fundamental misalignment between standard training objectives and global structural integrity. Importantly, models show substantially improved adherence when discourse-level structural objectives are explicitly optimized."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="hong-etal-2026-knowing">
<titleInfo>
<title>Knowing but Not Doing: Discourse-level Constraints Adherence of Large Language Models</title>
</titleInfo>
<name type="personal">
<namePart type="given">Zhonghua</namePart>
<namePart type="family">Hong</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Sihang</namePart>
<namePart type="family">Jiang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Keyi</namePart>
<namePart type="family">Wang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Xiaojun</namePart>
<namePart type="family">Meng</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jiansheng</namePart>
<namePart type="family">Wei</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jiaqing</namePart>
<namePart type="family">Liang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yanghua</namePart>
<namePart type="family">Xiao</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>Instruction following in large language models (LLMs) is commonly evaluated at the sentence or surface level, leaving discourse-level constraints underexplored. We study discourse-level instruction following by formalizing it as a structured constraint adherence problem and constructing both explicit and implicit benchmarks for evaluation. Experiments are conducted across models of different scales, revealing a systematic recognition–adherence gap: models can identify discourse constraints but frequently fail to adhere to them during generation with specific failure modes. This failure stems from a fundamental misalignment between standard training objectives and global structural integrity. Importantly, models show substantially improved adherence when discourse-level structural objectives are explicitly optimized.</abstract>
<identifier type="citekey">hong-etal-2026-knowing</identifier>
<identifier type="doi">10.18653/2026.findings-emnlp.1083</identifier>
<location>
<url>https://aclanthology.org/2026.findings-emnlp.1083/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>22934</start>
<end>22949</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Knowing but Not Doing: Discourse-level Constraints Adherence of Large Language Models
%A Hong, Zhonghua
%A Jiang, Sihang
%A Wang, Keyi
%A Meng, Xiaojun
%A Wei, Jiansheng
%A Liang, Jiaqing
%A Xiao, Yanghua
%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 hong-etal-2026-knowing
%X Instruction following in large language models (LLMs) is commonly evaluated at the sentence or surface level, leaving discourse-level constraints underexplored. We study discourse-level instruction following by formalizing it as a structured constraint adherence problem and constructing both explicit and implicit benchmarks for evaluation. Experiments are conducted across models of different scales, revealing a systematic recognition–adherence gap: models can identify discourse constraints but frequently fail to adhere to them during generation with specific failure modes. This failure stems from a fundamental misalignment between standard training objectives and global structural integrity. Importantly, models show substantially improved adherence when discourse-level structural objectives are explicitly optimized.
%R 10.18653/2026.findings-emnlp.1083
%U https://aclanthology.org/2026.findings-emnlp.1083/
%U https://doi.org/10.18653/2026.findings-emnlp.1083
%P 22934-22949
Markdown (Informal)
[Knowing but Not Doing: Discourse-level Constraints Adherence of Large Language Models](https://aclanthology.org/2026.findings-emnlp.1083/) (Hong et al., Findings 2026)
ACL