@inproceedings{zhao-etal-2026-syntactic,
title = "Syntactic Prediction through Reinforcement Learning",
author = "Zhao, Jinman and
Wang, Yining and
Liu, JIahe and
Zhang, Xueyan and
Cao, Linbo and
Huang, Jiakang and
Zhao, Yiren and
Cai, Renyi and
Ding, YiTian and
Penn, Gerald",
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.283/",
pages = "6477--6492",
ISBN = "979-8-89176-459-0",
abstract = "Syntactic prediction requires models to generate outputs that are globally consistent and satisfy hierarchical structural constraints. Although large language models have been applied to such tasks, existing approaches still rely mainly on prompting or supervised fine-tuning, and their performance generally remains below that of specialized neural syntactic models. Our experiments show that supervised fine-tuning reaches a plateau, and additional supervised training does not further improve prediction quality. We therefore propose a two-stage post-training framework for syntactic prediction. The first cold-start stage adapt the model to a unified syntax generation format, and then applies reinforcement learning on a disjoint data subset to directly optimize structure-level objectives. Experiments on constituency parsing and CCG supertagging show that Syntax-RL consistently outperforms supervised LLM baselines and achieves state-of-the-art results. Further analyses suggest that its gains come primarily from the reinforcement learning stage. These findings identify reinforcement learning as an effective post-training mechanism for improving hierarchical syntactic prediction in large language models."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="zhao-etal-2026-syntactic">
<titleInfo>
<title>Syntactic Prediction through Reinforcement Learning</title>
</titleInfo>
<name type="personal">
<namePart type="given">Jinman</namePart>
<namePart type="family">Zhao</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yining</namePart>
<namePart type="family">Wang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">JIahe</namePart>
<namePart type="family">Liu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Xueyan</namePart>
<namePart type="family">Zhang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Linbo</namePart>
<namePart type="family">Cao</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jiakang</namePart>
<namePart type="family">Huang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yiren</namePart>
<namePart type="family">Zhao</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Renyi</namePart>
<namePart type="family">Cai</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">YiTian</namePart>
<namePart type="family">Ding</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Gerald</namePart>
<namePart type="family">Penn</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>Syntactic prediction requires models to generate outputs that are globally consistent and satisfy hierarchical structural constraints. Although large language models have been applied to such tasks, existing approaches still rely mainly on prompting or supervised fine-tuning, and their performance generally remains below that of specialized neural syntactic models. Our experiments show that supervised fine-tuning reaches a plateau, and additional supervised training does not further improve prediction quality. We therefore propose a two-stage post-training framework for syntactic prediction. The first cold-start stage adapt the model to a unified syntax generation format, and then applies reinforcement learning on a disjoint data subset to directly optimize structure-level objectives. Experiments on constituency parsing and CCG supertagging show that Syntax-RL consistently outperforms supervised LLM baselines and achieves state-of-the-art results. Further analyses suggest that its gains come primarily from the reinforcement learning stage. These findings identify reinforcement learning as an effective post-training mechanism for improving hierarchical syntactic prediction in large language models.</abstract>
<identifier type="citekey">zhao-etal-2026-syntactic</identifier>
<location>
<url>https://aclanthology.org/2026.emnlp-main.283/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>6477</start>
<end>6492</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Syntactic Prediction through Reinforcement Learning
%A Zhao, Jinman
%A Wang, Yining
%A Liu, JIahe
%A Zhang, Xueyan
%A Cao, Linbo
%A Huang, Jiakang
%A Zhao, Yiren
%A Cai, Renyi
%A Ding, YiTian
%A Penn, Gerald
%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 zhao-etal-2026-syntactic
%X Syntactic prediction requires models to generate outputs that are globally consistent and satisfy hierarchical structural constraints. Although large language models have been applied to such tasks, existing approaches still rely mainly on prompting or supervised fine-tuning, and their performance generally remains below that of specialized neural syntactic models. Our experiments show that supervised fine-tuning reaches a plateau, and additional supervised training does not further improve prediction quality. We therefore propose a two-stage post-training framework for syntactic prediction. The first cold-start stage adapt the model to a unified syntax generation format, and then applies reinforcement learning on a disjoint data subset to directly optimize structure-level objectives. Experiments on constituency parsing and CCG supertagging show that Syntax-RL consistently outperforms supervised LLM baselines and achieves state-of-the-art results. Further analyses suggest that its gains come primarily from the reinforcement learning stage. These findings identify reinforcement learning as an effective post-training mechanism for improving hierarchical syntactic prediction in large language models.
%U https://aclanthology.org/2026.emnlp-main.283/
%P 6477-6492
Markdown (Informal)
[Syntactic Prediction through Reinforcement Learning](https://aclanthology.org/2026.emnlp-main.283/) (Zhao et al., EMNLP 2026)
ACL
- Jinman Zhao, Yining Wang, JIahe Liu, Xueyan Zhang, Linbo Cao, Jiakang Huang, Yiren Zhao, Renyi Cai, YiTian Ding, and Gerald Penn. 2026. Syntactic Prediction through Reinforcement Learning. In Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing, pages 6477–6492, Budapest, Hungary. Association for Computational Linguistics.