A Systematic Study of Compositional Syntactic Transformer Language Models
Yida Zhao, Hao Xve, Xiang Hu, Kewei Tu
Abstract
Syntactic language models (SLMs) enhance Transformers by incorporating syntactic biases through the modeling of linearized syntactic parse trees alongside surface sentences. This paper focuses on compositional SLMs that are based on constituency parse trees and contain explicit bottom-up composition of constituent representations. We identify key aspects of design choices in existing compositional SLMs and propose a unified framework encompassing both existing models and novel variants. We conduct a comprehensive empirical evaluation of all the variants in our framework across language modeling, syntactic generalization, summarization, and inference efficiency. Based on the experimental results, we make multiple recommendations on the design of compositional SLMs. Our code is released at https://github.com/zhaoyd1/compositional_SLMs.
BibTeX
@inproceedings{zhao-etal-2025-systematic,
title = "A Systematic Study of Compositional Syntactic Transformer Language Models",
author = "Zhao, Yida and
Xve, Hao and
Hu, Xiang and
Tu, Kewei",
editor = "Che, Wanxiang and
Nabende, Joyce and
Shutova, Ekaterina and
Pilehvar, Mohammad Taher",
booktitle = "Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)",
month = jul,
year = "2025",
address = "Vienna, Austria",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2025.acl-long.350/",
doi = "10.18653/v1/2025.acl-long.350",
pages = "7070--7083",
ISBN = "979-8-89176-251-0"
}