# An Empirical Study of Compound PCFGs

arXiv: https://arxiv.org/abs/2103.02298 · HTML: https://mlsanity.com/paper/2103.02298/

## Key facts

- Cited by 6 papers in the map (data through 14 Oct 2026).
- Builds on 3 of its 13 references that are in the map.
- First posted on arXiv 3 Mar 2021.

## Builds on (most cited first, 3 of 3)

- [Compound Probabilistic Context-Free Grammars for Grammar Induction](https://mlsanity.com/paper/1906.10225/) (2019)
- [Torch-Struct: Deep Structured Prediction Library](https://mlsanity.com/paper/2002.00876/) (2020)
- [Visually Grounded Compound PCFGs](https://mlsanity.com/paper/2009.12404/) (2020)

## Built on by (most cited first, 6 of 6)

- [GFlowNet-EM for learning compositional latent variable models](https://mlsanity.com/paper/2302.06576/) (2023)
- [Sequence-to-Sequence Learning with Latent Neural Grammars](https://mlsanity.com/paper/2109.01135/) (2021)
- [PCFGs Can Do Better: Inducing Probabilistic Context-Free Grammars with Many Symbols](https://mlsanity.com/paper/2104.13727/) (2021)
- [Contextual Distortion Reveals Constituency: Masked Language Models are Implicit Parsers](https://mlsanity.com/paper/2306.00645/) (2023)
- [On the Transferability of Visually Grounded PCFGs](https://mlsanity.com/paper/2310.14107/) (2023)
- [On Eliciting Syntax from Language Models via Hashing](https://mlsanity.com/paper/2410.04074/) (2024)

## Abstract (arXiv)

Compound probabilistic context-free grammars (C-PCFGs) have recently established a new state of the art for unsupervised phrase-structure grammar induction. However, due to the high space and time complexities of chart-based representation and inference, it is difficult to investigate C-PCFGs comprehensively. In this work, we rely on a fast implementation of C-PCFGs to conduct an evaluation complementary to that of~citetkim-etal-2019-compound. We start by analyzing and ablating C-PCFGs on English treebanks. Our findings suggest that (1) C-PCFGs are data-efficient and can generalize to unseen sentence/constituent lengths; and (2) C-PCFGs make the best use of sentence-level information in generating preterminal rule probabilities. We further conduct a multilingual evaluation of C-PCFGs. The experimental results show that the best configurations of C-PCFGs, which are tuned on English, do not always generalize to morphology-rich languages.