# Do Attention Heads in BERT Track Syntactic Dependencies?

arXiv: https://arxiv.org/abs/1911.12246 · HTML: https://mlsanity.com/paper/1911.12246/

## Key facts

- Cited by 88 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 27 Nov 2019.

## Built on by (most cited first, 30 of 88)

- [Probing Classifiers: Promises, Shortcomings, and Advances](https://mlsanity.com/paper/2102.12452/) (2021)
- [Pre-Trained Models: Past, Present and Future](https://mlsanity.com/paper/2106.07139/) (2021)
- [Causal Analysis of Syntactic Agreement Mechanisms in Neural Language Models](https://mlsanity.com/paper/2106.06087/) (2021)
- [Knowledgeable or Educated Guess? Revisiting Language Models as Knowledge Bases](https://mlsanity.com/paper/2106.09231/) (2021)
- [How Do Transformers Learn Topic Structure: Towards a Mechanistic Understanding](https://mlsanity.com/paper/2303.04245/) (2023)
- [A Primer on the Inner Workings of Transformer-based Language Models](https://mlsanity.com/paper/2405.00208/) (2024)
- [Same Pre-training Loss, Better Downstream: Implicit Bias Matters for Language Models](https://mlsanity.com/paper/2210.14199/) (2022)
- [On the Role of Attention Heads in Large Language Model Safety](https://mlsanity.com/paper/2410.13708/) (2024)
- [Attention Satisfies: A Constraint-Satisfaction Lens on Factual Errors of Language Models](https://mlsanity.com/paper/2309.15098/) (2023)
- [Mixture of Attention Heads: Selecting Attention Heads Per Token](https://mlsanity.com/paper/2210.05144/) (2022)
- [Explainable and Interpretable Multimodal Large Language Models: A Comprehensive Survey](https://mlsanity.com/paper/2412.02104/) (2024)
- [Incorporating Residual and Normalization Layers into Analysis of Masked Language Models](https://mlsanity.com/paper/2109.07152/) (2021)
- [What Do They Capture? -- A Structural Analysis of Pre-Trained Language Models for Source Code](https://mlsanity.com/paper/2202.06840/) (2022)
- [Artificial Text Detection via Examining the Topology of Attention Maps](https://mlsanity.com/paper/2109.04825/) (2021)
- [GlobEnc: Quantifying Global Token Attribution by Incorporating the Whole Encoder Layer in Transformers](https://mlsanity.com/paper/2205.03286/) (2022)
- [Quantifying Context Mixing in Transformers](https://mlsanity.com/paper/2301.12971/) (2023)
- [Grad-SAM: Explaining Transformers via Gradient Self-Attention Maps](https://mlsanity.com/paper/2204.11073/) (2022)
- [Sparse Attention with Linear Units](https://mlsanity.com/paper/2104.07012/) (2021)
- [The Impact of Depth on Compositional Generalization in Transformer Language Models](https://mlsanity.com/paper/2310.19956/) (2023)
- [AlignAtt: Using Attention-based Audio-Translation Alignments as a Guide for Simultaneous Speech Translation](https://mlsanity.com/paper/2305.11408/) (2023)
- [Attention as a Guide for Simultaneous Speech Translation](https://mlsanity.com/paper/2212.07850/) (2022)
- [Transformers are uninterpretable with myopic methods: a case study with bounded Dyck grammars](https://mlsanity.com/paper/2312.01429/) (2023)
- [Mechanistic?](https://mlsanity.com/paper/2410.09087/) (2024)
- [Language Models as Models of Language](https://mlsanity.com/paper/2408.07144/) (2024)
- [The Case for Translation-Invariant Self-Attention in Transformer-Based Language Models](https://mlsanity.com/paper/2106.01950/) (2021)
- [The Life Cycle of Knowledge in Big Language Models: A Survey](https://mlsanity.com/paper/2303.07616/) (2023)
- [Relational Composition in Neural Networks: A Survey and Call to Action](https://mlsanity.com/paper/2407.14662/) (2024)
- [Attention Can Reflect Syntactic Structure (If You Let It)](https://mlsanity.com/paper/2101.10927/) (2021)
- [Learning Relation Alignment for Calibrated Cross-modal Retrieval](https://mlsanity.com/paper/2105.13868/) (2021)
- [How Much Does Attention Actually Attend? Questioning the Importance of Attention in Pretrained Transformers](https://mlsanity.com/paper/2211.03495/) (2022)

## Abstract (arXiv)

We investigate the extent to which individual attention heads in pretrained transformer language models, such as BERT and RoBERTa, implicitly capture syntactic dependency relations. We employ two methods---taking the maximum attention weight and computing the maximum spanning tree---to extract implicit dependency relations from the attention weights of each layer/head, and compare them to the ground-truth Universal Dependency (UD) trees. We show that, for some UD relation types, there exist heads that can recover the dependency type significantly better than baselines on parsed English text, suggesting that some self-attention heads act as a proxy for syntactic structure. We also analyze BERT fine-tuned on two datasets---the syntax-oriented CoLA and the semantics-oriented MNLI---to investigate whether fine-tuning affects the patterns of their self-attention, but we do not observe substantial differences in the overall dependency relations extracted using our methods. Our results suggest that these models have some specialist attention heads that track individual dependency types, but no generalist head that performs holistic parsing significantly better than a trivial baseline, and that analyzing attention weights directly may not reveal much of the syntactic knowledge that BERT-style models are known to learn.