mlsanityData to 7 Oct 2026

Probing for Constituency Structure in Neural Language Models

arXiv 2204.06201 ↗First posted 13 Apr 2022cs.CLcited by 13

Not on the map. Why?

Abstract (arXiv)

In this paper, we investigate to which extent contextual neural language models (LMs) implicitly learn syntactic structure. More concretely, we focus on constituent structure as represented in the Penn Treebank (PTB). Using standard probing techniques based on diagnostic classifiers, we assess the accuracy of representing constituents of different categories within the neuron activations of a LM such as RoBERTa. In order to make sure that our probe focuses on syntactic knowledge and not on implicit semantic generalizations, we also experiment on a PTB version that is obtained by randomly replacing constituents with each other while keeping syntactic structure, i.e., a semantically ill-formed but syntactically well-formed version of the PTB. We find that 4 pretrained transfomer LMs obtain high performance on our probing tasks even on manipulated data, suggesting that semantic and syntactic knowledge in their representations can be separated and that constituency information is in fact learned by the LM. Moreover, we show that a complete constituency tree can be linearly separated from LM representations.

Key facts. Cited by 13 papers in the map (data through 7 Oct 2026). Builds on 31 of its 46 references that are in the map. First posted on arXiv 13 Apr 2022.

Builds on

31 of 46 references.

References · 31

  1. 2018BERT: Pre-training of Deep Bidirectional Transformers for Language Understandingcited by 40,204
  2. 2019RoBERTa: A Robustly Optimized BERT Pretraining Approachcited by 13,266
  3. 2018GLUE: A Multi-Task Benchmark and Analysis Platform for Natural Language Understandingcited by 4,854
  4. 2019DistilBERT, a distilled version of BERT: smaller, faster, cheaper and lightercited by 4,243
  5. 2019XLNet: Generalized Autoregressive Pretraining for Language Understandingcited by 2,801
  6. 2019SuperGLUE: A Stickier Benchmark for General-Purpose Language Understanding Systemscited by 1,718
  7. 2016Understanding intermediate layers using linear classifier probescited by 1,279
  8. 2019BERT Rediscovers the Classical NLP Pipelinecited by 1,046
  9. 2021Probing Classifiers: Promises, Shortcomings, and Advancescited by 831
  10. 2019Designing and Interpreting Probes with Control Taskscited by 608
  11. 2018What you can cram into a single vector: Probing sentence embeddings for linguistic propertiescited by 452
  12. 2019What do you learn from context? Probing for sentence structure in contextualized word representationscited by 436
19 more
  1. 2019Linguistic Knowledge and Transferability of Contextual Representationscited by 382
  2. 2016Fine-grained Analysis of Sentence Embeddings Using Auxiliary Prediction Taskscited by 216
  3. 2019Visualizing and Measuring the Geometry of BERTcited by 215
  4. 2018Colorless green recurrent networks dream hierarchicallycited by 210
  5. 2018Dissecting Contextual Word Embeddings: Architecture and Representationcited by 146
  6. 2018What Is One Grain of Sand in the Desert? Analyzing Individual Neurons in Deep NLP Modelscited by 124
  7. 2020Perturbed Masking: Parameter-free Probing for Analyzing and Interpreting BERTcited by 109
  8. 2020Analyzing Individual Neurons in Pre-trained Language Modelscited by 97
  9. 2021Neuron-level Interpretation of Deep NLP Models: A Surveycited by 81
  10. 2021Conditional probing: measuring usable information beyond a baselinecited by 61
  11. 2021Probing BERT in Hyperbolic Spacescited by 50
  12. 2021Refining Targeted Syntactic Evaluation of Language Modelscited by 41
  13. 2020A Tale of a Probe and a Parsercited by 40
  14. 2018Deep RNNs Encode Soft Hierarchical Syntaxcited by 34
  15. 2019On the Linguistic Representational Power of Neural Machine Translation Modelscited by 34
  16. 2018NeuroX: A Toolkit for Analyzing Individual Neurons in Neural Networkscited by 31
  17. 2021Do Syntactic Probes Probe Syntax? Experiments with Jabberwocky Probingcited by 28
  18. 2020Parsing as Pretrainingcited by 19
  19. 2020On the Branching Bias of Syntax Extracted from Pre-trained Language Modelscited by 3

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2022: 2; 2023: 4; 2024: 3; 2025: 3; 2026: 1.

13 papers.

Sort:

By area:

  1. 2023Physics of Language Models: Part 1, Learning Hierarchical Language Structurescited by 83
  2. 2023Do Transformers Parse while Predicting the Masked Word?cited by 48
  3. 2023Evaluating Neuron Interpretation Methods of NLP Modelscited by 8
  4. 2023Morphosyntactic probing of multilingual BERT modelscited by 8
  5. 2022Probing with Noise: Unpicking the Warp and Weft of Embeddingscited by 3
  6. 2024Metric Learning Encoding Models: A Multivariate Framework for Interpreting Neural Representationscited by 3
  7. 2025The Compositional Architecture of Regret in Large Language Modelscited by 3
  8. 2026Context-Free Recognition with Transformerscited by 3
  9. 2022Reconstruction Probingcited by 2
  10. 2024Topic Aware Probing: From Sentence Length Prediction to Idiom Identification how reliant are Neural Language Models on Topic?cited by 2
  11. 2025Geospatial Mechanistic Interpretability of Large Language Modelscited by 2
  12. 2024What Makes Two Language Models Think Alike?cited by 1
1 more
  1. 2025Cross-Layer Discrete Concept Discovery for Interpreting Language Modelscited by 1