mlsanityData to 7 Oct 2026

Investigating BERT's Knowledge of Language: Five Analysis Methods with NPIs

arXiv 1909.02597 ↗First posted 5 Sep 2019cs.CLcited by 58

Not on the map. Why?

Abstract (arXiv)

Though state-of-the-art sentence representation models can perform tasks requiring significant knowledge of grammar, it is an open question how best to evaluate their grammatical knowledge. We explore five experimental methods inspired by prior work evaluating pretrained sentence representation models. We use a single linguistic phenomenon, negative polarity item (NPI) licensing in English, as a case study for our experiments. NPIs like "any" are grammatical only if they appear in a licensing environment like negation ("Sue doesn't have any cats" vs. "Sue has any cats"). This phenomenon is challenging because of the variety of NPI licensing environments that exist. We introduce an artificially generated dataset that manipulates key features of NPI licensing for the experiments. We find that BERT has significant knowledge of these features, but its success varies widely across different experimental methods. We conclude that a variety of methods is necessary to reveal all relevant aspects of a model's grammatical knowledge in a given domain.

Key facts. Cited by 58 papers in the map (data through 7 Oct 2026). First posted on arXiv 5 Sep 2019.

Builds on

No reference list yet.

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2021: 17; 2022: 11; 2023: 14; 2024: 6; 2025: 4; 2026: 6.

58 papers.

Sort:

By area:

  1. 2022Winoground: Probing Vision and Language Models for Visio-Linguistic Compositionalitycited by 523
  2. 2021Dynabench: Rethinking Benchmarking in NLPcited by 447
  3. 2021Masked Language Modeling and the Distributional Hypothesis: Order Word Matters Pre-training for Littlecited by 176
  4. 2021Knowledgeable or Educated Guess? Revisiting Language Models as Knowledge Basescited by 107
  5. 2022State-of-the-art generalisation research in NLP: A taxonomy and reviewcited by 97
  6. 2023Language Model Behavior: A Comprehensive Surveycited by 92
  7. 2023Language models are not naysayers: An analysis of language models on negation benchmarkscited by 73
  8. 2021Structural Persistence in Language Models: Priming as a Window into Abstract Language Representationscited by 44
  9. 2023CompA: Addressing the Gap in Compositional Reasoning in Audio-Language Modelscited by 43
  10. 2021Turning Tables: Generating Examples from Semi-structured Tables for Endowing Language Models with Reasoning Skillscited by 39
  11. 2022An Analysis of Negation in Natural Language Understanding Corporacited by 39
  12. 2024Filtered Corpus Training (FiCT) Shows that Language Models can Generalize from Indirect Evidencecited by 35
38 more
  1. 2022SBERT studies Meaning Representations: Decomposing Sentence Embeddings into Explainable Semantic Featurescited by 33
  2. 2021Language Models Use Monotonicity to Assess NPI Licensingcited by 30
  3. 2021Do Syntactic Probes Probe Syntax? Experiments with Jabberwocky Probingcited by 28
  4. 2023Dissociating language and thought in large language modelscited by 27
  5. 2023ViLMA: A Zero-Shot Benchmark for Linguistic and Temporal Grounding in Video-Language Modelscited by 26
  6. 2021Structural Guidance for Transformer Language Modelscited by 25
  7. 2023The Life Cycle of Knowledge in Big Language Models: A Surveycited by 23
  8. 2024RuBLiMP: Russian Benchmark of Linguistic Minimal Pairscited by 23
  9. 2021Transformers Generalize Linearlycited by 19
  10. 2023Verb Conjugation in Transformers Is Determined by Linear Encodings of Subject Numbercited by 12
  11. 2023Can Language Models Be Tricked by Language Illusions? Easier with Syntax, Harder with Semanticscited by 11
  12. 2021Language Modelling as a Multi-Task Problemcited by 10
  13. 2023Morphosyntactic probing of multilingual BERT modelscited by 8
  14. 2024KnowledgeVIS: Interpreting Language Models by Comparing Fill-in-the-Blank Promptscited by 7
  15. 2025Different types of syntactic agreement recruit the same units within large language modelscited by 7
  16. 2022Evaluation Benchmarks for Spanish Sentence Representationscited by 6
  17. 2023Does Deep Learning Learn to Abstract? A Systematic Probing Frameworkcited by 6
  18. 2023Revealing the Unwritten: Visual Investigation of Beam Search Trees to Address Language Model Prompting Challengescited by 6
  19. 2023Assessing Step-by-Step Reasoning against Lexical Negation: A Case Study on Syllogismcited by 6
  20. 2021Uncovering Constraint-Based Behavior in Neural Models via Targeted Fine-Tuningcited by 5
  21. 2022Getting BART to Ride the Idiomatic Train: Learning to Represent Idiomatic Expressionscited by 4
  22. 2021Using Integrated Gradients and Constituency Parse Trees to explain Linguistic Acceptability learnt by BERTcited by 3
  23. 2021Analysis and Prediction of NLP Models Via Task Embeddingscited by 3
  24. 2022What do Large Language Models Learn beyond Language?cited by 3
  25. 2022Do Charge Prediction Models Learn Legal Theory?cited by 3
  26. 2023Decouple knowledge from parameters for plug-and-play language modelingcited by 3
  27. 2025FLUKE: A Linguistically-Driven and Task-Agnostic Framework for Robustness Evaluationcited by 3
  28. 2025Let's CONFER: A Dataset for Evaluating Natural Language Inference Models on CONditional InFERence and Presuppositioncited by 3
  29. 2021COMBO: State-of-the-Art Morphosyntactic Analysiscited by 2
  30. 2022DECK: Behavioral Tests to Improve Interpretability and Generalizability of BERT Models Detecting Depression from Textcited by 2
  31. 2022Reconstruction Probingcited by 2
  32. 2024What Makes Language Models Good-enough?cited by 2
  33. 2025Steering Prepositional Phrases in Language Models: A Case of with-headed Adjectival and Adverbial Complements in Gemma-2cited by 2
  34. 2026Modelling the Morphology of Verbal Paradigms: A Case Study in the Tokenization of Turkish and Hebrewcited by 2
  35. 2021Evaluating Contextualized Language Models for Hungariancited by 1
  36. 2023Topics in Contextualised Attention Embeddingscited by 1
  37. 2026The Grammar of Transformers: A Systematic Review of Interpretability Research on Syntactic Knowledge in Language Modelscited by 1
  38. 2026Do Language Models Know Theo Has a Wife? Investigating the Proviso Problemcited by 1

50 most cited of 58 All 58 (JSON)