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

arXiv: https://arxiv.org/abs/1909.02597 · HTML: https://mlsanity.com/paper/1909.02597/

## Key facts

- Cited by 58 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 5 Sep 2019.

## Built on by (most cited first, 30 of 58)

- [Winoground: Probing Vision and Language Models for Visio-Linguistic Compositionality](https://mlsanity.com/paper/2204.03162/) (2022)
- [Dynabench: Rethinking Benchmarking in NLP](https://mlsanity.com/paper/2104.14337/) (2021)
- [Masked Language Modeling and the Distributional Hypothesis: Order Word Matters Pre-training for Little](https://mlsanity.com/paper/2104.06644/) (2021)
- [Knowledgeable or Educated Guess? Revisiting Language Models as Knowledge Bases](https://mlsanity.com/paper/2106.09231/) (2021)
- [State-of-the-art generalisation research in NLP: A taxonomy and review](https://mlsanity.com/paper/2210.03050/) (2022)
- [Language Model Behavior: A Comprehensive Survey](https://mlsanity.com/paper/2303.11504/) (2023)
- [Language models are not naysayers: An analysis of language models on negation benchmarks](https://mlsanity.com/paper/2306.08189/) (2023)
- [Structural Persistence in Language Models: Priming as a Window into Abstract Language Representations](https://mlsanity.com/paper/2109.14989/) (2021)
- [CompA: Addressing the Gap in Compositional Reasoning in Audio-Language Models](https://mlsanity.com/paper/2310.08753/) (2023)
- [Turning Tables: Generating Examples from Semi-structured Tables for Endowing Language Models with Reasoning Skills](https://mlsanity.com/paper/2107.07261/) (2021)
- [An Analysis of Negation in Natural Language Understanding Corpora](https://mlsanity.com/paper/2203.08929/) (2022)
- [Filtered Corpus Training (FiCT) Shows that Language Models can Generalize from Indirect Evidence](https://mlsanity.com/paper/2405.15750/) (2024)
- [SBERT studies Meaning Representations: Decomposing Sentence Embeddings into Explainable Semantic Features](https://mlsanity.com/paper/2206.07023/) (2022)
- [Language Models Use Monotonicity to Assess NPI Licensing](https://mlsanity.com/paper/2105.13818/) (2021)
- [Dissociating language and thought in large language models](https://mlsanity.com/paper/2301.06627/) (2023)
- [Do Syntactic Probes Probe Syntax? Experiments with Jabberwocky Probing](https://mlsanity.com/paper/2106.02559/) (2021)
- [ViLMA: A Zero-Shot Benchmark for Linguistic and Temporal Grounding in Video-Language Models](https://mlsanity.com/paper/2311.07022/) (2023)
- [Structural Guidance for Transformer Language Models](https://mlsanity.com/paper/2108.00104/) (2021)
- [RuBLiMP: Russian Benchmark of Linguistic Minimal Pairs](https://mlsanity.com/paper/2406.19232/) (2024)
- [The Life Cycle of Knowledge in Big Language Models: A Survey](https://mlsanity.com/paper/2303.07616/) (2023)
- [Transformers Generalize Linearly](https://mlsanity.com/paper/2109.12036/) (2021)
- [Verb Conjugation in Transformers Is Determined by Linear Encodings of Subject Number](https://mlsanity.com/paper/2310.15151/) (2023)
- [Can Language Models Be Tricked by Language Illusions? Easier with Syntax, Harder with Semantics](https://mlsanity.com/paper/2311.01386/) (2023)
- [Language Modelling as a Multi-Task Problem](https://mlsanity.com/paper/2101.11287/) (2021)
- [Morphosyntactic probing of multilingual BERT models](https://mlsanity.com/paper/2306.06205/) (2023)
- [Different types of syntactic agreement recruit the same units within large language models](https://mlsanity.com/paper/2512.03676/) (2025)
- [KnowledgeVIS: Interpreting Language Models by Comparing Fill-in-the-Blank Prompts](https://mlsanity.com/paper/2403.04758/) (2024)
- [Evaluation Benchmarks for Spanish Sentence Representations](https://mlsanity.com/paper/2204.07571/) (2022)
- [Does Deep Learning Learn to Abstract? A Systematic Probing Framework](https://mlsanity.com/paper/2302.11978/) (2023)
- [Revealing the Unwritten: Visual Investigation of Beam Search Trees to Address Language Model Prompting Challenges](https://mlsanity.com/paper/2310.11252/) (2023)

## 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.