mlsanityData to 14 Oct 2026

Mission: Impossible Language Models

arXiv 2401.06416 ↗First posted 12 Jan 2024cs.CL cs.AI cs.LGcited by 32

Not on the map. Why?

Abstract (arXiv)

Chomsky and others have very directly claimed that large language models (LLMs) are equally capable of learning languages that are possible and impossible for humans to learn. However, there is very little published experimental evidence to support such a claim. Here, we develop a set of synthetic impossible languages of differing complexity, each designed by systematically altering English data with unnatural word orders and grammar rules. These languages lie on an impossibility continuum: at one end are languages that are inherently impossible, such as random and irreversible shuffles of English words, and on the other, languages that may not be intuitively impossible but are often considered so in linguistics, particularly those with rules based on counting word positions. We report on a wide range of evaluations to assess the capacity of GPT-2 small models to learn these uncontroversially impossible languages, and crucially, we perform these assessments at various stages throughout training to compare the learning process for each language. Our core finding is that GPT-2 struggles to learn impossible languages when compared to English as a control, challenging the core claim. More importantly, we hope our approach opens up a productive line of inquiry in which different LLM architectures are tested on a variety of impossible languages in an effort to learn more about how LLMs can be used as tools for these cognitive and typological investigations.

Key facts. Cited by 32 papers in the map (data through 14 Oct 2026). Builds on 32 of its 76 references that are in the map. First posted on arXiv 12 Jan 2024.

Builds on

32 of 76 references.

References · 32

  1. 2017Attention Is All You Needcited by 60,361
  2. 2018BERT: Pre-training of Deep Bidirectional Transformers for Language Understandingcited by 40,399
  3. 2020Stanza: A Python Natural Language Processing Toolkit for Many Human Languagescited by 723
  4. 2019What do you learn from context? Probing for sentence structure in contextualized word representationscited by 438
  5. 2021Causal Abstractions of Neural Networkscited by 422
  6. 2023The Impact of Positional Encoding on Length Generalization in Transformerscited by 327
  7. 2019Theoretical Limitations of Self-Attention in Neural Sequence Modelscited by 288
  8. 2018Targeted Syntactic Evaluation of Language Modelscited by 242
  9. 2022Neural Networks and the Chomsky Hierarchycited by 217
  10. 2023Finding Alignments Between Interpretable Causal Variables and Distributed Neural Representationscited by 209
  11. 2021Masked Language Modeling and the Distributional Hypothesis: Order Word Matters Pre-training for Littlecited by 177
  12. 2020A Systematic Assessment of Syntactic Generalization in Neural Language Modelscited by 165
20 more
  1. 2019Neural Language Models as Psycholinguistic Subjects: Representations of Syntactic Statecited by 138
  2. 2018On the Practical Computational Power of Finite Precision RNNs for Language Recognitioncited by 117
  3. 2018What do RNN Language Models Learn about Filler-Gap Dependencies?cited by 86
  4. 2020Out of Order: How Important Is The Sequential Order of Words in a Sentence in Natural Language Understanding Tasks?cited by 73
  5. 2020A Formal Hierarchy of RNN Architecturescited by 62
  6. 2023Call for Papers -- The BabyLM Challenge: Sample-efficient pretraining on a developmentally plausible corpuscited by 62
  7. 2020RNNs can generate bounded hierarchical languages with optimal memorycited by 59
  8. 2019Sequential Neural Networks as Automatacited by 54
  9. 2020How Can Self-Attention Networks Recognize Dyck-n Languages?cited by 51
  10. 2023Injecting structural hints: Using language models to study inductive biases in language learningcited by 35
  11. 2022Word Order Does Matter (And Shuffled Language Models Know It)cited by 32
  12. 2019Using Priming to Uncover the Organization of Syntactic Representations in Neural Language Modelscited by 29
  13. 2022Causal Proxy Models for Concept-Based Model Explanationscited by 29
  14. 2022When classifying grammatical role, BERT doesn't care about word order... except when it matterscited by 24
  15. 2023Pushdown Layers: Encoding Recursive Structure in Transformer Language Modelscited by 19
  16. 2018Context-Free Transductions with Neural Stackscited by 14
  17. 2021Syntactic Perturbations Reveal Representational Correlates of Hierarchical Phrase Structure in Pretrained Language Modelscited by 14
  18. 2023What makes a language easy to deep-learn? Deep neural networks and humans similarly benefit from compositional structurecited by 13
  19. 2018Depth-bounding is effective: Improvements and evaluation of unsupervised PCFG inductioncited by 4
  20. 2021Modeling Profanity and Hate Speech in Social Media with Semantic Subspacescited by 4

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2024: 9; 2025: 15; 2026: 8.

32 papers.

Sort:

By area:

  1. 2025How Linguistics Learned to Stop Worrying and Love the Language Modelscited by 51
  2. 2024Surveying the Effects of Quality, Diversity, and Complexity in Synthetic Data From Large Language Modelscited by 35
  3. 2025Generative Multimodal Pretraining with Discrete Diffusion Timestep Tokenscited by 28
  4. 2025Can Language Models Learn Typologically Implausible Languages?cited by 23
  5. 2024Endless Jailbreaks with Bijection Learningcited by 19
  6. 2025Both Direct and Indirect Evidence Contribute to Dative Alternation Preferences in Language Modelscited by 18
  7. 2024gzip Predicts Data-dependent Scaling Lawscited by 16
  8. 2024Language in Vivo vs. in Silico: Size Matters but Larger Language Models Still Do Not Comprehend Language on a Par with Humans Due to Impenetrable Semantic Referencecited by 15
  9. 2024Kallini et al. (2024) do not compare impossible languages with constituency-based onescited by 15
  10. 2024Beyond the Doors of Perception: Vision Transformers Represent Relations Between Objectscited by 13
  11. 2024Incremental Comprehension of Garden-Path Sentences by Large Language Models: Semantic Interpretation, Syntactic Re-Analysis, and Attentioncited by 11
  12. 2025Large Language Models as Proxies for Theories of Human Linguistic Cognitioncited by 10
20 more
  1. 2025Shaping Shared Languages: Human and Large Language Models' Inductive Biases in Emergent Communicationcited by 10
  2. 2024Linguistic Structure from a Bottleneck on Sequential Information Processingcited by 5
  3. 2026Bringing Up a Bilingual BabyLM: Investigating Multilingual Language Acquisition Using Small-Scale Modelscited by 5
  4. 2025Unnatural Languages Are Not Bugs but Features for LLMscited by 4
  5. 2025Reading the unreadable: Creating a dataset of 19th century English newspapers using image-to-text language modelscited by 3
  6. 2025Removing Noise, not Finding Gold: Quality Filtering for Large-Scale Pretrainingcited by 3
  7. 2024Scrambled text: training Language Models to correct OCR errors using synthetic datacited by 2
  8. 2026Baby Scale: Investigating Models Trained on Individual Children's Language Inputcited by 2
  9. 2026Structure Before Collapse: Transient semantic geometry in next-token predictioncited by 2
  10. 2025Probability Consistency in Large Language Models: Theoretical Foundations Meet Empirical Discrepanciescited by 1
  11. 2025Pragmatics beyond humans: meaning, communication, and LLMscited by 0
  12. 2025Language models as tools for investigating the distinction between possible and impossible natural languagescited by 0
  13. 2025Large language models have learned to use languagecited by 0
  14. 2025Large language models are not about natural languagecited by 0
  15. 2025Linguists should learn to love speech-based deep learning modelscited by 0
  16. 2026Are Language Models Models?cited by 0
  17. 2026Large Language Models and Impossible Language Acquisition: "False Promise" or an Overturn of our Current Perspective towards AIcited by 0
  18. 2026Across the Levels of Analysis: Explaining Predictive Processing in Humans Requires More Than Machine-Estimated Probabilitiescited by 0
  19. 2026You Can't Fight in Here! This is BBS!cited by 0
  20. 2026Not-So-Strange Love: Language Models and Generative Linguistic Theories are More Compatible than They Appearcited by 0